Geomagnetism: An Introduction and Overview: M Kono, Tokyo Institute of Technology, Tokyo, Japan

5.
01 Geomagnetism: An Introduction and Overview

M Kono, Tokyo Institute of Technology, Tokyo, Japan
ã 2015 Elsevier B.V. All rights reserved.
5.01.1 Geomagnetism in Perspective 1

5.01.1.1 Early History 1
5.01.1.1.1 Attractive force of the magnets 2
5.01.1.1.2 Early Chinese compasses 2
5.01.1.1.3 Magnetic compass in European documents 4
5.01.1.1.4 Epistola of Petrus Peregrinus 4
5.01.1.1.5 European recognition of declination 5
5.01.1.1.6 Inclination 6
5.01.1.1.7 De Magnete of William Gilbert 7
5.01.1.2 Further Developments in Observations 8
5.01.1.2.1 Secular variation 8
5.01.1.2.2 Short-term variations 8
5.01.1.2.3 Magnetic charts 8
5.01.1.2.4 Measurement of intensity 9
5.01.1.2.5 Magnetic observatories 10
5.01.1.3 Deciphering the Past Using Remanent Magnetization 10
5.01.1.3.1 Early observation of the remanence of rocks 11
5.01.1.3.2 Polarity reversals 11
5.01.1.3.3 Seafloor spreading 13
5.01.1.3.4 Long-term behaviors of the geodynamo 13
5.01.1.4 Descriptions of the Earth’s Magnetic Field 15
5.01.1.4.1 Spherical harmonic analysis 15
5.01.1.4.2 International geomagnetic reference field 15
5.01.1.4.3 Satellite measurements 15
5.01.1.4.4 Geomagnetic spectrum 16
5.01.1.4.5 Inverse problem 18
5.01.1.5 Observational Constraints 19
5.01.1.5.1 Geometric attenuation 20
5.01.1.5.2 Accuracy of data 20
5.01.1.5.3 Signal and noise in magnetic field measurements 20
5.01.1.6 Magnetic Fields of the Planets and Satellites 21
5.01.1.6.1 Example of a flyby observation 22
5.01.1.6.2 Magnetic mapping of planets 22
5.01.1.6.3 Planetary magnetic fields: A comparative study 22
5.01.1.6.4 The magnetic field of the Sun 24
5.01.1.7 Introduction to the Other Chapters 26
5.01.1.7.1 The present field and its components 26
5.01.1.7.2 Data sources and their reliability 27
5.01.1.7.3 Field behaviors in various time windows 28
5.01.1.7.4 Excursions, reversals, and polar wander 29
5.01.1.8 Conclusions 29
Acknowledgments 30
References 30
5.01.1 Geomagnetism in Perspective observation of the geomagnetic field has been carried out only
in the last few centuries, but with indirect measurements, we can
5.01.1.1 Early History
understand the field behavior millions of years back in time. In
The Earth has its own magnetic field (the geomagnetic field), this extended time frame, there is evidence that the polarity of
which is confined by the action of the solar wind into a volume the magnetic field reversed frequently and that the magnetic
called the magnetosphere. This field is not steady, but varies dipole axis in very ancient times was significantly displaced
with time due partly to the interaction with the solar wind, but from the present rotational axis (the north and south geographic
more importantly by its own physical processes. Direct poles).
Treatise on Geophysics, Second Edition http://dx.doi.org/10.1016/B978-0-444-53802-4.00095-6 1

It is of considerable interest how such knowledge was Ancient Chinese people made similar observations, but the
acquired over several centuries. We will take a brief tour of records are somewhat later than the corresponding Greek ones.
the historical events that provided important steps in formu- In 呂氏春秋 Lü Shih Chhun Chhiu (Master Lü’s Spring and
lating our understanding of the geomagnetic field. In doing so, Autumn Annals), written in the late third century BC, it is
we have to rely solely on the written records, which is the said in Chapter 45 that
reason why only the European and Chinese histories are
referred. There are many works on this topic; some of the 慈石召鐵. 或引之也.
important references are Mitchell (1932–46), Harradon The lodestone calls the iron to itself, or attracts it.
(1943–45), Needham (1962), and Yamamoto (山本義隆)
(2003). Among them, Needham’s work is the most compre- After that, references to magnets appear abundantly in the
hensive about Chinese contributions in the early history. Such Chinese literature. Examples of old documents with such
matters as the discovery of the north (or south)-pointing prop- descriptions are 淮南子 Huai Nan Tzu (The Book of Huai
erty of the magnet, the magnetic declination, and the creation Nan) in the first century BC and 論衡 Lun Hêng (Discourses
of the oldest magnetic compass are very thoroughly discussed Weighed in the Balance) in AD 83.
there. In this seminal monograph, Needham convincingly The attractive force that magnets exert on iron was a wonder
concluded that these matters were described in Chinese docu- in ancient times, and it was often attributed to magical power.
ments considerably earlier than in the corresponding Euro- Its full understanding had to wait until the nineteenth century
pean documents. Accordingly, the following descriptions when the magnetic force was explained by physical theorems
related to the Chinese old history are taken from Needham such as Ampère’s and Gauss’s laws in the framework of elec-
(1962). Thus, the English translations of Chinese literature and tromagnetic theory.
the alphabetical expressions of the names of persons and books
are as given in Needham (1962). But since the Chinese original 5.01.1.1.2 Early Chinese compasses
sentences were not given there, they were taken from the The fact that magnets have the property to align in the north–
Japanese translation of this book (Needham (ニーダム), 1977). south direction was discovered by the ancient Chinese. From
When we talk about the earliest recognition of the magne- about the second century AD, there are many Chinese texts
tism of the Earth, we should be careful to discriminate two referring to ‘south-pointing carriage,’ which, in many
separate issues, that is, the attractive force exerted by a magnet instances, were described as guiding the soldiers in thick fogs
on iron and the north (or south)-seeking property of the to the right direction to beat the enemies. Many people
magnet. The former can be taken as the forerunner to the thought that this was a device that used the property of mag-
science of magnetism, while the latter is the basis for appreci- nets. However, it is now considered to be some mechanical
ation of the magnetic field associated with the Earth. Our main device made up of gears and axles rather than an instrument
interest is in the geomagnetic field, but it is necessary to look similar to a magnetic compass (Needham, 1962). A more
into magnets first. interesting description appears in the earlier-mentioned
論衡 Lun Hêng, written by 王充 Wang Chhung in AD 83. In
Chapter 52, it is said that
5.01.1.1.1 Attractive force of the magnets
The earliest observation of the natural magnets (lodestone or 司南之杓、投之於地、其柢指南.
loadstone) is attributed to the Greek philosopher Thales of When the south-controlling ladle is thrown upon the ground, it
Miletus (624–546 BC). Thales did not leave any writings of comes to rest pointing at the south.
his own, but Aristotle (384–332 BC) wrote about him in De
Anima (‘On the soul’) about two centuries later. According to 王振鐸 Wang Chen-To (1948) suggested that the first two
this, Thales taught that the lodestone has a soul, because it letters 司南 south-controlling were changed in the process of
could set another body (iron) in motion. Diogenes Laertius hand copying from the original 指南 south-pointing, the
(c. AD 200–250) also wrote in his work ‘Lives and Opinions of fourth letter 杓 ladle means a spoon worked out from a lode-
Eminent Philosophers’ that Thales admitted that souls exist stone into that shape, and the eighth letter 地 ground actually
even in nonliving matter based on the observation that mag- indicates 式 (or 栻) diviner’s board. Now, a diviner’s board was
nets and ambers can attract things. This suggests that Thales used in ancient China for the purpose of telling fortunes, and it
knew not only about the attractive force of magnets but also is inscribed with the constellation of Great Bear in the center
about the force due to the static electricity of amber, which can and the names of 24 directions on the circle around it. With
be seen when they are rubbed by clothes (Mitchell, 1937). these interpretations, the sentence can be taken to describe an
References to the attractive force of magnets appear quite instrument for seeking south using a magnet! Note that the
often in Greek manuscripts (e.g., Plato, Aristotle, Democritus, Great Bear is the symbol of the north pole and the spoon also
and Lucretius), and there is no doubt that this force was well has a shape reminiscent of its form.
known to the ancient Greeks. This may be because the attrac- Wang (1948) went further to show the credibility of his
tive force appeared to them as a very remarkable phenomenon interpretation, by actually remaking a model of this instru-
since it can act on materials that are not in contact. In these, the ment, with a bronze base plate and a spoon cut from lodestone
magnets were referred to mostly as the rock of Magnesia (liyοB (see Figure 1). A photo of the actual instrument is shown in
magnsi). Magnesia is the name of a place in either Needham (1962). When Needham visited China, he was
Macedonia, Crete, or Asia Minor. The names of magnetism shown by Wang himself the experiment in which the lodestone
and magnetite (Fe3O4) were derived from this Greek word. spoon gradually rotated to the southward direction and settled
Geomagnetism: An Introduction and Overview 3
This is the earliest written record about the magnetic com-

pass using a magnetic (magnetized) needle.
Regarding the oldest compass, the existence of an even
earlier record was pointed out also by Wang (1948). The text
was found in 武經總要 Wu Ching Tsung Yao (Collection of the
Most Important Military Techniques), which is a compendium
of military technology edited by 曾公亮 Tsêng Kung-Liang and
completed in 1044. In Chapter 15, it is said that
若遇天景噎霾. 夜色瞑黒. 又不能辨方向. 則當縦老馬前行. 令識道

路. 或出指南車及指南魚. 以辨所向. 指南車世法不傳. 魚法以薄鐡
葉剪裁. 長二寸闊五分. 首尾鋭如魚形.
When troops encountered gloomy weather or dark nights, and the
directions of space could not be distinguished, they let an old horse
go on before to lead them, or else they made use of the south-
pointing carriage, or the south-pointing fish to identify the direc-
tions. Now the carriage method has not been handed down, but in
the fish method a thin leaf of iron is cut into the shape of a fish two
inches long and half an inch broad, having a pointed head and tail.
After that, how to make this fish is described.
置炭火中焼之. 候通赤. 以鐡鈐. 鈐魚首出火. 以尾正對子位. 蘸水盆

中. 沒尾數分則止. 以密器收之. 用時置水碗於無風處. 平放魚在水面
令浮. 其首常南向午也.
This is then heated in a charcoal fire, and when it has become
thoroughly red-hot, it is taken out by the head with iron tongs and
placed so that its tail points due north. In this position, it is
quenched with water in a basin, so that its tail is submerged for
Figure 1 An ancient Chinese compass with the lodestone cut in the several tenths of an inch. It is then kept in a tightly closed box. To use
shape of a spoon, restored by Wang (1948). Reproduced from Needham it, a small bowl filled with water is set up in a windless place, and the
J (1962) Science and civilisation in China, Physics and Physical fish is laid as flat as possible upon the water-surface so that it floats,
Technology, Part 1 Physics, vol. 4, pp. 434. Cambridge: Cambridge whereupon its head will point south.
University Press.
Apparently, this magnetic pointer (fish) is given a thermo-
remanent magnetization (TRM) by quenching from high tem-
there. Although this effort is very impressive, it is rather doubt- perature and keeping it in the north–south direction. The
ful if the Chinese at this early age really used an instrument that record is very convincing as the description is detailed and
can be identified as the ancestor of the magnetic compass. The correct. This communication can be taken as the first descrip-
interpretation, as suggested by Wang, is not completely con- tion of the magnetic compass of wet type. Figure 2 shows
vincing. Moreover, there is a conspicuous absence of references Wang’s reconstruction of this compass.
to compass-like instruments for about a thousand years The dry pivoted compass was described in 事林廣記 Shih
afterward. Lin Kuang Chi (Guide through the Forest of Affairs), an ency-
A well-known early record about the magnetic compass is clopedia compiled around 1150 by 陳元靚 Chhen Yuan-
in 夢溪筆談 Mêng Chhi Pi Than (Dream Pool Essays) written by Ching. In Chapter 10, Chhen wrote about the wet compass:
沈括 Shen Kua at about AD 1088. In this book, it is said in
Chapter 24 that 以木刻魚子位一个. 如拇指大. 開腹一竅. 陥好磁石一塊子. 却以﨟塡
滿. 用針一半. 僉従魚子口中鈎入令沒. 放水中. 自然指南. 以手撥轉.
又復如初.
方家以磁石磨針鋒. 則能指南. 然常微偏東. 不全南也.
They (the magicians) cut a piece of wood into the shape of a fish, as
Magicians rub the point of a needle with the lodestone; then, it is
big as one’s thumb, and make a hole in its belly, into which they
able to point to the south. But it always inclines slightly to the east
neatly fit a piece of lodestone, filling up the cavity with wax. Into this
and does not point directly at the south.
was a needle bent like a hook is fixed. Then when the fish is put in
the water it will of its nature point to the south, and if it is moved
The text further explains how to make a magnetic needle by with the finger it will return again to its original position.
rubbing a needle with the lodestone, its south-seeking prop-
erty, and moreover the fact that there is a slight difference But in this case, a lodestone magnet is inserted in a wooden
between the true south and its pointing direction (i.e., the fish that is put in the water. After that, Chhen continues to
first mention of the declination). Shen Kua further says that describe a different compass:
以芥子許蝋綴於針腰. 無風處懸之. 則針常指南. 以木刻龜子一个. 一如前法製造. 但於尾邊敲針入去. 用小板子. 上安

It is best to suspend it by a single cocoon fiber of new silk attached 以竹針子如箸尾. 大龜腹下. 微陥一穴. 安針子上. 撥轉常指北. 須是
to the centre of the needle by a piece of wax the size of a 釘尾後.
mustard-seed – then, hanging in a windless place, it will always They also cut a piece of wood into the shape of a turtle, and arrange
point to the south. it in the same way as before, only that the needle is fixed at the tail
Nature of Things) written about 1190, he described the use of

the magnetic needle in navigation to indicate north and that
the needle is put on a pivot that may be the form of a primitive
compass (Mitchell, 1932). Guyot de Provins of France
(1184–1210) wrote a poem called La Bible around 1205, in
which he described a floating compass. Jacques de Vitry of the
Kingdom of Jerusalem (1165–1240) left a similar document
(c. 1218). These people were all monks or priests, and they
only referred to the compass as having the noble property
(to point always to the same direction). It is therefore natural
to think that the properties of the compass were known to
mariners well before it became popular so that the priests
could use it for allegory in these writings (Mitchell, 1932).
5.01.1.1.4 Epistola of Petrus Peregrinus

Petrus Peregrinus (Roman name for Pierre Pèlerin, active in the
thirteenth century) wrote Epistola (Epistola Petri Peregrini de Mar-
icourt ad Sygerum de Foucaucourt, militem, de magnete; Letter of
Peter Peregrinus of Maricourt to Sygerus of Foucaucourt, Soldier,
on the Magnet) in 1269, while he took part in the siege of the
southern Italian town of Lucera, by the army of Charles, the
Count of Anjou. Maricourt and Foucaucourt are names of towns
in Picardy, France. It has the form of a letter (Epistola) to a soldier
called Sygerus of Foucaucourt. Although it was not published in
printed form until 1558, many hand copies were circulated
widely in western Europe in medieval times. In this booklet
(Figure 4), Peregrinus explained various properties of the mag-
Figure 2 Illustration of a Chinese fish compass by Wang (1948). net (lodestone) based on experimental observations. In fact, this
Reproduced from Needham J (1962) Science and civilisation in China, can be regarded as the first scientific treatise describing observa-
Physics and Physical Technology, Part 1 Physics, vol. 4, pp. 434. tions and experiments carried out for the purpose of clarifying a
Cambridge: Cambridge University Press. natural phenomenon. The conclusions were derived logically
based on observations and experiments.
end. A bamboo pin about as thick as the end of a chopstick is set up
The Epistola is composed of two parts. The first part is made
on a small board, and sustains the turtle by the concave under- up of ten chapters, in which the properties of magnets are
surface of its body, where there is a small hole. Then when the turtle discussed. The second part covers the technical use of the
is rotated, it will always point to the north, which must be due to the magnets, such as the construction of a magnetic compass.
needle at the tail.
The second part also contains a discussion of a perpetual
motion machine using magnets. This is certainly invalid from
In this compass, a wooden turtle has a tail made of a
the present-day knowledge of physics, but it cannot be blamed
magnetic needle. A thin bamboo stick stands from the base-
as an error, because it was written long before the concepts of
board and holds the turtle at the hole made in its belly. The
thermodynamics or energy conservation were formed.
turtle rotates and points to the north because of the magnetic
All of the material written in Epistola may not have been
needle (Figure 3).
discovered by Peregrinus himself (Harradon, 1943), but this
does not decrease the importance of Epistola as the first scien-
tific paper in human history. The most important properties of
5.01.1.1.3 Magnetic compass in European documents
magnets described by Peregrinus were as follows:
It is not clear when the knowledge of the magnetic compass
reached Europe and when it was first used in navigation. 1. Finding the two magnetic poles: To show that a magnet has
Gilbert wrote that it was brought to Europe by the Venetian two poles, grind and polish the magnet into a spherical
Marco Polo, but there is evidence that the compass was used shape. Next, put a small needle-shaped iron on this sphere
well before his return to Europe in 1295. It is often thought and write a line along its direction dividing the sphere into
that the knowledge of the compass came from China through two equal parts (this defines a magnetic meridian). Move
the intermediary of the Islam civilization. There is no written the needle to a different position and write the second
evidence, however, and the appearance of the compass is ear- division line. Peregrinus noted that, even if the previously
lier in European documents than in Islamic ones (Mitchell, mentioned process is repeated at many places, all the lines
1932; Needham, 1962). meet at the same point on the sphere. The intersection of
The earliest record of the north- and south-seeking property great circles defines two poles, the north and south mag-
of the compass in Europe appears to be that of Alexander netic poles.
Neckam (1157–1217) of St. Albans, England. In two treatises, 2. Determination of the polarity of the poles: To determine which
De Utensibibus (On Instruments) and De Rerum Naturis (On the is the north pole and which is the south pole, place a
Figure 3 Illustration of a Chinese turtle compass by Wang (1948). Clockwise from top left: plan view, lengthwise section, transverse section, and
side view. Reproduced from Needham J (1962) Science and civilisation in China, Physics and Physical Technology, Part 1 Physics, vol. 4, 434 pp.
Cambridge: Cambridge University Press.
magnet on a wooden plate that floats on the water surface (instead of a magnetized needle). In addition, the instrument
in a large-enough container. After some time, the magnet has a ruler with a pin at each end. To determine the direction of
settles into a north–south direction, and thus, the north the Sun or the Moon, the ruler is rotated until the shadow of a
and south magnetic poles can be determined. This method pin falls along its length. Peregrinus also showed another
is quite similar to the method of using a natural magnet as a instrument in which an iron needle magnetized by a magnet
compass (to put on water a wooden fish containing a is used in place of the magnet in this figure.
natural magnet) in Chinese documents (see Figure 2). Per- Many of the previously mentioned findings have usually
egrinus concluded that the magnet rotates so that the two been attributed to Gilbert. For instance, points 1, 2, and 4 are
poles are in the same direction as the celestial poles. described very similarly to the description in Epistola but in
3. Forces between two magnetic poles: Using two magnets with more detail in Part I, Chapters 3–5 of De Magnete (Gilbert,
poles marked as in the preceding text, one magnet floating 1600). As an example, Figure 6 shows an illustration from
on the water and another held by a hand, it can be shown De Magnete that describes point 4: cut a magnet in half, and
that the two poles attract each other if the S pole of the the poles of opposite polarities appear at the new edges.
second is brought near the N pole of the first, or vice versa. A rather curious fact is that Gilbert did not refer to Peregrinus
On the other hand, they repel each other if the two poles are in these descriptions. The neglect of the contribution of Pere-
of the same polarity (N to N or S to S). grinus in later years is perhaps caused by the popularity of
4. A magnetic pole cannot be isolated: To show this, Peregrinus Gilbert’s work and the lack of proper citation in it.
described an experiment of cutting the magnet into two
halves. Then, new poles appear at the cut end. The polarities
5.01.1.1.5 European recognition of declination
of these new poles are opposite to the one at the other end
Before Needham (1962) showed the precedence of Chinese
of the cut pieces, which can be seen by repeating the third
discovery, it was widely believed that Christopher Columbus
experiment with halved magnets. This is the first realization
discovered the declination on his first journey to the West
of the dipolar nature of the magnet and the fact that
Indies in 1492. According to the journal of the ship, quoted
monopoles do not exist.
in a book by Las Casas and in the biography of Columbus by
Figure 5 is the illustration of an instrument to determine his son Fernando, magnetic direction changed ‘from northwest
the azimuth of the Sun and others shown by Peregrinus. to northeast’ on September 13. However, Mitchell (1937)
A floating magnet (Magnes) determines the north direction. argued convincingly that the declination must have been
This part is a magnetic compass using a natural magnet known in Europe at the time of the voyages of Columbus.
Figure 6 Experiment of cutting lodestone to show the appearance of

magnetic pole at the new edges as described in De Magnete.
to the east of true north, which shows that the declination in

west Central Europe at that time was compensated in the latter
compass. From the speed of information transfer in those
times, it is deemed not probable that this knowledge is the
outcome of the first journey of Columbus. The artisans of
Flanders must have known the difference between true north
and magnetic north for some time.
It remains true, however, that the description by Columbus
is the first scientific record of declination measurement and
that important discoveries were made of (1) the variability of
declination from place to place and (2) the position of the
agonic line (the line connecting places where D ¼ 0 ) at the end
of the fifteenth century.
Figure 4 The title page of Epistola.
5.01.1.1.6 Inclination
The fact that the magnetic needle deflects not only from the
horizontal north–south direction but also from the horizontal
plane was first recognized by Georg Hartmann and Robert
Norman. Hartmann’s findings were written only briefly in a
letter (1544) that was mostly unknown until it was rediscov-
ered in the nineteenth century, and his value of the inclination
(9 ) is too shallow for Europe at that time. On the other hand,
Norman published his measurement of the inclination and
discussion of the results in 1581 as The Newe Attractive (The
Newe Attractive, Containyng a short discourse of the Magnes or
Lodestone: and amongst other his vertues of a newe discovered secret
and subtill propertie, concernyng the Declinyng of the Needle,
touched therewith under the plaine of the Horizon. Now first founde
out by Robert Norman Hydrographer).
Norman knew about the magnetic devices quite well since
he was a sailor for 20 years, and his later job was the construc-
tion and sales of navigational instruments. He was aware of the
tendency of the magnetic needle to make some angle with the
horizontal plane from his vast experience in making magnetic
compasses. He usually corrected this by putting a small weight
Figure 5 The compass using a magnet shown in Epistola. on the needle, but at some time, he decided to study this
phenomenon in more detail. He constructed a special instru-
In the biography written by Fernando, it is said that on the ment to measure the deflection from the horizontal plane by
second voyage (1493), Columbus carried compasses from preparing a magnetic needle from carefully balanced iron and
Genoa and from Flanders. In the Genoan type, the magnetic making it freely rotate around the horizontal pivot (Figure 7).
needle was fixed in the N–S direction of the compass card, but This is the first dip circle, and similar instruments were used in
in the Flemish type, the needle was fixed at one point (11.25) observatories until very recent times. By repeating experiments,
90 80
70
60
50
DLE
40
NEE
30
20
10
THE HORIZON
10
20
THE
30
40
50
0
70 6
90 80
Figure 8 The title page of De Magnete.

Figure 7 The dip circle of Robert Norman.
he found out that this angle (the inclination) was 71 500 in
London. As the title of the book indicates, Norman not only
carried out the measurement but also gave discussions about
the nature of this force. In it, he denied the old idea that there
are attractive points in heaven (such as the celestial poles of
Petrus Peregrinus) or in the Earth (the magnetic mountain in
the Arctic, etc.), which is important in the development of the
notion of remotely acting forces (Yamamoto, 2003).
Figure 9 Illustration of versorium, a pivoted iron needle used by
Gilbert to measure the direction of the field around terella.
5.01.1.1.7 De Magnete of William Gilbert

William Gilbert (1544–1603) was a well-established medical
about magnetism. His most important contributions may be
doctor and was the president of the Royal College of Physi-
summarized as follows:
cians. He was famous in his time because he became the
physician to Queen Elizabeth for the period 1601–03, but he 1. Magnets and iron are the same: This may sound a bit queer,
is now considered as one of the founders of the physical but lodestone and iron were considered to belong to differ-
sciences by the publication of De Magnete (De Magnete, magne- ent kinds of matter (‘rock’ and ‘metal’) from the time of
ticisque corporibus, et de magno magnete tellure; Physiologia nova, Aristotle. Gilbert concluded that they are the same because
plurimis & argumentis, & experrimentis demonstrata; On magnets, most of the best magnets were found in an iron mine,
magnetic materials, and the Earth as a large magnet; New magnets heated in the furnace produced good-quality
natural philosophy proved by arguments and experiments). iron, etc.
This book (Figure 8) had a profound influence in Europe 2. Similarity of the spherical magnet and the Earth: Gilbert
and prepared for Newton’s theory of gravity by establishing formed a lodestone into a sphere and called it terella. He
the fact that the magnetic force is a remotely working interac- used a small magnetic needle suspended at the pivot
tion (Yamamoto, 2003). Gilbert also founded the science (versorium; see Figure 9) to find the direction of magnetic
of electricity by the experiments described in De Magnete. force on this sphere. Like Peregrinus, he identified poles,
Here, we will only discuss what was made clear by him meridians, and the equator on the sphere. From the results
of this model experiment, he concluded that the Earth is a large error in Borough’s measurement. Henry Gellibrand
magnetic body and thus a magnet (Globus terrae est mag- (1597–1636), Gunter’s successor at Gresham College, went
neticus & magnes). back in 1635 to Deptford, repeated the measurement, and
3. Inclination is determined by the (magnetic) latitude: Gilbert found a value of about 4 E. This marks the discovery of secular
argued that if terellas are made from lodestones of different variation.
strength, the inclination might be larger in one because of
the stronger attraction of the magnet, yet the actual inclina- 5.01.1.2.2 Short-term variations
tions were the same. This observation strengthened his con- George Graham (1673–1751), a London clockmaker, observed
clusion that the Earth is a large spherical magnet (Figure 10). declination very frequently in 1722–23 using a very sensitive
(12 in.) compass. He discovered that the declination changed
His experiment using a spherical magnet was very similar to
even in 1 day, and sometimes, the magnitude of change reached
the one described by Peregrinus. However, the conclusion was
up to half a degree. Following this, Anders Celsius and Olof
quite different: Peregrinus found correspondence between the
Hjorter performed the same experiments in Uppsala in
spherical magnet and the celestial globe. Gilbert, on the other
1740–47. They not only confirmed Graham’s results but also
hand, concluded that it is the same as the Earth, or the Earth is
found out that the activity of the northern lights (aurora bor-
a spherical magnet. The way to reach this idea based on exper-
ealis) was accompanied by a large change in declination. Later,
iments is not quite perfect in the standards of present-day
they cooperated with Graham in London to make simultaneous
science, but Gilbert reached the correct conclusion for the
measurements and found out that large disturbances occurred at
first time that the Earth has a dipole field that is aligned
the same time in two places on 5 April 1741. These are the
roughly parallel to the rotational axis.
discoveries of magnetic storms and solar quiet-day variations.
5.01.1.2.3 Magnetic charts

5.01.1.2 Further Developments in Observations
After the voyage of Columbus, the practical utility of magnetic
Isaac Newton (1642–1727) published his Philosophiae Natur- measurements was recognized because of their importance in
alis Principia Mathematica (Mathematical Principles of Natural navigation. A precise knowledge of the position of the ship and
Philosophy) in 1668, which firmly established modern phys- the direction it is heading is very important for navigation.
ics. It is clear that Newton’s work received a big influence from The latitude part of the position and direction is relatively
Gilbert’s De Magnete, especially the idea of the forces between easy. The angle between the horizon and the pole star (which
objects that are not in contact (Yamamoto, 2003). Geomagne- is the astronomical latitude if the diurnal motion of the pole
tism as a physical science took shape at about the same time, star is adjusted) can be measured with enough accuracy using
and it is remarkable that progress was made through observa- sextants. The pole star shows the north direction at night, and
tions even from these old times. in the daytime, the mean of the two measurements of the Sun’s
azimuth gives the south direction, if they are obtained at times
5.01.1.2.1 Secular variation when the Sun’s height is the same (e.g., at the sunrise and
In October 1580, William Borough (1536–99), who later sunset).
served as a naval officer under Captain Drake, measured the On the other hand, the determination of longitude is quite
declination at Limehouse in London and recorded that it was difficult. If a clock can keep very accurate time after the ship has
11½ E. Edmund Gunter (1581–1626), professor of astron- left a port, the difference between the clock time and the local
omy at Gresham College, measured the declination at Dept- time as determined from the position of the Sun (e.g., 12:00
ford and Limehouse in June 1622 and obtained the values of noon for southing) gives the longitude difference between the
6 150 and 5 560 E (Limehouse value was the mean of eight current ship’s position and the port. But the needed accuracy
observations). Gunter pointed out that temporal change was could not be achieved by the clocks of those times. Since it was
very likely, but he was cautious to admit the possibility of a found by the voyages of Columbus that the declination chan-
ged from NE to NW as the ship crossed the Atlantic from east to
west, it was considered that the longitude may have a simple
relation with the magnetic declination.
At the end of the seventeenth century, Edmund Halley
(1656–1742) organized an expedition devoted to the determi-
nation of the declination in various parts of the Atlantic. He
thus published in 1701 the first isogonic map of the Atlantic
(General Chart of the Variation of the Compass (1701)). This was
the first magnetic chart, and its importance in the study of
geomagnetism cannot be overstated. Ironically, however, this
map did not solve the problem of the longitude, because of the
unlucky coincidence that the isogonic lines (the traces on
which the declination is equal) ran mostly east–west in the
northern Atlantic at that time (Figure 11).
For further discussion of magnetic data obtained by the
measurements at sea and their use in the study of geomagne-
Figure 10 Gilbert’s terella, or the spherical magnet. tism, see Chapter 5.12.
Figure 11 Halley’s 1701 Atlantic chart showing the lines of equal declination.
5.01.1.2.4 Measurement of intensity strength of magnetic field increased. This property was used
The determination of the intensity of the magnetic field started by the explorer Alexander von Humboldt (1769–1859), who
much later than that of the direction. First came relative mea- measured the oscillation period of a dip circle at many places
surements. It was noticed that if the needle of a dip circle was while he traveled in South America with Aimè Bonpland in
displaced from its rest position, the needle oscillated about the 1798. He could show that the intensity systematically
equilibrium position with a period that decreased as the increased as he went further south from the magnetic
equator, which was near Cajamarca in Peru. Actually, the Next, near another suspended magnet (B in Figure 12(b)),
period of oscillation (T ) depends not only on the strength of he placed the magnet (A) in the direction perpendicular to the
the magnetic field (H) but also on the moment of inertia (I) magnetic meridian. The magnetic field at the suspended mag-
and magnetic moment (M) of the magnetic needle: net is rotated by an angle H0 /H, where H0 is the magnetic field
pffiffiffiffiffiffiffiffiffiffiffiffiffi due to the magnet:
T ¼ 2p I=MH [1] " #

M 3 l 2
Thus, it is difficult to assume that the change in the period is H0 ¼ 3 1 + [2]
d 8 d
entirely due to the change in the field strength and not due to
changes in the property of the magnet. where d is the distance between the centers of the two magnets
Carl Friedrich Gauss (1777–1855) devised a method of and l is the length of the magnet with the magnetic moment M.
absolute intensity determination in 1832. His method was to Observing the angular deflection for different values of d and
use a suspended magnet (A in Figure 12(a)) and make it again utilizing the method of least squares, it is possible to
oscillate in the horizontal plane on both sides of the magnetic obtain M/H. With the values of MH and M/H, Gauss could
meridian. As the moment of inertia of the magnet is not determine the horizontal field intensity in G€ ottingen as
known, he added weights (W) to the magnet so that the 17.8 mT. Gauss’s method of intensity determination was
moment of inertia can be changed by a precisely known quickly taken up by other people, such as Hansteen and
amount. By observing the period T for various increments in Sabine, for measurements in other parts of the world. Present
the moment of inertia, it is possible to determine the product knowledge of the intensity of the magnetic field is discussed in
MH (through least-squares method, which is another inven- Chapter 5.13.
tion of Gauss).
5.01.1.2.5 Magnetic observatories
More or less continuous observations of D started in London
about 1652 and in Paris in 1663. By the time of Gauss, there
were already a number of such observatories, and it was nec-
T
essary to coordinate them to utilize the magnetic data. In the
F 1830s, Gauss and Humboldt organized the G€ ottinger Magne-
C
tischer Verein (Gottingen Geomagnetic Union), which can be
S regarded as the first international geophysical organization.
About 50 observatories (among them 15 were outside Europe)
M took part in this collaboration to carry out standardized
observations, for example, daily observation at the assigned
W
time, and intensive observations of D on the six selected days
per year.
R
A
W 5.01.1.3 Deciphering the Past Using Remanent
(a) Magnetization
Geomagnetism has a unique position in all the Earth sciences
because it can go back in time and treat changes that occurred
T
over millions of years or an even longer timescales. In other
F branches of geophysics, there are some efforts to go back in
history, such as paleoseismology based on the estimation of
S past movements of active faults uncovered by trench excava-
tion. Geology is in a sense a study of all the processes operating
M on the Earth since its formation. However, in all of these
studies, the quality of data is much inferior to those that are
obtained for the present observations. There are many reasons
for this. For instance, the amount of information used in the
B analysis of earthquakes today (e.g., waveforms recorded on
seismograms at the observatories) is many orders of magnitude
A larger than the information from active fault traces. Geologic
processes themselves modify the original information into
quite different forms (e.g., the fossils are different from the
living organisms and irreversible processes such as metasoma-
(b) tism change the properties of sediments).
Magnetic field itself cannot be preserved over time, which is
Figure 12 (a, b) Principle of Gauss’s method of intensity determination. similar to the case of seismic waves. Only the remanent magne-
Modified from Malin SRC (1987) Historical introduction to tization in rocks and other materials may survive, and they are
geomagnetism. In: Jacobs JA (ed.) Geomagnetism, vol. 1, pp. 1–49. also subject to geologic processes that can destroy or change
them. Nevertheless, the relation between the magnetic field (B) 5.01.1.3.2 Polarity reversals
and the remanent magnetization (M) acquired in that field is David (1904) and Brunhes (1906) reported the discovery of
very straightforward; their directions are usually parallel and NRM, which was roughly antiparallel to the present field direc-
their intensities are more or less proportional to each other tion at the sampled locality. They found that the directions of
with the proportionality constant determined by the magnetic NRMs in baked clays and in overlying lava flows that heated
properties of rocks and minerals. Remanent magnetizations can them were essentially the same. These observations strongly
survive some of the changes that affect rocks, for example, suggested that the Earth’s magnetic field had reversed in the
compaction, consolidation, and sometimes even metasomatism past. Matuyama (1929) studied Quaternary and late Tertiary
in sediments. The direct relation between B and M and the volcanic rocks in Japan, Korea, and Manchuria (NE China). He
stability of magnetization over millions of years make it possible found both normally and reversely magnetized rocks but
to study the magnetic field in the remote past with some confi- noted that the reversely magnetized rocks were older than
dence. The properties of rocks and minerals that are the basis of normally magnetized rocks, in all the localities (Figure 13).
magnetic memories are discussed in Chapter 5.08. Here, we will Thus, he concluded that the Earth’s magnetic field was reversed
discuss the major developments in paleomagnetism. in the early Quaternary. A similar comparison was attempted
earlier by Mercanton (1926) on rocks from widely separated
5.01.1.3.1 Early observation of the remanence of rocks areas such as Greenland, Iceland, and Scotland, but he could
The fact that some rocks carry strong magnetization was known not draw a firm conclusion because the inferred ages of the
for a few hundred years prior to the actual measurements of rocks were too widely scattered.
their remanent magnetization because a magnetic compass After World War II, the study of paleomagnetism was greatly
needle is sometimes deflected near such rocks. Humboldt intensified in many countries. In a few places, the measurements
attributed these effects to lightning strikes. The first measure- of NRMs were done in an effort to understand changes in a
ments of the natural remanent magnetization (NRM) carried by sequence of lava flows or sediments. Roche (1951) concluded
rocks were done by people like Achille Delesse (1817–81) and that the last reversal occurred in the Early Pleistocene based on
Macedonio Melloni (1796–1854) around 1850. These workers his study of lavas of Massif Central in France. Hospers (1953)
noted that some of the rocks carry magnetization that is parallel suggested that it is possible to use the changes in the NRM
to the Earth’s magnetic field. Folgerhaiter (1899) studied the polarity for stratigraphic correlation in Icelandic lavas.
magnetizations in bricks and potteries and suggested that the Khramov (1958) studied sedimentary sequences in Turkmeni-
record of direction changes of the Earth’s magnetic field may be stan and proposed the possibility of building a reversal timescale
obtained by measurements of such objects, knowing the place that could be applied to the whole world.
and time of firing at a kiln. This opened up the possibility of Even with these efforts, it was not quite obvious at that time
archaeomagnetism. The current knowledge in centennial and if the Earth’s magnetic field really reversed its polarity in the
millennial field changes is discussed in Chapter 5.09. past. There were speculations that magnetization itself could
-90
-60 -60
-30 -30
W 60 30 N 30 60 E 120 150 S 150 120 W
30 30
60 60
90
First group Second group
Present field Genbudo
Yakuno Zaitokusan
Fushun Other sites
Other sites
Figure 13 The magnetic directions of basalts in Japan, Korea, and NE China observed by Matuyama (1929). Vertical and horizontal directions
correspond to the inclination and declination. Reproduced from Matuyama M (1929) On the direction of magnetization of basalt in Japan, Tyosen and
Manchuria. Proceedings of the Imperial Academy Japan 5: 203–205.
orient in the direction opposite to the magnetic field. Néel occurred at the same time at widely separated places, such as
(1951) showed how self-reversal in the acquisition of TRM Alaska, California, Iceland, and Hawaii (Cox et al., 1963;
could occur. Soon afterward, self-reversal of artificially induced McDougall and Tarling, 1963). Allan Cox and his colleagues
TRM was actually discovered in a dacite lava from Mount developed a convention of calling the long and short intervals
Haruna, Japan (Nagata et al., 1951). The controversy about of the same polarity, ‘polarity epochs’ and ‘polarity events,’
field reversal versus self-reversal continued for several years, respectively (these were replaced by chrons and subchrons as
because it was difficult to settle one way or another. There was the internationally defined nomenclature later). With the dis-
even a suggestion that some self-reversal mechanisms cannot covery of the Jaramillo event in the Brunhes epoch (Doell and
be reproduced in the laboratory because of the long length of Dalrymple, 1966), the polarity timescale for the last 4 million
time involved (Verhoogen, 1956). Experimentally, a strong years was essentially completed (Figure 14). There is no doubt
correlation between the oxidation state of magnetic minerals that before the present normal polarity interval (Brunhes
and the NRM polarity was found in Scottish and other lava chron, 0–0.8 Ma), the field was reversed for more than 1 My
flows (Ade-Hall et al., 1968). (Matuyama chron, 0.8–2.6 Ma) with occasional short episodes
The reality of field reversal was finally established by show- of normal polarity, and before that, another period was dom-
ing the synchronicity of reversals over the world. Two groups, inated by normal polarity (Gauss chron). The study of polarity
one at the US Geological Survey, Menlo Park, and the other at reversals was vigorously continued to the present time, and
the Australian National University, combined K–Ar dating increasingly complex details are uncovered as described in
with paleomagnetism and showed that the reversals indeed Chapters 5.10 and 5.12.
K-Ar age (106 years)

0 1.0 2.0 3.0 4.0
JUNE 15, 1963 COX, DOELL, and DARYMPLE
OCT. 5, 1963 McDOUGALL and TARLING
OCT. 18, 1963 COX, DOELL and DALRYMPLE
FEB. 1964 EVERDEN, SAVAGE, CURTIS, and JAMES
APRIL 11, 1964 McDOUGALL and TARLING
JUNE 26, 1964 COX, DOELL, and DALRYMPLE
JAN. 15, 1966 DOELL, DALRYMPLE, and COX
MAY 20, 1966 DOELL and DALRYMPLE
JUNE 18, 1966 CHAMALAUN and MCDOUGALL
JULY 1966 MCDOUGALL and WENSINK
SEPT. 1966 COX, HOPKINS, and DALRYMPLE
DEC. 15, 1966 MCDOUGALL , ALLSOPP, and CHAMALAUN
DEC. 24, 1966 MC DOUGALL and CHAMALAUN
MAY 15, 1967 COX and DALRYMPLE
MAY 1967 COX and DALRYMPLE
OCT. 31, 1988 COX, DOELL, and DALRYMPLE
JAN. 1969 COX

Laschamp event Gilsa event Kaena event Cochili event
Jarmillo event Olduvai event(s) Mammoth event Nunivak event
Brunhes Gauss
Matuyama Gilbert
normal normal
reversed epoch reversed epoch
epoch epoch
Figure 14 Early development of the geomagnetic polarity timescale. Reproduced from Dalrymple GB (1972) Potassium-argon dating of geomagnetic
reversals and North American glaciations. In: Bishop WW and Miller JA (eds.) Calibration of Hominid Evolution, pp. 107–134. Edinburgh: Scottish
Academic Press.
5.01.1.3.3 Seafloor spreading Juan de Fuca Ridge 46⬚ N

In the 1950s, the biggest topic in paleomagnetism was conti- Profile reversed
nental drift rather than polarity reversals. Two groups from
England led by Keith Runcorn and Patrick Blackett showed
that paleomagnetic data from different continents indicated
that large-scale movement of landmasses occurred in the past
(Blackett et al., 1960; Runcorn, 1962). This revived the once-
rejected theory of continental drift by Wegener (1929). How-
ever, acceptance of continental drift was not immediate in
many parts of the geophysics community, because of scant
evidence. The reality and the importance of geomagnetic polar-
ity reversals were not enthusiastically received until later.
The situation changed dramatically in the few years
between 1963 and 1966. During this time, the combined
magnetic–geochronological data of Cox and others became 500 γ
compelling. However, the decisive step was put forward by
Vine and Matthews (1963) by their theory of tape recorder-
like acquisition of magnetization at the mid-ocean ridges.
Based on the measurement of magnetic anomalies across the
Mid-Atlantic Ridge and Indian Ocean ridge and the seafloor
spreading hypothesis of Hess (1962) and Dietz (1961), they Model 2.9 cm year−1
proposed that the oceanic crust acquires normal or reversed
remanence depending on the polarity of the magnetic field at
that time and that the crust moves away from the ridges,
causing the alternating normal–reverse magnetic anomalies.
Vine (1966) further gave very convincing evidence of the real-
ity of the Vine–Matthews theory (Figure 15).
Subsequently, the plate tectonics model was pioneered by
Morgan (1968), McKenzie and Parker (1967), and Le Pichon
(1968). The directions of current plate movements were 100 0 100 km
inferred from the source mechanisms of earthquakes occurr- S.L.
3.3
ing at the plate boundaries and the trends of the transform 5 km
faults separating two plates. Also, the past plate motion was
Figure 15 The magnetic anomaly across the Juan de Fuca Ridge as
strongly constrained by the magnetic anomalies through the
interpreted by normal–reverse magnetization in the oceanic crust.
Vine–Matthews mechanism. It was this definitive property
Reproduced from Vine FJ (1966) Spreading of the ocean floor: New
of the magnetic anomaly data that was one decisive factor in evidence. Science 154: 1405–1415, with permission. Copyright 1966
the quick establishment of the plate tectonics (compared AAAS.
with, for instance, the continental drift theory). For the inter-
play between ancient tectonic motions and the magnetic
record.
magnetic anomalies were absent. Later, Larson and Pitman
(1972) succeeded in establishing the polarity timescale for
5.01.1.3.4 Long-term behaviors of the geodynamo the parts of older oceans beyond the quiet zone. With Ceno-
Geomagnetism provided a very useful tool for recovering past zoic and Mesozoic timescales, it is possible to determine the
plate motions through the analysis of oceanic magnetic anom- age of the all ocean floors (Figure 16).
alies. There was a great reward for this service; the resulting The polarity timescale has been refined and updated many
intense study of seafloor spreading in the world oceans led to times. The current standard timescale is the one presented by
the establishment of the polarity reversal timescale. Cande and Kent (1992, 1995). Polarity timescales are the
When plate tectonics was initiated, the polarity timescale source of very important information about the nature of the
was available only for the last 4 My from the paleomagnetic geomagnetic field. The quiet zone is the part of the ocean
and geochronological studies of the volcanic rocks on land. where magnetic anomalies are absent because polarity rever-
Following Vine’s first attempt to extend this timescale to tens of sals did not occur. The long period without reversals is called
millions years in the past (Vine, 1966), Heirtzler et al. (1968) the Cretaceous Long Normal (CLN), and it is speculated that
produced a timescale that covered all the Cenozoic Era, based the core dynamo state was significantly different at that time
on the comparison of the magnetic anomaly patterns in the from both before (Jurassic and Early Cretaceous) and after that
Atlantic, Pacific, and Indian Oceans. This was a tremendous time (Tertiary and Quaternary).
step for exploring the evolution of oceanic plates, because it From the analysis of the polarity timescale, it was con-
provided a way to date the ocean floor, relying mostly on the cluded that reversals occur without memories of the past
pattern of magnetic anomalies. The reversal timescale ended (Poisson process). The reversal process is not steady; the prob-
abruptly at about 80 Ma due to the ‘magnetic quiet zone’ where ability of a reversal increases monotonically from the end of
0⬚ 30⬚ 60⬚ 90⬚ 120⬚ 150⬚ 180⬚ 210⬚ 240⬚ 270⬚ 300⬚ 330⬚ 0⬚
60⬚
30⬚
0⬚
−30⬚
−60⬚
Chron 5 6 13 18 21 25 31 34 M0 M4M10N M16 M21 M25
(a) 0.0 9.7 20.1 33.1 40.1 47.9 55.9 67.7 83.5 120.4 126.7 131.9 139.6 147.7 154.3 180.0 Ma
60⬚
30⬚
0⬚
−30⬚
−60⬚
0⬚ 30⬚ 60⬚ 90⬚ 120⬚ 150⬚ 180⬚ 210⬚ 240⬚ 270⬚ 300⬚ 330⬚ 0⬚
(b) 0 1 2 3 4 5 6 7 8 9 10 Ma
Figure 16 Digital isochrons of oceans based on magnetic chronology. Reproduced from Mueller RD, Roest WR, Royer J-Y, Gahagan LM, and Sclater
JG (1997) Digital isochrons of the world’s ocean floor. Journal of Geophysical Research 102: 3211–3214.
the CLN to the present (McFadden, 1984). The typical time- Sabine’s for total intensity). The coefficients of the potential
scale of the dynamo process in the core is considered to be up to degree 4 (24 coefficients) were calculated in such a way
about 15 000 years (e.g., Kono and Roberts, 2002). The geo- that the sum of the squares of differences between the observed
magnetic polarity timescale (GPTS) provides the record of field and calculated values of the field components takes the mini-
behavior for time intervals much longer than this time. Thus, mum value of all the possible cases (the least-squares method).
there is a strong possibility that the change in the dynamo Through this analysis, Gauss could show that (1) the Earth’s
behavior over such long times represents the changes in the magnetic field can be divided into various modes represented
environment in which the dynamo process operates by the spherical harmonics Pm l (cos y)(cos, sin)mf, (2) the
(McFadden and Merrill, 1984). source of this magnetic field lies within the Earth, (3) the
dipole term is dominant in the geomagnetic field, and (4) the
geomagnetic field can be well represented by low-degree field
5.01.1.4 Descriptions of the Earth’s Magnetic Field components. These are the fundamental properties of the geo-
5.01.1.4.1 Spherical harmonic analysis magnetic field, which require no modification even today.
Perhaps Gauss’s most important legacy to geomagnetism is the Since Gauss’s time, the SHA became the standard method
invention of spherical harmonic analysis (SHA) and its appli- of analysis of the geomagnetic field. Other methods exist that
cation to the Earth’s magnetic field. Because of Gauss’s law of can also uniquely describe the magnetic field, such as the
magnetism and the fact that the atmosphere is a very poor current loops or distributed dipole sources on the surface of
conductor of electricity, the magnetic field B is both the core. However, the SHA is in most cases preferred because
divergence-free (— B ¼ 0) and curl-free (— B ¼ 0) at the sur- the components obtained by this analysis (dipole, quadrupole,
face of the Earth. With these conditions, it can be shown that etc.) have definite physical meaning and because these can be
the magnetic field can be expressed as the gradient of a scalar conveniently continued to the surface of the core in which the
potential W, and this potential satisfies the Laplace equation dynamo process generates the magnetic field. After Gauss, SHA
was repeatedly applied to the magnetic observations of later
r2 W ¼ 0 [3] periods by other authors. In these, degree 6 (with 48 coeffi-
cients) was about the maximum the modelers could go to until
For the spherical geometry of the Earth, the geomagnetic
about 1960. This limit was imposed by the accuracy of data
potential can be expressed as
and the computational loads to derive coefficients.
X l l + 1
lmax X

a
W ¼a Plm ðcos yÞ glm cos mf + hm
l sin mf [4]
l¼1 m¼0
r 5.01.1.4.2 International geomagnetic reference field
With the advent of electronic computers, larger and larger
where a is the mean radius of the Earth (6371 km); r, y, and f
computations became possible. If enough data with good
are the spherical coordinates (radius, colatitude, and
accuracy are available, there is practically no limit in the trun-
longitude); Pml is the Schmidt-normalized associated Legendre
cation level of the models with the computer power of today.
functions of degree l and order m; and gnl and hm l are Gauss’s
As an example, the gravitational potential has a form quite
coefficients. The components of the magnetic field on the
similar to the geomagnetic potential (except for the l ¼ 0
surface of a sphere of radius r are, by the relation B ¼ rW,
term, which is absent in the latter), and the gravitational
X l l + 2
lmax X
dPlm potential deduced from satellite measurements goes up to
a
X¼ glm cos mf + hm
l sin mf [5] 200 or more in degrees (>40 000 coefficients). The limit for
l¼1 m¼0
r dy
the SHA of the geomagnetic potential comes from other rea-
X l l + 2
lmax X
mPlm ðcos yÞ sons (see later sections).
a
Y¼ glm sin mf hm
l cosmf [6] As more and more models were produced by different
r sin y
l¼1 m¼0 researchers, it became necessary to integrate all of these efforts
lmax X
X al + 2 to obtain the best model for that time. An internationally
l m
Z¼ ðl + 1Þ glm cosmf hm
l sin mf Pl ðcos yÞ
coordinated effort was carried out under the aegis of the Inter-
l¼1 m¼0
r national Association of Geomagnetism and Aeronomy (IAGA)
[7] and produced the International Geomagnetic Reference Field
(IGRF) for the year 1965. This was a model based on the results
where X, Y, Z are positive for the northward, eastward, and
of a dozen or so research groups around the world and the
downward components, respectively. If the flattening of the
truncation level was set to degree 10. Since then, the IGRF has
Earth can be neglected, the values at the surface of the Earth are
evolved into a series of models at 5-year intervals covering the
obtained by setting r ¼ a in the previously mentioned expres-
period 1900 to the present. By constant check with newly
sions. In the preceding text, we assumed that the magnetic field
acquired data and with the combined effort of research groups,
is produced entirely by sources within the Earth. If there are
the IGRF is considered to be the best model to describe the
external sources, the potential and the forces also contain terms
geomagnetic field in this period.
of rl and rl–1, respectively.
To determine the geomagnetic potential for 1835, Gauss
(1838) read the field values at 84 regularly spaced points (at 5.01.1.4.3 Satellite measurements
every 30 longitude and 0 , 20 , 40 , and 60 N and Good global data coverage was difficult in the past because of
S latitudes) from the magnetic charts available to him various problems on land but more so because of the oceanic
(Barlow’s for declination, Horner’s for inclination, and area that occupies two-thirds of the surface area. The use of a
magnetic satellite in a near-polar orbit solved this problem and examples are (1) the magnetic equator moving northward in
provided data with a uniform quality. The earlier satellites Southeast Asia and southward near Peru in South America, (2)
(such as POGO) measured only field intensities, which was the two positive lobes near the north pole, and (3) the single
not ideal for the purpose of the main-field analysis because of negative patch between Australia and Antarctica. These essen-
the nonuniqueness problem called the Backus effect (Backus, tial properties of the geomagnetic field have not changed in the
1970). Magsat and later satellites were equipped with vector last two centuries.
and scalar magnetometers, with good internal consistency Actually, the biggest difference in the two models lies in
among these measurements. their resolution. Whereas the model of Gauss could express the
One of the most important problems in satellite magnetic spatial variation with the shortest wavelength of about
measurements of the vector field is the determination of the 10 000 km (l ¼ 4), the most recent models can go down to
absolute orientation. The controlling factor on the accuracy of about 3000 km (l 13). However, because the short-
the data is not the magnetometer error but the error in orien- wavelength features have very small amplitudes at the surface
tation. Errors in the angle of 1 min and 1 s (when the field of the Earth, their effect is quite small if we compare two
intensity is 50 000 nT) correspond to 0.03% (15 nT) and models there. If we continue the potential field down to the
15 ppm (0.25 nT), respectively. Instrumental accuracy of this core–mantle boundary (CMB), the higher degree terms are
level is easily reached by most magnetometers. Magsat data greatly enhanced by the (a/r)l+2 factor appearing in the expres-
(1980) had an absolute accuracy of about 6 nT, while the sion for the force. The vertical component at the CMB would
more recent Ørsted satellite (2000) achieved 1.2 nT accuracy. show that the contributions from the shorter-wavelength
The reduction of errors for the Ørsted satellite reflects the components are quite significant (Figure 18). The Mercator
improvement in orientation techniques (use of star sensor, projection version of Figure 18 appears on the cover of this
etc.). With these satellites, models with degrees and orders up volume.
to about 13 were constructed (Langel et al., 1980; Olsen et al.,
2000). This is a great improvement compared with the ground-
based models, which can go up to about degree 10 (e.g., the 5.01.1.4.4 Geomagnetic spectrum
IGRF models). Global observation and description of the mag- The mean square of the field value over some area or volume
netic field is further detailed in Chapter 5.02. (hB2i in the case of the magnetic field) is called the power. The
Figure 17 compares the vertical component of the magnetic power can be defined as a single number or by parts if the field
field between the 1835 model of Gauss and the 2000 model can be decomposed by some appropriate method. The power
based on Ørsted measurements. It is remarkable that the two of the magnetic field on a spherical surface (Lowes, 1974) for
models show essentially the same features as already described the spatial degree l can be written as
by Gauss (see the last section). The primary features of both
l h
a2l + 4 X 2 m 2 i
figures are the dominance of the dipole field. The similarity is
Rl ¼ glm + hl [8]
also quite apparent even in the secondary features. Some r m¼0
80 800
40 400
0 0
−40 −400
(a) −800
(a) −80
800
80
400
40
0
0
−400
−40
(b) −800
(b) −80
Figure 18 Vertical component of the magnetic field at the core–mantle
Figure 17 Vertical component of the magnetic field at the surface of the boundary (CMB) surface based on (a) 1835 (Gauss model) and
Earth in (a) 1835 (Gauss model) and (b) 2000 (Ørsted model). (b) 2000 (Ørsted model).
which shows that the power decreases as r2l4 with change in to tackle this problem, but the vector field data from satellites
the radius of the sphere. In a process that has a large amount has enough resolution. The analysis of Magsat data (Langel and
of freedom, it is often found that the powers due to various Estes, 1982) showed that the spatial (Lowes) spectrum is nearly
degrees l are not much different from each other. One well- flat to about degree 13 if it is continued down to the CMB
known example is the classic problem of black-body radiation. (Figure 19). At the higher degrees, the field has undoubtedly
In essence, there are a very large number of independent modes sources in the crust.
in the system, which interact with each other, accompanied by Since full-scale dynamo simulations became possible
efficient energy transfer among various modes. The conse- (Glatzmaier and Roberts, 1995; Kageyama et al., 1995), the
quence is that the equipartition of energy is achieved among power spectra of many dynamo models have been analyzed. It
different modes. The power spectrum for such a process will be was found that most of the dynamo results also show the
nearly white. Of course, the spectrum cannot continue to be flat nearly flat spectrum at the surface of the fluid sphere (e.g.,
to infinity (remember the quantum effect of Planck). In fluid Kono and Roberts, 2002). The nearly flat spectrum continues
motion, it is expected that the spectrum decays rapidly, once to very high degrees such as the truncation level (l ¼ 40 100
the scale becomes small enough for diffusion to be effective. for most models). In a very intensive calculation (with no
This principle appears to apply to the dynamo process in hyperdiffusivity in the model), Roberts and Glatzmaier
the core. The resolution of the earlier models was not enough (2000) found that the trend continues to about l ¼ 200 and
1010 1015
n
Rn = (n + 1) ∑ [(gnm)2 + (hnm)2]
m=0
109 1014
108 1013
Surface of earth
107 1012
Rn, (nT)2, Core–mantle boundary

Rn, (nT)1, Surface of earth
106 1011
105 1010
Core–mantle
boundary
104 109
Core: Rn = 1.349 × 109 (0.270)n (nT)2

103 108
102
Sum
10
Crust: Rn = 37.1(0.974)n (nT)2
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
n
Figure 19 Spatial magnetic field spectrum (Lowes spectrum) based on Magsat data. Reproduced from Langel RA and Estes RH (1982) A geomagnetic
field spectrum. Geophysical Research Letters 9: 250–253.
then a steep decay with l seems to start. This corresponds to the data that are nonlinear field elements (I, D, F), the simple
(core surface) wavelength of about 200 km. method is no longer sufficient. One reason is that the local
This conclusion does not exclude the possibility that the minimum is no longer the global minimum. Another reason is
power of a particular degree l is different from the general trend that there is a fundamental nonuniqueness in some of the
by a factor of O(1). We may take note of the fact that Uranus inverse mapping of these components (Backus, 1970; Hulot
and Neptune have magnetic fields in which the dipoles are et al., 1997; Kono, 1976; Proctor and Gubbins, 1990).
much smaller than the quadrupoles. In the present geomag- As an example, we will discuss the case of declination. It can
netic field, the dipole term (l ¼ 1) is about two to three times readily be seen that since the axially symmetrical terms (g0l ) do
above and the quadrupole term (l ¼ 2) is by a similar factor not generate a Y component of the field, the potential com-
below the general trend. Similar departures have been posed only of these terms will give the declination either 0 or
observed in the simulation of various dynamo models. It is 180 everywhere. It is easy to form a combination of these terms
speculated that the departure of the l ¼ 1 term above the trend that gives D ¼ 0 everywhere, which is the same as the case of an
indicates that the dipole is somehow preferred in the dynamo axial dipole (g01). This is too easy, but we can show some more
process in the Earth. The behaviors of the geodynamo in a long complicated cases in which nonuniqueness appears.
time span are discussed in Chapter 5.11. To calculate the declination, we need to know the horizon-
tal components X and Y given by [5] and [6]. Because they will
be evaluated at r ¼ a, the factor (a/r)l+1 in the potential [4] may
5.01.1.4.5 Inverse problem be equated to 1. Consider two potentials:
Geophysical problems are often characterized by a combination
of three elements, namely, (a) a source for the process, (b) a W1 ¼ aðbcos y + csin ysinfÞ [11]
physical mechanism to induce some observable phenomena, n
Wn ¼ aðW1 =aÞ [12]
and (c) measurements of these quantities. An example is the
occurrence of an earthquake that has (a) geometry of the fault Note that W1 represents a geocentric dipole inclined by
plane (width, height, azimuth, and dip) and source function tan1(c/b) in the 90 meridian, that is, (g01, g11, h11) ¼ (b, 0, c).
(a time–space description of the fault movement), (b) excitation For this dipole field, the horizontal components are
and propagation of the seismic waves (which can be described by
1 dW1
a theory), and (c) waveforms recorded at seismic observatories. X1 ¼ ¼ bsiny + c cos ysin f [13]
When the source is known, it is relatively easy to predict what a dy
occurs and what will be observed. This is called the forward 1 dW1
Y1 ¼ ¼ c cos f [14]
problem. Most geophysical problems have a different nature. It asin y df
is more common that we have to guess the source when a set of
Similarly, the potential Wn gives the horizontal components
observations is available. This is called the inverse problem. The
determination of seismic source (e.g., centroid moment tensor Xn ¼ nðbsiny + c cos ysinfÞ
solution) is an example of the inverse problem. [15]
ðbcos y + c sinysin fÞn1
In the case of the magnetic field, there are (c) observations
(field elements such as X, Y, Z or I, D, F at many points over the Yn ¼ nccos fðb cos y + c sin ysin fÞn1 [16]
surface of the Earth), and we know (b) that the field is
expressed by a potential and how the field components can It can be seen from these that Yn/Xn ¼ Y1/X1 showing that
be derived from it. We seek (a) the source (a scalar potential tan D is the same for the two potential fields. By using D
given by Gauss coefficients) that causes the observed magnetic (jDj p) instead of tan D, the nonuniqueness may be removed
field. It is thus a typical inverse problem. in some cases, since the factor (b cos y + c sin y sin f)n1 can
Gauss’s analysis of the 1835 field is the first of such inver- take both positive and negative values. However, for odd n,
sion efforts. Suppose that a field element (one of the linear even the signs of horizontal components are the same. As an
elements X, Y, Z) at a point i is designated by yi, Gauss coeffi- example of odd n, the case of n ¼ 3 gives
cients (gm m
l , hl ) by xj, and the factors for each xj depending on
the position (yi, fi) by Aij. The difference between the observa- W3 6 11
¼ cos y ð3cos y + 5cos3yÞ
tion and the value derived from the potential is a 5 58
X 6
+ sinysin f +
33
ðsin y + 5sin3yÞsin f [17]
ei ¼ yi Aij xj or e ¼ y Ax [9] 5 10 8
j 1 15 1 15
ðcos y cos 3yÞ ð3siny sin 3yÞ
giving an estimate of error for this observation. The method of 10 4 60 4
Gauss was to minimize the square of errors jej2. To obtain the Thus, W3 is a combination of dipole and octupole fields
minimum, the standard approach is to seek the local mini- with nonzero coefficients
mum for each xj by equating the derivative to zero. This leads pffiffiffi pffiffiffiffiffiffi pffiffiffiffiffiffi
0 1 0 1 2 3 6 6 13 6 15 10
to the ordinary least-squares solution g1 , h1 , g3 , h3 , g3 , h3 ¼ , , , ,
5 5 5 5 5 10
1
x ¼ AT A AT y [10] [18]
where AT and A1 indicate the transpose and the inverse of A, Figure 20 shows two potential fields, W1 and W3, which are
respectively. For the analysis of the present field, this simple quite different, and yet the declination is the same everywhere on
method usually works (with the addition of the weight factor the surface of a sphere. The arbitrary linear combination of Wn
to take care of the reliability of individual data). In the case of has the same property. It can be concluded that the inverse
1.0
0.8
0.6
0.4
0.2
X 0.0
−0.2
−0.4
−0.6
−0.8
−1.0
1.0
0.8
0.6
0.4
0.2
Y 0.0
−0.2
−0.4
−0.6
−0.8
−1.0
2.0
1.6
1.2
0.8
0.4
Z 0.0
−0.4
−0.8
−1.2
−1.6
−2.0
200
160
120
80
40
D 0
−40
−80
−120
−160
−200
Dipole Dipole + octupole
Figure 20 An example of an inclined dipole field (left) and a dipole + octupole field (right) that are quite different in X, Y, Z components, and yet, the
values of declination D are the same everywhere on the surface of the sphere.
mapping of the declination to the potential is severely nonuni- physically plausible constraints imposed by the function (e.g.,
que (Kono, 1976). minimum energy requirement). Although the penalty function
Serious treatments of geophysical inversion were initiated and the hyperparameter are chosen arbitrarily, the application
by the seminal paper of Backus and Gilbert (1967). The idea of the stochastic inversion over many years has established that
was to minimize the sum of errors and some auxiliary function. this method gives reasonable models as solutions.
In their original treatment, the function was chosen so that the Stochastic inversion can treat both linear and nonlinear
obtained solution had the best resolution (the ‘deltaness’). data elegantly. Gubbins, Bloxham, and various colleagues con-
This would be optimal for problems such as obtaining the tinued to collect inclination and declination data back in time
mantle structure from seismic observations. However, it may and inverted them to describe the secular variation before the
not be appropriate for the geomagnetic inversion because the time of Gauss (Bloxham et al., 1989). This endeavor culmi-
parameters to be determined are discrete numbers and not a nated in the creation of the gufm1 model of Jackson et al.
continuous function of radius (say). Gubbins (1983) intro- (2000), which is based on a large data set composed of field
duced the use of stochastic inversion for geomagnetic analysis. directions compiled by Jonkers et al. (2003) and covers the
In this case, the function to be minimized is the sum of the interval AD 1590–1995. These models are very important
misfit (M) and a ‘regularization’ or ‘penalty’ function (P): achievements that expand the time span for which we know
some details of the magnetic field behavior.
S ¼ M + lP [19]
where l is the hyperparameter indicating the relative impor-

5.01.1.5 Observational Constraints
tance of the two terms. The ordinary least-squares method is
the limit of small l. When l becomes large, the inversion In the long history of geomagnetism and in the last one and
chooses models for which P is small, that is, that satisfy some a half centuries in particular, the methods of measurements
and analyses of the global field have undergone tremendous relative field intensity in a direction in which the core is
progress. Improvement in the sensitivity and accuracy of aligned. In most cases, there are three sensors placed orthogo-
measurements have been matched by more and more elabo- nal to each other in a fluxgate magnetometer.
rate modeling of the magnetic field. It is doubtful, however, For both types of instruments, it is not very difficult to
if this trend of improvement will continue into the future. attain the sensitivity and accuracy of 0.1 nT, with proper facil-
The reason for this lies in the nature of the Earth’s magnetic ity for calibration in the case of the fluxgate magnetometer.
field itself. Errors in the vector field accuracy of 6 nT (Magsat) and 1.2 nT
(Ørsted) are primarily caused by inaccuracy in orientation
5.01.1.5.1 Geometric attenuation determination. More details of the modern magnetic instru-
As seen earlier in the text, the geomagnetic spectrum at the ments can be found in Chapter 5.04.
surface of the Earth decays in power with the degree l roughly
as (Rc/a)2l+1, where Rc is the radius of the core. Since the power 5.01.1.5.3 Signal and noise in magnetic field
is the mean square value of the magnetic field by definition, measurements
the field components can be seen to decay as (Rc/a)l+(1/2). The magnetic field we observe over the surface of the Earth is
Using this relation, the dipole field is attenuated by a factor composed of three distinctly different components. The first is
of 0.407, while the quadrupole and octupole fields are reduced the internal magnetic field produced by the dynamo process in
to 0.224 and 0.123, respectively, compared with the value at the core. This is also called the main field. The main field can
the CMB. This effect damps the shorter-wavelength signal quite be described by Gauss coefficients with degrees up to about 13.
drastically at the surface of the Earth. Assume that the spectrum There must be shorter-wavelength components in the field
is completely flat with the amplitude Rl ¼ (100 mT)2 at the generated in the core, but they cannot be determined
CMB. The field at the surface contains about 41 mT dipole with any confidence because of the reasons discussed later in
and 22 mT quadrupole component, etc. The contributions the text.
from l ¼ 15 and from l ¼ 25 fields are only about 9.3 and The second is the short-term fluctuations of the field. There
0.02 nT, respectively. Considering the fact that the intensity are various types of such fluctuations. The largest of these is
of the field due to the crustal source is roughly 10 nT, the that accompanied with magnetic storms, which can sometimes
measurement of the field generated in the core having short exceed 1000 nT. However, these are rather rare events, and the
wavelength is practically impossible (Figure 21). data corresponding to these occurrences can be excluded in the
analysis of the magnetic field that originates within the Earth.
5.01.1.5.2 Accuracy of data Variations of the field on the timescale of a day when solar
The magnetic field is measured by either scalar or vector instru- activity is quiet (Sq) are small (about a few tens of nT), and
ments. A typical example of scalar magnetometers is the their effect can be made even smaller by averaging many
proton-precession magnetometer. In this magnetometer, a observations or by taking values at local midnight.
strong field is applied to protons (which are in the molecules The third is the field produced by the magnetization in the
of water contained in the cylinder) and afterward removed. The crust. The most famous example in this category is the oceanic
magnetic spins are aligned in the direction of the applied field, magnetic anomalies, which are due to the alternatingly nor-
and upon the removal of the strong field, they precess around mally and reversely magnetized layers produced by seafloor
the direction of the (remaining) ambient field with a frequency spreading from the mid-oceanic ridges. The average rate of
proportional to the field strength. The relation between the reversals of the geomagnetic dipole moment is about five
field strength B and the signal frequency f can be expressed as times per million years in the past few million years, and it is
2pf ¼ gpB. The factor gp is called the gyromagnetic ratio, and it is lower before. The present spreading rates are 1–10 cm year1 at
the constant gp ¼ m/ℏI where m and ℏI are the magnetic moment various ridges. Thus, the width of oceanic crust with either
and angular momentum of the proton. The measurement of normal or reverse magnetization is about a few to tens of
the frequency of the signal therefore gives the absolute magni- kilometers. This is much shorter than the wavelength of the
tude of the magnetic field. Optical pumping magnetometers, main-field fluctuations.
such as the rubidium vapor magnetometer, are similar. There are also anomalies with longer wavelengths, as
On the other hand, the fluxgate magnetometer is widely shown by the spectrum produced from Magsat observations
used to obtain vector values of the field. It uses the nonlinear (Figure 19). The surface spectrum is nearly flat above about
hysteresis property of a material such as mu-metal (a Ni-rich l ¼ 13, which suggests that this part of spectrum is produced by
alloy). Because of its very high magnetic permeability, such sources that are close to the observation locations.
sensors can easily be saturated in a weak magnetic field such as The bulk of the mantle cannot be the source because the
that of the Earth. In this magnetometer, an alternating current ambient temperature is much higher than the Curie points of
is applied to a coil wound around the mu-metal core. The magnetic minerals. Thus, the crust (and the uppermost man-
signal induced by the hysteretic property of the core has twice tle) appears to be responsible for this part of the spectrum. For
the original frequency, which is related to the strength of the more details about the magnetic properties of the mantle and
ambient field. Through a negative feedback mechanism, elec- crust, see Chapters 5.07 and 5.08. The intensity for many of
tric current is supplied to oppose the ambient field, until the the l > 13 terms is about 10 nT, and it is likely that they are
sum of the fields becomes negligibly small. The compensation mostly due to the sources in the crust (this is the subject of
current is thus proportional to the magnetic field, but the Chapter 5.06). It is therefore not possible to measure the core
proportionality depends on the core material and the shape contribution for degrees l > 13, as the level of signal is signifi-
of the coil. Thus, a fluxgate magnetometer measures the cantly lower than that.
L=6
L = 12
L = 18
L = 24
L = 43
(a) (b)
Figure 21 The vertical magnetic component due to a dynamo model (a) at the surface of the Earth and (b) at the CMB when the spherical harmonics
are truncated at various degrees. Modified from Kono M and Roberts PH (2002) Recent geodynamo simulations and observations of the
geomagnetic field. Reviews of Geophysics 40: RG1013. http://dx.doi.org/10.1029/2000RG00102.
radiation). With the advent of the space age, direct measure-

5.01.1.6 Magnetic Fields of the Planets and Satellites
ments by spacecraft showed that many planets and a few
The Earth is not unique in having an intrinsic magnetic field. satellites also have their own magnetic fields. In this section,
The magnetic fields of the Sun and Jupiter were detected by a brief summary of the current understanding of the magnetic
spectroscopic methods (Zeeman effect and kilometric fields associated with the planets in the solar system is
provided. Many good review papers are available on this sub- whole. Undoubtedly, our understanding of planetary dynamos
ject; among them, some important ones are Schubert and will be given a strong boost from these observations since
Soderlund (2011) and Stevenson (2010) for general summary, conditions in Mercury’s core are quite different from those in
Langereis et al. (2010) for the fields due to remanences, and the Earth’s core.
Stanley and Glatzmaier (2010) for dynamo models to explain
the observed planetary fields. 5.01.1.6.2 Magnetic mapping of planets
These fields can originate from either active dynamos, The magnetic fields of the Moon and Mars have already been
induced fields, or ancient dynamos. Spacecraft observations measured globally (Langereis et al., 2010; Tsunakawa et al.,
have been carried out at various levels of sophistication. The 2010), and a nearly global magnetic map of Mercury is also
first step is the ‘flyby,’ in which a spacecraft approaches a available (Uno et al., 2009). A few spacecraft have circulated in
planet (satellite) and observes the magnetic signature while polar orbit around Venus, but there is no strong magnetic
it is close enough. In the next step, a spacecraft becomes an signal from this planet. The Moon and Mars lack an active
artificial satellite of the target body with low orbital inclina- dynamo, but very conspicuous magnetic anomalies have
tion. This allows observations of the magnetic field at low been observed on both of them.
latitudes. Finally, a spacecraft will be inserted into a near- On the Moon, many magnetic signatures are correlated
polar orbit, in which case most of the surface of the body with surface features such as craters, indicating their possible
can be observed. relation with the impact events. Meteorite impacts are associ-
The distinction between planets and satellites is not ated with a very high temperature, enough not only to vaporize
important in the consideration of their magnetic fields. We there- rocks but also to ionize them. In such a situation, it is possible
fore ignore the division and call all these bodies ‘planets’ in the that an electric current is induced by the impact, which in turn
following. Among the terrestrial planets, the Moon (Clementine generates a magnetic field, providing the cause of the rema-
and Lunar Prospector), Venus (Magellan), Mars (many space- nences in rocks. However, since it is almost certain that a liquid
crafts in polar orbits), and Mercury (MESSENGER) have been core existed in the early history of the Moon, it is quite plausi-
observed by polar orbiting spacecraft. Magnetic fields of outer ble that the Moon had once an active dynamo but it shut down
planets have been known mostly from flyby observations. How- later (Schubert and Soderlund, 2011).
ever, more detailed measurements are being carried out by more Observations of the magnetic field on Mars revealed an
recent spacecrafts; that is, Galileo flew in a near-equatorial orbit astonishing feature. Very strong magnetic anomalies reaching
encountering Jupiter’s moons in a number of flybys, and Cassini 1500 nT (at 200 km altitude) were observed in a rather wide
covered low to middle latitudes of Saturn and made many flybys area in the Southern Hemisphere (Connerney et al., 1999).
of Titan and other satellites. With these, more detailed images of Moreover, the anomalies appear as zones of alternating polar-
the planetary magnetic fields are being constructed (see the ity. They are elongated in the east–west direction (Figure 23),
succeeding text). reminiscent of the anomalies associated with seafloor spread-
ing on the Earth. However, they are quite different: The anom-
5.01.1.6.1 Example of a flyby observation aly bands are much wider (typically 200 km) and their
As a typical flyby observation by a spacecraft, Figure 22 shows remanence intensity much higher (20 A m1 assuming a layer
the data from Mercury. Mariner 10 flew to Venus and used its thickness of 30 km) than the typical oceanic anomalies on the
gravity field to accelerate itself to reach Mercury’s orbit. On its Earth (5–50 km and 1–10 A m1). In addition, there is no
first flyby (29 March 1974), a coherent field structure was other evidence of seafloor spreading on Mars. Although it is
discovered on the antisunward side of Mercury, with the max- almost certain that the present anomaly field is due to the once
imum field strength of about 100 nT at the closest approach of active dynamo, more studies seem to be necessary before the
723 km (Ness et al., 1975). From this observation, it was origin of Martian magnetic anomalies can be understood
concluded that Mercury has an intrinsic magnetic field. The satisfactorily.
magnetic disturbance that occurred in the magnetosphere of
Mercury after Mariner 10’s closest approach resembles a sub- 5.01.1.6.3 Planetary magnetic fields: A comparative study
storm on the Earth. The planetary bodies in the solar system (both planets and
Mercury is the only terrestrial planet in the solar system satellites) can conveniently be categorized by their primary
other than the Earth in which an intrinsic magnetic field has composition. Terrestrial planets such as the Earth are mainly
been observed. The source mechanism for this magnetic field composed of rock (silicates and oxide with some metallic
has not been completely settled yet. There are some arguments iron). Gas giants (Jupiter and Saturn) have rock cores as
that the field may be due to remanent magnetization or ther- much as ten times the mass of the Earth, but their main
moelectric currents. But the majority opinion is that Mercury constituents are hydrogen and helium, and the rock part is
has an intrinsic dynamo. Until recently, Mercury had been not important in their dynamo processes. The two other giants
visited only by Mariner 10. But now, NASA’s MESSENGER (Uranus and Neptune) also have a thick hydrogen envelope,
spacecraft (launched 8 March 2004) has successfully entered but the main constituents are probably ‘ice’ (meaning nongas-
into an elliptical orbit around Mercury, and the magnetic field eous states) of H2O, CH4, and NH3 and the ice layer may
was observed and found to be expressible by a strongly offset extend to about 80% of their radii. The distribution of these
dipole model (Anderson et al., 2011). Another spacecraft, bodies indicate the condition under which the planets accu-
BepiColombo, will also be sent to Mercury around 2015 in a mulated in the early solar system; in the inner part, large
joint effort of European countries and Japan. These orbiting amounts of volatiles could not accrete due to high tempera-
spacecrafts can give us a picture of Mercury’s magnetic field as a ture, while gas and ice were the most abundant where
(a)
View from the Sun

Magnetopause ZSE
Magnetosphere 42 Sec B averages
Magnetosheath III Bow shock obs.
−3 +2
YSE
Quiet MMer
= 7 × 10−4
Mo
Scaled
20g
−3 Magnetopause Bow
shock
CA
−3 +3(RMer)
YSE
Mercury
+2 To Sun
XSE
Mariner 10 mercury I encounter
(b) 29 March 1974
Figure 22 (a) Mercury (courtesy of NASA, photo PIA03103.jpg) and (b) its magnetic field observed by Mariner 10 flyby. Modified from Ness NF,
Behannon KW, Lepping RP, and Whang YC (1975) The magnetic field of Mercury, I. Journal of Geophysical Research 80: 2708–2716.
temperature was low enough for them to accrete. Many of the hydrogen becomes metallic under the very high pressure and
satellites of Jupiter and other giants also contain large amounts has a high conductivity. There seems to be no problem of
of ice. supplying energy sufficient to maintain the convection. In the
There are three possibilities for the origin of planetary bod- icy giants, (dirty) water under very high pressure may provide
ies with magnetic fields: (1) dynamo, (2) remanence, and (3) the high conductivity necessary for a dynamo. Dynamo process
induction. Exploration of the solar system with spacecraft may be confined to the thin outer shell, which may be respon-
showed us all three of these are indeed operative in some of sible for the large tilt of the dipole axis observed for Uranus and
the planets (including satellites). We shall give a brief discus- Neptune (Hubbard et al., 1995).
sion of these occurrences. In the terrestrial planets, only the metallic core has the
A convective region with sufficiently large electrical conduc- possibility of providing dynamo generation. These cores are
tivity is necessary for maintaining a dynamo. In the gas giants, mainly composed of iron but contain small but significant
0⬚
20 −30⬚
15
−60⬚
06
13
−90⬚
240⬚ 210⬚ 180⬚ 150⬚ 120⬚
West longitude
Br (nT)
−1500 0 1500
Figure 23 Magnetic field map of one-third of the Southern Hemisphere of Mars. Note the east–west trending bands of strong anomalies.
Reprinted from Connerney JEP, Acuna MH, Wasilewski PJ et al. (1999) Magnetic lineations in the ancient crust of Mars. Science 284: 794–798, with
permission. Copyright 1999 AAAS.
amounts of light elements such as Si, O, H, or S. The melting another. Among them, the fields at Europa and Callisto seem
temperature of iron is considerably lower than that of silicate to be of induction origin ( Jia et al., 2010; Khurana et al.,
rocks (which is a primary reason why cores form to begin 1998). Although they are farther away from Jupiter compared
with). Inclusion of light elements (especially sulfur) lowers with the Moon from the Earth, the Jovian magnetic field is still
the melting temperature, so it is possible that the cores of strong enough to produce a clear signal of the induced mag-
not-too-small planets can remain largely molten for 4.6 billion netic field. The inductive region must be the ‘dirty ocean’ that
years since their formation, even with the secular cooling of the exists below the surface ice cover. Io shows a complex magnetic
silicate mantle overlying them. The molten state cannot guar- signature, but it does not have an intrinsic magnetic field.
antee, however, that convection in the core is vigorous enough Khurana et al. (2011) concluded that Io should have a global
to sustain a dynamo. In the Moon and Mars, the liquid part, if magmatic ocean beneath the crust, which causes a strong
it still exists, is perhaps too small to generate a significant induced field in the presence of the Jovian magnetic field.
dynamo. Ganymede, on the other hand, seems to have an intrinsic
Venus, on the other hand, must have a liquid core similar to field and consequently an active dynamo (Kivelson et al,
that of the Earth. The lack of an intrinsic magnetic field in 1996). The surface field is of the order of 1 mT, but since
Venus is perhaps not due to its core but more because of its Ganymede has the terrestrial-style interior beneath the ice
mantle. The surface temperature is quite high on Venus, so that shell of about 900 km thick, it might be more appropriate to
the temperature gradient within the mantle is not as large as in consider the field strength at the ice–rock boundary. In this
the Earth. Also, the volatile content of the upper mantle is quite case, the magnetic field becomes 3 mT, which is about an order
low, inhibiting the formation of an asthenosphere, which of magnitude larger (smaller) than Mercury’s (Earth’s) surface
lubricates the mantle and makes plate tectonics possible on field (Schubert and Soderlund, 2011). Ganymede is thought to
the Earth. As a result, mantle convection on Venus is of the type have a metallic core with a fluid component, and thus, it is
with a thick lid on top, and heat transfer is not sufficiently likely that the field is produced by dynamo action. However,
high. The mantle does not draw high-enough heat from the there remain uncertainties such as the size of the core and how
core, so the core may be inactive with a subadiabatic temper- the energy source can be maintained in such a small body.
ature gradient. The state of Venus may have been different in
the remote past; it may have had times when plate tectonics
was in operation. Then, the heat transfer was more efficient, 5.01.1.6.4 The magnetic field of the Sun
which resulted in convection in the core needed for dynamo In the previous section, we discussed the observed facts about
action. Unfortunately, the very high surface temperature pro- planetary magnetic fields. Before closing this topic, it is infor-
hibits a crustal remanence field to remain, and so there can be mative to consider the magnetic field of the Sun. Direct obser-
no evidence of dynamo action in Venus’ past, unlike the cases vation of the Sun’s magnetic field became possible around
of the Moon and Mars. 1950 when the solar magnetograph was constructed. Measure-
Observations of Jupiter’s moons by the Galileo spacecraft ments are done by observing the shift of spectral lines due to
brought us some shocks about their magnetic fields. All of the the Zeeman effect. At first, the strong magnetic field associated
Galileo satellites show magnetic signatures in one way or with sunspots (0.2 T) was detected. With an increase in
sensitivity, it became possible to observe the global field of component of the background field is positive (negative) in
about 104 T, which is more or less of dipole type. the Northern Hemisphere, which is similar to the reverse (nor-
These observations provided us with a rich view of the solar mal) polarity on the Earth, the radial component accompanied
magnetic field. The Sun has a dipole field roughly aligned with with a sunspot pair will have polarities from the west to east
its rotational axis, which reverses polarity every 11 years or so in the Northern Hemisphere ( in the Southern Hemisphere).
(the solar cycle). In one solar cycle, the sunspots first appear at When the background field is reversed, the polarity of the
middle latitudes (about 30 ). The area where sunspots are sunspot pairs also reverses (see Hathaway, 2010).
abundant migrates with time toward the equator. When the There are two particularly wonderful things about the solar
activity is confined to the equatorial region, a new group of magnetic field. First, we can directly see what occurs in the
sunspots appears in middle latitudes, and a new solar cycle dynamo region, at least near its surface. This is not possible for
begins (Figure 24). It is now known that the sunspots are an all the planetary dynamos, including that of the Earth. What we
expression of the toroidal magnetic field floating up from the see from radiation of various wavelengths is quite complex.
surface of the Sun; the sunspots often form pairs that show the The convection is chaotic and has characteristic length scales of
places where the field exits and reenters the Sun’s photosphere. 106 m (granules) and 3 107 m (supergranules), and it even
The polarity of the dipole can be seen not only from the weak occurs on the global scale (7 108 m). The magnetic field
background field but also in the polarity of the radial field in generation is also chaotic and shows every complexity (Weiss,
the sunspot pair (Figure 25). For instance, if the radial 2001). It is considered that the solar magnetic field is produced
Sunspot area in equal area latitude strips (% of strip area) > 0.0% > 0.1% > 1.0%
90 N
30 N
EQ
30 S
90 S
1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Date
Average daily sunspot area (% of visible hemisphere)
0.5
0.4
0.3
0.2
0.1
0.0 12 13 14 15 16 17 18 19 20 21 22 23
1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Date
Figure 24 The butterfly diagram of the sunspots. Top: Time change in latitudinal distribution of sunspots. Bottom: Area of the solar surface occupied
by sunspots. Reproduced from Hathaway DH (2010) The solar cycle. Living Reviews in Solar Physics 7:1.
- 10G - 5G 0G + 5G + 10G
90 N
30 N
Latitude
EQ
30 S
90 S
1975 1980 1985 1990 1995 2000 2005 2010
Date
Figure 25 The magnetic butterfly diagram. The radial component of the magnetic field is longitudinally averaged. Note that the character of the figure is
very similar to the ordinary butterfly diagram. Reproduced from Hathaway DH (2010) The solar cycle. Living Reviews in Solar Physics 7:1.
by the so-called ao dynamo process. This is consistent with the Dynamics.’ Core–mantle interaction is also discussed in that
strong toroidal field evidenced in the sunspots. volume.
Second, the sunspots have almost continuously been The study of geomagnetism, like many other disciplines of
observed since the time of Galileo, and the solar cycle can be natural science, is composed of measurements and observa-
traced in the past. Because of the correspondence between the tions to obtain data of the magnetic field and analysis of such
sunspots and the magnetic field, these records can be inter- data to investigate the cause and consequences of various
preted to represent the magnetic activity of the Sun. The level of phenomena. In this section, I shall write introductions to the
activity (which can be measured by the number of sunspots) other chapters in this volume by briefly referring to their con-
between the dipole reversals (the sunspot or solar cycles) tents. In writing the introduction, I will use these two items,
shows some changes over the observed period, but the most observation and analysis, as the basic concepts for arrange-
notable aspect of the record is that there was some period ment, rather than the order in which each of the chapters
(1647–1715) when the sunspots were absent (the Maunder appear.
Minimum; see Figure 26). This coincides with the time called
the ‘Little Ice Age’ when the temperature was exceptionally low 5.01.1.7.1 The present field and its components
in Europe. It is still controversial but some people speculate Chapter 5.02 ‘The Present and Future Geomagnetic Field’ by
that the Sun’s activity was low when sunspots were few or none Hulot, Olsen, Sabaka, and Fournier describes the data
and that the diminished amount of solar radiation caused the obtained from observations made by scientific instruments
change in the climate. (magnetometers) and analysis of these data. Broadly speaking,
In summary, the dynamo process in the Sun is quite differ- the ‘present’ in geomagnetism means since the time of the great
ent from that of the Earth. It is in the turbulent regime and can German mathematician and physicist Carl Friedrich Gauss. As
better be described by an ao process (the Earth’s dynamo is explained in this chapter, the measurements of magnetic field
thought to be closer to an a2 process; e.g., Christensen et al., directions were possible and carried out in some places, but the
1998). Still, the magnetic field of the Sun gives important clues measurement of absolute intensity became possible only after
for considering the magnetic field of the Earth, clues that are the invention by Gauss (about 1830) of the method based on
otherwise unattainable. The surface of the solar dynamo region the oscillation of a magnetic needle in the field. He also orga-
can be directly observed. The process is so rapid that 400 years nized a consortium of geomagnetic observatories and thus laid
of observations are enough to cover tens of polarity reversals. the basis of the geomagnetism as a discipline of natural science.
Using the observatory data thus collected and the field values
read from the magnetic charts, Gauss succeeded in making a
global field model of the geomagnetic field at that time, using
5.01.1.7 Introduction to the Other Chapters
the SHA that utilizes the basic property of the magnetic field,
The primary objective of geomagnetism is to understand vari- that is, it can be described by a potential satisfying the Laplace
ous properties of the geomagnetic field, how it is created, how equation. Together with the least-squares method, another
it changes in space and with time, what effects the field invention of Gauss, used in this analysis, the data collection
imposes on other phenomena taking place on the Earth, etc. and the analysis by SHA are still the main tools and methods
Besides these, geomagnetism also provides very strong and used in geomagnetism today.
effective tools for other branches of geosciences; for example, The biggest recent progress in observation is the deploy-
the geomagnetic polarity timescale is the most powerful ment of magnetic satellites such as Magsat and CHAMP. Satel-
method in dating various past events back to about 200 mil- lite measurements revolutionized observations in many ways.
lion years ago. For older ages, magnetic data give important In the presatellite era, the data distribution was heavily biased
inferences about the state of the surface of the Earth through to the Northern Hemisphere and to the land areas. Now, the
the study of polar wander. Almost all of these subjects are data are evenly distributed over the whole Earth. Also, the
covered in one or the other of the following chapters. The distribution in time used to be sporadic, as even the observa-
exception is the creation mechanism of the magnetic field, tories were only required to report hourly mean data. There are
that is, dynamo theory, which is treated in Volume 8 ‘Core certain drawbacks in satellite data; a satellite cannot provide
Yearly averaged sunspot numbers 1610–2010

200
Sunspot number
150
100
50 Maunder
minimum
0
1600 1650 1700 1750 1800 1850 1900 1950 2000 2050
Date
Figure 26 The yearly sunspot number for the period 1600–2000. Note the conspicuous absence of sunspots for half a century starting from about
1645 (the Maunder Minimum). Reproduced from Hathaway DH (2010) The solar cycle. Living Reviews in Solar Physics 7:1.
continuous measurement in one place since it moves in a solar wind. Because of the very high temperature where the
latitudinal direction so as to circle the Earth in about 90 min interaction takes place, almost all the atoms are ionized so that
and also in a longitudinal direction to cover the entire Earth. the carriers of the electric currents are plasma particles. In the
However, these drawbacks are very minor compared with the case of the ionosphere, there are also neutral particles. But in
merit of obtaining high accuracy data continuously and all all cases, currents flow because of the motion of plasma parti-
over the Earth. The availability of dense and continuous (for cles and generate the external magnetic fields. It is therefore
more than 10 years for some satellites) data made it possible to important to understand the properties and motions of
produce very sophisticated and detailed models such as the plasmas under the influence of electric and magnetic fields.
comprehensive modeling (CM) approach, which can express There are currents that are typical of the low or middle lati-
not only the internally generated field (core dynamo origin tudes, such as the Sq current, the equatorial electrojet, and the
plus crustal origin fields) but also external fields such as iono- ring current. These currents contribute to daily Sq variation,
spheric and magnetospheric fields. The current models CM4 enhancement of the field near the magnetic equator, and
and CHAOS are perhaps the best available models of this P01-type field generation. On the other hand, there are magne-
category. Another recent development is the use of assimilation tospheric and ionospheric convection currents and substorm
techniques in the modeling. This is widely used in such fields currents, which strongly influence the magnetic fields at high-
as meteorology, but its introduction in geomagnetism is rather latitude locations. These external fields are usually taken as the
recent. This technique may give us the possibility of predicting noise when we try to extract the internal magnetic fields. The
future magnetic field changes. advent of comprehensive models, however, makes it possible
As discussed fully in Chapter 5.02, the geomagnetic field is to describe the internal and external fields at the same time.
composed of various elements. The first important division is The currents flowing outside of the Earth not only cause the
between the field generated within and outside of the surface external field but also induce an internal component through
where observations are made (usually the surface of the Earth). electromagnetic induction. This field is again a noise if we
The former is called the internal field and is further divided intend to study the internal field, but it provides useful infor-
into the core field and the crustal field, while the latter is the mation for other purposes. This is the subject of Chapter 5.07
field generated by the currents flowing outside the Earth. The ‘Geomagnetic Induction Studies’ by Steven Constable. Induc-
core field is the one created by the dynamo process operating tion effects occur since the Earth is not a perfect insulator.
in the core and is also called the main field (which is the main Because of the finite electrical conductivity of Earth materials,
subject of Chapter 5.02). The magnetic field can also be electric currents are induced in the mantle, which generate
divided into intrinsic and induced components. The induced some magnetic fields in turn. There are two ways of analyzing
field occurs because of the electromagnetic induction effect in these phenomena: the magnetotelluric (MT) method and geo-
electrically conducting materials. The inducing field is mostly magnetic depth sounding (GDS). In the MT measurements, the
the currents flowing in the upper atmosphere of the Earth, but magnetic and electric field variations are measured in the
artificial current sources can also induce a magnetic field. orthogonal directions. The magnetic field represents the induc-
The crustal field is treated in Chapter 5.06 ‘Crustal Magne- ing field, while the electric field can be taken as the response.
tism’ by Purucker and Whaler. This field is generated by rem- The relation between these two fields can be analyzed by the
anent and induced magnetization in the crust and the response function, which is the ratio of the two in frequency
uppermost mantle where temperature is lower than the Curie domain. Various external field changes can be used as the
point of the magnetic minerals. The spherical harmonic degree inducing field in the MT method, for example, the daily vari-
of about 14 is the boundary where the dominant internal field ation, the magnetospheric ring current variations, and pulsa-
changes; the core field (the crustal field) dominates at lower tions. In most cases, its frequencies are 10 days or less. In the
(higher) harmonic degrees. Thus, the crustal field has wave- GDS methods, changes occurring in longer intervals are uti-
length shorter than a few thousand kilometers. Large-scale lized on the assumption that the field has degree 1 geometry,
anomalies such as continent-wide features can best be studied that is, P01. In this case, the penetration of the field change is
by satellite magnetic observations. More detailed (small-scale) deeper than in the MT method (because of the skin effect), and
features are observed and described based on aeromagnetic the electrical conductivities of the deeper parts of the mantle
surveys. The crustal field is due to either the magnetic suscep- can be deduced from such studies. The MT and GDS methods
tibility or remanent magnetization of the rocks, and their provide the estimates of electrical conductivity in the mantle,
separation is possible by making plausible assumptions. With which can be supported by the measurements of electrical
various techniques described in this chapter, the magnetic conductivity of rocks and minerals under high pressures and
anomalies associated with such geologically interesting fea- high temperatures.
tures as Chicxulub impact structure or kimberlite pipes can
be drawn. These provide very useful information in exploring 5.01.1.7.2 Data sources and their reliability
the continental structure and lithology. Geomagnetism is based on the observation of the magnetic
The sources of the external fields are mainly in the iono- field generated within the Earth and its interaction with other
sphere and magnetosphere, where electric currents flow phenomena. How these observations are done is the theme
because of the motion of ionized atoms and electrons. These discussed in Chapter 5.04 ‘Observation and Measurement
are described in Chapter 5.03 ‘Magnetospheric Contributions Techniques’ by Turner, Rasson, and Reeves. Various magne-
to the Terrestrial Magnetic Field’ by Baumjohann and Naka- tometers are employed in order to measure the field accurately;
mura. The most important source of the external magnetic field the principles of magnetometers and the measurement pro-
is the interaction between the Earth’s magnetic field and the cedures are described in this chapter. In the magnetometers
used today, absolute and relative instruments are important. cover different time intervals, with the data of mostly the direc-
The absolute instruments are based on some physical law, such tional information, that is, declination and inclination. Chapter
as Zeeman effect or gyromagnetic effect, and the measured 5.14, on the other hand, treats intensity variations. A separate
values can readily be translated to the magnetic field values chapter is required because intensity measurements are quite
without the need for empirical calibration. Most of these abso- difficult and form a unique branch in paleomagnetism.
lute instruments are also scalar magnetometers, such as the Jackson and Finlay are the authors for Chapter 5.05, which
proton-precession magnetometer and the optical-pumping has the title ‘Geomagnetic Secular Variation and Its Applica-
magnetometer. The relative instruments such as fluxgate mag- tions to the Core.’ The scientific measurements of various
netometers operate by compensating the field by flowing electric components of the geomagnetic field started about 1830
current in the coils. They need calibration to translate the when the great physicist and mathematician Carl Friedrich
observed value to absolute units, but they are so versatile that Gauss organized G€ ottinger Magnetischer Verein. There are,
these relative instruments are very widely used. Also discussed however, older documents that record the measurements of
are the measurement techniques and various methods to ascer- mostly declination, in the preceding centuries. These are in the
tain the reliability of data in the study of remanent magnetiza- ships’ log books collected in national archives in England,
tion in rocks and sediments. Holland, France, and others. After very assiduous efforts by
Paleomagnetism is the study of the magnetic field in the many researchers, these records are collected in the databases
past when there were no observations of the magnetic field ( Jonkers et al., 2003). This database covers four centuries from
itself. The direct measurement of the magnetic field cannot go 1590 to 1990 and consists of more than 150 000 declination,
back more than six centuries, even if we include all the prim- 19 000 inclination, and 16 000 intensity (either H or F) mea-
itive measurements done by the use of magnetic compasses. surements. From this data set, a field model called gufm1 was
Thus, we have to rely on the magnetic remanence contained in constructed ( Jackson et al., 2000), which covers the four cen-
rocks, sediments, or artifacts left by our ancestors. The natural turies. This is a monumental work and extended the period for
remanent magnetization (NRM) in these materials is measured which we know the behavior of the field about threefold from
and treated as representing the ancient magnetic field, with the Gauss to the present time. This chapter describes these efforts
assumption that the NRM is proportional to the magnetic field in detail. Moreover, this chapter provides us with what can be
when these rocks or sediments were formed. How good is this deduced from these field models about the conditions in the
assumption is the subject of Chapter 5.08 ‘Magnetizations in core, where the main field is generated. In particular, it gives
Rocks and Minerals’ by Dunlop and Ozdemir. The magnetic the theoretical basis for describing the core surface motions
properties of these materials therefore are an important matter based on secular variation of the magnetic field.
for the study of geomagnetism. The authors discuss domain Chapter 5.09 ‘Centennial- and Millennial-Scale Geomag-
structure in these materials. It appears that single-domain and netic Field Variations’ by Cathy Constable and Korte treats the
pseudo-single-domain states are the most important in getting field variations in about 10 000 years or less. This time interval
information of old magnetic fields. However, many of the is similar to that covered by what was conventionally called
materials used in paleomagnetism contain not only single- ‘archaeomagnetism.’ Whereas the usual practice in arch-
domain or pseudo-single-domain grains but also various aeomagnetism was to make measurements of artifacts or
other states that may cause problems in magnetic stability. lavas erupted repeatedly in the time interval of interest and
The composition and mineralogy of the magnetic minerals then proceed to compare the variation with the results
are also important in these problems, since the stability of obtained from other results from nearby areas, the new
the minerals to various outside conditions depend on these. approaches are characterized by their global view of the mag-
Different properties of various types of magnetic remanence netic field, very similar to the studies of the past few centuries
are explained, as well as the importance of the size of magnetic described in Chapter 5.05. This was made possible by two
grains that control the domain state, and consequently the main factors: the construction of a world database of observa-
reliability and other properties of remanence. The authors tional data for this interval and the application of inverse
also discuss the merits and shortcomings of the routinely techniques with regularization in both spatial and temporal
used experimental procedures such as stepwise demagnetiza- dimensions. By constructing global models spanning several
tions and paleointensity determination. Overall, it is a very thousand years, it became possible to discuss the persistence of
complicated problem, but we are lucky enough to have very such features as the Pacific dipole low or westward drift more
reliable paleomagnetic results for most of the geologic past! or less confined in the Atlantic hemisphere.
Johnson and McFadden treat a much longer time span of a
5.01.1.7.3 Field behaviors in various time windows million years or so in Chapter 5.11 ‘Time-Averaged Field and
The geomagnetic field before the instrumental observation of Paleosecular Variation.’ First, they show that the geocentric axial
the magnetic field is treated in four chapters. Division into dipole (GAD) approximation, which is the basic assumption in
multiple chapters is necessary because the nature of the mag- most of paleomagnetic works of older material, is indeed satis-
netic field data, the reliability of the assigned ages and the field fied by the data covering some appropriate time interval (105 or
values, and/or spatial or temporal coverage of available data is 106 years). However, the GAD indicates only the dominance of
quite different. What can be inferred from the data is also quite the axial dipole component, and so there can be other compo-
different, because of these differences. The chapters in this cate- nents when we look at the TAF. The departure from the GAD or
gory are Chapter 5.05 (field changes in a few centuries), TAF is studied by paleosecular variation (PSV) analysis, in
Chapter 5.09 (the same for centuries to millennia), and which the most frequently used quantity is the measure of
Chapter 5.11 (up to a few million years). These three chapters scatter of the virtual geomagnetic pole (VGP) about the
rotational axis or the mean VGP. But the introduction of the The three youngest of these are Mono Lake (33 ka), Laschamp
giant Gaussian process (GGP) model of Constable and Parker (41 ka), and Blake (120 ka) excursions. Because of the very
(1988) changed the situation drastically. The former PSV narrow age range of the excursions, they have potential to
models used to attribute the cause of the scatter of VGP to give very precise ages for certain strata. In the Matuyama
just two or three sources, such as dipole wobble and nondipole chron and older ages, there are also a large number of excur-
contributions, which were chosen by the author in an ad hoc sions reported, but their existence is not so well documented as
way. In the GGP model, it is assumed that Gauss coefficients in the Brunhes chron, so that the utility for the use of age
defined at the surface of the core behave like zero-mean ran- control seems to be somewhat limited at present.
dom variables obeying a normal distribution. The exceptions to Compared to the excursions, the normal–reverse polarity
this rule are the axial dipole (g01) and axial quadrupole (g02), transitions extracted from the oceanic magnetic anomalies in
which have means. With this very simple assumption, the GGP the direction of seafloor spreading give a very well-established
model can predict many of the behaviors of the real paleomag- GPTS back to 160 Ma. Gee and Kent describe the GPTS in
netic field. ‘Source of Oceanic Magnetic Anomalies and the Geomagnetic
Unlike the previous three chapters, Chapter 5.13 ‘Paleoin- Timescale.’ Their discussion extends to the behavior of the
tensities’ by Tauxe and Yamazaki is not confined by the interval magnetic field such as tiny wiggles (cryptochrons), intensity
of time. Instead, this chapter treats the intensity determination fluctuations, skewness, magnetizations in the source layers,
for the past geomagnetic field by paleomagnetic methods. The and crustal accretion process at the oceanic ridges. These
special treatment of paleointensities is the result of the difficul- are certainly interesting themes of their own merit, but it is
ties inherent in the experiments for intensity determination. the provision of the accurate timescale for nearly 200 Ma that
There are lots of complicated factors that hinder the retrieval is the most important. In a way, this is one of the best
of reliable intensity estimates by experiments, which the authors services done by geomagnetism for all the branches of Earth
discuss at length. However, some consensus in the minimum sciences.
requirement in the experimental procedures does now exist, Raub, Kirschvink, and Evans discuss polar wander and
including the built-in checks of reliability in the experiments. related problems in Chapter 5.14 ‘True Polar Wander: Linking
Absolute paleointensities are determined from rocks of which Deep and Shallow Geodynamics to Hydro- and Biospheric
the NRM is acquired as a TRM, because this remanence can be Hypotheses.’ Polar wander is a long-standing concept in paleo-
reproduced in the laboratory to a satisfactory degree. In addi- magnetism, but it was quite a surprise that some data from the
tion, there are relative paleointensities obtained from sediments, Proterozoic and Cambrian were shown to indicate very fast
which deposited in a relatively unchanging conditions so that pole movement and also a possibility that the surface of the
the variation in remanence intensity can be attributed to the Earth was completely frozen at some time. The latter possibil-
field intensity changes. Even in such cases, some normalization ity, for example, the snowball Earth, is now discussed with the
of remanence intensity is necessary because it is not sure that the possible link with the Ediacaran and Cambrian explosion seen
remanence is determined solely by the field intensity. Often from the fossil records. The authors discuss how the paleomag-
used normalizers are saturation isothermal remanent netic data can be interpreted in these respects. Although still
magnetization, anhysteretic remanent magnetization, and sus- somewhat speculative, the study of polar wander by paleomag-
ceptibility (w). Recently, stacks of sedimentary intensities (such netic methods is a manifestation of the vast possibility of
as Sint-800) have been created and gained some confidence as geomagnetic sciences. Future developments may provide
representing the global changes in magnetic field intensity. more decisive data or interpretation on this very stimulating
Thus, it seems that perhaps the intensity estimates from volcanic problem.
rocks and sediments will be integrated to provide absolute
intensity changes in a continuous manner.
5.01.1.7.4 Excursions, reversals, and polar wander 5.01.1.8 Conclusions

There are paleomagnetic data that are quite useful for the stud-
ies of other branches of Earth sciences, not particularly related Magnetism seems to have a peculiar property that it is both very
to geomagnetism. Examples are the timescales that use the exotic and very practical, at different times or at the same time. It
geomagnetic excursions (Chapter 5.10) or the polarity reversals is sometimes just a target of curiosity, but at other times, it is
(Chapter 5.12). Because the geomagnetic polarity transitions quite useful for daily life. Magnets have remained a source of
occur simultaneously everywhere on the surface of the Earth, wonder for thousands of years in human history. Today’s life
they provide a very useful measure of absolute time when the without magnets is difficult to conceive (just look around and
magnetic direction of geologic formations is studied. Another you will see motors, tape recorders, hard disks, etc.).
example is the true polar wander (Chapter 5.14), which The situation is very similar with the Earth’s magnetic field.
describes the change in the rotational axis of the Earth and The north-seeking property of the magnetic compass fascinated
consequently changes in the coordinate system of the Earth. people in medieval times. It then became a very practical tool
In Chapter 5.10 ‘Geomagnetic Excursions,’ Laj and Chan- to measure direction (and sometimes position). All the ships in
nell examine the observations of excursions in the Brunhes the world used magnetic compasses for navigational aid. Now,
chron and Matuyama chron and older periods. In the 790 ka such days are gone (although most ships still carry compasses
of the Brunhes chron, some 10 to 15 excursions have been for contingency purposes), and it is the Global Positioning
proposed, among which the authors choose seven as well System (GPS) that gives you precise position and bearing. As
documented after detailed examination of the original reports. to the target of curiosity, Albert Einstein counted the origin of
the Earth’s magnetism as one of the important problems of References

twentieth-century physics.
From the time of Gauss, the study of geomagnetism has Ade-Hall JM, Khan MA, Dagley P, and Wilson RL (1968) A detailed opaque petrological
and magnetic investigation of a single Tertiary lava flow from Skye, Scotland. 1.
steadily progressed and many of the wonders are now
Iron-titanium oxide petrology. Geophysical Journal of the Royal Astronomical
explained in terms of physical processes. The observational Society 16: 375–388.
constraints, such as the geometric decay and the crustal field Anderson BJ, et al. (2011) The global magnetic field of Mercury from MESSENGER
hiding the short-wavelength part of the main field, are now orbital observations. Science 333: 1859–1962.
well understood. Since 1995, when dynamo models based on Backus GE (1970) Nonuniqueness of the external geomagnetic field determined by
surface intensity measurements. Journal of Geophysical Research 75: 6339–6341.
first-principles calculations appeared, our understanding of Backus GE and Gilbert JF (1967) Numerical applications of a formalism for geophysical
field generation has been rapidly growing. inverse problems. Geophysical Journal of the Royal Astronomical Society
Is our curiosity going to be satisfied soon? Far from it. The 13: 247–276.
present is just a small window in the entire spectrum of what a Blackett PMS, Clegg JA, and Stubbs PHS (1960) An analysis of rock magnetic data.
Proceedings of the Royal Society of London A 256: 291–322.
magnetic field can do. Paleomagnetists used to be called
Bloxham J, Gubbins D, and Jackson A (1989) Geomagnetic secular variation.
‘paleomagicians’ because they tended to speak of unwieldy Proceedings of the Royal Society of London A 329: 415–502.
things such as ‘continental drift’ and ‘polarity reversals.’ The Brunhes B (1906) Recherches sur la direction de l’aimantation de roches volcanique.
discovery of the Vine–Matthews mechanism, which combined Journal de Physique 5: 705–724, 4e sér.
magnetic reversals with seafloor spreading, changed Earth sci- Cande SC and Kent DV (1992) A new geomagnetic polarity timescale for the Late
Cretaceous and Cenozoic. Journal of Geophysical Research 97: 13917–13951.
ences forever. Magnetism became the best tool to examine Cande SC and Kent DV (1995) Revised calibration of the geomagnetic polarity timescale
plate tectonics evolution since the Jurassic. Dynamo models for the Late Cretaceous and Cenozoic. Journal of Geophysical Research
can produce polarity reversals that are in many ways very 100: 6093–6095.
similar to paleomagnetic observations (Glatzmaier et al., Christensen U, Olson P, and Glatzmaier GA (1998) A dynamo model interpretation of
geomagnetic field structures. Geophysical Research Letters 25: 1565–1568.
1999; Kono and Roberts, 2002). But these computer models
Christensen UR, Aubert J, Cardin P, et al. (2001) A numerical dynamo benchmark.
include parameters that are quite different from those in the Physics of the Earth and Planetary Interiors 128: 25–34.
core of the Earth (particularly the viscosity). Moreover, the data Connerney JEP, Acuna MH, Wasilewski PJ, et al. (1999) Magnetic lineations in the
from paleomagnetism are far from satisfactory; for instance, we ancient crust of Mars. Science 284: 794–798.
are not at all sure of the field configurations during reversals. Constable CG and Parker RL (1988) Statistics of the geomagnetic secular variation for
the past 5 m.y. Journal of Geophysical Research 93: 11569–11581.
We need more and better observations. Likewise, theoretical Cox A, Doell RR, and Dalrymple GB (1963) Geomagnetic polarity epochs and
advance is needed to ascertain that the computer simulation Pleistocene geochronometry. Nature 198: 1049–1051.
results apply to the Earth’s dynamo. Dalrymple GB (1972) Potassium-argon dating of geomagnetic reversals and North
Another consideration is that the mode of dynamo opera- American glaciations. In: Bishop WW and Miller JA (eds.) Calibration of Hominid
Evolution, pp. 107–134. Edinburg: Scottish Academic Press.
tion may have also changed with time. The existence of mag-
David P (1904) Sur la stabilité de la direction d’aimantation dans quelque roches
netic anomalies on the Moon and Mars indicates that such volcanique. Comptes Rendus de l’Académie des Sciences 138: 41–42.
smaller bodies may have had self-sustaining dynamos early in Dietz RS (1961) Continent and ocean basin evolution by spreading of the sea floor.
their histories. Observations of Mercury’s and Ganymede’s Nature 190: 854–857.
magnetic fields suggest that there may be a mechanism by Doell RR and Dalrymple GB (1966) Geomagnetic polarity epochs – A new polarity event
and the age of the Brunhes–Matuyama boundary. Science 152: 1060–1061.
which the magnetic field can be actively sustained in these Folgerhaiter G (1899) Sur les variations séculair de l’inclinaison magnétique dans
small bodies in the solar system. Venus does not show any l’antiquité. Journal de Physique 8: 5–16.
intrinsic magnetic field, despite the fact that it has an internal Gauss CF (1838) Allgemeine Theorie des Erdmagnetismus in Resultate aus den
structure quite similar to that of the Earth. Unfortunately, we Beobachtunger des magnetischen Vereinsim Jahre 1836–1841, 6 Hefte, G€ottingen
und Leipzig. (Reprinted in Werke, Bd. 5, 121–193).
cannot learn about the past of the Venusian dynamo, because
Gilbert W (1600) De Magnete. New York: Dover, translated by Mottelay PF (1958).
we still do not know if some remanence field exists on Venus, Glatzmaier GA and Roberts PH (1995) A three-dimensional self-consistent computer
despite the high surface temperatures there. In the present simulation of a geomagnetic field reversal. Nature 377: 203–209.
Venus, mantle convection is of thick lid type and quite differ- Glatzmaier GA, Coe RS, Hongre L, and Roberts PH (1999) The role of the Earth’s mantle
ent from plate tectonics, in which heat transfer is very effective in controlling the frequency of geomagnetic reversals. Nature 401: 885–890.
Gubbins D (1983) Geomagnetic field analysis-I. Stochastic inversion. Geophysical
in the mantle and can draw high heat flow from the core. It is Journal of the Royal Astronomical Society 73: 641–652.
speculated, however, that Venus may have had a time in the Harradon HD (1943) Some early contributions to the history of geomagnetism.
past when the mantle convected in the plate tectonics regime, Terrestrial Magnetism and Atmospheric Electricity 48: 3–17, 79–91, 127–130,
and the core dynamo was active. A comparison of the magnetic 197–199, 200–202, 1943; ibid. 49: 185–198, 1944; ibid. 50: 63–68.
Hathaway DH (2010) The solar cycle. Living Reviews in Solar Physics 7: 1. (Online
fields of the planets with that of the Earth, together with the
Article): cited 2013/07/04, http://www.livingreviews.org/lrsp-2010-1.
knowledge of other physical properties, will provide a better Heirtzler JR, Dickson GO, Herron EM, Pitman WC III, and Le Pichon X (1968) Marine
understanding of the origin and evolution of the geomagnetic magnetic anomalies, geomagnetic field reversals, and motions of the ocean floor
field. and continents. Journal of Geophysical Research 73: 2119–2136.
Hess HH (1962) History of ocean basins. In: Engel AEJ, James HL, and Leonard BL
(eds.) Petrologic Studies, pp. 599–620. Boulder, CO: Geological Society of
America.
Hospers J (1953) Reversals of the main geomagnetic field, I.II. Proceedings of the
Acknowledgments Koninklijke Nederlands Academic van Wetenschappen B56: 467–491.
Hubbard WB, Podolak M, and Stevenson DJ (1995) The interior of Neptune.
In: Cruickshank B (ed.) Neptune, pp. 109–138. Tucson, AZ: University of Arizona Press.
I would like to thank Ron Merrill and Jerry Schubert for their
Hulot G, Khokhlov A, and Le Mouël J-L (1997) Uniqueness of mainly dipolar magnetic
constructive criticism on earlier versions of the manuscript. fields recovered from directional data. Geophysical Journal International
Their critical comments were quite helpful in revising the draft. 129: 347–354.
Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular Morgan WJ (1968) Rises, trenches, great faults, and crustal blocks. Journal of
variation from historical records. Philosophical Transaction of the Royal Society of Geophysical Research 73: 1959–1982.
London A 358: 957–990. Mueller RD, Roest WR, Royer J-Y, Gahagan LM, and Sclater JG (1997) Digital isochrons
Jia X, Kivelson MG, Khurana KK, and Walker RJ (2010) Magnetic fields of the satellites of the world’s ocean floor. Journal of Geophysical Research 102: 3211–3214.
of Jupiter and Saturn. Space Science Reviews 152: 271–305. Nagata T, Akimoto S, and Uyeda S (1951) Reverse thermo-remanent magnetism.
Jonkers ART, Jackson A, and Murray A (2003) Four centuries of geomagnetic data from Proceedings of the Japan Academy 27: 643–645.
historical records. Reviews of Geophysics 41(2): 1006. http://dx.doi.org/ Needham J (1962) Science and Civilisation in China. Physics and Physical Technology,
10.1029/2002RG000115. Part 1 Physics, vol. 4, pp. 434. Cambridge: Cambridge University Press.
Kageyama A, Sato T, and Complexity Simulation Group (1995) Computer simulation of Needham J. (ニーダム, ジョゼフ) (1977) 中國の科学と文明, 物理学 Science
a magnetohydrodynamic dynamo, II. Physics of Plasmas 2: 1421–1431. and Civilization in China, vol. 4, Part 1: Physics, Vol. 7, pp. 515. Tokyo: Shisaku-
Khramov AN (1958) Paleomagnetic Correlation of Sedimentary Strata. Leningrad: sha(東京: 思索社).
Gostoptekhizdat, 218 pp. Néel L (1951) L’inversion de l’aimantation permanente des roches. Annales de
Khurana KK, Jia X, Kivelson MG, Nimmo F, Schubert G, and Russell CT (2011) Geophysique 7: 90–102.
Evidence of a global magma ocean in Io’ s interior. Science 332: 1186–1189. Ness NF, Behannon KW, Lepping RP, and Whang YC (1975) The magnetic field of
Khurana KK, Kivelson MG, Stevenson DJ, et al. (1998) Induced magnetic fields as Mercury, I. Journal of Geophysical Research 80: 2708–2716.
evidence for subsurface oceans in Europa and Callisto. Science 395: 777–780. Olsen N, et al. (2000) Ørsted initial field model. Geophysical Research Letters
Kivelson MG, Khurana KK, Russell CT, et al. (1996) Discovery of Ganymede’s magnetic 27: 3607–3610.
field by the Galileo spacecraft. Nature 384: 537–541. Proctor MRE and Gubbins D (1990) Analysis of geomagnetic directional data.
Kono M (1976) Uniqueness problems in the spherical harmonic analysis of the geomagnetic Geophysical Journal International 100: 69–77.
field direction data. Journal of Geomagnetism and Geoelectricity 28: 11–29. Roberts PH and Glatzmaier GA (2000) A test of the frozen-flux approximation using a
Kono M and Roberts PH (2002) Recent geodynamo simulations and observations of the new geodynamo model. Philosophical Transactions of Royal Society
geomagnetic field. Reviews of Geophysics 40: RG1013. http://dx.doi.org/ A358: 1109–1121.
10.1029/2000RG00102. Roche A (1951) Sur les inversion des l’aimantation remanente des roches volcaniques
Langel RA and Estes RH (1982) A geomagnetic field spectrum. Geophysical Research dans les monts d’Auvergne. Comptes Rendus de l’Académie des Sciences
Letters 9: 250–253. 233: 1132–1134.
Langel RA, Estes RH, Mead GD, Fabiano EB, and Lancaster ER (1980) Initial geomagnetic Runcorn SK (1962) Palaeomagnetic evidence for continental drift and its geophysical
field model from Magsat vector data. Geophysical Research Letters 7: 793–796. cause. In: Runcorn SK (ed.) Continental Drift, pp. 1–40. New York: Academic Press.
Langereis B, Lesur V, Purucker ME, Connerney JEP, and Mandea M (2010) Crustal Schubert G and Soderlund KM (2011) Planetary magnetic fields: Observations and
magnetic fields of terrestrial planets. Space Science Reviews 152: 223–249. models. Physics of Earth and Planetary Interiors 187: 92–108.
Larson RL and Pitman WC III (1972) World-wide correlation of Mesozoic magnetic Stanley S and Glatzmaier GA (2010) Dynamo models for planets other than Earth. Space
anomalies and its implications. Geological Society of America Bulletin Science Reviews 152: 617–649.
83: 3645–3662. Stevenson DJ (2010) Planetary magnetic fields: Achievements and prospects. Space
Le Pichon X (1968) Sea-floor spreading and continental drift. Journal of Geophysical Science Reviews 152: 651–664.
Research 73: 3661–3697. Uno H, Johnson CL, Anderson BJ, Korth H, and Solomon SC (2009) Modeling
Lowes FJ (1974) Spatial power spectrum of the main geomagnetic field, and Mercury’s internal magnetic field with smooth inversions. Earth and Planetary
extrapolation to the core. Geophysical Journal of Royal Astronomical Society Science Letters 285: 328–339.
36: 717–730. Tsunakawa H, Shibuya H, Takahashi F, et al. (2010) Lunar magnetic field observation
Malin SRC (1987) Historical introduction to geomagnetism. In: Jacobs JA (ed.). and initial global mapping of Lunar magnetic anomalies by MAP-LMAG onboard
Geomagnetism, vol. 1, pp. 1–49. London: Academic Press. SELENE (Kaguya). Space Science Reviews 154: 219–225.
Matuyama M (1929) On the direction of magnetization of basalt in Japan, Tyosen and Verhoogen J (1956) Ionic ordering and self-reversal of magnetization in impure
Manchuria. Proceedings of the Imperial Academy Japan 5: 203–205. magnetites. Journal of Geophysical Research 61: 201–209.
McDougall I and Tarling DH (1963) Dating of polarity zones in the Hawaiian Islands. Vine FJ (1966) Spreading of the ocean floor: New evidence. Science 154: 1405–1415.
Nature 200: 54–56. Vine FJ and Matthews DH (1963) Magnetic anomalies over oceanic ridges. Nature
McFadden PL (1984) Statistical tools for the analysis of geomagnetic reversal 199: 947–949.
sequences. Journal of Geophysical Research 89: 3363–3372. Wang (王振鐸) CT (1948) Discovery and application of magnetic phenomena in
McFadden PL and Merrill RT (1984) Lower mantle convection and geomagnetism. China. 1. The lodestone spoon of the Han. Chinese Journal of Archaeology 3: 119.
Journal of Geophysical Research 89: 3354–3362. Wegener A (1929) Die Entstehung der Kontinente und Ozeane. In: The Origin of
McKenzie DP and Parker RL (1967) The North Pacific: An example of tectonics on a Continents and Oceans, 4th edn. Branschweig: Vieweg und Sohn, translation from
sphere. Nature 216: 1276–1280. German by J. Biram, pp. 246 New York: Dover 1966.
Mercanton PL (1926) Inversion de l’inclinaison magnetique terrestre aux ages Weiss N (2001) Turbulent magnetic fields in the Sun. Astronomy and Geophysics
geologiques. Terrestrial Magnetism and Atmospheric Electricity 31: 187–190. 42: 3.10–3.17.
Mitchell AC (1946) Chapters in the history of terrestrial magnetism. Terrestrial Yamamoto Y (山本義隆) (2003) 磁力と重力の発見 Discovery of Magnetic and
Magnetism and Atmospheric Electricity 51: 323–351, 37: 105–146, 1932; ibid. 42: Gravitational Forces, vols. 1–3, 947 pp. みすず書店, 東京 Tokyo: Misuzu
241–280, 1937; ibid. 44: 77–80, 1939. Publishers.
5.02 The Present and Future Geomagnetic Field
G Hulot, Université Paris Diderot, Paris, France
TJ Sabaka, NASA Goddard Space Flight Center, Greenbelt, MD, USA
N Olsen, Technical University of Denmark, Lyngby, Denmark
A Fournier, Université Paris Diderot, Paris, France
This chapter is a revision of the previous edition chapter by G. Hulot, T. J. Sabaka, N. Olsen, Volume 5, pp. 33–75, © 2007, Elsevier B.V.
5.02.1 Introduction 33
5.02.2 Magnetic Field Data Used in Modeling 35
5.02.2.1 Magnetic Elements, Coordinates, and Time 35
5.02.2.1.1 Geodetic and geocentric coordinates and components 35
5.02.2.1.2 Dipole coordinates and components 35
5.02.2.1.3 Quasi-dipole coordinates 36
5.02.2.1.4 Time 37
5.02.2.1.5 Coordinate systems for describing magnetospheric sources 37
5.02.2.2 Ground Data 38
5.02.2.3 Satellite Data 39
5.02.2.3.1 POGO (OGO-2, OGO-4, and OGO-6) 39
5.02.2.3.2 Magsat 40
5.02.2.3.3 rsted and SAC-C 40
5.02.2.3.4 CHAMP 41
5.02.2.3.5 Swarm 41
5.02.2.3.6 Calibration and alignment of satellite magnetometers 41
5.02.2.4 Indices 42
5.02.2.4.1 Kp, an index of global geomagnetic activity 42
5.02.2.4.2 Dst, an index of magnetospheric ring-current strength 42
5.02.3 Global Models of the Earth’s Magnetic Field 43
5.02.3.1 Spherical Harmonic Representation of the Field 43
5.02.3.2 Models of the Field of Internal Origin 44
5.02.3.2.1 The international geomagnetic reference field 45
5.02.3.2.2 More advanced models 45
5.02.3.3 Models of the Core Field 47
5.02.3.4 Models of the Crustal (Lithospheric) Field 49
5.02.3.5 Comprehensive Models 51
5.02.3.5.1 Modeling philosophy 51
5.02.3.5.2 Early comprehensive models 52
5.02.3.5.3 Recent comprehensive models 52
5.02.3.5.4 Predictions from CM4 53
5.02.4 The Present Main Field 56
5.02.4.1 Spatial Power Spectra and Timescales 57
5.02.4.2 The Field at the Earth’s Surface 60
5.02.4.3 The Field at the Core Surface 64
5.02.5 Predicting the Future Main Field 66
5.02.5.1 Data Assimilation Concepts 68
5.02.5.2 Closed-Loop Geomagnetic Data Assimilation Experiments 70
5.02.5.3 First Geomagnetic Applications of Data Assimilation 73
Acknowledgments 74
References 74
5.02.1 Introduction remains confined. Beyond this boundary is indeed the realm of
the interplanetary magnetic field (IMF) produced by the Sun and
The Earth’s magnetic field (also known as the geomagnetic field) transported by a plasma known as the solar wind. In addition,
is usually defined as the magnetic field produced by all sources and at least as far as this chapter is concerned, it is also gener-
within and outside the solid Earth up to the so-called magne- ally understood that the term Earth’s magnetic field only encom-
topause, a boundary within which the Earth’s magnetic field passes the magnetic field produced by natural phenomena in

34 The Present and Future Geomagnetic Field
the 0 (permanent) to typically a few hertz frequency range. This field, produced by the magnetized rocks (which indeed mainly,
rather arbitrary limit amounts to dismiss phenomena such as but not exclusively, lie in the crust), is weaker on average. But it
electromagnetic waves (and many other high-frequency can vary substantially as a function of location, from fractions of
phenomena, which can indeed be observed with appropriate a nanotesla to thousands of nanotesla at the Earth’s surface. This
instruments) and is largely dictated by the band-pass of the strong spatial variability of the crustal field results from the
magnetometers traditionally used to investigate the permanent complex history of its magnetized sources and from the various
and slowly changing components of the Earth’s magnetic field. abilities of rocks to carry magnetization. As a result, the crustal
Bearing this definition in mind, the sources of the Earth’s field is characterized by a spatial spectrum with fairly compara-
magnetic field can then be understood as falling into two main ble contributions at all length scales at the Earth’s surface. The
categories: magnetized media and electric currents. Obviously, sum of the core field and of the crustal field then makes what is
magnetized sources can only be found inside the solid Earth. usually referred to as the internal field (or field of internal origin),
They occur in the form of rocks that have been magnetized in since both have their sources inside the solid Earth. The internal
the past (permanent magnetization) but that also bear an addi- field is dominated by the core field at large scales and by the
tional magnetization proportional to the present ambient crustal field when scales on the order of less than typically
magnetic field (induced magnetization). Clearly also, such 2000 km are considered at the Earth’s surface.
rocks can only be found in regions of the solid Earth, where As already stated, the Earth’s magnetic field also has sources
the temperature is less than the Curie temperature of the min- above the neutral atmosphere (some electric currents are to be
erals ultimately carrying the magnetization. This restricts found in the neutral atmosphere, but they are very weak, except
magnetized rocks to lie in the uppermost layers of the Earth. at times of thunderstorms, the magnetic signature of which is
All other sources of the Earth’s magnetic field are electric cur- however traditionally not considered to be within the direct
rents. Those can be found in most regions of the Earth: inside scope of Earth’s magnetism). The field produced by such exter-
the metallic core, in the mantle and crust, in the oceans, and, nal sources is known as the external field (or field of external
finally, above the neutral atmosphere, in the ionosphere and origin). It is produced by electric current systems in the iono-
magnetosphere (cf. Figure 1). sphere and the magnetosphere, the physical origin of which
Sources of the Earth’s magnetic field thus differ in nature and can crudely be described in terms of currents driven by dynamo
location. The fields they produce also widely differ in magnitude processes in the plasma of the ionosphere, which moves within
and spatiotemporal behavior. By far the most intense field (on the main field, and currents produced by the movement of
the order of 30 000 nT at the equator and 60 000 nT at the charged particles traveling within the magnetosphere. Those
poles) is the one produced within the core through a self- two current systems are very different in physical nature but are
sustaining dynamo process (the geodynamo). This field is coupled to each other. The field produced by ionospheric
known as either the core field or the main field (given its domi- currents is known as the ionospheric field, and that produced
nance). It changes on secular timescales, and its time derivative by magnetospheric currents, as the magnetospheric field. Their
is therefore referred to as the secular variation. By extension, its magnitude can strongly vary with time. A weak large-scale
second time derivative is also often referred to as the secular magnetospheric field (dominated by an external axial dipole)
acceleration, although, as we shall later see, changes in these on the order of 20 nT at the Earth’s surface is found at all time.
time derivatives can occur on much shorter timescales. The During so-called magnetically quiet times, a weak ionospheric
core field is mainly dipolar but also has significant multipolar field can also be found, which varies with local time (LT). Its
terms, decreasing in strength as higher orders (and therefore magnitude is minimum at night, peaks during day time, and is
smaller spatial scales) are considered. The so-called crustal on the order of a few tens of nanotesla at ground level. But at
other times, known as disturbed magnetic times, both the
magnetospheric and ionospheric fields can become much
more dynamic. The resulting external field can then occasionally
Magnetosphere reach magnitudes on the order of thousands of nanotesla and
vary on timescales from a fraction of a second to several days.
Electric currents can also be found in the crust and mantle
Sat
Coupling currents
ellit
es of the Earth. Those arise because of the weak electric conduc-
(FAC)
tivity of rocks, which reacts to the time-varying core and
Ionosphere external fields described earlier. The resulting induced currents
rvato
ries produce what is usually referred to as induced fields. Currents
bse
do
Gr
oun Lithosphere induced by the core field are difficult to directly assess and are
Oc
ea
ns thought to likely screen out high-frequency core field signals
and prevent them from being observable at the Earth’s surface.
Mantle
By contrast, currents induced by the external field produce
Induced currents
so-called externally induced fields that can easily be observed
Fluid core (and reach magnitudes of a fraction of the external inducing
field). Similar induced currents are of course also to be found
within the salty (and therefore conducting) waters of the
Solid core
oceans. In that case, however, additional currents can also be
Figure 1 Sketch of the various sources contributing to the near-Earth found, which are generated by yet another dynamo process
magnetic field. linked to the tidal and oceanic flows of the oceanic waters
The Present and Future Geomagnetic Field 35
within the main field. Those however again produce very weak elements are the angle between geographic north and the
fields, on the order of a few nanotesla at ground level. (horizontal) direction in which a compass needle is pointing,
The Earth’s magnetic field is thus the result of the superpo- denoted as declination D ¼ arctan Y/X; the angle between the
sition of many fields produced by a great variety of sources. local horizontal plane and the field vector, denoted as inclina-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Unfortunately, magnetometers can only measure the resulting tion I ¼ arctan Z/H; horizontal intensity H ¼ X2 þ Y 2 ; and total
field, and a single measurement cannot distinguish the contri- pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
intensity F ¼ X 2 þ Y 2 þ Z2 . For the numerical calculation of
butions from each source. However, provided enough mea- declination D, it is recommended to use the ‘four-quadrant
surements are being acquired and processed in appropriate inverse tangent’ atan2, that is, D ¼ atan2(Y, X), to avoid the p
ways, a separation of the contributions from each type of ambiguity of the arctan function. Note that the vertical com-
field can be achieved and global models of the Earth’s ponent Z is given with respect to the local plumb line (i.e., with
magnetic field constructed. The purpose of this chapter is to respect to the geoid) but interpreted as if it was given with
provide an up-to-date review of such global models of the respect to the reference ellipsoid. The difference between the
present field and briefly discuss recent developments aiming two directions can be up to tens of arcsecs.
at improving our ability to forecast its evolution. Sec- In contrast to magnetic observations taken at or near
tion 5.02.2 first describes the various types of magnetic data ground, satellite data are typically provided in the geocentric
currently used to investigate the present magnetic field. Sec- coordinate system as spherical components Br, By, Bf.
tion 5.02.3 next reviews the models. Special attention is being
paid in Section 5.02.4 to the current behavior of the main
field. Finally, Section 5.02.5 briefly reports on the emerging 5.02.2.1.1 Geodetic and geocentric coordinates and
field of data assimilation in geomagnetism, aiming at both components
improving our understanding of the current state of the geo- Since spherical components (and coordinates) are used for
dynamo and forecasting its future evolution. geomagnetic field modeling, it is necessary to transform the
locations of surface data, and the data themselves, from the
geodetic to the geocentric coordinate system. This can be done
in the following way:
5.02.2 Magnetic Field Data Used in Modeling The geodetic coordinates (h, a, f), where h is altitude, a is
geodetic latitude, and f is longitude, are transformed into the
5.02.2.1 Magnetic Elements, Coordinates, and Time spherical geocentric coordinates (r, y, f) by means of
Magnetic measurements taken at the Earth’s surface, including pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h a2 cos 2 a þ b2 sin 2 a þ b2
shipborne and airborne data, are typically given in a local cot y ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tan a [1]
topocentric (geodetic) coordinate system (i.e., relative to a h a2 cos 2 a þ b2 sin 2 a þ a2
reference ellipsoid as approximation for the geoid), shown in pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r 2 ¼ h2 þ 2h a2 cos 2 a þ b2 sin 2 a
Figure 2. The magnetic elements X, Y, Z are the components of
a4 cos 2 a þ b4 sin 2 a
the field vector B in an orthogonal right-handed coordinate þ 2 [2]
a cos 2 a þ b2 sin 2 a
system, the axes of which are pointing towards geographic
north, geographic east, and vertically down. Derived magnetic r02
[3]
1 þ sin 2 a
e2
Zenith with a ¼ 6378.137 km and b ¼ 6356.752 km as the equatorial
South West and polar radii, respectively, of the Earth ellipsoid as
defined by the World Geodetic System 1984 (WGS84),
Y e2 ¼ (a2 b2)/a2 ¼ 6.69438 103 as the eccentricity of the
ellipsoid, and r0 as the equatorial radius of the ellipsoid at
D
X height h (r0 ¼ a at the surface).
H The transformation of the local topocentric (geodetic) compo-
I
nents X, Y, Z to spherical geocentric components Br, By, Bf is done
East North
by means of
0 1 0 10 1
Br sin c 0 cos c X
@ By A ¼ @ cos c 0 þ sin c A@ Y A [4]
Bf 0 1 0 Z
with c ¼ y 90 þ a.
Z
5.02.2.1.2 Dipole coordinates and components
F The first three Gauss coefficients, g01, g11, h11, of a spherical har-
Nadir monic expansion of the internal scalar magnetic potential (cf.
Section 5.02.3.1) represent the field of a dipole with origin at
the Earth’s center, but inclined with respect to the axis of
Figure 2 The magnetic elements in the local topocentric coordinate rotation. The dipole axis intersects the Earth’s surface at the
system, seen from north–east. dipole poles, also known as the geomagnetic poles; geocentric
colatitude y0 and longitude f0 of the north geomagnetic pole is c ¼ atan2ð sin y0 sin ðf f0 Þ,
thus given by cos y0 sin y sin y0 cos y cos ðf f0 ÞÞ [10]

y0 ¼ 180 arccos g10 =m0 [5] is the angle between the geographic and the dipole meridian at
the location in consideration. The dipole equator (geomagnetic
f0 ¼ 180 þ atan2 h11 =g11 [6]
equator, not to be confused with the dip equator (magnetic
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2 2 equator), the line where inclination I ¼ 0 ) is defined as the
where m0 ¼ ðg10 Þ þ ðg11 Þ þ ðh11 Þ is the dipole strength and
line for which yd ¼ 90 ; the dipole poles are then the points
the output of the functions arcos and atan2 is given in degrees. for which yd ¼ 0 , resp. 180 . A grid of constant dipole co-
Using the Gauss coefficients of the most recent international latitude and longitude is shown by the blue curves in Figure 4.
geomagnetic reference field (IGRF) (IGRF-11, Finlay et al.,2012)
gives as geocentric coordinates for the north geomagnetic pole
y0 ¼ 9.99 , f0 ¼ 72.22 (the north geomagnetic pole used to 5.02.2.1.3 Quasi-dipole coordinates
be located at y0 ¼ 10.46 , f0 ¼ 71.57 in 2000, as provided by Many ionospheric phenomena are naturally organized with
DGRF 2000; note incidentally that Table 3 of Finlay et al. respect to the geometry of Earth’s magnetic field due to its
(2012), which provides such past locations of geomagnetic influence on the motion of charged particles. As a conse-
poles, shows these in geodetic coordinates, as is usually done quence, ionospheric conductivity is highly anisotropic, result-
when providing location at ground level, and not geocentric, as ing in values that are so high parallel to the field that the
given here). A coordinate system with pole at (y0, f0) is called magnetic field lines are nearly equipotential lines. Therefore,
dipole or geomagnetic coordinate system (cf. Figure 3). Dipole (or it is often convenient to work in a coordinate system that
geomagnetic) colatitude and longitude are defined as follows the morphology of the magnetic main field, such
yd ¼ arccos ð cos y cos y0 þ sin y sin y0 cos ðf f0 ÞÞ [7] as the quasi-dipole (QD) coordinates (yq, fq) proposed by
Richmond (1995). The basic idea is to trace a field line
fd ¼ atan2ð sin y sin ðf f0 Þ, (given by IGRF of a specific epoch) from the point in consid-
cos y0 sin y cos ðf f0 Þ sin y0 cos yÞ [8] eration outward to the field line apex (i.e., its highest point
above surface). The longitude of that point defines QD longi-
The transformation of the spherical geocentric components tude fq. QD colatitude yq is found by following an axial dipole
Br, By, Bf, to those in the dipole frame, Br0 , By0 , Bf0 , is then done field line from the apex downward to the ionospheric E-layer
by means of (typically assumed to be at an altitude of 115 km); the co-
0 0 1 0 10 1 latitude of the intersection of the dipole field line and that
Br 1 0 0 Br
B B0 C B CB C altitude defines QD colatitude yq. The dip equator (magnetic
@ y A ¼ @ 0 þ cos c þ sin c A@ By A [9] equator) – the line where the magnetic field lines are horizon-
B0f 0 sin c þ cos c Bf tal, that is, inclination I ¼ 0 – is given by yq ¼ 90 ; the dipole
poles (for which the magnetic field lines are vertical) are
where
defined by the points for which yq ¼ 0 , resp. yq ¼ 180 .
G
f-f0
f0
f q
ia h
q0
id ic
er w
n
M een
r
G
D
fd y
qd
P
G
M eom
er ag
idi n
an et
of ic
Ge ridian
Me
P
og
rap of P
hic
Figure 3 Relationship between spherical geocentric coordinates y, f, dipole (geomagnetic) coordinates yd, fd, and the spherical geocentric
coordinates of the north dipole (geomagnetic) pole, y0, f0. G is the north geographic pole, D is the north dipole pole, and P is the location of the point in
consideration.
90
60
30
-30
-60
-90
-180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180
Figure 4 Map showing equidistant grids in latitude–longitude for various coordinate systems for epoch 2000 (as defined by DGRF 2000, note that
this map for more recent epochs would be almost indistinguishable). Geocentric coordinates are shown in black, dipole coordinates in blue, and
quasi-dipole (QD) coordinates in red. The respective equators (for which the respective colatitude is equal to 90 ) are indicated with thicker lines. The
blue dots indicate the location of the dipole (geomagnetic) poles (for which yd ¼ 0 , resp. 180 ), whereas the red dots indicate the dip poles (for
which yq ¼ 0 , resp. 180 ).
A grid of constant QD colatitude and longitude is shown by the where Td is DLT in hours and fd and fd,s are dipole longitudes (in
red curves in Figure 4. degrees) of the location in consideration and of the subsolar
Contrary to the dipole coordinate system, the QD coordi- point, respectively. (Note that this equation is given incorrectly
nate system is not orthogonal and therefore mathematical in Langel (1987) and Langel and Hinze (1998).) We use the
differential operations (like curl, grad, and div) are nontrivial. notation DLT rather than magnetic local time (MLT) to distin-
Richmond (1995) presented equations for these differential guish the previously mentioned definition from that used in
operators in the QD system. space physics, where MLT is defined with respect to a non-
orthogonal coordinate system related to the IGRF. DLT is
defined using only the first three Gauss coefficients of a field
5.02.2.1.4 Time model.
Local Time (LT), T, is defined as the difference in (geographic) Dipole universal time (DUT), td, is defined as the DLT of the
longitude (in hours) between the subsolar point and the posi- dipole prime meridian, that is, the meridian from which dipole
tion in consideration: longitude is rendered. Hence, td (in hours) is given by

T ¼ ð180 þ f fs Þ=15 [11] td ¼ 180 fd, s =15 [15]
where f and fs are the (geographic) longitudes (in degrees) of

the location in consideration and of the subsolar point, 5.02.2.1.5 Coordinate systems for describing
respectively. Universal time (UT), t, is defined as the LT of the magnetospheric sources
Greenwich meridian: It is advantageous to use different coordinate systems for
describing the various current systems contributing to the mag-
t ¼ ð180 fs Þ=15 [12] netospheric field. The geocentric solar magnetospheric (GSM)
frame is a Cartesian coordinate system, which is preferably
The connection between UT, t, and LT, T, follows as used to describe the interaction of the solar wind with the
T ¼ t þ f=15 [13] magnetosphere, especially magnetopause and tail currents
(which are located at distances > 8 Earth radii). The origin of
Dipole (geomagnetic) local time (DLT) is defined relative to the GSM frame is the center of the Earth. The x-axis is pointing
the dipole system towards the Sun; the z-axis lies in the plane formed by the
x-axis and the geomagnetic pole and points northward; the
Td ¼ 180 þ fd fd, s =15 [14] y-axis completes a right-handed coordinate system.
The solar magnetic (SM) frame is especially useful for geodetic reference frame, it is presently not possible to take
describing the electrodynamics in the near-Earth space, for advantage of that (absolute) accuracy in field modeling, since
instance, the magnetic effect of the ring current. SM is a Carte- the observatory data contain (unknown) contributions from the
sian coordinate system with origin in the Earth’s center. Its local crustal field of scale length shorter than those that present
z-axis is parallel with the dipole axis, x is in the plane defined field models can resolve. Following Langel et al. (1982), this
by the subsolar point and the dipole axis, and y completes a local field contribution is accounted for by observatory biases for
right-handed system. The spherical coordinates of the SM sys- each observatory and each component. Joint analysis of obser-
tem are identical to a dipole colatitude/DLT coordinate system. vatory and satellite data makes it possible to estimate these
A detailed definition of SM and GSM is given in Appendix 3 biases. As a consequence of this procedure, only the time-
of Kivelson and Russell (1995). varying part of observatory data is used, and not their absolute
(static) level. The need for an absolute baseline calibration of
observatory data is therefore less important compared to an
5.02.2.2 Ground Data
accurate determination of its time changes (for instance, drift).
The Earth’s magnetic field is continuously monitored by a Recognizing this would simplify measurements, especially for
number (presently about 150) of geomagnetic observatories. ocean-bottom magnetometers, for which the exact determina-
Ground observations were the only data source for modeling tion of true north is an extremely difficult (and expensive) task.
the geomagnetic field for periods without satellite measure- Historically, observatory data are provided as annual mean
ments. The distribution of the present observatory network is, values. The availability of hourly mean values and 1 min values
however, very uneven, as can be seen from the dots in the left has, however, increased significantly recently, as can be seen
part of Figure 5, which shows observatories that provided data from Figure 6, which shows the number of geomagnetic obser-
(hourly mean values) between 1997 and 2012. This consider- vatories providing annual means, hourly means, and 1 min
ably hampers determination of the global pattern of the mag- values, respectively, through time. More recently, and as a
netic field using observatory data. Satellite data provide result of the success of the various recent satellite missions,
excellent global coverage, but the spacecraft movement makes which provided 1 Hz magnetic field measurements (see
direct comparisons of satellite and observatory data difficult. Section 5.02.2.3 later), observatories (roughly 40 as of
Although geomagnetic observatory measurements are impor- today) have also started deploying systems capable of provid-
tant also for periods for which satellite measurements are ing 1 Hz data (e.g., Chulliat et al., 2009; Worthington et al.,
available, their role has changed: from being the only source 2009). Real-time availability of such preliminary data is now a
for describing the static and time-varying part of Earth’s mag- priority, just as much as the production of so-called quasi-
netic field to providing information on its temporal change definitive data (e.g., Peltier and Chulliat, 2010) that can be
(secular variation as well as external field contributions and made available more readily and continuously than the official
their induced counterparts) at a fixed location. The joint anal- definitive data (only released at the beginning of the next year,
ysis of ground and satellite data combines the strengths of the after an annual baseline correction). This will allow for novel
two data sources: good global coverage (satellite data) and use of observatory data in combination with satellite data, for
good temporal coverage (observatory data). advanced modeling of the time changes of external field vari-
Although geomagnetic observatories aim at measuring the ations, a particularly interesting prospect in the context of the
magnetic field components with an accuracy of 1 nT in the recently launched Swarm mission (see later text).
90°
60°
30°
0°
i
-30°
-60°
-90°
-180° -120° -60° 0° 60° 120° 180°
Figure 5 Left: Observatories that provided data to the WDC/INTERMAGNET system between 1997 and 2012 (red dots) and ground track of 24 h of the
rsted satellite on 2 January 2001 (blue curve). The satellite starts at 00 UT at 57 S and 72 E, moves northward on the morning side of the Earth,
and crosses the equator at 58 E (large black arrow). It continues crossing the polar cap and moves southward on the evening side. 50 min after the first
equator crossing, the satellite crosses again the equator, at 226 E (white arrow), on the dusk side of the orbit. The next equator crossing (after
additional 50 min) is at 33 E (small black arrow), 24 westward of the first crossing 100 min earlier while moving again northward. Right: The path of a
satellite at inclination i in orbit around the Earth.
200
Annual means
IQSY
IGY/C
Hourly means
180
1−min values
Preparation for Ørsted

160
IMS
140
Number of observatories
120
2nd IPY
100
80
1st IPY
Göttingen Magnetic Union
60
40
20
0
1800 1820 1840 1860 1880 1900 1920 1940 1960 1980 2000
Year
Figure 6 Number of geomagnetic observatories providing annual means, hourly means, and 1 min values, as a function of time. International
campaigns, like the Göttingen Magnetic Union, the 1st and 2nd International Polar Year (IPY), the International Geophysical Year (IGY/C), the
International Quiet Solar Year (IQSY), and the International Magnetospheric Study (IMS), and the preparation of the rsted satellite mission have
stimulated observatory data processing and encouraged the opening of new observatories.
Observatory data are provided through the INTERMAGNET 1962). Satellite data for global field modeling were first taken
network (http://www.intermagnet.org) and through the World by the POGO satellites (data from the earlier satellites Cosmos
Data Center (WDC) system (http://www.ngdc.noaa.gov/geo 26 and Cosmos 49 in 1964 were of much poorer quality).
mag/data.shtml, http://wdc.kugi.kyoto-u.ac.jp or http://www. The use of magnetic satellite data for modeling Earth’s
space.dtu.dk/English/Research/Scientific_data_and_models/ magnetic field is different from that of ground data in several
World_Data_Center_for_Geomagnetism.aspx). aspects: Firstly, satellites sample the magnetic field over the
They are also regularly complemented by additional data entire Earth (apart from the polar gap, a region around the
collected at magnetic repeat stations. These are permanently geographic poles that is left unsampled if the inclination of
marked sites where high-quality magnetic observations are the satellite orbit is different from 90 ). Secondly, the data are
made every few years for a couple hours or even days obtained over different regions with the same instrumentation.
(MacMillan, 2007; Newitt et al., 1996). Their main purpose is Thirdly, measuring the magnetic field from an altitude of
to provide regional measures of the secular variation. Unlike 400 km or so corresponds roughly to averaging over an area
geomagnetic observatories, repeat stations do not monitor the of this dimension. Thus, the effect of local magnetic heteroge-
field continuously. But they provide a better spatial resolution neities (for instance, due to magnetization of rocks) is reduced.
to interpolate the secular variation between epochs. They offer And finally, since the satellite moves (with a velocity of about
a relatively inexpensive safety net in case satellite data are not 8 km s1 for an altitude of 400 km), it is not directly possible
available, especially now that more advanced ways of carrying to decide whether an observed magnetic field variation is due
these surveys are becoming available (e.g., Lalanne et al., to a temporal or spatial change.
2013). More information about ground data can otherwise Table 1 lists key parameters of satellites that have been used
be found in Matzka et al. (2010). (or are currently to be launched) for modeling the geomagnetic
field. In what follows, we briefly describe each of these
satellites.
5.02.2.3 Satellite Data
Ground-based magnetic measurements were the only data 5.02.2.3.1 POGO (OGO-2, OGO-4, and OGO-6)
source before the first spaceborne magnetic measurements The Polar Orbiting Geophysical Observatories (POGO) were the
were taken by the Sputnik 3 satellite in 1958 (Dolginov et al., first satellites to measure the field intensity of Earth’s magnetic
Table 1 High-precision magnetic satellites, with key parameters (in the case of the three satellite Swarm mission, the left/right numbers refer to the
single-high/lower-pair satellites)
Satellite Operation Inclination Altitude Data
OGO-2 Oct 1965–Sep 1967 87 410–1510 km Scalar only

OGO-4 Jul 1967–Jan 1969 86 410–910 km Scalar only
OGO-6 Jun 1969–Jun 1971 82 400–1100 km Scalar only
Magsat Nov 1979–May 1980 97 325–550 km Scalar and vector
rsted Feb 1999– 97 650–850 km Scalar and vector
CHAMP Jul 2000–Sep 2010 87 250–450 km Scalar and vector
SAC-C/rsted-2 Jan 2001–Dec 2004 97 698–705 km Scalar only
Swarm Launched November 22, 2013 87.75 /87.35 510/< 462 km Scalar and vector
GPS POD antenna GPS limb sounding

antenna array
Star sensors
(backside)
Optical bench with Solar panels
fluxgate magnetometers
and star sensors
Scalar Overhauser magnetometer (OVH)
Nadir pointing
GPS altimeter antenna
Vector magnetometer (CSC)
Laser retro reflector (nadir surface)
and star imager (SIM) Overhauser Langmuir probe
magnetometer and digital ion
Accelerometer (at CoG)
driftmeter (DIDM)
S-band antenna
8 m deployable boom Optical bench (vector field Absolute scalar

magnetometer + star tracker magnetometer
assembly) magnetic field
Magnetic field vector components magnitude
GPS patch
antennas
Electric field Thermal ion imager

instrument
EF vector
components,
plasma density,
Accelerometer LRR ion drift velocity
Satellite body air drag vector, ion, and
electron temperature
Langmuir
GPS, STR, LRR : position, velocity, time, attitude probes
Figure 7 Magsat (left; see Purucker, 2007), rsted (middle; see Olsen, 2007), CHAMP (upper right; see Maus, 2007), and one of the three Swarm
satellites (lower right; see Friis-Christensen et al., 2002, 2006, 2008, 2009).
field globally. They were equipped with optically pumped sampled the ambient field at 16 Hz with a resolution of
rubidium vapor absolute magnetometers. Three satellites of 0.5 nT. Attitude was measured using two star trackers on the
the POGO series are of interest for field modeling: OGO-2 spacecraft; transformation of attitude determined by these star
measured the field between October 1965 and September trackers to the vector magnetometer on the end of the boom was
1967; OGO-4 between July 1967 and January 1969, with a done using a complicated optical system. Attitude errors limit
few weeks of data overlap with OGO-2; and OGO-6 operated the vector data accuracy to about 4 nT rms. See Purucker (2007)
between June 1969 and June 1971. Intrinsic measurement for more information on the satellite; data are available
error of all three satellites is believed to be below 1 nT, but at http://www.space.dtu.dk/English/Research/Scientific_Data_
contribution due to position uncertainty results in an effective and_Models/Magnetic_Satellites.aspx.
magnetic error of about 7 nT. See Cain (2007) for more infor-
mation on the POGO satellites. Data are available at http://
www.space.dtu.dk/English/Research/Scientific_Data_and_ 5.02.2.3.3 rsted and SAC-C
Models/Magnetic_Satellites.aspx. The Danish rsted satellite was the first satellite mission after
Magsat aiming at high-precision mapping of the Earth’s mag-
netic field (Figure 7). It was launched on 23 February 1999
5.02.2.3.2 Magsat into a near-polar orbit. Being the first satellite of the Interna-
The US satellite Magsat (October 1979 to June 1980; Figure 7) tional Decade of Geopotential Research, the satellite and its mag-
made the first precise, globally distributed measurements of the netic field instrumentation have since been a model for other
vector magnetic field near Earth. The satellite (of mass 180 kg) and forthcoming missions like CHAMP and Swarm.
flew at an altitude of 300–550 km, in a near-polar dawn–dusk rsted weighs 62 kg, measures 34 45 72 cm, and has an
orbit of inclination of 97 . Magsat carried a scalar (cesium 133 8 m long boom, deployed shortly after launch, carrying the
vapor optically pumped) magnetometer and a triaxial fluxgate magnetic field instruments. The satellite is gravity gradient-
magnetometer (both at the tip of a 6 m long boom), which stabilized and attitude maneuvers could be performed using
magnetic torquers. The rsted orbit has an inclination of accurate high-precision orbit determination. In addition,
96.5 , a period of 100.0 min, a perigee at 650 km, and an CHAMP carried laser retroreflectors and an accelerometer to
apogee at 860 km. The orbit plane is slowly drifting, the LT of both improve this orbit determination and measure non-
the equator crossing decreasing by 0.91 min per day, starting gravitational acceleration such as air drag. The overall design
from an initial LT of 02:26 on 23 February 1999 for the south- of the satellite was also slightly different, profiled and with a
going track. Nominal lifetime of the mission was 14 months, much larger mass (of 522 kg at launch). Finally, CHAMP also
but quite remarkably, rsted delivered high-precision scalar carried a digital ion drift meter and a Langmuir probe to make in
and vector data over more than 14 years and is still providing situ measurements of the ion distribution and its moments
scalar data. within the ionosphere. All these additional instruments proved
A proton precession Overhauser magnetometer (OVH), very useful also for magnetic field investigations.
measuring the magnetic field intensity with a sampling rate See Reigber et al. (2005) and Maus (2007) for more infor-
of 1 Hz and an accuracy better than 0.5 nT, is mounted at the mation on the satellite; data are available at http://isdc.gfz-
tip of the deployable boom. At a distance of 6 m from the potsdam.de/champ/.
satellite body is the optical bench with the CSC (compact
spherical coil) fluxgate vector magnetometer mounted closely 5.02.2.3.5 Swarm
together with the star imager (SIM). rsted also carries a The Swarm constellation mission was proposed by a consor-
charged particle detector, to measure the flux of fast electrons, tium led by E. Friis-Christensen (DTU Space, Denmark),
protons, and alpha particles around the satellite, and GPS H. Lühr (GFZ, Germany), and G. Hulot (IPGP, France) (Friis-
receivers, to accurately determine the satellite’s position and Christensen et al., 2002) and selected by the European Space
monitor the atmospheric pressure, temperature, and humidity Agency (ESA) in 2004 (Figure 7). It aims at providing the best
profile on the path between rsted and GPS satellites through ever survey of the geomagnetic field and its temporal evolu-
atmospheric occultation. The CSC samples the magnetic field tion. Initially planned for 2010 and finally launched on
at 100 Hz (burst mode, at polar latitudes) or 25 Hz (normal November 22, 2013, the mission comprises a constellation of
mode) with a resolution better than 0.1 nT and is calibrated three satellites, with two spacecraft flying side by side at lower
using the field intensity measured by the OVH. After calibra- altitude (at 462 km initial altitude) with a separation in lon-
tion, the agreement between the two magnetometers is better gitude of about 1.4 (about 150 km at the equator), thereby
than 0.33 nT rms. Due to attitude errors, the accuracy of the measuring the east–west gradient of the magnetic field. The
vector components (Br, By, Bf) is limited to 2–8 nT (4 nT rms), third satellite is at higher altitude (510 km) and in a different
depending on component. See Neubert et al. (2001) and Olsen drifting LT compared to the lower satellite pair. The design of
(2007) for more information on the satellite; data are available the three identical spacecrafts is derived from that of CHAMP.
at http://www.space.dtu.dk/English/Research/Scientific_Data_ High-precision and high-resolution measurements of the
and_Models/Magnetic_Satellites.aspx. strength (1 Hz), direction, and variation of the magnetic field
A copy of the rsted boom and payload (but with a scalar (1 and 50 Hz) will be provided with similar instrumentation
helium magnetometer instead of the OVH, a package some- (except for the new absolute scalar magnetometer, with
times referred to as rsted-2) was launched in November 2000 enhanced experimental abilities, based on helium 4 optical
onboard the Argentinean satellite SAC-C. However, due to a pumping, built by CEA-Leti and provided by CNES; Léger
broken connection in a coaxial cable, no high-precision atti- et al., 2009), complemented by electric field measurements
tude data (and hence no reliable vector data) have been (at minimum 2 Hz sampling, using a new EFI suite of instru-
available. Only scalar data up to December 2004 could ments designed by the University of Calgary and provided by
be used. the Canadian Space Agency; see, e.g., Marchand et al., 2010).
Just like CHAMP, each satellite also carries a GPS receiver, laser
5.02.2.3.4 CHAMP reflectometers, and an accelerometer. These provide the neces-
The German CHAMP satellite (Figure 7) was launched on 15 sary observations that are required to separate and model the
July 2000 onto an almost circular, near-polar (inclination 87.3 ) various sources of the geomagnetic field and investigate the
orbit with an initial altitude of 454 km, which progressively ionospheric environment. For more details, see Friis-
decreased to almost 320 km by 2009, when the last maneuver Christensen et al., 2002, 2006, 2008, 2009 and http://www.
brought it back to about 330 km before a final decay to 250 km esa.int/Our_Activities/Observing_the_Earth/The_Living_
and end of operation (and reentry in the atmosphere) on 19 Planet_Programme/Earth_Explorers/Swarm.
September 2010. The satellite advanced 1 h in LT within 11
days. Magnetic field instrumentation was very similar to that 5.02.2.3.6 Calibration and alignment of satellite
of rsted; however, attitude was obtained by combining mea- magnetometers
surements taken by two SIM heads, to minimize attitude error Vector magnetometers, like those flown on Magsat, rsted, and
anisotropy. Accuracy of the (1 Hz) scalar measurements was CHAMP, are not absolute instruments and hence have to be
similar to that of rsted (better than 0.5 nT), while that of vector calibrated on a routine basis (e.g., Olsen et al., 2003). This is
components (both 1 and 50 Hz) was better than 2 nT when done in orbit by comparing instrument output with the magnetic
attitude was measured by both SIM heads (which happened for intensity measurements of a second, absolute scalar, magnetom-
more than 60% of the time); otherwise, the same accuracy as for eter (OVH in the case of rsted and CHAMP). In addition to this
rsted was achieved. Like rsted, CHAMP had these instru- calibration of the vector magnetometer, the relative rotation
ments installed on a boom (4 m long) and carried a GPS between the coordinate system of the vector magnetometer and
receiver, to both exploit atmospheric occultation and provide that of the attitude determination instrument (SIM in case of
rsted and CHAMP) has to be determined with an accuracy of a the magnetic data (the regular daily variation being defined as
few arcsecs, a procedure known as the alignment of the magne- the mean daily variation of the 5 quietest days of each month).
tometer. The three Euler angles (alignment parameters) describing The range is then converted to a K value, taking values between
this rotation are estimated in orbit by comparing the magnetic 0 (quietest) and 9 (most disturbed) on a quasi-logarithmic
components measured by the calibrated vector magnetometer, scale. Using conversion tables based on statistical properties,
combined with the attitude information provided by the SIM, the local K values are then converted to a standardized number,
with the magnetic components predicted from a geomagnetic denoted as Ks. This number is given in a scale of thirds, ranging
field model, usually directly coestimated with the Euler angles through 28 grades in the order 0 , 0þ, 1, 1 , 1þ, 2, 2 , 2 þ,. . .,
(e.g., Olsen et al., 2006b). It is important to recognize that the 8 , 8þ, 9, 9 . The planetary activity index Kp is the mean of
alignment of spaceborne magnetometers is rather different from the Ks value of a number of ‘Kp stations’ (originally 11, pres-
the alignment of observatory magnetometers. The latter is possi- ently 13).
ble without any field model, by turning the magnetometer by, The 3 h index ap is a linearized version of Kp. For a station at
respectively, 180 around the three magnetometer axes and taking about 50 dipole latitude, ap may be regarded as the range of
additional measurements. Proper combination of the magnetom- the most disturbed of the two horizontal field components,
eter readings taken during this procedure, which is performed at expressed in the unit of 2 nT. The daily average of the six 3 h
the same position (no spatial change of the ambient magnetic values of ap is denoted as Ap.
field) and almost instantaneously (within a few minutes, to min- Kp, ap, and Ap are calculated biweekly and are provided at
imize the influence of temporal changes of the ambient field; the http://www-app3.gfz-potsdam.de/kp_index/.
remaining temporal field changes are corrected for by subtracting Only geomagnetic data taken during geomagnetic quiet
the field changes measured by a nearby variometer), removes the conditions are typically used for geomagnetic field modeling.
ambient magnetic field, and hence, no knowledge of the field An often used criterion is Kp 2 , which is fulfilled for about
strength and direction of the ambient field is required. 50% of all data. However, this is the average percentage of
Such a procedure is, however, not possible for satellite quietness; during years of solar maximum, considerably less
magnetometers in orbit, due to movement of the satellite, data satisfy the condition Kp 2 . As an example, only about
and therefore, satellite magnetometer alignment requires a 30% of the solar maximum year 2003 has Kp 2 , while the
model of the ambient field. See Olsen et al. (2003, 2006b) percentage is about 60% for the solar minimum year 1996.
for a description of satellite magnetometer alignment.
5.02.2.4.2 Dst, an index of magnetospheric ring-current

5.02.2.4 Indices
strength
Several indices have been developed for characterizing the level Dst is the north component of the axially symmetrical part of
of magnetic field contributions from ionospheric and magne- the equatorial disturbance field at the Earth’s surface caused
tospheric sources. In the following, we will concentrate on the by the magnetospheric ring current and its induced counter-
two indices Kp and Dst. These are particularly relevant for part. It is determined on an hourly basis from the H compo-
characterizing global geomagnetic disturbances that affect non- nent of the four low-latitude observatories Honolulu, San
polar latitudes and are mostly used for data selection and Juan, Hermanus, and Kakioka, after removal of secular varia-
external field parameterization when deriving models of the tion (defined as the value during magnetically quiet condi-
geomagnetic field. It should be pointed out, however, that tions) and the regular daily variation SR. This procedure is
some modelers also select data at polar latitudes according to somewhat subjective and ignores a possible offset of Dst (i.e.,
indices such as AE or PC, or to the state of the IMF (see Section the contribution of the magnetospheric ring current for
5.02.3). But this is not as common practice, due to the com- Dst ¼ 0), which was determined by Langel and Estes (1985a)
plexity of geomagnetic disturbances in polar regions. For a to be of about 20 nT. This offset and – probably more impor-
detailed description of all these indices, the reader is referred tant – baseline instabilities of the Dst index (e.g., Olsen et al.,
to, for example, Mayaud (1980) and Siebert and Meyer (1996). 2005) make data selection based on (the absolute value of )
Dst questionable. Instead, the first time derivative (which is
5.02.2.4.1 Kp, an index of global geomagnetic activity much less influenced by baseline instabilities compared to
Kp is a worldwide 3 h index that aims at describing the level of Dst itself ) has been used recently. Olsen et al. (2006b), for
‘all irregular disturbances of the geomagnetic field caused by instance, recommended data selection according to |dDst/
solar particle radiation within the 3-h interval concerned’ dt| 2 nT h. Similar to Kp, the distribution of dDst/dt
(Siebert and Meyer, 1996). The concept of this index was first depends also on the level of solar activity, although to a
introduced by Julius Bartels in 1938 and next formalized in smaller extent. The influence of various data selection criteria
1949. It was adopted by the IATME (International Association of on the quality of field models has been investigated by
Terrestrial Magnetism and Electricity, now International Associa- Thomson and Lesur (2007).
tion of Geomagnetism and Aeronomy, IAGA) in 1951. Kp is Note that Dst does contain not only the magnetic signal of
derived from (currently) 13 subauroral stations. For each of the magnetospheric ring current but also contributions due to
these observatories, the local disturbance level is determined induction in the Earth’s interior. Hence, it contains both exter-
by measuring the range (difference between the highest and nal (magnetospheric) and Earth-induced contributions; they
lowest values) during three hourly time intervals for the most can, however, be separated by means of a model of electric
disturbed horizontal magnetic field component. First, how- conductivity of the Earth’s mantle (Maus and Weidelt, 2004;
ever, the regular daily variation, SR, has to be removed from Olsen et al., 2005).
5.02.3 Global Models of the Earth’s Magnetic Field An alternative approach based on the use of combined
regional field representations has also been proposed by
Most global models of the Earth’s magnetic field rely on a Thébault (2006). The idea consists in stitching together a
spherical harmonic representation of the field. Those are the dense coverage of regional models, the resolution of which
models we will describe in some detail. But other global rep- can be adjusted to the regional density of data. This philoso-
resentations can be and have been used, which we also ought phy can in principle be applied to any classical type of
to briefly mention. regional representations, such as the rectangular harmonic
Those first include the so-called natural orthogonal analysis introduced by Alldredge (1981) (see Chapter 5.06
components method extensively used by the Russian school of the present volume for a review of those regional represen-
in the 1970s (a detailed description of which can be found in tations). Thébault (2006) however focused on the use of the
Langel, 1987) and methods based on either dipole or mono- so-called spherical cap harmonic analysis (SCHA). This
pole equivalent sources distributed over a source surface (see, regional representation was first introduced by Haines
e.g., Langel, 1987; Mayhew and Estes, 1983; O’Brien and (1985) and subsequently improved by De Santis (1991),
Parker, 1994). The latter have however found more applica- Korte and Holme (2003), and Thébault et al. (2006) who
tions in the context of crustal field modeling, and the interested successfully overcame the major drawback of the original
reader is therefore referred to Chapter 5.06 of the present SCHA approach, which relied on an incomplete set of non-
volume for more information. orthogonal basis functions and failed to properly describe the
Modeling efforts have also been made to address the rela- altitude (radial) dependency of the field. Thébault et al.
tively poor adequacy of spherical harmonic representations (2006) exhibited the correct and complete set of orthogonal
with respect to uneven geographic data distributions. When a basis functions. This so-called revised SCHA (R-SCHA) now
high concentration of data is to be accounted for in a specific has all the appropriate mathematical properties for efficiently
geographic area, high-degree spherical harmonics are required, building a global field model from a combination of regional
which will represent the field very well in that area but will field models. This technique has been applied to CHAMP
unfortunately generate ringing effects (i.e., oscillations at the data (Thébault, 2006) and will be used in the context of the
corresponding small length scales) wherever else the data den- Swarm mission (Thébault et al., 2013).
sity is lower. Such ringing can be avoided by the use of appro- Spherical harmonic representations however keep many
priate strategies, such as the one developed by Shure et al. advantages. In particular, they are perfectly suited when the
(1982) and Parker and Shure (1982), which consists in relying data to be modeled are reasonably well distributed over the
on so-called harmonic splines. Those splines, which are linear globe, as is the case of the satellite data that make the bulk of
combinations of spherical harmonics, make it possible to fit the contemporary data. Also, they make it easy to accurately
the data at the right level while keeping the model to the separate the field of internal origin from the field of external
minimum complexity required by the local distribution of origin. For all those reasons, we will now review in more detail
the data (by using a Lagrange multiplier type of approach). important examples of recent models based on this represen-
A description of how this method can be implemented is tation (for a review of the many models that have been pub-
provided by Langel (1987) to which the reader is referred for lished in the more distant past, see the well-named Parade of
more details. Models Section 4 of Langel (1987) and Langel and Hinze
When the geographic data distribution is particularly (1998); for other more recent reviews, see also Gillet et al.,
uneven, and/or very localized sources need to be accounted 2010b; Lesur et al., 2011).
for (as is the case for some of the local ionospheric sources
when dealing with satellite data), other types of linear com-
binations can be more appropriate, either for modeling or for
5.02.3.1 Spherical Harmonic Representation of the Field
introducing localized constraints (as proposed by, e.g., Lesur,
2006). One possibility then is to rely on wavelets. Because All those models rely on the following unique decomposition
these have a local geographic support, they indeed make it of a magnetic field B on a sphere of radius r ¼ R surrounded by
possible to keep the number of model parameters low sources (Backus, 1986; see also Backus et al., 1996; Sabaka
by introducing small-scale wavelets only where needed. et al., 2010):
Interesting wavelet frames have, for instance, been introduced
in the context of global magnetic field modeling by BðR; y; fÞ ¼ BiR ðR; y; fÞ þ BeR ðR; y; fÞ þ Btor ðR; y; fÞ [16]
Holschneider et al. (2003) and Chambodut et al. (2005), and where BiR(R,y,f) and BeR(R,y,f) are the potential fields
Mayer and Maier (2006) made a very convincing case of produced on the sphere r ¼ R by all sources, respectively,
the potential of such methods for modeling and analyzing the below and above r ¼ R. Those are in fact the values
small-scale crustal field. Over the past few years, however, those taken for r ¼ R by the potential fields BiR(r,y,f) and
methods have rather been applied to the investigation of BeR(r,y,f), which may be defined more generally for,
the gravity field (see, e.g., Hayn et al., 2012). Another interesting respectively, r R and r R, with the help of BiR ¼ ∇ViR
possibility that has attracted some interest in the recent years and BeR ¼ ∇VeR, with
is that provided by so-called spherical Slepian functions (see,
e.g., Wieczorek and Simons, 2005), which are particularly useful X n
1 nþ1 X
a
when it comes to modeling or characterizing the field over ViR ðr; y; fÞ ¼ a gnm, R Ynmc ðy; fÞ þ hm
n, R Yn ðy; fÞ
ms
r
sharply defined regions, such as continents or oceans (see, e.g., n¼1 m¼0
Beggan et al., 2013). [17]

1 n X
X r n
with
VeR ðr; y; fÞ ¼ a n, R Yn ðy; fÞ þ sn, R Yn ðy; fÞ
qm mc m ms
n 2
a
n¼1 m¼0 a 2nþ4 X 2
[18] Wni ðRÞ ¼ ðn þ 1Þ gnm, R þ hmn, R , [23]
R m¼0
where the index R of the Gauss coefficients is here to recall 2n2 Xn 2 2
R
that those coefficients describe the fields produced by all Wne ðRÞ ¼ n qm
n, R þ s m
n, R , [24]
a m¼0
sources, respectively, below (gm m m m
n,R,hn,R) and above (qn,R,sn,R)
n h 2 i
r ¼ R. The additional field Btor(R,y,f) is then the toroidal nðn þ 1 Þ X 2
field produced on the sphere r ¼ R by the local poloidal cur- WnT ðRÞ ¼ Tnm, c ðRÞ þ Tnm, s ðRÞ : [25]
2n þ 1 m¼0
rents. It is defined by
Equations [21]–[22] then show that each type of field (the
X
1 X
n
Btor ðr; y; fÞ ¼ Tnm, c ðr ÞCnm, c ðy; fÞ þ Tnm, s ðr ÞCnm, s ðy; fÞ , potential field produced by all sources above r ¼ R, the poten-
n¼1 m¼0 tial field produced by all sources below r ¼ R, and the non-
[19] potential (toroidal) field produced by the local (poloidal)
sources on r ¼ R), and within each type of field, each degree n
where (in fact, each elementary field of degree n and order m, as is
Cm , ðc;sÞ ðy; fÞ ¼ ^r ∇ Y m, ðc;sÞ ðy; fÞ ¼ ∇ Y m, ðc;sÞ ðy; fÞr [20] further shown by eqns [23]–[25]) contributes independently
n S n n
to the average squared magnitude hB2(R,y,f)iS(R) on the
and the Ym,c m,s
n (y,f) and Yn (y,f) are the Schmidt quasi- sphere r ¼ R. Hence, plotting Wni(R) (respectively, Wne(R),
normalized real surface spherical harmonics (e.g., Langel, WTn(R)) as a function of n provides a very convenient mean of
1987) (^r being the unit radial vector). In all those equations, identifying which sources, and within each type of source,
the Earth’s radius a is taken as the reference radius. which degrees n, most contribute on average to the magnetic
When R ¼ a, eqn [16] reduces to the sum of the two poten- field B(R,y,f) on the sphere r ¼ R. Such plots are known as
tial fields Bia(a,y,f) and Bea(a,y,f), since the neutral atmo- spatial power spectra.
sphere has no local currents. Then the (gm m
n,a,hn,a) are the sum In the special case when one is only interested in the con-
of the Gauss coefficients describing the field of internal origin tribution of the field of internal origin, the corresponding
(simply denoted (gm m
n ,hn )) and of the analogous Gauss coeffi- spectrum, then known as the Lowes–Mauersberger spectrum,
cients describing the externally induced fields. becomes
When R corresponds to a sphere within the F-region iono-
n h
a 2nþ4 X 2 2 i
sphere where satellites evolve (typically between 400 and Wni ðRÞ ¼ ðn þ 1Þ gnm þ hm , [26]
n
1000 km), local currents are to be found and Btor(R,y,f) is R m¼0
m
nonzero. Then, the (gn,R ,hmn,R) describe the sum of the field of
and can be computed for any value of R a (above the
internal origin, of the externally induced fields, and of the
sources). In the even more special case when one is only
potential field produced by all ionospheric sources located
interested in the core field, exactly the same formula can be
below r ¼ R (the main source of which is the E-region, located
used for R b (where b is the core radius), provided one uses
at about 110 km). The (qm m
n,R,sn,R) then essentially describe the
only core field Gauss coefficients.
potential magnetospheric field and are simply denoted (qm m
n ,sn ).
In discussing the various components of the Earth’s mag-
netic field, and in particular the way each component contrib-
utes on average to the observed magnetic field, it will prove 5.02.3.2 Models of the Field of Internal Origin
useful to deal with the concept of spatial power spectra. This
concept was introduced by Mauersberger (1956) and populari- A first series of models aims at describing the field of internal
zed by Lowes (1966, 1974), both in the case of potential fields. origin only. Such models assume the field to be defined as the
However, it is quite straightforward to introduce those also in gradient of a scalar potential of internal origin. Each model is
the case of nonpotential fields. then defined by a set of Gauss coefficients (gnm(t), hnm(t)), where
Indeed, consider the sphere S(R) of radius r ¼ R and assume some time dependency is introduced to account for the fact
the most general case when this sphere is surrounded by that the field of internal origin varies with time. Those models
sources. Then, the field can be written in the form of eqn do not intend to describe the field of external origin, their
[16], and its average squared magnitude over S(R) can be associated externally induced fields, and other nonmodeled
written in the form fields possibly contributing to the data used for the modeling.
ð ð Yet, those need to be taken into account. This is done by using
2 1 2p p
B ðR; y; fÞ SðRÞ ¼ BðR; y; fÞ BðR; y; fÞ sin y dy df selecting and correcting procedures, by allowing for some sim-
4p 0 0 plified parameterization of the remaining external fields, and
¼ W i ðRÞ þ W e ðRÞ þ W T ðRÞ by using appropriate damping and error covariance matrices,
[21] all in various combinations. Models thus differ because of the
where combination of tricks used to deal with those nonmodeled
fields, because of the type of data used, because of the maxi-
X
1 X
1 X
1
mum degree N of the involved Gauss coefficients, and because
W i ðRÞ ¼ Wni ðRÞ, W e ðRÞ ¼ Wne ðRÞ, W T ðRÞ ¼ WnT ðRÞ,
n¼1 n¼1 n¼1 of the way the temporal behavior of those Gauss coefficients is
[22] described.
5.02.3.2.1 The international geomagnetic reference field 2005.0). The predictive secular variation model that had been
Probably the most used models are those from the IGRF/DGRF used in the last 5 years is then also abandoned and simply
series, regularly published by the International Association of Geo- replaced by a linear interpolation between this DGRF model
magnetism and Aeronomy (IAGA). Those are (mostly empirical) (DGRF 2005.0) and the latest IGRF model (IGRF 2010.0), for
representations of the slowly changing large-scale part of the the period between those two models (while, of course, the
field of internal origin. Their maximum degree used to be new secular variation model published with the new IGRF is
N ¼ 10, but this was changed to N ¼ 13 in 2000.0 for all models used to predict the future evolution of the field).
published since. Likewise, whereas Gauss coefficients used to be The most recent IGRF release (IGRF-11; see Finlay et al.,
given to within 1 nT accuracy, these are now given within 2012) hence consists in 20 DGRF models for epochs 1900.0 to
0.1 nT. Each model is defined for a specific epoch and therefore 1995.0 up to degree N ¼ 10 (one every 5 years), two DGRF
gives an instantaneous picture of the field every 5 years, starting models for epochs 2000.0 and 2005.0 up to degree N ¼ 13, and
in 1900.0 (i.e., 1 January 1900). Estimates of the field between one IGRF for epoch 2010.0 up to degree N ¼ 13, with a predic-
those epochs can then be derived by simple linear interpolation. tive secular variation model up to degree N ¼ 8. The next revi-
Every time a new IGRF model is computed (the most recent sion will occur in 2015.
version is IGRF-11, computed for epoch 2010.0; see Finlay The great advantage of the IGRF/DGRF series is their sim-
et al., 2012), all the data covering the past few years are taken plicity and predictive nature. But the chosen temporal and
into account, including geomagnetic measurements from sat- spatial parameterization is limited and one needs to be aware
ellites (when available), geomagnetic observatories, and repeat of those limitations. A retrospective analysis of IGRF 2005.0
stations. Surveys taken by ships and aircrafts have also some- (compared to the a posteriori DGRF 2005.0 model released
times been used. Candidate models in the form of Gauss coeffi- with IGRF-11; see Finlay et al., 2010a) reveals a formal root
cients (gm m
n ,hn ) are then produced by various independent mean square error over the Earth’s surface of only 12 nT. But
teams, usually using simple classical techniques. However, this ignores all contributions of the field beyond degree 13 and
when satellite data are available, as is the present case, such local sources can easily amount to several 100 nT, if not much
models are often truncated (and perhaps preliminary) versions more (see, e.g., Cohen et al., 1997). Obviously, one should
of more advanced models of the type described in the next expect similar limitations to affect IGRF 2010.0 released with
sections. Those candidate models are next assessed (see IGRF-11. An analogous analysis of the 2005.0–2010.0 IGRF-10
Chulliat and Thébault, 2010; Finlay et al., 2010a) and the predictive secular variation model (compared to the a poster-
IGRF model is finally derived as a (weighted) mean of those iori DGRF 2005.0–IGRF 2010.0 field evolution; see again
candidate models. This procedure is specific to the IGRF series Finlay et al., 2010a) further revealed a formal root mean square
of models and amounts to some sort of best guess approach, error over the Earth’s surface of 20 nT year1, again indicative
based on the empirical experience of the IAGA working group of the likely errors affecting the predictive secular variation
in charge of producing the model. released with IGRF-11. In practice, this means that the IGRF-
The main purpose of the IGRF models is practical (see 11 prediction for epoch 2015.0 (when IGRF-12 will have to be
Meyers and Davis, 1990 for a review of the many, sometimes released) will then likely be in error by more than 100 nT at the
unexpected, applications of the IGRF models). They are Earth’s surface.
meant to predict the field of internal origin (in fact, the Model coefficients and software to synthesize field values at
main field, given the low degrees modeled) not only at a given time and position are available at the World Data
the epoch of validity (say, 2010.0) but also for the 5 years to Center (WDC) system (e.g., http://www.ngdc.noaa.gov/
come. This is possible thanks to the secular variation model IAGA/vmod/index.html). More details about IGRF/DGRF

(i.e., the g_ m ; h_ models can be found in Langel (1987) and Macmillan and
m
n n coefficients), published together with the
Finlay (2011). Note that a very similar model (up to N ¼ 12 for
fixed epoch Gauss coefficients. This secular variation model is both the main field and the predicted secular variation), with
also a mean of candidate secular variation models, usually co- similar specifications, but more specifically designed for
estimated with the field candidate models (because the data US/UK Defense and NATO use, is also concurrently released
used usually cover a significant time span and some temporal every 5 years (see http://www.ngdc.noaa.gov/geomag/WMM/
corrections are anyway needed). This model, currently of DoDWMM.shtml and Maus et al., 2010a).
maximum degree N ¼ 8 (but this could change in the future;
see, e.g., Finlay et al., 2010a; Silva et al., 2010), is thus an
estimate of the average temporal variation of the field in the 5.02.3.2.2 More advanced models
few years preceding the release of the model, next extrapo- Taking advantage of data obtained by the recent magnetic
lated to predict the field over the next 5 years, that is, until the satellites rsted, CHAMP, and SAC-C, many geomagnetic
next IGRF model is released. field models of increasing complexity and accuracy have now
IGRF models are bound to be temporary. Once the next also been derived.
model has been released (some 5 years later), the previous The rsted Initial Field Model (OIFM) of Olsen et al. (2000a)
IGRF is reconsidered, data that have become available in the was one of the first to be published. Similar to the IGRF model
mean time are taken into account, and a new Definitive Geo- series, it represents a snapshot of the magnetic field at a given
magnetic Reference Field (DGRF) is published. This new model epoch t0 (in this case t0 ¼ 2000.0), which was estimated from a
is no longer a temporal extrapolation, but an interpolation of few weeks of quiet nighttime rsted data around 1 January
the field. It is again computed for the same epoch and super- 2000. Data were selected according to the following criteria
sedes the old IGRF (DGRF 2005.0 recently superseded IGRF (recall Section 5.02.2.4 for definitions and purposes of
indices): LT at about 22:00 to minimize ionospheric contribu- not model the temporal variation of the field of internal origin
tions; Kp, the index of global geomagnetic activity, and Dst, the (such as OIFM), or only model a constant secular variation.
ring-current index, have to satisfy Kp 1 þ for the time of But as modern high-quality satellite data became available over
observation, Kp 2 for the previous 3 h interval, and |Dst| < longer and longer periods of time (they now have been avail-
10 nT and |d(Dst)/dt| < 3 nT h1, to ensure the data are col- able for more than 14 years), it became clear that this time
lected at magnetic quiet times. An additional criterion based variation had to be modeled in a more sophisticated way.
on the state of the IMF (as provided by Advanced Composition Some models have simply relied on a piecewise linear time
Explorer – ACE – NASA/NOAA spacecraft) was used to further dependence analogous to the IGRF, but with closer time spac-
reduce contributions from polar cap ionospheric currents. ing (typically on the order of a year). This is the case for the
Finally, only intensity data have been used at high magnetic so-called MEME08 model (Thomson et al., 2010) and its latest
latitudes (|90 yd| 50 ), to minimize the nonmodeled sig- update (used as a parent model for an IGRF candidate;
nature of field-aligned currents (FAC). Hamilton et al., 2010), both of which have rather focused on
Spatial resolution of the OIFM model goes beyond that of improving data weighing schemes (see also Lesur et al., 2011,
the IGRF, since spherical harmonics up to degree N ¼ 19 are for a discussion). Others, however, have been using a higher-
included. Some simple parameterization of the magneto- order Taylor expansion around a specific epoch t0, assuming
spheric field is also included, which is modeled as a permanent
1 m
potential field of external origin up to degree 2 and as a fast gnm ðt Þ ¼ gnm jt0 þ g_ m € j ðt t0 Þ2 . . .
n jt0 ðt t0 Þ þ g [27]
2 n t0
time-changing degree 1 potential field of external origin with
induced degree 1 potential field, the temporal variability of (and similar for hm
n ), or a spline representation of the type
which is imposed as an independent information via the
X
L
observatory-based Dst index (as a measure of the strength of gnm ðt Þ ¼ gnm, l Ml ðt Þ [28]
the magnetospheric ring current), following an earlier sugges- l¼1
tion of Langel and Estes (1985b). Finally, covariance matrices
taking into account correlated attitude errors have been used with Ml(t) as basic B-spline functions (De Boor, 2001;
(based on Holme and Bloxham, 1996). The inverse method is Schumaker, 1981) and L as number of basis functions. Such
a simple iterative least-square fit (iterative, since scalar data are B-splines have first been introduced in the context of historical
included) and does not involve any damping. Note also that field modeling (Bloxham and Jackson, 1992; Jackson et al.,
this model does not involve any secular variation (the data 2000; see also Chapter 5.05 of the present volume). Indeed,
used covering only a very short period of a few weeks and while a Taylor expansion including linear and quadratic terms
having been propagated to model epoch t0 ¼ 2000.0 by may be adequate for describing field changes over short time
means of an a priori model of linear secular variation). spans of a few years, a spline representation is now known to
Similar models were subsequently published by Langlais usually be preferable for periods longer than, say, 5 years.
et al. (2003) and Olsen (2002), but using much more (night- Recent models using a Taylor expansion (eqn [27]), up to
time) data. Langlais et al. (2003) used 14 months of rsted quadratic terms, include the POMME model series (Maus et al.,
data (between March 1999 and May 2000) and computed a 2005, 2006b, 2010b). In its latest version (POMME-6; Maus
model up to degree N ¼ 29 in very much the same way as the et al., 2010b; see also http://geomag.org/models/ for possible
OIFM was computed, except for the fact that a secular variation later updates), this model uses nighttime (except for high
model was then also coestimated (up to degree N ¼ 13). Olsen latitudes, where scalar data at all LTs were used) CHAMP vector
(2002) used even more data (between March 1999 and and intensity data between July 2000 and August 2009. It goes
September 2001). This model is also computed up to degree up to degree N ¼ 60 but only Gauss coefficients up to degree

N ¼ 16 include nonzero first g_ m ; h_ and second g€m ; h€
m m
N ¼ 29, with a secular variation up to degree N ¼ 13. But a n n n n
slightly different modeling approach was used, involving an
time derivatives. Also, some damping was introduced in the
improved statistical treatment of data errors – relying on an
inverse procedure to keep those time derivatives well behaved.
iteratively reweighed least-squares (IRLS) inverse scheme with
This damping was imposed on degrees 14–16 for the first time
Huber weights (Constable, 1988; Holland and Welsch, 1977;
derivative and on degrees 9–16 for the second derivatives. One
Walker and Jackson, 2000) – and a more advanced way of
additional particularity of POMME-6 is that it in fact consists
dealing with external field contributions. Permanent large-
in two successive models, with a discontinuity in the field
scale magnetospheric contributions were still estimated up to
acceleration at epoch 2005.0 (i.e., with two different sets for
degree 2, but additional annual and semiannual periodicities
g€m ; h€ in eqn [27] before and after 2005.0), illustrating the
m
were included in their zonal terms. Also, fast-changing degree 1 n n
coefficients (with induced counterparts), previously modu- limits of relying on a single Taylor expansion to describe the
lated by the Dst index, were modulated with a new, more field at times of rapid changes in the field acceleration (see also
appropriate RC~ index also estimated from globally distributed Thébault et al., 2010a). This model is also more sophisticated
geomagnetic observatories. Exactly the same procedure was than the models described so far in several other respects. In
later used to compute the first global model of the field of addition to data selection criteria fairly similar to those already
internal origin based on CHAMP data (Holme et al., 2003). described earlier, Maus et al. (2010b) relied on some compar-
This CO2 model used CHAMP and rsted vector and scalar isons with the predecessor POMME-5 to identify and reject
data, between August 2000 and December 2001, and SAC-C ‘noisy’ satellite tracks (POMME-5 itself was built by relying
scalar data throughout 2001. on an additional selection criteria proposed by Stolle et al.
The models discussed so far (and others that we did not (2006) to detect and exclude postsunset low-latitude iono-
describe in detail, such as that of Lesur et al., 2005) either do spheric plasma instabilities, which produce nonmodeled
local poloidal fields). Corrections in the Euler angles describ- CHAOS-4a (see http://www.spacecenter.dk/files/magnetic-
ing the transformation from the magnetometer frame to the models/CHAOS-4/). This model has been built in the same
SIM frame were introduced (following a procedure analogous way as CHAOS-3, except that it now uses more than 11 years of
to the one first introduced by Olsen et al. (2000b) and also data (and spans 1997.0–2011.0) and that its static high-degree
used in the context of the CHAOS series of models; see later coefficients (above n ¼ 25) have been removed and replaced by
text). Corrections for predictable fields were also implemented, those from a second independent model that only uses the
namely, the magnetic effect of the eight major oceanic tidal final low-altitude 2009–2010 CHAMP data and a simple linear
components (Kuvshinov and Olsen, 2005) and the diamagnetic secular variation parameterization. This made it possible to
effect (Lühr et al., 2003) (but not the signature of gravity-driven recover more static spherical harmonic coefficients (up to
currents in the ionospheric F-layer, as originally proposed by degree N ¼ 100). This model is the CHAOS model we will
Maus and Luhr (2006) and tentatively implemented in the later use for illustration purposes.
earlier POMME-3 version of Maus et al. (2006b)). Magneto- A second comparable series of models has also recently
spheric sources have also been taken into account: first by sub- been initiated under the acronym of GRIMM by Lesur et al.
tracting predictions from a quiet-time magnetospheric field (2008), the second version of which is GRIMM-2 (Lesur et al.,
model expressed in appropriate coordinate systems (i.e., SM 2010b). These models also rely on B-splines (of order 5 in
coordinates for describing near magnetospheric currents like the case of GRIMM and order 6 for GRIMM-2) to describe
the ring current and GSM for describing far magnetospheric the time-varying field up to degrees 14 (GRIMM) and
current systems like the tail currents) and using independent 16 (GRIMM-2), with higher degrees assumed constant up to
data as input (essentially the Dst index and the IMF, but see degrees 60 (GRIMM) and 30 (GRIMM-2). Note, indeed, that
Lühr and Maus, 2010 for details) and next by coestimating GRIMM-2 has been derived with the goal of more specifically
daily baseline corrections for the Dst index (along the lines of improving the description of the time-varying component of
Lesur et al., 2005 and Olsen et al., 2005). the field, in a two-step process that involved first subtracting an
The CHAOS series of models, initiated by Olsen et al. initial lithospheric field model. Details of the methods used to
(2006b), provides examples of models based on the alternative build these models can be found in both Lesur et al. (2008,
spline approach (eqn [28]). Its last published version, 2010b) and Lesur et al. (2011), where differences (substan-
CHAOS-3 (Olsen et al., 2010b), was derived from more than tially more technical than can be reported here) in these
10.5 years of high-precision geomagnetic measurements from methods and the methods used to build the CHAOS series
the three satellites rsted, CHAMP, and SAC-C taken between and the MEME08 model are also discussed. Just like the
March 1999 and December 2009. It also uses annual differ- CHAOS and POMME series of models, GRIMM is currently
ences of observatory monthly means between January 1997 evolving, and the interested reader is invited to consult http://
and June 2009 time interval, corrected for ionospheric contri- www.gfz-potsdam.de/portal/gfz/Struktur/Departments/Dep
butions (using the CM4 model; see later text) and magneto- artmentþ2/sec23/topics/models/GRIMM for possible updates.
spheric contributions (using CHAOS-2, Olsen et al., 2009,
predecessor of CHAOS-3). Satellite data selection criteria were
5.02.3.3 Models of the Core Field
similar to those described so far. The model goes up to degree
N ¼ 60, but all degrees above degree 20 are assumed static. The All models discussed so far are models of the field of internal
time change of the low-degree (n 20) coefficients is described origin. But this field is the sum of the low-frequency core field,
by order 6 B-splines (with 0.5 year knot separation). In that which reaches the Earth’s surface (and makes the main field)
case, also some damping is used to keep time derivatives well and of the crustal field. Fortunately, we know that the crustal
behaved, by minimizing the squared magnitude at the core sources can only lie very near the Earth’s surface (above the
surface of (1) the third time derivative of the field over the Curie isotherm), in layers with reasonably well-known mag-
1997–2010 time period and (2) the second time derivative of netic rock properties. This makes it possible to predict the likely
the field at the two model end points (1997 and 2010). The contribution of the crustal field to the observed field of internal
modeling approach also takes advantage of lessons learned origin. Such predictions can be made either in a statistical way
from earlier models. In particular, the magnetometer vector (e.g., Jackson, 1994) or in a deterministic way (e.g., Purucker
data are taken in the instrument frame, and the Euler angles et al., 2002). In both cases, the conclusion is the same: The
describing the transformation from the magnetometer frame to crustal field can easily explain the small scales of the field of
the SIM frame (i.e., the so-called alignment parameters; recall internal origin but cannot explain its largest scales. The transi-
Section 5.02.2.3) are coestimated, avoiding the inconsistency tion occurs around the degree 14 of the spherical harmonic
of using vector data that have been aligned using a different representation of the field and shows up nicely as a strong
(preexisting) field model. The bending of the CHAMP optical inflection point in the spatial Mauersberger–Lowes spectrum
bench connecting magnetometer and SIM is also accounted for (recall eqn [26]) of the field of internal origin at the Earth’s
by estimating CHAMP Euler angles in 10 day segments. Finally, surface (see Figure 13(a) in Section 5.02.4.1, where this spec-
in addition to describing magnetospheric fields using the SM trum is further discussed; see also Chapter 5.06 of the present
and GSM coordinate systems, degree 1 external fields are co- volume, where predictions of the crustal field contributions are
estimated separately for every 12 h (for the axial q01 term) and illustrated). This then shows that the crustal field is dominating
5-day (for q11 and q11) intervals. The same improved statistical the small scales (say, degree N ¼ 15 and above) of the field of
treatment of data errors, involving an IRLS inverse approach internal origin, to which it can be identified, while the core
with Huber weights as in Olsen (2002), is otherwise used. field is dominating the large scales (say, degree N ¼ 13 and
Since its publication, CHAOS-3 has already partly been below), to which it can also be identified. This of course also
updated into a preliminary CHAOS-4 version, known as means that the smallest scales of the core field are permanently
screened by some unknown crustal field and that the largest used to infer the final model. Finally, the inverse scheme also
scales of the crustal field cannot be observed. When consider- uses some regularization procedure based on some a priori
ing the first (and higher) time derivative of the core field, information with respect to quantities such as the Ohmic
however, crustal screening is not as severe. Although some dissipation in the core and to the temporal behavior of the
temporal changes can occur in the large-scale crustal field field (analogous to those used in the context of historical field
(because of the contribution of the magnetization induced by modeling; see Chapter 5.05). These regularizations and the
the core field), their contribution to the observed secular var- B-splines used are slightly different in the initial and extended
iation is unlikely to dominate the core field signal before C3FM model, which uses order 6 B-splines and minimizes the
degree N ¼ 22 (Hulot et al., 2009). Recovering core field secular squared magnitude of the third time derivative of the field at
variation even beyond degree 13 and up to degree 22 is thus the core surface (similar to what was done for CHAOS-3; recall
possible, at least in principle. earlier text), both of which have proved beneficial to better
Because the transition at degree 14 in the spatial capture fast changes in the core field, its secular variation, and
Mauersberger–Lowes spectrum is so sudden, identifying the its acceleration.
large-scale core field by simply truncating the field of internal As an alternative to the extended C3FM core field model
origin at degree 13 is not an unreasonable thing to do. But this built in the way described earlier (to which the authors refer as
has some potential drawback. When plotting maps, this will model 1), Wardinski and Lesur (2012) also built a second
distort large-scale core field features that can only be described model (model 2), using exactly the same data set, essentially
by Gauss coefficients around degree 13. Obviously also, time the same modeling methodology, but adapted (following
variations of the core field above degree 13 are then lost. These Lesur et al., 2010a) to force the field temporal evolution to
considerations led several authors to try and derive core field be driven by a time-varying core surface flow model under the
models directly from the data, relying on a variety of methods so-called frozen-flux approximation (this amounts to neglect
to avoid crustal field contamination and yet properly recover as the role of magnetic field diffusion in defining the core field
much information as possible. evolution over the time period under consideration; see
Such methods have first been used in the context of histor- Chapter 5.05 of the present volume). The core surface flow
ical field modeling, the distribution of historical data being model is then simultaneously recovered with the core field
anyway not good enough to accurately recover the field of model. This model is a first simple example of how some a
internal origin beyond degree 13. Examples of historical core priori information on the equations driving the field evolution
field models produced in this way can be found in Chapter can be taken into account to constrain the field evolution (see
5.05 of the present volume. However, quite a few models of the Section 5.02.5, for more advanced examples within the frame-
recent and present core field have now also been published, work of data assimilation). It also provides an interesting way
introducing, using, and testing a number of interesting ideas. of testing the role of magnetic diffusion in the core field evo-
The first series of models, known as C3FM, is the one lution over the period of interest (see also Chulliat and Olsen,
initiated by Wardinski and Holme (2006) and covering the 2010 and Jackson et al., 2007b, for alternative approaches over
1980–2000 time period, recently extended by Wardinski the same period of time and over historical times, and Chapter
and Lesur (2012) to cover the 1957–2008 time period. 5.05 of the present volume).
These models only have Gauss coefficients up to degree A second series of core field models of interest is the one
N ¼ 15 (N ¼ 14 for the extended version), the temporal varia- initiated by Jackson (2003) and Jackson et al. (2007a), who
tions of which are described with the help of B-splines. An specifically focused on the problem of improving the recovery
important characteristic is that magnetic observatory monthly of core field structures of intermediate spatial scales (described
means, annual means, and repeat station measurements are by Gauss coefficients with degrees both below and above
used in such a way that only field changes in time are taken degree 13) that are partly hidden by the crustal field. In these
into account. Satellite data also are used but only in an indirect initial studies, a pair of models for epochs 1980 and 2000 were
way, by requesting the field model to be as close as possible to built. Data used were a few months of Magsat data for epoch
truncated versions of a Magsat model (Cain et al., 1989) in 1980 and December 1999 to January 2000 rsted data for
1980 and of either the rsted model of Olsen (2002) in 2000 epoch 2000, which are short enough periods for the secular
(for C3FM) or the GRIMM model of Lesur et al. (2008) in 2004 variation to be ignored in the model (the data being simply
(for the extended C3FM). The underlying idea is that in doing propagated to model epoch with the help of IGRF). The data
so, the crustal signal is entirely removed from the data used, selection process is fairly standard for both the Magsat and
while it already is minimized in the low degrees of the satellite rsted data. Few corrections are introduced, only for the exter-
data-based models used as constraints. The main source of nal field, essentially following the philosophy of the OIFM
error linked to nonmodeled field sources then remains the model of Olsen et al. (2000a) described earlier. Contributions
field of external origin with its induced counterpart. Contrary from the crustal field, in particular, are kept minimum by
to the models discussed earlier, none of those fields are either simply using a coarse enough density of satellite data (1600
corrected for or even crudely modeled. They are dealt with as a Z components from Magsat and 3684 vector data from
source of error via an original two-step procedure. An interim rsted).
model assuming this error is uncorrelated (among the compo- The way these models are computed is otherwise quite
nents of each vector data) is first computed. This model is then different from what we have seen so far. Although the models
used to study residuals and infer a more sophisticated error are eventually expressed in the form of Gauss coefficients, they
covariance matrix, taking the so-inferred correlations among are directly computed at the core surface, which is tessellated
components into account. This new covariance matrix is next into 1442 almost equally spaced nodes and 2880 spherical
triangles (enough to describe the field up to degree N ¼ 30). models of the field of internal origin. This only requires remov-
This then makes it easy to implement a so-called maximum ing core field-dominated Gauss coefficients, up to, say, degree
entropy method (Buck and Macaulay, 1991), which is well 14. This procedure removes the large-scale crustal field, which
suited to solve such image reconstruction type of problems. we anyway cannot observe. It also unfortunately distorts
Rather than looking for a core field model crudely truncated at crustal field features of intermediate length scales (which
a given degree with no regularization (which would distort would need to be described by Gauss coefficients around
partly resolved medium-scale features), or for a model pro- degree 14). This is a well-known issue in the community
duced with the help of some regularization (which would involved in the study of crustal magnetism (see Chapter 5.06
smooth out those features), this approach makes it possible of the present volume). Obviously, small-scale crustal field
to look for as many high-degree Gauss coefficients as tolerated features will not be affected by those issues. The concern in
by the data, to sharpen those features. It is most appropriate if that case has more to do with the limited spatial resolution of
one feels that core field features that are only partly resolved those models. The models discussed so far only go to maxi-
because of the overlying crustal signal are intrinsically sharp. mum degree N ¼ 60 (as is the case for the POMME-6.0 model
That may well be the case, because dynamo simulations do of Maus et al. (2006b), the CHAOS-3 model of Olsen et al.
show such sharp features (e.g., Christensen et al., 1999). One (2010b), and the GRIMM model of Lesur et al. (2008)). In the
should however keep in mind that such a method may very quest of even better spatial resolutions, spherical harmonic
well sharpen features that are either not so sharp in reality or models that only attempt to describe the crustal field beyond
even more complex than reconstructed by the method. All degree 15 have also recently been derived. (Note that the
methods have their drawbacks. They simply provide alternative reason model CHAOS-4a, already described earlier, goes up
views. Interestingly, we note that hardly any significant to N ¼ 100 is because it combines two models, with one spe-
changes seem to have occurred in the so-called unsigned flux cifically targeting the crustal field; recall Section 5.02.3.2.)
between 1980 and 2000, as inferred from these Jackson (2003) An interesting series of such models has been derived by S.
and Jackson et al. (2007a) models. As further explained in Maus and coworkers, mainly using CHAMP data. The last
Chapter 5.05 of this volume, this is yet another (global) indi- published version of that series, MF6 (Maus et al., 2008b),
cation that diffusion currently contributes little to the global goes up to degree N ¼ 120 and uses CHAMP scalar and vector
evolution of the core field. measurements from January 2004 to September 2007. Build-
Recently, this maximum entropy method was also extended ing on the experience of past similar models (in particular MF4
to time-varying models of the core field directly described in (Maus et al., 2006a) and MF5 (Maus et al., 2007)), consider-
terms of time-varying Gauss coefficients. This was first done in able care has been taken to avoid unmodeled field contribu-
the context of historical field modeling (see Gillet et al., 2007 tions, first via selection and correction procedures comparable
and Chapter 5.05 of this volume) and next applied to the to that used for POMME-6 model previously described (which
2000–2010 time period (Finlay et al., 2012). As a matter of was produced by the same team) and next by relying on a
fact, these authors produced three alternative models, to which number of dedicated procedures.
they refer as the gufm-sat series, which all use exactly the same The data were selected based on Kp 1þ and LT between
data set and correction procedures (largely based on those used 22:00 and 05:00. At high latitudes, the effects of postsunset
for the CHAOS-3 model described earlier, plus the removal of a low-latitude ionospheric plasma instabilities were identified
(Stockmann et al., 2009) preexisting model of the known part and excluded by an automatic detection approach (Stolle
of the global crustal field) and same parameterization (order 6 et al., 2006), and data have been used only if the merging
B-splines, spherical harmonics up to degree N ¼ 24) but rely on electric field at the magnetopause satisfies Em < 0.8 mV m1
different regularization schemes. Two models use schemes (following a suggestion of Ritter and Lühr, 2006). Additional
analogous to those discussed previously: for gufm-sat-Q2: selection criteria were also used, based on the status of the star
square of the radial component of the core field averaged cameras onboard CHAMP and on some comparison with pre-
over the core surface and over time as a spatial regularization dictions from the POMME-4 model. The data have next been
together with square of the radial component of the second corrected for the magnetic effect of the eight major oceanic
time derivative of the core field averaged in the same way as a tidal components (Kuvshinov and Olsen, 2005) and the dia-
temporal regularization; for gufm-sat-Q3: – same spatial magnetic effect (Lühr et al., 2003). Nonpolar data are also
regularization but averaged square of the radial component corrected for the effect of the polar electrojets as was done for
of the third time derivative of the core field, plus end point MF5 (see Maus et al., 2007). Since the model only intends to
regularization for the second time derivative (as for CHAOS-3). describe the crustal field starting from degree 16, contributions
The last model, gufm-sat-E3, is the one built using maximum from lower degrees of the field of internal origin had to be
entropy instead of spatial regularization and the same tempo- removed. This was done by removing predictions from the
ral regularization as gufm-sat-Q3. All these models provide very POMME-4.0 model (an unpublished predecessor of POMME-
useful complementary views of how the core field actually 6; see http://geomag.org/models/pomme4.html) up to degree
behaved over the recent past. 15. Magnetospheric contributions (with their induced counter
parts) as parameterized in POMME-4.0 were also removed. For
vector data, all tracks have next been high-pass-filtered by
5.02.3.4 Models of the Crustal (Lithospheric) Field
removing a degree 1 (external plus internal) field on an orbit-
Spherical harmonic models of the crustal field (also referred to by-orbit basis. In addition, a set of 3-Euler angle corrections,
as the lithospheric field, when global models are considered, accounting for a remaining uncertainty in the satellite attitude,
such as here) can also straightforwardly be recovered from have been estimated for each orbit of CHAMP vector data.
For scalar data, a simpler procedure was used. Finally, a line- of the best pictures so far of small-scale crustal (lithospheric)
leveling algorithm, which minimizes the distance-weighted features resolvable from satellite data. However, the reader
misfit of the residuals against MF5, between all nearest pairs should be aware that some north–south trending features of
of measurements for all pairs of tracks, has been applied. This the crustal field could have been removed by the processing
selected and corrected data set has then been used to compute scheme, in particular by the high-pass along-track filtering,
Gauss coefficients up to degree N ¼ 120, using a standard least- which has been shown to also possibly introduce additional
squares approach. Coefficients above degree n ¼ 80, however, artifacts (Thébault et al., 2012). Illustrations of these issues can
were regularized. Finally, since models built using along-track be found in Thébault et al. (2010b), who compared MF5 and
filtering have been shown to suffer from some loss of signal, MF6, produced in this way, with CM4, the comprehensive
the initial model built in the way described earlier was eventu- model of Sabaka et al. (2004), which we introduce in some
ally merged with the POMME-4.0 model, the degrees 16–54 of detail in the next section. Also worth mentioning in this respect
which were simply substituted, to build the final MF6 model is the recent alternative and promising approach developed by
extending from degree 16 to degree 120 (for more details on Stockmann et al. (2009) that avoids along-track filtering and
this fairly complex sequence of treatments, see Maus et al., uses a regularized spherical triangle tessellation together with
2008b). the maximum entropy approach already introduced in the
Most recently, yet another update of the MF series was context of core field modeling by Jackson (2003) and Jackson
produced in a very similar way but using CHAMP scalar and et al. (2007a) (see Section 5.02.3.3 earlier).
vector measurements from the latest May 2007 to April 2010 Of course, many more detailed regional information about
time period (to take advantage of a lower flight altitude). It the crustal field can also be obtained from local magnetic
now formally goes up to degree N ¼ 133. This MF7 model is surveys, and a lot of efforts have been put in the recent years
unpublished, but available from http://geomag.org/models/ to patch this information together and with global models
MF7.html. such as those we just described. It should be stressed, however,
MF7 further improves on MF6 and is arguably better than that a very significant gap still exists between the scales of the
its predecessors, as is demonstrated in Figure 8. It provides one lithospheric field that can recovered from satellite data and
MF1 MF3 MF5 MF7
-200 -150 -100 -50 0 50 100 150 200

Vertical component of magnetic field (nT)
Figure 8 Improvement in resolution from MF1 to MF7 for India (top) and the region between Antarctica and Australia (bottom). Courtesy of S. Maus.
those that can be recovered from magnetic surveys (for reviews, E-region ionosphere (see, e.g., Lühr et al., 2002; Maus and
see, e.g., Thébault et al., 2010b and Chapter 5.06 of the present Lühr, 2006; Stolle et al., 2006).
volume). This gap is one of the motivation of the Swarm Finally, the potential field ultimately identified as being
mission, the design of which precisely aims at reducing this produced by sources below the Earth’s surface is still anyway
gap (see, e.g., Olsen et al., 2010a). the sum of the real field of internal origin (i.e., the sum of the
core field and of the crustal field) and of externally induced
fields.
5.02.3.5 Comprehensive Models
Fortunately, all those limitations can be overcome, thanks
Over the past two decades, a very different approach to geomag- to a number of favorable circumstances the comprehensive
netic field modeling has been developed jointly by the mag- modeling approach judiciously takes advantage of.
netics groups at Goddard Space Flight Center (NASA/GSFC) and First is the fact that the most limiting fast changes in the
DTU Space (formerly, Danish National Space Center, DNSC). field of external origin arise during magnetic disturbed times.
This approach, known as the comprehensive modeling (CM) Like all models we have seen so far, the comprehensive models
approach, takes advantage of the spatiotemporal properties of therefore currently rely on some indices (in particular Kp and
the various components of the near-Earth magnetic field, to Dst) to select quiet-time data and minimize those fast-changing
make the best of observatory and satellite data, and coestimate signals. Note that as a result, the present comprehensive models are
models of the internal, magnetospheric, and ionospheric fields quiet-time magnetic field models.
(with their secondary, Earth-induced counterparts), in one huge This selection procedure has two additional advantages.
inversion process. As a result, the comprehensive modeling One, already recognized in previous modeling approaches, is
approach differs in many ways from what we have seen so far. that the remaining fast-changing field will essentially take the
Indeed, all previous models relying on satellite data essentially form of a dipolar contribution to the magnetospheric field,
used observatory data via the Kp and Dst indices and often the variability of which can be parameterized by the Dst index.
ignored the much richer information provided by the raw obser- The other is that the remaining magnetospheric, and all of the
vatory data. None took advantage of the theoretical possibility ionospheric, quiet-time fields are then essentially fixed with
of modeling the ionospheric field by ‘sandwiching’ it between respect to the Earth–Sun line. They therefore produce regular
ground-based and spaceborne observations. The comprehensive temporal variations with restricted known temporal periodicities.
modeling approach precisely intends to take advantage of this More advantage can then also be taken of the spatial prop-
possibility. erties of the quiet-time ionospheric and magnetospheric fields.
First, the E-region sources can indeed reasonably be assumed
5.02.3.5.1 Modeling philosophy to take the form of a sheet current at fixed (110 km) altitude.
In principle, if one assumes the ionospheric sources to essen- Next, those fields can be expressed with the help of relatively
tially lie in the form of a current sheet in the E-region (at about few parameters, provided appropriate coordinate systems are
110 km altitude), and in the form of poloidal sources in the being used (QD for the ionospheric field and dipole for the
F-region (up to 1000 km) where they then only produce local magnetospheric field). The smallest modeled scales are then
toroidal fields, and if the field can be simultaneously deter- essentially those associated with the strong dayside E-region
mined everywhere at the Earth’s surface and on a sphere of ionospheric current produced by the equatorial electrojet (EEJ)
radius r ¼ R within the F-region, then the potential field pro- flowing along the magnetic equator and F-region currents, all
duced by all sources below the Earth’s surface, the potential of which are low-order features in QD coordinates. Although
field produced by the E-region ionospheric sources, the toroi- the associated fields then need to be represented with the help
dal field locally produced by the F-region ionospheric sources of high-degree (n) spherical harmonics, only low-order (m)
at altitude r ¼ R, and the potential field produced by all sources terms are required, which keeps the number of parameters
above r ¼ R can all be recovered. low. Also, and this of course is important, those features have
But can observatory and satellite data provide enough infor- a geometry the distribution of observatories and a few days of
mation at the Earth’s surface and on a sphere r ¼ R, to achieve satellite data are able to resolve.
such a separation of the various fields in practice? Yes, pro- A smart parameterization can thus be used to describe the
vided a fairly elaborate approach is being used, because of a field of external origin as the sum of a fast-changing essentially
number of practical limitations. dipolar magnetospheric field modulated by the Dst index and
One is that observatories do not measure the field every- of almost Sun-synchronous, fixed geometry quiet-time
where at the Earth’s surface. Their actual geographic distribu- ionospheric (both for E- and F-regions) and magnetospheric
tion is far from perfect (recall Section 5.02.2.2), and this limits modes. Those modes are not exactly Sun-synchronous because
the spatial resolution with which the Earth’s surface is sampled. at any instant, they are best represented in Earth-fixed dipole
Another is that satellites need time to complete a decent and QD coordinates. When converted into the standard
coverage of the Earth. A single orbit is completed in about spherical harmonic Earth-related geographic representation
90 min and a few days are typically needed to recover data (i.e., the solid Earth’s reference frame), they appear as oscillat-
with a spatial distribution commensurate with that of the ing modes with diurnal and a few higher harmonic periods.
observatories (recall Figure 5 in Section 5.02.2.2). This is a Annual and semiannual periodicities also have to be included
problem because both the ionospheric and the magneto- to account for seasonal effects.
spheric fields can evolve significantly in the mean time. This temporal behavior strongly contrasts with that of the
Yet another difficulty is that some toroidal currents do flow field of internal origin, which changes on much longer time-
in the F-region ionosphere, between the satellite’s orbit and the scales within the Earth’s reference frame. The core field mainly
changes on secular timescales, while the crustal field changes time. Annual observatory means were also added, for the entire
very little (see Hulot et al., 2009). They can simply be repre- 1960–1985 period covered by the models. The key is that,
sented using the same type of potential field spherical because of their a priori known temporal variability, the geom-
harmonic representations as in any of the previous models, etry of the external and induced fields could be constrained by
together with a B-spline temporal expansion (eqn [28]) for the observatory and satellite data when jointly available, while
lowest degrees. observatory data in the intermediate period provided the infor-
What about externally induced fields? We noted that those mation needed to model the slow variation of the field of
fields cannot be distinguished from the field of internal origin, internal origin.
both being produced by sources below the Earth’s surface, that
is, however, if we assume all fields to only be known at a given 5.02.3.5.3 Recent comprehensive models
instant. But the purpose of a comprehensive model is not to The third model of the series, CM3 (Sabaka et al., 2002), still
describe the field only at such a given instant, but throughout a covers the same 1960–1985 time period but is much more
certain amount of time. Advantage can then be taken of the very sophisticated and relies on more data. In particular, OHMs
different temporal behaviors of the field of internal origin and have been used for the quietest day of each month, as deter-
of the externally induced fields, which can be represented with mined by Kp, and at two sampling rates: the OHM values
similar temporal modes as the inducing fields of external closest to 01:00 LT, over the entire 1960–1985 span of the
origin. model (allowing for the determination of broad-scale secular
The only truly limiting issue that then remains is that of the variation), and OHMs every 2 h, during the POGO and Magsat
toroidal currents possibly flowing in the F-region between the mission envelopes (which is sufficient for analyzing external
satellite and the E-region ionosphere. This issue could partly be and induced contributions with periods down to a 6 h diurnal
dealt with using a selection procedure of the type proposed by period). More POGO data were also used, especially quiet-time
Stolle et al. (2006) (and already used by a number of core field passes.
models; recall earlier text). However, none of the published The scope of CM3 was much wider and its attention to
comprehensive models have yet implemented such proce- detail much higher than its predecessors. The internal spherical
dures, and toroidal currents in the F-region have thus so far harmonic expansion was extended to degree N ¼ 65 in order to
been ignored. Their contribution, together with contributions account for the crustal field as measured at satellite altitude,
from all other nonmodeled field sources and intrinsic mea- and crustal biases used only to account for smaller-scale crustal
surement errors, is dealt with as noise. field contributions at observatory locations. Temporal changes
of coefficients up to N ¼ 13 were still described by cubic
5.02.3.5.2 Early comprehensive models B-splines, but with some regularization to ensure a smooth
Sabaka and Baldwin (1993) and Langel et al. (1996) described secular variation. The ionospheric and magnetospheric fields
the first two versions (CM1 and CM2) of the comprehensive were better described using, respectively, QD and dipole
model (CM) series. In those early versions, not all fields coordinates and including four (24 h, 12 h, 8 h, and 6 h) diur-
were modeled. CM2, slightly more elaborate than CM1, only nal periods, with annual and semiannual modulations. Addi-
modeled the field of internal origin up to degree N ¼ 13, with tional solar activity influence was also taken into account
B-splines. The crustal field was not solved for, except for through an amplification factor for the E-region ionospheric
so-called crustal biases at observatories (i.e., for offsets, to field (based on a 3-monthly moving mean of absolute F10.7
account for the fact that taken together, all crustal sources solar radio flux values; Olsen, 1993). The sources of this
produce significant constant fields at each observatory loca- E-region ionospheric field were otherwise still considered
tion). In the same way, not all external fields were solved for. as a sheet-current density (at 110 km altitude). As another
F-Region currents were ignored altogether, and the classical improvement to CM2, F-region toroidal fields were also taken
trick, consisting in only retaining intensity satellite data at into account and modeled (assuming radial currents). Their
high magnetic latitudes to avoid field-aligned currents, was temporal variability (diurnal and annual) was however kept
used. Also, both the magnetospheric and the ionospheric simple to account for the fact that Magsat flew at relatively
E-region fields were modeled in simpler ways than later ver- fixed LTs (dawn and dusk; recall that only vector satellite data
sions. One interesting specificity of those early models we sense those fields).
should also mention is that, except for the component induced Another significant change with respect to CM2 is that
by the Dst variations, externally induced fields were modeled externally induced fields were no longer modeled indepen-
independently of the inducing external fields (i.e., without dently of their inducing external fields. Rather, it was recognized
making any assumption with respect to the electric structure that induced fields must be proportionate to the inducing fields
of the Earth). in some way governed by the electric properties of the solid
Perhaps the biggest merit of those early models is that they Earth. An a priori four-layer 1-D radially varying conductivity
actually established the practical validity of the comprehensive model derived independently from European observatory data
modeling approach. In particular, it was not obvious at the was thus introduced for that purpose (Olsen, 1998). This sim-
onset that such an approach would work, given the awkward plification allowed the externally induced fields to be formally
availability and properties of the various data sets. Both coupled to their inducing fields, thereby reducing the number of
models relied on essentially the same data, which consisted free parameters in the model.
in September 1965–August 1971 POGO scalar data, November This model was much improved. But it still had one major
1979–May 1980 Magsat scalar and vector data, and quiet-day drawback. It was only constrained by relatively few satellite
observatory hourly means (OHMs) during the same periods of vector data.
The most recent version of the model series, CM4 (Sabaka expected (recall that only quiet-time data were used to build
et al., 2004), precisely aimed at correcting for this. It is a natural the model), but also is able to make remarkable predictions of
extension of CM3, spanning the years 1960–2002 and taking the field at quite disturbed magnetic times.
advantage of additional CHAMP (scalar) and rsted (vector Let us first consider time series covering the entire time span
and scalar) data, along with additional OHM values closest to (1960–2002) the model is supposed to account for. Figure 9
01:00 LT on the quietest day of each month, up to 2000. shows observed monthly mean values of the X (north) com-
Additional Magsat data were also considered. This prompted ponent at the Niemegk (NGK) observatory in Germany (black
some slight modifications in the parameterization of the line) together with various predictions from the model. The red
F-region ionospheric toroidal fields, to account for the fact curve shows the sum of the predicted contributions from the
that Magsat and rsted did not sample the F-region at the field of internal origin (core field and crustal field). Adding the
same altitude and at the same LTs (contrary to Magsat, rsted predicted contribution from the magnetospheric field with its
covers all LTs, and not only dawn and dusk). This parameteri- induced counterpart leads to the green curve. Also adding the
zation was also improved to no longer restrict the correspond- predicted contribution from the ionospheric field with its
ing poloidal currents to only be radial (they must more induced counterpart leads to the blue curve. The final predic-
generally lie in a QD meridian). tion of the CM4 model would then require that we also add a
In total, CM4 consists of 25 243 model parameters, of constant value, corresponding to the local crustal bias, which is
which 8840 are used to model the core and lithospheric field, the local contribution from the small-scale crustal field not
5520 for describing the ionospheric field, 800 for the magne- described by the spherical harmonic gauss coefficients of the
tospheric field, 8848 for the F-region coupling currents, and model (which only goes to N ¼ 65). This crustal bias shows up
1635 for observatory biases. As a final improvement, these here as an offset and is on the order of 160 nT, a fairly typical
model parameters were estimated with the help of an itera- value. This offset is one parameter of the CM4 model, because
tively reweighted least-squares approach with Huber weights. data from Niemegk have been used to infer CM4. But note that
in general, like all global field models, CM4 can only predict
5.02.3.5.4 Predictions from CM4 the field to within such a constant value at the Earth’s surface
Perhaps the best way to show the success and limits of the CM4 (unless the local crustal bias is known, of course). Note also
model is to illustrate its ability to account for observations. This that by contrast, no such macroscopic offset will be found
can be done in a great variety of ways, and many such illustra- between the field predicted by CM4 and the observed field at
tions can be found in the original paper of Sabaka et al. (2004). satellite altitude, because biases are produced by the very small-
Here, only a few examples will be provided, mainly to show scale crustal field, the magnitude of which decreases fast with
how each type of fields contributes to the temporal variations of altitude.
the geomagnetic field observed at a fixed location, as testified by Figure 9 illustrates the ability of the CM4 model to predict
observatories. Interestingly, and as we shall see shortly, the most of the observed time variations on these timescales. Of
comprehensive model not only succeeds at explaining most of particular interest is the fact that both magnetospheric and
the geomagnetic field variations at quiet times, as would be ionospheric field contributions are important to achieve this
18900
nT
18800
18700
18600
18500
18400
Observed
Core + crust ...
... + Magnetosphere (primary and induced) ...
... + Ionosphere (primary and induced)
18300
1960 1965 1970 1975 1980 1985 1990 1995 2000
Year
Figure 9 Observed and predicted (from CM4) monthly mean values of the X component at Niemegk, Germany.
success. They account not only for the strong annual (and semi- by simultaneous ground and satellite data acquired during the
annual) variations but also for significant irregularities, which POGO and Magsat missions at other times.
have been captured by the Dst and F10.7 parameterization. Figures 9–11 are typical illustrations of the fit CM4 can
Figure 10 provides a complementary way of understanding achieve at middle- and low-latitude observatories (with a misfit
the contribution of each type of fields to the observed geomag- typically ranging from 6 to 8 nT). Note, however, that misfits at
netic field on the same timescale and over the same time polar observatories are somewhat larger (about twice as large).
period. It shows time series of monthly mean first differences The reader interested in such details is invited to consult
(smoothed with a 12-month running average) again not only Sabaka et al. (2004) and particularly its Table 3, which lists
at Niemegk (NGK) in Germany but also at Hermanus (HER) in all residual statistics.
South Africa. In each case, and for each field component (X, Y, Figure 12 shows maps of the equivalent sheet-current func-
Z; recall Section 5.02.2.1 for definitions), the observed signal tion c (which defines the surface currents Js ¼ ^r ∇C) of the
is shown (black dots), together with predictions from just the regular E-region ionospheric field sources at 110 km as inferred
core field (red line), from the core field plus the externally from CM4, on a typical quiet day (21 March 1980) when Magsat
induced fields (green line), or from all fields (blue line). data were available. Those currents are essentially fixed with
Note that this time, no more offset is to be seen. This illustrates respect to the Earth–Sun line (maxima are always located near
the fact that the crustal bias is indeed a constant value, which the central meridian line, below which the Earth rotates, when
has been removed by taking the time derivatives. Those curves the DUT changes), but their detailed morphology follows
otherwise illustrate the contribution of externally induced important features of the main field. Note in particular that
fields and show that the X, _ and especially Y,
_ horizontal com- the current is amplified at the dip equator (indicated by the
ponents are least affected by the external fields and their red curve), which represents the equatorial electrojet, and that
induced counterparts. This well-known feature is the reason the latitudes of the Sq foci in the northern and southern hemi-
why X_ and especially Y_ series have been extensively used in the sphere follow that of the dip equator. Such main field-related
past to investigate changes of trends in the field of internal features were not captured by previous models of the regular
origin. As first noted by Courtillot et al. (1978), those series daily variation (see, e.g., Campbell et al., 1989; Winch, 1981,
indeed reveal an intriguing piecewise pattern, with periods of and Chapter 5.03 of the present volume).
stable trends, interrupted by sudden changes in those trends Similar maps of the regular E-region ionospheric field
(as indicated by the arrows in Figure 10). Those sudden sources can be predicted from CM4 for April 1990 (which
changes are known as geomagnetic jerks (see also Chapter CM4 does by adjusting for seasonal effects and using the rele-
5.05 of the present volume and Chapter 8.04 of the present vant F10.7 index). The rotation of the Earth below those sources
Treatise). When and how exactly they occur has been the subject is then the cause of the diurnal periodicity of the quiet-time
of many recent studies, some of which rely on quite elaborate ionospheric field (and of its associated externally induced field,
techniques (see, e.g., Alexandrescu et al., 1996, 1999; De which arises precisely because of this periodic excitation)
Michelis and Tozzi, 2005; Mandea et al., 2010 for a recent observed in observatories. Similar periodic signals are also pro-
review and more references). Most of those studies have how- duced by the essentially Sun-synchronous quiet-time
ever relied on direct analysis of the observatory series, which do magnetospheric field and its induced counterpart, which CM4
not guarantee that the suddenness and timing inferred for those can also predict. All those fields add up to make for the periodic
jerks are not partly contaminated by the field of external origin component of the predictions in Figure 11. The last nonperiodic
and its induced counterpart. CM4 very clearly illustrates this contribution arises from the highly variable ring-current signal
issue. With induced and external contamination removed, the (and its induced counterpart), the variability of which CM4
core field contribution from CM4 indeed reveals the well- predicts using the Dst index. As can be seen, the resulting pre-
known jerks in the Y_ components at NGK and HER. But local diction is quite successful. Note in particular that CM4 still
extrema due to the contribution of other sources exist in the data manages to capture most of the field variations at NGK and
record near the times of each event. Those could lead to incor- HER, even when a storm starts on the 10th day and the magnetic
rectly time jerks by perhaps as much as 1 or 2 years. By better conditions become severe (as testified by the Dst index, which
separating the core field from other fields, comprehensive goes way above the limit of 20 nT used to define quiet times in
models such as CM4 can provide an interesting alternative way the modeling procedure). It should however be mentioned that
of investigating geomagnetic jerks (see, e.g., Chambodut and less success is achieved at such magnetically disturbed times,
Mandea, 2005 and Sabaka et al., 2004). when predicting field variations at higher latitudes, or closer to
What about short-term geomagnetic field variations? the equator (see Sabaka et al., 2004 or Chapter 5.05 of the
Figure 11 shows 1-month (April 1990) series of observed hourly present volume, where CM4 is also briefly discussed).
mean values of X, Y, Z components at the same NGK and HER As the reader will have noticed, CM4, which was published
observatories as before, together with CM4 predictions. This in 2004, is only valid up to 2002. But an update is currently
figure is particularly interesting because only a single point of under construction, to take advantage of the data collected since
each series was actually used to build CM4. April 1990 was 2002 thanks to rsted and CHAMP, in particular. This update
indeed a month no satellite data were available, and as stated will also take advantage of a number of developments made in
earlier, only the OHM value closest to 01:00 LT for the quietest the context of extensive tests conducted during the preparation
day of the month was then used in CM4 (to assess the contri- of the Swarm mission, the results of which can be found in
bution of the core and crustal fields). The fast variations pre- Sabaka and Olsen (2006) and Olsen et al. (2006a). The
dicted here are thus true predictions of the CM4 model, CM4 comprehensive modeling approach is indeed one that is
having its external field parameters independently constrained planned to be used to produce models from the Swarm mission
NGK HER
40
0
20 -20
Core
0 -40
-60
-20 -80
-40 -100
dx/dt (nT year-1)
40
0
Core + Ind
20 -20
0 -40
-60
-20 -80
-40 -100
Core + Ind + Ext

40
0
20 -20
0 -40
-60
-20 -80
-40 -100
1960 1970 1980 1990 2000 1960 1970 1980 1990 2000
60
40 40
Core
20
20 0
-20
0
dy/dt (nT year-1)
60
Core + Ind
40 40
20
20 0
-20
0
Core + Ind + Ext

60
40 40
20
20 0
-20
0
1960 1970 1980 1990 2000 1960 1970 1980 1990 2000
40 100
Core
20 80
0 60
dz/dt (nT year-1)
Core + Ind
40 100
20 80
0 60
Core + Ind + Ext
40 100
20 80
0 60
1960 1970 1980 1990 2000 1960 1970 1980 1990 2000
Year Year
Figure 10 Comparison of smoothed monthly mean first differences, X, _ Y,
_ and Z,
_ from observations (black symbols) and from predictions using a
progression of sources from CM4 (colored lines) at NGK and HER from 1960 to 2002. Red, green, and blue lines present core, core plus induced,
and core plus induced plus external sources, respectively. Arrows indicate approximate jerk locations from visual inspection of the core contribution to
_ Four jerks previously identified, in 1969, 1979, 1992, and 2000, are shown. The presence of an additional event in 1997 is also suggested.
Y.
Reproduced from Figure 8 of Sabaka TJ, Olsen N, and Purucker M (2004) Extending comprehensive models of the Earth’s magnetic field with rsted
and CHAMP data. Geophysical Journal International 159: 521–547, http://dx.doi.org/10.1111/j.1365-246X.2004.02421.x.
NGK HER
18850 10100
18750 10000
X (nT )
18650 9900
18550 9800
0 5 10 15 20 25 30 0 5 10 15 20 25 30
200 −4250
150 −4300
Y (nT )
100 −4350
50 −4400
0 −4450
0 5 10 15 20 25 30 0 5 10 15 20 25 30
45050 −24800
44950 −24900
Z (nT )
44850 −25000
44750 −25100
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Days after April 1, 1990 Days after April 1, 1990
Figure 11 Observed (red dots) and CM4 predicted (black curves) field values at the observatories Niemegk (left) and Hermanus (right), for April 1990.
00 DUT 06 DUT
NGK NGK
HER HER
12 DUT 18 DUT
NGK
NGK
HER HER
Figure 12 Global maps of ionospheric E-region equivalent current function c for different universal times (DUT), td (cf. eqn [15]) during March 21,
1980 and a value of F10.7 ¼ 140.0 1022 W m2 Hz1, which is an average over the time span of the model. Contour interval is 20 kA; positive values are
shown with solid, negative with dashed lines. The red curve represents the dip equator. Locations of the Niemegk (NGK) and Hermanus (HER)
observatories are also shown.
(Sabaka et al., 2013), in the context of the Swarm Satellite 5.02.4 The Present Main Field
Constellation Application and Research Facility (SCARF)
(Olsen et al., 2013), which will also be in charge of producing As the previous section amply illustrated, recent satellite mis-
alternative dedicated models of the core field (Rother et al., sions boosted our ability to model and identify all components
2013), crustal field (Thébault et al., 2013), external fields of the Earth’s magnetic field. Many breakthroughs have been
(Alken et al., 2013; Chulliat et al., 2013, Hamilton, 2013; made, which we briefly described or alluded to: considerable
Ritter et al., 2013), and other so-called level 2 products. improvement of our knowledge of the crustal field, better
understanding of the ionospheric and magnetospheric fields dominated by the crustal field for degrees above 14 and by
and of their associated induced fields, and detection of previ- the core field for degrees below 14, both fields probably con-
ously undetected or poorly known field components, such as tributing by similar amounts at degree 14 (recall Section
the fields produced by oceanic currents at satellite altitude or 5.02.3.3). As a result, the main field and therefore the core
those locally produced in the F-region of the ionosphere. field, at any given epoch such as here in 2000 and 2010, can
Those breakthroughs are not only important for the informa- only be recovered with some precision up to degree 13.
tion they provide. They also contributed to our improved Figure 13(a) also illustrates the overall excellent agreement
ability to investigate the main field and, most importantly, its between various models of the field of internal origin, espe-
temporal variation, the so-called secular variation. It is the cially as far as the core field is concerned. Most disagreements
purpose of the present section to report on the recent progress are indeed to be found in the crustal field branch. Those likely
and to provide an overview and brief discussion of the most trace back to the different ways models deal with external field
important spatiotemporal characteristics of the present main contributions, CHAOS-4a relying on a selection process (recall
field and of the core field it testifies for. We will focus on the Section 5.02.3.2) and CM4 modeling most of it (recall Section
1980 to present time period but will also occasionally refer to 5.02.3.5). It is unavoidable that at least a fraction of the
the known behavior of the main field over historical and arch- external and unmodeled fields will be misinterpreted as pos-
aeological time periods, which are otherwise much more sible crustal field contributions in both models, but in differ-
extensively addressed in Chapters 5.05 and 5.09 of the present ent ways, resulting in such disagreements. Taking extreme
volume (see also Volume 8). steps to get rid of those may also have some drawbacks, as
was discussed in Section 5.02.3.4 and is illustrated by the
comparison of the spectrum of the crustal field model MF7
5.02.4.1 Spatial Power Spectra and Timescales
and that of CHAOS-4a. This spectrum (also shown in
A useful way of characterizing the present global spatio- Figure 13(a)) shows that the estimated crustal field becomes
temporal behavior of the Earth’s main field, beyond the well- weaker beyond degree 70. By emphasizing the need to
known fact that it is mainly dipolar and evolving on secular remove the external field, the modeling approach used for
timescales, is to make use of the concept of spatial power MF7, which relied on high-pass along-track filtering, may
spectra introduced in Section 5.02.3.1. have ‘erased’ part of the crustal field signal.
Applying this concept to the field of internal origin, as Plotting the analogous Lowes–Mauersberger spectrum for
predicted by, for example, CM4 for epoch 2000 and the the first time derivative of the field of internal origin at the
CHAOS-4a for epoch 2010, and plotting the Lowes–Mauers- Earth’s surface provides a very interesting complementary pic-
berger spectrum at the Earth’s surface (eqn [26] for R ¼ a), leads ture. As discussed earlier, we know that this time derivative, the
to Figure 13(a). This figure (plotted in semi-log scale, beware) secular variation, testifies for the first time derivative of the core
shows that the field of internal origin is indeed dominated by field (at least up to degree N ¼ 22; Hulot et al., 2009).
its dipole (n ¼ 1) component. But it also shows that the spec- Figure 13(b), which shows such spectra predicted by several
trum is made of two almost linear branches, with a transition models (a number of which have been discussed in Section
around degree 14. This dichotomy (first noted by Langel and 5.02.3), is also an opportunity to illustrate the spectacular
Estes (1982) based on a 1980 Magsat-derived spectrum) is the improvements the recent satellite missions have brought.
signature of the fact that the field of internal origin is Whereas Langel and Estes (1985b) could only recover the
CM4 Langel and Estes (1985)

CHAOS−4 Langel et al. (1987)
MF7 103 Olsen (2002)
108 Difference CHAOS−4 − MF7 Maus et al. (2005)
CHAOS−4
Mean SV (Hulot et al., 2002)
Mean SV (OSVM−Magsat)
102
[(nT/year)2] or [(nT/year2)2]
106
101
(nT2)
First time derivative

104
100
102 Second time derivative

10-1
100 10-2
0 10 20 30 40 50 60 70 80 90 0 2 4 6 8 10 12 14 16 18
(a) Degree n (b) Degree n
Figure 13 a: Spatial power spectrum of the field of internal origin at Earth’s surface, as described by models CM4 for epoch 2000 and CHAOS-4a for
epoch 2010; also shown is the spectrum of the MF7 crustal (lithospheric) field model. b: Spatial power spectrum of secular variation and
secular acceleration at Earth’s surface for various models and epochs (see text for details).
1980 secular variation up to degree 6 or so, using data from 6 and the OSVM model of Olsen (2002) in 2000). This shows
months of Magsat satellite data and 91 observatories (the ‘flat’ that the recent improvements have brought the spectra of the
behavior above degree 6 being indicative of a noise of about instantaneous secular variation closer and closer to the average
70 nT2 year2), a resolution Langel et al. (1988) could only secular variation spectrum, further suggesting that those
slightly increase by the additional use of data from the DE-2 improvements are real. In fact, Figure 13(b) would even suggest
satellite between September 1981 and January 1983 (reducing that the instantaneous secular variation model is now better
the noise level to about 30 nT2 year2), the advent of rsted determined than the average secular variation from Hulot et al.
and CHAMP dramatically improved the situation. With just 2 (2002), the behavior of which suggests a noise contamination
years of rsted observations, the noise level was reduced to on the order of 0.06 nT2 year2, affecting degrees above 15.
about 1 nT2 year2, making it possible to resolve the 2000 Note however that a perfect match between those spectra should
secular variation up to degree 11 or so (Olsen, 2002). Yet not be expected since we also know that the secular variation is
another order of magnitude in noise reduction was achieved not stationary (as is clear from the fact that all models now call
by the model of Maus et al. (2005) (derived by a combined for a secular acceleration over the past years and even clearer
analysis of rsted and CHAMP data spanning more than 5 from the direct observation of geomagnetic jerks at
years), resulting in a determination of the secular variation up observatories – recall Figure 10 in Section 5.02.3.5).
to about degree 13. Finally, a noise level lower than 0.01 nT2 Plotting the analogous spectra for the core field, its secular
year2 could be achieved by the CHAOS-4a model, allowing a variation, and the secular acceleration at the core–mantle
determination of the 2010 secular variation possibly up to boundary (CMB) is also important, since this is where the
degree 16. In fact, and as was already pointed out in Section core field originates from.
5.02.3.3, rsted and CHAMP observations even made it possi- Figure 14(a) shows the core field spectrum at the CMB for
ble to simultaneously infer the second time derivative of the epoch 2005.0 computed from the CHAOS-4a model (all other
field of internal origin, the secular acceleration, the spectrum of models would predict virtually the same spectrum). The model
which is also shown in Figure 13(b) for the model of Maus et al. was truncated at degree 13 to avoid the crustal field (in a slightly
(2005) (average acceleration over the 2000–2005 time period) rough way; recall the discussion in Section 5.02.3.3) and eqn
and the CHAOS-4a model (instantaneous acceleration for the [26] was used with R ¼ b, where b is the core radius, which
2005.0 epoch, though smeared in time due to the temporal amounts to down-continue the core field to the CMB as a poten-
regularization used in the model). tial field (which we may do, since the core sources lie below the
All those spectra typically reflect average estimates of the time CMB and the mantle can be assumed to be a source-free
derivatives over only a few years, that is, almost instantaneous medium; recall Section 5.02.3.3). This spectrum reveals that
estimates of the present time secular variation and acceleration. the core field is still dominated by its dipole component at the
But Magsat flew long enough to recover high-degree models of CMB. But it now also shows that higher-degree components
the field of internal origin in 1980, which can be compared to (and therefore smaller scales) play a much more prominent
present time high-degree models derived from rsted and role at the CMB than at the Earth’s surface. Their contributions
CHAMP data, to produce average secular variation models. The only very weakly decrease on average as the degree increases.
spectra of two such models are also plotted in Figure 13(b) (as The important role played by small scales at the CMB shows
inferred by Hulot et al. (2002) from Magsat (1980) and rsted up in an even more striking way when plotting the secular
(2000) models computed by Langlais et al. (2003) and as variation spectrum at the CMB. Figure 14(b) shows three
inferred here from the model of Cain et al. (1989) in 1980 such spectra: the instantaneous 2010 secular variation spectrum
1011 109
CHAOS−4 in 2010 at CMB CHAOS−4 SV in 2010 at CMB
CHAOS−4 SA in 2005 at CMB
Mean SV (OSVM −CM4 )
2000 1980
108 Mean SV (Hulot et al., 2002)
[(nT/year)2] or [(nT/year2)2]
107
Static field
First time derivative
106
(nT2)
1010
105
104
Second time derivative

103
109
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
(a) Degree n (b) Degree n
Figure 14 a: Spatial power spectrum of the core field at the core surface as described by model CHAOS-4a for epoch 2010. b: Spatial power spectrum
of secular variation and secular acceleration at the core surface for various models and epochs (see text for details).
as inferred from the CHAOS-4a model and the two average A final very useful spectral quantity one can next easily
1980–2000 secular variation spectra already shown at the derive is the correlation time associated with a given degree n
Earth’s surface in Figure 13(b). All spectra have been boldly and is defined as
plotted up to degree 18, which we are theoretically entitled to vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiu Xn h m 2 m 2 i
do, since the secular variation is dominated by the first time u gn þ h n
Wn ðr Þ u m¼0
tðnÞ ¼ ¼uX
m 2 m 2
derivative of the core field. In doing so, we also make the [29]
Wn ðr Þ t n
0
assumption that all degrees up to 18 are properly recovered. m¼0
g_ n þ h_ n
This however is not so obvious as the divergence between the
two average secular variation spectra above degree 15 shows. where Wn(r) and W0 n(r) are the degree n contributions to the
As already noted, this divergence likely reflects noise contam- Lowes–Mauersberger spectra of, respectively, the core field and
ination in the Hulot et al. (2002) average secular variation. But its secular variation at radius r (as defined by eqn [26]). This
similar (weaker) contamination could also affect the other quantity, which is independent of r, is best understood in
spectra, as the steeper trend beyond degree 15 would suggest. statistical terms as a measure of the time it would take for the
Those spectra show that contrary to the core field itself, the degree n core field to significantly evolve, were the core field to
secular variation is generally not dominated by its largest scales behave as if produced by a random statistical process (Hulot
at the CMB. In contrast, a clear increasing trend is now indeed and Le Mouël, 1994). Although the core field is of course
to be seen, revealing a very strong small-scale dynamics, which produced by a deterministic dynamo process within the core,
could not be witnessed in such detail before the advent of the it turns out that such a statistical interpretation is quite consis-
rsted and CHAMP missions. tent with both the known temporal evolution of the large
Plotting the secular acceleration spectrum at the CMB con- scales of the core field recovered from historical and arch-
firms the important dynamic role played by small scales. aeomagnetic data (see Hongre et al., 1998; Hulot and Le
Figure 14(b) shows such a spectrum, as computed from Mouël, 1994) and the behavior of the field produced by
CHAOS-4a for epoch 2005.0. Just like the secular variation, numerical dynamo simulations (Bouligand et al., 2005).
the secular acceleration displays a clear increasing trend up to Figure 15 shows such estimates as inferred from the
degree 11. Beyond that degree, however, it sharply drops. This CHAOS-4a model by simply making use of the two CHAOS-
consequence of the temporal regularization introduced in the 4a core field and secular variation spectra at epoch 2010.0
CHAOS modeling process (recall Section 5.02.3.2) testifies for shown in Figure 13(a) and 13(b). Of course, any other choice
the fact that the secular acceleration is hardly constrained by the of models for the same epoch would essentially lead to the
observations beyond this degree. As a matter of fact, the exact same estimates. Of particular interest is the decreasing trend
amount of secular acceleration that can be recovered from recent showing that the correlation time gets all the smaller that the
satellite data is still a matter of much debate and another strong degree considered increases, reaching values on the order of
motivation for the Swarm mission. Most authors would agree 30 years for the largest degrees such correlation times can be
that current data are unlikely to provide any significant con- estimated. Small scales of the core field can change very fast.
straint beyond degree 8 and perhaps not even beyond degree 6 Several interpretations of the core field and secular
(see, e.g., Finlay et al., 2012; Lesur et al., 2011). variation spectra, as well as of the resulting correlation times,
103 CHAOS−4
t = 425 / n years
t (years)
102
101
0 2 4 6 8 10 12 14 16
Spherical harmonic degree n
Figure 15 Correlation time t (n) (eqn [29]), as a function of n, estimated from the CHAOS-4a model for epoch 2010, together with the t(n) ¼ tSV/n law
(with tSV ¼ 425 years, based on the CHAOS-3 predecessor model) proposed by Lhuillier et al. (2011b) for the nondipole field (n>1).
have been proposed (see, e.g., Holme and Olsen, 2006; magnetic conditions (watching out not only for strongly
McLeod, 1996; Voorhies, 2004; and references therein). All magnetized local rocks or man-made structures but also, of
those interpretations, however, predict fairly similar spectral course, for any possible ongoing geomagnetic storm – recall
behaviors, and it is difficult to tell which the observed spectra Section 5.02.1).
would favor. The issue is further complicated by the fact that Figures 16 and 17 show maps of the declination (D) for
the observed core field and secular variation spectra can be epochs 2010 and 1980, while Figures 18 and 19 show maps of
expected to fluctuate in time. the inclination (I) for the same two epochs. Those maps,
The statistical formalism introduced by Hulot and Le Mouël which are all plotted using the same Hammer and polar
(1994), however, provides a means of quantifying the fluctua- (north and south) projections, illustrate the current morphol-
tions one may expect. By formally linking the core field and ogy of the main field at the Earth’s surface. A pure axial dipole
secular variation spectra at a given epoch to the observed core field would lead to zero declination everywhere and to axisym-
field evolution, it makes it possible to assess whether an metric isoclines (lines of equal inclination), with the dip equator
observed property can qualify as long-standing. Using this (defined by I ¼ 0 ) lying on the geographic equator and the
approach, Hongre et al. (1998), for instance, showed that the north and south dip poles (also known as the north and south
distinct ‘V’ shape made by the first three degrees in the core magnetic poles and defined as the locations where I ¼ 90 and
field spectrum (Figure 14(a)), which has been around for at I ¼ 90 ) located at the north and south geographic poles. But
least the past 2000 years, is a robust long-standing feature. This the Earth’s main field is not an axial dipole, and this is the
is consistent with what is seen in all dipole-dominated numer- reason for both the distortion of the isoclines and the complex
ical dynamo simulations, which display such ‘V’ shape spectra pattern of the isogones (lines of equal declination). Much of this
(see, e.g., Christensen and Olson, 2003; Dormy et al., 2000; complexity is due to the tilt of the dipole component of the
and Volume 8 of this treatise). Similarly, Lhuillier et al. main field, but not all. This is best illustrated by the fact that
(2011b) recently showed that the observed correlation times the dip poles are not antipodal (which they would be, were the
t(n) for the nondipole field could be fitted by a simple tSV/n main field to be that of an inclined dipole). Indeed, using
law (see Figure 15) to within the fluctuations expected from CHAOS-4a and CM4, one can easily estimate the (geocentric)
the Hulot and Le Mouël (1994) formalism. Similar analysis of locations of those dip poles: 84.83 N, 131.71 W versus 64.08 S,
data produced from numerical dynamo simulations led to 137.18 E for the north and south dip poles in 2010 and
exactly the same conclusion, in an even more robust way 76.45 N, 101.2 W versus 65.05 S, 139.15 E for the north and
(given the much larger amount of data that can then be south dip poles in 1980. Those locations are also interesting to
analyzed), confirming the surprisingly general validity of this compare with the (geocentric) locations of the strictly antipodal
tSV/n law. This law provides a very robust way of scaling geomagnetic poles (derived from eqns [5] and [6]; recall Section
secular variation timescales in dynamo simulations (see 5.02.2.1): 80.02 N, 72.21 W versus 80.02 S, 107.79 E for the
Section 5.02.5 later and Chapter 5.05 of the present volume north and south geomagnetic poles inferred from CHAOS-4a in
for more discussion). 2010 (marginally different from the location inferred from IGRF
2010) and 78.8 N, 70.8 W versus 78.8 S, 109.2 E for the north
and south geomagnetic poles inferred from CM4 in 1980. The
5.02.4.2 The Field at the Earth’s Surface
very significant departure of the dip pole locations from those of
Let us now consider the present main (core) field at the Earth’s the matching geomagnetic poles is a direct consequence of the
surface. To also illustrate the way this field has been changing nondipolar components of the main field.
since it was first observed in detail by a high-precision satellite Figures 16–19 also illustrate the impact of secular variation.
(Magsat), we in fact focus on two epochs, 2010.0 and 1980.0. Although those maps look similar in 1980 and 2010, obvious
For epoch 2010.0, we rely on the CHAOS-4a model, which is changes can be found. Those are not trivial. In places such as in
one of the most recent models capable of describing that epoch Europe, Africa, and over the Atlantic Ocean, those crudely
without having to resort to some temporal extrapolation (con- amount to a westward drift of the main field (as is best seen
trary to the official IGRF 2010 of Finlay et al. (2012); recall by focusing on, e.g., the agonic D ¼ 0 line), a drift that is also
Section 5.02.3.2). For epoch 1980.0, we otherwise choose the seen on historical timescales (see Chapter 5.05 of the present
CM4 model of Sabaka et al. (2004). In all cases, only degrees volume). But this westward drift is not universal. Isogones in
up to N ¼ 13 have been used, since as we saw, higher degrees the Northern Pacific East of the United States clearly moved
correspond to crustal field contributions. In fact, it is worth eastward. Also, whereas the south dip pole hardly moved (as is
noting that had we used the full resolution of each model (up also confirmed by a series of direct local observations in 1986
to degree N ¼ 100 for CHAOS-4a, and N ¼ 65 for CM4) and and 2000; Barton, 2002), the north dip pole, which only
plotted the predicted field of internal origin, the resulting maps moved at < 10 km year1 over the past two centuries, started
would have anyway looked very much the same, since crustal accelerating very suddenly towards the north during the 1990s
field contributions up to such degrees only amount to a few and is currently moving at the impressive pace of 50 km year1
tens of nanotesla (i.e., to < 1% of the main field contribution). (as is again confirmed by a series of direct local observations, in
Note nevertheless that nonmodeled smaller scales can contrib- 1984, 1994, 2001, and 2007; Chulliat et al., 2010a; Newitt
ute much more (as illustrated by the crustal bias responsible for et al., 2009). All those changes reflect changes in the core field
the shift between the observed and predicted field in Figure 9 produced at the CMB, the morphology and evolution of which
of Section 5.02.3.5). The reader should therefore be warned will be illustrated in the next section.
not to make use of any of those maps to infer local values Maps of the field intensity at the Earth’s surface reveal
of magnetic field components without checking the local interesting additional features. Figure 20 shows such maps
00 00
20
0–7
–4
––11630
00 0 –10 0
–1 0
10 20 302–002–03–5 20 3
0 –10
100 –40 10 0
10 1 –10
–
0
20 20 –30 10 20 0
0 –1 0
0 0
10 –2 –10
10
–2
10
0
0
–1
–10
0
10
0
10
–10
10
0
–1
0
0
0
–2 0
0 –1
10 0
10
–10
–20
–10
–2
0
–1
–10 –20
0 10
0
10
–20
0
–30
–30
–2
–1 –20 –40
0
20 0 10
20 0
–50 –50
–4
10
0
0
–20
20
–1
20
–30
–1 –60 00
0
0
–5 –40
30 30 0 10
–3
0 –6 –70 –70 2
1 –80 20 30
0
40 2 0 –90 0 –
40 50 4300 0 0 9 0 – 1 00200 00101 60
1
3
–0
11
2
40
2
3
–
6
5
4
1
7
8
9
2 0050
01–1
8 1 3 4 5 4
–0 10
60
70
0
6500 0 –7– – –10–71 –1 –1––1–01
–071–20710
–10105–05
–16
63 104
70
80
90
100 –60020034 0
0–––1
5 1060––000
72
4
10
0 2
19
4
3
5
6
0–7
–3
7
8
100
1
2
4
6
3 0100
–1
51 00–7901–08–0
–5
110
120 ––1
1–8
–90 1–7
70 0–1
1––2 01–0
–1 1
6
–6
–1
–3
16 15 14013
0–1
–1
015 41 –20 –10
10 50
–1 –20
0
0 –30–20 –
–20 1 0
10
–10
–30
–4
–10
–1
0
10
20 –50
0
–4
0
0
0
–30
–20
0 –60
–10
–50
10 10
–10
0
0
20 30 –10 20 10 –5
–70 –80
10 –2–10 30 0
3000 20 10 0
–60
20
–40
0 ––50 340006–1
0–1
30 –1000
11– –1–1230
0
40 50
––4197
–10
00
10
–4
–90
–70
140
910
00–0
–80
–230
– 0
0
–1–75010
10
80
40
0
18–
116–1––1761–15004 50
–
–2
5060
0
20
–3
0
70 60
154000130–1 –0020 10 0 – –1
–10
–9
0
0
1120
110 30030––6 0102100–4
10 00
9876 03
00 0
4500
20
0
50
20
–20 10 0
30
40
10
0 30
–1 20
0
20
10
10
10
Figure 16 Declination D (contour lines every two degrees) at Earth’s surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 13.
for epoch 2010. Those maps again differ markedly from those geomagnetic eccentric dipole, see, e.g., Langel, 1987; Lowes,
of an axial dipole field (which would imply axisymmetric lines 1994). But it provides a convenient way of understanding the
of equal intensity, or isodynamic lines, with one maximum at consequence of the SAA for the near-Earth magnetospheric
each geographic pole and a minimum isodynamic line at the environment within which low-Earth orbit (LEO) satellites
geographic equator). In particular, two remarkable features (and astronauts) evolve.
show up that a dipole, even inclined, cannot account for. This near-Earth magnetospheric environment is a region
One is the contrast between the relatively weak (about above the ionosphere, where trapped particles are to be
60 000 nT) almost double northern maximum and its strong found, which oscillate back and forth along the main field
(66 670 nT) southern counterpart. The other is the very low- lines, slowly drifting westward if positively charged (mainly
intensity (as low as 22 600 nT) region extending over South protons), eastward otherwise (electrons) (see Chapter 5.03 of
America and the South Atlantic, known as the South Atlantic the present volume). Those particles can have very different
Anomaly (SAA). This SAA is in fact mainly (but not only) due to energies. The most numerous but less energetic make for the
the additional contribution of the quadrupole (degree 2) com- magnetospheric ring current the reader must now be familiar
ponent of the main field. Adding this contribution to the with, which is most intense at several Earth radii close to the
dipole field results in a field, which can be approximated by a equatorial plane. But much more energetic particles (protons
so-called geomagnetic eccentric dipole, that is, a dipole with and electrons typically above 1 MeV) are also to be found,
the same magnitude as the dipole component of the main which contribute little to the electric currents (and to the
field, but mathematically located slightly away from the center magnetic field) but have strong penetrating properties. Those
of the Earth (presently about 563 km, in the direction opposite make for what is then referred to as the radiation belts or Van
to the location of the SAA). As far as the main field itself is Allen belts (see, e.g., Kivelson and Russell, 1995 for details).
concerned, this location does not have any special meaning Electrons, which are most dangerous (because most
(for a definition and a discussion of the concept of penetrating), are in fact to be found in two belts: an outer belt,
032
–6
5–00 0 0
10–7
1–0–10053170060 100
10
1
03
–141512 1203
7–0 800
–96
1
4 2030 –10 0
20 3040 – 4–
–0–
6
8
41–1–00
9
10 503210070 –10 –10
0– –60 0
20
–50 0 0
20 10
0
10 30 –4 0
–3
0
20 10 –10
–3 0
10
–2
–1
–2 0
0
0
0
0
–1
0
–1
20
10 10
0
–20
10 0
0
0
–10
–10
0
10 0
10
–2 0
10
0
–1
0
–10
10 10
0
–2 –10
0
–10
10
–20
–1
10 –20
0
–20 –30
–30
–2
0
0
–1
20 –40
10
0
10 –20 0
20 –4
0
–50
–30
–3
20 20
0
0
–1
–1 –6
30 0 –60 0 100
30 0 0
–5 –70 –2 0 2
0
1 –80
–4
30
2 0
–30 –20
–7
40
0
–90
300
–4
0
50 0
0 –7 0 0 102 00 10200000
4 0 –6 –8–9 ––1––11–10 ––10–3
060–0
10 14–130
71–01–650 2–101
0
161 15015403
–1 4
8
7
2
5
6
9
60 6500 –42
11 1–51
0–10
–7
0
70 70 6 07
80
90
100
110
120 5––01–1–10–1
0
2013045
00
600
70–
–1
70
–1–1
60
–3
50 60
20
–
30
1
40 1
5
–
0 11
900
4
7
1403
5
8
1
6
2 5
2–00709–00780–4
60
3
0
4
–10
130 –90
–80–4
–7 0 –6 –1
–1 –1
–2
0 0
–1
170–1
–10 1151
–1
0
0 –20
10
10
–20 –3
0
20 0
–10 –10
–1 –2030
–40
0
–1
–3
0
0
– 40
30 0
0
0
20
–
10
0
–10
10 –5
4
500 –10
0
10201102000 1 0
–50
10
–4
20
–6
607 0 0 –70
–098320 20 –6
0
10 320 900 30 0 –80
–10
00–40
–1
40–8
0–7
–10
–70
1
30
–11
0 –6 00 10 50 ––1
––1100
1
3–090–0–
0 0
–4 –60
–20
00
0 –2 –11230
0
–50
0
–1
910
–90
–3
400
40
–528–00
–3
80
0
––11–0–1107
20
0
07
0
–870
1–7–1–1
5600700
70 60
–4 1106164–5143016200
0–
50
–9
0 543
11120 000010
0 1–30 – 0
10
50 110
190 87060000 0 –2 –10
20
3
200
4500
30
–30 10
0
0
–20
0 40
–2
30
0
20
0
–1
10
20
10
0
Figure 17 Declination D (contour lines every two degrees) at Earth’s surface in 1980.0 as given by the CM4 model up to degree N ¼ 13.
which corresponds to electrons oscillating along main field damaged or degraded in performance to some extent (this
lines between typically 60 latitude north and south, reaching has even been the cause of some problems with the electronic
4–6 Earth radii at the equator, and an inner belt, with electrons devices on rsted).
traveling along low-latitude (up to only 30 ) field lines about Looking at the way this SAA, and more generally the inten-
the equator. Those field lines are near horizontal and force this sity of the main field, has been evolving in the recent years is of
axisymmetric inner belt to lie very close to the Earth. For a course also of substantial interest. Figure 21 shows maps of the
purely dipolar field of the present strength, its lower boundary changes experienced by the main field intensity between 1980
would typically be only one thousand kilometers away from (from CM4) and 2010 (from CHAOS-4a). As can be seen, the
the Earth’s surface. But because the main field is better biggest changes occurred over Northern America, where the
described by an eccentric dipole than by a simple dipole, and field intensity decreased by about 3600 nT, and over the Indian
because the radiation belts follow the geometry imposed by the Ocean, where the field increased by about 2000 nT. Clearly,
main field, this lower boundary is brought even closer to the also the average change corresponds to a global decrease, most
Earth’s surface in the direction opposite to the shift, that is, of which is related to the average decrease (by almost 500 nT)
above the SAA where the main field intensity is the weakest. As of the axial dipole component of the main field over that
a result, the inner radiation belt is presently only a few hundred period of time. This decrease, however, is not the dominant
kilometers away from the Earth’s surface above the SAA. This signal in those maps, which reveal quite some structure. Of
has many unpleasant consequences for the more and more particular interest is the fact that relatively little changes
numerous LEO satellites, the instruments and electronics of occurred in the Pacific and that most of the decrease occurred
which can severely suffer from radiation. Indeed, as reported in the vicinity of the SAA, which has been slightly growing in
by, for example, Heirtzler et al. (2002), nearly all spacecrafts size (a measure of this growth can also be given in terms of the
crossing this area at altitudes of 100–1000 km have been shift of the equivalent eccentric dipole with respect to the
80
80 80
80
70
70 70
70 70
60 70 60 60 60
60 60
50
50 50 50
50 60 50 40
40 40 40
40 50 40 30
30 40 30 30 30
50 30 30 20
20 20 20 20
30 40 10 20 10
20 40 10 0 10 10 10
30 0 0 0 0
20 30 –10 –10 0
10 20 –20 –10 –10 –10 –10
0 10 20 10 –20 –30 –20 –20 –20
0 –30 –20
–10 0 10 0 –30 –40 –40 –30 –30
–1 –30
–10 0 0 –5 0 –50 –40 –40
–20 0 –4 –40
–20 –10 –2 –60 –50 –50
–30 0 0 –50
–30 –20 –3 –6 –60
0
–40
–5
–40 –30 –60 –60
0
–50 –40
–4
–50 0 –70
–7
–60 –50 –6
–60 0 –60
–50 –80
–70 –60 0
–70 –60 –7
–60 0
–8
–70 –80 –3–05
10
020
30 10 0
–80 –70 –70 –60 0 –40
40 20 –1 0 0
30 0 ––340 –6 –3
2030 40 –05
0
50
–5
50
40 10 0
–6
10
60
–2
0
0
0 2
–30 –2
0–4
60
50
60
0 30
0
70
–1
0
50
0
04
–30–560
–40
–60
230
–30 –20 0 –1
70 70
–2100
–50
–
40
80
0
–20–40
60
80
0 20–1
–70
10
50
60
–70
0
–6
40
50
80
320
0 10 30
20
0
–40
–10
70
70
0
20 0–3
–20
–50
80 –80
0 – ––650
–80
40
0
60 50 0 00
–70
80
–70
10
–20
0
–60
–8
60
0 –4
70
10 0
0– 2
4 32
50
70
–3100
–2 30
–80
–
–3
–4
60
0 50
–2 – 0
0
–5
–6
40
–70
10
50 –70
0
40 60 –2 –6 0
0
50 0 0
23100 20 –4
0 40 30 –10 –1–030 –40 –5 –60 0 30
0 1 0 –20 0 0 –5 –
2
–10 –2 –1 0 0 –30 –40 –20
–20 –10
0 0
Figure 18 Inclination I (contour lines every two degrees) at Earth’s surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 13. Also
shown, the location of the magnetic (dip) poles.
center of the Earth, which was of 490 km in 1980 and was of field already discussed can still be found. Note in particular
563 km in 2010). Finding out whether this SAA will go on the two northern high-latitude lobes, the single southern lati-
growing in the next decades, bringing the inner radiation belt tude polar maximum, and the signature of the SAA. These
closer to the ionosphere and increasing the hazard to LEO maps make it possible to identify the field lines at the CMB
satellites, is one of the important challenges the community that cause those features, as we shall later see.
currently faces. Precisely for the same reasons, maps of the first time deriv-
Inclination, declination, and intensity maps such as those ative of the radial component of the main field are also pro-
shown in Figures 16–21 provide interesting information about vided (Figure 23). Note that those maps do not illustrate
the current morphology of the present main field at the Earth’s changes between 1980 and 2010 as in Figure 21, but the radial
surface. But unless considered simultaneously, they only pro- component of the instantaneous secular variation in 2010, as
vide partial information. Plotting the radial component of inferred from the CHAOS-4a model. Comparing Figure 23 to
the field Br provides a more synthetic and complete image. Figure 21 shows that the present secular variation essentially
Figure 22 shows such maps for the main field in 2010, com- keeps with the trend of the past 30 years, producing fields of
puted from the same CHAOS-4a model and up to the same polarity opposite to that of the main field over the Americas
degree N ¼ 13 as in Figures 16, 18, and 20. Interestingly, and (locally resulting in a decrease in intensity) and producing
somewhat paradoxically, those Br maps seem to contain less fields of correct polarity over India and the Indian Ocean
information than maps of inclination, declination, and inten- (locally resulting in a weak increase in intensity). Note also
sity. This, however, is only because of the overwhelming con- the relatively weak secular variation in the Pacific Ocean,
tribution of the dipole component of the field at the Earth’s where, indeed, the intensity has only weakly changed over
surface. The information is present, but only in the details. the past 30 years. Secular variation, however, is far from
Indeed, the signature of all the basic features of the main being stationary. It changes as a result of secular acceleration,
80
80
80
80
70
80 70 70 70
70
60 60 60
60 70 60
50
60 50 50
50 50 50 40
60 40 40 40 30
40 40
30 50 30 30 30 30
50 40 20 20 20
40 30 20 20 10
30 20 10 10 10 10
20 40 10 0 0 0
30 –10 0 0
20 30 0 –10 –10
10 –10 –10
10 20 20 –10 –20 –20 –20 –20 –20
0
0 10 –20 –30 –30
–30 –30
–10 10 0 0 –40 –40 –30
–10 0 0 –3 –50 –40 –40
–20
–20 –10 –1 –50 –50 –50 –40
0 0 –60
–30 –20 –2 –4 –50
–30 –60
–40 –30 30 –60 –60
–40 –
0
–6
–50
–50 –40
–50 –70
–50 –40 –7
0
–60
–60
–60 –50 –80
–70 0 0
–70 –6 –7
–60 0
–70 –8 0
–8
10 20 30 0 10 –80 –70 –70 –3
–500
20 –60 –40
40 0 –3
30 40 50
30 –100 –3 –50
–50 0
–4
0 50 0
02 40 10
0
–2
1
–6
60
0
–4
20
60
–30 10
–02
60 50
50
–6
–
0–3–5 –60
70
40
30 4
30
0
–10
70
–20 0
00
–40
20
70
10 10
80
10 20 30
0
0
80
60
–5
–4–2
0
60
–7
0 – 20
–20
–70
–6
50
00
80
0
50
40
30
70
–30
30
–10
–5–0
70
–40
10 0
–80
20 40
–50
80
–8
10
–20
0–2– –60
–70
80
0
60 504030
–60
040
0
60
–20
–7
–10
0
50
70 70 –8 –8
0–1 30
–3
–3
0
0
0
–5
–4 20
0 –
60 –70 0
40
–6
10
–5
50 60 –6
0 –70
0
50 2200 0
40 –1 –240
2300 30 0 – 0 – –60 0–
4 0
0– 1 0 30 –40 –50 –50 –3
10 20 200 10 –20 0
0 –10
–10 –2 –10 –30 ––4
20
Figure 19 Inclination I (contour lines every two degrees) at Earth’s surface in 1980.0 as given by the CM4 model up to degree N ¼ 13. Also shown, the
location of the magnetic (dip) poles.
which itself also displays a lot of dynamics, the very reason why N ¼ 13, assuming no sources of any kind are to be found
geomagnetic jerks occur. Maps of the second time derivative of between the Earth’s surface and the CMB. This amounts to
the radial component of the main field, which we plot here in neglect the arguably small contribution of the crustal field to
2005.0 (Figure 24, again as inferred from CHAOS-4a), illus- degrees <13 (recall Section 5.02.3.3) and to ignore electric
trate this point. Epoch 2005.0 is an epoch of particularly currents possibly induced in the mantle by the core field (not
intense secular acceleration, a consequence of an acceleration an unreasonable assumption, but see, e.g., Alexandrescu et al.,
pulse, that started in 2003 with a first geomagnetic jerk and 1999; Backus, 1983; Benton and Whaler, 1983). Figure 25 thus
ended in 2007 with a second, opposite, geomagnetic jerk (see shows the radial component of the core field, which escapes
Chulliat et al., 2010b for details). The (currently) unpredict- the core at the CMB and is responsible for the main field
able occurrence of such events is one of the main reasons IGRF observed at the Earth’s surface.
predictive secular variation models have a hard time properly Not surprisingly, those maps reveal a lot more details than
predicting the main field evolution, even over as little as 5 years the analogous maps shown in Figure 22. This is because of the
(recall Section 5.02.3.2). geometric attenuation, which affects small scales (large degrees
n of the spherical harmonic representation eqn [17]) much
more strongly than large scales (small degrees n), when mov-
5.02.4.3 The Field at the Core Surface
ing away from the sources of the field. In particular, and as
We now consider the present field at the core surface. Figure 25 could be anticipated from the spectrum (recall Figure 14(a)),
shows maps of the radial component of this field in 2010.0. the field is far less dipolar at the core surface than at the Earth’s
Those maps correspond to exactly the same field as the one that surface.
led to Figures 16, 18, 20, and 22 at the Earth’s surface. They Several striking features show up. Most remarkable is the
have been computed using the CHAOS-4a model, up to degree large reverse patch that extends below the southern edges of
20000 30000 40000 50000 60000 70000

(nT )
Figure 20 Scalar field intensity F at Earth’s surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 13. Contour lines every 5000 nT.
Southern America and Africa. Also remarkable are the two scale secular variation at the core surface. Note also how the
high-intensity lobes found under Arctic Canada and Siberia relatively ‘quiet’ Pacific hemisphere at the Earth’s surface is
and the somewhat less intense, partly merged lobes found associated with a similarly quiet hemisphere at the core sur-
under Antarctica. Those features are the cause of the features face, even when small scales are taken into account. Note
we had already identified in the main field at the Earth’s finally the other quiet region below the south geographic
surface. In particular, the reverse patch at the core surface is pole, which strongly contrasts with two strong patches of pos-
the cause of the SAA, the growth of which is directly related to itive and negative secular variation lying close to the north
the growth of this reverse patch. geographic pole. It is the recent dynamics of these two patches
Maps of the first time derivative of the radial component of that produced the sudden acceleration of the north dip pole at
the field can of course also be plotted at the core surface. the Earth’s surface mentioned earlier, while, not surprisingly,
Figure 26 shows such maps plotted from the CHAOS-4a the south dip pole hardly moved (Chulliat et al., 2010a).
model for epoch 2010. In that case, for the reasons already Figure 27 completes this overview and shows maps of the
outlined in Section 5.02.4.1, maps can be plotted with higher second time derivative of the radial component of the field, the
spatial resolution. Here, we chose to plot the secular variation secular acceleration, at the core surface. Despite the limited
up to N ¼ 15, a relatively conservative choice. Again, and not resolution with which this secular acceleration can be recovered
surprisingly, those maps reveal a lot more details than the (recall Section 5.02.4.1; here, we chose to plot maps up to
analogous maps plotted at the Earth’s surface (Figure 23). N ¼ 8), these maps again reveal more details than the analogous
The important role played by the small scales, which dominate maps plotted at the Earth’s surface (Figure 24). In particular,
the signal, is obvious. Note in particular how the relatively they again highlight localized regions of strong secular
large-scale regions of strong secular variation found over the acceleration running in a rough belt in mainly one hemisphere.
Americas and the Indian Ocean at the Earth’s surface are in fact Many of those striking features can also be identified in
produced by regions of equally intense but contrasted small- maps of the historical field (though with much less detail,
–4000 –2000 0 2000 4000

(nT )
Figure 21 Difference of scalar field intensity F between 2010 (CHAOS-4a) and 1980 (CM4) up to degree N ¼ 13. Contour lines every 200 nT.
especially in the far past), as is being done in Chapter 5.05 of (recall Lesur et al., 2010a, Wardinski and Lesur, 2012 and
the present volume, to which the reader is referred. This then Section 5.02.3.3; see also Chapter 8.04 of this treatise). Such
reveals that the growth of the reverse patch responsible for the exercises require quite a few important assumptions (see
SAA probably started several centuries ago, that the Pacific Chapter 5.05 of the present volume). But they can provide
Hemisphere has essentially remained ‘quiet,’ and that the very useful insight into the processes responsible for some of
high-latitude lobes seem to have been fairly stationary over the most striking characteristics of the contemporary Earth’s
the same period of time. magnetic field, such as its hemispheric asymmetry, the current
Understanding the origin of those remarkable features is a growth of the SAA, the current overall decline of the dipole
problem of prime interest. But it also is a formidable task, as it field (see, e.g., Hulot et al., 2002), or the still poorly under-
requires some advanced understanding of the magnetohydro- stood occurrence of geomagnetic jerks (see, e.g., Bloxham
dynamics (MHDs) that governs the dynamo process responsi- et al., 2002; Silva and Hulot, 2012).
ble for the field generation within the core (see Chapter 8.03 of
this treatise). Very substantial progress has already been made
in the numerical simulation of this process (Chapter 8.08 of 5.02.5 Predicting the Future Main Field
this treatise), but direct simulation of the current behavior of
the Earth’s dynamo, best approached via data assimilation Building accurate models of the geomagnetic field is important for
techniques (see next section), is still in its infancy. Neverthe- identifying and studying its sources as well as understanding the
less, many interesting studies have already been carried out by present and past behaviors of the field. As previous sections
jointly analyzing the core field and its first (and sometimes illustrated, very significant progress has already been made in
second) time derivative, to also infer the core surface flows this respect over the past decades. However, an even bigger chal-
accounting for the secular variation, in ways similar to what lenge now to be faced is that of improving the still very rough
is being been done in some core field modeling approaches methods used to forecast the main field. As we saw (Section
–70000 0 70000
(nT )
Figure 22 Radial field component, Br, at Earth’s surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 13. Contour lines every 5000 nT.
5.02.3), the method currently used in the context of the IGRF possibility of assuming a steadily accelerated flow; see also
model is very simple. It consists in assuming that the secular Lesur and Wardinski, 2009; Maus et al., 2009) and more
variation observed over the years preceding the release of such a recently by Beggan and Whaler (2009, 2010) (who directly
model remains stationary over the next 5 years. Such forecasts are tested the forecasting approach), this can bring some slight
very shortsighted (the latest IGRF-11 release provides such fore- improvement to 5-year forecasts. Given our current under-
casts only up to 2015) and not even always accurate. Yet, predict- standing of the core field short-term dynamics, however, it is
ing the evolution of features such as the SAA over the next decades not clear why such alternative assumptions would be more
would be crucial for many applications, starting with spaceborne appropriate than the simple stationary secular variation
technology (recall Section 5.02.4). assumption used for IGRF forecasts. The issue here is very
Simple alternatives to the IGRF forecasting method have similar to that faced by early methods of weather forecasts
already been tested, which explicitly recognize that the main (see, e.g., Kalnay, 2003). Fortunately, taking advantage of the
field is the core field and assume that the observed secular remarkable methodological progress made by weather forecast
variation can be accounted for, and therefore predicted by methods since their early days, and of the no less remarkable
core surface flows under the frozen-flux assumption. The progress made in dynamo simulations (see Chapter 8.08 of
method consists in computing such flows from recent past this treatise) and the modeling of present and past geomag-
secular variation, assuming they (rather than the secular netic data (see Chapters 5.05, 5.09, and 5.11 of the present
variation itself ) remain stationary over the next 5 years, and volume; see also Hulot et al., 2010a, for a recent general
using them to forward the field evolution. This amounts to overview), it now has become possible to envision much
replacing the stationary secular variation assumption by two more advanced forecasting methods, based on data assimila-
new assumptions: the frozen-flux and steady flow assumption techniques. These are designed to combine in an optimal
tions. As shown by Maus et al. (2008a) (who tested the method fashion the information contained in observations and in the
using an equivalent hindcast approach and also tested the physical laws that describe the dynamics of a system.
–250 0 250
(nT year–1)
Figure 23 Secular variation of the radial field component, dBr/dt, at Earth’s surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 15.
Contour lines every 20 nT year1.
In what follows, after a short introduction to key data dynamic trajectory of the system is then entirely controlled
assimilation concepts, we briefly review the current status of by a set of initial conditions (at t ¼ ts) and, possibly, by a set
such data assimilation approaches in the context of of static control parameters. The data can then be used to
geomagnetism and comment on their ability to improve both correct a first estimate of the trajectory to provide an optimal
forecasts of the core field and our knowledge of the geody- fit to the data (in a general least-squares sense). This correction
namo. For complementary and more in-depth reviews, the can be implemented using either a variational or a sequential
reader is invited to consult Fournier et al. (2010) and Kuang approach (Figure 28). The variational approach is rooted in the
and Tangborn (2011). mathematical theory of control and aims at correcting the
initial conditions (and possibly the set of static control param-
eters) by making use of all the data available over the entire [ts,
5.02.5.1 Data Assimilation Concepts te] interval. In contrast, the sequential approach rests on estima-
tion (or filtering) theory. In that case, the stream of observa-
From an inverse problem perspective, data assimilation con- tions is assimilated sequentially, each time a new observation
sists in analyzing observations using the prior information becomes available at, say, t ¼ to E [ts, te]. This extra piece of
supplied by a prognostic numerical model of the physics information can then be used ( Jazwinski, 1970)
assumed to describe the system they sample (see Talagrand,
1997, which provides a very useful concise review on data • to update the past products of analyses at times t < to, which
assimilation). defines the smoothing (hindcast) problem;
Such an analysis is possible if some observations of the • to correct the current estimate of the state of the system, at
system are available over a finite time interval [ts, te] and if a t ¼ to, which defines the filtering problem, for which an
numerical model governing the time dependency of the system elegant solution was originally proposed by Kalman
over this interval is known. In a deterministic setting, the (1960) in a celebrated paper;
–25 0 25
(nT year–2)
Figure 24 Secular acceleration of the radial field component, d Br/dt2, at Earth’s surface in 2005.0 as given by the CHAOS-4a model up to degree
2
N ¼ 8. Contour lines every 2 nT year2.
• to make predictions about the state of the system at times In the case of the sequential approach, this leads to a class of
t > to, which defines the prediction (forecast) problem. methods known as the extended Kalman filter (EKF) and the
ensemble Kalman filter, the latter commonly known as the
The sequential and variational approaches yield the same EnKF (Evensen, 2009). In the variational framework, the most
results if the dynamics of the system is linear and errors affect- popular approach is the so-called 4D-Var approach, whose effi-
ing the data can be assumed to follow a Gaussian statistics. ciency rests on the implementation of the so-called adjoint
More generally, and regardless of their respective merits, both model (see Talagrand, 2010 for a recent review). Both the
approaches illustrate the same philosophy of combining data EnKF and 4D-Var have their supporters, for a variety of reasons
with physical models. Both are used on a daily basis for the (unfortunately too long to be listed here, please consult the
best-known problem of weather forecast, which requires sev- related references in Fournier et al., 2010). Hybrid methods
eral tens of millions of data to be assimilated every day into (such as ensemble variational assimilation) aiming at combin-
physical models of the atmosphere (and ocean), to first initiali- ing the benefits of both methods are also being tested in some
ze a state (or ensemble of states) of the atmosphere (and weather prediction centers (see, e.g., Clayton et al., 2012).
ocean) and subsequently generate weather forecasts (see, e.g., In the context of geomagnetism, efforts to implement such
Kalnay, 2003). data assimilation techniques have started a little less than a
The problem at hand may involve some nonlinearities, for decade ago and are thus not as advanced (Fournier et al., 2010;
instance, in the dynamics or in the relationships linking the state Kuang and Tangborn, 2011). Currently, most efforts aim at
of the system to the available observations. If so, both sequential assimilating data provided in the form of time-varying models
and variational approaches need be adapted. In practice, this of the core field of the type described in the previous sections.
amounts to perform a linearization at some stage in the analysis. Such models, indeed, provide estimates of the only direct
–1000 0 1000
(µT)
Figure 25 Radial field component, Br, at the core surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 13. Contour lines every 100 mT.
5.02.5.2 Closed-Loop Geomagnetic Data Assimilation

geodynamo signal one can reconstruct from magnetic observa-
Experiments
tions, that is, the radial component of the core field at the core
surface (recall Figure 25). Using such models and similar (but Studying the behavior and response of an assimilation system
all the less resolved as we move back in time) models covering using synthetic data in a well-controlled environment is an
the historical (e.g., Gillet et al., 2013; Jackson et al., 2000), important preliminary step in the development of such sys-
archaeological (e.g., Korte and Constable, 2005, 2011; Licht tems. Many geomagnetic data assimilation studies reported so
et al., 2013), and Holocene (e.g., Korte et al., 2011) time far are feasibility studies, in which so-called closed-loop (or
periods, the hope is that, eventually, our ability to forecast twin) experiments are conducted, using dynamo or MHD
the main field will be much improved, while backward- models of varying physical complexity.
propagating the quality of present-day main field models will In two proof-of-concept papers, Fournier et al. (2007) and
increase our knowledge of the past field (the smoothing/hind- Sun et al. (2007) studied two similar one-dimensional non-
casting problem) and help better reconstruct the internal struc- linear toy models, which coupled the evolution of a (pseudo-)
ture and dynamics of Earth’s core. magnetic field B and a (pseudo-)flow field u. They showed,
For this to be possible, however, an efficient transfer of the using, respectively, 4D-Var and an EnKF, that imperfect and
information provided by core surface field models to the core sparse knowledge of B can lead, in a data assimilation frame-
downward, and to other dynamo variables such as the core work, to an improved estimate of u, of which no direct obser-
flow, is a major prerequisite, and relevant numerical dynamo vation is available. This provided good examples of the benefit
models must be chosen. This is not a trivial task, given the coming from the dynamic coupling existing between the
current inability of full 3-D numerical dynamo codes to simul- ‘observable’ field and the ‘hidden’ flow.
taneously match all of the geodynamo nondimensional control Canet et al. (2009) introduced a much more geophysically
parameters (see, e.g., Chapter 8.08 of this treatise). relevant quasi-geostrophic model of core MHD, specifically
–25 0 25
–1)
(µT year
Figure 26 Secular variation of the radial field component, dBr/dt, at the core surface in 2010.0 as given by the CHAOS-4a model up to degree N ¼ 15.
Contour lines every 20 mT year1.
aiming at describing core processes on short secular variation torsional waves (see Volume 8 of this Treatise, Chapter 8.03
timescales. Within this framework, the MHD is still simplified. in particular) and that the 4D-Var developed by Canet et al.
It formally takes place in the equatorial plane of the outer core (2009) later served as the backbone of the study by Gillet et al.
and involves quadratic magnetic quantities averaged along the (2010a) on fast torsional waves in the Earth’s core (see later
direction of rotation of the Earth. In addition, thanks to the text).
strongly columnar nature of the quasi-geostrophic flow, the Studies based on full 3-D numerical dynamo models have
equatorial flow is projected to the core surface, where it inter- also been carried out. Such studies were initiated by Liu et al.
acts with Br (the radial component of the magnetic field) (2007) and Kuang et al. (2008), who resorted to a variant of
through the radial component of the magnetic induction equa- the EKF known as optimal interpolation (OI). OI assumes that
tion. That part of the model is what connects the dynamics of the forecast error statistics required for the correction of the
the model to the observed secular variation, Br then just acting dynamic trajectory (in the form of a covariance matrix Pf) are
as a passive tracer. Using a 4D-Var variational approach and stationary over [ts, te], while the theoretical implementation of
synthetic data (consisting of noisy, large-scale models of Br at the EKF method demands their update every time new infor-
the core surface, mimicking the information one could recover mation is supplied by the data. This choice is dictated by
from true core field models), Canet et al. (2009) showed with computational requirements, since the propagation of forecast
twin experiments that relying on the adjoint model could error statistics is simply out of reach using such 3-D dynamo
enable the retrieval of core state variables, which were not models. Indeed, if nx is the size of the state vector x (i.e., the
directly sampled at the core surface, but took part in the number of field variables describing the system on the compu-
dynamics. Note that the MHD model used by Canet et al. tational grid), then the propagation of Pf is nx times more
(2009) included a particular class of motions known as expensive than a simple model step. Since standard dynamo
–1 0 1
(µT year–2)
Figure 27 Secular acceleration of the radial field component, d2Br/dt2, at the core surface in 2005.0 as given by the CHAOS-4a model up to degree
N ¼ 8. Contour lines every 0.1 mT year2.
State
•
• •
•
•
• •
• •
• • •
• • •
Time
ts te
Sequential Variational
Adjust initial state
State State
Analysis
Analysis
• Analysis • •
• •
• • • • • • • • • • • •
• • • • • • •
• •
• • •
••
Analysis
Analysis
Time Time
ts te ts te
Figure 28 Top: Assimilation starts with an unconstrained model trajectory over the time window of interest (shown in pale blue). It aims at correcting
this initial model trajectory in order to provide an optimal fit to the available observations (squares), given their uncertainties (error bars). Bottom, left:
Sequential approach to data assimilation. Starting from the initial time, the model trajectory follows the initial forecast. As soon as the first observation is
available, the analysis is performed and the trajectory modified. The same cycle is repeated every time an observation is available, with the assimilated
trajectory deviating from the initial guess (the dashed line). Bottom, right: Variational approach to data assimilation. After adjustment of the initial condition
by means of an iterative minimization algorithm, the model trajectory is corrected over the entire time window, in order to provide an optimal fit to the
data (in a generalized least-squares sense). The dashed line corresponds again to the initial (unconstrained) guess of the model trajectory.
simulations have nx ≳ 106, the brute force propagation of error nonlinearities. In both instances, their algorithm appeared to
statistics is still currently impossible. Liu et al. (2007) used a be accurate and stable. Li et al. (2011) also proved that the
modeled Pf (characterized by an ad hoc correlation length) to adjoint dynamo system could be solved directly with the same
connect the surface of the core with its interior. They followed a order of computational complexity as that required to solve the
univariate approach, which means that the Pf they used forward problem, as is common with 4D-Var. This promising
described the connection of the surface poloidal field (the technology lays the ground for future geomagnetic data assim-
one that may escape the core) with the sole poloidal field in ilation over timescales of centuries.
the interior, leaving aside possible correlations with the other
components of the state vector (toroidal magnetic field, flow,
5.02.5.3 First Geomagnetic Applications of Data
and buoyancy; for explanations on the need to distinguish
Assimilation
poloidal from toroidal magnetic fields inside the core where
the dynamo acts, please refer to Chapter 8.03 of this Treatise). As of today, few data assimilation studies have reached the
Liu et al. (2007) then analyzed the behavior produced by stage of application to real geomagnetic data for either main
assimilating synthetic data (again, large-scale surface maps of field forecasting or reconstruction of the internal structure and
Br) into a model whose one control parameter was deliberately dynamics of the Earth’s core.
misspecified with respect to the reference case defining the Beggan and Whaler (2009, 2010) developed an interesting
synthetic ‘truth’ (in this instance, the Rayleigh number, mea- application of the EnKF method to the (very) simple frozen-
suring the vigor of convection; see Chapter 8.05 of this flux/stationary flow forecasting method proposed by Maus
Treatise). They reported a mitigation of the negative impact et al. (2008a) (recall earlier text). They generated an ensemble
of this misspecification in terms of the retrieval of the internal of core flows, which they used to advect Br at the top of the
magnetic structure and a much improved quality of the (syn- core, in a frozen-flux setting. The flows were assumed steady
thetic) forecast produced when assimilating data. over the 5-year long period of assimilation, and observations,
More recently, in a study also using 3-D dynamo models, provided as main field models, were assimilated every year. In
Aubert and Fournier (2011) stressed the importance of resort- their first study, Beggan and Whaler (2009) used an initial
ing to multivariate statistics to increase the accuracy and effi- steady flow computed from 2001.4 to 2004.5 geomagnetic
cacy of the retrieval of the internal dynamo state. Their results data, an initial main field model for epoch 2004, and next
were also obtained within an OI framework. The covariance assimilated high-quality main field models derived from satel-
matrix Pf they used was directly computed from a large number lite data over the 2004–2009 time period. In their subsequent
of snapshots stored during a preliminary model free run; this study, however, Beggan and Whaler (2010) also envisioned the
matrix displayed strong correlations between coefficients likely possibility that forecasting could also have to rely on
describing surface observations and coefficients describing assimilation of lower-quality models of the type derived from
internal fields sharing the same harmonic orders, a natural ground-based observatory data only. Not surprisingly, both of
consequence of the linear coupling of the governing dynamic the Beggan and Whaler (2009, 2010) studies reported much
equations and of the leading influence of the Coriolis forces. more successful forecasts of the main field when using this
Synthetic experiments performed with a weakly nonlinear assimilation scheme (even when assimilating lower-quality
model showed an excellent quantitative retrieval of the internal intermediate main field models) than when just letting the
dynamo state and demonstrated the superiority of predictions initial steady flow forward the initial main field. This paves
based on the OI scheme over predictions based upon linear the way for the possibility of improving on the IGRF forecast-
extrapolation, at least on forecast horizons of up to a few ing method, by assimilating geomagnetic data every year until
decades (when scaling time according to Lhuillier et al., the next IGRF is being produced, rather than by just relying on
2011b). Unfortunately, the use of a strongly nonlinear (and the official IGRF predictive secular variation model over the
more realistic, according to the criteria of Christensen et al., entire 5-year period. Note, however, that such a method would
2010) model resulted in somewhat less accurate estimations only improve predictions over the time period covered by the
and less valuable predictions. In the perspective of future oper- assimilation itself. Also, it explicitly assumes the core flow to be
ational geomagnetic data assimilation activities, this study thus steady over that period, whereas, as already noted, the true flow
highlighted the need for advanced estimation techniques to be may well have some dynamics of its own. Data assimilation
able to cope with the moderate to strong nonlinearities present techniques cannot compensate for the simplicity of the under-
in the geodynamo. lying model used to forward the field evolution.
Finally, and following a strategy complementary to that More ambitious attempts have been made in parallel by
adopted by the two sequential threads just discussed, Li et al. (Kuang et al., 2009, 2010), using 3-D dynamo models and the
(2011) developed the continuous adjoint forms of the general OI approach already implemented in the Liu et al. (2007) and
MHD equations governing convection in the Earth’s core. They Kuang et al. (2008) studies (recall earlier text). Kuang et al.
also developed the corresponding numerical algorithms for (2009) first investigated the response of one such 3-D dynamo
implementation in a fully spectral method and tested their model to the assimilation of 100 years of core field models,
method with two illustrative problems in a whole sphere between 1900 and 2000, as provided by the CM3 model (recall
(without any inner core): the so-called kinematic dynamo prob- Section 5.02.3.5) and earlier models. Both 5 years’ and
lem, where a full flow field is prescribed (see Chapter 8.03 of 20 years’ time steps were tested, and the results were compared
this Treatise), and the Hall-effect dynamo, not only relevant to with an unconstrained free model run to determine how the
neutron stars in astrophysics ( Jones, 1988) but also of interest magnetic and velocity fields were changed within the core and
to the geodynamo problem since it exhibits considerable how this might be related to any improvement in forecasts.
They showed that forecast errors gradually decreased over the radial component of B over a geostrophic cylinder. Observing
(short) course of assimilation, demonstrating that, indeed, the the detailed propagation of the waves, and using these as the
initial dynamo state could progressively assimilate the infor- data to be assimilated by a model of torsional waves (the
mation provided. These encouraging results subsequently led adjoint torsional wave model of Canet et al., 2009), thus
Kuang et al. (2010) to generalize the assimilation of data to a allowed the authors to recover an ensemble of admissible
much longer time period of 7000 years, using a suite of nested radial profiles of the magnetic field inside the core (which,
archaeomagnetic, historical, and CM4 core field models incidentally, is not a time-dependent field variable in the pre-
( Jackson et al., 2000; Korte and Constable, 2005; Sabaka sent case, but a static control parameter). This allowed Gillet
et al., 2004). Inspection of their initial results, however, led et al. (2010a) to estimate the rms field strength inside the core
them to slightly modify the initial assimilation scheme and to to be at least 3 mT, an order of magnitude larger than previ-
design a predictor–corrector scheme to apply upon the fore- ously thought.
cast, the goal of which was to reduce that component of the Such studies, and more advanced studies (Gillet et al.
forecast error due to the dynamo model itself. In essence, the (2010a) only assimilated selected torsional waves and not the
strategy of Kuang et al. (2010) was to consider a duet of original geomagnetic main field models), combined with the
sequences of analysis, one lagging the other by a time shift criteria of Christensen et al. (2010), could therefore help iden-
assumed to be much smaller than the typical timescale of tify which dynamo or MHD models are best suited for imple-
evolution of the dynamo model error. Considering the differ- mentation of an operational geomagnetic data assimilation
ence between the two sequences finally allowed Kuang et al. scheme to forecast the main field beyond the present frustrat-
(2010) to remove a substantial part of that error in their ing IGRF limits. Indeed, the hope to one day produce main
forecast. Applying their method to produce alternative predic- field forecasts with horizons of up to a couple of decades, with
tive secular variation models to those that had been published the same quality to what is currently possible to the much
for the official IGRF-8 and IGRF-9 models (i.e., to forecast the shorter horizon of 5 years, is justified. Unfortunately, however,
main field evolution over the 1995–2000 time period, only and just as is the case for weather forecasts, the nonlinear
using data prior to 1995, and over the 2000–2005 time period, nature of the MHDs that drives the dynamo will ultimately
only using data prior to 2000) allowed them to demonstrate limit this horizon. As recently shown by Hulot et al. (2010b)
some improvements on these predictive models. This finally and Lhuillier et al. (2011a), this nonlinear nature will intrinsi-
led them to also propose a candidate predictive secular varia- cally lead any initial error in the dynamo (and main field) state
tion model to the IGRF-11 for the 2010–2015 time period to double in size within roughly 20 years of forecasting, in the
(recall Section 5.02.3.2), based on data assimilated up to same way this occurs in meteorology within 1.5 days (Kalnay,
2010. This candidate model was indeed included in the pro- 2003), the very reason why meteorologic forecasts beyond 1
duction of the final IGRF-11 predictive secular variation model week remain very uncertain.
(see Finlay et al., 2010a, 2012). It is important to note, how-
ever, that in proceeding in this way, Kuang et al. (2010) only
used data assimilation to hopefully better guess what then Acknowledgments
remains a simple IGRF type of predictive secular variation,
with all the limitations that this implies (recall earlier text). The authors thank Mike Purucker, Susan Macmillan, and Ste-
Demonstrating the power of data assimilation for main field fan Maus for kindly providing the Magsat picture in Figure 7
forecasting and fully taking advantage of the dynamo (or and the material of Figures 6 and 8, Masaru Kono and
MHD) model used for it will require forwarding all of the Hisayoshi Shimizu for their careful reviews, and Joseph
dynamo states (field, flow, buoyancy, etc.), or at least its Ribaudo for pointing out mistakes in the very first edition of
demonstrably critical components, in a dynamically consistent this chapter. This is IPGP contribution 3448.
way. This major step is still ahead of us.
Efforts to take advantage of data assimilation techniques to
better understand the present state of the geodynamo also led
References
to very interesting results. Gillet et al. (2010a), in particular,
resorted to 4D-Var to estimate the strength of the magnetic Alexandrescu M, Gibert D, Hulot G, Le Mouël J, and Saracco G (1996) Worldwide
field deep inside the core, where it cannot be directly ob- wavelet analysis of geomagnetic jerks. Journal of Geophysical Research
served. They began by reconstructing an ensemble of quasi- 101: 21975–21994.
Alexandrescu MM, Gibert D, Le Mouël JL, Hulot G, and Saracco G (1999) An estimate
geostrophic core flow models accounting for the secular of average lower mantle conductivity by wavelet analysis of geomagnetic jerks.
variation observed over the past century. Analysis of this Journal of Geophysical Research 104: 17735–17745.
ensemble of flows next led them to identify the signature of a Alken P, Maus S, Vigneron P, Sirol O, and Hulot G (2013) Swarm SCARF equatorial
torsional wave with a 6-year period, which could also account electric field inversion chain. Earth, Planets and Space 65: 1309–1317.
Alldredge L (1981) Rectangular harmonic analysis applied to the geomagnetic field.
for a synchronous 6-year oscillation in the length-of-day time
Journal of Geophysical Research 86: 3021–3026.
series, via exchange of angular momentum (see also Chapter Aubert J and Fournier A (2011) Inferring internal properties of Earth’s core dynamics
8.04 of this Treatise). Such torsional waves express themselves and their evolution from surface observations and a numerical geodynamo
in the motions of neighboring geostrophic cylinders connected model. Nonlinear Processes in Geophysics 18(5): 657–674. http://dx.doi.org/
with each other by magnetic tension (see Chapter 8.03 of this 10.5194/npg-18-657-2011, ISSN 1023-5809.
Backus GE (1983) Application of mantle filter theory to the magnetic jerk of 1969.
Treatise). The stronger the tension, the larger the wave speed. Geophysical Journal of the Royal Astronomical Society 74: 713–746.
A proper derivation shows that the local wave speed varies Backus G (1986) Poloidal and toroidal fields in geomagnetic field modeling. Reviews of
inversely with the root mean squared average of the cylindrical Geophysics 24: 75–109.
Backus G, Parker R, and Constable C (1996) Foundations of Geomagnetism. New York: Cohen Y, Alexandrescu M, Hulot G, and Le Mouel J (1997) Candidate models for the
Cambridge University Press. 1995 revision of IGRF, a worldwide evaluation based on observatory monthly
Barton C (2002) Survey tracks current position of South magnetic pole. EOS. means. Journal of Geomagnetism and Geoelectricity 49: 279–290.
Transactions of the American Geophysical Union 83(27): 291. Constable CG (1988) Parameter estimation in non-Gaussian noise. Geophysical
Beggan CD, Saarimaeki J, Whaler KA, and Simons FJ (2013) Spectral and spatial Journal 94: 131–142.
decomposition of lithospheric magnetic field models using spherical Slepian Courtillot V, Ducruix J, and Le Mouël J-L (1978) Sur une accélération récente de la
functions. Geophysical Journal International 193(1): 136–148. http://dx.doi.org/ variation séculaire du champ magnétique terrestre. Comptes rendus hebdomadaires
10.1093/gji/ggs122, ISSN 0956-540X. des seances de l’Academie des sciences 287: 1095–1098.
Beggan CD and Whaler KA (2009) Forecasting change of the magnetic field using core De Boor C (2001) A practical guide to splines. Applied Mathematical Sciences (27).
surface flows and ensemble Kalman filtering. Geophysical Research Letters De Michelis P and Tozzi R (2005) A local intermittency measure (LIM) approach to
36: L18303. http://dx.doi.org/10.1029/2009GL039927. the detection of geomagnetic jerks. Earth and Planetary Science Letters 235: 261–272.
Beggan CD and Whaler KA (2010) Forecasting secular variation using core flows. Earth, De Santis A (1991) Translated origin spherical cap harmonic analysis. Geophysical
Planets and Space 62: 821–828. Journal International 106: 253–263.
Benton ER and Whaler KA (1983) Rapid diffusion of poloidal geomagnetic field through Dolginov S, Zhuzgov LN, Pushkov NV, Tyurmina LO, and Fryazinov IV (1962) Some
the weakly conducting mantle: A perturbation solution. Geophysical Journal of the results of measurements of the constant geomagnetic field above the USSR from the
Royal Astronomical Society 75: 77–100. third artificial Earth satellite. Geomagnetism and Aeronomy 2: 877–889.
Bloxham J and Jackson A (1992) Time-dependent mapping of the magnetic field at the Dormy E, Valet J-P, and Courtillot V (2000) Numerical models of the geodynamo and
core–mantle boundary. Journal of Geophysical Research 97: 19537–19563. observational constraints. Geochemistry, Geophysics, Geosystems 1: Paper
Bloxham J, Zatman S, and Dumberry M (2002) The origin of geomagnetic jerks. Nature number 2000GC000062.
420: 65–68. Evensen G (2009) Data Assimilation: The Ensemble Kalman Filter, 2nd edn. Berlin:
Bouligand C, Hulot G, Khokhlov A, and Glatzmaier GA (2005) Statistical palaeomagnetic Springer http://dx.doi.org/10.1007/978-3-642-03711-5.
field modelling and dynamo numerical dynamo simulations. Geophysical Journal Finlay CC, Jackson A, Gillet N, and Olsen N (2012) Core surface magnetic field
International 161: 603–626. evolution 2000–2010. Geophysical Journal International 189(2): 761–781. http://
Buck B and Macaulay VA (eds.) (1991) Maximum Entropy in Action. Oxford: Oxford dx.doi.org/10.1111/j.1365-246X.2012.05395.x.
University Press. Finlay CC, Maus S, Beggan CD, et al. (2010a) International geomagnetic reference field:
Cain JC (2007) POGO (OGO-2, -4 and -6 spacecraft). In: Gubbins D and Herrero-Bervera E The eleventh generation. Geophysical Journal International 183(3): 1216–1230.
(eds.) Encyclopedia of Geomagnetism and Paleomagnetism. Heidelberg: Springer. Fournier A, Eymin C, and Alboussière T (2007) A case for variational geomagnetic data
Cain JC, Wang Z, Kluth C, and Schmitz DR (1989) Derivation of a geomagnetic model to assimilation: Insights from a one-dimensional, nonlinear, and sparsely observed
n ¼ 63. Geophysical Journal 97: 431–441. MHD system. Nonlinear Processes in Geophysics 14: 163–180. http://dx.doi.org/
Campbell WH, Schiffmacher ER, and Kroehl HW (1989) Global quiet day field variation 10.5194/npg-14-163-2007.
model WDCA/SQ1. EOS. Transactions of the American Geophysical Union 70(5): Fournier A, Hulot G, Jault D, et al. (2010) An introduction to data assimilation and
66–74. predictability in geomagnetism. Space Science Reviews 155(1): 247–291.
Canet E, Fournier A, and Jault D (2009) Forward and adjoint quasi-geostrophic models Friis-Christensen E, Lühr H, and Hulot G (2002) Swarm – A constellation to study the
of the geomagnetic secular variation. Journal of Geophysical Research dynamics of the Earth’s magnetic field and its interactions with the Earth system.
114: B11101. http://dx.doi.org/10.1029/2008JB006189. DSRI Report 1/2002. Copenhagen: Danish Space Research Institute.
Chambodut A and Mandea M (2005) Evidence for geomagnetic jerks in comprehensive Friis-Christensen E, Lühr H, and Hulot G (2006) Swarm: A constellation to study the
models. Earth, Planets and Space 57: 139–149. Earth’s magnetic field. Earth, Planets and Space 58: 351–358.
Chambodut A, Panet I, Mandea M, Diament M, Holschneider M, and Jamet O (2005) Friis-Christensen E, Lühr H, Hulot G, Haagmans R, and Purucker M (2009)
Wavelet frames: An alternative to spherical harmonic representation of potential Geomagnetic research from space. EOS. Transactions of the American Geophysical
fields. Geophysical Journal International 163(8): 875–899. Union 90(25): 213–214.
Christensen UR, Aubert J, and Hulot G (2010) Conditions for Earth-like geodynamo Friis-Christensen E, Lühr H, Knudsen D, and Haagmans R (2008) Swarm – An Earth
models. Earth and Planetary Science Letters 296(3–4): 487–496. http://dx.doi.org/ observation mission investigating geospace. Advances in Space Research
10.1016/j.epsl.2010.06.009. 41: 210–216. http://dx.doi.org/10.1016/j.asr.2006.10.008.
Christensen UR and Olson P (2003) Secular variation in numerical geodynamo models Gillet N, Jackson A, and Finlay CC (2007) Maximum entropy regularization of
with lateral variations of boundary heat flow. Physics of the Earth and Planetary time-dependent geomagnetic field models. Geophysical Journal International
Interiors 138: 39–54. 171(3): 1005–1016. http://dx.doi.org/10.1111/j.1365-246X.2007.03521.x.
Christensen U, Olson P, and Glatzmaier G (1999) Numerical modelling of the Gillet N, Jault D, Canet E, and Fournier A (2010a) Fast torsional waves and strong
geodynamo: A systematic parameter study. Geophysical Journal International magnetic field within the earth’s core. Nature 465: 74–77. http://dx.doi.org/
138: 393–409. 10.1038/nature09010.
Chulliat A, Hulot G, and Newitt LR (2010a) Magnetic flux expulsion from the core as Gillet N, Jault D, Finlay CC, and Olsen N (2013) Stochastic modeling of the
a possible cause of the unusually large acceleration of the north magnetic pole earth’s magnetic field: Inversion for covariances over the observatory era.
during the 1990s. Journal of Geophysical Research 115: B07101. http://dx.doi.org/ Geochemistry, Geophysics, Geosystems 14: 766–786. http://dx.doi.org/
10.1029/2009JB007143. 10.1029/2012GC004355.
Chulliat A and Olsen N (2010) Observation of magnetic diffusion in the Earth’s outer Gillet N, Lesur V, and Olsen N (2010b) Geomagnetic core field secular variation models.
core from Magsat, rsted, and CHAMP data. Journal of Geophysical Research Space Science Reviews 155(1–4): 129–145.
115: B05105. http://dx.doi.org/10.1029/2009JB006994. Haines GV (1985) Spherical cap analysis. Journal of Geophysical Research
Chulliat A, Savary J, Telali K, and Lalanne X (2009) Acquisition of 1-second data in 90: 2583–2591.
IPGP magnetic observatories. In: Love JJ (ed.) Proceedings of the XIIIth IAGA Hamilton B (2013) Rapid modelling of the large-scale magnetospheric field from Swarm
Workshop on Geomagnetic Observatory Instruments, Data Acquisition and satellite data. Earth, Planets and Space 65(11): 1295–1308.
Processing, pp. 54–59, U.S. Geological Survey Open-File Report 2009-1226. Hamilton B, Macmillan S, and Thomson A (2010) The BGS magnetic field candidate
Chulliat A and Thébault E (2010) Testing IGRF-11 candidate models against CHAMP models for the 11th generation IGRF. Earth, Planets and Space 62(10): 737–743.
data and quasi-definitive observatory data. Earth, Planets and Space 62(10): http://dx.doi.org/10.5047/eps.2010.05.005.
805–814. http://dx.doi.org/10.5047/eps.2010.06.004. Hayn M, Panet I, Diament M, Holschneider M, Mandea M, and Davaille A (2012)
Chulliat A, Thébault E, and Hulot G (2010b) Core field acceleration pulse as a common Wavelet-based directional analysis of the gravity field: Evidence for large-scale
cause of the 2003 and 2007 geomagnetic jerks. Geophysical Research Letters undulations. Geophysical Journal International 189(3): 1430–1456. http://dx.doi.
37: L07301. http://dx.doi.org/10.1029/2009GL042019. org/10.1111/j.1365-246X.2012.05455.x.
Chulliat A, Vigneron P, Thébault E, Sirol O, and Hulot G (2013) Swarm SCARF Heirtzler J, Allen H, and Wilkinson D (2002) Ever-present South Atlantic Anomaly
dedicated ionospheric field inversion chain. Earth, Planets and Space damages spacecraft. EOS. Transactions of the American Geophysical Union 83(15):
65: 1271–1283. 165–172.
Clayton AM, Lorenc AC, and Barker DM (2012) Operational implementation of a hybrid Holland P and Welsch R (1977) Robust regression using iteratively reweighted
ensemble/4D-Var global data assimilation system at the Met Office. Quarterly least-squares. Communications in Statistics – Theory and Methods 6: 813–827.
Journal of the Royal Meteorological Society 139: 1445–1461. http://dx.doi.org/ Holme R and Bloxham J (1996) The treatment of attitude errors in satellite geomagnetic
10.1002/qj.2054. data. Physics of the Earth and Planetary Interiors 98: 221–233.
Holme R and Olsen N (2006) Core-surface flow modelling from high resolution secular Kuang W, Wei Z, Holme R, and Tangborn A (2010) Prediction of geomagnetic field with
variation. Geophysical Journal International 166: 518–528. data assimilation: A candidate secular variation model for IGRF-11. Earth, Planets
Holme R, Olsen N, Rother M, and Lühr H (2003) CO2: A CHAMP magnetic field model. and Space 62: 775–785.
In: Reigber C, Lühr H, and Schwintzer P (eds.) First CHAMP Mission Results for Kuvshinov AV and Olsen N (2005) 3D modelling of the magnetic field due to ocean
Gravity, Magnetic and Atmospheric Studies, pp. 220–225. Berlin: Springer. flow. In: Reigber C, Lühr H, Schwintzer P, and Wickert J (eds.) Earth Observation
Holschneider M, Chambodut A, and Mandea M (2003) From global to regional with CHAMP, Results from Three Years in Orbit. Berlin: Springer.
analysis of the magnetic field on the sphere using wavelet frames. Physics of the Lalanne X, Peltier A, Chulliat A, Telali A, and Heumez B (2013) A new measurement
Earth and Planetary Interiors 135(2): 107–124. method for magnetic repeat stations. In: Catalan M, Chulliat A, and Hejda P (eds.)
Hongre L, Hulot G, and Khokhlov A (1998) An analysis of the geomagnetic field over the Proceedings of the XVth IAGA Workshop on Geomagnetic Observatory
past 2000 years. Physics of the Earth and Planetary Interiors 106: 311–335. Instruments, Data Acquisition and Processing, San Fernando: Ministerio de Defensa.
Hulot G, Eymin C, Langlais B, Mandea M, and Olsen N (2002) Small-scale structure of Langel RA (1987) The main field. In: Jacobs JA (ed.) Geomagnetism, vol. 1,
the geodynamo inferred from rsted and Magsat satellite data. Nature pp. 249–512. London: Academic Press.
416: 620–623. Langel RA and Estes RH (1982) A geomagnetic field spectrum. Geophysical Research
Hulot G, Finlay CC, Constable CG, Olsen N, and Mandea M (2010a) The magnetic field Letters 9: 250–253.
of planet Earth. Space Science Reviews 152(1–4): 159–222. http://dx.doi.org/ Langel RA and Estes RH (1985a) Large-scale, near-Earth magnetic fields from external
10.1007/s11214-010-9644-0. sources and the corresponding induced internal field. Journal of Geophysical
Hulot G and Le Mouël JL (1994) A statistical approach to the Earth’s main magnetic Research 90: 2487–2494.
field. Physics of the Earth and Planetary Interiors 82: 167–183. http://dx.doi.org/ Langel RA and Estes RH (1985b) The near-Earth magnetic field at 1980 determined from
10.1016/0031-9201(94)90070-1. MAGSAT data. Journal of Geophysical Research 90: 2495–2509.
Hulot G, Lhuillier F, and Aubert J (2010b) Earth’s dynamo limit of predictability. Langel RA, Estes RH, and Mead GD (1982) Some new methods in geomagnetic field
Geophysical Research Letters 37: L06305. http://dx.doi.org/ modelling applied to the 1960–1980 epoch. Journal of Geomagnetism and
10.1029/2009GL041869. Geoelectricity 34: 327–349.
Hulot G, Olsen N, Thébault E, and Hemant K (2009) Crustal concealing of small-scale Langel RA and Hinze WJ (1998) The Magnetic Field of the Earth’s Lithosphere: The
core-field secular variation. Geophysical Journal International 177(2): 361–366. Satellite Perspective. Cambridge: Cambridge University Press.
http://dx.doi.org/10.1111/j.1365-246X.2009.04119.x. Langel RA, Ridgway JR, Sugiura M, and Maezawa K (1988) The geomagnetic field at
Jackson A (1994) Statistical treatment of crustal magnetization. Geophysical Journal 1982 from DE-2 and other magnetic field data. Journal of Geomagnetism and
International 119: 991–998. Geoelectricity 40: 1103–1127.
Jackson A (2003) Intense equatorial flux spots on the surface of the earth’s core. Nature Langel RA, Sabaka TJ, Baldwin RT, and Conrad JA (1996) The near-Earth magnetic field
424: 760–763. from magnetospheric and quiet-day ionospheric sources and how it is modeled.
Jackson A, Constable C, and Gillet N (2007a) Maximum entropy regularization Physics of the Earth and Planetary Interiors 98: 235–267.
of the geomagnetic core field inverse problem. Geophysical Journal Langlais B, Mandea M, and Ultré-Guérard P (2003) High-resolution magnetic field
International 171(3): 995–1004. http://dx.doi.org/10.1111/j.1365- modeling: Application to MAGSAT and rsted data. Physics of the Earth and
246X.2007.03530.x. Planetary Interiors 135: 77–91.
Jackson A, Constable CG, Walker MR, and Parker RL (2007b) Models of Earth’s main Léger JM, Bertrand F, Jager T, Le Prado M, Fratter I, and Lalaurie JC (2009)
magnetic field incorporating flux and radial vorticity constraints. Geophysical Swarm absolute scalar and vector magnetometer based on helium 4 optical pumping.
Journal International 171(1): 133–144. http://dx.doi.org/10.1111/j.1365- Procedia Chemistry 1: 634–637. http://dx.doi.org/10.1016/j.proche.2009.07.158.
246X.2007.03526.x. Lesur V (2006) Introducing localized constraints in global geomagnetic field modelling.
Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular Earth, Planets and Space 58: 477–483.
variation from historical records. Philosophical Transactions of the Royal Society of Lesur V, Macmillan S, and Thomson A (2005) A magnetic field model with daily
London, Series A 358: 957–990. variations of the magnetospheric field and its induced counterpart in 2001.
Jazwinski AH (1970) Stochastic Processes and Filtering Theory, vol. 63. New York: Geophysical Journal International 160(1): 79–88, URL: http://www.blackwell-
Academic Press. synergy.com/doi/abs/10.1111/j.1365-246X.2004.02479.x.
Jones PB (1988) Neutron star magnetic-field decay – Hall drift and Ohmic diffusion. Lesur V, Olsen N, and Thomson AWP (2011) Geomagnetic core field models in the
Monthly Notices of the Royal Astronomical Society 233(4): 875–885. satellite era. In: Mandea M and Korte M (eds.) Geomagnetic Observations and
Kalman RE (1960) A new approach to linear filtering and prediction problems. Models. IAGA Special Sopron Book Series, vol. 5, pp. 277–294. Netherlands:
Transactions of the ASME. Series D, Journal of Basic Engineering 82(1): Springer.
35–45. Lesur V and Wardinski I (2009) Comment on “Can core-surface flow models be used to
Kalnay E (2003) Atmospheric Modeling, Data Assimilation, and Predictability. improve the forecast of the Earth’s main magnetic field?” by Stefan Maus, Luis Silva,
Cambridge: Cambridge University Press. and Gauthier Hulot. Journal of Geophysical Research 114. http://dx.doi.org/
Kivelson MG and Russell CT (1995) Introduction to Space Physics. Cambridge: 10.1029/2008JB006188.
Cambridge University Press. Lesur V, Wardinski I, Asari S, Minchev B, and Mandea M (2010a) Modelling the Earth’s
Korte M and Constable CG (2005) The geomagnetic dipole moment over the last 7000 core magnetic field under flow constraints. Earth, Planets and Space 62(6): 503.
years – New results from a global model. Earth and Planetary Science Letters Lesur V, Wardinski I, Hamoudi M, and Rother M (2010b) The second generation of the
236(1): 348–358. GFZ Reference Internal Magnetic Model: GRIMM-2. Earth, Planets and Space
Korte M and Constable CG (2011) Improving geomagnetic field reconstructions for 0–3 62(10): 765–773. http://dx.doi.org/10.5047/eps.2010.07.007.
ka. Physics of the Earth and Planetary Interiors 188(3–4, SI): 247–259. http://dx. Lesur V, Wardinski I, Rother M, and Mandea M (2008) GRIMM: The GFZ reference
doi.org/10.1016/j.pepi.2011.06.017. internal magnetic model based on vector satellite and observatory data. Geophysical
Korte M, Constable CG, Donadini F, and Holme R (2011) Reconstructing the Holocene Journal International 173(2): 382–394. http://dx.doi.org/10.1111/j.1365-
geomagnetic field. Earth and Planetary Science Letters 312(3–4): 497–505. http:// 246X.2008.03724.x.
dx.doi.org/10.1016/j.epsl.2011.10.031. Lhuillier F, Aubert J, and Hulot G (2011a) Earth’s dynamo limit of predictability
Korte M and Holme R (2003) Regularization of spherical cap harmonics. Geophysical controlled by magnetic dissipation. Geophysical Journal International
Journal International 153(1): 253–262. 186: 492–508.
Kuang W and Tangborn A (2011) Interpretation of core field models. In: Mandea M Lhuillier F, Fournier A, Hulot G, and Aubert J (2011b) The geomagnetic secular-
and Korte M (eds.) Geomagnetic Observations and Models. IAGA Special variation timescale in observations and numerical dynamo models. Geophysical
Sopron Book Series, vol. 5, pp. 295–309. Netherlands: Springer, ISBN: 978-90- Research Letters 38: L09306, http://dx.doi.org/10.1029/2011GL047356.
481-9857-3, URL: http://dx,doi.org/10.1007/978-90-481-9858-0_12. Li K, Jackson A, and Livermore PW (2011) Variational data assimilation for the initial-
Kuang W, Tangborn A, Jiang W, et al. (2008) MoSST-DAS: The first generation value dynamo problem. Physical Review E 84(5, Part 2): 056321.
geomagnetic data assimilation framework. Communications in Computational Licht A, Hulot G, Gallet Y, and Thébault E (2013) Ensembles of low degree
Physics 3: 85–108. archeomagnetic field models for the past three millennia. Physics of the Earth and
Kuang W, Tangborn A, Wei Z, and Sabaka T (2009) Constraining a numerical Planetary Interiors 224: 38–67. http://dx.doi.org/10.1016/j.pepi.2013.08.007.
geodynamo model with 100-years of geomagnetic observations. Geophysical Liu D, Tangborn A, and Kuang W (2007) Observing system simulation experiments in
Journal International 179(3): 1458–1468. http://dx.doi.org/10.1111/j.1365- geomagnetic data assimilation. Journal of Geophysical Research 112:
246X.2009.04376.x. B08103. http://dx.doi.org/10.1029/2006JB004691.
Lowes FJ (1966) Mean-square values on sphere of spherical harmonic vector fields. Mayaud PN (1980) Derivation, Meaning, and Use of Geomagnetic Indices. Geophysical
Journal of Geophysical Research 71: 2179. Monograph, vol. 22. Washington, DC: American Geophysical Union.
Lowes F (1974) Spatial power spectrum of the main geomagnetic field, and Mayer C and Maier T (2006) Separating inner and outer Earth’s magnetic field from
extrapolation to the core. Geophysical Journal of the Royal Astronomical Society CHAMP satellite measurements by means of vector scaling functions and wavelets.
36: 717–730. Geophysical Journal International 167: 1188–1203. http://dx.doi.org/10.1111/
Lowes FJ (1994) The geomagnetic eccentric dipole: Facts and fallacies. Geophysical j.1365-246X.2006.03199.x.
Journal International 118: 671–679. Mayhew M and Estes R (1983) Equivalent source modeling of the core magnetic
Lühr H and Maus S (2010) Solar cycle dependence of quiet-time magnetospheric field using MAGSAT data. Journal of Geomagnetism and Geoelectricity
currents and a model of their near-Earth magnetic fields. Earth, Planets and Space 35: 119–130.
62(10): 843–848. http://dx.doi.org/10.5047/eps.2010.07.012. McLeod MG (1996) Spatial and temporal power spectra of the geomagnetic field.
Lühr H, Maus S, and Rother M (2002) First in-situ observation of night-time F region Journal of Geophysical Research 101(B2): 2745–2763.
currents with the CHAMP satellite. Geophysical Research Letters 29(10). http://dx. Meyers H and Davis WM (1990) A profile of the Geomagnetic Model users and abusers.
doi.org/10.1029/2001GL013845. Journal of Geomagnetism and Geoelectricity 42: 1079–1085.
Lühr H, Rother M, Maus S, Mai W, and Cooke D (2003) The diamagnetic effect of the Neubert T, Mandea M, Hulot G, et al. (2001) rsted satellite captures high-precision
equatorial Appleton anomaly: Its characteristics and impact on geomagnetic field geomagnetic field data. EOS. Transactions of the American Geophysical Union
modelling. Geophysical Research Letters 30(17): 1906. http://dx.doi.org/ 82(7): 81–88.
10.1029/2003GL017407. Newitt LR, Barton CE, and Bitterly J (1996) Guide for Magnetic Repeat Station Surveys.
MacMillan S (2007) Repeat stations. In: Gubbins D and Herrero-Bervera E (eds.) Warsaw: International Association of Geomagnetism and Aeronomy.
Encyclopedia of Geomagnetism and Paleomagnetism, pp. 858–859. Heidelberg: Newitt LR, Chulliat A, and Orgeval JJ (2009) Location of the North magnetic pole in
Springer. April 2007. Earth, Planets and Space 61(6): 703–710.
Macmillan S and Finlay C (2011) The international geomagnetic reference field. O’Brien MS and Parker RL (1994) Regularized geomagnetic field modelling using
In: Mandea M and Korte M (eds.) Geomagnetic Observations and Models. IAGA monopoles. Geophysical Journal International 118: 566–578.
Special Sopron Book Series, vol. 5, pp. 265–276. New York: Springer. Olsen N (1993) The solar cycle variability of lunar and solar daily geomagnetic
Mandea M, Holme R, Pais A, Pinheiro K, Jackson A, and Verbanac G (2010) variations. Annales Geophysicae 11: 254–262.
Geomagnetic jerks: Rapid core field variations and core dynamics. Space Science Olsen N (1998) Estimation of C-responses (3 h to 720 h) and the electrical
Reviews 155(1–4): 147–175. http://dx.doi.org/10.1007/s11214-010-9663-x. conductivity of the mantle beneath Europe. Geophysical Journal International
Marchand R, Burchill J, and Knudsen D (2010) Modelling electrostatic sheath effects on 133: 298–308.
Swarm electric field instrument measurements. Space Science Reviews 156(1–4): Olsen N (2002) A model of the geomagnetic field and its secular variation for epoch
73–87. http://dx.doi.org/10.1007/s11214-010-9735-y. 2000 estimated from rsted data. Geophysical Journal International 149(2):
Matzka J, Chulliat A, Mandea M, Finlay CC, and Qamili E (2010) Geomagnetic 454–462.
observations for main field studies: From ground to space. Space Science Reviews Olsen N (2007) rsted. In: Gubbins D and Herrero-Bervera E (eds.) Encyclopedia of
155(1–4): 29–64. Geomagnetism and Paleomagnetism. Heidelberg: Springer.
Mauersberger P (1956) Das Mittel der Energiedichte des geomagnetischen Hauptfeldes Olsen N, Friis-Christensen E, Floberghagen R, et al. (2013) The Swarm Satellite
an der Erdoberfläche und seine säkulare Änderung. Gerlands Beiträge zur Constellation Application and Research Facility (SCARF) and Swarm Data products.
Geophysik 65: 207–215. Earth, Planets and Space 65: 1189–1200.
Maus S (2007) CHAMP magnetic mission. In: Gubbins D and Herrero-Bervera E (eds.) Olsen N, Haagmans R, Sabaka TJ, et al. (2006a) The Swarm End-To-End mission
Encyclopedia of Geomagnetism and Paleomagnetism. Heidelberg: Springer. simulator study: Separation of the various contributions to Earth’s magnetic field
Maus S and Lühr H (2006) A gravity-driven electric current in the earth’s ionosphere using synthetic data. Earth, Planets and Space 58: 359–370.
identified in champ satellite magnetic measurements. Geophysical Research Letters Olsen N, Holme R, Hulot G, et al. (2000a) rsted initial field model. Geophysical
33: L02812, http://dx.doi.org/10.1029/2005GL024436. Research Letters 27(22): 3607–3610.
Maus S, Lühr H, Rother M, et al. (2007) Fifth generation lithospheric magnetic field Olsen N, Hulot G, and Sabaka TJ (2010a) Measuring the Earth’s magnetic field from
model from CHAMP satellite measurements. Geochemistry, Geophysics, space: Concepts of past, present and future missions. Space Science Reviews
Geosystems 8: Q05013. 155(1–4): 65–93. http://dx.doi.org/10.1007/s11214-010-9676-5.
Maus S, Manoj C, Rauberg J, Michaelis I, and Lühr H (2010b) NOAA/NGDC candidate Olsen N, Lühr H, Sabaka TJ, et al. (2006b) CHAOS – A model of Earth’s magnetic field
models for the 11th generation International Geomagnetic Reference Field and the derived from CHAMP, rsted, and SAC-C magnetic satellite data. Geophysical
concurrent release of the 6th generation POMME magnetic model. Earth, Planets Journal International 166: 67–75. http://dx.doi.org/10.1111/j.1365-
and Space 62(10): 729–735. http://dx.doi.org/10.5047/eps.2010.07.006. 246X.2006.02959.x.
Maus S, McLean S, Dater D, et al. (2005) NGDC/GFZ candidate models for the 10th Olsen N, Mandea M, Sabaka TJ, and Toffner-Clausen L (2009) CHAOS-2 –
generation International Geomagnetic Reference Field. Earth, Planets and Space A geomagnetic field model derived from one decade of continuous satellite data.
57: 1151–1156. Geophysical Journal International 179(3): 1477–1487. http://dx.doi.org/10.1111/
Maus S, Rother M, Hemant K, et al. (2006a) Earth’s crustal magnetic field determined to j.1365-246X.2009.04386.x.
spherical harmonic degree 90 from CHAMP satellite measurements. Geophysical Olsen N, Mandea M, Sabaka TJ, and Toffner-Clausen L (2010b) The CHAOS-3
Journal International 164: 319–330. http://dx.doi.org/10.1111/j.1365- geomagnetic field model and candidates for the 11th generation IGRF. Earth, Planets
246X.2005.02833.x. and Space 62(10): 719–727. http://dx.doi.org/10.5047/eps.2010.07.003.
Maus S, Rother M, Stolle C, et al. (2006b) Third generation of the Potsdam Magnetic Olsen N, Sabaka TJ, and Lowes F (2005) New parameterization of external and induced
Model of the Earth (POMME). Geochemistry, Geophysics, Geosystems 7: Q07008. fields in geomagnetic field modeling, and a candidate model for IGRF 2005.
Maus S, Silva L, and Hulot G (2008a) Can core-surface flow models be used to improve Earth, Planets and Space 57: 1141–1149.
the forecast of the Earth’s main magnetic field? Journal of Geophysical Research Olsen N, Sabaka TJ, and Tffner-Clausen L (2000b) Determination of the IGRF 2000
113: B08102. http://dx.doi.org/10.1029/2007JB005199. model. Earth, Planets and Space 52: 1175–1182.
Maus S, Silva L, and Hulot G (2009) Reply to comment by V. Lesur et al. on “Can Olsen N, Tffner-Clausen L, Sabaka TJ, et al. (2003) Calibration of the rsted vector
core-surface flow models be used to improve the forecast of the Earth’s main magnetometer. Earth, Planets and Space 55: 11–18.
magnetic field” Journal of Geophysical Research 114: B04105. http://dx.doi.org/ Parker RL and Shure L (1982) Efficient modeling of the Earth’s magnetic field with
10.1029/2008JB006242. harmonic splines. Geophysical Research Letters 9: 812–815.
Maus S and Weidelt P (2004) Separating the magnetospheric disturbance magnetic Peltier A and Chulliat A (2010) On the feasibility of promptly producing quasi-definitive
field into external and transient internal contributions using a 1D conductivity model magnetic observatory data. Earth, Planets and Space 62: e5–e8.
of the Earth. Geophysical Research Letters 31: L12614. http://dx.doi.org/ Purucker ME (2007) Magsat. In: Gubbins D and Herrero-Bervera E (eds.) Encyclopedia
10.1029/2004GL020232. of Geomagnetism and Paleomagnetism. Heidelberg: Springer.
Maus S, Yin F, Lühr H, et al. (2008b) Resolution of direction of oceanic magnetic Purucker M, Langlais B, Olsen N, Hulot G, and Mandea M (2002) The southern edge of
lineations by the sixth-generation lithospheric magnetic field model from CHAMP cratonic North America: Evidence from new satellite magnetometer observations.
satellite magnetic measurements. Geochemistry, Geophysics, Geosystems Geophysical Research Letters 29(15): 013645. http://dx.doi.org/
9: Q07021. http://dx.doi.org/10.1029/2008GC001949. 10.1029/2001GL013645.
Maus S, Macmillan S, McLean S, et al. (2010a) The US/UK World Magnetic Model for Reigber C, Lühr H, Schwintzer P, and Wickert J (2005) Earth Observation with CHAMP,
2010–2015. NOAA Technical Report NESDIS/NGDC. Results from Three Years in Orbit. Berlin: Springer.
Richmond AD (1995) Ionospheric electrodynamics using magnetic Apex coordinates. Talagrand O (1997) Assimilation of observations, an introduction. Journal of the
Journal of Geomagnetism and Geoelectricity 47: 191–212. Meteorological Society of Japan 75: 81–99.
Ritter P and Lühr H (2006) Search for magnetically quiet CHAMP polar passes and the Talagrand O (2010) Variational assimilation. In: Lahoz W, Khattatov B, and
characteristics of ionospheric currents during the dark season. Annales Menard R (eds.) Data Assimilation: Making Sense of Observations, pp. 41–67.
Geophysicae 24(11): 2997–3009. http://dx.doi.org/10.1029/2004JA010530. Berlin: Springer, ISBN: 978-3-540-74702-4. http://dx.doi.org/10.1007/978-
Ritter P, Lühr H, and Rauberg J (2013) Determining field-aligned currents with the 3-540-74703-1\_3.
Swarm constellation mission. Earth, Planets and Space 65: 1285–1294. Thébault E (2006) Global lithospheric magnetic field modelling by successive regional
Rother M, Lesur V, and Schachtschneider R (2013) An algorithm for deriving core analysis. Earth, Planets and Space 58: 485–495.
magnetic field models from Swarm data set. Earth, Planets and Space 65: 1223–1231. Thébault E, Chulliat A, Maus S, et al. (2010a) IGRF candidate models at times of rapid
Sabaka TJ and Baldwin RT (1993) Modeling the Sq magnetic field from POGO and changes in core field acceleration. Earth, Planets and Space 62(10): 753–763.
MAGSAT satellite and contemporaneous hourly observatory data: Phase I. Contract http://dx.doi.org/10.5047/eps.2010.05.004.
Report HSTX/G&G9302. Thébault E, Purucker M, Whaler KA, Langlais B, and Sabaka TJ (2010b) The magnetic
Sabaka TJ, Hulot G, and Olsen N (2010) Mathematical properties relevant to field of the earth’s lithosphere. Space Science Reviews 155(1–4): 95–127. http://dx.
geomagnetic field modeling. In: Freeden W, Nashed Z, and Sonar T (eds.) doi.org/10.1007/s11214-010-9667-6.
Handbook of Geomathematics, pp. 503–538. Heidelberg: Springer. Thébault E, Schott JJ, and Mandea M (2006) Revised spherical cap harmonic analysis
Sabaka TJ and Olsen N (2006) Enhancing comprehensive inversions using the Swarm (R-SCHA): Validation and properties. Journal of Geophysical Research 111(B1):
constellation. Earth, Planets and Space 58: 371–395. B01102, http://dx.doi.org/10.1029/2005JB003836.
Sabaka TJ, Olsen N, and Langel RA (2002) A comprehensive model of the quiet- Thébault E, Vervelidou F, Lesur V, and Hamoudi M (2012) The satellite along-track
time near-Earth magnetic field: Phase 3. Geophysical Journal International analysis in planetary magnetism. Geophysical Journal International 188(3):
151: 32–68. 891–907. http://dx.doi.org/10.1111/j.1365-246X.2011.05281.x.
Sabaka TJ, Olsen N, and Purucker M (2004) Extending comprehensive models of Thébault E, Vigneron P, Maus S, Chulliat A, Sirol O, and Hulot G (2013) Swarm
the Earth’s magnetic field with rsted and CHAMP data. Geophysical SCARF dedicated lithospheric field inversion chain. Earth, Planets and Space
Journal International 159: 521–547. http://dx.doi.org/10.1111/ 65: 1257–1270.
j.1365-246X.2004.02421.x. Thomson AWP, Hamilton B, Macmillan S, and Reay SJ (2010) A novel weighting
Sabaka TJ, Tffner-Clausen L, and Olsen N (2013) Use of the Comprehensive method for satellite magnetic data and a new global magnetic field model.
Inversion method for Swarm Satellite Data. Earth, Planets and Space Geophysical Journal International 181(1): 250–260. http://dx.doi.org/10.1111/
65: 1201–1222. j.1365-246X.2010.04510.x.
Schumaker LL (1981) Spline Functions: Basic Theory. New York: Wiley. Thomson A and Lesur V (2007) An improved geomagnetic data selection algorithm for
Shure L, Parker RL, and Backus GE (1982) Harmonic splines for geomagnetic global geomagnetic field modelling. Geophysical Journal International 169(3):
modelling. Physics of the Earth and Planetary Interiors 28: 215–229. http://dx.doi. 951–963. http://dx.doi.org/10.1111/j.1365-246X.2007.03354.x.
org/10.1016/0031-9201(82)90003-6. Voorhies CV (2004) Narrow-scale flow and a weak field by the top of earth’s core:
Siebert M and Meyer J (1996) Geomagnetic activity indices. In: Dieminger W, Evidence from rsted, MAGSAT, and secular variation. Journal of Geophysical
Hartman GK, and Leitinger R (eds.) The Upper Atmosphere, pp. 887–911. Berlin Research 109. B03106: http://dx.doi.org/10.1029/2003JB002833.
Heidelberg/New York: Springer. Walker MR and Jackson A (2000) Robust modelling of the earth’s magnetic field.
Silva L and Hulot G (2012) Investigating the 2003 geomagnetic jerk by simultaneous Geophysical Journal International 143: 799–808.
inversion of the secular variation and acceleration for both the core flow and its Wardinski I and Holme R (2006) A time-dependent model of the Earth’s magnetic
acceleration. Physics of the Earth and Planetary Interiors 198: 28–50. http://dx.doi. field and its secular variation for the period 1980–2000. Journal of Geophysical
org/10.1016/j.pepi.2012.03.002. Research 111: B12101. http://dx.doi.org/10.1029/2006JB004401.
Silva L, Maus S, Hulot G, and Thébault E (2010) On the possibility of extending the Wardinski I and Lesur V (2012) An extended version of the C3FM geomagnetic field
IGRF predictive secular variation model to a higher SH degree. Earth, Planets and model: Application of a continuous frozen-flux constraint. Geophysical Journal
Space 62(10): 815–820. http://dx.doi.org/10.5047/eps.2010.07.005. International 189: 1409–1429.
Stockmann R, Finlay CC, and Jackson A (2009) Imaging Earth’s crustal magnetic field Wieczorek MA and Simons FJ (2005) Localized spectral analysis on the sphere.
with satellite data: A regularized spherical triangle tessellation approach. Geophysical Journal International 162(3): 655–675. http://dx.doi.org/10.1111/
Geophysical Journal International 179(2): 929–944. http://dx.doi.org/10.1111/ j.1365-246X.2005.02687.x.
j.1365-246X.2009.04345.x. Winch DE (1981) Spherical harmonic analysis of geomagnetic tides. Philosophical
Stolle C, Lühr H, Rother M, and Balasis G (2006) Magnetic signatures of equatorial Transactions of the Royal Society of London, Series A 303: 1–104.
spread F as observed by the CHAMP satellite. Journal of Geophysical Research Worthington EW, Sauter EA, and Love JJ (2009) Analysis of USGS one-second
111: A02304, http://dx.doi.org/10.1029/2005JA011184. data. In: Love JJ (ed.) Proceedings of the XIIIth IAGA Workshop on
Sun Z, Tangborn A, and Kuang W (2007) Data assimilation in a sparsely observed one- Geomagnetic Observatory Instruments, Data Acquisition and Processing,
dimensional modeled MHD system. Nonlinear Processes in Geophysics 14(2): pp. 262–266. Washington, DC: U.S. Geological Survey, Open-File
181–192. http://dx.doi.org/10.5194/npg-14-181-2007. Report 2009-1226.
5.03 Magnetospheric Contributions to the Terrestrial Magnetic Field
W Baumjohann and R Nakamura, Austrian Academy of Sciences, Graz, Austria
This chapter is reproduced from the previous edition, Volume 5, pp. 77–92, © 2007, Elsevier B.V.
5.03.2 Geophysical Plasmas 79
5.03.2.1 Solar Wind 80
5.03.2.2 Magnetosphere 80
5.03.2.3 Ionosphere 80
5.03.2.4 Currents 81
5.03.3 Plasma Dynamics 81
5.03.3.1 Single-Particle Motion 82
5.03.3.2 Trapped Particles 83
5.03.3.3 Collisions and Conductivity 83
5.03.3.4 Convection and Merging 83
5.03.4 Low- and Midlatitude Currents 84
5.03.4.1 Sq Current 84
5.03.4.2 Equatorial Electrojet 84
5.03.4.3 Ring Current 85
5.03.4.4 Storms and Sudden Commencements 85
5.03.5 High-Latitude Currents 86
5.03.5.1 Magnetospheric Convection 86
5.03.5.2 Ionospheric Convection 86
5.03.5.3 Auroral Electrojets 87
5.03.5.4 Substorms 87
5.03.6 Geomagnetic Pulsations 88
5.03.6.1 Pc5 Pulsations 89
5.03.6.2 Pi2 Pulsations 89
5.03.7 Conclusions 89
References 90
5.03.1 Introduction of seconds, others on the scale of days, and typical disturbance
amplitudes on the ground range between a few nanoteslas and
The Earth’s magnetic field is created and governed by processes some hundred nanoteslas.
and material in the Earth’s interior. This field is not restricted to
the inside, the surface, or the atmosphere of the Earth, but
reaches far above the Earth into space. If that space were 5.03.2 Geophysical Plasmas
empty or only populated with neutral gases, there would be
no consequences. However, that space is not a vacuum but, A plasma is a gas of charged particles, which consists of equal
starting at a height of about 100 km, is filled with ionized gas. numbers of free positive and negative charge carriers. Having
The constituents of this ionized gas, a plasma of positively roughly the same number of charges with different signs in the
charged ions and negatively charged electrons, are not immo- same volume element guarantees that the plasma behaves
bile but rather move around under the influence of externally quasineutral. On average, a plasma looks electrically neutral
applied and internally generated forces. This motion of charge to the outside, since the randomly distributed particle electric
carriers often results in a significant current flow and thus in charge fields mutually cancel. However, because of its sensitiv-
the generation of magnetic field disturbances, which can be ity to electric and magnetic fields and its ability to carry electric
of the same magnitude as the field generated inside the Earth at currents and thus to generate magnetic fields, this fourth state
that altitude. If the spatial scale of such a current system is large of matter behaves quite different from a neutral gas.
enough or if it flows close to the Earth’s surface, as do for Similar to a gaseous medium, the charged plasma particles
ionospheric currents, it can generate significant magnetic var- are essentially free particles. Since the particles in a plasma
iations at the Earth surface. have to overcome the Coulomb coupling with their neighbors,
In this chapter, we will describe the main sources of external they must have thermal energies above some 105 K. Thus, a
magnetic field contributions. Nearly all of them result from an typical plasma is a hot and highly ionized gas. While only a few
interplay between the magnetic field of the Earth and that of natural plasmas, such as flames or lightning strokes, can be
the Sun. They are highly variable, some changing on the scale found near the Earth’s surface or below the ionosphere,
Treatise on Geophysics, Second Edition http://dx.doi.org/10.1016/B978-0-444-53802-4.00097-X 79

80 Magnetospheric Contributions to the Terrestrial Magnetic Field
plasmas are abundant in the universe. More than 99% of all oscillate back and forth between the two hemispheres
normal matter (baryonic matter, not including dark matter) is (see succeeding text). Typical electron densities and tempera-
in the plasma state. tures in the ring current are 1 cm3 and 5 107 K. The mag-
Extraterrestrial plasmas have a wide spread in their charac- netic field strength in this region is a few hundred nanoteslas.
teristic parameters like density, temperature, and magnetic Most of the magnetotail plasma is concentrated around the
field. Even in the Earth’s neighborhood, there are quite a tail midplane in an about 5–10 RE thick plasma sheet. Near the
number of different geophysical plasmas. Earth, it reaches down to the high-latitude auroral ionosphere
along the field lines. Average electron densities and tempera-
tures in the plasma sheet are 0.5 cm3 and 5 106 K, with
5.03.2.1 Solar Wind
magnetic fields of 10–20 nT.
The Sun emits a highly conducting plasma into interplanetary The outer part of the magnetotail is called the lobe. It is
space as a result of the supersonic expansion of the solar threaded by magnetic field lines originating in the polar caps
corona. This plasma is called the solar wind. It flows with and contains a highly rarefied plasma. Typical values for the
supersonic speed of about 500 km s1 and consists mainly of electron density, temperature, and the magnetic field strength
electrons and protons, with an admixture of 5% helium ions. are 102 cm3, 5 105 K, and 30 nT, respectively.
Because of the high conductivity, the solar magnetic field is
‘frozen’ in the plasma (as in a superconductor; see succeeding
text) and drawn outward by the expanding solar wind. Typical 5.03.2.3 Ionosphere
values for electron density and temperature in the solar wind
near the Earth are 5 cm3 and 105 K, respectively. The inter- The solar ultraviolet light impinging on the Earth’s atmosphere
planetary magnetic field strength is of the order of 5–10 nT ionizes a fraction of the neutral atmosphere. At altitudes above
near the Earth’s orbit.
When the solar wind impinges on the Earth’s dipolar mag-
Int mag
Magnetosheath
erp ne
netic field, it cannot simply penetrate it, but is slowed down
k
oc
lan tic
and, to a large extent, deflected around it. Since the solar wind
sh
eta fie
w
hits the obstacle with supersonic speed, a bow shock wave is
Bo
ry ld
generated (see Figure 1), where the plasma is slowed down and
a substantial fraction of the particles’ kinetic energy is con- Magnetosphere
verted into thermal energy. The region of thermalized subsonic
plasma behind the bow shock is called the magnetosheath. Its
plasma is denser and hotter than the solar wind plasma and the Solar wind
magnetic field strength has higher values in this region. T
ma erres
gne trial
tic fie
ld
5.03.2.2 Magnetosphere
Magnetopause
The shocked solar wind plasma in the magnetosheath cannot
easily penetrate the terrestrial magnetic field but is mostly
deflected around it. This is a consequence of the fact that the Figure 1 Solar wind interaction with the terrestrial magnetic field.
interplanetary magnetic field lines cannot penetrate the terres- Adapted from Baumjohann W and Treumann RA (1996) Basic Space
trial field lines and that the solar wind particles cannot leave Plasma Physics. London: Imperial College Press, with permission.
the interplanetary field lines due to the aforementioned frozen-
in characteristic of a highly conducting plasma.
The boundary separating the two different regions is called
magnetopause and the cavity generated by the terrestrial field
has been named magnetosphere (see Figures 1 and 2). The Lobe
kinetic pressure of the solar wind plasma distorts the outer part e
of the terrestrial dipolar field. On the dayside, it compresses the a us
top
field, while the nightside magnetic field is stretched out into a

ne
Plasmasphere
long magnetotail, which reaches far beyond the lunar orbit.
Mag
The plasma in the magnetosphere consists mainly of elec-

trons and protons. The sources of these particles are the solar
wind and the terrestrial ionosphere. In addition, there are
minor fractions of He+ and O+ ions of ionospheric origin Ring current Plasma sheet
(more prominent at lower altitudes) and some He2+ ions orig-
inating from the Sun. However, the plasma inside the magne-
tosphere is not evenly distributed, but is grouped into different
regions with quite different densities and temperatures.
Figure 2 depicts the topography of some of these regions. Figure 2 Plasma regions in the Earth’s magnetosphere. Note that ‘ring
The ring current lies on dipolar field lines between about 4 current’ and ‘plasmasphere’ partially overlap in reality. Adapted from
and 6 RE (1 Earth radius (RE) ¼ 6371 km). It consists of ener- Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics.
getic electrons and ions that move along the field lines and London: Imperial College Press, with permission.
Magnetospheric Contributions to the Terrestrial Magnetic Field 81
80 km, collisions are too infrequent to result in rapid recom- central plasma sheet, both of which are connected and form
bination and a permanent ionized population called the ion- a y-like current system, if seen from along the Earth–Sun line.
osphere is formed. Typical electron densities and temperatures Another large-scale current system, which mainly influences
in the midlatitude ionosphere are 105 cm3 and 103 K. The the configuration of the inner magnetosphere, is the ring cur-
magnetic field strength is of the order of 104 nT. rent. The ring current flows around the Earth in a westward
The ionosphere extends to altitudes of about a thousand direction at radial distances of several Earth radii and is carried
kilometers and, at low latitudes and midlatitudes, gradually by trapped energetic particles, which oscillate back and forth
merges into the plasmasphere. As depicted in Figure 2, the along the field lines. In addition to their bouncing motion, these
plasmasphere is a torus-shaped volume inside the ring current. particles drift around the Earth. Since the protons drift westward
It contains a cool but dense plasma of ionospheric origin, while the electrons move in the eastward direction, this consti-
which corotates with the Earth. In the equatorial plane, the tutes a net charge transport and thus a current.
plasmasphere extends out to about 4 RE, where the density A number of current systems exist in the conducting layers
drops down sharply to about 1 cm3. This boundary is called of the Earth’s ionosphere, at altitudes of 100–150 km. Most
the plasmapause. notable are the auroral electrojets inside the auroral oval, the
At high latitudes, plasma sheet electrons can precipitate Sq currents in the dayside midlatitude ionosphere, and the
along magnetic field lines down to ionospheric altitudes, equatorial electrojet near the magnetic equator.
where they collide with and ionize neutral atmosphere parti- In addition to the currents that flow across the magnetic
cles. As a by-product, photons emitted by this process create field lines, currents also flow along magnetic field lines. As
the polar light, the aurora. These auroras are typically observed shown in Figure 3, the field-aligned currents connect the mag-
inside the auroral oval, which is a 5–10 wide belt around 70 netospheric currents to those flowing in the polar ionosphere.
northern or southern magnetic latitude, containing the ‘foot- The field-aligned currents are mainly carried by electrons and
prints’ of those field lines that thread the plasma sheet. are essential for the exchange of energy and momentum
between these regions.
5.03.2.4 Currents
The plasmas discussed in the last section are usually not sta- 5.03.3 Plasma Dynamics
tionary but move under the influence of external forces.
Sometimes, ions and electrons move together, like in the The dynamics of a plasma is governed by the interaction of the
solar wind. But in other plasma regions, ions and electrons charge carriers with the electric and magnetic fields. If all the
often move in different directions, creating an electric current. fields were of external origin, the physics would be relatively
Such currents create magnetic fields, which distort the Earth’s simple. However, as the particles move around, they may create
internal field, most intensely at higher altitudes. local space charge concentrations and thus electric fields. More-
Actually, the distortion of the internal dipole field into the over, their motion can also generate electric currents and thus
typical shape of the magnetosphere is accompanied by electric magnetic fields. These internal fields and their feedback onto the
currents. As schematically shown in Figure 3, the compression motion of the plasma particles make plasma physics complex.
of the internal magnetic field on the dayside is associated with In general, the dynamics of a plasma can be described by
current flow across the magnetopause surface, the magneto- solving the equation of motion for each individual particle.
pause current. The taillike field of the nightside magnetosphere Since the electric and magnetic fields appearing in each equation
is accompanied by the current flowing on the tail magneto- include the internal fields generated by every other moving
pause surface and the cross-tail neutral sheet current in the particle, all equations are coupled and have to be solved simul-
taneously. Such a full solution is not only too difficult to obtain
but also of no practical use, since most of the time one is
Tail MP interested in knowing average quantities like density and tem-
current
perature rather than the individual velocity of each particle.
Therefore, one usually makes certain approximations suitable
to the problem studied. For studying the macroscopic interac-
tion between the solar wind and the Earth’s magnetosphere, two
approaches are most useful (the most developed theoretical
approach, the so-called kinetic theory of plasmas, is typically
needed for microphysical aspects of space plasma physics).
The simpler approach is the single-particle motion or
guiding-center description. It describes the motion of a particle
Ring current under the influence of external electric and magnetic fields.
Neutral sheet current This approach neglects the collective behavior of a plasma, but
Field-aligned currents is useful when studying a very low-density plasma, threaded by
Magnetopause current strong magnetic fields, like that found in the ring current.
The magnetohydrodynamic approach, on the other hand,
Figure 3 Magnetospheric current systems. MP current, magnetopause neglects all single-particle aspects, but includes collective
current. Adapted from Baumjohann W and Treumann RA (1996) Basic effects. The plasma is treated as a single conducting fluid with
Space Plasma Physics. London: Imperial College Press, with permission. macroscopic variables, like average density, velocity, and
temperature. The approach assumes that the plasma is able to and decelerated when moving parallel. But since their sense of
maintain local equilibriums and is suitable to study low- gyration is also opposite, their guiding centers drift into the
frequency wave phenomena in highly conducting fluids same direction.
immersed in magnetic fields. Up to now, we have assumed that the magnetic field is
homogeneous. This is definitely not the case in the magneto-
sphere, where the magnetic field has gradients and field lines
are curved. This inhomogeneity of the magnetic field leads to a
5.03.3.1 Single-Particle Motion
‘magnetic’ drift of charged particles. As visualized in Figure 6,
In a situation where the charged particles do not directly inter- in a magnetic field configuration with a gradient in field
act with each other and where they do not affect the external strength, the gyroradius of a particle decreases in the upward
magnetic field significantly, the motion of each individual direction, and thus, the gyroradius of a particle will be larger at
particle can be treated independently. This single-particle the bottom of the orbit than during the top half. As a result,
approach is only valid in very rarefied plasmas where collective ions and electrons drift into opposite directions, perpendicular
effects are negligible. Furthermore, the external magnetic field to both B and rB. Since ions and electrons gyrate in the
must be much greater than the magnetic field produced by the opposite sense, ions and electrons also drift in opposite direc-
electric current due to the charged-particle motion. tions. The gradient drift velocity is proportional to the perpen-
The equation of motion for a particle of charge q under the dicular gyratory energy of the particle, W┴ ¼ (1/2)mv2?: more
action of the Coulomb and Lorentz forces can be written as energetic particles drift faster, since they have a larger gyrora-
dius and experience more inhomogeneity of the field. The
dv
m ¼ qðE + v BÞ [1] opposite drift directions of electrons and ions lead to a trans-
dt
verse current.
where m represents the particle mass and v the particle velocity. The ‘gradient’ drift is only one component of the particle
Under the absence of an electric field and a homogeneous drift in an inhomogeneous magnetic field. When the field lines
magnetic field, eqn [1] describes a circular orbit of the particle are curved, a ‘curvature’ drift appears. Due to their parallel
around the magnetic field, with the sense of rotation depend- velocity, the particles experience a centrifugal force. The curva-
ing on the sign of the charge. The center of the orbit is called ture drift velocity is proportional to the parallel particle energy
the guiding center. The gyroradius of the particle orbit and perpendicular to the magnetic field and its curvature. It
increases with the particle’s momentum and decreases for again creates a transverse current since ion and electron drifts
stronger magnetic fields. A possible constant velocity of the have opposite signs.
particle parallel to the magnetic field will make the actual In a cylindrically symmetrical field, like in a dipole field,
trajectory of the particle three-dimensional and look like a gradient and curvature drifts can be combined to
helix (see Figure 4).
Taking the electric field into consideration will result in a
drift of the particle superimposed onto its gyratory motion. E lon
Since, due to the high mobility of electrons, parallel electric
fields can typically not be maintained in geophysical plasmas,
solving eqn [1] yields
EB
vE ¼ [2]
B2
The E B drift is independent of the sign of the charge, and
thus, electrons and ions move together with the same speed in
B Electron
the same direction.
Figure 5 shows the acceleration and deceleration effect of a Figure 5 Particle drifts due to an electric field. Adapted from
perpendicular electric field and explains the E B drift in an Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics.
intuitive way. An ion is accelerated in the direction of the London: Imperial College Press, with permission.
electric field, thereby increasing its gyroradius. But it is decel-
erated during the second half of its gyratory orbit, now with a Δ
decreasing gyroradius. The changing gyroradii shift the posi- B lon
tion of the guiding center in the E B direction. The electrons
are accelerated when moving antiparallel to the electric field
B Electron
Figure 4 Ion orbit in a uniform magnetic field. Adapted from Figure 6 Particle drifts due to a magnetic field gradient. Adapted from
Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics. Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics.
London: Imperial College Press, with permission. London: Imperial College Press, with permission.

1 B rB interaction between the particles has been neglected. Interaction
vB ¼ v r + v R ¼ v2jj + v2? [3]
2 og B2 in plasmas is, however, unavoidable and collective effects con-
stitute the very nature of plasma physics. The simplest kind of
where vr and vR are the gradient and curvature drift velocity, interaction between particles is a direct collision. The partially
og gives the gyrofrequency, and the subscripts ┴ and jj denote ionized plasma of the terrestrial ionosphere is a good example
components perpendicular and parallel to the ambient back- for such interactions. Here, collisions between charged and
ground field, respectively. The transverse current associated neutral particles create electric resistivity and current flow.
with this full magnetic drift creates the magnetospheric ring In the presence of collisions, a collisional term has to be
current, mentioned in the preceding text and further detailed added to the equation of motion [1] for a charged particle
later in the text. under the action of the Coulomb and Lorentz forces. Assuming
all collision partners are at rest, then
5.03.3.2 Trapped Particles dv

m ¼ qðE + v BÞ mvc v [4]
dt
In slowly varying magnetic and electric fields, not only the total
energy of a particle, W ¼ Wjj + W┴, is constant, but also the The collisional term on the right-hand side describes the
magnetic moment of a particle, m ¼ W┴/B, does not change momentum lost through collisions occurring at a frequency vc.
with time. Quantities like m are called adiabatic invariants It is often called frictional term since it impedes motion.
and are not absolute constants like total energy or total The friction term introduces a differential motion between
momentum, but change only very slowly. electrons and ions and thus a current, even in homogeneous
Since the magnetic moment is invariant and the total magnetic fields. In fact, the equation in the preceding text
energy is a constant of motion, only the ratio between perpen- reduces to a generalized Ohm’s law
dicular energy and parallel energy increases or decreases along
j ¼ sðE + v BÞ [5]
the guiding-center trajectory. In a converging magnetic field
geometry, a particle moving into regions of stronger fields will which is valid in all geophysical plasmas where the typical
have its transverse energy increasing at the expense of its collision frequencies are low and s is the plasma conductivity.
parallel energy. If there is a point along the field line where While treating the plasma conductivity as a scalar is war-
all of the particle energy is in W┴, the particle cannot penetrate ranted in the dilute magnetospheric plasma, there is one place
any further along the field line into the stronger field where we have to take the anisotropy introduced by the pres-
region. Actually, it does not stop, but is pushed back by the ence of a strong magnetic field into account. This is the lower
parallel component of the gradient force, the so-called mirror part, the so-called E region, of the partially ionized terrestrial
force, mrjjB, from this mirror point. ionosphere, at about 100–130 km altitude, where abundant
A dipole magnetic field has a field strength minimum at the collisions between the ionized part and the neutral part of the
equator and converging field lines in both hemispheres. In such upper atmosphere might even interrupt the cyclotron motion
a configuration, particles will be trapped and bounced back and of electrons and/or ions, leading to an anisotropic conductivity
forth between their mirror points in the Northern and Southern tensor and a different form of Ohm’s law:
Hemispheres (see Figure 7). The particles do not only gyrate and
j ¼ sjj Ejj + sP E? sH ðE? BÞ=B [6]
bounce but also undergo a slow azimuthal drift due to the
combined effect of the gradient and curvature of the dipole The Hall conductivity, sH, determines the Hall current in
magnetic field as described in eqn [3]. The ions drift westward the direction perpendicular to both the electric field and mag-
while the electrons move eastward around the Earth. It is the netic field. The Hall conductivity maximizes near 100 km alti-
current associated with this drift that constitutes the ring current. tude, where the ions collide so frequently with the neutrals that
they are essentially at rest, while the electrons already undergo
a somewhat impeded E B drift. The Pedersen conductivity,
5.03.3.3 Collisions and Conductivity sP, governs the Pedersen current in the direction of that part of
So far, only the motion of single particles in external and slowly the electric field, E┴, which is transverse to the magnetic field.
variable electromagnetic fields has been considered, but any The Pedersen conductivity maximizes near a height of 125 km,
since here the ions are scattered in the direction of the electric
field before they can start to gyrate about the magnetic field.
N The element sjj is called parallel conductivity since it governs
Trajectory of the magnetic field-aligned current driven by the parallel electric
trapped particle
field component, Ejj.
Mirror point
lons Electrons
Magnetic field line
5.03.3.4 Convection and Merging
S While collisions play an important role in the ionosphere,
most space plasmas are essentially collisionless. Hence, the
Figure 7 Trajectories of particles trapped on closed dipolar field lines. conductivity in the magnetospheric or in the solar wind plasma
Adapted from Baumjohann W and Treumann RA (1996) Basic Space is near infinite. As in a superconductor, magnetic field lines are
Plasma Physics. London: Imperial College Press, with permission. frozen in the plasma and both move together under the action
of external forces. In particular, under the influence of an magnetospheric particles, the ring current. It is concentrated in
external electric field, the so-called flux tubes, bundles of field the equatorial region of the Earth’s magnetosphere.
lines filled with plasma simply drift following eqn [2]. On the
other hand, if forces are exerted on the magnetic field lines
5.03.4.1 Sq Current
leading to a motion of the flux tubes, an electric field will be
generated. The latter is often called convection electric field. The relation between current, conductivity, electric field, and
However, there is an exception. Under certain conditions, neutral winds can be seen by replacing E┴ with E┴ + vn B in
especially in the thin and intense current sheets of the magne- the Ohm’s law given earlier in the text. For mid- and low-
topause and the magnetotail neutral sheet, strong plasma latitude dynamo currents, the dominant driving force for the
waves or inertial effects may substitute collisions and lower current is actually the E B field induced by the motion of
the conductivity to a finite value. In this case, the magnetic field ions, which are coupled to the neutral atmosphere via colli-
lines can diffuse through the plasma. This rarely has major sions and thus move with the neutral wind, across the mag-
consequences, except for a situation as depicted in Figure 8. netic field. (For auroral oval current systems discussed later, the
Consider a magnetic topology with antiparallel field lines neutral wind term is usually much smaller than the electric
frozen into the plasma, as depicted in the left-hand side of field term and can be neglected.)
Figure 8. If the flux tubes are stagnant and do not move, The most important dynamo effect at midlatitudes is the
nothing will happen. However, when plasma and field lines daily variation of the atmospheric motion caused by the tides
on both sides move toward each other, the situation may of the atmosphere, that is, the diurnal and semidiurnal oscil-
change. When the conductivity becomes finite in a small vol- lations, which are excited by the heating of the atmosphere due
ume of space, the magnetic field can vanish due to diffusion at to solar radiation. The current system created by this tidal
a particular point. This results in the X-type configuration motion of the atmosphere is called solar quiet or Sq current.
shown in the middle panel of Figure 8, with the magnetic This current system creates daily magnetic variations, records
field being zero at the center of the X, the magnetic neutral of which are obtained at many different magnetic observatories
point. The result will be the situation depicted on the right- distributed across the globe. These recordings can be used to
hand side of Figure 8. Plasma and field lines are being trans- construct the Sq current system. More sophisticated methods
ported toward the neutral point from either side. At the neutral use measured wind patterns, conductivities, and disturbance
point, the antiparallel field lines are cut into halves and the magnetic fields and calculate electric fields and currents based
field line halves from one side are reconnected with those from on Ohm’s law and the Biot–Savart law.
the other side. The merged field lines are then expelled from Figure 9 presents a global view of the average Sq current
the neutral point. The merged field lines will be populated by a system from above the terrestrial ionosphere: The lines give the
mixture of plasma from both sides. direction of the current while the distance between the lines is
inversely proportional to the height-integrated current density.
The Sq currents form two vortices, one in the Northern
Hemisphere and the other in the Southern Hemisphere, which
5.03.4 Low- and Midlatitude Currents touch each other at the geomagnetic equator. In accordance
with the day–night contrast in the low- and midlatitude E region
The ions and, to a lesser degree, also the electrons in the iono- conductivities, the Sq currents are concentrated on the dayside.
spheric E region are coupled by collisions to the neutral com-
ponents of the upper atmosphere and follow their dynamics.
Atmospheric winds and tidal oscillations of the atmosphere 5.03.4.2 Equatorial Electrojet
force the E region ion component to move across the magnetic At the geomagnetic equator, the Sq current vortices of the
field lines, while the electrons move much slower at right angles Southern and Northern Hemispheres touch each other and
to both the field and the neutral wind. The relative movement
constitutes an electric current and the separation of charge pro- N
duces an electric field, which in turn affects the current. Because 60°
of this, the E region bears the name dynamo layer, the generator
of which is the atmospheric wind motion. This wind-driven
30°
dynamo causes two current systems in the equatorial and mid-
latitude ionosphere whose ‘external’ magnetic variations alter
the geomagnetic field measured on the Earth’s surface. A third 0°
current system results from electric and magnetic drifts of 06 12 18 LT
30°
60°
S
Figure 9 Dayside view of the Sq current system. Adapted from
Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics.
Figure 8 Magnetic field line reconnection. London: Imperial College Press, with permission.
form an extended nearly jetlike current in the ionosphere, the Integrating over all energies, applying the Biot–Savart law,
equatorial electrojet. However, the electrojet would not be so and then integrating over all L-shells, several symmetries in the
strong if it were formed only by the concentration of the Sq equations lead to the simple expression
current. The special geometry of the magnetic field at the equa-
m0 3UR
tor together with the nearly perpendicular incidence of solar DBd ¼ [8]
4p BE R3E
radiation causes an equatorial enhancement in the effective
conductivity, which leads to an amplification of the jet current. field disturbance at the Earth’s center, where UR is the total
Since the magnetic field lines in the equatorial ionosphere energy of all ring current particles. The minus sign accounts for
are directed northward and parallel to the Earth’s surface, the the fact that the disturbance field of the westward ring current
eastward ionospheric electric field drives an eastward Sq is directed opposite the terrestrial dipole magnetic field.
Pedersen current and a Sq Hall current, which flows vertically The total magnetic field perturbation caused by the ring
downward at the equator. As shown in Figure 10, the latter current must also include the diamagnetic contribution due
causes a charge separation in the equatorial ionosphere with to the cyclotron motion of the ring current particles. Again,
negative charges accumulating on the top boundary and symmetries result in a simple expression:
positive charges accumulating at the bottom of the highly con-
m0 UR
ducting layer. This space charge distribution creates a secondary DBm ¼ [9]
polarization electric field, directed vertically upward. The polar- 4p BE R3E
ization electric field drives a vertical Pedersen current opposing This disturbance adds to the terrestrial dipole field, since
the Hall current until it compensates it. Since the Hall conduc- the Earth’s dipole moment and the magnetic moments of the
tivity is typically about 4 times higher than the Pedersen con- ring current particles are coaligned. The total magnetic field
ductivity, the polarization field must also be 4 times stronger depression caused by the ring current, DBr ¼ DBd + DBm, at the
than the primary electric field. Moreover, the polarization elec- Earth’s center is
tric field generates a secondary Hall current component flowing
eastward, about 16 times stronger than the primary eastward m0 UR
DBR ¼ [10]
Pedersen current, thus explaining the amplification of the equa- 2p BE R3E
torial electrojet current above the equator. This is the famous Dessler–Sckopke–Parker relation, which
The strong horizontal jet current causes a magnetic field directly relates the total energy contained in the ring current to
disturbance that weakens the horizontal component of the the magnetic variation measured on the Earth’s surface.
terrestrial magnetic field at the Earth’s surface over a distance
of about 600 km across the equator (similar to the effect of the
ring current field; see later text). Typical disturbance fields near 5.03.4.4 Storms and Sudden Commencements
the noon magnetic equator are of the order of 50–100 nT.
The ring current and its associated disturbance field is not a
stationary feature. At times, more particles than usual are
injected from the magnetotail into the ring current, mainly
5.03.4.3 Ring Current by an enhanced duskward solar wind electric field induced
The westward drift of trapped ions and the eastward drift of into the magnetotail. This way, the total energy of the ring
trapped electrons around the Earth, depicted in Figure 7, rep- current is increased and the additional depression of the sur-
resent a giant current loop of 1–10 MA that can significantly face magnetic field can clearly be seen in near-equatorial
alter the terrestrial field even at the Earth’s surface. magnetograms, as shown in Figure 11. For about 1 day, the
Applying the magnetic drift velocity given in eqn [3] to the equatorial terrestrial field was depressed by more than 150 nT.
Earth’s dipole field, one can calculate the current density Strong depressions of the terrestrial field, up to 2–3% of the
caused by n particles with energy W circulating around the total surface field in extreme cases, have been noticed in
Earth at a certain radial distance or particular L-shell:
50
3L2 nW
jd ¼ nevB ¼ [7]
Magnetic disturbance (nT)
BE RE
0
where L is measured in RE but is dimensionless, BE is the
equatorial magnetic field on the Earth’s surface, and jd results −50
as an azimuthal current flowing in the westward direction.
−100
East Epolarization
Polarization −150
currents
−200
Down 1 2 3 4 5
Primary
currents Time (days)
B Eprim
Figure 11 Magnetic field variation during a magnetic storm. Adapted
from Baumjohann W and Treumann RA (1996) Basic Space Plasma
Figure 10 Eastward current enhancement at the magnetic equator. Physics. London: Imperial College Press, with permission.
magnetograms long before one knew about the ring current the Earth’s fixed frame of reference will measure a convection
and have been called magnetic storms. electric field:
A magnetic storm has two distinct phases. For some hours
Ec ¼ vc B [11]
or days, an enhanced electric field injects more and more
particles into the inner magnetosphere, building up the strong Hence, the flow of the magnetized solar wind around the
storm-time ring current and the associated magnetic distur- magnetosphere represents an electric field in the Earth’s frame
bance field. After a day or two, the electric field amplitude of reference. Since the solar wind cannot penetrate the magne-
and the rate of injection return to the normal level. The distur- topause, this electric field cannot directly penetrate into the
bance field starts to recover, since the ring current loses more magnetosphere. However, when the interplanetary magnetic
and more storm-time particles. This recovery phase typically field has a southward component, the northward-directed ter-
lasts several days. restrial field lines at the dayside magnetopause can merge with
The depression of the terrestrial dipole field given in eqn [10] the interplanetary magnetic field.
is reflected in the Dst index. This index represents the average As depicted in Figure 12, when a southward-directed inter-
disturbance field at the Earth’s equator and is calculated on the planetary field line encounters the magnetopause, it can merge
basis of hourly averages of the northward horizontal compo- with a closed terrestrial field line, which has both ‘foot points’
nent recorded at four low-latitude observatories – Honolulu, on the Earth. The merged field lines will split into two open
San Juan, Hermanus, and Kakioka. All four observatories are field lines, each of which has one end connected to the Earth
20–30 away from the dipole equator to minimize equatorial and the other stretching out into the solar wind. Subsequently,
electrojet effects and are about evenly distributed in local time the solar wind will transport this field line across the polar cap
(longitude). down the tail, and due to the stiffness of the field line, the
At each observatory, a magnetic perturbation amplitude is magnetic tension, the magnetospheric part of the field line
calculated by subtracting from the hourly averages a quiet time (inside the shaded region), will also be transported tailward.
reference level and the Sq field, both of which vary with local At the nightside end of the magnetosphere, the two open field
time. All four magnetic disturbances are then averaged to fur- line halves will meet again and reconnect, leaving a closed but
ther reduce local time effects and multiplied with the averages stretched terrestrial field line in the magnetotail and an open
of the cosines of the observatories’ dipole latitudes, to obtain solar wind field line downtail of the magnetosphere. Due to
the value of the ring current field at the dipole equator. magnetic tension, the stretched tail field line will relax and
Magnetograms like in Figure 11 often also show a positive shorten in the earthward direction. During this relaxation, it
excursion of the horizontal field magnitude, right at the begin- transports the plasma, to which it is frozen, toward the Earth.
ning of the storm. This excursion is the magnetic signature of For an observer on the Earth, the sunward transport of
the solar wind impinging faster than usual onto the magneto- plasma in the magnetosphere caused by magnetic merging at
pause. The position of the dayside magnetopause is essentially the Earth’s magnetopause is equivalent to an electric field. The
determined as the surface of equilibrium between the magnetic total potential difference between the dawn magnetopause and
pressure of the terrestrial magnetic field and the kinetic energy dusk magnetopause or, equivalently, across the polar cap cor-
of the solar wind. Whenever the speed of the solar wind responds to about 50–100 kV. For a cross section of the mag-
increases, the terrestrial field has to be compressed, and thus, netosphere of about 30 RE, this amounts to a dawn-to-dusk
the magnetopause has to recede to a new equilibrium position. directed field of some 0.2–0.5 mV m1.
If such a sudden compression of the dayside magnetospheric
field occurs at the beginning of a magnetic storm, it is called
storm sudden commencement, whereas when it is not fol- 5.03.5.2 Ionospheric Convection
lowed by a storm, it is called sudden impulse.
The motion of the flux tubes across the polar cap due to
magnetic merging depicted in Figure 12 also moves the iono-
spheric foot point of the flux tube and the plasma tied to it
5.03.5 High-Latitude Currents
Intense ionospheric current systems are also flowing in the

high-latitude ionosphere. However, in this region, the mag-
netic field lines are oriented approximately perpendicular to
the ionospheric layers and the so-called field-aligned currents
connect the ionospheric currents to those flowing in the mag-
netosphere. Hence, the electrodynamics in the high-latitude E
region is coupled and even governed by the dynamics of the
magnetosphere at large.
5.03.5.1 Magnetospheric Convection

The concurrent drift of plasma and field lines as one entity is
called convection. Due to the infinite conductivity, the electric Figure 12 Reconnection and convection cycle in the magnetosphere.
field is zero in the frame of reference moving with the plasma Adapted from Baumjohann W and Treumann RA (1996) Basic Space
at a velocity vc. However, according to eqn [2], an observer in Plasma Physics. London: Imperial College Press, with permission.
across the polar cap to the nightside. Similarly, the sunward It is important to distinguish between the convection auro-
convection of magnetospheric flux tubes leads to a sunward ral electrojets, shown in the left-hand panel of Figure 14, and
convection of the foot points of these flux tubes in the dawn- the substorm electrojet on the right-hand side. The convection
and dusk-side high-latitude ionosphere, inside the auroral electrojets consist of eastward and westward electrojets. These
oval. This leads to a two-cell convection pattern in the polar are primarily Hall currents that originate around noon where
ionosphere, depicted in Figure 13. they are fed by downward field-aligned currents. Typical sheet
The convection pattern is equivalent to an electric potential current densities range between 0.5 and 1 Am1. The eastward
pattern. Cold particles will drift along these contours: drawing electrojet flows in the afternoon sector and terminates in the
equipotential contours and drawing E B drift trajectories of premidnight region where it partially flows up the magnetic
the plasma are equivalent. Hence, we can take the two-cell field lines and partially rotates northward, joining the west-
convection pattern as a two-cell pattern of equipotential con- ward electrojet. The westward electrojet flows through the
tours, which is equivalent to an ionospheric electric field that is morning and midnight sectors and typically extends into the
directed toward dusk in the northern polar cap. Inside the evening sector along the poleward border of the auroral oval
Northern Hemisphere auroral oval, the electric field is directed where it also diverges as upward field-aligned currents.
toward the pole on the dusk side, while it has a southward Similar to the ring current, which is ‘measured’ by the Dst
direction in the morning hours. index (see previous section), the auroral electrojet indices AE,
Since the ionospheric conductivity is a tensor with three AU, and AL were introduced as a measure of global auroral
different components (see previous section), three types of electrojet activity. The indices are based on 1 min samples of
currents will be generated by the convection electric field. The the northward component trace from auroral zone observato-
first type is the field-aligned currents flowing parallel to the ries located at 65–70 magnetic latitude with a longitudinal
magnetic field into and out of the ionosphere. Second, there spacing of 10–40 . Referenced to a quiet-day level, the data of
are the Pedersen currents that flow perpendicular to the mag- all observatories are plotted as a function of universal time. The
netic field lines and parallel to the ionospheric convection upper and lower envelopes are defined as AU and AL, while AE
field. Finally, Hall currents will flow perpendicular to both is defined as the separation between the upper and lower
the magnetic field and the electric field. envelopes. The upper and lower envelopes are thought to
represent the maximum eastward and westward electrojet
currents, respectively, while AE represents the total maximum
5.03.5.3 Auroral Electrojets
electrojet current.
Since particles precipitating into the auroral oval cause signif-
icant ionization, its conductivity is much higher than that of
the polar cap. As a result, the high-latitude current flow is 5.03.5.4 Substorms
concentrated inside the auroral oval, where it forms the auroral Convection is not a stationary process: magnetic merging
electrojets. between interplanetary and terrestrial field lines at the dayside
The auroral electrojets are the most prominent currents at magnetopause does not occur all the time, but only for
auroral latitudes. They carry a total current of some million southward-oriented interplanetary field lines, and is typically
amperes. This is the same order of magnitude as the total current not in equilibrium with reconnection in the magnetotail. Only
carried by the ring current, discussed in the previous section, but part of the flux transported into the tail is reconnected instanta-
since the auroral electrojets flow only 100 km above the Earth’s neously in the deep tail and convected back to the dayside. The
surface, they create the largest ground magnetic disturbance of remaining field lines are added to the tail lobes, where they
all current systems in the Earth’s environment. The disturbance increase the magnetic flux density and, hence, enhance the
fields have typical magnitudes of 100–1000 nT but may reach cross-tail current in the neutral sheet. After some tens of
3000 nT during the largest magnetic storms. minutes, these intermediately stored field lines are suddenly
reconnected at tail distances of 20–25 Earth radii and their mag-
netic energy is explosively released. The sudden reconnection of
Convection electrojets Substorm electrojet

Polar
cap
Dusk Dawn
E
Auroral
oval
Figure 14 Auroral electrojets. The green line and arrows in the
right-hand panel indicate the boundary of the westward- and
Figure 13 Convection and electric field in the high-latitude ionosphere. northward-expanding substorm auroral bulge. Adapted from
Adapted from Baumjohann W and Treumann RA (1996) Basic Space Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics.
Plasma Physics. London: Imperial College Press, with permission. London: Imperial College Press, with permission.
previously stored flux tubes has rather dramatic effects on the range. In the case of the substorm electrojet, the jet itself and
magnetospheric plasma and associated phenomena like aurora its field-aligned currents are much more concentrated in the
and magnetospheric and ionospheric currents. These effects, midnight sector, forming a current wedge as depicted in
which last for 1–2 h, are summarized as magnetospheric Figure 16. The effects of this current wedge, in particular the
substorm. magnetic disturbance associated with the field-aligned cur-
A substorm starts when the dayside merging rate is distinc- rents, can be seen also at midlatitudes.
tively enhanced, typically due to a southward turning of the
interplanetary magnetic field. The flux eroded on the dayside
magnetopause is transported into the tail. Part of the flux is
reconnected and convected back to the dayside magnetosphere. 5.03.6 Geomagnetic Pulsations
The enhanced convection causes enhanced current flow in the
convection electrojets and an associated growth of the AE index. As with any medium, a plasma carries waves in many different
The other part of the flux is added to the tail lobes. After frequency ranges, from as low as millihertz to as high as several
30–60 min, too much magnetic flux and thus magnetic energy tens of kilohertz. Typically, the higher-frequency waves can
have been accumulated in the tail. The tail becomes unstable only be observed in the plasma itself, but the ultralow-
and must release the surplus energy. This is the time of sub- frequency (ULF) waves, in particular, generate fast fluctuations
storm onset and the beginning of the substorm expansion of the Earth’s surface magnetic field in the frequency range
phase. At substorm onset, the aurora suddenly brightens and from a few millihertz up to a few hertz, corresponding to
fills the whole sky. During the following 30–60 min, rather oscillation periods from several hundred seconds to a fraction
dramatic changes are seen in the auroral zone currents. of a second. These are the so-called geomagnetic pulsations,
The sharp AE index seen in Figure 15 to values of about known of for about a century.
500 nT indicates that the ionospheric current flow is strongly In most cases, the pulsating disturbance fields observed are
enhanced. The unloading of magnetic flux previously stored in associated with shear Alfvén waves. These waves constitute the
the magnetotail leads to the formation of a substorm electrojet simplest wave solutions of the magnetohydrodynamic equa-
with strongly enhanced westward current flow in the midnight tions and represent simple stringlike oscillations of mass-
sector. The substorm electrojet is concentrated in the region of loaded magnetic field lines. Shear Alfvén waves are purely
active aurora and expands westward during the course of the transverse waves, that is, all variations have only components
expansion phase. In contrast to the convection electrojets, that are perpendicular to the ambient magnetic field. The
where any increase is caused mainly by an increasing convec- magnetic component of this type of wave is parallel to the
tion electric field, the strength of the substorm electrojet cur- plasma velocity variation while the wave electric field points
rent is mainly determined by a strong increase in ionospheric are perpendicular to the magnetic and velocity variations.
conductance due to strong particle precipitation in the bright An Alfvén wave may propagate parallel to the ambient field
substorm aurora. with the Alfvén velocity, vA, which is essentially a ‘magnetic
Since the substorm electrojet is governed by the strong sound’ velocity, given by
increase of the conductivities inside the region of bright aurora, B2
the situation is similar (except for directions) to that in the v2A ¼ [12]
m0 nm
equatorial electrojet described in the previous section. How-
ever, in the present case, the polarization electric field and thus In the Earth’s magnetosphere, typical Alfvén velocities
the enhancement of the westward current is not so strong, range from some hundreds to several thousands of kilometers
since field-aligned currents will remove part of the space charge per second.
deposited at the boundaries of the highly conducting channel. The ULF range and, hence, the pulsations are convention-
Another difference between the convection and substorm ally divided into five intervals, Pc1–Pc5, for continuous
electrojets is that in the case of the convection electrojets, the
field-aligned currents are distributed over a wide local time Field-aligned current
500
400
AE index (nT)
Onset
300
200
Neutral sheet
current
−45 0 45 90
Figure 16 Substorm current wedge. Adapted from Baumjohann W and
Minutes around substorm expansion onset
Treumann RA (1996) Basic Space Plasma Physics. London: Imperial
Figure 15 Variation of AE index during a substorm. College Press, with permission.
pulsations, and into two intervals, Pi1 and Pi2, for irregular Kelvin–Helmholtz instability excited by the flow of the solar
pulsations. The class of continuous pulsations covers quasi- wind around the magnetosphere and represent an evanescent
sinusoidal oscillations of narrow spectral bandwidth, as shown wave mode. Being surface waves, their amplitude is strongly
in Figure 17. They may have a comparably long duration from damped away from the magnetopause, yet they can still set the
several minutes up to hours. Pc pulsations can generally be field line with a matching eigenfrequency or resonance fre-
observed over a wide latitudinal and longitudinal range on the quency into oscillation. All other field lines, whose resonance
Earth’s surface and in the magnetosphere. The irregular pulsa- frequencies do not match, are only marginally excited and do
tions, in contrast, are shorter-lived, sometimes comprising not contribute to the pulsation.
only a few oscillations decaying in time. Pc4–Pc5 pulsations might also be excited by packets of
trapped particles bouncing up and down a field line, as long
as the bounce period of these particles, which depends mainly
5.03.6.1 Pc5 Pulsations on their energy, matches the eigenperiod of the field line.
The Pc5 pulsations shown in Figure 17 are caused by standing

oscillations of magnetospheric field lines. For a standing wave, 5.03.6.2 Pi2 Pulsations
the length of the field line between the two reflection points in The short-period irregular Pi2 pulsations are associated with
the ionosphere must be a multiple of half the parallel wave- the development of the substorm current wedge described in
length. Hence, each particular field line has a number of dis- the previous section. Whenever field-aligned currents are sud-
tinct eigenfrequencies or Alfvénic resonances. Since the length denly switched on somewhere in the magnetosphere, they
of the field lines increases with latitude, the resonance or must be transported to the ionosphere via Alfvén waves. Only
eigenfrequency decreases with latitude. For an average Alfvén this transverse magnetohydrodynamic wave mode can carry
velocity of 1000 km s1, the fundamental resonance frequency field-aligned current. Launched in the magnetosphere, the
on closed field lines ranges between 1 and 100 MHz and falls Alfvén waves are reflected back and forth between the iono-
into the Pc3–Pc5 range. sphere and the current generator in the tail until a stationary
Figure 18 schematically shows how the dipolar field con- equilibrium is reached.
figuration changes for two fundamental types of field line Figure 19 shows qualitatively the development of the mag-
resonances. The foot points of the field lines are fixed in the netic disturbance field and thus the field-aligned current flow
ionosphere, but the field lines may perform either a ‘breathing’ after switch on of a current generator in the magnetotail. At
motion (fundamental odd mode) or a ‘wobbling’ motion t ¼ 0, an Alfvén wave is launched that carries a current corre-
(even mode). In addition to these poloidal modes, the field sponding to the generator current and thus a particular mag-
lines can also exert toroidal oscillations, in which case the netic disturbance field. This wave reaches the ionosphere at
oscillation of the field line and the plasma bulk flow are purely t ¼ tA, the Alfvén wave travel time between magnetosphere and
azimuthal. Pc pulsations often are a mixture of poloidal and ionosphere, that is, some 30 s. Here, about 80% of its ampli-
toroidal oscillations. tude is reflected with the magnetic field of the reflected wave
Pc5 pulsations are resonant oscillations of field lines. They adding to the primary disturbance field.
are excited and driven via surface waves traveling along At t ¼ 2tA, the reflected wave comes back to the generator
the magnetopause. These surface waves are caused by a and launches a third wave, whose magnetic disturbance must
be of opposite polarity to decrease the total wave magnetic
field such that the total disturbance matches that caused by
the generator current at this time. Multiple bounces of the wave
lead to magnetic field disturbances that oscillate with a period
of 4tA until they finally converge to match the generator cur-
20 nT rent. The magnetic oscillations with periods of some 100 s are
readily observable as Pi2 pulsations. They are often used as a
0 10 20 30 40 50 60
good indicator for the onset of substorms.
Time (min)
Figure 17 Ground magnetic disturbance of a Pc5 pulsation.
5.03.7 Conclusions
N
This chapter gave a brief introduction into the fundamentals of
magnetospheric physics. The topics and phenomena discussed
were selected by their influence on the terrestrial field as mea-
Odd mode Even mode sured on the Earth’s surface. There are many more interesting
features that can be observed in magnetospheric plasmas,
let alone in other space plasmas, like solar or astrophysical
plasmas.
S However, all of them follow the same basic principles out-
Figure 18 Fundamental poloidal field line resonances. Adapted from lined in the preceding text. If one includes the Earth’s bow
Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics. shock (which was not discussed here, since its physics has little
London: Imperial College Press, with permission. influence on the terrestrial magnetic field), detailed in situ
is detailed in Hargreaves (1992) and Rishbeth and Garriot

(1969). A good description of solar wind–magnetosphere cou-
pling is given in a review article by Cowley (1982). Additional
material on high-latitude current systems and magnetosphere–
Generator current
ionosphere coupling is found in a monograph by Kamide and
Baumjohann (1993) and in a review article by Untiedt and
Baumjohann (1993). For a detailed account on substorms, see
Baker et al. (1996). An exhaustive elementary description of
fluid plasma waves is given in Bittencourt (1986) while further
information about pulsations is contained in Glassmeier (1995)
Magnetic disturbance
and McPherron (2005).
0 10 20 30 t/tA
References
Figure 19 Magnetic disturbance due to switch on of a current wedge.
Adapted from Baumjohann W and Treumann RA (1996) Basic Space Alfvén H and Fälthammar CG (1963) Cosmical Electrodynamics, Fundamental
Principles. Oxford: Clarendon Press.
Plasma Physics. London: Imperial College Press, with permission.
Baker DN, Pulkkinen TI, Angelopoulos V, Baumjohann W, and McPherron RL (1996)
Neutral line model of substorms: Past results and present view. Journal of
studies of the space plasmas in the Earth’s neighborhood are an Geophysical Research 101: 12975–13010.
effective means to understand many solar and astrophysical Baumjohann W and Treumann RA (1996) Basic Space Plasma Physics. London:
Imperial College Press.
phenomena, from which we have only sparse observational Bittencourt JA (1986) Fundamentals of Plasma Physics. Oxford: Pergamon Press.
information transmitted by electromagnetic radiation. Cowley SWH (1982) The causes of convection in the Earth’s magnetosphere: A review of
Hence, geomagnetism not only is the root of magneto- developments during the IMS. Reviews of Geophysics and Space Physics 30: 531.
spheric physics (ground-based observations of magnetic varia- Daglis IA, Thorne RM, Baumjohann W, and Orsini S (1999) The terrestrial ring current:
Origin, formation, evolution, and decay. Reviews of Geophysics 37: 407–438.
tions done by Humboldt and Gauss provided the first window
Glassmeier KH (1995) ULF pulsations. In: Volland H (ed.). Handbook of Atmospheric
into what was later called magnetosphere) but also, in a Electrodynamics, vol. 2, p. 463. Boca Raton, FL: CRC Press.
broader sense, that is, by measuring magnetic fields in the Hargreaves JK (1992) The Solar-Terrestrial Environment. Cambridge: Cambridge
Earth’s neighborhood, is still essential to understand the University Press.
plasma universe. Kamide Y and Baumjohann W (1993) Magnetosphere–Ionosphere Coupling.
Heidelberg: Springer.
A more exhaustive description of all possible external sources Kamide Y, Baumjohann W, Daglis IA, et al. (1998) Current understanding of magnetic
of geomagnetic field variations (and on their use in diagnosing storms: Storm–substorm relationship. Journal of Geophysical Research
magnetospheric dynamics) is presented in Nishida (1978). 103: 17705–17728.
A full description of theory and observations of space plasmas Lyons LR and Williams DJ (1984) Quantitative Aspects of Magnetospheric Physics.
Dordrecht: D. Reidel Publication Company.
in the Earth’s neighborhood can be found in Baumjohann and
McPherron R (2005) Magnetic pulsations: Their sources and relation to solar wind and
Treumann (1996). For those readers who want to know more geomagnetic activity. Surveys in Geophysics 26: 545–592.
about the guiding-center approach, we recommend reading the Nishida A (1978) Geomagnetic Diagnosis of the Magnetosphere. Heidelberg: Springer.
monograph by Alfvén and Fälthammar (1963) or that by Northrop TG (1963) The Adiabatic Motion of Charged Particles. New York: Interscience
Northrop (1963). More about the physics of trapped particles, Publishers.
Rishbeth JA and Garriot OK (1969) Introduction to Ionospheric Physics. New York:
the ring current, and magnetic storms can be found in Lyons Academic Press.
and Williams (1984), Kamide et al. (1998), and Daglis et al. Untiedt J and Baumjohann W (1993) Studies of polar current systems using the IMS
(1999). The physics of the ionosphere and ionospheric currents Scandinavian magnetometer array. Space Science Reviews 63: 245.
5.04 Observation and Measurement Techniques
GM Turner, Victoria University of Wellington, Wellington, New Zealand
J Rasson, Centre de Physique du Globe, Dourbes, Belgium
C Reeves, Earthworks, Delft, The Netherlands
5.04.2 Instrumentation 93
5.04.2.1 Vector Magnetometer Sensors 93
5.04.2.1.1 The induction coil 93
5.04.2.1.2 The fluxgate 94
5.04.2.1.3 The SQUID 95
5.04.2.2 Scalar Magnetometers 96
5.04.2.2.1 Proton precession magnetometers 96
5.04.2.2.2 Overhauser effect proton magnetometers 97
5.04.2.2.3 Optically pumped magnetometers 97
5.04.3 Magnetic Observatories 99
5.04.3.1 Special Magnetic Conditions in an Observatory 99
5.04.3.2 Geomagnetic Observations Using the Baseline Concept 100
5.04.3.2.1 Absolute measurements of field components 100
5.04.3.2.2 Measuring the angular orientation of the geomagnetic field 101
5.04.3.2.3 Variation measurements: Variometers 103
5.04.3.2.4 Instrument certification and calibration 103
5.04.3.2.5 Obtaining definitive absolute data 104
5.04.3.3 The INTERMAGNET Magnetic Observatory Network 105
5.04.3.4 Fully Automatic Magnetic Observatories 106
5.04.3.5 Magnetic Repeat Station Surveys 106
5.04.3.6 Products and Services Magnetic Observatories Can Provide 107
5.04.4 Magnetic Surveys for Geological Exploration 107
5.04.4.1 Mapping Magnetic Anomalies 108
5.04.4.2 The Origin of Magnetic Anomalies 108
5.04.4.3 Instrumentation Applied to Magnetic Surveying 111
5.04.4.3.1 Ground surveys 111
5.04.4.3.2 Airborne survey techniques 111
5.04.4.4 Design and Execution of Surveys 112
5.04.4.4.1 Practical airborne magnetometry 112
5.04.4.4.2 Elimination of nongeological influences (aircraft and temporal changes) 113
5.04.4.5 Data Presentation, Enhancement, and Interpretation Methods 114
5.04.4.6 The Link Between Magnetic Properties of Rocks and Regional Geology 116
5.04.4.7 Satellites and Space Probes 117
5.04.5 Paleomagnetic Methods 118
5.04.5.1 Sampling 119
5.04.5.1.1 Introduction 119
5.04.5.1.2 Consolidated rocks 119
5.04.5.1.3 Lake and deep-sea sediments 120
5.04.5.1.4 Archaeological materials 121
5.04.5.2 Rock Magnetometers 122
5.04.5.2.2 Astatic and parastatic systems 122
5.04.5.2.3 Spinner magnetometers 122
5.04.5.2.4 Cryogenic magnetometers 123
5.04.5.3 Progressive Demagnetization Techniques 125
5.04.5.3.2 Thermal demagnetization 125
5.04.5.3.3 Alternating field demagnetization 126
5.04.5.3.4 Paleointensity determination 127

92 Observation and Measurement Techniques
5.04.5.4 Data Analysis and Statistics 127

5.04.5.4.1 Progressive demagnetization data 127
5.04.5.4.2 The statistical treatment of paleomagnetic data 128
5.04.5.4.3 Field tests 130
5.04.5.4.4 Poles 131
References 133
5.04.1 Introduction In order to describe the (instantaneous) geomagnetic field

vector completely, three independent components must be
The geomagnetic field is a complex function of space and time, specified. The three coordinate systems commonly used are
with contributions of both internal and external origin. His- as follows:
torically, knowledge of the morphology and variability of the
field has been important in navigation, in understanding • Cartesian: X (north component), Y (east component), and
Z (vertical component, positive downward)
atmospheric and ionospheric processes, including radio trans-
mission, and in geophysical studies of the Earth’s interior: the • Cylindrical: D (magnetic declination angle, positive east of
north), H (horizontal component), and Z (as above)
core, mantle, and past movements of the tectonic plates that
make up the lithosphere. The purpose of this chapter is to • Spherical: D (as above), I (magnetic inclination angle, pos-
itive below horizontal), and F (intensity of magnetic field)
describe the instruments and practices used to observe and
measure the full range of features of the geomagnetic field. The relations between these different systems are illustrated
The significance of the predominantly dipole nature of the in Figure 2.
main field was recognized by Gilbert over 400 years ago;
secular variation was documented at about the same time.
The nondipole part of the field that originates from the core
manifests itself at the surface as features that typically extend
over thousands of kilometers. The International Geomagnetic 1011 years
Reference Field (IGRF, website) is a spherical harmonic model Age of the Universe
designed to accurately describe the spatial features of the field 1010
originating from the core. Much of the data incorporated into Age of Earth
the IGRF come from the global network of permanent geomag- 109
netic observatories (INTERMAGNET, website).
By contrast, the magnetization of crustal bodies, whether 108
induced by the present-day field or a remanent magnetization Superchrons
107
from an earlier geological time, typically results in localized
magnetic ‘anomalies’ from a few meters to several tens of 106 Polarity reversals
kilometers in extent. These are generally charted and studied
Internal origin
through regional or local magnetic surveys. 105
In addition, the field has been observed to undergo a very 104
broad spectrum of time variations, extending well beyond the
limit of real-time and historical observations (Figure 1). Perhaps 103 Secular variation
the most astounding of these is the phenomenon of polarity
reversals found in the paleomagnetic record and in magnetic 102
anomalies over the ocean floor. Time-varying signals reaching
101 Sunspot cycle
the Earth’s surface from the core are attenuated due to the
conductivity and permeability of the mantle, so that periodicities 1 year Annual variation
less than about 1 year cannot be detected. The shorter-period
External origin
magnetic field variations that are observed originate externally: 1 month

from variations in the solar wind and the interplanetary mag-
netic field and their interaction with the magnetosphere. 1 day Daily variation, magnetic storms
Although the majority of measurements of the geomagnetic
field are made on or near the Earth’s surface (e.g., in low-flying 1h
aircraft), magnetometers are also routinely installed in space
probes and satellites, giving worldwide spatial coverage but 1s Pulsations, whistlers, noise
with loss of resolution of the most closely spaced sources of
geological interest. Modern models of the main field incorpo-
rate satellite measurements, while aeromagnetic surveys make a Figure 1 Time variations of the geomagnetic field in relation to the age
huge contribution to modern geological mapping and resource of the Earth and the universe, indicating which are of internal and
exploration. which are of external origin.
Observation and Measurement Techniques 93
X Humboldt, made between 1798 and 1803, clearly show the

increase of intensity with latitude, which is expected of a pre-
H dominantly geocentric axial dipole (GAD) field.
D Many of the first generation of magnetic field measuring
Y instruments were based on pivoted or suspended magnet sys-
I tems. These include the quartz horizontal magnetometer
(QHM) and the balance magnétique zero (BMZ) instruments
used extensively in observatories, the astatic and parastatic rock
magnetometers used in many pioneering paleomagnetic studies,
the variometer-type instruments used in early magnetic surveys
F and exploration work, and the Gough–Reitzel magnetometer
used until recently in many early magnetometer array studies.
From the 1950s onward, these elegant but delicate instru-
ments have gradually been replaced by more robust electric and
electronic instruments such as the induction coil and the flux-
gate, which capitalizes on the high-permeability materials that
were developed around that time. Electric feedback systems have
played a major role in stabilizing the response of these instru-
ments. Later still, computer interfacing and data logging have
led to partial or complete automation of many operations.
Z The most recent generation of magnetometers has grown
out of research on magnetic properties at the atomic level and
Figure 2 Decomposition of the geomagnetic vector, showing
the development of superconducting materials and devices,
components used when describing the vector in Cartesian (X, Y, Z),
cylindrical (D, H, Z), and spherical (D, I, Z) coordinate system. which are, in principle, capable of counting individual quanta
of magnetic flux and thus reaching the theoretical limit of
resolution. These include the proton precession magnetometer
5.04.2 Instrumentation (PPM) and the Overhauser effect magnetometer, which are
used extensively in observatories, and the superconducting
The earliest known compass, made in China around the first quantum interference device (SQUID), which has revolution-
century AD, was a spoon-shaped piece of lodestone, balanced ized paleomagnetic measurements.
on a horizontal plate so that it aligned with the Earth’s mag- In this section, we describe the principles of operation of
netic field, the handle pointing south. Although originally the main types of sensors used in geomagnetism and paleo-
designed for the purpose of divining favorable orientations magnetism. Their incorporation into practical magnetometers
(feng shui), these early instruments were the forerunners of an is covered in the following sections on geomagnetic observa-
extensive range of measuring instruments and navigational tories, magnetic surveys, and paleomagnetism.
compasses based on pivoted or suspended magnets. The induction coil, fluxgate, and SQUID are vector sensors:
The magnetic compass is simply a needle balanced to rotate their orientation in space determines which component of the
in a horizontal plane until it comes to rest in the plane of the ambient field is measured. The proton precession, Overhauser,
magnetic meridian (with north, rather than south, now being and optically pumped magnetometers (OPMs), on the other
the principal sense). It was the first scientific instrument to hand, are insensitive to the direction of the ambient magnetic
employ a needle moving over a graduated scale, and though field and measure only its magnitude. These differences stem
most have by now been superseded by digital meters, the from the fundamental physical principles put to work in their
compass remains the simplest and easiest means of direction operation.
finding in remote locations, since it has no electronics and
needs no power supply.
Very early compass makers appear to have been unaware of
5.04.2.1 Vector Magnetometer Sensors
the vertical component of the geomagnetic field. However, by the
mid-sixteenth century, Hartmann and Norman had both noted 5.04.2.1.1 The induction coil
and measured the angle of inclination, using a magnetized needle The induction coil, or search coil magnetometer, is one of the
pivoted to rotate in the vertical north–south plane: a dip needle. simplest modern magnetometers. In principle, when the mag-
The dip needle was also used to make the first measure- netic flux, f, threading through a conducting circuit varies, an
ments of the intensity of the magnetic field. Small oscillations emf or voltage signal is induced in the circuit, in accordance
of the needle about its equilibrium position have a period
pffiffiffiffiffiffiffiffiffiffiffi with Faraday’s law of electromagnetic induction. The induced
given by T ¼ 2p I=mB, where I and m are the moment of voltage, Vi, is proportional to the rate of change of the
inertia and the magnetic moment of the needle, respectively, magnetic flux.
and B is the magnetic field. The first recorded intensity mea- In practice, the flux linkage is enhanced by winding many
surements were made by de Rossel on the d’Entrecasteaux turns of conducting wire on a core of high-permeability ferro-
expedition of 1791–94 (Lilley and Day, 1993). Both the magnetic material. For a solenoid of N turns, each of cross-
d’Entrecasteaux measurements and those of Alexander von sectional area A, and a core of relative permeability m,
f ¼ mNAB, where B is the component of the magnetic field fluxgate axis. A phase-sensitive detector, with its reference sig-
parallel to the axis of the solenoid. nal set to twice the frequency of the driving signal, is generally
For magnetic field variations of frequency o, along the axis, used to produce a DC output that is proportional to the ambi-
B ¼ B0eiot, and the theoretical induced voltage is ent field.
In practical applications, a feedback arrangement is almost
df
Vi ¼ ¼ NAmðioÞB0 eiot always employed to cancel the field being measured and,
dt
hence, linearize the response. Fluxgate sensors are highly direc-
This signal must be amplified, and the actual measured tional, measuring only the field component parallel to the
voltage depends also on the resistance, inductance and capac- fluxgate axis. An arrangement of three orthogonal fluxgates
itance of the coil as well as the size of any damping resistor and may therefore be used to obtain the intensity and direction of
the gain of the amplifier. the total field vector. If three independent units are used in
Due to their simplicity, induction coils are widely used for close proximity, for example, in a spacecraft, problems can
many applications, for instance, in geophysical prospecting arise due to the interference of the separate feedback systems.
and metrology. In magnetic observatories, they are used to This has been overcome in the compact spherical coil system,
monitor rapid variations of the geomagnetic field, such as as used in the ersted and CHAMP satellites, where all three
pulsations. With suitable circuitry, induction coils can be sensors see the whole feedback field (Nielsen et al., 1995).
designed to measure variations with frequencies between The first fluxgate-type device was built and described by
104 and 107 Hz and intensities from fractions of a femtotesla Aschenbrenner and Goubau (1936). Airborne fluxgates were
to tens of teslas (Korepanov et al., 2001). used extensively during the Second World War for submarine
detection. Subsequent developments mean that the modern
fluxgate has many applications: they are used for navigation
5.04.2.1.2 The fluxgate on land and sea and as vector magnetometers in almost all
The operation of a fluxgate sensor depends on the nonlinear space flights, in observatories, and in many medium-
relation between induced magnetization and magnetizing field resolution rock magnetometers.
in high-permeability, easily saturated materials such as ferrites,
permalloy, metallic glasses, and mu-metal.
Various different designs exist: most have two cores, either
as two separate linear cores or as two halves of a ring fluxgate. Coil/core 1 Coil/core 2
The general principle of operation is described below for
the case of the two-core Vacquier design shown in Figure 3.
Identical primary coils are wound around the two high-perme-
ability cores. The drive current (of frequency 50–1000 Hz) is of
sufficient amplitude to saturate the cores during most of each (a)
cycle. The primary windings are connected in series opposi-
tion, so when the drive current is applied, the axial magnetic
fields are out of phase (Figure 4(a)). If there is a steady ambi-
ent magnetic field, Bamb, parallel to the core axis, as shown in (b)
Figure 3, then when it augments the field in one coil, it reduces
the total field in the other, and vice versa (Figure 4(b)). The
effect of the ambient field is thus that, during any given half-
cycle, the core in which the field is augmented will be saturated
(c)
for more of the half-cycle, while the core with the reduced field
will be saturated for less of the half-cycle. The situation reverses
in the next half-cycle of the driving signal (Figure 4(c)). This
asymmetry between the magnetization of the two cores leads
directly to an asymmetry between the rates of change of flux in (d)
the secondary coils Figure 4(d)). The secondary coils are oppo-
sitely wound and series-connected as shown (Figure 3), so the
asymmetry results in a nonzero output twice per cycle, that is,
at twice the frequency of the driving signal (Figure 4(e)). For
(e)
small fields, the amplitude of this second harmonic is propor-
tional to the component of the ambient field parallel to the Figure 4 Signals involved in the operation of a fluxgate magnetometer.
(a) Sinusoidal drive current, frequency o. In the absence of the iron
cores, the drive signal results in antiphase magnetic fields in coils 1 and
Bamb
2; (b) the addition of the ambient field Bamb, as shown in Figure 3, results
Drive Coil/core 1 in asymmetry; (c) so core 1 is saturated for more of the first half-cycle
current than core 2, while core 2 is saturated for more of the second half-cycle
w Output signal 2w than core 1; (d) the magnetic flux changes in secondary coils 1 and 2 are
Coil/core 2 therefore not synchronous, resulting in induced emf pulses as shown; (e)
the output signal is the sum of the emf’s in secondary coils 1 and 2, which
Figure 3 A two-core fluxgate magnetometer (Vacquier design). has a dominant frequency of 2o.
5.04.2.1.3 The SQUID The first superconducting rock magnetometers used radio
The development of the SQUID as a magnetic field sensor frequency (RF)-driven SQUIDs (Goree and Fuller, 1976).
heralded a new era in sensitivity, particularly in paleomagnetic These are not strictly interference devices, but rather single
rock magnetometers. weak links contained in a ring or cylinder of superconducting
The superconductivity of mercury was discovered by material. The operation of a device similar to the Dayhem
Kamerlingh Onnes in 1911, when he found that, below bridge used in the first superconducting rock magnetometers
about 4 K, its electric resistivity effectively vanished. Other built by Superconducting Technologies Inc. (SCT) is illustrated
metal superconductors, more suitable for the construction of in Figure 5.
practical devices, include niobium (TC 9.2 K) and lead (TC The sensor is driven by an RF signal of frequency
7.2 K). More recently, superconducting ceramic materials, for 20–30 MHz, through a coil wound around the outside. The
example, oxides of lanthanum or yttrium, with barium, amplitude of the RF signal is sufficient to exceed the critical
strontium, or calcium and copper, have been discovered with current of the weak link. When this occurs, the loop momen-
transition temperatures of 90 K and higher. However, practical tarily becomes resistive and a flux quantum is admitted before
sensors utilizing these ‘high-temperature’ superconductors superconductivity is regained, as described in the text earlier.
have not yet been developed. This produces a pulse in the sensor coil. Figure 5(a) illustrates
Superconductivity may be explained by the BCS (Bardeen, the situation when the critical current corresponds to 0.75 flux
Cooper, and Schrieffer) theory (Bardeen et al., 1957). At suffi-
ciently low temperatures, a weak interaction causes electrons to
No bias field
form pairs. Such ‘Cooper pairs’ have boson-like properties: in
1.5
particular, they occupy a single quantum state and may be fo Drive signal
described by a single coherent wave function. The pairing of
0.75
electrons is critical to the theory. In the normal state, electric
resistivity results from collisions or scattering of conduction
electrons by the ions of the crystal lattice. In the superconduct- 0
ing state, one electron of the Cooper pair effectively distorts the
lattice, allowing the other free passage. -0.75
Another property of the superconducting state is the exclu- SQUID current
sion of magnetic flux from the bulk of the superconducting (a) -1.5
material (Meissner effect). If an external magnetic field exists 1.00 Sensor coil output
through a ring of superconductor, then, as it is cooled through
the critical temperature, superconducting currents will be gener-
ated to cancel it. However, if the current exceeds a certain value, 0.00
Ic, the superconductivity is lost, and the material reverts to its
normal resistivity. The moment this happens, the current drops
and flux enters the ring. With the lowering of the current, (b) -1.00
superconductivity is restored. However, to maintain the coher-
With bias field
ency of the electron-pair wave function, the current can drop
1.5
only by quantized amounts, and correspondingly, only quan-
fo Drive signal plus bias
tized amounts of flux may enter the ring. This flux quantum, fo,
0.75
is equal to h/2e ¼ 2.09 1015 Wb. In principle, the number of
flux jumps may be counted to give a measure of the strength of
0
the external magnetic field. This effect is utilized in the SQUID.
Josephson (1962) predicted that a supercurrent should be
able to pass through a thin layer of dielectric material sand- -0.75
wiched between two superconductors, if the wave functions of SQUID current
the electrons extend into the junction and join up coherently. (c) -1.5
A similar effect is achieved by creating a very narrow bridge or 1.00

Sensor coil output
‘weak link’ between two pieces of superconductor.
The term SQUID has been applied to two types of device
that have evolved from the basic Josephson junction. The 0.00
simpler concept is that of the DC SQUID. It consists of two
Josephson junctions or weak links arranged symmetrically in a
ring of superconductor. Magnetic flux through the ring will (d) -1.00
result in equal and opposite phase shifts of the wave functions
Figure 5 Signal processing by a weak-link SQUID sensor, without and
across the two junctions. Interference therefore occurs between with a direct biasing field. (a) Sinusoidal drive signal that causes the
the currents in the two branches of the device, in a manner supercurrent to exceed the critical value at 0.75 fo, at which point the
analogous to the interference of light waves in a double-slit flux through the SQUID jumps by fo. The bold curve shows the resulting
experiment. The current is maximum if the flux is equal to an SQUID current. (b) At each flux jump, a pulse is induced in the sensor
integral number of flux quanta but is reduced to zero for an coil. (c) and (d) correspond to the curves for the case of a direct bias field
odd number of half quanta. and show that the pulses in the sensor coil are offset by the direct field.
quantum: with no bias field, the output consists of two positive Hence, a measurement of the Larmor frequency, and division
pulses and two negative pulses per cycle. The addition of an by the proton–gyromagnetic ratio, leads directly to the inten-
external magnetic field biases the RF signal and changes the sity of the ambient magnetic field. In practice, the ‘shielded’
positions in the cycle at which the critical current is exceeded. proton–gyromagnetic ratio, g0 , is used instead of g. This takes
Figure 5(b) shows that the positions of the pulses have been account of the diamagnetism of water (or other proton-rich
shifted, and from these shifts, the direct field can be obtained. fluid) and the shape of the sample (usually taken to be spher-
A common arrangement to avoid problems of nonlinearity ical). For a spherical sample of water at 25 C (CODATA 2010,
and drift is to apply negative feedback, so the sensor is always NIST website),
at its most sensitive, and to monitor the feedback signal needed
to achieve this. g0 ¼ 2:675153268ð66Þ 108 T 1 s1
A PPM sensor is therefore ideally a spherical container

5.04.2.2 Scalar Magnetometers filled with water, or other proton-rich fluid. A surrounding
coil supplied with a direct current produces a direct field
5.04.2.2.1 Proton precession magnetometers
several orders of magnitude larger than the Earth’s field and
Thanks to the groundbreaking work on nuclear magnetic
aligns the magnetic moments of the protons. This is to
induction by Bloch (1946) and its application to the measure-
partially remove the random orientation due to the thermal
ment of weak magnetic fields by Packard and Varian (1954),
agitation of the protons. After the few seconds necessary to
the PPM became available from the late 1940s. It allows the
obtain sufficient alignment, this direct field is removed, and
absolute measurement of the scalar magnetic field at the push
the proton magnetic moments precess about the ambient
of a button. In the PPM, the magnetic field is found by mea-
(geomagnetic) field vector. An emf induced by the precessing
suring the frequency of the Larmor precession of the proton
protons is then observed for a few seconds by a pickup coil,
magnetic moment about the ambient field, the protons being
until thermal agitation drowns the signal again (Primdahl,
provided by a sample of water or other proton-rich fluid.
2002). The observed signal is an exponentially decaying sinu-
The spin angular momentum and magnetic moment vec-
soid at the Larmor frequency f. The emf signal in the geomag-
tors of the proton, L and mp, are parallel. Their ratio is a
netic field ranges from 850 to 3400 Hz and its amplitude is
scalar constant known as the proton–gyromagnetic ratio, gp
typically of a few microvolts. The signal-to-noise ratio is
(CODATA 2010, NIST, website).
dependent upon the strength of the field; therefore, a PPM
mp
gp ¼ ¼ 2:675222005ð63Þ 108 T1 s1 works better in strong fields. Often, the same coil is used for
L the polarizing and pickup operations. For maximum signal,
In an external magnetic field, B, the magnetic moment experi- the polarizing direct field should be applied at right angles to
ences a torque, and this results in a change in the angular the field to be measured, in order to maximize the rotating
momentum vector: component and the emf pickup by the coil. Cylindrical coils
will therefore have dead zones for certain orientations of the
dL
t ¼ mp B ¼ sensor, where the emf signals disappear in the noise. Toroidal
dt
coils have been used to eliminate this spatial dependency on
L is parallel to m, while the torque is perpendicular to m, so dL the signal-to-noise ratio.
is perpendicular to L, and the effect is to make L and m precess Efforts to improve the PPM have also addressed the way in
about the direction of B (Figure 6). This is Larmor precession, which the frequency measurement is performed. In early
and its frequency is given by PPMs, the frequency, f, of the output signal was multiplied by
an integer M using a phase-locked loop. The resulting signal at
oLarmor ¼ 2pf ¼ gp B
frequency Mf was then counted by a digital counter for a time
duration D. D was set by a custom-cut quartz crystal, serving as
a frequency reference, and whose frequency F was adjusted to
provide the correct scaling according to the value of gp0 . For
B instance, the Geometrics G816 PPM used M ¼ 64 and
F ¼ 2.857252 MHz. The interval D spanned 220 cycles of F,
L sin q
df about 0.37 s. The final count C was then equal to the magnetic
dL field, B, in units of nT:
L
64 220 2p
C¼ f ¼ 23:4872 f ¼ 0 f 109 nT
2857252 gp
Nowadays, PPM frequency-measuring schemes are assisted

by microcomputers and digital signal processors. This comput-
q
ing power allows the use of every part of the proton precession
signal, to adopt noise-canceling techniques and to extract qual-
Figure 6 Larmor precession of the proton angular momentum vector, itative information on the measurement ( Jankowski and
L, about the ambient magnetic field vector, B, in response to the Sucksdorff, 1996; Primdahl, 2002; Sapunov et al., 2001). Elab-
torque, m B, on the magnetic moment vector of the proton, m, which orate statistical analysis on the digitized time series of the
is parallel to L. See text for more details. decaying precession signal and Fourier techniques are used in
proprietary algorithms by PPM manufacturers in order to continuous polarization. An Overhauser PPM sensor should
increase the true resolution of their PPMs. consist of a container filled with a proton-rich fluid having at
Factors affecting the accuracy of the PPM include: the same time free electrons available for RF ESR. This is
obtained by dissolving a substance containing free radicals
(i) Erroneous frequency measurement. This is the most easily
such as Tempone, Proxyl, or trityl into the fluid (Primdahl,
avoided error as frequency standards are readily available
2002). A cavity resonator, supplied by an RF generator via a
for control either as precision oscillators or by broadcast-
coaxial cable, surrounds the fluid container. A coil also sur-
ing (time signals or global positioning system (GPS)).
rounds the container, which plays a similar pickup and polar-
The reference oscillator in the PPM’s electronics should
ization role as in the standard PPM.
regularly be checked against a frequency standard.
Three different polarization schemes are known and in use
(ii) Magnetically unclean sensors, which distort the field to be
at present, each requiring specific free radical substances and
measured. This contamination will result in ‘heading
RF excitation characteristics (Sapunov et al., 2001).
errors’ where the magnetometer readings will depend
The increased complexity in the chemistry of the fluid and
upon the sensor orientation.
the electronics of the sensor results in a much improved signal-
(iii) Failure to adhere to the conditions under which (symbol
to-noise ratio for the Overhauser instrument compared to the
for gamma dashed p) is defined, for example, a nonsphe-
standard PPM. Noise levels and a repeatability approaching
rical sample (Primdahl et al., 2005) or use of a proton-
1 pT are achieved, but the absolute accuracy is no better than
rich fluid other than water (Hrvoic, 2001).
the 0.2 nT of the standard PPMs, mainly because no testing
(iv) Mechanical rotation. Clockwise or anticlockwise sensor
and certification procedures exist to improve this figure (see
rotation around the magnetic field direction will add or
succeeding text).
subtract to the precession frequency. This effect is readily
By avoiding the power-hungry direct field polarization of
noticed in PPMs on board rotating platforms such as
standard PPMs, the Overhauser magnetometer can achieve
rockets or satellites (Alexandrov and Primdahl, 1993).
relatively low-power operation, which is an advantage for an
A drawback of the PPM is the low sampling rate, and the instrument intended to operate continuously in an unattended
dead times corresponding to the polarization phases, when no mode. A single reading uses about 1 W s (1 J) and the standby
data can be obtained. The low signal-to-noise ratio of the power supply can be as low as 50 mW.
standard proton precession signal makes it difficult to perform A drawback of the Overhauser PPM stems from the
geomagnetic field measurements with repeatability better than unknown long-term reliability of the device, compared with
0.1 nT. The latter limitations result in an instrument with an the tried and tested standard PPM. The single electron of the
overall accuracy of 0.2 nT in a field of about 50 000 nT. free radical makes it somewhat unstable chemically. The useful
PPMs are used extensively in ground, marine, and air- life span of the dissolved free radical is often specified as being
borne magnetic surveys; they are routinely carried on space between 5 and 10 years, and signal degradation in the best
probes and satellites and are almost ubiquitous in geomag- devices has been observed for continuous operation much
netic observatories. Provided its frequency reference is longer than this.
checked and adjusted regularly, the PPM is an excellent real-
ization of the International Magnetic Standard (IMS) (see 5.04.2.2.3 Optically pumped magnetometers
Section 5.04.3.2.1.1). Like proton magnetometers, OPMs are scalar instruments but,
unlike the standard PPM, they deliver a continuous stream
of data, in the form of a frequency that depends on the mag-
5.04.2.2.2 Overhauser effect proton magnetometers nitude of the magnetic field. OPMs are based on the Zeeman
In 1953, Overhauser predicted an effect now known as splitting of the electron energy levels of some alkali metal and
dynamic nuclear polarization (Overhauser, 1953a,b), which helium atoms in a magnetic field. The optical pumping scheme
enhances the initial alignment of the proton magnetic allows the measurement of this energy splitting – and therefore
moments considerably. His idea was to impose an appropriate the magnetic field – with very high resolution (Alexandrov and
RF signal on the atom and thereby excite the electronic spins to Bonch-Bruevich, 1992).
higher, nonthermal equilibrium states, a process now known The state of an alkali metal (or metastable 4He) atom is
as electron spin resonance or ESR. Because of a quantum determined primarily by its outermost (valence) electrons. In
mechanical coupling between the electron and the nuclear the presence of a magnetic field, atomic energy levels are split
spins, as the excited electron spins try to equilibrate to their by an amount that is proportional to the magnitude of the
lower states, they would reorient the nuclear spins. The nuclear field: this is the Zeeman effect. Optical ‘pumping’ (Kastler,
spin polarization achieved in this way would be increased by a 1950) refers to the populating of one of these Zeeman sub-
factor of about 1000, the ratio of the electronic to the nuclear levels at the expense of another.
magnetic moments. The idea was met with much skepticism: it The basic principle of OPMs is illustrated in Figures 7
was even deemed by some to contravene the second law of and 8. A monochromatic light beam of suitable polarization
thermodynamics, until, later in 1953, it was verified experi- is shone into a cell containing gaseous atoms of the alkali
mentally by Carver (Carver and Slichter, 1953, 1956). metal or helium. The wavelength and polarization are such
Dynamic nuclear polarization of the proton magnetic that atoms from only one of the Zeeman sublevels, A, are
moments is employed in the Overhauser PPM, where it excited to some higher state, C. After a very short time in state
increases the signal-to-noise ratio and lowers the power C, these excited atoms relax spontaneously: to A or B with
requirements, since little or no direct current is required for equal probability. Eventually, this process (excitation from
Excitation by pumping
Spontaneous decay
light beam
Zeeman A
sub-levels ΔE
B
RF
repopulation
Figure 7 Schematic representation of the energy states of an atom in an optically pumped magnetometer (OPM). A and B are Zeeman sublevels, split
by an amount D E in the presence of a magnetic field. C is a substantially higher energy state. See text for details of the optical pumping and RF
repopulation processes.
Magnetic field
Filter
Lens Photodiode
Lens
Synchr
Det.
Vapour cell
LoF
Lamp Oscill
Circular polarizer Tuned
Frequency or phase
modulator
RF
RF coil Oscill
Output
Figure 8 The operation principle of an Mz mode OPM. The RF frequency is usually in the hundreds of kilohertz while the low frequency, LoF, scans the
spectral line at 10–100 Hz by modulating the frequency of the RF signal.
only A, but relaxation to either A or B) leads to a situation with Table 1 OPM sensitivities and vaporization temperatures (depending
all the atoms in state B and none in A. The light beam will no on cell size)
longer be absorbed – the cell becomes transparent to it.
Substance Basic sensitivity Vaporization
Transmission of the light beam through the cell (to a photo-
(Hz nT1) temperature ( C)
diode) serves as a switch to a coil around the cell that transmits
an RF signal. The frequency of this RF signal is adjusted (via 23
Na 7.00 100–130
suitable feedback) until it corresponds to the Zeeman splitting 39
K 7.00 40–60
41
between A and B. When this occurs, A is repopulated and the K 7.00 40–60
87
light beam is once again able to excite atoms to state C, that is, Rb 7.00 25–35
85
light is again absorbed and the cell loses its transparency. This Rb 4.66 25–35
135
cycle therefore provides the basis of the feedback mechanism, Cs 3.50 20–30
4
while the RF frequency, f, provides the means to calculate B, the He 28.0 –
magnitude of the ambient magnetic field.
The Zeeman splitting is DE ¼ Dm B ¼ hf, where Dm is the
difference between the components of the atomic magnetic helium), the Zeeman spectrum is ‘resolved,’ and the polynomial
moment parallel to the magnetic field between the two sub- coefficients can be calculated directly from fundamental physi-
levels. This is calculated from the quantum mechanics of the cal constants: then, the OPM has absolute accuracy (Alexandrov
particular atom, using the Breit–Rabi polynomial formula. The and Bonch-Bruevich, 1992; Gilles et al., 2001).
linear term is roughly equal to 28/(2I 1) Hz nT 1, where I is The change in optical transmission upon the application of
the nuclear spin number, and determines the basic sensitivity of the resonating RF signal depends in detail on the direction of
the magnetometer (Table 1). In some cases (e.g., potassium and the incident light beam relative to the ambient magnetic field.
This has led to two categories of OPMs: Mz mode OPMs, where field. Measurements are made at intervals of between 1 h and
the quasi-stationary change of the transmitted light intensity is 1 s and are referred to universal time. A resolution of 1–0.01 nT
monitored, and Mx mode OPMs in which the modulation of is required. Vector measurements using, for example, fluxgate
an auxiliary light beam at the RF frequency is detected. instruments are often complemented by scalar measurements
OPMs have a high sensitivity: some approaching noise levels of intensity to improve absolute accuracy. Geomagnetic obser-
of 0.1 pT Hz1/2. This makes them attractive for magnetic pro- vatory data are collected, collated, and disseminated by the
specting and aeromagnetic surveying and for space-based obser- World Data Centers (WDC website). Nowadays, near real-
vations. Their high cost, due largely to the short lifetime of the time data are available through the INTERMAGNET website
gas discharge lamp, has limited their use in observatories. (see Section 5.04.3.4). Recently, the Virtual Global Magnetic
Rb and Cs are preferred by magnetometer designers, as their Observatory Network has made geomagnetic datasets available
low vaporization temperatures are easy to achieve and result in online via search engines and object-building Internet applica-
longer lifetimes: some have been measuring continuously for tions (Papitashvili et al., 2006).
more than 10 years. 135Cs, 85Rb, and 87Rb can all be used, but
only 87Rb has a well-resolved spectrum in the geomagnetic field
5.04.3.1 Special Magnetic Conditions in an Observatory
range, making it suitable for an absolute observatory magne-
tometer. The first digital magnetic observatory was based on a A magnetic observatory should be constructed so that only the
rubidium OPM (Alldredge and Saldukas, 1964). Cesium OPMs natural magnetic field is present. Therefore, all buildings
are used extensively in aeromagnetic survey work. 39K and 41K intended to house magnetic instruments are made from non-
also have resolved spectra in the geomagnetic field range and magnetic material (Figure 9). Additionally, the location is
produce instruments of high sensitivity. However, the chemical selected so that it is not situated on a local magnetic anomaly,
reactivity of potassium, together with its higher vaporization be it from geological or artificial origin. Consequently, the
temperature, reduces the life of the vapor cell considerably. magnetic field is very homogeneous: the magnetic field lines
Furthermore, to achieve a high sensitivity requires a cell are parallel. Magnetic field differences within the observatory
150 mm or more in diameter, making a potassium OPM rather are very small and spatial gradients are low (<1 nT m1). For
bulky. The fundamental physics of 4He is rather simpler; the observation procedures to be valid, temporal magnetic
however, there are technical problems to be overcome in its field variations must be identical (within the observational
application (Blinov et al., 1984; Gravrand et al., 2001). Helium error) inside the entire observation space.
OPMs have been flown on several satellites and space missions. As extremely high degrees of precision and accuracy are
required in both angular and component measurements (1
arc second and 0.1 nT), recording instruments must be installed
5.04.3 Magnetic Observatories on specially constructed, highly stable ‘pillars.’ Horizontal direc-
tions are referenced to true north: therefore, a target should be
A global network of geomagnetic observatories was proposed available whose azimuth is known to the same accuracy and
and initiated by Gauss and Weber in 1834: the original set of precision. Over time, these finely tuned conditions may degrade
observations providing the data for Gauss’ first spherical har- through secular changes in the environment or unauthorized
monic analysis of the geomagnetic field. Nowadays, geomag- introduction of magnetic material to the observatory premises.
netic observatories are charged with the task of maintaining Therefore, observatory conditions should be checked regularly
continuous permanent records of all three components of the and corrected if and when necessary.
Figure 9 The geomagnetic observatory of Dourbes, showing the construction of the pavilions housing the observing instruments. To the left is the
underground vault for the variometers: a high thermal inertia coupled with thick insulation ensures a low daily temperature variation in the vault.
In the center is the absolute measurement pavilion, housing the reference pillar for the observatory. To the right, a technical pavilion for
mains appliances and offices.
5.04.3.2 Geomagnetic Observations Using the Baseline 1. The magnet is first suspended from its center of mass and
Concept allowed to swing in the horizontal plane, about its equilib-
rium position in the magnetic meridian. The period of
A complete description of the geomagnetic field requires the
oscillation, T, is measured. T is given by
independent measurement of three angular or field components
(Section 5.04.1). The geomagnetic field varies with time, so I
T 2 ¼ 4p2
observatory instrumentation must be of the recording type. At mH
present, state-of-the-art component and orientation magnetom- where I is the moment of inertia of the magnet (about an axis
eters, while potentially very sensitive and fast (up to the pT through its center and perpendicular to its length) and can be
resolution at 100 Hz sampling rate), are not yet capable of calculated from measurements of its length and mass. The
measuring and recording all the field components continuously, value of the product mH can therefore be found in terms of
with the required absolute accuracy. Therefore, a two-step obser- T and I.
vatory measurement procedure is generally employed: 2. The magnet is next laid in an E–W orientation, and the
i. A variometer is used to measure the variation of the field compass is placed a distance r from its center, on the extrap-
components about baseline values, in a continuous and olation of its axis. The compass needle experiences the
unattended way, at the required sampling rate, say 1 min1. resultant of the horizontal component of the Earth’s mag-
ii. Absolute measurements are performed manually (say 1 per netic field, H, in a northerly direction and the axial field of
week) by an observer with adequate instrumentation the dipole, Hr, in an easterly direction.
(DIflux and proton magnetometer) to establish the values The field on the axis of a dipole, moment m, a (large) distance r
of the baselines mentioned in (i). Here, the expression from its center, is Hr ¼ 2m0m/4pr3.
‘absolute measurement’ means that the process of observa- The angle, a, of the compass needle from north is therefore
tion of the geomagnetic field must be fully traceable to given by
metrological SI standards and that the orientation of the
geomagnetic vector is measured with respect to the local Hr
tan a ¼
vertical and to the geographic north. H
If one can show that the baselines remain stable between from which m/H can be found. Once both the product and the
absolute measurements and the gradient between the variom- quotient of m and H are known, each can be found separately
eter and the absolute measurement site is constant, then post- and the absolute measurement task is complete.
processing can be carried out to merge the two datasets For the highest accuracy, it is necessary to consider further
and produce a final record having the accuracy of the parameters (Laursen and Olsen, 1971) such as torsion of the
absolute instrument, at the sampling rate of the variometer. suspension fiber, time changes in H, more accurate field patterns,
It is expected that in the future, fully automatic observatories and interactions of the axial magnet and compass needle.
will be available (see Section 5.04.3.4 on observatory automa-
tion) where the absolute measurement will also be made
5.04.3.2.1.3 Quasi-absolute measurement of H and Z using the
unattended.
QHM and BMZ
The QHM was invented by La Cour in 1934, it was modified
5.04.3.2.1 Absolute measurements of field components and developed by Lamont, and a Soviet version was developed
5.04.3.2.1.1 The International Magnetic Standard at IZMIRAN. Although not strictly fulfilling the requirements
Before the advent of truly absolute instruments such as the of an IMS, the QHM provided valuable measurements in many
PPM (Section 5.04.2.2), based on fundamental principles of observatories, particularly before the advent of the proton
physics, absolute measurement of the geomagnetic field inten- magnetometer. Its robustness, compactness, and practical
sity was complicated and prone to many sources of error. The design ensured its use until the 1980s, and the QHM is still
concept of an IMS was therefore introduced (Wienert, 1970). used in some parts of the world.
The IMS is an ideal, hypothetical instrument with no system- An axial magnet, of moment m, suspended by a quartz
atic error, against which observatories’ quasi-absolute instru- fiber, is able to swing in the horizontal plane. The upper
ments, such as quartz horizontal magnetometers (QHMs) and suspension point of the fiber is rotated until the mechanical
balances magnétiques zero (BMZs) (see Section 5.04.3.2.1.3), torsion of the fiber equals exactly 360 , using the magnetic
could be checked. moment of the horizontal field on the magnet for detaining
it. A total rotation of 360 þ y, where y is the angle through
which the magnet itself has turned, is read from a graduated
5.04.3.2.1.2 Absolute measurement of the horizontal component horizontal disk. If the torsion constant, t, of the fiber is known,
H by the Gauss method the horizontal component of the field, H, can be found by
Although rarely used nowadays, Gauss’ method, devised in equating the torques on the magnet due to the fiber and due to
1832, is of historical interest and importance. It enables mea- H (Laursen and Olsen, 1971):
surement of both H and the magnetic moment, m, of a magnet,
360t
using only the magnet and a compass and making measure- H¼
ments of only length, mass, time, and angle. In the past, this m sin y
procedure was the cornerstone of the absolute determination Measurements made with a QHM are only quasi-absolute
of magnetic field: because they depend on the recalibration of the torsion
constant at 1–2-year intervals. The elastic properties of the Fa (Figure 11), the magnitude of the resultant vector, Fr, which
quartz fiber and the magnetic moment of the magnet are will be measured by the proton magnetometer is given by
temperature-dependent, necessitating a temperature correction
to each measurement made with the QHM. Fr2 ¼ F 2 þ Fa2 þ 2FFa cos a ¼ F2 þ Fa2 þ 2Fa H
The key part of the balance magnétique zero, or BMZ, is a where H ¼ F cos a is the horizontal component of F, that is, the
magnet balanced on knife-edges so it is free to pivot about a component in the direction of the auxiliary field. Direct use of
horizontal axis. The vertical component of the ambient field is this equation to find H requires knowledge of F and Fa. F is
canceled by a large magnet above the housing and fine adjust- obtained simply from a PPM measurement with no auxiliary
ment of the position of a smaller magnet below the housing. field. Fa, however, requires accurate calibration of the coil
The null position of the pivoted magnet is achieved by moni- system, which is difficult to achieve to the same accuracy.
toring, through a telescope, the reflection of a light beam from Several methods of obtaining the horizontal H and vertical Z
a mirror mounted on the magnet. component of the geomagnetic field that avoid the need for
Like the QHM, the BMZ requires careful calibration, and this have been described. These generally involve switching the
measurements made with it must be corrected for temperature, direction of an accurately constant auxiliary field (Alldredge,
particularly of the upper magnet. It is therefore also a ‘quasi- 1960), increasing Fa in multiples (de Vuyst and Hus, 1966), or
absolute’ instrument. varying Fa to minimize Fr (Hurwitz and Nelson, 1960).
The crucial aspect of the proton vector magnetometer
(PVM) setup is the need for accurate spatial orientation of the
5.04.3.2.1.4 Absolute measurements with the proton vector current carrying coils – which can be quite bulky. The magnetic
magnetometer axes should be oriented exactly along the local vertical
Although the proton magnetometer is a scalar instrument, or horizontally toward the magnetic cardinal directions
meaning it measures only the magnitude of the magnetic (Figure 10). Orienting the mechanical symmetry axis of the
field vector, irrespective of direction (Section 5.04.2.2), its coils often is not enough, and special manipulations must be
accuracy and simplicity of use have motivated researchers to performed to compensate for the noncoincidence of the mag-
devise ways in which it can be used for field component netic and mechanical axes. Ultimately, a leveling precision of 3
measurements as well. The PPM is placed inside a system of arc seconds must be achieved for H accuracy measurements of
coils, by means of which known auxiliary horizontal or vertical 0.7 nT in a typical midlatitude geomagnetic field.
components can be added to the ambient geomagnetic field Another drawback of the PVM method is that it needs quiet
(Figure 10). Knowing the direction of the auxiliary field means field conditions – the measurements involve several steps,
that, in principle, information on the direction of the ambient during which the field should remain steady. It is therefore
field can be obtained from a comparison of PPM measure- difficult to implement in high-latitude observatories, subject to
ments made with and without the auxiliary field. A current frequent and large geomagnetic variations.
generator able to set the coil current to about 1 part in 106 is The PVM was being superseded by the DIflux (see
required for results aiming at the 0.1 nT level of accuracy. Section 5.04.3.2.2.3) at the end of the twentieth century, but
Considering the triangle defined by the ambient geomag- it may well be revived again in the future: with smaller and faster
netic field vector, F, and an applied horizontal auxiliary field, proton sensors and compact coil systems available, it can be
installed on the telescope of a nonmagnetic theodolite, alleviat-
ing the orientation and leveling problem (Sapunov et al., 2005).
5.04.3.2.2 Measuring the angular orientation of the

geomagnetic field
5.04.3.2.2.1 Declinometer
The magnetic declination is the angle between the geodetic and
magnetic meridian planes. The magnetic meridian plane can be
defined as the vertical plane containing the magnetic axis of a
magnet suspended so as to move freely in the horizontal plane.
Fa
H
a
Z
F Fr
Figure 11 Diagram illustrating the principles of the PVM absolute

Figure 10 A proton vector magnetometer, able to make absolute component measurement. F is the ambient magnetic field vector, with
measurements of the horizontal and vertical component of the horizontal and vertical components H and Z, respectively; Fa is the
geomagnetic field. Note the special adjustments to individually level and additional field (in this case horizontal) supplied by the auxiliary coil
orient the two sets of Helmholtz coils. system. Fr is the total resultant field. See text for discussion.
These definitions lead to a straightforward principle for the 5.04.3.2.2.3 DIflux (DIM)
absolute measurement of the magnetic declination: measuring For angular measurements, state-of-the-art instrumentation is
the azimuth of the magnetic axis of a freely horizontally sus- now provided by a device called the ‘DIflux’ or declination/
pended magnet. inclination magnetometer (DIM), which is assembled from a
This conceptually simple procedure is not easy to realize, nonmagnetic theodolite and a fluxgate sensor mounted on a
especially if accuracies of the order of the arc second are to be telescope. The magnetic axis of the fluxgate should be parallel
attained. Technical solutions (Laursen and Olsen, 1971) have to the optical axis of the telescope (Figure 12). The DIflux was
to be found in order to suspend the magnet freely in a hori- first described by Tenani (1941). By the 1970s, it had reached
zontal plane, to observe the orientation of its axis and to an advanced level of development (Meyer and Voppel, 1954;
measure angles between the meridians with the required accu- Serson and Hannaford, 1956; Trigg, 1970). Further develop-
racy, and to eliminate rheological effects, such as hysteresis and ments by Daniel Gilbert, Jacques Bitterly, and Jean-Michel
anelasticity, in the suspension fiber. Cantin at IPG Paris have resulted in levels of precision, accu-
As a result, the declinometer, or magnetic theodolite, was racy, resolution, and ease of use that make it currently the
developed. It consists of a torsion head suspending the magnet preferred instrument for angular measurements both in the
in the required position. This head is mounted together with a observatory and in the field (Bitterly et al., 1984).
telescope on a vertical axis rotation table, indexed by a gradu- The measurement principle takes advantages of the direc-
ated circle. The telescope performs the dual task of observing a tional properties of the fluxgate: only the projection of the
distant target with known azimuth and collimating on the geomagnetic vector on the fluxgate axis is measured. Using
magnet-end mirror. This mirror is mounted on the magnet the theodolite, an observer can orient the fluxgate in any
with its optical plane normal to the magnet’s magnetic axis. direction while monitoring its electronic output and keeping
Angles can typically be obtained to an accuracy of 0.1 arc track of its orientation in space via the theodolite’s graduated
minute or even 1 arc second. horizontal and vertical circles. Usually, the preferred orienta-
The declinometer has been superseded nowadays by the tions of the fluxgate correspond to its magnetic axis being
DIflux, which offers operational simplification and an perpendicular to the geomagnetic field vector: Then, the elec-
increased accuracy (see Section 5.04.3.2.2.3). tronic output is close to zero and a high signal amplifier gain
can be applied. The declination measurement involves setting
up the telescope axis into the horizontal plane, with the flux-
5.04.3.2.2.2 Inclinometer gate axis normal to the magnetic meridian, so as to obtain a
Historically, the first inclinometer – the ‘dip circle’ – was built null from the fluxgate electronics. For inclination, the horizon-
as a needle magnet free to move around a horizontal EW axis. tal axis of the theodolite is set normal to the geomagnetic
A vertical graduated circle, centered on this axis, allowed the meridian and a fluxgate null is again sought. The essence of
measurement of the angular position of the magnet in the the measurement is therefore to determine the orientation of a
magnetic meridian, and this gave the magnetic inclination. plane perpendicular to the geomagnetic field vector. The
Despite clever procedures to eliminate the effect of gravity plane’s orientation is defined by two orthogonal lines in it,
and to minimize collimation errors, the dip circle never one horizontal and one in the geomagnetic meridian.
achieved the high accuracy required in a magnetic observatory. In practice, a series of four measurements for declination
Another inclinometer, known as the ‘earth inductor,’ with and two for inclination (Lauridsen, 1985) are usually made, to
better metrological characteristics was introduced at the begin- minimize errors.
ning of the twentieth century. This device was the first in The accuracy of a measurement with a DIflux depends on
geomagnetic observatories to use an ‘electronic’ sensor as a the accuracy and the magnetic cleanliness of the theodolite.
null indicator. This apparatus uses the emf induced by the The accuracy of a theodolite can be assessed by appropriate
geomagnetic field in a coil rapidly spinning about one of its
diameters, as an indicator of its orientation relative to the field
vector. When the coil’s spin axis is collinear with the field, no
emf is generated. Small deviations from colinearity can be
amplified with high gain to give a restoring signal limited
only by system noise. The early amplifiers were galvanometers
deflecting light spots, and the rectification of the small alter-
nating emf’s by mechanical rotating contacts was not always
satisfactory. Parasitic DC offsets were eliminated by observa-
tion with the coil oriented in symmetrical positions. Play in the
bearings of the spin axis was probably the main limiting factor
of accuracy. A modern version of the earth inductor, the Tur-
bomag (Schnegg and Fischer, 1991), uses hydrodynamic bear-
ings and contactless transfer of the emf to the observer.
In principle, the earth inductor could have been used for
declination measurement also but, apart from the Turbomag, Figure 12 A DIflux with its electronic console. This instrument is able
this did not occur probably because the addition of a telescope to measure the geomagnetic inclination and declination. A single-axis
would have made it too bulky. Nevertheless, the earth inductor fluxgate sensor is located on top of the theodolite’s telescope so that the
paved the way for the development of the DIflux. magnetic axis is parallel to the telescope optical axis.
measurement techniques (Deumlich, 1980) and the magnetic pillar tilt, and being mechanical, they do not suffer from
cleanliness can also be measured and sometimes corrected to electric noise; however, their sensitivity and dynamic range
be below a given limit. We can therefore classify the ‘clean’ are limited, and they are now costly to maintain.
DIflux as an absolute instrument. In that sense, it belongs to
the family of the IMSs as defined in Section 5.04.3.2.1.1. 5.04.3.2.3.2 Digital variometers
Instrumental errors originate in imperfect parallelism of Many observatories now have electronic variometers with a
optical and magnetic axes of telescope and fluxgate sensor digital data acquisition system that may be uploaded directly
and in residual magnetization of the fluxgate core. These errors to the World Data Centers. Introduced from the late 1960s,
are eliminated when using the standard measurement proto- electronic variometers were expected to obviate the disad-
col. Since the geomagnetic field may vary during a series of vantages of photographic variometers, namely, the lack of
measurements, an external D and I variometer must keep track dynamic range and sensitivity. It was also felt that they would
of this change if we want to have a truly absolute spot mea- be more economical, being easier to install and offering the
surement of D or I. The proven best way to do that is by possibility of unattended operation. However, the introduc-
measuring the baseline of the variometer with the four D and tion of these new instruments has come with some drawbacks,
two I measurements of the DIflux. as the electronic components are less robust against electro-
The DIflux performs well in equatorial and midlatitude magnetic/electric disturbances than the mechanical configura-
zones. The DIflux can also be operated in polar regions tion of suspended magnets. Many digital variometer
where the geomagnetic field vector is close to vertical. installations have been lost as a result of a lightning strike on
However, because the horizontal components are small the observatory. Also, as the sensors are no longer suspended,
(<2000 nT), horizontal angular measurements are not conve- the new instruments are more sensitive to tilt. The acquisition
nient there. Direct measurements of the small X and Y compo- system samples an analog signal at discrete intervals, so low-
nents using the fluxgate, properly oriented along geodetic pass filtering at at least twice the Nyquist frequency is necessary
north–south and east–west and with a scale factor directly to avoid aliasing. This proves to be a problem with sensors
determined with a proton magnetometer, are preferred such as proton magnetometers.
(Gilbert et al., 1988). The sensors of digital variometers presently in use in mag-
A problem facing the DIflux nowadays is in the supply of netic observatories are mainly fluxgates (Rasmussen, 1990) or
nonmagnetic theodolites. Therefore, the future is uncertain for scalar magnetometers surrounded by backing-off coils
the DIflux, and the development and marketing of a fully (Alldredge and Saldukas, 1964; GEM website). Suspended
automatic DIflux would provide a solution (Section 5.04.3.4). (hanging or taut-fiber) magnets are still used in some designs,
notably Bobrov, with attitude-restoring feedback coils
5.04.3.2.3 Variation measurements: Variometers ( Jankowski et al., 1984). As the necessary filtering cannot be
In the present context, a variometer is a magnetometer applied to PPMs and as they are too slow to measure the high-
designed to monitor the time changes of a magnetic field frequency portion of the geomagnetic spectrum, OPMs should
component relative to a fixed baseline. The variometers be used for variometric multiaxis recording (Alexandrov et al.,
installed in magnetic observatories measure a variety of vecto- 2004; Gravrand et al., 2001).
rial components depending upon the reference system used Perhaps the biggest problem with the currently used vari-
(see Section 5.04.1). The most popular orientations are D, H, ometers is their vulnerability to temperature variations. An
Z and X, Y, Z. The latter ensures that the variometer orientation instrument with a temperature coefficient of 0.1–1.0 nT K1
will not have to be modified over time due to secular variation in an environment that undergoes temperature variations of up
of the geomagnetic field. to 10 K may result in errors of 10 nT, or 30% of the typical
The essential quality of a variometer is its ability to main- daily variation in a component of the field. Clearly, it is desir-
tain a stable baseline between two absolute measurements and able to use variometers with low-temperature coefficients, to
orthogonality of the measured components and scale factor minimize temperature changes, and to correct for temperature
accuracy. This means that it should have a very low drift over variation wherever possible.
time and a very small dependence on temperature, pressure,
humidity, etc. A variometer should respond to variations in the 5.04.3.2.4 Instrument certification and calibration
selected component only: contamination from other compo- The community of geomagnetic observatories has always been
nents must be eliminated. This often reduces to ensuring the concerned to set standards in order to maintain the quality of
correct orientation of the sensors. In the case of multiaxis geomagnetic data produced and collated.
sensors, the quality may be measured by the orthogonality of Not only should metrological standards be examined, but
the setup being close to perfection. The quality of the installa- also other parameters that affect instrument performance, for
tion of the variometer must also to be considered: tilts or example, temperature coefficients of variometers, should be
rotation of the pillar on which the variometer is installed will monitored.
appear in the variometer’s recording as a component variation.
5.04.3.2.4.1 Variometer certification
5.04.3.2.3.1 Classical magnet-based instruments The factors that affect a variometer’s certification can in prin-
Recording variometers originally used suspended magnets and ciple be deduced from an inspection of their baselines, which
photographic recording. They provided, and still do provide, are obtained from measurements with absolute instruments.
much of the geomagnetic data going to the magnetic observa- This is true for instrumental parameters like long-term stability,
tories data banks. Being suspended, they are not affected by orthogonality, and scale factor accuracy. However, except for
scalar measurements, the diurnal effects of temperature cannot corresponding uncertainty in declination or inclination (1
be detected unless half a dozen or so absolute measurements arc second in declination, in a horizontal field of 20 000 nT).
are taken per day. This is not yet routinely possible. Therefore, Another frequently used procedure for certifying DIfluxes is
the certification of a variometer is best obtained by intercom- through participation in the ‘Workshops on Geomagnetic
parison with a master variometer. This has been often realized Observatory Instruments, Data Acquisition and Processing’
at the ‘Workshops on Geomagnetic Observatory Instruments, already mentioned. One of the main activities in this kind of
Data Acquisition and Processing’ organized by IAGA every international workshop is the measurement of the baseline of a
other year since 1987. Even better is the continuous running stable onsite variometer by all the participating DIfluxes. As all
of two variometers set up in slightly different conditions at the devices should measure the same constant baselines, any devi-
same observatory. ation of an instrument indicates a fault either in the angle
Some variometers have sufficient dynamic range for mea- reading or in the magnetic hygiene of the DIflux.
suring the full field in all three axes. It is then possible to
perform an experiment where the measuring variometer is
oriented in various directions with respect to the field while a
scalar magnetometer is recording close by. One may then 5.04.3.2.5 Obtaining definitive absolute data
equate the calculated modulus obtained from the variometer As mentioned previously, definitive geomagnetic field data at a
measurements to a reading taken from the scalar magnetome- magnetic observatory are obtained from two streams of data:
ter. This will lead to a redundant set of equations from which a variometric measurements and absolute measurements. Merg-
rotation matrix describing the errors of the variometer (orthog- ing the two datasets is not a trivial or unique procedure and it
onality and scale factors) can be extracted. Subsequent correc- requires all the skill of the observatory staff to result in an
tion of the variometer with the matrix operator will then accurate time series, giving the absolute value of the vector at
provide a variometer reading with quasi PPM accuracy each variometric sample. Important factors that impact on the
(Gravrand et al., 2001; Merayo et al., 2000) for whatever procedure are
orientation in space. (i) baseline stability of the variometer,
(ii) accuracy of the absolute measurements,
5.04.3.2.4.2 Calibration of scalar magnetometers (iii) frequency and regularity of absolute measurements,
Most scalar magnetometers do not in principle need to be (iv) data gaps in the variometer time series.
calibrated, since their operation is traceable to fundamental Items (i) and (ii) have been dealt with in earlier sections.
physical constants. This is particularly true for the PPM, where, Items (iii) and (iv) lead directly to the main problem facing
in addition to an accurate, precise value of the proton– definitive data production: the adoption of baselines.
gyromagnetic ratio, only calibration of the frequency reference In the simplest case of spherical coordinates for both abso-
is required. The situation is more complicated for the absolute lute and variometric data, we have the fundamental magnetic
OPMs where several physical constants are required. Some, like observatory relationships between absolute (D, I, F) and vario-
the Landé factors for K, have not been determined to a high metric measurements (dD, dI, dF):
degree of precision, and these dominate the error budget of the
OPM. Instrumental effects like light shifts and phase errors in D ¼ D0 þ dD
electronic signals also need to be taken into account. The result I ¼ I0 þ dI
of this is that scalar OPMs are quoted with varying and some-
times conflicting accuracies. A consensus exists however that F ¼ F0 þ dF
the agreement between PPMs and OPMs is at the 0.1 nT level in
The subscript 0 indicates the baseline. In an ideal world with
a 50 000 nT field. This corresponds to 2 ppm.
perfect instruments, the baselines would be merely constants,
Two techniques exist for intercomparing scalar magnetom-
but instrumental drift, due to temperature effects, mechanical
eters. The magnetometer exchange procedure (Rasson, 2005) is
creep, and pillar instabilities makes the baselines wander in an
carried out in the ambient field and will therefore only give a
unpredictable way. Values of the baselines D0, I0, and F0,
difference for that value. The other technique uses artificial
which are known every time an absolute measurement is
fields in a field stabilizer (Shifrin et al., 2000) and will provide
taken (say 1 per week), vary slightly with time. But a value of
intercomparison over an extended range of fields.
the baselines must be available for every sample taken by the
variometer. Adopting a baseline therefore entails the interpo-
5.04.3.2.4.3 Certification of DIflux (DIM) lation of the baseline from a sampling at 1 per week to the
Since the DIflux performs angle measurements, we cannot variometer sampling typically at 1 min1. This can be conve-
strictly speak of absolute measurements, as there need not niently done by fitting a mathematical expression like a low-
exist a standard for angles. Nevertheless a critical assessment order polynomial or a spline function to the baseline data
of the accuracy can be made, by investigating the angular sampled at 1 per week (Figure 13).
accuracy (Deumlich, 1980) and the magnetic hygiene of the A final quality control should be made when the definitive
fluxgate-bearing theodolite. data series have been obtained, in order to ensure that no
Provided those two checks give results that comply with errors resulting from faulty data processing software have
specified levels of accuracy, the DIflux can be declared certified. occurred. A simple way of doing this is to go back to the
Note however that the two are not independent: an uncertainty absolute measurements performed during the data time series
(of, say 0.1 nT) in the fluxgate readings results in a under consideration: the absolute spot measurements and the
Manhay observatory inclination baseline of LAMA variometer measured with Diflux 117 on pillar MA2
65.800
65.795
65.790
Degrees
65.785
65.780
Inclination baseline from absolute measurement

65.775
Adopted inclination baseline
65.770
−30 0 30 60 90 120 150 180 210 240 270 300 330 360 390
Day of year 2005
Figure 13 Illustration of the baseline (here measuring the magnetic inclination, I) in the MAB observatory for a whole year. Note the sensitive vertical
scale of 0.005 per division, corresponding to about 5 nT per division in Z. In this case, a third-order polynomial, fit by a least squares procedure,
has been adopted as a smooth baseline. The differences between measured and adopted baseline are of the order of 0.001 .
definitive data should agree. The adopted baselines should also adjusted (corrected for artificial spikes and jumps), and defin-
agree with the absolute data baseline determinations. itive (reduced to baseline so that it has absolute accuracy). The
latter is made available a few months after the end of each year
at the earliest and finally with the production of a yearly
5.04.3.3 The INTERMAGNET Magnetic Observatory Network
CD-ROM.
The International Real-Time Magnetic Observatory Network People needing real-time data use the reported or adjusted
(INTERMAGNET) was created in order to establish a world- data, available from the GINs in daily ASCII files in either
wide cooperation of digital magnetic observatories. The INTERMAGNET imfv1.22 or IAGA 2002 format (email request
networked observatories agreed to adopt modern standard or ftp).
specifications for measuring and data-logging equipment and Access to recent definitive data is through the website
to make data available in close to real time. Moreover, INTER- (INTERMAGNET, website). Older data are available on the
MAGNET extends technical support for maintaining and CD-ROMs and at the website. Definitive data come in monthly
upgrading existing magnetic observatories and for establishing files in a dedicated binary code. Multiplatform java-based
new ones. browsers are available for easy perusal, inspection, and format
INTERMAGNET defines standards for the measurement conversion.
and recording of the geomagnetic field, considering the state The data are available at no cost for bona fide scientific
of the art. INTERMAGNET is constituted from existing groups users. If any commercial aspects are involved, the user should
who undertake geomagnetic observatory measurement. The contact the IMO directly for a financial arrangement.
acronym IMO is used to indicate an INTERMAGNET Magnetic Since 2008, two new data products have been available
Observatory (Green et al., 1998). from some IMOs: ‘quasi-definitive data (QD)’ and ‘one-second
Presently, INTERMAGNET data consist of time series of the data.’ QD are data very close to the definitive data and are
geomagnetic vector, sampled at the round minute and carefully prepared on a monthly basis, being available a few months
filtered to avoid aliasing effects. These data, collected at the after collection. The one-second data product, as specified by
IMOs represented in Figure 14 (crosses), are continuously strict INTERMAGNET measurement standards, provides 1 Hz
available from the Geomagnetic Information Nodes (GINs, samples of the geomagnetic field components. These two new
crossed circles) within 72 h. The data come in different accu- products address specific needs of the user mainly in the geo-
racies and delays: reported (as recorded – near real time), magnetic satellites and space weather communities.
INTERMAGNET observatories and GINs (2013)
Figure 14 INTERMAGNET magnetic observatories (crosses) and Geomagnetic Information Nodes (GINs) (crossed circles) representing the global
coverage of the network in 2013.
All practical details about the data and their access are 5.04.3.5 Magnetic Repeat Station Surveys
available from the INTERMAGNET website where the
From the point of view of secular variation measurement, a
technical manual is also available for download.
magnetic repeat station plays the role of a miniature magnetic
observatory, with elementary infrastructure and an observation
schedule reduced to one measurement session every few years.
5.04.3.4 Fully Automatic Magnetic Observatories This section on repeat surveys is therefore included with mag-
netic observatories, although we are strictly dealing with a
If the globe is one day to be covered by a uniform distribution
survey operation.
of magnetic observatories separated by no more than say
Instrumentation used during repeat surveys is nowadays
2000 km, it will be necessary to install some in very remote
similar to that used in an observatory. As pillars are seldom
and hard to access places such as on the deep-sea bottom
possible in the field, a nonmagnetic tripod is often used for
(Chave et al., 1995), on remote islands, and in inhospitable
setting up the DIflux and PPM.
deserts. This is unlikely to be accomplished with classical
For declination measurement, a device for measuring the
observatories running the traditional way. Therefore, there is
geodetic azimuth of a distant target is necessary, at least when
a need for fully automatic magnetic observatories.
the repeat station is installed for the first time. Provided there is
Technology already exists for some geomagnetic measure-
sunshine, a theodolite with a solar filter is useful for this since
ments to be executed automatically.
then the astronomically known position of the sun can provide
Variometers are able to work for long time spans in an
the target’s azimuth (Rasson, 2005). Often, the DIflux used
automatic mode, provided their baselines are regularly checked
for the magnetic measurements can be used for the sun sightings.
by absolute measurements.
If the sun is not visible, a gyrotheodolite can be used. The true
Provided electric power is available, the magnitude of the
north direction is then obtained by sensing the direction of
geomagnetic field can be measured easily and absolutely by a
the Earth’s rotation axis by way of a suspended gyroscope.
scalar PPM in an unattended way for long periods. This is
Differential GPS is also a possibility but needs the occupation
because a PPM does not require orientation or mechanical or
of both the repeat station and target by the measuring team.
electronic adjustments, in order to produce absolute data.
A good topographic map may also help obtain the azimuth of
Two approaches have been followed for obtaining automa-
the target.
tion of observatories.
When a measurement of the magnetic field is performed at a
The first is to set up a very precise (in terms of scale factor
repeat station, it is a spot value measurement that can be notably
and orthogonality), stable, and temperature-insensitive vari-
different from the mean value at the site. Mean values, normally
ometer in a fixed orientation, in a clean geomagnetic environ-
taken over one full year, are usually required in a repeat survey.
ment. Then, baseline checks as far apart as 1 year may be
The difficulty then is to obtain this mean from a single
acceptable (Gravrand et al., 2001).
measurement session at the station. Several procedures have
The second approach is to automate the full absolute mea-
been designed, where the spot measurement is linked in some
surement protocol, which comes down to automating the DIflux
way to a nearby observatory, where the annual mean is known.
measurements, since the modulus can be measured automati-
All methods try to eliminate or mitigate the main source
cally by a PPM (Rasson, 1996; van Loo and Rasson, 2006).
of error in this linkage: differential field variations – mainly
Some magnetometer setups flown on satellites have already
diurnal – between station and observatory:
reached a level of quality in baseline stability and precise
orientation with respect to the stars such that they can truly The magnetic field variations at the repeat station site are
be considered as automatic magnetic observatories in orbit recorded for a few days with a dedicated variometer. The
(Nielsen et al., 1995). spot value measurements are then used to establish the
baselines of the onsite variometer. Several daily means can magnetic field of internal and external origin and our under-
then be computed with absolute accuracy. Subtracting the standing of the Earth’s deep interior. Navigation is a notable
corresponding daily means at the nearby observatory will practical application, but there are also others like space
then produce the difference in field elements station–obser- weather, metrology, and magnetic signature determination.
vatory. The annual mean can then be obtained by simple Table 2 is an attempt to list, as completely as possible, the
addition. products and services a magnetic observatory is able to
The on-site spot value measurements are used to compute the provide.
baselines of a nearby magnetic observatory variometer. These
baselines are subtracted from the baselines obtained from
absolute measurements made at the observatory and give the
difference in field elements station–observatory. The annual
5.04.4 Magnetic Surveys for Geological Exploration
mean can then be obtained by simple addition. This method
The general aim of a magnetic survey is to improve the resolu-
supposes that the (diurnal) variations are similar at both the
tion of the IGRF through a greater sampling density and/or
station and observatory, at the time of the spot measure-
frequency. This might be at a regional level to better depict
ments. This method hence works well for a close-by obser-
high-order features of the main field. Such low-density
vatory or for measurements taken when the diurnal variation
regional surveys are often repeated (reoccupying the same
at both places is small: dusk, sunrise, or sunset.
sites as far as possible) at 5–10-year intervals to monitor secu-
On-site spot value measurements are performed at night,
lar variation and are discussed in Section 5.04.3.5.
when the diurnal variation is negligible. With such a mea-
At a more local level, magnetic surveys may be designed to
surement, it is considered that an undisturbed field is
map anomalies attributable to the rocks in the Earth’s crust
obtained and that it reflects truly the secular variation.
(e.g., at a scale of 1:250 000) and this has now been carried out
Direct comparison with a nearby observatory – also situated
routinely over most of the world’s land area to assist geological
in the dusk zone – is allowed.
mapping and frontier resource exploration. At a more detailed
Obviously, quiet magnetic conditions are required for accu- scale, investigations of local sites of archaeological or environ-
rate results. mental interest (e.g., 1:1000) may also be carried out using
IAGA has published a detailed guide for magnetic repeat magnetometers. Magnetic surveys may be conducted over land,
survey practice (Newitt et al., 1996), where the reader can find at sea, or from air. Usually, only scalar (intensity) measure-
all the necessary practical information to set up a magnetic ments are made but at fractional nT sensitivity. Absolute accu-
repeat station network. racy is of relatively little importance, so such surveys are
seldom rigorously linked to observatories but meticulous care
must be taken with the correction for temporal variations.
5.04.3.6 Products and Services Magnetic Observatories
PPMs are ideally suited to local survey work on the ground,
Can Provide
being inexpensive, quick and easy to use, and readily portable.
As mentioned in Section 5.04.1, magnetic observatories have a Airborne surveys, of which many millions of line-km are flown
very practical purpose and contribute to our knowledge of each year, now usually employ cesium vapor magnetometers.
Table 2 Products and services provided by geomagnetic observatories
Products/services Users
Value of the magnetic declination for various epochs and/or Topographic and cartographic services, oil and gas companies, harbors, air traffic
locations services, airports, military
Value of the secular variation of the magnetic declination Topographic and cartographic services
International Geomagnetic Reference Field generation, Topography, aeronautical mapping and safety services, maritime coastal and fluvial
magnetic charts mapping, military
Time series of geomagnetic field Aeromagnetic and marine surveyors. High-precision directional drilling for gas and oil,
Earth-orbiting satellite safety, space weather, military
Geomagnetically induced currents: nowcasting and National and international electricity mains providers and operators, other cable
forecasting network operators
High-precision in situ measurements of the magnetic Airports, military, topography
declination or other components
Setup and certification of compass roses (aircraft compass Airports
calibration and compensation facility)
Measurement of magnetic orientation of aircraft runways Airports
Measurement of magnetic signatures of equipment, Manufacturers of magnetic resonance, military
mechanical or electronic components
Calibration and check of magnetic compasses Leisure equipment industry, geometers, surveyors, antenna orientation for satellite TV
EM radiation standards of safety and security, electrosmog Government public safety agencies
Keeping of the magnetic induction metrological standards in Metrological community
the Earth’s field range
At sea, a proton magnetometer may be towed behind a correction for temporal variations and subtraction of an appro-
survey ship. priate IGRF, the anomaly recorded is then the component of
The main problem with making vector measurements at sea the local-source anomaly in the direction of the main geomag-
and from the air is the lack of a stable reference frame. Accurate netic field (Section 5.04.4.2). This is universally understood as
determination of the vertical is essential and this ultimately the ‘total field anomaly’ in exploration magnetometry. Mea-
limits the accuracy of results. So far, the added cost and compli- surements made ten times per second correspond to a sample
cation of measuring three components have not achieved com- spacing of about 7 m on the ground with a typical survey
mercial viability in the industry. Most applied magnetic surveys aircraft. Very often, surveys are now augmented by measure-
therefore measure only the scalar magnitude of the total geo- ments of horizontal gradients of the anomalous field using a
magnetic field, including the contribution of local anomalies, second or third magnetometer on the aircraft. Advantages
typically to a sensitivity of 0.1 nT. In the air, this is now com- include improved resolution of near-surface sources.
monly done using three magnetometers, one in each wingtip Magnetometers have also been mounted in earth-orbiting
and one in the tail. The differences recorded between them are satellites, most successfully in the ersted and CHAMP satellites
estimates of the gradient of the anomaly field across and along of recent years. However, at satellite altitudes, the amplitudes of
the flight line. This improves the accuracy of interpolation of the anomalies attributable to lithospheric sources are no more than
Laplacian field into the spaces between flight lines and onto the about 30 nT and the resolution of detail no better than the
regular rectangular grid of data points used for imaging the data. altitude of the satellite, setting a rather coarse limit of
Many useful ways of processing such a grid, relying on its 300–400 km for the scale of geological detail that is resolvable
Laplacian nature, are also available to enhance the magnetic (Chapter 1.07). The determination of a reliable and universal
features revealed (Reeves, 2005). ‘background’ against which local magnetic surveys may be fitted
has been a valuable contribution of satellite measurements.
The production of magnetic anomaly maps and images in
5.04.4.1 Mapping Magnetic Anomalies
general has evolved rapidly in the last 50 years as a result of (a)
Local magnetic disturbances attributable to certain types of improved – now exclusively electronic – magnetometer and
rock have long been recognized. Where these rock types are ancillary instrumentation; (b) computer software and hard-
of economic importance, their magnetic effect may be used as a ware to gather and process large volumes of data; and (c)
means of detecting them, even where the rocks are at depth or image-processing techniques to display results in image format
concealed below soil cover or overburden. Instrumentation to that lead to a largely intuitive appreciation of the geological
measure such local variations or ‘anomalies’ in the geomag- information content.
netic field in exploration mode – that is, carrying out a survey
by making observations repeatedly in a pattern of observation
points distributed systematically over a given search area – 5.04.4.2 The Origin of Magnetic Anomalies
evolved rapidly during the second half of the twentieth century
and eventually became the basis of an airborne geophysical While most rocks are only very weakly magnetic, any rock
survey industry that now sets the pace in geological reconnais- containing a quantity of a ferrimagnetic mineral such as mag-
sance worldwide. By comparison, even with the aid of aerial netite distributed through its mineral fabric may display mag-
photographs and satellite imagery, traditional geological map- netic properties. The magnetization may be a long-lived,
ping on the ground is prohibitively slow and expensive in many virtually permanent feature of the rock, having its origins in
large areas of the world remote from modern infrastructure – the geological history of the rock itself – for example, when it
and often devoid of outcropping geology. Aeromagnetic cooled through its Curie temperature (Section 5.04.5) – and
surveys enable the groundwork to be focused more selectively described as a remanent magnetization Mr with a direction that
and efficiently and therefore play a significant role in mineral is, in general, unknown. On the other hand, the rock may have
and petroleum exploration (Reeves, 2001). a certain magnetic (volume) susceptibility k, in which case it
Ground-based magnetometer surveys are still used to will acquire an induced magnetization simply due to its pres-
follow-up in more detail anomalies that have been selected ence in the present-day geomagnetic field. Such induced mag-
from airborne surveys, for example, prior to exploratory dril- netization, Mi, is proportional to the product of the magnetic
ling. Specialized systems have been developed for particular susceptibility and the inducing field and will be in the direc-
localized applications such as engineering site investigations, tion of the inducing field:
detection of unexploded ordnance and archaeology. At sea, Mi ¼ kF
magnetometers are often towed behind marine research ves-
sels. While this is slow and expensive on its own, cruises carried Both types of magnetization may be present and their relative
out primarily for other purposes can simultaneously conduct importance is quantified by the Koenigsberger ratio, Q:
magnetometer traverses at little additional cost. Hence, appli-
Q ¼ Mr =Mi
cations of surveying with magnetometers are to be found in the
air, on the ground, and at sea. Considering, for simplicity, a single body of magnetized rock
In recent decades, essentially all magnetometers employed isolated within a large volume of nonmagnetic country rock,
for work of this type have been electronic magnetometers that the magnetic rock body will possess a magnetic field that will
measure only the scalar magnitude of the magnetic field, be evident within its immediate vicinity and over a certain
regardless of its direction. This eliminates any need for precise volume of space thereabouts. This will be the case whether
orientational reference and so simplifies survey practice. After the magnetization of the rock is induced, remanent, or a
combination of both. A scalar magnetometer measuring the variations have been eliminated. Figure 15 illustrates that, to
total field strength within the vicinity of this body will experi- a close approximation, the anomaly defined as in the preced-
ence the magnitude of the vector sum of the geomagnetic field ing text is the component of the local-source field in the
and the local field due to the magnetic rock body (Figures 15 direction of the geomagnetic field. As a result of the variations
and 16). In the vast majority of cases, the latter is found to be at in the direction of the local magnetic field around the rock
least two orders of magnitude smaller than the former. The body, there will, in general, be areas where the local field tends
former is predictable at any given location from models of the to reinforce the geomagnetic field and others where it tends to
geomagnetic field; the model provided by the IGRF (IGRF-11, oppose it. It follows that, when mapped over a given area
website) being the one used almost universally in magnetic around the body or ‘source,’ some areas will show a positive
surveying. magnetic anomaly value while others will show a negative one.
What is measured by the surveying magnetometer is the A typical anomaly over a simple compact body is shown in
vector sum of the global and local contributions and, by con- Figure 17.
vention, the so-called total field magnetic anomaly is what is In general, the shape of any magnetic anomaly is a function
left when the magnitude of the predicted field (IGRF) is sub- of many factors, including the geomagnetic inclination and
tracted from the observed field magnitude, once temporal declination at the survey locality, the resultant direction of
total (induced þ remanent) magnetization within the body,
the geometry of the body itself, its attitude in the ground (dip
and strike), and the strength of its magnetization.
The shapes of anomalies over bodies of simple geometric
shape approximating to common geological bodies may be
calculated directly, a process known as forward modeling.
Such forward models help in interpreting the sources of
observed anomalies in terms of the possible depth, dip, and
geometric form of the source body. While the inverse problem
FIGRF
Fobserved is, in principle, underdetermined, reasonable geological
assumptions or other a priori information can usually be
brought to bear to add practical constraints in the inversion
process that is often used in anomaly interpretation.
D F cos q In practice, the flat-topped dipping dyke, extending down-
dip to infinity and with parallel sides, is one of the more useful
models for the geological sources of many anomalies. The
q DF anomaly to be expected over such a body with a vertical dip
for a range of magnetic inclinations is illustrated in Figure 18.
Approximations of magnetic sources by isolated poles or
Figure 15 By convention, the ‘total field anomaly’ is the difference
between the observed total field value, |Fobserved|, and the geomagnetic dipoles are found to be of little value in serious interpretation,
field value predicted by the IGRF, |FIGRF|. As long as the local anomaly is particularly now that computer algorithms are readily available
much smaller than the geomagnetic field, the total field anomaly so to compute accurately the effects of realistically dimensioned
defined is a close approximation to the component of the magnetic field geological bodies in two and three dimensions. In specific
of the local source in the direction of the geomagnetic field, DF cos y.
FIGRF
High
1050
S N
Low
Source
5
107
of
DF
1075
5 km
H
Figure 16 In the vicinity of a magnetized rock body, the passing
magnetometer experiences the influence of both the ambient Figure 17 A typical aeromagnetic anomaly pattern in the vicinity of a
geomagnetic field, modeled by the IGRF, FIGRF, and a local-source single, compact, and isolated geological source, showing both positive
field, DF. and negative parts (southern hemisphere). Contour interval 25 nT.
0
-9
90
5
-7
75
0
-6
60
5
45
-4
0
30
-3
5
15
-1
S N
0
0
Figure 18 The magnetic anomaly over a vertical dyke-like body striking east–west and magnetized solely by induction varies with the angle of
inclination I of the geomagnetic field at the locality of the body. The figure shows the curve family as I goes from þ90 to 0 (northern hemisphere)
as solid lines and from 0 to 90 (southern hemisphere) as dotted lines. Note that the positive lobe of the anomaly is displaced from above the
body in the direction of the equator in both hemispheres as the negative lobe grows on the side toward the nearest pole.
applications, such as understanding the structure and stratig-

raphy in a potential petroleum province, the interpretation of
gravity and magnetic data in terms of geological bodies and
horizons of specified densities and magnetizations can be a
helpful adjunct to understanding the stratigraphic and struc-
tural information contained in seismic sections. Since most
sedimentary rocks are essentially nonmagnetic, the anomalies
in such an area are usually of deep-seated origin, arising from
the igneous and metamorphic basement below the sediments.
Hence, a rather small number of deep-seated anomalies are
seen and their interpretation can help understand the structure
of these older rocks, the downfaulting and subsidence of which
led to the development of the sedimentary basin of interest for
petroleum resources.
In normal survey practice, particularly in areas of Precam-
brian basement typical of solid mineral exploration environ-
ments, many hundreds or even thousands of anomalies are
Figure 19 The magnetic anomalies in a certain area are displayed here
often detected (Figure 19). The extent to which any one of as a grayscale shaded-relief image. In many areas, the geology gives
them may be quantitatively interpreted depends to some rise to a plethora of anomalies that enable, in this case, the folds in
extent on the degree of interference from neighboring anom- the rock formations of the western half of the area to be traced
alies. In any event, the trailing off to zero at infinite distance, quite clearly. Some subtle dyke anomalies appear in the NE corner of
predicted mathematically for all anomalies, is seldom the area. Courtesy of Geoscience Australia.
observable to any great distance from the source. Often, the and correct for temporal variations in the main geomagnetic
interference from nearby anomalies is quite severe and the field. As is often the case with exploration surveying, the quan-
emphasis of the interpretation must then be mostly on qual- tity of observations and the uniformity of coverage of an area is
itatively tracing patterns of anomalies from one place to more important than the absolute accuracy of each individual
another and reconciling these patterns with the (often iso- reading, the idea being to cover ground and detect interesting
lated) exposures of certain suites of rock in the survey area. anomalies.
Even so, aeromagnetic anomaly mapping has proven to be a Improved performance in ground surveys was soon
powerful tool in extending geological mapping beyond areas achieved by the proton precession magnetometer (Section
of relatively well-exposed geology, in old (particularly Pre- 5.04.2.2) in the 1950s and 1960s with the added advantage
cambrian) terrains where the topography has become sub- that it was the magnitude of the total field that was being
dued and overlain by younger weathering products (see later measured, directly and absolutely to 1 nT accuracy, eliminating
sections). the need for any careful adjustment, orientation, or calibration
It should be noted that it is rarely the case that eye-catching of the instrument. This also brought ground surveys into line
magnetic anomalies arise directly from important ore bodies. with airborne surveys where only the scalar magnitude of the
While this may have been the initial hope in developing the total field has ever been measured from the early postwar days
aeromagnetic survey method, experience has shown that the of the airborne fluxgate. Only relatively recently has the addi-
real value of aeromagnetic surveying is its role in building up a tion of a second or third magnetometer sensor been exploited
picture of the regional geology. From this picture, and knowl- to obtain magnetic gradient estimates either vertically, along
edge of the types of geological environments in which certain track, or across track. Any possible theoretical advantages of
types of economic minerals occur, selections can be made of making vector (as opposed to scalar) measurements in explo-
limited areas in which more detailed and expensive investiga- ration practice have, even until the present, been outweighed
tions (such as ground geochemistry and drilling) may be by the additional technical complexity that would be entailed
worthwhile in order to isolate ore bodies of potential eco- when working from a moving platform.
nomic value. Even so, at a detailed scale of magnetic surveying In the 1980s, with the advent of inexpensive and portable
on the ground surface, it is often the case that magnetic computer systems, ground magnetometry in which the opera-
minerals occur in association with economic minerals tor had little to do except walk, push a button, and note down a
even where the latter are themselves nonmagnetic. In such reading became automated into integrated systems in which
cases, magnetic anomalies may serve as pathfinders to the x- and y-coordinates of each observation point and the time
economic targets and, in any case, provide an inexpensive of the observation (for temporal variation corrections) were
way of tracing known economic occurrences along strike, for recorded directly in computer memory for later retrieval, cor-
example. rection with time-synchronized base-station readings, and
plotting as contour maps or grayscale images. Eventually,
GPS readings became accurate enough that position could
5.04.4.3 Instrumentation Applied to Magnetic Surveying
also be recorded directly (rather than the observer following
5.04.4.3.1 Ground surveys presurveyed lines) and, in open terrain and small local areas,
The earliest ground magnetometer surveys used sensitive dip the use of multiple sensors could be employed to collect more
needles, the deflections of which were sufficient to detect large than one line of observations with each passage of the observer
deposits of magnetic iron ore. More sensitive mechanical or ground survey vehicle. Quad bikes, mountain bikes, and
instruments evolved, of which the Schmidt-type magnetic many other types of transport suited to the terrain being sur-
field balance became almost ubiquitous in mineral and petro- veyed have been pressed into service for this purpose. For
leum exploration in the 1940s and 1950s, prior to the preva- archaeological, unexploded ordinance and engineering site
lence of electronic magnetometers. They were designed to investigation purposes, the readings may be as little as 50 cm
measure the vertical (more rarely the horizontal) component apart, giving resolution at a submeter scale where the sources
of the geomagnetic field. The necessity to set up a tripod and are buried within the uppermost layers of the subsurface. Apart
level the instrument meant that progress was slow and surveys from ferrous metal objects that are magnetic, bricks and the
were consequently costly, despite a poor sensitivity of only a clay in the immediate vicinity of open fireplaces will have
few tens of nT. acquired a magnetization that is detectable in such surveys
The invention of the fluxgate magnetometer (Section and the method therefore offers considerable economy of
5.04.2.1) in the 1940s marked the arrival of the electronic effort in the excavation of sites of potential archaeological
age in magnetometry. Initially designed for airborne applica- interest. Figure 20 shows, by way of example, the magnetic
tion, it was some time before the electronics were sufficiently anomaly image over a settlement dated at 800–600 BC where
lightweight and compact for hand-carried operations. The flux- ancient excavations of the subsoil to build primitive habita-
gate element, if gimbaled to hang vertically in a damping tions have left their mark in the details of the local magnetic
medium such as oil, could measure the vertical component field.
of the magnetic field with a sensitivity sufficient to detect
variations of a few tens of nT. Instruments such as the Jalander 5.04.4.3.2 Airborne survey techniques
weighed only a few kilos and could be read in seconds so that Airborne magnetic surveying has its origins in the Second
long traverses (several kilometers) with frequent readings World War in more ways than one. The necessary fluxgate
(every 5–50 m) could be accomplished each working day, magnetometer technology was originally developed for mili-
even when combined with suitable survey practice to monitor tary applications such as submarine detection. But the war’s
nT
5.0
3.0
1.0
-1.0
-3.0
-5.0
50 m -7.0
Figure 20 A carefully executed ground magnetometer survey over an archaeological site in Siberia. The anomalies reveal the locations of excavations
for dwellings and ditches dated at 800–600 BC. The grid squares on the figure are of dimension 40 m, and magnetic anomaly readings were
made on a grid of 50 25 cm (Becker and Fassbinder, 1999).
legacy of aircraft and flying skills also contributed to a favor- poor or nonexistent topographic mapping) that early satellite
able environment for the application of the airborne magne- imagery (1970s) was able to play a significant role as base
tometer to exploration in peacetime. Early case histories were maps in many surveys. Offshore, the conventional application
documented during the late 1940s (Reford and Sumner, 1964) of a downward-looking 35 mm camera to record the flight
as early fluxgate magnetometers underwent postwar refine- path was of little use and necessitated pressing various elec-
ment to improve their resolution and efficiency. tronic navigation systems such as inertial navigation and
PPMs (Section 5.04.2.2) made their appearance in the late Doppler navigators into service. A summary of these tech-
1950s and early adaptations to airborne use and to digital niques, written at the time when they were about to be replaced
recording of the magnetic field readings are evident. Even so, by GPS, is given by Bullock and Barritt (1989).
the fluxgate magnetometer proved capable of further refine- To achieve good resolution of closely spaced near-surface
ment and continued in service for many years. OPMs (Section geological sources, the airborne magnetometer needs to be
5.04.2.2) first came into airborne service in the early 1960s, flown at a low ground clearance, typically 60 or 80 m these
but they did not seriously replace earlier types until the late days, and often only 150 m, even in the early years. For a
1980s when expiry of the original patent led to their more similar reason, close spacing of the parallel survey lines in the
general application. Since then, they have become almost grid of observations is essential if small features of geological
ubiquitous on account of their high sensitivity and fast reading importance are not to be missed. (Kimberlite pipes containing
rate. Helium, rubidium, cesium, and potassium types have all economic concentrations of diamonds, e.g., may be no more
been used, but cesium vapor types seem now almost ubiqui- than a hundred meters in diameter.) In the days of early
tous in the industry. SQUID-type magnetometers (Section reconnaissance work, the line spacing was often one kilometer
5.04.2.1) with vector capability are under development but or more, but in modern surveys, particularly over the hard rock
so far only approach the overall sensitivity of standard systems terrain typical of mineral exploration work, this is now com-
based on the cesium vapor instrument. monly reduced to 400 or even 200 m. Worldwide, many mil-
lions of kilometers have been flown already in the attempt to
achieve appropriate coverage of the geology of all continents.
5.04.4.4 Design and Execution of Surveys
The national coverage of Australia, for example, now includes
5.04.4.4.1 Practical airborne magnetometry about 28 million line-km of data gathered over almost 60 years
The development of suitable magnetometers is only part of the (Geoscience Australia, website).
story of aeromagnetic surveying. Many survey practicalities Typical survey aircraft fly at about 250 km h1 and operate
need to be addressed before the profiles of readings can be for periods of 4–6 h each day, typically recording 20 000 line-
combined into a contour map – a hand-drafted map being the km of data per month of operation.
end product of early surveys (pre-1975 approximately) after Magnetometer readings are gathered at appropriately timed
several manually intensive intermediate steps. These steps intervals – 1 s for the early proton magnetometer (equivalent
included the planning and execution of an appropriately reg- to about 70 m on the ground) and more recently 10 or even
ular grid of parallel flight lines and their recovery in map form. 100 times per second for modern cesium vapor instruments.
In many frontier areas, this was sufficiently challenging (due to Clearly, there is a need for the electronic data gathering that
made its appearance in the 1970s, first with dedicated custom-

built systems and more recently with standard laptop com-
puters and commercially available software.
Output of survey results in computer-generated anomaly
contours, as opposed to contours hand-drafted from analog
paper-chart records generated during each flight, gradually
became the norm in the late 1970s. Almost simultaneously,
new computer software, originally developed for use with dig-
ital satellite imagery, was adapted for the display of aeromag-
netic anomaly datasets as gray scale and color maps using, for
example, shaded-relief effects to emphasize the geological con-
tent of the anomaly datasets (e.g., Figure 19) (O’Sullivan, Figure 21 An Aero Commander Shrike aircraft fitted with a
1991). Individual aeromagnetic survey projects with more magnetometer in a tail stinger for aeromagnetic surveying. Courtesy of
than 1 million line-km of data have been carried out in recent Geoscience Australia.
years with the latest technology in Africa and Arabia. In Europe
(outside of Scandinavia) and North America, generally older
coverage to a lower standard is the norm, except where mount the magnetometer sensor as far away as possible from
upgraded for oil exploration – or, quite rarely now in Europe such effects, usually in a fairing attached to the tail or wingtips
at least, for mineral exploration purposes. Images of outstand- (Figure 21) and backing out as much as possible of any per-
ing quality and detail at scales compatible with regional manent magnetic effects.
geological maps (say 1:250 000) are now published routinely Magnetic effects due to the varying attitude of the aircraft in
from these surveys and often made freely available via the the geomagnetic field (the so-called maneuver noise) have
Internet. been greatly reduced by the development of the so-called auto-
Gamma-ray spectrometers are usually flown simulta- matic compensators. In calibration mode, such equipment
neously to add information streams on the abundances of records the variations in magnetometer reading as the aircraft
thorium, uranium, and potassium to the graphic output of executes a predetermined series of maneuvers in pitch, roll, and
many surveys. Airborne gravity anomaly mapping is some- yaw at high altitude (distant from geological sources) while
times carried out simultaneously with aeromagnetic readings. flying on each of the cardinal compass directions. When
This is slowly gaining ground as it is considerably more expen- switched to the survey mode, the instrument then automati-
sive, making it attractive only in rather limited situations such cally applies in real time the appropriate correction for the
as frontier oil exploration areas with difficult ground access actual attitude of the aircraft at any moment. A ‘figure of
(Reeves and Bullock, 2006). merit’ – the sum of the 12 terms, three maneuvers in each of
Arguably, the biggest single innovation in more than four cardinal compass directions – is an assessment of the
50 years of airborne magnetometry was the introduction of performance of the airborne system and is commonly now as
the GPS in the early 1990s. So many of the tedious and labor- low as about 1 nT.
intensive aspects of previous survey practice centered around From the point of view of revealing geological detail, a
navigation of the survey aircraft as closely as possible along major achievement in airborne survey practice has been the
predetermined lines and, subsequently, the recovery of the reduction of noise from all sources on the magnetic anomaly
flight path actually flown. Even after the arrival of the digital profile to as little as 0.1 nT. Since the interest for geological
era in the 1970s, this part of the process remained primarily mapping is not confined only to the most magnetic rocks, even
manual, up to the point where the flight path – recovered the most subtle anomalies are of potential value. So reducing
manually using the 35 mm strip films exposed during flight – the noise level lowers the threshold of what is detectable in
was digitized and merged with the digitally acquired geophys- anomaly mapping generally.
ical data. GPS provided quicker, better, and cheaper solutions Noise arises from various sources. Maneuver noise is dis-
to all these problems – in real time to an accuracy of about 5 m cussed above. The elimination of time variations in the geo-
after the introduction of differential GPS. The regularity of the magnetic field itself is achieved largely by monitoring temporal
pattern of flight lines so achieved was an added bonus to the variations at a fixed base station during survey flying. The limit
quality of the survey data, quite apart from the accuracy with to this is set by the fact that geomagnetic micropulsations vary
which individual features could be recovered on the ground. detectably from place to place, even over a few tens of kilome-
In addition, GPS facilitated the production of digital elevation ters, so the records of a single fixed base station do not perfectly
models of the surveyed area, thanks to the combination of GPS represent the temporal variations experienced by the magne-
(geocentric) heights for the aircraft and radar altimeter values tometer in the aircraft throughout a survey area, parts of which
for the aircraft’s ground clearance. are typically hundreds of kilometers distant. Nevertheless, the
time-synchronized subtraction of total field readings at the
5.04.4.4.2 Elimination of nongeological influences (aircraft fixed base-station magnetometer from the readings made in
and temporal changes) the surveying aircraft eliminates the bulk of the temporal var-
A reading of a magnetometer on board an aircraft must clearly iations. Corrections of several tens of nT are made typically in
be corrected for any effect of the aircraft itself – primarily the this way. The base-station magnetometer should be sited in a
engines and electric currents, since most other parts of aircraft magnetically quiet locality and its output carefully filtered to
are now of nonferrous materials. The first-line approach is to avoid the addition of (negative) noise when subtracted from
the airborne profiles. On days when geomagnetic activity is Positional accuracy for the aircraft of 5 m is achievable
high (magnetic storms), survey operations must be suspended. routinely with differential GPS, meaning that the location of
The magnetic base stations are rarely tied to a geomagnetic crossover points is well determined, but it may still be prudent
observatory. While this would be ideal, the remoteness of to exclude some crossovers from the error analysis where
many survey areas dictates that the additional logistic effort exceptionally high magnetic gradients are noted and errors of
required and the low-commercial value of the absolute back- a few nT may arise simply through a slight mispositioning of
ground level in any single survey (as opposed to the anoma- the crossover.
lies) usually make this prohibitive. As a result, tying together a With all these precautions, flight lines executed repeatedly
patchwork of hundreds of independently acquired surveys into over terrain with only very subtle anomalies show repeatability
a consistent national system with a credible datum requires to about 0.1 nT. This is comparable with the noise level on
some innovative approaches (Tarlowski et al., 1996). National the profiles in general in modern data.
grids of regional survey lines have been adopted (e.g., in At the end of the data-reduction and processing cycle, maps
Australia) to achieve the highest standards of accuracy for and images revealing geological detail on the scale of a 50 m
long wavelength features in recent years (Milligan et al., 2009). pixel are produced for interpretation. Commonly, these maps
The imperfections in the base-station subtraction process and digital datasets are published by national government
are almost always reduced further by flying surveys with a agencies as part of a nation’s geological mapping program,
number of control lines or tie lines at right angles to the and compilations of such maps over whole continents will,
main survey line direction. These are spaced such that an eventually, provide valuable new insights into the, often hid-
aircraft flying a survey line crosses a control line every few den, geology of all the continental areas of the world. Thanks
minutes – sufficiently often that departures from linearity in to modern communications, Internet, and software technol-
temporal variations are not severe. Once the aircraft magne- ogy, these datasets may be exchanged between organizations
tometer and compensator are properly calibrated, any differ- and clients worldwide, ultimately meaning that exploration
ence between the two magnetometer readings (flight line and decisions may be based on access to a maximum of factual
tie line) at each intersection point should be due only to x–y registered information from which intelligent interpreta-
temporal geomagnetic variations or, to be more precise, the tions of exploration priorities may be made.
imperfections of the base-station subtraction procedure in Over the world’s oceans, the magnetic anomaly data are
removing them. An iterative adjustment of each of the survey mostly of shipborne origin and lack the quality and resolution
lines with low-order polynomial corrections to minimize these of airborne surveys since marine research vessels travel com-
differences at all the hundreds or thousands of intersections in paratively slowly, remotely from good monitoring of temporal
a survey usually makes on average a few nT change to the near- variations, and seldom execute a regular pattern of closely
DC level of each of the profiles. spaced lines. A wealth of shipborne magnetometer data still
The prevalent use of image-processing procedures for the await inclusion in rational worldwide databases (Reeves et al.,
presentation of survey data as images (Section 5.04.4.5) 1998). Nevertheless, the mapping of oceanic magnetic anom-
means that any remaining line-related noise may still be visible alies has played a key role in establishing our understanding of
in certain image presentations. This residual can be reduced by the way in which oceans grow at midocean ridges with conse-
applying a type of directional filtering procedure often called quent separation of the once-adjacent continents and the
microleveling (Minty, 1992). Adjustments at this stage are at improved appreciation of geological evolution that came
the level of a few tenths of a nT, but still as large as, or larger with global tectonics. Over continental shelf areas critical for
than, the noise level on each of the profiles. This is essentially a petroleum exploration, airborne surveys will often be executed
cosmetic procedure, but it does significantly improve the to the same critical standard as described in the preceding text
signal-to-noise ratio in the final data when viewed as an image. for land areas.
Exposing geological detail requires a minimum separation
between airborne magnetometer and magnetic sources com-
5.04.4.5 Data Presentation, Enhancement, and
mensurate with operational safety of the aircraft. This leads to
Interpretation Methods
surveys with a terrain clearance of typically 60 or 80 m. This is
achievable with safety in areas of subdued topography but Since the data capture and processing stream of modern mag-
needs to be adjusted in more rugged areas or around isolated netic anomaly surveys is entirely digital, the final product is
hillocks due to the limited climbing capacity of any fixed-wing also a digital data file or database in which all the collected data
aircraft. Flying to a predetermined ‘drape’ on the topography are preserved and organized sequentially by flight lines and
means that similar altitudes are achieved at all the intersections where as many as necessary of the intermediate corrections are
of flight lines with tie lines in the survey, adding further preci- recorded against each reading. Assuming the data reduction
sion to the data-reduction process in exchange for a small loss and processing has been done correctly, a final corrected
of geological resolution where terrain clearance needs to be anomaly value will exist in this database for each of many
greater than the minimum. millions of observation points. For a large survey, this can
The free-air gradient of the undisturbed geomagnetic field involve gigabytes of data. Accessing small windows and indi-
with height is generally small (0.01–0.03 nT m1), meaning vidual profiles from such a database requires dedicated soft-
that height differences at intersections need to be 10 m or more ware if it is to be done conveniently and routinely. Commercial
for differences in magnetic values due to this gradient to exceed packages have been developed to enable users to do this on
the noise envelope in the profiles. This is comparable with the laptop PCs (e.g., Geosoft, website; Intrepid Geophysics,
height discipline normally achieved in survey operations. website).
For the human interpreter, the first priority is to visualize the data or to enhance certain types of feature, with a direc-
the data (Isles and Rankin, 2013). This may be done on a tional bias, for example. Calculating gradients across strike will
profile-by-profile basis to inspect individual anomalies, but clearly enhance features striking in one direction, while the
synoptic overviews of an entire survey are also necessary and same features may be largely subdued by calculating the gradi-
best obtained by means of graphic displays of maps and ent along strike. This is analogous to choosing a direction for
images. The original method of display of magnetic anomaly illumination in the shaded-relief map. Clearly, many different
data was contour maps. With the advent of personal comput- options are available.
ing power in the 1980s, contour displays have been largely A further suite of opportunities for data processing arises
replaced by images in which the anomaly values are converted from the Laplacian nature of the magnetic anomaly field and
to a gray scale or a color scale (Milligan et al., 1991; Reeves the ability to apply spectral analysis to a survey (Spector and
et al., 1997). The first step in this process is to interpolate the Grant, 1970). Measurements of any potential field on one plane
line-based data onto a regular raster or grid of values over the (if it is done thoroughly at all wavelengths, as is attempted in
survey area. Typically, the size of the grid element (pixel) is aeromagnetic surveys) are sufficient to calculate the field that
one-quarter to one-fifth of the flight-line spacing, so com- would be observed on any different plane, permitting what is
monly 50–100 m for most surveys. Such a grid cannot repre- called upward and downward continuation of the data. Conve-
sent the full information content of the original profiles along niently, this is done by way of a Fourier transform of the data into
line (with samples at 7 m for a 0.1 s sampling interval) and the wavenumber domain where upward and downward contin-
antialias filtering is needed to avoid the smallest anomalies uation operators are relatively simple filter functions that may be
being represented as broader ones in the grid. Across lines, applied to the wavenumber and phase spectra. Inverse Fourier
there is often an undersampling problem since the line spac- transformation restores the filtered result into a new grid in the
ing, for cost reasons, often does not meet the strictest sampling space domain for display using any of the previously mentioned
criteria. This can be solved in most cases by interpolation along grayscale or color raster techniques.
the predominant strike direction of the geology where anom- Similar operations may be performed for calculating gradi-
alies change most slowly from one flight line to the next. ents in the data (vertical or in any chosen horizontal direction)
Once available in grid format, many methods of visualization and, by calculating suitable low-pass and high-pass filters based
of the gridded data are available (Milligan et al., 1991; Reeves on the analysis of the wavenumber spectrum, regional and
et al., 1997). A grayscale representation may be the simplest of residual components of the anomaly field may be separated out.
these, with the highest magnetic values portrayed as white and The shape of a typical anomaly depends in part on
the lowest as black (or vice versa) with a suitably stretched gray the magnetic inclination at the survey locality (Figure 18).
scale between. Often more satisfactory is a grayscale shaded-relief Through adjustments to the phase spectrum in the Fourier
image in which the ‘topography’ of the anomaly field is dis- domain, anomalies may be transformed into the simple posi-
played as though illuminated by a light source in a given directive shape typical of bodies at the geomagnetic poles, bringing
tion (azimuth and altitude). Diffuse reflection from the surface the positive anomaly above its source body. This process is
of the magnetic topography is calculated such that slopes facing called migration (or reduction) to the pole and is often con-
the illumination source appear brighter than those sloping away sidered an advantage in interpretation, though it is not without
from it (Figure 19). This gives the human eye a clear impression some problems at low magnetic inclinations where north–
of the undulations in the anomaly field and has the added south trending bodies have very small amplitude anomalies
benefit that the average magnetic anomaly background level in theory. The desired result assumes that source bodies are
(which may be ill-determined and of little value in studying the vertical and that remanent magnetization plays an insignificant
near-surface geology) is effectively lost among the average gray role, neither of which is totally correct. Generally speaking, the
levels of the spectrum. results are nevertheless useful. The persistence of occasional
Color may also be used for display purposes, usually with a negative anomalies in a map processed in this way is not
natural color spectrum and with the highest magnetic values in disastrous and simply signals that one or both of these two
red and the lowest in blue. Such a presentation may even be assumptions are invalid locally.
combined with the grayscale shaded-relief effect, or the color The approach may be taken a step further by assuming the
saturations may be adjusted to give the eye an optimum geology to be made up of a raster of vertical prisms of size equal
impression of the pseudo-topography (Isles and Rankin, to the grid-cell size of the data and calculating the equivalent
2013; Milligan et al., 1991). magnetic susceptibility necessary for each prism to produce the
All these techniques evolved quickly in the early 1990s and observed magnetic anomalies. This, again, is seldom perfect in
have become universally accessible to users working with com- its application but is reasonably valid in many metamorphic
mercial software packages. Display of the aeromagnetic data terrains such as Precambrian shields. It can form an important
geographically registered with, for example, the preexisting step in the interpretation process that may be seen as attempt-
geological map (or any other geophysical or geochemical data- ing to go from the continuous Laplacian magnetic anomaly
set), is then a powerful tool for the interpreter to use, and this is field to the discontinuous geology where changes in lithology
readily available in the geographic information systems (GISs) and, for example, faults bring different rock types with differ-
that are in common use with individuals and groups working ent magnetic susceptibilities into direct contact.
with x,y-referenced geoscience data. Other techniques used routinely to process data and produce
Once the data are in a digital, gridded format, these are also outputs that help the extraction of geological information include
readily amenable to digital filtering processes that may be the calculation and plotting of Euler solutions that indicate the
operated in the space domain. These may be useful to smooth depth of source bodies based on the curvature of their anomalies
(Reid et al., 1990) and the calculation of analytic signal expres- third dimension of the geology that does not appear on con-
sions (Roest et al., 1992). The latter are the positive envelope to ventional geological maps.
the curve family shown in Figure 18 and so are independent of
magnetic inclination. Positive analytic signal values are also to be
5.04.4.6 The Link Between Magnetic Properties of Rocks
found over faults and contacts, however, so they do not neces-
and Regional Geology
sarily highlight only the main magnetic bodies in an area,
though in experienced hands, they do have considerable value. The fact that magnetic properties are retained at all tempera-
Despite all these modes of assistance, the translation of the tures below the Curie point means that virtually, the whole of
aeromagnetic anomalies into a geological map – interpretation the Earth’s crust contributes to magnetic anomalies and that it
sensu stricto – remains a largely intellect-driven endeavor at is the bulk of these igneous and metamorphic rocks that is
present. This is based on the physical constraints surrounding represented in the anomaly patterns, though in general, the
the origin of anomalies and the integration of quantitative effects of the shallower rocks will predominate over the dee-
interpretation results with more qualitative approaches and per ones. The fact that the wavelength of anomalies increases
the constraints provided by other data such as the preexisting with increasing vertical distance between source and magne-
geological map and other layers of geophysical (gamma-ray tometer means that processes of wavelength-based filtering
spectrometry and gravity) or geochemical information. An (e.g., after spectral analysis of anomaly patterns; see previous
illustration of magnetic anomalies over an area of Precambrian section) can be used to emphasize sources at different depths.
shield in Western Australia and their interpretation at the Often, for geological mapping, it is the shallowest sources
hands of a skilled geologist is given in Figure 22. An under- that are of most interest. These can be enhanced (and the
standing of the magnetic properties of rocks in general (next deeper sources suppressed) by high-pass filtering or the cal-
section) undoubtedly plays an important part in this process, culation of a residual or a vertical derivative field from the
though information in any given area on this latter point is observed field data (see Section 5.04.4.5). Vertical gradient
almost always incomplete. While a geological map shows the magnetometers similarly are more sensitive to the effects of
formations that outcrop, or at least suboutcrop below soils and shallow sources, so vertical gradient surveys (or the calcula-
overburden, the magnetic anomaly pattern also contains infor- tion of vertical gradients from a grids of values in the x,y
mation from more deeply buried geological sources. An inter- plane) tend to reveal the effects of near-surface rocks more
pretation map often, therefore, has to include elements of this clearly than total field surveys. Even quite magnetic rocks
Figure 22 (a) An area of the composite aeromagnetic anomaly map of Western Australia and (b) its geological interpretation. Courtesy of Geoscience
Australia and the Geological Survey of Western Australia.
disposed in thin, flat-lying layers at or near the Earth’s surface,

such as lava flows, have relatively little effect in comparison
with the ‘basement’ when it comes to geological mapping,
however. So the ability to map underlying crystalline rocks
below whatever surface formations is well entrenched in the
aeromagnetic method.
In the interpretation of anomalies, some understanding of
the magnetic properties of rocks in general is an essential part
of deriving useful solutions to the mathematically underdeter-
mined problem of inverting anomaly patterns in two dimen-
sions into models – even simplistic ones – of the subsurface in
three dimensions. While the geometric parameters of buried
rock bodies in a given area may be unknown, the magnetic
properties of representative rock types may be determined
independently in the field or laboratory. Experience gained
from such measurements on common rock types around the
world may be helpful, even in an area where the local geology
is essentially unknown.
Magnetic properties reside mainly – but not exclusively – in
the magnetite grains that rocks contain. Magnetite content is,
unfortunately, largely independent of the abundance of the
main minerals that determine rock lithology as described by
the geologist and petrologist. These basement rocks display a
wide range of magnetic properties that are not readily summa-
rized by simple rules that can be used to deduce rock types
from their anomalies (Grant, 1985). Figure 23 A grayscale shaded-relief image of a small area of Western
Apart from magnetite and the mixed oxides of iron and Australia showing the magnetic expression of several typical geological
titanium, only the sulfide mineral pyrrhotite contributes to units. A, granite pluton with highly magnetized rim; B and B0 , dyke with
magnetic anomalies. At low concentrations, a fairly simple, normal magnetization; C and C0 , dyke with reversed magnetization, D:
granite; E and E0 , greenstones. Courtesy of Geoscience Australia.
monotonic relationship is found between magnetite (or pyr-
rhotite) content and the magnetic susceptibility of a rock.
However, there is little relationship between magnetite content
and the gross mineralogy of rocks and hence their more con- These evident difficulties notwithstanding, there appears to
ventional petrologic classification. It is more useful to relate the be no letup in the application of aeromagnetic anomaly map-
anomaly patterns to geological units known in the area, or ping as a method of extending geological understanding of our
known from their typical expressions elsewhere. For example, continents and their resources toward a more universal cover-
long rather straight anomalies cutting through all other pat- age of magnetic anomaly maps at global scale. The first map of
terns are typical of dykes intruded through cracks or fissures this nature was published in 2007 (Korhonen et al., 2007) and
created in brittle country rock (Figure 23). Similarly, an oval the data are now being continuously upgraded (Maus
area of anomalies with a texture distinctly different from the et al., 2009).
surrounding rocks is typical of a granite diapir emplaced long
after the surrounding rock solidified.
5.04.4.7 Satellites and Space Probes
While the direct relation of magnetic anomalies to specific
igneous and metamorphic rock types is not possible in inter- The first satellite to carry a magnetometer into the Earth’s orbit
pretational exercises (except where independent evidence is was the Soviet Sputnik 3 in 1958. Since then, most satellites
available from ground studies), the fact that rock types have a and space probes have carried magnetometers of some sort.
bimodal distribution of susceptibility with the two popula- Early US missions were restricted by low payload requirements
tions separated by at least two orders of magnitude and used a single search coil design. Later on, single fluxgates
(Figure 24) means that many lithologic boundaries in such were employed, and by varying the fluxgate orientation,
geological terrains are coincident with magnetic property three-component vector measurements were possible. The
boundaries. Hence, even where two adjacent rock types are astronauts of the Apollo 12, 15, and 16 missions made vector
both unknown, their contact can often still be traced through measurements on the lunar surface, Apollo 16 using the newly
aeromagnetic mapping. Magnetic anomaly patterns therefore developed ring-core fluxgate.
provide a wealth of information on the general disposition of Relatively few satellites have been put into orbit specifically
rock types in an area – their fold patterns, the brittle faults to carry out high-resolution magnetic surveys from space.
separating terranes with different tectonic histories, and the MAGSAT (Langel et al., 1982) carried a purpose-built fluxgate,
intrusion of geologically younger features (such as dykes) together with a cesium vapor scalar instrument to calibrate the
that crosscut the older rocks. They have therefore become an intensity while in space. It took measurements during 1979
essential part of systematic and efficient geological mapping and 1980. Twenty years later, the Danish satellite ersted,
and mineral exploration in most parts of the world. launched in 1999, carried a three-component fluxgate
-1
Log10 Magnetic susceptibility (SI)

0.3 ‰
3‰ 1‰
6‰
-2 0.1 ‰
-3
-4
10 ‰
-5
2.50 2.65 2.80 2.95 3.10 3.25
Density (g cm-3)
Figure 24 A frequency distribution plot of magnetic susceptibility against density for almost 30 000 rock specimens from northern Scandinavia shows
a bimodal distribution between two populations of rocks with a difference in magnetic susceptibility of almost two orders of magnitude – effectively
‘magnetic’ and ‘nonmagnetic’ in terms of magnetic anomaly mapping. Note that, while some increase in susceptibility with density (and the higher
proportion of mafic minerals) is evident, many highly magnetic rocks are of low density and must, therefore, contain few of the dark minerals
found in mafic rocks. Reproduced from Henkel H (1991) Petrophysical properties (density and magnetisation) of rocks from the northern part of the
Baltic shield. Tectonophysics 192: 1–19.
mounted alongside a star camera for absolute orientation and whether Earth has always possessed a magnetic field, much
an Overhauser effect proton magnetometer. However, a better longer records, extending back into the geological past, are
resolution and better spatial and temporal coverage were required.
achieved by CHAMP (Challenging Minisatellite Payload), Paleomagnetism provides such records through the rema-
launched in July 2000 and managed by GFZ, Potsdam nent magnetization of natural materials and artifacts for which
(CHAMP, website). CHAMP was in a low-altitude, near-polar the age of the magnetization can be reliably estimated by
orbit, which made it ideal for high-resolution gravity and independent means.
magnetic measurements. The magnetometers carried by Most rocks and sediments contain small concentrations
CHAMP were similar to ersted’s, but the low-Earth orbit of ferrimagnetic minerals, for example, magnetite and hema-
and improved GPS technology resulted in much better imaging tite. The two main primary processes by which these can
of crustal anomalies over a period of almost 10 years. This attain a stable natural remanent magnetization parallel to
duration meant it was possible to begin to investigate short- the ambient geomagnetic field are thermoremanent magneti-
period features of the geomagnetic secular variation: sudden zation (TRM) and detrital remanent magnetization (DRM).
changes such as jerks and westward drift (Reigber et al., 2005). TRM is acquired as a lava, volcanic, or igneous rock, a fired
pot, or kiln cools; as magnetic grains pass through their Curie
temperatures, they acquire magnetic moments, and these sta-
5.04.5 Paleomagnetic Methods bilize on further cooling – through an often well-defined,
blocking temperature. At ambient temperatures, thermal
Direct measurements of the direction of the geomagnetic field relaxation times can be of geological duration. Theory and
exist only for the past 400 years, while measurements of inten- experiment show that the intensity of a TRM or a partial TRM
sity date back only 200 years. However, these records indicate is directly proportional to the intensity of the magnetizing
that many of the interesting features of the geomagnetic field field. Hence, it is, in principle, possible to retrieve informa-
operate on much longer timescales. To gain a complete under- tion on the paleointensity of the geomagnetic field by nor-
standing of the geomagnetic field, or even to be able to say malizing a TRM. Delesse (1849) and Melloni (1853) were the
first to study TRM in volcanic rocks. Reversely magnetized 5.04.5.1.2 Consolidated rocks
lavas and baked contacts were reported by David (1904) The most popular method of sampling consolidated rocks is by
and Brunhes (1906), sparking the suggestion of polarity the use of a high-speed drill with a diamond matrix-tipped drill
reversals, while Chevallier (1925) published remanent direc- stem, usually 2.5 cm in internal diameter and usually water-
tions from lava flows on Mt. Etna that showed evidence of cooled. Various models have been designed: modified chain-
secular variation. saw motors, petrol-fuelled, and modified electric drills, used
By contrast, in DRM, the grains already carry stable mag- either with rechargeable batteries or a portable generator.
netic moments, and, on the average, they align with the ambi- A petrol-fuelled drill, with pressurized water cooling, is
ent magnetic field during depositional and postdepositional shown in Figure 25(a). Outcrops are often remote and access
processes. DRM occurs in sediments, sedimentary rocks, and is frequently difficult or awkward, so portability is important,
sometimes in windblown deposits or loess and is generally, as is the availability of water.
though not always, weaker than TRM. Retrieval of the paleoin- Drilled samples are usually between 5 and 15 cm in
tensity of the geomagnetic field from a DRM is complicated by length, depending on the hardness and structural integrity
the many sedimentologic factors that affect the magnetization of the rock. Before removal from the outcrop, the sample
process: however, it is possible to obtain records of relative must be oriented with respect to north and the vertical. This
changes in paleointensity from sequences of sediments that are is usually achieved by measuring the plunge below the
extremely uniform in lithology and grain size. vertical and the azimuth of the core axis (or a pair of angles
The accuracy and age control of paleomagnetic records are simply related to these), with an inclinometer and a compass
very variable. In favorable situations, an accuracy of a few degrees (Figure 25(b)). A fiducial line is drawn lengthwise along the
can be attained in direction, but 10–20% is usually regarded as upper edge of the sample, the front (outside) face marked and
good for absolute paleointensity determinations. Whereas the sample clearly labeled following laboratory conventions
archaeological artifacts can often be dated to within decades or (Figure 25(c)). Every laboratory has its own set of conven-
even years, radiogenic isotope methods are most commonly tions for the orientation and labeling of samples, which have
used for natural materials, and uncertainties of a few percent of usually been carefully worked out over many field seasons.
the age are common. The sequential ordering of sedimentary Weakly magnetized rocks can usually be oriented with a
deposits and sedimentary features such as climate-induced cyclic geological magnetic compass; however, some basalts and
structures that can be tied to astronomical cycles can also provide other strongly magnetized rocks can influence the compass
invaluable constraints in dating sedimentary records in the past needle, and in such cases, a sun compass is preferable. If a
few million years. The fact that the main geomagnetic field has magnetic compass is used, the local magnetic variation (dec-
reversed its polarity at irregular intervals throughout geological lination) must be known, so a correction to geographic north
time more generally means that it can be used as the basis of a can be made. When a sun compass is used, the location, date,
unique and global timescale of immense value in dating and time of each measurement must be recorded, so that the
geological events. azimuth can subsequently be calculated from tabulated or
Natural materials may also be magnetized or remagnetized programmed positions of the sun.
through reheating, recrystallization, diagenetic, and other sec- Each drilled sample is normally cut into several 2.2 cm long
ondary physical or chemical changes. (Natural magnetization specimens for NRM measurement (Figure 25(d)).
processes and paleointensities are discussed in detail in An alternative sampling method is to remove oriented
Chapters 1.08 and 1.15, respectively.) blocks of rock by hand or by breaking from the exposure
with a hammer and to subsequently drill samples from this
block in the laboratory or workshop. The block is oriented
in the field by marking strike and dip lines on a flat surface
5.04.5.1 Sampling
and orienting these lines with an inclinometer and com-
5.04.5.1.1 Introduction pass. One advantage of this method is that the outcrop
High-resolution rock magnetometers (see Section 5.04.5.2) is not left with an unsightly array of holes. However, it is
are precision instruments, housed and operated in modern often difficult to find suitable outcrops, as it is difficult to
sophisticated laboratories. Natural and archaeological mate- remove block samples from rock that is completely firm,
rials must therefore be carefully sampled and brought to the intact, and devoid of joints or cracks. Further, accurate
paleomagnetic laboratory for measurement. A crucial part of orientation is much harder to achieve than with a
the sampling procedure is the recording of the orientation of drilled core.
the sample while in situ, that is, before removal from the Many unconsolidated sediments and soft materials such
outcrop, exposure, or site. Without such orientation, the mag- as ignimbrites and loess are too fragile to remain intact and
netization vector cannot be referred back to a geographic ref- keep their shape without being encased in some way.
erence frame. It is also important to establish whether the A popular practice with such materials is to carve a cube-
material is still in the orientation in which it was magnetized shaped sample and to cover it with a small plastic box (one
or, if not, whether a correction can be made for any distur- standard type is 8 cm3), which is then oriented in a
bance that has occurred since that time. In rocks, this usually manner modified from that described above for drill
means looking for an indicator of the paleohorizontal, for samples.
example, a bedding plane in sediments or possibly flow Most modern rock magnetometers are designed to accept
marks in lavas or ignimbrite flows, and careful measurement both 2.5 2.2 cm cylindrical samples and one of the standard-
of its attitude. sized cubes.
(b)
y
x
z Specimen A
(a)
z Specimen B
y
Sam
ple
ID
Specimen C
(c) (d)
Figure 25 (a) Paleomagnetic sampling using a water-cooled, petrol-fuelled drill with diamond-tipped drill stem (photo by S. Hüsing). (b) Before the
sample is broken loose, a line is marked on its exposed face and an orientation tool placed over it. A compass is used for azimuthal orientation of
weakly magnetized rocks; the inclinometer level and scale are obscured beneath the compass (photo by D. Michalk). (c) A method of marking a drill
sample once removed from the exposure; sample coordinate system. (d) Each sample is normally cut into a number of specimens.
5.04.5.1.3 Lake and deep-sea sediments The most appropriate method depends on the field environ-
Sedimentary sequences that represent periods of unbroken ment and the resolution required.
deposition often provide continuous records of geomagnetic In relatively calm environments and up to about 100 m
changes spanning thousands, hundreds of thousands, or even water depths, the Mackereth corer (Figure 26) has proved
millions of years. Continuous cores of lacustrine or deep-sea invaluable. Originally developed for geochemical studies
sediments may thus yield invaluable paleomagnetic informa- (Mackereth, 1958, 1969), versions of this corer have been
tion. Numerous coring devices have been designed for soft designed to collect single-drive cores between 1 and 18 m in
sediments. To obtain material suitable for paleomagnetic length, the most common being the 6 m corer. It is built entirely
work, core samplers should not twist, shear, or vibrate the from nonmagnetic materials and is operated pneumatically
sediment: a simple push mechanism is preferable. Any twisting from a small boat anchored near the coring site, using cylinders
or shearing of material will compromise the determination of of compressed air. Air is expanded above a piston (MP), forcing
declination, while vibrating may cause reorientation of mag- the core tube into the sediment in a single smooth drive; a
netic grains within the sediment. Corers should contain an second, fixed piston (KP) inside the core tube sits on top of
absolute minimum of ferrous materials: steel core barrels are the sediment throughout, preventing disturbance. At the com-
known to induce steep components of magnetization in drill pletion of the coring stroke, air is diverted into a large anchor
cores. There are very few devices that incorporate absolute chamber (AC), which, until this moment, has anchored the
orientation of the core sample or even relative orientation corer, but which now becomes buoyant, lifting the corer and
between consecutive drives or sections. Individual drives core to the surface. When properly deployed, a buoy at the water
should therefore be as long as possible, to minimize azimuthal surface and/or a float chamber near its upper end supports the
matching between sections and the possibility of lost material. corer, and the core tube penetrates the sediment vertically.
In deeper water or less clement conditions, gravity-driven

Boat
piston corers or mechanical drills are more commonly used.
In shallow water, corers pushed into the sediment on the
end of a system of connected rods are popular. This method
necessitates a stable platform: a well-anchored raft and a winter
platform of solid ice are two possible options. Such systems
and most drilling systems retrieve sections that are rarely more
C R P than 1.5 m long: necessitating matching to obtain declination
records.
The rotary corers traditionally used to core hard sediments
PV and igneous basement on the sea floor are not suitable for the
MP soft sediments that carry high-resolution paleomagnetic and
climatic records as they cause too much disturbance and shear-
ing of the material. The Ocean Drilling Program, which oper-
ates the ship JOIDES Resolution, which has drilled throughout
the world’s oceans for several decades, has developed an
Advanced Piston Corer to overcome the problem (Ocean Dril-
ling Program, website). The corer is dropped to the bottom of
the coring wire line and then operates in a manner similar to
that of the Mackereth corer. Once the corer is in place, safety
pins are broken by a burst of hydraulic pressure, which then
drives the 9.5 m inner core barrel into the sediment in a single
smooth stroke. The core section is brought to the surface for
recovery, the drill bit and bottom hole assembly driven a
further 9.5 m down, and the whole process repeated to recover
the next 9.5 m of sediment. The Advanced Piston Corer also
carries an orientation tool so each section can be oriented for
absolute paleomagnetic work. The Ocean Drilling Program
quotes a recovery rate of 1–4 sections per hour
(9.5–38.0 m h1), depending on water depth and the hardness
of the material cored.
5.04.5.1.4 Archaeological materials

OV Archaeomagnetism involves paleomagnetic work on archaeo-
logical materials, such as fired bricks, burnt clay from the floor
of an ancient hearth, or in situ cooking stones. It is usually
necessary to orient and remove samples carefully, causing min-
imal disturbance to the surrounding material or structure.
Often, cultural considerations dictate acceptable sampling
methods. Only in rare situations, it is possible to drill into a
solid structure using a conventional paleomagnetic rock drill.
KP
For relatively large artifacts that can be removed whole, such as
cooking stones, a common method is to make and level a cap
of plaster of paris on the object. The plaster is leveled using a
two-axis bubble level, and once set, it is marked with magnetic
north and/or a sun bearing (Figure 27). Back in the laboratory,
oriented specimens can be made by either drilling or sawing.
This is essentially the method used by Thellier in his early
AC
archaeomagnetic investigations. For smaller samples, a
2.5 cm diameter plastic disk may be fixed to the artifact with
epoxy and leveled, and for poorly consolidated material, a
2.5 cm plastic tube may be pushed into the material, oriented
before extraction, and sealed afterward (Clark et al., 1988).
Figure 26 The Mackereth lake sediment corer is operated sediment. The Kullenberg piston, KP, remains fixed with respect to the
pneumatically from a boat. Passing air down line P, through a one-way corer, minimizing disturbance to the uppermost sediment. At the
valve OV, and the open valve PV creates suction effect that pulls the completion of the coring stroke, air is diverted into AC, which becomes
anchor chamber AC into the sediment. Passing air down line C then buoyant, lifting the corer to the lake surface for retrieval. Air-line R allows
closes PV and pushes the main piston, MP, and core tube into the retraction of the core tube and emergency retrieval.
(a) (b)
Figure 27 Sampling for archaeomagnetic studies. (a) Cooking stones in an excavated hangi site, showing three stones selected for sampling;
(b) plaster of paris caps are molded on the tops of the selected stones, leveled and marked with magnetic north and a sun bearing before removal to
the laboratory for subsampling and measurements.
A comparison of these methods is given by Trapenese et al. of different positions with respect to the magnets, it is possible
(2008). to obtain all three components of its magnetic moment and,
hence, calculate its magnitude and direction. Blackett (1952)
detailed the design features, construction, and use of a highly
5.04.5.2 Rock Magnetometers sensitive astatic magnetometer, and this was then used in many
of the pioneering paleomagnetic studies of the 1950s.
5.04.5.2.1 Introduction Moderate-sensitivity astatic systems were relatively easy to
Since its early days in the 1950s, paleomagnetism has seen
set up in most laboratories. Such systems gave reliable results
three generations of rock magnetometers. The first astatic sys-
on standard-sized samples with magnetizations above about
tem was possibly a development of that used by Melloni
0.2 A m1: most TRM-bearing volcanic and igneous rocks.
(1853) in his pioneering investigation of the TRM of Vesuvian
However, their sensitivity was limited by vibration, field
lava flows. Astatic systems continued to be developed and used
gradients, and variability caused by moving objects such as
extensively for over 100 years, until post Second World War
lifts and vehicles.
developments in electronics led to the fluxgate, spinner
These factors were particularly problematic in noisy city
magnetometers, and online computer control and signal pro-
laboratories and led to the establishment, in the 1960s, of
cessing. This second generation of magnetometers proved rug-
many out-of-city laboratories, many of which are still in oper-
ged, reliable, and versatile, and most laboratories still employ
ation. High-sensitivity astatic magnetometers housed in these
spinner magnetometers for certain types of work. The ultimate
quiet environments, usually mounted on stable concrete slabs
in sensitivity is, in principle, provided by the cryogenic mag-
and set in the center of large sets of Helmholtz coils, were
netometers, which first appeared in the early 1970s following
capable, with patience, of accurate measurements to
Josephson’s discoveries in superconductivity, including quan-
0.001 A m1: weak sediments and limestones.
tum properties of the Josephson junction and its close cousins,
An alternative in magnetically noisy environments is a
the weak link, and the SQUID. Modern cryogenic systems offer
parastatic system, which employs three suspended magnets,
unprecedented sensitivity, fast response, no spinning or vibrat-
as shown in Figure 28(b). The upper and lower magnets each
ing, full automation, including some demagnetization proce-
have half the magnetic moment of the center magnet, and are
dures if desired, data logging, and presentation.
mounted antiparallel to it. This arrangement is less sensitive to
changes in the gradient of a horizontal field. Parastatic systems
5.04.5.2.2 Astatic and parastatic systems were first described by Thellier (1933) and were further devel-
A single suspended magnet suffers from time variations of the oped into the 1960s (Pozzi and Thellier, 1963). Larochelle and
torque given by the time-varying ambient magnetic field and is Christie (1967) quote an accuracy of 1 in the measurement of
really only suitable for rough measurements on samples with a sample of 103 A m1, in a measurement time of 6 min.
magnetizations strong enough to exceed this variability.
The principle of an astatic system is to compensate for time 5.04.5.2.3 Spinner magnetometers
variations in a uniform horizontal field by using two magnets The basic principle of all spinner magnetometers is that of
of equal moment, mounted antiparallel, one above the other, electromagnetic induction: when the magnetic moment of a
from the same suspension (Figure 28). The total moment of sample is rotated in or near either a coil or a fluxgate system, an
the system is weak, and a weak torsional suspension can be alternating voltage signal is induced in the coil or fluxgate
used. When a sample is brought up symmetrically beneath the circuitry. For this reason, they have also been called ‘rock-
lower magnet, a torque is exerted on the system that is propor- generator’ magnetometers (e.g., Nagata, 1961). In general,
tional to the magnetic moment of the sample (and depends the amplitude of the induced signal is proportional to the
also on a number of geometric factors); this results in an component of the magnetic moment perpendicular to the
angular deflection that is measured by means of an optical rotation axis, and the phase can be used to determine
lever. By measuring the deflection with the sample in a number the direction of this component with respect to some reference
Several different designs of spinner magnetometers have

been developed commercially. The Princeton Applied Research
design includes a compensated system of two coils connected
in series opposition to eliminate field variations, while the
Schonstedt magnetometer contains a single fluxgate housed
in a mu-metal shield to eliminate external magnetic fields.
Agico, Prague, also manufactures a high-speed (85 Hz) spinner
with a pair of uncompensated pickup coils, housed in a three-
layer mu-metal shield.
The Digico instrument (Molyneux, 1971; Bartington, web-
site) incorporated an innovative ring-shaped fluxgate, within
which the sample was rotated at about 4 Hz. It was also the first
magnetometer to be controlled from a computer terminal and
to employ online signal processing: features that, a decade
later, had become virtually taken for granted but that made
the Digico the most advanced magnetometer of its day. The
basic operating principles, which were later miniaturized in the
portable Minispin magnetometer, are illustrated in Figure 29.
The sample rotates in the circular fluxgate. The signal is sam-
pled 128 times per revolution – this is controlled by a photo-
gate arrangement mounted over a slotted disk that rotates with
the sample, at the lower end of the drive shaft. The signals pass
through an analog-to-digital converter to the computer, where
they are averaged over a preset number of revolutions. When
z z the measurement is complete, the output signal is Fourier-
analyzed and the first harmonic gives the magnetic moment
y y in the plane of the fluxgate: its components are calculated with
x x respect to a reference direction on the sample holder, printed
(a) (b) out, and saved. Once the sample has been spun in 4 or 6
different orientations, the direction and intensity of the mag-
Figure 28 (a) Principle of the astatic magnetometer: the two magnets
have equal and opposite moments. The angular deflection is measured netic moment are calculated, together with a measure of uncer-
for a number of different orientations of the specimen, one of which tainty. Further calculations can be programmed to perform
is shown. (b) Parastatic magnetometer system: the central magnet has a field and/or bedding corrections and so output declination
moment equal to twice that of the upper and lower magnets and is and inclination in various reference frames. By increasing the
oppositely oriented: this compensates for gradients in the horizontal number of revolutions over which each measurement is aver-
magnetic field. aged, the user can increase the precision of the result.
Collinson (1983) quoted that the practical lower limit of sen-
sitivity of the Digico magnetometer is about 104 A m1, with
direction in the sample. The sample must be spun about at a measurement time of between 10 and 30 min and a direc-
least two axes to obtain three orthogonal components of the tional accuracy of about 5 .
magnetic moment and so compute the total vector. Measure-
ments are often made in 4, 6, or 12 different orientations, not 5.04.5.2.4 Cryogenic magnetometers
only creating a redundancy of information but also enabling Rock magnetometers incorporating superconducting (SQUID)
the calculation of a statistical measure of confidence in the sensors (Section 5.04.2.1) were developed in the early 1970s
result. This can provide valuable information on the homoge- and began to appear in paleomagnetic laboratories around the
neity of magnetization in a sample and its short timescale world later that decade. Cryogenic magnetometers offer enor-
viscosity. mous improvements in both the sensitivity and speed of
In induction coil-based systems, the induced voltage is measurement while causing minimum disturbance to delicate
actually a function of rotation frequency and the number of samples. Sensor design and signal processing are not the only
turns and dimensions of the coils and the component of the technical problems in the development of a cryogenic magne-
magnetic moment of the sample perpendicular to the rotation tometer: maintaining the extremely low temperatures at which
axis. High-rotation speeds (100–400 Hz) are required, and the superconducting components operate is an equally impor-
these are sometimes achieved by means of an air turbine, to tant challenge.
avoid magnetic field interference. However, rapid rotation is The first cryogenic rock magnetometers were built by Super-
not always ideal for delicate samples. conducting Technologies (SCT) in the United States and by
Fluxgate sensors avoid the need for rapid rotation, since, in Cryogenic Consultants Ltd. (CCL) in the United Kingdom and
addition to the sample moment, the induced signal depends incorporated RF-driven, weak-link SQUID sensors. The practi-
only on the fluxgate sensitivity (Section 5.04.2.1) and the cal design of the sensors is dictated by constraints on the size of
distance of the sample from it. Rotational frequencies of less the superconducting loop, which, in order to mitigate the
than 10 Hz are common. effects of thermal noise, must have a very small inductance
Mu metal shield
Circular fluxgate
Specimen
Analog Analog
signal 6 Hz to digital
Fluxgate
processor converter
drive signal Fluxgate
6.25 kHz output
12.5 kHz
Specimen orientation data,
bedding and elevation data
Microprocessor
and/or
PC N, E, Down components
of magnetic moment,
declination, inclination, intensity
Photogates of magnetization
Slotted disk 1 pulse/rev
128 pulses/rev
Spin motor
~6 Hz
Figure 29 Principle of the Digico fluxgate magnetometer (Molyneux, 1971). The specimen is spun at the center of a single-axis ring-core fluxgate;
signal processing is carried out by a microprocessor or computer, which calculates and outputs results as components of magnetic moment or as
required by the user.
and have a diameter of 1–2 mm (Fuller, 1987). This is an order region of the magnetometer is deep within a magnetically
of magnitude smaller than the standard paleomagnetic speci- shielded, heavily insulated liquid helium cryostat.
men, so the most convenient way to link the specimen’s mag- Over the past 30 years, several modifications and improve-
netic flux to the sensor is indirectly, via a transformer-like ments have been made to commercially available cryogenic
coupling. The specimen is inserted into a sensor coil, which is magnetometers. These include (i) replacement of RF-driven
connected by to a more tightly wound field transfer coil, and SQUIDs with DC SQUIDs, which effectively count individual
the flux of this coil is measured by the SQUID detector. Both flux quanta: improving sensitivity to c. 1012 A m2 for a stan-
coils and the connections are superconducting: the arrange- dard diameter access (2G Enterprises, website), and (ii) design
ment therefore acts somewhat like a DC transformer, the cou- of a horizontal axis machine, with room temperature access at
pling depending on the actual flux linkage rather than its rate both ends of the cryostat, allowing long samples to be passed
of change. Figure 30 illustrates the arrangement of pickup coils right through the sensor region. Whole or half cores of sedi-
and SQUID sensors in the original SCT machine, which mea- ment can thus be measured intact, by stepping them through
sured two components of magnetic moment, vertical and hor- the sensors. Whole sediment cores typically have diameters of
izontal. The Helmholtz configuration of the sensor coils 5 cm or more, so this necessitates a wider access, which lowers
ensures optimum sampling of a homogeneous specimen the sensitivity for smaller samples. A further development, the
placed at their center: the horizontal axis coils are less sensitive U-channel, solves this problem. A long rectangular plastic
to sample position than the vertical axis set. Typically, a spec- trough, up to 1.5 m long and 20 mm by 20 mm cross section –
imen is inserted into the sensor region, and measurements are the ‘U-channel’ – is pressed into the flat surface of a half core,
made at 90 steps as it is rotated about a vertical axis. This gives the resulting sample removed, and the open face of the
four estimates of the z component and two of each of the x and U-channel sealed with a lid. The U-channel sample can be
y components of magnetic moment. The specimen may then passed continuously through a standard small-access magne-
be inverted and the process repeated to give four more esti- tometer. At other times, the same magnetometer can be used
mates of z and two more of x and y. Analyzing these, together for discrete sample measurement without loss of precision.
with background and holder measurements, allows calculation Both whole-core and U-channel measurements inevitably
of the magnetic moment vector and estimation of the internal result in a convolution of the magnetization signal with the
consistency of the overall measurement. The specimen holder response function(s) of the sensor(s), which has a smoothing
is usually made of lightweight plastic or mylar, and the sensor effect on the output. Methods of deconvolution are available
Output A relatively new development in cryogenics is the ‘pulse-

signal tube cryocooler,’ which completely circumvents the need for
liquid helium in order to reach temperatures of 4.2 K. The
pulse-tube cryocooler first appeared in 1999 and is now incor-
Weak link sensor porated into rock magnetometers made by 2G Enterprises (2G
Field transfer coil Enterprises, website).
5.04.5.3 Progressive Demagnetization Techniques

5.04.5.3.1 Introduction
The NRM of a paleomagnetic sample generally comprises sev-
eral components that are carried by grains in different parts of
the blocking temperature spectrum. The paleomagnetist usu-
ally wants to isolate and determine the primary or characteris-
tic component, acquired at the time of formation of the rock.
In a straightforward situation involving only thermal activa-
tion/relaxation, but not chemical alteration of the magnetic
z
y axis pick up coils minerals, the oldest component of magnetization will reside in
the most stable grains, that is, those with the highest blocking
y temperatures and longest relaxation times. This will be the
x primary component unless the whole spectrum of grains
z axis pick up coils
has been remagnetized, for example, by a lightning strike or
by heating to above the Curie temperature. Secondary partial
thermoremanent magnetizations (pTRMs) and viscous
remanent magnetization (VRM) components will be carried
by less-stable grains.
Such a situation is illustrated by the hypothetical example
in Figure 31. Only those grains with blocking temperatures
Figure 30 Arrangement of horizontal axis (black) and vertical axis
above T2 retain a record of the primary magnetization. At some
(white) superconducting pickup coils, field transfer coils, and weak-link
(SQUID) sensors in the early SCT cryogenic magnetometer. time in the rock’s history, all the grains with blocking temper-
atures below T2 have been remagnetized by a heating event and
have gained a secondary pTRM in a different direction. Further,
the least stable grains, those with TB < T1, have gained a VRM,
(e.g., Jackson et al., 2010) but must be tailored to individual probably parallel to the present day field at the site. A single
machines. measurement of the NRM will yield the vector sum of these
Until very recently, the only practical means of maintaining components, as shown in Figure 31(b). In real situations,
the very low temperatures necessary to operate superconduct- there is almost always some overlap between the grains carry-
ing sensors was by means of liquid helium. The first cryogenic ing each of the components of magnetization.
magnetometers were insulated by means of high vacuums, The association of the oldest component of magnetization
superinsulation (layers of reflective aluminized mylar sepa- with the most stable grains breaks down in more complex
rated by fiberglass sheets), heat shields cooled by contact situations, for example, when chemical alteration has pro-
with the evaporated helium gas, and/or an outer dewar filled duced new, secondary ferrimagnetic minerals. In such a case,
with liquid nitrogen. It is crucial to ensure that helium gas is one might find a secondary chemical remanent magnetization
excluded from the vacuum spaces, as even small amounts can carried by authigenic grains that have grown to such a size that
result in the conduction of heat into the sensor region and loss they have the highest blocking temperatures in the sample
of superconductivity. By these means a boil-off rate of about and that have formed at the expense of primary grains with
2.5 l of liquid helium per day could be achieved, or 10–12 high TB’s.
working days per 30 l fill. Partial or progressive demagnetization procedures are
Huge improvements in liquid helium retention may be designed to incrementally remove the magnetization of a
achieved by incorporating an active cooling device or ‘cryocoo- sample – working systematically through the blocking temper-
ler’ in the magnetometer. Unfortunately, the mechanical vibra- ature or coercivity spectrum. This enables the separation and
tion and magnetic noise of such devices mean they must be determination of components carried by grains in different
physically separated from the magnetometer. A common intervals of the spectrum. Further information will be required
arrangement is to place the compressor outside the building to identify which is the primary or characteristic component
or outside the magnetically shielded space or room that houses and to piece together the magnetization history of the sample
the magnetometer. The cryocooler cools a thermal shield completely.
around the helium reservoir, thus inhibiting evaporation.
A modern magnetometer with a dewar of 50–90 l, and a well- 5.04.5.3.2 Thermal demagnetization
maintained cryocooler, will typically run for between 500 and This method appeals directly to the blocking temperature spec-
900 days between fills. trum of the ferrimagnetic grains in a rock sample.
Proportion of
ferrimagnetic
grains
Most stable grains

retain primary TRM
Least stable grains

acquire viscous
remanent magnetization
T1 T2 Blocking
Intermediate temperature, TB
grains carry
(a) secondary pTRM
Secondary pTRM:
intermediate grains
Primary TRM:
most stable grains
VRM:
least stable grains
(b) Resultant natural remanent magnetization

Figure 31 (a) Schematic diagram of the stability spectrum of ferrimagnetic grains in a typical rock sample, showing portions of the spectrum carrying
the primary (TRM), secondary (pTRM), and viscous (VRM) components of remanent magnetization. (b) Vector diagram showing the addition of the
various components of magnetization to yield the resultant natural remanent magnetization (NRM).
After measurement of NRM, a specimen is heated to a pre- stage that it is most important to maintain as low a magnetic
determined temperature, TH, and then cooled back to room field as possible. Some designs incorporate a separate chamber
temperature in a carefully controlled, zero magnetic field envi- into which the samples are passed for cooling, while a second
ronment. This procedure randomizes the magnetic moments set of samples may be heated in the already hot oven. In other
of all grains whose blocking temperatures are exceeded during designs, the samples are cooled in the same space, without
the heating. A subsequent measurement of remanence will moving them. Some furnaces incorporate the additional facility
now give the vector sum of the moments of only those grains to cool samples in a preset magnetic field, thus growing a
with blocking temperatures above TH. Vector subtraction of (partial) TRM in part of the blocking temperature spectrum,
this from the original NRM will give the removed magnetiza- rather than demagnetizing. This is particularly useful for the
tion: that carried by grains with TB < TH. In principle, a series of estimation of paleointensities (see Chapter 5.13).
incremental heating steps, alternated with measurements of It is important to allow sufficient time for the specimens to
the magnetization remaining after each (zero field) cooling, equilibrate both at the maximum temperature and back to
will enable a complete analysis of the components of magne- room temperature before measurement: typically 20–30 min
tization carried by a specimen. for each stage.
A number of thermal demagnetizers are available commer-
cially. In addition, many paleomagnetic laboratories have 5.04.5.3.3 Alternating field demagnetization
purpose-built equipment. The basic requirements of a thermal Working progressively through the coercivity spectrum of a
demagnetizer are a magnetic field-free furnace and cooling specimen provides an alternative way of separating different
chamber. Most furnaces employ noninductively wound electric components of magnetization. In alternating field demagneti-
heating elements, though gas-fired furnaces are also used. Var- zation, the specimen is placed inside a solenoid, which is
ious designs take between 12 and 100 specimens at a time: connected to a variable AC supply while being housed in an
either arranged around the axis of a cylindrical furnace inside environment of zero direct field. The amplitude of the alter-
multiple, nested mu-metal shields or in a layered arrangement nating current is increased to a preset maximum corresponding
at the center of a large Rubens coil setup. It is during the cooling to a desired maximum magnetic field: it is then smoothly and
slowly ramped back down to zero. When the current is at it with the strength of a TRM grown in a known field in the
maximum amplitude, the magnetic moments of all grains laboratory.
TRM
with coercivities below the corresponding magnetic field fol- Specifically, Bpaleo ¼ TRMpaleo
lab
Blab or for each interval of the
pTRM
low the oscillations back and forth along the axis of the sole- blocking temperature spectrum, Bpaleo ¼ pTRMpaleo Blab .
lab
noid. As the peak field falls below the coercivity of each grain, In practice, a specimen is heated to a predetermined tem-
its magnetic moment becomes blocked: ideally, equal numbers perature TH and cooled in zero field, removing a portion of the
of grains become blocked in each opposing direction. All NRM (pTRMpaleo). It is then reheated to the same temperature
directions in the specimen should be exposed equally to the again and, this time, cooled in a known laboratory field, gain-
demagnetizing field of the solenoid. This may be achieved by ing a component of magnetization pTRMlab. Measurements
successively aligning each of the three axes of the specimen made before and after each cooling step enable the calculation
with the axis of the solenoid while the alternating magnetic of each pTRMpaleo and each pTRMlab. As with progressive
field is cycled up and down (static method). Alternatively, the demagnetization, this process is repeated, each time increasing
specimen may be tumbled simultaneously about two perpen- the temperature, until the whole blocking temperature spec-
dicular axes, by means of a small motor and belt-driven turn- trum has been covered. Temperature steps are generally
table, while the field is cycled just once. There are advantages decided after an examination of the blocking temperature
and disadvantages associated with each method. Certain spec- spectrum, for example, from thermomagnetic or thermal
imens are susceptible to acquisition of a spurious magnetiza- demagnetization experiments.
tion during the tumbling procedure: a rotational remanent The paleointensity is obtained from the gradient of the
magnetization (RRM) antiparallel to the inner rotation axis linear part of a graph of the NRM remaining after each demag-
was first documented by Wilson and Lomax (1972). RRM netization step against the pTRM gained in the same step, often
seems to be related to the more general phenomenon of gyro- called an Arai plot (Nagata et al., 1963).
remanent magnetization (GRM), which may be acquired by The relationship between magnetization and paleointensity
single-domain grain-bearing specimens when they are rotated is more complicated in the case of a sedimentary material
in steady or alternating magnetic fields (Stephenson, 1980, carrying a DRM, where the degree of alignment of ferrimag-
1981). Methods to correct for RRM and GRM have been pro- netic grains depends on external factors that are much more
posed by Hillhouse (1977) and Giddings et al. (1997). How- difficult to quantify. The intensity of DRM is nevertheless still
ever, these corrections are of an empirical nature, and to avoid predicted to be proportional to the prevailing magnetic field,
any suspicion of doubt in results, many paleomagnetists prefer and methods to retrieve logs of relative paleointensity from
to place their specimens in the coil successively in three orthog- uniform sequences of sediments are often successful. Such
onal orientations or to use thermal demagnetization in speci- methods usually employ normalization of a portion of the
mens that are likely to develop RRM or GRM. NRM with a laboratory-produced remanence, often anhystere-
tic or isothermal remanent magnetization.
5.04.5.3.4 Paleointensity determination

Methods for the determination of the intensity of the geomag- 5.04.5.4 Data Analysis and Statistics
netic field from paleomagnetic measurements are covered in Most rock magnetometers return the components of the mag-
detail in Chapter 1.15, and only the rudimentary principles are netic moment of a specimen along the x, y, and z axes of the
given here. In general, the intensity of magnetization (TRM, specimen. The operating software usually offers the provision
DRM, CRM, etc.) of a sample will depend on many factors in to enter the field orientation of the specimen, and this enables
addition to the strength of the prevailing field. These include transformation of the result to the present-day geographic
intrinsic magnetic properties of the material such as the con- reference frame: northerly, easterly, and downward compo-
centration of the remanence-bearing mineral(s), their satura- nents of magnetic moment or declination, inclination, and
tion magnetization, and granulometry. In seeking the total magnetic moment. The intensity of magnetization is fur-
paleointensity, one must firstly identify the portion of the ther obtained by dividing the total magnetic moment by the
ferrimagnetic grain spectrum that carries the primary or char- specimen volume. Finally, if the bedding attitude is known,
acteristic component of magnetization (ChRM) and then nor- this may be entered and the magnetization vector referred to
malize the intensity of magnetization of this portion to the pretilting coordinate frame.
compensate for the intrinsic magnetic composition of the sam-
ple as described earlier.
In principle, absolute paleointensity determinations can be 5.04.5.4.1 Progressive demagnetization data
made from some TRM-bearing materials. Progressive demagnetization of a specimen yields a series of
Néel’s theory (1949, 1955) of TRM carried by single- vectors, describing the magnetization left after each step of
domain grains suggests that the strength of TRM should be demagnetization. Three different plots are useful in interpret-
proportional to the strength of the magnetic field in which it ing such data. Examples are shown in Figure 32.
was acquired. Further, subject to some basic assumptions, this
applies also to the portion of the magnetization (partial TRM 5.04.5.4.1.1 Intensity of magnetization against demagnetization
or pTRM) carried in each discrete interval of the blocking level
temperature spectrum (see Chapter 1.15 for more details). In If the direction of remanence changes little during demagneti-
principle then, one can obtain an estimate of the paleointen- zation, then this plot gives an approximation to the blocking
sity of the field at the time of acquisition of TRM by comparing temperature or coercivity spectrum of the specimen. For
N N
Mmax = 4.7 mA m-1
M/Mmax
Lower
Upper 1.0
E
UP 0.5
0.0
0 100 200 300 400 500
Temperature (⬚ C)
(a) (b) (c)
Figure 32 An example of progressive thermal demagnetization data for a (southern hemisphere) specimen carrying a primary magnetization of
reversed polarity, overprinted by a normal polarity thermoviscous secondary component. (a) Vector component plot showing the projection onto the
horizontal plane (N vs. E) and the projection onto the vertical NS plane (N vs. UP), as crosses and diamonds, respectively; (b) an equal angle
stereographic projection of the sequence of remanent directions; (c) the intensity of magnetization plotted against demagnetization level (temperature).
multicomponent magnetizations, its interpretation is less component plot. In practice, however, the data are invari-
straightforward. ably scattered and estimation of the best straight line is often
difficult and subjective. In such cases, principal component
5.04.5.4.1.2 Stereographic projection analysis, or PCA (Kirschvink, 1980), provides a rigorous,
Stereographic projection of the directions of the remanence objective way of estimating components of magnetization.
remaining at each step contains no information regarding In PCA, the best-fit direction is calculated to a preselected
the intensity of magnetization. For a single-component sequence of demagnetization data (vectors), which may be
magnetization, the direction does not change during demag- ‘anchored’ to the origin if desired, by analysis of a 3 3
netization and all the data points overlie each other. For a two- matrix made up from the vector components: the
component magnetization, the initial NRM is the vector sum ‘orientation tensor.’ The principal eigenvector gives the
of the two components. As the lower blocking temperature best-fit direction, while a statistic of the quality of the fit,
component is progressively removed, the tip of the remanent the ‘maximum angular deviation’ or MAD, is derived from
unit vector moves along a great circle path toward the direction the eigenvalues.
of the underlying more stable component. The direction of the
secondary component lies on the backward extrapolation of 5.04.5.4.1.5 Remagnetization circles
this great circle. This process is best seen in Figure 32(b), where In some situations involving multicomponent magnetization,
the primary magnetization is of reversed polarity (to the south it is impossible to isolate the ChRM by progressive demagne-
and downward, since the specimen is from a southern hemi- tization. In such cases, techniques involving extrapolation
sphere site), and the secondary magnetization is close to the of trends in the demagnetization data (remagnetization
present-day field at the site (D ¼ 20 , I ¼ 66 ). circles) from a number of different samples until they intersect
can be used to infer a common underlying component
5.04.5.4.1.3 Vector component plots of magnetization or hypothetical endpoint (Halls, 1978;
Vector component plots combine both intensity and direc- McFadden and McElhinny, 1988). As with any extrapolation
tion data, by projecting the successive remanent vectors onto technique, extreme care is required in the interpretation of
two different planes and presenting them in a single diagram directions obtained from remagnetization circle analysis.
(Zijderveld, 1967). The actual planes chosen vary: in
Figure 32(a), the horizontal plane (N vs. E) and the vertical 5.04.5.4.2 The statistical treatment of paleomagnetic data
north–south plane (N vs. Up) are shown, with the northerly When analyzing paleomagnetic data, it is usual to treat the
component plotted on the same (y) axis in both cases. Linear direction and intensity of the vectors separately. This is because
segments correspond to the removal of single components of the retrieval of the paleointensity of the geomagnetic field from
magnetization, and the direction of such a component can be magnetization intensity is a more complex task and frequently
measured directly from the diagram. The component residing cannot be achieved with the same degree of confidence as the
in the grains with highest blocking temperatures or coerciv- paleodirection.
ities (often the primary component) should yield a linear
segment terminating at the origin. 5.04.5.4.2.1 Intensity of remanence and geomagnetic
paleointensity
5.04.5.4.1.4 Principal component analysis The intensity of remanence (and also magnetic susceptibility and
In theory, the direction of a component of magnetization other, laboratory-induced magnetizations) depends on the com-
can be measured directly from a linear section of the vector position and concentration of the magnetic minerals and their
granulometry, that is, grain size and shape distributions. Most parameter and reflects the dispersion of the distribution: it is
such natural distributions do not follow a normal or Gaussian analogous to the invariance, or the reciprocal of the variance,
form, which is symmetrical about the mean value, but are highly of a Gaussian distribution. k ranges between zero for a random
skewed, with a long tail toward high values. It is often found that distribution (maximum scatter) and infinity for perfect align-
the logarithms of the values are normally distributed however. ment of the entire population.
The ‘log-normal’ distribution is found to give a good representa- A paleomagnetic dataset is a subset of N samples taken from
tion of natural remanent intensities and magnetic susceptibilities the whole population. For N ≳ 6 and k ≳ 3, the best (minimum
in rocks (Irving et al., 1966; Tarling, 1966). variance, unbiased) estimate of k is given by k ¼ (N 1)/
(N R) (Fisher, 1953; McFadden, 1980). Naturally, as more
5.04.5.4.2.2 Paleomagnetic directions: the Fisher distribution of the population is sampled, that is, as N increases, k becomes
The analysis of remanent directions usually reduces to dealing a better estimate of k.
with a set of unit vectors obtained from a set of N samples. The angular standard deviation, or angular dispersion, S, is
Each unit vector M ^ a useful measure of scatter, for example, in studies where
i can be expressed in terms of its declination secular variation should be adequately sampled. For a Fisher
and inclination or as direction cosines li, mi, and ni, which are
the northerly, easterly, and downward components of the unit distribution, S is the semiangle of the cone that includes 63%
vector (Figure 33(a)): of the population. For low scatter, a good approximation is
given by
mi
li ¼ cosðDeci ÞcosðInci Þ Deci ¼ tan1 81
li y63 ¼ pffiffiffi ffi S
mi ¼ sinðDeci ÞcosðInci Þ Inci ¼ sin1 ni k
ni ¼ sinðInci Þ The closeness of the mean of a given dataset to the true
The mean direction is obtained from the vector sum R mean of the population is clearly of crucial importance in
(Figure 33(b)): paleomagnetism. The ideal situation is obviously a large data-
! set that samples the population randomly.
X
N X
N X
N For k ≳ 10, the confidence that can be placed in the sample
R¼
li , mi , ni ¼ Rx ; Ry ; Rz mean is quantified by the relation
i¼1 i¼1 i¼1
( 1=ðN1Þ !)
0 1
NR 1
að1PÞ ¼ cos1 1 1
R B R C R P
Mean Dec ¼ tan1 Mean Inc ¼ tan1 @
y z
1=2 A
Rx
R2x þ R2y
where a is the semiangle of the cone around the sample mean
The length R ¼ (R2x þ R2y þ R2z )1/2 of the vector sum R forms within which there is a probability (1 P) of the true mean of
the basis of the statistical analysis. If all N unit vectors the population lying. The levels of confidence most commonly
are perfectly aligned, then R ¼ N; the greater the scatter in quoted are 95% and 63%. The best estimates of a95 and a63 are
the directions of the unit vectors, the smaller is the value of R. given by
The two-dimensional probability density distribution func- 140 81
tion generally assumed to apply to paleomagnetic unit vectors a95 ffi pffiffiffiffiffiffiffi and a63 ffi pffiffiffiffiffiffiffi
kN kN
is the Fisher distribution (Fisher, 1953). It is, in many ways,
equivalent to a 2-D Gaussian or normal distribution and is where a63 is the direct analog of the standard error in the mean
described mathematically by the function of a Gaussian distribution.
k
P ð’Þ ¼ expðk cos ’Þ 5.04.5.4.2.3 Comparison of directions
4p sinh k In order to assess whether a paleomagnetically determined
where ’ is the angle between a direction and the true mean direction differs significantly from a direction that, by compar-
direction of the whole population. k is known as the precision ison, is completely specified, for example, the direction of the
l North North (x)

Dec
Inc
m
East ( y)
East
n
ˆ
M R
Down
(a) (b) Down (z)
^
Figure 33 (a) The direction cosines l, m, and n of the unit magnetization vector M
of a sample and their relationship to declination and inclination. (b)
^ ^2 M ^
Vector addition of unit magnetization vectors from a number of samples, M 1 M 3 . . ., to yield the vector R (see text).
present-day field, a95, may be used directly. If the known within confidence levels. It is then possible to conclude that
direction lies outside the cone of 95% confidence of the paleo- the magnetization predates the folding or tilting episode. An
magnetic direction, then there is a 95% probability that the unsuccessful or partially successful fold test can also yield
true directions differ. information about the magnetization history of a unit.
It is frequently necessary to compare two paleomagnetically
determined directions, each with its own a95. It is often stated
5.04.5.4.3.3 Reversals test
that the two directions are distinct (at the 95% level of confi-
When a study involves sampling through a number of geomag-
dence) if their cones of confidence do not overlap and indis-
netic polarity reversals, for example, for the calculation of
tinguishable if they do. While the first statement is true, the
paleomagnetic pole positions (see Section 5.04.5.4.4) or for
second is not strictly correct. The cones may overlap by a small
magnetostratigraphy, the resulting dataset will contain records
amount while the means are still distinct. More rigorous sta-
of both normal and reversed polarity. According to the GAD
tistical tests have been devised and are discussed by McFadden
hypothesis (explained in Section 5.04.5.4.4.3), if secondary
and Lowes (1981) and Watson (1983).
components of magnetization have been completely removed
so that the records are accurate reflections of the original
5.04.5.4.3 Field tests
geomagnetic field, and if sufficient time intervals have been
Progressive demagnetization experiments enable the analysis
sampled to average over secular variation (104–105 years), then
of an NRM in terms of components carried by grains in differ-
the averages of the normal polarity directions and the reversed
ent parts of the blocking temperature or coercivity spectrum.
polarity directions should be antipodal. Figure 34(b) illus-
Statistical methods allow a meaningful comparison to be made
trates data from a magnetostratigraphic study in NZ. After
between determined paleomagnetic directions. However, nei-
progressive thermal demagnetization and principal compo-
ther gives any information regarding the ages of the various
nent analysis to estimate characteristic magnetization direc-
components of magnetization nor can prove that the charac-
tions, the mean normal and reversed directions are antipodal
teristic magnetization was acquired at the time the rock
to better than 95% confidence. The difference between the
formed. The application of some or all of the field tests dis-
overall mean direction and the GAD direction for the site is a
cussed in the following sections can help order events in the
combination of an anomaly due to the movement of the
geological and magnetization histories of a rock and so help
Australian plate over the past 3 Ma and a local vertical axis
date components of magnetization.
rotation associated with the proximity of the site to the Austra-
lian/Pacific Plate boundary (Turner et al., 2005).
5.04.5.4.3.1 Consistency test
Demonstration of the reproducibility of results is an important
aspect of all branches of science. In paleomagnetism, it is 5.04.5.4.3.4 Baked contact test
usually necessary to show that the characteristic magnetization When an igneous magma or volcanic lava is emplaced adjacent
is of geomagnetic origin and not produced by some spurious or to a country rock, a temperature gradient develops in the
local effect such as a lightning strike or physical disturbance. country rock as heat is conducted away and the magma or
Replication of a signal in rocks or sediments from different lava cools and solidifies. As the temperature falls through the
provenances or of different lithologies is a strong argument in blocking temperatures of its constituent ferrimagnetic grains,
favor of a regional or global magnetization process. Multiple the igneous or volcanic rock gains a TRM, which (ideally) is
records also provide data for more sophisticated levels of sta- parallel to the ambient geomagnetic field. The maximum tem-
tistical analysis. perature reached in the country rock will be a decreasing func-
tion of distance from the magma or lava, and the magnetic
5.04.5.4.3.2 Fold test moments of ferrimagnetic grains with blocking temperatures
Paleomagnetists often sample beds or sediments that are no up to this maximum will be reset, that is, the baked contact will
longer flat-lying. It is usual to assume that such sediments acquire a secondary pTRM. This will be parallel to the field at
originally accumulated on a horizontal surface and have later the time of emplacement of the lava or magma and hence
been tilted or folded by tectonic processes of some sort. If a parallel to the TRM of the lava or magma. In parts of the
stable characteristic magnetization predates the tilting or fold- country rock closest to the magma, the entire blocking temper-
ing, then the magnetization vector will have been tilted with ature spectrum might be affected, but further away, only those
the beds (Figure 34(a)), and a correction to restore the beds to grains with lower blocking temperatures will become reset,
the horizontal will also restore the magnetization to its original while grains with the higher blocking temperatures will retain
orientation. To this end, it is routine to measure the attitude of the primary magnetization of the country rock. Such a scenario
all sedimentary strata sampled during fieldwork. In cases of results in a positive baked contact test.
simple bedding tilt, a single rotation about the horizontal In the first part of the twentieth century, baked contact tests
strike of the beds by an angle equal to the dip is required. In provided valuable evidence for the theory of polarity reversals.
cases where folding is evident, two steps are necessary: restora- The opposing theory held that rocks found with reversed mag-
tion of the fold axis to horizontal followed by untilting of the netizations had somehow become magnetized in the opposite
beds in the limbs of the fold. This requires the additional field direction to the ambient field (self-reversal). A worldwide com-
measurement of the plunge of the fold axis. A fold test is pilation made by Wilson (1962a) listed 48 positive and 3
successful if, after these field corrections, the characteristic negative baked contact tests: 11 years later, McElhinny (1973)
magnetization vectors from individual sites move closer listed 154 positive results and still only 3 negative ones. As the
together or site average vectors become indistinguishable number of positive baked contact tests accumulated, the
(ii)
(i)
(iii)
(a)
Dexp
Turakina River, NZ
3.5–2.5 Ma
Mean normal direction

Dec = 10.9°
Inc = −57.7°
α-95 = 1.9°
N = 65
Mean reversed direction

Dec = 187.7°
Inc = 58.4°
α-95 = 4.3°
N = 36
(b) Dexp
Figure 34 Paleomagnetic field tests: (a) Block diagram showing (i) the rotation of primary magnetization vectors during the folding of a rock unit (fold
test); (ii) the randomization of the primary magnetization vectors of clasts incorporated in a conglomerate unit (conglomerate test); (iii) the
remagnetization of country rocks immediately adjacent to an intruded dyke (baked contact test). Adapted from Graham JW (1949) The stability and
significance of magnetism in sedimentary rocks. Journal of Geophysical Research 54: 131–167. (b) Example of a positive reversals test: equal
area stereographic projection showing paleomagnetic directions from 101 sites sampled in a magnetostratigraphic study of the Turakina River valley,
New Zealand. The mean normal and reversed polarity directions are shown by stars, surrounded by an oval showing the cone of 95% confidence
in the mean (alpha-95). Reproduced from Turner GM, Kamp PJJ, McIntyre AP, Hayton S, McGuire DM, and Wilson GS (2005) A coherent
middle Pliocene magnetostratigraphy, Wanganui Basin, New Zealand. Journal of the Royal Society of New Zealand 35: 197–227.
likelihood of widespread self-reversal rapidly diminished and carried by the pebbles and by the matrix. If the pebbles retain
the idea of polarity reversals became generally accepted by the a ChRM dating back to their previous history, then these
geomagnetism community. Documented self-reversal mecha- vectors should have been randomized in the process of depo-
nisms invariably involve particular, uncommon mineralogies, sition, and no coherency should be observable in the data
which are unlikely to be found simultaneously in large num- (Figure 34(a)). Any coherent component of magnetization
bers of lavas and adjacent country rocks. An unusual example found in the pebbles or between the pebbles and matrix raises
of a double-contact test was described by Wilson (1962b). the possibility of postdepositional remagnetization.
A lava flow and a dyke were successively emplaced in a laterite,
which consequently carries secondary components of magne-
tization from both heating episodes. Both are of reversed 5.04.5.4.4 Poles
polarity, but they can be separated by progressive demagneti- 5.04.5.4.4.1 Geomagnetic poles
zation since their directions differ significantly. It is virtually The present-day geomagnetic field can, to a first approxima-
inconceivable that this situation could have occurred through tion, be modeled as a geocentric dipole. In 2010, the axis of the
self-reversal. best-fitting geocentric dipole intersected the Earth’s surface at
80.0 N, 287.8 E and 80.0 S, 107.8 E (IGRF-11, 2010). These
5.04.5.4.3.5 Conglomerate test locations are called the geomagnetic poles. This tilted geocentric
A conglomerate test may be applied when the rock (or sedi- dipole model (Figure 35(b)) describes about 80–90% of the
ment) of interest contains pebbles or clasts of a different lithol- present field at the Earth’s surface – the remaining ‘nondipole’
ogy. It involves comparing the components of magnetization part may be described by the higher-order components of a
Pole
lp fp
Site D
I
ls fs
fs fp
(a) (b) (c)
Figure 35 (a) The magnetic field due to a geocentric axial dipole; (b) the magnetic field due to a tilted geocentric dipole; (c) the principle of calculation
of a virtual geomagnetic pole from an observed or recorded magnetic field direction at site (ls, fs).
spherical harmonic representation, that is, quadrupole, octu- mapping sequences of transitional directions from polarity
pole, etc. reversals, when the field was almost certainly anything but
dipolar!
5.04.5.4.4.2 Virtual geomagnetic poles
If the Earth’s magnetic field could be modeled completely by a 5.04.5.4.4.3 The GAD hypothesis
geocentric dipole aligned with the rotation axis (a geocentric There is considerable evidence to indicate that, when the geo-
axial dipole, or GAD), then, at all points on the globe, its magnetic field is averaged over sufficiently long intervals of time
horizontal component would be exactly northward, that is, (omitting intermediate directions recorded when the field was
the declination would be zero, and the inclination would be actually in the process of reversal), the mean positions of the
a simple monotonic function of latitude (l): tan(I) ¼ 2 tan(l). geomagnetic poles coincide with the north and south poles of
A corollary of this is that from any observed magnetic field the rotation axis. For example, when the VGPs of a global
direction (or paleomagnetically recorded direction), one can distribution of rocks dated between 0 and 5 Ma are averaged,
calculate the orientation of the geocentric dipole that would the mean coincides with the rotation axis to within about 2
produce it, and the corresponding virtual geomagnetic poles (McElhinny et al., 1996). This leads to the GAD hypothesis: that
(VGPs). Imagine a globe: from the site of the observation, draw the time-averaged geomagnetic field is a GAD field. There is
a great circle in the direction of the horizontal component of some debate as to the minimum time period required to average
the magnetic field or magnetization vector: the virtual pole out the secular variation (of both nondipole and dipole fields):
must fall on this great circle. The angle (at the center of most sources quote between 10 000 and 100 000 years. The
Earth) between the site and the pole (or the virtual geomag- GAD hypothesis is fundamental to many applications of paleo-
netic latitude) is calculated from the recorded inclination magnetism, including providing crucial support for plate tec-
according to the relation noted above and corresponds to a tonic theory and continental reconstructions.
distance measured along the great circle. This process is illus-
trated in Figure 35(c). 5.04.5.4.4.4 Paleomagnetic poles and apparent polar wander
Alternatively, the latitude and longitude of the VGP (lp, ’p) paths
may be calculated from the observed paleomagnetic direction The term paleomagnetic pole is applied to a pole calculated by
(D, I) and the site latitude and longitude (ls, ’s) using the averaging a paleomagnetic dataset that spans a time interval
following equations: long enough to average secular variation. The GAD hypothesis
implies that such a pole gives the position of the rotation axis
sin lp ¼ sin ls sin l þ cos ls cos l cos D with respect to the sampling location at the time the magneti-
fp ¼ fs þ b if sin l sin ls sin lp zation was acquired. Paleomagnetic poles for the past 5
million years are coincident with the rotation axis. However,
fp ¼ fs þ b þ 180 if sin l < sin ls sin lp older poles differ significantly, the discrepancy generally
increasing with age. The time sequence of paleomagnetic
where poles from a given continent or stable cratonic block is called
tan l ¼ ð1=2Þtan I an apparent polar wander path (APWP). The concept of an
APWP was introduced by Creer et al. (1954). Originally, it was
and thought that, while the orientation of the rotation axis remains
cos l sin D fixed in space, APWPs were produced by rotation of the whole
sin b ¼ Earth with respect to its rotation axis: this is now termed true
cos lp
polar wander. The steady accumulation of data over the years
A pole calculated in this way is virtual in the sense that the soon showed that APWPs from different continents differ sig-
geocentric dipole model employed in the calculation cannot nificantly. However, it was also found that sometimes, the
be justified for an observation from a single location and a application of a finite rotation brought a segment of one
single point of time: it has no real physical meaning. VGPs, APWP into coincidence with the contemporaneous section of
however, remain a popular way of presenting paleomag- another. If one imagines an APWP tied to the continent from
netic directions and are used extensively, for example, in which it is derived, the implication is that there have been
periods of time when some continents have moved together, Fisher RA (1953) Dispersion on a sphere. Proceedings of the Royal Society of London,
producing similar APWPs, and other periods when they have Series A 217: 295–305.
Fuller M (1987) Experimental methods in rock magnetism and palaeomagnetism.
moved separately over the Earth’s surface, causing separation
In: Sammis CG and Henyey TL (eds.) Methods of Experimental Physics,
and new segments of their APWPs. This is the basis of the pp. 303–471. Orlando: Academic Press.
theory of plate tectonics, whereby the continents move on Giddings JW, Klootwijk CT, Rees J, and Groenewoud A (1997) Automated
thin lithospheric plates in response to constructive and AF-demagnetization on the 2G-Enterprises cryogenic magnetometer. Geologie en
destructive processes at plate boundaries. Mijnbouw 76: 35–44.
Gilbert G, Cantin JM, Bitterly J, Schlich R, and Folques J (1988). Absolute
measurements of the Earth’s magnetic field in French Observatories: Results
obtained with the portable theodolite fluxgate magnetometer for the period
1979–86. Proceedings of the International Workshop on Magnetic Observatory
References Instruments, Geomagnetism Series 32. Geological Survey of Canada, paper
88-17.
Alexandrov EB and Bonch-Bruevich VA (1992) Optically pumped magnetometers after Gilles H, Hamel J, and Chéron B (2001) Laser pumped 4He magnetometer. Review of
three decades. Optical Engineering 31: 711–717. Scientific Instruments 72: 2253–2260.
Alexandrov EB and Primdahl F (1993) On gyro-errors of the proton magnetometer. Goree WS and Fuller MD (1976) Magnetometers using RF-driven SQUIDs and their
Measurement Science and Technology 4: 737–739. application in rock magnetism and palaeomagnetism. Reviews of Geophysics and
Alexandrov EB, Balabas MV, Kulyasov VN, et al. (2004) Three-component variometer Space Physics 14: 591–608.
based on a scalar potassium sensor. Measurement Science and Technology Graham JW (1949) The stability and significance of magnetism in sedimentary rocks.
15: 918–922. Journal of Geophysical Research 54: 131–167.
Alldredge LR and Saldukas I (1964) An automatic standard magnetic observatory. Grant FS (1985) Aeromagnetics, geology and ore environments, I. Magnetite in
Journal of Geophysical Research 69: 1963–1970. igneous, sedimentary and metamorphic rocks: An overview. Geoexploration
Alldredge LR (1960) A proposed automatic standard magnetic observatory. Journal of 23: 303–333.
Geophysical Research 65: 3777–3786. Gravrand O, Khokhlov A, Le Mouël JL, and Léger JM (2001) On the calibration of a
Aschenbrenner H and Goubau G (1936) Eine Anordnung Registrieung rascher vectorial 4He pumped magnetometer. Earth, Planets and Space 53: 949–958.
magnetischer Storungen. Hochfreg Tech. Elektroakust 47: 178–181. Green AW, Coles RL, Kerridge DJ, and LeMouël J-L (1998) INTERMAGNET, today and
Bardeen J, Cooper LN, and Schrieffer JR (1957) Theory of superconductivity. Physical tomorrow. In: Proceedings of the VIIth Workshop on Geomagnetic Observatory
Review 108: 1175–1204. Instruments, Data Acquisition and Processing’96, Scientific Technical Report
Becker H and Fassbinder JWE (1999) Magnetometry of a Scythian settlement in Siberia STR98/21, GFZ-Potsdam, pp. 277–286.
near Cicah in the Baraba Steppe. In: Fassbinder JWE and Irlinger WE (eds.) 3rd Halls HC (1978) The use of converging remagnetization circles in paleomagnetism.
International Conference on Archaeological Prospection. Physics of the Earth and Planetary Interiors 16: 1–11.
Bitterly J, Cantin JM, Schlich R, Folques J, and Gilbert D (1984) Portable magnetometer Henkel H (1991) Petrophysical properties (density and magnetisation) of rocks from the
theodolite with fluxgate sensor for earth’s magnetic field component measurements. northern part of the Baltic shield. Tectonophysics 192: 1–19.
Geophysical Surveys 6: 233–239. Hillhouse JW (1977) A method for the removal of rotational remanent magnetization
Blackett PMS (1952) A negative experiment relating to magnetism and the earth’s acquired during alternating field demagnetization. Geophysical Journal of the Royal
rotation. Philosophical Transactions of the Royal Society A 245: 309–370. Astronomical Society 50: 29–34.
Blinov VE, Ginzburg BI, Zhitnikov RA, and Kuleshov PP (1984) Alkali-helium magnetometer Hrvoic I (2001) Standardization of total field magnetometers. Contributions to
with optically oriented potassium atoms. Soviet Physics: Technical Physics 29(2): Geophysics & Geodesy 31: 93–97.
168–171, See also (in Russian): Zhurnal Tekhnicheskoi Fiziki 54: 287–292. Hurwitz L and Nelson JH (1960) Proton vector magnetometer. Journal of Geophysical
Bloch F (1946) Nuclear induction. Physical Review 70: 460–474. Research 65: 1759–1765.
Brunhes B (1906) Récherches sur les directions d’aimentation des roches volcaniques. Irving E, Molyneux L, and Runcorn SK (1966) The analysis of remanent intensities and
Journal de Physique 5: 705–724. susceptibilities of rocks. Geophysical Journal of the Royal Astronomical Society
Bullock SJ and Barritt SD (1989) Real-time navigation and flight-path recovery of aerial 137: 451–464.
geophysical surveys: A review. In: Garland GD (ed.) Proceedings of Exploration 87, Isles D and Rankin L (2013) Geological Interpretation of Aeromagnetic Data. ASEG-SEG
Third Decennial International Conference on Geophysical and Geochemical publication (E-book).
Exploration for Minerals and Groundwater. Ontario Geological Survey, Special Jackson M, Bowles JA, Lascu I, and Slheid P (2010) Deconvolution of U channel
Volume 3, pp. 960. magnetometer data: Experimental study of accuracy, resolution, and stability of
Carver TR and Slichter CP (1953) Polarization of nuclear spins in metals. Physical different inversion methods. Geochemistry, Geophysics, Geosystems 11: Q07Y10.
Review 92: 212–213. http://dx.doi.org/10.1029/2009GC002991.
Carver TR and Slichter CP (1956) Experimental verification of the Overhauser nuclear Jankowski J and Sucksdorff C (1996) Guide for Magnetic Measurements and
polarization effect. Physical Review 102: 975–980. Observatory Practice. Boulder: NOAA/International Association of Geomagnetism
Chave AD, Green AW, Evans RL, et al. (1995). Report of a Workshop on Technical and Aeronomy.
Approaches to Construction of a Seafloor Geomagnetic Observatory. Technical Jankowski J, Marianiuk J, Ruta A, Sucksdorff C, and Kivinen M (1984) Long-term
Report WHOI-95-12, Woods Hole Oceanographic Institution, Woods Hole, USA. stability of a torque-balance variometer with photoelectric converters in observatory
Chevallier R (1925) L’aimentation des laves de l’Etna et l’orientation du champ terrestre practice. Geophysical Surveys 6: 367–380.
en Sicile du XIIe au XVIIe siècle. Annales de Physique 4: 5–162. Josephson BD (1962) Possible new effects in superconducting tunnelling. Physics
Clark AJ, Tarling DH, and Noel M (1988) Developments in Archaeomagnetic dating in Letters 1: 251–253.
Britain. Journal of Archaeological Science 15: 645–667. Kastler A (1950) Quelques suggestions concernant la production optique et la détection
Collinson DW (1983) Methods in Rock Magnetism and Palaeomagnetism: Techniques optique d’une inégalité de population des niveaux de quantification spatiale des
and Instrumentation. New York: Chapman and Hall. atomes. Journal de Physique et Le Radium 11: 255–265.
Creer KM, Irving E, and Runcorn SK (1954) The direction of the geomagnetic field in Kirschvink JL (1980) The least-squares line and plane and the analysis of
remote epochs in Great Britain. Journal of Geomagnetism and Geoelectricity palaeomagnetic data. Geophysical Journal of the Royal Astronomical Society
6: 163–168. 62: 699–718.
David P (1904) Sur la stabilité de la direction d’aimentation dans quelques roches Korepanov V, Berkman R, Rakhlin L, et al. (2001) Advanced field magnetometers
volcaniques. Comptes Rendus de l’Académie des Sciences de Paris 138: 41–42. comparative study. Measurement 29: 137–146.
Delesse A (1849) Sur le magnétisme polaire dans les minéraux et dans les roches. Korhonen JV, Fairhead JD, Hamoudi M, et al. (2007) Magnetic Anomaly Map of the
Annales de Chimie et de Physique 325: 194–209. World, Scale 1:50 000 000, 1st edn. Commission for the Geological Map of the
Deumlich F (1980) Instrumentenkunde der Vermessungstechnik. Berlin (Ost): VEB World/UNESCO.
Verlag für Bauwesen. Langel R, Ousley G, Berbert J, Murphy J, and Settle M (1982) The MAGSAT mission.
De Vuyst AP and Hus J (1966) Généralisation de la Mesure de l’Intensité des Geophysical Research Letters 9: 243–245.
Composantes Verticale et Horizontale du Champ Magnétique Terrestre avec le Larochelle A and Christie KW (1967) An automatic 3-magnet or biastatic magnetometer.
Magnétomètre à Protons. Annales de Geophysique 22: 119–127. Geological Survey of Canada, Paper 67-26, 28 pp.
Lauridsen KE (1985) Experiences with the DI-fluxgate magnetometer: Inclusive theory of Primdahl F (2002) Resonance magnetometers. In: Ripka P (ed.) Magnetic Sensors and
the instrument and comparison with other methods. Danish Meteorological Institute Magnetometers, pp. 267–304. Boston and London: Artech House.
Geophysical Papers, R-71. Primdahl F, Merayo JMG, Brauer P, Laursen I, and Risbo T (2005) Internal field of
Laursen V and Olsen J (1971) Classical methods of geomagnetic observations. homogeneously magnetized toroid sensor for proton free precession magnetometer.
In: Flugge S and Rawer K (eds.) Encyclopedia of Geophysics, Geophysics III/3, Measurement Science and Technology 16: 590–593.
vol. XLIX/3, pp. 276–322. Berlin: Springer Verlag. Rasmussen O (1990) Improvements in fluxgate magnetometers at Danish
Lilley FEM and Day AA (1993) D’Entrecasteaux 1792: Celebrating a bicentennial in Meteorological Institute’s Magnetic Observatories. In: Proceedings of the
geomagnetism. EOS. Transactions of the American Geophysical Union 74: 97, International Workshop on Geomagnetic Observatory Data Acquisition and
102–103. Processing. Finnish Meteorological Institute Geophysical Publications No. 15.
Mackereth FJH (1958) A portable core sampler for lake deposits. Limnology and Rasson JL (1996) Progress in the design of an automatic Diflux. In: Rasson JL (ed.)
Oceanography 3: 181–191. Proceedings of the VIth Workshop on Geomagnetic Observatory Instruments, Data
Mackereth FJH (1969) A short core sampler for subaqueous deposits. Limnology and Acquisition and Processing, Publ. Sci. et Techn. No 003, Institut Royal
Oceanography 14: 145–151. Meteorologique de Belgique, Brussels, pp. 190–194.
Maus S, Barckhausen U, Berkenbosch H, et al. (2009) EMAG2: A 2-arc min resolution Rasson JL (2005) About absolute geomagnetic measurements in the observatory and in
earth magnetic anomaly grid compiled from satellite, airborne and marine magnetic the field. Publication Scientifique et Technique de l’Institut Royal Météorologique de
measurements. Geochemistry, Geophysics, Geosystems 10(8). http://dx.doi.org/ Belgique, 40, pp. 42.
10.1029/2009GC002471. Reeves CV (2001) The role of airborne geophysical reconnaissance in exploration
McElhinny MW (1973) Palaeomagnetism and Plate Tectonics. Cambridge: Cambridge geoscience. First Break 19(9): 501–508.
University Press. Reeves CV (2005) Aeromagnetic Surveys: Principles, Practice and Interpretation.
McElhinny MW, McFadden PL, and Merrill RT (1996) The time-averaged Geosoft Inc, 155 pp. (e-book: http://www.geosoft.com/media/uploads/resources/
palaeomagnetic field. Journal of Geophysical Research 101: 25007–25027. technical-papers/Aeromagnetic_Survey_Reeves.pdf).
McFadden PL (1980) The best estimate of Fisher’s precision parameter k. Geophysical Reeves CV and Bullock SJ (2006) Airborne Exploration. Fugro Airborne Surveys
Journal of the Royal Astronomical Society 60: 397–407. Limited (unpublished).
McFadden PL and Lowes FJ (1981) The discrimination of mean directions drawn from Reeves CV, Macnab R, and Maschenkov S (1998) Compiling all the world’s magnetic
Fisher distributions. Geophysical Journal of the Royal Astronomical Society anomalies. EOS. Transactions of the American Geophysical Union 79: 338.
67: 19–33. Reeves CV, Reford SW, and Milligan PR (1997) Airborne geophysics: Old methods, new
McFadden PL and McElhinny MW (1988) The combined analysis of remagnetization images. In: Gubbins AG (ed.) Proceedings of Exploration 97, Fourth Decennial
circles and direct observations in paleomagnetism. Earth and Planetary Science International Conference on Mineral Exploration, GEO F/X, pp. 13–30.
Letters 87: 161–172. Reford MS and Sumner JS (1964) Aeromagnetics. Geophysics 29: 482–516.
Melloni M (1853) Du magnétisme des roches. Comptes Rendus de l’Académie des Reid AB, Allsop JM, Gransser H, Millett AJ, and Somerton IW (1990) Magnetic
Sciences de Paris 37: 966–968. interpretation in three dimensions using Euler deconvolution. Geophysics 55: 80–91.
Merayo MG, Brauer R, Primdahl F, Petersen JR, and Nielsen OV (2000) Scalar calibration Reigber C, Lühr H, Schwintzer P, and Wickert J (2005) Earth Observation with CHAMP:
of vector magnetometer. Measurement Science and Technology 11: 120–132. Results from Three Years in Orbit. Berlin: Springer, 628 pp.
Meyer O and Voppel D (1954) Ein Theodolit zur Messung des Erdmagnetischen Feldes Roest WR, Verhoef J, and Pilkington M (1992) Magnetic interpretation using the 3-D
mit der Förstersonde als Nullindicator. Deutsche Hydrographische Zeitschrift analytic signal. Geophysics 57: 116–125.
12: 73–77. Sapunov V, Denisov A, Denisova O, and Saveliev D (2001) Proton and Overhauser
Milligan PR, Morse MP, and Rajagopalan S (1991) Pixel map preparation using the magnetometers metrology. Contributions to Geophysics and Geodesy 31: 119–124.
HSV colour model. Exploration Geophysics 23: 219–224. Sapunov V, Rasson J, Denisov A, et al. (2005) Theodolite-borne vector Overhauser
Milligan P, Minty B, Richardson M, and Franklin R (2009) The Australia-wide airborne magnetometer: DIMOVER. In: Yumoto E (ed.) Proceedings of the XIth Workshop on
geophysical survey – Accurate continental magnetic coverage. ASEG Extended Geomagnetic Observatory Instruments. Data Acquisition and Processing. Kakioka.
Abstracts. doi:10.1071/ASEG2009ab075. Schnegg PA and Fischer G (1991) The TURBOMAG vector magnetometer. Münchner
Minty BRS (1992) Simple micro-levelling for aeromagnetic data. Exploration Geophysikalische Mitteilungen 5: 121–130.
Geophysics 22: 591–592. Serson PH and Hannaford WLW (1956) A portable electrical magnetometer. Canadian
Molyneux LA (1971) A complete result magnetometer for measuring the remanent Journal of Technology 34: 232–243.
magnetization of rocks. Geophysical Journal of the Royal Astronomical Society Shifrin VYa, Alexandrov EB, Chikvadze TI, et al. (2000) Magnetic flux density standard
24: 429–433. for geomagnetometers. Metrologia 37: 219–227.
Nagata T (1961) Rock Magnetism, 2nd edn. Tokyo: Maruzen. Spector A and Grant FS (1970) Statistical models for interpreting aeromagnetic data.
Nagata T, Arai Y, and Momose K (1963) Secular variation of the geomagnetic total Geophysics 35: 293–302.
force during the last 5000 years. Journal of Geophysical Research 68: 5277–5281. Stephenson A (1980) Rotational remanent magnetization and the torque exerted on a
Néel L (1949) Théorie du trainage magnétique des ferromagnétiques aux grains fins rotating rock in an alternating magnetic field. Geophysical Journal of the Royal
avec applications aux terres cuites. Annales de Geophysique 5: 99–136. Astronomical Society 62: 113–132.
Néel L (1955) Some theoretical aspects of rock magnetism. Advances in Physics Stephenson A (1981) Gyromagnetic remanence and anisotropy in single domain
4: 191–243. particles, rocks and magnetic recording tape. Philosophical Magazine B
Newitt LR, Barton CE, and Bitterly J (1996) Guide for Magnetic Repeat Station Surveys. 44: 635–644.
Boulder: NOAA/International Association of Geomagnetism and Aeronomy. Tarling DH (1966) The magnetic intensity and susceptibility distributions in some
Nielsen OV, Petersen JR, Primdahl F, et al. (1995) Development, construction and Cenozoic and Jurassic basalts. Geophysical Journal of the Royal Astronomical
analysis of the “Oersted” fluxgate magnetometer. Measurement Science and Society 11: 423–432.
Technology 6: 1099–1115. Tarlowski C, McEwin AJ, Reeves CV, and Barton CE (1996) Dewarping the composite
O’Sullivan KN (1991) A map for all reasons: The role of image processing in mineral aeromagnetic anomaly map of Australia using control traverses and base stations.
exploration. Minerals Industry International, pp. 8–14 (January 1991). Geophysics 61: 696–705.
Overhauser AW (1953a) Polarization of nuclei in metals. Physical Review 91: 476 Tenani M (1941) Nuovo Metodo di misura della declinazione e della inclinazione
(abstract only). magnetica. La Ricerca Scientifica 20: 1135–1140.
Overhauser AW (1953b) Polarization of nuclei in metals. Physical Review 92: 411–415. Thellier E (1933) Magnétometre insensible aux champs magnétiques troubles des
Packard ME and Varian RH (1954) Free nuclear induction in the Earth’s magnetic field. grandes villes. Comptes Rendus de l’Académie des Sciences de Paris
Physical Review 93: 941 (abstract only). 197: 224–234.
Papitashvili VO, Petrov VG, Saxena AB, Clauer CR, and Papitashvili NE (2006) A virtual Trapenese A, Batt CM, and Schnepp E (2008) Sampling methods in archaeomagnetic
global magnetic observatory network: VGMO.NET. Earth Planets Space 58: 765–774. dating: A comparison using case studies from Wörterberg, Eisenerz and Gams
Pozzi JP and Thellier E (1963) Sur des perfectionments récents apportés aux Valley (Austria). Physics and Chemistry of the Earth 33: 414–426.
magnétometre a très haute sensibilité utilisés en minéralogie magnétique et en Trigg DF (1970) A portable D and I magnetometer. Geomagnetic Laboratory Report
paléomagnétisme. Comptes Rendus de l’Académie des Sciences de Paris n 70-3, Direction de la Physique du Globe, Energie, Mines et Ressources, Ottawa,
257: 1037–1040. Canada.
Turner GM, Kamp PJJ, McIntyre AP, Hayton S, McGuire DM, and Wilson GS (2005) Relevant Websites
A coherent middle Pliocene magnetostratigraphy, Wanganui Basin, New Zealand.
Journal of the Royal Society of New Zealand 35: 197–227. http://www.2genterprises.com – 2G Enterprises.
van Loo SA and Rasson JL (2006) Development of an automatic declination-inclination http://www.bartington.com – Information about Bartington and Molspin
magnetometer. In: Rasson JL and Delipetrov T (eds.) Geomagnetics for Aeronautical instrumentation.
Safety: A Case Study in and Around the Balkans. Nato Science Series, Sub-Series http://www.ga.gov.au – Geoscience Australia.
CNetherlands: Springer. http://www.gfz-potsdam.de/pbl/op/champ – CHAMP.
Watson GS (1983) Statistics on Spheres. University of Arkansas Lecture Notes in the http://www.gemsys.ca/ – GEM Systems.
Mathematical Sciences, New York: Wiley. http://www.geosoft.com – Geosoft.
Wienert KA (1970) Notes on Geomagnetic Observatory and Survey Practice. Brussels: http://home.ural.ru/ant/index.html – Quantum Magnetometry Laboratory of Ural State
UNESCO. Technical University.
Wilson RL (1962a) The palaeomagnetic history of a doubly baked rock. Geophysical http://www.intermagnet.org – INTERMAGNET.
Journal of the Royal Astronomical Society 6: 397–399. http://www.intrepid-geophysics.com – Intrepid Geophysics.
Wilson RL (1962b) The palaeomagnetism of baked contact rocks and reversals of the http://www.ngdc.noaa.gov/IAGA/vdat/iagaformat.html – IAGA 2002 Data Exchange
earth’s magnetic field. Geophysical Journal of the Royal Astronomical Society Format.
7: 194–202. http://www.ngdc.noaa.gov/IAGA/vmod/igrf.html – IGRF-11.
Wilson RL and Lomax R (1972) Magnetic remanence related to slow rotation of http://www.ngdc.noaa.gov/wdc/wdcmain.html – World Data Center Systems.
ferromagnetic material in alternating magnetic fields. Geophysical Journal of the http://www-odp.tam.edu/publications/ – Ocean Drilling Programme.
Royal Astronomical Society 30: 295–303. http://physics.nist.gov/cuu/Constants/index.html – NIST: Values of physical constants
Zijderveld JDA (1967) AC demagnetization of rocks: Analysis of results. In: Collinson DW, recommended by CODATA, 2010.
Creer KM, and Runcorn SK (eds.) Methods in Palaeomagnetism. Amsterdam: Elsevier.
5.05 Geomagnetic Secular Variation and Its Applications to the Core
A Jackson, ETH Z€urich, Z€urich, Switzerland
C Finlay, Technical University of Denmark, Kgs. Lyngby, Denmark

5.05.1.1 Historical Background 138
5.05.1.2 Early Theories of the Secular Variation 138
5.05.2 Data 138
5.05.2.1 Catalogs and Compilations of Data 138
5.05.2.2 Surveys, Repeat Stations, and Marine Data 140
5.05.2.3 Observatory Data 140
5.05.2.4 Satellite Data 145
5.05.3 Time-Dependent Models of the Main Field 145
5.05.3.1 Methodologies 146
5.05.3.1.1 Taylor series models 147
5.05.3.1.2 Two-step models 147
5.05.3.1.3 Time-dependent models based on B-splines 147
5.05.4 Historical Field Evolution: Long-Term Secular Variation 151
5.05.4.1 Field Evolution at the Earth’s Surface 151
5.05.4.1.1 The westward drift 152
5.05.4.1.2 Hemispherical asymmetry 154
5.05.4.1.3 Axial dipole decay 154
5.05.4.1.4 Timescale associated with different wavelengths (spherical harmonic degrees) 154
5.05.4.1.5 Evolution of integrated rate of change of vertical field at the Earth’s surface 157
5.05.4.1.6 Geomagnetic jerks 161
5.05.4.2 Evolution of Radial Field at the Core Surface 163
5.05.4.2.1 High latitude, approximately stationary flux lobes 164
5.05.4.2.2 Reversed flux patches 164
5.05.4.2.3 Low-latitude, westward drifting field features 164
5.05.5 Interpretation in Terms of Core Processes 168
5.05.5.1 Maxwell’s Equations and Moving Frames 168
5.05.5.2 The Induction Equation in a Spherical Earth 174
5.05.5.3 The Navier–Stokes Equation 175
5.05.5.4 The Frozen-Flux Approximation 176
5.05.5.4.1 The neglect of magnetic diffusion and its physical consequences 176
5.05.5.4.2 Application of the frozen-flux hypothesis to the generation of secular variation at the core surface 177
5.05.5.4.3 Consequences of frozen-flux approximation at the core surface 177
5.05.5.4.4 Attempts to test the frozen-flux approximation using geomagnetic observations 177
5.05.5.4.5 Theoretical issues concerning the frozen-flux hypothesis 179
5.05.5.5 Other Invariants 180
5.05.6 Summary 181
Acknowledgments 181
References 181
5.05.1 Introduction The source of our knowledge on the timescales we deal with
is primarily the direct historical observations of the field; we
The purpose of this chapter is to review the origins of our review the available data, followed by the treatment of the data
current knowledge of the secular variation of the magnetic to generate mathematical models of the field in space and time.
field, that is, the slow changes that occur on timescales of We then discuss interpretations of these models in terms of
years to centuries. There is clearly an overlap with the descrip- some of the physical processes occurring at the core surface. We
tion of the present geomagnetic field (Chapter 5.02) and also stop short of describing the actual calculation of models of
with Chapter 5.09, which treats changes in the field from fluid flow at the core surface, as this is covered in detail in
centuries to millennia. Volume 8 (Core dynamics), but we lay the groundwork by

138 Geomagnetic Secular Variation and Its Applications to the Core
developing an exposition of the governing equations and the Hutcheson (1990) (see also Hutcheson and Gubbins, 1990)
approximations that are frequently used. and Barraclough (1985), and Jackson et al. (2000) (hereinafter
JJW2000); in addition, the latter authors also developed a sta-
tistical theory for accounting for imprecision in longitude.
5.05.1.1 Historical Background
The Greenwich meridian was adopted as an international
This section gives a very brief overview of the development of longitude standard only in 1884, and some national conven-
geomagnetism and does not purport to be comprehensive. tions remained in use later than that date. Consequently, care
Fuller treatments of the history can be found in various places, must be taken as to which of the particular national conven-
for example, relevant chapters of Merrill et al. (1998) or tions of Paris and the observatory at Pulkovo (Leningrad),
Chapman and Bartels (1940), Malin (1987), or Stern (2002). Washington, or San Fernando were being used. One example
A detailed account of geomagnetism up to 1500 can be found of French marine data measuring longitude from Paris until at
in Crichton Mitchell (1932, 1937, 1939); recent articles on least 1895 has been given by Jackson (1989); this difference of
nineteenth-century geomagnetism are those of Good (1985, 2 130 of longitude between Paris and Greenwich is small, but
1988). Excellent discussions of geomagnetic instruments can extremely significant.
be found in McConnell (1980) and Multhauf and Good
(1987). An authoritative source on virtually every aspect of
5.05.1.2 Early Theories of the Secular Variation
geomagnetic history is the epistle by Jonkers (2000).
It is generally acknowledged that the Chinese were the first Beginning with the seminal works Epistola de Magnete by
to discover the directive property of lodestone, almost certainly Peregrinus (1269) and the better-known De Magnete by
in the AD first century. Its development as a primitive naviga- Gilbert (1600), various authors have sought to explain the
tional device was slow, though the declination had almost Earth’s magnetic field by models, some physical, some mathe-
certainly been discovered in the ninth century and compasses matical. Though Gilbert’s model explained a considerable part
were certainly in use in the eleventh century; early observations of the static field, after the discovery of the secular variation, a
of declination are given by Needham (1962) and Smith and whole new dimension was opened up, requiring explanation.
Needham (1967). The first recorded observation of declination It is not our purpose to adumbrate the numerous models
in Europe was by George Hartmann in 1510; inclination was created over time to explain the temporal variation of the
discovered by Robert Norman in 1576. The fact that the field field. However, recently, Jonkers (2000) has provided just
underwent slow changes with time (the secular variation) was such a list, comprising a remarkable compilation of theories
not discovered until 1635: by comparing a series of records of the field up to 1800, starting with Peregrinus (1269) and
taken at London previously, Henry Gellibrand showed that ending with Churchman’s (1794) petitions to the English
secular variation was a real effect. Relative intensities of the Board of Longitude, requesting acceptance of his theories for
field were measured at the end of the eighteenth century by La use in determining longitude. A shorter description can also be
Perouse, D’Entrecasteaux, and Humboldt, by comparing the found in Jonkers (2003).
periods of oscillation of a magnetic needle at different places.
Measurements of the absolute intensity of the field were not
made until a method was devised by Gauss in 1832 (see, e.g., 5.05.2 Data
Malin, 1982). Gauss published his method in Gauss (1833a);
an English translation of the abstract of a paper read in We refer the reader to Chapter 5.04 for information on how
G€ottingen in December 1832 can be found in Gauss (1833b). measurements of the field are taken and for definitions of the
While early observations of the field are extremely valuable, quantities that are typically reported: the declination D, the
some problems do exist. For example, before the discovery of inclination I, the horizontal and total intensities (H and F,
secular variation, some observations are undated as the need to respectively), and the Cartesian components X, Y, and Z in
record the date was not apparent. The accuracy with which an the easterly, northerly, and downward directions. The avail-
observer’s position was known is also a source of error. ability of different data types varies as a function of time,
Although the measurement of latitude was precise even by the chiefly as a result of the needs for navigation, followed by the
late fifteenth century (e.g., an accuracy of 10 min of arc was drive of scientific curiosity in the eighteenth and nineteenth
claimed by 1484 ( John II’s Commission, 1509)), the measure- centuries. It should be noted that until Gauss’ invention of a
ment of longitude at sea remained a problem until approxi- method for the determination of absolute intensity in 1832,
mately 1770 with the introduction of accurate chronometers only the morphology of the field can be determined from
by Harrison. The result of this poor knowledge of longitude direct measurements.
led to the practice of ‘running down the parallel,’ or sailing to
the correct line of latitude before sailing due east or west along
5.05.2.1 Catalogs and Compilations of Data
that parallel to the desired location. Although this practice
meant that the ship’s company often arrived at their desired The earliest catalogs, of Stevin (1599), Kircher (1641), and
destination, it does mean that large navigation errors could Wright (1657), are deficient in that they contain undated
occur in the quoted positions of magnetic observations. To a observations. Around 1705, the French hydrographer
large extent, these errors can be alleviated by examination of the Guillaume Delisle compiled some 10 000 observations
original ship’s log and plotting the positions on a modern chart. (mostly of declination) in his notebooks, trying to establish
This procedure has been performed for sixteenth-, seventeenth-, regularity in secular acceleration; these were never published
and eighteenth-century data by Bloxham (1985, 1986), but still exist in the Archives Nationales in Paris. The next
Geomagnetic Secular Variation and Its Applications to the Core 139
important compilation of magnetic data was made by Finally, one of the largest single compilations, featuring
Mountaine and Dodson (1757) who claimed to have based over 28 000 data points of all types, is the Catalogue of magnetic
their tables of declination at different points on the Earth on determinations in USSR and in adjacent countries from 1556 to
over 50 000 original observations of the field. The original 1926 in three volumes. It was compiled and published by
observations of this enormous collection were never recorded Russian physicist B. P. Veinberg in 1929–33 (Veinberg,
and are lost: the work merely printed grids of averaged data 1929–1933) and contained original data from Russia and
with no reference to sources, numbers of data, etc. This claim neighboring states, mostly obtained in the first decades of the
regarding the number of data involved has attracted some twentieth century. A review of the previous compilations of
skepticism; however, the work of Jonkers et al. (2003) indicates magnetic data that have been produced over time can be found
that it is undoubtedly the case that the authors’ claim for the in Barraclough (1982).
number of data is true. The early work of Mountaine and Another category of sources comprises time series for spe-
Dodson should almost certainly receive more prominence cific locations, normally major cities where investigators
than it does, representing probably the first large-scale attempt have set up permanent instruments, for instance, at national
to describe the morphology of the field. Maps based on the astronomical observatories. Past observers include Graham,
data were subsequently produced. London clockmaker and the discoverer of diurnal variation
The main era of printed compilations of geomagnetic data and Gilpin in England, academics Celsius and Hi€ orter in
was the nineteenth century, featuring the work of Hansteen, Sweden (who studied the correlation of needle disturbance
Becquerel, Sabine, and van Bemmelen. In 1819, the Norwegian with the occurrence of auroras), MacDonald in Sumatra
astronomer and physicist Christopher Hansteen published (eighteenth century), and, in France, many scholars and
Untersuchungen u€ber den Magnetismus der Erde, which listed astronomers summarized in Alexandrescu et al. (1996a). But
data from land surveys and 73 nautical voyages from 1589 to despite their achievements, a mere handful of cities can boast a
1816. His collection includes many of the great scientific expe- series of more or less regular observations spanning over a
ditions during the latter half of the eighteenth century, includ- century prior to 1800. A review of recent efforts to make
ing Cook’s three voyages, contributing over 6500 declination these data series available to a modern audience is given by
and 1200 inclination observations. Another valuable addition Alexandrescu et al. (1996a), who also list all early geomagnetic

was made by A. C. Becquerel’s Traité Expérimental de l’Electricité observations made in Paris (1541–1883, based in part on
et du Magnétisme (1840), which contains the only comprehen- earlier work by Raulin (1867) and Rayet (1876)); see Figure 1.
sive collection of relative intensities. Several Phil. Trans. papers Other capitals with a sustained tradition of geomagnetic obser-
by astronomer Edward Sabine span the period 1818–70 with vations include London (Barraclough et al., 2000; Malin and
exceptionally good coverage, although, as various authors have Bullard, 1981; see Figure 18), Rome (Cafarella et al., 1992),
noted, they are far from comprehensive. Finally, in the 1890s, and Edinburgh (Barraclough, 1995).
Dutch physicist Willem van Bemmelen processed 165 nautical It should be noted that only when observations in compil-
sources prior to 1741 (Bemmelen, 1899). ations can be confidently ascribed to individual observations
Paris declination Paris inclination

15
74
10
72
5
Declination/degrees
Inclination/degrees
0 70
-5
68
-10
66
-15
-20
64
1600 1700 1800 1900 2000 1700 1800 1900 2000

Year Year
Figure 1 Declination (D) and inclination (I) in Paris during historical times. Declination and inclination data from Paris, based on Alexandrescu
et al. (1996b) and more recent observatory data. The data have been reduced to the current site of Chambon-la-Forêt; early observations are made
by many different observers at several different sites.
(with well-defined times and locations and not derived by made on an expedition was by Weymouth in Frobisher Bay,
interpolation) will they be included in the database of histor- Canada, in 1602. For the first inclination observations south of
ical observations of Jonkers et al. (2003) (hereinafter JJM2003) the equator, one had to wait until 1680 when Benjamin Harry
and used historical field models such as gufm1 described in took observations on board the Berkeley Castle en route to the
JJW2000. far east ( Jackson, 2014).
The nineteenth century saw burgeoning scientific expedi-
tions on land, which included the measurement of intensity as
5.05.2.2 Surveys, Repeat Stations, and Marine Data
well as D and I – thus giving the first vector measurements of
Recent interest in historical secular variation has led to original the field. De Rossel’s measurements of the oscillation time of a
observations being compiled for other time periods; a compre- dip needle in 1791 provided the first measurements of relative
hensive review of available data has been given recently by intensity between several places on the Earth. Humboldt and
JJM2003. It is impossible to detail all characteristics of this Erman also provided well-known relative intensity measure-
dataset. Suffice it to say that the largest part of the dataset ments prior to Gauss’ (1833a) invention of a method to deter-
originates in marine observations of the declination, typically mine absolute intensity (see Malin, 1982). The net result of the
taken for the purposes of navigation. It has been possible to collation of the known historical observations as described in
characterize the accuracy with which observers measured the JJM2003 is a dataset that is summarized in Table 1. The avail-
declination at sea; it is better than half a degree for the able data are summarized in geographic plots in Figures 2–10.
seventeenth and eighteenth centuries taken as a whole. When A vast source of magnetic field data for the twentieth century
one compares this accuracy with modern measurements, it was prepared by the US Coast and Geodetic Survey for the 1965
transpires that the old measurements have a signal/noise world charts (Hendricks and Cain, 1963) and is accessible in
ratio that is not too bad compared to modern measurements, machine-readable form from the World Data Centres. The cutoff
for the simple reason that both types of measurements suffer date for the collection was arbitrarily set at 1900. This dataset has
the contaminating effect of the crustal magnetic field. This been used by many authors over the years and has been the basis
contributes 60s/H degrees of error in declination where s is for many International Geomagnetic Reference Field models; we
the root mean square (rms) horizontal crustal field and H the shall not dwell on a description of this dataset, as it has been
local horizontal field strength; for s 200 nT and described numerous times (e.g., Bloxham et al., 1989; Sabaka
H ¼ 20 000–40 000 nT, the error contribution is 0.3–0.6 et al., 1997). Its temporal distribution dominates the temporal
degrees and thus commensurate with the observational error. distribution of the data used to create gufm1 during the twentieth
An accurate position is of course a crucial part of any century, which is shown in Figure 11. Twentieth-century data are
magnetic measurement, and thus, latitude and longitude characterized by a constant improvement in measurement
need to be known. This poses no problems on land, but at accuracy (see Chapter 5.02). Marine surveys continued in the
sea, the determination of position can be challenging. From twentieth century, the most notable being the voyages of the
the sixteenth century onward, the backstaff provided a method nonmagnetic surveying ship the Carnegie. A new type of data
for latitude determination, often said to be accurate to 10 min emerged with the advent of aeromagnetic surveys, most notably
of arc; the empirical findings of JJW2000 agree with this. For Project Magnet; for details, see Langel (1993).
early data, a well-known difficulty is the imprecision in longi-
tude prior to the invention of the marine chronometer. A very
detailed study of this was undertaken in JJW2000, who showed 5.05.2.3 Observatory Data
that navigational error generally generated a Brownian motion The establishment of the G€ ottingen Magnetic Union
type effect, such that the errors increased with the square root (Magnetische Verein) in 1834 by Gauss and Weber heralded
of voyage duration. Empirically, the data suggest that a typical the establishment of an observatory network at sites around
25-day voyage might accumulate 2 degrees of error (though the world where observations of the magnetic field would be
voyages often achieved much better than this). The effect can made with regularity. With the adoption of the ‘Magnetic
be ameliorated in most voyages by using the fact that the Crusade’ of Sabine, Herschel, and Lloyd by the British learned
voyage arrived at a known location – thus, the whole voyage bodies in 1838, Germany and Britain took the lead in driving
can be retrospectively corrected for the accumulated errors, forward observational geomagnetism (Cawood, 1977, 1979).
giving a lower error. The appropriate model for the errors The number of observatories gradually grew and their distribu-
becomes the so-called Brownian bridge, and the error at the tion increased toward the distribution of today (Figure 12),
midpoint of the journey is reduced to exactly half what it although some former observatories have closed due to a
would have been using the simple Brownian motion model –
hence typically, it is 1 degree for a 25-day journey. The effect of
Table 1 Temporal distribution of data-based geomagnetic
this imprecision in location is to increase the error budget by
measurements; a single record may contain a land sighting and/or up to
an amount depending on the gradient of declination with
three types of measurement (D, I, and intensity (H or F))
respect to longitude – a fairly representative figure is less than
1 degree of declination change per degree of longitude. Hence, Period Records D I H F Total
one can see that even early data have contributions to their
error budget (from observational imprecision, crustal magnetic 1510–89 162 160 2 0 0 162
fields, and longitudinal inaccuracy) that are not too dissimilar. 1590–1699 13 673 12 001 53 0 0 12 054
1700–99 85 070 68 076 1747 0 36 69 859
Inclination was initially measured on land at London (1576);
1800–1930 78 162 71 323 17 723 11 404 4779 105 229
the next extant measurement was taken shortly after at Ura-
Total 177 067 151 560 19 525 11 404 4815 187 304
nienborg (1584) by Brahe. The first example of a measurement
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 2 Geographic data distribution of declination observations made before 1590; n ¼ 160; some points may overlap; cylindrical equidistant projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 3 Geographic data distribution of declination observations made in 1590–1699; n ¼ 12 001; some points may overlap; cylindrical
equidistant projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 4 Geographic data distribution of declination observations made in 1700–99; n ¼ 68 076; some points may overlap; cylindrical equidistant
projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 5 Geographic data distribution of declination observations made in 1800–1930; n ¼ 71 323; some points may overlap; cylindrical
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
-30˚ -30˚
-60˚ -60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 6 Geographic data distribution of inclination observations made in 1590–1699; n ¼ 53; some points may overlap; cylindrical equidistant projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
-30˚ -30˚
-60˚ -60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 7 Geographic data distribution of inclination observations made in 1700–99; n ¼ 1747; some points may overlap; cylindrical equidistant projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 8 Geographic data distribution of inclination observations made in 1800–1930; n ¼ 17 723; some points may overlap; cylindrical
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 9 Geographic data distribution of intensity observations made in 1700–99; n ¼ 36; some points may overlap; cylindrical equidistant projection.
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
60˚ 60˚
30˚ 30˚
0˚ 0˚
−30˚ −30˚
−60˚ −60˚
180˚ 210˚ 240˚ 270˚ 300˚ 330˚ 0˚ 30˚ 60˚ 90˚ 120˚ 150˚ 180˚
Figure 10 Geographic data distribution of intensity observations made in 1800–1930; n ¼ 16 183; some points may overlap; cylindrical equidistant
projection.
25 000
20 000
Frequency
15 000
10 000
5000
0
1600 1700 1800 1900 2000
Date
Figure 11 The overall number of historical data (as described in JJM2003) together with observatory data, twentieth-century survey and repeat station
data, and satellite data used in the construction of gufm1 ( Jackson et al., 2000). Note that this depicts a subset of data available, as some data
selection has taken place, based on criteria designed to avoid the effect of the correlation in errors due to the crust.
Geomagnetic observatory locations 1980–2005
Observatory location
Figure 12 The distribution of observatories operating at some point at some point between 1980 and 2005.
multitude of factors. Some of the history of the growth of

the observatory network can be found in Chapman and Table 2 Types of observations classed as annual means and used for
Bartels (1940). secular variation modeling
From the point of view of studies of the secular variation Flag Annual means derived from Percentage
due to the core, the most important product derived from the
continuous monitoring of the observatories is the so-called 1 Data for all days 72.4
annual mean, representing the yearly averaged value of the 2 Data for quiet days 4.0
geomagnetic elements. Although the current definition of an 3 Preliminary data 1.9
annual mean is a mean overall data, there has, in the past, been 4 Absolute observations only 9.0
some variability in exactly what is reported as an annual mean. 5 Incomplete data (<12 months, but 6 months) 3.4
6 Very incomplete data (<6 months) 1.4
For example, occasionally mean of all data reported from the
7 Limited absolute control (introduced in 1996) 0.4
five International Quiet Days every month have been used as
0 Unknown 7.6
an annual mean. Table 2 shows the frequency of the different
types of data that are included in the definitive annual means observatory in the X, Y, and Z directions) as unknowns at the
data file, held by the World Data Centre for Geomagnetism at same time as solving for the magnetic field. This technique
the British Geological Survey (Edinburgh). This inconsistency, continues to be adopted in the comprehensive series of field
which cannot be corrected retrospectively (since the original models (see in the succeeding text) and works very effectively.
data no longer exist), leads to inevitable difficulties in treating The second approach is to desensitize the observatory data to
the data, because the data contain different amounts of exter- the presence of the bias (see, e.g., Bloxham and Jackson,
nal magnetic field contribution. Compromises are always 1992). An effective way of doing this is to work with the
required in treating historical magnetic data, and so far, these rate of change of the field from the observatory, and hence,
data have been treated as if they were homogeneously first differences of observatory data are used in the ufm and
recorded; perhaps it will be possible to treat them in a way gufm series of models (see in the succeeding text). There
that recognizes their different characteristics in the future. appears to be very little difference in the results of the two
Figure 13 shows the distribution of observatory annual mean approaches.
data through time, from the first observations originating with
the formation of the G€ ottingen Magnetic Union to the
present day.
5.05.2.4 Satellite Data
It is straightforward to treat single observations of the field
(such as those made by surveys or satellites) as being inde- Satellite data play a crucial role in determining a detailed
pendent measurements that can be fitted simultaneously in a global picture of the secular variation. However, an extensive
least-squares process. Some words are in order regarding the discussion of the special character of satellite data can be
treatment of observatory data in time-dependent field model- found in Chapter 5.02, and we shall not duplicate that here.
ing. Observatories obviously supply critical data on the secu- Nevertheless, in Table 3, we list some of the satellites that have
lar variation, and indeed, the accuracy of many of the modern been used for magnetic field determination over time and their
field models rests on the observatory time series. A problem different characteristics.
that must be recognized, however, is the fact that the obser-
vatories are subject to a (quasi-)constant field associated with
the magnetization of the crust in the region that they are
located. If observatory data are mixed with other types of 5.05.3 Time-Dependent Models of the Main Field
data (survey and satellite data), this so-called observatory
bias must be recognized; otherwise, it will bias the solution We now turn to the use that is made of the datasets that have
for the main field because an observatory time series essen- been described in the previous section. The tool that has
tially records it many times. Two approaches have developed been most commonly applied has been spatial spherical
for dealing with this. The first approach, developed by Langel harmonic analysis, first applied by Gauss (1839). His analysis
et al. (1982), is to solve for the observatory biases (three per demonstrated the predominantly internal origin of the field.
600
500
No. of observations per year
400
300
200
100
0
1840 1860 1880 1900 1920 1940 1960 1980 2000
Year
Figure 13 Distribution of observatory annual means through time, reflecting availability as of October 2006. The falloff in recent times is due to the
delay in observatories reporting definitive data to the World Data Centres.
Table 3 Satellite missions of relevance for measurement of the core secular variation
Name Inclination Dates Altitude (km) Accuracy (nT) Remarks
Cosmos 49 50 1964 261–488 22 Scalar

OGO-2 87 1965–67 413–1510 6 Scalar
OGO-4 86 1967–69 412–908 6 Scalar
OGO-6 82 1969–71 397–1098 6 Scalar
Magsat 97 1979–80 325–550 6 Vector
DE-1 90 1981–91 568–23 290 ? Vector (spinning)
DE-2 90 1981–83 309–1012 30(F)/100 Low accuracy vector
POGS 90 1990–93 639–769 ? Scalar, timing problems
UARS 57 1991–94 560 ? Vector (spinning)
Ørsted 97 1999 600–850 a
Vector
CHAMP 87 2000–10 350–460 a
Vector
SAC-C 98 2000–04 702–709 a
Scalar
Swarm 87 2013 460 and 510 a
Three identical satellites
Accuracies refer to the intrinsic accuracies of the instrumentation, combined with the positional and orientation accuracy. The two accuracies given for DE-2 refer to scalar and vector
data, respectively. ? means that the overall accuracy of the missions is difficult to determine.
a
For accuracies of the present missions, see Chapter 5.02. Inclination is measured as the angle at which the satellite crosses the equator while passing from the southern hemisphere
to the northern hemisphere.
The principles of spherical harmonic analysis are described been constructed by interpolating the original data by hand.
in Chapter 5.02. They were applied by many authors in the Useful descriptions of these types of model can be found in
nineteenth and twentieth centuries, with various amendments Barraclough (1978) or Langel (1987).
in order to deal with the fact that primarily nonlinear functions
of Gauss’ coefficients were being measured, namely, D, I, H,
5.05.3.1 Methodologies
and F; such developments are described fully in Barraclough
(1978). As one such example, consider how to treat measure- Chapter 5.02 discusses the mathematical foundations for the
ments of declination in the spherical harmonic inverse prob- determination of static models of the field. In this section, we
lem. We have that will review a selection of the most widely used time-dependent
field models and the techniques used to derive them. We
Y
D ¼ tan 1 [1] restrict attention to models that have been produced specifi-
X
cally as time-dependent; only passing reference is made to
and the northerly (X) and easterly (Y) components are linearly models designed to describe either the static magnetic field or
related to Gauss’ coefficients {gm m
l ; hl } forming the model vec- its rate of change (secular variation) at a particular point in
tor m. Let us write these relations as X ¼ ATx m and Y ¼ ATy m. If we time; for models of this type, we refer the reader to Chapter
rearrange eqn [1] into the form 5.02. Note that a different flavor of time-dependent field
models now exists, in the form of models created as a result
X sin D ¼ Y cos D [2]
of data assimilation. The interested reader is referred to Section 5
one can form a linear constraint on m of the form of Chapter 5.02 of the present volume.
T Our description focuses on models of the magnetic field B
ðsin DÞAx ðcos DÞAy m ¼ 0 [3] that are simultaneously models of its spatial ((r, y, f) in
spherical coordinates) dependence and the temporal depen-
This can be fit in a least-squares sense, but note that the data
dence (t denotes time). The standard technique that is com-
enter in defining the linear relation, rather than as a target
mon to all analyses we will describe is to employ the spherical
for the prediction. Numerous other schemes for dealing
harmonic expansion of the field in terms of Gauss’ coefficients
with nonlinear data are described in Barraclough (1978).
{gm m
l ; hl } for the internal field; some of the most recent models
With the advent of significant computer power, the need to
also incorporate coefficients representing the external field. All
deal with nonlinear data in such a way has diminished, and
the models will employ the Schmidt quasi-normalization com-
iterative schemes, as described in Chapter 5.02, are more
mon in geomagnetism.
commonplace.
A time-dependent model of the field necessarily must be
In the years following the early applications of Gauss’
built using a dataset spanning a period of time, denoted herein
method, the technique was applied to the field at different
[ts, te]. In order that a spherical harmonic analysis can be
epochs, the interest being primarily in the evolution of global
performed, a parameterization is required for the temporal
averages such as the dipole moment. Being before the advent of
variation of the field. The unifying idea, common to all ana-
modern computers, it was impossible to deal with true mea-
lyses, is to use an expansion for Gauss’ coefficients of the form
surements of the field without some preliminary reduction of
X
the data – thus, the source for the spherical harmonic analyses glm ðt Þ ¼ gl fi ðt Þ
i m
[4]
was field values at regular intervals read from charts that had i
where fi are a set of basis functions and the igm l are a set of be required in eqn [7], and it no longer remains an attractive
unknown coefficients. (A similar expansion is of course used method because of numerical instabilities and lack of flexibil-
for hml .) The different models that have been produced over the ity of the parameterization.
last few decades differ in their choice of the fi(t). With an
expansion of the form [4], the unknown coefficients
l ; hl } are denoted as a model vector m, and when linear
{igm i m
5.05.3.1.2 Two-step models
data such as the elements (X, Y, Z) are required to be synthesi- A variety of models have been made by a two-step process:
zed (denoted by vector d), the resulting forward problem is first, making a series of spatial models at particular epochs
linear and of the form and, second, making a series of spatial models by some form
d ¼ Am [5] of interpolation. For example, the International Geomagnetic
Reference Fields (IGRFs) and Definitive Geomagnetic Refer-
where A is often termed the equations of condition or design ence Fields (DGRFs) are strictly snapshot models of the field
matrix and describes how model parameters are combined to for particular epochs, but they can be used to calculate the
give predictions that can be compared to the data. magnetic field at times intermediate between two epochs by
The inverse problem of finding the coefficients m is gen- linear interpolation between the models. As a result, it is
erally solved by finding a model minimizing the least- possible to evaluate the DGRF models at any point in time
squares difference between the model predictions and the between 1900 and the present day, though from a purist
data, sometimes together with a measure of the field com- point of view, they are not strictly time-dependent models
plexity to help resolve the issue of nonuniqueness (see of the magnetic field. The stepping stone between such
Chapter 5.02). More generally, when I, D, F, and H data two-step models and the more sophisticated approach of
are involved so that the relation between the model param- using a spline representation of temporal behavior (see
eters and the data is a nonlinear function (which we write as Section 5.05.3.1.3) was the pioneering paper of Langel
d ¼ f(m)), the model must be found iteratively. If [A]ij ¼ @fi/@ et al. (1986). These authors used a spline temporal basis to
mj, and if Ce is the data covariance matrix, then the model interpolate between single-epoch secular variation models.
solution is sought by an iterative scheme, such as the quasi-
Newton method:
1 T 1 5.05.3.1.3 Time-dependent models based on B-splines
mi + 1 ¼ mi + AT C1
e A A Ce ðd f ðmi ÞÞ [6]
After the mid-1980s, more flexible representations of the time
In eqn [6], mi stands for the model at the ith iterate, and in dependency were introduced. Beginning with Bloxham
principle, the matrix A should be recomputed at every iterate. (1987), who used Legendre polynomials, a variety of functions
Such methods converge very rapidly since the effect of the have been employed. The most commonly used and referenced
nonlinearity is very mild. time-dependent field models along with their timespan and
modeling approach are summarized in Table 4.
The methods employed by different workers have gradually
converged toward the use of B-splines as temporal basis func-
5.05.3.1.1 Taylor series models
tions following the example of Bloxham and Jackson (1992),
The earliest time-dependent models used a Taylor series expan-
who were heavily influenced by the approach of Langel et al.
sion for Gauss’ coefficients of the form
(1986). The initial models used fourth-order (or cubic)
ðt t0 Þ2 B-splines, whereas the order of the splines has gradually risen
glm ðt Þ ¼ glm ðt0 Þ + g_m
m ðt0 Þðt t0 Þ + g€l ðt0 Þ
m
+ [7] to sixth-order (e.g., Lesur et al., 2010) for satellite models
2!
seeking to determine the secular acceleration adequately.
about some central epoch here denoted t0. This expansion is of There are two reasons for the popularity of the B-spline
the form [4], with the identification fi(t) ¼ (t t0)n/n ! and
method. Firstly, when global basis functions such as Legendre
gl ¼ (@ t)ngm
i m
l (t0), the nth time derivative at the central epoch. or Chebyshev expansions are used (see, e.g., Bloxham (1987)
In the case of the Taylor expansion A is a dense matrix. The or Bloxham and Jackson (1989)), the design matrix remains
first models to be produced this way were those of Cain et al.
dense and requires considerable memory for its storage,
(1965, 1967), who produced models GSFC(4/64) and whereas a B-spline basis is a local basis, meaning that the
GSFC(12/66) with temporal expansions truncated at first- basis functions are zero outside a small range (see Figure 14).
derivative and second-derivative terms, respectively. The trun- This fact leads to a design matrix that is sparse (in fact, it is
cation level was subsequently raised to third-derivative terms banded), and storage requirements are minimized. Secondly,
in the model GSFC(9/80) of Langel et al. (1982). More the B-splines provide a flexible basis for smoothly varying
recently, Taylor series expansion techniques have been used descriptions of data. One can show that of all the interpolators
to provide time-dependent models of satellite data, covering passing through a time series of points i), i ¼ 1, N), an
(say, f(t
the only short intervals of a few years. For example, Olsen expansion in B-splines of order 2n f^ðt Þ say is the unique
(2002) used a first-order expansion and Maus et al. (2005) interpolator that minimizes a particular measure of roughness
used a second-order expansion; for such models of satellite
Rn (see, e.g., De Boor, 2002):
data covering only a few years, Taylor series expansion models
are reasonable. ð te " n ^ #2
@ f ðt Þ
When one wishes to produce a model of the field spanning Rn ¼ dt [8]
ts @t n
a long time period, it is clear that a large number of terms will
Table 4 Characteristics of widely used models of the time-varying magnetic field
Model L N Time period Expansion Regularized? References
GSFC(4/64) 5 2 1940–63 Taylor No Cain et al. (1965)

GSFC(l2/66) 10 3 1900–66 Taylor No Cain et al. (1967)
GSFC(9/80) 13 4 1960–80 Taylor No Langel et al. (1982)
MFSV/1900/1980/OBS 14 8 1900–80 Legendre Yes Bloxham (1987)
14 10 1820–1900, 1900–80 Chebyshev Yes Bloxham and Jackson (1989)
ufm1, ufm2 14 63 1690–1840, 1840–1990 B-spline Yes Bloxham and Jackson (1992)
gufm1 14 163 1690–90 B-spline Yes Jackson et al. (2000)
CM3 13 14 1960–85 B-spline integrals Yes Sabaka et al. (2002)
CM4 13 24 1960–2002.5 B-spline integrals Yes Sabaka et al. (2004)
CHAOS 18 10 1999–2006 B-spline and Taylor Yes Olsen et al. (2006)
CHAOS-4 20 38 1997–2013.5 B-spline Yes Olsen et al. (2014)
POMME 15 3 2000–10 Taylor Partly Maus et al. (2005)
POMME-6 16 3 2000–10 Taylor Partly Maus et al. (2010)
GRIMM 14 10 2001–06.7 B-spline Yes Lesur et al. (2008)
GRIMM-2 16 15 2001.0–09.5 B-spline Yes Lesur et al. (2010)
COV-OBS 14 90 1840–2010 B-spline Prior used Gillet et al. (2013)
gufm-sat 24 45 2000–10 B-spline Yes Finlay et al. (2012)
L is the maximum degree of the internal secular variation; N is the number of temporal basis functions used for each Gauss’ coefficient. Models covering the recent satellite era,
the last 15 years or so, often have sophisticated parameterization schemes that cannot be adequately described in the table. In the case of models that have a sequence of
models (CHAOS, POMME, and GRIMM), only the original and latest published sibling are listed.
B-spline basis (10 splines)
0.6
0.5
B-spline value B(t)
0.4
0.3
0.2
0.1
0.0
0 1 2 3 4 5 6 7
Time (t)
Figure 14 B-splines of order 4 (cubic B-spines). Local temporal basis of cubic B-splines used in the construction of time-dependent geomagnetic
field models.
ð te h i2
The idea of attempting to construct a smooth representa- ðn Þ ðn2 Þ
N¼ rh 1 @ t Br dOdt [9]
tion in time is an application of ‘Occam’s razor’ that there ts
should be no extra detail in the representation than that truly
demanded by the data. This idea of ‘regularization’ has been where Br is the radial field on the CMB. The models produced by
employed in many of the models of Table 4 from that of Bloxham et al. use n1 ¼ 0 and n2 ¼ 2 in one norm and n1 ¼ 1 and
Bloxham (1987) onward. Those models that employ regulari- n2 ¼ 0 (approximately, to be precise the ohmic heating norm of
zation typically minimize a combination of norms N on the Gubbins (1975) is used) in a second norm; this is slightly
core–mantle boundary (CMB) of the form different to the choices made by Sabaka et al. in their
comprehensive models and the CHAOS time-dependent model Comprehensive models take into account not only the
of satellite data (see the succeeding text). A rather different form time-varying core magnetic field (out to degree 13) but also
of regularization was recently proposed by Jackson (2003) that the static crustal field from degree 14 to degree 65. Because a
involves maximizing the entropy of the field model rather than model of the lithospheric field to this degree captures only a
penalizing spatial or temporal gradients. This new method has small proportion of the total lithospheric signal, it is necessary
so far been used to produce single-epoch models, but it can also to also solve for 1635 observatory biases, generally three com-
be applied to provide both spatial and temporal regularizations ponents at each observatory. The novel features of the model
of time-dependent model; see Gillet et al. (2007). arise in its very sophisticated treatment of the external mag-
Regularized field models are found by minimizing an netic fields, and we will discuss these in some detail.
objective function consisting of a measure (often the L2 least- The ionospheric field is modeled as currents flowing in a
squares norm) of the misfit of the time-dependent model to thin shell at an altitude of 110 km. This leads to magnetic fields
the data along with spatial and temporal norms measuring the that are derived from potentials below and above this layer,
field complexity. The relative weights of the spatial and tem- which influence the observatory and satellite data, respectively
poral norms are scaled by the sizes of the so-called damping (since all the satellites fly above this layer). In quasi-dipole
parameters lS and lT. The choice of the damping parameters is coordinates, the currents are allowed to vary with 24, 12, 8,
made by trading off the desire that the data be fit within their and 6 h periods, as well as annually and semiannually.
estimated errors, the desire that the spatial complexity of the Induced fields are accounted for by assuming that the conduc-
time-dependent model at the core surface be compatible with tivity distribution of the Earth varies only in radius, which
accurate single-epoch models, and the requirement that no means that an external spherical harmonic can only excite its
unnecessary temporal oscillations be introduced. The pub- corresponding internal spherical harmonic. The magneto-
lished models satisfy each of these criteria. spheric field is also parameterized in a similar way, with both
daily and seasonal periodicities, but also a modulation is
allowed based on the Dst index. In order to take into account
5.05.3.1.3.1 The ufm1, ufm2, and gufm1 models the poloidal F-region currents through which the satellites fly,
The ufm1/ufm2 and gufm1 field models share a common aim, a parameterization is made in terms of a toroidal magnetic
namely, to model the long-term secular variation at the core field, which also has periodic time variations.
surface as accurately as possible over the past few centuries. They The model is estimated by an iteratively reweighted least-
were built using the cubic B-spline basis with knots every 2.5 squares method, using Huber weights, and the core contribu-
years and from the largest datasets possible at the time: ufm1/ tion is regularized as in eqn [9] using n1 ¼ 2 and n2 ¼ 1 in one
ufm2 used over 250 000 data originating from old ships’ logs, norm and n1 ¼ 0 and n2 ¼ 2 in another. This difference from
survey data, observatories, and satellite missions. A description the ufm/gufm method simply represents a different approach;
of the oldest data can be found in Bloxham (1986b) and the fundamental quantity in the comprehensive models is the
Bloxham et al. (1989). The gufm1 model was built from similar secular variation @ tBr, which has an expansion in B-splines, and
data from the twentieth century but a vastly expanded historical the main field Br is found as the integral of this using the 1980
dataset, described in Jonkers et al. (2003) – the model contains value as the offset or integration constant. All the other param-
over 365 000 data and consists of 36 512 parameters. Figure 11 eters are regularized in a similar way, by smoothing on spheres
shows the time distribution of the data used in gufm1. No at different altitudes, representing the physical locations of the
account is explicitly taken of external fields in these models. sources. In total, CM4 consists of 25 243 free parameters. At the
time of writing, a new version of the comprehensive model,
CM5, was in preparation (Olsen, personal communication).
5.05.3.1.3.2 The comprehensive models
An effort began in the early 1990s to build a comprehensive
5.05.3.1.3.3 Field models of recent satellite data
series of field models that took account of many effects that are
Over the period 1999 to the present, the satellites Ørsted,
recorded in geomagnetic data in addition to the core secular
CHAMP, and SAC-C have provided unprecedented coverage of
variation. The first model was reported by Sabaka and Baldwin
the Earth’s magnetic field. The wish to exploit these data has
(1993); Sabaka et al. (2002) described the most recent model
spawned a series of field models covering only this recent satel-
formulation in detail, while Sabaka et al. (2004) discussed its
lite era that have been regularly updated by the authors; these
extension to include Ørsted and CHAMP satellite data. We will
are the CHAOS, GRIMM, and POMME models. In Table 4, the
specifically report on the CM4 model of Sabaka et al. (2004).
first and last of each series are noted. A recent review of these
In general terms, the model includes representations of the
models can be found in Gillet et al. (2010). Sophisticated meth-
main field, its secular variation, and both local-time (Sun-
odologies, which often include solving simultaneously for the
synchronous) and seasonal modes of the magnetospheric
Euler angles relating the instrument to the satellite, are beyond
and ionospheric fields, as well as describes ring-current varia-
the scope of this chapter (see Chapter 5.02).
tions through the Dst index and internal fields induced by
time-varying external fields. The data used in creating the
model consist of POGO, Magsat, Ørsted, and CHAMP satellite 5.05.3.1.3.4 Comparison between CM3/CM4 and gufm1
data (totaling over 1.6 million observations) and over 500 000 and discussion
observatory data; the latter consist of either a 1.00 a.m. obser- To illustrate the fidelity with which the present field models are
vation (actually an hourly mean) on the quietest day of the able to model observatory data, we show in Figure 15 a compar-
month during the 1960–2002.5 period or observations every ison of model gufm1’s predictions with some observatory annual
2 h on quiet days during the POGO and Magsat missions. mean datasets.
CLF
40
X (nT year–1)
30
20
10
0
1940 1950 1960 1970 1980 1990 2000
Time / years
60
50
Y (nT year–1)
40
30
20
10
1940 1950 1960 1970 1980 1990 2000
Time / years
50
40
Z (nT year–1)
30
20
10
0
1940 1950 1960 1970 1980 1990 2000
(a) Time / years
ABG
80
X (nT year–1)
40
-40
1920 1940 1960 1980 2000

Time/years
20
Y (nT year–1)
-20
-40
1920 1940 1960 1980 2000
Time/years
120
Z (nT year–1)
80
40
0
-40
1920 1940 1960 1980 2000
(b) Time/years
Figure 15 Comparison of secular variation models and first differences of annual means. Two observatories (a) Chambon-la-Forêt, France, and
(b) Alibag, India, are shown comparing observed field rate of change with predictions from the model gufm1 (solid lines). The symbols show the rate of
change of the field, as obtained from first differences of annual means. The X, Y, and Z components are in the northerly, easterly, and downward
directions, respectively. Because the post-1990 data were not used in the creation of gufm1, there is a small mismatch at the end of the data series – this
shows the difficulty in predicting the secular variation.
To show CM4’s performance on very short timescales, In the next section, we move on to describe the characteristics
Figure 16 compares the model to hourly mean values for the of secular variation as observed on the Earth’s surface and
month of April 1990, data that were not used in deriving the inferred at the CMB. We will ultimately (Section 5.05.5) describe
model. It is clear that the model is capable of predicting vari- possible underlying physical mechanisms in terms of core hydro-
ations rather well, though with more difficulty at the Antarctic magnetics. Most of the results shown in the next section (unless
station SBA (Scott Base). explicitly stated otherwise) are derived from the gufm1 field
Table 5 compares the performance of models gufm1 and model of Jackson et al. (2000), which, as we have described,
CM3 against observatory data, showing almost identical per- provides a good representation of the historical field evolution.
formance. This comes about principally because of the large
intrinsic variance of the data at some observatories, which
neither field model is able to capture. 5.05.4 Historical Field Evolution: Long-Term Secular
Figure 17 shows a comparison of the model predictions for Variation
the variation in the first 6 Gauss’ coefficients over century and
5.05.4.1 Field Evolution at the Earth’s Surface
decade timescales. Although small differences exist, particu-
larly in estimates of the instantaneous secular variation, it is The magnetic field at the Earth’s surface has changed signifi-
apparent that modeling has reached a stage where there is cantly over the past 400 years. This can clearly be seen, for
considerable consensus between the models. example, in the long times series of measurements cataloged by
CTA
31 400
X (nT)
31 300
31 200
4300
Y (nT)
4250
4200
-38 250
-38 275
Z (nT)
-38 300
5 10 15 20 25 30
(a) April 1990 (days)
Figure 16 Comparison of 1 month (April 1990) of hourly mean data. Observed X, Y, and Z components (dots) from selected observatories and
predictions (solid line) from CM4. (a) Charters Towers,
(Continued)
KAK
29 900
X (nT) 29 800
29 700
-3550
-3600
Y (nT)
-3650
35 000
Z (nT)
34 950
34 900
5 10 15 20 25 30
(b) April 1990 (days)
Figure 16 (Continued) (b) Kakioka,
Malin and Bullard (1981) (Figure 18). In fact, it was such 5.05.4.1.1 The westward drift
measurements of changes in declination by Henry Gellibrand Perhaps the most striking aspect of the geomagnetic secular
in 1634 that first indicated the existence of geomagnetic secular variation over the past 400 years is the westward motion of the
variation. Note that in this figure, since declination and incli- field at the Earth’s surface. This phenomenon has been recogni-
nation are nonlinearly related to the model m, it is not possible zed since the time of Halley (1683, 1692) and was first analy-
to account for observatory crustal biases. zed in detail by Bullard et al. (1950), who concluded that the
The best way to appreciate global field changes (i.e., secular nondipole part of the field had moved westward at a rate of
variation) is for the reader to study contour maps of different 0.18 degrees per year during the first half of the twentieth
field components and compare how they have evolved. In century. Bullard et al. (1950) and later Yukutake (1962) sug-
Figure 19, the declination (D) at the Earth’s surface is shown gested that the westward drift was not globally constant, but
in AD 1590 and in AD 1990, while Figure 20 shows the rather depended on latitude; subsequently, Yukutake and
inclination anomaly (Ia) (defined as the difference between Tachinaka (1969) realized that it could be better explained by
the observed inclination and that of a geocentric axial dipole) separating the field into standing and drifting parts. The latitu-
at the same epochs. Figures 21–23 catalog the evolution of the dinal dependence of the drift rate was conclusively demon-
vertical component of the field (which is much larger than the strated by Jault et al. (1988).
horizontal components except at low latitudes) at 1590, 1690, The westward motion of the field is most easily seen by
1790, 1890, and 1990 A.D. following the motion of the agonic lines (where D ¼ 0) in
SBA
-9600
-9800
X (nT)
-10 000
-10 200
4600
Y (nT)
4400
4200
-66 800
Z (nT)
-67 000
5 10 15 20 25 30
(c) April 1990 (days)
Dst index
0
-40
-80
-120
Dst (nT)
-160
-200
-240
-280
5 10 15 20 25 30
(d) April 1990 (days)
Figure 16 (Continued) (c) Scott Base, and (d) the Dst index for April. Note the commencement of a magnetic storm on the tenth day. The Dst index is
used in the synthesis of predictions at individual observatories; this is particularly noticeable in the predictions of the X component in (a) and (b).
Table 5 Comparison of rms differences (in nT) between observatory annual means and predictions from the models gufm1 and CM3, the latter with
or without its external contribution
Component Number of data gufm1 CM3 (all) CM3 (no external)
X 4047 17.71 17.48 18.09

Y 4047 21.27 21.45 21.47
Z 4047 24.55 24.49 24.53
Figure 19. It can be seen that in 1590, one agonic line bisected (2008) revisited this analysis and found, using a Bayesian pro-
the African continent, running through the Cape of Good cedure, that no change in the dipole moment was the most
Hope (which at this time was named Cape Agulhas (‘Needle parsimonious description over this period while also noting the
Cape’) by sailors due the coincidence of the directions of true large error bars on the inferred rate. Rejecting some of the
and magnetic north there); fast forwarding 400 years to AD archeomagnetic data as outliers and using an alternative data
1990, we find that the same agonic line has now moved error estimate scheme, Suttie et al. (2011) more recently found
westward so that it now bisects southern America. The maxima a rate prior to 1840 largely consistent with that post-1840. Their
and minima of inclination anomalies centered on low lati- favored decay rate is 12 nT year1. Figure 24 shows the extrap-
tudes can also be tracked westward, for example, the inclina- olation of Barraclough (1974) used by Jackson et al. (2000), the
tion anomaly high that was present over Africa in 1590 now result of Gubbins et al. (2006), and the result of Suttie et al.
lies on the western edge of South America. Contour maps of (2011). The variability in the rate of change of the axial dipole
the vertical component of the magnetic field are dominated by over the past four centuries remains an open issue, demanding
the axial dipole component of the field that is unchanged by more well-dated archeomagnetic data for its resolution.
westward motion due to its axisymmetric nature; however, the
westward motion of nonaxial dipole parts of the field can still 5.05.4.1.4 Timescale associated with different wavelengths
be discerned in the maps of Figures 21–23, especially by (spherical harmonic degrees)
following long-lived distortions in the magnetic equator. A useful statistical estimate of how changes in the Earth’s
A southwest to northeast trending element of the magnetic magnetic field at the surface depend on the length scale
equator can be followed from its initial location at the Indian under consideration is the reorganization (or correlation)
Ocean in 1590, to Africa in 1790 and the Atlantic in 1890, to time t(l) introduced by Hulot and LeMouël (1994):
the eastern edge of South America in 1990. vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u X m 2 m 2
u g + hl
tðlÞ ¼ u m l
tX 2 m 2 [10]
g_m + h_
m l l
5.05.4.1.2 Hemispherical asymmetry
The description of westward motion of field features in the This quantity is a measure of how long it takes for power at
previous section focused on high-amplitude features moving spherical harmonic degree l to be completely changed (altered by
across the Atlantic hemisphere (longitude 90 E to 90 W); in an amount equal to its current value) given its present rate of
contrast, the field evolution in the Pacific hemisphere is change. Physically, this corresponds to the time taken to
characterized by lower amplitude features and a lack of system- completely reorganize field features of a particular size. In order
atic secular variation. The asymmetry between the hemispheres to calculate t(l), one requires only a model of the main field and
was first discussed by Fisk (1931), and it has been suggested its time derivative at a given time. In Figure 25, t(l) derived from
that this could be a consequence of the influence of lower the CHAOS-3 model (Olsen et al., 2010a) is presented.
mantle inhomogeneities on the dynamo in the core (Doell The CHAOS-3 model contains global datasets from the
and Cox, 1971). This interpretation is not necessarily unique Ørsted, CHAMP, and SAC-C satellites (see earlier description)
however, as it is known that asymmetrical field morphologies and is broadly similar to other satellite models in Table 4.
are transiently possible during highly supercritical core convec- Numerous studies beginning with Olsen et al. (2006) and
tion, even in the absence of inhomogeneous boundary condi- Holme and Olsen (2006) studied functional forms for t(l)
tions (see Volume 8). and found that power laws of the form
tðlÞ ¼ t1 lg [11]

5.05.4.1.3 Axial dipole decay
The westward drift is only part of the observed secular variation. provided a good fit to the calculated values of t(l) when the
The largest contribution to present-day secular variation comes two free parameters t1 and g were chosen appropriately.
from the decay of the axial dipole part of field. The axial dipole Lhuillier et al. (2011) have shown that one should be careful
has decayed rapidly at an average rate of 5% per century since in deriving models for this quantity: if Gauss’ coefficients are
the first direct measurements of intensity (Barraclough, 1974). independent and identically distributed zero-mean Gaussian
Gubbins et al. (2006) have used paleointensity measurements processes, then the probability density function (pdf ) for t can
from the database of Korte et al. (2005) along with estimates of be calculated. This predicts a mean and a skewed dispersion, to
field directions from gufm1 to infer g01 for the interval which the observations can be compared. They showed that a
1590–1840 and found that field decay rate was much slower parsimonious explanation of the data is in the form of a
(almost constant) during this earlier interval. Indeed, Finlay distribution
g01 g11 h11
6000
−30 000 −1800
−2000
−31 000 5000
−2200
Coefficient/nT
Coefficient/nT
−32 000 −2400
4000
−2600
−33 000
−2800
3000
−34 000 −3000
−3200
−35 000 2000
−3400
−36 000 −3600
1000
1600 1700 1800 1900 2000 1600 1700 1800 1900 2000 1600 1700 1800 1900 2000
Geomagnetic Secular Variation and Its Applications to the Core

Time/years Time/years Time/years
g02 g12 h12

1000
3000
500 2000
2800
0
1000
Coefficient/nT
Coefficient/nT
2600
−500
0
2400
−1000
2200 −1000
−1500
−2000 2000 −2000
1600 1700 1800 1900 2000 1600 1700 1800 1900 2000 1600 1700 1800 1900 2000
(a) Time/years Time/years Time/years
Figure 17 Comparison of model values for the first 6 Gauss’ coefficients. (a) 1590–1990.
(Continued)
155
156
g01 g11 h11
−29 600 −1700 5800
−29 700 5700

−29 800 −1800
5600
Coefficient/nT
Coefficient/nT
−29 900
−1900 5500
−30 000
−30 100 5400
−2000
−30 200 5300
−30 300 −2100
5200
−30 400
5100
1960 1970 1980 1990 2000 1960 1970 1980 1990 2000 1960 1970 1980 1990 2000
Time/years Time/years Time/years
−1500 g02 g12 h12

3070
−1600 −2000
3060
−1700
3050 −2100
Coefficient/nT
−1800
Coefficient/nT
3040 −2200
−1900
3030
−2000 −2300
3020
−2100
3010 −2400
−2200
3000 −2500
−2300
2990
1960 1970 1980 1990 2000 1960 1970 1980 1990 2000 1960 1970 1980 1990 2000
(b) Time/years Time/years Time/years
Figure 17 (Continued) (b) 1960–2002.5. Solid is gufml, dashed is CM4, and the triangles are DGRFs. In (a), g01 has been fixed to decrease at a rate of 15 nT year1 prior to 1840; in the absence of intensity
data, it is necessary to fix the amplitude of the solutions.
London declination
London inclination
10
5 74
Declination/degrees
Inclination/degrees
72
-5
-10 70
-15
68
-20
-25 66
1600 1700 1800 1900 2000
1600 1700 1800 1900 2000
(b) Year
(a) Year
Figure 18 Declination (D) and inclination (I) in London during historical times. Declination and inclination data from London (Malin and Bullard, 1981)
and the fit by gufm1 of Jackson et al. (2000) (line). The data were not used in the construction of gufm1 and provide an independent check of its
fidelity. Note that the inclination prior to 1700 is very poorly constrained, and the model gufm1 differs from some archaeomagnetic measurements of
inclination in Europe (Le Goff, personal communication).
tðlÞ ¼ t1 =l [12] GRIMM-3, Christensen et al. (2012) concluded that tSV was
approximately 13 years for degrees between 2 and 10. Intrigu-
which contains only one free parameter t1. Of note are the facts ingly, they found similar behavior in numerical geodynamo
that one must take into account the width of the theoretical pdf models, provided the magnetic Reynolds number Rm 1000,
when comparing with data and that a one-parameter fit is and traced this timescale back to the timescale of convective
preferable to the two-parameter fit [11] provided that it satisfies flow accelerations. They concluded that convection of the type
the data. Lhuillier et al. (2011) showed that models CHAOS-3, found in current geodynamo simulations is sufficient to repro-
GRIMM-2, and POMME-6 all gave values of t1 in the range duce the magnitude of the (large-scale) rapid changes of secular
406–425 years. There is theoretical justification for the model variation found in recent field models of the satellite era.
[12], since it was shown to be representative of numerical Gillet et al. (2013) have carried out synthetic tests related
dynamo models (Christensen and Tilgner 2004). In analyzing to this topic and suggested that caution should be exercised
the time average of gufm1 over the last 150 years, Christensen in such comparisons. For tSV(l) to be correctly determined,
and Tilgner (2004) found a value of t1 of 535 years and an the secular acceleration power at that degree must first be
excellent fit to the equivalent dynamo model values. This corre- accurately known. Unfortunately, temporal regularization of
spondence is valuable because it offers a way to determine the the form employed in most existing field models (i.e., mini-
magnetic Reynolds number in the core (see Section 5.05.5.4.2); mization of the second or third time derivatives) can strongly
Christensen and Tilgner (2004) found Rm 1200. influence the inferred instantaneous secular acceleration
More than a decade of continuous monitoring by satellite power. If very rapid time changes of the core field take place,
has facilitated broad consensus on timescales for changes in these will be smoothed by the modeling procedure, and artifi-
the magnetic field [10], at least up to spherical harmonic cially long tSV(l) mistakenly inferred. tSV(l) should not be
degree 10. This agreement has encouraged workers to push regarded an instantaneous quantity but rather as a time-
further and to investigate the timescale on which the secular averaged quantity, due to the resolution kernel implicit in the
variation itself has changed, that is, the square root of the ratio modeling (e.g., Silverman, 1984), with the amount of time-
of the secular variation power per degree to the secular accel- averaging varying with degree l. The same form of temporal
eration power per degree smoothing should be applied to dynamo model output when
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi calculating tSV(l), if one wishes to compare directly with a field
uX m 2
u
u m g_l + h_l
m 2 model, and this places restrictions on the minimum timescale
tSV ðlÞ ¼ u
tX 2 m 2 [13] of flow accelerations that can be compared.
m l
g€m + h€l For further discussion on the core processes underlying
secular variation, see Chapter 8.04.
Lesur et al. (2008) were the first to explicitly discuss this
quantity. Holme et al (2011), investigating it for a preliminary 5.05.4.1.5 Evolution of integrated rate of change of vertical
version of the CHAOS-4 model (Olsen et al., 2010b), found it field at the Earth’s surface
to be approximately 10 years independent of spherical har- It is also interesting to consider the evolution of a global
monic degree. Considering a number of field models, including measure of the amplitude of the instantaneous rate of change
(a) −180.0 −160.0 −140.0 −120.0 −100.0 −80.0 −60.0 −40.0 −20.0 0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0
(b) −180.0 −160.0 −140.0 −120.0 −100.0 −80.0 −60.0 −40.0 −20.0 0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0
Figure 19 Historical change in declination at the Earth’s surface. Declination D at the Earth’s surface in (a) AD 1590 and (b) AD 1990 from the model
gufm1 of Jackson et al. (2000). Plots are Lambert equal area projections of the Atlantic and Pacific hemispheres. Color bars are at 20 intervals,
red being negative and blue positive. Note the westward displacement of the agonic lines where D ¼ 0.
of vertical field at the Earth’s surface. A suitable measure is the century but gradually increased resolution of the rapid varia-
rms of B_r integrated over the surface. This quantity is plotted in tions in the models produced most recently, with COV-OBS
Figure 26 from 1840 to 2010, for the ufm1, gufm1, and COV- being the most rapidly varying. These changes are known to
OBS models. be robust as all models are a good representation of the
Dramatic changes in the integrated instantaneous rate of globally averaged field evolution at the core surface, as is
Br are observed to have occurred during the twentieth century. evident in comparisons of the model prediction with
Notice in Figure 26 the same basic patterns for the twentieth observed secular variation. Particularly dramatic is the 20%
(a) −45.00−40.00−35.00−30.00−25.00−20.00−15.00−10.00 −5.00 0 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00
(b) −45.00−40.00−35.00−30.00−25.00−20.00−15.00−10.00 −5.00 0 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 45.00
Figure 20 Historical change in inclination anomaly at the Earth’s surface. Inclination anomaly Ia at the Earth’s surface in (a) AD 1590 and (b) AD 1990
from the model gufm1 of Jackson et al. (2000). Plots are Lambert equal area projections of the Atlantic and Pacific hemispheres. Inclination
anomaly is the inclination of the field minus that expected for a geocentric axial dipole. Color bars are at 5 intervals, red being negative and blue positive.
increase in the amplitude of the secular variation at the start flow speeds altered precisely in the required way for observed
of the twentieth century. This is thought to be associated with decadal length of day changes to be explained by geostrophic
an increase in the rms core flow velocity; Hulot et al. (1993) core motions ( Jackson, 1997). It is also remarkable that there
inferred by inversion of the observed secular variation (see are a number of local maxima and minima in Figure 26
Volume 8) that the rms flow speed increased at this time, and occurring throughout the twentieth century. These extrema
it appears to be the case that zonal (axisymmetric) core seemingly mark reorganizations of the global secular
(a) −90 000 −80 000 −70 000 −60 000 −50 000 −40 000 −30 000 −20 000 −10 000 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
(b) −90 000 −80 000 −70 000 −60 000 −50 000 −40 000 −30 000 −20 000 −10 000 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
Figure 21 Historical change in Bz at the Earth’s surface: AD 1590 and AD 1690. Vertical magnetic field Bz at the Earth’s surface in (a) AD 1590 and
(b) AD 1690 from the model gufm1 of Jackson et al. (2000). Plots are the Mollweide projection; each color bar represents a 10 000 nT increment.
(a) –90 000 –80 000 –70 000 –60 000 –50 000 –40 000 –30 000 –20 000 –10 000 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
(b) –90 000 –80 000 –70 000 –60 000 –50 000 –40 000 –30 000 –20 000 –10 000 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
Figure 22 Historical change in Bz at the Earth’s surface: AD 1790 and AD 1890. Vertical magnetic field Bz at the Earth’s surface in (a) AD 1790 and
(b) AD 1890 from the model gufm1 of Jackson et al. (2000). Plots are the Mollweide projection; each color bar represents a 10 000 nT increment.
variation and at least some of them appear coincident with 5.05.4.1.6 Geomagnetic jerks
the so-called geomagnetic jerks that are discussed in the next Geomagnetic jerks or secular variation impulses are abrupt
section. Note that the situation for the nineteenth century is changes in the second time derivative of the geomagnetic field at
far less clear. the Earth’s surface (see, e.g., Courtillot and Le Mouël, 1984).
–90 000 –80 000 –70 000 –60 000 –50 000 –40 000 –30 000 –20 000 –10 000 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000
Figure 23 Historical change in Bz at the Earth’s surface: AD 1990. Vertical magnetic field BZ at the Earth’s surface in AD 1990 from the model gufm1 of
Jackson et al. (2000). Plots are the Mollweide projection; each color bar represents a 10 000 nT increment.
-25 000
-30 000
g01/nT
-35 000
-40 000
-45 000
1600 1700 1800 1900 2000

Date
Figure 24 Change in the axial dipole component (g01) in past 400 years. g01 in nT as determined by archaeomagnetic measurements (Suttie, personal
communication). The measurements are transformed into estimates of g01 by using the known morphology of the field given by model gufm1 of
Jackson et al. (2000). The figure shows a box and whisper plot, where the gray box represents the interquartile range (IQR) of the data and the whisper
extends to the last datum within 1.5 times the IQR. The triangle shows the median of the data. Data have been binned in 40 year bins. Also drawn
is the variation of g01 post-1840 (solid line) as given by the model gufm1, and prior to 1840, three curves are drawn: Solid is the dipole variation
of 12 nT year1 as determined by Suttie et al. (2011), dashed is the dipole variation of 2 nT year1 as determined by Gubbins et al. (2006), and dotted is
the 15 nT year1 extrapolation of Barraclough (1974).
During the twentieth century, they were found to separate intervals 1996; Mandea et al., 2000). The signature of jerks can be seen
of linearly changing secular variation and have been unambigu- particularly clearly at European observatories, for example, in
ously identified as having occurred in 1901, 1913, 1925, 1969, Figure 27, which shows the evolution the secular variation of the
1978, 1991, and 1999 (Alexandrescu et al., 1995; Macmillan, eastward component of the geomagnetic field (Y)_ in Niemegk.
that are observed are not simultaneous; Alexandrescu et al.

103 (1996b) noted that, for example, in 1969, the signature of
the jerk tended to be observed later in the southern hemisphere
(see Figure 28).
The physical processes causing jerks, as well as geographic
variations in their detectability and delays, are not well under-
t1
stood. Bloxham et al. (2002) suggested that jerks might be the

102 surface manifestation of a superposition of torsional oscilla-
tions (a special class of axisymmetric, geostrophic, hydromag-
netic waves likely to be present in the Earth’s outer core – see
Core Dynamics volume) and that variations in their detectabil-
ity might be the result of variation in the field morphology at
100 101 the core surface. Alexandrescu et al. (1999) and Nagao et al.
1 (2003) have suggested that variations in mantle conductivity
Spherical harmonic degree
could explain the observed delays in jerk observations. Much
Figure 25 Estimated reorganization timescales t. The reorganization work remains to be carried out in understanding the physical
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
X 2 2 X m 2 m 2 mechanisms involved and in testing the various hypotheses.
(or correlation) time tðl Þ ¼ g m
+ h m
= g_ + h_
m l l m l l
Variations in the main geomagnetic field have their origin
(Hulot and LeMouël, 1994) is derived from the CHAOS-3 field model for in the Earth’s core. Most insight into the physical mechanisms
the year 2005, giving an instantaneous estimate of the time taken for causing the field evolution can therefore be obtained by exam-
power at spherical harmonic degree l of the field to be completely
ining the patterns of field evolution at the core surface. To
changed or renewed. Stars shows the CHAOS-3 estimates; solid line
determine the core field, we adopt the approximation of treat-
shows a prediction in the form of eqn [12] along with 90% statistical
dispersion bars, with t1 taken to be 425 years (see Lhuillier et al., 2011). ing the mantle as a perfect insulator. This approximation has
Data courtesy of F. Lhuillier (personal communication). been studied by Benton and Whaler (1983), who show that
when variations are considered whose periods are longer than
annual, the error introduced is small when the mantle has an
electrical conductivity structure as currently believed (i.e., from
102 to 10 Sm1). In the next section, the patterns of field
Models ufm1, gufm1 and COV– OBS 〈∂t Br〉 at r = a
evolution that result from such an approach are discussed in
70 detail.
1914
1925 1978
65 1997
<∂t Br>a (nT year-1)
1971
1950
5.05.4.2 Evolution of Radial Field at the Core Surface
60 1963
The evolution of the geomagnetic field at the core surface over
the past few centuries was first described in detail by Bloxham
55
and Gubbins (1985) and Bloxham et al. (1989) by considering
1938 1958
a series of single-epoch models. The picture they described has
50
1887 been borne out by the more recent time-dependent field
models ufm1 (Bloxham and Jackson, 1992) and gufm1
45 ( Jackson et al., 2000), so we shall reiterate their findings here
before discussing more recent developments. Contour plots of
1840 1880 1920 1960 2000
the historical evolution of the vertical field at the core surface
Time/years
are found in Figures 29–31.
Figure 26 Evolution of root-mean-square value of B_ r averaged over the The structure of the vertical field at the core surface is
Earth’s surface since 1840. RMS Br integrated over the Earth’s considerably more complicated than at the surface, because
surface from the ufm1 (solid), gufm1 (dashed), and COV-OBS (dotted) higher spherical harmonics are amplified more (by a factor
time-dependent field models. Although the models agree for the (a/c)(l+2) where a is the radius of the Earth, c is the core
twentieth century, for the nineteenth century, there are variability and radius, and l is the spherical harmonic degree) during the
disagreement between models.
downward continuation procedure. This is one reason why it
is preferable to downward continue regularized field models
rather than those that have been simply truncated. Downward
A 12-month running average filter has been applied to the continuing truncated field models also unfortunately
central differences of monthly mean data to produce this time introduce the possibility of unwanted Gibbs ringing effects
series, following the methodology of Mandea et al. (2000). The due to the sharp cutoff in spectral space (see, e.g., Gubbins,
jerk events are captured (at least in a smoothed manner) by the 1983; Shure et al., 1982; Whaler and Gubbins, 1981). The
internal field representation of global models such as CM4, gufm1 model results presented and discussed here have been
COV-OBS, and gufm1 – this can been seen, for example, in regularized so the power spectrum for the model has decayed
Figure 15 that compares model results to observatory annual substantially before the nominal cutoff at spherical harmonic
means. Jerks are not always observed at all locations and those degree l ¼ 14 is reached.
NGK: SV in Y component f
80
75
70
65
60
dY/dt (nT year -1) 55
50
45
40
35
30
25
20
15
10
5
0
1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Time of centre of sliding window (year)
Figure 27 First differences of 12-month averaged monthly means of Y_ at Niemegk. Central differences of monthly means of Y_ at the Niemegk
observatory processed using a 12-month moving average filter.
5.05.4.2.1 High latitude, approximately stationary flux decay of the axial dipole field observed since 1840. The signif-
lobes icance of the changes in the flux through these patches will be
Probably the most prominent feature in the maps of the field at discussed in Section 5.05.5.4. If taken at face value, the growth
the core surface are the high intensity flux lobes (by which we of the South Atlantic patch implies a failure of a particularly
mean the areas of flux maxima, of either sign) under Arctic attractive approximation for the core, the so-called frozen-flux
Canada, Siberia, and the eastern and western edges of Antarctica; hypothesis, which consequently means that it is very difficult
they can be seen particularly clearly in Figure 31. These lobes are to retrieve fluid motions at the core surface. It is important to
responsible for the predominantly axial dipole field structure recognize that the increase in quality, quantity, and distribu-
observed at the surface and have remained approximately station- tion of data throughout time leads to increased complexity in
ary (wobbling slightly about a mean position) over the past four the field models, and it is very difficult to disentangle this effect
centuries. Gubbins and Bloxham (1987) identified these high- from true diffusional effects; we refer the reader to the discus-
latitude flux lobes as the signature of columnar convection rolls in sion in Section 5.05.5.4.
the core (Busse, 1975), which are thought to be a major ingredient
in the geodynamo process (Kono and Roberts, 2002). They pro- 5.05.4.2.3 Low-latitude, westward drifting field features
posed that flow convergence associated with downwelling in the Bloxham and Gubbins (1985) noted the presence of a number
convection rolls is responsible for producing the observed field of rapidly westward moving field concentrations at low- and
concentrations. Bloxham and Gubbins (1987) ascribed the rela- midlatitudes, especially clear in the Atlantic hemisphere.
tive stationarity of these flux lobes to the influence of heat flow Bloxham et al. (1989) noted that beneath Europe and the
inhomogeneities at the CMB associated with the structure of Atlantic ocean during the twentieth century, there was a west-
mantle convection. More detailed studies using geodynamo sim- ward moving sequence of field highs and lows and referred to
ulations (Bloxham, 2002; Olson and Christensen, 2002) have this as a midlatitude polar wave. They suggested this could be a
confirmed the feasibility of this mechanism. wave with azimuthal wavenumber between m ¼ 5 and m ¼ 9.
Jackson (2003) examined very high-resolution images of the
5.05.4.2.2 Reversed flux patches field at the core surface in 1980 and 2000 constructed using
The presence of reversed flux features at the core surface is a high-quality satellite data and utilizing a maximum entropy
major difference to the field structure observed at the Earth’s regularization technique. He showed that the wavelike feature
surface. Most prominent of these are the patch that is found identified in the northern hemisphere by Bloxham et al.
close to the geographic north pole throughout most of the past (1989) has a counterpart at low latitude on the other side of
400 years and the large feature that extends from under the geomagnetic equator that had a considerably higher ampli-
Southern Africa across to under southern America that has tude than was evident in previous studies.
been formed by the coalescence of two earlier patches. Since these drifting features are moving essentially east to
Gubbins (1987) and Gubbins et al. (2006) have linked the west, their motion can be tracked using plots of field amplitude
growth and migration of the South Atlantic patch to the rapid as a function of time and longitude (TL plots). TL plots of the
˚
180
˚
150
˚
1 20
90˚
60˚
30˚
0˚
˚
330
90˚
˚
300
60˚
˚
270
30˚
˚
240
0˚
˚
−30
˚
210
˚
−60
˚
180
1 2 3
˚
−90
1969.88
Figure 28 Geographic distribution of occurrence time for 1969 jerk. Geographic distribution of the times of occurrence of the 1969 jerk measured by
Alexandrescu et al. (1996b). A linear combination of X and Y field components was analyzed using wavelet ridge functions and the jerk onset
time estimated. Blue bars represent negative delays relative to the mean occurrence time (1969.88) and correspond to earlier jerks, while red bars
represent positive delays relative to the mean (later jerks). The scale bar varies from 0 to 3 years. Green squares represent locations where
jerks were not detected.
radial component of the field at the core surface between 1590 (see, e.g., at latitudes 60 N between longitudes 120 and 90
and 1990 (from the gufm1 model) at latitudes 60 N, 40 N, and between +90 and +120, similarly at 60 S, near
and 20 N, at the equator, at 20 S and at 60 S are presented in longitudes 90 and +120). At lower latitudes, for example,
Figure 32. at 20 N after 1900, at the equator between longitudes
In TL plots, vertical lines of high field intensity represent 0 and +90 and at 20 S, there are some hints of diagonal lines of
stationary flux features such as the high-latitude flux features high field intensity that represent azimuthally moving field
-10.0 -8.75 -7.5 -6.25 -5.0 -3.75 -2.5 -1.25 0 1.25 2.5 3.75 5.0 6.25 7.5 8.75 10.0 Z / 105 nT
(a)
-10.0 -8.75 -7.5 -6.25 -5.0 -3.75 -2.5 -1.25 0 1.25 2.5 3.75 5.0 6.25 7.5 8.75 10.0 Z / 105 nT
(b)
Figure 29 Historical change in Bz at the core surface: AD 1590 and AD 1690. Vertical magnetic field Bz at the core surface in (a) AD 1590 and
(b) AD 1690 from the model gufm1 of Jackson et al. (2000). Plots are the Mollweide projection; units are nT.
features. Unfortunately, it is rather difficult to analyze these fea- Figure 33. Note that the first and last 40 years of the record have
tures because they are swamped by stationary features that are not been disregarded to eliminate filter warm-up effects, namely, the
of interest in this context. To get round this problem, Finlay and fact that the edges of the time series affect the filtered output.
Jackson (2003) high-pass-filtered the radial field from the gufm1 The filtering reveals clear westward moving, wavelike, sig-
model, removing the time-averaged axisymmetric field and all nals at low latitudes (between 20 N and S and particularly
field components varying on timescales longer than the 400 striking at the equator). No clear wavelike motions were
years to obtain a field that they denoted by Ber . The result of this found at higher latitudes indicating that such patterns of secu-
processing is shown in TL plots at the same latitudes as before in lar variation are confined to low latitudes on timescales shorter
-10.0 -8.75 -7.5 -6.25 -5.0 -3.75 -2.5 -1.25 0 1.25 2.5 3.75 5.0 6.25 7.5 8.75 10.0 Z / 105 nT
(a)
-10.0 -8.75 -7.5 -6.25 -5.0 -3.75 -2.5 -1.25 0 1.25 2.5 3.75 5.0 6.25 7.5 8.75 10.0 Z / 105 nT
(b)
Figure 30 Historical change in Bz at the core surface: AD 1790 and AD 1890. Vertical magnetic field Bz at the core in (a) AD 1790 and (b) AD 1890
from the model gufm1 of Jackson et al. (2000). Plots are the Mollweide projection; units are nT.
than 400 years. By measuring the power traveling at different hemisphere, there are weak signals probably associated with
angles in the TL plots at all latitudes, it is possible to construct the wobbles of the high-latitude flux lobes – these motions are
latitude–azimuthal speed (LAS) plots that summarize the rela- both eastward and westward and appear most clearly when
tive strength, location, and rate of azimuthal secular variation long timescale field variations are retained in Figure 34(a) and
processes. Such plots constructed when the radial field from 34(b). Next, there is the strong equatorially confined signal
gufm1 is high pass-filtered with thresholds of 2500, 600, 400, with speed of approximately 17 km year1 westward as
and 200 years are shown in Figure 34. described by Finlay and Jackson (2003). This is the dominant
The LAS power plots suggest that three distinct types of signal when only field variations with timescales shorter than
azimuthal secular variation have been operating during the 400 years are considered and appears in TL plots to have the
past four centuries. At mid- to high latitudes in the northern form of a wavelike disturbance. Finally, on all timescales, there
-10.0 -8.75 -7.5 -6.25 -5.0 -3.75 -2.5 -1.25 0 1.25 2.5 3.75 5.0 6.25 7.5 8.75 10.0 Z / 105 nT
Figure 31 Historical change in Bz at the core surface: AD 1990. Vertical magnetic field Bz at the core surface in AD 1990 from the model gufm1 of
Jackson et al. (2000). Plots are the Mollweide projection; units are nT.
is a strong westward signal in the southern hemisphere, which rB ¼ 0 Absence of free magnetic monopoles [14]
is particularly clear when the filter threshold is much longer
r E ¼ @t B Faraday’s law of magnetic induction [15]
than the record length. It seems to be associated with the
westward motion of reversed flux features and is particularly 1
ðr BÞ ¼ J Ampere’s law of magnetostatics [16]
strong in the twentieth century. m0
No in-depth study of meridional motions of field features
where m0 is the magnetic permeability of free space that is
at the core surface has yet been carried out. Such a study would
applicable to nonferromagnetic fluids. It should be noticed
be of interest especially considering the possible links between
that these equations are somewhat simpler than the usual,
meridional motions and proposed reversal mechanisms
most general form of Maxwell’s equations described in, for
(Gubbins, 1987; Wicht and Olson, 2004).
example, Jackson (1999) or Backus et al. (1996). The fact that
the liquid metal flows we are interested in have speeds juj c
(the speed of light) has enabled the well-known displacement
5.05.5 Interpretation in Terms of Core Processes
current term in Ampere’s law to be neglected and allowed
(decoupled) Gauss’ law of electrostatics to be dispensed with.
The observed evolution of the internally generated part of the
This powerful simplification is known as the MHD approxi-
Earth’s magnetic field is a consequence of the motions in the
mation. In this scenario, Ohm’s law for the electrically con-
liquid metal outer core. In order to understand and model
ducting and moving fluid takes the form
the mechanisms underlying these changes, we must employ
the mathematical framework of magnetohydrodynamics J ¼ sðE + u BÞ [17]
(MHDs) – the marriage of Maxwell’s laws of electromagnetism where s is the electrical conductivity of the fluid. The mathe-
and the principles of hydrodynamics or fluid mechanics. In this matical formalism can be further compacted by realizing that
section, equations describing the evolution of the core magnetic eqns [14]–[17] can be combined to yield a single prognostic
field and the generation of core fluid motions will be derived, equation governing the evolution of magnetic fields. Substitut-
and useful approximations will be discussed; we will stop short ing from eqn [16] into eqn [17] gives
of describing attempts to invert field observations for core fluid
motions at the CMB, which is the territory of Volume 8. r B ¼ m0 sðE + u BÞ [18]
Taking the curl of this,

5.05.5.1 Maxwell’s Equations and Moving Frames
r ðr BÞ ¼ m0 sðr E + r ðu BÞÞ [19]
Maxwell’s equations for an electrically conducting fluid mov-
ing with a velocity u in the presence of a magnetic field B, an Substituting from eqn [15] and using the vector identity
electric field E, and an electric current density J are that r (r B) ¼ r(r B) r2B, this becomes
Br /105 nT Br/105 nT
1990 1990
10 10
1950 1950
8 8
1900 6 1900 6
4 4
1850 1850
2 2
Time (years)
Time (years)
1800 1800
0 0
1750 −2 1750 −2
−4 −4
1700 1700
−6 −6
1650 1650
−8 −8
−10 −10
1590 1590

−150 −120 −90 −60 −30 0 30 60 90 120 150 −150 −120 −90 −60 −30 0 30 60 90 120 150
(a) Longitude (degrees east) (b) Longitude (degrees east)
Br /105 nT
1990
10
1950
8
1900 6
4
1850
2
Time (years)
1800
0
1750 −2
−4
1700
−6
1650
−8
−10
1590
−150 −120 −90 −60 −30 0 30 60 90 120 150
(c) Longitude (degrees east)
Figure 32 Time-longitude (TL) plots of Br from gufm1. TL plots of the unfiltered radial magnetic field Br from the field model gufm1 at latitudes 60 N in (a), at 40 N in (b), at 20 N in (c),
169
(Continued)
170
B̃r /104 nT B̃r /104 nT
1950 1950
10 10

8 8
1900 1900
6 6
1850 4 1850 4
2 2
Time (years)
Time (years)
1800 1800
0 0
−2 −2
1750 1750
−4 −4
1700 −6 1700 −6
−8 −8
1650 1650
−10 −10
−150 −120 −90 −60 −30 0 30 60 90 120 150 −150 −120 −90 −60 −30 0 30 60 90 120 150
(d) Longitude (degrees east) (e) Longitude (degrees east)
B̃r /104 nT
1950
10
8
1900
6
1850 4
2
Time (years)
1800
0
−2
1750
−4
1700 −6
−8
1650
−10
−150 −120 −90 −60 −30 0 30 60 90 120 150
(f) Longitude (degrees east)
Figure 32 (Continued) at the equator in (d), at the 20 S in (e), and at the 60 S in (f ).

B̃r /104 nT B̃r /104 nT
1950 1950
10 10
8 8
1900 1900
6 6
1850 4 1850 4
Time (years)
2 2
Time (years)
1800 1800
0 0
−2 −2
1750 1750
−4 −4
1700 −6 1700 −6
−8 −8
1650 1650
−10 −10

−150 −120 −90 −60 −30 0 30 60 90 120 150 −150 −120 −90 −60 −30 0 30 60 90 120 150
(a) Longitude (degrees east) (b) Longitude (degrees east)
B̃r /104 nT
1950
10
8
1900
6
1850 4
2
Time (years)
1800
0
−2
1750
−4
1700 −6
−8
1650
−10
−150 −120 −90 −60 −30 0 30 60 90 120 150
(c) Longitude (degrees east)
171
Figure 33 TL plots of Ber from gufm1. TL plots of the processed radial magnetic field Ber with time-averaged axisymmetric component subtracted and high-pass-filtered with cutoff period 400 years, from the
field model gufm1 at latitudes 60 N in (a), at 40 N in (b), at 20 N in (c),
172
Br /105 nT Br /105 nT
1990 1990
10 10
1950 1950

8 8
1900 6 1900 6
4 4
1850 1850
2
Time (years)
2
Time (years)
1800 1800
0 0
1750 −2 1750 −2
−4 −4
1700 1700
−6 −6
1650 1650
−8 −8
−10 −10
1590 1590
−150 −120 −90 −60 −30 0 30 60 90 120 150 −150 −120 −90 −60 −30 0 30 60 90 120 150
(d) Longitude (degrees east) (e) Longitude (degrees east)
Br /105 nT
1990
10
1950
8
1900 6
4
1850
Time (years)
2
1800
0
1750 −2
−4
1700
−6
1650
−8
−10
1590
−150 −120 −90 −60 −30 0 30 60 90 120 150
(f) Longitude (degrees east)
Figure 33 (Continued) at the equator in (d), at 20 S in (e), and at 60 S in (f ).

~ ~
(Br)R2dz/1016(nT)2 (Br)R2dz/1015(nT)2
-60 3.5 -60

12
3
-30 -30 10
2.5
Latitude (°)
Latitude (°)
8
0 2 0
6
1.5
30 30
1 4
0.5 2
60 60
-60 -40 -20 0 20 40 60 -60 -40 -20 0 20 40 60

(a) Azimuthal speed (km(year)-1) (b) Eastward zonal phase speed (km(year)-1)

~
~ (Br)R2dz/1014(nT)2
(Br)R2dz/1015(nT)2
7 -60 6
-60
6
5
-30 -30
5
4
Latitude (°)
Latitude (°)
4
0 0
3
3
30 2
2 30
1 1
60
60
-60 -40 -20 0 20 40 60
-60 -40 -20 0 20 40 60
(c) Eastward zonal phase speed (km(year)-1)
(d) Azimuthal speed (km(year-1))
Figure 34 Latitude–azimuthal speed (LAS) power plots of Ber from gufm1. Summing the power traveling at different angles in TL plots using a Radon transform method, LAS power plots are constructed
173
for the high-pass-filtered Br from gufm1. (a) shows the result when the high-pass-filter threshold is 2500 years, (b) when it is 600 years, (c) when it is 400 years (the case shown for the TL plots
in Figure 33), and (d) when it is 200 years.
@B 1 @T 1 @T
rðrBÞ r2 B ¼ m0 s + r ðu B Þ [20] BT ¼ 0, , [24]
@t r sin y @f r @y
Using eqn [14], rearranging and defining the magnetic and

diffusivity ¼ m1s yields
0 L2 P 1 @ 2 P 1 @2P
BP ¼ , , [25]
@t B ¼ r ðu BÞ + r2 B magnetic induction equation [21] r 2 r @r@y r sin y @r@f
The second term represents the changes in magnetic fields where L is the angular momentum operator of quantum
due to dissipative Joule heating effects that are the consequence mechanics, defined by
of the flow of electric currents in a material with finite

resistivity. 1 @ @ 1 @2
L2 ¼ sin y + [26]
sin y @y @y sin 2 y @f2
We can use the same representation for the fluid flow u, if

5.05.5.2 The Induction Equation in a Spherical Earth we assume incompressibility. While this is not strictly true (the
In this section, we discuss the interpretation of the secular density increase with pressure across the outer core is approx-
variation signal in terms of core processes. We believe that imately 20%), it is a fair approximation.
convection in the core is the primary agent that generates the We are now in a position to revisit some earlier comments
magnetic field and the resulting secular variation; approxima- about the finite conductivity of the mantle. We revise eqn [21]
tions can be made to simplify the underlying equations and we (having first set u to zero) to account for the fact that in the
will discuss their applicability. mantle may be variable; this gives the diffusion equation for a
We note here that we take no account of the electrical heterogeneous :
conductivity of the mantle: we approximate the mantle as @B
an insulator. If the mantle has a spherically symmetrical ¼ r ðr BÞ [27]
@t
distribution of electrical conductivity that is of moderate
When we analyze its toroidal and poloidal ingredients, we
amplitude, this may not be too poor an approximation. At
find that in general there can be poloidal to toroidal conver-
the present time, very little is known about the three-
sion and vice versa; however, in the case that there is spherical
dimensional heterogeneity in conductivity, and in the deep
symmetry ( ¼ (r)), a remarkable simplification occurs: the
mantle, even the spherically symmetrical part is poorly
toroidal and poloidal ingredients of B obey
known. Thus, our approximation may be in considerable
error.
@T 1 d @ @P
We begin with the induction eqn [21] governing changes ¼ r2 T + ðrT Þ; ¼ r2 P [28]
@t r dr @r @t
with time in the magnetic field due to the effects of magne-
toadvection and magnetic diffusion. We will assume for the There is thus complete decoupling between the poloidal
moment that the flow u is given and examine its effects on and toroidal ingredients of the field, even when the mantle
the field, the so-called kinematic problem; the question of the conductivity is radially varying. Inasmuch as the spherically
dynamics will be examined in the succeeding text. It is useful to symmetrical assumption is valid, this simplifies the task of
introduce the toroidal–poloidal, or Mie (Backus, 1986; Backus understanding the secular variation.
et al., 1996), representation: At the Earth’s surface, we have electrically insulating bound-
ary conditions and thus the toroidal magnetic field must van-
B ¼ BT + BP ¼ r ðT^
rÞ + r r ðP^
rÞ [22] ish; this arises because Ampere’s law relates fields B to currents
where T(r, y, f) and P(r, y, f) are the toroidal and poloidal J; thus,
scalars, respectively, defining the toroidal BT and poloidal BP
ingredients of B. This representation is valid for any solenoidal m0 J ¼ r B ¼ r BT + r BP
vector field satisfying ¼ r r ðT^
rÞ + r r r ðP^
rÞ [29]
rB ¼ 0 [23] rÞ + r ½ðr P Þ^
¼ r r ðT^ r 2
The sphericity of the core suggests that we continue to Comparing this expression for the current with our defini-
work in spherical polar coordinates, though note that it is tion of the toroidal–poloidal decomposition, it is clear that
the case that the core is ellipsoidal with equatorial radius poloidal field results from toroidal currents and toroidal field
greater by approximately one part in 400 than the polar from poloidal currents.
radius. In terms of approximating the surface for the purposes If J ¼ 0, as in an insulator, then BT ¼ 0, while BP does not
of plotting fields or fluid motion, the ignorance of the oblate necessarily vanish, although P must satisfy r2P ¼ 0 (in which
spheroidal nature of the core can be seen to introduce negli- case, we call it a potential field). In principle, we can find the
gible errors; note, however, that the ellipticity might be toroidal field within the mantle from measurements of the
important in a dynamical context, because of the coupling electric field at the sea bottom (e.g., Lanzerotti et al., 1993;
that it can generate between mantle and core. In spherical Runcorn, 1955; Shimizu and Utada, 2004; Shimizu et al.,
polar coordinates (r, y, f), the toroidal and poloidal ingredi- 1998), but the electric field from the toroidal field from the
ents may be written in terms of the toroidal and poloidal core is likely to be small compared with the field from other
scalars in the form sources, especially that from ocean currents.
We are therefore only able to monitor the poloidal part of represent the radial component of magnetic field at the top of
the magnetic field at the Earth’s surface. Turning now to the the free stream. The same cannot be said to be true for the
CMB, it is extremely unlikely that the toroidal magnetic field horizontal components, but we omit a discussion on the
vanishes there on account of the finite (and possibly large) possible jumps in Bh mainly because the induction equation
conductivity there. Several factors come to our rescue to ame- for Bh involves the toroidal field, which we are unlikely to
liorate what otherwise would seem like a hopeless situation. ever know.
We temporarily assume that the CMB is a free-slip boundary,
so that u^r ¼ 0, but u ¼ uh 6¼ 0. Then following Bullard and
Gellman (1954), we can show that 5.05.5.3 The Navier–Stokes Equation
We adopt the hypothesis that on macroscopic length scales,
½r ðuh BÞ P ¼ ½r ðuh BP Þ P [30]
core fluid can be well approximated as a continuum (see, e.g.,
from which we obtain the poloidal induction equation at Batchelor, 1967), suppose that it is to first approximation
the CMB incompressible, obeys Newtonian laws of viscosity, and is
uniformly rotating. Then, in a frame of reference rotating
@BP with the fluid, the conservation of momentum is encapsulated
¼ ½r ðuh BP Þ P + r2 BP [31]
@t in the Navier–Stokes equation, which under the Boussinesq
In the preceding text, []P stands for the poloidal part of the approximation (e.g., Gubbins and Roberts, 1987) reads
equation. We discover, somewhat counterintuitively, that the
@u
poloidal secular variation depends only on the poloidal mag- r0 + uru + 2O u ¼ rp + r0 g + J B + r0 vr2 u [34]
@t
netic field! This result hangs entirely on the fact that radial
motions are zero at the CMB. The result would not be true where r0 and r0 are the hydrostatic density and departure from
elsewhere in the core, where radial motions are crucial for the hydrostatic density, respectively, O is the Earth’s rotation vec-
production of poloidal field from toroidal magnetic field, in tor, p is the nonhydrostatic part of the pressure, g is the accel-
order for the dynamo to operate. Note also that the toroidal eration due to gravity, n is the kinematic viscosity, and J is the
part of the induction equation does not separate so easily and current density. The Boussinesq approximation is a simplifica-
that the rate of change of toroidal field depends on both the tion frequently adopted for the core and ignores variations in
toroidal field and the poloidal field. Horizontal flow in this density except those that are responsible for thermal buoyancy
case shears both poloidal and toroidal magnetic fields in order through the term r0 g; its applicability to the core is under
to create toroidal secular variation. current scrutiny in the field of numerical simulation of core
It is a fortuitous fact that when one analyzes the different dynamics, since the compressibility of the core does cause a
components of the poloidal induction equation, the radial part change of approximately 20% in the core density between the
has a particularly simple form and gives an equation that will inner and outer core boundaries. It suffices as an approxima-
be central to much of our discussion. The radial induction tion for our purposes, as we shall predominantly be limiting
equation reads our discussion to the surface of the core.
Considerable simplification can be made if one analyzes
@t Br + uh rh Br + Br rh uh ¼ r2 ðrBr Þ [32] the likely sizes of the terms in the equation, concentrating on
r
the flow in the main body of the core (outside the boundary
demonstrating that only radial fields, and their derivatives, layers). Firstly, the Rossby number
need to be known for radial secular variation to be calculable
when a particular flow is prescribed. U
Ro ¼ ’ 4 106 [35]
It is incumbent on us to realize the shortcomings in the OL
aforementioned analysis. We applied the condition u^r ¼ 0 compares the nonlinear advective term on the left-hand side of
rather than the true nonslip condition u ¼ 0 at the CMB. In eqn [34] with the Coriolis term. We take L 3 106 m as a
reality, there is a boundary layer over which the flow adjusts to characteristic length scale for the core. Then, the estimate
the nonslip boundary condition, and we really apply earlier is based on values for U (roughly half a millimeter per
the induction equation at the top of the free stream, second) gleaned from the analysis of the secular variation (e.g.,
the bottom of the boundary layer. We need to know the differ- Volume 8 and Section 5.05.4.2.3), and hence (as in much of
ence in the values of B across this boundary layer, denoted [B]. our analysis), there is a slight sense of circularity. Similarly, the
Various analyses of the boundary layer have been carried Ekman number (the ratio of viscous forces to the Coriolis
out, and no consensus has been reached (e.g., Backus, 1968; force) is given by
Hide and Stewartson, 1973; Jault and LeMouël, 1991). The
v
important issue for our purposes is that the radial derivatives E¼ [36]
in all three components of B are expected to be much bigger OL2
than the horizontal derivatives. When one uses this fact along where O is the rotation rate. If we take L as before and
with the divergence-free constraint on the field, we find n 106 ms2, we find the classic value of E 1015, indicat-
ing that viscous effects are negligible in the main body of
@Br @Br
+ rh Bh ¼0 [33] the core if a laminar value for n is adopted. The inertial term
@r @r
is somewhat more difficult – there is a mode of oscillation
This leads to the conclusion that [Br] ¼ 0 so maps of the in the core that can occur on decade timescales, the so-called
radial component of field immediately above the CMB also torsional oscillation (see Volume 8) that may not be
negligible. It is easily excited and it is inappropriate to com- 5.05.5.4 The Frozen-Flux Approximation
pare it to the Coriolis force because it is unaffected by it. We
5.05.5.4.1 The neglect of magnetic diffusion and its physical
will neglect the inertial term on the grounds that it is only
consequences
significant when the period approaches the diurnal period,
The interpretation of secular variation using the induction eqn
except in the force balance when averaged over cylinders
[21] is often simplified by neglecting the contribution of mag-
coaxial with the rotation axis. This leads us to a very useful
netic diffusion. In the limit of a perfectly electrically conducting
approximation in core studies, the so-called magnetostrophic
fluid (zero magnetic diffusivity), the induction equation becomes
approximation
ð@t + urÞB ¼ ðBrÞu [44]
r0 ð2O uÞ ¼ rp + r0 g + J B [37]
The left-hand side of this equation is the advective deriva-
The difficult term in this equation is the last term on the tive describing how the magnetic field changes as one moves
right-hand side, the Lorentz force L. We write it in the form along with the fluid, while the right-hand side tells us that such
changes occur through the stretching of the magnetic field by
1 fluid motions.
L ¼JB ¼ ðr BÞ B
m0 Further intuition follows if we think about how a velocity
[38]
¼
1 1
rB2 + ðBrÞB field u would advect a material line element dl. We imagine the
m0 2 line element being drawn in the fluid at some instant and
subsequently moved along and stretched by the fluid motions.
An approximation called the tangential geostrophy approx- The total rate of change of dl is then u(r + dl) u(r), where r
imation, proposed independently by Hills (1979) and and r + dl are position vectors at the two ends of dl. The equa-
LeMouël (1984), would neglect the horizontal components tion describing the evolution of dl therefore has the form
of this term when compared to all others; to aid our
development, we write the horizontal and radial components ð@t + urÞdl ¼ uðr + dlÞ uðr Þ ¼ ðdlrÞu [45]
of L as Lh and Lr, respectively. For tangential geostrophy, we Inspection of eqns [44] and [45] reveals they have precisely the
require that same form. This simple example demonstrates that because mag-
jLh j jðBrÞBh j netic fields evolve in an identical manner to material line elements
M¼ ¼ 1 [39] in a fluid, a field line found on a particular fluid element at some
2r0 OU 2m0 r0 OU
initial instant must continue to lie on that element at all subse-
There are two contributions to (B r)Bh: quent times. The magnetic field effectively appears to be frozen
@Bh into the fluid as it moves. This result is known as Alfvén’s theorem
Br and ðBh rh ÞBh [40] (part I) after Hannes Alfvén who first derived it; we shall discuss
@r
the second part of the theorem in a moment. Neglect of magnetic
We need to estimate jBhj and j@Bh/@rj. LeMouël (1984) diffusion in the induction equation and its consequences are most
argued that if the toroidal field is small at the CMB (if the commonly referred to as the frozen-flux approximation and we
mantle is a perfect insulator it must vanish) and its radial shall use the latter terminology.
gradient is small, then these terms are of order B2P/L, where BP Another important property that results from assuming that
is the size of the poloidal field at the CMB (’ 5 104 T), a fluid is a perfect electrical conductor can be demonstrated by
giving M ’ 103. returning to Faraday’s law of magnetic induction (eqn [15]).
If we adopt this approximation, we have Earlier, this was stated in its differential form. The integral form
2r0 ðO uÞ ¼ rp + r0 g + Lr [41] when applied to material curves of an electrically conducting
fluid in motion (see, e.g., Davidson, 2001) takes the form
Curling this equation, we obtain þ ð
d
E0 dl ¼ BdS [46]
2r0 ðOrÞu ¼ g rr0 + r Lr [42] C dt S
which, in the case that Lr ¼ 0 is the thermal wind equation, of where E0 ¼ E + u B is the total electric field in a reference frame
great importance in meteorology. The radial component gives moving along with dl at velocity u, C is a closed material curve
the so-called geostrophic constraint composed of line elements dl, and S is any surface that spans C.
Now, from Ohm’s law, J ¼ sE0 so
rh ðuh cos yÞ ¼ 0 [43] þ ð
1 d
which remains true regardless of whether Lr 6¼ 0. Given these Jdl ¼ BdS [47]
s C dt S
assumptions, we therefore have a strong constraint on the
types of allowed fluid motions at the CMB. The interested but under the frozen-flux approximation, we assume s ! 1;
reader can see in Volume 8 that output from self-consistent therefore,
geodynamo simulations tends to suggest that the tangential ð
d
geostrophy approximation is reasonably well obeyed. The sig- BdS ¼ 0 [48]
dt S
nificance of the constraint is that it vastly reduces the types of
allowable fluid motions when the inverse problem for u is In a perfect electrical conductor, the integrated magnetic
solved (for further details, consult chapter on core flow in field (or magnetic flux) through any material surface is thus
Chapter 8.04). always preserved. This is known as Alfvén’s theorem (part II).
5.05.5.4.2 Application of the frozen-flux hypothesis to the through a null-flux curve C (an example of a material curve) is
generation of secular variation at the core surface preserved. The simple derivation is as follows. We know that
Roberts and Scott (1965) were the first to suggest that the ð ð ð
d
frozen-flux approximation could be applied to the problem of Br dS ¼ @t Br dS + Br vn dl [54]
dt S S C
modeling secular variation. Many authors (see, e.g., Backus
et al., 1996) refer to this as the frozen-flux hypothesis. Roberts where dl is a line element along the null-flux curve C and vn is
and Scott argued that there are two distinct timescales associated the normal component of its velocity. Using eqn [52], the first
with the induction equation. Considering a length scale LB over term on the right-hand side is zero and, since C is a null-flux
which the magnetic field changes, a characteristic flow speed U, curve, the second term is also zero, giving
and the magnetic diffusivity , these timescales are defined as ð
d
Br dS ¼ 0 [55]
LB dt S
tadv ¼ advection timescale [49]
U
From eqn [55], it further follows that the sum of the
and unsigned flux over all null-flux curves must also be conserved:
ð
L2B d
tdif ¼ magnetic diffusion timescale [50] jBr jdS ¼ 0 [56]
dt S0
The ratio of these timescales Rm ¼ tdif =tadv ¼ ULB = is with the integration now over the entire core surface S0 . It
known as the magnetic Reynolds number and gives a crude should be noted that eqn [56] is a weaker constraint than
measure of the relative strength of advection to magnetic dif- eqn [55] on the signed flux through individual flux patches
fusion. Taking estimates of LB ¼ 106 m (the approximate scale because contributions from small patches will be swamped by
of the outer core container and of large-scale features in the those from the larger northern and southern hemisphere
Earth’s magnetic field), U ¼ 5 104 ms1 (speed of observed patches. The unsigned flux condition will therefore only be
westward field motions, thought to be caused by core flow), violated if there are large amounts of magnetic diffusion occur-
and ¼ 2m2s1 (from estimates of liquid iron electrical con- ring on a global scale; it could still be approximately obeyed
ductivity at core pressures and temperatures (Braginsky and even if magnetic diffusion was occurring locally. On the other
Roberts, 1995)) gives estimates of tadv ¼ 65 years and hand, it is less likely to be adversely affected by errors in the
tdif ¼ 1.6 104 years in the core (Roberts and Glatzmaier, field models so is a robust test of the global applicability of the
2000). On this basis, Roberts and Scott suggested that making frozen-flux approximation.
the frozen-flux assumption was a reasonable (though imper- Backus’ results additionally require that the topology of the
fect) approximation when modeling the core motions, causing field must be invariant so that null-flux curves cannot split or
large-scale secular variation. This theoretical argument has coalesce. In practice, this is a rather difficult condition to satisfy
been the subject of much comment and debate over the past as it requires only a small amount of diffusion in order to be
40 years; we will return later to the question of its validity. First, violated. It therefore seems unlikely that this condition would
we will give details on its formal consequences and review be satisfied by the magnetic field at the core surface, where
attempts to determine whether these are compatible with frozen flux is at best a useful approximation, so we shall not
observed secular variation. discuss this condition further.
5.05.5.4.3 Consequences of frozen-flux approximation 5.05.5.4.4 Attempts to test the frozen-flux approximation
at the core surface using geomagnetic observations
Backus (1968) described the conditions for the main field The conditions described in the previous section (which will be
morphology and secular variation to be consistent with a referred to collectively as the Backus conditions) have enabled
frozen-flux theory of core motions. Neglecting magnetic diffu- workers to test the validity of the frozen-flux approximation
sion, he showed that the radial part of the induction equation using models of the main field and secular variation constructed
reduces to (cf. eqn [32]) from geomagnetic observations. These tests have fallen into two
@t Br + rh ðuh Br Þ ¼ 0 [51] main categories: (i) attempts to estimate whether the Backus
conditions are violated in field models constructed without
He then deduced that this implies a set of conditions on any constraints on field evolution and (ii) attempts to build
null-flux points and curves (where Br ¼ 0). field models that not only are constrained to obey the Backus
The most important of these are conditions but also satisfactorily fit the observations.
ð The first category involves investigating whether observa-
@t Br dS ¼ 0 where S is a surface bounded by a null flux curve C tions are sufficiently accurate to observe diffusion. Booker
S
[52]
(1969) was the first to investigate this issue. He attempted to
estimate the integral of @ tBr through the null-flux curves repre-
and sented by the magnetic equator and found that the dipole part
of the time derivative was nonzero, possibly violating eqn [52].
@t Br ¼ 0 where two null flux curves C1 and C2 intersect [53]
However, the field models he used lacked the necessary reso-
From the first condition follows a condition closely related lution at high spherical harmonic degrees to definitively calcu-
to eqn [48] that states that the integrated radial magnetic field late the full secular variation integral and he was thus unable to
definitely identify the presence of diffusion. Gubbins (1983) approach to enforce zero radial secular variation through
repeated the calculation and arrived at a similar conclusion, null-flux curves (eqn [52]) and found that it was possible to
again being limited by the accuracy of observations. do this over the interval 1959–74. Bloxham and Gubbins
Hide and Malin (1981) and later Voorhies and Benton (1986) and Bloxham et al. (1989) produced field models
(1982) used the criteria for invariance of the unsigned flux at that satisfied the condition that the flux of Br through null-
the core surface (eqn [56]) as a method for calculating the flux curves be conserved (eqn [55]). They implemented this
outer core radius and found values that agreed well with the by imposing an additional penalty during the inversion pro-
seismologically determined value. These results indicated that cess; models with flux integrals differing from a predefined
the unsigned flux is not changing rapidly at the core surface reference model were heavily penalized. They reported that it
and has evidence in support in the application of the frozen- was possible to find models that satisfied these flux
flux approximation on large length scales. Benton and constraints and that their imposition often improved field
Voorhies (1987) later extended these analyses to show that models where the data quality was poor. However, they also
there had been little change in the unsigned flux over the reported that the constrained models had a slightly higher
interval 1945–85. Bloxham et al. (1989) considered the evolu- misfit than the unconstrained models, using this to support
tion of the unsigned flux over a much longer interval from their contention that frozen flux was in fact violated. This
1715 to 1980 and found it remained approximately constant argument is subject to similar caveats regarding errors in
over that interval. They concluded that this global requirement field models as the argument concerning changes in the flux
for frozen flux was satisfied by their field models but pointed integrals in unconstrained models. Benton et al. (1987) con-
out that changes in the flux of smaller null-flux patches would structed field models with constrained flux through null-flux
not be seen by this method. Similar results for the unsigned curves covering the interval 1977.5 to 1982.5 and were able to
flux were recently obtained by Holme and Olsen (2006) using demonstrate that the secular variation predicted by these
a high-quality, satellite-derived field model (CO2003). models for times outside the span of input data was an
The question of possible changes in the flux through indi- improvement on the predictions of unconstrained models.
vidual null-flux curves was addressed in a series of papers by They took this to be evidence in favor of the frozen-flux
Bloxham and Gubbins in the mid- to late 1980s including hypothesis.
Bloxham and Gubbins (1985), Gubbins and Bloxham (1985), Most recently, Constable et al. (1993), O’Brien et al.
Bloxham and Gubbins (1986), and Bloxham (1986a) and in the (1997), and Jackson et al. (2007) have attacked the same
landmark study of Bloxham et al. (1989) where the definitive problem but used a different parameterization of the field
results were reported. In the latter paper, a sequence of single- based on a spherical triangle tessellation at the core surface.
epoch, regularized field models spanning 1715–1980 were stud- Constable et al. (1993) constructed field models satisfying the
ied and changes of the flux though null-flux curves at the core flux constraints of eqn [55] for the epochs 1945.5 and 1980.0;
surface were calculated. Major flux changes through some curves O’Brien et al. (1997) did the same for the epochs 1915 and
were found; in particular, a patch that moved from the Indian 1980 while Jackson et al. (2007) managed to do so for the
Ocean to Southern Africa was found to increase its flux dramat- epochs 1882, 1915, 1945, 1980, and 2000. In all cases, it was
ically. Bloxham and Gubbins regarded this as conclusive evi- found that reasonable misfit levels to the observations could be
dence for the violation of frozen flux; however, others have achieved, and it was noted that in order to reject the frozen-flux
expressed doubts over the rigor of their arguments. In order for hypothesis, it would be necessary to demonstrate that no such
frozen flux to have been demonstrably violated, the changes in models could be found.
the flux must be larger than possible changes in the flux due to We remarked here that the published time-dependent field
errors in the field models. Backus (1988) has argued that the models of Bloxham and Jackson (1992) and Jackson et al.
error estimates of Gubbins and Bloxham (1985) and Bloxham (2000) both showed an obvious growth in the intensity of
and Gubbins (1986) are rather optimistic. He suggests that their reversed flux patches in the southern hemisphere in the
models do not fully solve the uniqueness problem because the twentieth century. Gubbins (1987) has pointed out that a
damping (regularization) parameter is arbitrarily chosen; their simple and physically appealing explanation of this phenom-
error estimates depend crucially on the value of this parameter enon is the expulsion of toroidal flux by upwelling core fluid
and are likely to actually be much larger than those quoted. (Allan and Bullard, 1966; Bloxham, 1986a; Drew, 1993). It is
Furthermore, O’Brien (1996) has demonstrated that even the therefore unfortunate that the present field observations seem
number of null-flux patches in field models based on excellent incapable of constraining field models sufficiently to discrim-
satellite data is difficult to definitively determine. It therefore inate between this mechanism and a mechanism involving
seems that inferences based on the changes in flux through frozen-flux advection.
individual null-flux patches in the far past (when there were Although our discussion has focused on the radial magnetic
significant variations in the distribution and accuracy of obser- field and its secular variation, a limited amount of work has
vations) should be viewed with a healthy degree of skepticism. been carried out on whether the horizontal components of the
The subject has been revisited anew by Chulliat and Olsen magnetic field are consistent with the frozen-flux hypothesis.
(2010) and Asari et al. (2010). Both sets of authors find evi- Assuming continuity of the horizontal magnetic field compo-
dence for violation of the Backus conditions. nents across the magnetic field boundary layer close to the core
The compatibility of observations with the Backus condi- surface (note that this assumption is questionable – see, e.g.,
tions for frozen flux has recently become apparent thanks to Jault and LeMouël, 1991), a further set of consistency condi-
the construction of field models that are constrained to obey tions can be constructed (Backus, 1968; Gubbins and Roberts,
these conditions. Gubbins (1984) used a Lagrangian constraint 1987). An attempt to test these conditions has been made by
Barraclough et al. (1989); they find no evidence that the con- use the frozen-flux approximation to understand the rapid
straints are violated, but confess that the field models used azimuthal motion of flux patches but not to understand
were not yet accurate enough to allow a stringent test. rapid growth and decay of field concentrations. Such reasoning
was used by Dumberry and Bloxham (2006) in their study of
5.05.5.4.5 Theoretical issues concerning the frozen-flux global azimuthal core flows on millennial timescales, derived
hypothesis from archaeomagnetic field models.
The simple scaling arguments of Roberts and Scott regarding Love (1999) raised an objection to the frozen-flux hypoth-
the plausibility of using the frozen-flux assumption to model esis in the case of a nearly steady dynamo. Although clearly
secular variation have been the subject of some debate, espe- correct under the conditions considered, the applicability of
cially since the claims in the 1980s that the signature of mag- his examples to the Earth is questionable since, as noted earlier,
netic diffusion had been detected. the geodynamo does not appear to be steady on any known
Gubbins and Kelly (1996) have pointed out that, for the timescales. Recent feasibility tests using more dynamically
special case of steady flows, the frozen-flux hypothesis is inva- plausible dynamo models (Rau et al., 2000; Roberts and
lid because the balance in the induction equation must be Glatzmaier, 2000) have, on the other hand, demonstrated
between secular variation and diffusion, with frozen-flux that the frozen-flux approximation is a useful construct.
effects being negligible. Although undoubtedly true, this obser- These important tests will be described in more detail in the
vation appears to be of little relevance when trying to model succeeding text. The important point to take away from the
changes in the Earth’s magnetic field that are time-dependent arguments of Love (1999) is that the original scaling argument
across a wide variety of timescales (even the dynamo process of Roberts and Scott was overly simplistic – in real secular
appears to be fundamentally time-dependent as seems appar- variation, there is not just one length scale and timescale of
ent from considering the frequency of geomagnetic excursions interest: both magnetic and velocity fields will contain power
over the past 700 000 years (Gubbins, 1999)). A more over a range of length scales and timescales. To address
worrisome objection raised by Gubbins and Kelly is that the whether the frozen-flux hypothesis is a useful approximation
frozen-flux approximation is a singular limit of the induction really requires experiments studying the induction effects of
equation. It lowers the differential order of the system from Earth-like velocity fields acting on Earth-like magnetic fields.
second order to first order, raising the possibility that physi- Unfortunately, it is still beyond our ability to accurately simu-
cally relevant solutions have been filtered out. late all aspects of the dynamics of the Earth’s core (Glatzmaier,
Concerns over how frozen-flux approximation arises as a 2002), though present geodynamo models already capture
limit of a magnetic advection–diffusion process have been many important aspects of the Earth’s magnetic field and its
addressed in detail by Gubbins (1996). Gubbins has devel- evolution (Christensen et al., 1998; Kuang and Bloxham,
oped a novel formalism for determining core motions that 1997) and are certainly useful tools when considering the
takes magnetic diffusion into account and shows that the validity of the frozen-flux hypothesis.
frozen-flux approximation can be formally retrieved in the Roberts and Glatzmaier (2000) were the first to use a
limit when (/o)1/2 tends to zero, where o is the frequency numerical geodynamo model in an attempt to evaluate the
of the secular variation. This implies that we should only value of the frozen-flux hypothesis. They calculated the varia-
expect the frozen-flux approximation to work well when field tions in the unsigned flux from a high-resolution simulation
variations are rapid and the concomitant fluid motions highly and found that it varied by c.3% over timescales estimated to
time-dependent. Gubbins used his formalism to derive an be equivalent to 150 years. This suggests that for plausible
estimate of the toroidal field gradient close to the core surface magnetic and velocity fields, the frozen-flux hypothesis is a
necessary to explain the amount of flux change through the good approximation on the global scale. Unfortunately, no
South Atlantic reversed flux patch between 1905.5 and 1965.5 attempt was made to investigate how much individual flux
as estimated by Bloxham and Gubbins (1985). He arrived at a integrals varied over the same interval. The authors further
plausible scenario by considering a 1.2 mT toroidal field at a remark that for frozen-flux approximation to be useful, it is
depth of 60 km in the core (diminishing at a rate of 20 nT m1 not required to be strictly true; rather, it must only involve
toward the surface) with a horizontal length scale of 106 m. errors smaller than those associated with our incomplete
When this field was acted on by an upwelling flow with a radial knowledge of the field at the core surface. They point out that
rate of change of 0.02 year1 in the presence of a magnetic the errors from truncation of the main field at degree 12 in
diffusivity of 1.6 m2 s1, it was found that it was possible to their model (only one aspect of the error present in
generate the required flux change of 500 MWb. observationally based field models) are much larger than the
This estimate, together with the earlier forward calculations amount by which the unsigned flux is varying. Again, a more
of Allan and Bullard (1966), Bloxham (1986a), and Drew stringent assessment would involve studying such issues for
(1993), has established the plausibility of toroidal flux expul- individual null-flux patches.
sion as a mechanism that could cause localized growth of pairs Rau et al. (2000) have performed a large number of tests on a
of reverse and normal flux patches. Such arguments, however, different suite of geodynamo models. The main focus of their
fail to convince on the need to apply a diffusive formalism study was to take the time-dependent magnetic field output by
globally. Rather, they serve as a caution against interpreting the model, carry out inversions for the underlying core surface
core flow inferred using the frozen-flux approximation without flow on the basis of the frozen-flux hypothesis, and compare the
first considering whether the specific local patterns of secular results to the known model flow. We are most interested in their
variation one wishes to explain might be produced by a diffu- preliminary test that aimed to determine how well the assump-
sive mechanism. For example, it might be very reasonable to tion of frozen flux was satisfied in the models. They observed
that the known secular variation and the secular variation pre- geomagnetic data is not capable of unambiguously detecting
dicted on the basis of the frozen-flux hypothesis using the the presence of diffusion. Yet, recent studies of short series of
known flow were similar, finding high correlation coefficients data, particularly from satellites, argue the opposite. Frozen
in the range 0.7–0.8. They observed that the diffusive contribu- flux remains undoubtedly a useful tool in understanding the
tion is significant in only a few isolated locations and that the advective motions that cause much of the observed secular
deviations from the frozen-flux assumption are not so large as to variation. It should, however, always be borne in mind that
preclude its use in the determination of core motions from the the frozen-flux assumption is only an approximation and that
magnetic data. It should be noted that their models have Rm in magnetic diffusion will inevitable be present at some level,
the range 118–320; Rm for the Earth is expected to be rather causing violation of flux constraints over long time intervals
larger than this (around 500 – see Roberts and Glatzmaier, or where field gradients are very large.
2000), and hence, the frozen-flux approximation might perform
even better in reality. Perhaps the most important finding of this
study is that even though the frozen-flux assumption is not 5.05.5.5 Other Invariants
perfectly satisfied (the formal necessary conditions would be We have concentrated on the frozen-flux approximation and
violated to an extent), useful information can be extracted con- its observable (and sometimes testable) consequences.
cerning fluid motions at the core surface and substantial parts of Depending on the additional approximations one is willing
the secular variation pattern can be explained using the frozen- to make, there are other invariants that can sometimes be
flux approximation. Errors in the field models (due to limits on testable. When one makes assumptions about the type of
the range of core field spherical harmonic coefficients that can fluid flow that may be occurring at the CMB, a number of
be determined by observations) and in the oversimplified other invariants arise – see, for example, those reviewed in
dynamical assumptions can apparently lead to more serious the cases of tangentially geostrophic or toroidal fluid flow in
problems in the inversions. Bloxham and Jackson (1991) and additional constraints
One final remark should be made concerning the inevitable described in Jackson and Hide (1996), Chulliat and Hulot
failure of the frozen-flux hypothesis (Backus and LeMouël, (2001), and Chulliat (2004). The possibility of using devia-
1986). We recollect the radial component of the induction tions from exact satisfaction of the constraints by field models
equation in its exact form: is discussed in Gubbins (1996) and Hulot and Chulliat (2003).
1 Here, we briefly describe an invariant that arises from per-
@t Br + rh ðuh Br Þ ¼ r2 ðrBr Þ [57]
r haps the most innocuous of assumptions additional to that of
frozen flux, that of a poorly conducting (or to be strict, a
It is clear that the frozen-flux hypothesis fails in all locations
perfectly insulating) mantle. The invariant arises from a con-
where jrh (uhBr)j
jr1r2(rBr)j. In particular, the approxi-
sideration of eqn [37], that is, the Navier–Stokes equation in
mation always fails on the curve S where
the magnetostrophic limit. We consider a null-flux curve (on
rh ðuh Br Þ ¼ 0 [58] which Br ¼ 0) and examine the form of the Lorentz force J B:
we can write the horizontal part of it as
Backus and LeMouël call these curves ‘leaky curves’ and the
region around them where the approximation fails the ‘leaky ½J B h ¼ Bh Jr Jh Br [59]
belt.’ The problem, of course, is that one cannot locate these
If the mantle is a sufficiently poor conductor that the radial
places a priori, because one does not know u, and much
current Jr is negligible, then all terms on the right-hand side
depends on the relative scales of variation of Br, u, including
vanish (because Br ¼ 0 also), and we discover that the horizon-
the radial variation of Br.
tal Lorentz force vanishes on null-flux curves. Null-flux curves
One thing is certain: any point where Br and rhBr vanish is
must therefore move as if they are governed by a tangentially
inevitably on the leaky curve. Backus called such places ‘touch
geostrophic force balance, even if the true force balance over
points’; it is at these places that null-flux curves can appear or
the core is one of magnetostrophic equilibrium. The repercus-
disappear. Of course, it is incredibly difficult, if not impossible,
sion of this is the following: because null-flux curves are mate-
to determine accurately the positions of such points on the
rial curves (Section 5.05.5.4.1), they must obey Kelvin’s
CMB (Backus, 1988), and hence, very little effort has been
celebrated theorem (e.g., Gill, 1982) and conserve their plan-
expended on such activity.
etary vorticity (we have ignored relative vorticity from the
outset by dropping the nonlinear advection term in the
5.05.5.4.5.1 Summary of applicability of the frozen-flux
Navier–Stokes equation). We therefore conclude that the fol-
approximation in core studies
lowing integral must hold:
After 40 years of study using observations, theoretical argu-
ments, and numerical tests, the worth and limitations of ð
d
using the frozen-flux approximation as an explanation of sec- cos ydS ¼ 0 [60]
dt S
ular variation are now apparent. The approximate invariance of
the unsigned flux over intervals of several hundred years in over any patch S that is a null-flux patch, where y is colatitude.
both observationally based field models and numerical simu- For more details, see Gubbins (1991) and Jackson (1996). The
lations demonstrates that flux is well conserved globally. On a interesting aspect of these constraints is that null-flux patches
regional level, the picture is much less clear. Field models were previously free to shrink in situ while conserving their flux
spanning over 100 years that satisfy flux conservation condi- (by concentrating the flux into a smaller area while increasing the
tions and satisfactorily fit observations tell us that this type of amplitude of Br). This new constraint places demands on their
area, so that they cannot shrink in situ; they must instead move Purucker and Mathieu Dunberry. We acknowledge the crucial
part of their area to a different latitude, either toward the pole role played by observatories and their staff in the long-term
(for shrinking) or away from the pole (for enlargement). These monitoring of the field.
constraints have been imposed on field models by Jackson et al.
(2007), who find that they can fit 100 years of historical data
adequately even when the constraints are imposed.
References
Alexandrescu M, Courtillot V, and LeMouël JL (1996a) Geomagnetic field direction in
5.05.6 Summary Paris since the mid-sixteenth century. Physics of the Earth and Planetary Interiors
98: 321–360.
We have described the basis of our knowledge of the secular Alexandrescu M, Gibert D, Hulot G, LeMouël JL, and Saracco G (1995) Detection of
variation of the magnetic field, starting from the basic datasets geomagnetic jerks using wavelet analysis. Journal of Geophysical Research
100(B10): 12557–12572.
that are available. We subsequently moved through the math- Alexandrescu M, Gibert D, Hulot G, LeMouël JL, and Saracco G (1996b) Worldwide
ematical techniques that are used for representing the four- wavelet analysis of geomagnetic jerks. Journal of Geophysical Research 101(B10):
dimensional vector B(r, t) and discussed the underlying fluid 21975–21994.
dynamics and electrodynamics of the core. We ended with a Alexandrescu M, Gibert D, LeMouël JL, Hulot G, and Saracco G (1999) An estimate of
average lower mantle conductivity by wavelet analysis of geomagnetic jerks. Journal
discussion of some of the controversies surrounding the
of Geophysical Research 104: 17735–17746.
approximations employed and what can be gleaned about Allan DW and Bullard EC (1966) The secular variation of the Earth’s magnetic field.
the physical state at the top of the core. We have stopped Proceedings of the Cambridge Philosophical Society 62(3): 783–809.
short of describing the industry of computing models of core Asari S, Lesur V, and Mandea M (2010) Geomagnetic secular variation violating the
velocities and the interesting conclusions that can be drawn frozen-flux condition at the core surface. Earth, Planets and Space 62: 693–709.
Backus G (1968) Kinematics of geomagnetic secular variation in a perfectly conducting
from them. For example, models of flow at the core surface core. Philosophical Transactions of the Royal Society of London, Series A
appear to have certain components linked to motions deeper 263: 239–266.
within the core: these are the axisymmetric zonal flows that Backus GE (1986) Poloidal and toroidal fields in geomagnetic field modeling. Reviews
apparently are linked to certain modes of oscillation in the of Geophysics 24: 75–109.
Backus G (1988) Bayesian inference in geomagnetism. Geophysical Journal of the
core, often termed torsional oscillations, that are most easily
Royal Astronomical Society 92: 125–142.
viewed as motions of nested cylinders (coaxial with the rota- Backus GE and LeMouël JL (1986) The region on the core-mantle boundary where a
tion axis) filling the core. These oscillations are almost cer- geostrophic velocity field can be determined from frozen-flux geomagnetic data.
tainly linked to changes in the rotation rate of the Earth, Geophysical Journal of the Royal Astronomical Society 85: 617–628.
often coined changes in the length of day. This observation Backus G, Parker R, and Constable C (1996) Foundations of Geomagnetism.
Cambridge: Cambridge University Press.
in itself is remarkable, but what is more exciting is the possi- Barraclough D (1974) Spherical harmonic analysis of the geomagnetic field for eight
bility of gleaning something about the state of the magnetic epochs between 1600 and 1910. Geophysical Journal of the Royal Astronomical
field within the core from exactly these oscillations, since they Society 36: 497–513.
are strongly affected by the component of the interior magnetic Barraclough DR (1978) Spherical harmonic analysis of the geomagnetic field.
Geomagnetic Bulletin of the Institute of Geological Sciences 8.
field perpendicular to the rotation axis. For a discussion of this,
Barraclough DR (1982) Historical observations of the geomagnetic field. Philosophical
see Volume 8 of the present series. Transactions of the Royal Society of London, Series A 306: 71–78.
The reader who has consulted Chapter 5.02 will see that Barraclough DR (1985) Halley’s Atlantic magnetic surveys. In: Schr€oder W (ed.)
satellites are beginning to provide models of secular variation Historical Events and People in Geosciences. New York: Peter Lang.
with unprecedented accuracy; there is a trade-off between Barraclough DR (1995) Observations of the Earth’s magnetic field in Edinburgh from
1670 to the present day. Transactions of the Royal Society of Edinburgh: Earth
accuracy and the short observing timespan. This means that Science 85: 239–252.
historical models still have a role to play, since the accumu- Barraclough DR, Carrigan JG, and Malin SRC (2000) Observed geomagnetic field
lated effects of slow processes can be seen over long times even intensity in London since 1820. Geophysical Journal International 141: 83–99.
when the signal/noise level is lower. A number of challenges Barraclough D, Gubbins D, and Kerridge D (1989) On the use of horizontal components
of magnetic field in determining core motions. Geophysical Journal International
lay ahead, not least the question of how to optimally incorpo-
98: 293–299.
rate the newest satellite, repeat station and observatory data in Batchelor GK (1967) An Introduction to Fluid Mechanics. Cambridge: Cambridge
a homogeneous way with the older data described herein. University Press.
Becquerel AC (1840) Traité experimental de l’electricité et du magnétisme. Paris: Fermin
Didot Fréres.
Benton ER, Estes RH, and Langel RA (1987) Geomagnetic field modeling incorporating
Acknowledgments constraints from frozen-flux electromagnetism. Physics of the Earth and Planetary
Interiors 48: 241–264.
We take this opportunity to record our indebtedness to Art Benton ER and Voorhies CV (1987) Testing recent geomagnetic field models via
Jonkers for his major contribution to the developments of magnetic flux conservation at the core-mantle boundary. Physics of the Earth and
Planetary Interiors 48: 350–357.
the historical dataset and model and for innumerable stimu-
Benton ER and Whaler KA (1983) Rapid diffusion of poloidal geomagnetic field through
lating conversations. We thank Mioara Mandea, Neil Suttie, the weakly conducting mantle: a perturbation solution. Geophysical Journal of the
and Florian Lhuillier for supplying data for some of the figures Royal Astronomical Society 75: 77–100.
and Terry Sabaka and Katia Pinheiro for supplying Figures 16 Bloxham J (1985) Geomagnetic Secular Variation. Unpublished PhD Thesis, University
and 28, respectively. The advice of Susan Macmillan regarding of Cambridge.
Bloxham J (1986a) The expulsion of magnetic flux from the Earth’s core. Geophysical
observatory data is gratefully acknowledged. Nils Olsen has Journal of the Royal Astronomical Society 87: 669–678.
constantly given early access to his models, for which we Bloxham J (1986b) Models of the magnetic field at the core-mantle boundary for 1715,
are grateful. We benefited greatly from reviews by Michael 1777, and 1842. Journal of Geophysical Research 91: 13954–13966.
Bloxham J (1987) Simultaneous stochastic inversion for geomagnetic main field and Courtillot V and Le Mouël JL (1984) Geomagnetic secular variation impulses. Nature
secular variation 1. A large scale inverse problem. Journal of Geophysical Research 311: 709–716.
92: 11597–11608. Crichton Mitchell A (1932) Chapters in the history of terrestrial magnetism I. Terrestrial
Bloxham J (2002) Time-independent and time-dependent behaviour of high-latitude Magnetism and Atmospheric Electricity 37: 105–146.
flux bundles at the core-mantle boundary. Geophysical Research Letters 29(18): Crichton Mitchell A (1937) Chapters in the history of terrestrial magnetism II. Terrestrial
1854–1858. Magnetism and Atmospheric Electricity 42: 241–280.
Bloxham J and Gubbins D (1985) The secular variation of Earth’s magnetic field. Nature Crichton Mitchell A (1939) Chapters in the history of terrestrial magnetism III.
317: 777–781. Terrestrial Magnetism and Atmospheric Electricity 44: 77–80.
Bloxham J and Gubbins D (1986) Geomagnetic field analysis IV – Testing the frozen Davidson P (2001) An Introduction to Magnetohydrodynamics. Cambridge: Cambridge
flux hypothesis. Geophysical Journal of the Royal Astronomical Society University Press.
84: 139–152. De Boor C (2002) A Practical Guide to Splines. New York: Springer.
Bloxham J and Gubbins D (1987) Thermal core-mantle interactions. Nature Doell RR and Cox A (1971) Pacific geomagnetic secular variation. Science
325: 511–513. 171: 248–254.
Bloxham J, Gubbins D, and Jackson A (1989) Geomagnetic secular variation. Drew S (1993) Magnetic field line expulsion into a conducting mantle. Geophysical
Philosophical Transactions of the Royal Society of London, Series A 329(1606): Journal International 115: 303–312.
415–502. Dumberry M and Bloxham J (2006) Azimuthal flows in the Earth’s core and changes in
Bloxham J and Jackson A (1989) Simultaneous stochastic inversion for geomagnetic the length of day on millennial timescales. Geophysical Journal International
main field and secular variation 2. 1820–1980. Journal of Geophysical Research 165: 32–46.
94: 15753–15769. Finlay CC (2008) Historical variation of the geomagnetic axial dipole. Physics of the
Bloxham J and Jackson A (1991) Fluid flow near the surface of the Earth’s outer core. Earth and Planetary Interiors 170: 1–14.
Reviews of Geophysics 29: 97–120. Finlay CC and Jackson A (2003) Equatorially dominated magnetic field change at the
Bloxham J and Jackson A (1992) Time-dependent mapping of the magnetic field at surface of Earth’s core. Science 300: 2084–2086.
the core-mantle boundary. Journal of Geophysical Research 97(B13): Finlay CC, Jackson A, Gillet N, and Olsen N (2012) Core surface magnetic field
19537–19563. evolution 2000–2010. Geophysical Journal International 189: 761–781.
Bloxham J, Zatman S, and Dumberry M (2002) The origin of geomagnetic jerks. Nature Fisk HW (1931) Isopors and isoporic movements. Section of Terrestrial Magnetism and
420: 685–687. Electricity of the International Geodetic and Geophysical Union, Stockholm. Bulletin
Booker JR (1969) Geomagnetic data and core motions. Proceedings of the Royal 8, pp. 280–292.
Society of London A 309: 27–40. Gauss CF (1833a) Intensitas vis magneticae terrestris ad mensuram absolutam
Braginsky SI and Roberts PH (1995) Equations governing convection in Earth’s core revocata. Gottingae: Sumtibus Dieterichianis.
and the geodynamo. Geophysical and Astrophysical Fluid Dynamics 79: 1–97. Gauss CF (1833b) Intensitas vis magneticae terrestris ad mensuram absolutam
Bullard EC, Freedman C, Gellman H, and Nixon J (1950) The westward drift of the revocata. Abstracts of the Papers Printed in the Philosophical Transactions of the
Earth’s magnetic field. Philosophical Transactions of the Royal Society of London, Royal Society of London 3: 166–174.
Series A 243: 67–92. Gauss CF (1839) Allgemeine Theorie des Erdmagnetismus: Resultate aus den
Bullard EC and Gellman H (1954) Homogeneous dynamos and terrestrial magnetism. Beobachtungen des magnetischen Verein im Jahre 1838. Leipzig.
Philosophical Transactions of the Royal Society of London, Series A 247: 213–278. Gilbert W (1600) De Magnete. London: P. Short (English translation, Chiswick Press,
Busse FH (1975) A necessary condition for the geodynamo. Journal of Geophysical London, 1900).
Research 80: 278–280. Gill AE (1982) Atmosphere–Ocean Dynamics. Cambridge: Cambridge University Press.
Cafarella L, De Santis A, and Meloni A (1992) Secular variation in Italy from historical Gillet N, Jackson A, and Finlay CC (2007) Maximum entropy regularization of time-
geomagnetic field measurements. Physics of the Earth and Planetary Interiors dependent geomagnetic field models. Geophysical Journal International
73: 206–221. 171: 1005–1016. http://dx.doi.org/10.1111/j.1365- 246X.2007.03521.x.
Cain JC, Daniels WE, Hendricks SJ, and Jensen DC (1965) An evaluation of the main Gillet N, Jault D, Finlay CC, and Olsen N (2013) Stochastic modelling of the earth’s
geomagnetic field, 1940–1962. Journal of Geophysical Research 70: 3647–3674. magnetic field: Inversion for covariances over the observatory era. Geochemistry,
Cain JC, Hendricks SJ, Langel RA, and Hudson WV (1967) A proposed model for the Geophysics, Geosystems 14: 766–786.
International Geomagnetic Reference Field – 1965. Journal of Geomagnetism and Gillet N, Lesur V, and Olsen N (2010) Geomagnetic core field secular variation models.
Geoelectricity 19: 335–355. Space Science Reviews 155: 129–145.
Cawood J (1977) Terrestrial magnetism and the development of international Glatzmaier GA (2002) Geodynamo simulations - how realistic are they? Annual Review
collaboration in the early nineteenth century. Annals of Science 34: 551–587. of Earth and Planetary Sciences 30: 237–257.
Cawood J (1979) The Magnetic Crusade: Science and politics in early Victorian Britain. Good GA (1985) Geomagnetics and scientific institutions in 19th century America. EOS
Isis 70(254): 493–518. Transactions American Geophysical Union 66: 521–526.
Chapman S and Bartels J (1940) Geomagnetism. vol. 2, p. 1049. London: Oxford Good GA (1988) The study of geomagnetism in the late 19th century. EOS Transactions
University Press. American Geophysical Union 69: 218–232.
Christensen UR, Olson P, and Glatzmaier G (1998) A dynamo model interpretation of Gubbins D (1975) Can the Earth’s magnetic field be sustained by core oscillations?
geomagnetic field structures. Geophysical Research Letters 25: 1565–1568. Geophysical Research Letters 2(9): 409–512.
Christensen UR and Tilgner A (2004) Power requirement of the geodynamo from ohmic Gubbins D (1983) Geomagnetic field analysis – I. Stochastic inversion. Geophysical
losses in numerical and laboratory dynamos. Nature 429: 169–171. Journal of the Royal Astronomical Society 73: 641–652.
Christensen UR, Wardinski I, and Lesur V (2012) Timescales of geomagnetic secular Gubbins D (1984) Geomagnetic field analysis – II. Secular variation consistent with a
acceleration in satellite field models and geodynamo models. Geophysical Journal perfectly conducting core. Geophysical Journal of the Royal Astronomical Society
International 190: 243–254. 77: 753–766.
Chulliat A (2004) Geomagnetic secular variation generated by a tangentially geostrophic Gubbins D (1987) Mechanism for geomagnetic polarity reversals. Nature
flow under the frozen-flux assumption – II. Sufficient conditions. Geophysical 326: 167–169.
Journal International 157(2): 537–552. Gubbins D (1991) Dynamics of the secular variation. Physics of the Earth and Planetary
Chulliat A and Hulot G (2001) Geomagnetic secular variation generated by a tangentially Interiors 68: 170–182.
geostrophic flow under the frozen-flux assumption – I. Necessary conditions. Gubbins D (1996) A formalism for the inversion of geomagnetic data for core motions
Geophysical Journal International 147(2): 237–246. with diffusion. Physics of the Earth and Planetary Interiors 98: 193–206.
Chulliat A and Olsen N (2010) Observation of magnetic diffusion in the Earth’s outer Gubbins D (1999) The distinction between geomagnetic excursions and reversals.
core from Magsat, Ørsted, and CHAMP data. Journal of Geophysical Research 115, Geophysical Journal International 173: F1–F3.
B05105. Gubbins D and Bloxham J (1985) Geomagnetic field analysis – III. Magnetic fields on
Churchman J (1794) The Magnetic Atlas or Variation Charts of the Whole Terraqueous the core mantle boundary. Geophysical Journal of the Royal Astronomical Society
Globe: Comprising a System of the Variation and Dip of the Needle, by Which, the 80: 695–713.
Observations Being Truly Made, the Longitude May Be Ascertained, 2nd edn. Gubbins D and Bloxham J (1987) Morphology of the geomagnetic field and
London. implications for the geodynamo. Nature 325: 509–511.
Constable CG, Parker RL, and Stark PB (1993) Geomagnetic field models incorporating Gubbins D, Jones AL, and Finlay CC (2006) Fall in Earth’s magnetic field is erratic.
frozen-flux constraints. Geophysical Journal International 113: 419–433. Science 312: 900–902.
Gubbins D and Kelly P (1996) A difficulty with using the frozen flux hypothesis to find Geochemistry, Geophysics, Geosystems 6(2): Q02H15. http://dx.doi.org/
steady core motions. Geophysical Research Letters 23: 1825–1828. 10.1029/2004GC000800.
Gubbins D and Roberts PH (1987) Magnetohydrodynamics of the Earth’s core. In: Kuang W and Bloxham J (1997) An Earth-like numerical dynamo model. Nature
Jacobs JA (ed.) Geomagnetism, vol. 2, pp. 1–183. Academic Press. 389: 371–374.
Halley E (1683) A theory of the variation of the magnetic compass. Philosophical Langel RA (1987) The main field. In: Jacobs JA (ed.) Geomagnetism, vol. I,
Transactions of the Royal Society of London 13: 208–221. pp. 249–512. London: Academic Press.
Halley E (1692) On the cause of the change in the variation of the magnetic needle, with Langel RA (1993) Types and Characteristics of Data for Geomagnetic Field Modeling.
a hypothesis of the structure of the internal parts of the Earth. Philosophical Washington, DC: NASA, NASA Conference Publication 3153.
Transactions of the Royal Society of London 17: 470–478. Langel RA, Estes RH, and Mead GD (1982) Some new methods in geomagnetic field
Hendricks SJ and Cain JC (1963) World magnetic survey data. NASA Goddard Space modelling applied to the 1960–1980 epoch. Journal of Geomagnetism and
Flight Centre, Technical Report, August 1963. Geoelectricity 34: 327–349.
Hide R and Malin SRC (1981) On the determination of the size of the Earth’s core from Langel RA, Kerridge DJ, Barraclough DR, and Malin SRC (1986) Geomagnetic temporal
observations of geomagnetic secular variation. Proceedings of the Royal Society of change: 1903–1982 – A spline representation. Journal of Geomagnetism and
London A 374: 15–33. Geoelectricity 38: 573–597.
Hide R and Stewartson K (1973) Hydromagnetic oscillations of the Earth’s core. Reviews Lanzerotti LJ, Chave AD, Sayres CH, Medford LV, and Maclennan CG (1993) Large-
of Geophysics 10(2): 579–598. scale electric-field measurements on the Earth’s surface – A review. Journal of
Hills RG (1979) Convection in the Earth’s Mantle Due to Viscous Shear at the Core- Geophysical Research: Planets 98(E12): 23525–23534.
Mantle Interface and Due to Large-Scale Buoyancy. PhD Thesis, New Mexico State LeMouël JL (1984) Outer core geostrophic flow and secular variation of Earth’s
University, Las Cruces, New Mexico. magnetic field. Nature 311: 734–735.
Holme R and Olsen N (2006) Core-surface flow modelling from high resolution secular Lesur V, Wardinski I, Hamoudi M, and Rother M (2010) The second generation of the
variation. Geophysical Journal International 166: 518–528. GFZ Reference Internal Magnetic Model: GRIMM-2. Earth, Planets and Space
Holme R, Olsen N, and Bairstow FL (2011) Mapping geomagnetic secular 62: 765–773.
variation at the core-mantle boundary. Geophysical Journal International Lesur V, Wardinski I, Rother M, and Mandea M (2008) GRIMM: The GFZ reference
186: 521–528. internal magnetic model based on vector satellite and observatory data. Geophysical
Hulot G and Chulliat A (2003) On the possibility of quantifying diffusion and horizontal Journal International 173: 382–394.
Lorentz forces at the Earth’s core surface. Physics of the Earth and Planetary Lhuillier F, Fournier A, Hulot G, and Aubert J (2011) The geomagnetic secular-variation
Interiors 135(1): 47–54. timescale in observations and numerical dynamo models. Geophysical Research
Hulot G, Le Huy M, and Le Mouël J-L (1993) Secousses (jerks) de la variation séculaire Letters 38: L09306.
et mouvements dans le noyau terrestre. Comptes Rendus de l’Académie des Love JJ (1999) A critique of frozen-flux inverse modelling of a nearly steady
Sciences de Paris 317: 333–341. geodynamo. Geophysical Journal International 138: 353–365.
Hulot G and LeMouël JL (1994) A statistical approach to Earth’s main magnetic field. Macmillan S (1996) A geomagnetic jerk for the early 1990’s. Earth and Planetary
Physics of the Earth and Planetary Interiors 82: 167–183. Science Letters 137: 189–192.
Hutcheson KA (1990) Geomagnetic Field Modelling. Unpublished PhD Thesis, Malin SRC (1982) Sesquicentenary of Gauss’s first measurement of the absolute
University of Cambridge. value of magnetic intensity. Philosophical Transactions of the Royal Society
Hutcheson KA and Gubbins D (1990) Earth’s magnetic field in the seventeenth century. A306: 5–8.
Journal of Geophysical Research 95: 10769–10781. Malin SRC (1987) Historical introduction to geomagnetism. In: Jacobs JA (ed.).
Jackson A (1989) The Earth’s Magnetic Field at the Core-Mantle Boundary. Geomagnetism, vol. I, pp. 1–49. London: Academic Press.
Unpublished PhD Thesis, University of Cambridge. Malin SRC and Bullard EC (1981) The direction of the Earth’s magnetic field at London,
Jackson A (1996) Kelvin’s theorem applied to Earth’s core. Proceedings of the Royal 1570–1975. Philosophical Transactions of the Royal Society of London, Series A
Society of London 452: 2195–2201. 299: 357–423.
Jackson A (1997) Time-dependency of tangentially geostrophic core surface motions. Mandea M, Bellanger E, and LeMouël JL (2000) A geomagnetic jerk for the end of the
Physics of the Earth and Planetary Interiors 103: 293–311. 20th century. Earth and Planetary Science Letters 183: 369–373.
Jackson JD (1999) Classical Electrodynamics, 3rd edn. New York: Wiley. Maus S, L€uhr H, Balasis G, Rother M, and Mandea M (2005) Introducing POMME the
Jackson A (2003) Intense equatorial flux spots on the surface of Earth’s core. Nature POtsdam magnetic model of the Earth. In: CHAMP Mission Results II, pp. 347–352.
424: 760–763. Berlin: Springer.
Jackson A (2014) Added credence for a late Dodo extinction date. Historical Biology Maus S, Manoj C, Rauberg J, Michaelis I, and Lhr H (2010) NOAA/NGDC candidate
26: 699–701. models for the 11th generation International Geomagnetic Reference Field and the
Jackson A, Constable CG, Walker MR, and Parker R (2007) Models of Earth’s main concurrent release of the 6th generation POMME magnetic model. Earth, Planets
magnetic field incorporating flux and radial vorticity constraints. Geophysical and Space 62: 729–735.
Journal International 171: 133–144. McConnell A (1980) Geomagnetic Instruments Before 1900: An Illustrated Account of
Jackson A and Hide R (1996) Invariants in toroidal and tangentially geostrophic frozen- Their Construction and Use. London: Harriet Wynter.
flux velocity fields. Geophysical Journal International 125: 925–927. Merrill RT, McElhinny MW, and McFadden PL (1998) The Magnetic Field of the Earth.
Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular London: Academic Press.
variation from historical records. Philosophical Transactions of the Royal Society of Mountaine W and Dodson J (1757) A letter to the Right Honourable the Earl of
London, Series A 358: 957–990. Macclesfield, President of the Council and Fellow of the Royal Society, concerning
Jault D, Gire C, and LeMouël JL (1988) Westward drift, core motions and exchanges of the variation of the magnetic needle; with a set of tables annexed, which exhibit
angular momentum between core and mantle. Nature 333: 353–356. the results of upwards of fifty thousand observations, in six periodic reviews, from
Jault D and LeMouël JL (1991) Physical properties at the top of the core and core the year 1700 to the year 1756, both inclusive; and are adapted to every 5
surface motions. Physics of the Earth and Planetary Interiors 68: 76–84. degrees of latitude and longitude in the more frequented oceans. Philosophical
John II’s Commission (1509) Regimento do astrolabe e do quadrante, Lisbon. Transactions of the Royal Society of London 50: 329–350.
Jonkers ART (2000) North by Northwest. Seafaring, Science, and the Earth’s Magnetic Multhauf RP and Good G (1987) A brief history of geomagnetism and a catalogue
Field (1600–1800). G€ottingen: Cuvillier Verlag. of the collections of the National Museum of American History. Smithsonian
Jonkers ART (2003) Earth’s Magnetism in the Age of Sail. Baltimore, MD: Johns Studies in History and Technology, vol. 48, p. 87. Washington, DC: Smithsonian
Hopkins University Press. Institution Press.
Jonkers ART, Jackson A, and Murray A (2003) Four centuries of geomagnetic data from Nagao H, Iyemori T, Higuchi T, and Araki T (2003) Lower mantle conductivity anomalies
historical records. Reviews of Geophysics 41: 1006. http://dx.doi.org/ estimated from geomagnetic jerks. Journal of Geophysical Research 108. http://dx.
10.1029/2002RG000115. doi.org/10.1029/2002JB001786.
Kircher A (1641) Magnes sive de Arte Magnetica (. . .), Rome. Needham J (1962) Science and Civilisation in China, vol. 4. Pt.1. Cambridge:
Kono M and Roberts PH (2002) Recent geodynamo simulations and observations of the Cambridge University Press.
geomagnetic field. Reviews of Geophysics 40: 1013. http://dx.doi.org/ O’Brien MS (1996) Resolving magnetic flux patches at the surface of the core.
10.1029/2000RG00102. Geophysical Research Letters 23: 3071–3074.
Korte M, Genevey A, Constable C, Frank U, and Schnepp E (2005) Continuous O’Brien MS, Constable CG, and Parker RL (1997) Frozen flux modelling for epochs
geomagnetic field models for the 7 millennia I: A new global data compilation. 1915 and 1980. Geophysical Journal International 128: 434–450.
Olsen N (2002) A model of the geomagnetic field and its secular variation estimated for Sabaka TJ, Olsen N, and Purucker ME (2004) Extending comprehensive models of the
epoch 2000. Geophysical Journal International 149: 454–462. Earth’s magnetic field with Oersted and CHAMP data. Geophysical Journal
Olsen N, Luhr H, Finlay CC, et al. (2014) The CHAOS-4 geomagnetic field model. International 159: 521–547.
Geophysical Journal International 197: 815–827. Shimizu H, Koyama T, and Utada H (1998) An observational constraint on the strength
Olsen N, L€uhr H, Sabaka TJ, et al. (2006) CHAOS – A model of Earth’s magnetic field of the toroidal magnetic field at the CMB by time variation of submarine cable
derived from CHAMP, Ørsted and SAC-C magnetic satellite data. Geophysical voltages. Geophysical Research Letters 25(21): 4023–4026.
Journal International 166: 67–75. Shimizu H and Utada H (2004) The feasibility of using decadal changes in the
Olsen N, Mandea M, Sabaka TJ, and Tøffner-Clausen L (2010a) The CHAOS-3 geoelectric field to probe Earth’s core. Physics of the Earth and Planetary Interiors
geomagnetic field model and candidates for the 11th generation IGRF. Earth, Planets 142(3–4): 297–319.
and Space 62: 719–727. Shure L, Parker RL, and Backus GE (1982) Harmonic splines for geomagnetic
Olsen N, Luhr H, Sabaka TJ, Michaelis I, Rauberg J, and Tøffner-Clausen L (2010b) modelling. Physics of the Earth and Planetary Interiors 28: 215–229.
CHAOS-4 – A high-resolution geomagnetic field model derived from low-altitude Silverman BW (1984) Spline smoothing: The equivalent variable kernel method. Annals
CHAMP data. AGU Fall Meeting Abstracts, December 2010. of Statistics 12: 898–916.
Olson P and Christensen UR (2002) The time averaged magnetic field in numerical Smith PJ and Needham J (1967) Magnetic declination in mediaeval China. Nature
dynamos with nonuniform boundary heat flow. Geophysical Journal International 214: 1213–1214.
151: 809–823. Stern DP (2002) A millennium of geomagnetism. Reviews of Geophysics 40: 1007.
Peregrinus P (1269) Epistola de magnete. http://dx.doi.org/10.1029/2000RG000097.
Rau S, Christensen U, Jackson A, and Wicht J (2000) Core flow inversion tested with Stevin S (1599) De Havenvinding. Leyden: Plantijn.
numerical dynamo models. Geophysical Journal International 141: 484–497. Suttie N, Holme R, Hill MJ, and Shaw J (2011) Consistent treatment of errors in
Raulin V (1867) Sur les variations seculaires du magnetisme terrestre. Actes de la archaeointensity implies rapid decay of the dipole prior to 1840. Earth and Planetary
Société linnéenne de Bordeaux 26: 1–92. Science Letters 304: 13–21.
Rayet MG (1876) Recherches sur les observations magnétiques faites à l’observatoire van Bemmelen W (1899) Die Abweichung der Magnetnadel: Beobachtungen, Säcular-
de Paris de 1667 à 1872, Volume 13 of Annales de l’Observatoire de Paris: Variation, Wert-und Isogonensysteme bis zur Mitte des XVIIIten Jahrhunderts,
Mémoires. supplement to Obsns. Roy. Mag. Met. Observatory, Batavia, vol. 21, pp. 1–109.
Roberts PH and Glatzmaier GA (2000) A test of the frozen flux approximation using a Veinberg BP (1929–1933) Catalogue of Magnetic Determinations in USSR and in
new geodynamo model. Philosophical Transactions of the Royal Society of London, Adjacent Countries from 1556 to 1926, vol. 3. Leningrad: Central Geophysical
Series A 358: 1109–1121. Observatory.
Roberts PH and Scott S (1965) On the analysis of the secular variation. 1. Voorhies CV and Benton ER (1982) Pole-strength from MAGSAT and the magnetic
A hydromagnetic constraint: Theory. Journal of Geomagnetism and Geoelectricity determination of the core radius. Geophysical Research Letters 9: 258–261.
17(2): 137–151. Whaler K and Gubbins D (1981) Spherical harmonic analysis of the geomagnetic field:
Runcorn SK (1955) The electrical conductivity of the Earth’s mantle. Transactions of the An example of a linear inverse problem. Geophysical Journal of the Royal
American Geophysical Union 36: 191–198. Astronomical Society 65: 645–693.
Sabaka TJ and Baldwin RT (1993) Modeling the Sq magnetic field from POGO and Wicht J and Olson P (2004) A detailed study of the polarity reversal mechanism in a
Magsat satellite and contemporaneous hourly observatory data. HSTX/G&G-9302, numerical dynamo model. Geochemistry, Geophysics, Geosystems 5: Q03H10.
Hughes STX Corp., 7701 Greenbelt Road, Greenbelt, MD. http://dx.doi.org/10.1029/2003GC000602.
Sabaka TJ, Langel RA, Baldwin RT, et al. (1997) The geomagnetic field 1900–1995, Wright E (1657) Certaine Errors of Navigation Detected and Corrected, with Many
including the large-scale field from magnetospheric sources, and the NASA Additions That Were Not in the Former Editions, 3rd edn. London: Joseph Moxton.
candidate models for the 1995 revision of the IGRF. Journal of Geomagnetism and Yukutake T (1962) The westward drift of the magnetic field of the Earth. Bulletin of the
Geoelectricity 49: 157–206. Earthquake Research Institute-University of Tokyo 40: 1–65.
Sabaka TJ, Olsen N, and Langel RA (2002) A comprehensive model of the quiet- Yukutake T and Tachinaka H (1969) Separation of the Earth’s magnetic field into the
time, near-earth magnetic field: Phase 3. Geophysical Journal International drifting and standing parts. Bulletin of the Earthquake Research Institute-University
151: 32–68. of Tokyo 47: 65–97.
5.06 Crustal Magnetism
ME Purucker, Goddard Space Flight Center/NASA, Greenbelt, MD, USA
KA Whaler, Grant Institute, Edinburgh, UK

5.06.1.1 Definition 187
5.06.1.2 Measurement 187
5.06.1.3 Governing Equations 187
5.06.1.4 Previous Reviews 187
5.06.1.5 Computer Software and Applications 187
5.06.1.6 Structure of the Remainder of the Chapter 188
5.06.2 Magnetic Petrology 188
5.06.3 Continental and Oceanic Magnetic Anomalies 189
5.06.3.1 Chicxulub 190
5.06.3.2 Dike Swarms 191
5.06.3.3 Cenozoic–Recent Faulting in Forearc Basins 192
5.06.3.4 Heat Flux Beneath the Antarctic Ice Sheet 192
5.06.3.5 Northern Canadian Kimberlite Province 193
5.06.3.6 Structural Control of the Urengoy Gas Field 193
5.06.4 Compilations and Models 193
5.06.4.1 Continental-Scale Compilations 194
5.06.4.2 WDMAM Compilation 196
5.06.4.3 Satellite Compilations of Crustal Magnetic Fields 196
5.06.4.4 Global Magnetization Models 196
5.06.5 The ‘Tools of the Trade’ 198
5.06.5.1 Survey Design and Resolution 198
5.06.5.2 Removal of Noncrustal Fields 199
5.06.5.3 Representations 199
5.06.5.3.1 Spherical harmonic analysis 200
5.06.5.3.2 Spectral, rectangular harmonic, and cylindrical harmonic analysis 201
5.06.5.3.3 Equivalent source modeling 201
5.06.5.3.4 Magnetic monopole modeling 201
5.06.5.3.5 HS modeling 202
5.06.5.3.6 Spherical Slepian function analysis 202
5.06.5.3.7 Minimum norm magnetization modeling 203
5.06.5.3.8 Spherical cap harmonic analysis 204
5.06.5.4 Transformations 204
5.06.5.4.1 Analytic continuation 204
5.06.5.4.2 Reduction to the pole 205
5.06.5.4.3 Pseudogravity 205
5.06.5.4.4 Spatial derivatives 206
5.06.5.4.5 Pie-crust filter 206
5.06.5.4.6 Analytic signal 206
5.06.5.4.7 Euler deconvolution 206
5.06.5.4.8 Tilt-depth method 207
5.06.5.5 Forward and Inverse Methods 207
5.06.5.5.1 Forward models 207
5.06.5.5.2 Inverse approaches 207
5.06.5.6 Resolving Interpretational Ambiguity 208
5.06.6 Spectral Overlap with Other Fields 208
5.06.7 Separation of Induced and Remanent Magnetizations 209
Acknowledgments 211
References 211

186 Crustal Magnetism
Glossary Magnetic susceptibility Expresses the proportionality factor

Analytic signal A transformation formed through the relating ambient field H to the induced magnetization.
combination of the horizontal and vertical gradients of a Magnetite Dominant magnetic mineral, often with some Ti,
magnetic anomaly. in the Earth’s crust.
CHAMP A low-earth orbiting satellite launched in 2001 MAGSAT The first satellite mission to map the Earth’s vector
designed to map the vector crustal magnetic field. magnetic field, including the long-wavelength crustal
Crust (lithosphere) The outer shell of the Earth formed by magnetic field.
differentiation under the influence of temperature and Matched filter Filter(s) that can then be used to decompose
gravity. The crust overlies the mantle, and the distinction can observed magnetic anomalies into estimates of the
be expressed in terms of seismic velocities, rock density, rock anomalies caused by sources at various depths. These depths
type, mineralogy, chemical composition, or magnetic are determined by a Fourier domain decomposition of the
properties. The lithosphere, a rheological term, includes the magnetic anomaly signal.
crust and the uppermost mantle. Observatory bias The difference between the magnetic
Curie depth The depth at which rocks lose their permanent components measured at a magnetic observatory and those
and induced magnetism by virtue of their elevated predicted by a geomagnetic model truncated at degree 13.
temperature. The Curie depth is a function of the geothermal This quantity is thought to reflect the higher-degree crustal
gradient within the Earth and the magnetic mineralogy. magnetic field contribution, in part.
Curie temperature The temperature at which a rock loses its Pseudogravity transformation Converts a magnetic
permanent and induced magnetism. anomaly into the gravity anomaly that would be caused by a
Euler deconvolution method A technique for estimating density distribution exactly proportional to the
source positions for magnetic anomalies, which relates the magnetization distribution.
magnetic field to its gradient through the specification of the Reduction to pole Transforms magnetic anomalies into the
arrangement of the magnetic sources. anomalies that would be caused by identical magnetic
Green’s function Used here to denote the relation between sources but with vertical magnetization and with
magnetic component data or the magnetic scalar potential measurement in a vertical magnetic field.
and the model, expressed as a continuous variable or Remanent magnetization One of the two types of
through model parameters. magnetization, the other being induced. The direction and
Harmonic spline A local basis function employed in intensity of a remanent magnetization is dependent on the
modeling the magnetic field. origin and history of a rock.
Induced magnetization One of two types of magnetization, Spherical Slepian function A local basis function
the other being remanent. Induced magnetizations are employed in modeling the magnetic field.
proportional in magnitude and generally parallel to the Structural index Expresses the rate of attenuation of a
ambient field H. magnetic anomaly. This attenuation is a function of the
International Geomagnetic Reference Field Values for the source geometry.
spherical harmonic coefficients from which the magnetic Tilt angle Inverse tangent of the ratio of vertical to
field can be calculated at any point in space and time. Each horizontal field derivatives, used to solve for the lateral
IGRF consists of a set of main field and secular variation positions and depths of sources.
coefficients covering a 5-year interval, thereby accounting Total field anomaly The signed scalar quantity, which
for the temporal evolution of the main field. is the most common measure of the Earth’s crustal
Koenigsberger ratio, Q, or Q-ratio Expresses the relation magnetic field.
between the strength of the induced and remanent Upward (downward) continuation Transforms observed
magnetizations. It is given by |Mr|/|Mi|. Hence, Q’s greater anomalies into the anomalies that would be observed at a
than unity indicate dominance by remanent magnetization; higher (lower) altitude.
Q’s less than unity indicate dominance by induced Werner deconvolution A depth to basement technique
magnetization. that assumes the magnetic body has a specific geometry
Magnetic basement The top of a layer of more strongly (dike-like) and solves for the depth to the top of the body
magnetized rocks, usually igneous and/or metamorphic based on four or more observations of the magnetic field
rocks, underlying more weakly magnetized sediments. over the body.
Nomenclature a Radius of Earth

t Temperature A Analytic signal
x Magnetic susceptibility B Magnetic induction
f Longitude d Magnetic source–observation distance
u Colatitude F Noncrustal magnetic field
m0 Permeability of free space G Green’s function
DT Total field anomaly gm m
n , hn Spherical harmonic coefficients
Crustal Magnetism 187
H Magnetic field intensity Q Koenigsberger ratio

M Magnetization si Structural index or attenuation rate
Ma Million years T Total field
Mi Induced magnetization t Time
Mr Remanent magnetization V Scalar potential
P Pressure v Volume
Pmn (cos q) Schmidt quasinormalized associated Legendre vf Volume fraction of magnetite
functions of degree n and order m G Gram matrix
5.06.1 Introduction total magnetic field without regard to its vector direction. The
total field anomaly (DT ) is then
5.06.1.1 Definition
DT ¼ jTj jFj [3]
Crustal magnetism is defined as magnetism originating from
rocks below their Curie temperature, in the Earth’s crust and where |T| is the magnitude of the magnetic field and |F| is the
uppermost mantle. The dominant magnetism is associated magnitude of the (largely) noncrustal field, determined from a
with igneous and metamorphic rocks, whereas sedimentary global or regional model. If vector data are available, the total
rocks generally have subordinate, but measurable, magnetism. field anomaly is calculated as
Magnetism of these ferri- and ferromagnetic materials is a
function of temperature, with a loss of magnetism as the mate- ^
DT ¼ FT [4]
rials approach their Curie temperature (typically 200–700 C). where F ^ is the unit vector in the direction of F. The geometry of
The increase of temperature with depth in the Earth means that a total field anomaly of a magnetic body dominated by
rocks below a certain depth, termed the Curie depth, will be induced magnetization is dependent on the geometry of the
nonmagnetic. This depth is typically in excess of 20 km in inducing field. At high latitudes, an induced magnetization
stable continental regions. will give a total field anomaly high (positive) over the source,
whereas at low latitudes, an induced magnetization will yield a
5.06.1.2 Measurement total field anomaly low (negative).
Crustal magnetism is measured with total and vector field

magnetometers and associated gradiometers. The total field 5.06.1.4 Previous Reviews
magnetometers exploit fundamental resonances (Primdahl, Book length reviews include those of Langel and Hinze (1998),
2000) to measure the magnitude of the field, whereas the Blakely (1995), Lindsley (1991), Hahn and Bosum (1986),
vector instruments typically utilize fluxgate magnetometers Grant and West (1965), the Encyclopedia of Geomagnetism
(Ripka, 2000). The magnetometers measure these fields from and Paleomagnetism (Gubbins and Herrero-Bervera, 2007),
borehole, ground-based, marine, aerial, balloon, or satellite and Hinze et al. (2013). Shorter chapters, within books or ency-
platforms. clopedias, have included Shive et al. (1992), Frost (1991a,b,c),
Reynolds et al. (1990c), Blakely and Connard (1989), Harrison
(1987), Bosum et al. (1985), Haggerty (1976), Zietz and
5.06.1.3 Governing Equations
Andreasen (1967), Purucker and Clark (2011), Langlais et al.
The magnetic induction B()(rj) of the component of (2009), Thébault et al. (2010), and Schott and Thébault (2011).
the magnetic field due to a magnetization distribution M is Reviews in journals include those of Mandea and Purucker
given by (2005), Nabighian et al. (2005), Nabighian and Asten (2002),
ð( ) Clark (1997, 1999), Phillips et al. (1991), Keller (1988),
ðÞ m0 1
BðÞ rj ¼ Îj ∇rj ∇s MðsÞ dv [1] Mayhew and LaBrecque (1987), Paterson and Reeves (1985),
4p r j s Grant (1985), Mayhew et al. (1985), Haggerty (1979), Hinze
v
(1979), Zietz and Bhattacharyya (1975), and Robinson et al.
ðÞ
where Îj is the unit vector in the direction, v is the volume of (2008). Bibliographies include Langel and Benson (1987) and
the magnetized crust, and the quantity in brackets is Green’s Hill (1991) and online at http://crustal.usgs.gov/geophysics/
function relating magnetization M to magnetostatic potential North_America.html.
V. The subscript on the gradient (∇) operator indicates whether
derivatives are with respect to observation point coordinates
(rj) or locations within the magnetized crust (s). The magneti-
5.06.1.5 Computer Software and Applications
zation M is the vector sum of remanent magnetization Mr and Computer software for crustal magnetic field applications
induced magnetization Mi: includes a collection of Fortran subroutines in Appendix B of
Blakely (1995). Another software resource is the potential field
M ¼ M i þ Mr [2]
software programs of the US Geological Survey (Phillips,
In the case of terrestrial crustal magnetic field observations, 1997). MATLAB and Fortran routines for the evaluation of
what is often measured is the total field, the magnitude of the spherical harmonic (SH) models are described in Olsen et al.
(2006a). Other Fortran-based tools for working with SH the governing relationship is of the form |M| ¼ w|H|. Magnetic
include SHTOOLS (cf. Wieczorek et al., 2012), an archive of susceptibility in many rocks is strongly controlled by their
Fortran 95-based software that can be used to perform SH magnetite content, and the empirically determined relation-
transforms and reconstructions, rotations, and multitaper ship (Shive et al., 1992) is
spectral analyses. The transforms are demonstrated to be accu-
w 0:2 4pvf [5]
rate through SH degree 2800 and have been optimized for
speed. Transforms are calculated using quadratures, via either where w is the susceptibility and vf is the volume fraction of
sampling theorem of Driscoll and Healy (1994) or Gauss– magnetite. Many authors use k for susceptibility.
Legendre quadrature. Programs for high-degree spherical har- Magnetic remanence, on the other hand, while also corre-
monic analysis (SHA) and synthesis (Adams and Swarztrauber, lated with titanomagnetite content, is strongly dependent on
1997) are associated with SPHEREPACK. Occasionally, the the grain size, shape, and microstructure of the magnetic min-
journal Computers & Geosciences includes chapters of relevance. erals. The Koenigsberger ratio (Q) measures the relative
The ‘Numerical Recipes’ books (Press et al., 1992, 1996, 1997) strengths of the induced and remanent magnetizations. It is
are another resource for inverse codes, sparse matrix theory, given by |Mr|/|Mi|. Hence, Q > 1 indicates dominance by rem-
wavelets, interpolation, and Fourier and spectral applications. anent magnetization; Q < 1 indicates dominance by induced
The publicly available generic mapping package (GMT) is use- magnetization. Representative tables and values of susceptibil-
ful for both producing maps and analyzing potential field data. ity and Q can be found in Clark (1997).
It is documented in Wessel and Smith (1998). Commonly The magnetic properties of igneous and metamorphic rocks
used commercial codes include Geosoft, MATLAB, and IDL. are a reflection of the partitioning of iron between oxide and
Online applications include those for the evaluation of silicate phases and do not correspond to standard petrologic
terrestrial and planetary magnetic fields of the solar system classifications. This partitioning occurs in the near-surface
(Nicholas et al., 2011), the International Geomagnetic Refer- realm (Clark, 1997) and probably also within the deep litho-
ence Field (IGRF) (Maus and Macmillan, 2005), and the Atlas sphere (cf. Wasilewski and Mayhew, 1992). Standard sedimen-
of Structural Geophysics (Jessell, 2001). tary rock classifications, on the other hand, do show a
correspondence with magnetic properties. Fe-rich chemical sed-
iments (e.g., banded iron formations) and immature clastic
5.06.1.6 Structure of the Remainder of the Chapter
sediments with abundant magnetite are two strongly magnetic
The remainder of this chapter begins with a summary of the sedimentary rock types, for example. Iron sulfide minerals, pos-
salient points of magnetic petrology. We then outline the sibly associated with hydrocarbon migration or abiological pro-
utility of crustal magnetism through a series of case studies cesses, may also produce subtle magnetic anomalies over
and discuss compilations to produce models at continental sedimentary basins (Reynolds et al., 1991, 1994), but the pro-
or larger scale. This is followed by details of the processing, cesses are still controversial and an active area of research (Stone
transformation, and modeling methods that are applied to et al., 2004).
crustal magnetic data to facilitate interpretation. The issue of Igneous and metamorphic rock types (e.g., granodiorite,
the separation of the various contributions to the measured rhyolite, and gabbro) often exhibit bimodal susceptibility dis-
magnetic field is then addressed, and we conclude with one of tributions, a reflection of ferromagnetic and paramagnetic
the key outstanding questions, identifying the induced and populations (cf. Figure 7 in Clark, 1999). This was first recog-
remanent components of magnetization. We have included nized as a consequence of the very large petrophysical sam-
within the references both papers that we have cited within pling program conducted on the Fennoscandian Shield (cf.
the text of the review and others, which, while important, we Korhonen et al., 1993). Iron in the paramagnetic population
were unable to discuss because of space limitations. is incorporated into silicate phases, whereas iron in the ferro-
magnetic population is typically in magnetite.
The magnetic petrology of granitic rocks provides an exam-
5.06.2 Magnetic Petrology ple of this bimodal distribution, with the relatively oxidized,
magnetite-rich, I-type granitoids contrasting with the relatively
An understanding of the processes that create, alter, and reduced, ilmenite-rich, S-type granitoids (Clark, 1999). These
destroy magnetic minerals in rocks is the province of magnetic granitoid types can often be distinguished by the presence of
petrology (Clark, 1997; Lindsley, 1991; Purucker and Clark, common minor minerals. Hornblende–biotite granodiorites
2011). This field integrates rock magnetism (see Chapter 5.08) are usually ferromagnetic, whereas muscovite–biotite granodi-
and petrology to address questions such as the effects of meta- orites are not (Clark, 1997). Economic mineralization (Cu, Au,
morphism, hydrothermal alteration, rock composition, and Mo, and Sn) also shows patterns (Ishihara, 1981) that are
redox state on magnetic properties. Magnetic minerals of controlled in part by this classification.
major importance to an understanding of crustal magnetism Although the rule of thumb that basic rocks are more
are the Fe–Ti spinel group (magnetite and titanomagnetite), magnetic than silicic rocks is often violated, rocks from within
the rhombohedral titanohematites, and monoclinic pyrrho- a single igneous province are more likely to show this tendency
tite. These minerals can possess remanent (permanent) or than are larger population samples. Hence, the interpretation
induced (in response to an inducing field) magnetizations. of magnetic surveys should include the investigation of the
Induced magnetization is, to first order, proportional to, magnetic properties of representative rock samples when pos-
and parallel to the direction of, the inducing field. The propor- sible. Within-province generalizations also find that basalts
tionality constant w is called the magnetic susceptibility, and have slightly higher susceptibilities than related andesites, but
phonolites are weakly magnetic. Rhyolites also exhibit a magnetization may form. Blakely et al. (2005) have explained
bimodal susceptibility distribution. Rhyolites that are under- the long-wavelength aeromagnetic and gravity fields above the
or oversaturated with respect to alumina, or that contain iron- Cascadia forearc as an example of this process, using matched
rich olivine, are likely to be weakly magnetic (Clark, 1999). filters (Section 5.06.5.5.2.1) to establish the depth of the
Rapidly chilled basaltic rocks are characterized by high Q source and a pseudogravity transform (Section 5.06.5.4.3) to
values, and the Q-ratio is strongly correlated to the distance center the magnetic fields over their source. Satellite magnetic
from the chilled margin. As long as the primary remanent anomalies over subduction zones are also common (cf. Clark
magnetization has not been chemically or thermally modified, et al., 1985; Maus et al., 2006; Purucker and Ishihara, 2005;
even relatively thick sills and dikes have high Q values. Vasicek et al., 1988) and may have a similar explanation.
Hydrothermal alteration generally destroys magnetite and A second mechanism for magnetizations in the mantle invokes
replaces it with paramagnetic phases like zeolites, clays, or the presence of metallic alloys, which have been detected in
more weakly magnetic minerals like titanohematite (Criss xenoliths originating from the upper mantle (Toft and
and Champion, 1984). One major exception to this generali- Haggerty, 1988). Significant amounts of metal alloys in the
zation is that serpentinization of olivine-rich ultramafic rocks upper mantle could impart magnetic behaviors to depths of
produces abundant magnetite with low Q values (Saad, 1969). almost 100 km. But questions remain about how representa-
Other notable exceptions include potassic alteration associated tive of the upper mantle these metal alloys are (Frost and Shive,
with magnetic felsic-intermediate intrusives (Sexton et al., 1989; Toft and Haggerty, 1989).
1995) and potassic and sodic alteration in deeper levels of While magnetic petrologic approaches have provided signif-
iron oxide copper–gold systems (Hitzman et al., 1992). The icant insights into the interpretation of crustal magnetism,
production of hydrothermal magnetite is enhanced in mafic there still remains the problem of extrapolating from field
protoliths. observations at micron scale to hand-sample scale to scales
Prograde and retrograde metamorphism can produce appropriate for aeromagnetic or satellite observation. For exam-
marked changes in magnetic properties, and these changes ple, even within the ferromagnetic population, the distribution
are dependent on the composition of the protolith and the of magnetization or susceptibility is usually log normal and
pressure (P), temperature (t), and time (t) path of the meta- exhibits high variability. Parker (1991) has developed an inverse
morphism. For mafic igneous protoliths undergoing regional approach, which incorporates this variability into the creation
metamorphism, primary magnetite remains unchanged during and testing of a magnetization model. Some of this high vari-
zeolite to prehnite–pumpellyite-grade metamorphism in the ability can also be ascribed to surface processes, such as light-
absence of hydrothermal fluids. Subsequent metamorphism to ning (Verrier and Rochette, 2002) and weathering, which may
greenschist grade converts the magnetite to chlorite, epidote, not be observable from nonground-based platforms.
and hematite. In turn, these minerals give way to biotite and
amphibole in the amphibolite facies of regional metamor-
phism. Magnetite is again created during granulite-grade meta- 5.06.3 Continental and Oceanic Magnetic Anomalies
morphism. At the highest metamorphic grade (eclogite), the
iron returns to silicates such as clinopyroxene and garnet. For Because magnetic oxide or sulfide-bearing phases are com-
sedimentary protoliths, the redox conditions prevailing during monly associated with other economic mineral phases, mag-
sedimentation and diagenesis, and the iron content of the netic measurements play a significant role in mineral
protolith, constrain the mineral assemblage produced during exploration. Mapping of the crustal magnetic field is a geologic
metamorphism. and exploration tool in the terrestrial environment and pro-
The magnetic state of the lower crust remains poorly vides a third dimension to surface observations of composition
known. Although P and t can be predicted, the protolith’s and structure. The magnetic method also contributes to plate
history and current compositions are the subject of specula- tectonic theory, oil and gas exploration, structural geology, and
tion. Because of the lower crust’s elevated temperature, geologic mapping. The generation of new seafloor at the ridge
induced and viscous remanent magnetizations are expected crest was established via the magnetic method. The symmetry
to be strong (Shive et al., 1992). Inferences from deep drilling of the magnetic patterns (Vine and Matthews, 1963) about the
and seismic constraints (Prodehl and Mooney, 2012) suggest a ridge crest is often cited as the breakthrough that led to the
generally mafic composition. A host of mineralogical and widespread acceptance of plate tectonics. The magnetic time-
magnetic changes may occur, with maximum magnetizations scale (Heirtzler et al., 1968), suitably calibrated with numerical
in the granulite-facies zone. Stable large remanence in ages, serves many purposes in the Earth sciences. Inferences
ilmenite–hematite intergrowths (McEnroe and Brown, 2000; from crustal magnetic fields, interpreted in conjunction with
McEnroe et al., 2001a,b) within granulite-facies rocks provides other geologic and geophysical information, can locate kim-
another mechanism for producing magnetic rocks within the berlite pipes, impact structures, plutons, ophiolites, and other
lower crust. geologic entities that have a magnetic contrast with their sur-
Two mechanisms have been suggested for large-scale mag- roundings. This permits extrapolation from, or interpolation
netizations within the mantle. The conversion of metabasalt to between, outcrops, drill holes, or regions of localized geophys-
eclogite within subducting oceanic crust releases large amounts ical measurements into areas where surficial materials may
of water into the surrounding upper-mantle peridotite and obscure the feature. Magnetic studies can locate faults, folds,
may produce serpentinite (Hyndman and Peacock, 2003). As and unconformities and describe their geometric properties.
long as the mantle wedge in the subduction zone is cooler than Magnetic measurements provide constraints on the amount of
the Curie temperature, it is possible that a significant sediment in a depositional basin by characterizing its depth
and dimensions. Magnetic measurements can be used to infer and its relevance in petroleum exploration. The proprietary
heat flux and the depth to the bottom of the magnetic crust, nature of this kind of work means that while there are no
because magnetic properties are temperature-dependent. shortage of chapters discussing depth to basement techniques
Finally, crustal magnetic fields can help delineate suture (e.g., Hsu et al., 1998; Ku and Sharp, 1983; Li, 2003;
zones or terrane boundaries and unravel the history of volcanic Mushayandebvu et al., 2001, 2004; Nabighian et al., 2005;
terranes. Naudy, 1971; Peters, 1949; Salem and Ravat, 2003; Silva and
In the sections that follow, we use a case study approach to Barbosa, 2003; Silva et al., 2001; Thompson, 1982; Thurston
illustrate the utility of the magnetic method. We begin with the and Smith, 1997; Thurston et al., 2002), there is only a single
Chicxulub impact structure, showing how it was first recog- volume (Gibson and Millegan, 1998) that focuses on the role
nized using a combination of aeromagnetic and gravity data of magnetics in an integrated hydrocarbon exploration
and how these data sets have been used to produce program.
3-dimensional (3-D) models of the structure. We proceed
then to review geodynamic interpretations of aeromagnetic
5.06.3.1 Chicxulub
data that have been derived from dike swarms and some of
the caveats that must be considered. We then discuss structural Located below, and straddling the coastline of the northwest
and tectonic interpretations of aeromagnetic maps over forearc Yucatan, Mexico, the Chicxulub impact structure (Figure 1) is
basins with Cenozoic to recent faulting and their role in asses- the world’s most widely known impact and produced major
sing earthquake risk. We next illustrate how magnetics has biological and environmental changes at the end of the Creta-
been used to infer heat flux under the Antarctic ice cap and ceous period 65 Ma. The enhanced porosity associated with the
how this may have applications in modeling ice flow and in collapse of nearby structures (Grieve and Therriault, 2000)
identifying undiscovered volcanic regions under the ice. In the from Chicxulub’s associated seismic events has been linked to
exploration arena, we summarize the role of aeromagnetic and the development of large hydrocarbon deposits in the Campe-
ground magnetic surveys in identifying diamond-bearing kim- che Bank region immediately to the NW. The impact site is
berlites from northern Canada. Finally, we review the struc- now covered by up to1 km of carbonate rock. First recognized
tural inferences drawn from magnetic and gravity surveys over by its circular and coincident magnetic (Figure 1) and gravity
the West Siberian basin and their relation to the world’s largest signatures in the aftermath of a 1978 survey (Penfield and
gas field, the Urengoy. A case study approach such as this might Camargo, 1981), the impact was subsequently tied to other
also have included a demonstration of the utility of magnetics diagnostic signatures such as an iridium anomaly and shocked
in determining the depth to basement in sedimentary basins quartz grains by direct drilling into the structure and dating of
24⬚
22⬚
21⬚ 22⬚
270⬚ 271⬚
20⬚
18⬚
270⬚ 272⬚ 274⬚
nT
-800 -400 -200 -100 -50 -20 20 50 100 200 400 800
Figure 1 Total field anomaly (DT ) over Chicxulub impact structure (Pilkington and Hildebrand, 2000), shown in an expanded view in the inset.
Coastline shown as a solid line. Data interpolated to a 1 km grid from digital data grids of the magnetic anomaly map of North America (Bankey et al.,
2002). Artificial illumination from the NNE and ESE. Mercator projection.
the crystallization age of the melt rocks (Hildebrand et al., assemblages. A magnetic low is frequently encountered, due to
1991; Sharpton et al., 1992). a reduction in magnetic susceptibility. Large structures such as
The magnetic signature consists of three concentric zones Chicxulub tend to exhibit a central high-amplitude anomaly.
(Pilkington and Hildebrand, 2000) with radii of 20, 45, and Imaging techniques that emphasize the edges of magnetic
80 km. The impact occurred in a carbonate sequence several bodies via derivatives or via artificial illumination in one or
kilometers thick characterized by much longer (hundreds of more directions (Wessel and Smith, 1998) are commonly
kilometers) and weaker amplitude magnetic anomalies. The employed adjuncts to magnetic survey interpretation of
innermost zone is characterized by a single, high-amplitude impacts. Specific extensions to impact, and other circular fea-
anomaly indicative of a single source. The middle zone consists tures (e.g., kimberlite pipes) within magnetic data, include
of numerous, intermediate amplitude dipolar anomalies. The circular sunshading as described by Cooper (2003) and
outermost zone consists of short-wavelength, low-amplitude Cooper and Cowan (2003) and fractional derivatives (Cowan
anomalies. The outermost zone is better defined by its gravity and Cooper, 2005) for better matching to the available data.
signature, and associated cenotes (freshwater caves), than by its Finally, impact structures on extraterrestrial bodies, such as the
magnetic signature. In their recent interpretation of the aero- Moon and Mars, often have a magnetic signature (Purucker
magnetic survey data, Pilkington and Hildebrand (2000) per- and Clark, 2011).
form 3-D modeling of the crater structure by inversion using a
two-layer model. The layers, at depths corresponding to the
5.06.3.2 Dike Swarms
melt sheet and the basement surface, are inverted individually
subsequent to separation via a wavelength filter (see The Earth hosts hundreds of radiating, arcuate, or linear mafic
Section 5.06.5.5.2.1 for the related concept of matched filter). dike swarms (Ernst et al., 1996) whose mapping has contrib-
The inner magnetized zones within the melt sheet are inter- uted to improved geodynamic models of the Earth. In southern
preted to result from hydrothermal activity at the edge of the Africa alone, one digital database (Mubu, 1995) has enumer-
central uplift and the collapsed disruption cavity. Although ated 14 000 dikes, mapped in large part because of their mag-
some lines of evidence (Snyder et al., 1999) suggest that Chic- netic expression. While some of these dikes are exposed, most
xulub may be a multiring impact structure, the magnetic data are not, and hence, the magnetic method has played a crucial
as currently modeled resolve only a single ring with a central role in their understanding. These magnetically defined dike
peak. Although the magnetic signature of Chicxulub is distinc- swarms have been used in global plate reconstructions and
tive, a variety of magnetic signatures are encountered in other locally to understand the kinematics of rifting. In addition,
terrestrial impact structures (Goussev et al., 2003; Grieve and dikes define fractures and shear zones (Figure 2), along
Therriault, 2000; Grieve et al., 2008; Pilkington and Grieve, which economic mineralization is often found.
1992; Shah et al., 2005), dependent on the target rocks, impact Mafic dikes provide evidence for magmatic activity, large
magnetizations, and subsequent evolution of these metastable igneous provinces, and mantle plumes (Ernst and Buchan,
-28⬚
-30⬚
-32⬚
120⬚ 122⬚ 124⬚ 126⬚
nT
-800 -400 -200 -100 -50 -20 20 50 100 200 400 800
Figure 2 Total field anomaly (DT ) over Australian dikes of the Archean Yilgarn craton. The E–W trending set seen here is part of the
Widgiemooltha dike swarm, dated at 2410 Ma. Map based on a 1 km grid rendition of the Magnetic Anomaly Grid Database of Australia (Milligan et al.,
2005). Artificial illumination from the east and southeast. The Kalgoorlie gold and precious metal district is located in the central portion of the
figure. Many of the ore deposits in this district are localized along fractures and shear zones (Weinberg et al., 2004). Mercator projection.
2001; Soederlund et al., 2010) and are especially useful in north to south, the package consists of a magnetic Miocene
older rocks where erosion has removed much of the other volcanic conglomerate, a thick sequence of nonmagnetic
evidence for igneous activity. In these older rocks, it is fre- marine and fluvial rocks, and variably magnetic volcanic and
quently only the dikes, representing the igneous plumbing sedimentary rocks of Eocene age. After accounting for rema-
system, that survive. Although dikes are often interpreted as nent magnetization, the magnetic contacts were picked objec-
paleostress markers, they can also reflect the preexisting struc- tively (Blakely and Simpson, 1986). Near-surface features of
ture of the lithosphere. The Jurassic dikes of southern Africa this magnetic survey have also been enhanced using a
(Chavez Gomez, 2001; Marsh, 2005; Reeves, 2000), one of the matched-filter approach (Phillips, 1997; Syberg, 1972; and
manifestations of the Karoo large igneous province, have been see Section 5.06.5.5.2.1). The deformation front of the Seattle
used to enumerate plate motion associated with the breakup of fault zone, as revealed by the seismic reflection data, lies
Gondwana. For example, Ernst and Buchan (1997) make the immediately north of, and locally coincident with, the mag-
case that the convergence point of these Jurassic dikes defines netic conglomerate. The aeromagnetic survey can also provide
the location of a paleoplume. The dikes here consist of four information on individual strands of the fault zone and
distinct swarms, the Okavango, the Save–Limpopo, the Oli- whether it is segmented (Blakely et al., 2002). The longer-
fants River, and the Lebombo. Dikes of both Jurassic age and wavelength information within these aeromagnetic surveys
Proterozoic age have been identified within the ESE-trending (Blakely et al., 2005; Finn, 1990; Wells et al., 1998) can be
Okavango dike swarm ( Jourdan et al., 2006), suggesting that used to provide a regional context for the tectonics of the
the Jurassic events represent the reactivation of a preexisting Cascadia forearc region that hosts these basins.
trend and calling into question Jurassic kinematic reconstruc-
tions made using these dikes. Many older dike swarms are now
5.06.3.4 Heat Flux Beneath the Antarctic Ice Sheet
dismembered, as in the well-documented Central Atlantic dike
swarm of Africa, North America, and South America (May, Using magnetic data to infer heat flux is possible because the
1971). Magnetic identification of dikes relies on simple pattern magnetic properties of rocks are temperature-dependent, and
matching from contour maps generated from simple source at the Curie temperature, rocks lose their magnetism. The
geometries (Vacquier et al., 1951). The depth to the top of geothermal heat flux is an important factor in the dynamics
dikes can be a valuable indicator of the kinematics of postdike of ice sheets and the occurrence of subglacial lakes and ice
faulting (Modisi et al., 2000). In Modisi et al.’s (2000) study, streams and may affect the mass balance. Direct heat flux
determinations of the depth to the tops of dikes were made measurements in ice covered regions are difficult; thus, Fox
using Euler’s homogeneity equation (see Section 5.06.5.4.7). Maule et al. (2005) developed a method using first-order fea-
Although the magnetic signature of a dike is usually easy to tures of the satellite magnetic data to estimate the heat flux
recognize, little attention has been directed to the important underneath the Antarctic ice sheet. They found that it varies
problem of magnetically recognizing dikes of common trend from 40 to 185 mW m2; that areas of high heat flux coincide
but dissimilar ages from within a single swarm. There are likely in part with known current volcanism, subglacial lakes, and ice
to be significant differences in magnetic signature, although streams; and that some areas landward of the Ronne Ice Shelf
the identification and mapping of these differences will require near the shoulder of the West Antarctic rift system may host
inputs from both field and laboratory studies. Dike swarms are active, but undiscovered, sub-ice volcanic regions.
also common on other planets and satellites within the solar Traditional methods for inferring heat flux, or the related
system, and in at least one case, within the South Pole–Aitken magnetic problem of inferring the bottom of the magnetic
basin of the Moon, they possess a significant magnetic contrast crust, have relied on the shape of radially averaged spectra
with the host rock (Purucker et al., 2012). from gridded aeromagnetic data sets (Bouligand et al., 2009;
Maus et al., 1997; Spector and Grant, 1970). To quote Blakely
(1995), “this calculation ranks among the most difficult in
5.06.3.3 Cenozoic–Recent Faulting in Forearc Basins
potential field inversion.” At all wavelengths, the contribution
Forearc basins around the Pacific Rim are the site of devastating from the bottom of the magnetic source is dominated by
earthquakes because of their proximity to large population contributions from the top. The top of the source must be
centers (Olsen et al., 2007). Three types of earthquakes (mega- also be known, in itself a difficult problem. The estimate of
thrust contact, deep intraslab, and shallow) are commonly the bottom focuses on the lowest wave numbers, which over-
encountered in these basins (Saltus et al., 2005). The faults lap with poorly known regional fields that may be unrelated to
that host earthquakes occurring along shallow crustal faults in the bottom of the magnetic bodies. There also exists a depen-
the overriding continental plate can sometimes be located with dence on the characteristic shape of the magnetic bodies and
high-resolution magnetic surveys. The Seattle fault zone, an an assumption about the magnetization distribution. Assum-
east-trending zone of reverse faulting extending through Seat- ing the magnetization is spatially uncorrelated (‘white’) is
tle, Washington, was the site of an M7 earthquake about 1100 common, although magnetic susceptibility distributions are
years ago (Bucknam et al., 1992) and is an example of such a often correlated (Pilkington and Todoeschuck, 1995).
fault. Mapped geologically and with an aeromagnetic survey The method of Fox Maule et al. (2005, 2009) uses a self-
(Blakely et al., 2002), and studied along several profiles with consistent compositional and thermal model of the mantle
seismic reflection surveys, this region hosts a tripartite package and crust (Nataf and Ricard, 1996) as a starting point, which
of rocks in close proximity to the fault zone. The package has a is then modified in an iterative fashion with the satellite data
distinct magnetic signature and allows the fault zone to be until the magnetic field predicted by the model matches the
traced in areas of poor exposure, or where it is covered. From observed magnetic field within some specified error. At the
scale of the surveys used (400 þ km wavelength), a unique 5.06.3.6 Structural Control of the Urengoy Gas Field
solution is guaranteed by assuming that induced magnetiza-
The West Siberian basin, one of the world’s largest sedimentary
tions dominate over remanent magnetizations in continental
basins developed on continental crust, hosts a super giant gas
crust and that vertical crustal thickness variations dominate
accumulation in the Urengoy field (Littke et al., 1999). The
over lateral susceptibility variations (Purucker and Ishihara,
hydrocarbons in the Urengoy are found in an anticlinal trap
2005). The resulting magnetic crustal thickness is then
defined by rejuvenated graben faults (Gibson, 1998a; Grace
used as one boundary condition in a thermal model of the
and Hart, 1990). Aeromagnetic (Makarova, 1974) and gravity
continental crust, assuming 1-D heat conduction and using a
(Arctic Gravity Project, 2002) mapping (Figure 3) over this
simple model to account for radioactive heat production in the
region reveals north–south trending positive anomalies that
crust. The largest complications and uncertainties in this
are fundamentally lithologic, originating in Permian–Triassic
approach are (1) uncertainties in determining the magnetic
basalt now found in rift basins. The basalt in these buried
field model in the dynamic, high-latitude auroral, subauroral,
grabens is of the same age (Reichow et al., 2002) as the bulk
and polar cap region; (2) the starting seismic and thermal
of the Siberian Traps exposed further east on the Siberian
model; (3) uncertainties in the upper- and lower-temperature
platform.
boundary conditions; (4) lateral variations in thermal
The Siberian Traps, part of the largest recorded terrestrial
conductivity; and (5) lateral variations in (viscous) remanent
flood basalt province, are contemporaneous with the end-
magnetization.
Permian extinction (Erwin, 1994), the largest mass extinction
of the Phanerozoic, although a causal relation between the two
has not yet been established (Elkins-Tanton and Bowring,
5.06.3.5 Northern Canadian Kimberlite Province 2006). The West Siberian rift basins define the base of the
sedimentary column, and subsequent postrift deposition
Diamond-bearing kimberlites were first recognized in rocks of
from the Jurassic to the Cretaceous consists of fluviatile and
cratonic North America more than 150 years ago. Exploration
marginal marine sediments. The boundary faults of these
interest focused on the Slave craton in Canada beginning in the
basins were reactivated later, and hence, the magnetic and
1970s, and the discovery of diamond-bearing kimberlites in
gravity anomalies serve to reveal indirectly the faults that
the early 1990s set off a mineral staking rush (Krajick, 2001).
define the hydrocarbon trap. The graben faults were rejuve-
By 2004, these deposits accounted for 15% of global diamond
nated in the Early Cretaceous and created broad arches in the
output by value. The exploration program relied on a comple-
Cretaceous sediments. Maturation of Jurassic source rocks was
mentary suite of geochemical and geophysical techniques,
followed by migration of hydrocarbons into traps located
of which the magnetic technique was one (Reed and
within the Pokur Formation of Cenomanian age.
Witherly, 2007). Exploration usually proceeded from a pro-
gram of indicator mineral sampling, to one of geophysical
surveys in favorable regions, finally to drilling in order to
prove the deposits (Power et al., 2004). Airborne total mag- 5.06.4 Compilations and Models
netic field and electromagnetic surveys, and follow-up ground
surveys, were the most common geophysical surveys per- Because maps of the crustal magnetic field are so useful for
formed ( Jansen and Witherly, 2004), although sometimes regional geologic understanding, and because magnetic sur-
gravity, ground-penetrating radar, and seismic techniques veys are usually acquired over small regions, there is a need
were used. The kimberlite host rock often exhibits a positive to assemble the individual magnetic surveys into larger com-
magnetic susceptibility contrast and a strong remanence, pilations. This assembly is often assisted by the addition of
compared to the surrounding country rock, commonly a longer, higher-altitude magnetic surveys that serve to tie the
high-grade metamorphic rock, or granite, in the Slave craton. individual surveys together and ameliorate the discontinuities
Kimberlite pipes are often found in geographically localized that occur at survey boundaries. Perhaps, the best known of
groups, frequently under lakes because of differential erosion, these higher-altitude surveys are a series of surveys flown in
and the remanence directions within those groups are often Australia (Tarlowski et al., 1996) for the purpose of leveling the
similar. Kimberlite pipes are often associated with diabase Australian magnetic map and the Project Magnet surveys
dikes (see previous section for a discussion of their magnetic (Coleman, 1992) of the US military. In North America, the
signature) and are also commonly intruded along preexisting 1970s saw the first regional compilations, followed by partial
zones of weakness (regional faults and geologic contacts), compilations of the entire continent in the 1980s, and more
many of which will have magnetic signatures. A completely complete compilations by 2003. Similar scenarios have played
preserved kimberlite pipe may be several hundred meters out in Australia, the former Soviet Union, China, South Asia,
wide and is often pipe- or carrot-shaped (Macnae, 1979). The Australia, the Arctic, and Antarctic; in Europe; and over the
resulting magnetic anomalies are usually circular in form world’s oceans. In contrast, Africa and South America are less
(because the area is near the magnetic pole; see Section advanced in terms of magnetic compilations, most of which
5.06.5.4.2), and data enhancement techniques are similar to have been led by industrial consortiums. The longest wave-
those used for impact craters (see Section 5.06.3.1). The use of lengths of the crustal magnetic field can be measured from
the analytic signal (see Section 5.06.5.4.6), and a pattern satellites in near-Earth orbits, and beginning in the 1960s,
recognition technique (Keating and Sailhac, 2004), has been Russian and US satellites began to measure those magnetic
shown to be of some use in identifying possible kimberlite fields. This effort continues today as an international effort,
target rocks. with the imminent launch of the ESA (European Space Agency)
Total field magnetic anomaly Free-air gravity
68⬚ 68⬚
66⬚ 66⬚
74⬚ 76⬚ 78⬚ 74⬚ 76⬚ 78⬚
nT mgal
-250 -100 -40 0 120 160 220 620 -80-18-14-11 -8 -3 0 3 7 35
Figure 3 Aeromagnetic total field anomaly (DT ) and free-air gravity maps of a portion of the West Siberian basin showing the correspondence of
magnetic and gravity lows with the Urengoy gas field. This coincidence is a consequence of both lithologic and structural factors (Gibson, 1998b). The
magnetic data are extracted from compilations of Makarova (1974) and Geological Survey of Canada (1995), the gravity data come from the Arctic
Gravity Project (2002), and the field boundaries of the Urengoy field are from Grace and Hart (1990). Lambert projection.
Swarm mission (Friis-Christensen et al., 2009). Earlier com- magnetic field, built on a deepening understanding of the
parisons (Schnetzler et al., 1985) suggested a difference in sources of the magnetic field. These models utilize both for-
amplitude between the crustal fields measured at or near the ward and inverse approaches and are frequently tested, and
surface and from satellites when the data sets were compared at enhanced, using data from the compilations.
the same altitude, with the satellite amplitude lower, but the
two approaches are beginning to converge (e.g., Ravat et al.,
2002). Upcoming satellite missions will use a gradiometer
5.06.4.1 Continental-Scale Compilations
configuration to go to SH degree 130 and beyond, and the
wavelength content of near-surface surveys is being enhanced The first experimental airborne total field magnetometer was
at both ends of the wavelength spectrum. There still remains flown in the USSR in 1936 (Gibson, 1998b), and in 1974, the
a gap in our knowledge of magnetic anomalies with wave- Ministry of Geology of the USSR published a mosaic series of
lengths from about 200 to 400 km. Only in Australia is this 18 sheets at 1:2.5 million scale showing the residual magnetic
gap partially filled. The upcoming missions will provide data intensity over the USSR and surrounding waters (Makarova,
for an updated World Digital Magnetic Anomaly Map 1974). These sheets were digitized in 1982 by the US Naval
(WDMAM), the first edition of which was released in 2007 Oceanographic Office, Stennis Space Center, Mississippi, in
(see Section 5.06.4.2). order to produce four regional one-arc-minute grids of mag-
In parallel with the development of compilations has been netic anomaly values covering the entire former Soviet Union.
the development of larger and more elaborate models of the These digital data were provided to the National Geophysical
Data Center (NGDC, 1996) for archival and public dissemina- model (Sabaka et al., 2004), has now recovered much of the
tion. The digitized data were made available on a 2.5 km grid. longer-wavelength information.
The first continental-scale compilation, of North America Magnetic observations of the North Atlantic and Arctic
(Hinze et al., 1988), was completed in preliminary form as part oceans, and adjacent landmasses, were compiled as part of a
of the Decade of North American Geology and released by the Geological Survey of Canada program (Macnab et al., 1995;
Committee for the Magnetic Anomaly Map of North America Verhoef et al., 1996; and Figure 4). The final data set, on a
in 1987. Consisting of the aeromagnetic surveys of Canada and 5 km grid, was merged from three subgrids of (1) digital airborne
the United States, and surrounding waters, the compilation observations, (2) digital shipborne observations, and (3) pre-
effort had been preceded by compilations of the United States existing grids or digitized maps. Only the shipborne observations
(Zietz, 1982) and Canada (Hood et al., 1985). The North showed some agreement with the satellite measurements of the
American compilation was released as a 2 km grid. The addi- crustal magnetic field, and as a consequence, all three subgrids
tion of aeromagnetic surveys over Mexico, and improved Cana- were filtered to remove wavelengths greater than 400 km prior to
dian and United States maps, led to a second-generation merging. An update of the North Atlantic and Arctic region
product (Bankey et al., 2002; Hernandez et al., 2001). The compilations, now including both magnetics and gravity, has
data grids comprising this map have a variety of wavelength recently been completed (Gaina et al., 2011). The updated com-
content, 1 km grid spacing, and show the total field at 1 km pilation preserves smaller-wavelength structures from the origi-
above the terrain. They are projected using a spherical trans- nal surveys by using a 2 km grid and includes updated data in the
verse Mercator with a central meridian of 100 W, base latitude NE Atlantic, the high Arctic, and north of Greenland. It also
of 0 , scale factor of 0.926, and Earth radius of 6 371 204 m. utilizes the long-wavelength component of the MF6 satellite-
Wavelengths greater than 150 km are poorly represented in based lithospheric magnetic model (Maus et al., 2008).
this compilation. A recent update (Ravat et al., 2009) using European magnetic observations, from northern, western,
the NURE data set, in conjunction with the comprehensive and eastern Europe, were compiled by Wonik et al. (2001) on a
-90⬚ 0⬚ 90⬚ 180⬚
60⬚ 60⬚
30⬚ 30⬚
0⬚ 0⬚
-30⬚ -30⬚
-60⬚ -60⬚
-180⬚ 0⬚
-90⬚ 0⬚ 90⬚ 180⬚
nT
-90⬚
-90⬚
90⬚
90⬚
-3384 0 8366
0⬚ -180⬚
Figure 4 Total field anomaly map (DT ) at 4 km above the WGS84 ellipsoid with Mercator and polar stereographic projections. Map is based on the
EMAG2 model (Maus et al., 2009). Ridges, fracture zones, and trenches are shown as white lines. Continental outlines are shown in black. The
compilation shows the absence of data from areas in white, either because the data do not exist or because they have not been released. The magnetic
dip equator is shown as a gray line extending across the entire equatorial region. The dip equator differs from the geographic equator because the
magnetic field is not exactly aligned with the spin axis of the Earth. The internal magnetic field is weakest, and horizontal, at the magnetic dip equator,
and these facts produce a host of interesting features in both the magnetic field of the crust (annihilators (Section 5.06.5.6) and induced
magnetization (Section 5.06.7)) and that of the ionosphere (equatorial electrojet; see Chapter 5.03).
5 km grid at an altitude of 3 km above mean sea level 2003; Reeves et al., 1998). The map utilized airborne, marine,
(Figure 4). Long wavelengths were retained in this survey, and satellite data to capture as many wavelengths as possible
although comparisons with satellite data suggest that wave- between 10 and 2200 km. It included all freely available major
lengths in excess of 300 km are poorly resolved. Questions digital national and regional anomaly data sets: Arctic–North
have been raised about the reliability of the registration of Atlantic, North America, Europe, South Asia, north East Asia,
this map. eastern Indian Ocean, Australia, and the Antarctic. It also
A compilation of magnetic maps of onshore and offshore included lower-resolution grids extracted from the proprietary
regions of China, Mongolia, and Russia with accompanying coverage of Getech (1996) for Africa and South America.
interpretation was produced by a team from the Geological Getech’s web page contains maps showing aeromagnetic cov-
Survey of Canada (1995). The data were on a 5 km grid, and erage worldwide in their holdings. This global magnetic field
wavelengths in excess of 400 km were removed from the map, may be quasistatic on a human timescale, but maps of this field
which is displayed with a transverse Mercator projection. continue to improve. We expect that there will be updates to
A digital compilation of marine and aeromagnetic data over this map series and releases of interim products from research
South Asia (Geological Survey of Japan, 2002) was produced groups. One such release from the NOAA group (Maus et al.,
on a 2 km grid. A Lambert azimuthal equal-area projection was 2009) is shown in Figure 4.
used with a central point at 15 N 120 E and a terrestrial radius
of 6377 km.
5.06.4.3 Satellite Compilations of Crustal Magnetic Fields
A digital compilation of aeromagnetic data over Australia
and the surrounding oceans is now in its fifth edition (Milligan Satellite models of crustal magnetic fields are commonly spher-
and Franklin, 2004; Milligan et al., 2005, 2009). The associated ical harmonic analyses of data often gathered during magnet-
database contains publicly available airborne magnetic grid ically quiet times but sometimes at all times. Two current
data for onshore and near-offshore Australia. Flight-line magmodels of this type are the MF series (MF7 uses the approach
netic data for each survey have been optimally gridded and the of Maus et al., 2008 on CHAMP data between 2007 and 2010)
grids matched in one inverse process. A composite grid at 80 m and CM4 (Sabaka et al., 2004). The two models reflect some-
grid spacing is available. Aeromagnetic traverses flown around what different design philosophies and hence have different
Australia during 1990 and 1994 are used both to control the strengths: MF7 is an inversion of data from which estimates of
quality of the grids they intersect and also to constrain grid other magnetic field sources have been removed, while CM4
merging by forcing grid data, where intersected, to the level of solves for all sources, suitably parameterized, simultaneously.
the traverse data. The map is displayed with a Lambert Thus, MF7 is a crustal field model only and extends from
conformal conic projection. degrees 16 to 133. The CHAMP magnetic field satellite input
A sparse grid of aeromagnetic and marine magnetic data, to MF7 has had removed an internal field model to degree 15, a
supplemented by satellite magnetic coverage, is available for simple magnetospheric model, and the predicted signatures
the Antarctic (Golynsky et al., 2002, 2013). The data set is from eight main ocean tidal components. Additional external
publicly available as a 5 km grid, referenced to a polar stereo- fields are subsequently removed in a track-by-track scheme.
graphic projection. Because of its design philosophy, the MF7 model can be con-
The first compilation of onshore and offshore magnetic sidered a minimum estimate of the crustal magnetic field, one
anomaly maps for China dates from the late 1980s (Chinese in which there will be some suppression of along-track mag-
National Aerogeophysics Survey and Remote Sensing Center, netic fields. Regularization has been applied to degrees higher
1989) and has been updated (2004) in digital form. than 80 to extract clusters of SH coefficients that are well
Industry-led consortia have produced magnetic compila- resolved by the data.
tions of Africa (Barritt et al., 1993), Arabia, India and the CM4, in contrast, is a comprehensive model, that is, it
Middle East (Reeves and Erren, 1994), and South America includes components of internal and external origin and toroi-
(Getech, 1996). dal fields, in addition to the crustal field (Figure 5). It is based
Oceanic data sets are held by the National Geophysical on data from all high-precision satellite magnetic field mis-
Data Center of NOAA. The most recent regional compilations sions, beginning with the POGO missions of the 1960s. It uses
are by Dyment et al. (2012), Quesnel et al. (2009), and an iteratively reweighted least-squares approach to solve for all
Ishihara (2004), where noncrustal magnetic field sources of the 25 000þ parameters using more than 2 million obser-
defined by the CM4 model of Sabaka et al. (2004) were sub- vations. Because of its design philosophy, the CM4 crustal field
tracted (see Section 5.06.5.2). component estimate is expected to have more power than that
of the MF series, both because no direct damping is applied to
the crustal field coefficients and because of the along-track
5.06.4.2 WDMAM Compilation
approach used in the MF series. CM4 has no suppression of
Although aeromagnetic data have been collected for more than along-track magnetic fields, and some of them, especially in
70 years, no worldwide compilation of them existed until the the vicinity of the dip equator, are of uncertain origin.
production of the WDMAM (Korhonen et al., 2007). This was
an initiative of the International Association of Geomagnetism
5.06.4.4 Global Magnetization Models
and Aeronomy, which had as its goal the production of a 5 km
grid of the crustal magnetic field at an altitude of 5 km. Global magnetization models often represent an integration of
The unveiling occurred at the General Assembly of IUGG/ compositional and thermal models of the crust and mantle
IAGA at Perugia, July 2007 (Korhonen et al., 2007; Ravat et al., with long-wavelength crustal magnetic field measurements
90
420 km altitude
20 Magnetosphere
0 60
nT -20
-40
30
-60
Ionosphere
0
0
-20
nT
-40 E
-30
W
20
A Lithosphere
-60
nT
F F
-20 -90
-90 -60 -30 0 30 60 90 120 130 140 150
Latitude (⬚) 10 Sep 2001 1200 LT
Figure 5 Residual progression and geographic latitude as magnetic fields from the four main source regions (core, crust, ionosphere, and
magnetosphere) are removed with the comprehensive model CM4 (Sabaka et al., 2004). This profile shows the total field T in the direction of the
main field F of a CHAMP descending (south-going) satellite pass on 10 September 2001. The pass is centered at 0300 UT and crosses the equator at
137 W and 1220 LT. Magnetically quiet conditions prevailed, with Kp ¼ 1þ for this period, Kp ¼ 0þ for the previous 3 h period, Dst ¼ 2 nT, and |d
(Dst)/dt| < 2 nT h1. Siebert and Meyer (1996) discussed magnetic indices, while Mandea and Purucker (2005) discussed their role in data selection
(see also Appendix in Chapter 1.04). For a given panel, the symbols represent residuals with respect to a main field (to SH degree 13) plus all fields
labeled in the panels above; the line is the prediction from the field source labeled in the current panel. The figure on the right shows the location of the
subsatellite point and includes a contour map of the total field anomaly (DT ) originating in the crust (to SH degree 60) from the comprehensive model
(contour interval ¼ 2 nT, dashed lines indicated negative DT ). The data from this profile were not included in the construction of the comprehensive
model. The equatorial electrojet (EE) can be seen (E) because it is most prominent around midday, following the magnetic dip equator. Although the
amplitude of the EE in the model and profile is similar, a slight amplitude offset and latitudinal shift results in a residual anomaly that might be mistaken
for a crustal anomaly. While the EE is a robust feature of the low-latitude ionosphere, it does exhibit significant variability on a day-to-day basis (Langel
et al., 1993; Lühr et al., 2004) and includes wavelengths shorter than the resolution of CM4 (SH degree 45 for the EE). The magnetic field originating in
the distant magnetosphere exhibits variations, which are not entirely accounted for by the Dst index, and this may account for some of the mismatch. In
contrast, the high-latitude current system (F) exhibits significant variability in time on a minute-to-minute basis and in space and CM4 does not attempt
to model it. Two significant crustal anomalies in Wilkes Land, Antarctica (W), and in southern Australia (A) are prominent in the profile. The frequency
content of these anomalies again exceeds the cutoff of CM4 (SH degree 65 for crustal fields). These two magnetic features (Mayhew and Johnson, 1987;
Purucker et al., 1999) were adjacent in prerift reconstructions of Gondwana (Von Frese et al., 1986).
from satellite. Both forward (Hemant and Maus, 2005a,b) and matches the observed magnetic field. A unique magnetic
inverse (Fox Maule et al., 2005) approaches have been crustal thickness solution is obtained by assuming (1) that
employed. For example, one approach (Purucker et al., 2007) induced magnetizations dominate in continental crust, (2)
has used the 3SMAC (Nataf and Ricard, 1996) compositional that the model of Dyment and Arkani-Hamed (1999) describes
and thermal model of the crust and mantle as a starting model, the oceanic remanence, and (3) that vertical thickness varia-
which is then modified in an iterative fashion with the satellite tions dominate over lateral susceptibility variations. A starting
data until the magnetic field predicted by the model (Figure 6) model is necessary for two reasons: (1) to constrain
Field from starting model
240⬚ 260⬚ 280⬚ 300⬚
Final magnetic thickness

Seismic starting model
60⬚
60⬚
Observed field
40⬚
40⬚
20⬚
20⬚
240⬚ 260⬚ 280⬚ 300⬚
240⬚ 260⬚ 280⬚ 300⬚
km
km 0 7 11 15 19 22 26 31 42 52
0 7 11 15 19 22 26 31 42 52
Unfit field: -0.8 to 0.9 nT
240⬚ 260⬚ 280⬚ 300⬚
nT
-20 -10 -6 -4 -2 -1 1 2 4 6 10 20
Figure 6 Magnetic crustal thickness map of North America (right), which reproduces satellite observations (bottom) from CHAMP, as represented
by MF4 (Maus et al., 2006). As a starting model, the seismic crustal thickness (left) from Chulick and Mooney (2002) over North America is used instead
of the global 3SMAC crustal and thermal model (Nataf and Ricard, 1996). The magnetic field is calculated from this starting model (top) under the
assumption of a constant magnetic susceptibility (w) of 0.04, and long-wavelength fields (SH degree <15) are removed, simulating a main field
subtraction. The observed (bottom) and modeled (top) fields are differenced, and the difference is inverted for a magnetic crustal thickness. The starting
model is then updated to reflect this change, and the process continues until convergence is achieved. The process is nonlinear because the total
anomaly field (DT ) is used and because of the high-pass filter. After three iterations of this technique, the residuals to the observations are less than
1 nT. Negative magnetic crustal thickness (shown in black, the minimum is 6 km) over a few regions in the ocean could be a consequence of
remanent magnetic fields. Large magnetic crustal thickness (shown in white; the maximum is 60 km) over parts of the midcontinent region could
be a consequence of inaccuracies in the starting model. Purucker et al. (2002) applied this approach over North America and found that if 3SMAC
alone was used as a starting model, negative crustal thickness was found over the southeastern United States landmass. Modification of 3SMAC to
place the major crustal thickness change near the Coastal Plain/Piedmont boundary resulted in more realistic crustal thickness.
wavelengths obscured by overlap with the core field and (2) to 5.06.5 The ‘Tools of the Trade’
ensure that most magnetic crustal thickness will be non-
5.06.5.1 Survey Design and Resolution
negative. An approach (Figure 6) such as this has been used
to define the thickness and thermal properties of cratonic Although magnetic surveys are frequently conducted as ‘mis-
North America (Purucker et al., 2002). sions of opportunity,’ where the mission design is largely dic-
A second approach (Hemant and Maus, 2005a,b, based on tated by the needs of the primary instrument, or the platform,
earlier work of Hahn and Bosum, 1986) uses the available there are many cases in which the collection of a magnetic
magnetic petrology, geologic age, and tectonic and seismic survey is the primary goal and consideration needs to be
crustal thickness information of the Earth’s crust and assigns given to optimizing the return from the survey. A recent exam-
magnetization strengths and directions to geologic units based ple of such a process, and its documentation, has been the
on their age and rock compositions. In this way, a global planning for the Swarm magnetic field satellite constellation
magnetization model of the Earth’s crust is computed. The (Olsen et al., 2006b). Survey design of aeromagnetic surveys,
model is used to predict the crustal magnetic field at satellite including the spacing and orientation of flight lines, their
altitude and compared with the observed crustal field measure- altitude, and the inclusion of tie lines, is discussed by Reid
ments. One can match the observed field by varying the (1980). For further details, refer to Chapter 5.04 on ‘Observa-
boundaries and composition of lower-crustal structures. tion and Measurement Techniques.’
5.06.5.2 Removal of Noncrustal Fields representations are not perfect. This is a particular problem
when trying to merge surveys to form larger compilations.
An important part of obtaining crustal anomalies suitable for
The most important data for IGRF modeling have been
further processing, modeling, and interpretation is adequate
permanent magnetic observatory nighttime values on magnet-
removal of noncrustal fields, primarily that arising from
ically quiet days, usually selected on the basis of geomagnetic
dynamo action in the core, and external fields due to solar–
activity indices (see Siebert and Meyer, 1996). Observatories
terrestrial interactions. The geodynamo-generated field, often
are located in areas of low spatial magnetic field gradients and
referred to as the main field, has large amplitude but varies
where crustal fields are a minimum. With the incorporation of
slowly, both temporally and spatially. External fields have
large numbers of data from orbiting satellites, and the avail-
much smaller amplitudes but have much more rapid temporal
ability of more powerful computational resources, different
and spatial variations. Time-varying external fields also induce
methods of analysis become appropriate. As described earlier,
subsurface magnetic fields throughout the crust and mantle,
rather than attempting to remove certain field sources from the
but their amplitudes are generally small compared to those of
data on a point-by-point basis prior to main field modeling,
typical crustal anomalies.
they can now be solved simultaneously. This approach has led
The effect of external fields can be minimized by collecting
to the series of ‘comprehensive models’ (e.g., Sabaka et al.,
data at magnetically quiet times, but this is frequently imprac-
2004) in which large numbers of parameters expressing the
tical, especially at higher magnetic latitudes. Many surveys are
main, crustal, external, and induced fields are coestimated (see
conducted with a continuously recording base station to mon-
Chapter 5.02). These models (Figure 7) should enable better
itor and correct for external variations. The base station is
estimates of the crustal field to be obtained from survey data
located at a site where the spatial field gradients are low (i.e.,
(Nabighian et al., 2005). The current version of the ‘compre-
not on a magnetic anomaly), ideally in roughly the center of
hensive model’ is CM4 (Sabaka et al., 2004), and the data
the survey area. It can be used to alert surveyors to magnetic
envelope extends from 1960 to July of 2002. Usage of the
storms, when data acquisition will be suspended, and as a
model outside of this time range entails two steps. First, the
means to judge the quality of the survey data. In periods of
user must update the values of Dst and F10.7, the indices used
normal activity, the temporal variations recorded at the base
for characterizing the state of the ionosphere and magneto-
station will be a reasonable approximation to the external field
sphere. Second, the internal, time-varying low-degree part of
throughout the survey area. External fields are a minimum at
the model must be replaced by a model that is valid over the
night in low latitudes and midlatitudes. Over several days (or
time span considered. This would mean the IGRF for data
longer), it is usually possible to identify a ‘nighttime quiet
collected prior to 1960 and a model such as the CHAOS
value’ (NTQV) from the base station record. The difference
model (Olsen et al., 2006a) for data collected after 2002.7.
between the NTQV and the base station value at a given time
The end result of this data reduction process should be
is an approximation to the external field at that time and is
point representations of our best estimates of the crustal mag-
removed from the survey data. The difficulty of this method
netic field, often referred to as the crustal anomaly field.
lies in the complicated behavior of the external field, combined
with the generally unknown conductivity structure of the
5.06.5.3 Representations
Earth. A second approach to the correction of external fields
is via a least-squares analysis of the mis-ties at intersecting Taking the cue from the way seismic trace information is
survey lines (Ray, 1985). The two approaches are often used displayed, crustal magnetic data collected along a profile, or
together, with the regression-type analysis used as a refine- even a series of profiles, can be represented as a set of
ment, to remove errors not removed by the first approach. ‘wiggles,’ with areas above the zero line filled or the areas
After correcting for external fields, the method for removing above and below the line colored differently. More often,
the main field depends on the size and scope of the survey. For data collected over an area are interpolated onto a regular
a small, ground-based survey, it is often sufficient to treat the grid for display as a color image and for further processing
main field as constant over the survey area. Its amplitude is and modeling. Various algorithms are suitable for gridding
likely to be well approximated by the average field (data mean) crustal magnetic data and when employed in a particular
or the NTQV. Airborne and satellite surveys typically cover instance will depend on how the data have been collected
much larger areas, over which it may not be reasonable to and the form of the crustal anomalies encountered. For exam-
assume the main field is constant, and therefore, more sophis- ple, aeromagnetic data usually have a much smaller interval
ticated main field removal methods are justified. An obvious between data points along flight lines than between them (see
extension is to remove the best-fitting line (for a 1-D survey) or Chapter 5.04), making bidirectional spline gridding
plane (2-D survey) through the data. More commonly, the (Bhattacharyya, 1969) appropriate. For more evenly spaced
predictions of a main field model such as the IGRF are sub- data, minimum curvature methods (Smith and Wessel, 1990)
tracted. IGRFs consist of values for the SH coefficients from are often applied. Widely spaced tie lines are often flown
which the field can be calculated at any point in space and time perpendicular to the survey direction, and this facilitates the
(see Chapter 5.02 for further details). Each IGRF consists of a ‘leveling’ of the survey (Ray, 1985). If the anomalies have a
set of main field and secular variation coefficients covering a 5- particular directionality to them (e.g., they arise from a series
year interval, thereby accounting for the temporal evolution of of parallel dikes), interpolations of the crustal anomaly field
the main field. However, mismatches can occur between can be improved by incorporating measured horizontal gra-
crustal anomaly fields in overlapping or abutting areas dients (Reford, 2006). Besides offering a visual image of the
obtained from surveys at different times because the IGRF data through imaging, these regular grids form the basis for all
90
40 400 km altitude
20 Magnetosphere
60
0
nT -20
-40 30
20
Ionosphere
0
0
nT
-20
-40 -30
Lithosphere
0
-60
nT
-20
-90
-90 -60 -30 0 30 60 90 0 10 20 30 40
Latitude (⬚) 9 Jan 2002, 1200 UT, 1310 LT
Figure 7 Residual progression versus geographic latitude as magnetic fields from the four main source regions (core, crust, ionosphere, and
magnetosphere) are removed with the comprehensive model CM4 (Sabaka et al., 2004). This profile shows the total field T in the direction of the main
field F of a CHAMP satellite pass on 9 January 2002, when the magnetic field was in a quiet state. The data from this profile were not included in
the construction of the comprehensive model. For a given panel, the symbols represent residuals with respect to a main field (to SH degree 13) plus all
fields labeled in the panels above; the line is the prediction from the field source labeled in the current panel. The figure on the right shows the location of
the subsatellite point and includes a contour map of the total field anomaly (DT ) originating in the crust (to SH degree 60) from the comprehensive
model (contour interval ¼2 nT; dashed lines indicated negative DT ). This figure illustrates the ionospheric/lithospheric separation that is possible with a
comprehensive model approach with the equatorial electrojet (E) in close proximity to the Bangui (B) crustal anomaly (Girdler et al., 1992). Note also the
Tornquist–Teisseyre zone (Taylor and Ravat, 1995) (T), a major lithotectonic structure in central Europe, and the Kursk (Taylor and Frawley, 1986)
anomaly (K), associated with a substantial banded iron formation.
the transformations (mostly using wave number domain 5.06.5.3.1 Spherical harmonic analysis
manipulation) discussed later that can be applied to the The most commonly employed global method is SHA, which
data. As a consequence, this is an area of continuing research has been described in Chapter 5.02 in the context of main field
(Cordell, 1992; Hansen, 1993; Keating, 1993; O’Connell modeling. The potential satisfying Laplace’s equation in spher-
et al., 2005; Ridsdill-Smith and Dentith, 1999). ical polar coordinates is expressed as
There are many methods of modeling the data that can be
used to interpolate between data points and extrapolate X
N max anþ1 X
n m
beyond the survey area. Some of these are only useful for V¼a gnm cos ðmfÞ þ hm
n sin ðmfÞ Pn ð cos yÞ
n¼1
r m¼0
local modeling; others are only applicable to data sets cover-
[6]
ing all, or at least a large part of, the Earth’s surface; and some
can be used for both local modeling and global modeling. where a is the radius of the Earth, r is the radial distance of the
Global methods are well summarized in Langel and Hinze’s observation from the center of the Earth, f denotes longitude
(1998) book, and they present most of the methods outlined and y colatitude, Pmn (cos y) are the Schmidt quasinormalized
in the succeeding text in a uniform notation. associated Legendre functions of degree n and order m, and the
gm m
n and hn are the SH coefficients to be estimated. The
difference for crustal field modeling is that the series needs to again, the basis support is local. The magnetized crust is
include much higher-harmonic-degree terms (Nmax) to repre- divided into blocks, each of which is assumed to have a mag-
sent the anomaly field adequately. SH, Pm n (cos y)cos(mf) and netic dipole at its center. In this case, the potential can be
Pmn (cos y)sin(mf), are orthogonal over the sphere. SH models expressed as
of the crustal field from satellite data (Maus et al., 2008; Sabaka
1
et al., 2004) now go to SH degrees as high as Nmax ¼ 133, V ¼ M ∇ [7]
d
corresponding to Nmax(Nmax þ 2) coefficients. This creates
computational difficulties (Cain et al., 1989; Lesur and where d is the distance between the dipole and the observation
Gubbins, 1999), and high-resolution data sets require enor- location and M is the dipole moment. The conversion factor,
mous numbers of SH coefficients to represent them ade- m0/4p, between SI and cgs units (cf. Blakely, 1995, p. 67) is
quately. SHA is not particularly well suited to global (or near- implicitly assumed to be included within eqn [7] and subse-
global) data sets of varying spatial density: since the basis quent equations relating the potential V to the magnetization
functions are themselves global, the SH series must extend up M. The model parameters of eqn [7] are the direction and
to the degree representing the shortest spatial wavelength in magnetization strength of the dipoles. However, magnetiza-
the data set (approximately c/n, where c is circumference and n tion is often assumed to be purely induced, meaning that the
is SH degree). The coefficients multiplying these degree terms dipole directions are known (parallel to the main field); the
will not be well constrained if only a limited area of the globe problem of inferring strength from vector component anomaly
has coverage at the spatial sampling rate appropriate to deter- data is then linear. The dipoles can be arranged with variable
mine them and, unless regularization is applied, their numer- density according to the data distribution, retaining the reso-
ical values may generate ringing over other parts of the globe. lution of the original data set. This can be a far more efficient
Basis functions with more local support, such as harmonic (i.e., fewer parameter) modeling method than SHA when the
splines (HSs) (Shure et al., 1982) or wavelet-like functions spatial resolution of the data set is uneven. Although the dis-
(Beggan et al., 2013; Lesur and Maus, 2006), are better suited tribution of magnetization in the crust reproducing the anom-
to data sets with variable resolution over the globe. aly data is highly nonunique (Runcorn, 1975; see Section
5.06.5.6), it can be interpreted geologically, especially if a
priori information has been incorporated in the modeling,
5.06.5.3.2 Spectral, rectangular harmonic, and cylindrical whether forward or inverse (see Section 5.06.5.5). ES dipole
harmonic analysis models (Dyment and Arkani-Hamed, 1998) are widely used
Spectral analysis can be applied to data collected along profiles
for forward modeling since they are intuitively accessible. They
or on a plane, with the usual techniques to avoid ringing,
can be used straightforwardly to predict the magnetic field at
edge effects, and spectral leakage (Lowe et al., 2001; Parker
any altitude on or above the Earth’s surface and so also provide
and O’Brien, 1997). This allows high-resolution data sets to be
an excellent tool for upward and downward (analytic) contin-
represented by small numbers of model parameters. Alldredge
uation (see Section 5.06.5.4.1).
(1981) introduced rectangular harmonic analysis, suitable
The crustal anomaly field at or above the Earth’s surface,
when the area covered is small enough for the flat Earth
even as high as typical orbiting satellite altitudes of a few
approximation to be appropriate. The method is based on
hundred kilometers, depends on the magnetization of only a
the solution to Laplace’s equation in a Cartesian geometry.
small volume of the crust directly beneath the observation
The data are first transformed into the local Cartesian coordi-
point – the footprint of an anomaly measurement is small.
nate system with origin at the center of the region, and then,
Thus, when a local basis is used to represent the anomaly field,
the coefficients are determined. They can be used to predict the
the matrix relating observations to model parameters is sparse.
field at any altitude and can also be transformed back to a
Using numerical methods for solving sparse matrix systems
spherical Earth coordinate system. Malin et al. (1996) intro-
then allows a large number of basis functions to be included,
duced extra coefficients to remove trends. In an analysis of
meaning that the resolution of the original data set can be
main field data, they concluded the method was only suitable
retained. An application of this to crustal anomaly modeling
for interpolation and not for extrapolation. Nakagawa and
was by Purucker et al. (1996), who applied the iterative con-
Yukutake (1985) used a cosine function weight near the
jugate gradient algorithm to ES dipole modeling of a satellite
edges of the region subject to rectangular harmonic analysis
crustal anomaly data set.
to reduce edge effects.
Alldredge (1982) introduced the related concept of cylin-
drical harmonic analysis, where the equation to be solved
5.06.5.3.4 Magnetic monopole modeling
is Laplace’s equation in cylindrical polar coordinates. He
Although nonphysical and therefore not suitable as an inter-
advocated this when the observations displayed cylindrical
pretation tool, crustal anomaly data can be represented by a
symmetry. Again, the size of the area to be modeled and
subsurface distribution of magnetic monopole sources
single-valuedness of the potential imposes constraints on the
(O’Brien and Parker, 1994). The number and positions of the
arguments.
monopoles on the source sphere are chosen to provide a good
representation of the data (again, a spatially variable mono-
5.06.5.3.3 Equivalent source modeling pole density can be used to represent spatially variable resolu-
A representation useful for both local modeling and global tion of the original data); the model parameters are then
modeling (typically in a Cartesian and spherical coordinate simply the monopole amplitudes (no assumptions concerning
system, respectively) is equivalent source (ES) dipoles, where directionality are required).
The potential is expressed as a sum of potential sources be constructed at any radius beyond which the solution con-
’k(r), k ¼ 1, . . ., K: verges, making analytic continuation straightforward.
HS is not used in crustal anomaly modeling because the
X
k
V ðr Þ ¼ ak ’ k ð r Þ [8] number of points in practical data sets is too large, but Whaler
k¼1 (1994) inverted a 2 2 grid of MAGSAT satellite crustal anom-
aly vector data using depleted basis HS, using it to downward
where ak are the monopole amplitudes to be determined. continue the field from satellite altitude to just above the
Monopoles at locations sk are represented by functions Earth’s surface. With the computational resources available
1 then, she was only able to retain every other basis point in
’ k ðr Þ ¼ [9]
j r sk j latitude and longitude even over a continental-sized area of the
globe (an 80 80 area centered on Africa), with consequent
The solution is calculated by minimizing loss of resolution. She also found that a sparse distribution of
2 points was required over the remainder of the globe to avoid
U ¼ C1 ðd GaÞ þ laT Ga [10]
ringing. Another disadvantage of depleted basis HS is that the
where C is the data covariance matrix, d is the data vector, G is arrangement of depleted basis points is subjective. However, it
the matrix of Green’s functions relating the monopoles to the simplifies the inversion of total field anomaly data, since the
measurements, l is a Lagrange multiplier, and aTGa is a qua- basis functions for their expansion can be chosen to be those
dratic form expressing the field complexity. ‖ ‖ denotes the for the vertical component (at a limited subset of the actual
Euclidean norm or length. G is known as the Gram matrix; its data points), in which case the matrix elements can be calcu-
(j,k)th element is the inner product of ’j and ’k. Thus, the first lated straightforwardly (Langel and Whaler, 1996).
term measures the fit to the data, and the second, the amount A more satisfactory application of HS uses sparse matrix
of structure in the resulting field model. This is an example of a techniques, allowing the full basis to be retained. Unpublished
regularized, or minimum norm, solution; by an appropriate models using the conjugate gradient algorithm based on satel-
choice of l, we can relax slightly the fit to the data such that it lite anomaly data are very similar to those obtained using other
does not attempt to model noise. The quadratic form (and global methods and also compare favorably with Whaler’s
definition of the inner product for the calculation of the (1994) depleted basis models over Africa. HS (and depleted
Gram matrix) is chosen to measure some global property of basis HS) coefficients can be converted into an infinite series of
the field such as its mean strength or lateral variability; useful SH coefficients that give power spectra similar to those
measures lead to closed-form, or at least easily calculable, obtained from SHA at low degree but typically have less
Gram matrix elements. The concept was introduced with power at higher degrees.
main field modeling in mind (see Chapter 5.02), where Achache et al. (1987) used the same basis functions to
some quantities that are expressible as quadratic norms can model satellite data but reduced the size of the linear system
be bounded theoretically or empirically. It is now widely used by recognizing that they fall to negligibly small values quickly
as a regularizing tool. For crustal modeling, it ensures that the with lateral distance from the point at which the solution is
models have minimum structure for a given fit to the data; if being calculated. They thus included only those related to data
the fit is acceptable, we can then argue that the real Earth has at points close to the point of interest, reducing the dimensions of
least as much structure as the model. Since eqn [10] minimizes the matrix to be inverted from the number of data to the
a global measure of complexity, it does not matter if the number of ‘nearby’ data points. Their recommendation is
monopole sources are distributed unevenly over the surface that points within a horizontal distance 3h, where h is satellite
to reflect the data coverage. altitude, be included. In addition, they used principal compo-
nent analysis to stabilize the inversion of the resulting (smal-
ler) matrix by including only those eigenvectors associated
5.06.5.3.5 HS modeling with the largest eigenvalues. The decision as to how many
HSs are local basis functions introduced by Shure et al. (1982) eigenvectors to include is subjective, but the eigenvalue spec-
for global main field modeling. They were the first to apply trum shows a rapid falloff for satellite data acquired above
minimum norm modeling to geomagnetism. Using Green’s 200 km altitude, making the choice relatively clear-cut. Previ-
function for the magnetostatic potential, the solution is con- ously, Langel et al. (1984) used principal component analysis
structed as a linear combination of the HS associated with each to stabilize the calculation of ES solutions.
data point, leading to the solution of a linear system of dimen-
sion the number of data points. This is impractically large even 5.06.5.3.6 Spherical Slepian function analysis
for main field modeling but naturally preserves the resolution Another regional modeling method producing coefficients that
of the original data set. To make the inversion of large data sets can be converted to their SH counterparts uses spherical Sle-
computationally tractable, Parker and Shure (1982) expanded pian functions, which provide the optimal spatial and spectral
the solution in terms of HS at only a subset of the data points, localization of a function on a region of the unit sphere (see
known as the depleted basis. The system then reduces to one of Simons et al., 2006, for a review). Spherical Slepian functions
the dimension of the number of data points. Tests based on can be used to model data over a region or to localize an
small data sets demonstrated that the depleted basis solution SH model onto a region. Like SH spherical Slepians are
had only a slightly larger norm than the minimum value orthogonal over the sphere, but in addition, they are orthogo-
obtained by HS. The resolution of the depleted basis solution nal over the region. The formulation was developed and most
depends on the spacing between the basis points. The field can problems tackled to date have been for potentials on a sphere
including the monopole coefficient, such as the geoid and the The resulting Gram matrix elements have closed-form
cosmic microwave background. In that case, an existing SH expressions involving elliptic integrals, but these can be
model to degree and order Nmax over a region of fractional approximated very accurately by expressions involving only
area A (in steradians) of the unit sphere is optimally repre- elementary functions since the thickness of the magnetized
sented by (Nmax þ 1)2 A/4p spherical Slepian functions, a layer is small in comparison with the Earth’s radius. A similar
quantity known as the Shannon number, S. The spherical simplification applies if depth-independent magnetization is
Slepian functions are obtained by the eigenvector–eigenvalue assumed; this is more appropriate for satellite data modeling,
decomposition of the symmetric ‘localization’ matrix formed since the thickness of the magnetized layer is very much smal-
from the integration over the region of interest of all possible ler than satellite altitude, so it is indistinguishable from a thin
products of SH of degree and order up to Nmax. For most sheet. The norm minimized was
regions, this must be undertaken numerically, and thus, for vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uð
large Nmax, it is a significant computational undertaking. u
kMk ¼ t M M dv [12]
Ranked by decreasing eigenvalue, the first S eigenvectors are v
the spherical Slepians that optimally project the SH coefficients
onto the region of interest. These eigenvectors are used to that is, the root-mean-square (RMS) magnetization amplitude
decompose the original SH coefficients into those optimally of the crust. The method makes no assumption about the
concentrating the potential over the region of interest and the magnetization direction, so it allows both remanent and
remainder into its complement, concentrated over the rest of induced magnetizations. Whaler and Langel (1996) chose
the unit sphere. The region of interest need not be connected, the same data set and distribution of depleted basis points
and Beggan et al. (2013) used the method to decompose Whaler (1994) used to downward continue the magnetic field
optimally MF7 up to degree and order 72 (beyond which to produce a magnetization model for Africa. For ease of
regularization was applied) into the continental and oceanic comparison with ES dipole models assuming purely induced
domains. They extended the original algorithm, which pro- magnetization, they displayed the model as components in
duced low-pass spherical Slepian functions, to produce their the direction of the main field (consistent with induced
band-pass counterparts, leading to a modified expression for magnetization or remanent magnetization acquired in today’s
the Shannon number. The band-pass spherical Slepian decom- main field), perpendicular to the main field in the meridian
position prevented energy in degrees 16 and above in MF7 from plane, and perpendicular to the meridian plane. Whaler
leaking into the lower-degree coefficients – set to zero – which and Langel (1996) note that by damping least-squares ES
represent the core field. Although the decomposition was per- inversion, the solution minimizes the same norm as they
formed on the potential representing the field, rather than field employed, that is, minimum RMS magnetization. The largest
components, it produced a good localization of the vertical component of magnetization was in the direction of the cur-
field. Beggan et al. (2013) found that the field over the conti- rent main field (or antiparallel to it), but the component of
nents dominates and that its power spectrum is significantly magnetization perpendicular to the main field in the meridian
different from that of the oceanic anomaly field – the plane was also significant in many areas. The smallest compo-
oceanic power spectrum is approximately flat, whereas the nent (perpendicular to the meridian plane) requires rotation
continental power increases with increasing harmonic degree. and translation of the magnetization vector from that which
Current research is developing methods for spherical Slepian would be recorded by rocks acquiring a contemporary rema-
analysis of vector components and producing optimal nent magnetization.
localization at any radius from analysis at any other (i.e., on A similar difficulty of loss of resolution of the solution and
analytic continuation of spherical Slepians). Spherical Slepian subjectivity of the choice of depleted basis points can be over-
analysis is most effective on quantities with significant power come in the same fashion as for HS: by employing the iterative
over a large range of SH degrees and that have an approximately conjugate gradient technique to solve the full data-by-data
white spectrum and is thus well suited to crustal anomaly field system of equations and taking advantage of the sparseness of
modeling. the matrix relating data to model parameters. Again, compar-
isons between Whaler and Langel’s (1996) depleted basis mag-
netization model for Africa and surrounds and the equivalent
5.06.5.3.7 Minimum norm magnetization modeling part of the global conjugate gradient model of Whaler et al.
Based on methods originally developed for modeling sea- (1996) are favorable. Whaler and Purucker (2005) had applied
mount magnetism (Parker et al., 1987), then adapted to this technique to Martian orbiting satellite data and compared
account for crustal magnetization when modeling the main the model to Langlais et al.’s (2004) ES dipole model. Mars
field ( Jackson, 1990, 1994), Whaler and Langel (1996) used no longer has an active dynamo so the magnetization direc-
a depleted basis minimum norm method to model crustal tion is unknown. Langlais et al. (2004) developed an iterative
magnetization from satellite anomaly field data sets. Data technique that allowed them to solve for both the amplitude
are related to magnetization varying continuously in a crust and the direction of ES dipoles. Convergence was difficult
of assumed constant thickness through eqn [1], and hence, to achieve, particularly as the dipole spacing was reduced, so
the solution is expressed as a linear combination of Green’s their final model had a coarser spacing than the separation
functions: between the data points. Nonetheless, there was good agree-
( ) ment between their model and Whaler and Purucker (2005)
ð Þ m0 1
Îj ∇rj ∇s [11] model, which was also derived from a slightly different
4p r j s data set.
5.06.5.3.8 Spherical cap harmonic analysis ð

1 ð
1
Mða; b; gÞ
Spherical cap harmonic analysis (SCHA) (Haines, 1985), and da db Mðx; y; gÞ*Rðx, y, z gÞ [15]
d
the related translated origin SCHA (TOSCA) (de Santis, 1 1
1991), has been developed to model the field over small
where * denotes convolution and
patches of the globe. As for global SHA, the potential is
expressed as a finite sum of SH but including harmonics of 1
Rðx, y, z gÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [16]
noninteger degree. Assuming the cap is centered on y ¼ 0 and x þ y þ ðz gÞ2
2 2
subtends an angle y0 at the center of the Earth, the possible
values of degree, nk, which is a function of order, m, are those Hence, the Fourier transform of V is
@P m ðyÞ
for which y0 is a zero of either Pnmk ðyÞ or n@yk . SCHA maps ð1
harmonics on the sphere to the spherical cap, so their effective e ðu; v; zÞ ¼ @
V e ðu; v; gÞ R
M eðu, v, z gÞdg [17]
wavelength is reduced accordingly. Thus, smaller-scale fea- @k0 0
tures can be represented over the cap with fewer coefficients where
than are required for global SHA. TOSCA moves the origin ð1 ð1
from the center of the Earth toward the surface along a line e ðu; v; gÞ ¼
M Mðx; y; gÞeiðuxþvyÞ dx dy [18]
joining the Earth’s center to the center of the cap. This adjusts 1 1
the wavelength represented by a given harmonics to be smal- is the Fourier transform of M, using the tilde symbol to denote
ler at the center of the cap than at the edge, an advantage if the Fourier-transformed quantities. The Fourier transform of R is
data distribution is concentrated toward the center of the
ð
1 ð pffiffiffiffiffiffiffiffiffi
region. Korte and Holme (2003) presented a method for 1 eðzyÞ u þv
2 2
iðuxþvyÞ
regularizing SCHA, pointing out that, unlike SHA, the basis qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi e dx dy ¼2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [19]
x2 þ y2 þ ðz gÞ2 u 2 þ v2
functions are not orthogonal. This means that it is not possi- 1
ble simultaneously to represent the potential for the vertical
and so
and horizontal field components exactly. Analytic continua-
ð
1 pffiffiffiffiffiffiffiffiffi
tion is prone to errors that increase with the upward or
e ðu; v; zÞ ¼ 2p @ eðzgÞ u þv
2 2
downward continuation distance, although Thébault et al. V e

Mðu; v; gÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dg [20]
@k0 u 2 þ v2
(2004) had reposed SCHA as a boundary value problem 0
within a cone extending above the reference surface in a
Let (l,m,n) be the direction cosines of M. Then,
method they call revised SCHA or R-SCHA, thereby allowing
satellite data to be downward continued to the Earth’s surface. @ @ @ @
¼l þm þn [21]
In a further extension, Thébault et al. (2006) used both @k0 @x @y @z
Neumann and Dirichlet boundary conditions in R-SCHA
and related the resulting coefficients to their SHA counter-
@f
But @x ¼ iu fe for any Fourier transform pair f , fe (and simi-
parts. Vervelidou (2013) had derived approximations that larly for differentiation with respect to y). Hence,
enable spatial power spectra to be calculated directly from pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
ð
ilu þ imv þ n u2 þ v2 e pffiffiffiffiffiffiffiffiffi
the R-SCHA coefficients, allowing it to be extended efficiently e
V ðu; v; zÞ ¼ 2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Mðu; v; gÞeðzgÞ u þv dg
2 2
to the shorter wavelengths that an R-SCHA can represent. u þv

2 2
0
[22]
We now need to relate the (Fourier-transformed) potential
5.06.5.4 Transformations V to the (Fourier-transformed) component of the magnetic
field being measured, usually that in the direction of F, which
The transformations of the next few subsections are applicable we take to have direction cosines (l0 ,m0 ,n0 ):
to 2-D data sets expressible in a rectangular geometry, such as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 0 0 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
those recorded in regional aeromagnetic surveys. They are most ilu þ imv þ n u2 þ v2 il u þ im v þ n u2 þ v2
△ Teðu; v; zÞ ¼ 2p pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
easily considered and performed in the wave number domain. u 2 þ v2
The development here follows that of Gunn (1975) closely, ð1 pffiffiffiffiffiffiffiffiffiffi
but beware of typographical errors in his manuscript. We begin e ðu; v; gÞeðzgÞ u2 þv2 dg
M [23]
0
with the expression for the magnetic scalar potential V result-
ing from a distribution of magnetization M within an infinite It is easily shown that l ¼ cos IF cos DF, m ¼ cos IF sin DF,
half-space. Assuming a uniform direction of magnetization, and n ¼ sin IF, where IF and DF denote the inclination and
the potential is (in Cartesian coordinates, with z positive declination of the main field, respectively, and similarly for
downward) (l0 ,m0 ,n0 ).
ð ð ð
@ 1 1 1 Mða; b; gÞ
V ðx; y; zÞ ¼ da db dg [13] 5.06.5.4.1 Analytic continuation
@k0 0 1 1 d
From eqn [23], it follows straightforwardly that analytic
where @/@k0 is the derivative in the direction of M and d is the (upward or downward) continuation involves convolving
source–observation distance: with a filter whose frequency response is eHk, that is, multiply-
ing by eHk in the wave number domain, where k2 ¼ u2 þ v2 is
d2 ¼ ðx aÞ2 þ ðy bÞ2 þ ðz gÞ2 [14]
the square of the wave number and H is the continuation
The a, b integral is a convolution, that is, height or depth (also measured positive downward). eHk is
sometimes referred to as the analytic continuation operator. 1 1

V ðPÞ ¼ M— p U ¼ MgM [27]
Thus, downward continuation is an amplifying and roughen- Gr Gr
ing operation (conversely for upward continuation). This is
where gM is the component of gravity in the direction of M.
analogous to the spherical case where the amplification factor
Equation [27] is Poisson’s relation. In fact, it is not necessary
is (a/r)nþ2, where here n is SH degree, a is the reference radius
for the potential and magnetization to be constant. We can
of the SH expansion, and r the analytic continuation radius.
consider a body to be composed of arbitrarily small volumes in
Examples of the uses of analytic continuation are to suppress or
which density and magnetization can be regarded as constant.
enhance short-wavelength features; reduce data collected at a
Since potentials add, eqn [27] applies to a body in which
variety of altitudes (e.g., different flying heights for airborne
density and magnetization vary in proportion. However, pseu-
surveys) to constant height, either above terrain or relative to
dogravity is defined as the gravity anomaly that would be
mean sea level; compare ground and airborne observations;
observed if the magnetization distribution were replaced by
and estimate depth to sources (e.g., Sections 5.06.5.4.6–
an identical density distribution, that is, M/r is a constant. In
5.06.5.4.8 and 5.06.5.5.2).
the wave number domain, this gives
e
geM ¼ C V [28]
5.06.5.4.2 Reduction to the pole
Reduction to the pole (RTP) is achieved by convolving with where C ¼ Gr
is a constant. Converting from the magnetic
M
a filter whose frequency response is potential to total field anomaly,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
u 2 þ v2 u2 þ v2 geM ¼ C pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi D Te [29]
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [24] il0 u þ im0 v þ n0 u2 þ v2
ilu þ imv þ n u þ v il u þ im v þ n0 u2 þ v2
2 2 0 0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(the factor u2 þ v2 in each term preserves dimensions), that Finally, converting from the component of gravity in the
is, eqn [24] is the RTP operator (or filter). It removes the factors direction of M to the vertical component gives the Fourier-
in eqn [23] associated with the direction of the main field and transformed pseudogravity, geps :
the direction of remanent magnetization. Its effect is to reduce
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the anomalies to those that would be observed at the magnetic u2 þv2
North Pole with a vertical remanent magnetization direction. geps ¼C pffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiD Te [30]
iluþimvþn u2 þv2 il0 uþim0 vþn0 u2 þv2
If the remanent magnetization direction is unknown, or mag-
netization can be assumed to be purely induced, the direction
Thus, the pseudogravity operator is
cosines of magnetization are replaced by those of the main
field. RTP becomes unstable as the magnetic equator is pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 2 þ v2
approached since the numerator of eqn [24] approaches zero
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi [31]
(recall u2 þ v2 ). At low magnetic latitudes, it is possible to ilu þ imv þ n u2 þ v2 il0 u þ im0 v þ n0 u2 þ v2
perform reduction to the equator (RTE) instead, but the form
of the resulting anomalies is not as simple as for RTP. In Note that, unlikep the expression for the RTP operator, there
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
particular, anomalies are smeared in the east–west direction, is only one factor u2 þ v2 in the numerator of eqn [31]
making north–south trending boundaries or structures diffi- because Poisson’s relation [27] relates the magnetic scalar
cult to identify. Silva (1986) has treated RTP as an inverse potential to the component of the gravitational field in the
problem, applying methods designed for stabilizing inversion direction of magnetization. This means that the pseudogravity
to stabilize it. Besides aiding interpretation through simplify- operator alters the frequency content of the signal, preferen-
ing the form of the anomalies and centering them over their tially amplifying the longer-wavelength components (in con-
causative structures, RTP eases the comparison of oceanic mag- trast to the spatial derivatives discussed in Section 5.06.5.4.4,
netic anomalies at different latitudes. which preferentially amplify the shorter-wavelength compo-
nents). Like the RTP operator, it can run into problems at low
magnetic field or magnetization inclinations. The constant C in
5.06.5.4.3 Pseudogravity eqn [30] means we can predict the pattern (but not the ampli-
The pseudogravity transformation follows from Poisson’s rela- tude) of the gravity anomalies that would be obtained over the
tion between the magnetic potential and the gravitational field. same structure. The pseudogravity transformation aids the
Consider a body with uniform magnetization (both strength comparison of magnetic and gravity anomalies, allows gravity
and direction) and density occupying a volume O. Then the methods to be used to interpret magnetic anomalies, and can
magnetic scalar potential is be used in conjunction with gravity data to determine the
ð direction of magnetization and the ratio of magnetization to
1
V ðPÞ ¼ M— p dO [25] density (Cordell and Taylor, 1971). Note that Poisson’s rela-
d
O tionship provides an example of the ambiguity in magnetiza-
where P is the observation point and d is distance from P, and tion modeling – there is no total field anomaly over a
the gravitational potential is uniformly magnetized sheet (regardless of the direction of
ð magnetization) since its gravity anomaly is constant. The result
1 extends to a spherical geometry, where more complicated mag-
U ðPÞ ¼ Gr dO [26]
d netization distributions can be shown to produce no external
O
magnetic field (Runcorn, 1975). These magnetic annihilators
where G is the gravitational constant. Combining the two, are discussed more fully in Section 5.06.5.6.
5.06.5.4.4 Spatial derivatives 5.06.5.4.6 Analytic signal

Derivatives are useful for enhancing smaller-scale features of a Another useful interpretational tool is the analytic signal,
data set, and anomalies caused by shallow bodies, and direc- defined as
tional derivatives for enhancing or suppressing features in a
@DT @DT @DT
given direction. The second vertical derivative is valuable Aðx; yÞ ¼ iþ jþi k [37]
@x @y @z
because it relates to second horizontal derivatives through
Laplace’s equation, which is satisfied by the total field anomaly where i, j, and k are unit vectors in the x-, y-, and z-directions,
as well as the scalar potential if the main field direction is respectively. The real and imaginary parts of its Fourier trans-
constant. This involves multiplying D Te by k2 and is thus a form are the horizontal and vertical derivatives of DT, respec-
significantly roughening operation. It is used to suppress tively. They form a Hilbert transform pair, as required for A to
regional gradients and to aid in the determination of source be an analytic signal, a property most easily demonstrated in
depth and the attitude of interfaces. As noted earlier, differen- the wave number domain (Roest et al., 1992). The amplitude
tiation with respect to x or y multiplies D Te by iu or iv, respec- of the analytic signal
tively. The horizontal derivative, that is, the derivative in
the direction of maximum change, therefore has a Fourier ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s
e This follows because 2 @DT 2 @DT 2 @DT 2
transform kD T. jAðx; yÞj ¼ þ þ [38]
@x @y @z
@DT @DT
— H DT ¼ iþ j [32]
@x @y is most often used for interpretation. It has maxima at magne-
tization contrasts, independent of the direction of the ambient
where i, j are unit vectors in the x-, y-directions, respectively.
magnetic field for 2-D sources and only weakly dependent on
Thus, in the wave number domain,
these directions for 3-D sources, offering a method for locating
— H D Te ¼ iuD Ti
e þ ivD Tj
e [33] the edges of magnetized bodies. If the edges are assumed
vertical, it can be used to determine depth to sources. Over a
and hence, 2-D vertical contrast in magnetization at depth d, the ampli-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tude of the analytic signal is proportional to 1/(x2 þ d2) where x
— H D Te ¼ — H D T*— e H D Te ¼ kD Te [34] is the horizontal distance from the contact (Nabighian, 1972).
Thus, the half-width of the curve gives d, the depth to the
where here * denotes complex conjugate. First derivatives are
contrast. Automatic methods for finding the positions of max-
also required in some methods of estimating depth to sources
ima of quantities such as the analytic signal are given by
(see, e.g., Section 5.06.5.4.7). Directional derivatives, for
Blakely and Simpson (1986). Other boundary edge finding
example, in the direction defined by an angle ’0 with the
methods include the terracing operator of Cordell and
x-axis, are obtained by taking the magnitude of cos ’0 times
McCafferty (1989) and the amplitude-of-horizontal-gradient
the x-derivative and sin ’0 times the y-derivative. Directions are
method of Cordell and Grauch (1985).
preserved on Fourier transformation: since the wave vector k
has components p u ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and v in the x- and y-directions, respectively,
its magnitude is u2 þ v2 and its phase is
5.06.5.4.7 Euler deconvolution
Euler’s homogeneity equation can be written as (e.g., Reid
v y et al., 1990)
tan ’0 ¼ ¼ [35]
u x
the latter by definition of ’0. Thus, the directional derivative ðxx0 Þ@DT=@xþðyy0 Þ@DT=@yþðzz0 Þ@DT=@z¼s1 DT [39]
enhances features making an angle ’0 with the x-axis; to sup-
press them, but still improve the definition of smaller-scale Here, (x,y,z) is the position at which the total field anomaly is
features, one could take the directional derivative in the △T, arising from a source at position (x0,y0,z0). si is the ‘struc-
orthogonal direction. tural index’ (SI), a measure of the rate of falloff of the field
with distance, which therefore reflects the source geometry
(Thompson, 1982). An SI of 3 corresponds to a point source
5.06.5.4.5 Pie-crust filter (dipole), while an SI of 2 is appropriate for extended line
An alternative method of eliminating anomalies in a particular sources, such as pipes and cylinders, a value of 1 for a step,
direction is to use the pie-crust filter or operator, given by thin dike, or sill edge, and values of 0.5 and 0 have been chosen
8 for faults and other contacts. If si is set to zero, the right-hand
< 0 ð’0 D’ < ’ < ’0 þ D’Þ
side of eqn [39] should be replaced by a constant, which
W ð’Þ ¼ 0 ð’0 D’ < ’ þ p < ’0 þ D’Þ [36]
: depends on the amplitude of the contrast in magnetization
1 ðotherwiseÞ
and the strike and dip of the contact (Reid et al., 1990).
This removes anomalies within an angle D’ either side of Euler deconvolution (which is not strictly a deconvolution)
’0 and preserves those at all other angles, without altering the has proved itself particularly useful in a regional context, when
frequency content. Anomalies within this wedge can be pre- grids or profiles of data can be inverted systematically for
served by swapping the values 0 and 1 in eqn [36]. A more source parameters and the background field (or the constant
sophisticated filter (with a taper between 0 and 1) would on the right-hand side of eqn [39] when si ¼ 0). As eqn [39]
reduce ringing when the filtered anomalies are transformed shows, the method requires derivatives of △T, which are usu-
back to the spatial domain. ally calculated in the wave number domain. These are treated
as data (along with the SI) in an inversion for the source averaged to account for nonvertical contacts and noise. Tilt-
positions and background field (or right-hand side constant). depth profiles for RTE data are the mirror image of their RTP
Obtaining useful solutions depends on careful choice of the equivalents (i.e., yRTE ¼ yRTP) so the same averaging provides
window size (i.e., how many adjacent points are included in a the depth estimate in this case (Fairhead et al., 2011). How-
single least-squares-type inversion). Each window produces ever, whereas tilt angles calculated from RTP data are azimuth-
a set of source parameters, and only those with standard devi- ally symmetric, their RTE equivalents are more complex,
ations below a specified threshold are retained. Even these making interpretation more difficult. Fairhead et al. (2011)
often plot as quasilinear features (‘strings of pearls’), that is, caution against relying on zero tilt-depth contours of RTE
the solutions tend to be defocused. This may be partly because anomalies to define the boundaries of structures or suscepti-
a region often includes sources represented by more than one bility contrasts.
SI and attempts have been made to develop a multiple-source
approach. Alternatively, it is possible to focus the solutions and
determine appropriate structural indices, based on methods 5.06.5.5 Forward and Inverse Methods
that apply similarity transforms to Euler’s equation (e.g., Transformations assist in the characterization of certain fea-
Gerovska et al., 2010). Other approaches include methods of tures of the magnetic source, thereby facilitating interpretation.
determining automatically, or also solving for, SI, for example, Forward and inverse methods take this characterization a
wavelet-based methods for estimating degree of homogeneity step further, determining attributes of the magnetic source.
(Sailhac et al., 2000). The usual methods of damping and Forward methods begin with one or more magnetic bodies
generalized inverses are useful in improving the performance whose salient features are selected a priori, on the basis of
of the method and interpretation of the results (e.g., geologic or geophysical knowledge. Magnetic fields are then
Mushayandebvu et al., 2004; Whaler et al., 1991), but damp- predicted for these bodies at the survey location, and model
ing tends to bias depth estimates. parameters are adjusted on the basis of the closeness of the fit
Several authors have pointed out that Euler deconvolution to the observation. This process continues until a sufficiently
can be applied to any homogeneous field or function. This close fit to the observations is achieved. Inverse methods, in
includes the horizontal gradient or analytic signal of a field contrast, allow for the direct determination of one or more
that is itself homogeneous, for example, the magnetic field or attributes of the magnetic source, usually through least-squares
its Hilbert transform; the appropriate SI for the horizontal or Fourier transform techniques.
gradient or analytic signal is then one larger than that of the
original field source. The advantage of deconvolving the ana-
lytic signal rather than the magnetic field itself is that its 5.06.5.5.1 Forward models
calculation effectively removes the background field. Procedures for calculating magnetic forward models involve
simplification of complex bodies into simpler ones, as collec-
tions of either rectangular prisms (Bhattacharyya, 1964), mag-
5.06.5.4.8 Tilt-depth method
netic dipoles (Dyment and Arkani-Hamed, 1998), polygonal
Another popular interpretation technique providing both lat-
laminae (Plouff, 1976; Talwani, 1965), or polyhedrons (Bott,
eral and depth source information that depends only on first
1963). The calculation can be made in either space or wave
derivatives of the field is the tilt-depth method (Salem et al.,
number domains. Parker (1973) gave a wave number domain-
2008), a generalization of the local phase (Miller and Singh,
based algorithm for the rapid calculation of the crustal mag-
1994; Verduzco et al., 2004). The tilt angle, or tilt derivative, is
netic field over sources defined by a magnetization contrast
defined as
over a topographic surface.
!
@DT
y ¼ tan 1 @z
@DT
@H 5.06.5.5.2 Inverse approaches
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Quantitative interpretations sought using an inverse approach
@DT
@DT 2 @DT 2 aim to estimate the causative body’s depth, dimension, and
where @H ¼ @x þ @y denotes the horizontal deriva-
magnetization contrast. In many applications, depth to the
tive. It is independent of field strength and thus acts as an magnetic source is the most important of these properties.
unbiased automatic gain control filter, preserving equally Depth-to-source determinations are of two types, (1) based
long- and short-wavelength anomalies. It is also indepen- on the shape of individual anomalies (beginning with Peters,
dent of susceptibility contrast between the source and its 1949) and (2) based on the statistical properties of ensembles
surroundings, but this (and strike) can be obtained by com- of anomalies (beginning with Spector and Grant, 1970) and
bined with, for example, the analytic signal. When applied to implemented in the spectral domain.
RTP anomaly fields to remove the inclination dependency, The first analytic approximation to determining the depth
the tilt derivative vanishes over the edge of the source, and to source was given by Werner (1955), who solved the problem
the depth can then be obtained from under the assumption that the source was a thin dike. Subse-
quent work has relaxed that limitation and allowed for other
h
y ¼ tan 1 source geometries (Ku and Sharp, 1983). The exploitation of
z
Euler’s homogeneity relation [39] led to a second class of
where h is the distance from the contact (assumed vertical) and analytic approximations (Reid et al., 1990). This approxima-
z is its depth. Thus, for instance, y is 45 at horizontal tion allows for a variety of sources to be treated successfully,
distances z from the source, and the two estimates can be as outlined in Section 5.06.5.4.7.
Wave number domain approaches to individual anomalies and Dittmer and Szymanski (1995) have applied simulated
include the methods of Naudy (1971), applicable to a vertical annealing to magnetic and resistivity data.
dike or thin plate and COMPUDEPTH (O’Brien, 1973).
5.06.6 Spectral Overlap with Other Fields

5.06.5.5.2.1 Fourier domain approaches to groups of anomalies
Matched filters (Phillips, 1997; Syberg, 1972) use the Fourier
The transition from core-dominated to crust-dominated pro-
domain properties (Spector, 1968; Spector and Grant, 1970)
cesses occurs as a relatively sharp break centered at SH degree
of the magnetic field to estimate the depths of the principal
14 (Alldredge et al., 1963; Cain et al., 1974; Langel and Estes,
sources. These depths are then used to design wavelength
1982) corresponding to wavelengths of 40 000/14 ¼ 2860 km.
filters, which are in turn used to decompose observed magnetic
This transition can be seen in a power spectrum of the static
anomalies into estimates of the anomalies caused by sources at
field (Figure 8). The crustal field is much stronger over the
those principal depths. The original Spector and Grant (1970)
continents than over the oceans (Arkani-Hamed and
method estimated the depth from the slope of the radially
Strangway, 1986; Beggan et al., 2013; Hinze et al., 1991), and
averaged power spectrum.
so we expect that the spectral overlap with the core field will be
different for continental than for oceanic crust. The crustal field
is expected to have power at wavelengths longer than degree
5.06.5.6 Resolving Interpretational Ambiguity
14 because of the markedly different characteristics of conti-
Some of the ways for resolving or better understanding inter- nental and oceanic crust (Meissner, 1986) and because of
pretational ambiguity include annihilators (Gubbins et al., long-wavelength oceanic magnetic anomalies (Dyment and
2011; Maus and Haak, 2003; Runcorn, 1975), ideal body Arkani-Hamed, 1999). These longest wavelengths are masked
analysis (Parker, 2003), Monte Carlo simulations (Sambridge by the dominant core fields (Hahn and Bosum, 1986), and
and Mosegaard, 2002), and, of course, the use of prior various forward (Cohen, 1989) and inverse (Purucker et al.,
information. 1998, 2002) approaches have been developed in an attempt
It has long been known that an infinite sheet with constant to include at least some notional idea of these fields. The
magnetization produces no magnetic field outside of the longest-wavelength crustal magnetic fields remain inaccessi-
sheet, although second-order effects usually ensure that some ble to direct observation, although Mayhew and Estes (1983)
magnetic fields escape. Only magnetization contrasts produce suggested that simultaneous modeling of core and crustal
magnetic fields. Runcorn (1975) demonstrated, in the case of fields using ES dipole arrays located within both the core
the moon, that a spherical shell of constant susceptibility and the crust might make possible the separation of the
linearly magnetized by an arbitrary internal field also pro- sources. The task they outlined is now computationally feasi-
duces no field outside of the shell. More recently, Maus and ble (Purucker et al., 1996), but there are reasons for suspect-
Haak (2003) illustrated another class of magnetization solu- ing that a full separation may not be possible. For example, it
tions that produce no external fields. Their example, based on is possible to represent magnetic fields of long wavelength
reasoning from SH is one defined by a magnetic susceptibility (say the dimension of a continent) by ES dipoles placed
profile in a dipolar field that is symmetric about the magnetic either at the surface or at the core–mantle boundary.
equator. In the case of the Earth, South America and Africa are A separation based solely on the radial position of the dipoles
approximately bisected by the magnetic dipole equator, and is thus likely to be ill-posed and depend on details of the
their shape can be approximated fairly well by an annihilator. parameterization, such as the tessellation used, its spacing,
Thus, if these continents possess a large-scale magnetic con- and the distance over which the observations are expected to
trast with the surrounding ocean, much of that contrast may influence the crustal dipoles. The debate over the existence of
be invisible. Further inquiry (Gubbins et al., 2011) has field-aligned current systems (Dessler, 1986) shows the diffi-
revealed additional information on annihilators. culty of interpreting physically an equivalent current/dipole
The theory of ideal bodies (Parker, 1974, 1975) systema- system deduced solely from observations of the magnetic
tizes the process of placing bounds on the parameters of the field.
source region, such as the depth of burial (Grant and West, On the other hand, coestimation of magnetic fields of
1965). The process for doing this involves the minimization of internal and external origin through the ‘comprehensive
the infinity norm of the magnetic intensity |M| within the model’ approach (Olsen et al., 2006a; Sabaka et al., 2002,
source region. Parker (2003) showed how such a process 2004) has been successful because of the different decay char-
could be used to determine the distribution of magnetization acteristics of the internal and external fields (Langel et al.,
that has the smallest possible intensity, without any assump- 1996). This approach utilizes magnetic field measurements
tions about its direction. from satellites (ionosphere, crust, and core are internal) and
The Metropolis algorithm and the Gibbs sampler the surface (crust and core are internal) and does not separate
(Mosegaard and Sambridge, 2002) are Monte Carlo techniques the internal fields of core and crustal origin, lumping them
for the exploration of the space of feasible solutions. They also instead into a SH series from SH degrees 1 to 60. These ‘-
provide measures of resolution and uncertainty. Although not comprehensive’ models are now widely touted (Nabighian
widely used in the field of crustal magnetism, the Metropolis et al., 2005) to replace the existing IGRF models in many
algorithm has been used by Rygaard-Hjalsted et al. (2000) to applications.
conduct resolution studies on fluid flow in the Earth’s outer Another, qualitative, approach to separation is to character-
core from geomagnetic field observations. Monte Carlo tech- ize visually the field and magnetization solutions deduced
niques can also be used to find globally optimal solutions, from that field, in the hope that features at small scales will
109
108 Earth: all internal sources
107
3000 2000 1300 990 800 660
6
10 km (Earth)
105
Rn (nT2)
104
103
102 Inflection point

Earth: induced magnetization
101
Earth: remanent magnetization
100
10-1
0 10 20 30 40 50 60
Degree n
Figure 8 Comparison of the Lowes–Mauersberger (Rn) spectra at the surface of the Earth for a variety of internal fields. The inflection point in the
terrestrial power spectra represents the sharp transition from core processes at low n to lithospheric processes at higher n. Rn is the mean square
amplitude of the magnetic field over a sphere produced by harmonics of degree n. The spectrum of all internal sources comes from Sabaka et al. (2004),
the induced spectrum is derived from Fox Maule et al. (2005), and the remanent magnetization spectrum (of the oceans, and hence a minimum
value) was derived from Dyment and Arkani-Hamed (1998).
provide clues into what is happening at the largest scales. ‘C’ pattern, and regions with magnetization less than
Because of the wider spectral content of recent satellite crustal 0.06 A m1 (gray regions) correspond to the ‘O’ pattern.
field models, Purucker and Whaler (2004) were able to recog- Intermediate values of magnetization (between 0.06 and
nize two patterns in the vertical component map (Figure 9) of 0.1 A m1, pink on Figure 10(a)) generally envelop regions
the crustal field of the North American region. displaying the ‘C’ pattern. In a general way, the ‘C’ and ‘O’
The first pattern, which they refer to as ‘C,’ encompasses the patterns correspond to regions of thick and thin magnetic
North American land mass, the Caribbean and Gulf of Mexico, crustal thickness, as defined by temperature and seismology
and northernmost South America. The peak-to-trough magni- in the 3SMAC model (Nataf and Ricard, 1996). There are
tude of anomalies in ‘C’ typically exceeds 50 nT, and the conspicuous exceptions to this generalization. Most of the
anomalies are either equidimensional or oriented in a direc- South American landmass south of the equator is characterized
tion subparallel to the nearest coastline or tectonic element. by the ‘O’ pattern, yet crustal thickness is typical of continental
The second pattern, which they refer to as ‘O,’ encompasses the crust. The other major exception is the Sahara desert, again
eastern Pacific, the Cocos plate, and the western Atlantic away characterized by the ‘O’ pattern but with typical continental
from continental North America. The peak-to-trough magni- crustal thickness. In both of these regions, seismic crustal
tude of anomalies in ‘O’ is typically less than 30 nT, and the thickness and heat flow are poorly known.
anomalies are commonly narrow and elongate in the direction Electrical conductivity contrasts (Grammatica and Tarits,
of the nearest spreading or subduction zone. The ‘C’ pattern 2002) and motional induction by ocean currents (Vivier et al.,
can also be discerned on global maps of the field, when 2004) can also produce quasistatic magnetic fields that overlap
account is taken of the higher altitude. The ‘C’ pattern is with the crustal field. Specific time-variable features of these
characteristic of much of the Asian landmass, a region centered fields make separation possible, but their low amplitude makes
on but more extensive than Australia, and two broad regions it difficult in practice, and no separation has yet been achieved.
within the African landmass. The ‘O’ pattern is seen in the
eastern Pacific, the North and South Atlantic, and the
Indian oceans. The ‘C’ and ‘O’ patterns are also evident in a 5.06.7 Separation of Induced and Remanent
magnetization model based on these observations (Purucker Magnetizations
and Whaler, 2004). A map of |M| shows these patterns
best (Figure 10). Regions with magnetizations greater than Separation of induced and remanent magnetizations is a
0.1 A m1 (red regions on Figure 10(a)) correspond to the major outstanding issue. Induced magnetizations in the crust
40⬚
20⬚
0⬚
-140⬚ -120⬚ -100⬚ -80⬚ -60⬚
nT
-200 -100 -50 -20 -10 -5 5 10 20 50 100 200
Figure 9 CHAOS-4a vertical field map of the crustal magnetic field (Olsen et al., 2006a, 2010) over North America, evaluated at 50 km altitude.
0.00 0.06 0.10 1.00 0 10 32 58

(a) Magnitude fo M (A m-1) (b) Crustal thickness (km)
Figure 10 Map showing the magnitude (a) of the magnetization inferred from a model based on MF3 (reproduced from Maus S, et al. (2006) Earth’s
lithospheric magnetic field determined to SH degree 90 from CHAMP satellite measurements. Geophysical Journal International 164: 319–330),
compared with a map of the magnetic crustal thickness. (b) (reproduced from Nataf H and Ricard Y (1996) An a priori tomographic model of the upper
mantle based on geophysical modelling. Physics of the Earth and Planetary Interiors 95: 101–122)
represent one of the largest time-variable geomagnetic fields of is a continuous function, and viscous remanent magnetiza-
internal, noncore origin, and the separation of induced from tions exist at all timescales.
remanent magnetization remains an area of active research. Both the spatial variability of the static field as a function of
McLeod (1996) predicted that crustal-source secular variation inducing field and an approach using time-variable geomag-
should dominate core-source secular variation for SH degrees netic fields as a probe have been used to estimate induced
in excess of 22. The separation of induced from remanent magnetizations. Maus and Haak (2002) investigated the long-
magnetization can be done without ambiguity only if the wavelength power of the crustal field as a function of magnetic
source material is available for rock magnetic tests (cf. dip latitude. Two of the crustal magnetic field models (Cain
Chapter 5.08) and if the timescale is specified. Although mag- et al., 1989; Sabaka et al., 2002) they examined showed the
netizations are often considered either remanent or induced, strong trend with latitude expected for induced magnetization,
the distribution of magnetic coercivities in rocks while the third model (Arkani-Hamed et al., 1994) they
examined was consistent with a purely remanent magnetiza- Acknowledgments

tion. It may be relevant that the models that were consistent
with induced magnetization were models of both internal David Clark, Richard Gibson, Terence Sabaka, and an anony-
(core and crust) and external fields, while the model that was mous reviewer provided feedback through their edits on the
consistent with remanence was a crustal-only model that first edition of this chapter. We further thank Terence Sabaka
employed along-track filtering to remove noncrustal(?) long- for the discussions about the comprehensive model and Misac
wavelength trends. Maus and Haak (2002) also showed that, Nabighian for providing additional references. We appreciate
even in the case of a purely remanent magnetization, DT is the patience of our editor, Masaru Kono. All graphs and
expected to increase by a factor of 2 between the equator and maps have been plotted using the GMT software (Wessel and
the pole, while for a purely induced magnetization, DT is Smith, 1998).
expected to increase by a factor of 32/3 (almost 11) between
the equator and the pole. This is due to the preferential sam-
pling of the weaker tangential part of the crustal field at the
equator and the stronger radial part at the poles. References
Approaches using time-variable geomagnetic fields can be
either direct, using the change of the main geomagnetic field, Achache JA, Abtout A, and Le Mouël JL (1987) The downward continuation of MAGSAT
crustal anomaly field over Southeast Asia. Journal of Geophysical Research
or indirect, using an EM induction response (Clark et al., 1998;
92: 11584–11596.
Goldstein and Ward, 1966; Yanovsky, 1938) to time-variable Adams JC and Swarztrauber PN (1997) Spherepack 2.0: A model development facility.
fields like Sq or micropulsations. Lesur and Gubbins (2000) NCAR Technical Note NCAR/TN-436-STR.
took a direct approach, with a regional study of 20 long- Alldredge LR (1981) Rectangular harmonic analysis applied to the geomagnetic field.
running European observatories. They calculated the difference Journal of Geophysical Research 86: 3021–3026.
Alldredge LR (1982) Geomagnetic local and regional harmonic analysis. Journal of
between the observatory annual mean and the core field model Geophysical Research 87: 1921–1926.
of Bloxham and Jackson (1992). The variability with time of Alldredge LR, Vanvoorhis GD, and Davis TM (1963) A magnetic profile around the
that difference, or observatory bias, was then examined to see if world. Journal of Geophysical Research 68: 3679–3692.
it was consistent with induced or remanent magnetization. Andreasen GE and Zietz I (1969) Magnetic fields for a 46 prismatic model. U.S.
Geological Survey Professional Paper 666, Washington, DC.
They found that for nine observatories, a time-dependent
Arctic Gravity Project (2002) 50 50 grids compiled by the International Association of
induced field fitted the data better than a steady remanent Geodesy from airborne, survey, and submarine surveys, http://earth-info.nga.mil/
field at the 99% confidence level. The other observatories GandG/wgs84/agp/readme.html.
yielded ambiguous results. Some external field signatures Arkani-Hamed J and Strangway D (1986) Band-limited global scalar magnetic anomaly
remain in observatory annual mean data, and only local dis- map of the Earth derived from Magsat data. Journal of Geophysical Research
91: 8193–8203.
tributions of induced and remanent magnetizations were con- Arkani-Hamed J, Langel R, and Purucker M (1994) Scalar magnetic anomaly maps of
sidered in the analysis. A global approach to observatory biases Earth derived from POGO and Magsat data. Journal of Geophysical Research
by Mandea and Langlais (2002), for observatories operating 99: 24075–24090.
while the Magsat and rsted missions were in operation, while Aspler LB, Pilkington M, and Miles WF (2003) Interpretations of Precambrian basement
based on recent aeromagnetic data, Mackenzie Valley, Northwest Territories.
not attempting to isolate the remanent and induced compo-
Geological Survey of Canada, Current Research 2003-C2: 11.
nents, illustrates some of the other difficulties that the direct Aydin I, Karat HI, and Kocak A (2005) Curie-point depth map of Turkey. Geophysical
approach entails. In addition to contamination by external Journal International 162: 633–640.
fields, changes in observatory location or measurement prac- Bankey V, Cuevas A, Daniels D, et al. (2002) Digital data grids for the magnetic anomaly
tice add to the uncertainty. One can estimate the magnitude of map of North America, USGS Open File Report 02-414. http://pubs.usgs.gov/of/
2002/ofr-02-414; http://crustal.usgs.gov/geophysics.
the change in the observatory bias between the Magsat and Baranov V (1957) A new method for interpretation of aeromagnetic maps; pseudo-
rsted epochs, assuming that it is caused solely by changes in gravimetric anomalies. Geophysics 22: 359–383.
the induced magnetization. Values can be as large as 77 nT, Baranov V and Naudy H (1964) Numerical calculation of the formula of reduction to the
with 14 of the 62 observatories having predicted changes in magnetic pole. Geophysics 29: 67–79.
Barbosa VCF, Silva JBC, and Medeiros WE (1999) Stability analysis and
excess of 5 nT. Such a large signal over only 20 years should be
improvement of structural index estimation in Euler deconvolution. Geophysics
measurable, assuming that it can be isolated. Although 64: 48–60.
Mandea and Langlais (2002) determined the biases a posteriori Barritt SA, Fairhead JD, and Misener DJ (1993) The African magnetic mapping project.
the model generation, they could also be solved for in a model, ITC Journal: 121–131.
as, for example, in the comprehensive model of Sabaka et al. Bean RJ (1966) A rapid graphing solution for the aeromagnetic anomaly of a two-
dimensional tabular body. Geophysics 31: 963–970.
(2002). Beggan CD, Saarimäki J, Whaler KA, and Simons FJ (2013) Spectral and spatial
The separation of induced and remanent magnetizations decomposition of lithospheric magnetic field models using spherical Slepian
can also be attempted from the spatial distribution of DT functions. Geophysical Journal International 193: 136–148. http://dx.doi.org/
over isolated magnetic bodies (Zietz and Andreasen, 1967). 10.1093/gji/ggs122.
Bhattacharyya BK (1964) Magnetic anomalies due to prism-shaped bodies with arbitrary
Zietz and Andreasen (1967) used the position and relative
polarization. Geophysics 29: 517–531.
intensity of the maximum and minimum anomaly to infer Bhattacharyya BK (1965) Two-dimensional harmonic analysis as a tool for magnetic
the declination and inclination of the magnetization vector interpretation. Geophysics 30: 829–857.
in the causative body. Schnetzler and Taylor (1984) evaluated Bhattacharyya BK (1966) Continuous spectrum of the total-magnetic-field anomaly due
this technique globally and found that the method was more to a rectangular prismatic body. Geophysics 31: 97–121.
Bhattacharyya BK (1969) Bicubic spline interpolation as a method for treatment of
sensitive at higher magnetic latitudes and that due to zero-level potential field data. Geophysics 34: 402–423.
uncertainties, inclination was very difficult to estimate suscep- Bhattacharyya BK (1980) A generalized multibody model for inversion of magnetic
tibility variations. anomalies. Geophysics 45: 255–270.
Bhattacharyya BK and Chan KC (1977a) Reduction of magnetic and gravity data on an Clark DA (1999) Magnetic petrology of igneous intrusions: Implications for exploration
arbitrary surface acquired in a region of high topographic relief. Geophysics and magnetic interpretation. Exploration Geophysics 30: 5–26.
42: 1411–1430. Clark SC, Frey H, and Thomas HH (1985) Satellite magnetic anomalies over subduction
Bhattacharyya BK and Chan KC (1977b) Computation of gravity and magnetic zones. Geophysical Research Letters 12: 41–44.
anomalies due to inhomogeneous distribution of magnetization and density in a Clark DA, et al. (1998) Remote determination of magnetic properties and improved drill
localized region. Geophysics 42: 602–609. targeting of magnetic anomaly sources by differential vector magnetometry.
Blakely RJ (1981) A program for rapidly computing the magnetic anomaly over digital Exploration Geophysics 29: 312–319.
topography. U.S. Geological Survey Open-File Report 81–298: 46 pp. Cohen Y (1989) Traitements et interpretations de donnees spatiales in geomagnetisme:
Blakely RJ (1988) Curie-temperature isotherm analysis and tectonic implications of etude des variations laterales d’aimantation de la lithosphere terrestre. Docteur es
aeromagnetic data from Nevada. Journal of Geophysical Research sciences physiques thèse, Paris, p. 95.
93: 11817–11832. Coleman R (1992) Project magnet high-level vector survey data reduction. In: Langel R
Blakely RJ (1995) Potential Theory in Gravity and Magnetic Applications. New York: and Baldwin R (eds.) Types and Characteristics of Data for Geomagnetic Field
Cambridge University Press. Modelling, 215–248, NASA Conference Publication 3153.
Blakely RJ and Connard GG (1989) Crustal studies using magnetic data. In: Pakiser LC Constable CG and Parker RL (1988) Smoothing, splines and smoothing splines.
and Mooney WD (eds.) Geophysical Framework of the Continental United States. Journal of Computational Physics 78: 493–508.
Geological Society of America Memoir, vol. 172, pp. 45–60. Boulder, CO: Cooper GRJ (2003) Feature detection using sun shading. Computers and Geosciences
Geological Society of America (Chapter 4). 29: 941–948.
Blakely RJ and Grauch VJS (1983) Magnetic models of crystalline terrane: Accounting Cooper GRJ and Cowan DR (2003) Sunshading geophysical data using fractional order
for the effect of topography. Geophysics 48: 1551–1557. horizontal gradients. The Leading Edge 22: 204.
Blakely RJ and Simpson RW (1986) Approximating edges of source bodies from Cordell L (1992) A scattered equivalent-source method for the interpolation and
magnetic or gravity anomalies. Geophysics 51: 1494–1498. gridding of potential-field data in three dimensions. Geophysics 57: 629–636.
Blakely RJ, Jachens RC, Simpson RW, and Couch RW (1985) Tectonic setting of the Cordell L and Grauch VJS (1985) Mapping basement magnetization zones from
southern Cascade Range as interpreted from its magnetic and gravity fields. aeromagnetic data in the San Juan basin, New Mexico. In: Hinze WJ (ed.) The Utility
Geological Society of America Bulletin 96: 43–48. of Regional Gravity and Magnetic Anomaly Maps, pp. 181–197. Tulsa, OK: Society
Blakely RJ, Wells RE, Yelin TS, Madin IP, and Beeson MH (1995) Tectonic setting of the of Exploration Geophysicists.
Portland-Vancouver area, Oregon and Washington: Constraints from low-altitude Cordell L and McCafferty AE (1989) A terracing operator for physical property mapping
aeromagnetic data. Geological Society of America Bulletin 107: 1051–1062. with potential field data. Geophysics 54: 621–634.
Blakely RJ, Wells RE, Tolan TL, Beeson MH, Trehu AM, and Liberty LM (2000) New Cordell L and Taylor PT (1971) Investigation of magnetization and density of a North
aeromagnetic data reveal large strike-slip faults in the northern Willamette Valley, Atlantic Seamount using Poisson’s Theorem. Geophysics 36: 919–937.
Oregon. Geological Society of America Bulletin 112: 1225–1233. Cowan DR and Cooper GRJ (2005) Enhancement of magnetic signatures of impact
Blakely RJ, Wells RE, Weaver CS, and Johnson SY (2002) Location, structure, and structures. In: Kenkmann T, Horz F, and Deutsch A (eds.) Large Meteorite Impacts.
seismicity of the Seattle fault zone, Washington: Evidence from aeromagnetic GSA Special Paper, vol. 384, pp. 51–65. Boulder, CO: Geological Society of
anomalies, geologic mapping, and seismic-reflection data. Geological Society of America.
America Bulletin 114: 169–177. Cowan DR and Cowan S (1993) Separation filtering applied to aeromagnetic data.
Blakely RJ, Brocher TM, and Wells RE (2005) Subduction zone magnetic anomalies and Exploration Geophysics 24: 429–436.
implications for hydrated forearc mantle. Geological Society of America Bulletin Craig M (1996) Analytic signals for multivariate data. Mathematical Geology
33: 445–448. 28: 315–329.
Bloxham J and Jackson A (1992) Time-dependent mapping of the magnetic field at the Criss RE and Champion DE (1984) Magnetic properties of granitic rocks from the
core–mantle boundary. Journal of Geophysical Research 97: 19537–19563. southern half of the Idaho Batholith: Influences of hydrothermal alteration and
Bosum W, Pucher R, and Roeser H (1985) Crustal anomalies and their causes. implications for aeromagnetic interpretations. Journal of Geophysical Research
In: Fuchs K and Soffel H (eds.) Geophysics of the Solid Earth, the Moon and the 89: 7061–7076.
Planets, vol. V2, subvolume B, pp. 74–99. Berlin: Springer (Chapter 4.2.1). Davies J, Mushayandebvu MF, and Smith R (2004) Magnetic detection and
Bott MHP (1963) Two methods applicable to computers for evaluating magnetic characterization of tertiary and quaternary buried channels. SEG Expanded Abstracts
anomalies due to finite three dimensional bodies. Geophysical Prospecting 23: 734.
11: 292–299. De Santis A (1991) Translated original spherical cap harmonic analysis. Geophysical
Bouligand C, Glen J, and Blakely R (2009) Mapping Curie temperature depth in the Journal International 106: 253–263.
western United States with a fractal model for crustal magnetization. Journal of Dentith M, Cowan D, and Tompkins L (2000) Enhancement of subtle features in
Geophysical Research 114: B11104. http://dx.doi.org/10.1029/2009JB006494. aeromagnetic data. Exploration Geophysics 31: 104–108.
Bucknam RC, Hemphill-Haley E, and Leopold EB (1992) Abrupt uplift within the past Dessler AJ (1986) The evolution of arguments regarding the existence of field-aligned
1700 years at southern Puget Sound. Science 258: 1611–1614. currents. In: Potemra T (ed.) Magnetospheric Currents. Geophysical Monograph,
Cady JW (1980) Calculation of gravity and magnetic anomalies of finite-length right vol. 28, pp. 22–28. Washington, DC: American Geophysical Union.
polygonal prisms. Geophysics 45: 1507–1512. Dimri VP (1998) Fractal behaviour and detectability limits of geophysical surveys.
Cain JC, et al. (1974) An N ¼ 22 model of the geomagnetic field. EOS. Transactions of Geophysics 63: 1943–1946.
the American Geophysical Union 56: 1108. Dittmer JK and Szymanski JE (1995) The stochastic inversion of magnetics and
Cain JC, et al. (1989) Derivation of a geomagnetic model to n ¼ 63. Geophysical resistivity data using the simulated annealing algorithm. Geophysical Prospecting
Journal 97: 431–441. 43(3): 397–416.
Campos-Enriquez JO, et al. (1996) Chicxulub – Subsurface structure of impact crater Driscoll JR and Healy DM (1994) Computing Fourier transforms and convolutions on
inferred from gravity and magnetic data. The Leading Edge 15: 357–359. the 2-sphere. Advances in Applied Mathematics 15: 202–250.
Caress DW and Parker RL (1989) Spectral interpolation and downward continuation of Dyment J and Arkani-Hamed J (1998) Equivalent source magnetic dipoles revisited.
marine magnetic anomaly data. Journal of Geophysical Research 94: 17393–17407. Geophysical Research Letters 25: 2003–2006.
Chavez Gomez S (2001) A Catalogue of Dykes from Aeromagnetic Surveys in Eastern Dyment J and Arkani-Hamed J (1999) Contribution of lithospheric remanent
and Southern Africa. Enschede: ITC, Publication No. 80. magnetization to satellite magnetic anomalies over the world’s oceans. Journal of
Chenot D and Debeglia N (1990) Three-dimensional gravity or magnetic constrained Geophysical Research 103: 15423–15441.
depth inversion with lateral and vertical variation of contrast. Geophysics 55: 327–335. Dyment J, Hamoudi M, Choi Y, et al. (2012) Equivalent magnetization over the World
Chinese National Aerogeophysics Survey and Remote Sensing Center (1989) Ocean. In: EOS Fall Meeting Abstracts, T42B-01.
Aeromagnetic Anomaly Map of China and Adjacent Sea. Beijing: Geological Eaton D and Vasudevan K (2004) Skeletonization of aeromagnetic data. Geophysics
Publishing House (2nd digital edition in 2004). 69: 478–488.
Chulick GS and Mooney WD (2002) Seismic structure of the crust and uppermost Elkins-Tanton L and Bowring S (2006) Report on the workshop on the Siberian flood
mantle of North America and adjacent oceanic basins: A synthesis. Bulletin of the basalts and the end-Permian extinction. NSF-Sponsored Workshop, September
Seismological Society of America 92: 2478–2492. 2005.
Clark DA (1997) Magnetic petrophysics and magnetic petrology: Aids to geological Ernst RE and Buchan KL (1997) Giant radiating dyke swarms: Their use in identifying
interpretation of magnetic surveys. AGSO Journal of Australian Geology and Pre-Mesozoic large igneous provinces and mantle plumes. In: Mahoney JJ and
Geophysics 17: 83–103. Coffin MF (eds.) Large Igneous Provinces: Continental, Oceanic, and Planetary
Flood Volcanism. Geophysical Monograph, vol. 100, pp. 297–333. Washington, Fault Traps. AAPG Treatise of Petroleum Geology, Atlas of Oil and Gas Fields,
DC: American Geophysical Union. A-109: pp. 309–335. Tulsa, OK: AAPG.
Ernst RE and Buchan KL (2001) The use of mafic dike swarms in identifying and Grammatica N and Tarits P (2002) Contribution at satellite altitude of
locating mantle plumes. In: Ernst RE and Buchan KL (eds.) Mantle Plumes: Their electromagnetically induced anomalies from a three-dimensional heterogeneously
Identification Through Time, pp. 247–265. Boulder, CO: Geological Society of conducting Earth, using Sq as an inducing source field. Geophysical Journal
America (Special Paper 352). International 151: 913–923.
Ernst RE, et al. (1996) Diabase (dolerite) dike swarms of the world: First edition, Grant FS (1985) Aeromagnetics, geology and ore environments. I. Magnetite in
1:35,000,000 map, 104 pp. Geol. Surv. Can. Open File 3241. igneous, sedimentary and metamorphic rocks: An overview. Geoexploration
Erwin DE (1994) The Permo-Triassic extinction. Nature 367: 231–236. 23: 303–333.
Fairhead JD, Salem A, Cascone L, Hammill M, Masterton S, and Samson E (2011) New Grant FS and West GF (1965) Interpretation Theory in Applied Geophysics. New York:
developments of the magnetic tilt-depth method to improve structural mapping of McGraw-Hill.
sedimentary basins. Geophysical Prospecting 59: 1072–1086. Grauch VJS (1987) A new variable-magnetization terrain correction method for
Finn C (1990) Geophysical constraints on Washington convergent margin structure. aeromagnetic data. Geophysics 52: 94–107.
Journal of Geophysical Research 95: 19533–19546. Grieve R and Therriault A (2000) Vredefort, sudbury, chicxulub: Three of a kind? Annual
Fox Maule C, Purucker M, Olsen N, and Mosegaard K (2005) Heat flux anomalies in Review of Earth and Planetary Sciences 28: 305–338.
Antarctica revealed by satellite magnetic data. Science 309: 464–467. Grieve R, Reimold W, Morgan J, Riller U, and Pilkington M (2008) Meteoritics and
Fox Maule C, Purucker M, and Olsen N (2009) Inferring magnetic crustal thickness and Planetary Science 43(5): 855–882.
geothermal heat flux from crustal magnetic field models, Danish Climate Centre Gubbins D and Herrero-Bervera E (eds.) (2007) Encyclopedia of Geomagnetism and
Report 09-09, Danish Meteorological Institute, 33 pp. Paleomagnetism, 1054 pp. Berlin: Springer.
Friis-Christensen E, Lühr H, Hulot G, Haagmans R, and Purucker M (2009) Gubbins D, Ivers D, Masterton SM, and Winch DE (2011) Analysis of lithospheric
Geomagnetic research from space. EOS. Transactions of the American Geophysical magnetization in vector spherical harmonics. Geophysical Journal International
Union 90(25): 213–214. http://dx.doi.org/10.1111/j.1365-246X.2011.5153.x.
Frost BR (1991a) Magnetic petrology: Factors that control the occurrence of magnetite Gunn PJ (1975) Linear transformations of gravity and magnetic fields. Geophysical
in crustal rocks. In: Lindsley DH (ed.) Oxide Minerals: Petrologic and Magnetic Prospecting 23: 300–312.
Significance, vol. 25, pp. 489–509. Blacksburg, VA: Mineralogical Society of Airborne magnetic and radiometric surveys. Gunn PJ (ed.) (1997) AGSO Journal of
America (Chapter 14). Australian Geology and Geophysics 17: (No. 2).
Frost BR (1991b) Stability of oxide minerals in metamorphic rocks. In: Lindsley DH Haggerty SE (1976) Opaque mineral oxides in terrestrial igneous rocks. In: Rumble D
(ed.) Oxide Minerals: Petrologic and Magnetic Significance, vol. 25, pp. 469–488. (ed.) Oxide Minerals, vol. 3, pp. 1–98. Blacksburg, VA: Mineralogical Society of
Blacksburg, VA: Mineralogical Society of America (Chapter 13). America (Chapter 4).
Frost BR (1991c) Introduction to oxygen fugacity and its petrologic importance. Haggerty SE (1979) The aeromagnetic mineralogy of igneous rocks. Canadian Journal
In: Lindsley DH (ed.) Oxide Minerals: Petrologic and Magnetic Significance, vol. 25, of Earth Sciences 16: 1281–1293.
pp. 1–8. Blacksburg, VA: Mineralogical Society of America (Chapter 1). Hahn A and Bosum W (1986) Geomagnetics: Selected Examples and Case Histories.
Frost BR and Shive PN (1989) Comment on “Limiting Depth of Magnetization in Berlin: Geopublication Associates.
Cratonic Lithosphere” Geophysical Research Letters 16: 477–479. Hahn A, Ahrendt H, Meyer J, and Hufen JH (1984) A model of magnetic sources within
Gaina C, Werner SC, Saltus R, Maus S, and The CAMP-GM group (2011) Circum- the Earth’s crust compatible with the field measured by the satellite Magsat.
arctic mapping project: New magnetic and gravity anomaly maps of the arctic. Geological Jahrbuch A75: 125–156.
Geological Society of London’s Memoirs 35: 39–48. http://dx.doi.org/10.1144/ Haines GV (1985) Spherical cap harmonic analysis. Journal of Geophysical Research
M35.3. 90: 2583–2591. http://dx.doi.org/10.1029/JB090iB03p02583.
Geological Survey of Canada (1995) Magnetic anomalies and tectonic elements of Hanna WF (ed.) (1987) Geologic applications of modern aeromagnetic surveys.
Northeast Eurasia. GSC Open-File Report 2574. In: USGS Bulletin 1924.
Geological Survey of Japan (2002) Magnetic anomaly map of East Asia. Geological Hansen RO (1993) Interpretive gridding by anisotropic kriging. Geophysics
Survey of Japan Digital Geoscience Map P-3, 2nd edn. 58: 1491–1497.
Gerovska D, Araúzo-Bravo MJ, Whaler K, Stavrev P, and Reid A (2010) Three- Hansen RO (2005) 3D multiple-source Werner deconvolution for magnetic data.
dimensional interpretation of magnetic and gravity anomalies using the finite Geophysics 70: L45–L51.
difference similarity transform. Geophysics 75: L79–L90. Hansen RO and Miyazaki Y (1984) Continuation of potential fields between arbitrary
Getech (1996) http://www.getech.com/data. surfaces. Geophysics 49: 787–795.
Gibson RI (1998a) Interpretation of rift-stage faulting in the West Siberian Basin from Hansen RO and Pawlowski RS (1989) Reduction to the pole at low latitudes by Wiener
magnetic data. In: Gibson RI and Millegan PS (eds.) Geologic Applications of filtering. Geophysics 54: 1607–1613.
Gravity and Magnetics. Geophysical Reference Series, vol. 8, pp. 79–81. Tulsa, OK: Hansen RO and Simmonds M (1993) Multiple-source Werner deconvolution.
Society of Exploration Geophysicists. Geophysics 58: 1792–1800.
Gibson RI (1998b) Gravity and magnetics in oil exploration: A historical perspective. Hansen RO and Suciu L (2002) Multiple-source Euler deconvolution. Geophysics
Geologic Applications of Gravity and Magnetics: Case Histories. Geophysical 67: 525–535.
Reference Series, vol. 8. Tulsa, OK: Society of Exploration Geophysicists. Hansen RO and Wang X (1988) Simplified frequency-domain expressions for
Gibson RI and Millegan PS (eds.) (1998) Geologic Applications of Gravity and potential fields of arbitrary three-dimensional bodies. Geophysics 53: 365–374.
Magnetics: Case Histories. In: Geophysical Reference Series, vol. 8. Tulsa, OK: Harrison C (1987) The crustal field. In: Jacobs J (ed.) Geomagnetism, vol. 1,
Society of Exploration Geophysicists. pp. 513–610. London: Academic Press (Chapter 5).
Girdler RW, Taylor PT, and Frawley JJ (1992) A possible impact origin for the Bangui Heirtzler JR, Dickson GO, Herron EM, Pitman WC, and Le Pichon X (1968) Marine
magnetic anomaly (Central Africa). Tectonophysics 212: 45–58. magnetic anomalies, geomagnetic field reversals, and motions of the ocean floor
Goldhaber MB and Reynolds RL (1991) Relations among hydrocarbon reservoirs, and continents. Journal of Geophysical Research 73: 2119–2136.
epigenetic sulfidization, and rock magnetization: Examples from the south Texas Hemant K and Maus S (2005a) Geological modeling of the new CHAMP magnetic
coastal plain. Geophysics 56: 748–757. anomaly maps using a Geographical Information System technique. Journal of
Goldstein NE and Ward SH (1966) The separation of remanent from induced magnetism Geophysical Research 110: B12. http://dx.doi.org/10.1029/2005JB003837.
in situ. Geophysics 31: 779–796. Hemant K and Maus S (2005b) Why no anomaly is visible over most of the continent-
Golynsky AM, et al. (2002) “ADMAP – Magnetic anomaly map of the Antarctic”, ocean boundary in the global crustal magnetic field. Physics of the Earth and
1:10 000 000 scale map. In: Morris PR and von Frese R (eds.) BAS (Misc.), Planetary Interiors 149: 321–333.
vol. 10. Cambridge: British Antarctic Survey. Hernandez I, et al. (2001) Aeromagnetic map of Mexico: An exploration approach for the
Golynsky AM, Bell R, Blankenship D, et al. (2013) Air and shipborne magnetic surveys new millennium – A progress report. Revista Geofisica Instituto Panamericano de
of the Antarctic into the 21st century. Tectonophysics 585: 3–12. Geografia e Historia 55: 33.
Goussev SA, Charters RA, Peirce JW, and Glenn WE (2003) The meter reader – Hildebrand AR, et al. (1991) Chicxulub crater: A possible Cretaceous–Tertiary boundary
Jackpine magnetic anomaly: Identification of a buried meteorite impact structure. impact crater on the Yucatan peninsula. Geology 19: 867–871.
The Leading Edge 22: 740–741. Hill PL (1991) Bibliographies and location maps of publications on aeromagnetic and
Grace JD and Hart GF (1990) Urengoy gas field – USSR West Siberian basin, Tyumen aeroradiometric surveys. U.S. Geological Survey Open-File Report 91-370A-F,
District. In: Beaumont EA and Foster NH (eds.) Structural Traps III, Tectonic Old and Reston, VA.
Hinze W (1979) Continental magnetic anomalies. Reviews of Geophysics and Space Korhonen JV, et al. (1993) One hundred seventy eight thousand petrophysical
Physics 17: 257–273. parameter determinations from the regional petrophysical programme. In: Autio S
Hinze WJ (ed.) (1985) The Utility of Regional Gravity and Magnetic Anomaly Maps. (ed.) Geological Survey of Finland, Current Research 1991–1992, Geological
Tulsa, OK: Society of Exploration Geophysicists. Survey of Finland. Special Paper 18: 137–141.
Hinze WJ and Zietz I (1985) The composite magnetic-anomaly map of the conterminous Korhonen J, et al. (2007) Magnetic anomaly map of the world, and associated DVD,
United States. In: Hinze WJ (ed.) The Utility of Regional Gravity and Magnetic Commission for the Geological Map of the World, UNESCO, Paris, France, Scale:
Anomaly Maps, pp. 1–24. Tulsa, OK: Society of Exploration Geophysicists. 1:50,000,000.
Hinze W, et al. (1988) Magnetic anomaly map of North America. The Leading Edge Korte M and Holme R (2003) Regularization of spherical cap harmonics. Geophysical
7: 19–21. Journal International 153: 253–262.
Hinze W, Von Frese R, and Ravat D (1991) Mean magnetic contrast between oceans and Koulomzine T, Lamontagne Y, and Nadeau A (1970) New methods for the direct
continents. Tectonophysics 192: 117–127. interpretation of magnetic anomalies caused by inclined dikes of infinite length.
Hinze W, Von Frese RB, and Saad AH (2013) Gravity and Magnetic Exploration: Geophysics 35: 812–830.
Principles, Practices, and Application. Cambridge: Cambridge University Press. Krajick K (2001) Barren Lands: An Epic Search for Diamonds in the North American
Hitzman MW, Oreskes N, and Einaudi MT (1992) Geological characteristics and tectonic Arctic. New York: Henry Hold.
setting of Proterozoic iron oxide (Cu–U–Au–REE) deposits. Precambrian Research Ku CC and Sharp JA (1983) Werner deconvolution for automated magnetic
58: 241–287. interpretation and its refinement using Marquardt inverse modeling. Geophysics
Hood PJ, McGrath PH, and Teskey DJ (1985) Evolution of Geological Survey of Canada 48: 754–774.
magnetic-anomaly maps: A Canadian perspective. In: Hinze WJ (ed.) The Utility of LaBrecque JL and Ghidella ME (1997) Depth of magnetic basement, and sediment
Regional Gravity and Magnetic Anomaly Maps, pp. 62–87. Tulsa, OK: Society of thickness estimates from aerogeophysical data over the western Weddell Basin.
Exploration Geophysicists. Journal of Geophysical Research 102: 7929–7945.
Hornby P, Boschetti F, and Horowitz FG (1999) Analysis of potential field data in the LaBrecque JL and Raymond CA (1985) Seafloor spreading anomalies in the MAGSAT
wavelet domain. Geophysical Journal International 137: 175–196. field of the North Atlantic. Journal of Geophysical Research 90: 2565–2575.
Hsu S-K, Sibuet J-C, and Shyu C-T (1996) High-resolution detection of geologic Langel R and Benson BV (1987) The Magsat bibliography. NASA Tech Memo. TM
boundaries from potential-field anomalies: An enhanced analytic signal technique. 87822.
Geophysics 61: 373–386. Langel RA and Estes RH (1982) A geomagnetic field spectrum. Geophysical Research
Hsu S-K, Coppens D, and Shyu C-T (1998) Depth to magnetic source using the Letters 9: 250–253.
generalized analytic signal. Geophysics 63: 1947–1957. Langel R and Hinze W (1998) The Magnetic Field of the Earth’s Lithosphere.
Hyndman RD and Peacock SM (2003) Serpentinization of the forearc mantle. Earth and Cambridge: Cambridge University Press.
Planetary Science Letters 212: 417–432. Langel RA and Whaler KA (1996) Maps of the magnetic anomaly field at Earth’s surface
Ishihara S (1981) The granitoid series and mineralization. Economic Geology: from scalar satellite data. Geophysical Research Letters 23: 41–44.
458–484, 75th Anniversary Issue. Langel RA, Slud VE, and Smith PJ (1984) Reduction of satellite magnetic anomaly data.
Ishihara T (2004) Application of CM3 model in compilation of marine magnetic Journal of Geophysics 54: 207–212.
anomaly data of North Pacific. EOS. Transactions of the American Geophysical Langel RA, et al. (1993) The equatorial electrojet and associated currents as seen in the
Union 85(47) (Fall Meet. Suppl. Abstract GP11D-0882). Magsat data. Journal of Atmospheric and Terrestrial Physics 55: 1233–1269.
Jackson DD (1972) Interpretation of inaccurate, insufficient and inconsistent data. Langel RA, et al. (1996) The near-Earth magnetic field from magnetospheric and quiet-
Geophysical Journal of the Royal Astronomical Society 28: 97–109. day ionospheric sources and how it is modelled. Physics of the Earth and Planetary
Jackson A (1990) Accounting for crustal magnetization in models of the core magnetic Interiors 98: 235–267.
field. Geophysical Journal International 103: 657–673. Langlais B, Purucker M, and Mandea M (2004) Crustal magnetic field of Mars.
Jackson A (1994) Statistical treatment of crustal magnetization. Geophysical Journal Journal of Geophysical Research 109: E02008. http://dx.doi.org/
International 119: 991–998. 10.1029/2003JE002048.
Jackson A (1996) Bounding the long wavelength crustal magnetic field. Physics of the Langlais B, Lesur V, Purucker M, Connerney J, and Mandea M (2009) Crustal magnetic
Earth and Planetary Interiors 98: 283–302. field of terrestrial planets. Space Science Reviews. http://dx.doi.org/10.1007/
Jansen J and Witherly K (2004) The Tli Kwi Cho Kimberlite Complex, Northwest s11214-009-9557-y.
Territories, Canada: A Geophysical Case Study. 74st Annual Meeting Technical Leaman DE (1994) Criteria for evaluation of potential field interpretations. First Break
Program (Ext. Abstr). 23. Denver, CO: Society of Exploration Geophysicists, pp. 12: 181–191.
1147–1149. LeSchack LA and Van Alstine DR (2002) High-resolution ground-magnetic (HRGM) and
Jessell MW (2001) Three-dimensional modeling of potential-field data. Computers & radiometric surveys for hydrocarbon exploration: Six case histories in Western
Geosciences 27: 455–465. Canada. In: Schumacher D and LeSchack LA (eds.) Surface Exploration Case
Jessell MW and Fractal Geophysics Pty Ltd (2002) An atlas of structural geophysics II. Histories: Applications of Geochemistry, Magnetics, and Remote Sensing. AAPG
Journal of the Virtual Explorer 5. Studies in Geology No. 48 and SEG Geophysical Reference Series No. 11,
Johnson SY, et al. (1996) The southern Whidbey Island fault: An active structure in the pp. 67–156. Tulsa, OK: AAPG.
Puget Lowland. Geological Society of America Bulletin 108: 334–354. Lesur V and Gubbins D (1999) Evaluation of fast spherical transforms for geophysical
Johnson SY, Dadisman SV, Mosher DC, Blakely RJ, and Childs JR (2001) Active applications. Geophysical Journal International 139: 547–555.
tectonics of the Devils Mountain fault and related structures, northern Puget Lesur V and Gubbins D (2000) Using geomagnetic secular variation to separate
Lowland and eastern Strait of Juan de Fuca region, Pacific Northwest. U.S. remanent and induced sources of the crustal magnetic field. Geophysical Journal
Geological Survey Professional Paper 1643, 45 pp, 2 plates. International 142: 889–897.
Jourdan F, Fraud G, Bertrand H, Watkeys MK, Kampunzu AB, and Le Gall B (2006) Lesur V and Maus S (2006) A global lithospheric magnetic field model with reduced
Basement control on dyke distribution in Large Igneous Provinces: Case study of noise in the Polar regions. Geophysical Research Letters 33. http://dx.doi.org/
the Karoo triple junction. Earth and Planetary Science Letters 241: 307–322. 10.1029/2006GL025826.
Kanasewich ER and Agarwal RG (1970) Analysis of combined gravity and magnetic Li X (2003) On the use of different methods for estimating magnetic depth. The Leading
fields in wavenumber domain. Journal of Geophysical Research 75: 5702–5712. Edge 22: 1090–1099.
Keating P (1993) The fractal dimension of gravity data sets and its implication for Li Y and Oldenburg DW (1996) 3-D inversion of magnetic data. Geophysics
gridding. Geophysical Prospecting 41: 983–994. 61: 394–408.
Keating P (1995) A simple technique to identify magnetic anomalies due to kimberlite Li Y and Oldenburg DW (2003) Fast inversion of large-scale magnetic data using
pipes. Exploration Mining Geology 4: 121–125. wavelet transforms and logarithmic barrier method. Geophysical Journal
Keating P and Sailhac P (2004) Use of the analytic signal to identify magnetic anomalies International 152: 251–265.
due to kimberlite pipes. Geophysics 69: 180–190. Lindsley DH (ed.) (1991) In: Oxide Minerals: Petrologic and Magnetic Significance,
Keating P and Zerbo L (1996) An improved technique for reduction-to-the-pole at low vol. 25. Blacksburg, VA: Mineralogical Society of America.
latitudes. Geophysics 61: 131–137. Littke R, et al. (1999) Gas generation and accumulation in the West Siberian Basin.
Keller GR (1988) The development of gravity and magnetic studies, emphasizing articles AAPG Bulletin 83: 1642–1665.
published in the GSA Bulletin. Geological Society of America Bulletin Lowe DAJ, Parker RL, Purucker M, Constable E, and Constable CG (2001) Estimating
100: 469–478. the crustal power spectrum from vector Magsat data. Journal of Geophysical
Kellogg OD (1953) Foundations of Potential Theory. New York: Dover. Research 106: 8589–8598.
Lühr H, Maus S, and Rother M (2004) Noon-time equatorial electrojet: Its spatial McEnroe SA, Robinson P, and Panish P (2001a) Aeromagnetic anomalies, magnetic
features as determined by the CHAMP satellite. Journal of Geophysical Research petrology and rock magnetism of hemo-ilmenite- and magnetite-rich cumulates
109: A01306. http://dx.doi.org/10.1029/2002JA009656. from the Sokndal Region, South Rogaland, Norway. American Mineralogist
Machel HG and Burton EA (1991) Chemical and microbial processes causing 86: 1447–1468.
anomalous magnetization in environments affected by hydrocarbon seepage. McEnroe SA, et al. (2001b) Effect of fine-scale microstructures in titanohematite on the
Geophysics 56: 598–605. acquisition and stability of natural remanent magnetization in granulite facies
Macnab R, et al. (1995) New database documents the magnetic character of the Arctic metamorphic rocks, southwest Sweden: Implications for crustal magnetism. Journal
and North Atlantic. EOS. Transactions of the American Geophysical Union 76(45): of Geophysical Research 106: 30523–30546.
449–458. McLeod MG (1996) Spatial and temporal power spectra of the geomagnetic field.
Macnae JC (1979) Kimberlites and exploration geophysics. Geophysics Journal of Geophysical Research 101: 2745–2763.
44: 1395–1416. Meissner R (1986) The Continental Crust: A Geophysical Approach. International
Makarova ZA (ed.) (1974) Map of the Anomalous Magnetic Field of the Territory of the Geophysical Services, vol. 34. San Diego, CA: Academic Press.
U.S.S.R. and Adjacent Marine Areas, Scale 1:2,500,000 (18 sheets). Leningrad: U.S. Miller HG and Singh V (1994) Potential field tilt: A new concept for location of potential
S.R. Ministry of Geology, VSEGEI. field sources. Journal of Applied Geophysics 32: 213–217.
Malin SRC, Duzgit Z, and Baydemir N (1996) Rectangular harmonic analysis revisited. Milligan PR and Franklin R (2004) Magnetic Anomaly Map of Australia. Canberra:
Journal of Geophysical Research 101: 28205–28209. Geoscience Australia, 1:25000000.
Mandea M and Langlais B (2002) Observatory crustal magnetic biases during MAGSAT Milligan PR, et al. (2005) Fourth edition Magnetic Anomaly Map of Australia, derived
and Orsted satellite missions. Geophysical Research Letters 29. http://dx.doi.org/ from a new-generation Magnetic Anomaly Grid Database of Australia (MAGDA).
10.1029/2001GL013693. In: International Association of Geomagnetism and Aeronomy, IAGA2005-A-01186,
Mandea M and Purucker M (2005) Observing, modeling, and interpreting magnetic Toulouse, France.
fields of the solid earth. Surveys in Geophysics 26: 415–459. http://dx.doi.org/ Milligan PI, Minty BRS, Richardson M, and Franklin R (2009) The Australia-wide
10.1007/s10712-005-3857-x. Airborne Geophysical Survey – Accurate continental magnetic coverage. Preview
Marsh JS (2005) Discussion: The geophysical mapping of Mesozoic dyke swarms in 138: 70.
southern Africa and their origin in the disruption of Gondwana. Journal of African Modisi MP, Atekwana EA, Kampunzu AB, and Ngwisanyi TH (2000) Rift kinematics
Earth Sciences 35: 525–527. during the incipient stages of continental extension: Evidence from the nascent
Masterton SM, Gubbins D, Muller RD, and Singh KH (2013) Forward modelling of Okavango rift basin, northwest Botswana. Geology 28: 939–942.
oceanic lithospheric magnetization. Geophysical Journal International Mooney WD, Laske G, and Masters TG (1998) Crust 5.1: A global crustal model at
192: 951–962. http://dx.doi.org/10.1093/gji/ggs063. 5 5. Journal of Geophysical Research 103: 727–748.
Maus S and Dimri VP (1996) Depth estimation from the scaling power spectrum of Moreau F, Gibert D, Holschneider M, and Saracco G (1997) Wavelet analysis of
potential fields? Geophysical Journal International 124: 113–120. potential fields. Inverse Problems 13: 165–178.
Maus S and Haak V (2002) Is the long wavelength crustal magnetic field dominated by Mosegaard K and Sambridge M (2002) Monte Carlo analysis of inverse problems.
induced or by remanent magnetisation? Journal of Indian Geophysical Union Inverse Problems 18: R29–R54.
6: 1–5. Mubu MS (1995) Aeromagnetic mapping and interpretation of mafic dyke swarms in
Maus S and Haak V (2003) Magnetic field annihilators: Invisible magnetisation at the southern Africa. MSc Thesis, ITC Delft, Netherlands, 63 pp.
magnetic equator. Geophysical Journal International 155: 509–513. Mushayandebvu MF, van Driel P, Reid AB, and Fairhead JD (2001) Magnetic source
Maus S and Macmillan S (2005) 10th generation international geomagnetic reference parameters of two-dimensional structures using extended Euler deconvolution.
field. EOS. Transactions of the American Geophysical Union 86: 159. Geophysics 66: 814–823.
Maus S, Gordon D, and Fairhead JD (1997) Curie temperature depth estimation Mushayandebvu MF, Lesur V, Reid AB, and Fairhead JD (2004) Grid Euler
using a self-similar magnetisation model. Geophysical Journal International deconvolution with constraints for 2D structures. Geophysics 69: 489–496.
129: 163–168. Nabighian MN (1972) The analytic signal of two-dimensional magnetic bodies with
Maus S, et al. (2006) Earth’s lithospheric magnetic field determined to spherical polygonal cross-section: Its properties and use for automated interpretation.
harmonic degree 90 from CHAMP satellite measurements. Geophysical Journal Geophysics 37: 507–517.
International 164: 319–330. Nabighian MN (1984) Toward a three-dimensional automatic interpretation of potential
Maus S, et al. (2007) Fifth-generation lithospheric magnetic field model from CHAMP field data via generalized Hilbert transforms: Fundamental relations. Geophysics
satellite measurements. Geochemistry, Geophysics, Geosystems 8. http://dx.doi. 49: 780–786.
org/10.1029/2006GC001521. Nabighian MN and Asten MW (2002) Metalliferous mining geophysics – State of the art
Maus S, Yin F, Luhr H, et al. (2008) Resolution of direction of oceanic magnetic in the last decade of the century and the beginning of the new millennium.
lineations by the sixth generation lithospheric magnetic field model from CHAMP Geophysics 67: 964–978.
satellite magnetic measurements. Geochemistry, Geophysics, Geosystems 9. http:// Nabighian MN and Hansen RO (2001) Unification of Euler and Werner deconvolution in
dx.doi.org/10.1029/2008GC001949. three dimensions via the generalized Hilbert transform. Geophysics 66: 1805–1810.
Maus S, Barckhausen U, Berkenbosch H, et al. (2009) EMAG2: A 2-arc min resolution Nabighian MN, et al. (2005) The historical development of the magnetic method in
Earth Magnetic Anomaly Grid compiled from satellite airborne and marine magnetic exploration. Geophysics 70: 33–61.
measurements. Geochemistry, Geophysics, Geosystems 10(8): 1–12. http://dx.doi. Naidu PS (1970) Fourier transform of large scale aeromagnetic field using a
org/10.1029/2009GC002471. modified version of Fast Fourier Transform. Pure and Applied Geophysics
May PR (1971) Pattern of Triassic–Jurassic diabase dikes around the North Atlantic in 81: 17–25.
context of predrift position of the continents. Geological Society of America Bulletin Nakagawa I and Yukutake T (1985) Rectangular harmonic analysis of geomagnetic
82: 1285–1292. anomalies derived from Magsat data over the area of the Japanese islands. Journal
Mayhew M (1982) Application of satellite magnetic anomaly data to Curie isotherm of Geomagnetism and Geoelectricity 37: 957–977.
mapping. Journal of Geophysical Research 87: 4846–4854. Nataf H and Ricard Y (1996) An a priori tomographic model of the upper mantle based
Mayhew M and Estes R (1983) Equivalent source modeling of the core on geophysical modelling. Physics of the Earth and Planetary Interiors
magnetic field using Magsat data. Journal of Geomagnetism and Geoelectricity 95: 101–122.
35: 119–130. Naudy H (1971) Automatic determination of depth on aeromagnetic profiles.
Mayhew M and Johnson BD (1987) An equivalent layer magnetization model for Geophysics 36: 717–722.
Australia based on Magsat data. Earth and Planetary Science Letters 83: 167–174. Nelson JB (1988) Calculation of the magnetic gradient tensor from total field gradient
Mayhew M and LaBrecque J (1987) Crustal geologic studies with Magsat and surface measurements and its application to geophysical interpretation. Geophysics
magnetic data. Reviews of Geophysics 25: 971–981. 53: 957–966.
Mayhew M, Johnson BD, and Wasilewski PJ (1985) A review of problems and progress NGDC (1996) Magnetic anomaly data of the former U.S.S.R., U.S. Naval Oceanographic
in studies of satellite magnetic anomalies. Journal of Geophysical Research Office, in collaboration with the Ministry of Geology of the U.S.S.R, 1 CD-ROM.
90: 2511–2522. Nicholas JB, Purucker ME, Johnson CL, et al. (2011) Magnetic fields of the solar
McEnroe SA and Brown LL (2000) A closer look at remanence-dominated anomalies: system: A comparative planetology toolkit. In: Fall AGU Meeting, GP21B-0992.
Rock-magnetic properties and magnetic mineralogy of the Russell Belt microcline- O’Brien DP (1973) CompuDepth – New method for depth-to-magnetic-basement
sillimanite gneisses, Northwest Adirondacks Mountains, New York. Journal of computation. Geophysics 38: 187, Soc. Expl. Geophy. 42nd Annual Mtg Tech Prog
Geophysical Research 105: 16437–16456. (Abstr), Anaheim, CA.
O’Connell MD, et al. (2005) Gridding aeromagnetic data using longitudinal and Pilkington M and Crossley DJ (1986) Determination of crustal interface topography
transverse gradients with the minimum curvature operator. The Leading Edge from potential fields. Geophysics 51: 1277–1284.
24: 142–145. Pilkington M and Grieve RAF (1992) The geophysical signature of terrestrial impact
O’Brien MS and Parker RL (1994) Regularized geomagnetic field modeling using craters. Reviews of Geophysics 30: 161–181.
monopoles. Geophysical Journal International 118: 566–578. Pilkington M and Hildebrand AR (2000) Three-dimensional magnetic imaging of the
Olsen N, Lühr H, Sabaka TJ, et al. (2006a) CHAOS-A model of Earth’s magnetic field Chicxulub crater. Journal of Geophysical Research 105: 23479–23491.
derived from CHAMP, Orsted, and SAC-C magnetic field data. Geophysical Journal Pilkington M and Keating P (2004) Contact mapping from gridded magnetic data –
International 166: 67–75. A comparison of techniques. Exploration Geophysics 35: 306–311.
Olsen N, et al. (2006b) The Swarm end-to-end mission simulator study: Pilkington M and Todoeschuck JP (1995) Scaling nature of crustal susceptibilities.
A demonstration of separating the various contributions to Earth’s magnetic field Geophysical Research Letters 22: 779–782.
using synthetic data. Earth, Planets and Space 58: 359–370. Plouff D (1976) Gravity and magnetic fields of polygonal prisms and application to
Olsen KB, Stephenson WJ, and Geisselmeyer A (2007) 3D crustal structure and long- magnetic terrain corrections. Geophysics 41: 727–741.
period ground motions from a M9.0 megathrust earthquake in the Pacific Northwest Power M, Belcourt G, and Rockel E (2004) Geophysical methods for kimberlite
region. Journal of Seismology 12: 145–159. http://dx.doi.org/10.1007/s10950- exploration in northern Canada. The Leading Edge 23: 1124–1129.
007-9082-y. Press WH, Teukolsky SA, Vetterling WT, and Flannery BP (1992) Numerical Recipes in
Olsen N, Luehr H, Sabaka T, Michaelis I, Rauberg J, and Toffner-Clausen L (2010) Fortran: The Art of Scientific Computing, 2nd edn. Cambridge: Cambridge Press.
CHAOS-4 – A high resolution geomagnetic field model derived from low-altitude Press WH, Teukolsky SA, Vetterling WT, and Flannery BP (1996) Numerical Recipes in
CHAMP data. In: AGU Fall Meeting 2010, San Francisco, Poster GP21A-0992. Fortran 90: The Art of Scientific Computing. Cambridge: Cambridge Press.
Parker RL (1972) Inverse theory with grossly inadequate data. Geophysical Journal of Press WH, Teukolsky SA, Vetterling WT, and Flannery BP (1997) Numerical Recipes in
the Royal Astronomical Society 29: 123–138. C: The Art of Scientific Computing, 2nd edn. Cambridge: Cambridge Press.
Parker RL (1973) The rapid calculation of potential anomalies. Geophysical Journal of Prieto C and Morton G (2003) New insights from a 3D earth model, deepwater Gulf of
the Royal Astronomical Society 31: 447–455. Mexico. The Leading Edge 22: 356–360.
Parker RL (1974) Best bounds on density and depth from gravity data. Geophysics Primdahl F (2000) Resonance magnetometers. In: Ripka P (ed.) Magnetic Sensors and
39: 644–649. Magnetometers, pp. 267–304. Boston, MA: Artech.
Parker RL (1975) The theory of ideal bodies for gravity interpretation. Geophysical Prodehl C and Mooney WD (2012) Exploring the Earth’s Crust: History and results of
Journal of the Royal Astronomical Society 42: 315–334. controlled source seismology. Geological Society of America Memoirs 208: 764.
Parker RL (1986) Harmonic splines in geomagnetism. In: Function Estimation. Purucker M (1990) The computation of vector magnetic anomalies: A comparison of
Proceedings of the American Mathematical Society – Contemporary Mathematics techniques and errors. Physics of the Earth and Planetary Interiors 62: 231–245.
59, pp. 63–76, Ann Arbor, MI. Purucker M and Clark D (2011) Interpretation and mapping of the lithospheric magnetic
Parker RL (1988) A statistical theory of seamount magnetism. Journal of Geophysical field. In: Mandea M and Korte M (eds.) Geomagnetic Observations and Models.
Research 93: 3105–3115. IAGA Special Sopron Book Series 5: pp. 311–338. Springer.
Parker RL (1991) A theory of ideal bodies for seamount magnetism. Journal of Purucker M and Ishihara T (2005) Magnetic images of the Sumatran region crust.
Geophysical Research 96: 16101–16112. Transactions of the American Geophysical Union 86(10): 101–102.
Parker RL (1997) Coherence of signals from magnetometers on parallel paths. Journal Purucker M and Mandea M (2005) New research directions based on the World
of Geophysical Research 102: 5111–5117. Digital Magnetic Anomaly Map, and Swarm. In: International Association of
Parker RL (2003) Ideal bodies for Mars magnetics. Journal of Geophysical Research Geomagnetism and Aeronomy International Meeting, Toulouse, France, 18–29 July.
108(E1): 5006. http://dx.doi.org/10.1029/2001JE001760. Purucker M and Whaler K (2004) Recognizing and interpreting the longest wavelength
Parker RL and Huestis SP (1974) Inversion of magnetic anomalies in the presence of lithospheric magnetic fields obscured by overlap with the core field. Fall Meeting
topography. Journal of Geophysical Research 79: 1587–1593. Suppl., Eos, Transactions American Geophysical Union 85: 47, Abstract GP31A-
Parker RL and Klitgord KD (1972) Magnetic upward continuation from an uneven track. 0821.
Geophysics 37: 662–668. Purucker M, Sabaka T, and Langel R (1996) Conjugate gradient analysis: A New tool for
Parker RL and O’Brien MS (1997) Spectral analysis of vector magnetic field profiles. studying satellite magnetic data sets. Geophysical Research Letters 23: 507–510.
Journal of Geophysical Research 102: 24815–24822. Purucker M, Langel R, Rajaram M, and Raymond C (1998) Global magnetization
Parker RL and Shure L (1982) Efficient modeling of the earth’s magnetic field with models with a priori information. Journal of Geophysical Research
harmonic splines. Geophysical Research Letters 9: 812–815. 103: 2563–2584.
Parker RL and Shure L (1985) Gravitational and magnetic fields of some simple solids Purucker M, Von Frese RRB, and Taylor PT (1999) Mapping and interpretation of
of revolution. Geophysical Journal of the Royal Astronomical Society 80: 631–647. satellite magnetic anomalies from POGO data over the Antarctic region. Annali Di
Parker RL, Shure L, and Hildebrand J (1987) The application of inverse theory to Geofisica 42: 215–228.
seamount magnetism. Reviews of Geophysics 25: 17–40. Purucker M, Langlais B, Olsen N, Hulot G, and Mandea M (2002) The southern edge of
Paterson NR and Reeves CV (1985) Applications of gravity and magnetic surveys: The cratonic North America: Evidence from new satellite magnetometer observations.
state-of-the-art in 1985. Geophysics 50: 2558–2594. Geophysical Research Letters 29(15): 8000. http://dx.doi.org/
Pedersen LB (1977) Interpretation of potential field data – A generalized inverse 10.1029/2001GL013645.
approach. Geophysical Prospecting 25: 199–230. Purucker ME, Sabaka T, Le G, Slavin JA, Strangeway RJ, and Busby C (2007) Magnetic
Pedersen LB (1978) Wavenumber domain expressions for potential fields from arbitrary field gradients from the ST-5 constellation: Improving magnetic and thermal models
2-, 2½-, and 3-dimensional bodies. Geophysics 43: 626–630. of the lithosphere. Geophysical Research Letters 34. http://dx.doi.org/
Pedersen LB (1979) Constrained inversion of potential field data. Geophysical 10.1029/2007GL031739.
Prospecting 27: 726–748. Purucker M, Head J, and Wilson L (2012) Magnetic signature of the lunar South Pole–
Pedersen LB (1991) Relations between potential fields and some equivalent sources. Aitken basin: Character, origin and age. Journal of Geophysical Research
Geophysics 56: 961–971. 117: E05001. http://dx.doi.org/10.1029/2011JE003922.
Pedersen LB and Rasmussen TM (1990) The gradient tensor of potential field Quesnel Y, Catalan M, and Ishihara T (2009) A new global marine magnetic anomaly
anomalies: Some implications on data collection and data processing of maps. data set. Journal of Geophysical Research 114: B04106. http://dx.doi.org/
Geophysics 55: 1558–1566. 10.1029/2008JB006144.
Penfield GT and Camargo ZA (1981) Definition of a major igneous zone in the central Rasmussen R and Pedersen LB (1979) End corrections in potential field modeling.
Yucatan with aeromagnetics and gravity. 51st Annual Meeting Technical Program Geophysical Prospecting 27: 749–760.
(Abstract), 37: Los Angeles, CA: Society of Exploration Geophysicists. Ravat D (1996) Analysis of the Euler method and its applicability in environmental
Peters LJ (1949) The direct approach to magnetic interpretation and its practical magnetic applications. Journal of Environmental Engineering and Geophysics
application. Geophysics 14: 290–320. 1: 229–238.
Phillips JD (1997) Potential-field geophysical software for the PC-version 2.2. U.S. Ravat D, Whaler K, Pilkington M, Sabaka T, and Purucker M (2002) Compatibility of
Geological Survey Open-File Report 97-725, 34 pp. high-altitude aeromagnetic and satellite altitude magnetic anomalies over Canada.
Phillips JD, Reynolds RL, and Frey H (1991) Crustal structure interpreted from magnetic Geophysics 67: 546–554.
anomalies. Reviews of Geophysics Supplement 1: 416–427. Ravat D, Hildenbrand TG, and Roest W (2003) New way of processing near-surface
Pilkington M (1997) 3-D magnetic imaging using conjugate gradients. Geophysics magnetic data: The utility of the comprehensive model of the magnetic field. The
62: 1132–1142. Leading Edge 22: 784–785.
Ravat D, Finn C, Hill P, et al. (2009) A preliminary, full spectrum, magnetic anomaly Salem A, et al. (2008) Interpretation of magnetic data using tilt-angle derivatives.
grid of the United States with improved long wavelengths for studying continental Geophysics 73. http://dx.doi.org/10.1190/1.2799992.
dynamics: A website for distribution of data, USGS OF09-1258. Saltus RW, et al. (2005) Utility of aeromagnetic studies for mapping of potentially active
Ray RD (1985) Correction of systematic error in magnetic surveys: An application of faults in two forearc basins: Puget Sound, Washington, and Cook Inlet, Alaska.
ridge regression and sparse matrix theory. Geophysics 50: 1721–1731. Earth, Planets and Space 57: 781–793.
Reed LE and Witherly KE (2007) 50 years of kimberlite geophysics: A review. Sambridge M and Mosegaard K (2002) Monte Carlo methods in geophysical inverse
In: Milkereit B (ed.) Proceedings of Exploration 07: Fifth Decennial International problems. Reviews of Geophysics 40: 1009. http://dx.doi.org/
Conference on Mineral Exploration, pp. 679–689. 10.1029/200RG00089.
Reeves CV (2000) The geophysical mapping of Mesozoic dyke swarms in southern Schlinger CM (1985) Magnetization of lower crust and interpretation of regional
Africa and their origin in the disruption of Gondwana. Journal of African Earth magnetic anomalies: Example from Lofoten and Vesteralen, Norway. Journal of
Sciences 30: 499–513. Geophysical Research 90: 1484–1504.
Reeves CV and Erren H (1994) AAIME: Aeromagnetics of Arabia, India, and the Middle Schnetzler CC and Taylor PT (1984) Evaluation of an observational method for
East, International Institute for Aerospace Survey and Earth Science, Unpublished. estimation of remanent magnetization. Geophysics 49: 282–290.
Reeves CV, Macnab R, and Maschenkov S (1998) Compiling all the world’s magnetic Schnetzler CC, Taylor PT, Langel RA, Hinze WJ, and Phillips JD (1985) Comparison
anomalies. Eos, Transactions of the American Geophysical Union 79(28): 338–339. between the recent U.S. composite magnetic anomaly map and Magsat anomaly
Reford S (2006) Gradient enhancement of the total magnetic field. The Leading Edge data. Journal of Geophysical Research 90: 2543–2548.
25: 59–66. Schott JJ and Thébault E (2011) Modeling the earth’s magnetic field from global to
Reichow MK, et al. (2002) Ar–Ar Dates from the West Siberian Basin: Siberian Flood regional scales. In: Mandea M and Korte M (eds.) Geomagnetic Observations and
Basalt Province Doubled. Science 296: 1846–1849. Models. IAGA Special Sopron Book Series, vol. 5, pp. 229–264. Dordrecht:
Reid AB (1980) Aeromagnetic survey design. Geophysics 45: 973–976. Springer.
Reid AB, Allsop JM, Granser H, Millett AJ, and Somerton IW (1990) Magnetic Sexton MA, et al. (1995) The Mt Leyshon magnetic anomaly. Exploration Geophysics
interpretation in three dimensions using Euler deconvolution. Geophysics 26: 84–91.
55: 80–91. Shah AK, et al. (2005) New surveys of the Chesapeake Bay impact structure suggest
Reynolds RL (1990) Rock magnetism, the distribution of magnetic minerals in the melt pockets and target-structure effect. Geology 33: 417–420.
earth’s crust, and aeromagnetic anomalies. In: Hanna WF (ed.) Geologic Sharpton VL, et al. (1992) New links between the Chicxulub impact structure and the
Applications of Modern Aeromagnetic Surveys, pp. 24–45. Denver, CO: U.S. Cretaceous/Tertiary boundary. Nature 359: 819–821.
Geological Survey Bulletin 1924. Shive PN, et al. (1992) Magnetic properties of the lower crust. In: Fountain DM,
Reynolds RL, et al. (1990a) Iron sulphide minerals at Cement oilfield, Oklahoma: Arculus R, and Kay RW (eds.) Continental Lower Crust. Developments in
Implications for magnetic detection of oilfields. Geological Society of America Geotectonics, vol. 23, pp. 145–178. New York: Elsevier.
Bulletin 102: 368–380. Shuey RT and Pasquale AS (1973) End corrections in magnetic profile interpretation.
Reynolds RL, et al. (1990b) Magnetic forward models of Cement oil field, Oklahoma, Geophysics 38: 507–512.
based on rock magnetic, geochemical and petrological constraints. Geophysics Shure L, Parker RL, and Backus GE (1982) Harmonic splines for
55: 344–353. geomagnetic modelling. Physics of the Earth and Planetary Interiors 28:
Reynolds RL, Fishman NS, and Hudson MR (1991) Sources of aeromagnetic anomalies 215–229.
over Cement oil field (Oklahoma), Simpson oil field (Alaska), and the Wyoming- Siebert M and Meyer J (1996) Geomagnetic activity indices. In: Dieminger W,
Idaho-Utah thrust belt. Geophysics 56: 606–617. Hartmann GK, and Leitinger R (eds.) The Upper Atmosphere: Data Analysis and
Reynolds RL, et al. (1994) Magnetization and geochemistry of greigite-bearing Interpretation, pp. 887–911. Berlin: Springer (Chapter V.3).
Cretaceous strata, North Slope Basin, Alaska. American Journal of Science Silva JBC (1986) Reduction to the pole as an inverse problem and its application to low-
294: 485–528. latitude anomalies. Geophysics 51: 369–382.
Ridsdill-Smith TA and Dentith MC (1999) The wavelet transform in aeromagnetic Silva JBC and Barbosa VCF (2003) 3D Euler deconvolution: Theoretical basis for
processing. Geophysics 64: 1003–1013. automatically selecting good solutions. Geophysics 68: 1962–1968.
Rigoti A, Padilha AL, Chamalaun FH, and Trivedi NB (2000) Effects of the equatorial Silva JBC, Barbosa VCF, and Medeiros WE (2001) Scattering, symmetry, and bias
electrojet on aeromagnetic data acquisition. Geophysics 65: 553–558. analysis of source position estimates in Euler deconvolution and its practical
Ripka P (2000) Fluxgate magnetometers. In: Ripka P (ed.) Magnetic Sensors and implications. Geophysics 66: 1149–1156.
Magnetometers, pp. 75–128. Boston, MA: Artech. Simons FJ, Dahlen FA, and Wieczorek MA (2006) Spatiospectral concentration on a
Robinson D, et al. (2008) Advancing process-based watershed hydrological research sphere. SIAM Review 48: 504–536.
using near-surface geophysics: A vision for, and review of, electrical and magnetic Sleep NH (2005) Evolution of the continental lithosphere. Annual Review of Earth and
geophysical methods. Hydrological Processes 22: 3604–3635. Planetary Sciences 33: 369–393.
Roest WR and Pilkington M (1993) Identifying remanent magnetization effects in Smith WHF and Wessel P (1990) Gridding with continuous curvature splines in tension.
magnetic data. Geophysics 58: 653–659. Geophysics 55: 293–305.
Roest WR, Verhoef J, and Pilkington M (1992) Magnetic interpretation using the 3-D Snyder DB, Hobbs RW, and ChicxulubWorkingGroup (1999) Ringed structural zones
analytic signal. Geophysics 57: 116–125. with deep roots formed by the Chicxulub impact. Journal of Geophysical Research
Runcorn SK (1975) On the interpretation of lunar magnetism. Physics of the Earth and 104: 10743–10755.
Planetary Interiors 10: 327–335. Soederlund U, Hofmann A, Klausen M, Olsson J, Ernst R, and Persson P (2010)
Rygaard-Hjalsted C, Mosegaard K, and Olsen N (2000) Resolution studies of fluid flow Towards a complete magmatic barcode for the Zimbabwe craton: Baddeleyite U–Pb
models near the core-mantle boundary through Bayesian inversion of geomagnetic dating of regional dolerite dyke swarms and sill complexes. Precambrian Research
data. In: Hansen PC, Jacobsen BH, and Mosegaard K (eds.) Methods and 183: 388–398.
Applications of Inversion: Proceedings of the IIC98 Conference (Copenhagen, Spector A (1968) Spectral analysis of aeromagnetic data. PhD Thesis, University of
1998), pp. 255–275, Copenhagen, Denmark. Toronto.
Saad AH (1969) Magnetic properties of ultramafic rocks from Red Mountain, California. Spector A and Grant FS (1970) Statistical models for interpreting aeromagnetic data.
Geophysics 34: 974–987. Geophysics 35: 293–302.
Sabaka TJ, Olsen N, and Langel RA (2002) A comprehensive model of the quiet-time, Stockmann R, Finlay CC, and Jackson A (2009) Imaging Earth’s crustal magnetic field
near-the-earth magnetic field: Phase 3. Geophysical Journal International with satellite data: A regularized spherical triangle tessellation approach.
151: 32–68. Geophysical Journal International 179: 929–944.
Sabaka TJ, Olsen N, and Purucker ME (2004) Extending comprehensive models of the Stone VCA, Fairhead JD, and Oterdoom WH (2004) Micromagnetic seep detection in
Earth’s magnetic field with rsted and Champ data. Geophysical Journal the Sudan. The Leading Edge 23: 734–737.
International 159: 521–547. Syberg FGR (1972) A Fourier method for the regional-residual problem of potential
Sailhac P, Galdeano A, Gibert D, Moreau F, and Delor C (2000) Identification of sources fields. Geophysical Prospecting 20: 47–75.
of potential fields with the continuous wavelet transform: Complex wavelets and Talwani M (1965) Computation with the aid of a digital computer of magnetic anomalies
application to aeromagnetic profiles in French Guiana. Journal of Geophysical caused by bodies of arbitrary shape. Geophysics 30: 797–817.
Research 105: 19455–19475. Tanaka A, Okubo Y, and Matsubayashi O (1999) Curie point depth based on spectrum
Salem A and Ravat D (2003) A combined analytic signal and Euler method (AN-EUL) for analysis of the magnetic anomaly data in East and Southeast Asia. Tectonophysics
automatic interpretation of magnetic data. Geophysics 68: 1952–1961. 306: 461–470.
Tarlowski C, et al. (1996) Dewarping the composite aeromagnetic anomaly Voorhies CV, Sabaka TJ, and Purucker M (2002) On magnetic spectra of Earth and
map of Australia using control traverses and base stations. Geophysics Mars. Journal of Geophysical Research: Planets 107(E6): 5034. http://dx.doi.org/
61: 696–705. 10.1029/2001JE001534.
Taylor PT and Frawley JJ (1986) Magsat anomaly data over the Kursk region, USSR. Wang X and Hansen RO (1990) Inversion for magnetic anomalies of arbitrary three-
Physics of the Earth and Planetary Interiors 45: 5–15. dimensional bodies. Geophysics 55: 1321–1326.
Taylor PT and Ravat D (1995) An interpretation of the Magsat anomalies of central Warner RD and Wasilewski PJ (1997) Magnetic petrology of arc xenoliths from Japan
Europe. Applied Geophysics 34: 83–91. and Aleutian Islands. Journal of Geophysical Research 102: 20225–20244.
Teskey DJ, Hood PJ, Morley LW, et al. (1993) The aeromagnetic survey Wasilewski PJ and Mayhew MA (1992) The Moho as a magnetic boundary revisited.
program of the Geological Survey of Canada: Contribution to regional Geophysical Research Letters 19: 2259–2262.
geological mapping and mineral exploration. Canadian Journal of Earth Sciences Weinberg RF, Hodkiewicz PF, and Groves DI (2004) What controls gold distribution in
30: 243–260. Archean cratons? Geology 32: 545–548.
Thébault E, Schott JJ, Mandea M, and Hoffbeck JP (2004) A new proposal Wells RE, Weaver CS, and Blakely RJ (1998) Fore-arc migration in Cascadia and its
for spherical cap harmonic modelling. Geophysical Journal International neotectonic significance. Geology 26: 759–762.
159: 83–103. Werner S (1955) Interpretation of magnetic anomalies of sheet-like bodies, Sver. Geol.
Thébault E, et al. (2006) Revised spherical cap harmonic analysis (R-SCHA): Validation Undersok. Serv. C, Arsbok 43, No. 6.
and properties. Journal of Geophysical Research 111. http://dx.doi.org/ Wessel P and Smith W (1998) New, improved version of Generic Mapping Tools
10.1029/2005JB003836. released. EOS. Transactions of the American Geophysical Union 79: 579.
Thébault E, Purucker M, Whaler K, Langlais B, and Sabaka T (2010) The magnetic field Whaler KA (1994) Downward continuation of Magsat lithospheric anomalies to the
of the Earth’s lithosphere. Space Science Reviews 155: 95–127. http://dx.doi.org/ Earth’s surface. Geophysical Journal International 116: 267–278.
10.1007/s11214-010-9667-6. Whaler KA and Langel RA (1996) Minimal crustal magnetizations from satellite data.
Thompson DT (1982) EULDPH: A technique for making computer-assisted depth Physics of the Earth and Planetary Interiors 48: 303–319.
estimates for magnetic data. Geophysics 47: 31–37. Whaler K and Purucker M (2005) A spatially continuous magnetization model for Mars.
Thurston JB and Smith RS (1997) Automatic conversion of magnetic data to Journal of Geophysical Research 110(E9): E09001. http://dx.doi.org/
depth, dip, and susceptibility contrast using the SPI™ method. Geophysics 10.1029/2004JE002393.
62: 807–813. Whaler KA, Neil C, and Reid AB (1991) Extension to Euler’s method for three-
Thurston JB, Smith RS, and Guillon J-C (2002) A multimodel method for depth dimensional potential field interpretation. In: EAEG 53rd Meeting and Technical
estimation from magnetic data. Geophysics 67: 555–561. Exhibition, Florence, Italy, pp. 416–417, Expanded Abstracts.
Toft PB and Haggerty SE (1988) Limiting depth of magnetization in cratonic lithosphere. Whaler KA, Langel RA, Jackson A, and Purucker ME (1996) Non-uniqueness in
Geophysical Research Letters 15: 530–533. magnetization: Crustal models from satellite magnetic data. EOS. Transactions of the
Toft PB and Haggerty SE (1989) Reply: Limiting depth of magnetization in cratonic American Geophysical Union 77: F172.
lithosphere. Geophysical Research Letters 16: 480–482. Wieczorek MA, Weiss BP, and Stewart ST (2012) An impactor origin for lunar magnetic
Treitel S, Clement WG, and Kaul RK (1971) The spectral determination of depths to anomalies. Science 335: 1212–1215.
buried magnetic basement rocks. Geophysical Journal of the Royal Astronomical Wonik T, et al. (2001) Magnetic anomaly map for northern, western, and eastern
Society 24: 415–428. Europe. Terra Nova 13: 203–213.
Trumbull RB, et al. (2004) Aeromagnetic mapping and reconnaissance geochemistry of Yanovsky BM (1938) Variations in elements of terrestrial magnetism in an anomalous
the Earth Cretaceous Henties Bay-Outjo dike swarm, Etendeka Igneous Province. field. Transactions of the Main Geophysical Observatory 17(3).
Journal of African Earth Sciences 40: 17–29. Zietz I (1982) Composite magnetic anomaly map of the United States: Part A,
Vacquier V, Steenland NC, Henderson RG, and Zietz I (1951) Interpretation of conterminous United States: U.S. Geological Survey Geophysical Investigations
Aeromagnetic Maps. Geological Society of America, Memoir 47. Map GP-954-A, pp. 59, 2 sheets, scale 1:2,500,000.
Vallée MA, Keating P, Smith RS, and St-Hilaire C (2004) Estimating depth and model Zietz I and Andreasen GE (1967) Remanent magnetization and aeromagnetic
type using the continuous wavelet transform of magnetic data. Geophysics interpretation. Mining Geophysics, vol. 2, pp. 569–590. Tulsa, OK: Society of
69: 191–199. Exploration Geophysicists.
Vasicek JM, Frey HV, and Thomas HH (1988) Satellite magnetic anomalies and the Zietz I and Bhattacharyya BK (1975) Magnetic anomalies over the continents and their
Middle America trench. Tectonophysics 154: 19–24. analyses. Reviews of Geophysics and Space Physics 13: 176–215.
Verduzco B, Fairhead JD, Green CM, and MacKenzie C (2004) New insights into
magnetic derivatives for structural mapping. The Leading Edge 23: 116–119.
Verhoef J, et al. (1996) Magnetic anomalies of the Arctic and North Atlantic Oceans and
adjacent land areas. Geological Survey of Canada Open File Report 3125. Relevant Websites
Dartmouth, NS: Geological Survey of Canada.
Verrier V and Rochette P (2002) Estimating peak currents at ground lightning impacts http://virtualexplorer.com.au/special/noddyatlas – Atlas of Structural Geophysics.
using remanent magnetization. Geophysical Research Letters 29. http://dx.doi.org/ http://www.dnsc.dk/Oested/Field_models – Magnetic field models, Denmark.
10.1029/2002GL015207. http://www.spacecenter.dk/files/magnetic-models/CHAOS-4 – DTU field models.
Vervelidou F (2013) Contribution à la modélisation et à l’interprétation multi-échelle du http://gmt.soest.hawaii.edu – GMT.
champ magnétique de la lithosphère terrestre. PhD Thesis, Université Paris Diderot. http://geomag.org – IGRF.
Vine FJ and Matthews DH (1963) Magnetic anomalies over oceanic ridges. Nature http://www.ngdc.noaa.gov/geomag – NGDC magnetic field models.
199: 947–949. http://www.nr.com – Numerical Recipes.
Vivier F, Maier-Reimer E, and Tyler RH (2004) Simulations of magnetic fields generated http://planetary-mag.net – Planetary magnetic field evaluation.
by the Antarctic circumpolar current at satellite altitude: Can geomagnetic http://www.ipgp.fr/wieczor/SHTOOLS/SHTOOLS.html – SHTOOLS.
measurements be used to monitor the flow? Geophysical Research Letters http://www.cisl.ucar.edu/css/software/spherepack – Spherepack.
L10306. http://dx.doi.org/10.1029/2004GL019804. http://geoweb.princeton.edu/people/simons/software.html – Spherical Slepian function
Von Frese RRB, Hinze WJ, Olivier R, and Bentley CR (1986) Regional magnetic anomaly software.
constraints on continental breakup. Geology 14: 68–71. http://pubs.usgs.gov/fs/fs-0076-95/FS076-95.html – USGS potential field software.
5.07 Geomagnetic Induction Studies
S Constable, University of California San Diego, La Jolla, CA, USA
5.07.1 Geomagnetic Induction Studies 220

5.07.1.1 Historical Beginnings 220
5.07.1.2 The Earth’s Geomagnetic Environment 220
5.07.1.3 Earth/ionosphere cavity 221
5.07.1.4 Daily variation and Sq 222
5.07.1.5 The magnetospheric ring current 222
5.07.1.6 Pulsations 223
5.07.2 Geomagnetic Sounding 223
5.07.2.1 Introductory Theory 223
5.07.2.2 The MT Method 224
5.07.2.3 MT in Practice 225
5.07.2.4 Equipment 225
5.07.2.4.1 Electric field measurements 225
5.07.2.4.2 Magnetic field measurements 226
5.07.2.5 MT Data Processing 227
5.07.2.6 Global Conductivity Studies 228
5.07.2.7 Global Response Functions 228
5.07.2.7.1 Using magnetic satellites 229
5.07.3 Interpretation of GDS and MT Data 230
5.07.3.1 Forward Model Studies 230
5.07.3.2 Parameterized Inversion 230
5.07.3.3 Regularized Inversion 231
5.07.3.4 Analytic Least Squares Solutions 231
5.07.3.5 MT over a Layered Earth 231
5.07.3.5.1 Weidelt’s transformation 231
5.07.3.6 Forward Modeling in Higher Dimensions 231
5.07.3.7 Numerical Inversion of Geomagnetic Data 232
5.07.3.8 Analytic Inversion and Estimation of Bounds 233
5.07.4 Electric Conductivity of Earth Materials 234
5.07.4.1 Moving an Electric Charge 234
5.07.4.1.1 Conduction in native metals 234
5.07.4.1.2 Semiconduction 235
5.07.4.1.3 Ionic, or point defect, conduction 235
5.07.4.1.4 Pressure 236
5.07.4.1.5 Aqueous electrolytic conduction 237
5.07.4.1.6 Clay minerals 237
5.07.4.1.7 Magma 237
5.07.4.2 Binary Mixing Laws 237
5.07.4.3 Polarization 238
5.07.4.4 Anisotropy 238
5.07.4.4.1 Crystallographic anisotropy 238
5.07.4.4.2 Textural anisotropy 238
5.07.4.4.3 Structural anisotropy 238
5.07.4.5 Laboratory Measurement of Conductivity 238
5.07.4.6 High-Pressure Measurements 241
5.07.4.7 Thermopower Measurements 241
5.07.4.8 Conductivity of Mantle Minerals 242
5.07.4.9 Olivine Conductivity 242
5.07.5 Global Conductivity Structure 244
5.07.5.1 The Oceans and Crust 244
5.07.5.2 The Mantle 247
5.07.5.2.1 Melt 248
5.07.5.2.2 Carbon 248
5.07.5.2.3 Strain-induced defects 249

220 Geomagnetic Induction Studies
5.07.5.2.4 Water 249

5.07.5.2.5 Oxides and sulfides 249
5.07.5.3 The Core 250
5.07.6.1 3D Conductivity from Satellite Data 251
5.07.6.2 Water in the Mantle 251
5.07.6.3 Anisotropy Measurements 252
Acknowledgment 252
References 252
5.07.1 Geomagnetic Induction Studies Induction effects associated with magnetic storms were first
seen in telegraph cables, the storm of 1838 being seen on
The interaction of temporal fluctuations in the Earth’s external Norwegian telegraph cables; in telegraph lines in England,
magnetic field with the electrically conductive rocks of the spontaneous currents reported by W.H. Barlow (1849); and
planet generates internal secondary electric and magnetic fields. in an observation by Clement (1859) that the aurora of
The measurements of the primary and secondary fields may 29 August disrupted telegraphy. The critical mathematical
then be used to probe the conductivity structure over depths developments started in 1839 with Gauss’ spherical harmonic
from a few meters to the lower mantle. Conductivity in turn may expansion of the main magnetic field, which allows the sepa-
be used to infer physical and chemical properties, such as tem- ration of internal and external magnetic sources, and later
perature, melt content, water and volatile content, mineral type, (1864) with Maxwell’s equations of electromagnetism. In
and anisotropic fabric. There are three basic steps necessary to 1865, the Greenwich Observatory started to observe Earth
obtain useful information from geomagnetic induction studies: potentials on 15 km grounded lines.
the estimation of electromagnetic impedances from observa- By early in the twentieth century, the relationship between
tions of the electric and magnetic fields, the modeling of these magnetic activity and Earth currents was well known. Schuster
impedances in both forward (conductivity to impedance) and and Lamb (1889) observed the relationship between the diur-
inverse (impedance to conductivity) directions, and, finally, the nal magnetic variation and Earth potentials, and Chapman
use of laboratory studies to relate conductivity to the relevant (1919) inferred that the Earth’s interior must be more conduc-
physical and chemical properties. Each of these steps represents tive than crustal rocks and modeled the daily variation with a
significant scientific challenges, and all of them are the subject of conductive sphere of smaller radius than the Earth’s. Although
current research and recent progress. Clearly, the results of each Chapman and Bartels (1940) lamented that a more quantita-
of these components depend on the quality of the preceding tive relationship between electric conductivity and magnetic
step, and conclusions about the state of the Earth’s interior field variation was not yet available, progress in this direction
depend on the reliability of all three parts of the process. was being made. Lahiri and Price (1939) modeled the internal
Electromagnetic induction covers a very broad field that and external parts of the magnetic field using a radial conduc-
includes applications in exploration, environmental and min- tivity profile to a depth of nearly 1000 km. Later, Banks (1969)
ing geophysics, and studies of the Earth’s deep interior. proposed that the harmonics of the 27-day solar rotation were
Although many of the underlying principles span the entire dominantly of P01 spherical harmonic geometry, associated
range of applications, in the context of Treatise on Geophysics, with the ring of current in the magnetosphere above the geo-
we will concentrate on the ‘whole-Earth’ aspects of geomag- magnetic equator, and produced a conductivity profile down
netic induction. For a good discussion of exploration and to nearly 2000 km. In order to get an estimate of the electric
crustal applications, the reader is referred to the book by conductivity of the lowermost mantle, McDonald (1957)
Simpson and Bahr (2005). modeled the outward propagation of an inferred geomagnetic
secular variation signal originating in the core. Figure 1 sum-
marizes the results of this early work.
These observations represent the beginnings of the geomag-
5.07.1.1 Historical Beginnings
netic depth-sounding (GDS) method, in which magnetic fields
The history of geophysical electromagnetic induction is alone are used to probe the Earth’s conductivity. Another highly
entwined with the development of the fundamental physics important method, magnetotelluric (MT) sounding, began in
of electromagnetic energy, closely following Oersted’s 1820 the early 1950s when Tikhonov (1950) and Cagniard (1953)
observation that an electric current deflected a magnet and quantified the relationship between induced electric currents
Ampere’s quantification of this phenomenon in the 1820s. and the magnetic field. Both the GDS and MT methods are in
Ampere also noted that an electric current exerts a force on a extensive use today, and we will discuss them in this chapter.
second electric current and that a coil behaved like a magnetic
dipole. In 1831, Faraday observed that moving a magnet
through a coil produces an electric current, and in 1832, he
5.07.1.2 The Earth’s Geomagnetic Environment
predicted that water moving through a magnetic field should
produce an electric field, an effect that was observed by the Temporal variations in the Earth’s magnetic field exist on all
British Admiralty much later in 1918. timescales from radio frequencies to the reversal record
Geomagnetic Induction Studies 221
(Figure 2). Electromagnetic induction methods can exploit become smaller as period increases, periods of a few weeks or
only those fields that are externally generated, which in prin- shorter for MT methods. As mentioned in Section 5.07.1, the
ciple means periods shorter than about 1 year for GDS propagation of internally generated fields of the secular varia-
methods (with the possible exception of the 11-year sunspot tion upward through the mantle may be used to estimate
cycle) and, because the induced electric field components conductivity (Alexandrescu et al., 1999; McDonald, 1957;
Pinheiro and Jackson, 2008), but the lack of an independent
estimate of the source-field timing limits this approach
(Backus, 1983).
103
McDonald (1957) 5.07.1.3 Earth/ionosphere cavity

102 Starting at the high-frequency end of the spectrum, the con-
ductive ionosphere and the conductive Earth are separated by
the resistive atmosphere to form a cavity. The electric conduc-
Conductivity (S m-1)
101
Lahiri and Price (1939), model d tivity of the ground is of order 103 S m1, while the atmo-
sphere near the Earth’s surface is about 1014 S m1. What
Banks (1969)
100 little conductivity there is results from the ionization of oxygen
and nitrogen by cosmic rays. The mean free path of these ions
increases with decreasing atmospheric density, and by an alti-
10-1 tude of 100 km, the start of the ionosphere, conductivity is
about the same as that of the solid Earth. The ionosphere is
several hundred kilovolts positive with respect to ground, and
10-2 a leakage current of about 1000 A flows between the iono-
sphere and ground through an integrated transverse resistance
Upper mantle Lower mantle
10-3 of about 200 O. This leakage current supports a vertical electric
500 1000 1500 2000 2500 3000 field of 100–300 V m1 at the ground’s surface. The Earth/
Depth (km) atmosphere/ionosphere acts as a large capacitor with a value
Figure 1 Early electric conductivity profiles of the Earth. These models of about 1 F and a time constant of a few thousand seconds.
all feature an increase in conductivity at about the upper/lower mantle Thus, there must be a return current to sustain the charge on
transition, a feature of most modern models. the ionosphere.
Grand spectrum
105
Reversals
103
Cryptochrons?
101
Secular variation
10-1
Amplitude (T/√Hz)
?
?
Annual and semi-annual
10-3 Solar rotation (27 days)

Daily variation
10-5
10-7 Quiet days

Storm activity
10-9
Powerline noise
Radio
10-11
1 thousand years
10 million years
Schumann resonances
10-13
1 month
10 kHz
1 year
1 day
1 min
10-15
1h
1s
10-15 10-11 10-7 10-3 101 105

Frequency (Hz)
Figure 2 An approximate amplitude spectrum for the Earth’s magnetic field (to get a power spectrum, double the exponents, and change the
units to T2/Hz). While somewhat schematic, this figure illustrates that the source and morphology of the field variations depend on frequency.
In particular, at periods shorter than about a year, the field is dominated by sources external to the Earth. Reproduced from Constable CG and Constable
SC (2004a) Satellite magnetic field measurements: Applications in studying the deep Earth. In: Sparks RSJ and Hawkesworth CT (eds.) The State
of the Planet: Frontiers and Challenges in Geophysics, Geophys. Mono. 150, pp. 147–159. Washington, DC: American Geophysical Union.
The major contributor to the return current is thunderstorm magnetic field is seen at the surface of the Earth, amounting to
activity, in which convection carries positively charged parti- a few tens of nanotesla.
cles to the tops of thunderclouds and cloud-to-ground negative
charge flow in lightning completes the circuit. Thunderstorms
preferentially form in the tropics, in the afternoon, and over 5.07.1.5 The magnetospheric ring current
land. Since Africa is the dominant equatorial land mass, there The largest component of external field variations comes
is an uneven temporal distribution of lightning activity that from a ring of current circulating roughly around the magnetic
peaks as the Earth rotates through afternoon in equatorial equator at a distance of 2–9 Earth radii. The solar wind
Africa, or about 16:00 UT. This recharging of the global capac- injects charged particles into the ring current, mainly positively
itor can be seen as an increased fair-weather electric field over charged oxygen ions. Sudden increases in the solar wind
the oceans anywhere in the world at these times, first observed associated with coronal mass ejections cause magnetic storms,
from the research vessel Carnegie and now known as the characterized by a sudden commencement, a small but sharp
Carnegie curve. increase in the magnetic field associated with the sudden
The atmospheric cavity is excited by lightning strikes and increased pressure of the solar wind on the Earth’s magneto-
resonates with a characteristic frequency of about 8 Hz and sphere. Following the commencement is a period of fluctuating
harmonics, a phenomenon called the Schumann resonance. magnetic field called the initial phase. The main phase of the
Man-made sources of electromagnetic energy such as power- storm is associated with a large decrease in the magnitude of
line noise at 50 and 60 Hz and the entire radio frequency the fields as a westward circulating ring current is energized – the
spectrum also propagate within the resistive atmosphere effect is to cancel the Earth’s main dipole field slightly. Finally,
(Figure 3). there is a recovery phase in which the field returns quasi-
exponentially back to normal. All this can happen in a couple
of hours or may last days for a large storm.
5.07.1.4 Daily variation and Sq Although complicated in detail, the ring current behaves
The dayside ionosphere, heated by the sun, is more conductive somewhat like a large single turn of wire around the Earth,
than the nightside and has a pattern of two circulating current creating fields at the Earth’s surface of predominantly simple P01
systems (one in each hemisphere) that are stationary in solar spherical harmonic geometry. The frequency content is huge:
time, with an intensified, normally eastward-flowing current at from minute-by-minute fluctuations during a storm, the
the magnetic equator called the equatorial electrojet. Since the hours-long duration of a storm, a large peak at the 27-day
Earth rotates beneath these current systems, a daily variation in rotation period of the sun (and harmonics), a semiannual
The geomagnetic earth
Satellite
Ionosphere
rre ric
s
cu he
nt
Atmosphere
Sq nosp
Io
+ Schumann resonances
Solar wind
Magnetosp
heric
e
induction Inner
cor
Lower
on core Magnetospheric
Uppe
mantle
ter
ucti
q ind ring current
Ou
S
rm
–
an
tle
Magnetosphere
Tangent cylinder Crust
Secular variation
Figure 3 A pictorial representation of the electromagnetic environment of the Earth. Reproduced from Constable CG and Constable SC (2004a)
Satellite magnetic field measurements: Applications in studying the deep Earth. In: Sparks RSJ and Hawkesworth CT (eds.) The State of the Planet:
Frontiers and Challenges in Geophysics, Geophys. Mono. 150, pp. 147–159. Washington, DC: American Geophysical Union.
Dst index (blue, nT) and 10.7 cm radio flux (red, 10-22 J/s/m2/Hz)
400
300
200
100
−100
−200
−300
−400
−500
−600
1980 1982 1984 1986 1988 1990 1992 1994 1996 1998
Year
Figure 4 A record of the Dst index (blue) for 19 years between 1979 and 1998. The large negative spikes are individual magnetic storms. To illustrate
the relationship between storm activity and the 11-year sunspot cycle, the 10.7 cm solar radio flux is plotted in red.
line associated with the geometry of the ecliptic and the sun’s where J is the current density (A m2), E is the electric field
equatorial plane, and, finally, the 11-year solar cycle. (V m1), B is the magnetic flux density or induction (T ), and H
The strength of the ring current is characterized by the ‘Dst’ is the magnetic field intensity (A m1). We neglect displace-
(disturbance storm time) index, an index of magnetic activity ment currents, as they are not significant at the frequencies and
derived from a network of low- to midlatitude geomagnetic obser- conductivities relevant to geomagnetic induction (for a rigor-
vatories that measures the intensity of the globally symmetrical ous justification of this, see Section 2.4 of Backus et al., 1996).
part of the equatorial ring current (Figure 4). The current method We can use the identity ∇ • ∇ A ¼ 0 to show that
for computing Dst is described in IAGA Bulletin No. 40, a report
by Masahisa Sugiura that presents the values of the equatorial ∇ B ¼ 0 and ∇ J ¼ 0
Dst index for 1957–86. Honolulu, Hermanus, San Juan, and
in regions free of sources of magnetic fields and currents. B and
Kakioka are the currently contributing observatories. The actual
H are related by magnetic permeability m and J and E by
morphology of the ring current fields is asymmetrical about the
conductivity s
day/night hemisphere, as injection of ions occurs preferentially
as a function of solar time (e.g., Balasis and Egbert, 2006). B ¼ mH J ¼ sE
(the latter equation is Ohm’s law), and so

5.07.1.6 Pulsations
The interaction of the solar wind and magnetosphere can @H
∇ E ¼ m
generate resonances and waves that have relatively narrow- @t
band frequency content, creating quasiperiodic oscillations of ∇ H ¼ sE
the magnetic field called pulsations. The period of oscillation If we take the curl of these equations and use ∇
varies between about 1 s and several hundred seconds, and the (∇ A) ¼ ∇ (∇ • A) ∇2A, for constants s and m, we have
nomenclature for pulsations distinguishes between various
frequency bands and whether the variations are continuous @ @E
∇2 E ¼ m ð∇ HÞ ¼ ms
for a time or form a damped oscillation (see Parkinson, 1983). @t @t
@H
∇2 H ¼ sð∇ EÞ ¼ ms
@t
5.07.2 Geomagnetic Sounding which are diffusion equations. In air and very poor conductors
where s 0, or if o ¼ 0, the equations reduce to Laplace’s
5.07.2.1 Introductory Theory
equation. Now, if we consider sinusoidally varying fields of
One of the key concepts in geomagnetic induction is that of the angular frequency o,
skin depth, the characteristic length over which electromagnetic
@E
fields attenuate. We can derive the skin depth starting with Eðt Þ ¼ Eo eiot ¼ ioE
@t
Faraday’s law
@H
Hðt Þ ¼ Ho eiot ¼ ioH
@B @t
∇E¼
@t and so
and Ampere’s law
∇2 E ¼ iomsE
∇H¼J ∇2 H ¼ iomsH
If we further consider fields that are horizontally polarized in in terms of electric conductivity with depth. The estimation of
the xy directions and are propagating vertically into a half- purely geomagnetic response functions and their interpreta-
space, in Cartesian coordinates, these equations decouple to tion in terms of mantle electric conductivity structure date
from the end of the nineteenth century, but the additional
@2E
þ k2 E ¼ 0 use of electric field measurements only started with Cagniard’s
@z2 work in the 1950s.
@2H One can see that the core is effectively a perfect conductor
þ k2 H ¼ 0
@z2 into which even the longest-period external magnetic field
with solutions cannot penetrate. However, 1-year variations can penetrate
into the lower mantle. The very large range of conductivities
E ¼ Eo eikz ¼ Eo eiaz ebz found in the crust (sea, sediment, and igneous rock) indicates
H ¼ Ho eikz ¼ Ho eiaz ebz the need for a corresponding large range of frequencies in
electromagnetic sounding of that region.
where we have defined a complex wave number
Continuing with our half-space formulation, we can define
pffiffiffiffiffiffiffiffiffiffiffi
k ¼ ioms ¼ s ib an impedance by taking the ratio of orthogonal field
components
and an attenuation factor, which is called a skin depth,
sffiffiffiffiffiffiffiffiffiffiffi Ex mo
z¼ ¼
2 Hy k
zs ¼ 1=a ¼ 1=b ¼
smo o
where we recall that the complex wave number
The skin depth is the distance that field amplitudes are 1 p
k ¼ a ib ¼ ðsmoÞ2 ei 4
reduced to 1/e, or about 37%, and the phase progresses one
radian, or about 57 . In practical units of meters, the skin and so we can see that Ex leads Hy by 45 and that the half-
depth is space resistivity is given by
pffiffiffiffiffiffiffiffiffiffi
zs 500 1=sf 1 1 T Ex 2
r¼ ¼ jZj2 ¼
s 2pf m 2pm Hy
where circular frequency f is defined by o ¼ 2pf.
The skin depth concept underlies all of inductive electro- for fields with period T. This is the MT formula made famous
magnetism in geophysics. Substituting a few numbers into the by Cagniard (1953). Perhaps more usefully (since measure-
equation shows that skin depths cover all geophysically useful ments tend to be made in B rather than H),
length scales from less than a meter for conductive rocks and
m Ex 2
kilohertz frequencies to thousands of kilometers in mantle r ¼
rocks and periods of days (Table 1). Skin depth is a reliable o By
indicator of maximum depth of penetration, but one should
Another quantity called the c-response (Weidelt, 1972)
not think of skin depth as a resolution length. A thin conduc-
(also called admittance by Parker, 1994) is also used, given by
tive layer can have an effect at periods associated with skin
depths much larger than its width, and all external magnetic Ex
c¼
field variations have to propagate through surface layers. The ioBy
skin depth describes attenuation associated with induction,
(note that c is complex) from which it can be seen that
but electric and magnetic fields are also modified by variations
in conductivity through Ohm’s law even in the DC limit, an r ¼ omjcj2
effect often described as ‘galvanic’ rather than inductive.
If the Earth really were a homogeneous half-space, the
resistivity obtained at all periods would be the same and
5.07.2.2 The MT Method
equal to the true resistivity of the Earth, and the phase between
Because the skin depth generates a natural scale length for time- B and E would be constant at 45 . Of course, this is never the
varying fields, the response of the Earth to time variations in case, but resistivities are computed nevertheless and described
the externally generated geomagnetic field can be interpreted as ‘apparent resistivities.’ Solutions for computing the apparent
Table 1 Approximate skin depths for a variety of typical Earth environments and a large range of frequencies
Material s (S m1) 1 year 1 month 1 day 1h 1s 1 ms
Core 5 105 4 km 1 km 209 m 43 m 71 cm 23 mm

Lower mantle 10 894 km 258 km 47 km 9 km 160 m 5m
Seawater 3 1600 km 470 km 85 km 17 km 290 m 9m
Sediments 0.1 9000 km 2600 km 470 km 95 km 1.6 km 50 m
Upper mantle 0.01 3 104 km 8000 km 1500 km 300 km 5 km 160 m
Igneous rock 1 105 106 km 2 105 km 5 104 km 9500 km 160 km 5 km
resistivity response of layered 2D and even 3D models exist attenuation provides an indication of the limits to deepest
and are used to interpret the MT data. The phase relationship sensitivity (see Figure 5(b)). The type of instrumentation
between E and B is also estimated and is useful because it is used to record the E and B fields will depend on the periods
independent of the amplitudes of E and H. Complicated near- required, and the time over which data are collected will deter-
surface structure can change the amplitude of the electric fields mine the longest period available. Horizontal spacing of MT
very easily, but because these distortions are not inductive, the sites can vary from continuous electric field dipoles for shallow
phase relationships are not altered. crustal studies, to individual sites spaced tens of kilometers
apart for regional profiles, to single long-period soundings
for mantle conductivity studies (e.g., Egbert and Booker,
5.07.2.3 MT in Practice 1992; Tarits et al., 2004).
Figure 5(a) shows a schematic layout of an MT measurement;

the magnetic and electric fields are measured orthogonal to
5.07.2.4 Equipment
each other. To a first approximation, the magnetic field is the
EM source and the electric field is the EM response. For a 1D 5.07.2.4.1 Electric field measurements
Earth, this would be the only measurement required; for a 2D Metal electrodes are too noisy for MT measurements, except at
or 3D Earth, measurements are made in both directions, and radio frequencies. A nonpolarizing electrode, or porous pot, is
perhaps, even the vertical magnetic field would be measured. almost always used, in which a reversible electrochemical reac-
In Figure 5(b), we provide a quantitative example of this tion is restricted to that of a single chemistry. Electric connec-
behavior, plotting magnetic field, electric field, and current tion is made to a metal rod dipped into an aqueous saturated
density for 0.5 Hz energy propagating into a model containing solution of the metal salt (Figure 6). Contact is made with the
a buried conductive layer. All the fields decay quasi- ground through a semipermeable material such as unglazed
exponentially within the Earth, but inside the conductor, the porcelain, porous plastic, or plaster. Copper–copper sulfate is a
magnetic fields can be seen to attenuate more rapidly due to common chemistry for land use, as is lead–lead chloride sta-
induced electric currents, which peak inside this layer. At a bilized in plaster of Paris or mud made with lead chloride
higher frequency (5 Hz), magnetic fields attenuate before the solution (Petiau, 2000). (For marine use, silver–silver chloride
conductive layer is reached because of the shorter skin depth. provides a low-noise electrode, the silver chloride being com-
Without the conductive layer, fields propagate more deeply patible with seawater and very low solubility (Filloux, 1987;
into the Earth, but importantly, the electric field at the surface Webb et al., 1985).) On land, such a system produces a low-
depends on structure at depth. This is the principle of the MT noise electric contact with an impedance of typically a few
method. kiloohm. Electrodes are buried at a depth of around 30 cm or
MT measurements can be made from the low end of the more to limit temperature variations, which can generate a
radio frequency spectrum, using radio stations and lightning as time-varying potential on the electrode. Dipole lengths are
sources of energy, to periods of several days. The table of skin about 50–200 m. Since signals are of order mV km1, a gain
depths gives some indication of the relationship between con- of 10–1000 is applied before digitization. The amplifier needs
ductivity, period, and penetration scale. Structure can pro- to be a true differential amplifier because there is no natural
foundly affect the way MT fields propagate, but skin-effect concept of ‘ground’ when making telluric measurements.
0
ity ds
ns fiel
200 Hz de lE
E-sensor t5 nt od
e
a rre M
B cu
400 el
s
d
ield
B-sensor Mo
Bf
600
del
Mo
800
E
J,
Depth (m)
1000 0.5 Hz
1.2 Wm
1200
0.25 Wm
B
1400
1.2 Wm
1600
Half-space E fields
1800
2000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
(a) (b) Amplitude (nT, mV km-1, A km-2)
Figure 5 (a) Schematic of an magnetotelluric (MT) measurement. Orthogonal components of the magnetic and electric field are measured at the
surface of the Earth. In conductive layers (red), electric fields and currents are induced whose secondary magnetic fields attenuate the primary
magnetic field more rapidly than in the background material. (b) In a quantitative example of this behavior, magnetic field, electric field, and current
density are plotted for 0.5 Hz energy propagating into a model containing a buried conductive layer (solid lines). For comparison, magnetic fields for a
5 Hz source field are shown (broken red line). The dotted blue line shows the electric fields without the conductive layer.
Often, optical isolation is used to ensure no ground path exists used because then a is limited by the core geometry, rather than
between the recording system and the electrode system. mr, thus removing any temperature dependence in mr from the
sensor. Since most of the flux is trapped in the core material, A
becomes the cross-sectional area of the core, rather than the
5.07.2.4.2 Magnetic field measurements area of each turn of the wire. For large N, achieved by making
Two types of magnetic field sensor are in common use, both the windings out of a very fine wire, the dominant source of
necessarily vector instruments (which rules out nuclear preces- noise becomes the thermal resistance noise, or Johnson noise,
sion sensors). For periods shorter than a few thousand sec- of the wire:
onds, induction coils are used, and for periods longer than
about 10 s, fluxgate magnetometers provide a lower-noise V 2J ¼ 4kTR
measurement. where k is the Boltzmann constant (1.4 1023 J K1), T is the
Induction coils (Figure 7) are simply an application of temperature, and R is the resistance. Here, we see the second
Faraday’s law reason for a long, thin magnetometer. Piling windings upon
þ windings increases the diameter of the turns, and so R starts to
dF
E dl ¼ increase more rapidly than V.
C dt
A good induction coil magnetometer has a noise level of
so integrating E around a single loop of wire of area A gives a about 108 nT2 Hz1 at around 1 Hz (i.e., less than a picote-
potential difference sla), with a red noise spectrum at lower frequencies because of
dB the dB/dt loss of sensitivity. Long-period response is limited to
V ¼ A about 5000 s, while high-frequency response is limited by
dt
capacitive and inductive losses in the core material. This is
We see that V increases with frequency, but since variations addressed by laminating the core material and, ultimately,
in B are of order pico- to nanotesla, unless A is huge, this does going to air-cored coils for the highest (low radio) frequencies.
not produce much of a voltage. Two approaches are used to Because the Earth’s main field is around 40 mT, minute rota-
boost V. The first is to simply wind many (say, N) turns in tion of an induction coil magnetometer couples this field into
series and the second is to increase the flux through the coil by the sensor. For a subpicotesla noise floor, this corresponds to a
inserting a core material of high relative permeability mr, such rotation of only a nanoradian, or 1 mm in 1000 km. For these
as mumetal or Permalloy. The permeable core provides an reasons, coils are buried 10 cm or so to avoid wind motion,
effective increase in flux through the coil by a factor of a, so and even so MT sites near trees or coastlines will be observed to
dB have high noise levels.
V ¼ aNA The fluxgate magnetometer (Figure 8) measures the total
dt
magnetic field and so is useful for long-period MT studies and
The relative permeability of mumetal alloys is of order 104– magnetic observatory recording. In a fluxgate magnetometer,
105, so a can be made quite high. In practice, a long, thin coil is two primary coils are wound around two identical cores of
permeable material and connected in series so that a current
Contact point passed through the primary circuit generates magnetic fields in
the two cores that oppose each other. A 50–1000 Hz primary
Saturated electrolyte Insulating stopper current is used of sufficient strength to saturate the cores. In the
Porous material absence of an external field, the magnetization in the cores is
Metal rod equal and opposite, so there is no net magnetic field for the
(ceramic or wood)
two-core system. In the presence of an external field, the core
being magnetized in the direction of the external field saturates
sooner than the opposite core. Thus, an asymmetry develops in
the magnetization of the cores, leading to a time-varying net
magnetization. A secondary coil wound around the entire
system measures the rate of change of this magnetization as a
series of induced voltage spikes. These spikes are rectified and
amplified to produce a voltage signal that is proportional to
Figure 6 Porous pot, or nonpolarizing, electrode. the magnetic field along the axis of the sensor cores. Note that
Amplifier
Output
B
Permalloy bar
Windings
Figure 7 Induction coil magnetometer.
the fundamental harmonic of the sensing voltage is twice the Figure 2 of Constable, 2013). However, the noise stays below
fundamental harmonic of the primary current. the geomagnetic spectrum until about 1 Hz, making the flux-
Because permeability and saturation magnetization will gate an adequate induction sensor for periods longer than
depend on temperature, one common approach to building a about 10 s.
fluxgate sensor is to operate it as a null instrument. A solenoid
around the fluxgate sensor carries an electric current that cre- 5.07.2.5 MT Data Processing
ates a magnetic field equal and opposite the Earth’s field (in
that direction). The output of the fluxgate should then be zero The raw measurements made in the field are time series of
(and so need not be calibrated). Any nonzero measurements magnetic and electric fields, usually in two orthogonal direc-
from the fluxgate are fed back into the solenoid current to tions. An example is given in Figure 9. The objective of MT data
return the measurement to zero. The current in the solenoid, processing is to obtain the frequency domain response of the
rather than the fluxgate output, is then the measurement of the coherent part of these signals.
magnetic field. Another common modification is to use a The frequency domain transfer function between the hori-
permeable core in the shape of a ring with a toroidal excitation zontal magnetic and electric field is called the MT impedance
winding. Further details can be found in Nielsen et al. (1995). tensor Z:
Because the main magnetic field is so large (around 40 mT), " #
dynamic range and sensitivity issues limit the noise floor of the Ex ðoÞ Zxx ðoÞ Zxy ðoÞ Hx ðoÞ
¼
fluxgate to about 0.01 nT, and the noise curves of fluxgates Ey ðoÞ Zyx ðoÞ Zyy ðoÞ Hy ðoÞ
and induction coils cross at around 500 s period (see, e.g.,
I Current in primary
Primary circuit
Time
B Magnetization, no external field
Core 1
Core 2
B
Magnetization in an external field
Secondary Total induction

circuit
Permeable cores
Voltage in secondary
Figure 8 Principle of operation for a fluxgate magnetometer. Two permeable cores are driven into saturation in opposite directions at frequency f.
A bias associated with an external field along the axis of the cores produces an asymmetry in the internal fields and a total internal field of frequency 2f,
detected in a secondary winding by a phase-locked amplifier.
Site t03 from GoM 2003: 15 min at 32 Hz sampling
Hx
Hy
Ex
Ey
17:46 17:47 17:48 17:49 17:50 17:51 17:52 17:53 17:54 17:55 17:56 17:57 17:58 17:59
Time (HH:MM) on 01/12/03
Figure 9 Sample of MT time-series measurements, in arbitrary units. Note that Hx and Ey are anticorrelated, and Hy and Ex are correlated.
(we will drop the o dependence for clarity). The standard more than a hundred years, there are few observatories collect-
treatment of a transfer function would consider H the input ing electric field data. Secondly, if one examines the MT
and E the output with noise є in the output measurement only relationship

E ¼ ZH þ e T E2
ra ¼
2pm B
which in the classic least squares approach is estimated by
{ one observes that for a given magnetic field and resistivity, the
H E amplitude of the induced electric field decreases with period
b¼
Z
H{ H rffiffiffiffiffiffiffiffiffiffiffiffiffi
2pmra
E¼B
where hH{Ei is the cross spectrum between H and E. The prob- T
lem is that there is noise in H and E, which will bias hH{Hi up which is exacerbated by a falling ra with depth and period. At
and bias Z down. The remote reference technique was intro- long periods, the assumption that the source-field morphology
duced (Gamble et al., 1979) to reduce bias associated with noise is uniform over scales comparable to the depth of penetration
in H by making a second measurement of the magnetic field, Hr, in the Earth breaks down, particularly for the daily variation
sufficiently far away (remote) that the noise sources are inde- and harmonics (e.g., Bahr et al., 1993). Finally, there will be an
pendent, relying on the fact that the magnetic source field is ambiguity in the absolute value of mantle conductivity associ-
coherent over very large distances. ated with unknown static effects in the electric data.
Then,
{
H E 5.07.2.7 Global Response Functions
b¼ r
Z
H{r H One can derive other types of electric response functions,
and with any luck, hH{r Hi is now unbiased. The least squares similar in nature to the MT response, using only the three
method also assumes that noise is normally distributed. components of the magnetic field recorded by geomagnetic
Non-Gaussian noise can be handled with an iterative robust observatories (GDS, responses). Much of this work is still
weight W based on techniques developed by Banks (1969). Figure 10
{ shows an example of overlapping GDS and MT response
Hr WE functions.
b
Z¼ {
Hr WH The vector magnetic field can be written as a gradient of a
scalar potential in the usual fashion
which reduces the effect of outliers. Both the effects of non-
B ¼ mo ∇O
Gaussian noise and nonstationary source-field spectra can be
handled by various robust estimation techniques (e.g., Chave and, following Gauss, the scalar potential can be expanded in
et al., 1987; Egbert and Booker, 1986). These approaches were turn into coefficients of internal (in) and external (en) origin for
generalized into an eigenvalue decomposition for multiple MT spherical harmonics Pn :
sites by Egbert (1997).
The nature of Z depends on the apparent dimensionality of Apparent resistivity
the geology: 102
For 1D, Zxx ¼ Zyy ¼ 0; Zxy ¼ Zyx

For 2D, Zxx ¼ Zyy ¼ 0; Zxy ¼ 6 Zyx
W (m)
For 3D, Zxx , Zyy , Zxy , Zyx 6¼ 0: 101
For 2D structure, Zxx ¼ Zyy ¼ 0 only when the coordinate

system of the measurement is aligned to the 2D geometry of
the geology. With a prior understanding of structural trends or 100
100 101 102
geologic strike, the experimenter will normally arrange for this Phase
to be approximately so. Otherwise, the tensor can be rotated
mathematically to minimize Zxx and Zyy. Conventionally, the 80
impedance with electric field along strike is called the trans- MT
verse electric, or TE mode, and the impedance with across- 60
Degrees
strike electric fields is called the transverse magnetic, or GDS

40
TM mode.
20
0
5.07.2.6 Global Conductivity Studies 100 101 102
If one wants to look deep into the Earth using electric Period (days)
techniques and to study the properties of the deeper mantle, Figure 10 A combination of MT and geomagnetic depth-sounding
then the MT method is not the ideal tool. First, one needs very (GDS) response functions from Egbert and Booker (1992).
long time series of the EM fields, spanning perhaps years, and Magnetotelluric data are available to a maximum of 10-day period
while magnetic observatories have been collecting records for (about 106 s), while GDS data can extend this to periods of many months.
X ao nþ1
n
r If one can assume a simple degree-one field geometry, the
O ¼ ao in þ en Pn ð cos yÞ
n
r a o earlier-mentioned discussion of response functions simplifies
greatly. W becomes simply the ratio of vertical to horizontal fields
where n is the degree of spherical harmonic, ao is the radius of with a colatitudinal trigonometric term. The c-response becomes
the Earth, y is geomagnetic colatitude, and r is the radius of simply c ¼ aoW/2. Banks looked at the geometry of simultaneous
observation. If one chooses geomagnetic coordinates, then the observatory records as a function of geomagnetic colatitude to
analysis is zonal in terms of the main dipole field. The in and en demonstrate the P01 geometry. Having shown this, one can take
can be functions of time or frequency although, like MT, this records from a single observatory and, assuming P01, compute a
kind of analysis is usually done in the frequency domain. geomagnetic response function. Since some observatories have
One can thus define a geomagnetic response for a radially been recording for over a hundred years, this allows response
symmetrical Earth as simply the ratio of induced (internal) to functions to be computed out to periods of at least 6 months.
external fields: The daily variation Sq and harmonics can also be used to
infer geomagnetic response functions, but this is proportion-
in ðoÞ
Qn ðoÞ ¼ ately more difficult because of the more complicated geometry
en ðoÞ
of the source fields. The nature of the source field cannot be
where the frequency domain is again made explicit and will be assumed a priori, but rather must be estimated using an array
implicit in the following. of observatory sites. For example, Olsen (1992) used nearly
In certain circumstances, such as satellite observations, 100 observatories and a spherical harmonic representation of
one has enough data to fit the en and in directly. However, degree 10 to estimate geomagnetic responses for the first 6
most of the time, one just has horizontal (H) and vertical (Z) daily harmonics.
components of B as recorded by a single observatory. We can
obtain H and Z from the appropriate partial derivatives of O
to obtain 5.07.2.7.1 Using magnetic satellites

The use of magnetic satellites to examine global induction
1 @O potentially allows the continents and oceans to be examined
H ¼ mo
r @y r¼ao together. Because satellites sample the entire Earth, one can
X @P n ð cos yÞ transcend the limitations of the sparse and uneven observatory
¼ mo AH, n
n
@y distribution and also sample source-field morphology better
than for a single-site GDS measurement (although simulta-
and neous processing of data from the global observatory network

would sample the source field better than a single satellite
@O
z ¼ mo r ¼ ao could; e.g., Balasis and Egbert, 2006, in practice the observato-
X @r
¼ mo AZ, n P n ð cos yÞ ries are often processed individually). However, because
n satellites move through the static parts of the Earth’s magnetic
field, folding spatial variations into temporal variations
where we have defined new expansion coefficients
within the satellite frame of reference, considerable effort is
AH, n ¼ in þ en required to remove the Earth’s main field, secular variation,
AZ, n ¼ nen þ ðn þ 1Þin crustal field, and daily variation/equatorial electrojet (if one
wishes to process only ring current induction). One tool that
Thus, we can define a new electromagnetic response
attempts to model all these phenomena simultaneously is the
AZ, n comprehensive model (Sabaka et al., 2004). Even so, it is
Wn ¼
AH, n difficult to remove the effects of field-aligned currents in the
auroral zones, and so data are typically processed between
related to Q by
50 geomagnetic latitude, and residual effects of the daily
in n AZ, n =AH, n variation are usually avoided by using only nighttime data.
Qn ¼ ¼
en n þ 1 þ AZ, n =AH, n To obtain response functions from satellite data after
removal of nonmagnetospheric effects, we return to our spher-
The inductive scale length or c-response is, in turn,
ical harmonic expansion of associated Legendre polynomials
ao W n Pml , with Schmidt quasi-normalized spherical harmonic coeffi-
cn ¼ cients representing the internal im m
nðn þ 1Þ l (t) and external el (t) mag-
netic fields as a function of time t:
where the MT apparent resistivity and phase are
X1 X l a lþ1
ra ¼ omo jicn j2 f ¼ arg ðomo icn Þ Fðr; y; fÞ ¼ ao l ðt Þ
im
o
l¼1 m¼l
r
Note that for causal systems, the real part of c is positive,
and the imaginary part is always negative.
l
r
Banks showed that the previously mentioned analysis could l ðt Þ
þem l ð cos yÞe
Pm imf
ao
be simplified because, except for the daily and annual varia-
tions, the magnetic field was dominantly of P01 geometry as a Keeping only the P01 contribution and with r, y, f in geo-
result of the nature of the ring current. magnetic coordinates
a 2

o r conductivity distributions in the Earth. Predicting the response
F01 ðr; yÞ ¼ ao i01 ðt Þ þ e01 ðt Þ P 01 ð cos yÞ
r ao functions that would be observed over a given conductivity
model is the ‘forward problem,’ and estimating conductivity
and so the magnetic induction B is derived from the negative of
from response function data is the ‘inverse problem.’
the gradient in the usual manner
Bðr; y; fÞ ¼ ∇F01 ðr; y; fÞ
5.07.3.1 Forward Model Studies
or expressed as components Br, By, Bf of a spherical coordinate
system: If the forward modeling computation is fast enough, and the
a 3 models simple enough, then it is perfectly reasonable to guess
Br ¼ e01 þ 2i01 cos ðyÞ a conductivity model, compute and compare the predicted
r
a 3 response to the data visually, and estimate ways in which the
By ¼ e01 þ i01 sin ðyÞ model needs to be modified to improve the fit. Forward
r
Bf ¼ 0 modeling requires some understanding of how variations in
the model affect the predicted data but allows geologic preju-
For an analysis using a single satellite, these equations can be dices to be incorporated directly into the process. This
fit to data from each equatorial pass having different values of approach can be automated through stochastic inversion
altitude (r) and geomagnetic colatitude (y), as was done by methods, which test thousands or even millions of models
Didwall (1984) in the first analysis of POGO data, and subse- for compatibility with the data and derive statistical properties
quent studies by Olsen et al. (2003) and Constable and Constable of the acceptable model space. Because it is extremely ineffi-
(2004b), providing estimates of i01(t) and e01(t) at about 100 min cient to test truly random models (using Monte Carlo
intervals. For studies using multiple satellites (e.g., Kuvshinov and methods), various schemes are used to generate models, such
Olsen, 2006; Olsen et al., 2003), a parameterized fit to higher- as simulated annealing (e.g., Dosso and Oldenburg, 1991),
order coefficients can be made. These time series may then be evolutionary or genetic algorithms (e.g., Everett and Schultz,
Fourier transformed into the frequency domain to derive a com- 1993), and Markov chains (e.g., Grandis et al., 1999).
plex geomagnetic response function of frequency: Another role of forward model studies is hypothesis testing.
If the number of possible geologic models is limited, and each
Q01 ðoÞ ¼ i01 ðoÞ=e01 ðoÞ
model is likely to have a distinct geophysical signature, then
Figure 11 provides a comparison of several satellite GDS geophysical models can be built for each hypothesis and com-
responses. puted to see if the predictions are compatible with the
observed data.
5.07.3 Interpretation of GDS and MT Data

5.07.3.2 Parameterized Inversion
The subjects of this section are the mechanisms by which one If it is sensible to approximate the true complexity of geology to
can relate MT and GDS response functions to electric a model having only a few parameters, then there is a whole
1500 Olsen (1999) 1500 Constable and Constable (2004a,b)

Kuvshinov et al. (2005) Kuvshinov and Olsen (2006)
Kuvshinov and Olsen (2006) Kuvshinov and Olsen, corrected
1000 1000
Re{C} Re{C}
C (km)
C (km)
500 500
Im{C} Im{C}
0 0
-500 -500
104 105 106 107 104 105 106 107

Period (s) Period (s)
Figure 11 Long-period geomagnetic response functions computed from magnetic satellite records. Kuvshinov and Olsen (2006) attributed the difference
in the long-period imaginary response between their data and those of Constable and Constable (2004b) as an artifact of Constables’ time-series
analysis. Their correction removes the 3D effects of the oceans on the 1D response functions. Modified from Kuvshinov A and Olsen N (2006) A global model
of mantle conductivity derived from 5 years of CHAMP, Orsted, and SAC-C magnetic data. Geophysical Research Letters 33: L18301.
class of methods that will automatically estimate the value of into a conductivity profile in a spherical Earth s(r) of radius R
these parameters (see, e.g., Gill et al., 1981). In electric and using
electromagnetic methods, the problems are almost always !
nonlinear, so fully nonlinear or linearized approaches need ðr=RÞn ðr=RÞnþ1
sðr Þ ¼ f4 ðr=RÞ
s R
to be chosen. ð2n þ 1Þf ðr=RÞ
where
5.07.3.3 Regularized Inversion
ðn þ 1Þðr=RÞn þ nðr=RÞnþ1
f ðr=RÞ ¼
If one does not have an a priori justification for parameterizing 2n þ 1
the subsurface conductivity model, or such a parameterization
which for n ¼ 1 reduces to
is impractical, then some other constraint on the model
!
space must be imposed to make the problem well posed. ðR=r Þ ðr=RÞ2
A popular, effective, and useful approach is to generate smooth sðr Þ ¼ f4 ðr=RÞ
s R
3f ðr=RÞ
models using some sort of regularization (see, e.g., Constable
et al., 1987). and
2R=r þ ðr=RÞ2
f ðr=RÞ ¼
5.07.3.4 Analytic Least Squares Solutions 3
A true least squares solution for a given model dimensionality We see from Figure 12 that the correction is only significant
is a rare thing; usually, either the forward models must be below depths of about 2000 km or, for typical mantle resistiv-
trivially simple, or the nonlinear forward solution must be ities, periods of about 1 day or longer.
recast as a linear functional. Such a solution exists for the 1D
resistivity-sounding problem (Parker, 1984) and for the 1D MT
problem (Parker and Whaler, 1981).
5.07.3.6 Forward Modeling in Higher Dimensions
Once the realm of 1D or radially symmetrical models is left,
analytic solutions are no longer available for general conduc-
5.07.3.5 MT over a Layered Earth tivity structure, and one is forced to solve the electromagnetic
The simplest extension of the half-space apparent resistivity is a governing equations numerically. For interpretation of MT
horizontally layered Earth. For an N-layered structure num- data on a local scale, a 2D model is useful, especially if the
bered downward, we can define a complex wave number for data are collected along a single profile (as is often the case),
each layer and avoids the complexity of model construction and compu-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi tation in three dimensions. However, for global induction
kj ¼ iomo sj i ¼ 1, . . . , N problems, we are forced to take the step from 1D directly to
3D. A review of 3D modeling is provided by Avdeev (2005).
with layer impedance
Rewriting Maxwell’s equations for frequency o and consid-
mo o ering an applied source of current J0 or magnetic field H0 , we
Gi ¼
ki have
Then, the total impedance of the layers i,. . .,N is given by Zi ∇ E ¼ iomðH þ H0 Þ
and defined for the top of each layer by ∇ H ¼ sE þ J0
Ziþ1 þ Gi tanh ðiki hi Þ which can be solved if the differential operators are approxi-
Zi ¼ Gi
Gi þ Ziþ1 tanh ðiki hi Þ mated numerically. The most straightforward approach is the
finite difference scheme in which E and H are discretized over a
where hi are the layer thicknesses (Schmucker, 1970). Thus, Z1
rectilinear grid of nodes, resulting in a linear system of equa-
is the impedance observed at the surface and is easily com-
tions Ax ¼ b where b represents the boundary conditions and
puted using a recurrence relationship starting with GN ¼ ZN.
applied source fields, x is the vector of field values on the
nodes, and A is a matrix of difference operators. A is large,
5.07.3.5.1 Weidelt’s transformation but sparse, making the inversion of the matrix tractable for
An analytic solution exists for a layered, spherically even large problems if appropriate sparse matrix solutions are
symmetrical Earth conductivity model (Srivastava, 1966; see used. Alternatively, one can use algorithms that do not require
also Parkinson, 1983, p. 313). However, given the greater the storage of the entire matrix A (e.g., Weiss, 2001).
simplicity of the layered MT solution (in the preceding text) A disadvantage of the finite difference approach is that the
and the various inverse solutions available for the flat-Earth grid of nodes must be uniformly rectilinear, making it difficult
approximation (in the succeeding text), it is often desirable to to mesh complex surfaces or structure having a large range of
analyze global data using flat-Earth solutions. Weidelt (1972) length scales. A solution to this is the finite element approach,
provided a compact transformation between the two geome- in which E and H are defined on the edges of triangular (2D) or
tries. If we interpret a GDS c-response c(o) obtained for tetrahedral (3D) elements and interpolated across the elements
spherical harmonic degree n using a layered model to get a using basis functions. A linear system of equations again results
conductivity-depth relationship sðzÞ, this can be transformed in a sparse matrix to be solved. Model construction is perhaps
100
Depth
3000
Model depth, z (km)

10-5
2000
σ/σ
s /s
10-10
1000
0 10-15
0 2000 4000 6000 km
True depth (r)
Figure 12 Weidelt’s transformation for depth and conductivity, for a P01 source-field geometry.
more difficult using finite elements, but automated meshing magnetic storm. In this case, a time-domain approach to
schemes are becoming available (e.g., Key and Weiss, 2006). modeling is required (Hamano, 2002; Kuvshinov and Olsen,
Usually, the equations are solved for E and then a discrete 2004; Velimsky et al., 2003).
approximation to Faraday’s law is used to recover H. A com-
mon approach to numerical modeling of electromagnetic
5.07.3.7 Numerical Inversion of Geomagnetic Data
fields is to break the problem into primary (E0, H0, and s0)
and secondary (Es, Hs, and ss) components: The first thing one needs for an inversion scheme is a measure
of how well a given model fits the data. For practical and
E ¼ E0 þ Es , H ¼ H0 þ Hs , s ¼ so þ ss
theoretical reasons, the sum-squared misfit is favored:
where the primary fields have an analytic solution, such as a
XM
1
whole-space, half-space, or layered model. This primary/sec- w2 ¼ 2
s 2 ½di f ðxi ; mÞ
ondary separation not only has many advantages, particularly i¼1 i
if there is a singularity associated with a man-made source
where
field, but also provides the basis for the third approach to
high-dimensional modeling, the integral equation method. If d ¼ ðd1 ; d2 ; d3 . . . ; dM Þ
the region over which ss is nonzero is a finite volume V, then
ð are M observed data values,
Es ¼ GE ðr; r0 Þss Eðr0 Þ dv0 x ¼ ð x 1 ; x2 ; x3 . . . ; x M Þ
V
and are independent data variables (electrode spacing, frequency,

ð etc.), and
Hs ¼ GH ðr; r0 Þss Eðr0 Þ dv0 m ¼ ðm1 ; m2 . . . ; mN Þ
V
where G are the 3 3 tensor Green’s functions relating the fields are model parameters (layer or block conductivities, layer or
at r to current elements at r0 . The discretization of these equa- block sizes, etc.). Equivalently, the misfit may be written as
tions leads to a linear system where A is now dense, but much 2
b
smaller than for the finite element and finite difference w2 ¼ Wd wd
methods. All of these approaches can be projected onto spher- b is the predicted response from the model
where d
ical geometries in order to model global electromagnetic induc-
tion (e.g., Everett and Schultz, 1996; Grammatica and Tarits, b ¼ f ðx; mÞ
d
2002; Kuvshinov et al., 2002; Martinec, 1999; Uyeshima and
Schultz, 2000; Weiss and Everett, 1998; Yoshimura and and W is a diagonal matrix of reciprocal data errors
Oshiman, 2002).
W ¼ diagð1=s1 , 1=s2 , . . . , 1=sM Þ
A frequency domain approach, which assumes a stationary
excitation of the source field at a given frequency o, may not be The least squares approach attempts to minimize w2 with
appropriate if one wants to consider the effect of a discrete respect to all the model parameters simultaneously. If the data
errors si are Gaussian and independent, then least squares of layers significantly, to approximate the realistic case of
provides a maximum likelihood and unbiased estimate of m, a continuously varying conductivity, or to invert a 2D or 3D
and w2 is chi-square distributed with M N degrees of freedom. model with thousands or millions of parameters will result
To apply linear inverse theory to nonlinear electromagnetic in wildly fluctuating parameters followed by failure of the
data, we need to linearize the problem by expanding f around algorithm. This is because there will be columns of J that
an initial model guess m0 are small, reflecting the fact that large variations in a
model parameter may relate to negligible effects on the
b ¼ f ðm1 Þ ¼ f ðm0 þ DmÞ f ðm0 Þ þ JDm
d combined data.
where J is a matrix of derivatives of data with respect to the One approach, suggested by Backus and Gilbert (1967), is
model to minimize Dm, but this and related algorithms converge
extremely slowly and are called by Parker (1994) ‘creeping
@f ðxi ; m0 Þ methods.’ In any case, the true least squares solutions can be
J ij ¼
@mj pathologically rough – for the 1D MT case, it is known that the
least squares solution is composed of delta functions in con-
(often called the Jacobian matrix) and
ductivity. Almost all high-dimensional inversion today incor-
Dm ¼ ðdm1 , dm2 , . . . , dmN Þ porates some type of regularization, an approach suggested by
Tikhonov and Arsenin (1977), which explicitly penalizes bad
is a model parameter perturbation about m0. Now, our expres- behavior in the model. For example, instead of minimizing w2,
sion for w2 is we minimize an unconstrained functional
w2 kWd Wf ðm0 Þ þ WJDmk2
U ¼ kRm1 k2 þ m1 kWd Wf ðm1 Þk2 w2∗
which we minimize in the usual way by setting the derivatives
where w2∗ is a target misfit that is greater than the minimum
of w2 with respect to Dm equal to zero to get N simultaneous
possible, but statistically acceptable, and Rm is some measure
equations:
of roughness in the model, often taken to be first differences
b ¼ aDm between adjacent model parameters and easily generated by a
where matrix R consisting of (1, 1) entries on the diagonal. Mini-
mizing U has the effect of minimizing model roughness and
b ¼ ðWJÞT W ðd f ðm0 ÞÞ
how far the data misfit is from being acceptable. We substitute
a ¼ ðWJÞT WJ: our linearization around f(m0) and differentiate U with respect
The matrix a is sometimes called the curvature matrix. This to m1 to accomplish this, rearranging the result to get m1
system can be solved for Dm by inverting a numerically, and a directly:
second model m1 ¼ m0 þ Dm found. Because J depends on m, 1
one needs to solve this repeatedly and hope for convergence to m1 ¼ mRT R þ ðWJÞT WJ ðWJÞT W ðd f ðm0 Þ þ Jm0 Þ
a solution. Near the least squares solution, this method will
We need only to choose the trade-off (Lagrange) multiplier
work, but any significant nonlinearity will result in likely fail-
m. The approach of Constable et al. (1987) was to note that for
ure. An algorithm to compensate for this behavior was sug-
each iteration w2 is a function of m and to use 1D optimization
gested by Marquardt (1963). The diagonal terms of the
(simply a line search) to minimize w2 when w2 > w2∗ and to find
curvature matrix are increased by a factor l:
m such that w2 ¼w2∗ otherwise. They called this approach
ajk ¼ ajk ð1 þ lÞ for j ¼ k ‘Occam inversion.’ Although the Occam algorithm is reliable
ajk ¼ ajk for j ¼
6 k and has good convergence behavior, the computation and
storage of J for large models can be limiting, and numerous
For small l, this method obviously reduces to the linearized other regularization approaches using conjugate gradient algo-
inversion. For large l, the diagonal terms dominate and the rithms and approximate derivatives exist (e.g., Mackie and
method reduces to a gradient algorithm that chooses a path in Madden, 1993; Newman and Alumbaugh, 2000; Smith and
the direction of maximum reduction in w2 (the method of Booker, 1991).
steepest descent). The gradient algorithm is robust to nonli-
nearity but is very inefficient in the parabolic region near the
solution (because the gradient gets small; ultimately zero).
5.07.3.8 Analytic Inversion and Estimation of Bounds
By adjusting l to be large far from the solution and small
as we approach the minimum, we have a fairly stable method The 1D MT problem is one of the few problems for which an
of determining a model composed of a small number of analytic least squares solution exists. Parker and Whaler (1981)
parameters. (see also Parker, 1994) showed that the c-response (also called
For nonlinear problems that are truly parameterized (e.g., admittance by Parker, since the electric field is a response to the
finding the concentration and mobility of a charge carrier in a forcing magnetic field) can be cast as a linear functional of a
mineral), the Marquardt method is pretty hard to beat. It also spectral function that is in turn nonlinearly related to conduc-
works fairly well for problems where the number of degrees tivity. Linear inverse theory can be used to find a least squares
of freedom is large, given by MN when the M data are truly solution for a given data set in terms of the spectral function,
independent, and the starting model provides a rough fit to which can then be used to recover a conductivity distribution.
the data. In practice, this means that 1D models consist of a The math is nontrivial, and interested readers are referred to
small number of layers. Any attempt to increase the number the original works, but the result is that least squares solutions
to the 1D MT problem are delta functions of conductivity in an Table 2 Approximate conductivities and resistivities of various Earth
infinitely resistive half-space, a model space called Dþ by materials
Parker. This is one of the reasons why layered least squares
Ionic conduction through water
solutions are not stable as the layer number increases past Seawater at 20 C 5 S m1 0.2 O m
the point of restricting variations in conductivity to be few Seawater at 0 C
3 S m1 0.3 O m
and far between. It may appear that such a pathological Marine sediments 1–0.1 S m1 1–10 O m
solution is not useful, but being able to put a lower Land sediments 0.001–0.1 S m1 10–1000 O m
bound on the possible 1D misfit is very useful when choos- Igneous rocks 105–102 S m1 100–100 000 O m
ing desired misfits for regularized inversions, and the inver- Good semiconductors
sion machinery of Dþ also allows the estimation of a Graphite 10 000–1 000 000 S m1 104–106 O m
maximum depth of resolution available for a given data Galena 10–100 000 S m1 101–105 O m
Pyrite 1–1000 S m1 103–1 O m
set. Also, the algorithm can be used to test the compatibility
Magnetite 20 000 S m1 5 105 O m
of resistivity and phase data and put bounds on missing
Thermally activated conduction
data (Parker and Booker, 1996). Olivine at 1000 C 0.0001 S m1 10 000 O m
Another approach to interpreting response function data Olivine at 1400 C 0.01 S m1 100 O m
is to put bounds on average conductivity over a given region Tholeiite melt 3 S m1 0.3 O m
of the Earth. The earliest attempt at this by Backus and Silicate perovskite 1 S m1 1Om
Gilbert (1968) suffers from being a linearized theory; this
may not be too much of a restriction for seismology, but in
geomagnetic induction, the range of possible conductivities 5.07.4.1 Moving an Electric Charge
is so large as to make this approach essentially useless.
Electric conduction in a material is accomplished by a move-
A nonlinear approach using electromagnetic data was first
ment of charge, which in practice means the movement of
presented via the funnel functions of Oldenburg (1983).
electrons, holes (the absence of an electron), ions, or ion
More recently, Medin et al. (2006) generalized this by using
vacancies. Materials may be classified on the basis of the nature
quadratic programming to solve for electric conductivity and
of the charge carriers and the type of movement. Generally,
the forward responses simultaneously, by making both the
conductivity in a material is given by
data and the forward solution equations constraints in the
optimization. The flexibility of the optimization approach s ¼ nqm
allows many other constraints, such as monotonicity of con-
ductivity with depth and bounds on average conductivity, to where q is the charge of the mobile species, n is the concentra-
be applied. By systematically solving for various averages of tion of the charged species, and m is the mobility or drift
conductivity over depth, the bounds of acceptable averages velocity per applied electric field, |u/E|.
can be discovered.
To date, attempts to put bounds on average conductivity 5.07.4.1.1 Conduction in native metals
have been restricted to the 1D problem. One might try to Conduction in native metals is by means of valence, or con-
extend the machinery to higher dimensions, but bounded duction, electrons that do not take part in crystal bonding
solutions to the 2D MT problem may not exist (Robert Parker, and are loosely bound to atoms. This form of conduction is
personal communication, 2013). very efficient, resulting in room temperature conductivities
for metals of about 107 S m1. The conductivity of metals
decreases with increasing temperature because of thermal agi-
tation of the conduction electrons, which impedes their move-
5.07.4 Electric Conductivity of Earth Materials
ment in response to an electric field and therefore lowers the
drift velocity. Native metals are fairly rare in surface rocks, but
There are three basic conduction regimes for crustal and man-
metallic iron is responsible for high conductivity in the core.
tle rocks:
Metallic conduction electrons have long mean free paths
(1) Ionic conduction of crustal rocks containing water. The because they do not get deflected by the periodic arrangement
mineral grains making up most rock-forming minerals of ion cores and only rarely get scattered off other conduction
are essentially insulators (e.g., quartz’s conductivity at sur- electrons. For metals,
face temperatures is about 1012 S m1), and so the higher
eEt
conductivity of water in pores and cracks determines the u¼
m
conductivity of the rock.
(2) Conduction dominated by minerals that are good semicon- where t is the mean free time between collisions and m is the
ductors or metals, such as graphite, magnetite, and pyrite, or mass of the electron. For metals, n is of order 1028–1029 m3.
native metals. These minerals may occur in trace amounts The mean free path (ut) is of order 1 cm at 4 K, 108 m at room
that nevertheless form a well-connected pathway, or in temperature. Thus, conductivity from metals is thus given by
extreme cases, they may account for the bulk of the rock,
ne2 t
in the case of ore bodies. s¼
(3) Thermally activated conduction in subsolidus or molten sil- m
icates at elevated temperature, either in the deep mantle or This can be thought of as an acceleration under an electric
crustal volcanic systems (Table 2). field (e/m) of charge of density ne for a time period t.
Scattering (which determines mobility m and mean free move ions through a perfect lattice, but defects in the lattice
time t) is by means of phonons (lattice vibrations) and impu- allow the motion of ions and vacancies. Schottky defects are
rities. Phonons go to zero at 0 K and are proportional to lattice vacancies and Frenkel defects are ions in interstitial
temperature above some threshold, so positions. Impurities also constitute defects in the lattice,
both in themselves and because they can promote vacancies
r∝T
to maintain charge balance. These are called point defects,
(That threshold is the Debye temperature, below which compared with dislocations, line, and plane defects, which
quantization of energy causes some vibration modes to be tend not to contribute substantially to conduction (Figure 14).
‘frozen’ out. It is between 100 and 400 K for most metals.) Schottky and Frenkel defects are thermally induced, so the
number of defects, and hence conductivity, increases with
5.07.4.1.2 Semiconduction increasing temperature:
Semiconduction is the result of behavior in between metallic ns ¼ N eEu =kT
conduction and conduction in insulators. It is typical of ioni-
cally bonded binary compounds, such as PbS, ZnS, and MgO. (Schottky defects or vacancies)
There are no free electrons, as such, in a semiconductor, but an
nf ¼ ðNN 0 Þ
1=2 Ei =2kT
applied electric field can supply enough energy to move elec- e
trons from the valence band, across the band gap Eg, into the (Frenkel defects or interstitials) where N is the total number of
conduction band at higher energy levels. As a result of these atoms and N0 is the total number of interstitial sites. Again,
few electrons and the holes they leave behind in the lower however, defects can be generated by impurities. Thus, a diva-
energy levels, moderate conductivities are achieved, between lent cation on a monovalent site induces a cation vacancy to
about 105 and 105 S m1. Sulfide minerals are often semi- maintain charge neutrality.
conductors with quite low resistivities. The mobility of point defects can be related to diffusion, the
For semiconductors, flux JN of atoms crossing unit area in unit time under the
s ¼ neme þ pemh
where n and p are the carrier densities of electrons and holes
(typically around 1019–1023 m3) and me and mh are the respective
mobilities. Carrier concentration is thermally activated and is
n ∝ T 3=2 eEg =2kT

where k is the Boltzmann constant (1.380 1023 J K1 or Intrinsic
8.617 105 eV). Mobility is also only proportional to T, so regime
Conductivity
the exponential term dominates. Such thermally activated pro-

cesses are often termed Arrhenius relationships and produce
linear plots when log(s) is plotted against 1/T. Extrinsic regime
Charge carriers can also be generated by impurities or dop-
ing; for example, a divalent cation substituting on a trivalent
site will have an excess electron that is available for the con-
duction band. These two conduction regimes, one determined
by statistical mechanics and one determined by crystal impu-
rities, are typical of solid-state conduction. The first, purely
thermally activated, variation in conductivity is the same for a
given material and temperature and is called ‘intrinsic’ conduc- Temperature
tion. The second regime, which also has some temperature Figure 13 Extrinsic and intrinsic regimes in semiconductors. Extrinsic
dependence but will vary from sample to sample with the conduction occurs at lower temperature and is a result of electrons or
amount of impurity, is called ‘extrinsic’ conduction. In the holes generated by impurities (or doping). Intrinsic conduction is a result
intrinsic regime, holes must equal electrons because the eleva- only of thermally activated charge carriers associated with the pure
tion of every electron into the conduction band will leave crystal.
behind a hole. Impurities, however, can donate either holes
or electrons to conduction (Figure 13).
Dislocation Impurity
5.07.4.1.3 Ionic, or point defect, conduction

Conduction by point defects dominates when the band gap is
too large for semiconduction and is the result of moving ions Frenkel defect
and ion vacancies through a crystal structure. It is very difficult
to move ionically bonded ions through a crystal lattice, so the
Schottky defect
resistivities of silicate minerals are very high; quartz has resis-
tivities between 1010 and 1014 O m and rock salt has a resistiv-
ity of 1012 O m or so. It would, in fact, be close to impossible to Figure 14 Defects in crystals.
influence of a concentration gradient ∇N. The diffusivity, or Fe•Mg is a Fe3þ ion occupying a Mg2þ site resulting in a net
diffusion constant, D is defined by Fick’s law: charge for the defect of þe.
In spite of the variety in defect types, the dominance of the
JN ¼ D∇N
exponential term in the expression for conductivity means that
Diffusivities are found to vary with temperature, again as a at a given temperature, one mechanism is likely to dominate
thermally activated Boltzmann process: over another. As temperature changes, different activation
energies will cause the dominant conduction mechanism to
D ¼ Do eEd =kT change, so conductivity is often expressed as
where Ed is another activation energy for diffusivity. X
s¼ si eEi =kT
Mobilities are related to diffusivities by the Nernst–Einstein i
equation
The Boltzmann relationship between temperature and con-
kTm ¼ qD
ductivity can be linearized by taking log(s) as a function of 1/T,
so m ¼ qD/kT and the conductivity of, for example, Schottky to form an Arrhenius plot. For a single conduction mechanism,
defects becomes data plotted in this way will be a straight line. For mixed
conduction, a number of lines may be observed as a function
s ¼ ns qm of temperature. Figure 16 shows an example of two conduc-
¼ NeEu =kT qqD=kT tivity mechanisms being expressed as a sample is heated from
¼ Nq2 Do =kTeEu =kT eEd =kT 500 to 1200 C.
(much of the earlier-mentioned material can be found in

Kittel, 1986). The exponential terms dominate, and so to a
first approximation, conductivity is given by 5.07.4.1.4 Pressure
Pressure can also have an effect on conductivity, which may be
s ¼ so eE=kT modeled as
where E is an apparent activation energy associated with both s ¼ so eðEþpDV Þ=kT
defect concentration and mobility. Often, there are several
point defects in the mineral; Figure 15 presents a model of where P is pressure and DV is called the activation volume.
11 different charge-carrying defects in silicate perovskite, and Pressure usually has a minor effect compared with tempera-
olivine would have a similar suite of defects. The description of ture, unless it changes the mineral phase and causes a more
the defects is handled by the Kröger–Vink notation, in which mobile or numerous defect to dominate.
the main character is the actual species involved (a ‘V’ denotes a One of the biggest effects on defect chemistry is oxygen.
vacancy), the subscript is the crystal site for a normal lattice (in Oxidation and reduction are associated with an exchange of
interstitial site is denoted by ‘I’), and the superscript is negative, charge, so oxidation state changes the number of charged
positive, or neutral charge shown as 0 , •, or x, respectively. Thus, defects in a mineral. The model for perovskite in Figure 15
28 x
FeMg
²
FeSi
VO• •
Log (concentration, m-3)
•
24 FeMg e¢
MgSi²
¢
Fe ²
Si VMg
••
SiMg ²²
VSi
•
h
20
OI²
16
-40 -30 -20 -10 0 10
Log (fO2, Pa)
Figure 15 A defect model for silicate perovskite, showing concentrations of various defects as a function of oxygen activity (fO2). Modeled
temperature is 2500 K (2227 C) and pressure is 40 GPa. Redrawn from Hirsch LM and Shankland TJ (1991) Point defects in (Mg, Fe)SiO3 perovskite.
Geophysical Research Letters 18: 1305–1308.
0 5.07.4.1.7 Magma
1200°C
1000°C
800°C
600°C
Magma conductivity is approximately that of seawater and is
associated with diffusion of ions in a highly polymerized struc-
-1 ture. So, it is not surprising that again, one observes a thermally
activated Boltzmann-type conductivity, which for tholeiitic
melt above 10 kbar pressure is approximately (Tyburczy and
Waff, 1983)
Log10 (conductivity, S m-1)
-2
s ¼ 105:332 e1:533 eV=kT
-3
How well a mixture of melt and rock conducts electricity
depends on melt fraction, temperature, composition, and the
geometry of the partial melt (Roberts and Tyburczy, 1999). The
232,000 S m-1 e-1.86eV/kT relationship between geometry and conductivity of two-
-4
component systems is discussed in the next section.
-5
0.435 S m-1 e-0.75eV/kT
5.07.4.2 Binary Mixing Laws
We have seen that rocks are often made up of two components,
-6 one very much more conductive than the other. For example, a
5 7 9 11 13 gabbro could be mostly nonconducting silicate minerals with
Reciprocal temperature (104/K) minor amounts of conductive magnetite. A sandstone can be
Figure 16 Electric conductivity data for cobalt olivine (Co2SiO4), nonconducting quartz saturated with water.
showing two conduction mechanisms expressed as Arrhenius A mantle rock can be subsolidus silicates of several hundred
relationships of two different activation energies. Redrawn from Hirsch ohmmeters with conductive melt between the grains. Rather
LM (1990) Electrical conduction in Co2SiO4. Physics and Chemistry of than make measurements on every possible combination,
Minerals 17: 187–190. binary mixing laws can be used to predict the conductivity of
the bulk rock as a function of the conductive fraction. As one
gives some idea of how defects in a mantle mineral depend might imagine, the physical geometry of the conductive mate-
greatly on oxidation state. rial very much determines how conductive the bulk rock ends
up.
5.07.4.1.5 Aqueous electrolytic conduction The classic binary mixing relationship is Archie’s law, which
Aqueous conduction is associated with ions dissociated in (in spite of the name) is an entirely empirical relationship
water and is arguably the most important conduction mecha- developed to explain the conductivity of sedimentary rocks
nism in crustal rocks. The dominant dissolved salt is sodium containing water:
chloride, and it is clear that the number of charge carriers s ¼ ss þ ðsf ss Þbm
depends simply on the concentration of salt. Mobility increases
with temperature because the viscosity of water decreases with Here, sf is the fluid, or water conductivity, and b is the fluid
temperature, and for seawater concentration (32 g l1), con- fraction. In the case of a saturated rock, b is the porosity, but it
ductivity is approximately is easy enough to express fluid content in terms of saturation
and porosity. Here, we have included the conductivity of the
s ¼ 3 þ T=10 S m1 solid, ss, although for silicate minerals at crustal temperatures,
for temperature T in Celsius. However, for water under pres- this is usually taken to be zero. The exponent m can be derived
sure, this linear relationship breaks down at around 200 C, experimentally from a suite of samples for a given rock type but
and saltwater conductivity peaks at between 300 and 400 C is usually taken to be 2 for sediments. A smaller m corresponds
(Quist and Marshall, 1968). to more efficiently connected pore space.
A model of fluid-filled tubes might be a reasonable repre-
sentation of well-connected melt in a volcanic system:
5.07.4.1.6 Clay minerals
Clays have an important effect on aqueous conduction in 1
s ¼ bsf þ ð1 bÞss
crustal rocks, because they have a surface charge, usually neg- 3
ative, associated with their crystal structure. This creates a Models for binary systems can become quite complicated;
so-called ‘double layer’ of charge, the negative charge on the various geometries are presented by Schmeling (1986).
crystal creating a diffuse layer of positive charge in the electro- However, all possible models must lie between the Hashin–
lyte that is mobile and can contribute to electric conduction. Shtrikman (HS) bounds:
This effect is greatest for low-salinity groundwater and, in

practice, limits the effective resistivity of pore water. A good 1 1 b 1
HS ¼ ss þ b þ
description of this surface electrochemistry is found in Morgan sf ss 3ss

et al. (1989) and a comprehensive treatment of surface con- 1 b 1
HSþ ¼ sf þ ð1 bÞ þ
ductance by Revil and Glover (1997). ss sf 3sf
Normally, when one measures the electric conductivity of were originally the subhorizontal directions (i.e., foliation
fluid-filled crustal rocks as a function of pressure, increasing direction) and lower in the direction across the beds. Anisot-
pressure decreases pore volume, and so conductivity decreases. ropy ratios (sy/sz) of up to 2–10 are possible in this way.
However, if the dominant conduction is by conductive min-
erals, such as graphite or magnetite, increasing pressure can
5.07.4.4.3 Structural anisotropy
press these minor minerals into better electric contact, in which
Structural anisotropy is caused by macroscopic features that
case conductivity will increase with pressure. This is an impor-
would normally be resolved by geologic mapping but that are
tant diagnostic of conduction mechanism (Shankland
too small in relation to the electric measurements to be
et al., 1997).
resolved. For example, interbedded, meter-thick sedimentary
horizons of different resistivity that are buried 10–100 m deep
5.07.4.3 Polarization will appear anisotropic to a surface-based electric method but
not to borehole logging.
We have discussed various conduction mechanisms – metallic, Although in principle all three principal directions of con-
semiconduction, ionic conduction, and electrolytic conduction – ductivity can vary, if rocks are anisotropic, it is usually uniaxial
all of which can be found in rocks and minerals and within the anisotropy – that is, only one of the three directions differing
apparatus we used to make measurements. When one attempts from the other two. It is interesting to consider the relationship
to pass electric current between two different types of conductors, between the mathematical description and the physical inter-
there will be a polarization at the boundary. The classic example pretation of the six possibilities associated with uniaxial anisot-
of this will be a metallic electrode making contact with the ropy, shown in Figure 17.
ground or a laboratory sample. Electrons flowing through the
circuit set up an electric field in the electrolyte. Ions drift through
the fluid in response to this field until they encounter the elec- 5.07.4.5 Laboratory Measurement of Conductivity
trodes, where they have to stop. The initial effect of this is to
increase the impedance of the fluid (and change the phase of the While the fundamental physical property of rocks and min-
conduction current) by depleting the ions available for transport erals is conductivity (which may be expressed as equivalent
and the distance over which they can migrate. A second effect is resistivity), our laboratory equipment is designed to measure
that the ions may then participate in reversible and irreversible circuit resistance. Resistivity r and resistance R for a regular
electrochemical reactions with the electrodes. prism are related by
R ¼ rL=A
5.07.4.4 Anisotropy where L is the length of the sample and A is the cross-sectional
area of the electrode-covered faces. It can be seen that if one
So far, we have only considered the case in which conductivity
takes a cubic meter of material (L ¼ 1; A ¼ 1), the series resis-
is isotropic. In general, conductivity may depend on direction:
tance in ohms across two opposite faces of the cube will be the
2 3
sx same number as the resistivity in ohmmeter.
s¼4 sy 5
sz x s s s s
y z
where most commonly x, y, and z correspond to two horizontal
and the vertical direction, although any general case can be s
rotated into three principal axes. s
As with porosity, the nature of anisotropy depends on scale, s
which can loosely be classified as crystallographic anisotropy,
textural anisotropy, and structural anisotropy as the scale
length gets progressively larger. s
s
s
5.07.4.4.1 Crystallographic anisotropy
Crystallographic anisotropy is associated with variations in
conductivity of the mineral itself. The principal components s
of crystallographic properties must lie along the crystal axes, s
and so it is sufficient to measure and specify conductivity along s
these directions.
Figure 17 Relationship between geologic and mathematical
5.07.4.4.2 Textural anisotropy descriptions of anisotropy. Here, the more conductive material is shown
in gray. Rodlike conductive structure (left) corresponds to the
Textural anisotropy is associated with the fabric of the rock.
background material (s//) being less conductive, and sheetlike conductive
A common example of this type of anisotropy is found in structure (right) corresponds to the background material being more
sedimentary rocks that have well-developed bedding, either conductive. Modified from Everett ME and Constable S (1999) Electric
as a consequence of platy or tabular mineral grains that pref- dipole fields over an anisotropic seafloor: Theory and application to the
erentially fall flat during deposition or by fine-scale interbed- structure of 40 Ma Pacific Ocean lithosphere. Geophysical Journal
ding. In this case, conductivity will always be higher in what International 136: 41–56.
Such a measurement can be made in practice on a regular two-electrode measurements will separate from the impedance
prism (such as a rock core or mineral sample) when electric of the sample. Grain surface impedance can also be separated
current is passed through the sample by means of two metal from grain-interior impedance this way (e.g., Tyburczy and
plates pressed to the end faces. This is called the ‘two-electrode Roberts, 1990; Figure 20).
method,’ but it suffers from the limitation that the contact When measurements are made at high temperature to pre-
resistance of the electrodes must be much lower than the dict the behavior in the deep Earth and to establish an activa-
resistance of the sample. If L is small compared with A and tion energy for the conducting species, care must be taken to
the sample is of relatively high r, this approximation is accept- control the chemical environment of the sample so that it is
able, because the contact resistance of the electrodes will be less not altered during measurement and that the conditions are
than the resistance through the sample. Otherwise, the solu- applicable to the natural environment of the sample. The most
tion is the ‘four-electrode method’ (Figure 18). This solves the obvious problem is oxidation during heating; even laboratory-
problem of contact resistance at the electrodes by measuring a grade inert gases have enough contaminant oxygen to place
potential difference across the center of the rock rather than the some minerals outside their stability fields.
ends. Because modern voltmeters have very high input imped- For low-pressure experiments, a controlled mixture of CO:
ances, the contact resistance on the potential measurement CO2 or H:CO2 is passed over the sample. For a given temper-
circuit does not have a significant impact on the potential ature and gas mix ratio, there is a known equilibrium concen-
across the sample, and the resistivity can be computed from tration of oxygen developed; any excess is consumed by the
r ¼ DVA/(Is) where s is the potential electrode spacing. Practi- reducing agent (CO or H), and any deficiency is liberated by
cal considerations often force the issue in favor of the two- the oxidizing agent (CO2). For example,
electrode method (e.g., measurements made on small samples
2CO þ O2 Ð 2CO2
at high temperature and pressure).
Polarization at the sample-electrode interface is controlled so even if the starting gases contained contaminant oxygen, it
by using an AC current, usually of around 1 kHz. Since polar- will be consumed until only the equilibrium amount remains.
ization creates a capacitive response, it is wise to measure the Other problems associated with the chemical state of the sam-
phase, as well as the magnitude, of the conduction current ple are iron loss to platinum electrodes and migration of
versus applied potential. A technique called impedance spec- chemical species such as silicon between constituents of a
troscopy exploits this capacitive effect to examine the sample multiphase sample.
more closely, by varying the frequency of the applied voltage Since the two-electrode method is usually used for relatively
and monitoring both the in-phase (real) and out-of-phase resistive samples, and often at high pressure, care has to be
(quadrature or imaginary) current. An impedance diagram is taken to ensure that the dominant conduction path is through
constructed, plotting imaginary versus real impedance for var- the sample and not through the sample holder and surround-
ious frequencies (such a diagram is also called a Cole–Cole ing apparatus. Figure 21 shows an example of a furnace
plot). An equivalent circuit of a resistor and capacitor in paral-
lel will create a semicircular arc on such a plot (Figure 19). If
several conduction mechanisms are present, several arcs will be
R2 R3
generated, and in particular, electrode contact impedance in
R1
Two-electrode method Four-electrode method C2 C3
C1
V V
-Imaginary impedance
Grain
I I Grain interiors boundaries Electrodes
Rc Rr Rc Rc+1/3Rr 1/3Rr Rc+1/3Rr
ω
V
Rc Rc ωR2C2=1 ωR3C3=1
ωR1C1=1
V
I I
Real impedance R1 R1+R2 R1+R2+R3
Figure 18 Two- and four-electrode methods of measuring rock
resistivity, showing equivalent circuit elements. Rc is the contact or Figure 19 Impedance spectroscopy can identify the different series of
electrode resistance; Rr is the rock or sample resistance. In the conduction paths in the sample because they can be associated with
two-electrode method, the potential drop across the current electrodes is different capacitive effects. The top diagram shows an equivalent series
measured along with the potential drop across the sample. In the circuit, and the lower diagram shows the ideal measurement that
four-electrode method, the potential drop across the current electrodes is would be made on a rock sample. Frequency increases toward the origin.
not part of the voltage measuring circuit, and the potential electrode Redrawn from Tyburczy JA and Roberts JJ (1990) Low frequency
impedance simply adds to the already high input impedance of electrical response of polycrystalline olivine compacts: Grain boundary
the voltmeter. transport. Geophysical Research Letters 17: 1985–1988.
15
Imag. (Z), kW
Log10(Hz)
10
4 1 -3
0 -1 -2
2
5
3
0
0 10 20 30 40 50 60
Real (Z), kW
Figure 20 Example of impedance spectroscopy on a sample of San Carlos olivine at 1200 C and an fO2 of 10–15 Pa. Data are from Tyburczy and
Roberts (1990). We have highlighted the highest frequency (103–104 Hz) mechanism, interpreted as grain-interior conduction, with a 12 kO radius arc
centered on the real axis. Intermediate frequency (10–2 to 102 Hz) impedance is associated with grain boundary conduction, and lowest frequency
impedance as polarization on the electrode surfaces.
Rubber stopper
Thermocouple wires
Upper nylon guide assembly
Compression spring
CO–CO2 outlet
'o' rings
Water cooling jacket Alumina thermocouple tube
Thermocouple Upper alumina electrode rod

wires
Pt -Pt10%Rh thermocouple
Iridium electrodes
Platinum
resistance Sample
furnace
Lower elecrode rod
Primary alumina Alumina thermocouple tube

tube
Water cooling jacket
Lower nylon
guide CO–CO2 inlet
Thermocouple wires
Figure 21 Gas-mixing furnace for making electric conductivity measurements at room pressure and high temperature (after Duba et al., 1990). The
mixture of CO and CO2 is controlled by separate mass flow controllers for each gas. The vertical size is about 60 cm.
designed to control the oxygen activity surrounding the sample 5.07.4.7 Thermopower Measurements
and to minimize conduction leakage around the sample. The
In a gas-mixing furnace such as shown in Figure 21, the assem-
electrodes in contact with the sample are made of iridium,
bly can be moved up or down to produce a temperature
rather than platinum, to minimize iron loss. The thermocou-
gradient across the sample that can then be measured using
ple leads, which allow the temperature to be measured on both
the two thermocouples. The resulting thermoelectric effect,
sides of the sample (important for thermopower measure-
also known as the Seebeck effect, can be measured as an
ments), also serve as electrode leads for a 2-electrode measure-
potential difference between the two electrode wires. The ther-
ment. For thermopower measurements, the entire assembly
mopower Q is defined as
can be moved up or down in the furnace/cooling system to
create a temperature gradient across the sample. DV
Q ¼ lim
DT!0 DT
The mobility of defects is less on the cold side of the sample,
5.07.4.6 High-Pressure Measurements and so an excess ‘condenses’ there. For a charge-carrying defect,
this results in an electric field across the sample that can be
The surface pressure gas-mixing furnace is unsurpassed in
measured with a suitable high-impedance voltmeter. The sign
terms of making measurements under finely controlled fO2,
convention for Q is such that the sign of the thermopower
but if one wishes to examine electric conductivity of lower
mantle phases (minerals), these need to be made within a
high-pressure apparatus using solid chemical buffers instead ZrO2
of gas mixes.
There are three basic types of high-pressure apparatus. The
first is the piston and cylinder apparatus, which is exactly Shield (Mo)
what it says and is suitable only for relatively low pressures, MgO
Electrode
although it does have the advantage that gas mixes could be wire
used to control fO2. The second is the multianvil cell Sample
(Figure 22), which uses a set of carefully shaped steel, tung- Electrode (Mo)
sten carbide, and/or sintered diamond anvils to turn the force Furnace
of a 1000 þ ton uniaxial press into a triaxial confining pres-
sure of up to several tens of gigapascals (40 GPa is about Thermocouple/ Al2O3
1000 km deep). Figure 23 shows a sample capsule used for electrode wire
conductivity measurements of high-pressure olivine phases in Mo Alumina
a multianvil cell. Finally, the diamond anvil cell allows sim- cement
ilar or greater pressures to the multianvil cell with a much
smaller experimental setup and the possibility of optical inter- Figure 23 High-pressure conductivity cell for use in a multianvil press.
action with the specimen. X-ray experiments can be made Reproduced from Xu YS, Poe BT, Shankland TJ, and Rubie DC (1998)
through the diamond, and lasers used to locally heat the Electric conductivity of olivine, wadsleyite, and ringwoodite under
sample (Figure 24). upper-mantle conditions. Science 280: 1415–1418.
(i) (ii) (iii) Tungsten

Uniaxial
Upper guide block carbide
press
anvil
Guide blocks Split-sphere

anvils Anvils 4
6 Second-stage
assembly Presure medium
2000 Ton Jack Spacer

2 1
Pre-formed
gasket
Lower guide block
32 mm
Figure 22 The multianvil cell. From left to right: (i) the uniaxial press and the split-sphere anvils, (ii) close-up of the split-sphere anvils
showing the location of the second-stage assembly, and (iii) the second-stage assembly of eight tungsten carbide cubic anvils and an octahedral
pressure medium. Reproduced from Libermann RC and Wang Y (1992) Characterization of sample environment in a uniaxial split-sphere
apparatus. In: Syono Y and Manghnani MH (eds.) High Pressure Research: Applications to Earth and Planetary Sciences, pp. 19–31. Washington, DC:
American Geophysical Union.
Laser or X-ray 0
Orthopyroxene
Force
Support
1200 °C
Clinopyroxene
Garnet
Diamond
anvils or 2.5 mm 200 Olivine
cutlets
Gasket
Sample (0.1– 0.2 mm)
10
Support
Figure 24 A diamond anvil setup designed to measure precise
Pressure (GPa)
strains using an optical technique (from Meade and Jeanloz, 1987).
Depth (km)
The size of the sample area is less than 1 mm in diameter. 400
1400 °C
agrees with charge-carrying defect. Thermoelectric power is

inversely proportional to the concentration of charge carriers Wadsleyite
and is equivalent to the entropy per charge carrier. Expressions Majorite
for this are of the form (Tuller and Nowick, 1977)

k ð1 cÞ S 20
Q¼ ln b þ Ringwoodite
q c k 600
Here, c is the fraction of sites that contain a charge carrier, q is 1600 – 2000 °C
the charge of the defect, b is a degeneracy factor (for electrons,
b is usually taken to be equal to 2 to account for the two Ca–P
Magnesio-
wüstite
possible spin states), S is the vibrational entropy associated Perovskite
with the ions (usually considered negligible), and k is the Al–P
Boltzmann constant. Because thermopower is dependent +
St
only on concentration, and not on mobility, it provides an
important tool for the study of charge transport in minerals. 0.0 0.2 0.4 0.6 0.8 1.0
Volume fraction
5.07.4.8 Conductivity of Mantle Minerals Figure 25 Mantle mineralogy, after Ito and Takahashi (1987). Olivine is
the major component of the upper mantle. It is a orthorhombic
Mantle mineralogy depends to a large extent on the increase in magnesium iron silicate, (Mg1–xFex)2SiO4, where x is typically 0.1 in
pressure with depth and is inferred from a combination of the mantle. Wadsleyite and ringwoodite are high-pressure phases
mantle xenoliths, models of seismic velocity, estimates of of olivine. Garnets are cubic aluminum silicates of the general composi-
bulk chemistry, and high-pressure laboratory studies. Figure 25 tion X3Al2Si3O12, where X can be magnesium (pyrope), divalent iron
shows such a model of mantle mineralogy in pictorial form. (almandine), manganese (spessartine), or calcium (grossular)
Figure 26 shows laboratory studies of water-free electric (neglecting garnets where the Al is replaced by other trivalent cations).
conductivity for all the major minerals in the mantle, repre- Majorite is a high-pressure tetragonal phase of (Mg,Fe)SiO3 garnet.
sented as Arrhenius fits to the original data. Most of these data Orthopyroxene is typified by enstatite, (Mg1–xFex)SiO3, a chain silicate
of orthorhombic structure. Clinopyroxene is typified by the
were collected in a multianvil press, where large (by compari-
monoclinic diopside series Ca(Mg,Fe,Mn)Si2O6, another chain silicate.
son with diamond anvil cells) samples can be made and solid-
Magnesiowüstite (also called ferropericlase) is magnesium iron
state buffers used to control fO2. Most minerals have activation oxide (Mg1–xFex)O. Stishovite (St) is a high-pressure phase of silica
energies between 0.5 and 2 eV (1 eV ¼ 1.6 1019 J mol1). It (quartz), or SiO2. Silicate perovskite is a high-pressure distorted cubic
is interesting to note that upper mantle minerals have activa- form of magnesium iron silicate, Mg1–xFexSiO3. Ca–P and Al–P are
tion energies of around 1.5 eV, while lower mantle minerals CaO- and Al2O3-rich phases, respectively.
(magnesiowüstite and silicate perovskite) have activation ener-
gies of 0.7 eV. Whether there is a reason or not for the high- the next most abundant upper mantle mineral, clinopyroxene,
pressure phases to have lower activation energies, the implica- has a conductivity similar to olivine, it is fairly safe to build
tion is that the lower mantle will be more uniform in conduc- models of mantle conductivity on studies of olivine.
tivity given similar variations in temperature. Olivine, an orthorhombic mineral, is mildly anisotropic.
Figure 27 shows measurements of olivine anisotropy, made on
samples from mantle xenoliths found in San Carlos Apache
5.07.4.9 Olivine Conductivity
Indian Reservation, Arizona. The [0 0 1], or c-axis, is the most
Olivine attracts special attention from the geophysics commu- conductive but still barely more than a factor of 2 greater than
nity because it dominates the mineralogy of the upper mantle, the other axes. The [1 0 0] (a) axis is of intermediate conduc-
a region that is easiest to study and closest to the crust. Because tivity. This is, perhaps, unfortunate, because seismic and
Temperature (°C)
2000 1800 1600 1400 1200 1000 800
Ferro
peric
lase (m
agne
siowü
100 stite)
Pero
vskit
Electrical conductivity (S m-1)
e
10-1
Wa
ds
ley
ite
/rin
Cl gw
ino oo
10-2 py
ro
dit
e
xe
ne
Ma
jor
ite
10-3 Ga
rne
t
Or
10-4 Ol
th
op
ivi yro
ne xe
SE ne
O3
10-5
5 6 7 8 9 10
Reciprocal temperature (104/K)
Figure 26 Laboratory studies of conductivity in nominally anhydrous mantle minerals, expressed as Arrhenius model fits to data. Solid lines represent
the temperatures over which measurements were made; the broken and dotted lines are extrapolations to higher and lower temperatures.
Ferropericlase data (dark blue) are from Dobson and Brodholt (2000). Perovskite results (dark blue) are from Xu et al. (1998, broken line) and Yoshino
et al. (2008a, dotted line). Ringwoodite (black) is from Yoshino et al. (2008b, broken line) and wadsleyite (also black) is from Yoshino and Katsura
(2012, dotted line). Majorite garnet (red) is from Yoshino et al. (2008c) for a pyrolite composition (dotted line) and mid-ocean ridge basalt
composition (broken line). Garnet (light-blue) results are from Xu and Shankland (1999, broken line) and Dai and Karato (2009a, dotted line).
Orthopyroxene (green) is from Xu and Shankland (1999, broken line) and Zhang et al. (2012, dotted line). Clinopyroxene (pink) is from Xu and
Shankland (1999). Olivine SEO3 is from Constable (2006) for iron–wüstite (black, dotted line) and quartz–fayalite–magnetite (black, broken line) models
of mantle oxygen fugacity.
The SO2 model is

Log10(Conductivity, S m-1)
-2.0 T = 1100 °C [001]

[100] s ¼ 102:40 e1:6 eV=kT þ 109:17 e4:25 eV=kT S m1
-2.5 [010]
where the first term corresponds to low-temperature behavior
-3.0 and the second term corresponds to a conduction mechanism
that becomes dominant above 1300 C. SO2 served as a useful
-3.5 benchmark for some time but was often found to underpredict
mantle conductivity. A new model, SEO3, was created based on
-4.0 our understanding of the conduction mechanisms in olivine
-11 -9 -7 -5 -3 -1 1 3
(Constable, 2006). The relationship between the major defect
Log10(fO2, Pa) population in olivine and fO2 is thought to be described by
Figure 27 Data (symbols) and model fits (lines) for electric 00 00 00
conductivity of San Carlos olivine along the three crystallographic axes as 8Fe
Mg þ 2O2 Ð 2VMg þ VSi þ 4OO þ FeMg :
a function of fO2 at a temperature of 1200 C (data from Du Frane et al.,
2005). Anisotropy is not strong, and the crystal axis most likely to be Many studies indicate that the dominant charge carriers are
00
aligned in the mantle, [1 0 0], is of intermediate conductivity. Fe•Mg and VMg. For samples buffered by pyroxene, this model
00
predicts that [Fe•Mg] and [VMg] will each have a dependence on
deformational models of the upper mantle suggest that this fO2 of the form
axis will be aligned in the direction of plate motion. h i h 00
i
Fe Mg ¼ aFe f O2 ;
1=6 1=6
SO2 (Constable et al., 1992) is a quantitative model VMg ¼ aMg f O2
of olivine conductivity based on a geometric average of high-
temperature single-crystal measurements and lower-temperature where the preexponential terms a are constant and [] denotes
measurements on polycrystalline olivine rocks (which would concentration. For mixed conduction, total conductivity is
melt above 1250 C, fluxed by the pyroxene content). given by
s ¼ sFe þ sMg SO2 (Figure 29), because it is based on the Boltzmann rela-
h i h 00 i tionships rather than parametric fits to data. Furthermore, it
¼ Fe Mg mFe e þ 2 VMg mMg e allows explicit designation of the fO2 conditions; although the
quartz–fayalite–magnetite buffer is often used for the upper
where m are the respective mobilities for the charge carriers and mantle, it is possible that at depth fO2 may be lower and close
e is the charge on the electron or hole. to the iron–wüstite buffer (McCammon, 2005). The effect of
Thermopower measurements provide a mechanism of esti- pressure is neglected in these models, but Xu et al. (2000)
mating concentration independently of mobility, and showed that it is small compared with the sample-to-sample
Constable and Roberts (1997) created a model of conduction variations associated with high-pressure measurements. (In
in olivine based on a combined thermopower/conductivity Figure 29, the QFM curves are very slightly different than in
data set. Figure 28 shows the data set used for this – a dunite Constable (2006) because they use the fugacity model of
from San Quintin in Baja California, Mexico. Natural rocks are O’Neill (1987) to replace a simpler approximation used in
useful samples because the crystallographic orientations are the original paper.)
usually randomized, and there are enough minor phases to
buffer silica. It can be seen that the effect of fO2 has the same
magnitude as a 100 K change in temperature. Iron content also 5.07.5 Global Conductivity Structure
has an effect on polaron conduction, but Fe content in mantle
olivines is quite constant at about 10%. This model supports In this section, we will examine what is known about the
earlier predictions (Schock et al., 1989) that below 1300 C, electric conductivity of the Earth’s interior, from both field
small polaron hopping (Fe•Mg) dominates the conduction measurements and laboratory studies. Field studies include
00
mechanism and, above 1300 C, magnesium vacancies (VMg) long-period MT soundings using temporary instrument instal-
account for conductivity. lations, GDS studies using magnetic observatory data, and,
Constable (2006) used the mobilities and concentrations as most recently, induction studies using magnetic satellite data.
a function of temperature and fO2 from the joint conductivity/
thermopower model to create a quantitative model of conduc-
tivity that is perhaps more reliable at high temperature than 5.07.5.1 The Oceans and Crust
The most significant conductivity feature on the surface of the
Earth is the world ocean. Fortunately, the ocean is also the
500 most easy feature to characterize. Global maps of bathymetry
1000 °C are available, and seawater conductivity is well defined as
a function of salinity (nearly constant) and temperature.
Thermopower (mV K-1)
400
1100 °C Figure 30 shows an example of the electric conductivity struc-
300 ture of the upper 1000 m of a site in the Pacific Ocean. The
surface water is warm and more than 5 S m1, but both tem-
200 perature and conductivity drop rapidly below the thermocline,
in this case at a depth of about 70 m. By a depth of 1000 m, the
1200 °C
100 conductivity is close to that observed at full ocean depth, and a
value of 3.3–3.5 S m1 is usually used to represent ocean
0 conductivity.
The rocks of the oceanic crust are, by and large, also fairly
10-8 10-6 10-4 10-2 100
easy to characterize because for the most part, the genesis of
oceanic crust is uniform and the metamorphic history fairly
simple. Boreholes to a depth of a kilometer or so, along with
1/6
marine-controlled source EM studies, provide in situ measure-
1200 °C ments of conductivity, which, combined with our knowledge

10-3 of oceanic geology gleaned from dredging and ophiolite stud-
1100 °C ies, allows us to paint a fairly good picture of electric structure
in the oceanic crust (Figure 31). Away from ridges, the con-
ductivity is basically a measure of the porosity of the volcanic
1000 °C and igneous rocks that make up the crust. Near ridges conduc-
10-4 tivity is increased by the presence of magma and hydrothermal
fluids in the crust.
In stark contrast, the electric structure of the continental crust
10-8 10-6 10-4 10-2 100 (which includes the continental shelves) is very much more
Oxygen partial pressure (Pa) complicated, undoubtedly because it is better studied than the
Figure 28 Data (symbols) and model fits (lines) for thermoelectric oceanic crust, but mostly because the geologic history is much
power and conductivity data for San Quintin dunite, as a function of fO2 more complicated and the variation in water content (and
for three temperatures. Reproduced from Constable S and Roberts salinity) is much larger. Data for crustal conductivities on the
JJ (1997) Simultaneous modeling of thermopower and electrical continents come mainly from broadband and long-period MT
conduction in olivine. Physics and Chemistry of Minerals 24: 319–325. soundings, with some data from large-scale controlled-soure
Reciprocal temperature (104/K) Temperature (C)

10 9 8 7 6 16001500 1400 1300 1200 1100 1000 900 800
10-1 101
FM
SE
-Q B
O3
O3
-
SE
QF
M
10-2 2 102
SO
10-3 103
Resistivity (Wm)
3-IW SE
S EO O3
-IW
10-4 SO 104
2
10-5 105
10-6 106
700 800 900 1000 1100 1200 1300 1400 1500 1600 5 6 7 8 9 10
(a) Temperature (C) (b) Reciprocal temperature (104/K)
Figure 29 The SEO3 olivine conductivity–temperature model for two different fO2 conditions, quartz–fayalite–magnetite (SEO3-QFM) and iron–wüstite
(SEO3-IW) (Constable 2006). For comparison, the SO2 model of Constable et al. (1992) is also shown. The models are represented in both 1/T
space (b) and temperature space (a).
XBT cast, East pacific rise 9° North
Temperature (C)
0 4 8 12 16 20 24 28
0
200
400
Depth (m)
600 Temperature Conductivity
800
1000
Deep seafloor values

1200
0 1 2 3 4 5 6 7
Figure 30 Expendable bathythermograph (XBT) taken in the Pacific Ocean near the East Pacific Rise at 9 N. The temperature profile has been
converted to electric conductivity. The ‘o’ symbols show the values near the seafloor in the deeper ocean (4000–5000 m).
Everett et al. (2003) formalized this approach to generate a

electromagnetic (CSEM) and resistivity studies. However,
global crustal conductance map, using the global sediment
neglecting important details and ignoring unusually conductive,
thickness map of resolution of 1 1 compiled by Laske and
volumetrically small, and often economically significant
Masters (1997). This map contains sediments discretized in up
minerals, we can make some gross generalizations:
to three layers: a surface layer up to 2 km thick, an underlying
• Continental shields: 103–106 Om layer up to 5 km thick, and, where necessary, a third layer to
• Continental sediments: 10–100 Om make up total sediment thickness, which often exceeds 10 km.
• Marine sediments: 1–100 Om. The sediment map was augmented by a 1 1 topographic/
Electrical resistivity (Wm)

0.1 1 10 102 103 104 105
Seawater
0
Extrusives
1 Intrusives
3 Gabbro
Kilometers below seafloor Ridge Off-axis (magma at MOR)
4
5
6
Lithospheric mantle
Break in scale (depleted olivine,
7 pyroxene)
100
200 Asthenospheric mantle
(olivine, pyroxene)
300
400
500 Transition zone
(high pressure olivine phases,
600 garnet, pyroxene)
700
Lower mantle
800 (perovskite, wüstite)
Figure 31 Electric conductivity structure of the oceanic crust and mantle, based on marine CSEM to depths of about 50 km and marine MT below that.
Global surface conductance map, logarithmic scale [log10 S]
4.5
3.5
2.5
1.5
Figure 32 Global electric conductance generated by the crustal algorithm of Everett et al. (2003).
bathymetric map. Continental and oceanic sediments were conductance, but in areas such as the Gulf of Mexico, Arctic
mapped heuristically to conductivity and summed over depth Ocean, and Mediterranean/Caspian/Black Sea, accumulated
to create a surface conductance map (Figure 32). The sediment has a conductance comparable to the oceans and
previously mentioned algorithm is undoubtedly arbitrary and sediments drastically alter the shape of the continent/ocean
could be improved upon but improves on simple ocean depth function.
as a proxy for conductance. Overall, the sedimentary sections For frequencies where the skin depth is large compared with
defined in this way contribute only 10% to the total surface the thickness of a layer, a surface layer can be characterized by
its conductance only, because its thickness cannot be resolved color restrict the models to be monotonically increasing in
and attenuation of EM energy is determined entirely by the conductivity with depth.
conductivity–thickness product, or conductance. Computa- These models are representative of the many mantle con-
tions in three dimensions often exploit this phenomenon ductivity profiles published in the literature and are in good
using a thin-sheet algorithm (e.g., Kuvshinov et al., 1999). agreement with the anhydrous mineral conductivities shown
in Figure 26. One of the most well-resolved features is the jump
in conductivity to 2–3 S m1 in the lower mantle, entirely
consistent with the laboratory measurements on silicate perov-
5.07.5.2 The Mantle
skite. The order of magnitude increase in conductivity in the
As can be seen from Figure 11, global response functions from lowermost mantle is just within the range of thermally activated
satellite and observatory measurements are in broad general conduction in perovskite or could be the influence of other
agreement. This is further illustrated in Figure 33, which shows mineral phases. However, this increase in model conductivity
the data set used in an inversion study by Medin et al. (2006). is poorly constrained, relying on the longest-period measure-
These data consist of a combination of stacked global observa- ments having largest uncertainty, and probably represents a
tory GDS responses (Constable, 1993a), which relied on lower bound on conductivity because the smoothness penalty
assumptions of P01 geometry, and European observatory minimizes slope. The order of magnitude increase of conduc-
responses of Olsen (1999), which allowed more complicated tivity in the transition zone is in general agreement with the
source-field geometry and thus a broader bandwidth. Medin more recent water-free laboratory measurements on the high-
et al. excluded four admittances that failed to satisfy the pressure phases wadsleyite and ringwoodite.
assumption of one dimensionality, three of which they had The uppermost mantle above the transition zone is likely to
reason to believe did not meet the source-field requirements. be quite heterogeneous and dependent on local tectonic set-
Figure 34 shows various inversions of this data set. The ting, and although Medin et al.’s data are all derived from land
green lines represent the Dþ model of Parker and Whaler observatories and thus avoid the oceans, the continental crust
(1981), scaled by 105 and fitting the data to RMS 1.044. The can be locally quite conductive. This part of the model is the
red line is a smooth, first-derivative regularized inversion fit- least likely to have a monotonically increasing conductivity,
ting the data to RMS 1.2, and the blue line is a similar inversion especially if fO2 decreases with depth, but the bounds on
but allowing unpenalized jumps at the 440 and 670 km seis- average conductivity span the range for dry olivine conductiv-
mic discontinuities. Finally, the yellow boxes are bounds on ity. While one-dimensional global response functions are in
average conductivity at the 90% confidence level over the general agreement with each other and with laboratory studies,
depth intervals 0–418, 418–672, and 672–1666 km, derived individual response functions vary considerably (e.g., Egbert
by Medin et al. (2006). The broader bounds put no restrictions and Booker, 1992; Lizarralde et al., 1995; Schultz et al., 1993;
on the models, and the tighter bounds shown in the darker Tarits et al., 2004), with nearly two orders of magnitude
3000
2500
Real
2000
1500
C (km)
1000
500
Imaginary
0
-500
-1000
4 5 6 7 8 9
log (Period, s)
Figure 33 Response function used by Medin et al. (2006), consisting of the slightly edited data sets from Olsen (1999, blue symbols) and Constable
(1993a, green symbols). The error bars are as reported by the authors. The red line is the response of the smooth model shown in Figure 34,
fitting the data to a conservative RMS 1.2.
103
Perovskite at 2000 °C
Core
102
Lower mantle
101
Transition zone
Bounds for monotonic models

100
Upper mantle
Bounds on all models
10−1 Ringwoodite at 1500 °C
Wadsleyite at 1500 °C
10−2 SEO3-QFM at 1400 °C
SEO3-IW at 1400 °C
10−3
0 500 1000 1500 2000 2500 3000
Depth (km)
Figure 34 Models fitting the data set shown in Figure 33. Green lines are the Dþ model divided by 105, red line is a first-derivative maximally smooth
model, and the blue line is a smooth model that is allowed to jump at 440 and 670 km. The yellow boxes represent bounds on average
conductivity from Medin et al. (2006). Values of selected mineral conductivities from Figure 26 are also shown.
variation in conductivity inferred for the upper mantle. This, of silicate melts (Gaillard et al., 2008; Yoshino et al., 2012).
course, begs the question of the reliability of the response Furthermore, carbonate melts have low dihedral angles,
functions, particularly for single-site MT soundings where gal- which results in high wetting of grain surfaces and high inter-
vanic distortions of the electric field and effects of 2D and 3D connection, and the presence of CO2 decreases the melting
structure are not all that unusual. Certainly, early single-site MT point of peridotite, allowing melting to occur at greater depths
measurements carried out in the deep ocean basins were dis- in the mantle upwelling beneath mid-ocean ridges. This all
torted by the effect of coastlines (Heinson and Constable, means that significant increases in deep mantle conductivity
1992). However, there are several factors that could modify are possible from small fractions of carbonate melting in peri-
the conductivity of the upper mantle from the averages pre- dotite. Key et al. (2013) claimed to have observed such an
sented earlier in the text and that may become the dominant increase at a depth of 140 km beneath the mid-ocean ridge
conduction mechanism locally. that separates the Pacific and Cocos plates.
5.07.5.2.1 Melt 5.07.5.2.2 Carbon

Since magma is very much more conductive than subsolidus Graphite is extremely conductive (104–106 S m1), and so it
olivine, one mechanism for increasing conductivity is to would take only a small amount to increase mantle conductivity
invoke a mantle permeated by partial melt. Tholeiitic melt over that of dry olivine, as long as it is well connected. The idea
conductivity is more than 1 S m1. For small melt fractions to that carbon might contribute to mantle conductivity was
be an effective mechanism for increased conduction, the melt first put forward by Duba and Shankland (1982). There is a
must be well connected. However, well-connected melt is grav- significant amount of carbon in the mantle (several hundred
itationally unstable and will migrate upward, and so while parts per million) that can exist as CO2 fluids or graphite.
partial melt is clearly an important mechanism in volcanic Movement between the two phases depends on fO2, and so it
systems and mid-ocean ridges, it must be replenished on a is conceivable that CO2 fluids permeating grain boundaries
timescale comparable to the residency time in the mantle for could be reduced to form a connected network of graphite.
it to contribute to permanent enhancement of conductivity or However, recent results of Watson et al. (2010) for 0.1 wt.%
be trapped by a freezing horizon. Naif et al. (2013) inferred carbon in polycrystalline olivine suggest that at these
just such trapping of a few percent melt at the lithosphere– concentrations, carbon does not form a connected network or
asthenosphere boundary in the 23 Myr old Cocos plate off- film and, indeed, slightly increases resistivity by acting in series
shore Nicaragua, based on an interpretation of marine MT with olivine conduction. Another limitation on carbon as a
data. It is unlikely that melt will persist in much older oceanic conduction mechanism for mantle rocks is that below about
mantle, but this result suggests that it may be common under 160 km, graphite turns into diamond, which is a nonconductor.
young oceanic lithosphere. Note also that carbon has a low thermal activation energy, so
It has been recently shown that carbonate melts in perido- we would not expect to see an increase in conductivity with
tite are more than an order of magnitude more conductive than temperature and depth for carbon-enhanced conductivity.
5.07.5.2.3 Strain-induced defects study area could be difficult to identify without adequate data
Since the early marine MT studies inferred a zone of increased (e.g., Heinson and White, 2005).
conductivity at the depth of the asthenosphere, one possibility Direct laboratory evidence for enhanced conduction by
that was considered was that straining generated defects that hydrogen was slow to appear and initial reports were in poor
contribute to conduction. This idea was dispelled by careful agreement with each other and with the diffusion data (e.g.,
measurements by Hirsch and Wang (1986), and now, it is clear Poe et al., 2005; Shankland and Duba, 1997; Wang et al., 2006;
that marine MT measurements may be interpreted in terms of Yoshino et al., 2006), suggesting that hydrogen does not
lower conductivities by taking into consideration the effect of behave as a simple charged defect in olivine. More recent
the coastlines (Heinson and Constable, 1992; Key and results have not completely resolved this issue, and there is
Constable, 2011). an order of magnitude or more discrepancy between various
studies (Figure 35(a)). Yoshino et al. (2009) argued that the
higher temperatures used by Wang et al. resulted in dehydra-
5.07.5.2.4 Water tion and increased conductivity from water on grain bound-
Like carbon, water is known to exist in the mantle – estimates aries or the sample cell. Poe et al. (2010) also cited differences
are that several Earth oceans are still resident somewhere in the in the way water content is estimated and the type of calibra-
deep Earth. Where, and in what state, is less well understood. tion used for this, arguing that a factor of 4 difference with
There are three ways to store water in the mantle: free water, Yoshino et al.’s result could be explained in this way. However,
hydrous minerals, and as defects in nominally anhydrous min- Wang et al. (2006) used the same analytic method as Yoshino
erals. Free water could exist for a period of time in the coolest et al. (2006, 2009) and still obtained differing results. Poe et al.
parts of the mantle but, at even modest temperatures, will also noted other differences in sample type and pressure.
eventually react with olivine to create serpentine minerals. Because the activation energy for the hydrated samples is
This mechanism is interesting in that magnetite is excluded lower than for dry olivine, the influence of water will decrease
during the reaction, potentially creating halos around relic with temperature (and depth). The few results on other
olivine grains of fine magnetite that potentially can be electri- hydrated upper mantle minerals, garnet and pyroxene
cally connected to lower the resistivity of the rock considerably (Figure 35(b)), show a similar pattern.
(Stesky and Brace, 1973). However, hydrous minerals are not High-pressure measurements on the transition zone min-
otherwise conductive (consider mica). At some temperature/ erals ringwoodite and wadsleyite suggest that water solubility
pressure, depending on the mineral but generally at depths in these minerals is very much higher than in upper or lower
shallower than 125 km, hydrous minerals will decompose to mantle minerals, leading Bercovici and Karato (2003) to sug-
produce free water, which will usually lower the melting point gest a water-saturated transition zone (410–670 km deep) and
sufficiently to generate a partial melt into which water dis- pooling of water and melt at the 410 km discontinuity. Fur-
solves, and the melt in turn will be gravitationally unstable thermore, Huang et al. (2005) showed that conductivity is
and migrate upward. This mechanism is largely responsible for enhanced by the addition of water in these minerals. This
volcanism at subduction zones. presents the exciting possibility that electromagnetic methods
The idea that water dissolved as point defects in olivine can provide important constraints on deep Earth processes.
could enhance electric conductivity was originally proposed Based on conductivity bounds obtained from a rigorous anal-
by Karato (1990), based on observations that chemical diffu- ysis of global response function data, Medin et al. (2006)
sion of hydrogen in olivine was very high, particularly along estimated that the global average water content in the transi-
the a-axis. The activation energy of diffusion is 1.3 eV, similar tion zone is < 0.27%, less than the 0.4% required for pooling
to other upper mantle conduction mechanisms. Although ini- at the 410 km discontinuity. However, there remains the pos-
tially put forward as an explanation for early marine MT mea- sibility that the effect of water is heterogeneous and more
surements, the idea found a broad and enthusiastic following significant locally (e.g., Toffelmier and Tyburczy, 2007), and
in the EM community. Since the olivine a-axis is considered to Medin et al.’s inferences depend critically on the early labora-
be aligned with mantle flow, conduction by hydrogen as pro- tory data. More recent laboratory work (Figure 35(c) and
posed by Karato would create significant anisotropy (only 35(d)), however, echoes the olivine story; different groups
about 100 ppm hydrogen is required to increase a-axis con- and laboratories produce results that differ by up to an order
ductivity by an order of magnitude if Karato’s diffusion model of magnitude. Again, Yoshino et al. produced lower conduc-
was correct), and evidence of electric anisotropy in the mantle tivity results and caution against making measurements at high
was often interpreted as support for the hydrogen conduction temperatures where water is more labile.
hypothesis (e.g., Simpson and Tommasi, 2005).
However, the effect of hydrogen on upper mantle conduc- 5.07.5.2.5 Oxides and sulfides
tivity rapidly became a matter for debate. Karato (1990) only Mantle rocks typically contain about one percent oxides of
considered fast diffusion along the a-axis, but the effect is not iron, chromium, and titanium, but as discrete crystals, these
as dramatic when the other two axes are taken into account; do not contribute to bulk conductivity in spite of their high
adding about 1000 ppm hydrogen per silicon atom predicts mineral conductivities. It is possible to increase the magnetite
a conductivity based on diffusivity only slightly above that of content of mantle rocks by hydration, as discussed in the
dry olivine (Constable, 1993b). While observation of anisot- preceding text, but there is no field evidence of increased
ropy in MT data has quickly led to support for the idea of mantle conductivity as a result of this alteration.
hydrogen conduction, many things can result in anisotropic Ducea and Park (2000) interpreted high mantle conductiv-
MT responses; for example, structural anisotropy outside the ities under western North America as connected sulfide melts
Temperature (°C) Temperature (°C)

1900 1600 1400 1200 1000 800 700 600 1900 1600 1400 1200 1000 800 700 600
Olivine Garnet and enstatite

0 0
10 10
-1 -1
Electrical conductivity (S m )
10 10
-1
-2 -2
10 10
-3 -3
10 10
-4 -4
10 10
-5 -5
10 Poe et al. (2010) 10
Wang et al. (2006) Dai and Karato (2009a) (garnet)
Yoshino et al. (2009) Zhang et al. (2012) (enstatite)
-6 -6
10 10
5 6 7 8 9 10 11 12 5 6 7 8 9 10 11 12
(a) (b)
1900 1600 1400 1200 1000 800 700 600 1900 1600 1400 1200 1000 800 700 600
Wadsleyite Ringwoodite
0 0
10 10
-1 -1
Electrical conductivity (S m )
-1
10 10
-2 -2
10 10
-3 -3
10 10
-4 -4
10 10
Huang et al. (2005)
Romano et al. (2009)
-5 -5
10 Dai and Karato (2009b) 10
Yoshino and Katsura (2012) Huang et al. (2005)
Yoshino et al. (2008b) Yoshino et al. (2008b)
-6 -6
10 10
5 6 7 8 9 10 11 12 5 6 7 8 9 10 11 12
(c) Reciprocal temperature (10 4/K) (d) Reciprocal temperature (104/K)
Figure 35 Laboratory studies of conductivity of mantle minerals under pressure, containing 0.1 wt.% water (red) or 1 wt.% water (blue) as point
defects, and shown as Arrhenius model fits to data. Green curves are for anhydrous models. Solid lines represent the temperatures over which
measurements were made; other lines are extrapolations to higher and lower temperatures. In all plots, the black lines are the SEO3 models shown in
Figure 26, shown to provide a common reference. (a) Olivine models are from Wang et al. (2006), Poe et al. (2010), and Yoshino et al. (2009).
Anhydrous data are from Yoshino et al. (2009). (b) Other upper mantle minerals are from Dai and Karato (2009a) and Zhang et al. (2012). (c) Wadsleyite
is from Huang et al. (2005), Romano et al. (2009), Dai and Karato (2009b), Yoshino et al. (2012), and Yoshino et al. (2008b). Dry models are from
Romano et al. (2009), Yoshino et al. (2008b), and Yoshino et al. (2012). (d) Ringwoodite is from Huang et al. (2005) and Yoshino et al. (2008b).
at concentrations of less than 0.4 vol.%. Watson et al. (2010) 1.6 106 O m in the inner core. Although this refinement
showed that even at 1 vol.%, iron sulfide in polycrystalline on the previously accepted values (1–3 106 O m) is not
olivine was largely unconnected, consistent with estimated terribly significant in terms of geoelectromagnetism, it is
percolation threshold of 4–6 vol.%. However, they did observe important with respect to thermal conductivity, which is line-
an increase in conductivity to about 0.1 S m1 at 900 C, arly related to electric conductivity.
inferred to be the result of a very small amount of intercon- The resistivity of metals increases with temperature
nection or an increase in the surface conduction of olivine because lattice vibrations (phonons) can be considered to
grains. This is sufficient to explain many observations of scatter electrons. Similarly, adding impurity atoms also
increased mantle conductivity that have traditionally been increases scattering and resistivity. This is important because
interpreted as melt or the effect of water. we know the outer core is alloyed with 10–25% light ele-
ments for the alloyed iron to match density estimates. Stacey
and Anderson argued that the effects of impurities are inde-
5.07.5.3 The Core
pendent of temperature and indeed somewhat independent
The state of the art with regard to conductivity of the core is of the actual impurity chemistry; the effects of Si and Ni
nicely summarized in a paper by Stacey and Anderson (2001). (both probably present in the core) are considered similar
Their estimates are 2.1 106 O m in the outer core and when measured in weight percent.
Iron behaves as an ideal metal in that the effect of pressure 5.07.6.1 3D Conductivity from Satellite Data
acts to decrease lattice vibrations in a way that affects melting
While there has been a convergence of globally averaged
point and resistivity similarly, and so iron at the melting point
response functions, the ultimate goal of using satellite data is
TM is expected to have the same resistivity (1.35 106 O m)
to obtain information about 3D structure in the mantle. This
whether at zero pressure or at core pressure. At other temper-
presents challenges both for response function estimation and
atures T, for a given pressure, resistivity is assumed to be
inversion of data. The induction signal in satellite data is
proportional to T/TM. Thus, at the core–mantle–boundary
buried deeply in signals from the main field, lithosphere, etc.,
(CMB), temperature is estimated to be 3750 K and iron solidus
and the data themselves are not always nicely behaved, having
as 4147 K, so the component of pure iron resistivity is
significant gaps and sometimes not having the best control on
r ¼ 3750/4147 1.35 106 O m or 1.22 106 O m.
sensor orientation. One can average through these problems
At the inner core boundary, temperature is estimated to be
when assumptions of P01 geometry are made and a global 1D
4971 K and iron solidus as 6000 K, so the component of pure
response is sought, but the generation of 3D response func-
iron resistivity is 1.12 106 O m.
tions will require a finer control over data quality and response
Because iron converts from body-centered cubic (bcc) to
function estimation and dealing with significant complications
hexagonal close-packed (hcp) under pressure, the measure-
of source-field morphology, such as the effect of the iono-
ments of iron with impurities have been carried out using
sphere and non-P01 behavior of the magnetosphere. The multi-
shock-wave experiments, in which a gas/explosive gun or
satellite SWARM experiment should assist greatly in this regard.
laser is fired at a sample. The shock front, moving at velocity
Also, one needs inversion schemes that are not only fully 3D,
us, separates material at the starting density, pressure, and
but can handle the surface conductivity structure accurately.
temperature (ro, Po, To) with material at the shocked condi-
The problem is that the average conductance of the upper
tions (r, P, T ). The shock front is supersonic and so moves
mantle (0.01 S m1 400 km ¼ 4000 S) is less than that of
faster than the actual material, or sound, speed up. Conserva-
the average ocean (3 S m1 4 km ¼ 12 000 S). The more resis-
tion of mass across the shock front gives
tive continents provide windows into the mantle, and regions

us ro ¼ r us up such as mid-ocean ridges, subduction zones, and plumes are
likely to be more conductive than the average, but the existence
so the density increase produced by the shock is of a surface conductance that varies by nearly four orders of

magnitude puts a great demand on the combination of
r up 1
¼ 1 response function estimation and modeling that is needed
ro us before 3D mantle structure can be determined.
and the pressure is given by the rate of change of momentum
per unit area of shock front, which for material starting at rest
and room pressure is 5.07.6.2 Water in the Mantle
P ¼ ro us up As can be seen from Figure 34, while bounds on average

mantle conductivity encompass the measured conductivity of
Hence, knowing the ratio and product of the two velocities anhydrous minerals, an increase in the conductivity of the
for various-sized shocks gives density as a function of pressure upper mantle and transition zone (perhaps a factor of 2 or 3)
for the material. The shock compression is not adiabatic because is compatible with the data, if not required by them. Water is
the material gains kinetic energy. The temperature curve that the commonly invoked to account for this, but since the laboratory
material follows during shocking is called a Hugoniot. experiments on hydrous minerals differ by up to an order of
Stacey and Anderson (2001) fitted linear models to shock- magnitude (Figure 35), it is difficult to make a quantitative
wave data to conclude that a silicon impurity of volume frac- interpretation of the observations. There have also been recent
tion X would contribute 3.0X 106 O m to resistivity. They papers suggesting that carbonate melts or sulfide minerals
expected the silicon to exsolve in a similar way to nickel and so might easily be responsible for increased conductivity in the
took averages of the two components to finally come up with upper mantle, obviating the need to invoke conduction by
an impurity contribution of 0.9 106 O m for a total resistiv- hydrogen. It is important to note that one cannot invoke
ity of 2.1 106 O m. They estimated a lower impurity content water concentrations of more than a few hundred parts per
for the inner core to derive a total resistivity of 1.6 106 O m. million to explain mantle conductivity without then consider-
ing the likelihood and effect of consequent melting (e.g., Ardia
et al., 2012).
5.07.6 Conclusions The current proposals for water pooling in the transition
zone provide a great opportunity for geomagnetic induction
General agreement between various globally averaged response methods to make fundamental contributions to our under-
functions estimated from both satellite and observatory data standing of the mantle. Although, based on average conductiv-
(Figures 11 and 33) has produced a convergence in our under- ity, a case can be made that water pooling does not occur, there
standing of radial electric conductivity structure, represented to is every reason to believe that the distribution of water is not
a large extent by Figure 34. The challenges for the future lie in homogeneous and that locally water may reach the critical
extending this work to higher dimensions and bounding threshold. Long-period MT soundings, if they can be made
the amount of water that resides in the upper mantle and free of distortions due to large-scale crustal structures, could
transition zone. provide important constraints on this hypothesis. Satellite and
observatory response functions, if generated in a way that does Chapman S (1919) The solar and lunar diurnal variations of terrestrial magnetism.
not reject 3D structure, could also be brought to bear on this Philosophical Transactions of the Royal Society of London (A) 218: 1–118.
Chapman S and Bartels J (1940) Geomagnetism. Oxford University Press.
problem.
Chave AD, Thomson DJ, and Ander ME (1987) On the robust estimation of power
spectra, coherences, and transfer-functions. Journal of Geophysical Research
5.07.6.3 Anisotropy Measurements 92: 633–648.
Clement KT (1860) Das grosse Nordlicht in der Nacht zum 29. Aug. 1859 und die
A measurement of bulk conductivity tells us only a small Telegraphenverwirrug in Nord-Amerika und Europa, p. 121. Hamburg.
amount about the interior of the Earth; the trade-offs between Constable SC (1993a) Constraints on mantle electrical conductivity from field and
laboratory measurements. Journal of Geomagnetism and Geoelectricity
composition and temperature are large and often poorly con- 45: 707–728.
strained. A measurement of electric anisotropy, however, Constable SC (1993b) Conduction by mantle hydrogen. Nature 362: 704.
unfolds another dimension in our understanding. For exam- Constable S (2006) SEO3: A new model of olivine electrical conductivity. Geophysical
ple, strain induced by plate motion could generate anisotropy Journal International 166: 435–437.
Constable S (2013) Review paper: Instrumentation for marine magnetotelluric and
associated either with mineral texture and fabric or by aligning
controlled source electromagnetic sounding. Geophysical Prospecting
crystallographic axes. Some conduction mechanisms, such as 61: 505–532.
the diffusion of hydrogen in olivine, predict significant anisot- Constable CG and Constable SC (2004a) Satellite magnetic field measurements:
ropies in the upper mantle. Combining the interpretation of Applications in studying the deep Earth. In: Sparks RSJ and Hawkesworth CT (eds.)
electric measurements with other geophysical data, such as The State of the Planet: Frontiers and Challenges in Geophysics, Geophys. Mono.
150, pp. 147–159. Washington, DC: American Geophysical Union.
seismic velocity, has always been a good idea, and the combi- Constable S and Constable C (2004b) Observing geomagnetic induction in magnetic
nation of electric and seismic anisotropy could prove to be a satellite measurements and associated implications for mantle conductivity.
powerful tool for understanding Earth processes. The reader is Geochemistry, Geophysics, and Geosystems 5: Q01006. http://dx.doi.org/
referred to a review by Eaton and Jones (2006) and the papers 10.1029/2003GC000634.
Constable SC, Parker RL, and Constable CG (1987) Occam’s Inversion: A practical
in that volume. However, the challenge here is to make a
algorithm for generating smooth models from EM sounding data. Geophysics
meaningful measurement of mineralogical or fabric anisotropy 52: 289–300.
and not become confused with anisotropy associated with Constable S and Roberts JJ (1997) Simultaneous modeling of thermopower
larger-scale structures. and electrical conduction in olivine. Physics and Chemistry of Minerals
24: 319–325.
Constable SC, Shankland TJ, and Duba A (1992) The electrical conductivity of an
isotropic olivine mantle. Journal of Geophysical Research 97: 3397–3404.
Acknowledgment Dai L and Karato S (2009a) Electrical conductivity of pyrope-rich garnet at high
temperature and high pressure. Physics of the Earth and Planetary Interiors
This contribution was supported in part by funding from NASA 176: 83–88.
Dai L and Karato S (2009b) Electrical conductivity of wadsleyite at high temperatures
grant NAG5-13747 and NSF grant EAR-1112861.
and high pressures. Earth and Planetary Science Letters 287: 277–283.
Didwall EM (1984) The electrical conductivity of the upper mantle as estimated from
satellite magnetic field data. Journal of Geophysical Research 89: 537–542.
References Dobson DP and Brodholt JP (2000) The electrical conductivity of the lower mantle
phase magnesiowustite at high temperatures and pressures. Journal of Geophysical
Alexandrescu MM, Gibert D, Le Mouel JL, Hulot G, and Saracco G (1999) An estimate Research 105: 531–538.
of average lower mantle conductivity by wavelet analysis of geomagnetic jerks. Dosso SE and Oldenburg DW (1991) Magnetotelluric appraisal using simulated
Journal of Geophysical Research 104: 17735–17745. annealing. Geophysical Journal International 106: 379–385.
Ardia P, Hirschmann MM, Withers AC, and Tenner TJ (2012) H2O storage capacity of Du Frane WL, Roberts JJ, Toffelmier DA, and Tyburczy JA (2005) Anisotropy of
olivine at 5–8 GPa and consequences for dehydration partial melting of the upper electrical conductivity in dry olivine. Geophysical Research Letters 32: L24315.
mantle. Earth and Planetary Science Letters 345–348: 104–116. Duba AG, Schock RN, Arnold E, and Shankland TJ (1990) An apparatus for
Avdeev DB (2005) Three-dimensional electromagnetic modelling and inversion from measurement of electrical conductivity to 1500 C at known oxygen fugacity.
theory to application. Surveys in Geophysics 26: 767–799. In: Duba AG, Durham WB, Handin JW, and Wang HF (eds.) The Brittle-Ductile
Backus GE (1983) Application of mantle filter theory to the magnetic jerk of 1969. Transitions in Rocks. The Heard Volume, Geophys. Monogr. Ser, vol. 56,
Geophysical Journal of the Royal Astronomical Society 74: 713–746. pp. 207–210. Washington, DC: AGU.
Backus GE and Gilbert JF (1967) Numerical applications of a formalism. Geophysical Duba AG and Shankland TJ (1982) Free carbon and electrical conductivity in the Earth’s
Journal of the Royal Astronomical Society 13: 247–276. mantle. Geophysical Research Letters 9: 1271–1274.
Backus GE and Gilbert JF (1968) The resolving power of gross Earth data. Geophysical Ducea MN and Park SK (2000) Enhanced mantle conductivity from sulfide
Journal of the Royal Astronomical Society 16: 169–205. minerals, southern Sierra Nevada, California. Geophysical Research Letters
Backus G, Parker R, and Constable C (1996) Foundations of Geomagnetism. 27: 2405–2408.
Cambridge: Cambridge University Press. Eaton DW and Jones A (2006) Tectonic fabric of the subcontinental lithosphere:
Bahr K, Olsen N, and Shankland TJ (1993) On the combination of the magnetotelluric Evidence from seismic, magnetotelluric and mechanical anisotropy. Physics of the
and the geomagnetic depthsounding method for resolving an electrical conductivity Earth and Planetary Interiors 158: 85–91.
increase at 400 km depth. Geophysical Research Letters 20: 2937–2940. Egbert GD (1997) Robust multiple-station magnetotelluric data processing. Geophysical
Balasis G and Egbert GD (2006) Empirical orthogonal function analysis of magnetic Journal International 130: 475–496.
observatory data: Further evidence for non-axisymmetric magnetospheric sources Egbert GD and Booker JR (1986) Robust estimation of geomagnetic transfer-functions.
for satellite induction studies. Geophysical Research Letters 33: L11311. Geophysical Journal of the Royal Astronomical Society 87: 173–194.
Banks RJ (1969) Geomagnetic variations and the electrical conductivity of the upper Egbert GD and Booker JR (1992) Very long period magnetotellurics at Tucson-
mantle. Geophysical Journal of the Royal Astronomical Society 17: 457–487. observatory – Implications for mantle conductivity. Journal of Geophysical Research
Barlow WH (1849) On the spontaneous electrical currents observed in wires of the 97: 15099–15112.
electric telegraph. Philosophical Transactions of the Royal Society of London Everett ME and Constable S (1999) Electric dipole fields over an anisotropic seafloor:
139: 61–72. Theory and application to the structure of 40 Ma Pacific Ocean lithosphere.
Bercovici D and Karato SI (2003) Whole-mantle convection and the transition-zone Geophysical Journal International 136: 41–56.
water filter. Nature 425: 39–44. Everett ME, Constable S, and Constable CG (2003) Effects of near-surface conductance
Cagniard L (1953) Basic theory of the magneto-telluric method of geophysical on global satellite induction responses. Geophysical Journal International
prospecting. Geophysics 18: 605–635. 153: 277–286.
Everett ME and Schultz A (1993) 2-Dimensional nonlinear magnetotelluric inversion Laske G and Masters G (1997) A global digital map of sediment thickness. Eos,
using a genetic algorithm. Journal of Geomagnetism and Geoelectricity Transactions of the American Geophysical Union 78(46): F483, Fall Meet. Suppl.,
45: 1013–1026. Abstract S41E-01.
Everett M and Schultz A (1996) Geomagnetic induction in a heterogeneous sphere, Libermann RC and Wang Y (1992) Characterization of sample environment in a uniaxial
azimuthally symmetric test computations and the response of an u 660-km split-sphere apparatus. In: Syono Y and Manghnani MH (eds.) High Pressure
discontinuity. Journal of Geophysical Research 101: 2765–2783. Research: Applications to Earth and Planetary Sciences, pp. 19–31. Washington,
Filloux JH (1987) Instrumentation and experimental methods for oceanic studies. DC: American Geophysical Union.
In: Jacobs JA (ed.) Geomagnetism, pp. 143–248. New York: Academic Press. Lizarralde D, Chave A, Hirth G, and Schultz A (1995) Northeastern Pacific mantle
Gaillard F, Marki M, Iacono-Marziano G, Pichavant M, and Scaillet B (2008) conductivity profile from long-period magnetotelluric sounding using Hawaii to
Carbonatite melts and electrical conductivity in the asthenosphere. Science California submarine cable data. Journal of Geophysical Research
322: 1363–1365. 100: 17837–17854.
Gamble TD, Clarke J, and Goubau WM (1979) Magnetotellurics with a remote magnetic Mackie RL and Madden TR (1993) 3-Dimensional magnetotelluric inversion using
reference. Geophysics 44: 53–68. conjugate gradients. Geophysical Journal International 115: 215–229.
Gill PE, Murray W, and Wright MH (1981) Practical Optimization. London: Academic Marquardt DW (1963) An algorithm for least-squares estimation of non-linear
Press. parameters. Journal of the Society for Industrial and Applied Mathematics
Grammatica N and Tarits P (2002) Contribution at satellite altitude of 11: 431–441.
electromagnetically induced anomalies arising from a three-dimensional Martinec Z (1999) Spectral-finite element approach to three-dimensional
heterogeneously conducting earth, using Sq as an inducing source field. electromagnetic induction in a spherical Earth. Geophysical Journal International
Geophysical Journal International 151: 913–923. 136: 229–250.
Grandis H, Menvielle M, and Roussignol M (1999) Bayesian inversion with Markov McCammon C (2005) The paradox of mantle redox. Science 308: 807–808.
chains-I. The magnetotelluric one-dimensional case. Geophysical Journal McDonald KL (1957) Penetration of the geomagnetic secular field through a mantle with
International 138: 757–768. variable conductivity. Journal of Geophysical Research 62: 117–141.
Hamano Y (2002) A new time-domain approach for the electromagnetic induction Mead C and Jeanloz R (1987) High precision strain measurements at high pressures.
problem in a three-dimensional heterogeneous Earth. Geophysical Journal In: Manghnani MH and Syono Y (eds.) High-Pressure Research in Mineral Physics,
International 150: 753–769. pp. 41–51. Washington, DC: Am. Geophys. Union.
Heinson G and Constable SC (1992) The electrical conductivity of the oceanic upper Medin AE, Parker RL, and Constable S (2006) Making sound inferences
mantle. Geophysical Journal International 110: 159–179. from geomagnetic sounding. Physics of the Earth and Planetary Interiors
Heinson GS and White A (2005) Electrical resistivity of the northern Australian 160: 51–59.
lithosphere: Crustal anisotropy or mantle heterogeneity? Earth and Planetary Morgan FD, Williams ER, and Madden TR (1989) Streaming potential properties of
Science Letters 232: 157–170. Westerly Granite with applications. Journal of Geophysical Research
Hirsch LM (1990) Electrical conduction of Co2SiO4. Physics and Chemistry of Minerals 94: 12449–12461.
17: 187–190. Naif S, Key K, Constable S, and Evans RL (2013) Melt-rich channel observed at the
Hirsch LM and Shankland TJ (1991) Point-defects in (Mg, Fe)SIO3 perovskite. lithosphere-asthenosphere boundary. Nature 495: 356–359.
Geophysical Research Letters 18: 1305–1308. Newman GA and Alumbaugh DL (2000) Three-dimensional magnetotelluric inversion
Hirsch LM and Wang CY (1986) Electrical conductivity of olivine during high- using non-linear conjugate gradients. Geophysical Journal International
temperature creep. Journal of Geophysical Research 91: 429–441. 140: 410–424.
Huang XG, Xu YS, and Karato SI (2005) Water content in the transition Nielsen OV, Petersen JR, Primdahl F, et al. (1995) Development, construction and
zone from electrical conductivity of wadsleyite and ringwoodite. Nature analysis of the Orsted fluxgate magnetometer. Measurement Science and
434: 746–749. Technology 6: 1099–1115.
Ito E and Takahashi E (1987) Ultrahigh-pressure phase transformations and the O’Neill HStC (1987) Quartz–fayalite–iron and quartz–fayalite–magnetite equilibria and
constitution of the deep mantle. In: Manghnani MH and Syono Y (eds.) High- the free energy of formation of fayalite (Fe2SioO4) and magnetite (Fe3O4). American
Pressure Research in Mineral Physics, pp. 221–229. Washington, DC: American Mineralogist 72: 67–75.
Geophysical Union. Oldenburg DW (1983) Funnel functions in linear and non-linear appraisal. Journal of
Karato S (1990) The role of hydrogen in the electrical conductivity of the upper mantle. Geophysical Research 88: 7387–7398.
Nature 347: 272–273. Olsen N (1992) Day-to-day c-response estimation for Sq from 1-cpd to 6-cpd
Key K and Constable S (2011) Coast effect distortion of marine magnetotelluric data: using the Z-Y-method. Journal of Geomagnetism and Geoelectricity
Insights from a pilot study offshore northeastern Japan. Physics of the Earth and 44: 433–447.
Planetary Interiors 184: 194–207. Olsen N (1999) Long-period (30 days–1 year) electromagnetic sounding and the
Key K, Constable S, Liu L, and Pommier A (2013) Electrical image of passive mantle electrical conductivity of the lower mantle beneath Europe. Geophysical Journal
upwelling beneath the northern East Pacific Rise. Nature 495: 499–502. International 138: 179–187.
Key K and Weiss C (2006) Adaptive finite-element modeling using unstructured grids: Olsen N, Vennerstrm S, and Friis-Christensen E (2003) Monitoring magnetospheric
The 2D magnetotelluric example. Geophysics 71: G291–G299. contributions using ground-based and satellite magnetic data. In: Reigber C,
Kittel C (1986) Introduction to Solid State Physics, sixth edn. New York: Wiley. Luehr H, and Schwintzer P (eds.) First CHAMP Mission Results for Gravity,
Kuvshinov AV, Avdeev DB, and Pankratov OV (1999) Global induction by Sq and Dst Magnetic and Atmospheric Studies, pp. 245–250. Berlin: Springer.
sources in the presence of oceans: Bimodal solutions for non-uniform spherical Parker RL (1984) The inverse problem of resistivity sounding. Geophysics
surface shells above radially symmetric earth models in comparison to 49: 2143–2158.
observations. Geophysical Journal International 137: 630–650. Parker RL (1994) Geophysical Inverse Theory. Princeton, NJ: Princeton University
Kuvshinov AV, Avdeev DB, Pankratov OV, Golyshev SA, and Olsen N (2002) Press.
Modelling electromagnetic fields in 3-D spherical earth using fast integral equation Parker RL and Booker JR (1996) Optimal one-dimensional inversion and bounding of
approach. In: Zhdanov MS and Wannamaker PE (eds.) Three Dimensional magnetotelluric apparent resistivity and phase measurements. Physics of the Earth
Electromagnetics, Methods in Geochemistry and Geophysics 35, pp. 43–54. and Planetary Interiors 98: 269–282.
Amsterdam: Elsevier. Parker RL and Whaler K (1981) Numerical methods for establishing solutions to the
Kuvshinov AV and Olsen N (2004) Modelling the coast effect of geomagnetic storms at inverse problem of electromagnetic induction. Journal of Geophysical Research
ground and satellite altitude. In: Reigber C, Luhr H, Schwintzer P, and Wickert J 86: 9574–9584.
(eds.) Earth Observation with CHAMP. Results from ‘Three Years in Orbit’, Parkinson WD (1983) Introduction to Geomagnetism. Amsterdam: Elsevier.
pp. 353–359. Berlin: Springer. Petiau G (2000) Second generation of lead-lead chloride electrodes for geophysical
Kuvshinov A and Olsen N (2006) A global model of mantle conductivity derived from 5 applications. Pure and Applied Geophysics 157: 357–382.
years of CHAMP, Orsted, and SAC-C magnetic data. Geophysical Research Letters Pinheiro K and Jackson A (2008) Can a 1-D mantle electrical conductivity model
33: L18301. generate magnetic jerk differential time delays? Geophysical Journal International
Lahiri BN and Price AT (1939) Electromagnetic induction in non-uniform conductors, 173: 781–792.
and the determination of the conductivity of the Earth from terrestrial magnetic Poe B, Romano C, Nestola F, and Rubie D (2005) Electrical conductivity of hydrous
variations. Philosophical Transactions of the Royal Society of London (A) single crystal San Carlos olivine. Eos, Transactions of the American Geophysical
237: 509–540. Union 86(52), Fall Meet. Suppl., Abstract MR41A-0895.
Poe TB, Romano C, Nestola F, and Smyth JR (2010) Electrical conductivity anisotropy charge transport and melt structure. Journal of Geophysical Research
of dry and hydrous olivine at 8 GPa. Physics of the Earth and Planetary Interiors 88: 2413–2430.
181: 103–111. Uyeshima M and Schultz A (2000) Geoelectromagnetic induction in a heterogeneous
Quist AS and Marshall WL (1968) Electrical conductances of aqueous sodium chloride sphere, a new 3-D forward solver using a staggered-grid integral formulation.
solutions from 0 to 800 and at pressures to 4000 Bars. Journal of Physical Geophysical Journal International 140: 636–650.
Chemistry 71: 684–703. Velimsky J, Everett ME, and Martinec Z (2003) The transient Dst electromagnetic
Revil A and Glover PW (1997) Theory of ionic-surface electrical conduction in porous induction signal at satellite altitudes for a realistic 3-D electrical conductivity in the
media. Physical Review B 55: 1757–1773. crust and mantle. Geophysical Research Letters 30. http://dx.doi.org/
Roberts JJ and Tyburczy JA (1999) Partial-melt electrical conductivity: Influence of melt 10.1029/2002GL016671.
composition. Journal of Geophysical Research 104: 7055–7065. Wang D, Mookherjee M, Xu Y, and Karato S (2006) The effect of water on the electrical
Romano C, Poe BT, Tyburczy J, and Nestola F (2009) Electrical conductivity of hydrous conductivity of olivine. Nature 443: 977–980.
wadsleyite. European Journal of Mineralogy 21: 615–622. Watson HC, Roberts JJ, and Tyburczy JA (2010) Effect of conductive impurities on
Sabaka TJ, Olsen N, and Purucker ME (2004) Extending comprehensive models of the electrical conductivity in polycrystalline olivine. Geophysical Research Letters
Earth’s magnetic field with Orsted and CHAMP data. Geophysical Journal 37: L02302. http://dx.doi.org/10.1029/2009GL041566.
International 159: 521–547. Webb SC, Constable SC, Cox CS, and Deaton T (1985) A seafloor electric field
Schmeling H (1986) Numerical models on the influence of partial melt on elastic, instrument. Journal of Geomagnetism and Geoelectricity 37: 1115–1130.
anelastic, and electrical properties of rocks. 2. Electrical conductivity. Physics of the Weidelt P (1972) The inverse problem in geomagnetic induction. Zeitschrift für
Earth and Planetary Interiors 43: 123–136. Geophysik 38: 257–289.
Schmucker U (1970) Anomalies of geomagnetic variations in the Southwestern United Weiss CJ (2001) A matrix-free approach to solving the fully 3D electromagnetic
States. Bulletin of the Scripps Institution of Oceanography 13: 69. induction problem. In: Contributed Paper at SEG 71st Annual Meeting, San Antonio,
Schock RN, Duba A, and Shankland TJ (1989) Electrical conduction in olivine. Journal USA.
of Geophysical Research 94: 5829–5839. Weiss CJ and Everett ME (1998) Geomagnetic induction in a heterogeneous sphere,
Schultz A, Kurtz RD, Chave AD, and Jones AG (1993) Conductivity discontinuities in fully three-dimensional test computation and the response of a realistic distribution
the upper mantle beneath a stable craton. Geophysical Research Letters of oceans and continents. Geophysical Journal International 135: 650–662.
20: 2941–2944. Xu YS, Poe BT, Shankland TJ, and Rubie DC (1998) Electrical conductivity of olivine,
Schuster A and Lamb H (1889) The diurnal variation of terrestrial magnetism. wadsleyite, and ringwoodite under upper-mantle conditions. Science
Philosophical Transactions of the Royal Society of London 45: 481–486. 280: 1415–1418.
Shankland TJ and Duba AG (1997) Correlation of olivine electrical conductivity change Xu YS and Shankland TJ (1999) Electrical conductivity of orthopyroxene and its high
with water activity. Eos, Transactions of the American Geophysical Union 78(46), pressure phases. Geophysical Research Letters 26: 2645–2648.
Fall Meet. Suppl., Abstract T52D-10. Xu YS, Shankland TJ, and Duba AG (2000) Pressure effect on electrical conductivity of
Shankland TJ, Duba AG, Mathez EA, and Peach CL (1997) Increase of electrical mantle olivine. Physics of the Earth and Planetary Interiors 118: 149–161.
conductivity with pressure as an indicator of conduction through a solid phase in Yoshimura R and Oshiman N (2002) Edge-based finite element approach to the
midcrustal rocks. Journal of Geophysical Research 102: 14741–14750. simulation of geoelectromagnetic induction in a 3-D sphere. Geophysical Research
Simpson F and Bahr K (2005) Practical Magnetotellurics. Cambridge: Cambridge Letters 29. http://dx.doi.org/10.1029/2001GL014121.
University Press. Yoshino T and Katsura T (2012) Re-evaluation of electrical conductivity of anhydrous
Simpson F and Tommasi A (2005) Hydrogen diffusivity and electrical anisotropy of a and hydrous wadsleyite. Earth and Planetary Science Letters 337–338: 56–67.
peridotite mantle. Geophysical Journal International 160: 1092–1102. Yoshino T, Manthilake G, Matsuzaki T, and Katsura T (2008a) Dry mantle transition
Smith JT and Booker JR (1991) Rapid inversion of 2-dimensional and 3-dimensional zone inferred from the conductivity of wadsleyite and ringwoodite. Nature
magnetotelluric data. Journal of Geophysical Research 96: 3905–3922. 451: 326–329.
Srivastava SP (1966) Theory of the magnetotelluric method for a spherical conductor. Yoshino T, Matsuzaki T, Shatskiy A, and Katsura T (2009) The effect of water on
Geophysical Journal of the Royal Astronomical Society 11: 373–387. the electrical conductivity of olivine aggregates and its implications for the
Stacey FD and Anderson OL (2001) Electrical and thermal conductivities of Fe–Ni–Si electrical structure of the upper mantle. Earth and Planetary Science Letters
alloy under core conditions. Physics of the Earth and Planetary Interiors 288: 291–300.
124: 153–162. Yoshino T, Matsuzaki T, Yamashita S, and Katsura T (2006) Hydrous olivine unable to
Stesky RM and Brace WF (1973) Electrical conductivity of serpentinized rocks to 6 account for conductivity anomaly at the top of the asthenosphere. Nature
kilobars. Journal of Geophysical Research 78: 7614–7621. 443: 973–976.
Tarits P, Hautot S, and Perrier F (2004) Water in the mantle: Results from electrical Yoshino T, McIsaac E, Laumonier M, and Katsura T (2012) Electrical conductivity of
conductivity beneath the French Alps. Geophysical Research Letters 31: L06612. partial molten carbonate peridotite. Physics of the Earth and Planetary Interiors
Tikhonov AN (1950) Determination of the electrical characteristics of the deep strata of 194–195: 1–9.
the Earth’s crust. Doklady Akademii NaukSSSR 73: 295–297. Yoshino T, Nishi M, Matsuzaki T, Yamazaki D, and Katsura T (2008b) Electrical
Tikhonov AN and Arsenin VY (1977) Solutions of Ill-Posed Problems. New York: John conductivity of majorite garnet and its implications for electrical structure in
Wiley and Sons. the mantle transition zone. Physics of the Earth and Planetary Interiors
Toffelmier DA and Tyburczy JA (2007) Electromagnetic detection of a 410-km-deep melt 170: 193–200.
layer in southwestern United States. Nature 447: 991–994. Yoshino T, Yamazaki D, Ito E, and Katsura T (2008c) No interconnection of
Tuller HL and Norwick AS (1977) Small polaron electron transport in reduced CeO2 ferro-periclase in post-spinel phase inferred from conductivity measurement.
single crystals. Journal of Physics and Chemistry of Solids 38: 859–867. Geophysical Research Letters 35: L22303. http://dx.doi.org/
Tyburczy JA and Roberts JJ (1990) Low frequency electrical response of polycrystalline 10.1029/2008GL035932.
olivine compacts: Grain boundary transport. Geophysical Research Letters Zhang B, Yoshino T, Wu X, Matsuzaki T, Shan S, and Katsura T (2012) Electrical
17: 1985–1988. conductivity of enstatite as a function of water content: Implications for the
Tyburczy JA and Waff HS (1983) Electrical conductivity of molten basalt ad electrical structure in the upper mantle. Earth and Planetary Science Letters
andesite to 25 kilobars pressure: Geophysical significance and implications for 357–358: 11–20.
5.08 Magnetizations in Rocks and Minerals
DJ Dunlop and Ö Özdemir, University of Toronto, Toronto, ON, Canada

5.08.1.1 Diamagnetism and Paramagnetism 256
5.08.1.2 Ferromagnetism 256
5.08.1.3 Antiferromagnetism and Ferrimagnetism 257
5.08.1.4 Magnetocrystalline Anisotropy 258
5.08.1.5 Magnetostriction and Magnetoelastic Anisotropy 258
5.08.1.6 Magnetic Quantities, Units, and Conversions 259
5.08.2 Domains and the Magnetization Process 259
5.08.2.1 Demagnetizing Energy and Domain Structure 259
5.08.2.2 Domain Walls and Multidomain Magnetization 261
5.08.2.3 Domain Wall Width and Energy 261
5.08.2.4 Equilibrium Domain Structures 262
5.08.2.5 Observations of Domains 262
5.08.2.6 Micromagnetic Modeling 264
5.08.2.7 Single-Domain Grains 264
5.08.2.8 Metastable SD Grains 265
5.08.2.9 Pseudo-Single-Domain Grains 265
5.08.2.10 Superparamagnetism 267
5.08.3 Magnetic Minerals and Their Properties 268
5.08.3.1 Thermomagnetic Curves, Curie Temperatures, and SD Ranges 268
5.08.3.2 Magnetite 269
5.08.3.3 Titanomagnetites 272
5.08.3.4 Maghemite and Titanomaghemites 273
5.08.3.5 High-Temperature Oxidation 274
5.08.3.6 Hematite 275
5.08.3.7 Titanohematites (Hemoilmenites) 275
5.08.3.8 Iron Oxyhydroxides 276
5.08.3.9 Iron Sulfides 277
5.08.3.10 Iron–Chrome Spinels (Chromites) 278
5.08.3.11 Iron Carbonates 278
5.08.3.12 Iron and Iron–Nickel 278
5.08.3.13 ‘Magnetic’ Silicates 279
5.08.3.14 Biogenic Magnetic Minerals 280
5.08.4 Induced and Remanent Magnetization 281
5.08.4.1 SD Hysteresis and Susceptibility 281
5.08.4.2 MD Hysteresis and Susceptibility 283
5.08.4.3 Temperature Effects: Hopkinson Peak and Koenigsberger Ratio Q 283
5.08.4.4 Magnetostatic Interactions 284
5.08.5 NRM and Paleomagnetic Stability 285
5.08.5.1 AF Demagnetization 285
5.08.5.2 Low-Temperature Demagnetization 286
5.08.5.3 Thermal Demagnetization of SD Grains 286
5.08.5.4 Thermal Demagnetization of MD Grains 287
5.08.5.5 Resolving Multivectorial NRM 287
5.08.5.6 Thellier Paleointensity Determination 288
5.08.5.7 Stability and Domain-State Tests 288
5.08.6 Remanent Magnetization Processes in Nature 290
5.08.6.1 Thermoremanent Magnetization 290
5.08.6.2 Viscous and Thermoviscous Magnetization 293
5.08.6.3 Chemical Remanent Magnetization 296
5.08.6.4 Detrital and Postdepositional Remanent Magnetizations 297
5.08.7 Summary 299
Appendix A Developments in Rock and Mineral Magnetism 2006–12 299
References 302
Further-reading 308

256 Magnetizations in Rocks and Minerals
5.08.1 Introduction Table 1 Susceptibilities k and w of some common minerals
Mineral k (106 SI) w

All matter is magnetic, but in geophysics, we are usually
(108 m3 kg1)
concerned with minerals that are either ‘ferromagnetic’ like
iron–nickel in meteorites, ‘ferrimagnetic’ like magnetite, or Diamagnetic
‘antiferromagnetic’ like hematite. These minerals are orders of Quartz (SiO2) 16.4 0.62
magnitude more magnetic than diamagnetic minerals like Orthoclase (KAlSi3O8) 14.9 0.58
quartz or paramagnetic minerals like micas. They possess Calcite (CaCO3) 13.6 0.48
both a reversible ‘induced magnetization’ (which changes Forsterite (Mg2SiO4) 12.5 0.39
when the inducing field changes, e.g., reverses its polarity) Paramagnetic
Troilite (FeS) 0.6–1.7 103 13–35
and a permanent ‘remanent magnetization’ (which preserves
Pyrite (FeS2) 1.5 103 30
a memory of paleomagnetic fields).
Siderite (FeCO3) 4.9 103 123
Ilmenite (FeTiO3) 4.7–5.2 103 100–110
Ulvöspinel (FeTi2O4) 4.8 103 100
5.08.1.1 Diamagnetism and Paramagnetism Orthopyroxene ((Fe,Mg)SiO3) 1.55–1.8 103 43–50
Fayalite (Fe2SiO4) 5.53 103 126
‘Diamagnetism’ is a property of all atoms. It originates in the Olivine, intermediate 1.56 103 36
speeding up or slowing down of electrons in orbital motion ((Fe,Mg)2SiO4)
when the magnetic Lorentz force is added to the electrostatic Serpentinite (Mg3Si2O5(OH)4) 3 103a 120a
Coulomb force. The direction of the resulting magnetic moment Amphiboles 0.5–2.7 103a 16–94a
per unit volume, or magnetization, M is opposite to that of H, Biotites 0.5–1.15 103 17–38
the magnetic field applied, giving a negative susceptibility: 1.5 103 30
1.5–2.9 103a 50–97a
kd ¼ dM=dH < 0 [1] Clay minerals (illite and 0.33–0.41 103 13–15
montmorillonite)
The diamagnetic susceptibility is extremely weak and is
Ferromagnetic, ferrimagnetic, and antiferromagnetic
overshadowed if any other forms of magnetism are present. Iron, multidomain (aFe) 3.9 106 5.0 104
However, it is temperature-independent and establishes a base- Pyrrhotite (Fe7S8) 3.2 106 6.9 104
line to the magnetic signal at high T, for example, above the (Fe9S10) 0.17 106 0.38 104
Curie points of ferromagnetic minerals. Hematite (aFe2O3) 0.5–40 103 10–760
‘Paramagnetism’ is a more powerful effect arising from the Maghemite, multidomain 2.0–2.5 106 4.0–5.0 104
partial alignment by H of permanent magnetic dipole (gFe2O3)
moments of atoms, which in turn arise from unpaired electron Magnetite, multidomain 3.0 106 5.8 104
spins. Every odd-numbered element has at least one unpaired (Fe3O4)
Titanomagnetite, x ¼ 0.6 0.13–0.62 106 0.25–1.2 104
spin and therefore a permanent magnetic moment. The tran-
(TM60)
sition elements, in particular, Fe, Co, and Ni, have multiple
Titanomaghemite 2.8 106 5.7 104
unpaired 3d electrons and correspondingly stronger moments. Goethite (aFeOOH) 1.1–12 103 25–280
Paramagnetic susceptibility is temperature-dependent
a
because the aligning effect of H is opposed by thermal disor- Higher values are due to magnetite inclusions.
dering. The Langevin theory predicts that M is parallel to H and
varies as H/T (the Curie law), giving a positive susceptibility: susceptibility, rocks that are virtually barren of ferrimagnetic
kp ðT Þ ¼ @MðH, T Þ=@H 1=T [2] minerals may still have a measurable anisotropy because of the
preferential alignment of paramagnetic mineral axes.
At ordinary temperatures, paramagnetic susceptibilities
are 1–2 orders of magnitude larger than typical diamagnetic
susceptibilities (Table 1) but 2–3 orders of magnitude smaller
5.08.1.2 Ferromagnetism
than the susceptibilities of ferrimagnetic minerals, for
example, magnetite with a (per unit mass) susceptibility Iron, nickel, cobalt, and a few other elements differ from dia-
w 5.8 104 m3 kg1. For more extensive tabulations of vol- magnets and paramagnets in retaining a ‘remanent’
ume and mass susceptibilities of rocks and minerals, see Hunt magnetization or ‘remanence,’ Mr, in the absence of a field H.
et al. (1995) or Carmichael (1982). In addition, even if the remanence is minor, the ‘induced’
Paramagnetism comes into prominence in three situations. magnetization kH is strong and readily measurable for rela-
The first is at very low temperatures, where alignment of atomic tively weak fields like the Earth’s. Magnetic anomalies and the
moments becomes substantial. Low-T superconducting quan- paleomagnetic method both depend on these two basic prop-
tum interference device magnetometry is becoming common erties, which in turn reflect coupling of electron spins of neigh-
in rock magnetism, and paramagnetic minerals are sometimes boring atoms.
detectable by their 1/T signals below 20 K. The second situ- For ferromagnetic minerals, weak-field initial susceptibili-
ation where paramagnetism becomes visible is at above the ties k0 and w0 are listed.
Curie temperature TC where ferromagnetic and ferrimagnetic Figure 1 illustrates a generic ‘hysteresis’ or M versus H curve.
minerals lose their spin coupling (see the next section). Third, (In industrial magnetism, flux density B ¼ m0(H þ M) is plotted
in determining rock fabric by the anisotropy of magnetic instead of M. The curves are indistinguishable because M H
Magnetizations in Rocks and Minerals 257
only common fields in nature strong enough to produce SIRM

M (A m–1)
Ms result from lightning currents, which produce their own fields
in directions unrelated to the Earth’s field. Nevertheless, SIRM
is frequently measured as part of the characterization of a
Initial sample, as is the ‘coercive force’ (more properly, ‘coercive
curve field’) Hc, which is the axis-crossing field in a hysteresis loop,
usually the saturation loop. Another field commonly reported
Mrs is the ‘remanent coercive force’ Hcr, which is the (opposed)
field whose application and removal reduce Mr to zero.
Both B and H are called the magnetic field in common
–Hc
Mr
parlance. Outside a magnetic material, they can be used inter-
H (A m–1) changeably. H is preferred as the measure of the output of a
ko
–Hcr Hc field-producing device like a solenoid and B (or M) as the
–Mr magnetic effect of that field, hence B–H loops. In recent
years, Bc and Bcr have begun to be quoted in the rock magnetic
–Mrs literature because B converts more conveniently than H
Minor
loop between cgs and SI units (see Section 5.08.1.6). This is incor-
rect. H is the abscissa in all hysteresis loops; Bc and Bcr imply
Major
loop measurements of an internal field, essentially M. If conve-
nience of units is an issue, m0Hc and m0Hcr can be used. Other
situations where we must be careful to use H and not B are in
–Ms calculating induced magnetization, Min ¼ kH, and in dealing
with internal fields like the self-demagnetizing field Hd (see
Figure 1 Initial magnetization curve and minor and major hysteresis
Section 5.08.2.1).
loops. Standard parameters are ko (initial susceptibility), Ms
Ferromagnetic ordering results from the Heisenberg
(saturation magnetization), Mrs (saturation remanence), Mr (isothermal
remanence, minor loop), Hc (coercive force – for either major or exchange or superexchange coupling between the spins of 3d
minor loop), and Hcr (remanent coercive force). Reversing the field electrons in neighboring or next-nearest neighboring atoms, a
at Hcr brings the sample to zero net magnetization (dashed line). quantum mechanical manifestation of electrostatic interaction
between the electrons. The same mechanism is responsible for
covalent bonding. For a discussion, see Dunlop and Özdemir
for ferromagnets, making B essentially equal to M, apart from (1997, Chapter 2). Before the mechanism of spin coupling was
the rescaling required by the permeability of free space, known, Weiss (1907) proposed that a powerful internal field,
m0 ¼ 4p 107Hm1.) Hysteresis, the irreversibility of the mag- which he termed the molecular field, must exist in iron and
netization process, ensures that M is not a single-valued func- magnetite to produce alignment of their atomic moments. This
tion of H. However, provided that maximum fields of equal approach is very fruitful and generates an immediate under-
strength H are applied in opposite directions, a field cycle standing of the ‘spontaneous magnetization’ Ms of ferromag-
results in a closed M–H loop. nets and its temperature dependence and ultimate
Hysteresis originates in the energy expended in either disappearance at the Curie point TC. ‘Thermomagnetic’ or
increasing the size of one set of ‘domains’ of coupled spins at Ms(T ) curves are one of the most basic methods of determining
the expense of another set or else rotating spins past unfavor- the magnetic mineralogy of a rock (see Section 5.08.3.1).
ably oriented axes in the crystal. Because these processes
expend similar amounts of energy for fields in opposite direc-
5.08.1.3 Antiferromagnetism and Ferrimagnetism
tions, a complete cycle returns the system to its original state of
magnetization but with the expenditure of an amount of In ferromagnets, the moments of neighboring atoms are coupled
energy proportional to the area enclosed by the loop. This parallel, and the origin of spontaneous magnetization is obvious.
hysteresis loss is important in industrial applications but not But of the common magnetic minerals, only iron and iron–
in nature, where field changes are quite modest. nickel are ferromagnetic and they are found mainly on other
A ‘minor’ loop measures changes in M in small fields, for planets (Moon and meteorites). Magnetite and hematite, Earth’s
example, in the course of two successive reversals of the Earth’s commonest magnetic minerals, are, respectively, ‘ferrimagnetic’
field. Note that the ferromagnetic susceptibility kf ¼ dM/dH is and ‘antiferromagnetic,’ terms invented by Néel (1948) to
not constant but generally increases with increasing H (until H describe new states of matter embodying ‘negative’ exchange
reverses). The isothermal remanence or IRM, Mr, in such a weak- coupling. A schematic picture of these states appears in Figure 2.
field process is very small and furthermore is paleomagnetically Negative coupling of atomic moments does not obviously
useless because it will change with each and every field reversal lead to any net magnetization, however strong the coupling
and with secular changes in field intensity between reversals. may be. This is not actually the case, however, because the ‘up’
The ‘saturation remanence’ or SIRM, Mrs, produced by and ‘down’ spins may not belong to the same type of atom or
departing from a field sufficient to saturate all domain enlarge- ion and therefore may have different magnetic moments,
ments or rotations, is much more substantial and changes little represented in Figure 2 by different lengths of the arrows.
with subsequent small changes (positive or negative) in the This is the situation in magnetite (Fe2þFe3þ 2 O4), for example,
field. However, it is paleomagnetically irrelevant because the where the antiparallel magnetic sublattices have different
Ferromagnetism Antiferromagnetism Ferrimagnetism Spin-canted AFM
net
Ms
Figure 2 Schematic representation of spin coupling in the ferromagnetic, antiferromagnetic, and ferrimagnetic states of matter. Spin canting in the
antiferromagnetic state imparts a weak transverse ferromagnetic moment to hematite. Ms is the net spontaneous magnetization per spin lattice site (or
A þ B spin sublattice sites, in the case of ferrimagnetism).
numbers of Fe2þ and Fe3þ ions at individual sites. The ferri- 1

magnetism of magnetite, although only a net imbalance Magnetite
between two competing sublattices, creates an Ms 1/3 that (Fe3O4)
of ferromagnetic iron, where all atomic moments add.
In antiferromagnetism, as exemplified by hematite Ms(T )
(Fe3þ 3þ
2 O3) and its hydrated precursor goethite (Fe OOH), the
up and down sublattices have identical numbers and species of l(T )
ions. Each sublattice has a spontaneous magnetization, but in
zero field, these cancel because they are oppositely directed. An
antiferromagnet in theory has no remanence, but a field K1(T )
applied perpendicular to the sublattices will deflect both sets
of spins, giving rise to an antiferromagnetic susceptibility ka.
A strong field is required to produce any substantial angular Tc
deflection, but on the other hand, the moments being
deflected are collectively strong, with the result that ka is com- 0
0 100 200 300 400 500
parable to kp at ordinary temperatures, although the exchange
Temperature ( C)
coupling and thermal disordering mechanisms involved are
entirely different. The temperature dependences reveal the dif- Figure 3 Normalized temperature dependences of spontaneous
ferent mechanisms, ka, varying only weakly with T. Indeed, magnetization Ms, direction-averaged magnetostriction constant l, and
before Néel elucidated its origin, antiferromagnetism was first-order magnetocrystalline anisotropy constant K1 of magnetite. All
known as temperature-independent paramagnetism. three decrease to zero at the Curie point TC, but at very different rates.
Hematite actually has a weak permanent magnetization,
enabling it to carry a remanence. The fundamental mechanism axes. In iron, K1 > 0 and h100i are the easy axes. For magnetite,
is spin canting, a zero-field deflection of the sublattices out of K1 < 0 and h111i are the easy axes. K1 and K2 are measured from
antiparallelism by a fraction of a degree. As explained by torque curves of a magnetically saturated single crystal or from
Dzyaloshinsky (1958), spin canting occurs only for certain ferromagnetic resonance of a polycrystalline sample.
crystal symmetries and disappears when the symmetry Minerals with a single-crystalline easy axis of magnetization
changes, as it does when the sublattices are parallel to the are not common in nature (hematite at low temperatures is
rhombohedral c-axis instead of being perpendicular to it. one example), but ‘uniaxial anisotropy’ is often used as a
Hematite has two phases, truly antiferromagnetic at tempera- simplified model in magnetic structure calculations. The
tures below the Morin transition TM (at 15 C) and weakly anisotropy energy is then
ferromagnetic above TM. Even in its ferromagnetic phase, Ms is
Ek ¼ K u1 Vsin 2 y þ K u2 V sin 4 y [4]
only 0.5% that of magnetite.

y ¼ 0 or 180 being the two preferred orientations of Ms that
define the easy axis. Uniaxial anisotropy can arise from ‘mag-
5.08.1.4 Magnetocrystalline Anisotropy netoelastic anisotropy’ due to magnetostriction
(Section 5.08.1.5) and ‘shape anisotropy’ of the magnetostatic
Magnetocrystalline anisotropy is a manifestation of atomic energy of elongated grains (Section 5.08.2.1).
spin–orbit coupling. Electron orbitals tend to lie along pre- Magnetocrystalline anisotropy decreases with temperature
ferred or ‘easy axes’ in a crystal, and spin magnetic moments T much more rapidly than does Ms (Figure 3). It is generally
are linked to these axes via the orbital coupling. negligible in the T range near the Curie point where cooling
Macroscopically, magnetocrystalline anisotropy energy EK is igneous rocks acquire most of their remanence.
created when Ms deviates from an easy axis. In a cubic crystal,

EK ¼ K 1 V a1 2 a2 2 þ a2 2 a3 2 þ a3 2 a1 2 þ K 2 Va1 2 a2 2 a3 2 [3] 5.08.1.5 Magnetostriction and Magnetoelastic Anisotropy
where K1 and K2 are the anisotropy constants, V is the volume, and When a crystal is magnetized, its dimensions change sponta-
ai are the direction cosines of Ms with respect to the h100i crystal neously. A positive ‘magnetostriction’ corresponds to an
expansion in the direction of the magnetization M. Magnetite labeled Bc or Bcr. They should be called m0Hc or m0Hcr if the
expands in the easy direction h111i with maximum strain convenience of using T rather than A m1 is compelling.
l111 ¼ þ 72.6 106 and contracts in the hard direction h100i More serious are quantities determined ‘inside’ a magnetic
with maximum strain l100 ¼ 19.5 106. The magnetostric- material, such as the internal field Hi of a magnetized grain.
tion averaged over all directions, appropriate for a polycrystal- One reason for using H rather than B for internal fields is that
line material, is ls ¼ 0.4l100 þ 0.6l111. H is defined even at an atomic level (even though seldom
If a uniform strain is imposed through an external or inter- calculated at this scale because of its extreme spatial fluctua-
nal stress s and isotropic magnetostriction is assumed, the tion), while M and B are mesoscopic averages over thousands
crystalline anisotropy energy changes by an amount of atoms. B should always be treated as a derivative field, a
combination of H and M, unless one is in free space far from all
Es ¼ lsV 1:5ls sVsin 2 y [5] magnetized bodies. For example, in a B–H loop, although
which has the form of a uniaxial ‘magnetoelastic anisotropy’; measurements are made outside the magnetized body, by the
y ¼ 0 is the axis of tension. In practice, individual magnetic nature of the measurement procedure, it is B inside the body
grains in rocks experience highly variable local strains whose that is determined, not m0H.
magnetic effects are difficult to describe in a simple way. Another situation where H must be used, not B, is in deter-
Magnetostriction constants decrease with increasing tem- mining susceptibility, k. By definition, k ¼ dM/dH. Although k is
perature less rapidly than K1 or K2 (Figure 3). Magnetoelastic dimensionless, it has different values in SI and cgs: a k value in SI
anisotropy may thus survive as an important factor in rema- is numerically larger by 4p than the cgs k value. By the same
nence acquisition to higher T than magnetocrystalline token, demagnetizing factors N (Section 5.08.2.1) that relate Hi
anisotropy. to M are numerically smaller in SI than in cgs by a factor 4p.
5.08.1.6 Magnetic Quantities, Units, and Conversions 5.08.2 Domains and the Magnetization Process
The basic relation between B, H, and M is B ¼ m0 (H þ M) in SI,
5.08.2.1 Demagnetizing Energy and Domain Structure
with m0 ¼ 4p 107 Hm1, and B ¼ H þ 4pM in cgs emu.
Sometimes, the symbol J is used for magnetization in cgs; In exact analogy to a polarized dielectric, at the surface of a
this is not recommended because J in electromagnetism is magnetized body, bound magnetic poles appear with a density
current density. Traditionally, cgs was the system of choice in s ¼ M n, where n is the outward surface normal (Figure 4). In
magnetism, and it remains widely used today, largely for prag- turn, these poles act as sources and sinks for lines of the
matic reasons. B, H, and M have the same dimensions, internal ‘self-demagnetizing field’ Hd, so called because its
although their units have different names (Table 2), and con- direction is opposite to M. If M is a remanent magnetization,
versions between B and H are effortless. In SI, B and H, both Hd is the total internal field, but if an external applied field Ho
usually called the magnetic field, differ numerically by a factor is necessary to maintain an induced magnetization M, the
of 106. For geomagnetists, cgs units were particularly natural internal field is
because the Earth’s magnetic field is 1 G or Oe. The corre-
Hi ¼ Ho þ Hd ¼ Ho NM [6]
sponding figures in SI are awkwardly large or small:
80 A m1 for H and 100 mT for B. N is the demagnetizing factor and usually has different values
Because B converts from cgs to SI with a simple factor of for different directions in a crystal, dependent on its shape. For
104, while the conversion for H is more cumbersome, fields a cube, for example, iron, N ¼ 1/3 (4p/3 in cgs) for M parallel
in SI are seldom quoted in A m1, even when they should be. to a h100i axi. For an octahedral crystal, for example, magne-
One example was cited in Section 5.08.1.2, where H axis- tite, N ¼ 1/3 for M along a h111i axis. For platy crystals, for
crossing parameters in hysteresis loops are increasingly being example, hematite, N 0 for M in the plane (ordinary
Table 2 Units and conversions for magnetic quantities
Quantity SI unit cgs unit Conversion
Magnetic moment, m Am2 emu 1 Am2 ¼ 103 emu

Magnetization, M (¼m/V) Am1 ‘emu cm3’ 1 Am1 ¼ 103emucm3
Magnetization per unit mass Am2 kg1 ‘emug1’ 1 Am2 kg1 ¼ 1 emug1
Magnetic field, H Am1 Oersted 1 Am1 ¼ 4p 103 Oe
Magnetic field or induction, B Tesla Gauss 1 T ¼ 104 G
B ¼ m0(H þ M) in SI
B ¼ H þ 4pM in cgs
Susceptibility, k (¼dM/dH) (dimensionless) ‘SI unit’ ‘cgs unit’ 1 SI unit ¼ (1/4p) cgs
Susceptibility per unit mass, w m3 kg1 ‘emug1 Oe1’ 1 m3 kg1 ¼ (103/4p) cgs
Demagnetizing factor, N (dimensionless) 1 (in SI) ¼ 4p (in cgs)
Source: Moskowitz BM (1995) Fundamental physical constants and conversion factors. In: Ahrens TJ (ed.) Rock Physics and Phase Relations: A Handbook of Physical Constants,
vol. 1. Washington, DC: American Geophysical Union, pp 346–55, Chapter 1.02, Table 1; Dunlop DJ and Özdemir Ö (1997) Rock Magnetism: Fundamentals and Frontiers.
New York: Cambridge University Press, Table 2.1.
temperatures) or 1 transverse to the plane (low tempera- The shape dependence of Hd and the ‘demagnetizing
tures). For elongated grains, N 0 for M parallel to the long energy’ Ed ¼ (1/2) m0VNM2 is large enough in strongly mag-
axis (the magnetostatically preferred orientation) and 1/2 for netic minerals like magnetite and iron that a slight deviation
transverse M (hard direction). from perfectly cubic shapes produces a ‘shape anisotropy’
1
– + Ed ¼ m0 VDNM2 sin 2 y [7]
2
– + which outweighs the rather small magnetocrystalline anisot-
s =
– + ropy. Equation [7] applies only to ellipsoidal grains, which
M Hd M·n have uniform internal fields, but is a useful first-order model
– + for real crystals in nature. DN is the difference between N for M
– + along the axis of intermediate hardness (usually near y ¼ 90 )
n n and N when the grain is magnetized along its length (the easy
– + axis, y ¼ 0).
– + Measured remanent and induced magnetizations are quite
small compared to Ms for weak fields Ho. How is it that shape
( Ho ) anisotropy is so powerful? The answer lies in domain structure
(Figure 5). In very small particles,
100 nm for magnetite and
Figure 4 The (almost uniform) self-demagnetizing field Hd of a material even smaller for iron, exchange coupling successfully main-
with uniform magnetization M. M may be induced by an externally applied
tains a uniform magnetization Ms throughout the crystal.
field Ho or it may be a permanent (remanent) magnetization. Magnetic
Such particles contain a ‘single magnetic domain.’ In larger
poles with surface density s ¼ M n, the exact analogs of bound charges
associated with electric dipoles in dielectrics, appear wherever M cuts the particles, Ed for M ¼ Ms becomes so large that there are energy
surface. These poles can be viewed as the sources of field lines Hd savings to be had by subdividing the crystal into two or more
within the crystal and of the leakage or stray fields outside the crystal by domains with their Ms vectors antiparallel (or at other angles
which domain boundaries are imaged in the Bitter colloid method. dictated by magnetocrystalline easy axes, as in Figure 5). The
Ms1 -
[001] [111]
90° M --
sc [111]
90° --
[010] [112]
Iron Magnetite
Ms2
Msc
Closure
45° 45° domain
45° Msc 90°
45° walls
135° 135°
Ms1 Ms2
180°
wall
Body domains
Figure 5 Spontaneous magnetization vectors within body domains (Ms1 and Ms2) and closure domains (Msc). For materials like iron or high-Ti
titanomagnetites, with h100i easy axes, Msc is at 90 to Ms1 and Ms2, as sketched. For materials like magnetite and low-Ti titanomagnetites, with h111i
easy axes, Msc is at 70.5 or 109.5 to Ms1 and Ms2. Domain walls bisect the angle between Ms vectors in adjacent domains, in order to eliminate
magnetic poles on the boundary (e.g., Ms1 n þ Msc n ¼ 0). Modified from Özdemir Ö, Xu S, and Dunlop DJ (1995) Closure domains in magnetite.
domains are sheetlike, with Ms in the plane of the sheet, so that regions of rotated Ms are called ‘domain walls’ (Figure 6).
N is reduced for each domain compared to N for a single They have a definite thickness, dictated by the balance between
domain. However, M ¼ Ms within each domain and DN when reducing Eex and Ed and increasing anisotropy energy (next
the domains are pulled away from the long axis is merely section). They migrate easily because each spin in the wall
scaled down relative to DN for rotation of a single domain. needs only to rotate by a small angle in response to Ho to
Shape anisotropy remains important. cause ‘domain wall displacement.’ Spins at one edge of the
If the interior of a ferromagnet is magnetized to saturation, wall are added to the enlarged domain. New spins from the
globally (single-domain or SD grain) or locally (multidomain less favored domain are added to the opposite edge of the wall.
or MD grain), how does it come about that the measured The wall itself has the same structure it had before moving; its
induced or remanent magnetization M of a rock is propor- energy is unchanged. Only the changing shapes of the domains
tional to the applied field Ho that caused it? An individual SD tilt the energy balance.
grain has no demagnetized or low-M state. The only possible The only reason walls are hindered at all in their progress is
response to Ho is rotation of its total magnetic moment VMs that they tend to become anchored at crystal imperfections
toward Ho. This added component of magnetization in the such as surface pits or interior voids and cracks, which reduce
direction of Ho explains induced magnetization – but not the wall’s energy by the magnetostatic effect of poles that
remanence. When Ho disappears, so does M: all the grain appear on the crack or void boundaries or at pileups of dislo-
moments return to their original easy (long) axes, oriented at cations, which interact magnetoelastically with the wall. For a
random. A remanence can only be produced if some moments detailed treatment, see Özdemir and Dunlop (1997). The same
are pulled by Ho past the hard axis, ultimately reversing their pinning centers hinder the wall’s return motion when Ho is
moments when Ho is removed. Because this takes considerable reduced or removed. However, the pinning is not symmetrical.
energy, SD grains are hard to magnetize using field alone. On The net Hd in any grain favors a return to an overall demagne-
the other hand, once SD moments have been rotated past a tized state. This is not necessarily exactly the same as the initial
shape anisotropy barrier, by whatever means, they are hard to state, but on average, self-demagnetization opposes any rema-
‘demagnetize’ under ordinary conditions of temperature and nence and helps walls jump past pins on their way to a low-
field. They are ideal paleofield recorders. magnetization state.
By analogy with the macroscopic hysteresis loop, an indi-
vidual wall can be thought of as having one or several coercive
5.08.2.2 Domain Walls and Multidomain Magnetization forces Hc that describe how much field is required to impel it
The magnetization process is entirely different in MD grains. past a barrier or barriers to motion. These are often called
Instead of rotating domain moments, a small field Ho will microcoercivities. The Hc values are generally small compared
cause domains with VMs close (<90 ) to Ho to enlarge at the to fields necessary to rotate SD moments past shape anisotropy
expense of unfavorably oriented domains. This process is barriers, both because pinning by defects is inherently weak
almost reversible and should result in a large susceptibility. and because Hd aids so effectively in demagnetizing any rem-
There is, however, a price to be paid in increased Ed because the anence. For these reasons, MD grains are usually poor paleo-
enlarged domains have increased their N. (The domains that field recorders. Their remanence is a small fraction of that of an
shrink decrease their N but the net effect is still an increase in equal volume of SD grains, although this is offset by the high
Ed, as Section 5.08.4.3 shows.) Self-demagnetization applies a proportion of MD grains in the ferromagnetic fraction of most
severe brake on increased M. Observed susceptibilities of rocks. More importantly, their coercivities are low and do not
magnetite-bearing rocks are essentially controlled by N, not stabilize the remanence against changing fields later in the
by the ease of enlarging domains. rock’s life.
The mechanism by which domains enlarge or shrink resem-
bles the migration of a dislocation when a crystal deforms. The
5.08.2.3 Domain Wall Width and Energy
boundaries between domains are not marked by abrupt
changes in the direction of Ms because exchange energy favors The width of a wall is an important parameter because it affects
a gradual rotation from one domain to its neighbor. The the ease of observing domain structures. Along a line of m spins
f wall
Plane o
Bod
y do
mai
n
180
B loch
dom
ain
wal
l
Bod
y do
mai
n
Figure 6 Progressive rotation of spins across a 180 wall between two body domains. In the Bloch wall shown, spins rotate in the plane of the wall so
as to eliminate interior magnetic poles. A real domain wall contains a hundred or more spins, not the few shown here.
normal to a wall, Ms rotates through small angles Dy/m from 1=2 1=2
deq ¼ m0 NSD M2s =2gw D [11]
one domain to the other (Figure 6). Exchange energy is
reduced by making the rotation as gradual as possible: In multiaxial materials, there is a choice of several crystal-
Eex ¼ const. JeS2Dy2/m, where Je is the exchange integral and S line easy axes and closure domains can form. Ed is eliminated
atomic spin. However, EK increases when spins do not lie along or greatly reduced because Ms n ¼ 0 on all internal boundaries
crystalline easy axes: DEk ¼ Ka3m, where a is the interatomic (180 and 90 walls in iron and 180 , 109.5 , and 70.5 walls
spacing and K (¼1.64 10 J m3 for magnetite) averages over in magnetite) and is small if not zero on surface boundaries of
the line of spins (see Dunlop and Özdemir, 1997, Chapter 5, body and closure domains (Figure 5). A simple calculation for
for a discussion). Minimizing the sum of the two energies gives magnetite that considers two main energies, the energy of 180
a wall width walls and the magnetoelastic energy Es due to incompatible
magnetostrictions across 109.5 and 70.5 walls, gives
dw ¼ meq a ¼ psin ðDy=2ÞðA=K Þ1=2 [8] (Özdemir et al., 1995)
11 1
in which the exchange constant A (¼1.33 10 J m for deq ¼ ð4gw =9l111 c44 Þ1=2 D1=2 [12]
magnetite at 20 C) is JeS2/a.
From [8], dw is predicted to be 0.28 mm for 180 walls in in which c44 is an elastic constant.
magnetite, compared with an experimentally measured dw of The domain width predicted by eqn [12] is several times
0.18 mm (Moskowitz et al., 1988). The rather crude theoretical larger than that according to [11]. Closure domains effectively
approach is therefore acceptable if only a rough estimate is eliminate Ed and so long, narrow body domains are no longer
needed. imperative.
The wall energy per unit area is needed to calculate equilib- Experimental data testing the prediction that deq D1/2
rium domain widths. It is appear in Figure 7. For magnetite at least, the prediction is
verified most convincingly for grain sizes of 10 mm and less.
gw ¼ 2psin ðDy=2ÞðAK Þ1=2 [9] The proportionality constant is appropriate to eqn [12], not
2 [11], for reasonable values of gw, yet relatively few of these
giving a 180 wall energy in magnetite of gw ¼ 0.93 J m . The
small grains exhibited closure domains (Figure 8). Large mag-
experimental value is gw ¼ 0.91 J m2 (Özdemir and Dunlop,
netite crystals do have well-developed surface and interior
1993a). The close agreement may be deceptive. In the calcula-
closure domains (Figure 9), yet their domain widths are even
tion, the demagnetizing field and energy of the wall were
larger and more discrepant with the equilibrium theory
ignored. While it is true that no poles appear on the interfaces
(Figure 7). That is, they contain fewer domains than what the
between a wall and its bounding domains, because spins rotate
equilibrium theory predicts.
in the plane of the wall, poles do appear at the edges of the wall
where it meets the crystal surface. One way of eliminating this
energy is through surface closure domains (see Figure 5), dis-
cussed in the succeeding text. 5.08.2.5 Observations of Domains
The domains in Figures 8 and 9 were imaged by the Bitter
technique. A colloidal suspension of magnetite nanoparticles
5.08.2.4 Equilibrium Domain Structures gathers over places in a polished section through a crystal
where magnetic flux leaks out of the surface. The colloidal
A lattice of spins spontaneously subdivides itself into domains particles are in essence tracers of magnetic field lines, similar
to reduce demagnetizing energy Ed. To a good approximation, in principle to iron filings used to map the field lines of a bar
both the effective demagnetizing factors N and Ed are inversely magnet, but on a greatly reduced size scale. Field lines leave a
proportional to the number of domains n (Dunlop and crystal wherever Ms cuts the surface. Sectioning the crystal in a
Özdemir, 1997, Figure 5.5). Increasing the number of domains plane that contains two sets of easy axes, for example, {110} in
thus reduces Ed to (1/n) of the single-domain energy (Ed)SD. magnetite, containing two sets of h111i axes, cuts across both
However, the process of subdivision cannot continue indefi- body and closure domains. The Ms vectors in the domains are
nitely, because each new domain wall adds its energy to the surface-parallel, so Ms n ¼ 0 and there are no poles to act as
total. In addition, there are other ways of reducing Ed besides sources for leakage fields. However, Ms vectors in the walls
making individual domains longer and narrower. The most bounding the domains rotate in the plane of the wall and do
effective way is by forming flux-closure domains either in the cut the surface. The poles thus formed attract colloid particles,
body of the crystal or at its surface (Figure 5). making the walls visible. The 180 walls are more easily imaged
The simplest domain structure is that of uniaxial materials, than 109.5 and 70.5 walls because of their greater width
which have no closure domains. They form only 180 walls (Williams et al., 1992; cf. eqn [8]).
and the domains terminate at the crystal surface. The energy is Magnetic force microscopy (MFM), like its sister technique,
then the sum of Ed of the n domains and Ew of the n 1 walls: atomic force microscopy (AFM), measures the force exerted on
E ¼ ð1=2Þm0 LWDN SD M2s ð1=nÞ þ gw LW ðn 1Þ [10] the tip of a nanometer-scale cantilever as it scans over the
surface of a magnetic substance (Foss et al., 1998; Pokhil and
where L and W are the length and width of the grain and of the Moskowitz, 1997). Companion MFM and AFM images are
plate-like walls, respectively, and D ¼ nd is the third dimension needed to separate forces due to magnetic leakage fields from
of the grain, subdivided into n domains of width d. Minimizing the atomic Coulomb forces. Although resolutions
10 nm are
E by setting dE/dn ¼ 0 yields in principle achievable, clear magnetic images like those in
1 10 100 1000
60 60
40
This study
Özdemir and Dunlop (1993),
20 Özdemir et al. (1995)
Other studies
10 10
Domain with (mm)
1 1
0.7
0.5
0.3 0.3
1 2 4 7 10 20 40 70 100 200 400 700 1000
Grain size or domain length (mm)
Figure 7 Average width of domains in magnetite as a function of grain size (or in large grains, domain length). The dot-dash line is based on the
equilibrium domain theory (eqns [11] and [12]). Most data fall above this line, indicating fewer than the predicted number of domains. Modified from
Özdemir Ö and Dunlop DJ (2006) Magnetic domain observations on magnetite crystals in biotite and hornblende grains. Journal of Geophysical
Research 111: B06103. (http://dx.doi.org/10.1029/2005JB004090).
polarization that rotated a small amount by the surface mag-

netic field. Figure 11 illustrates the domain wall displacement
process described in Section 5.08.2.2. Domains with favorably
or unfavorably oriented Ms vectors enlarge or shrink with
changes in the applied magnetic field. The angle of rotation
of the light polarization is extremely small in magnetite; image
processing was needed to achieve the contrast seen in Figure 11
(Heider and Hoffmann, 1992).
Another technique that images the surface magnetization
directly rather than through leakage fields is spin-polarized
scanning electron microscopy (SEM). It has submicron
resolution and so can detect fine structures that Bitter or
MOKE imaging techniques cannot image. Secondary electrons
ejected from the surface by the primary electron beam are
analyzed at detectors for their spin polarization, which is pro-
portional to the magnetization in the uppermost 1 nm of the
50 mm section. Magnetizations parallel and perpendicular to the sur-
face can be determined in the same scan, and no corrections for
Figure 8 Lamellar body domains and wedge-shaped surface closure surface topography are needed, unlike MFM. In examples of
domains imaged by the Bitter colloid method on a {110} surface of a spin-polarized SEM images given by Haag and Allenspach
single crystal of magnetite. Modified from Özdemir Ö, Xu S, and Dunlop (1993), titanomagnetite has clear lamellar domain patterns,
DJ (1995) Closure domains in magnetite. Journal of Geophysical while titanohematite has rather vague and confused patterns.
Research 100: 2193–2209. A third technique that directly images the in-plane compo-
nent of B, which inside a ferromagnet is essentially M
Figure 10 (Pan et al., 2002) are generally seen only on microm- (Section 5.08.1.2), is electron holography transmission elec-
eter scales and for scan heights of 0.1 mm. The images are of tron microscopy (TEM). The resolution is a few tens of nano-
poorer quality when viewed 0.4 mm above the crystals because meters, as Figure 12 shows (Harrison et al., 2002). Off-axis
of the divergence of the field lines (Frandsen et al., 2004). electron holography measures the amplitude and phase shift of
A technique that images domains themselves rather than the electron wave that passes through a sample in TEM. From
their boundaries is the magneto-optical Kerr effect (MOKE). the magnetic phase image, a map of the surface-parallel mag-
The light reflected from a magnetic medium has its plane of netization is generated. Figure 12 illustrates a finely exsolved
109
71
0
109
18
71
0
18
0
18
75 mm –
111
–
111
h
Figure 9 Closure domains iobserved
h oni a {110} surface in a magnetite single crystal. The boundary (top) is a deep crack. The body and closure domains
are magnetized along 111 and 111 easy directions, respectively. The 70.5 and 109.5 walls bisect the angles between easy axes, as predicted.
Modified from Özdemir Ö, Xu S, and Dunlop DJ (1995) Closure domains in magnetite. Journal of Geophysical Research 100: 2193–2209.
titanomagnetite grain with closely spaced magnetite blocks Metastable magnetic states are often found as solutions.
separated by paramagnetic ulvöspinel. The contours trace out These local energy minimum (LEM) states are an impediment
magnetic flux lines, which close after linking together a num- if they stall the calculations and prevent the lowest energy or
ber of neighboring magnetite crystals. The individual grains global energy minimum (GEM) state from being found. On
may be essentially single domains, but they interact so strongly the other hand, LEM states have their own interest. They are
that they collectively form superdomains of correlated Ms vec- possible states of the system, albeit less stable than the GEM
tors. Overall, flux patterns resemble those of a vortex or a two- state. A major change in thinking in magnetism came with the
domain grain with two closure domains. realization that in a system of many particles, many different
LEM states are occupied, as well as the GEM state, and further-
more that each particle can transform its structure from one
LEM state to another with the aid of thermal energy. Thus, the
5.08.2.6 Micromagnetic Modeling
equilibrium state of the system is a Boltzmann partition among
Interacting arrays of magnetic crystals like those in Figure 12 all LEM states, not the lowest energy state.
can be modeled theoretically (Muxworthy et al., 2004) and so
can structures within noninteracting crystals. These ‘micromag-
netic models’ are more sophisticated versions of the energy
5.08.2.7 Single-Domain Grains
minimization process used in Sections 5.08.2.3 and 5.08.2.4
to find domain and wall widths. Instead of imposing a struc- SD grains are of special interest because they have a very strong
ture of uniformly magnetized domains with straight bound- remanence compared with MD grains, and their remanence is
aries, these ab initio calculations begin from randomly extremely stable because to change it requires rotating the
perturbed spin structures and evolve through numerous itera- grain’s moment past a large anisotropy barrier. MD grains by
tions to a minimum energy state. The convergence may be slow contrast change remanence rather easily by low-energy motion
but can be speeded by using fast Fourier transforms. If the of domain walls. Figure 10 shows that only very small grains
dynamic behavior of the evolving spin system is included in can be SD; these 5–13 mm magnetite crystals contain two or
addition to simple energy minimization, computation time is more walls. Grain shape is another factor: an elongated grain
considerably increased but the solutions are more stable. All may maintain an SD state because of its needlelike shape,
relevant energies are considered, including Eex, EK, Es, and Ed, which reduces Ed, while a grain of similar size but of more
but of these, Ed is by far the most time-consuming to compute regular shape will nucleate a wall. The maximum size for
because it requires evaluating the magnetostatic interaction of crystals of a particular mineral to resist subdividing into
every pair of spins in the model. domains is called the critical SD size d0.
The remanent critical size (d0)r is usually quoted but (d0)c

for coherent reversal is more important paleomagnetically
because incoherent reversals greatly reduce the coercive force
and the long-term stability of remanence. Figure 14 illustrates
how different (d0)r and (d0)c are for magnetite grains of differ-
ent elongations (Newell and Merrill, 1999). Butler and
Banerjee (1975a) calculated that (d0)r increases by almost
an order of magnitude in grains with 5:1 elongation, and
Figure 14 confirms this. Elongated grains can remain SD in
(a) (b) the sense of having a strong and uniform remanence to much
larger sizes than equidimensional grains. However, the shape
5 µm 8 µm dependence is much weaker for (d0)c, which changes very little
up elongations of 5:1. SD stability – the property of reversing
coherently with high coercivity – is hardly dependent on shape
at all for geologically reasonable elongations.
5.08.2.8 Metastable SD Grains

Even (d0)r is not as straightforward as it seems. Figure 15
reveals that starting around 70 nm, energy minimizations
with an initial vortex state yield an ultimate low-remanence
state (Fabian et al., 1996). The GEM state is vortex. On the
(c) (d) other hand, an initial SD (flower) state does not transform
during minimization to the GEM state until a much larger
12 mm 13 mm size, around 140 nm. In the 70–140 nm size range, the rema-
nent state of a real grain depends on prior history. If the grain
began in an SD state, it will remain metastably SD unless
driven over the energy barrier between LEM states by some
[100]
external influence (field, long exposure to thermal fluctua-
tions, etc.). The region of possible metastable SD behavior is
theoretically quite broad in magnetite for all elongations
[110] (Figure 16). It is noteworthy that the magnetosomes manufac-
tured by magnetotactic bacteria for navigational use have sizes
(e)
and shapes well outside the ‘true’ SD region, covering almost
exactly the predicted metastable SD region.
Figure 10 Magnetic force microscope (MFM) images of domains in
250 nm thick epitaxially grown magnetite crystals. The smallest crystal Because SD sizes for magnetite are <1 mm, independent
(5 mm) contains three body domains and the largest (13 mm) five experimental proof of metastable SD behavior has been diffi-
domains. Growth on a substrate promotes [110] uniaxial anisotropy cult (see, however, Boyd et al., 1984). The same is not true for
and may be the reason for the lack of closure domains. Modified from high-Ti titanomagnetite and pyrrhotite, both of which have
Pan Q, Pokhil TG, and Moskowitz BM (2002) Domain structures in (d0)r values around 1 mm (Table 3). The TM60 (titanomagne-
epitaxial Fe3O4 particles studied by magnetic force microscopy. Journal tite containing 60 mol% Ti) grain in Figure 17 is (d0)r but
of Applied Physics 91: 5945–5950. has failed to nucleate a domain wall after exposure to a satu-
rating field (Halgedahl and Fuller, 1983). It is metastably SD.
However, quite a small reverse field of 1.2 mT causes a wall to
In reality, this definition of d0 is ambiguous. Figure 13 nucleate and then rapidly spread across the grain. Thus, as
illustrates the various remanent LEM states in a cube (Rave argued in the last section for somewhat different reasons, a
et al., 1998). The structures are of three types: (1) SD, in grain can have SD-like remanence but relatively low stability.
which Ms is uniform, except for a ‘flowering’ of the vector at
the corners and edges; (2) lamellar, conventional slab-like
5.08.2.9 Pseudo-Single-Domain Grains
domains, appropriate if anisotropy is uniaxial and strong;
and (3) vortex, with a continuous curling of the Ms vector, The possible choice of LEM states should be fairly narrow for
appropriate if anisotropy is multiaxial and relatively weak. sizes close to (d0)r (Figure 13). Experimentally, however, grains
A two-domain structure with two closure domains is a com- have a wider choice of remanent LEM states than theory would
promise between (2) and (3). But it is possible for an SD flower suggest. The TM60 grain in Figure 18 is an example. In the SIRM
structure to reverse to a similar state with an intermediate state, it contains three domains, one of them much larger than
structure of curling spins that resembles a vortex state. Such a the others, but by judicious field cycling, a five-domain state
reversal is called ‘incoherent’ in contrast to a ‘coherent reversal’ appears. In other TM60 grains, cycled repeatedly to Tc and back
in which the structure remains uniform throughout. Thus, d0 in a weak field to produce ‘thermoremanent magnetization
for a uniform remanent state is larger than d0 for coherent (TRM),’ Halgedahl (1991) observed one to nine domains in
reversal (Enkin and Williams, 1994). the remanent state. These grains were between 20 and 40 mm
(a) (b)
(c) (d)
Figure 11 Magneto-optical Kerr effect (MOKE) images of body domains in a 70 mm magnetite crystal. A maximum of six domains were observed
(top left). At saturation remanence, one large domain fills most of the crystal (top right). In reverse fields, a domain wall propagates across the
crystal (lower left) and ultimately the grain saturates in the opposite polarity (lower right). Modified from Heider F and Hoffmann V (1992)
Magneto-optical Kerr effect on magnetite crystals with externally applied magnetic fields. Earth and Planetary Science Letters 108: 131–138.
(a) (b)
(c) (d)
Figure 12 Chemical maps (blue: Fe; red: Ti) of a magnetite–ulvöspinel intergrowth and the magnetic microstructure of the same regions imaged by
off-axis electron holography in the transmission electron microscope. Magnetic field lines link neighboring magnetite crystals across intervening
nonferromagnetic ulvöspinel. Isolated crystals have vortex (c) and SD (d) structures, but interactions are so strong that groups of particles form
‘superdomains.’ In (c), the spins of three crystals link in a supervortex, while in (d), the three SD grains in the middle are flux-coupled with their moments
mutually antiparallel. Modified from Harrison RJ, Dunin-Borkowski RE, and Putnis A (2002) Direct imaging of nanoscale magnetic interactions in
minerals. Proceedings of the National Academy of Sciences of the United States of America 99: 16556–16561.
1.0
0.9
0.8
0.7
0.6
Mnet /Ms
0.5
Single-domain Two-domain states
0.4
or flower state Symmetrical vortex Twisted vortex
(a) (b) (c) 0.3
Figure 13 Schematic drawings of the energetically favored
0.2
micromagnetic states in small cubic particles with uniaxial anisotropy.
Vortex states evolve from structures with two body and two closure 0.1
domains, forming a closed internal flux linkage. Modified from Rave W,
Fabian K, and Hubert A (1998) Magnetic states of small cubic 0.0
particles with uniaxial anisotropy. Journal of Magnetism and Magnetic 0 100 200 300
Materials 190: 332–348. Grain edge length (nm)
Figure 15 Theoretical micromagnetic states of a uniaxial magnetite
grain as a function of grain size. An SD/flower structure with high M/Ms is
100
the minimum energy state at very small sizes and a vortex structure
Butler and Banerjee (1975a)
L00 with low M/Ms at large sizes. Between 70 and 140 nm, there is a
choice between SD and vortex LEM states. The state adopted by
Cube root of volume (mm)
the model grain then depends on the initial structure chosen: vortex
(triangles) or SD (squares). Modified from Fabian K, Kirchner A, Williams
rem
LSD W, Heider F, Leibl T, and Hubert A (1996) Three-dimensional
-1
10 coerc micromagnetic calculations for magnetite using FFT. Geophysical Journal
LSD International 124: 89–104.
are another (Fabian and Hubert, 1999; Halgedahl, 1991).

LSPM The right-hand domain in the grain of Figure 18 accounts
for more than half the area viewed and presumably for more
-2
10 than half the SIRM. Grains like these with domain walls far
0 0.2 0.4 0.6 0.8 1 from a uniform distribution could certainly explain why
1/q = X/Z the remanence does not plummet at the SD threshold (d0)r.
Figure 14 Theoretical stable single-domain and superparamagnetic The continuity of coercive force and other measures of
threshold sizes for magnetites of different sizes and elongations. stability is not so obvious. As Figure 17 demonstrates, even
The usual SD threshold is the upper limit for uniform magnetization in the a grain with a metastable SD remanent state has fairly low
remanent state, but this is often much larger than the limit for coercivity. Either a new domain wall (or walls) will nucleate
uniform magnetization throughout reversal (‘coherent rotation’), and spread through the grain or existing walls will be driven
which determines coercivity. Modified from Newell AJ and Merrill RT by the self-demagnetizing field Hd to equalize the size of
(1999) Single-domain critical sizes for coercivity and remanence. domains.
in size; theoretically, (d0)r is 1 mm or less. It is remarkable that

5.08.2.10 Superparamagnetism
such large grains could possess an SD LEM state at all, much less
exhibit such an unfavorable state with fair frequency. Such states Any magnetic structure can become unstable and transform if
were never seen after alternating fields (AFs) were applied. there is sufficient thermal energy available. The resulting ther-
The experimental evidence is clear that the SD threshold is mally activated changes in magnetization are at the heart of
not at all sharp in terms of the variation of properties like TRM acquisition and the thermal demagnetization of rema-
coercive force Hc and SIRM Mrs with grain size (Figure 19). nences of all types (Sections 5.08.3.2 and 5.08.5.3). If a system
Unlike the precipitous decrease predicted theoretically for contains relatively few coupled spins, the probability of corre-
SIRM (Figure 15), experimental magnetic properties, for mag- lated thermal impulses (spin waves) sufficient to rotate SD
netite at least, decrease gradually with increasing grain size. The moments or effect other changes in structure (e.g., SD ! vortex
survival above (d0)r of remanence intensities and stabilities or two-domain) may become significant even at room temper-
much higher than those of MD grains is called pseudo-single- ature. The size below which the anisotropy barrier to magneti-
domain (PSD) behavior. zation changes can easily be surmounted by thermal energy on
Metastable SD grains are one possible source of PSD a specified (usually laboratory) timescale is called the super-
behavior. Disproportionately large domains in MD grains paramagnetic (SP) threshold ds. ‘SP’ refers to the thermal
4 Extrapolation
1 MD
0.8 dmax Sp
ac
Length (mm)
ing To
0.6 uch
dmin /le i ng
SD+MD n gt
0.4 h=
I nt 0 .05 Séd (chain)
N er a Itaipu-1
oi c ti
nt n g
er
0.2 ac
tio
ns HCM12
SD
0.1 Malom-tó Increasing
(scattered) interactions
0.05
0 0.2 0.4 0.6 0.8 1
Axial ratio (width/length)
Figure 16 Minimum and maximum critical SD particle lengths, dmin and dmax, as a function of particle elongation. Strong particle interactions move
dmax to larger sizes, broadening the metastable SD region. Some interacting magnetosomes (Séd chain) fall within this predicted broader SD region.
Modified from Muxworthy AR and Williams W (2006) Critical single-domain/multidomain grain sizes in noninteracting and interacting elongated
magnetite particles: Implications for magnetosomes. Journal of Geophysical Research 111: B12S12. (http://dx.doi.org/10.1029/2006JB004588).
Table 3 Critical size limits d0 and ds for thermally stable single- ds ¼ ð25kT=K Þ1=3 [13]
domain behavior (room-temperature, equidimensional grains)
K can be due to any combination of crystalline, magnetoelastic,
Mineral Superparamagnetic Single-domain and shape anisotropies.
size, ds (nm) critical size, d0 (nm) Figure 14 indicates that ds is essentially equal to (d0)r or
(d0)c for equidimensional magnetite grains. Elongated grains
Iron (aFe) 8–26 17–23
have a significant but still narrow stable SD range. It is this very
Magnetite (Fe3O4) 25–30 50–84
Maghemite (gFe2O3) 60 restricted SD range that makes PSD remanence so important
Titanomagnetite, 80 200–600 for magnetite. In minerals less magnetic than magnetite, there
x ¼ 0.6 (TM60) is a larger difference between ds and d0 (Table 3).
Titanomaghemite
x ¼ 0.6, z ¼ 0.4 50 750 5.08.3 Magnetic Minerals and Their Properties
x ¼ 0.6, z ¼ 0.7 90 2400
Hematite (aFe2O3) 25–30 15 000 5.08.3.1 Thermomagnetic Curves, Curie Temperatures,
Pyrrhotite (Fe7S8) 1600 and SD Ranges
Source: Dunlop DJ and Özdemir Ö (1997) Rock Magnetism: Fundamentals and Thermomagnetic curves of saturation magnetization Ms versus
Frontiers. New York: Cambridge University Press, Table 5.1 temperature T (Figure 20) are often used in determining mag-
netic mineralogy. The Ms(T ) function itself plays a major role
in TRM acquisition and thermal demagnetization of a particu-
lar mineral (Sections 5.08.5.3 and 5.08.6.1). The Curie tem-
equilibrium orientation of grain moments in an applied mag- perature TC marks the ferromagnetic ! paramagnetic
netic field, which is analogous to paramagnetism except that transition, at which Ms ! 0.
grain moments contain hundreds or thousands of coupled Iron, hematite, and magnetite have TC values in the range
atomic moments, hence ‘super.’ 580–765 C (Table 4). Natural remanent magnetizations
In practical terms, ds represents the lower limit to stable SD (NRMs) of these minerals have high thermal stabilities, some-
behavior. SP is not a new state of matter but a limiting case of times remaining essentially unchanged in heating to 10–20 C
ferromagnetism or ferrimagnetism. Exchange coupling is not below TC except for reversible decreases dictated by Ms(T ).
overcome (that occurs at TC), but remanence and coercivity Among other things, this means that they remain sources of
drop to zero on any reasonable timescale of observation, caus- magnetic anomalies even when they occur at depths of tens of
ing a drop-off in Hc and Mtr at grain sizes smaller than those in kilometers in planetary lithospheres. Pyrrhotite has an interme-
Figure 19. To make a rough estimate of ds for equidimensional diate TC of 320 C and intermediate thermal stability. The
SD grains, we can equate the thermal energy of 25kT avail- 300 C isotherm in the crust marks the base of a magnetic layer
able to the system over times of seconds to minutes to the due to pyrrhotite. High-Ti titanomagnetite (Fe24Ti06O4 or
anisotropy barrier VK that must be surmounted for reversal TM60) and goethite (FeOOH) have TC 120–150 C and are
(normally coherent reversal for very small SD grains): effective magnetic sources only in the uppermost crust.
14 mm
10 mm
(a) (b) (c)
(d) (e)
Figure 17 Observed metastable SD saturation remanence state in a TM60 grain (a) and subsequent backfield nucleation (b) and propagation of a
domain wall (c–e) to a reversed near-saturation state. Reproduced from Halgedahl SL and Fuller M (1983) The dependence of magnetic domain
structure on magnetization state with emphasis on nucleation as a mechanism for pseudo-single-domain behavior. Journal of Geophysical Research 88:
6505–6522.
+H 10 mm
(a) +200 mT ( b) +2.5 mT ( c) IRMs ( d) –3 mT (e) ±H cycling
Figure 18 Nucleation of a domain wall in a TM60 grain in a small forward field (b) and alternative remanence states (c, e). The magnetization is
dominated by one large domain in (c) and (d); a single wall displacement transforms one state into the other and reverses the grain’s moment. Modified
from Halgedahl SL and Fuller M (1983) The dependence of magnetic domain structure on magnetization state with emphasis on nucleation as a
mechanism for pseudo-single-domain behavior. Journal of Geophysical Research 88: 6505–6522.
Relative magnitudes of Ms are also important: they deter- Magnetic signal, whether NRM or induced magnetization,
mine the critical SD size d0 (Table 3 and Figure 21). Because depends on more than the domain state. The high Mrs/Ms and
demagnetizing energy Ed is proportional to M2 (eqn [7]), broad SD range of minerals like hematite and goethite are
grains of weakly magnetic minerals like hematite and goethite offset by their weak Ms. A small fraction of strongly magnetized
(Ms 2 kA m1) have much lower Ed than similar-size grains magnetite of SD/PSD size will usually dominate the NRM even
of strongly magnetic minerals like magnetite if a rock also contains large amounts of hematite. On the other
(Ms ¼ 480 kA m1) or iron (Ms ¼ 1715 kA m1). They remain hand, thermal demagnetization will ultimately erase the mag-
SD, that is, magnetized to saturation (M ¼ Ms), at large sizes netite remanence, leaving a substantial part of the hematite
(up to 15 mm for hematite), while magnetite subdivides into NRM intact above 600 C.
domains above 0.07 mm and iron above 0.02 mm. All three
minerals have similar SP thresholds, 0.025 mm. Therefore,
5.08.3.2 Magnetite
much of the broad size spectrum of hematite in nature falls
in the SD range and has large Mrs/Ms (Section 5.08.4.1). Nat- The most important terrestrial magnetic minerals are oxides of
ural examples of SD magnetite are not so common and require iron and titanium. There are three series: (1) stoichiometric
a mechanism for producing needlelike grains or subgrains with titanomagnetites with spinel structure, solid solutions between
a very restricted size distribution (see Figures 14 and 16 and end-members magnetite (Fe2þFe3þ 2 O4) and ulvöspinel
Sections 5.08.3.2, 5.08.3.13, and 5.08.3.14). Naturally occur- (Fe2þ 4þ
2 Ti O4); (2) nonstoichiometric (oxidized) titanomagne-
ring SD iron is rare. On the Moon, only impact breccias with a tites or titanomaghemites, in which some Fe2þ ions migrate to
specific range of annealing temperatures develop SD iron the surface where they are converted to Fe3þ, leaving ordered
(Section 5.08.3.12). With this one exception, lunar rocks and vacancies in the spinel lattice; and (3) titanohematites (or
meteorites contain MD iron with very mobile domain walls – hemoilmenites) with rhombohedral structure, solid solutions
the worst imaginable material for paleomagnetic recording. between hematite (aFe3þ 2þ 4þ
2 O3) and ilmenite (Fe Ti O3). The
100 100
50
Crushed grains
Day et al. (1977)
20
Glass ceramic
Coercive force,µo Hc (mT)

10 + Worm and Market (1987) 10
5 +×
µo Hc (mT)
2 Oxalate decomposition
Schmidbauer and Schembera (1987) ×
1 × 1
Hydrothermally grown crystals ×
0.5 + Levi and Merrill (1978)
Özdemir and Banerjee (1982) ×
Dunlop (1986b) × ×
0.2 × Heider et al. (1987)
Amin et al. (1987)
0.1 Argyle and Dunlop (1990) 0.1
Özdemir and Dunlop (1995)
0.05
0.01 0.05 0.1 0.5 1 5 10 50 100 5001000
(a) Grain diameter, d (µm)
0.1 1 10 100 1000

1.0
Reduced saturation remanence, Mrs /Ms
Crushed grains
Day et al. (1977)
Glass ceramic
Worm and Market (1987)
Grown crystals
Heider et al. (1987) 0.1
0.1 Simple
Other data (after
domain
Argyle and Dunlop, 1990)
theory
Özdemir and Dunlop (1995)
Micromagnetic
theory
0.01 ‘Vortex’ 0.01
LEM
‘Diagonal’
state
LEM state
(after Newell et al., 1993b)
0.001 0.001
0.01 0.1 1 10 100 1000
(b) Grain size, d (µm)
Figure 19 Grain-size dependence of (a) coercive force Hc and (b) saturation remanence ratio Mrs/Ms for equidimensional magnetite grains of various
origins. Rather than the rather abrupt decrease at the SD threshold predicted theoretically, gradual decreases in properties are observed over
several decades of grain diameter. Modified from Dunlop DJ (1986) Hysteresis properties of magnetite and their dependence on particle size: A test of
pseudo-single-domain remanence models. Journal of Geophysical Research 91: 9569–9584; Dunlop DJ (1995) Magnetism in rocks. Journal of
Geophysical Research 100: 2161–2174; Argyle KS and Dunlop DJ (1990) Low-temperature and high-temperature hysteresis of small multidomain
magnetites (215–540 nm). Journal of Geophysical Research 95: 7069–7083.
1 Hematite
(aFe2O3)
M (normalized)
Iron
(aFe)
Magnetite
(Fe3O4)
Pyrrhotite
S
(Fe7S8)
Titanomagnetite
(Fe2.4Ti0.6O4)
0
20 200 400 600 800
Temperature, T (°C)
Figure 20 Normalized Ms(T ) dependences for five common magnetic minerals. Hematite is unique in having a ‘blocky’ Ms(T ) curve with a steep
descent just below the Curie temperature.
Table 4 Some magnetic properties of the more important magnetic minerals
Mineral Ms (kA m1) TC ( C) K1 (J m3) l 106
Iron (aFe) 1715 765 4.8 104 21 (l111)

21 (l100)
Magnetite (Fe3O4) 480 580 1.35 104 72.6 (l111)
19 6 (l100)
Maghemite (gFe2O3) 380 590–675 4.65 103 8.9
Titanomagnetite, x ¼ 0.6 125 150 2.02 103 142.5 (l100)
95.4 (l111)
Hematite (aFe2O3) 2.5 675 1.2 106 (c-axis) 8
Pyrrhotite (Fe7S8) 90 320 104 (c-plane) <10
d0 d0 d0
SD Non- SD Non- SD Non-

uniform uniform uniform
Titanomagnetite
(Fe Ti O ) Hematite
Magnetite
(aFe O )
2.4 0.6 4
(Fe O ) Pyrrhotite 2 3
3 4
(Fe S8)
7
Optical limit
0.01 0.1 1 10 100

Crystal size, d (mm)
Figure 21 Approximate SD remanence threshold sizes for magnetite, TM60/pyrrhotite, and hematite.
compositions of all three series are conveniently represented in weight of Fe3O4, close to the theoretical moment of Fe2þ of
a ternary FeO–Fe2O3–TiO2 diagram (Figure 22). 4mB. This translates into a spontaneous magnetization
Magnetite is the single most common magnetic mineral on the Ms 480 kA m1 at ordinary temperatures and about half
Earth. It occurs as a primary magmatic mineral in plutonic rocks this amount at 500 C.
and as an end product of either deuteric oxidation or maghemi- Below the ‘Verwey transition’ at TV 120 K, the magnetite
tization followed by inversion (see the succeeding text) in mafic lattice distorts slightly from cubic to monoclinic. The main
lavas, dikes, and sills. It is equally common as a primary or magnetic effect is a greatly increased magnetocrystalline anisot-
secondary mineral in sedimentary and metamorphic rocks. ropy below TV. In addition, at a temperature TK slightly above
Magnetite is cubic with inverse spinel structure (Dunlop TV, the easy axes switch from h111i to h100i. This transition is
and Özdemir, 1997, Figure 3.4). The O2 ions form a slightly called the ‘isotropic point’ because the first magnetocrystalline
distorted cubic close-packed lattice with Fe2þ and Fe3þ cations anisotropy constant K1 momentarily passes through zero in
in interstitial sites. Each cation is surrounded by either four both heating and cooling. The effect on SD remanence is
O2 ions in tetrahedral coordination (A sites) or six O2 ions usually minor (although not always) because the anisotropy
in octahedral coordination (B sites). A unit cell contains barrier is mainly due to shape anisotropy and Ms is continuous
32 O2 ions, 8 Fe2þ on B sites, and 16 Fe3þ shared equally at both TV and TK. Unpinning of domain walls can be consid-
between A and B sites. erable, however. This effect forms the basis of ‘low-temperature
The exchange coupling between Fe cations is indirect (super- demagnetization (LTD),’ described in Section 5.08.5.2.
exchange), with two O2 2p orbitals providing the linkage Magnetite, being a cubic mineral, prefers equant crystal
between the Fe 3d orbitals. The tetrahedral bonds are in h111i forms (cubes, octahedra, and dodecahedra). Crystals of these
directions and the octahedral bonds are in h100i directions. forms have little or no shape anisotropy, and their magneto-
Exchange coupling is relatively ineffectual for bond angles near crystalline anisotropy leads to modest Hc values of 10–15 mT.
90 . For this reason, A–A and B–B exchange couplings are weak, These values are not much higher than typical coercivities for
and the dominant exchange interaction is the negative A–B MD magnetite. Other paleomagnetically undesirable traits of
coupling, which gives magnetite its ferrimagnetic character. equidimensional crystals are their narrow SD ranges
The A and B sites can be thought of as two different magnetic (Figure 14) and low resistance to LTD when cooled through
sublattices, with oppositely directed cation moments. TK and TV.
The inverse spinel structure, with trivalent cations equally Fortunately, there are several natural mechanisms that pro-
partitioned between A and B sites and the Fe2þ ions confined duce at least limited quantities of elongated magnetite crystals
to the B sublattice, produces a spontaneous magnetic moment or chains of crystallites. First is the process of high-temperature
at low temperatures of 4.1mB (Bohr magnetons) per formula oxidation, which generates a lath-like intergrowth of ilmenite
TiO2 (i.e., Ti4+)

rutile, anatase
1 1
Fe2TiO4 FeTi2O5
3 0 3
100
200
300
400 0
1
FeTi2O5 45 0
50
3
550
600
1
FeTiO3 FeO 1 1
3 Fe3O4 Fe2O3
ilmenite 3 2
(b)
Z Pseudobrookite
1
Fe2TiO4 0 0.2 0.4 0.6 0.8 1
3 Fe2TiO5
ulvöspinel 3
Tit Tit
an an 0.8
o
om Titano hem Maghemites
0.6
ag ati
ne tes X
tite 0.4
s
0.2
0
FeO 1 1
Fe3O4 Fe2O3
(i.e., Fe2+) 3 2
magnetite (i.e., Fe3+)
hematite
(a) maghemite
Figure 22 Ternary diagram showing the iron–titanium oxides and their solid solutions. Horizontal dashed lines are titanomaghemite oxidation
trends at constant Ti content x; cross cutting dashed lines are contours of oxidation parameter z. Titanomaghemite Curie temperatures are contoured
over the entire oxidation field (upper right). Modified from O’Reilly W (1976) Magnetic minerals in the crust of the Earth. Reports on Progress
in Physics 39: 857–908.
lamellae on conjugate magnetite {111} planes. Ilmenite Titanomagnetites have the same inverse spinel structure as
extracts Ti from the titanomagnetite, leaving a Ti-poor phase magnetite but the A–B exchange interaction weakens, and the
(‘near-magnetite’ with TC usually 500 C) in the form of Curie temperature therefore falls, more or less linearly, with
elongated prisms sandwiched between the ilmenite lamellae. increasing Ti content x. For compositions beyond x 0.8, TC is
The second process is precipitation of Fe at high temperatures below room temperature and the oxides are paramagnetic. The
from the lattices of silicate minerals like plagioclase, pyroxenes, replacement of Fe on octahedral sites by nonmagnetic Ti4þ
amphiboles, micas, and olivine (Section 5.08.3.13). In some causes the spontaneous magnetization to decrease as x
cases, crystallographically oriented nanometer-size needles of increases. The low-temperature spontaneous moment drops
magnetite are formed, most commonly in clinopyroxene and almost linearly from 4mB in magnetite to 0 in ulvöspinel.
plagioclase (Feinberg et al., 2005). The third mechanism is These variations of TC and Ms are useful in determining x.
biogenic production of crystallographically aligned nanocrys- However, they are not diagnostic because cations like Al, Mg,
tals of magnetite in elongated chains, for example, by magne- and Mn often substitute for Fe in the lattice.
totactic bacteria (Section 5.08.3.14). The individual SD-size On the Earth, two compositions are commonly found as
crystals interact strongly, so that the chains act as collective SD stoichiometric titanomagnetites. Fe2.4Ti0.6O4 (called TM60
particles, although reversing noncoherently. because it contains 60 mol% Ti) is the primary oxide in mid-
ocean ridge basalts (MORBs) and in subaerial basalts (plumes
and large igneous provinces). Actual compositions range from
5.08.3.3 Titanomagnetites
TM50 to TM70. The Curie point is 150 C Ms 125 kA m1,
There is a complete solid solution between magnetite and ulvös- about one-fourth that of magnetite (Table 4). Shape anisotropy
pinel at temperatures well above the magnetite Curie point. In is correspondingly weaker than in magnetite, and magnetocrys-
principle, any intermediate composition TixFe3xO4 could be talline anisotropy (with positive K1 and h100i easy axes) is also
preserved metastably at ordinary temperatures by quenching in less than that of magnetite. The magnetostriction constants l100
an oxygen-poor environment from the melt. In massive flows, and l111 are two to three times those of magnetite, however. The
stoichiometric titanomagnetites may survive essentially unoxi- coercivity of TM60 is in large part magnetoelastic. This stress
dized for millions of years. In submarine pillow lavas, however, sensitivity of TM60 makes domain observations difficult.
maghemitization begins almost immediately after extrusion, TM10–TM30 occur in andesitic and dacitic volcanic
probably aided by iron-leaching bacteria in the seawater rocks. The iron-rich titanomagnetites studied by Özdemir
(Carlut et al., 2007). For this reason, linear magnetic anomalies and Dunlop (2003) had TC ¼ 460–490 C, indicating
over the oceans are strongest over spreading ridges and rapidly x 0.10–0.15. Even 10 mol% Ti was enough to depress TK to
decrease in amplitude away from the ridges. 40–55 K (Figure 23), making LTD an inconvenient treatment
(since TK is well below the liquid nitrogen temperature, 77 K). 5.08.3.4 Maghemite and Titanomaghemites
Sixty mol% Ti suppresses the transition, so that remanences
Partially oxidized titanomagnetites are commoner in nature
produced at either T0 or low T change continuously between
than stoichiometric oxides. They are called ‘titanomaghemites’
50 and 200 K (Moskowitz et al., 1998).
by association with ‘maghemite’ (gFe2O3), the fully oxidized
end product of magnetite low-temperature oxidation. Maghe-
mitization is an iron-leaching process, usually occurring in the
2.0
presence of water, that leaves the spinel lattice intact but one-
MSIRM (emu g-1 or A m2 kg-1)
sixth of the octahedral sites vacant. Ms of maghemite is there-

1.6 fore somewhat lower than that of magnetite although TC is
higher, intermediate between those of magnetite and rhombo-
1.2 H-8 hedral hematite (aFe2O3).
Titanomagnetite Lattice mismatch between the oxidized surface and unoxi-
single crystal
0.8 dized core of partially oxidized (or cation-deficient) magnetite
haruna dacite
produces a strained transition region between them. The result-
0.4 ing internal stress leads to coercivities higher than those of either
gFe2O3 or Fe3O4 separately. The Verwey transition is also broad-
ened (Figure 24), a diagnostic fingerprint of partial oxidation of
0
0 50 100 150 200 250 300 magnetite (Özdemir et al., 1993). Maghemite itself has no Ver-
(a) Temperature, T(K) wey transition. In nature, maghemitization often accompanies
soil formation. In rocks, it suggests a nonprimary NRM resulting
0.25 from chemical processes (Section A.5.3).
(emu g-1 or A m2 kg-1)
Although even a secondary NRM can be of interest paleo-

H-8
Single crystal magnetically, maghemite NRM is difficult to work with. At
0.20
temperatures of 250–550 C, maghemite inverts to aFe2O3 by
restacking the lattice and loses most of its magnetism. Ther-
0.15
mal demagnetization of a maghemitized rock is usually a
losing proposition. The mineralogy changes irreversibly
0.10
Titanomagnetite before the NRM components can be ‘cleaned’ and isolated
(Sections 5.08.5.3–5.08.5.5), except when aluminum is
SIRM
0.05 Haruna dacite

incorporated into the lattice (Wilson, 1961), leading to inver-
M
sion temperatures as high as 700 C. Aluminum-substituted

0 maghemites are common in soils.
0 50 100 150 200 250 300
(b) Temperature, T(K) Titanomaghemites formed by oxidation of TM60 have Ms
values that decrease but TC values that increase with increasing
Figure 23 Zero-field temperature cycling of 20 K SIRM (a) and 300 K
oxidation parameter 2 (Figure 25). Coercive force Hc at first
SIRM (b) of a crystal of iron-rich titanomagnetite (x 0.15) from Mt.
rises and then falls sharply as synthesized SD TM60 is oxidized
Haruna, Japan. The isotropic point around 45 K is well marked in (a) but
less sharply defined in (b). Modified from Özdemir O and Dunlop DJ (Figure 26). Thus, low-temperature oxidation of TM60 in
(2003) Low-temperature behavior and memory of iron-rich MORB may ultimately destabilize the NRM. This, rather than
titanomagnetites (Mt. Haruna, Japan and Mt. Pinatubo, The Philippines). decreasing Ms, is probably why magnetic anomaly amplitudes
Earth and Planetary Science Letters 216:193–200. decay within at most 1 Ma after the formation of MORB.
0 50 100 150 200 250 300

0 50 100 150 200 250 300 1.0 1.0
1.0 1.0
Oxidized Fe3O4
0.8 0.8
0.8 Stoichiometric Fe3O4 0.8
0.6 0.037 mm 0.6

0.6 0.6
M/Mo
M/Mo
0.076 mm
0.4 0.10 mm 0.4
0.4 0.4
0.037 mm
0.22 mm
0.076 mm
0.2 0.2 0.2 0.2
0.10 mm
0.22 mm
0 0 0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
( a) Temperature, T(K) Temperature, T(K)
(b)
Figure 24 Zero-field warming curves of 5 K SIRM for (a) stoichiometric and (b) surface oxidized (maghemitized) magnetites. The Verwey
transition is blurred and ultimately suppressed by oxidation. Modified from Özdemir Ö, Dunlop DJ, and Moskowitz BM (1993) The effect of oxidation on
the Verwey transition in magnetite. Geophysical Research Letters 20: 1671–1674.
500 Synthetic titanomaghemite

200 x = 0.6, z = 0.3
400
Curie temperature, Tc (°C)
150
Ms (kA m–1)
300 Oxidized TM60
100
200
50
100 0
20 100 200 300 400 500
Temperature, T(C)
20
Figure 27 Characteristic irreversible thermomagnetic curve of a
0 0.2 0.4 0.6 0.8 1.0 partially oxidized TM60. Modified from Özdemir Ö and O’Reilly W (1982a)
Oxidation parameter, z Magnetic hysteresis properties of synthetic monodomain
Figure 25 The Curie temperature as a function of maghemitization for titanomaghemites. Earth and Planetary Science Letters 57: 437–447.
TM60. Data: circles, Özdemir Ö and O’Reilly W (1982a) Magnetic
hysteresis properties of synthetic monodomain titanomaghemites. Earth a Ti-rich rhombohedral phase (‘near-ilmenite’). The rhombo-
and Planetary Science Letters 57: 437–447; triangles, Brown K and hedral phase grows in {111} planes of the spinel, eventually
O’Reilly W (1988) The effect of low-temperature oxidation on the producing a trellis-like texture of intergrown phases very like
remanence of TRM-carrying titanomagnetite Fe2 4Ti0 6O4. Physics of the that resulting from ‘high-temperature oxidation’ (next section).
Earth and Planetary Interiors 52: 147–154. The signature of inversion is an irreversible thermomag-
netic curve, in air or in vacuum (Figure 27). Above 250 C
but before the titanomaghemite reaches its TC, Ms begins to
×105
increase and only drops to zero above 500 C at the Curie
Hc
1.5 temperature of a considerably more Fe-rich spinel. During
cooling, Ms is much higher than during heating, sometimes
c
by a factor of four to five. This process is capable of producing a
magnetic layer in the seafloor extending to greater depth (the
Coercive force, Hc (A m–1)
500 C isotherm) than fresh MORB.
1.0
5.08.3.5 High-Temperature Oxidation
0.8
Subaerial basalts exhibit a wide range of oxidation states in the
Initial susceptibility, Xo
TM60 interior of thick flows (Figure 28). Oxidation occurs quite

ATM60/10 0.6 rapidly in the melt, producing a phase assemblage similar
ATM60/20 (for moderate oxidation) to that accompanying inversion of
0.5 titanomaghemite. However, cation-deficient phases are not
0.4
part of the assemblage, except possibly during initial single-
phase oxidation. The observed phases represent the equilib-
0.2 rium under high-temperature oxidizing conditions if cooling is
not too rapid.
The ‘oxyexsolution’ intergrowth texture (Figure 29), if
0 0
0 0.2 0.4 0.6 0.8 1.0 sufficiently finely divided, can generate magnetite rods of SD
Oxidation parameter, z size. Although individual magnetite crystals are separated by
nonferromagnetic ilmenite, they do not behave independently.
Figure 26 Coercive force and initial susceptibility of SD pure and
Al-substituted TM60 as a function of maghemitization. Modified from The strong interactions among crystals are similar to those
Özdemir Ö and O’Reilly W (1982a) Magnetic hysteresis properties of shown in Figure 12 for unoxidized titanomagnetite exsolved
synthetic monodomain titanomaghemites. Earth and Planetary Science into magnetite þ ulvöspinel.
Letters 57: 437–447. For sufficiently high oxygen fugacities and/or long-term
heating, the ultimate phase assemblage is hematite þ
If titanomaghemites are heated, by burial beneath later pseudobrookite (Fe2TiO5) and rutile (TiO2). The hematite
flows on the seafloor, for example, they become unstable and (next section) imparts long-term and high-temperature stability.
‘invert.’ This is the analog process to inversion of Pseudobrookite and especially rutile testify to terminal oxida-
gFe2O3 ! aFe2O3. However, the inversion products are more tion. The presence of pseudobrookite, which cannot form below
varied and so are their intergrowth textures. The usual phase 585 C, also guarantees that oxidation did not continue below
assemblage is a Ti-poor titanomagnetite (‘near-magnetite’) and TC of magnetite. Any magnetite present carries a primary TRM.
1.0
Saturation magnetization, M (normalized) Hawaii

(quenched
1.0 from < 350° C)
1.0
500 600
Iceland
(oxidation
index 1.8)
300 400 500 600
Hawaii
(quenched
from >950° C)
0
0 100 200 300 400 500 600
Temperature, T (C)
Figure 28 Characteristic vacuum thermomagnetic curves of natural

stoichiometric and high-temperature oxidized titanomagnetites, showing Figure 29 Electron micrograph (magnification 10 400) of the
a single TM60 Curie point (lower), a single near-magnetite Curie point oxyexsolution intergrowth texture of high-temperature oxidized
(upper, high oxidation state), and both Curie points (middle, intermediate titanomagnetite. Modified from Gapeev AK and Tsel’movich VA (1983)
oxidation state). Modified from Gromme CS, Wright TL, and Peck DL Microstructure of synthetic titanomagnetite oxidized at high partial
(1969) Magnetic properties and oxidation of iron–titanium oxide minerals pressures of oxygen (translated from Russian), Izvestiya. Physics of the
in Alae and Makaopuhi lava lakes, Hawaii. Journal of Geophysical Solid Earth 19: 983–986.
Research 74: 5277–5293.
More vital in most applications is the stability of hematite
5.08.3.6 Hematite NRM at and above room temperature. Magnetocrystalline
anisotropy in the c-plane is small, but magnetoelastic anisotropy
Hematite (aFe2O3), with TC ¼ 675 C, is the fully oxidized end leads to high coercivities in SD crystals (Özdemir and Dunlop,
product of magnetite. It is both thermally and chemically 2002). MD (>15 mm) hematites are less frequent in most rocks
stable. Its rhombohedral crystals nucleate and grow topotacti- and sediments, but they too have adequate coercivities to pre-
cally in magnetite {111} planes. The quantity of hematite on serve long-term NRM (Figure 30). The Ms(T ) curve of hematite
the surface of the Earth and other planets with oxidizing atmo- is unusually ‘blocky’ in form (Figure 20), resulting in TRM
spheres, notably Mars, far exceeds that of magnetite, but hema- blocking temperatures clustered within 20–30 C of TC for all
tite is seldom dominant magnetically because of its small Ms. but the finest nanometer-size grains. Thus, hematite has unpar-
At ordinary temperatures, hematite is a slightly imperfect alleled stability against changes in both field and temperature.
antiferromagnet with Ms 2.5 kA m1. The spin sublattices lie
in the c-plane, also called the basal plane because of the tabular
form of single crystals. Within the c-plane, the sublattices are
5.08.3.7 Titanohematites (Hemoilmenites)
slightly tilted or canted with respect to each other, giving rise to
a tiny but persistent magnetic moment perpendicular to the Titanohematite TiyFe2yO3 is a rhombohedral mineral with
spins (Figure 2). This moment rotates relatively freely in the bulk composition intermediate between hematite and ilmen-
c-plane but not at all out of the plane until the crystal is cooled ite. As with titanomagnetites, single-phase titanohematites of
below the Morin transition TM 260 K. At this point, the spin intermediate composition can only be preserved metastably at
sublattices rotate to the c-axis, and with this new crystal sym- room temperature by quenching from the melt. The cation
metry, spin canting in principle disappears. distribution ideally is ordered (Fe2þ and Ti4þ ions segregated
Curiously enough, a fraction (usually but not always small) on alternate c-planes) for y 0.5 and disordered (Fe2þ and Ti4þ
of hematite’s ferromagnetic moment survives below TM. This ions in equal numbers on all c-planes) for y
0.5.
low-temperature moment is usually ascribed to either chemical The ordered distribution gives rise to ferrimagnetism, but
impurities or structural imperfections, hence the name ‘defect’ TC drops below room temperature if y > 0.7 (Figure 31), while
moment. Whatever its origin, the defect moment serves to for y < 0.7, cation ordering is only partial. Ferrimagnetic
renucleate a fractional memory of any remanence that existed hemoilmenites are uncommon in nature. Even andesites and
above TM when the crystal is reheated to room temperature. It dacites with intermediate Fe/Ti ratios usually contain more
has only recently been discovered that cycling through TM is titanomagnetites than titanohematites.
incidental to memory: a remanence produced ‘below’ TM gener- Disordered titanohematites are weakly ferromagnetic
ates a much larger remanence, in the same direction, above TM antiferromagnets like hematite. The unique phenomenon of
(Özdemir and Dunlop, 2005). The importance of these obser- ‘self-reversing TRM’ can occur if the bulk hemoilmenite com-
vations is that hematite cycled repeatedly through TM in nature, position is in the ferrimagnetic range (y 0.5) and the oxides
for example, on Mars, will retain a fraction of its original NRM. are exsolved on a microscale into intergrown ordered and
disordered phases (Harrison et al., 2000). The phases are neg- show full self-reversals (Ozima and Funaki, 2003). This may be
atively exchange-coupled. When cooled in a small magnetic because the compositions or interlayering of the hemoilme-
field, the weakly magnetic disordered phase is the first to nites is not optimal for strong coupling or because tita-
acquire TRM (in the field direction) because of its higher TC. nomagnetites dominate the whole-rock TRM.
With further cooling through TC of the ordered phase, this A recently discovered phenomenon in slowly cooled rocks
strongly ferrimagnetic phase acquires a much larger remanence where exsolution of titanohematite has gone to completion is
negatively coupled to the TRM of the first phase, essentially ‘lamellar magnetism’ (Robinson et al., 2002). The bulk com-
ignoring the external field. The net result is a self-reversed TRM position of these oxides is in the antiferromagnetic range
(Lagroix et al., 2004; Prévot et al., 2001). (y
0.5). The finer the scale of the exsolution lamellae, the
Since its discovery by Uyeda (1958), self-reversal has fasci- stronger is the magnetism, which is the outcome of imbalance
nated rock magnetists. The fact is that it is a rarity in nature. between oppositely directed Fe2þ spins at interfaces between
Samples from Uyeda’s type locality, the Haruna dacite, seldom lamellae (Figure 32). According to Robinson et al., Ms values
as large as 55 kA m1 are possible by this mechanism. What is
certain is that titanohematites with compositions around
y ¼ 0.25 can impart NRMs as large as 30 A m1 to anorthositic
M (A m2 kg-1)
and noritic rocks (McEnroe et al., 2004), on the same order as
0.4
the TRM of fresh MORB.
MS = 0.36 A m2 kg-1
0.3
MRS/ MS = 0.87
HC = 8.7 mT 0.2
HCR/HC = 1.07 5.08.3.8 Iron Oxyhydroxides
0.1
The most important natural hydrous iron oxide is orthorhom-
bic goethite (aFeOOH), a common weathering product and
-200 -100 100 200
precursor to hematite in sediments and soils. Goethite is anti-
H (mT) ferromagnetic with a weak superimposed ferromagnetism, but
aFe2O3 its Curie temperature is only 120 C. Unlike hematite, the
Single crystal sublattice magnetizations lie along the c-axis and so does the
H//(0001) weak ferromagnetism. TRM produced parallel to the c-axis is
20 times larger than TRM perpendicular to the c-axis (Özdemir
and Dunlop, 1996). Even the c-axis TRM is weaker than TRM of
Figure 30 Typical MD hysteresis curve of a large hematite crystal. The
hematite but it is extremely hard. Typical coercivities are several
intrinsic magnetic properties are revealed because hematite is weakly
magnetic and lacks the self-demagnetizing fields that control the teslas, higher even than those of hematite and far beyond the
hysteresis response of strongly magnetic minerals like magnetite (see capabilities of AF demagnetizers (Rochette et al., 2005a).
Figure 38(a)). Modified from Özdemir Ö and Dunlop DJ (2005) Blocking temperatures of TRM are usually above 90 C,
Thermoremanent magnetization of multidomain hematite. Journal of enough to preserve goethite NRM in nature but very low com-
Geophysical Research 110: B09104. (http://dx.doi.org/10.1029/ pared with those of other minerals, making goethite the first
2005JB003820). mineral to be cleaned in thermal demagnetization.
40
600 Titanohematite
Ferrimagnetic 30
400
Ms (A m2 kg–1)
Tc (C)
200 20
To = 20 C
0
10
–200
0
0 0.2 0.4 0.6 0.8 1.0
Composition parameter, y
Figure 31 The Curie temperature and spontaneous magnetization of titanohematites as a function of Ti content y. TC drops below room temperature
around y ¼ 0.7. The ferrimagnetic region (shaded) is therefore rather restricted. Modified from Hunt CP, Moskowitz BM, and Banerjee SK (1995)
Magnetic properties of rocks and minerals. In: Ahrens TJ (ed.) Rock Physics and Phase Relations: A Handbook of Physical Constants, vol. 3, ch. 14,
Washington, DC: American Geophysical Union, pp. 189–204.
(001) layers
Fe3+
Fe3+
Fe3+
Fe3+ Antiferromagnetic
Fe3+ hematite
Fe3+
Fe3+
Fe3+
Fe2+ + Fe3+ Contact layer
Ti
Fe2+
Ti
Fe2+ Paramagnetic
Ti ilmenite
Fe2+
Net Ti
ferrimagnetic Fe2+
moment Ti
Fe2+ + Fe3+ Contact layer
Fe3+
Fe3+
Fe3+
Fe3+ Antiferromagnetic
Fe3+ hematite
Fe3+
Fe3+
Fe3+
Fe3+
Figure 32 Monte Carlo simulation of cation ordering and resulting magnetic signal of hematite-rich and ilmenite-rich phases in exsolved titanohematite.
‘Lamellar magnetism’ is associated with boundary layers between the exsolved phases. Modified from Robinson P, Harrison RJ, McEnroe SA, and
Hargraves RB (2002) Lamellar magnetism in the haematite–ilmenite series as an explanation for strong remanent magnetization. Nature 418: 517–520.
Goethite may dehydrate under natural conditions, usually common in soils. Ferrihydrite’s main interest is as a tracer of
with mild heating, to form hematite. Goethite’s needlelike biogeochemical cycling of Fe, not as a major remanence carrier.
crystal form is preserved but each original crystal consists of Also found in soils and sediments is lepidocrocite (gFeOOH),
many hematite crystallites. The effective grain size is often which is antiferromagnetic at low temperatures and paramag-
below 30 nm, the SP threshold of hematite. Hematite formed netic at room temperature. Its main interest comes from the fact
in this way carries no NRM. Magnetic parameters useful for that it dehydrates when heated above 200 C to strongly mag-
discriminating between hematites and goethites of different netic maghemite (Gendler et al., 2005; Özdemir and Dunlop,
grain sizes have been developed by Maher et al. (2004). 1993b). This poses no problem in zero-field heating but could
Ferrihydrite is a poorly crystalline iron hydroxide or lead to a substantial chemical remanent magnetization (CRM) in
hydrous ferric oxide that lacks long-range order. It is wide- experiments like the Thellier paleointensity determination
spread in aquatic sediments, where it generally forms with (Section 5.08.5.5), which involve in-field heating–cooling
the participation of Fe-oxidizing bacteria, for example, near cycles. However, materials containing lepidocrocite are unlikely
hydrothermal vents on mid-ocean ridges. In terrestrial soils to be used in paleointensity work.
such as tropical laterites, ferrihydrite is a common intermediate
weathering product of iron oxides or sulfides, later recrystalliz-
5.08.3.9 Iron Sulfides
ing to goethite or hematite. At room temperature, ferrihydrite
has an extremely weak ferromagnetism, with an SIRM Pyrrhotite (Fe1xS) is common in terrestrial rocks as an acces-
Mrs 104 Am2 kg1 (Pannalal et al., 2005). This is three sory phase, but it seldom dominates the magnetic properties.
orders of magnitude less than that of hematite or goethite. The situation is different in Martian rocks, for example, the
However, the grain-size dependence of ferrihydrite’s magnetic SNC meteorites (Rochette et al., 2005b). Pyrrhotite is the stable
behavior is dramatic (Guyodo et al., 2006), and even 1–2% iron sulfide for oxygen fugacities fO2 < FMQ (fayalite
silica substitution produces highly ordered ferrihydrite with a (FeSiO4) $ magnetite þ quartz), whereas paramagnetic pyrite
Curie point above room temperature (Berquó et al., 2007). (FeS2) is dominant under the more oxidizing conditions in
This is of importance because ferrihydrite is the ubiquitous the Earth’s crust (Lorand et al., 2005). Martian meteorites
first product of silicate diagenesis and silica substitution is contain about as much pyrrhotite as magnetite.
TC 330 C, close to TC of troilite (FeS) and monoclinic pyrrho-

RV-06-D tite. Despite a crystalline anisotropy constant of only 103 Jm3,
Susceptibility, ko (10–6 SI) 200 Riverbank sediments the coercivities of SD greigite are large enough to preserve a stable
NRM (Diaz Ricci and Kirschvink, 1992). A hallmark of natural
100 SD greigite is its high ratio of Ms/kf compared with pyrrhotite and
magnetite of similar sizes (Roberts, 1995).
0
5.08.3.10 Iron–Chrome Spinels (Chromites)
–100
Chromites, members of the solid-solution series FeCr2O4–
Fe3O4 (also containing variable amounts of Mg, Al, and Ti),
are cubic spinel minerals found in nature in such diverse
0 50 100 150 200 250 300 350
occurrences as meteorites (Weiss et al., 2002), lunar igneous
Temperature, T (C) rocks, terrestrial andesitic pumice (Yu et al., 2001), and sub-
Figure 33 Hopkinson peak (Section 5.08.4.3) in initial susceptibility ko marine gabbro and peridotite (Dunlop and Prévot, 1982);
of riverbank sediments from Taiwan just below the 320 C Curie TC ¼ 185 C for FeCr2O4 and increases with decreasing Cr
point of pyrrhotite, Fe7S8. The overall negative susceptibility is due to content. A few naturally occurring chromites carry a significant
diamagnetism of the sample holder (see Section 5.08.1.1). Modified NRM. Examples are the Kurokami pumices of Mt. Sakurajima,
from Horng C-S and Roberts AP (2006) Authigenic or detrital origin of Japan (Yu et al., 2001), where the NRM is shared with titano-
pyrrhotite in sediments? Resolving a paleomagnetic conundrum. Earth magnetite, and Martian meteorite SaU 008 (Yu and Gee,
and Planetary Science Letters 241: 750–762.
2005), where the chromite dominates the NRM. SaU 008 is
an unusual shergottite of plutonic origin, perhaps akin to the
Natural terrestrial pyrrhotite is a mixture of monoclinic Fe7S8 gabbro/peridotite affinities of terrestrial chromites.
and hexagonal Fe9S10 and Fe11S12. The deficiency of Fe relative
to S in these minerals results in lattice vacancies. In Fe7S8, these
5.08.3.11 Iron Carbonates
vacancies are confined to one of the two magnetic sublattices,
giving rise to ferrimagnetism with Ms 80–90 kAm1 and The most common iron carbonate is siderite (FeCO3), a fre-
TC 320 C (Figure 33) and a monoclinic distortion of the quent constituent of carbonate sediments and rocks on Earth.
basically hexagonal lattice. Magnetic properties as a function of Siderite has a low-temperature transition at 30–35 K (Housen
grain size are reported by Dekkers (1988, 1989). A distinctive et al., 1996), distinctively different from that of pyrrhotite
low-temperature transition around 34 K is diagnostic of the (Section 5.08.3.9). Siderite is paramagnetic at ordinary tem-
presence of monoclinic pyrrhotite in a rock or sediment peratures and carries no NRM. Its interest is as a source of
(Dekkers et al., 1989; Rochette et al., 1990). diagenetic magnetite. On Earth, the oxidation of siderite at
The c-axis is hard in pyrrhotite, as in hematite above TM, with room temperature and accompanying CRM acquisition is
sublattice magnetizations confined to the c-plane. Within the well documented (Hus, 1990). The process becomes rapid
c-plane, triaxial magnetocrystalline anisotropy is much stronger and even explosive at higher temperatures, and the magnetite
than in hematite and overshadows any stress-induced anisot- CRM effectively masks any remaining NRM of the carbonate
ropy. For this reason, pyrrhotite is the only mineral whose rock being thermally demagnetized. Decomposition of siderite
domain patterns can easily be viewed on a {0001} crystal in the crust of Mars is more hypothetical but is certainly pos-
plane without removing the stressed layer due to polishing. sible given adequate heat sources (Scott and Fuller, 2004). It
Pyrrhotite generally becomes unstable on heating, for could provide a potent source of magnetite crustal
example, during thermal demagnetization, and oxidizes or magnetization.
decomposes to magnetite (Bina and Daly, 1994). The transfor-
mation occurs well above 320 C, so that no information
5.08.3.12 Iron and Iron–Nickel
about the pyrrhotite NRM is lost, but below TC ¼ 585 C of
magnetite. If the field is not perfectly zeroed, a CRM will be Iron and iron–nickel are the principal NRM carriers in lunar rocks
produced, corrupting the magnetite NRM. and most meteorites. Chondritic meteorites also contain varying
Hexagonal pyrrhotite (Fe9S10) is basically antiferromag- amounts of magnetite. Body-centered cubic kamacite (aFe) is
netic but has a restricted ferrimagnetic range due to thermally ferromagnetic with Ms ¼ 1715 kA m1 and TC ¼ 765 C, making
activated vacancy ordering above the l-transition, so called it the epitome of intense and thermally stable magnetization.
because of the shape of the Ms(T ) curve between the onset of There are two drawbacks to iron as a remanence carrier, however.
ferrimagnetism around 200 C and its loss at TC 265 C The first is its extremely narrow SD range: d0 and ds are theoreti-
(Schwarz and Vaughan, 1972). In rapid cooling, the vacancy cally and experimentally very close (23 and 8 nm, respectively)
ordering and ferrimagnetism can be preserved metastably (Butler and Banerjee, 1975b; Kneller and Luborsky, 1963;
below 200 C. Table 3). Only elongated particles are likely to be SD, and iron,
Both pyrrhotite and greigite (Fe3S4) are frequent diagenetic being cubic, does not readily form such particles.
minerals in anoxic, for example, sulfate-reducing, sedimentary The second problem is the transformation during heating
environments (Roberts and Turner, 1993). Greigite has the same from kamacite to face-centered cubic taenite (gFe). The trans-
inverse spinel structure as magnetite but is less magnetic and formation occurs below 765 C if more than 5–10% Ni is
less anisotropic than its oxide cousin: Ms 125 kAm1 and alloyed with Fe. In this case, the NRM, a phase transformation
g Fe
800 (taenite) Fe–Ni
Tc(a)
Temperature, T (C)
600 a
g
aFe (h
a (heating), eatin Tc(g)
(kamacite) g)
g (cooling)
400
g ooli
a
(c
200 Tetrataenite
ng
)
0
0 10 20 30 40 50 60
(Fe) Mole percent Ni
Figure 34 Part of the phase diagram of Fe–Ni. Body-centered cubic kamacite (a-Fe) does not exist for >30 mol% Ni. Face-centered cubic taenite
(g-Fe) is the stable phase at high temperature and at all T for >30 mol% Ni. The ordered phase tetrataenite can exist for 50–55 mol% Ni.
CRM rather than TRM, is destroyed during thermal demagne-

tization before the kamacite Curie point is reached. The trans-
formation is reversible but sluggish. During cooling, taenite 10 mm
may survive 150–200 C below the a ! g transition tempera-
ture during heating (Figure 34).
When the Ni content passes 30%, kamacite is no longer
stable above room temperature. However, taenite becomes
ferromagnetic in this range. The iron in lunar rocks is seldom
Ni-rich unless it is of meteoritic origin. Magnetic properties of Amph.
Ni–Fe in different classes of meteorites and synthetic analogs
are reviewed by Nagata (1979) and Wasilewski (1981).
Néel et al. (1964) discovered that the random (or disor-
dered) arrangement of Fe and Ni atoms in taenite can become Plag.
ordered in the presence of a strong magnetic field along a h100i
axis for compositions of 50–55% Ni. Ordering occurs sponta-
neously in nature below 300 C over very long times, gener- Figure 35 Photomicrograph of two sets of crystallographically
ating a natural mineral called tetrataenite (Clarke and Scott, oriented magnetite inclusions in a clinopyroxene crystal in gabbro. The
1980; Nagata, 1983), which is common in chrondritic mete- magnetite laths do not continue into the amphibole rim nor are they
orites and is also known from lunar rocks. Ordered tetrataenite present in the adjacent plagioclase crystal. Modified from Renne
has rather different magnetic properties from disordered tae- PR, Scott GR, Glen JMG, and Feinberg JM (2002) Oriented inclusions
of magnetite in clinopyroxene: Source of stable remanent magnetization
nite of the same composition, including high coercivities
in gabbros of the Messum Complex, Namibia. Geochemistry,
(30–600 mT) and a ‘blocky’ Ms(T ) curve like that of hematite Geophysics, Geosystems 3(12): 1079. (http://dx.doi.org/10.1029/
(cf. Figure 20) with TC 550 C. 2002GC000319).
Because of the complications introduced by phase trans-
formations in Ni–Fe and uncertainty about the nature of NRM,
paleointensity work on meteorites is best confined to chon-
5.08.3.13 ‘Magnetic’ Silicates
drites and achondrites containing only kamacite or carbona-
ceous chondrites containing magnetite (Nagata, 1983). Silicate minerals have no inherent ferromagnetism, but they
Iron in lunar rocks has four distinct origins: primary igne- can precipitate ferrimagnetic particles during cooling from
ous Ni-poor iron, often in kamacite–taenite intergrowths; the melt (producing primary TRM) or as a result of later
coarse-grained Ni-rich meteoritic iron; secondary iron gener- oxidation (producing secondary CRM). The most common
ated by the impact melting of troilite (FeS); and nanometer- magnetic precipitate or inclusion in silicates is magnetite (see
size pure Fe spheres in impact-melted glasses. Of these forms of Section 5.08.3.2). The intergrowth texture between the precip-
iron, only the last contains any SD fraction. The other forms itated crystals and the host silicate(s) is the best guide to the
are paleomagnetically unappealing, having weak and unstable nature of the NRM, but this is not easily observed when the
NRMs. Viscous remagnetization is so pronounced in some of precipitates are submicroscopic.
the nearly SP fine grains that the advantages of an SD fraction Figure 35 illustrates crystallographic control of the orienta-
are largely lost. Lunar rocks as a whole are discouraging to work tions of needlelike magnetite laths, which lie along two direc-
with, and any primary paleomagnetic record is tenuous. tions in the host clinopyroxene crystal (Renne et al., 2002).
The magnetite laths do not continue into an adjacent amphi- inclusion is not pure magnetite but a high-temperature
bole grain. These magnetites must have precipitated at a high exsolution intergrowth of magnetite and ulvöspinel
temperature during crystallization, and cooling of the pyrox- (Section A.2.1). Magnetic domains within the magnetite are
ene and the NRM is almost certainly a TRM. The remanence, truncated by the crosscutting ulvöspinel bands (cf. textures in
confined to two crystallographic directions, is markedly aniso- Figure 29). It is interesting that despite this disruption, the
tropic. TRM directions can be in error by tens of degrees in domains prefer to align with the short dimension of individual
individual crystals, although such errors will average out on the magnetite bands, lining up across the gaps to form ‘superdo-
scale of paleomagnetic samples. mains’ continuing across adjacent grains parallel to the long
Figure 36 is an AFM/MFM (Section 5.08.2.5) image of a dimension of the entire inclusion. A similar example of super-
single inclusion (Feinberg et al., 2005). In this case, the domains crossing several grains appears in Figure 12.
The earliest work on crystallographically oriented inclu-
sions in pyroxene is that of Evans and Wayman (1970). They
used TEM of an etched surface to reveal the arrays of inclusions.
(a) These were assumed to be magnetite on the basis of their
coercivities, which ranged up to the limit expected for shape
anisotropy in magnetite (Evans and McElhinny, 1969). Murthy
et al. (1971) and Davis (1981) subsequently demonstrated
that arrays of crystallographically oriented magnetite needles
occur in plagioclase in anorthosites and oceanic gabbros
and inferred that they carry primary TRM that dominates
the NRM in the particular rocks they studied. Plagioclase crys-
tals extracted from volcanic and plutonic rocks are now
being used to refine paleointensity studies (Cottrell and
Tarduno, 2000). The magnetites they contain are equidimen-
sional, not needlelike, and record paleomagnetic directions
faithfully.
1 µm Biotite and amphiboles like hornblende and actinolite can
also be quite magnetic. In biotite, magnetite generally occurs in
cracks or between the mica sheets but is not so clearly cogenetic
with the silicate host. Relatively large magnetite crystals with
(b) MD properties tend to dominate the magnetic properties
although a population of finer magnetites seems to be present
as well, perhaps extending down to SD sizes (Dunlop et al.,
2006). Amphibole NRMs are generally MD-like, with rather
unstable NRM directions and nonlinear paleointensity behav-
ior (Dunlop et al., 2005; Wu et al., 1974).
5.08.3.14 Biogenic Magnetic Minerals

Biomagnetism is the science of magnetic fields and magnetic
materials in living organisms (Kirschvink et al., 1985). Its main
intersection with geophysics is the presence of fossil magneto-
tactic bacteria and their magnetosomes (chains, intact or dis-
1 µm rupted, of magnetic crystals) in sediments, where they carry an
NRM of distinctively high stability. The reason for this stability
is clear from Figure 37. Bacteria, as well as higher organisms,
Figure 36 Atomic force microscope (AFM, upper) and magnetic have evolved the ability to produce magnetite (less commonly
force microscope (MFM, lower) images of a titanomagnetite inclusion greigite) whose crystals are of strictly SD size. When crystals are
in clinopyroxene. The titanomagnetite is exsolved into a boxwork strung together in chains, the added shape anisotropy renders
structure (AFM image; see also Figures 12 and 29) of magnetite the magnetosome even harder magnetically.
prisms (high relief ) separated by ulvöspinel (two sets of crosscutting Intact chains seldom survive in sediments because cell walls
channels, white arrows). The individual magnetite prisms appear to deteriorate rapidly after death. However, Petersen et al. (1986)
be SD (MFM image), with Ms vectors parallel to the long axis of and Vali et al. (1987) had detected what appear to be fossil
each prism and of the inclusion but transverse to the short axis of
biomagnetic crystals and chains in deep-sea sediments as old
bands of magnetite prisms. Because of strong magnetostatic
as 150 Ma using SEM. Carbonate globules from 4-Ga-old
interactions, the Ms vectors line up across the gaps between bands
but alternate in polarity along individual bands, similar to the patterns Martian meteorite ALH84001 contain elongated prismatic
seen in Figure 12. Modified from Feinberg JM, Scott GR, Renne magnetite crystals whose size and morphology resemble
PR, and Wenk H-R (2005) Exsolved magnetite inclusions in those of modern bacterial magnetites (Thomas-Keprta et al.,
silicates: Features determining their remanence behavior. Geology 2000). Whether or not these are magnetofossils testifying to
33: 513–516. the presence of early life on Mars is an open question.
(b)
(a)
200 nm
(c) (d)
100 nm 100 nm
400 400
Fe O
200 200
100 nm
0 0
Figure 37 Transmission electron micrograph of part of a single cell of a magnetotactic bacterium, showing a chain of magnetosomes (b), which are
imaged by off-axis electron holography in (a). Elemental maps for Fe (c) and O (d) indicate that the magnetosomes are composed of magnetite.
Magnetic field lines link the individual magnetosomes along the chain (a), demonstrating strong magnetostatic interactions. As a result, the chain
behaves as an elongated SD particle. Modified from Dunin-Borkowski RE, McCartney MR, Frankel RB, Bazylinski DA, Pósfai M, and Buseck PR (1998)
Magnetic microstructure of magnetotactic bacteria by electron holography. Science 282: 1868–1870.
5.08.4 Induced and Remanent Magnetization wall between domains whose spins remain in easy directions.
As explained in Section 5.08.2.7, complete reversals of M
5.08.4.1 SD Hysteresis and Susceptibility sometimes occur incoherently: for example, ‘flower’ SD states
Self-demagnetizing fields and the magnetizing process for can reverse by curling, the transitory passing of a wall-like
SD and MD grains were discussed in a general way in vortex across the grain.
Sections 5.08.2.1 and 5.08.2.2. Figure 38 illustrates medium- The simplest model of SD magnetization is coherent rota-
field hysteresis loops synthesized by measuring the difference in tion against a uniaxial anisotropy (Néel, 1949; Stoner and
total area between positive (field-aligned) and negative domains Wohlfarth, 1948). The model is simplistic but surprisingly
observed under the microscope at different stages of field cycling powerful and is easily adapted to fit more realistic situations
(Soffel and Appel, 1982). A moderately large grain (30 mm) (Egli and Lowrie, 2002; Lanci and Kent, 2003). The model SD
exhibits distinctively MD hysteresis. The loop is narrow because grain (Figure 39) has its easy axis, y ¼ 0, at angle f to the
lattice defects impeding wall motion produce only slight irre- applied field Ho, which pulls Ms through an angle y away
versibility in wall positions in increasing and decreasing fields. from the easy axis. The energies involved are due to (shape)
M saturates when Ho reaches the maximum value of the internal anisotropy and the applied field (Figure 39): Ed ¼ 1/2 m0V D
demagnetizing field, (Hd)max ¼ NMs (eqn [6]). This state corre- ANM2s sin2y (eqn [7]) and EH ¼ m0VMs Ho ¼ m0VMsHo cos
sponds to zero net internal field, Hi ¼ 0. (f y). DN is the difference between the demagnetizing factors
A smaller grain (5 10 mm), near-magnetite in composi- for M along the hard and easy axes.
tion, has a broad SD-like loop, with considerable irreversibility When the sum E ¼ Ed þ EH is minimized by setting
between the magnetization processes in increasing and dE/dy ¼ 0, the elementary hysteresis loops, shown in Figure 40,
decreasing H0 the maximum available field m0Ho 25 mT. result. When Ho is applied along the easy axis (y ¼ 0), the loop
A broad minor (nonsaturating) loop in intermediate fields is is perfectly rectangular. There is no susceptibility because M
a sign of SD-like behavior, although this grain has a size d0 remains fixed at y ¼ 0 until Ho reaches a critical value HK, when
for magnetite (cf. Section 5.08.2.9 and Table 3). M abruptly rotates past the hard axis (y ¼ 90 ) and reverses to
An SD grain is magnetically hard (high Hc) because it the opposite easy axis (y ¼ 180 ). This elementary rectangular
contains no wall. The coupled spins of the entire grain must loop is sometimes called a hysteron. Its simplicity makes it
be bodily rotated away from an easy axis of anisotropy in ideal for further modeling, for example, in the presence of
response to Ho, and this takes more energy than moving a interaction fields (Section 5.08.4.4).
1.0
x = 0.515
M/Ms
0.5 x = 0.11
(30 mm × 30 mm) 0.5
−20 −10 (5 mm × 10 mm)

0 10 20
m0 Ho (mT)
−20 −10 0 10 20
−0.5
−0.5
−1.0
(a) (b)
Figure 38 Hysteresis loops synthesized by measuring the areas of ‘þ’ and ‘’ domains imaged by the Bitter colloid patterns at different field strengths:
(a) MD-type saturating loop; (b) SD-type minor loop. Modified from Soffel HC and Appel E (1982) Domain structure of small synthetic
titanomagnetite particles and experiments with IRM and TRM. Physics of the Earth and Planetary Interiors 30: 348–355.
Ed = 1 VN(q )Ms2
2
q
Energy
Ms f q
Ho 0 p/2 p
EH = –VMs Ho
•
(a) (b)
Figure 39 Coherent rotation in a spheroidal SD grain. (a) Orientation of Ms without and with a field Ho applied at angle f the easy axis y ¼ 0.
(b) Contributions of Ed and EH to the total energy as a function of y.
f=0 f= p f=p all f

2 4
M
Ms
1M
2 s
Ho
–HK HK HK HK = 1 HK(f = 0)
2
(a) (b) (c) (d)
Figure 40 Theoretical hysteresis loops for the SD grains of Figure 39. The rectangular loop ((a) f ¼ 0) and ramp ((b) f ¼ 90 ) are used as
hysterons in the Preisach–Néel diagram (Figure 43). The angle-averaged loop (d) resembles the loop for f ¼ 45 (c) and incorporates both reversible
and irreversible changes in magnetization.
When Ho is applied along the hard axis, there is in effect no linearly with Ho, giving a constant susceptibility kf (or just k) ¼
barrier to pass and so no hysteresis. M rotates reversibly at all Ms/HK.
Ho, reaching y ¼ 90 when Ho ¼ HK. At this point, M (measured When Ho is along an intermediate axis, for example, 45 ,
in the direction of Ho) ¼ Ms. This magnetization function is M at first rotates reversibly and then abruptly and irreversibly
also particularly simple and can be used as a hysteron. M varies switches at a critical field, which is always less than HK.
Averaging over an ensemble of grains with their easy axes those of SD grains of the same shape, but the difference
randomly oriented relative to Ho gives a loop that is reminis- decreases rapidly with increasing n. Large MD grains with
cent of the experimental SD-like loop in Figure 38. The coer- many domains ‘look like’ SD grains with M Ms.
cive force is 1/2HK ¼ 1/2DNMs and the SIRM is exactly 1/ Equation [15] is exact if there is no resistance to wall dis-
2Ms. placement, that is, if the intrinsic susceptibility of the material,
The saturation hysteresis loop for randomly oriented grains ki ¼ dM/dHi, is infinite. This is close to being the case in MD
in Figure 40 appears to have no demagnetized state. That is iron, but in magnetic oxides and sulfides, there is considerably
certainly true for individual SD grains in the ensemble, but the more resistance to wall motion. Then, solving the equation
ensemble itself possesses a demagnetized state (not accessible M ¼ ki(Ho – NM) for M,
by applying continuously increasing or decreasing Ho) in
ko ¼ M=Ho ¼ ki =ð1 þ Nki Þ [17]
which grain moments are randomly distributed in space
when Ho ¼ 0. Starting from this state, which can be achieved For a large equidimensional MD grain of magnetite, con-
in the laboratory by a sequence of positive and negative fields taining many domain walls, an estimate of ki is 12 (SI units),
(AF demagnetization, Section 5.08.5.1) or by thermal ran- giving ko 12/5 ¼ 2.4 compared with ko ¼ 3 from eqn [15].
domization (thermal demagnetization, Section 5.08.5.3), the Note that the so-called screening factor 1/(1 þ Nki) is indepen-
ensemble has a low-field or initial susceptibility ko ¼ (2/3) Ms/ dent of the system of units; for large MD grains of magnetite, it
HK and an SIRM Mrs ¼ 1/2Ms. is about 1/5.
5.08.4.2 MD Hysteresis and Susceptibility

5.08.4.3 Temperature Effects: Hopkinson Peak
MD grains respond to a small applied field Ho by domain wall and Koenigsberger Ratio Q
displacement. Only in large fields do the domains rotate as in
MD susceptibility ko increases with rising temperature T
SD grains. The amount of wall displacement is determined by
because wall pinning decreases, causing ki to increase. Walls
how strongly a particular wall is pinned at various locations
may be pinned magnetoelastically (by the stress fields of
along its path and by the orientation of Ho relative to the
dislocations) or may become anchored on vacancies or actual
domain Ms vectors. The overall magnetostatic energy due to
voids in the crystal, in which case the wall energy is lowered by
the external field is
a combination of magnetocrystalline and demagnetizing
EH ¼ m0 VM Ho ¼ m0 VMs ðx=dÞH0 cos c [14] effects. Since l, K, and Ms all decrease as T rises (Figure 3),
resistance to wall motion also weakens. For magnetite, the
where x is the average wall displacement in a grain size d, increase in ko is only 20%, from 2.4 SI at room temperature
M ¼ (x/d)Ms is the net magnetization of the grain, and c is T0 to 3 near the Curie point TC, because of the strong braking
the angle between Ms and Ho. Since net magnetization governs effect of Hd. In minerals like TM60 and pyrrhotite with stronger
EH, the walls can be thought of as interacting with one another: wall pinning and weaker Ms and Hd, ko(T ) increases more and
if one wall moves only a small distance, another wall will move reaches a peak (the Hopkinson peak) just below TC.
a larger distance to produce the desired M. The Koenigsberger ratio Q ¼ Mr/koHo between remanent
Ultimately, how far walls can be driven by Ho depends on magnetization Mr (normally TRM or NRM assumed to be of
the opposition of the internal self-demagnetizing field TRM origin) and induced magnetization Min ¼ koHo produced
Hd ¼ NM created by M. The total field inside the grain is the by the same (weak) field Ho is often used in magnetic explo-
sum of Ho and Hd. Wall motion will cease when Hd grows large ration. Although most magnetic anomaly interpretation is
enough to balance Ho: done using room-temperature values of Mr and ko, this is not
Hi ¼ Ho NM ! 0 or ko ¼ dM=dHo ¼ 1=N [15] appropriate for long-wavelength anomalies having their source
in the deep crust, where ambient temperatures are hundreds of
This expression, which ignores both resistance to wall Celsius above T0.
motion by lattice defects and orientation of Ho relative to Ms, A crude picture of TRM can be obtained by assuming that
works surprisingly well as a first-order approximation. TRM is due to walls that are strongly pinned below a blocking
Before deriving a more exact expression, we need to con- temperature TB (Stacey, 1958). The magnetization at TB is given
sider the meaning of N in an inhomogeneously magnetized by eqn [15]: M(TB) ¼ Ho/N. In cooling to room temperature,
grain with a set of domains whose sizes and orientations this is magnified due to the reversible increase in Ms: M(T0) ¼
change with changing Ho. Micromagnetic calculations demon- M(TB) (Ms0/MsB). However, when Ho is removed at T0, weakly
strate that to a very good approximation, the demagnetizing pinned walls adjust their positions under the influence of Hd to
energy is given by partially cancel or screen M. The final TRM is then M(T0)
multiplied by the screening factor 1/(1 þ Nki). For MD magne-
Ed ¼ Ed0 þ 1=2m0VNM2 [16]
tite, Ms0/MsB 3 and 1/(l þ Nki) 1/5, thus, 3/5 at T0.
with N NSD (1 1/n), where n is the number of domains and We can treat the case of deep crustal anomaly sources by
NSD is the single-domain demagnetizing factor (see Dunlop interrupting cooling at some T below TB but above T0 and setting
and Özdemir, 1997, Chapter 5.02). Thus, N remains nearly a Ho ! 0. The TRM acquired will be less than before by the factor
constant despite the existence of domain structure, so that ko in Ms(T )/Ms0, with a further reduction due to the decrease in the
eqn [15] is constant, independent of the field Ho and also screening factor with increasing ki. When induced magnetization
independent of temperature unless the domain structure alters. is produced by a later Earth’s field Ho, Q will therefore be <0.6
Two-domain grains have demagnetizing factors about one-half (Figure 41). Very deep-seated rocks may remain above their
2.0 82.0
Sample: 8c
1.8
1.6
Susceptibility (10 –3 SI)

1.4
Qt or ko (normalized)
ko
1.2
1.0
0.8
0.6
0.4 Qt
0
–200 –100 0 100 200 300 400 500 600 700
0.2 Curie Temperature, T (°C)
point
0 Figure 42 Susceptibility ko versus temperature for a monzonite
0 100 200 300 400 500 600 sample containing large grains of magnetite. There are a small
Hopkinson peak at 520–540 C and a more prominent peak near the
480 Verwey transition (153 C). Modified from Dunlop DJ, Schmidt PW,
Özdemir Ö, and Clark DA (1997) Paleomagnetism and
paleothermometry of the Sydney Basin. 1. Thermoviscous and
Slow change Fast
chemical overprinting of the Milton Monzonite. Journal of Geophysical
Ms (kA m–1)
in Ms
Research 102: 27271–27283.
TB
Range TC) moderate fields. This comes about because the coupled
spins in an SD grain are much less perturbed by thermal
Tcurie
agitation than are individual atomic moments. The resulting
SP susceptibility in weak fields
0 100 200 300 400 500 600
kSP ¼ m0 VM2s =3kT [19]
Temperature, T (C)
can be orders of magnitude larger than ordinary SD suscepti-
Figure 41 Sketches of the temperature dependences of initial
bility due to rotation of moments against an anisotropy barrier
susceptibility ko and Koenigsberger ratio Qt ¼ Mtr/koHo of SD or SD-like
grains of magnetite. When heated to the main blocking temperature (Section 5.08.4.1) or MD susceptibility. As a result, SD grains
range above 500 C, Ms and therefore Mtr and Qt decrease strongly, while have a pronounced Hopkinson peak in ko at any temperature
ko rises to a Hopkinson peak just below the Curie point. Multidomain where anisotropy drops to low values, not only near TC but also
grains do not display any substantial Hopkinson peak because ko is around the magnetite isotropic point Tk 130 K (Figure 42).
limited by self-demagnetization to a value
1/N. For an SD grain of volume V, the onset of SP occurs quite
suddenly in heating above the TRM blocking temperature TB.
At TB, M is described by an equation similar to [18]:
blocking temperatures and record no TRM from earlier magnetic
fields (Q ¼ 0). However, their induced magnetizations are still MðHo , T Þ ¼ Ms tan hðm0 VMs :H0 =kT Þ [20]
limited by self-demagnetization. Any Hopkinson peak is minor that describes thermal equilibrium between populations of
if MD magnetite is the mineral involved. grains aligned along two anisotropy-favored easy axes
The same is not true for SD grains. Each grain is magnetized (Section 5.08.4.1). This distribution appropriate to (Ho, TB)
to saturation, and the only limit on induced magnetization is remains in frozen equilibrium during cooling below TB and
the number of grains whose moments remain oriented oppo- when Ho is removed at T0. Thus, M in eqn [20] is the SD TRM.
site to Ho. At sufficiently high T, all barriers to rotation of Just as SP susceptibility kSP is much higher than anisotropy-
moments (shape, crystalline, or magnetoelastic anisotropy) limited susceptibility ko, SD TRM is SD-induced magnetiza-
disappear, and only thermal energy prevents perfect alignment tion koHo at ordinary temperatures. Thus, Q 1 for SD grains.
of VMs of a grain with Ho. This condition is called SP because However, as T increases, TRM decreases reversibly up to TB,
Min obeys the paramagnetic Langevin and Curie laws: while ko increases. Therefore, Q drops steadily with increasing
if MðHo , T Þ ¼ Ms Lðm0 VMs Ho =kT Þ T, particularly above TB (in reality, not a single temperature but
a range of TB’s for different values of V; Figure 41).
m0 VMs 2 H0 =3kT [18]
if m0 VMs Ms Ho =kT 1
5.08.4.4 Magnetostatic Interactions
L(a) being the Langevin function coth(a) 1/a.
In contrast to paramagnetism, SP magnetization Just as internal self-demagnetizing fields greatly affect the
approaches saturation for ordinary temperatures and (near response of MD grains to an external applied field Ho, the
M(Ho) response of an SD grain is modified by magnetostatic 5.08.5 NRM and Paleomagnetic Stability
interaction fields due to surrounding magnetized grains
(Figures 12, 36, and 37). Usually, neighboring grains are the 5.08.5.1 AF Demagnetization
most influential, but because dipole fields drop off as 1/r3 Magnetic anomaly interpretation usually focuses on induced
while volume increases as r3, the effect can be over a long magnetization in the present geomagnetic field. This field has a
range if the grains are fairly uniformly spaced. known direction and intensity at a specified location on Earth.
A commonly used approximation (Néel, 1954; Preisach, More challenging is determining the direction and, in favor-
1935) is the local-field model. The interaction field Hi acting able cases, the intensity of an ancient geomagnetic or planetary
on an SD grain is taken to be constant in direction and mag- magnetic field as recorded by the NRM of rocks. To be success-
nitude, independent of the magnetization state of the assem- ful, the paleomagnetic method must be able to separate vari-
blage of grains in the sample. This is not very realistic but works ous generations of NRM acquired by the same rock at different
reasonably well if interactions are due to a few nearby grains times in its history, including relatively modern ‘overprints’
that do not switch their moments in the field range of interest; that constitute noise. This separation of NRM components is
the rest of the grains then provide a background M, which can accomplished by one or a combination of ‘cleaning’ or ‘step-
be accounted for separately by a demagnetizing field. In the wise demagnetization’ techniques.
special case where both Hi and Ho are parallel to the easy axis ‘AF demagnetization,’ which subjects a sample to fields of
of a uniaxial SD grain, the rectangular loops of Section 5.08.4.1 alternating polarity whose amplitude gradually diminishes to
become displaced along the field axis, with modified switching zero, is universally used. Raising the initial AF amplitude in a
fields a ¼ Hc Hi and b ¼ Hc – Hi (Figure 43). These provide a series of steps has the effect of demagnetizing grains with
set of hysterons for Preisach or FORC (first-order reversal increasing microcoercivities Hc. Individual SD grains do not
curve) diagrams, which are widely used to characterize domain literally demagnetize because they have asymmetrical hystere-
state, coercivity distribution, and interactions of experimental sis loops (Figure 43) and will always be left in the same state
samples (Carvallo et al., 2005; Roberts et al., 2000). (either þ Ms or Ms, depending on Hi) after AF treatment.
Ms Ms
Hi a
b
–Hc Hc H H
–Ms –Ms
Hb
a Hu
b
a
b
b a
Ha
a
b
b a
b a
a =b
Preisach space
–Hb Hc
FORC space
Figure 43 Top: Elementary hysteresis loops of an SD grain with and without a constant interaction field Hi acting parallel to the applied field
Ho and the easy axis y ¼ 0 (cf. Figure 40). Bottom: Elementary reversible and irreversible magnetization cycles, based on the hysterons of Figure 40
shifted by Hi, for switching fields (a and b) in different sectors of the Preisach/FORC diagram. Modified from Carvallo C, Dunlop DJ, and Özdemir
Ö (2005) Experimental comparison of FORC and remanent Preisach diagrams. Geophysical Journal International 162: 747–754.
However, if the distribution of interaction fields g(Hi) is 0.014 SIRM

symmetrical with mean zero, equal numbers of grains will be 6 µm
magnetized in opposite directions, and the net moment will be annealed
zero for the population of grains with Hc – Hi less than the peak 0.012
AF applied in a particular step.
It is not self-evident that higher-microcoercivity grains nec-
Magnetization, M (emu)
essarily have more ancient NRMs. This depends on the mech- 0.01
anism of NRM. Their NRMs are, however, generally more
stable or resistant to resetting over long intervals of time (see
Section 5.08.6.2). 0.008
If a unidirectional bias field is added to the AF, each ensem-
ble of identical SD grains will acquire a net magnetization
rather than being (statistically) demagnetized. This process is 0.006
called ‘anhysteretic remanent magnetization (ARM)’ and is
sometimes useful as a laboratory analogue to TRM. On the
other hand, higher-order harmonics leading to asymmetrical 0.004
AFs can produce unwanted ARMs that contaminate rather than
clean the NRM. Some older AF demagnetizers produce serious
ARMs above 60 mT. 0.002
100 150 200 250 300
Temperature, T (K)
5.08.5.2 Low-Temperature Demagnetization Figure 44 Stepwise low-temperature demagnetization of saturation

isothermal remanent magnetization (SIRM) produced at 300 K in 6 mm
Cooling magnetite grains through the isotropic point magnetite grains. Cooling–warming cycles in zero field to 200, 150, 130,
(TK 130 K) has the effect of momentarily broadening domain 120, and 110 K produced similar decrements in remanence. Modified
walls, which are then less strongly pinned by defects. The effect on from Dunlop DJ (2003a) Stepwise and continuous low-temperature
SD grains is less important because shape anisotropy is governed demagnetization. Geophysical Research Letters 30(11): 1582. (http://dx.
by Ms, which is continuous at TK and almost continuous at the doi.org/10.1029/2003GL017268).
Verwey transition (TV ¼ 120 K) (Özdemir et al., 2002). As nor-
mally applied, LTD is a one-shot technique designed to reduce
MD remanence by cooling in zero field to below TV and back to Meq ðHo , T Þ m0 VMS0 2 ms 2 ðT ÞHo =kT [22]
T0. Stepwise AF or thermal cleaning generally follows LTD.
It is possible to carry out LTD in stepwise fashion because in which ms(T ) ¼ Ms(T )/Ms0. The relaxation time in seconds is
K1 of magnetite decreases continuously below T0, particularly given by
below 200 K, not discontinuously at Tk (Dunlop, 2003a).
t 1010 eDE=kT ¼ 1010 exp m0 VHc0 Ms0 m2s ðT Þ=2kT [23]
Figure 44 demonstrates that saturation IRM of 6 mm magnetite
grains is reduced to 25% of its initial value in five approxi- where DE is the activation energy or barrier to be crossed,
mately equal decrements using stepwise LTD. here by reversals of SD moments (Section 5.08.4.1), and
Hc0 ¼ Hc(T0) is the room-temperature microcoercivity for this
process. Shape anisotropy is assumed, in which case Hc(T )/
5.08.5.3 Thermal Demagnetization of SD Grains Hc0 ¼ ms(T ). For magnetocrystalline anisotropy, for example,
in hematite, Hc(T ) varies as a higher power of ms(T ).
The simplest NRM mechanism in igneous and metamorphic The equilibrium magnetization Meq is proportional to
rocks is a TRM. Even when factors like time and chemical m2s (T ), and Ms(T ) decreases strongly between 500 C and
change enter, cooling through a blocking temperature TB is TC ¼ 580 C in magnetite. However, the relaxation time t also
often a useful first-order model of the NRM process. Reciprocal decreases with increasing t. These two effects oppose each
to the TRM blocking process is ‘thermal demagnetization,’ a other, the second promoting a closer approach of induced
zero-field heating–cooling cycle through TB. Since there is a magnetization M to Meq, while the first causing Meq to decrease
range of TB’s in any sample, in practice, thermal demagnetiza- with heating. The exponential dependence in eqn [23] makes
tion occurs continuously over some range of heating the second effect dominant. M is thus enhanced over the
temperatures. blocking temperature range and only drops off within
For SD grains, Néel (1949) produced a satisfying theory 5–10 C of Tc. This results in a strong SD Hopkinson peak in
that links the roles of time t and temperature T in magnetiza- ko (Figure 41).
tion changes. Any out-of-equilibrium magnetization DM ¼ Blocking temperature implies a sudden rather than a grad-
M(t) Meq decays exponentially with time: ual change in M. This is indeed the case for individual grains.
Because of the exponential dependence of t on T, if T < TB, t t
DMðt Þ ¼ DMð0Þet=t [21]
where t is the time measured on a laboratory or a geologic
with Meq being thermal equilibrium magnetization and t relax- scale, depending on the situation. Then, according to [21], M
ation time. In the case of SD grains, following eqn [20] and remains ‘blocked’ at its initial value M(0). This is the origin of
assuming m0VMs/kT 1, TRM. Grains acquire a remanence Mtr when they are cooled
through TB in a field Ho because M no longer responds to a curve are blocked and unresponsive to Ho at T. Grains inter-
changes in Ho on the timescale of observation. secting the curve are at their blocking temperature TB.
On the other hand, when grains are above their blocking The connection between AF and thermal demagnetization
temperatures, t t and M ! Meq during time t. This is the is provided by eqn [24]. For grains of a particular volume V,
‘unblocked’ or SP state, ‘super’ alluding to the giant suscepti- coercivity Hc0 increases in proportion to TB/m2s (TB). All other
bility that results from equilibrium magnetization Meq being things being equal, SD grains with high blocking temperatures
achieved in a weak field Ho (cf. eqn [19]). SP is responsible for will also have high AF coercivities. However, if there are a wide
the Hopkinson peak in ko ¼ Min/Ho sketched in Figure 41, as spectrum of grain sizes or a mixture of remanence-carrying
explained in Section 5.08.4.3. minerals, this correspondence is not clear-cut. Note also that
When grains are heated through TB in zero field, they pass for grains of fixed size and shape (i.e., VHc0 ¼ const.), ln t plays
from the blocked to the unblocked state and lose all their a role similar to T in magnetization and demagnetization. This
remanent magnetization. Because no magnetization is induced equivalence is sometimes exploited in zero-field ‘storage tests,’
above TB in zero field, they do not acquire any new remanence but only very low-TB remanence can be demagnetized in prac-
when cooled back through TB. This process of thermal demag- tical storage times of days to weeks. The ln t – T equivalence is
netization occurs only in the laboratory; zero-field conditions also at the root of thermoviscous magnetization
are never encountered in nature, at least on Earth. (Section 5.08.6.2).
For individual SD grains or ensembles of identical grains,
the demagnetization is abrupt and occurs over an interval of a
few degrees Celsius around TB. From eqn [23], setting 5.08.5.4 Thermal Demagnetization of MD Grains
t ¼ 1.78 T at blocking (the factor 1.78 ensures agreement with
Although one often speaks of TB of an MD grain (e.g., in
more exact theories), we have
Section 5.08.4.3), the reality is that individual walls have
different blocking temperatures and the set of TB’s changes
VHc0 ¼ ½2kð23 þ ln 1:78tÞ=Ms0 T B =ms 2 ðT B Þ [24]
each time one of the walls moves because of the change in
Hd. This distribution of TB values is particularly marked during
an implicit equation for TB.
thermal demagnetization. Figure 46 illustrates how the spec-
The magnetic grains in any rock have a spectrum of volumes
trum of thermal demagnetization temperatures broadens with
V and critical fields Hc0. Grains with TB > T will retain their
increasing grain size. For the larger sizes, demagnetization
remanence, while those with TB < T will be demagnetized. For
begins much below the TRM blocking temperature, even in
increasing values of T, the rectangular hyperbola described by
the first heating step above T0, and continues essentially up to
eqn [24] sweeps through a set of fixed points (V, Hc0), each
Tc. Thermal demagnetization of MD remanence is not the clean
representing a grain or population of similar grains (Figure 45).
efficient procedure that it is for SD grains.
Because of the nonlinear T/m2s (T ) dependence, the blocking
curve sweeps out small areas up to 500 C (for magnetite)
but increasingly large areas above 500 C. Grains below and to
5.08.5.5 Resolving Multivectorial NRM
the left of a particular curve are SP and are in thermal equilib-
rium with Ho at the T indicated. Grains above and to the right of A frequent problem in paleomagnetism is to separate NRMs
acquired over different T intervals at different times in the
rock’s history. In the simplest case, these are ‘partial TRMs’
Grain representative
points acquired over nonoverlapping ranges of TB. If two such com-
ponents of NRM have sufficiently different ages, they will also
have different directions because of plate motion and polarity
reversals. The composite NRM is said to be ‘multivectorial,’ and
how efficiently the components are resolved can be judged by
Grain volume, V
measuring the resultant NRM vector direction in the course of

T
550 °C stepwise demagnetization.
Increasing
SD grains have ‘sharp’ TRM blocking and thermal demag-
netization temperatures. Their partial TRMs obey the Thellier
500 °C (1938) laws:
400 °C 1. Partial TRMs produced in nonoverlapping T intervals cov-

ering (T0 and Tc) sum to give the total TRM (Tc and T0)
t (T ) = t (additivity law).
20 °C 2. Partial TRMs produced in nonoverlapping T intervals are
independent of one another in direction and magnitude
Microscopic coercivity, Hc (independence law).
3. A partial TRM produced by cooling in a field from T1 to T2 is
Figure 45 The Néel diagram showing representative points (V and Hc0) unaffected by reheating in zero field to T2 but is completely
of SD grains in a hypothetical sample. With increasing T, the blocking destroyed by reheating to T1 (reciprocity law).
curve (hyperbolas, eqn [24]) sweeps ever more rapidly through the
points, producing the strongly nonlinear variations of ko and Qt portrayed The clean thermal demagnetization of small grains evident
in Figure 41. in Figure 46 (1 mm grains) is evidence of reciprocity:
1.0
+ +
× + 350 370 °C
1 mm Partial TRMS
×
6 mm+
0.8 110 20 mm +
× µm
135 mm × +
0.6 ×
M/Mo
50% +
×
0.4 50%
+ ×
0.2
+
×
+
0 +×
0 100 200 300 400 500
Temperature, T(°C)
Figure 46 Stepwise thermal demagnetization of partial TRM produced over a narrow TB range (370–350 C) in magnetites of different grain sizes.
In the finest grains, the partial TRM demagnetizes mainly over the original blocking range, but 135 mm MD grains demagnetize over a broad
range from room temperature to the Curie point. Open and filled dots, squares, and crosses are arbitrary and used to mark the data for the 1 mm
(solid dots), 6 mm (upright crosses), 20 mm (open squares), 110 mm (x’s), and 135 mm (open dots) samples. Modified from Dunlop DJ
and Özdemir Ö (2001) Beyond Néel’s theories: Thermal demagnetization of narrow-band partial thermoremanent magnetizations. Physics of the
Earth and Planetary Interiors 126: 43–57.
unblocking is exactly reciprocal to blocking in SD grains. In To make this experiment the basis of a practical technique
thermal cleaning of multivectorial NRM, there is a sharp junc- for determining a priori unknown field intensities, Ho requires
tion between the demagnetization trajectories of two adjacent two additional properties of SD TRM. First is linearity: TRM
partial TRMs (Figure 47). intensity must be proportional to Ho for weak fields. Second is
MD grains have smeared or distributed unblocking temper- constancy of blocking/unblocking temperatures: TB must be
atures and do not give sharp junctions in demagnetization independent of Ho, again for weak fields. In practice, these
trajectories. In extreme cases, the result is that neither NRM conditions are met if Ho
1 Oe or 80 A m1 approximately.
component can be correctly resolved. In Figure 48, the esti- Paleointensities are much more difficult to determine if MD
mated direction of the lower-T partial TRM is within 10 of grains are the TRM carriers. Thermal demagnetization begins
the correct direction, but the direction of the higher-T partial immediately above T0, as explained in the last section, but
TRM is 30 in error. Errors of this magnitude are unaccept- partial TRM tends to be acquired in higher T ranges (Xu and
able and render MD grains generally unusable as paleomag- Dunlop, 2004). The result is a ‘sagging’ Arai plot in which more
netic directional recorders. TRM is lost than regained in the lower T steps, but acquisition
outstrips loss of TRM in the higher T steps (Figure 49).
Although such a plot is perfectly reproducible if caused solely
by MD material, it generally leads to immediate rejection on
5.08.5.6 Thellier Paleointensity Determination
grounds of nonlinearity because most Arai plots of this form
The Thellier laws are the basis of the standard method for result from chemical alteration of the sample during the Thel-
determining paleofield intensity. The reciprocity law is again lier experiment. For practical details, see Chapter 5.13.
the key. If a TRM carried by SD grains is heated twice to the
same temperature and cooled to room temperature, the first
time in zero field and the second time in a field equal to Ho
5.08.5.7 Stability and Domain-State Tests
used to impart the TRM, the remanence lost in the first
heating–cooling will be exactly restored in the second. The There are many tests that use magnetic parameters to classify
usual way of presenting such double-heating data is in an samples according to their domain state and implied paleo-
Arai plot (Nagata et al., 1963) of NRM remaining after the magnetic stability. Among these are the stepwise demagnetiza-
first heating versus partial TRM gained in the second. For the tion techniques described in Sections 5.08.5.1–5.08.5.4,
experiment just described, the Arai plot for ascending T steps is which deal with NRM directly. The two tests described in this
a descending straight line of slope 1. section use instead hysteresis data, which may or may not give
323 N,Up N
600 Sample: 7d2
300 (mA m-1)
334 NRM
408 290 600

NRM 242
193
265 222
440
LT
HT 620
168
505 290 635
CRM1 265 CRM2 660
242 NRM 575
323
265
580 CRM1
290
300 580
E LT 20–222 °C
323 620 HT 222–323 °C
CRM2
440
LT
CRM1 323–620 °C
242 CRM2 >620 °C
HT
168
193
222
(a) (b)
Figure 47 Multivectorial NRM in a monzonite sample, consisting of two partial TRMs (LT and HT) in adjacent TB ranges (20–222 C and 222–323 C:
pyrrhotite) and two CRMs (magnetite and hematite). Stepwise thermal demagnetization results are displayed as orthogonal projections of the full
magnetization vector (left) and as unit vectors on a stereographic projection (right). Modified from Dunlop DJ, Schmidt PW, Özdemir Ö, and
Clark DA (1997) Paleomagnetism and paleothermometry of the Sydney Basin. 1. Thermoviscous and chemical overprinting of the Milton Monzonite.
a fair representation of the mixture of minerals and grain sizes confined to the Hc axis. Increasing interactions cause spreading
carrying NRM. parallel to the Hu axis. In PSD grains, the distribution peak
Very popular is the Day diagram (Day et al., 1977), a plot of moves toward the origin and ultimately merges with the Hu
the saturation hysteresis parameters Mrs/Ms versus Hcr/Hc (see axis. MD grains have distribution contours almost paralleling
Section 5.08.1.2 for definitions of these quantities). The Day the Hu axis at low Hc.
diagram is well documented and easily interpretable only for It has recently been verified that the Preisach–Néel inter-
magnetite and TM60 (Dunlop, 2002a,b). For magnetite, it pretation does make sense for SD FORC distributions
gives a fairly decisive separation into SD þ SP, PSD or (Carvallo et al., 2004; Muxworthy and Dunlop, 2002), by
SD þ MD, and MD regions (Figure 50). The underlying theory showing that the distribution parallel to the Hu axis contracts
is too complicated to review here, but it is based on results like with increasing T as Ms(T ), as expected for magnetostatic inter-
Mrs/Ms ¼ 0.5 for uniaxial SD grains (Section 5.08.4.1) and self- actions, and the distribution parallel to the Hc axis contracts in
demagnetization of MD grains (Section 5.08.4.3). the same way as the bulk coercive force, Hc(T ). These results
The venerable Preisach diagram, which has been used in place the FORC method on a firmer physical foundation. Hu in
many forms by the magnetic recording community over the the MD case presumably reflects the effect of self-
past four decades, has in recent years been popularized in demagnetizing fields (‘domain interactions’) rather than inter-
geophysics as the FORC diagram (Roberts et al., 2000). FORC particle interaction.
stands for first-order reversal curve and describes the method The decomposition of magnetic mixtures into their constit-
used to derive the distribution plotted in the diagram. Starting uent mineral and domain-state fractions is a developing field
from saturation, a series of descending partial hysteresis loops with much promise. The Day diagrams are ambiguous in some
are interrupted at different fields, and minor loops (the regions, magnetite þTM60 mixtures and mixtures of SP, SD,
FORCs) are traced in increasing fields. Rather than simply and PSD magnetites being indistinguishable (Dunlop, 2002b).
differencing results at various fields, a quadratic function is For simple mixtures of magnetites with two different domain
fitted to the data in the immediate surroundings of each states, Carter-Stiglitz et al. (2001) have devised a decomposi-
point in order to reduce noise. tion algorithm for hysteresis data that uniquely determines the
The FORC diagram plots Hu (equivalent to Hi) versus Hc, proportions of the phases. FORC diagrams have the potential
thus using the Preisach–Néel hysterons of Figure 43 to inter- to unravel bimodal mixtures of two soft phases or hard and
pret the data. Pragmatically, FORC diagrams for different soft phases (Muxworthy et al., 2005). The Preisach models
domain and interaction states are distinctive (Figure 51). SD without the FORC formalism have also proved useful in ana-
grains have a single peak in their distribution, well separated lyzing IRM acquisition curves, themselves widely used to infer
from the origin. If interactions are minimal, the distribution is components of mixtures (Egli, 2003; Fabian and von
(to 2.84, 7.33 at 20 °C) sufficient accuracy before doing the experiment, and the
Up
slope NRM/TRM must be assumed to be HA/HL. Fortunately,
linearity seems to be a good assumption for almost all Earth
151 materials for fields
100 mT.
Observed
4 vector plot, One would anticipate such linearity on the basis of the SD
135 mm equation for TRM:
magnetite
Mtr ðHo , T o Þ ¼ Ms0 tanhðm0 VMs ðT B ÞH0 =kT B Þ
201
m0 VMs0 2 ms ðT B ÞH0 =kT B
Magnetic moment (10-4 A m2)
Vector pTRM
3 sum (400,20) for m0 VMsB H0 =kT B 1 [25]
250
which is an extension of eqns [20] and [22]. For purposes of
testing, since V can be difficult to estimate, it is useful to recast
pTRM 300 [25] in terms of Hc0 and TB by substituting from eqn [24] and
2
(Tc, 400) 330 setting 23 þ ln 1.78t 25, giving (Dunlop and Kletetschka, 2000)
362 Mtr ðHo , T 0 Þ ¼ Ms0 tan h½50Ho =Hc0 ms ðT B Þ [26]

401
The TRM intensity predicted by [25] or [26] is often much
451 higher than experimental TRMs of SD magnetites, a discrep-
1 ancy explained by Dunlop and West (1969) in terms of particle
500
530 Expected interactions. Magnetostatic interaction is much less important
550 single-domain for hematite because of its low Ms, and SD hematites do indeed
560 vector plot
575 570 have TRMs in close agreement with theory (Dunlop and
400 North Kletetschka, 2000; Özdemir and Dunlop, 2002).
It is usually assumed that any discrepancy between pre-
0 1 2 3
dicted and measured TRMs in the weak-field region is irrele-
Magnetic moment (10-4 A m2)
vant to paleointensity determination because NRM and TRM
Figure 48 Projection of the full NRM vector in the course of are affected identically. This is certainly true if the comparison
stepwise thermal demagnetization for a sum of orthogonal partial TRMs is between total NRM and total TRM: HA ¼ (NRM/TRM)HL.
in 135 mm magnetite. Instead of reproducing the orthogonal pattern of The theoretical field dependence is not used in determining
the actual partial TRM sum, the thermal data approximate to a sum HA. All that is required is proportionality between Mtr and HA.
of linear demagnetization trajectories with angular errors of about 10o However, matters are not so clear for a Thellier-type compari-
and 30o, respectively. Modified from Dunlop DJ (2003b) Partial
son between sets of partial TRMs and their corresponding NRM
thermoremanent magnetization: Louis Néel’s legacy in rock
fractions because the theoretical proportionality factor,
magnetism. Journal of Applied Physics 93: 8236–8240.
m0VM2s0ms(TB)/kTB, is a strong function of V and TB and will
be different for each partial TRM. If the actual proportionality
Dobeneck, 1997; Heslop et al., 2002, 2004). Finally, Fabian
factor depends on additional parameters (e.g., Hi), different
(2003) had suggested several useful parameters that can be
Thellier steps could give different results. The main effect is
obtained with little additional effort from hysteresis and back-
likely to be a stretching of portions of the Arai graph relative to
field measurements, and Leonhardt (2006) had devised a soft-
others, rather than nonlinearity.
ware for analysis of parameters from a broad variety of rock
The fundamental basis of TRM linearity in weak fields is less
magnetic measurements.
obvious for MD grains. The Stacey (1958) theory
(Section 5.08.4.3) predicts a TRM
5.08.6 Remanent Magnetization Processes in Nature Mtr ðHo , T 0 Þ ¼ ½ms ðT B Þð1 þ Nki Þ1 ðHo =N Þ [27]
5.08.6.1 Thermoremanent Magnetization The proportionality between Mtr and Ho in [27] is an illu-
sion, however, because TB is itself a function of Ho. The Néel
We have touched on many of the properties of SD and MD
(1955) theory, on the other hand, gives an explicit field depen-
TRM in Sections 5.08.4.3 and 5.08.5.3–5.08.5.6. An exhaus-
dence. TB(Ho) is accounted for by introducing the temperature
tive review of TRM has recently been given by Özdemir (2007).
dependence of microcoercivity Hc, which through wall pinning
Only a limited number of aspects are dealt with here, mostly
or nucleation controls TB: Hc(T )/Hc0 ¼ hc(T ) ¼ msn(T ). For
those of practical paleomagnetic importance.
magnetite, n ¼ 1–2, while for hematite, n ¼ 3 is usually
First is the question of how TRM intensity depends on the
assumed. The ultimate result is
strength Ho of the magnetic field applied during cooling from h ih i
TC. Only if TRM is proportional to Ho will paleointensity Mtr ðHo , T 0 Þ ¼ n=ðn 1Þ11=n Hc0 1=n =ð1 þ Nki Þ
determination (Section 5.08.5.6) be feasible. If it was possible

to choose the laboratory field HL to exactly match the paleo- Ho 11=n =N [28]
field HA, the Arai plot would have a slope of 1 and the TRM
field dependence would be irrelevant. In practice, it is difficult which is more intricate than Stacey’s result and highly non-
or impossible to guess HA for a particular sample with linear for small values of n. Equation [28] is quite successful in
1.0
0.8
Single-domain
(thermal activation)
NRM remaining 0.6 theory
1 mm
0.6 mm
0.4 6 mm
20 mm
135 mm
Multidomain
0.2 (field blocking)
theory
0
0 0.2 0.4 0.6 0.8 1.0
pTRM gained
Figure 49 Simulated Thellier paleointensity experiments carried out on magnetite grains of various sizes, compared to the predictions of SD and MD
theories. With increasing grain size, the data ‘sag’ increasingly below the ideal SD line of slope 1. The symbols are arbitrary and a different set is used
to mark the data for 0.6, 1, 6, 20, and 135 mm samples, respectively. Modified from Xu S and Dunlop DJ (2004) Thellier paleointensity theory and
experiments for multidomain grains. Journal of Geophysical Research 109: B07103 (http://dx.doi.org/10.1029/2004JB003024).
predicting TRM intensities for MD hematite (Dunlop and Fabian’s model to devise an improved experimental protocol
Kletetschka, 2000; Kletetschka et al., 2000a,b; Özdemir and that substantially reduces the curvature of Arai plots.
Dunlop, 2005; Figure 52), which are nonlinear below Finally, Kletetschka et al. (2004) had reported a linear
100 mT, and TRM intensities of PSD magnetites (Dunlop and relationship between log Mtr/Mrs and log Ho, which is
Argyle, 1997), which are quasilinear in the same field region. ‘universal,’ that is, has the same slope for all minerals and
Experimental TRMs for MD magnetite and TM60 are linear domain states. In view of the complexities of TRM and the
or close to it for fields below about 1 mT and vary approxi- vastly different blocking mechanisms of SD and MD grains, it
o (appropriate for n ¼ 2) at higher fields (Dunlop
mately as H1/2 is unlikely that their parallel data trends reveal an aspect of
and Waddington, 1975; Tucker and O’Reilly, 1980). The gen- some universal truth, delightful though that would be. The
erally observed weak-field linearity of TRM inspired Néel log–log scale expands and highlights the weak-field region,
(1955) to modify his field-blocking theory (eqn [28]) by pos- the crucial field range for paleointensities. The observed paral-
tulating a different blocking condition when Ho < Hf(TB), Hf lelism implies that for the materials tested, TRM was approxi-
representing the unpinning effect of thermal fluctuations on mately linear with field in this range.
domain walls. Then, if n ¼ 2, Offsets between Kletetschka et al.’s lines are in some cases
substantial. The offsets reflect the very different fields at which
h i TRM rises toward saturation (Mtr ! Mrs) for different minerals
Mtr ðHo , T 0 Þ ¼ Hc0 1=2 Hf ðT B Þ1=2 =ð1 þ Nki , Þ ðHo =N Þ [29]
and domain states. Figure 52 shows that for hematite, MD
TRM is almost saturated in a weak field (100 mT), while SD
Although the physical reasoning is sound, Néel’s result is TRM is far from saturation even at 1000 mT. Just the reverse is
unsatisfying because Hf(TB) involves TB and furthermore is true for magnetite, SD TRM rising rapidly but MD TRM theo-
experimentally inaccessible, so that [29] is untestable. retically reaching saturation only in fields 200 times larger than
Many other attempts have been made to improve our the TRM saturation field for MD hematite. These startling
understanding of MD TRM. Dunlop and Xu (1994) and Xu differences are due entirely to the 200-fold contrast between
and Dunlop (1994, 2004) extended the Néel field-blocking the Ms values of the two minerals. The self-demagnetizing
theory to partial TRMs and were able to predict quite closely fields that strongly oppose TRM in MD magnetite are negligible
the curvature of MD Arai plots. Fabian (2000, 2001) proposed in MD hematite. The contrast between the SD TRMs for the two
a phenomenological model closely paralleling the Preisach– minerals (Figure 52) is also due to their contrast in Ms, the
Néel model (Figure 43), which has since been extended by ultimate Mrs toward which TRM is rising being 200 times larger
Leonhardt et al. (2004). Biggin and Böhnel (2003) had used in magnetite.
Theoretical day plot curves for magnetite

1 2 3 4 5 6 8 10 20 30 40 50 70 100
1
SD
0.5 0 10%
20%
0.4 20% 0 SP saturation envelope
30%
20% 40% 30%
0.3 40%
40% 30 nm
50%
40% Langevin 50%
0.2 60% calculation 60% 25 nm
60%
SP + SD 60%
70% 70%
20 nm
80% PSD
0.1 80% 80%
80% 10 nm 15 nm
0.08
Mrs
Ms
90%
0.06
90%
0.05 SD + MD
0.04 95%
95%
0.03 98%
0.02
100%
100%
0.01 MD
0.008
0.006
0.005
1 2 3 4 5 6 8 10 20 30 40 50 70 100
Hcr /Hc
Figure 50 Theoretical Day plot curves for magnetites in various domain states (SD, PSD, and MD) and mixtures of different states (SP þ SD and
SD þ MD). Modified from Dunlop DJ (2002a) Theory and application of the Day plot (Mrs/Ms versus Hcr/Hc), 1. Theoretical curves and tests
using titanomagnetite data. Journal of Geophysical Research 107(B3): 2056 (http://dx.doi.org/10.1029/2001JB000486).
W (0.3 mm) W (1.7 mm) W (7 mm) W (11 mm)
60
40
20
Hu (mT)
–20
–40
–60
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

(a) Hc (mT) (b) Hc (mT) (c) Hc (mT) (d) Hc (mT)
Figure 51 Experimental first-order reversal curve (FORC) distributions for magnetites of various grain sizes. As size increases, the central
distribution peak moves to smaller coercivities Hc and ultimately merges with the axis. At the same time, the contours increasingly spread parallel to the
Hu axis, a result of self-demagnetizing fields. Modified from Muxworthy AR and Dunlop DJ (2002) First-order reversal curve (FORC) diagrams for
pseudo-single-domain magnetites at high temperature. Earth and Planetary Science Letters 203: 369–382.
0.01 0.1 1 10 100

2.0 2.0
Mrs
SD
magnetite
1.5 Data: 1.5

, Dunlop (1968,1971)
, Kletetschka et al. (2000b)
TRM, Mtr (A m kg )
-1
2
1.0 1.0
MD
magnetite
Earth’s
magnetic
0.5 field 0.5
MD hematite
Mrs
SD
hematite
0 0
0.01 0.1 1 10 100
Magnetic field, Ho (mT)
Figure 52 Experimental and theoretical (dashed) dependences of TRM intensity on applied magnetic field Ho for SD and MD magnetite and hematite.
Self-demagnetization is the reason why TRM saturates for MD magnetite in fields three orders of magnitude larger than for MD hematite.
Modified from Dunlop DJ and Kletetschka G (2000) Multidomain hematite: A source of planetary magnetic anomalies? Geophysical Research Letters
28: 3345–3348.
These diametrically opposite grain-size dependences of 5.08.6.2 Viscous and Thermoviscous Magnetization
TRM in magnetite and hematite are illustrated further in
The equivalence between the effect of T in thermal demagneti-
Figure 53. The TRM of magnetite drops more than 2 orders
zation and TRM acquisition and the effect of an equivalent
of magnitude between the SD threshold around 0.1 mm and
change in ln t, which produces the so-called viscous
true MD sizes of 50–100 mm. Furthermore, it decreases contin-
magnetization, was pointed out in Section 5.08.5.3. If T and
uously, not abruptly. This is one of the defining features of PSD
ln t both change by comparable amounts, for example, during
behavior. The TRM of hematite increases over the same size
very slow cooling in nature, ‘thermoviscous magnetization’
range, reaching a constant and unexpectedly high (for such a
results. Viscous magnetization and thermoviscous magnetiza-
weakly magnetic mineral) level above its SD threshold around
tion have recently been reviewed by Dunlop (2007b).
15 mm.
Equation [24] implies that a change in T by a fixed amount
The grain-size dependence of TRM in titanomagnetites with
DT has a disproportionate effect on the motion of (V, Hc0)
various Ti contents and in pyrrhotite has been reported by Day
contours on the Néel diagram (Figure 45) at high T where
(1977), O’Donovan et al. (1986), and Menyeh and O’Reilly
ms(T ) is changing rapidly (>500 C in the case of magnetite).
(1998). Over the 1–30 mm range, there is a general decrease in
Hence, blocking temperatures tend to be concentrated close to
TRM intensity, but not as marked as in magnetite. There are
TC. The same is true of a fixed change D(ln t) ¼ Dt/t. Viscous
also measurements of TRM in SD TM60, which link with those
magnetization is usually minor near T0 but large at high T,
for the larger sizes (Özdemir and O’Reilly, 1982b).
0.1 1 10 102 3
10 Granulite sample 103M1
0.5
22 °C
10
4
10
4 50 °C
do (aFe2O3) 200 °C
0.4 400 °C
TRM in 0.1 mT (A m )
do
–1
–1
(Fe2O3) d
3 3 0.3
10 10
0.2
Viscous magnetization, Mv (A m-1)

d
2 2
10 10
0.1
10 0
10 10 100 1000 10 000
2 3
0.1 1 10 10 10
Grain diameter, d (mm) Granulite sample 246B3b
1.2
Figure 53 The contrasting grain-size dependences of TRM intensity in
magnetite and hematite. TRM of magnetite decreases steadily with 22 °C inital
1.0 22 °C final
increasing grain size above the critical SD size d0, initially as d1. TRM of 200 °C
hematite increases with grain size up to a constant plateau (saturation 400 °C
0.8
remanence) above the SD–MD transition size d0 (aFe2O3). Modified
from Dunlop DJ and Arkani-Hamed J (2005) Magnetic minerals in the
Martian crust. Journal of Geophysical Research 110: E12S04. (http://dx. 0.6
doi.org/10.1029/2005JE002404).
0.4
especially near TC (Figure 54). This effect amounts to a ther- 0.2

moviscous Hopkinson peak, which augments the peak due to
SP magnetization. In Figure 55, for temperatures above 0
100 C, the extrapolated viscous magnetization Mv due to 10 100 1000 10 000
exposure to the Earth’s field for 700 ka (the length of the Time, t (min)
Brunhes polarity chron) equals Min due to short-term exposure Figure 54 Viscous magnetization as a function of time measured at
to the same field, doubling the overall induced magnetization. temperatures up to 400 C for two continental mafic granulites. Samples
In the Hopkinson peak region above 500 C, the increase is by are much more viscous at 400 C than at room temperature. Modified
more than a factor 2.5. from Kelso PR and Banerjee SK (1994) Elevated temperature viscous
The rocks illustrated in Figures 54 and 55 are not excep- remanent magnetization of natural and synthetic multidomain magnetite.
tional. Almost any paleomagnetic sample has a measurable Earth and Planetary Science Letters 122: 43–56.
viscous remanent magnetization (VRM) produced by exposure
to the geomagnetic field during the Brunhes at ambient tem-
Normalized viscous magnetization, Mv/Min
peratures. The question that then arises is the most efficient 1.6 400 s 3h 1 year
means of erasing this contaminant NRM component. In the 100 ka 700 ka
1.4
case of hematite especially, AF demagnetization is not the
answer (see Dunlop and Özdemir, 1997, Chapter 10.3). Ther- 1.2
mal cleaning, on the other hand, is very efficient because of the
1.0
reciprocal roles of time and temperature in SD magnetization
relaxation. VRM carried by MD grains is less amenable to 0.8
thermal cleaning because of the spreading of the blocking
0.6
temperature spectrum by self-demagnetizing fields, but
fortunately, MD grains tend to be less viscous and the VRMs 0.4
to be cleaned are usually minor.
0.2
A useful graphic depiction of SD thermoviscous effects is
the Pullaiah diagram (Pullaiah et al., 1975). Equation [24] can 0.0
be used to find when blocking occurs for the same SD ensem- 0 100 200 300 400 500 600
ble, specified by VHc0, under different combinations (TB1 and Temperature, T (°C)
t1) and (TB2 and t2) of temperature and time. The result is
Figure 55 Viscous magnetization measured after 400 s and 3 h and
extrapolated assuming a ln t dependence to 1 year, 100 ka, and 700 ka for
ð23:6 þ ln t 1 Þ T B1 =ms 2 ðT B1 Þ
an oceanic serpentinized peridotite. Mv is much larger at 120 C and
¼ ð23:6 þ ln t 2 Þ T B2 =ms 2 ðT B2 Þ [30] above than at room temperature and increases to a Hopkinson peak in the
TB range around 500 C. Modified from Pozzi J-P and Dubuisson
for magnetite. For hematite, the appropriate factor is m4s (TB). G (1992) High temperature viscous magnetization of oceanic deep
When plotted on a time–temperature diagram, the contours crustal and mantle-rocks as a partial source for Magsat magnetic
of Figure 56 result. Each curve is the locus of all combinations anomalies. Geophysical Research Letters 19: 21–24.
Magnetite
50 150 250 350 450 550
1 Ga 1 Ga
100 Ma 100 Ma
10 Ma 10 Ma
1 Ma 1 Ma
100 ka 100 ka
10 ka 10 ka
1 ka 1 ka
100 year 100 year
Time, t
10 year 10 year
1 year 1 year
1 month 1 month
1 day 1 day
1h 1h
100 s 100 s
10 s 10 s
1s 1s
50 150 250 350 450 550
(a) Temperature, T (°C)
Hematite
20 50 100 150 200 250 300 350 400 450 500 550 600 650 675
1 Ga 1 Ga
100 Ma 100 Ma
10 Ma 10 Ma
1 Ma 1 Ma
100 ka 100 ka
10 ka 10 ka
1 ka 1 ka
100 year 100 year
Time, t
10 year 10 year
1 year 1 year
1 month 1 month
1 day 1 day
1h 1h
100 s 100 s
10 s 10 s
1s 1s
50 100 150 200 250 300 350 400 450 500 550 600 650 675
(b) Temperature, T (°C)
Figure 56 Time–temperature blocking contours for SD grains of magnetite (a) and hematite (b), calculated from the Néel theory (eqn [30] or
eqns [31]–[35]). Along each contour, the product VHc0 is constant and each combination (t and T ) produces 5%, 57%, or 95% blocking (upper, middle,
or lower curves, respectively, in each set). The blocking temperature is strongly t-dependent at low T but almost independent of t near the Curie
point where Ms(T ) changes rapidly. For details of the calculations, see the text.
(t and T ) that produce blocking for a specified ensemble VHc0. weak applied magnetic field Ho sin ot. Néel (1949, eqn 73)
For example, if t1 ¼ 700 ka, TB1 ¼ T0 (Brunhes VRM), it is easy showed that
to calculate the thermal demagnetization temperature TB2 that
@ko =@ ln o ¼ ð1=Ho Þð@Mr =@ ln t Þ [31]
will erase the VRM on a laboratory heating timescale
(t2 15 min, say) by following one of the blocking curves. Frequency-dependent susceptibility focuses on grains at
SD magnetites follow the Pullaiah contours closely, while their blocking condition, passing from stably magnetized to
larger grains require systematically higher temperatures to SP. By varying T, the Néel diagram can be scanned, a method
demagnetize their VRMs (Dunlop and Özdemir, 2000). of ‘magnetic granulometry,’ that is, of determining (V, Hc0).
Another way of studying viscous magnetization is to mea- Usually, ultrafine grains are targeted; these are SP at T0 but
sure how initial susceptibility depends on the frequency o of a become stably magnetized at low T ( Jackson et al., 2004).
In discussing TRM and thermal demagnetization, we treated (Figure 56). At lower T, as the t – T contours become more
TB as a sharp temperature. In reality, both processes are thermo- oblique, the blocking range DTn ¼ T0.05 – T0.95 widens. At high
viscous rather than truly thermal because natural heating or T, where m2s (T ) controls blocking in [24] or [35], Tn or TB are
cooling times even for a rapidly chilled lava flow are atomic almost independent of t, so the contours are steep and DTn is
reorganization times (1010 s). Blocking occurs over a range of narrow. To be specific, a magnetite partial TRM with
temperatures, indicated on Figure 56 as 5% and 95% blocking TB 550 C blocks over about a 3 C interval, whatever the
contours flanking the average contour specifying TB. cooling rate, while for TB 300 C, blocking requires cooling
The bounds on blocking are calculated in the following way. through 15–25 C, depending on dT/dt.
York (1978a,b) derived an expression for blocking temperature As an example of the use of the Pullaiah diagram in assessing
TB during slow cooling at a rate dT/dt. This same TB can be magnetization or remagnetization during slow cooling, consider
obtained from an isothermal magnetization equation, for exam- the case of Mars (Dunlop and Arkani-Hamed, 2005). Mars’ core
ple, [24], if t is identified with a characteristic cooling time t*: dynamo is thought to have ceased by 4.0 Ga. Did this early and
short-lived field allow enough time for Mars’ crust to become
t * ¼ kT B 2 =DEðT B Þ ðdT=dt Þ1 stably magnetized by TRM? For TB 500–550 C, TRM blocking
[32]
¼ T B = 23 þ ln 1:78* ðdT=dt Þ1 requires about 5 C of cooling or about 10 Ma at an assumed
cooling rate dT/dt 0.5 C/Ma. To cover the entire
(Dunlop and Özdemir, 1997, p. 480), the second expres- 500–550 C TB range would require about 100 Ma, which is
sion being obtained by substituting for DE(TB) from eqn [23] well within the lifetime of Mars’ dynamo. A second question is
and substituting t ¼ 1.78t. An alternative approach that makes whether the primary TRM could have resisted thermoviscous
no approximations is given by Dodson and McClelland-Brown demagnetization over 4 Ga since the field shut off. This question
(1980). can be answered by following one of the average blocking
To find 5% and 95% limits on blocking, note that accord- contours. The grains most susceptible to demagnetization are
ing to eqn [21], those with the lowest TB values, namely, 500 C, measured on a
laboratory timescale of a few minutes. The corresponding tem-
DMðt Þ=DMð0Þ ¼ exp½t=tðT n Þ ¼ n perature for a timescale of 4 Ga is 350–400 C. Provided that
[33]
if tðT n Þ ¼ t= ln ð1=nÞ cooling to this temperature level occurred before the field
ceased, all TRM would in principle have survived to the present
where Tn is the temperature at which DM is reduced to n ¼ 0.05 day. This point is crucial in interpreting the origin of the intense
or 0.95. Specifically, we have magnetic anomalies on Mars.
Cooling rate affects the intensity of TRM and its blocking, a
n ¼ 0:05,tðT 0:05 Þ ¼ 0:334t
vital point in determining paleointensities in slowly cooled
n ¼ 0:57,tðT B Þ ¼ 1:78t [34] orogens. Dodson and McClelland-Brown (1980) and
n ¼ 0:95, tðT 0:95 Þ ¼ 19:5t Halgedahl et al. (1980) predicted using the Néel SD theory
that the intensity of TRM in SD grains should increase for
The middle line of [34] shows that TB as defined in longer cooling times by as much as 40% between laboratory
Section 5.08.5.3 corresponds to 57% blocking. The blocking and geologic settings. Their predictions are based on the
equation corresponding to eqn [24] for a general n is change in TB with t in eqn [25], but this equation often fails
to predict accurately the proportionality factor between TRM
VHc0 ¼ ½2k=Ms0 f23 þ ln ½t= ln ð1=nÞg T n =m2s ðT n Þ (35)
and Ho. Experimentally, Fox and Aitken (1980) found for
The Pullaiah diagrams (Figure 56) are computed as follows: baked clay samples containing SD magnetite about a 7%
increase in the intensity of TRM when the cooling time chan-
1. Choose values of V, Hc0, and TB. Determine the value of
ged from 3 min to 2.5 h. The only serious attempt to apply a
ms(TB) for the chosen TB. Calculate the matching value of t
cooling-rate correction in paleointensity work is by Selkin et al.
from [24].
(2000). In their study of the Archean Stillwater Complex, the
2. For the same t, determine the values of T0.05 and T0.95 that
uncorrected paleointensity was estimated to be 45% higher
satisfy [35]. This is a trial-and-error process. Even if ms(T )
than the true paleofield recorded during 0.8 Ma cooling.
can be represented analytically (e.g., Dunlop et al., 2000),
Bowles et al. (2005) found that submarine basaltic glasses
the resulting transcendental equation is best solved by
cool in nature at about the same rate as in the laboratory, so
iteration.
minimal or no cooling-rate corrections were necessary.
3. Find the corresponding cooling rate by substituting Tb and
t* ¼ t in [32].
5.08.6.3 Chemical Remanent Magnetization
For outside limits 10 s
t*
1 Ga and 300 K
TB
950 K,
5
t*(dT/dt)
35. Thus, t* can be interpreted as 5–35 times VRM is not the only source of secondary NRM in rocks. Equally
the time to cool through 1 C. For example, for a typical common, and serious because of its often high coercivities and
magnetite blocking temperature of 540 C ¼ 813 K and a cool- TBs, is CRM, produced by magnetization blocking in a growing
ing timescale t* ¼ 1 Ma, t* 15(dt/dT ) or the time to cool magnetic phase. In Figure 47, two CRMs are present in the
through about 15 C if (dT/dt) is constant. rock, and both have higher blocking temperatures (since they
We must not confuse t* with the time to pass from 5% to reside in magnetite and hematite) than the paleomagnetically
95% blocking. The blocking interval Dt ¼ t0.05 – t0.95 is of con- interesting pyrrhotite partial TRMs. In this particular example,
stant width on a logarithmic scale for short or long t the CRM directions mimic those of the partial TRMs, effectively
dating the CRMs. Generally, however, CRMs are difficult to unmixing were intentionally produced during experiments
date with any precision and are merely noise obscuring the (Draeger et al., 2006) concerns paleointensity determination on
primary paleomagnetic signal. basalts, which very frequently reequilibrate chemically in the
One setting in which CRM is useful, indeed vital, is the course of a Thellier experiment. Draeger et al. found that CRM
seafloor. Primary TM60 in pillow lavas produced at mid- (produced near ambient temperature) and thermochemical rem-
ocean ridges is soon oxidized in seawater to titanomaghemite anence (TCRM) produced at higher T are virtually indistinguish-
(Sections A.2.1 and 5.08.3.4). Oxidation proceeds inward able in their Arai plots from pure TRM. However, there is no
from the surface of grains, probably aided by iron removal by reason to suppose that the proportionality between remanence
sulfate-reducing bacteria (Carlut et al., 2007). Exchange and field is the same for them as it is for pure TRM. Since basalts
coupling seems to remain unbroken across the moving phase are by far the most frequently studied rock in paleointensity work,
boundary between the two spinel phases. Coupling is aided by this finding is sobering. Paleofield estimates could be seriously in
the diffuse nature of the boundary region, which has a gradient error despite apparently high-quality Arai plots.
of lattice vacancies extending inward from the surface. CRMs are particularly common in chemically active environ-
Nevertheless, the lattice mismatch between the two minerals ments such as soils and recently deposited sediments. Their
ultimately causes cracking of the surface titanomaghemite and detection and elimination are a whole science in itself (e.g.,
a considerable reduction in effective grain size. This in turn Rowan and Roberts, 2006; see also Verosub and Roberts, 1995).
makes the titanomaghemite CRM considerably more coercive
than the parent TM60 TRM and ensures stability of the rema-
5.08.6.4 Detrital and Postdepositional Remanent
nence over the lifetime of an ocean basin. Linear magnetic
Magnetizations
anomalies over the oceans, the hallmark of seafloor spreading,
are usually thought of in terms of TRM, but their actual source Detrital remanent magnetization (DRM) is produced when pre-
is CRM. Geophysically speaking, it is fortunate indeed that the viously magnetized mineral grains settle through a water column
CRM direction ‘remembers’ that of the parent TRM. The oxida- in the geomagnetic field. The partial alignment of the magnetic
tion cannot be pushed beyond about z ¼ 0.65, where multi- moments of the grains may subsequently be altered by changes
phase oxidation products begin to appear (Section 5.08.3.5). in the water-rich sediment as it is buried beneath later sediments,
The CRM direction then increasingly deviates from the TRM compressed, and reworked by burrowing organisms. The altered
direction (Özdemir and Dunlop, 1985; Figure 57). The fidelity remanence is called postdepositional remanent magnetization
of magnetic stripes over the oceans suggests that oxidation of (PDRM). For a review, see Kodama (1992).
the seafloor typically stays below this level. The key question for paleomagnetic directional studies is
Controlled CRM experiments are difficult. Conditions in whether or not DRM and PDRM faithfully record the geomag-
nature are hard to reproduce in the laboratory, and if heating is netic field at the time of deposition or shortly thereafter. In
used to accelerate the kinetics, undesired additional changes may order for sediments and the sedimentary rocks that eventually
occur. One recent study in which additional oxidation and form from them to be useful as recorders of the paleofield
40
Soft
component
Deflection of Mcr relative to Mtr (deg)
30 Mtr
20 Total
CRM,
Mcr
10
Hard
component
0
0 0.2 0.4 0.6 0.8 1.0
Oxidation parameter, z
Figure 57 Direction of CRM formed by maghemitization of single-domain TM60 in a field applied perpendicular to the initial TRM direction. CRM forms
parallel to initial TRM until multiphase oxidation products begin to appear around z ¼ 0.6. Modified from Özdemir Ö and Dunlop DJ (1985) An
experimental study of chemical remanent magnetizations of synthetic monodomain titanomaghemites with initial thermoremanent magnetizations.
intensity, DRM/PDRM must in addition be proportional to the analogous in its coercivity spectrum to DRM/PDRM. Although
field acting during deposition. This question is the subject of there is no a priori reason to expect such equivalence, the success
lively debates. Since DRM is strongly dependent on the grain of the pseudo-Thellier method (e.g., Brachfeld and Banerjee,
size, composition, and concentration of the magnetic phase(s) 2000) justifies the assumption.
and the properties of the nonmagnetic matrix, there is no Yu et al. (2002a,b, 2003) had investigated whether or not
simple answer to either question. partial ARMs in magnetite obey laws of additivity, reciprocity,
Almost any theory indicates that the alignment of magnetic and independence analogous to the Thellier laws for partial
moments during settling in a magnetic field occurs rapidly TRMs (Section 5.08.5.5). This is a necessary condition for suc-
(within seconds for
100 mm magnetite grains) or not at all cess of the pseudo-Thellier method. They found the following:
(for millimeter-size grains) (see Dunlop and Özdemir, 1997,
1. For SD, PSD, and MD grains, partial ARMs produced in
Chapter 1.17). However, at the bottom of the water column,
nonoverlapping Hc intervals covering the AF interval (0,
elongated grains will tend to rotate until the long axis is hori-
Hmax) sum to give the total ARM (0, Hmax) (additivity law).
zontal. Since magnetic moments of elongated grains prefer to
2. For SD and PSD grains, a partial ARM produced by applying
be parallel to the long axis to reduce self-demagnetizing fields,
a steady field while ramping down the AF from Hc2 to Hc1 is
this rotation produces a shallowing of the DRM vector, called
AF demagnetized mainly over the interval (Hc1, Hc2) (reci-
the inclination error. DRM with no substantial post-
procity law). In the case of MD grains, however, >50% of the
depositional reworking does exhibit an inclination error in
partial ARM is demagnetized below Hc1 (nonreciprocity).
both natural and laboratory redepositional settings (Tauxe
3. As a result of 2, for SD and PSD grains, partial ARMs
and Kent, 1984). So do the remanences of deep-sea sediments
produced in nonoverlapping Hc intervals are independent
redeposited in the laboratory and compacted under hydro-
of one another in direction and magnitude (independence
static pressure. The origin of the inclination shallowing in
law). For MD grains, nonreciprocity results in overlap
this case seems to be locking of magnetite grains to clay parti-
between AF demagnetization ranges such that the direction
cles during formation of clay fabric (Sun and Kodama, 1992).
of the higher-coercivity partial ARM is recoverable, but the
It may be possible to ‘undo’ inclination shallowing using a
direction of the lower-coercivity partial ARM is spurious.
correction based on measured magnetic fabric anisotropy
On the other hand, the intensity of the lower-coercivity
(Hodych and Bijaksana, 1993; Kodama and Sun, 1992).
partial ARM is relatively well estimated by pseudo-Thellier
Postdepositional reworking of a sediment may reduce or
analysis of the multivectorial total ARM, but the intensity of
remove the inclination error. Although demonstrated in labora-
the higher-coercivity partial ARM is always underestimated.
tory experiments (e.g., Kent, 1973), it is less clear that this is
always the case under natural conditions. Within a fluid layer at As a result of 2 and 3, pseudo-Thellier plots are increasingly
the top of the sediment column, magnetic grains continue to nonlinear for larger grain sizes (Figure 58).
rotate in water-filled voids until either lack of pore space or Finally, although magnetite is typically the primary magnetic
adhesion to clay particles causes lock-in. Generally, slow deposi- mineral in clastic sediments, it is perfectly possible for hematite
tion in the deep sea means that there could be a time lag of to acquire DRM and PDRM. Hematite-rich redbeds are of
several thousand years between initial deposition and lock-in.
This estimate is based on continuous deposition experiments in
the laboratory (e.g., Lvlie, 1976), which suggest not only a time 1.0
lag but also some blurring or loss of resolution of rapid geomag-
netic field changes. The complex array of factors that potentially 0.8
affect PDRM has led to increasingly intricate and sophisticated
ARM remaining
modeling (e.g., Heslop et al., 2004; Witt et al., 2004). Ideal SD

The question of linearity of DRM/PDRM intensity with 0.6
applied field Ho is far from clear despite decades of experimen-
tation. In classic studies, Johnson et al. (1948) redeposited 0.4
varved clays and found linear DRM intensities for Ho
1 Oe or 0.065 mm
80 A m1, while there was essentially no linear region for PDRM 0.24 mm
0.2 1.06 mm
in sediments redeposited by Tucker (1980). Tauxe et al. (2006) 18.3 mm
concluded from their experiments that DRM intensity is not
linearly related to field strength under many natural conditions.
0.0
Apart from the uncertainty about linearity of sedimentary 0.0 0.2 0.4 0.6 0.8 1.0
paleointensities, there is another difficulty. Because the DRM/ pARM gained
PDRM process cannot really be reproduced in the laboratory,
sediments and sedimentary rocks yield only relative paleointen- Figure 58 Simulated pseudo-Thellier paleointensity experiments
comparing the remaining fraction of an initial ARM after AF
sities. Tauxe et al. (1995) developed a pseudo-Thellier method
demagnetization with partial ARM gained in the same AF for magnetites
that compares different coercivity fractions of the NRM with
of various grain sizes. As with the simulated Thellier experiments of
corresponding fractions of a laboratory-produced ARM. ARM Figure 49, the results increasingly ‘sag’ below the ideal SD line of slope
avoids heating and alteration of the material and for this reason 1 as grain size increases. Modified from Yu Y, Dunlop DJ, and
is sometimes used as an assumed analogue of TRM in determin- Özdemir Ö (2002b) Partial anhysteretic remanent magnetization in
ing (absolute) paleointensities of igneous rocks (see Chapter magnetite. 2. Reciprocity. Journal of Geophysical Research 107(B10):
5.13). In the pseudo-Thellier method, ARM is taken to be 2245 (http://dx.doi.org/10.1029/2001JB001269).
considerable importance in the paleomagnetic record; indeed, when a narrow range of TB has been isolated as a partial TRM
they dominated early studies because of their typically strong (e.g., Figure 46).
NRMs. The ultrafine-grain hematite pigment that colors redbeds Rock magnetism has mainly concentrated on serving the
is secondary and carries CRM (or no remanence at all if the grains needs of paleomagnetism, explaining how NRM is produced
are SP). AF treatment generally fails to demagnetize the CRM and stabilized over geologic history, and inventing techniques
completely because of SD hematite’s high coercivities, but ther- for characterizing and selectively enhancing the signal of the
mal demagnetization may succeed because the very fine grain most desirable of the remanence carriers. What of induced
sizes lead to relatively low TB’s (eqn [24]). The remaining primary magnetization? Most continental magnetic anomalies proba-
DRM/PDRM signal in coarser detrital hematite grains, if any, is bly originate in the induced magnetization of MD magnetite,
then revealed. Most paleomagnetists would feel uncomfortable although there are notable exceptions (e.g., McEnroe et al.,
in accepting any hematite NRM as primary unless supported 2004). Oceanic magnetic anomalies are more likely to have
by field evidence, for example, correlation between NRM and an SD or SD-like source because of the oxidation/granulation
bedding tilt. processes the minerals have undergone on the seafloor or in
the shallow crust. Even at high T in the deep crust, MD grains
exhibit only slight susceptibility enhancement, but SD grains
5.08.7 Summary can have a sizable ko(T ) (Hopkinson) peak, which is further
increased, possibly even doubled, by thermoviscous effects
The induced and remanent magnetizations of rocks are the over the Brunhes epoch (Figure 55).
result of a complex interplay among factors like mineralogy, Extremely fine grains are often of interest in paleoclimate
grain size, domain structure, temperature, time, and ambient studies using loess–paleosol successions or in other environmen-
geomagnetic field. For a given mineral, material properties like tal applications (e.g., Evans and Heller, 2003). Here, susceptibil-
Ms, K, and l are slowly varying functions of T except in the ity is generally used simply as a measure of the quantity of
vicinity of the Curie point or other phase transitions. It is magnetic material, and its grain-size dependence is ignored. It
relatively sharp changes in properties that are mainly responsi- is true that ko is an insensitive probe of size variation in the
ble for the recording and preservation of a stable paleomagnetic SD ! MD range (Heider et al., 1996) but this is not the case for
remanence, however. One such transformation is from single- ultrafine grains, which are SP or nearly SP (viscous) at ordinary
domain to nonuniform domain structures with changing grain temperatures. These grains have strongly size-dependent ko,
size. Although this transition is not as abrupt as imagined in which may be orders of magnitude greater than ko for SD or
classical domain theory, the first appearance of a spin vortex or MD grains. Scanning ko(T ) at low T is an effective granulometric
other embryonic wall nucleus signals the end of high-coercivity probe since kSP is proportional to V for small fields (eqn [19]).
coherent rotation of the entire coupled spin structure of a grain. The classic interpretation (the ‘Chinese model,’ based on
Reversal by spin curling or, in larger grains, by wall displace- observations from the rainy east-central Chinese loess plateau)
ment compromises paleomagnetic stability. A single-domain– is that SP magnetite forms primarily in response to high rain-
multidomain dichotomy may be unrealistic, but SD or SD-like fall. However, the magnetic enhancement of modern topsoil is
coercivity is still the paleomagnetic ideal. due rather to 3–30 nm cuboidal magnetite, likely produced by
The other crucial change in properties occurs during NRM iron-reducing bacteria (Banerjee, 2006). Furthermore, detailed
acquisition and is truly sharp on ordinary timescales. This is work on loess–paleosol sequences from northern and western
blocking the passage from a thermal equilibrium state to a China, Alaska, and Argentina is also calling into question the
frozen out-of-equilibrium state that may survive unchanged generality of the old Chinese model for magnetic enhancement
for geologic lengths of time. Blocking is not the result of any (e.g., Carter-Stiglitz et al., 2006).
change in domain structure. It is due simply to the exponential In this chapter, we have attempted to summarize briefly the
dependence of relaxation time on grain properties like volume most useful aspects of rock magnetism, emphasizing recent
and coercivity and on T (eqn [23]). Just below TC, where SD developments. A more comprehensive coverage with separate
blocking temperatures are concentrated, TRM blocking typi- treatments of igneous, sedimentary, and metamorphic rocks
cally occurs over an interval of a few degrees Celsius. CRM appears in Dunlop and Özdemir (1997). Magnetic mineralogy
blocking requires only a few percent increase in V. Of course, is covered thoroughly by O’Reilly (1984) and the physics of
blocking is only sharp in a time sense for SD grains with rock magnetism by Stacey and Banerjee (1974). These sources,
reasonably fast rates of cooling or grain growth. although older and less linked to applications, are often
Any MD grain has a multiplicity of magnetization states and invaluable. A recent brief but useful primer on rock magnetism
consequently a set of both blocking and demagnetization tem- is found in Tauxe (1998).
peratures. The whole concept of blocking becomes blurred. When
one wall in a grain moves, all the other walls seek new equilib-
rium positions so as to minimize the overall self-demagnetizing
field. Therefore, blocking is conditional: a domain wall is only Appendix A Developments in Rock and Mineral
locked in place as long as none of its neighbors moves. During Magnetism 2006–12
cooling, blocking! remobilization! blocking will occur repeat-
edly. This complexity has so far defeated efforts to model MD In the period 2006–12 since the first edition of this book, there
grains containing more than a single wall (Néel, 1955) in any have been great advances in both fundamental and applied
direct way. The set of unblocking temperatures during zero-field rock and mineral magnetism, particularly in the areas reviewed
heating is easier to model and is well studied experimentally in the succeeding text. A number of state-of-the-art reviews and
topical collections of papers have appeared; among them are (Rivas-Sanchez et al., 2009) are balanced by theoretical studies of
the following: critical sizes for the onset of single-domain and SP behavior in
Fundamental and Frontier Research in Rock Magnetism (32 elongate chains of crystals, with implications for interacting
papers, Journal of Geophysical Research, vol. 111, no. B12, 2006, magnetosomes (Muxworthy and Williams, 2006; Newell,
eds. Jackson MJ, Williams W, and Smirnov A) covers the spec- 2009). Fundamental to Néel’s model of single-domain behavior,
trum from fundamental rock magnetism to paleomagnetic and reversal modes at different angles to the applied field have been
other applications. Some papers are referenced individually. documented experimentally (Milne and Dunlop, 2006).
Magnetism, iron minerals, and life on Mars
(Astrobiology,Rochette et al., 2006).
Paleofield intensity determination using single silicate crys-
tals (Reviews of Geophysics, Tarduno et al., 2006a,b and Ele-
A.2 Magnetic Minerals and Their Properties
ments, Tarduno, 2009). A.2.1 Titanomagnetites
Paleointensity determinations using submarine basaltic
Curie temperatures (Lattard et al., 2006) and low-temperature
glasses (Physics of the Earth and Planetary Interiors, Tauxe, 2006).
magnetic properties (Carter-Stiglitz et al., 2006; Engelmann
Magnetism from Atomic to Planetary Scales (18 papers,
et al., 2010) have been determined for titanomagnetites with
Geochemistry Geophysics Geosystems theme collection
Ti contents that were previously uninvestigated.
2008–2010, eds. Moskowitz BM, Feinberg JM, Florindo F, and
Roberts AP). The title is indicative of the breadth of coverage of
these papers, some of which are referenced individually.
Magnetism of Volcanic Materials (21 papers, A.2.2 Maghemite and Titanomaghemites
Goguitchaichvili et al., 2009). Contrary to the title, the topics A new weapon has been added to the arsenal aimed at the
range widely. Some of the papers are referenced individually. difficult task of recognizing partial low-temperature oxidation
Magnetic properties of greigite Fe3S4 (Reviews of Geophysics, of magnetite and titanomagnetite. Spreading and blurring of
Roberts et al., 2011). the Verwey transition has been the usual tool of choice for
TRM and the Thellier paleointensity method (Physics of the diagnosing maghemitization of magnetite or near-magnetite,
Earth and Planetary Interiors, Dunlop, 2011). but Özdemir and Dunlop (2010) demonstrated that humps in
Rock magnetic recording (Physics Today, Dunlop, 2012). the cooling curves of room-temperature remanence are both
diagnostic and quantitative, allowing an estimate of the oxida-
tion parameter z to be made.
A.1 Domains and the Magnetization Process
A.1.1 Domain Walls and Multidomain Magnetization
A.2.3 Hematite
Domains and domain wall pinning have been observed using
electron holography on magnetite below its Verwey transition Hematite has proven to have distinctive FORC diagrams
(120 K) by Kasama et al. (2010) and Church et al. (2011). (Roberts et al., 2006), a grain size-varying Morin transition
Previous documentation of domains in the low-temperature temperature TM (Özdemir et al., 2008), and an unanticipated
phase of magnetite was tenuous. low-temperature basal-plane magnetism that can nucleate a
much-larger room-temperature magnetic moment in warming
through TM (Özdemir and Dunlop, 2006a,b).
A.1.2 Observations of Domains
The frontier of observations continues to be electron microscopy,
particularly off-axis electron holography and MFM because of A.2.4 Titanohematites (Hemoilmenites)
the information they offer about vector magnetization. How-
From single crystals from the Ecstall pluton, BC (Brownlee
ever, Uehara et al. (2010) emphasized recent advances in
et al., 2011), to fine exsolution intergrowths responsible for
magneto-optical imaging applied to rock magnetism.
lamellar magnetism (McEnroe et al., 2007; Robinson et al.,
2006), now investigated by Lorentz microscopy (Kasama
A.1.3 Micromagnetic Modeling et al., 2009), the hematite–ilmenite solid-solution series has
proven a rich field of investigation. The classic Ti-rich compo-
The past 6 years have seen micromagnetic modeling advance
sitional range for self-reversing TRM, where Ti-ordered and Ti-
from calculations on prismatic crystals of simple regular shapes
disordered regions are juxtaposed by rapid chilling, has also
to sophisticated treatments of polyhedra with irregular shapes
been examined recently by Fabian et al. (2011).
and nonplanar surfaces (Fukuma and Dunlop, 2006;
Muxworthy et al., 2006b; Williams et al., 2010, 2011; Yu and
Tauxe, 2008).
A.2.5 Iron Sulfides
Fundamental room-temperature and low-temperature proper-
A.1.4 Single-Domain Grains and Superparamagnetism
ties of greigite (Fe3S4) have been established in new work by
Applications like new methods of detecting single-domain Chang et al. (2008, 2009). Earlier, Roberts et al. (2006) charac-
assemblages by magnetic measurements (e.g., Mitra et al., terized the FORC diagrams of both greigite and pyrrhotite
2011) and the discovery of magnetite nanoparticles in iron ores (Fe7S8).
A.2.6 ‘Magnetic’ Silicates effect of multidomain and PSD remanences (Biggin and
Poidras, 2006; Paterson, 2011; Shaar et al., 2011).
Magnetic inclusions and precipitates in nonmagnetic host min-
erals have been studied in basaltic silicate glasses (Bowles et al.,
2011), plagioclase (Wenk et al., 2011), and olivines (Belley A.4.3 Stability and Domain-State Tests
et al., 2009) where they potentially could record pre-
Among significant advances in using the Day diagram, a cross
accretionary remanence (Lappe et al., 2011). The internal struc-
correlation plot of two well-recognized hysteresis parameters
ture of magnetic inclusions has been investigated by electron
are rigorous testing using controlled mixtures of magnetites of
holography (Feinberg et al., 2006). Minerals that lack iron in
different sizes and domain states (Dunlop and Carter-Stiglitz,
their lattice on occasion can contain enough impurity iron to be
2006) and a procedure for finding best-fit binary mixing lines
detectably (and usefully) magnetic: quartz (Tarduno, 2009)
(Heslop and Roberts, 2012a). Enkin et al. (2007) used the Day
and even diamonds (Clement et al., 2008) are in this category.
plot to gain insights into variable diagenetic alteration in sed-
iments hosting gas hydrates.
A.2.7 Biogenic Magnetic Minerals FORC diagrams continue to be a major focus of research,
extending to new minerals (Roberts et al., 2006) and develop-
Electron holography (Kasama et al., 2006), ferromagnetic res-
ing insights into hallmarks of various sometimes subtle
onance spectroscopy (Kopp et al., 2006), and FORC diagrams
domain states and mixtures of states (Carvallo and
(Carvallo et al., 2009; Chen et al., 2007) have all been used to
Muxworthy, 2006; Carvallo et al., 2009; Chen et al., 2007;
characterize biogenic magnetic particles, magnetotactic bacte-
Church et al., 2011; Egli et al., 2010).
ria, and their magnetofossils in sediments. A more exotic appli-
The analysis and evaluation of hysteresis data using all
cation is to magnetic material in brain tissue (Hirt et al., 2006).
available tools are reviewed by Jackson and Solheid (2009).
Magnetic mixtures and the best approach to unmixing their
superimposed magnetic data remain a lively and practically
A.3 Induced and Remanent Magnetization
important topic (e.g., Frank and Nowaczyk, 2008; Gong
A.3.1 MD Hysteresis and Susceptibility et al., 2009; Heslop and Roberts, 2012b). A new granulometry
technique, thermal fluctuation tomography, potentially reveals
The AC susceptibility of magnetite at low temperatures as a
the detailed particle size and shape distribution of a sample
function of both field and temperature has been measured by
(Jackson et al., 2006a,b).
Özdemir et al. (2009) and modeled based on observed domain
wall pinning by Church et al. (2011).
A.5 Remanent Magnetization Processes in Nature
A.3.2 Magnetostatic Interactions
A.5.1 Thermoremanent Magnetization
Magnetostatic particle interactions have been studied theoret-
ically in the context of FORC diagrams (Egli, 2006) and exper- A sobering discovery (Selkin et al., 2007) is that TRM intensity
imentally in magnetite–ulvöspinel intergrowths (Evans et al., is not always proportional to field strength in the weak-field
2006), in sediment suspensions (Heslop et al., 2006), and in region (fields comparable to geomagnetic field strengths),
magnetite particle arrays produced by electron-beam lithogra- undermining the basic assumption of all paleointensity
phy (Muxworthy et al., 2006; Krása et al., 2011). methods – not just the Thellier method.
The effect of cooling rate on TRM intensity has been looked
into in greater detail in recent years (Biggin et al., 2013; Ferk
A.4 NRM and Paleomagnetic Stability et al., 2010; Leonhardt et al., 2006; Yu, 2011). The consensus
seems to be that the corrections required are in many cases not
A.4.1 Thermal Demagnetization of MD Grains as large as had been feared.
Continuous demagnetization and stepwise thermal demagne- Partial TRMs and their demagnetization ‘tails’ have been the
tization give equivalent results for SD grains but not for MD focus of numerous studies, among them Biggin and Poidras
grains (Dunlop, 2009). An investigation of Néel’s thermal (2006), Yu and Dunlop (2006), and Yu and Tauxe (2006).
fluctuation field in the context of basalt magnetization was
carried out by Muxworthy et al. (2009). A.5.2 Viscous and Thermoviscous Magnetization
A fascinating application of thermoviscous magnetization is to
A.4.2 Resolving Multivectorial NRM and Thellier
magnetic field changes observed to accompany volcanic erup-
Paleointensity Determination
tions (Hashimoto et al., 2008). On a much longer timescale
One of the problems with the classic Thellier method is that it and a different planet (the Moon), Garrick-Bethell and Weiss
only works in the ideal case of single-domain carriers of TRM (2010) have worked out a Pullaiah-type time–temperature
(Shcherbakov and Zhidkov, 2006) and even these have been nomogram applicable to kamacite.
questioned (Morales et al., 2011). Much effort has thus been
enlisted in devising domain state-independent paleointensity
A.5.3 Chemical Remanent Magnetization (CRM)
methods (Dekkers and Böhnel, 2006; Dunlop, 2011; Fabian
and Leonhardt, 2010; Muxworthy, 2010; Muxworthy et al., Chemical alteration in nature and the remanence changes it
2011) or in adapting Thellier methodology to mitigate the imparts lurk in the background of many a paleomagnetic
study. In submarine basalts, alteration can lead to an actual Banerjee SK (2006) Environmental magnetism of nanophase iron minerals: Testing the
reversal of the NRM, clouding the pattern of magnetic stripes biomineralization pathway. Physics of the Earth and Planetary Interiors
154: 210–221.
that record field reversals and seafloor spreading (Carvallo
Belley F, Ferré EC, et al. (2009) Magnetic properties of natural and synthetic (Fex,
et al., 2010; Doubrovine and Tarduno, 2006). Even alteration Mg1x)2SiO4 olivines. Earth and Planetary Science Letters 284: 516–526.
of the very young crust by microorganisms affects the NRM Berquó TS, Banerjee SK, Ford RG, and Lee Penn R (2007) High crystallinity
(Carlut et al., 2007). Self-reversal in a different setting, Precam- Si-ferrihydrite: As insight into its Néel temperature and size dependence of magnetic
brian basalts, is attributable to the oxidation of magnetite to properties. Journal of Geophysical Research 112: B02102. http://dx.doi.org/
10.1029/2006JB004583.
martite (hematite) (Swanson-Hysell et al., 2011). Biggin AJ, Badejo S, et al. (2013) The effect of cooling rate on the intensity of
Fabian (2009) drew attention to the possibility that TCRM thermoremanent magnetization (TRM) acquired by assemblages of pseudo-single-
could lead to overestimation of paleofields. Draeger et al. domain, multidomain and interacting single-domain grains. Geophysical Journal
(2006) demonstrated experimentally that CRM and TCRM International. http://dx.doi.org/10.1093/gji/ggt078.
Biggin AJ and Böhnel HN (2003) A method to reduce the curvature of Arai plots
can successfully mimic TRM in basalts, making this a very
produced during Thellier palaeointensity experiments performed on multidomain
real possibility. grains. Geophysical Journal International 155: F13–F19.
Biggin AJ and Poidras T (2006) First-order symmetry of weak-field partial
thermoremanence in multidomain (MD) ferromagnetic grains. 1. Experimental
evidence and physical implications. Earth and Planetary Science Letters
A.5.4 Detrital and Postdepositional Remanent 245: 438–453.
Magnetizations Bina M and Daly L (1994) Mineralogical change and self-reversed magnetizations in
pyrrhotite resulting from partial oxidation; geophysical implications. Physics of the
Deposition experiments (Carter-Stiglitz et al., 2006a; Heslop Earth and Planetary Interiors 85: 83–99.
et al., 2006) and DRM investigations generally have been less Bowles J, Gee J, Burgess K, and Cooper R (2011) Timing of magnetite formation in
active than other areas recently. Nevertheless, two important basaltic glass: Insights from synthetic analogs and relevance for geomagnetic
paleointensity analyses. Geochemistry, Geophysics, Geosystems 12: Q02001.
theoretical treatments have appeared in the last 2 or 3 years:
http://dx.doi.org/10.1029/2010GC003404.
Mitra and Tauxe (2009) and Shcherbakov and Sycheva (2010). Bowles J, Gee JS, Kent DV, Bergmanis E, and Sinton J (2005) Cooling rate effects on
paleointensity estimates in submarine basaltic glass and implications for dating
young flows. Geochemistry, Geophysics, Geosystems 6: Q07002. http://dx.doi.org/
A.5.4.1 Additional areas of interest 10.1029/2004GC000900.
The study of soils, paleosols and loess, and environmental mag- Bowles JA, Hammer JE, and Brachfeld SA (2009) Magnetic and petrologic
characterization of synthetic Martian basalts and implications for the surface
netism generally has exploded in the past decade and is beyond
magnetization of Mars. Journal of Geophysical Research 114: E10003. http://dx.doi.
the capability of this chapter to summarize. Two examples of org/10.1029/2009JE003378.
recent work of particular rock magnetic interest are Geiss and Boyd JR, Fuller M, and Halgedahl SL (1984) Domain wall nucleation as a controlling
Zanner (2006) on pedogenic magnetite in loess and Zhao and factor in the behaviour of fine magnetic particles in rocks. Geophysical Research
Roberts (2010) on the magnetization process of Chinese loess. Letters 11: 193–196.
Brachfeld SA and Banerjee SK (2000) A new high-resolution geomagnetic relative
Magnetism and magnetic minerals in the deep crust and paleointensity record for the North American Holocene: A comparison of
possibly the upper mantle have been of great interest recently sedimentary and absolute intensity data. Journal of Geophysical Research
as possible sources of observed (but poorly explained) long- 105: 821–834.
wavelength magnetic anomalies at the Earth’s surface and at Brachfeld SA and Hammer J (2006) Rock magnetic and remanence properties of
synthetic Fe-rich basalts: Implications for Mars crustal anomalies. Earth and
satellite altitudes. Papers in this area either presenting data for
Planetary Science Letters 248: 599–617.
minerals brought to the surface from great depths or summa- Brown K and O’Reilly W (1988) The effect of low-temperature oxidation on the
rizing the candidate minerals, and their known or inferred remanence of TRM-carrying titanomagnetite Fe2.4Ti06O4. Physics of the Earth and
properties include Strada et al. (2006), Liu et al. (2008), Planetary Interiors 52: 147–154.
Qingsheng et al. (2008), Zhao and Tominaga (2009), Brownlee S, Feinberg J, et al. (2011) Magnetic properties of ilmenite–hematite single
crystals from the Ecstall pluton near Prince Rupert, British Columbia. Geochemistry,
Dunlop et al. (2010), and Ferré et al. (2013). Geophysics, Geosystems 12: Q07Z29. http://dx.doi.org/10.1029/2011GC003622.
Finally, planetary and solar system magnetic records from Butler RF and Banerjee SK (1975a) Theoretical single-domain grain-size range in
meteorites (Martian, lunar, and asteroidal) and from returned magnetite and titanomagnetite. Journal of Geophysical Research 80: 4049–4058.
lunar samples, as well as experiments on materials that might Butler RF and Banerjee SK (1975b) Single-domain grain-size limits for metallic iron.
simulate the most magnetic crust on Mars, have experienced a
Carlut J, Horen H, and Janots D (2007) Impact of microorganisms activity on the natural
resurgent interest not matched since the Apollo era. A small remanent magnetization of the young oceanic crust. Earth and Planetary Science
selection includes Brachfeld and Hammer (2006), Acton et al. Letters 253: 497–506.
(2007), Lawrence et al. (2008), Bowles et al. (2009), Garrick- Carmichael RS (1982) Magnetic properties of minerals and rocks. In: Carmichael RS
Bethell et al. (2009), Thomas-Keprta et al. (2009), Rochette (ed.) Handbook of Physical Properties of Rocks, vol. 2, pp. 229–287. Boca Raton,
FL: CRC Press, ch. 2.
et al. (2010), Lappe et al. (2011), and Uehara et al. (2011). Carter-Stiglitz B, Banerjee SK, Gourlan A, and Oches E (2006a) A multi-proxy study of
Argentina loess: Marine oxygen isotope stage 4 and 5 environmental record from
pedogenic hematite. Palaeogeography, Palaeoclimatology, Palaeoecology 239: 45–62.
Carter-Stiglitz B, Moskowitz B, and Jackson M (2001) Unmixing magnetic assemblage
References and the magnetic behavior of bimodal mixtures. Journal of Geophysical Research
106: 26397–26411.
Acton G, Yin Q, et al. (2007) Micromagnetic coercivity distributions and interactions in Carter-Stiglitz B, Moskowitz B, et al. (2006b) Low-temperature magnetic behavior of
chondrules with implications for paleointensities of the early Solar System. Journal multidomain titanomagnetites: TM0, TM16, and TM35. Journal of Geophysical
of Geophysical Research 111: B03S90. http://dx.doi.org/10.1029/2006JB004655. Research 111: B12S05. http://dx.doi.org/10.1029/2006JB004561.
Argyle KS and Dunlop DJ (1990) Low-temperature and high-temperature hysteresis of Carter-Stiglitz B, Valet J-P, and LeGoff M (2006c) Constraints on the acquisition of
small multidomain magnetites (215–540 nm). Journal of Geophysical Research remanent magnetization in fine-grained sediments imposed by redeposition
95: 7069–7083. experiments. Earth and Planetary Science Letters 245: 427–437.
Carvallo C, Dunlop DJ, and Özdemir Ö (2005) Experimental comparison of Dunlop DJ (2002a) Theory and application of the Day plot (Mrs/Ms versus Hcr/Hc). 1.
FORC and remanent Preisach diagrams. Geophysical Journal International Theoretical curves and tests using titanomagnetite data. Journal of Geophysical
162: 747–754. Research 107(B3): 2056.
Carvallo C, Hickey S, Faivre D, and Menguy N (2009) Formation of magnetite in Dunlop DJ (2002b) Theory and application of the Day plot (Mrs/Ms versus Hcr/Hc). 2.
Magnetospirillum gryphiswaldense studied with FORC diagrams. Earth, Planets and Application to data for rocks, sediments, and soils. Journal of Geophysical Research
Space 61: 143–150. 107(B3): 2057.
Carvallo C and Muxworthy A (2006) Low-temperature first-order reversal curve (FORC) Dunlop DJ (2003a) Stepwise and continuous low-temperature demagnetization.
diagrams for synthetic and natural samples. Geochemistry, Geophysics, Geophysical Research Letters 30(11): 1582. http://dx.doi.org/
Geosystems 7: Q09003. http://dx.doi.org/10.1029/2006GC001299. 10.1029/2003GL017268.
Carvallo C, Özdemir Ö, and Dunlop DJ (2004) First-order reversal curve (FORC) Dunlop DJ (2003b) Partial thermoremanent magnetization: Louis Néel’s legacy in rock
diagrams of elongated single-domain grains at high and low temperatures. Journal magnetism. Journal of Applied Physics 93: 8236–8240.
of Geophysical Research 109: B04105. http://dx.doi.org/10.1029/2003JB002539. Dunlop DJ (2007a) Transition warming and cooling remanences in magnetite.
Carvallo C, Sainctavit P, et al. (2010) Self-reversal of magnetization in oceanic Journal of Geophysical Research 112: B11103. http://dx.doi.org/
submarine basalts studied with XMCD. Geophysical Research Letters 37: L11306. 10.1029/2007JB005233.
http://dx.doi.org/10.1029/2010GL043390. Dunlop DJ (2007b) Magnetization, viscous remanent. In: Gubbins D and Herrero-
Chang L, Roberts AP, et al. (2008) Fundamental magnetic parameters from pure Bervera E (eds.) Encyclopedia of Geomagnetism and Paleomagnetism. Dordrecht,
synthetic greigite (Fe3S4). Journal of Geophysical Research 113: B06104. http://dx. The Netherlands: Springer.
doi.org/10.1029/2007JB005502. Dunlop DJ (2009) Continuous and stepwise thermal demagnetization: Are they
Chang L, Roberts AP, et al. (2009) Low-temperature magnetic properties of greigite equivalent? Geophysical Journal International 177: 949–957.
(Fe3S4). Geochemistry, Geophysics, Geosystems 10: Q01Z04. http://dx.doi.org/ Dunlop DJ (2011) Physical basis of the Thellier–Thellier and related paleointensity
10.1029/2008GC002276. methods. Physics of the Earth and Planetary Interiors 187: 118–138.
Chen AP, Egli R, and Moskowitz BM (2007) First-order reversal curve (FORC) diagrams Dunlop DJ (2012) Magnetic recording in rocks. Physics Today 65: 31–37.
of natural and cultured biogenic magnetic particles. Journal of Geophysical Dunlop DJ and Argyle KS (1997) Thermoremanence, anhysteretic remanence and
Research 112: B08S90. http://dx.doi.org/10.1029/2006JB004575. susceptibility of submicron magnetites: Nonlinear field dependence and
Church NS, Feinberg JM, and Harrison RJ (2011) Low-temperature domain wall variation with grain size. Journal of Geophysical Research
pinning in titanomagnetite: Quantitative modeling of multidomain first-order 102: 20199–20210.
reversal curve diagrams and AC susceptibility. Geochemistry, Geophysics, Dunlop DJ and Arkani-Hamed J (2005) Magnetic minerals in the Martian crust.
Geosystems 12: Q07Z27. http://dx.doi.org/10.1029/2010GC00. Journal of Geophysical Research 110: E12S04. http://dx.doi.org/
Clarke RS and Scott ERD (1980) Tetrataenite-ordered FeNi, a new mineral in meteorites. 10.1029/2005JE002404.
American Mineralogist 65: 624–630. Dunlop DJ and Carter-Stiglitz B (2006) Day plots of mixtures of superparamagnetic,
Clement BM, Haggerty S, and Harris J (2008) Magnetic inclusions in diamonds. Earth single-domain, pseudo-single-domain and multidomain magnetites. Journal of
and Planetary Science Letters 267: 333–340. Geophysical Research 111: B12S09. http://dx.doi.org/10.1029/2006JB004499.
Cottrell RD and Tarduno JA (2000) In search of high-fidelity geomagnetic Dunlop DJ and Kletetschka G (2000) Multidomain hematite: A source of planetary
paleointensities: A comparison of single plagioclase crystal and whole rock magnetic anomalies? Geophysical Research Letters 28: 3345–3348.
Thellier–Thellier analyses. Journal of Geophysical Research Dunlop DJ and Özdemir Ö (1997) Rock Magnetism: Fundamentals and Frontiers.
105: 23579–23594. New York: Cambridge University Press.
Davis KE (1981) Magnetite rods in plagioclase as the primary carrier of stable NRM in Dunlop DJ and Özdemir Ö (2000) Effect of grain size and domain state on thermal
ocean floor gabbros. Earth and Planetary Science Letters 55: 190–198. demagnetization tails. Geophysical Research Letters 27: 1311–1314.
Day R (1977) TRM and its variation with grain size. Journal of Geomagnetism and Dunlop DJ and Özdemir Ö (2001) Beyond Néel’s theories: Thermal demagnetization of
Geoelectricity 29: 233–265. narrow-band partial thermoremanent magnetizations. Physics of the Earth and
Day R, Fuller M, and Schmidt VA (1977) Hysteresis parameters of titanomagnetites: Planetary Interiors 126: 43–57.
Grain size and composition dependence. Physics of the Earth and Planetary Interiors Dunlop DJ, Özdemir Ö, Clark DA, and Schmidt PW (2000) Time-temperature
13: 260–267. relations for the remagnetization of pyrrhotite (Fe7S8) and their use in
Dekkers MJ (1988) Magnetic properties of natural pyrrhotite. I. Behaviour of initial estimating paleotemperature. Earth and Planetary Science Letters
susceptibility and saturation-magnetization-related rock-magnetic parameters in a 176: 107–116.
grain-size dependent framework. Physics of the Earth and Planetary Interiors Dunlop DJ, Özdemir Ö, and Costanzo-Alvarez V (2010) Magnetic properties of rocks of
52: 376–393. the Kapuskasing Uplift (Ontario, Canada) and origin of long-wavelength magnetic
Dekkers MJ (1989) Magnetic properties of natural pyrrhotite. II. High-and low- anomalies. Geophysical Journal International 183: 645–658.
temperature behaviour of Jrs and TRM as a function of grain size. Physics of the Dunlop DJ, Özdemir Ö, and Rancourt DG (2006) Magnetism of biotite crystals. Earth
Earth and Planetary Interiors 57: 266–283. and Planetary Science Letters 243: 805–819.
Dekkers MJ and Böhnel HN (2006) Reliable absolute palaeointensities independent of Dunlop DJ and Prévot M (1982) Magnetic properties and opaque mineralogy of drilled
magnetic domain state. Earth and Planetary Science Letters 248: 508–517. submarine intrusive rocks. Geophysical Journal of the Royal Astronomical Society
Dekkers MJ, Mattei J-L, Fillion G, and Rochette P (1989) Grain-size dependence of the 69: 763–802.
magnetic behaviour of pyrrhotite during its low-temperature transition at 34 K. Dunlop DJ, Schmidt PW, Özdemir Ö, and Clark DA (1997) Paleomagnetism and
Geophysical Research Letters 16: 855–858. paleothermometry of the Sydney Basin. 1. Thermoviscous and chemical
Diaz Ricci JC and Kirschvink JL (1992) Magnetic domain state and coercivity overprinting of the Milton Monzonite. Journal of Geophysical Research
predictions for biogenic greigite (Fe3S4): A comparison of theory with magnetosome 102: 27271–27283.
observations. Journal of Geophysical Research 97: 17309–17315. Dunlop DJ and Waddington ED (1975) The field dependence of
Dodson MH and McClelland-Brown E (1980) Magnetic blocking temperature of single thermoremanent magnetization in igneous rocks. Earth and Planetary Science
domain grains during slow cooling. Journal of Geophysical Research Letters 25: 11–25.
85: 2625–2637. Dunlop DJ and West GF (1969) An experimental evaluation of single domain theories.
Doubrovine PV and Tarduno JA (2006) Alteration and self-reversal in oceanic basalts. Reviews of Geophysics 7: 709–757.
Journal of Geophysical Research 111: B12S30. http://dx.doi.org/ Dunlop DJ and Xu S (1994) Theory of partial thermoremanent magnetization in
10.1029/2006JB004468. multidomain grains. 1. Repeated identical barriers to wall motion (single
Draeger U, Prévot M, Poidras T, and Riisager J (2006) Single-domain chemical, microcoercivity). Journal of Geophysical Research 99: 9005–9023.
thermochemical and thermal remanences in a basaltic rock. Geophysical Journal Dunlop DJ, Zhang B, and Özdemir Ö (2005) Linear and nonlinear Thellier paleointensity
International 166: 12–32. behavior of natural minerals. Journal of Geophysical Research 110: B01103. http://
Dunin-Borkowski RE, McCartney MR, Frankel RB, Bazylinski DA, Pósfai M, and dx.doi.org/10.1029/2004JB003095.
Buseck PR (1998) Magnetic microstructure of magnetotactic bacteria by electron Dzyaloshinsky I (1958) A thermodynamic theory of ‘weak’ ferromagnetism of
holography. Science 282: 1868–1870. antiferromagnetics. Journal of Physics and Chemistry of Solids 4: 241–255.
Dunlop DJ (1986) Hysteresis properties of magnetite and their dependence on particle Egli R (2003) Analysis of the field dependence of remanent magnetization curves.
size: A test of pseudo-single-domain remanence models. Journal of Geophysical Journal of Geophysical Research 108(B2): 2081. http://dx.doi.org/
Research 91: 9569–9584. 10.1029/2001JB002023.
Egli R (2006) Theoretical aspects of dipolar interactions and their appearance in first- Frandsen C, Stipp SLS, McEnroe SA, Madsen MB, and Knudsen JM (2004) Magnetic
order reversal curves of thermally activated single-domain particles. Journal of domain structures and stray fields of individual elongated magnetite grains revealed
Geophysical Research 111: B12S17. http://dx.doi.org/10.1029/2006JB004567. by magnetic force microscopy (MFM). Physics of the Earth and Planetary Interiors
Egli R, Chen AP, Winklhofer M, Kodama KP, and Horng C-S (2010) Detection of 141: 121–129.
noninteracting single-domain particles using first-order reversal curve diagrams. Frank U and Nowaczyk NR (2008) Mineral magnetic properties of artificial samples
Geochemistry, Geophysics, Geosystems 11: Q01Z11. http://dx.doi.org/ systematically mixed from haematite and magnetite. Geophysical Journal
10.1029/2009GC002916. International 175: 449–461.
Egli R and Lowrie W (2002) Anhysteretic remanent magnetization of fine magnetic Fukuma K and Dunlop DJ (2006) Three-dimensional micromagnetic modeling of
particles. Journal of Geophysical Research 107(B10): 2209. http://dx.doi.org/ randomly oriented magnetite grains (0.03–0.3 mm). Journal of Geophysical
10.1029/2001JB000671. Research 111: B12S11.
Engelmann R, Kontny A, and Lattard D (2010) Low-temperature magnetism of synthetic Gapeev AK and Tsel’movich VA (1983) Microstructure of synthetic titanomagnetite
Fe–Ti oxide assemblages. Journal of Geophysical Research 115: B12107. http://dx. oxidized at high partial pressures of oxygen (translated from Russian). Izvestiya.
doi.org/10.1029/2010JB000865. Physics of the Solid Earth 19: 983–986.
Enkin RJ, Baker J, Nourgaliev D, Iassonov P, and Hamilton TS (2007) Magnetic Garrick-Bethell I and Weiss BP (2010) Kamacite blocking temperatures and applications
hysteresis parameters and Day plot analysis to characterize diagenetic alteration in to lunar magnetism. Earth and Planetary Science Letters 294: 1–7.
gas-hydrate-bearing sediments. Journal of Geophysical Research 112: B06S90. Garrick-Bethell I, Weiss BP, Shuster DL, and Buz J (2009) Early lunar magnetism.
http://dx.doi.org/10.1029/2006JB004638. Science 323: 356–359.
Enkin RJ and Williams W (1994) Three-dimensional micromagnetic analysis of stability Geiss CE and Zanner CW (2006) How abundant is pedogenic magnetite? Abundance
in fine magnetic grains. Journal of Geophysical Research 99: 611–618. and grain size estimates for loessic soils based on rock magnetic analyses. Journal
Evans ME and Heller F (2003) Environmental Magnetism: Principles and Applications of Geophysical Research 111: B12S21. http://dx.doi.org/10.1029/2006JB004564.
of Enviromagnetics. London: Academic Press. Gendler TS, Shcherbakov VP, Dekkers MJ, Gapeev AK, Gribov SK, and McClelland E
Evans ME, Krása D, Williams W, and Winklhofer M (2006) Magnetostatic interactions in (2005) The lepidocrocite-maghemite-haematite reaction chain – I. Acquisition of
a natural magnetite-ulvöspinel system. Journal of Geophysical Research chemical remanent magnetization by maghemite, its magnetic properties and
111: B12S16. http://dx.doi.org/10.1029/2006JB004454. thermal stability. Geophysical Journal International 160: 815–832.
Evans ME and McElhinny MW (1969) An investigation of the origin of stable remanence Goguitchaichvili A, Calvo-Rathert M, and Camps P (eds.) (2009) Magnetism of volcanic
in magnetite-bearing rocks. Journal of Geomagnetism and Geoelectricity materials. Earth, Planets and Space 61(1), published a special issue containing
21: 757–773. 21 papers.
Evans ME and Wayman ML (1970) An investigation of small magnetic particles by Gong Z, Dekkers MJ, Heslop D, and Mullender TAT (2009) End-member modeling of
means of electron microscopy. Earth and Planetary Science Letters 9: 365–370. isothermal remanent magnetization (IRM) acquisition curves. Geophysical Journal
Fabian K (2000) Acquisition of thermoremanent magnetization in weak magnetic fields. International 178: 693–701.
Geophysical Journal International 142: 478–486. Grommé CS, Wright TL, and Peck DL (1969) Magnetic properties and oxidation of iron-
Fabian K (2001) A theoretical treatment of paleointensity determination experiments on titanium oxide minerals in Alae and Makaopuhi lava lakes, Hawaii. Journal of
rocks containing pseudo-single or multidomain magnetic particles. Earth and Geophysical Research 74: 5277–5293.
Planetary Science Letters 188: 45–58. Guyodo Y, Banerjee SK, Lee Penn R, et al. (2006) Magnetic properties of synthetic six-
Fabian K (2003) Some additional parameters to estimate domain state from isothermal line ferrihydrite nanoparticles. Physics of the Earth and Planetary Interiors
magnetization measurements. Earth and Planetary Science Letters 213: 337–345. 154: 222–233.
Fabian K (2009) Thermochemical remanence acquisition in single-domain particle Haag M and Allenspach R (1993) A novel approach to domain imaging in natural Fe/Ti
ensembles: A case for possible overestimation of the geomagnetic paleointensity. oxides by spin-polarized scanning electron microscopy. Geophysical Research
Geochemistry, Geophysics, Geosystems 10: Q06Z03. http://dx.doi.org/ Letters 20: 1943–1946.
10.1029/2009GC002420. Halgedahl SL (1991) Magnetic domain patterns observed on synthetic Ti-rich
Fabian K and Hubert A (1999) Shape-induced pseudo-single-domain remanence. titanomagnetites as a function of temperature and in states of thermoremanent
Geophysical Journal International 138: 717–726. magnetization. Journal of Geophysical Research 96: 3943–3972.
Fabian K, Kirchner A, Williams W, Heider F, Leibl T, and Hubert A (1996) Three- Halgedahl SL, Day R, and Fuller M (1980) The effect of cooling rate on the intensity of
dimensional micromagnetic calculations for magnetite using FFT. Geophysical weak-field TRM in single domain magnetite. Journal of Geophysical Research
Journal International 124: 89–104. 85: 3690–3698.
Fabian K and Leonhardt R (2010) Multiple-specimen absolute paleointensity Halgedahl SL and Fuller M (1983) The dependence of magnetic domain structure on
determination: An optimal protocol including pTRM normalization, domain-state magnetization state with emphasis on nucleation as a mechanism for pseudo-
correction, and alteration test. Earth and Planetary Science Letters 297: 84–94. single-domain behavior. Journal of Geophysical Research 88: 6505–6522.
Fabian K, Miyajima N, et al. (2011) Chemical and magnetic properties of rapidly cooled Harrison RJ, Becker U, and Redfern SAT (2000) Thermodynamics of the R3 to R3c
metastable ferri-ilmenite solid solutions: Implications for magnetic self reversal and phase transition in the ilmenite-hematite solid solution. American Mineralogist
exchange bias – I. Fe–Ti order transition in quenched synthetic ilmenite 61. 85: 1694–1705.
Geophysical Journal International 186: 997–1014. Harrison RJ, Dunin-Borkowski RE, and Putnis A (2002) Direct imaging of nanoscale
Fabian K and von Dobeneck T (1997) Isothermal magnetization of samples with stable magnetic interactions in minerals. Proceedings of the National Academy of Sciences
Preisach function: A survey of hysteresis, remanence, and rock magnetic of the United States of America 99: 16556–16561.
parameters. Journal of Geophysical Research 102: 17659–17677. Hashimoto T, Hurst T, et al. (2008) The role of thermal viscous remanent magnetization
Feinberg JM, Harrison RJ, et al. (2006) Effects of internal mineral structures on the (TVRM) in magnetic changes associated with volcanic eruptions. Journal of
magnetic remanence of silicate-hosted titanomagnetite inclusions: An electron Volcanology and Geothermal Research 176: 610–616.
holography study. Journal of Geophysical Research 111: B12S15. http://dx.doi.org/ Heider F and Hoffmann V (1992) Magneto-optical Kerr effect on magnetite crystals with
10.1029/2006JB004498. externally applied magnetic fields. Earth and Planetary Science Letters
Feinberg JM, Scott GR, Renne PR, and Wenk H-R (2005) Exsolved magnetite inclusions 108: 131–138.
in silicates: Features determining their remanence behavior. Geology 33: 513–516. Heider F, Zitzelsberger A, and Fabian K (1996) Magnetic susceptibility and remanent
Ferk A, von Aulock FW, Leonhardt R, Hess KU, and Dingwell DB (2010) A cooling rate coercive force in grown magnetite crystals from 0.1 mm to 6 mm. Physics of the
bias in paleointensity determination from volcanic glass: An experimental Earth and Planetary Interiors 93: 239–256.
demonstration. Journal of Geophysical Research 115: B08102. http://dx.doi.org/ Heslop D, Dekkers MJ, Kruiver PP, and van Oorschot IHM (2002) Analysis of
10.1029/2009JB006964. isothermal remanent magnetisation acquisition curves using the expectation-
Ferré EC, Friedman SA, et al. (2013) The magnetism of mantle xenoliths and potential maximisation algorithm. Geophysical Journal International 148: 58–64.
implications for sub-Moho magnetic sources. Geophysical Research Letters Heslop D, McIntosh G, and Dekkers MJ (2004) Using time-and temperature-dependent
40: 105–110. http://dx.doi.org/10.1029/2012GL054100. Preisach models to investigate the limitations of modeling isothermal remanent
Foss S, Moskowitz BM, Proksch R, and Dahlberg ED (1998) Domain wall structures in magnetisation acquisition curves with cumulative log Gaussian functions.
single-crystal magnetite investigated by magnetic force microscopy. Journal of Geophysical Journal International 157: 55–63.
Geophysical Research 103: 30551–30560. Heslop D and Roberts A (2012a) Estimating best fit binary mixing lines in the Day plot.
Fox JMW and Aitken MJ (1980) Cooling-rate dependence of thermoremanent Journal of Geophysical Research 117: B01101. http://dx.doi.org/
magnetisation. Nature 283: 462–463. 10.1029/2011JB008787.
Heslop D and Roberts A (2012b) A method for unmixing magnetic hysteresis loops. Kodama KP and Sun WW (1992) Magnetic anisotropy as a correction for compaction-
Journal of Geophysical Research 117: B03103. http://dx.doi.org/ caused palaeomagnetic inclination shallowing. Geophysical Journal International
10.1029/2011JB008859. 111: 465–469.
Heslop D, Witt A, Kleiner T, and Fabian K (2006) The role of magnetostatic Kopp RE, Weiss BP, et al. (2006) Chains, clumps, and strings: Magnetofossil
interactions in sediment suspensions. Geophysical Journal International taphonomy with ferromagnetic resonance spectroscopy. Earth and Planetary
165: 775–785. Science Letters 247: 10–25.
Heslop D, Witt A, von Dobeneck T, Huhn K, Fabian K, and Bleil U (2004) Discrete Krása D, Muxworthy A, and Williams W (2011) Room-and low-temperature magnetic
element modeling of postdepositional remanent magnetization acquisition: First properties of 2-D magnetite particle arrays. Geophysical Journal International
results. EOS. Transactions of the American Geophysical Union 85(47), Fall Meeting 185: 167–180.
suppl., Abstract GP21C–08 (available online). Lagroix F, Banerjee SK, and Moskowitz BM (2004) Revisiting the mechanism of self
Hirt AM, Brem F, Hanzlik M, and Faivre D (2006) Anomalous magnetic properties of reversed thermoremanent magnetization based on observations from synthetic
brain tissue at low temperature: The 50 K anomaly. Journal of Geophysical Research ferrian ilmenite (y ¼ 0.7). Journal of Geophysical Research 109: B12108.
111: B12S06. http://dx.doi.org/10.1029/2006JB004570. Lanci L and Kent DV (2003) Introduction of thermal activation in forward modeling of
Hodych JP and Bijaksana S (1993) Can remanence anisotropy detect paleomagnetic hysteresis loops for single-domain magnetic particles and implications for the
inclination shallowing due to compaction? A case study using Cretaceous deep-sea interpretation of the Day diagram. Journal of Geophysical Research 108(B3): 2142.
limestones. Journal of Geophysical Research 98: 22429–22441. http://dx.doi.org/10.1029/2001JB000944.
Horng C-S and Roberts AP (2006) Authigenic or detrital origin of pyrrhotite in Lappe S, Church N, et al. (2011) Mineral magnetism of dusty olivine: A credible
sediments? Resolving a paleomagnetic conundrum. Earth and Planetary Science recorder of pre-accretionary remanence. Geochemistry, Geophysics, Geosystems
Letters 241: 750–762. 12: Q12Z35. http://dx.doi.org/10.1029/2011GC003811.
Housen BA, Banerjee SK, and Moskowitz BM (1996) Low-temperature properties of Lattard D, Engelmann R, Kontny A, and Sauerzapf U (2006) Curie temperatures of
siderite and magnetite in marine sediments. Geophysical Research Letters synthetic titanomagnetites in the Fe–Ti–O system: Effects of composition, crystal
23: 2843–2846. chemistry, and thermomagnetic methods. Journal of Geophysical Research
Hunt CP, Moskowitz BM, and Banerjee SK (1995) Magnetic properties of rocks and 111: B12S28. http://dx.doi.org/10.1029/2006JB004591.
minerals. In: Ahrens TJ (ed.) Rock Physics and Phase Relations: A Handbook of Lawrence K, Johnson C, Tauxe L, and Gee J (2008) Lunar paleointensity measurements:
Physical Constants, vol. 3, pp. 189–204. Washington, DC: American Geophysical Implications for lunar magnetic evolution. Physics of the Earth and Planetary
Union, chapter 14. Interiors 168: 71–87.
Hus JJ (1990) The magnetic properties of siderite concretions and the CRM of their Leonhardt R (2006) Analyzing rock magnetic measurements: The RockMagAnalyzer 1.0
oxidation products. Physics of the Earth and Planetary Interiors 63: 41–57. software. Computers and Geosciences 32: 1420–1431.
Jackson MJ, Carter-Stiglitz B, Egli R, and Solheid P (2006) Characterizing the Leonhardt R, Krása D, and Coe RS (2004) Multidomain behavior during Thellier
superparamagnetic grain distribution f(V, Hk) by thermal fluctuation tomography. paleointensity experiments: A phenomenological model. Physics of the Earth and
Journal of Geophysical Research 111: B12S07. http://dx.doi.org/ Planetary Interiors 147: 127–140.
10.1029/2006JB004514. Leonhardt R, Matzka J, Nichols ARL, and Dingwell DB (2006) Cooling rate correction of
Jackson MJ and Solheid P (2009) On the quantitative analysis and evaluation of paleointensity determination for volcanic glasses by relaxation geospeedometry.
magnetic hysteresis data. Geochemistry, Geophysics, Geosystems 11: Q04Z15. Earth and Planetary Science Letters 243: 282–292.
http://dx.doi.org/10.1029/2009GC002932. Lorand J-P, Chevrier V, and Sautter V (2005) Sulphide mineralogy and redox
Jackson M, Slheid P, Carter-Stiglitz B, Rosenbaum J, and Till J (2004) Tiva Canyon conditions in some shergottites. Meteoritics and Planetary Science 40: 1257–1272.
Tuff: Superparamagnetic samples available. Institute of Rock Magnetism Quarterly Lvlie R (1976) The intensity pattern of post-depositional remanence acquired in some
(University of Minnesota) 14(3): 1 and 9–11. marine sediments deposited during a reversal of the external magnetic field. Earth
Jackson MJ, Williams W, and Smirnov A (eds.) (2006) Fundamental and Frontier and Planetary Science Letters 30: 209–214.
Research in Rock Magnetism, a special issue containing 32 papers published in Maher BA, Karloukovski VV, and Mutch TJ (2004) High-field remanence
Journal of Geophysical Research 111: B12. properties of synthetic and natural submicrometre haematites and goethites:
Johnson EA, Murphy T, and Torreson OW (1948) Pre-history of the Earth’s magnetic Significance for environmental contexts. Earth and Planetary Science Letters
field. Terrestrial Magnetism and Atmospheric Electricity 53: 349–372. 226: 491–505.
Kasama T, Church NS, Feinberg JM, Dunin-Borkowski RE, and Harrison RJ (2010) McEnroe SA, Langenhorst F, Robinson P, Bromiley GD, and Shaw CSJ (2004) What is
Direct observation of ferrimagnetic/ferroelastic domain interactions in magnetite magnetic in the lower crust? Earth and Planetary Science Letters 226: 175–192.
below the Verwey transition. Earth and Planetary Science Letters 297: 10–17. McEnroe SA, Robinson P, et al. (2007) Magnetization of exsolution intergrowths of
Kasama T, Dunin-Borkowski RE, et al. (2009) The application of Lorentz transmission hematite and ilmenite: Mineral chemistry, phase relations, and magnetic properties
microscopy to the study of lamellar magnetism in hematite–ilmenite. American of hemo-ilmenite ores with micron-to nanometer-scale lamellae from Allard Lake,
Mineralogist 94: 262–269. Quebec. Journal of Geophysical Research 112: B10103. http://dx.doi.org/
Kasama T, Posfai M, et al. (2006) Magnetic properties, microstructure, composition, 10.1029/2007JB004973.
and morphology of greigite nanocrystals in magnetotactic bacteria from electron Menyeh A and O’Reilly W (1998) Thermoremanence in monoclinic pyrrhotite particles
holography and tomography. American Mineralogist 91: 1216–1229. containing few domains. Geophysical Research Letters 25: 3461–3464.
Kelso PR and Banerjee SK (1994) Elevated temperature viscous remanent magnetization Milne GA and Dunlop DJ (2006) Angular variation of the magnetic properties and
of natural and synthetic multidomain magnetite. Earth and Planetary Science Letters reversal mode of aligned single-domain iron nanoparticles. Journal of Geophysical
122: 43–56. Research 111: B12S08. http://dx.doi.org/10.1029/2006JB004530.
Kent DV (1973) Post-depositional remanent magnetization in deep-sea sediment. Mitra R and Tauxe L (2009) Full vector model for magnetization in sediments. Earth and
Nature 246: 32–34. Planetary Science Letters 286: 535–545.
Kirschvink JL, Jones DS, and MacFadden BJ (eds.) (1985) Magnetite Biomineralization Mitra R, Tauxe L, and Gee J (2011) Detecting uniaxial single domain grains with a
and Magnetoreception in Organisms: A New Biomagnetism. New York: Plenum. modified IRM technique. Geophysical Journal International 187: 1250–1258.
Kletetschka G, Acuña MH, Kohout T, Wasilewski PJ, and Connerney JEP (2004) An Morales JA, Goguitchaichvili A, et al. (2011) Are ceramics and bricks reliable absolute
empirical scaling law for acquisition of thermoremanent magnetization. Earth and geomagnetic intensity carriers? Physics of the Earth and Planetary Interiors
Planetary Science Letters 226: 521–528. 187: 310–321.
Kletetschka G, Wasilewski PJ, and Taylor PT (2000a) Unique thermoremanent Moskowitz BM, Feinberg JM, Florindo F, and Roberts AP (eds.) (2008–10) Magnetism
magnetization of multidomain hematite: Implications for magnetic anomalies. Earth from atomic to planetary scales: Physical principles and interdisciplinary
and Planetary Science Letters 176: 469–479. applications in geoscience, a theme collection of 18 papers published in
Kletetschka G, Wasilewski PJ, and Taylor PT (2000b) Hematite vs. magnetite as the Geochemistry Geophysics Geosystems.
signature for planetary magnetic anomalies. Physics of the Earth and Planetary Moskowitz BM, Halgedahl SL, and Lawson CA (1988) Magnetic domains on unpolished
Interiors 119: 259–267. and polished surfaces of titanium-rich titanomagnetite. Journal of Geophysical
Kneller EF and Luborsky FE (1963) Particle size dependence of coercivity and Research 93: 3372–3386.
remanence of single-domain particles. Journal of Applied Physics 34: 656–658. Moskowitz BM, Jackson M, and Kissel C (1998) Low-temperature behavior of
Kodama KP (1992) Depositional remanent magnetization. In: Nierenberg WA (ed.) titanomagnetites. Earth and Planetary Science Letters 157: 141–149.
Encyclopedia of Earth System Science, vol. 2, pp. 47–54. San Diego, CA: Academic Murthy GS, Evans ME, and Gough DI (1971) Evidence of single-domain magnetite in
Press. the Michikamau anorthosite. Canadian Journal of Earth Sciences 8: 361–370.
Muxworthy AR (2010) Revisiting a domain-state independent method of palaeointensity Özdemir Ö and Dunlop DJ (2003) Low-temperature behavior and memory of iron-rich
determination. Physics of the Earth and Planetary Interiors 179: 21–31. titanomagnetites (Mt. Haruna, Japan and Mt. Pinatubo, The Philippines). Earth and
Muxworthy AR and Dunlop DJ (2002) First-order reversal curve (FORC) diagrams for Planetary Science Letters 216: 193–200.
pseudo-single-domain magnetites at high temperature. Earth and Planetary Science Özdemir Ö and Dunlop DJ (2005) Thermoremanent magnetization of multidomain
Letters 203: 369–382. hematite. Journal of Geophysical Research 110: B09104. http://dx.doi.org/
Muxworthy A, Heslop D, and Michalk DM (2009) Thermal fluctuation fields in basalts. 10.1029/2005JB003820.
Earth, Planets and Space 61: 111–117. Özdemir Ö and Dunlop DJ (2006a) Magnetic memory and coupling between spin-
Muxworthy A, Heslop D, Paterson G, and Michalk D (2011) A Preisach method for canted and defect magnetism in hematite. Journal of Geophysical Research
estimating absolute paleofield intensity under the constraint of using only 111: B12S03. http://dx.doi.org/10.1029/2006JB004555.
isothermal measurements: 2. Experimental testing. Journal of Geophysical Research Özdemir Ö and Dunlop DJ (2006b) Magnetic domain observations on magnetite
116: B04103. http://dx.doi.org/10.1029/2010JB007844. crystals in biotite and hornblende grains. Journal of Geophysical Research
Muxworthy A, Heslop D, and Williams W (2004) Influence of magnetostatic interactions 111: B06103. http://dx.doi.org/10.1029/2005JB004090.
on first-order-reversal-curve (FORC) diagrams: A micromagnetic approach. Özdemir Ö and Dunlop DJ (2010) Hallmarks of maghemitization in low
Geophysical Journal International 158: 888–897. temperature remanence cycling of partially oxidized magnetite nanoparticles.
Muxworthy AR, King JG, and Heslop D (2005) Assessing the ability of first-order Journal of Geophysical Research 115: B02101. http://dx.doi.org/
reversal curve (FORC) diagrams to unravel complex magnetic signals. Journal of 10.1029/2009JB006756.
Geophysical Research 110: B01105. http://dx.doi.org/10.1029/2004JB003195. Özdemir Ö, Dunlop DJ, and Berquó TS (2008) Morin transition in hematite: Size
Muxworthy AR, King JG, and Odling N (2006) Magnetic hysteresis properties of dependence and thermal hysteresis. Geochemistry, Geophysics, Geosystems
interacting and noninteracting micron-sized magnetite produced by electron beam 9: Q10Z01. http://dx.doi.org/10.1029/2008GC002110.
lithography. Geochemistry, Geophysics, Geosystems 7: Q07009. http://dx.doi.org/ Özdemir Ö, Dunlop DJ, and Jackson M (2009) Frequency and field dependent
10.1029/2006GC001309. susceptibility of magnetite at low temperature. Earth, Planets and Space
Muxworthy AR and Williams W (2006) Critical single-domain/multidomain grain sizes 61: 125–131.
in noninteracting and interacting elongated magnetite particles: Implications for Özdemir Ö, Dunlop DJ, and Moskowitz BM (1993) The effect of oxidation on the Verwey
magnetosomes. Journal of Geophysical Research 111: B12S12. http://dx.doi.org/ transition in magnetite. Geophysical Research Letters 20: 1671–1674.
10.1029/2006JB004588. Özdemir Ö, Dunlop DJ, and Moskowitz BM (2002) Changes in remanence, coercivity
Nagata T (1979) Meteorite magnetism and the early solar system magnetic field. and domain state at low temperature in magnetite. Earth and Planetary Science
Physics of the Earth and Planetary Interiors 20: 324–341. Letters 194: 343–358.
Nagata T (1983) Meteorite magnetization and paleointensity. Advances in Space Özdemir Ö and O’Reilly W (1982a) Magnetic hysteresis properties of synthetic
Research 2: 55–63. monodomain titanomaghemites. Earth and Planetary Science Letters 57: 437–447.
Nagata T, Arai Y, and Momose K (1963) Secular variation of the geomagnetic total force Özdemir Ö and O’Reilly W (1982b) An experimental study of the intensity and stability of
during the last 5000 years. Journal of Geophysical Research 68: 5277–5281. thermoremanent magnetization acquired by synthetic monodomain titanomagnetite
Néel L (1948) Propriétés magnetiqués des ferrites: Ferrimagetisme et substituted by aluminium. Geophysics Journal of the Royal Astronomical Society
antiferromagnétisme. Annales de Physique 3: 137–198. 70: 141–154.
Néel L (1949) Théorie du traı̂nage magnetique des ferromagnetiques en grains fins avec Özdemir Ö, Xu S, and Dunlop DJ (1995) Closure domains in magnetite. Journal of
applications aux terres cuites. Annales de Geophysique 5: 99–136. Geophysical Research 100: 2193–2209.
Néel L (1954) Remarques sur la théorie des propriétés magnétiques des substances Ozima M and Funaki M (2003) Hemoilmenite as a carrier of SRTRM in dacitic pumice
dures. Applied Scientific Research B 4: 13–24. from Akagi, Ontake and Sambe volcanoes, Japan. Earth and Planetary Science
Néel L (1955) Some theoretical aspects of rock magnetism. Advances in Physics Letters 213: 311–320.
4: 191–243. Pan Q, Pokhil TG, and Moskowitz BM (2002) Domain structures in epitaxial Fe3O4
Néel L, Pauleve J, Pauthenet R, Laugier J, and Dautreppe D (1964) Magnetic properties particles studies by magnetic force microscopy. Journal of Applied Physics
of an iron-nickel single crystal ordered by neutron bombardment. Journal of 91: 5945–5950.
Applied Physics 35: 873–876. Pannalal SJ, Crowe SA, Cioppa MT, Symons DTA, Sturm A, and Fowle DA (2005)
Newell AJ (2009) Transition to superparamagnetism in chains of magnetosome crystals. Room-temperature properties of ferrihydrite: A potential magnetic remanence
Geochemistry, Geophysics, Geosystems 10: 1108. http://dx.doi.org/ carrier? Earth and Planetary Science Letters 236: 856–870.
10.1029/2009GC002538. Paterson G (2011) A simple test for the presence of multidomain behavior during
Newell AJ and Merrill RT (1999) Single-domain critical sizes for coercivity and paleointensity experiments. Journal of Geophysical Research 116: B10104. http://
remanence. Journal of Geophysical Research 104: 617–628. dx.doi.org/10.1029/2011JB008369.
O’Donovan JB, Facey D, and O’Reilly W (1986) The magnetization process in Petersen N, von Dobeneck T, and Vali H (1986) Fossil bacterial magnetite in deep-sea
titanomagnetite (Fe2.4Ti0.6O4) in the 1–30 mm particle size range. Geophysical sediments from the South Atlantic Ocean. Nature 320: 611–615.
Journal of the Royal Astronomical Society 87: 897–916. Pokhil TG and Moskowitz BM (1997) Magnetic domains and domain walls in pseudo-
O’Reilly W (1976) Magnetic minerals in the crust of the Earth. Reports on Progress in single-domain magnetite studied with magnetic force microscopy. Journal of
Physics 39: 857–908. Geophysical Research 102: 22681–22694.
O’Reilly W (1984) Rock and Mineral Magnetism. New York: Blackie, Glasgow and Pozzi J-P and Dubuisson G (1992) High temperature viscous magnetization of oceanic
Chapman & Hall. deep crustal and mantle-rocks as a partial source for Magsat magnetic anomalies.
Özdemir Ö (2007) Magnetization, thermoremanent. In: Gubbins D and Herrero- Geophysical Research Letters 19: 21–24.
Bervera E (eds.) Encyclopedia of Geomagnetism and Paleomagnetism. Dordrecht, Preisach F (1935) Über die magnetische Nachwirkung. Zeitschrift für Physik
The Netherlands: Kluwer. 94: 277–302.
Özdemir Ö and Dunlop DJ (1985) An experimental study of chemical remanent Prévot M, Hoffman KA, Goguitchaicvili A, Doukhan J-C, Shcherbakov V, and Bina M
magnetizations of synthetic monodomain titanomaghemites with initial (2001) The mechanism of self-reversal of thermoremanence in natural hemoilmenite
thermoremanent magnetizations. Journal of Geophysical Research crystals: New experimental data and model. Physics of the Earth and Planetary
90: 11513–11523. Interiors 126: 75–92.
Özdemir Ö and Dunlop DJ (1993a) Magnetic domain structures on a natural single Pullaiah G, Irving E, Buchan KL, and Dunlop DJ (1975) Magnetization changes caused
crystal of magnetite. Geophysical Research Letters 20: 1835–1838. by burial and uplift. Earth and Planetary Science Letters 28: 133–143.
Özdemir Ö and Dunlop DJ (1993b) Chemical remanent magnetization during gFeOOH Qingsong Liu, et al. (2008) Magnetic study of mafic granulite xenoliths from the
phase transformations. Journal of Geophysical Research 98: 4191–4198. Hannuoba basalt, north China. Geochemistry, Geophysics, Geosystems 9: Q06008.
Özdemir Ö and Dunlop DJ (1996) Thermoremanence and Néel temperature of goethite. http://dx.doi.org/10.1029/2008GC001952.
Geophysical Research Letters 23: 921–924. Rave W, Fabian K, and Hubert A (1998) Magnetic states of small cubic particles
Özdemir Ö and Dunlop DJ (1997) Effect of crystal defects and internal stress on the with uniaxial anisotropy. Journal of Magnetism and Magnetic Materials
domain structure and magnetic properties of magnetite. Journal of Geophysical 190: 332–348.
Research 102: 20211–20224. Renne PR, Scott GR, Glen JMG, and Feinberg JM (2002) Oriented inclusions of
Özdemir Ö and Dunlop DJ (2002) Thermoremanence and stable memory of single- magnetite in clinopyroxene: Source of stable remanent magnetization in gabbros of
domain hematites. Geophysical Research Letters 29(18): 1877. http://dx.doi.org/ the Messum Complex Namibia. Geochemistry, Geophysics, Geosystems 3(12):
10.1029/2002GL015597. 1079. http://dx.doi.org/10.1029/2002GC000319.
Rivas-Sanchez ML, Alva-Valdivia LM, et al. (2009) Natural magnetite nanoparticles of northern Sardinia (Italy). Journal of Geophysical Research 111: B12S26. http://
from an iron-ore deposit: Size dependence of magnetic properties. Earth, Planets dx.doi.org/10.1029/2006JB004574.
and Space 61: 151–160. Sun WW and Kodama KP (1992) Magnetic anisotropy, scanning electron
Roberts AP (1995) Magnetic properties of sedimentary greigite (Fe3S4). Earth and microscopy, and X-ray pole figure goniometry study of inclination shallowing
Planetary Science Letters 134: 227–236. in a compacting clay-rich sediment. Journal of Geophysical Research
Roberts AP, Chang L, Rowan C, Horng C-S, and Florindo F (2011) Magnetic properties 97: 19599–19615.
of sedimentary greigite (Fe3S4): An update. Reviews of Geophysics 49: RG1002. Swanson-Hysell NL, Feinberg J, Berquó T, and Maloof A (2011) Self-reversed
http://dx.doi.org/10.1029/2010RG000336. magnetization held by martite in basalt flows from the 1.1-billion-year-old
Roberts AP, Liu QS, et al. (2006) Characterization of hematite (a-Fe2O3), goethite Keweenawan rift, Canada. Earth and Planetary Science Letters 305: 171–184.
(a-FeOOH), greigite (Fe3S4), and pyrrhotite (Fe7S8) using first-order reversal curve Tarduno JA (2009) Geodynamo history preserved in single silicate crystals: Origins and
diagrams. Journal of Geophysical Research 111: B12S35. http://dx.doi.org/ long-term mantle control. Elements 5: 217–222.
10.1029/2006JB004715. Tarduno JA, Cottrell RD, and Smirnov A (2006) The paleomagnetism of single silicate
Roberts AP, Pike CP, and Verosub KL (2000) First-order reversal curve diagrams: crystals: Recording geomagnetic field strength during mixed polarity intervals,
A new tool for characterizing the magnetic properties of natural samples. Journal of superchrons, and inner core growth. Reviews of Geophysics 44: RG1002. http://dx.
Geophysical Research 105: 28461–28475. doi.org/10.1029/2005RG000189.
Roberts AP and Turner GM (1993) Diagenetic formation of ferrimagnetic iron sulphide Tarduno JA, Cottrell RD, Watkeys MK, and Bauch D (2006) Geomagnetic field
minerals in rapidly deposited marine sediments, South Island, New Zealand. Earth strength 3.2 billion years ago recorded by single silicate crystals. Nature
and Planetary Science Letters 115: 257–273. 446: 657–660.
Robinson P, Harrison RJ, McEnroe SA, and Hargraves RB (2002) Lamellar magnetism Tauxe L (1998) Paleomagnetic Principles and Practice. Dordrecht: Kluwer.
in the haematite–ilmenite series as an explanation for strong remanent Tauxe L (2006) Long-term trends in paleointensity: The contribution of DSDP/ODP
magnetization. Nature 418: 517–520. submarine basaltic glass collections. Physics of the Earth and Planetary Interiors
Robinson P, Heidelbach F, Hirt AM, McEnroe SA, and Brown LL (2006) 156: 223–241.
Crystallographic-magnetic correlations in single-crystal haemo-ilmenite: New Tauxe L and Kent DV (1984) Properties of a detrital remanence carried by
evidence for lamellar magnetism. Geophysical Journal International 165: 17–31. haematite from study of modern river deposits and laboratory redeposition
Rochette P, Fillion G, Mattéi J-L, and Dekkers MJ (1990) Magnetic transition at experiments. Geophysical Journal of the Royal Astronomical Society
30–34 kelvin in pyrrhotite: Insight into a widespread occurrence of this mineral in 76: 543–561.
rocks. Earth and Planetary Science Letters 98: 319–328. Tauxe L, Pick T, and Kok YS (1995) Relative paleointensity in sediments: A pseudo-
Rochette P, Gattacceca J, Chevrier V, Mathé PE, and Menvielle M (2006) Magnetism, Thellier approach. Geophysical Research Letters 22: 2885–2888.
iron minerals, and life on Mars. Astrobiology 6: 423–436. Tauxe L, Steindorf JL, and Harris A (2006) Depositional remanent magnetization:
Rochette P, Gattacceca J, Ivanov AV, Nazarov MA, and Bezaeva NS (2010) Magnetic Toward an improved theoretical and experimental foundation. Earth and Planetary
properties of lunar materials: Meteorites, Luna and Apollo returned samples. Earth Science Letters 244: 515–529.
and Planetary Science Letters 292: 383–391. Thellier E (1938) Sur l’aimantation des terres cuites et ses applications
Rochette P, Gattacceca J, et al. (2005) Matching Martian crustal magnetization and géophysiques. Annales de l’Institut de Physique du Globe de l’Université de
magnetic properties of Martian meteorites. Meteoritics and Planetary Science Paris 16: 157–302.
40: 529–540. Thomas-Keprta KL, Bazylinski DA, et al. (2000) Elongated prismatic magnetite crystals
Rochette P, Mathé P-E, et al. (2005) Non-saturation of the defect moment of goethite in ALH84001 carbonate globules. Geochimica et Cosmochimica Acta
and fine-grained hematite up to 57 teslas. Geophysical Research Letters 32: L0. 64: 4049–4081.
http://dx.doi.org/10.1029/2005GL024196. Thomas-Keprta KL, Clemett SJ, McKay DS, Gibson EK, and Wentworth SJ (2009)
Rowan CJ and Roberts AP (2006) Magnetite dissolution, diachronous greigite Origins of magnetite nanocrystals in Martian meteorite ALH84001. Geochimica et
formation, and secondary magnetizations from pyrite oxidation: Unravelling Cosmochimica Acta 73: 6631–6677.
complex magnetizations in Neogene marine sediments from New Zealand. Earth and Tucker P (1980) A grain mobility model of post-depositional realignment. Geophysical
Planetary Science Letters 241: 119–137. Journal of the Royal Astronomical Society 63: 149–163.
Schwarz EJ and Vaughan DJ (1972) Magnetic phase relations of pyrrhotite. Journal of Tucker P and O’Reilly W (1980) The acquisition of thermoremanent magnetization by
Geomagnetism and Geoelectricity 24: 441–458. multidomain single-crystal titanomagnetite. Geophysical Journal of the Royal
Scott ERD and Fuller M (2004) A possible source for the Martian crustal magnetic field. Astronomical Society 60: 21–36.
Earth and Planetary Science Letters 220: 83–90. Uehara M, Gattacceca J, Leroux H, Jacob D, and van der Beek CJ (2011) Magnetic
Selkin PA, Gee JS, and Tauxe L (2007) Nonlinear thermoremanence acquisition and microstructures of metal grains in equilibrated ordinary chondrites and
implications for paleointensity data. Earth and Planetary Science Letters implications for paleomagnetism of meteorites. Earth and Planetary Science
256: 81–89. Letters 306: 241–252.
Selkin PA, Gee JS, Tauxe L, Meurer WP, and Newell AJ (2000) The effect of remanence Uehara M, van der Beek CJ, Gattacceca J, Skidanov VA, and Quesnel Y (2010) Advances
anisotropy on paleointensity estimates: A case study from the Archean Stillwater in magneto-optical imaging applied to rock magnetism and paleomagnetism.
Complex. Earth and Planetary Science Letters 183: 403–416. Geochemistry, Geophysics, Geosystems 11: Q05Y09. http://dx.doi.org/10.1029/
Shaar R, Ron H, Tauxe L, Kessel R, and Agnon A (2011) Paleomagnetic field GC002653.
intensity derived from non-SD: Testing the Thellier IZZI technique on MD Uyeda S (1958) Thermo-remanent magnetism as a medium of paleomagnetism, with
slag and a new bootstrap procedure. Earth and Planetary Science Letters special reference to reverse thermoremanent magnetism. Japanese Journal of
310: 213–224. Geophysics 2: 1–123.
Shcherbakov VP and Sycheva N (2010) On the mechanism of formation of depositional Vali H, Förster O, Amarantidis G, and Petersen N (1987) Magnetotactic bacteria and
remanent magnetization. Geochemistry, Geophysics, Geosystems 11: 213. http://dx. their magnetofossils in sediments. Earth and Planetary Science Letters
doi.org/10.1029/2009GC002830. 86: 389–400.
Shcherbakov VP and Zhidkov GV (2006) Multivectorial paleointensity determination by Verosub K and Roberts AP (1995) Environmental magnetism: Past, present, and future.
the Thellier method. Journal of Geophysical Research 111: B12S32. http://dx.doi. Journal of Geophysical Research 100: 2175–2192.
org/10.1029/2006JB004504. Wasilewski P (1981) Magnetization of small iron-nickel spheres. Physics of the Earth
Soffel HC and Appel E (1982) Domain structure of small synthetic titanomagnetite and Planetary Interiors 26: 355–377.
particles and experiments with IRM and TRM. Physics of the Earth and Planetary Weiss P (1907) L’hypothèse du champ moleculaire et la propriété ferromagnétique.
Interiors 30: 348–355. Journal de Physique 6: 661–690.
Stacey FD (1958) Thermoremanent magnetization (TRM) of multidomain grains in Weiss BP, Vali H, Baudenbacher FJ, Kirschvink JL, Stewart DL, and Shuster DL (2002)
igneous rocks. Philosophical Magazine 3: 1391–1401. Records of an ancient Martian magnetic field in ALH84001. Earth and Planetary
Stacey FD and Banerjee SK (1974) The Physical Principles of Rock Magnetism. Science Letters 201: 449–463.
Amsterdam: Elsevier. Wenk H, Chen K, and Smith R (2011) Morphology and microstructure of magnetite and
Stoner EC and Wohlfarth EP (1948) A mechanism of magnetic hysteresis in heterogeneous ilmenite inclusions in plagioclase from Adirondack anorthositic gneiss. American
alloys. Philosophical Transactions of the Royal Society of London A 240: 599–642. Mineralogist 96: 1316–1324.
Strada E, Talarico FM, and Florindo F (2006) Magnetic petrology of variably Williams W, Enkin RJ, and Milne G (1992) Magnetic domain wall visibility in Bitter
retrogressed eclogites and amphibolites: A case study from the Hercynian basement pattern imaging. Journal of Geophysical Research 97: 17443–17448.
Williams W, Evans ME, and Krása D (2010) Micromagnetics of paleomagnetically Yu Y, Dunlop DJ, and Özdemir Ö (2002b) Partial anhysteretic remanent
significant mineral grains with complex morphology. Geochemistry, Geophysics, magnetization in magnetite. 2. Reciprocity. Journal of Geophysical Research
Geosystems 11: Q02Z14. http://dx.doi.org/10.1029/2009GC002828. 107(B10): 2245.
Williams W, Muxworthy A, and Evans ME (2011) A micromagnetic investigation of Yu Y, Dunlop DJ, and Özdemir Ö (2003) Testing the independence law of partial ARMs:
magnetite grains in the form of platonic polyhedra with surface roughness. Implications for paleointensity determination. Earth and Planetary Science Letters
Geochemistry, Geophysics, Geosystems 12: Q10Z31. http://dx.doi.org/ 208: 27–39.
10.1029/2011GC003560. Yu Y, Dunlop DJ, Özdemir Ö, and Ueno H (2001) Magnetic properties of Kurokami
Wilson RL (1961) Palaeomagnetism in Northern Ireland. 1. The thermal pumices from Mt. Sakurajima, Japan. Earth and Planetary Science Letters
demagnetization of natural magnetic moments in rocks. Geophysical Journal of the 192: 439–446.
Royal Astronomical Society 5: 45–58. Yu Y and Gee JS (2005) Spinel in Martian meteorite SaU 008: Implications for Martian
Witt A, Heslop D, von Dobeneck T, Huhn K, Fabian K, and Bleil U (2004) Discrete magnetism. Earth and Planetary Science Letters 232: 287–294.
element modelling of post-depositional remanent magnetization acquisition: Yu Y and Tauxe L (2006) Effect of multi-cycle heat treatment and pre-history
Numerical principles. EOS. Transactions of the American Geophysical Union dependence on partial thermoremanence (pTRM) and pTRM tails. Physics of the
85(47), Fall Meeting suppl. Abstract GP23A-0172 (available online). Earth and Planetary Interiors 157: 196–207.
Wu YT, Fuller M, and Schmidt VA (1974) Microanalysis of N.R.M. in a granodiorite Yu Y and Tauxe L (2008) Micromagnetic models of the effect of particle
intrusion. Earth and Planetary Science Letters 23: 275–285. shape on magnetic hysteresis. Physics of the Earth and Planetary Interiors
Xu S and Dunlop DJ (1994) Theory of partial thermoremanent magnetization in 169: 92–99.
multidomain grains. 2. Effect of microcoercivity distribution and comparison with Zhao X and Roberts AP (2010) How does Chinese loess become magnetized? Earth and
experiment. Journal of Geophysical Research 99: 9025–9033. Planetary Science Letters 292: 112–122.
Xu S and Dunlop DJ (2004) Thellier paleointensity theory and experiments for Zhao X and Tominaga M (2009) Paleomagnetic and rock magnetic results from lower
multidomain grains. Journal of Geophysical Research 109: B07103. http://dx.doi. crustal rocks of IODP site U1309: Implications for thermal and accretion history of
org/10.1029/2004JB003024. the Atlantis Massif. Tectonophysics 474: 435–448.
York D (1978a) A formula describing both magnetic and isotopic blocking
temperatures. Earth and Planetary Science Letters 39: 89–93.
York D (1978b) Magnetic blocking temperature. Earth and Planetary Science Letters
39: 94–97.
Yu Y (2011) Importance of cooling rate dependence of thermoremanence in Further-reading
paleointensity determination. Journal of Geophysical Research 116: B09101. http://
dx.doi.org/10.1029/2011JB008388. Dunlop DJ (1995) Magnetism in rocks. Journal of Geophysical Research
Yu Y and Dunlop DJ (2006) Testing the independence of partial thermoremanent 100: 2161–2174.
magnetizations of single-domain and multidomain grains of magnetite. Journal Moskowitz BM (1995) Fundamental physical constants and conversion factors.
of Geophysical Research 111: B12S31. http://dx.doi.org/10.1029/2006JB004434. In: Ahrens TJ (ed.) Rock Physics and Phase Relations: A Handbook of Physical
Yu Y, Dunlop DJ, and Özdemir Ö (2002a) Partial anhysteretic remanent magnetization in Constants, vol. 1, pp. 346–355. Washington, DC: American Geophysical Union
magnetite. 1. Additivity. Journal of Geophysical Research 107(B10): 2244. (chapter 23).
5.09 Centennial- to Millennial-Scale Geomagnetic Field Variations
C Constable, University of California at San Diego, La Jolla, CA, USA
M Korte, Deutsches GeoForschungsZentrum – GFZ, Potsdam, Germany
This chapter is a revision of the previous edition chapter by C Constable, Volume 5, pp. 337–372, © 2007, Elsevier B.V.

5.09.2 Data Types and Methods 310
5.09.2.1 Geomagnetic Directional Information 311
5.09.2.2 Paleointensity Data and Uncertainties 312
5.09.2.3 Age Controls 313
5.09.3 Local and Regional Secular Variation Studies 313
5.09.4 Global Data Compilations: Geographic and Temporal Sampling 316
5.09.5 Global Geomagnetic Field Reconstructions 319
5.09.5.1 Virtual (Axial) Dipole Moment and VGP Reconstructions 319
5.09.5.2 Spherical Harmonic Models 321
5.09.6 The Average Field and Its Secular Variation on Millennial Timescales 323
5.09.6.1 The Average Field 323
5.09.6.2 Dipole Moment and Axis Variations 325
5.09.6.3 Jerks, Drifts, and Waves 328
5.09.7 The Geomagnetic Spectrum 331
5.09.8 Applications 332
5.09.8.1 Calibrating Relative Geomagnetic Intensity Variations 332
5.09.8.2 Cosmogenic Isotope Production Rates: The Global View Linking Geomagnetic and Climate Studies 333
5.09.8.3 Surface Exposure Dating and Local Variations 334
5.09.8.4 Historical Applications: Building Orientations, Navigation, and Auroras 334
5.09.8.5 High-Resolution Magnetostratigraphy 335
5.09.9 Outstanding Problems and Scope for Future Progress 335
Appendix: Electronic Links 336
Acknowledgments 336
References 336
5.09.1 Introduction dating materials for archeological, sedimentary, or other geo-

logical applications.
The historical record contains comparatively few direct obser- The idea that archeological artifacts could be used as a
vations of the geomagnetic field prior to the sixteenth century, means of extending the geomagnetic record was known by
when the expanding number of global marine expeditions led the last decade of the nineteenth century. Folgerhaiter (1899)
to a great increase in the use of the magnetic compass. summarized the method and his earlier results on magnetic
Although ideas about relative variations in field strength were inclinations, and subsequently, Chevallier (1925) applied the
used much earlier, absolute measurements of magnetic field same ideas to oriented historical lavas from Mount Etna and
strength were not routinely available until the middle of the published a directional record of secular variation. Thellier
nineteenth century when Gauss initiated systematic measure- (1938) used archeological materials in the development of
ments at a number of geomagnetic observatories. Earlier obser- his now well-known technique for recovering magnetic field
vations of the geomagnetic field must be supplemented by strength (Thellier and Thellier, 1959; see Chapter 5.13 for
indirect measurements, made using a range of paleomagnetic more on paleointensities) and went on to produce a record
techniques on geological materials or archeological artifacts of intensity and directional changes in France that extended
that preserve a record of the field as either a thermal or (post) over 2000 years (Thellier, 1981). The first varved sedimentary
depositional magnetic remanence. The primary focus of this records of secular variations were attempted in the middle of
chapter is on records for the time interval ranging from about the twentieth century (Griffiths, 1953; Johnson et al., 1948).
0 to 10 ka, and it includes a discussion of archeomagnetic Subsequently, there have been numerous studies of sediments,
records preserved in man-made objects, paleomagnetic work greatly facilitated by the use of portable coring technology that
on young volcanic materials, and rapidly deposited lake and has allowed routine sampling of unconsolidated lake
marine sediments. To be useful as a geomagnetic record, the sediments and of high-accumulation-rate marine sediments.
age of magnetization must be obtained by a method that is Although data were often only available in a regional con-
independent of the geomagnetic field, but in some studies, the text, early studies of paleo- and geomagnetic secular variation
magnetic direction or field intensity is used as a mechanism for were routinely described in terms of motion of the

310 Centennial- to Millennial-Scale Geomagnetic Field Variations
geomagnetic dipole axis, westward drift of geomagnetic high-resolution magnetostratigraphy contribute to global
sources, and variations in the geomagnetic dipole moment. interpretations about the magnetic field. It is now feasible to
These results are well summarized in Chapter 4 of the mono- use these data to look at issues such as the evolution and
graph by Merrill et al. (1996). This was a reasonable approach longevity of the South Atlantic Anomaly, persistence of flux
given the predominantly dipolar structure in the geomagnetic lobes seen in the historical field, dynamo waves, and symme-
field combined with the sparse spatial and temporal data cov- tries in the millennial-scale geomagnetic secular variation. The
erage and lower accuracy generally available for millennial- remainder of this chapter discusses the status of these recent
scale paleomagnetic data compared with direct observations. magnetic field data and how they are currently being used.
Changes in dipole moment are not only of interest for geo-
magnetic studies (Genevey et al., 2008; Korte and Constable,
2005a; Valet et al., 2008; Yang et al., 2000) but also of impor- 5.09.2 Data Types and Methods
tance in other areas of Earth science: The geomagnetic field
effectively screens the Earth’s surface from cosmic radiation, Fundamental observations in archeomagnetic and paleomag-
and the magnitude of the dipole moment influences radio- netic work involve the recovery of one or more geomagnetic
genic nuclide production (Elsasser et al., 1956; Lal and Peters, elements (declination, D, inclination, I, or geomagnetic field
1967), with corresponding implications for climate studies strength, B or F; see Figure 1) from the time of acquisition of
that model solar activity (Usoskin et al., 2006) and/or ocean magnetization. Such data can be combined with an age esti-
circulation in the past (e.g., Bard et al., 2000; Muscheler et al., mate (either through absolute chronological methods or per-
2004a,b) and surface exposure dating (e.g., Dunai, 2001; haps via inferred stratigraphic or proxy relationships to other
Gosse and Phillips, 2001; Lal, 1991; Lifton et al., 2005). dated materials) to provide useful independent input for past
Long-term secular variation causes temporal changes in the geomagnetic field reconstructions. Alternatively, other inde-
location of the dipole axis and associated changes in auroral pendent knowledge of geomagnetic field variations may be
sightings that can be tracked through the historical record combined with new observations to allow geomagnetic dating,
(Oguti, 1993; Oguti and Egeland, 1995; Siscoe and Siebert, drawing inferences about age constraints on the material from
2002; Siscoe et al., 2002; Willis and Stephenson, 2001). As which undated geomagnetic elements are derived. Paleomag-
more data with better age constraints have become available, it netic data usually derive from an archeological structure or
has become possible to detect and map regional departures artifact or some geological material that is the object of study,
from simple dipolar field behavior (Constable et al., 2000; and knowledge of the context in which this object is found
Donadini et al., 2009; Genevey et al., 2008; Johnson and should play an important role in assessing the uncertainties
Constable, 1998; Korte et al., 2005). tied to the observations. The geographic location to be associ-
Historical field behavior provides a fairly detailed view of ated with the field elements is in many cases known extremely
geomagnetic field changes for the past 400 years. The current well: In others, it is not recorded with the desired accuracy, and
understanding, supported by a wealth of survey, observatory, in some cases, there may be confusion about the place where
and satellite data and elaborated in Chapters 5.02, 5.04, and specific artifacts originated. To acquire a paleomagnetic direc-
5.05 of this volume, is that the radial magnetic field at high tion requires knowledge of the relationship between the coor-
latitudes shows relatively stable structures often characterized dinate system in which measurements are made and the
as persistent flux lobes. The westward drift observed by Halley geographic system in which the magnetization was originally
and often considered a ubiquitous part of secular variation is acquired. If a structure, sediment, or lava flow has been dis-
pronounced in the Atlantic hemisphere and has been inter- turbed after the magnetization was acquired or its original
preted in terms of dynamo waves in the core (Finlay and orientation was unknown, this can add a systematic bias to
Jackson, 2003), but the secular variation in the Pacific hemi- the uncertainty already present in translating directions from
sphere is different in style and appears to be dominated by sample to geographic coordinates. Age constraints come in an
large-scale longer-term variations. Differences between the enormous range of flavors: Some give only broad stratigraphic
Atlantic and Pacific hemisphere secular variations may well relationships, while in the best-case scenario, they can be very
be controlled by thermal structure at the core–mantle bound- accurate, for example, when there is a correctly recorded eye-
ary (CMB) (Bloxham, 2002; Bloxham and Gubbins, 1987; witness account of some historical spectacle like a volcanic
Olson and Christensen, 2002). The Atlantic hemisphere also
shows a large region of low field strength known as the South
〉
x, North
Atlantic Anomaly that presents a radiation hazard for low-
earth-orbiting satellites (Heirtzler, 2002; Pinto et al., 1992; D Bh
Vernov et al., 1967) and has been tied to ongoing decreases
〉
in geomagnetic dipole moment. The longevity of this feature y, East

remains uncertain, but it certainly contributes to north–south I
hemispheric asymmetries in dipole moment changes for the
B
past few hundred years (Gubbins et al., 2006).
The acquisition of enough reliable paleomagnetic data to
〉
represent the global magnetic field is a painstaking business, z, Down

but for the time interval 0–10 ka, enormous progress has been Figure 1 Geomagnetic elements in a local coordinate system. D is
made in recent years. Long-term regional and global secular the declination, I the inclination, B the magnetic field strength (also
variation studies combined with archeomagnetic dating and known as F), Bh the horizontal component of magnetic field.
Centennial- to Millennial-Scale Geomagnetic Field Variations 311
eruption or for a specific style of pottery that can be closely tied gufm1 field model of Jackson et al. (2000) and suggest that a
to a specific time and place. General paleomagnetic techniques conservative estimate of the error is 2.5–3.0 for Italian lavas.
are discussed in Chapter 5.04, but some consideration needs A wide range of laboratory protocols continue to be used in
to be given here to the influence of specific sampling strategies archeomagnetic work, and there is no overall consensus on
and kinds of materials in the context of assessing uncertainties. quality criteria. Gallet et al. (2002) noted in a recent compila-
tion of western European results that laboratory cleaning
efforts range from none to the kind of complete demagnetiza-
5.09.2.1 Geomagnetic Directional Information
tion that allows principal component analysis. It should be
Sampling of young lava flows usually follows standard tech- noted that the latter is now widely accepted as the standard
niques described in Chapter 5.04, with multiple samples expected for high-quality paleomagnetic work. The same
drilled from a single flow using a handheld portable drill or remark could equally be applied to sedimentary directional
occasionally hand samples taken. Orientation data are col- data. Where available, it is probably reasonable to suppose
lected using sun or magnetic compass to allow transfer of that the Fisherian a95, the estimate of precision parameter k,
directions from the sample coordinate system into local coor- and the number of samples treated can provide some idea of
dinates as in Figure 1. In the absence of the sun, some quality control. Several studies have attempted to ground truth
researchers will tie the sample orientation into a global (GPS; results from historical lava flows with somewhat mixed results
Cromwell et al., 2013; Lawrence et al., 2009) or local (Baag et al., 1995; Castro and Brown, 1987; Rolph and Shaw,
(Mankinen and Cox, 1988) geodetic network providing a 1986; Tanguy and Le Goff, 2004; Valet and Soler, 1999). Some
direct tie into geographic coordinates rather than relying on have inferred large distortions of the field at individual sites
global geomagnetic models like the International Geomagnetic from local magnetic effects, but when large number of samples
Reference Field (IGRF) for local declination corrections. Each have been taken distributed along the volcanic unit as in the
sample may be cut into multiple specimens. Averaging of Mount Etna study by Tanguy and Le Goff (2004) and studies of
directions from independently oriented samples helps reduce the 1960 Hawaiian flow (discussed further later in the text), it
uncertainty in the paleomagnetic directional results and pro- seems that any large-scale influences are undetectable.
vides a means of evaluating internal consistency in the results. A variety of coring devices are used in acquiring sedimen-
Archeomagnetic sampling methods are more heterogeneous tary samples ranging from box cores (Klovan, 1964) designed
and often follow a hierarchy of the kind described by Lanos et al. to sample the topmost 0.5–1 m of sediment to the hydraulic
(2005). It is quite common to stabilize fragile material and piston cores used for large-scale projects like the Ocean Dril-
provide a suitable surface for orientation by forming some kind ling Program and drilling lakes like Lake Baikal. Most uncon-
of plaster or other cap over the structure to be sampled. solidated lake sediments are collected with some kind of
Kostadinova et al. (2004) evaluated the effects of water glass portable piston coring device, for example, the Mackereth
impregnation on magnetic properties of baked clay, while corer (Barton and Burden, 1979; Mackereth, 1958) or Living-
Schnepp et al. (2008) described some improvements in the stone corer (Livingstone, 1955). Freeze coring (Rymer and
basic sampling techniques used by Thellier (1938) and a soft Neale, 1981) can be used to recover material near the
coring strategy for baked clay. Similarly, Trapanese et al. (2008) sediment–water interface, which is frequently disturbed in
compared results from hand samples, disks, tubes, or plaster the piston coring process, but in practice, a multiple corer is
hand blocks noting that unreliable results in some cases might the most widely used apparatus for recovering undisturbed
be related to the suitability of material sampled. In some cases, material near the sediment–water interface. Freeze cores are
very large samples are collected (e.g., Tanguy et al., 1999). The not generally used as paleomagnetic samples, but they can
large sample size allows for high precision in the initial orienta- provide important age constraints. A common limitation in
tion process. Later, a core can be drilled from the oriented paleomagnetic records from sediment cores is that disturbance
sample, or sometimes, it is treated whole and measured in a or loss of material near the top of the core means that many
large inductometer (e.g., LeGoff, 1975). Other researchers collect records do not extend to the present day. It is also the case that
standard paleomagnetic drill samples of 1 in. diameter, when unless some reliable horizontal azimuthal orientation method
this is acceptable for the structure in question. The large sample has been employed, only relative declination variations can be
method has also been applied to volcanic rocks, and its pro- acquired. If the core does not penetrate vertically, then some
ponents (Arrighi et al., 2004; Tanguy et al., 2003) argue that with correction may also be needed to the measured inclination
this method, it is possible to achieve uncertainties of <2 in values. These issues are discussed in some detail by Snowball
averaged results for a single site. This is lower than what is usually and Sandgren (2002) (see also Figure 4) and Constable and
achievable using drill cores, and it has been suggested by Arrighi McElhinny (1985).
et al. (2004) that the large sample method is preferable for high- Once a sediment core has been acquired, the material is
resolution work. In contrast, Speranza et al. (2005) argued that generally subsampled in some way so that a sequence of mag-
laboratory and data processing techniques also play an impor- netization measurements and magnetic cleaning can be per-
tant role and that work by Lanza et al. (2005) demonstrates that formed in the laboratory. U-channel samples (Tauxe et al.,
the drill sample estimates provide a more realistic assessment of 1983) taken from the center of a split core can be measured
the uncertainty for both the large samples and drill core sites in a pass-through magnetometer, or closely spaced individual
when compared to the historical geomagnetic field directions. specimens (typically a few cubic centimeters in volume) can be
Arrighi et al. (2005) noted that there are few direct field obser- used to generate a stratigraphic record of magnetization within
vations, from the relevant time intervals: Lanza et al. (2005) the core. In either case, there is an inherent limitation to the
comparisons in large part depend on predictions from the temporal resolution attainable in the paleomagnetic results:
Subsamples represent an average over the time interval sam- Rolph and Shaw, 1986; Yamamoto et al., 2003). Recent studies
pled, and the pass-through magnetometer is limited by the by Mochizuki et al. (2004), Oishi et al. (2005), and Yamamoto
physical width of the instrument response. A variety of et al. (2003) suggest that in some cases, high mean paleointen-
methods have been described for improving the resolution of sities are obtained with the modified Thellier method depend-
pass-through records by deconvolution of the instrument ing on the degree of deuteric oxidation in the sample. They
response (Constable and Parker, 1991; Jackson et al., 2010; suggested using a modified Shaw method with double heating
Oda and Shibuya, 1996; Weeks et al., 1993). Parker (2000), and low-temperature demagnetization instead. Other checks
Parker and Gee (2002), and Jackson et al. (2010) have carefully on the Thellier method are discussed in Chapter 5.13. Detailed
analyzed the uncertainties for these procedures. Following any and specific information of this kind is often not available for
preliminary processing, each core is correlated with others prehistoric flows, but one can assess the bias expected overall
from nearby, using suitable stratigraphic markers (which may by combining historic data obtained using a broad variety of
involve the identification of characteristic sediment, volcanic, methods. Love and Constable (2003) combined 86 directional
or rock magnetic properties), and eventually placed on an and 95 intensity data available from the 1960 Kilauea flow
appropriate chronology. With appropriate correlation tools in on the island of Hawaii and used maximum likelihood to
place, magnetic records from multiple cores are often stacked estimate the mean and variance of the resulting field vector.
in attempts to improve the signal-to-noise ratio. In general, one Comparisons with the field values predicted from Bloxham
might expect paleomagnetic directions acquired from sedi- and Jackson’s (1992) field model (a predecessor to gufm1)
ment cores to have lower accuracy than volcanic sites because indicate reasonable agreement with the field values, but the
orientation errors (from core twisting and nonvertical core analysis of the distribution of observations reveals that the
penetration and from subsequent subsampling) can be signif- uncertainties are quite large: the standard deviation in F is
icant for individual cores. Stacked records also tend to be between 16% and 20% of the average value for a single flow
temporally blurred by the lack of resolution in correlating that should provide a spot reading of the field. In principle,
multiple cores within a given lake or region. this large uncertainty can be reduced by averaging results
from multiple specimens, but the time-consuming nature of
Thellier-type measurements combined with the usually rather
5.09.2.2 Paleointensity Data and Uncertainties
low success rate means that in the past, the number of speci-
The theoretical and practical basis for recovering paleomag- mens treated has generally been rather small. Recent studies
netic intensity estimates is described in Chapter 5.13, and the seem to suggest that the variance is lower for more homoge-
details of the many variations and combinations of absolute neous materials like submarine basaltic glass (Pick and Tauxe,
methods developed by (Thellier and Shaw, 1974) and micro- 1993) and archeomagnetic artifacts (Donadini et al., 2006;
wave strategies and the varied relative intensity methods will Korhonen et al., 2008; Ziegler et al., 2008), which are often
not be duplicated here. However, there are some important clay or ceramic. However, even in the best of cases, it currently
considerations in assessing the reliability of paleointensity data seems optimistic to expect to recover the field with a Thellier
for geomagnetic field studies. We make a major distinction paleointensity experiment with less than 10% standard error.
between absolute and relative paleointensity measurements These results are broadly in agreement with those presented by
and consider them as distinct entities requiring different eval- Korte et al. (2005) who compared nineteenth- and twentieth-
uation techniques. century results with predictions from the gufm1 field model
Absolute paleointensity observations are in principle based ( Jackson et al., 2000).
on the thermal remanent magnetization (TRM) preserved in a The results of the Kilauea study did not suggest any system-
sample after it has been cooled through some range of mag- atic bias in the mean, but two important effects can result in
netic blocking temperatures in a magnetic field, and it is widely systematic errors to individual results, namely, anisotropy in
believed on the basis of Neel (1955) theory that, for single- the material being studied and the influence of cooling rate on
domain and pseudo-single-domain magnetic grains, this TRM the TRM acquired (Aitken et al., 1981; Dodson and
acquisition can be at least approximately replicated in the McClelland-Brown, 1980; Fox and Aitken, 1980; Halgedahl
laboratory. One of the greatest hazards is that the magnetic et al., 1980). Samples generally cool more rapidly in the labo-
minerals present can alter during the paleointensity experi- ratory than in firing pots, bricks, or kilns or in cooling of a lava
ment in the laboratory, and numerous strategies have been flow (submarine basaltic glass being an exception), and this
developed for attempting to detect such alteration. Neverthe- leads to an overestimate of the field in which the original
less, it is to be expected that some changes will go undetected remanence was acquired. In principle, this can be corrected,
during the experiment. An important question for the purpose but one needs an estimate of the original cooling rate.
of recovering geomagnetic field behavior in the past (recently A number of careful studies of glass give independent measures
summarized by Lanos et al., 2005) is whether the experiment of cooling rate from relaxation geospeedometry, and these can
can be regarded as unbiased in the statistical sense: If it is be used for corrections (Bowles et al., 2005; Leonhardt et al.,
unbiased, then even observations with large uncertainties are 2006). Genevey et al. (2003) conducted some empirical stud-
in principle useful for studying geomagnetic behavior, but if ies for French pottery by experimental cooling of pots at dif-
systematic effects result in persistently high or low records from ferent rates in a special kiln designed to mimic the original
specific regions, this will clearly pose a problem. conditions. But such attention to detail is rare. In many cases,
A number of studies have been conducted on historical lava crude estimates are used to correct for cooling rate, while in
flows with a view to assessing the accuracy and reproducibility others, no correction is performed or it is not recorded. For
of the paleomagnetic method (e.g., Bohnel et al., 1997; Castro thin lavas and archeomagnetic artifacts, this might impart a
and Brown, 1987; Mochizuki et al., 2004; Oishi et al., 2005; 10% upward bias to the results. Very thick flows or intrusions
may suffer worse effects, and this might be one contributing the strictest constraints; nevertheless, the quality can vary con-
factor to the wide range of results found in different positions siderably depending on the specific study location, era, and
for the extensively studied Xitle flow in Mexico (Bohnel material available. The topic of archeomagnetic dating is dis-
et al., 1997). cussed further in Section 5.09.8.
Magnetic anisotropy is also known to impart bias to individ- Radiocarbon dating of organic material is widely used to
ual results (Rogers et al., 1979). The need for anisotropy correc- provide age constraints for both archeological sites and lake
tions has often been avoided in archeomagnetic studies, by sediments. For uniformity, uncalibrated radiocarbon years
applying the laboratory field so the TRM is acquired parallel to before present (BP) are conventionally measured relative to
Natural Remanent Magnetization (NRM) (Aitken et al., 1981). AD 1950. The ages must be corrected to account for the fact
However, Veitch et al. (1984) initiated the idea of correcting for that the production rate in the atmosphere varies with time, as
anisotropy using the TRM anisotropy tensor (e.g., Chauvin et al., seen in Figure 2 where the relationship between 14C radiometric
2000; Selkin and Tauxe, 2000). Sometimes, anisotropy of mag- age BP and calendar age is plotted. The data are from dendro-
netic susceptibility (AMS) has been used instead, but the fact that chronologically dated tree ring samples and form the basis for
different magnetic mineral fractions generally contribute to AMS the standard IntCal04 terrestrial radiocarbon age calibration
and NRM makes this a questionable approach. Cooling rate (Reimer et al., 2004). It is readily seen that there are substantial
corrections would be expected to lower the overall intensities, differences in age prior to about 3000 years ago, reaching as
but one might hope that the effects of anisotropy would average much as 2600 years by about 14 000 BC. The fact that the curve
out, so that when corrections have not been applied, this would is not monotonic makes the calibration of radiocarbon ages
just result in an overall increase in scatter when data from differ- inherently ambiguous in some time intervals. However, calibra-
ent sources are combined. tions have been steadily improving since an early curve pro-
Time series of relative paleointensity variations are produced posed by Clark (1975) (see also Stuiver and Reimer, 1993), and
using a variety of normalization techniques discussed in Chapter it is important to know the details of the calibration scheme in
5.13. Such estimates range from crude ratios of NRM to bulk order to assess the reliability of the age estimates. A separate
Anhysteretic Remanent Magnetization (ARM), Isothermal Rem- calibration is available for marine environments (Hughen et al.,
anent Magnetization (IRM), or susceptibility to methods that 2004) that takes account of the variable mixing times for oceans
mimic the Thellier method in a so-called pseudo-Thellier and atmosphere. The role of the ocean reservoir is reviewed by
method devised by Tauxe et al. (1995) for marine sediments Key (2001). The IntCal04 and Marine04 have subsequently
but applied, for example, by Snowball and Sandgren (2004) to been updated as described in Reimer et al. (2009 and 2013) to
varved lake sediments in Sweden. Many factors apart from geo- extend the curves to 50 000 calendar years BP, but with only
magnetic paleointensity variations can affect the normalized minor changes in the 0–12 cal kBP interval. Electronic links to
records, including changes in magnetic mineralogy, hiatuses or software packages for translating radiocarbon to calendar ages
other variations in sedimentation leading to incomplete records, are given in the ‘Appendix.’ They all draw on the same raw data
inclination errors or other anisotropies in the record, and tem- compilations, which are also available from the websites.
poral smoothing of the geomagnetic field record. There is no One concern in using radiocarbon dating of organic material
currently accepted general theory of how depositional or post- in lake sediments is the age of material being incorporated into
depositional remanence is acquired: consequently, there is no the sediment. Mixing of old organic carbon into the sediment
analog for the quality checks used in the Thellier-type absolute can cause systematic age biases; such influences can sometimes
paleointensity experiments. Although data selection criteria be identified by dating of the surface interface or independent
based on uniform rock magnetic properties have been in place age comparisons. Another possibility for introducing both age
for some time (see King et al., 1983; Tauxe, 1993; Tauxe and bias and temporal smoothing of the paleomagnetic signal lies in
Yamazaki, 2007), these remain ad hoc approaches to the prob- the lock-in depth for the remanence: This may vary over time and
lem that offer no guarantees of reliability. The best quality assess- it is not unusual for it to generate a smoothing effect over
ment in many cases comes from regional compatibility among a 100–200 years (e.g., Sagnotti et al., 2005), with corresponding
broad range of different records. uncertainties in the ages assigned to the magnetization. Addi-
tionally, the amount of dateable material in the sediment and
personnel and financial capacity limit the number of radiocar-
5.09.2.3 Age Controls
bon dates that can be obtained for a given sediment core. The
A wide variety of dating techniques are applied to volcanic, time intervals spanned by dated tie points can be several thou-
archeological, and sedimentary materials, including radiomet- sand years, and methods to interpolate them vary. Ages can differ
ric ages, varve counting, archeological ages, thermolumines- by centuries in some intervals depending on whether, for exam-
cence (Aitken, 1998; Liritzis, 2011), various stratigraphic ple, piecewise linear interpolation, polynomial or spline fits
correlations, and other relative methods. An overview of the are used. Recent developments towards using Bayesian methods
various absolute and relative methods used in quaternary geo- to construct age–depth models (Blaauw and Christen, 2011;
chronology is provided in various chapters in Noller et al. Parnell et al., 2008; Ramsey, 2008) can provide improved age
(2000), but the most widely used ‘absolute’ method in the uncertainty statistics for sediment records.
Holocene interval remains radiocarbon dating, which was
developed by Libby (1955) and coworkers following the
1934 interpretation by F. N. D. Kurie at Yale that certain 5.09.3 Local and Regional Secular Variation Studies
cloud chamber observations could be interpreted as evidence
for the existence of 14C (Kamen, 1963; Key, 2001). In general, As was noted in Section 5.09.1, Thellier was a pioneer of devel-
varve counting and archeological methods are able to provide oping the archeomagnetic record in France and established
400
350
2000
300
4000
250
D14C (per mil)
14C years BP
6000
200
150 8000
100
10000
50
12000
0
−50 14000
−14000 −12000 −10000 −8000 −6000 −4000 −2000 0 2000
Year BC/AD
Figure 2 Reconstruction of atmospheric concentration of 14C for the past 16 ky (black) from Stuiver et al. (1998) and Reimer et al. (2004) and
of radiocarbon age versus calendar age (blue). Dashed blue line indicates one–one relationship. Error bars are 1s. Note that 14C years BP are counted
relative to AD 1950, so the offset between the two age scales at 14 000 BC is about 2600 years.
techniques that are acknowledged as fundamental today. latter goal include first data from Argentina (Goguitchaichvili
Europe and surrounding regions extending eastward and south- et al., 2011), Brazil (Hartmann et al., 2010), Senegal and Mali
ward to the Far East are natural regions in which to conduct (Mitra et al., 2013), and South Africa (Neukirch et al., 2012).
archeomagnetic studies, because of the long history of human Building a regional secular variation curve often requires the
structures and associated archeological artifacts. Extensive his- reduction of widely distributed data to a central location, and
torical records can make dating more accurate, and there are established strategies for accomplishing this. Directional
consequently, there are a number of well-established regional data are commonly adjusted for the gross geographic variations
reference curves for secular variation, some with multiple revi- by mapping through the virtual geomagnetic pole (VGP), using a
sions and improvements as, for example, in France (Bucur, method described by Shuey et al. (1970), while intensity data are
1994; Chauvin et al., 2000; Gallet et al., 2002; Genevey and equivalently adjusted by mapping through the virtual dipole
Gallet, 2002; Thellier, 1981). An overview of recent reference moment (VDM) (see, e.g., Daly and Le Goff, 1996).
curves for several countries is given in Table 1. Outside of the A variety of smoothing techniques have been used in con-
European area, reference curves have been published for North structing reference curves for both sediment and archeomag-
America (Hagstrum and Blinmann, 2010; LaBelle and Eighmy, netic data. These range from scalar averaging of individual
1997; Lengyel, 2010; Lengyel and Eighmy, 2002), Japan geomagnetic elements to more sophisticated bivariate moving
(Yoshihara et al., 2003), and Korea (Yu et al., 2010). In many average approaches (Daly and Le Goff, 1996). The topic of
other parts of the world, in particular in the southern hemi- consistent and optimal construction of reference curves for
sphere, the archeomagnetic record is not so well established secular variation has been addressed in some detail in a paper
because the human record is sparse and less well documented. by Lanos (2004) laying out and applying a Bayesian approach
Places where fired archeological structures and artifacts are to incorporating all the chronological and archeomagnetic
sparse can sometimes provide numerous if sporadic volcanic information and their uncertainties. The method produces a
records instead, with Hawaii and the Western United States penalized spherical spline smoothing of the vector observa-
providing two of the best-studied examples. Japan and Italy are tions that takes account of data uncertainties and the uneven
well represented by mixed archeological and volcanic records. temporal distribution of observations. A functional envelope
There remains considerable scope for integrated analyses of for the 95% confidence limit is defined. Figure 3 shows as an
existing data (see studies compiled at http://geomagia.ucsd. example the Austrian reference curve by Schnepp and Lanos
edu/geomagia/studies.php) to provide well-defined regional (2006). Undated or poorly constrained archeomagnetic direc-
curves and for acquiring new data that could fill some holes in tions can then be assigned a probabilistic archeomagnetic age
geographic and temporal coverage. Recent efforts towards the using a Bayesian reference curve. The method has been
successfully applied to a number of European studies (see from multiple cores are often stacked together and smoothed
Table 1) and is sufficiently flexible that as the number of in an attempt to improve signal-to-noise ratio in the record.
archeomagnetic data available continues to grow, it should This requires identifying equivalent stratigraphic markers and
be possible to compute custom reference curves for specific interpolation onto common timescales. When the chronology
localities rather than relying on existing national curves. is poorly constrained, stacking often seems to result in an
The Bayesian approach has recently been contrasted with a apparent loss of resolution. In principle, Bayesian approaches
hierarchical extension of the moving average bivariate analysis like the one of Lanos (2004) can be applied to stack lake
used by Daly and Le Goff (1996). Lanos et al. (2005) had sediment data, and there is some work in progress, but nothing
conducted a detailed analysis of a hierarchical bivariate mov- in this regard has been published yet.
ing average modeling procedure for producing reference curves Many of the lake records show good internal consistency at a
and concluded that it is appropriate for very well-dated data local level, and an example from Snowball and Sandgren (2002)
that are evenly distributed in time but that such data sets are is shown in Figure 4. Sarsjön and Frängsjön are two closely
only rarely achievable. spaced lakes with varved sediments situated in northern Sweden
Sediment cores directly provide time series of geomagnetic (locations at 64 020 N, 19 360 E, and 64 010 N, 19 420 E, respec-
field variations from individual locations. Lake sediment data tively). In Figure 4, black triangles and circles correspond to data
Table 1 Regional archeomagnetic reference curves
Country Latitude Longitude Reference Time (AD) No. of data Components Method
Austria 47.38 N 13.45 E Schnepp and Lanos (2006) 1000 to 1900 170 D, I CS Bayesian
Bulgaria Kovacheva et al. (1998) 6150 to 1850 D, I, F CS
France 48.9 N 2.3 E Gallet et al. (2002) 950 to 1830 D, I RM
Germany 51.53 N 9.93 E Schnepp and Lanos (2005) 1000 to 1900 166 D, I CS Bayesian
Greece 37.97 N 23.72 E Marco et al. (2008) 5700 to 1900 330 F CS Bayesian
Hungary 47.5 N 19.1 E Márton (2010) 300 to 1800 217 D, I CS Bayesian
Italy 42.45 N 12.03 E Tema et al. (2006) 1200 to 1600 65 D, I CS Bayesian
Japan 35.0 N 135.8 E Yoshihara et al. (2003) 0–2000 68 F CS
Korea 37.6 N 127.0 E Yu et al. (2010) 1100 to 1790 26 D, I
Spain 40.4 N 3.7 W Gómez-Paccard et al. (2006) 875 to 1959 134 D, I CS Bayesian
United Kingdom 52.43 N 1.64 W Zananiri et al. (2007) 2000 to 1700 620 D, I CS Bayesian
North America 35.3 N 108.45 W Hagstrum and Blinmann (2010) 1588 to 1986 533 D, I RM
D, declination; I, inclination; F, intensity; CS, continuous, splines; RM, running means.
40
30
Declination (⬚)
20
10
0
−10
−20
−30
80
75
Inclination (⬚)
70
65
60
55
50
−800 −400 0 400 800 1200 1600 2000

Age (years AD)
Figure 3 D and I plotted versus timescale (2s or archeological estimate for age, 95% confidence limit for D, I) for relocated data used to construct the
Austrian reference curve. Open symbols are from France, Germany, and Switzerland, closed from Austria, Bosnia, and Hungary. Bold black lines
give the reference curves; gray area is the 95% confidence band. Reproduced from Schnepp E and Lanos P (2006) A preliminary secular
variation reference curve for archeomagnetic dating in Austria. Geophysical Journal International 166: 91–96. (http://dx.doi.org/10.1111/j.1365-246X.
2006.03012.x).
90
85
80
Inclination (deg)
75
70
65
60
55
50
−6000 −4000 −2000 0 2000
Calendar years (−BC/+AD)
150
100
Declination (deg)
50
−50
−100
−150
−6000 −4000 −2000 0 2000

Calendar years (−BC/+AD)
Figure 4 Secular variation records obtained from Swedish lakes with varved sediments (Snowball and Sandgren, 2002); black triangles and
circles are from two Sarsjön cores, and blue from Frängsjön.
from two separate cores from Sarsjön and blue from Frängsjön. to the lowest-degree model term (g00) and then when scaled by
In general, cores from the same lake produce records that agree to absolute intensity data in SCHA.DI.00-F still did not consider
within a few degrees. The occasional intervals with high scatter the full vector field information consistently. These are super-
highlight the need for replicate records with good internal seded by SCHA.DIF.3K, spanning the interval 1000 BC to AD
consistency. 1900. Model SCHA.DIF.8k extends further back, spanning the
The regional consistency between archeomagnetic or lava interval 6000–1000 BC. Due to the scarcity of archeomagnetic
data and sediment records varies. While general features often data for these times, it includes 24 lake sediment records, ten of
agree, more detailed discrepancies are readily found at both which have relative paleointensity information. Figure 6
the individual and reference curve levels. Figure 5 shows shows data, compiled regional curves, and a comparison of
directional results from individual Hawaiian lavas compared different model predictions for the locations of Paris in France
with the Lake Waiau record of Peng and King (1992) and and Viterbo in Italy.
individual German archeomagnetic results and the Bayesian Data, regional reference curves, and model predictions in the
archeomagnetic reference curve Schnepp and Lanos (2005) European region confirm that the structure of secular variation is
compared with the composite Eifel maars sediment record of broadly similar across length scales of up to a few thousand
Stockhausen (1998). The main sources of discrepancies kilometers. However, regional correlations break down on
include stronger smoothing of the secular variation signal large scales, and Asian records cannot easily be reconciled with
inherent to most sediment records, uncertainties in age– European versions. Similarly, there are regional similarities across
depth models or dating of archeomagnetic artifacts, and occa- North America, but not between Europe and North America. This
sional inclination flattening in sediments. breakdown in correlation on large scales can be attributed to
The increasing number of data and easy availability from systematic nondipole contributions to the geomagnetic field.
databases and global compilations (Section 5.09.4) have
spurred the development of both regional and global models
(Section 5.09.5). A series of regional models (Table 2) have 5.09.4 Global Data Compilations: Geographic
been developed for Europe, the region with the best data and Temporal Sampling
coverage (Pavón-Carrasco et al., 2008a,b, 2009, 2010). They
are all based on spherical cap harmonic techniques and arche- In addition to the regional efforts described in the previous
omagnetic data. The first versions were initially based on direc- section, there have been a number of initiatives directed at
tional data (SCHA.DI.00) and thus modeled variations relative acquiring global collections of observations for further
60 80
75
50
Inclination (deg) 70
Inclination (deg)
40
65
30 60
55
20
50
10
45
40
−8000 −4000 0 −6000 −4000 −2000 0 2000
Years (−BC/+AD) Years (−BC/+AD)
40 40
Declination (deg)
Declination (deg)
20 20
0 0
−20 −20
−40 −40
−8000 −4000 0 −6000 −4000 −2000 0 2000

Years (−BC/+AD) Years (−BC/+AD)
(a) (b)
Figure 5 I and D from (a) Hawaiian lavas in red and Lake Waiau sediment data from Peng and King (1992) in blue, (b) from German
archeomagnetic data in red, from Eifel maars composite sediment record by Stockhausen (1998) in blue, and from the German reference curve
(Schnepp and Lanos, 2005) with uncertainty estimates in black. The lava and archeomagnetic data are compiled in the GEOMAGIA50 database and are
shown with one standard deviation error bars. All Hawaiian data are reduced to the location of Lake Waiau, and all German data and the reference
curve are reduced to the location of Göttingen.
Table 2 Regional European archaeomagnetic field models
Model Reference Time (AD) Components Data Method
SCHA.DI.00 Pavón-Carrasco et al. (2008b) 0–1900 AM, 5 Bayesian curves D, I Relative SCHA
SCHA.DI.00- Pavón-Carrasco et al. (2008a) 0–1900 AM, 5 Bayesian curves þ 611 F D, I, F Relative SCHA, scaled by
F absolute, sup. by SCHA.DIF.3K
SCHA.DIF.3K Pavón-Carrasco et al. (2009) 1000 to 1900 4329 AM þ 3 hist. records D, I, F SCHA
SCHA.DIF.8K Pavón-Carrasco et al. (2010) 6000 to 1000 959 AM, 24 LS, incl. 10 RPI D, I, F R-SCHA2D
AM, archaeomagnetic data; LS, sediment data; RPI, relative paleointensity; D, declination; I, inclination; F, intensity; SCHA, spherical cap harmonic analysis; R-SCHA2D,
revised-spherical cap harmonic analysis in 2-D (Thébault, 2008). See text for details.
synthesis. In the late 1980s, the International Association for (1998) extended the data collection to include archeomagnetic
Geomagnetism and Aeronomy (IAGA) encouraged the devel- data from Peru, sediment data from Argentina and New Zeal-
opment of a series of paleomagnetic databases. These included and, and lava data from Hawaii and Sicily, giving a slightly
ARCHEO, containing archeomagnetic directional data com- better global coverage with these 14 localities. Johnson and
piled by Donald Tarling of Southampton University, and Constable (1998) and Constable et al. (2000) extended the
SECVR, with lake sediment records compiled by McElhinny temporal span to 1000 BC and produced 24 data series based
and Lock (1996). These data sets are archived at the National on 100-year averages of D and I using the previously listed
Geophysical Data Center (http://www.ngdc.noaa.gov/seg/geo directional sites but grouped slightly differently and aug-
mag/paleo.shtml) but have been largely superseded by others mented with globally distributed lake sediments and further
described later in the text, which incorporated them. Daly and archeomagnetic data including new sites from China, Mongo-
Le Goff (1996) compiled available archeomagnetic intensity lia, Australia, and New Zealand. This data set, known as
and directional data for the past 2000 years, sorted them into PSVMOD1.0, provided the first reasonable global distribution
nine geographic regions, corrected the data to a standard refer- of directional observations suitable for global field modeling.
ence location within each region, and produced smoothed Korte et al. (2005) assembled data from a variety of published
curves presenting D, I, and F at 25-year intervals where data sources for the explicit purpose of developing a time-varying
were available. Five sites had more-or-less complete directional global field model, using the paleomagnetic databases
data for the past 2000 years; only three also had good coverage Archeo00 and Secvr00 archived at the World Data Center in
with paleointensity. All were in the northern hemisphere, and Boulder to acquire digital directional observations where pos-
most were confined to midlatitude regions. Hongre et al. sible, and supplementing these with information acquired
40 40
30 30
20 20
Declination (deg)
Declination (°)
10 10
0 0
-10 -10
-20 CALS3k.4 ARCH3k.1 -20 CALS3k.4 ARCH3k.1
CALS10k.1b SCHA.DIF.3K CALS10k.1b SCHA.DIF.3K
-30 Reference curve SCHA.DIF.8K -30 Reference curve SCHA.DIF.8K
-40 -40
-6000 -4000 -2000 0 2000 -6000 -4000 -2000 0 2000
80 80
75 75
70 70
Inclination (deg)
Inclination (°)
65 65
60 60
55 55
50 50
45 45
40 40
-6000 -4000 -2000 0 2000 -6000 -4000 -2000 0 2000
100 100
90 90
80 80
Intensity (mT)
Intensity (mT)
70 70
60 60
50 50
40 40
30 30
20 20
-6000 -4000 -2000 0 2000 -6000 -4000 -2000 0 2000
(a) Year (-BC/+AD) (b) Year (-BC/+AD)
Figure 6 Archeomagnetic data from (a) France and (b) Italy in gray as provided in the GEOMAGIA50 database, reduced to Paris, France, and Viterbo,
Italy, respectively, and a comparison of different models and regional curves for the same locations. Reference curves (brown) by Gallet et al.
(2002) and Tema et al. (2006), respectively; regional models SCHA.DIF.8K (orange) (Pavón-Carrasco et al., 2010) and SCHA.DIF.3K (red)
(Pavón-Carrasco et al., 2009); and global models CALS3k.4 (black) (Korte and Constable, 2011), CALS10k.1b (blue) (Korte et al., 2011), and
ARCH3k.1 (purple) (Korte et al., 2009) described in Section 5.09.5.
directly from the investigators involved in data collection. newly published data. It provides the most comprehensive
Archeomagnetic intensity data were obtained from the original global data set currently available in a web-based searchable
literature or directly from the authors. Three kinds of data are format (see ‘Appendix’ for URLs).
distinguished in this data set: (a) directional records from lake The global sedimentary record compilation of Korte et al.
and rapidly deposited marine sediments, (b) archeomagnetic (2005) has been complemented by relative paleointensity
directional data, and (c) archeomagnetic intensity data. records (Korte and Constable, 2006b) and regularly updated
No intensity data from sediments were included, because at by newly published or earlier missed records (Donadini et al.,
the time, no systematic evaluation of their validity and internal 2009). The most recent version of this compilation is given by
consistency had been undertaken. All included data have a chro- Korte et al. (2011), and work to incorporate this collection of
nology that is independent from the magnetic measurements. files into a GEOMAGIA-like database with metadata informa-
Two archeointensity data compilations for the past 50 ky tion is in progress.
with comprehensive metadata information were developed in The MagIC (Magnetics Information Consortium) PMAG
parallel at about the same time: the ArcheoInt data compila- Portal (http://earthref.org/MAGIC) provides access to paleo-
tion by Genevey et al. (2008) and the GEOMAGIA50 database magnetic data from a broad range of studies and timescales.
by Donadini et al. (2006) (see also Korhonen et al., 2008). In The entries range from individual measurements to specimen,
the meantime, they have been merged, all archeomagnetic sample or site level results, and include a wide variety of
directional information from the Korte et al. (2005) compila- metadata like associated rock magnetic measurements. All
tion has been included, and this GEOMAGIA50 database GEOMAGIA50 (http://geomagia.ucsd.edu/geomagia/) entries
maintained by Fabio Donadini is regularly updated to include are included and linked to other comprehensive information
in the underlying publications, and the same is planned for the homogeneous time series from single locations, which are
GEOMAGIA-like sedimentary database. marked in Figure 7(d)–7(f ). Archeomagnetic data in contrast
The numbers of data in the latest archeomagnetic data and rarely form a time series at a single site, and each individual
sediment record compilations for most of the Holocene period result is plotted in Figure 7(a)–7(c) without conveying any-
(10 000 BC to present) are given in Table 3, and their spatial thing about temporal coverage of different regions. The general
distribution is shown in Figure 7. The dominance of northern temporal distribution of the different data types and compo-
hemisphere data is obvious. Sediments provide more-or-less nents is illustrated in Figure 8, from which it is obvious that
the archeomagnetic results are concentrated strongly towards
the most recent two millennia and that intensity information is
somewhat underrepresented.
Table 3 Numbers of globally compiled Holocene archeo- and
paleomagnetic data
Component All Archaeomagnetic Sediment

5.09.5 Global Geomagnetic Field Reconstructions
Inclination
All 40 522 5187 35 335 5.09.5.1 Virtual (Axial) Dipole Moment and VGP
N. Hemisphere 34 297 5134 29 163 Reconstructions
S. Hemisphere 6225 53 6172
The geomagnetic field observed at the Earth’s surface is clearly
Declination
All 34 300 3692 30 608 dominated by a tilted geocentric dipole geometry. Neglecting
N. Hemisphere 28 075 3639 24 436 all more complex field contributions, the dipole field evolu-
S. Hemisphere 6225 53 6172 tion can be reconstructed by virtual axial dipole moment
Intensity (VADM) estimates and VGPs, obtained from archeointensity
All 17 779 4149 13 630 and archeo- and/or paleomagnetic directional data, respec-
N. Hemisphere 14 686 3904 10 782 tively. As described in Section 5.13, a VADM is the equivalent
S. Hemisphere 3093 245 2848 geocentric axial dipole moment that would produce the
All components observed intensity at that latitude, and a VDM is the equivalent
All 92 601 13 028 79 573
moment of a tilted geocentric dipole, obtainable if inclination
N. Hemisphere 77 058 12 677 64 381
is known in addition to intensity. The piercing point of the
S. Hemisphere 15 543 351 15 192
tilted dipole axis is the VGP, which can be obtained from
(a) (d)
(b) (e)
(c) (f)
Figure 7 Locations of globally compiled Holocene archeomagnetic and paleomagnetic data. Sites of archeomagnetic and lava data included in the
GEOMAGIA50 database on the left and sediment record locations on the right, for declination (a, d), inclination (b, e), and absolute archeomagnetic (c) or
relative sedimentary intensity (f ).
400 Archeomagnetic
all
Declination
Inclination
No. data/50 year

300 Intensity
200
100
0
−10 000 −8000 −6000 −4000 −2000 0 2000
(a) Year (−BC/+AD)
600 Sediments
all
500 Declination
Inclination
No. data/50 year
Intensity
400
300
200
100
0
−10 000 −8000 −6000 −4000 −2000 0 2000
(b) Year (−BC/+AD)
Figure 8 Numbers of (a) archeomagnetic þ lava and (b) sediment paleomagnetic data over time for declination (green), inclination (blue), absolute or
relative intensity (red), and the sum of all component data (black) in 50-year bins. Archeomagnetic data as currently included in the GEOMAGIA50
database; sediment data from the global compilation by Korte et al. (2011).
declination and inclination data at a given location. criteria (Genevey et al., 2008) are selected. The range across
Obviously, any nondipole field contributions will distort indi- individual VADMs of all intensity data presently included in
vidual VADM, VDM, or VGP results compared to the true the GEOMAGIA50 database illustrates the range of potential
dipole moment and geomagnetic pole. In the present-day bias contributed to the dipole estimate by the nondipole field
field, this effect can bias, for example, VADMs by up to at single locations, and the increased uncertainty estimates in
50%. Averaging of globally distributed VADMs, VDMs, the Knudsen et al. (2008) reconstruction in the early millennia
and VGPs can mitigate such distortions efficiently for well- with sparse data coverage indicate the potential for geographic
distributed data, but care has to be taken with inhomogeneous bias in globally averaged VADMs. Small uncertainty estimates
data distributions (e.g., Korte and Constable, 2005b). Tempo- in recent millennia might be deceptive: The available data
ral averaging can also reduce nondipole field contributions, distribution is extremely biased towards the northern hemi-
which in general vary faster than the dipole part. However, it sphere (cf. Figure 7(c)), and this cannot be overcome by
remains unclear whether any geographically insufficient data bootstrap resampling.
distribution can efficiently eliminate all nondipole field Globally averaged VGP reconstructions are based on a larger
influences. variety of data sets. For example, Merrill and McElhinny (1983)
Table 4 provides an overview over a number of dipole field used archeomagnetic data from the past 2000 years, and Ohno
reconstructions, and the recent ones are shown in Figure 9. The and Hamano (1992) used sediment records from up to
reconstructions by McElhinny and Senanayake (1982) and 11 regions to calculate global VGPs as 500-year averages for
Yang et al. (2000) have a low temporal resolution from aver- the past 10 000 years. More recently, Valet et al. (2008) used the
ages binned in 500–1000-year intervals and differ mainly in updated compilation of archeomagnetic directional data as
the number of data used. More recent global VADM recon- now included in the GEOMAGIA50 database to determine
structions based on the data compilations described earlier in 25-year averages of VGPs from three to five regions for the
the text and constructed as running average curves differ little past 2000 years, and Nilsson et al. (2010) obtained global
(Figure 9(a)), regardless of whether all available data VGPs for the past 9000 years as running 100-year means of
(Knudsen et al., 2008) or only data fulfilling certain quality three to five well-distributed sediment records selected by
Table 4 Holocene global geomagnetic field reconstructions
Model Reference Time (AD) Data type No. of data Method
Dipole Reconstructions
VDM/VADM McElhinny and Senanayake 10 000 to 1980 AM int. 1175 1000–500-year avg. SS
(1982)
VDM/VADM Yang et al. (2000) 10 000 to 1995 AM int. 3243 1000–500-year avg. SS
VDM/VADM Genevey et al. (2008) 6000 to 1950 AM int. 1839 sel. 100-year avg. RM
VADM Valet et al. (2008) 0–1900 AM int. 100-year avg. RM
VADM Knudsen et al. (2008) 10 000 to 1750 AM int. 3517 1000–500-year avg. RM
VGP Ohno and Hamano (1992) 8000 to 1950 LS dir. 11 rec. 500-year avg. SS
VGP Valet et al. (2008) 0–1900 AM dir. 3–5 regions 25-year. avg.
VGP Nilsson et al. (2010) 7000 to 1500 LS dir. 3–5 LS rec. 100-year avg. RM
SH models
HHK Hongre et al. (1998) 0–1700 AM, LS, no RPI 14 smoothed rec. 25-year SS, lmax ¼ 2 þ (l ¼ m ¼ 3)a
ALS3K Constable et al. (2000) 1000 to 1800 AM, LS, dir. 24 smoothed rec. 100-year SS, SR, fixed ADM
CALS3K.1 Korte and Constable (2003) 1000 to 1950 AM, LS, dir. 24 smoothed rec. CS, STR, prescribed ADM
CALS3K.2 Korte and Constable (2005a) 1000 to 1950 AM, LS, no RPI 19 376 CS, STR, sup. by CALS3k.4
CAKS7K.2 Korte and Constable (2005a) 5000 to 1950 AM, LS, no RPI 32 353 CS, STR, sup. by CALS10k.1bb
CALS3k.3 Korte et al. (2009) 1000 to 1990 AM, LS, incl. RPI 29 980 CS, STR, sup. by CALS3k.4
ARCH3k. 1 Korte et al. (2009) 1000 to 1990 AM 9605 CS STR, strong north hemisphere Biab
SED3k.1 Korte et al. (2009) 1000 to 1990 LS incl. RPI 20 375 CS, STR, sup. by CALS10k.1bb
CALS3k.4 Korte and Constable (2011) 1000 to 1990 AM, LS, incl. RPI 35 777 CS, STR
CALS3k.4b Korte and Constable (2011) 1000 to 1990 AM, LS, incl. RPI 35 800 CS, STR, BA
CALS10k.1b Korte et al. (2011) 8000 to 1990 AM, LS, incl. RPI 87 000 CS, STR, BA
HFMb Panovska (2012) 8000 to 1990 AM, LS, incl. RPI 85 000 CS, STR
AM, archeomagnetic data; LS, sediment data; SS, snapshots; RM, running mean; CP, continuous, polynomial; CS, continuous, splines; ADM, axial dipole moment; RPI, relative
paleointensity; BA, bootstrap average model; avg., average; int., intensity data; dir., directional data; sel., selected by quality criteria; SR, spatial regularization; STR, spatial and
temporal regularization; sup., superseded; rec., sediment record or composite archeomagnetic curve.
a
Gauss coefficients up to degree (l) and order (m) 2 and g33, h33.
b
Four models using same data as CALS10k.1b with modified weighting and different regularization constraints.
data quality criteria. These two geomagnetic pole reconstruc- of the past 3 ky and construct extended versions going back
tions (Figure 9(b)) show broadly similar variations of 7 and 10 ky.
dipole tilt agreeing within the uncertainty estimates, but differ- The basic method for building spherical harmonic global
ences in details indicate potential influences from characteris- models of millennial-scale geomagnetic field variations has
tics of underlying data sets and data distribution. Further followed that used for the present and historical field. The
discussion of the geomagnetic dipole evolution is given in parameterization is in terms of a truncated spherical harmonic
Section 5.09.6.2. expansion of a scalar potential representing the magnetic field
originating in the Earth’s core:
1 X
X l lþ1 m
a m
5.09.5.2 Spherical Harmonic Models cðr, y, fÞ ¼ a l ðt Þcosmf þ hl ðt Þsin mf P l ðcos yÞ
gm
l¼1 m¼0
r
Attempts to build global spherical harmonic geomagnetic field
models on archeological timescales have been going on since where Pm l are partially normalized Schmidt functions of degree
the 1970s (e.g., Braginskiy and Burlatskaya, 1979; Márton, l and order m and position is specified in a geocentric spherical
1970; Ohno and Hamano, 1993; Sakai, 1980), and a recent coordinate system with radius r, y as colatitude, and f as
resurgence of interest in such activities has followed from the longitude. A magnetic field model is a listing of the Gauss
much improved data sets that have been compiled over the coefficients gm m
l , hl up to some maximum degree L: the coeffi-
past two decades. These investigations began with smoothed cients may or may not have an explicit parameterization in
time series of archeomagnetic directional data (Daly and Le time, t, that expresses the temporal dependence arising from
Goff, 1996), then extended to include lake sediment direc- secular variation. When no explicit temporal parameterization
tional data for the past 3 ky (Constable et al., 2000; Hongre is specified, the model either is usually an average over some
et al., 1998; Johnson and Constable, 1998; Korte and time interval or can be regarded as a snapshot model of a
Constable, 2003), and more recently have used growing num- specific moment in time. The gradient of the potential provides
bers of original archeomagnetic directional and intensity the magnetic field B(r, y, f, t) ¼ ∇c(r, y, f, t), and as
results together with sediment directional (Korte and appropriate, the secular variation can be evaluated from @B/@
Constable, 2005a; Korte et al., 2005) and relative intensity t. The approximation that there are no magnetic sources in
records (Korte and Constable, 2006b, 2011; Korte et al., the Earth’s (more-or-less) insulating mantle means that
2009, 2011) to improve reliability and resolution of models field models built from observations at the Earth’s surface
30 excluded from the model. The maximum degree for least-

squares modeling is usually much less than in regularized inver-
25
sions, which typically extend to spherical harmonic degree and
20 order ten and may have spline representations with knots every
1022 Am2
10–50 years. The actual spatial and temporal resolutions for

15
these regularized millennial-scale models are much lower than
10 this and not exactly known. Synthetic experiments with gufm1
or a numerical dynamo simulation and data distributions for
5
regularized inversion models demonstrated broad-scale agree-
ment between the input and recovered models (Korte and
-8000 -4000 0 Constable, 2003, 2008). Korte and Constable (2008) infer
(a) Calendar years (-BC/+AD) from the rapid falloff in the spatial power spectrum that the
90 resolution of regularized models with the available data distri-
88 bution at that time can be no better than spherical harmonic
86 degree four and estimate the temporal resolution to be no better
Pole latitude (deg)
84 than in the order of 100 years. The absence of power in the

82 secular variation at periods shorter than a few hundred years
80 suggests that averaging in the time domain might even be two to
78
three times larger (Korte and Constable, 2006a). These results
76
are in broad agreement with experiments on data distributions
74
72 conducted by Hatakeyama (2007). The recent amplifications in
70 numbers of available data have led to some refinements in the
-8000 -4000 0 models, but not yet to a significant increase in achievable spatial
(b) Calendar years (-BC/+AD) and temporal resolution.
Figure 9 Global virtual (axial) dipole moment reconstructions by Uncertainties in archeo- and paleomagnetic data and in
Genevey et al. (2008) in black, Valet et al. (2008) in yellow, and Knudsen their dating are clearly much higher than errors in direct obser-
et al. (2008) in red and individual VADMs from all archeointensity data vations of the geomagnetic field and have to be considered in
at present included in the GEOMAGIA50 database in gray (a). Global modeling. A standard method is to weight the data by their
virtual geomagnetic pole reconstructions by Valet et al. (2008) in black uncertainties, but inhomogeneity in the kind of uncertainties
and Nilsson et al. (2010) in red. Archeomagnetic data are shown reported hampers a rigorous treatment of errors. Some recent
with one standard deviation error bars, and the red dashed lines indicate progress in evaluating uncertainty has been achieved by
model uncertainties as provided by the authors.
Panovska et al. (2012) using a cross validation approach to
assess random variability in sediment records. The results sug-
can be downward continued to map the radial magnetic field gest that in general uncertainties may have been underesti-
at the CMB. However, the accompanying amplification of mated in previous studies.
small-scale features and noise means that such activities A hypothesis testing approach using several different kinds
require appropriate caution when conducted on already uncer- of roughness penalty can be useful in evaluating whether par-
tain models. ticular features in a given model are robust (Panovska, 2012).
As with historical field modeling, activities on millennial- An alternative strategy is to explore the sensitivity of the model
scale data have gradually evolved from least-squares fits of the to the specific data used via bootstrap techniques (Korte and
data with low-degree spherical harmonic representations at Constable, 2011; Korte et al., 2009, 2011). Such evaluations
fixed or averaged intervals (Hongre et al., 1998; Johnson and tend to highlight the fact that there are large regions with few or
Constable, 1998; Márton, 1970) to regularized snapshot inver- no data, and others may have mutually inconsistent observa-
sions (Constable et al., 2000) and more recently to regularized tions, so that the reliability of the models can be quite variable
time-varying representations with the temporal variations for different times and places. It is worth noting that a flaw in
parameterized by cubic splines (e.g., Korte and Constable, this approach is that regions with no data will tend to produce
2003). The hazards of making spherical harmonic models very consistent minimum complexity results, even though the
with least squares and poor data coverage are well known and absence of data means that little is known about the field in
the results must be treated with appropriate caution. One strat- this region. Although regularized inversions will minimize the
egy in using least squares is to truncate the expansion at suffi- complexity in poorly sampled regions, it should be kept in
ciently low degree that the results are stable for the given data mind that even under the best circumstances, the data uncer-
distribution, determining the resolution a priori by truncation tainties and poor geographic coverage discussed in earlier sec-
level rather than from information in the data. The rationale for tions of this chapter make these models considerably less
regularized field modeling has been discussed elsewhere (see reliable than their historical or recent counterparts.
Parker, 1994), but loosely speaking, one can say that it recog- Table 4 catalogs some basic information about recent
nizes that there is some acceptable measure of misfit to the millennial-scale field models, derived from data compilations
observations and seeks a model that trades the misfit for finding discussed in Section 5.09.4. The earlier of these models are
a model with minimal complexity. The resolution is determined largely superseded by the more recent versions, as numbers of
by the data, and unnecessary structure, of the form specified by available data have increased notably and modeling strategies
the particular complexity or roughness penalty in use, will be have been refined. Models HHK, ALS3K, and CALS3k.1 are all
based on rather small numbers of smoothed data series, and models, ARCH3k.1 might give the best field predictions at the
they are limited by one or more of the following characteristics Earth’s surface for the areas of the northern hemisphere with
of modeling methodology: strong truncation of the spherical good data coverage but that it is not suitable for global studies.
harmonic series, no temporal continuity, and scaling of all The level of agreement between ARCH3k.1, the regional model
coefficients by a fixed or prescribed value of the axial dipole SCHA.DIF.3K, and the national reference curves in Figure 6
moment when no intensity information was used. All of the mostly supports this assessment, but considering the data
CALSxk.n models are constructed using the same inversion strat- uncertainties, it is not obvious that the fit to the data is in general
egy (described in Korte and Constable, 2003) with spatial regu- significantly better than that of CALS3k.4, for example, in the case
larization under the diffusion norm and temporal regularization of inclination at the Italian location around AD 1300.
of the second derivative of the field. The acronym stands for Several of the regional and global models are electronically
Continuous Archeomagnetic and Lake Sediment models span- available and links to relevant websites are listed in the
ning x ky and model version n. CALS3K.2 and CALS7K.2 were ‘Appendix.’ In most cases, some evaluation tools and/or ani-
the first models to consider each magnetic element (D, I, or B) at mations showing the field evolution for different field compo-
its appropriate place in time and space, rather than reducing to a nents are provided along with the model coefficients.
common epoch or reference location. These kinds of data sets
are not only more comprehensive than earlier ones but also
more heterogeneous in quality, and the heavy regularization 5.09.6 The Average Field and Its Secular Variation
will tend to produce models with considerably less structure on Millennial Timescales
than the real geomagnetic field. The early CALSxk.n models are
5.09.6.1 The Average Field
outdated now mainly due to the small numbers of underlying
data. Models ARCH3k.1 and SED3k.1 use the same modeling Figure 11 maps inclination anomalies (left) and declination
methodology as the CALS3k.n models but only archeomagnetic (right) at the Earth’s surface for temporal averages of the field
and only sediment data, respectively. Relative sediment intensity models gufm1, CALS3k.4, and CALS10k.1b over their respective
records have been scaled to absolute values by earlier models time spans. These averages reflect the differences in data distri-
here and in subsequent CALSxk.n models, and the recent model bution and quality among the models discussed in the preced-
ends have been constrained at different levels to agree with the ing text and the range of averaging times from 400 to 10 000
gufm1 model by Jackson et al. (2000) spanning four centuries years. The millennial-scale model directional anomalies are
from AD 1590 to AD 1990. These and later CALSxk.n models distinct from those found in gufm1: the temporal averaging
include uncertainty estimates based on bootstrap techniques, has a powerful effect, in not only reducing their scale but also
where the standard deviation is determined from a large number altering their distribution. The large-scale geographic varia-
of models using data sets varying in spatial distribution and tions in directions are broadly similar in structure for
within data and age uncertainties. To increase robustness and CALS3k.4 and CALS10k.1b, with longitudinal asymmetry in
reliability, at the cost of a loss in resolution, models CALS3k.4b both DI and D. Generally, negative inclination anomalies
and CALS10k.1b are averaged models from such a bootstrap emerge in equatorial regions as the averaging time lengthens,
approach, rather than realizations based on an individual data as one might expect for a field with a persistent axial quadru-
set. Panovska (2012) used the data compilation underlying pole contribution, but a positive anomaly persists in the Indian
CALS10k.1b, applied statistical uncertainty estimates to the Ocean near Australia and Indonesia. It is paired with a strong
sediment data (Panovska et al., 2012) and thus changed the negative anomaly in the Pacific region, and in animations of
weighting of these data in the modeling, modified the strategy the time-varying field, this can be seen to be a very active
for calibrating relative paleointensity records, and explored region. This is also an anomalous area for the average declina-
different regularization constraints. Comparatively, small differ- tion that is generally small aside from the polar regions (which
ences from CALS10k.1b confirm that the results of spherical are strongly influenced by the low horizontal field). In general,
harmonic millennial-scale models are mostly determined by CALS10k.1b has larger directional anomalies than the 3 ky
data numbers, distribution, and quality rather than differences model, associated with the diminished magnitude of g01 in
in modeling methodology. the early part of the record (see Figure 10).
Nearly all spherical harmonic models show roughly similar Figure 11 provides a useful means of assessing the size of
features but differ in various details as a consequence of their directional anomalies expected in various regions, but a more
individual properties, with data distribution and level of smooth- direct view of the average geodynamo including poloidal field
ing being the strongest contributors to both regional differences strength is given by taking account of the field strength F and
as seen, for example, in Figure 6, and discrepancies in the tem- the radial component of the field, Br. Figure 12 shows the field
poral evolution of Gauss coefficients. Dipole and quadrupole strength at the Earth’s surface (left) and the radial field com-
coefficients of several models are shown in Figure 10. Offsets ponent after downward continuation to the CMB (right). Note
between CALS7K.2 and CALS10k.1b coefficients, in particular that both gufm1 and CALS10k.1b are substantially weaker on
prior to 1000 BC, can be attributed to a lack of intensity infor- average than CALS3k.4, in the first case reflecting the recent
mation in the earlier model, where only absolute archeomagnetic decrease in the axial dipole and the latter the lower values
intensity data were included (cf. Figure 8(a)). Notable differences during the interval 8000–1500 BC. The millennial-scale
in the ARCH3k.1 coefficients compared to the other models result models suggest the presence of quasi-stationary high-latitude
from insufficient global data coverage due to the scarcity of magnetic flux patches in both hemispheres, as seen in gufm1
southern hemisphere archeomagnetic data (cf. Figure 7(a)– and other historical models, but the morphology is not con-
7(c)). Korte et al. (2009) considered that among the global sistent across all averaging times. Experiments with numerical
4
-20 2
-25 0
g11 [ mT]
g10 [mT]
-30 -2
ARCH3k.1 -4
-35
CALS3k.3 CALS3k.4
-40 -6
CALS7k.2 CAL10k.1b
1
4 0
-1
0 -2
g20 [mT]
h11 [mT]
-4 -3
-4
-8 -5
4
4
2 2
h21 [mT]
g21 [mT]
0 0
-2
-2
-4
4
2
2
0
g22 [mT]
h22 [mT]
0
-2
-2
-4 -4
-8000 -4000 0 -8000 -4000 0
Year Year
Figure 10 Comparison of degree l ¼1, 2 Gauss coefficients for field models CALS10k.1b (blue), CALS7K.2 (gray), ARCH3k.1 (black), CALS3k.3
(brown), and CALS3k.4 (red). All coefficients are given in microtesla at the Earth’s surface.
dynamo simulations produce similar time-dependent and boundary achieved with the whole CALS10k.1b data set is
time-averaged field morphology when a heat flow related to illustrated in Figure 14 and compared to that available for
lower mantle seismic tomography is imposed at the CMB the earlier CALS7K.2 model. The figure shows enormous
(Bloxham, 2002; Olson and Christensen, 2002), lending sup- improvement in geographic sampling since the CALS7K.2
port to the idea that these patches are the result of core flows model was produced, although the bias to midnorthern lati-
influenced by thermal interactions between the core and tudes remains.
mantle. The South Atlantic Anomaly is notably absent in the pre-
In the southern hemisphere, the millennial models have gufm1 models, and the flux lobes are highly attenuated. The
generally lower-average field values than in the northern hemi- lowest field strength is no longer in the Atlantic hemisphere,
sphere, and it should be questioned whether this is a conse- but (Figure 12) lies in the highly active region near Indonesia
quence of data distribution giving rise to poor model and Australia.
resolution or a sign of genuine persistent north–south hemi- The absence of the South Atlantic Anomaly might be taken
spheric asymmetry reflected, for example, in the generally neg- as an indication that this is a transient feature, but we should
ative values for g02 in Figure 10. Linearized data kernels for keep in mind the limitations imposed by the temporal and
changes to an axial dipole field structure shown in Figure 13 spatial distribution of the available data, especially in the
give some guidance on how geomagnetic data elements sam- southern hemisphere. There are very few archeomagnetic direc-
pled at various locations are sampling structures at the CMB. tional or intensity data to constrain the field in this region
From these, it can be seen that even though the data coverage is (Figure 7). Sediment directional data in general have relatively
spartan, large-scale features at the CMB will contribute to sur- poorly constrained age scales compared with the best arche-
face observations at distant locations, rather than just to those omagnetic records, and the age constraints for the African data
directly above. The relative sampling of the core–mantle are the weakest. In many cases, the largest contribution to the
16
12
8
4
0
-4
-8
-12
-16
gufm1 gufm1
16
12
8
4
0
-4
-8
-12
-16
CALS3k.4 CALS3k.4
16
12
8
4
0
-4
-8
-12
-16
CALS10k.1b CALS10k.1b
Figure 11 Inclination anomalies (left) and declination (right) for temporally averaged field models gufm1, CALS3k.4, and CALS10k.1b. Averaging
times are 400, 3000, and 10 000 years, respectively.
data uncertainty comes from poor age constraints (and associ- radial component of the magnetic field at the Earth’s surface
ated secular variation in the magnetic field), and the assigned is plotted after removing the contribution due to the axial
uncertainties in the observations average 4.0 for inclination, dipole part of the field. This emphasizes geographic variations
6.7 for declination, and 11$\m$T for intensity. The regularized in structure that represent departures from the geocentric
inversion results in a trade-off between misfit to the data and axial dipole field. The different maps in Figure 15 represent
minimum complexity in the model, and with these large averages over vastly different timescales. The two panels on
uncertainties, we expect the model to lack both temporal and the left side are for AD 2010 and the average over the past 400
spatial resolutions. The experimental procedures used for years, while on the right, we have the 3 ky and 10 ky average
deriving intensity data lead to uncertainties in the form of models. There are broad similarities in structure for the two
percentage errors and are usually larger for high field values. left-hand models. The lower panel looks like an attenuated
This can lead to a small systematic bias to positive values in the version of the upper left, as might be expected when small-
distributions of intensity residuals from regularized models. scale rapidly changing features are averaged over time. The
For example, in the case of CALS7K.2, the bias was reported same holds true when the two right-hand panels are com-
as about 3.6 mT (Korte and Constable, 2005a). For the direc- pared with one another, but these longer-term averages have
tional data, there are no significant biases in the residual features that are quite distinct from the historical record. On
distributions. the longer time intervals, the largest contributions appear to
The low average intensity structure extending from the be latitudinal variations, but there is also a systematic non-
western equatorial Pacific to East Africa in Figure 12 should zonal contribution.
also be viewed with some caution, although if it is not
required by the observations, it should have been eliminated
5.09.6.2 Dipole Moment and Axis Variations
by the inversion strategy using quadratic regularization. It is
possible that it is real geomagnetic behavior, and this view is Spherical harmonic paleofield models derived from both
supported by comparisons made in Figure 15, where the directional and intensity data allow the recovery of a direct
90 700
85 600
80 500
75 400
70 300
65 200
100
60 0
55 -100
50 -200
45 -300
40 -400
35 -500
30 -600
25 -700
gufm1 gufm1
90 700
85 600
80 500
75 400
70 300
65 200
100
60 0
55 -100
50 -200
45 -300
40 -400
35 -500
30 -600
25 -700
CALS3k.4 CALS3k.4
90 700
85 600
80 500
75 400
70 300
65 200
100
60 0
55 -100
50 -200
45 -300
40 -400
35 -500
30 -600
25 -700
CALS10k.1b CALS10k.1b
Figure 12 Strength of the magnetic field at the Earth’s surface (left) and radial field component downward continued to CMB (right) for temporally
averaged field models gufm1, CALS3k.4, and CALS10k.1b. Averaging times are 400, 3000, and 10 000 years, respectively.
(c) Declination 60 degrees (f) Inclination 60 degrees (i) Intensity 60 degrees

−4
´ 10
1.2
1
0.8
0.6
0.4
0.2
0
−0.2
−0.4
−0.6
−0.8
−1
−1.2
(b) Declination 30 degrees (e) Inclination 30 degrees (h) Intensity 30 degrees
(a) Declination 0 degrees (d) Inclination 0 degrees (g) Intensity 0 degrees
Figure 13 Data kernels for individual samples of D, I, and B at various locations.

0.7 0.7
0.63 0.63
0.56 0.56
0.49 0.49
0.42 0.42
0.35 0.35
0.28 0.28
0.21 0.21
0.14 0.14
0.07 0.07
0 0
F, kernel sum, CALS10k F, kernel sum, CALS7k
0.8 0.8
0.72 0.72
0.64 0.64
0.56 0.56
0.48 0.48
0.4 0.4
0.32 0.32
0.24 0.24
0.16 0.16
0.08 0.08
0 0
I, kernel sum, CALS10k I, kernel sum, CALS7k
1.65 1.65
1.5 1.5
1.35 1.35
1.2 1.2
1.05 1.05
0.9 0.9
0.75 0.75
0.6 0.6
0.45 0.45
0.3 0.3
0.15 0.15
0 0
D, kernel sum, CALS10k D, kernel sum, CALS7k
2.4 2.4
2.2 2.2
2 2
1.8 1.8
1.6 1.6
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
All kernel scaled sum, CALS10k All kernel scaled sum, CALS7k
Figure 14 Comparison of the sampling of the CMB achieved by the CALS10k.1b and CALS7k.2 data sets.
estimate of the geomagnetic dipole moment and poles rather intensity data in combination with data uncertainties and the
than relying on the proxy obtained by averaging VADMs and regularization scheme (Korte and Muscheler, 2012; Korte
VGPs. Figure 16(a) shows the dipole moment variation for et al., 2011).
four of the models described earlier in the text and the VADM The short-term variability shown by the different reconstruc-
reconstruction by Knudsen et al. (2008). The dipole moment tions depends on the averaging interval for the VADM curve, on
for CALS7K.2 is lower than the VADM estimates by about 19%, the choice of regularization parameter for the spherical har-
while more recent spherical harmonic model results mostly lie monic models in general, and on a smoothing effect of the
somewhere in between. Korte and Constable (2005b) treatment of the sediment series in the bootstrap averaging
considered VADM proxies from presently available archeomag- applied for CALS10k.1b. The high variability shown by
netic data distributions to be biased high due to nondipole CALS3k.3 and CALS3k.4 is required by the data set, unless the
field effects arising from insufficient global coverage by inten- treatment of uncertainties in the modeling is inadequate. It is
sity data and data quality influences. It is possible that inade- compatible with the information we can gain via model gufm1
quate cooling rate corrections play a role here. The dipole from the time series of direct intensity observations going back
moment of CALS7K.2 is considered to be biased low, in par- no further than 1840. In fact, it is unclear how long the dipole
ticular prior to 1000 BC, due to the underrepresentation of moment has been decreasing at nearly linear rate without
24 24
21 21
18 18
15 15
12 12
9 9
6 6
3 3
0 0
−3 −3
−6 −6
−9 −9
−12 −12
−15 −15
−18 −18
−21 −21
−24 −24
(a) IGRF, 2010 AD non-axial-dipole Br at r = a (d) CALS3k.4, 0−3 ka average non-axial-dipole Br at r = a
24 24
21 21
18 18
15 15
12 12
9 9
6 6
3 3
0 0
−3 −3
−6 −6
−9 −9
−12 −12
−15 −15
−18 −18
−21 −21
−24 −24
(b) GUFM1, 1590−1990 AD average non-axial-dipole Br at r = a (c) CALS10k.1b, 0−10 ka average non-axial-dipole Br at r = a
Figure 15 (a) Vertical component of the non-axial-dipole field in microtesla evaluated at the Earth’s surface (r ¼6371.2 km) using the geomagnetic
field models IGRF for AD 2000, (b) gufm1 averaged over 400 years, (c) CALS3k.4 averaged over 3 ky, and (d) CALS10k.1b averaged over 10 ky.
interruption, as illustrated by Figure 16(b). Model CALS7K.2 A comparison of different reconstructions in Figure 17 shows
suggested substantial deviations from the linear trend around that agreement is variable over time and uncertainties are large.
1800, and Gubbins et al. (2006) used archeointensity data The present dipole tilt of about 11 is not extraordinarily large if
combined with the historical directional data used for gufm1 uncertainty estimates are taken into account. In particular, it is
to confirm that the original gufm1 extrapolation was inaccurate. likely that CALS10k.1b underestimates the magnitude of past
Finlay (2008) created a variety of gufm1-like models incorporat- tilts as a consequence of limited temporal resolution.
ing archeomagnetic intensity data. Depending on the weighting
of the archeomagnetic data, he found a wide range of decay
5.09.6.3 Jerks, Drifts, and Waves
rates, including no decrease at all, which was given as the
preferred solution. Suttie et al. (2011) investigated a large set The westward drift of geomagnetic features noted by Halley
of recent archeointensity data in detail and concluded that during the seventeenth century is widely interpreted to reflect
systematic error in the data can lead to underestimation of the aspects of the fluid motion at the surface of the Earth’s
field magnitude and that the decay rate from 1590 to 1840 was core (Bullard et al., 1950; Yukutake and Tachinaka, 1969).
similar to that after 1840, in rough agreement with predictions Two possible dynamical sources for these effects have received
by recent CALSxk.n model versions. Differences in absolute level attention: The first discussed by Braginskiy (1972, 1974) is the
among CALS3k.3, CALS3k.4, and CALS10k.1b are indicative of superposition of propagating MAC (magnetic–Archimedean–
substantial uncertainties in the calibration of relative paleoin- Coriolis) waves at the core surface resulting in phase propaga-
tensity records, which will have to be investigated further. tion of the field and reflecting a balance among Lorentz,
Despite these differences, all reconstructions agree that Coriolis, and buoyancy forces; the second interpretation is
the dipole moment was clearly higher than at present that the drifts of the field reflect mean azimuthal flow at the
between 1000 BC and AD 1000 and about 30% lower than core surface. Core surface flows have been mapped under
the maximum values for at least the time interval 7000–3000 BC. various constraints about the field evolution in addition to
Although nonnegligible uncertainties in the exact level of dipole the frozen flux approximation, which specifically requires
moment and variability remain, the current decrease in dipole that the magnetic field is advected with the material flow.
moment does not appear anomalous compared with other Holo- Conservation of angular momentum also provides a firm
cene field variations. Constable and Korte (2006) used this and theoretical foundation linking decadal geomagnetic secular
other data to argue that a geomagnetic reversal cannot be consid- variation to changes in the length of day (LOD). These changes
ered imminent. Gubbins et al. (2006) had tied the historical are linked to torsional oscillations in the core fluid as it adjusts
decrease in dipole moment to the growth of the low-field South to small departures from the Taylor state and have been tied to
Atlantic Anomaly and associated north–south hemispherical the sharp accelerations in geomagnetic field records known as
asymmetry. Any extension of such analyses for the paleofield geomagnetic jerks (Bloxham et al., 2002). No such firm theo-
will need to take careful account of geographic variations in retical footing exists to link longer-term changes in LOD to
resolution and accuracy of millennial-scale global field models. millennial-scale core flows, as there is little justification for
Nilsson et al. (2011) discussed a potential 1350-year cyclicity supposing that the core motions are without axial shear over
in the variation of the dipole axis, in particular the dipole tilt. these longer timescales. Nevertheless, Dumberry and Bloxham
12
10
1022 Am2
6
4 CALS10k. 1b
CALS3k.4
CALS3k.3
2 CALS7k.2
VADM
−8000 −6000 −4000 −2000 0 2000

(a) Year
12
11
10
9
1022 Am2
7
VADM
CALS3k.3
6 CALS3k.4
CALS10k. 1b
Gufm1
5
Dashed line: Finlay (2008)
Gubbins et al. (2006)
4
1000 1200 1400 1600 1800 2000
(b) Year
Figure 16 Geomagnetic dipole evolution during (a) the past 10 000 years and (b) the past 1000 years, as predicted by spherical harmonic global
models CALS7K.2 (gray), CALS3k.3 (purple), CALS3k.4 (brown), and CALS10k.1b (blue) and the global VADM reconstructions by Knudsen et al. (2008)
(red). Panel (b) additionally shows the dipole moment of model gufm1 ( Jackson et al., 2000) for historical times with the original extrapolation
between 1840 and 1590 (solid black line), the preferred solution by Finlay (2008) (dashed black line), and the extrapolation suggested by Gubbins
et al. (2006) (green). The recent CALSxk.n models are designed to agree with gufm1 from 1840 onward. All uncertainty estimates were
obtained by bootstrap resampling.
90
88
86
84
Latitude (deg)
82
80
78
76
74
CALS10k. 1b
72 Nilsson et al. (2010)
CALS7k.2
70
−7500 −5000 −2500 0
Year
Figure 17 Tilt of the dipole axis as determined from the geomagnetic pole latitude of spherical harmonic global models CALS7K.2 (gray) and
CALS10k.1b (blue) and the global virtual geomagnetic pole (VGP) reconstruction by Nilsson et al. (2010) (red). Uncertainty estimates from bootstrap
resampling for CALS10k.1b and from the averaging, reflecting the spread of data, for the VGP curve.
(2006) had made some substantial preliminary efforts to assess Gallet et al. (2003) had recently drawn attention to coinci-
the contributions of millennial-scale geomagnetic field dent features in the Western European archeomagnetic direc-
changes to the longer-term changes in LOD now documented tional secular variation curve and archeomagnetic intensity
back to 2700 BP (Morrison and Stephenson, 2001; Stephenson changes in France and the Middle East. Changes in curvature
and Morrison, 1995; Stephenson et al., 1984). in the directional variations on Bauer plots are inferred to
Studies of regional westward drift in distinct paleomagnetic occur at the same time as local maxima in intensity variations.
records are generally limited by inaccuracies in the assigned They call these coincident features archeomagnetic jerks and
timescales for records that are separated by a few thousand raise questions about their regional versus global significance
kilometers or less (e.g., Lund, 1996). In more distant records, and whether such relationships are a general characteristic of
the identification of the same magnetic features at different short-term geomagnetic field variations. It should be noted
times has proved challenging, probably reflecting more com- that despite the similarity in terminology, they should not be
plex variations than simple westward drift. An assessment of thought of as similar to the geomagnetic jerks observed by
drift in individual paleomagnetic records has often been made direct observations for the most recent century. Instead as out-
on the basis of the interpretation of Bauer (1895) plots, gen- lined later in the text, these archeomagnetic jerks seem to be
erally of declination versus inclination centered on the axial associated with broad regional changes in directions of drift
dipole or mean field direction at a site. Occasionally, the path and wave motions as recently shown in a global assessment by
traced by the VGP is used. According to the rule of Runcorn Dumberry and Finlay (2007).
(1959), clockwise motion is taken to represent westward drift The advent of time-varying millennial-scale geomagnetic
of an underlying magnetic source, although it is well known models enables analyses of drift to move from a regional to a
that this interpretation is nonunique (e.g., Dodson, 1979; global context, and there have been several investigations of
Skiles, 1970). Nevertheless, the general idea can be supported this type to investigate the predominant direction of drift
by an analysis of the gufm1 historical field model shown in (Dumberry and Finlay, 2007; Wardinski and Korte, 2008)
Figure 18. Regional VGP trajectories mainly show clockwise and to determine whether oscillating mean azimuthal flows
motion, and the motion is largest in amplitude in the regions are the predominant cause of the observed drift (Dumberry
where westward drift of the magnetic field is most pronounced. and Bloxham, 2006). Dumberry and Finlay (2007) used the
Local application of Runcorn’s rule has led to numerous global time-varying CALS7K.2 model to conduct a global study
reports of eastward drift in millennial-scale magnetic records of episodes of eastward and westward drift during the past 3 ky.
(e.g., Constable and McElhinny, 1985; Snowball and They found both eastward and westward motion at mid to
Sandgren, 2002). high latitudes in the northern hemisphere, corresponding to
180
270 90 1590.0 − 1990.0 VGP tracers

70N

Figure 18 VGP trajectories calculated on a 15 geographic grid using field predictions from gufm1. Pluses indicate the geographic location for which

the VGP is calculated and form the center for each local VGP projection with scale as at upper left, where 70 N is the inner circle and 50 N the outer
one. Paths are for the time interval AD 1590–AD 1990 with arrowheads indicating the position in 1990. Note the broad clockwise looping in
the Atlantic/African region.
displacements and changes in the two major quasi-stationary with the limited temporal and geographic resolution achieved
high-latitude magnetic flux patches. Poor resolution in that in the current time-varying global paleomagnetic models.
model may prevent similar motions from being discovered in
the southern hemisphere (Korte and Holme, 2010). Dumberry
and Finlay (2007) noted that the direction changes are associ-
5.09.7 The Geomagnetic Spectrum
ated with the times of the archeomagnetic jerks identified by
Gallet et al. (2003): These direction changes are moderately
The temporal and spatial power spectra are two statistical
rapid (centennial timescales), but there is no evidence for a
properties of the geomagnetic field that are important, for
sharp change in the geomagnetic field. Dumberry and
example, because they should be reproducible by adequate
Bloxham (2006) showed that the observed drift is consistent
numerical simulations of the geodynamo (Kono and Roberts,
with the motions being caused by advection of magnetic field
2002; Kono et al., 2000; McMillan et al., 2001). Korte and
features by azimuthal flows, although it is worth noting that
Constable (2005a) noted that both the spatial power spectra
their arguments clearly spell out that this is a plausible rather
Rl(t) and the secular variation spectrum Sl(t) for CALS7K.2
than a unique interpretation of the observations. The quality of
have greatly diminished power above spherical harmonic
the CALS7K.2 model made it difficult to determine whether
degree 4 or 5 compared with the present field. This is also
westward drift is a persistent feature of the field for 0–3 ka. In a
true for all subsequent models and reflects their limited spa-
more recent analysis using CALSSk.3, Amit et al. (2011) used
tial and temporal resolution, imposed by the distribution and
an algorithm to detect centers of intense flux patches and track
quality of the data through the regularization in the modeling
their time evolution, finding that quasi-stationary periods with
method. The long-term temporal variation of Rl with time
small oscillations of the patches appear to occur more often
(here designated Pl) can be measured by the variance of the
than drifts. Average patch lifetimes of 300 years are found and
Gauss coefficients about their mean value, yielding informa-
both eastward and westward drifts are detected in that study,
tion about the power in the secular variation for comparison
which concluded that results were compatible with the idea
with long-term (million-year) statistical models of paleosecu-
that the time-averaged zonal core flow may be driven by ther-
lar variation (PSV). For CALS7K.2, we estimate Pl directly
mal core–mantle coupling producing both the high-latitude
from the standard deviation in the estimated Gauss
flux lobes and a regime producing westward drift in the
coefficients,
Atlantic hemisphere. However, the westward drift seen in the
millennial-scale models should be distinguished from that l
X 2 2
seen in the modern field, which tends to be dominated by Pl ¼< sB! sB! > r a ¼ ðl þ 1Þ sgml þ shml
l l
m¼0
rapid motions in equatorial regions: these cannot be resolved
Power spectra of PSV models at Earth’s surface
100
108
CJ98 10−1
Reversal
107 CALS7K.2 variance rate changes
TK03.gad
10−2 Average reversal,
106 OSVM Sec Var ′n crypto-chron and
PSD (x1044 A2m4) Myr
excursion rate
10−3
Power (nT2 or (nT/yr)2)
105
10−4
PSV, lengths
104 of reversals and
10−5 excursions
103
10−6
102
10−7
101
10−8
100
10−9
2 4 6 8 10 12 10−2 10−1 100 101 102 103 104
(a) Spherical harmonic degree (b) Frequency Myr−1
Figure 19 (a) Spatial power spectrum showing the variance as a function of spherical harmonic degree for PSV models CJ98 and TK03 and for
CALS7K.2. Spectrum of secular variation from OSVM is also shown. (b) Composite paleomagnetic spectrum for dipole moment variations (reproduced
from Constable C and Johnson C (2005) A paleomagnetic power spectrum. Physics of the Earth and Planetary Interiors 153: 61–73, http://dx.doi.org/10.
1016/j.pepi.2005.03.015): the spectrum is derived from the 0–160 Ma reversal record (CK95 (black)), 0–83 Ma reversal record including cryptochrons
(CK95cc (gray)), and various marine sediment records from site 522 (blue), VM93 (red), site 983 (green), site 984 (brown), Sint-800 (orange), and
CALS7K.2 (pink). See text for details.
90
PEP, LEB STL
80 SAR,FRG,POH,NAU,BYE BJO, GAR

MEZ BIR BAI
ON1−ON4 BAR, EAC
70
PAD,LAR ESC
60 −g01
mT
50
40
30
20
10
−5000 −4000 −3000 −2000 −1000 0 1000 2000
Year
Figure 20 Predictions from CALS7K.2 of field intensity variations at a variety of locations for which relative intensity records are available. See Korte
and Constable (2006a) for details.
where Pl is shown in Figure 19 along with the equivalent CALS7K.2 (pink). CALS7K.2, which is the highest-frequency
spectra for two candidate PSV models for the time interval contributor to the paleomagnetic power spectrum, fits well
0–5 Ma, CJ98 and TK03.gad, of Constable and Johnson in the frequency overlap with the spectrum derived from
(1999) and Tauxe and Kent (2004), respectively. Both TK03. the high-sedimentation-rate marine sediments at the North
gad and CJ98 are based on the premise that the spectrum of Atlantic sites 983 and 984. The individual 983 and 984 records
geomagnetic variations is essentially flat at the CMB for l >2, show substantially more power than the global average Sint-
and they are quite similar in this range. But otherwise there is 800, which is to be expected from dating uncertainties among
rather poor agreement among the three models. The differ- the cores contributing to the stack. Further comparisons are
ences in CALS7K.2 for l >5 can be readily explained by the needed with records from a broader range of geographic areas
limited resolution in CALS7K.2. But both 0 – 5 Ma models for the older part of the record and with historical field models
have more power in quadrupole variations than CALS7K.2, a like gufm1. Korte and Constable (2006a) had undertaken a
feature that is worth further investigation. Also plotted in more detailed study of the frequency content of higher-degree
Figure 19(a) is Sl for the OSVM model for epoch AD 2000. spherical harmonic terms in CALS7K.2. The average temporal
OSVM has most power in the secular variation at l ¼2 and resolution of CALS7K.2 is no better than a few centuries, but a
diminished power in the dipole. The low dipole power is to study of higher-frequency dipole moment variations in more
be expected from the longer time constant associated with recent models has not yet been completed.
dipole variations. But in CALS7K.2, P2 drops below the general
trend, contrasting strongly with CJ98 and OSVM. The disagree-
ment among these models highlights our lack of knowledge 5.09.8 Applications
about long-term (large-scale) field variations. If the low power
5.09.8.1 Calibrating Relative Geomagnetic Intensity
at l ¼2 in CALS7K.2 proves to be robust in future modeling
Variations
efforts, this may be an important feature of long-term geomag-
netic secular variations. For several decades, researchers have been acquiring records of
The best-known part of the temporal spectrum of variations relative variations in geomagnetic field strength from lacustrine
is that due to dipole moment variations, and on very long and marine sediments (e.g., Constable and McElhinny, 1985;
timescales, even that is rather poorly known. Figure 19(b) Levi and Banerjee, 1976; Tauxe, 1993). Calibrating these data
(after Constable and Johnson, 2005) shows a composite spec- to absolute field strengths remains a thorny problem
trum derived from the Cande and Kent (1995) 0–160 Ma (Constable and Tauxe, 1996; Valet, 2003), and most current
reversal record (the CK95 (black)) and 0–83 Ma reversal record strategies rely on the idea that average global VADMs provide a
including cryptochrons (CK95cc (gray) portion) and various suitable dipole moment for scaling individual records. The
marine sediment records from site 522 (blue; Constable et al., CALSxk.n models can provide a direct prediction of intensity
1998), VM93 (red; Valet and Meynadier, 1993), site 983 variations at any geographic location for the past 7 ky, and it
(green; Channell et al., 1998), site 984 (brown; Channell, can be seen from the CALS7K.2 example in Figure 20 that there
1999), Sint-800 (orange; Guyodo and Valet, 1999), and are substantial geographic variations. The gray curve shows the
Paleoclimate studies
350
300
250 After stuiver et al. (1998)

D14C (per mil)
200
150
100
50
−14000 −12000 −10000 −8000 −6000 −4000 −2000 0 2000 D14C is converted to normalized production rate
Year BP
Q/Q0. Theoretical relations among normalized Q,
M and solar modulation are then used to drive
relative variations in solar flux.
1.6
Muscheler et al. (2005)
1.4
1.2
Q/Q0
1.0
0.8
0.6
–10000 –8000 –6000 –4000 –2000 0 2000

Calendar age (−BC/+AD)
Figure 21 Reconstruction of atmospheric concentration of 14C for the past 10 ky (top). This is converted to a normalized production rate like that given
by Muscheler et al. (2005) in the lower part of the figure.
variation expected from the axial dipole field as g01 and cor- increasing latitude) as shown in the early work by Elsasser
responds to the intensity expected at the equator from that et al. (1956) who used neutron monitor data and archeomag-
term alone. Comparisons of these predictions with relative netic observations made by Thellier to initially infer an approx-
intensity time series then allow a calibration that can take imately inverse square root relationship between magnetic
account of nondipole field contributions that have not been dipole moment and nuclide production rate in the upper
averaged out over millennial timescales. Korte and Constable atmosphere. This led to the idea that reconstructions of past
(2006b) carried out this calibration for a series of 22 distinct atmospheric radiocarbon concentrations might be used to
records and concluded that many of them contain significant derive a so-called radiocarbon dipole moment as a means of
information about geomagnetic paleointensity variations. Cal- studying past geomagnetic field behavior (Merrill et al., 1996).
ibration of relative paleointensity records by scaling to pre- Changes in solar activity (i.e., modulation of the so-called solar
dictions from a starting model has been applied in the constant) affect the incident cosmic ray flux and are also sig-
derivation of all subsequent CALSxk.n models. nificant in affecting production rates in the upper atmosphere:
More recently, Panovska (2012) and Ziegler et al. (2011) such changes are well known for the 11-year solar cycle
had explored with some success the strategy of direct calibra- variations and have been inferred at longer periods using prox-
tion of relative paleointensity data as part of the inversion ies such as sunspot numbers.
process used for constructing time-varying models. This is an In situ production of 14C, 10Be, and 36Cl can be measured in
area that would benefit from further work. ice cores (e.g., Lal et al., 2005; Muscheler et al., 2004b), and at
high latitudes, the variation in cosmic ray flux is minimally
influenced by the geomagnetic field, making these records
5.09.8.2 Cosmogenic Isotope Production Rates: The Global most sensitive to solar modulation. Lower-latitude proxy
View Linking Geomagnetic and Climate Studies records drawn from tree rings, corals, speleothems, and sedi-
It is well known that the geomagnetic field acts as a shield ment cores (e.g., Laj et al., 2002; Reimer et al., 2004; Snowball
against cosmic radiation, preventing particles with insufficient and Sandgren, 2002) are influenced by the geomagnetic field
energy from penetrating to the Earth’s atmosphere and pro- and are also affected by the Earth’s climate, making the recon-
ducing cosmogenic nuclides such as 14C, 10Be, and 36Cl. The struction of the paleoproduction rates a nontrivial task. The
shielding effect is latitude-dependent (diminishing with individual isotopes have distinct chemistries with differing
residence times and mixing effects in the atmosphere, ocean, or events may be related to the mechanism for centennial-scale
sediment sources so that the measured series may reflect quite climate change; in particular, they seem to view geomagnetic
different response to climate variations and recording variations as a potential trigger for climate change, suggesting
environment. that the geomagnetic field may have a smaller axial dipole
Reconstructions of past atmospheric radiocarbon concen- component during archeomagnetic jerks. The connection to
tration are usually presented as D14C defined as per mil devi- centennial climate change would come from changing interac-
ations from the National Institute of Standards and tions between the internal geomagnetic field and the solar
Technology 14C standard, after correction for decay and frac- wind. This view that the geomagnetic field might be a partial
tionation. Figure 21 (upper part) shows the variation with driving force for climate change is opposite to one suggested
time for the past 10 ky based on the terrestrial dendrochrono- for longer timescales (Yokoyama and Yamazaki, 2000), where
logical data set compiled for IntCal98 (Stuiver et al., 1998). the possibility has been raised that long-term (around 100 ky)
One of the major influences on this concentration curve is the orbital changes influencing climate may provide a partial
change in the Earth’s dipole moment over the past 10 ky, but energy source for the geodynamo. A clear need exists for
this is far from the only one. Changes in solar activity are also improved understanding and physical modeling that could
significant in affecting production and are clearly apparent in support the perceived correlations in this area.
the highest-frequency variations in Figure 21. 14C produced in
the atmosphere also plays an important role in the global
5.09.8.3 Surface Exposure Dating and Local Variations
carbon cycle as it exchanges between the atmosphere, bio-
sphere, and ocean. The ocean is the largest reservoir in the The calculation of cosmogenic nuclide production rates is also
14
C cycle, with a variable climate-driven residence time of the important in surface exposure studies that focus on processes
same order as the half-life for 14C. The conversion of atmo- operating on timescales from 103 to 105 years (Gosse and
spheric concentration to the normalized production rate in the Phillips, 2001), usually involving the production of radioiso-
lower panel is achieved using a box diffusion carbon cycle topes on or near the surface of rocks that have either been
model (Siegenthaler et al., 2006) that is essentially constant exposed since their formation or recently uncovered. The expo-
over time – considered a reasonable approximation for much sure time can be inferred from careful study of an appropriate
of the Holocene (for details, see Muscheler et al., 2005). radioisotope. A major component of these studies involves
The interplay among climate, geomagnetic intensity, and determining the geographic distribution of cosmic radiation
solar modulation is complicated, and it is unlikely that cosmo- and how to account for scaling effects on the production due to
genic isotopes alone can be used to unravel details of past losses in the atmosphere. First, scaling models were proposed
geomagnetic field behavior. However, paleomagnetic studies by Lal (1991) and allowed the calculation of production
provide an important independent record of field strength rates as a function of geomagnetic latitude and altitude.
variations that can be used to model the geomagnetic influence Subsequently, attention has focused more on geographic vari-
on production rate by studying relationships among produc- ability and the effects of changes in geomagnetic field strength
tion rate, solar flux, and dipole moment (e.g., Masarik and and nondipolar magnetic field structure on local production
Beer, 1999; Solanki et al., 2004; Usoskin et al., 2006). Previ- rates (Dunai, 2001; Lifton et al., 2005; Pigati and Lifton, 2004).
ously, corrections have been largely based on 500- or 1000- The most recent of these studies include the idea of using
year averages of the VADMs, like those published by Yang et al. cosmic ray tracing (Shea et al., 1965) and geomagnetic field
(2000), but, for example, Usoskin et al. (2006) used spherical models to determine cumulative surface exposure rates. It
harmonic model dipole moments instead. These show sub- remains to be seen whether millennial-scale magnetic field
stantial changes over 1000 years and less. The dipole moment models can significantly improve the accuracy in Holocene
results presented in Figure 16 point to the need for continued surface exposure dating.
improvements in temporal resolution and accuracy for
millennial-scale geomagnetic field models like CALS10k.1b.
5.09.8.4 Historical Applications: Building Orientations,
These are especially important for paleoclimate studies that
Navigation, and Auroras
make use of production rates of cosmogenic nuclides (14C,
10
Be, and 36Cl) in the upper atmosphere to infer the solar The past motion of the geomagnetic dipole axis can be mapped
flux and associated changes in insolation (e.g., Damon and from Holocene field models, and this is of some interest to
Peristykh, 2004; Muscheler et al., 2004a) or that use estimates scholars who find references to auroral sightings in medieval
of the cosmic ray-induced ionization of the troposphere and ancient texts (Siscoe and Verosub, 1983; Silverman, 1998;
(e.g., Usoskin et al., 2008, 2010). Willis and Stephenson, 2001). Several archaeological studies
It is of some interest to note that relative paleointensity (Charvátová et al., 2011; Downey, 2011; Klokoĉnı́k, 2011;
variations derived in several Holocene studies have been linked Klokoĉnı́k et al., 2007) have used paleomagnetic regional sec-
to millennial- and even centennial-scale variability within cos- ular variation curves or the CALSxk.n models to assess the
mogenic isotope records (Snowball and Sandgren, 2002; possibility that ancient buildings in China, Greece, and Crete
St-Onge et al., 2003), raising the possibility of links between were oriented using a magnetic compass, and this has some
climate change and geomagnetic variability, with either one bearing on the early history of the use of the magnetic com-
possibly serving as the driving force. On a regional level, Gallet pass. Historians interested in mapping the paths taken by early
et al. (2005) also proposed that the so-called archeomagnetic maritime explorers have compared the records from early
jerks in the European archeomagnetic record can be further ships’ logs with the declination values predicted by the models
associated with cooling events in Western Europe and that such (Keith Pinkerton, personal communication).
5.09.8.5 High-Resolution Magnetostratigraphy • Data remain unevenly distributed in time and space and of
mixed quality.
From the discussion of local and regional secular variation
studies, it should already be clear that paleointensity and direc-
• Resolution in current models remains poor beyond degree
four or five resulting in steep drop-offs in the spatial power
tional changes in high-accumulation-rate sediments are con-
spectra.
sidered useful for high-resolution magnetostratigraphy and
that regional secular variation curves based on archeomagnetic
• Power in the secular variation is underestimated because of
limited temporal resolution.
and volcanic data find widespread application in providing age
constraints, for both lake and archeological studies. A number
• Dipole moments derived from average VADMs for the
recent millennia appear to have been overestimated.
of studies have continued to exploit the PSV record to unravel
the eruptive history of southern Italian volcanoes. Although
• The secular variation at degree two has lower power in
CALS7K.2 than might be anticipated from both longer-
the timing of eruptions is well documented in the historical
term and present field secular variation models.
record, in some cases, their locations are poorly known, lead-
ing to misidentification of specific flows and resulting in tim-
• The 3000- and 10 000-year averages for the non-axial-
dipole field differ substantially from both the present field
ing errors in the secular variation record. Recent work by
and the average for AD 1590–1990 and look much more
LeGoff et al. (2002) (see also Tanguy et al., 2003) is specifically
like the structure inferred for the 0–5 Ma field (see Chapter
directed at assessing the limits of archeomagnetic dating in this
5.11, Figure 23).
region. Others (including Arrighi et al., 2004; Arrighi et al.,
2005; Incoronato et al., 2002; Principe et al., 2004; Speranza
• In addition to the geomagnetic interest, numerous applica-
tions for global and regional models exist in other fields,
et al., 2005) have used recent paleomagnetic studies to assess
including historical, archeological, stratigraphic, and paleo-
the eruptive history of various volcanoes.
climate studies.
Two other magnetostratigraphic applications exploit recent
improvements in knowledge about Holocene secular varia- The previously mentioned list focuses on results for the past
tions. Hagstrum et al. (2004) and St-Onge et al. (2004) were 10 ky. The number of archeomagnetic data drops off rapidly
able to recover paleomagnetic directions from sediments asso- with increasing age, and the spatial distribution is in any case
ciated with paleoseismic excavation in Northwestern United limited to areas where human development has provided suit-
States and Quebec, respectively, and used these to make infer- able materials. Recent lava flows help augment the data
ences about the timing of the earthquakes studied. Bowles et al. distribution but, in general, have less tightly controlled age
(2005, 2006) used the paleointensity recorded by submarine constraints. The episodic nature of both developments of civ-
basaltic glass and comparison with trends in the record from ilizations and volcanic eruptions mean that the temporal cov-
VADMs to constrain the ages of young flows on the East Pacific erage is uneven. However, there have been substantial efforts to
Rise, with corresponding implications for the timing and acquire new high-quality data with tight age controls and well-
extent of eruptive activity. documented laboratory procedures. Public archive of these
To date, most of these high-resolution stratigraphic applica- results in databases like GEOMAGIA50 and MagIC allows an
tions have invoked regional records (but see Brachfeld et al., assessment of systematic influences related to material types,
2003), and it is mainly on longer timescales that global or inter- data corrections, and experimental methods. Realistic assess-
hemispheric comparisons have been made with corresponding ments of age uncertainties are particularly important for mak-
lack of resolution because of lower sedimentation rates (see Laj ing further progress in field modeling.
et al., 2000, 2004; McMillan and Constable, 2006; McMillan et al., There is substantial potential for acquiring more sedimen-
2002, 2004; Stoner et al., 2000, 2002). However, millennial-scale tary data, which can augment geographic coverage and also
field models have started to develop into useful magnetostrati- offer the possibility of extending the global view further back in
graphic dating tools (Lodge and Holme, 2008; Pavón-Carrasco time. Relative paleointensity records can be calibrated by inclu-
et al., 2009) regionally. With appropriate assessment of the sion in the modeling procedure, and it is apparent that a 100
resolution, attainable future global or regional models might ky time-varying field model is within reach. A higher-
also provide this functionality for regions other than Europe. resolution model may yet be feasible for a time period like
0–3 ka that only uses the highest-quality data with tight age
constraints. One goal here would be to detect variations in
5.09.9 Outstanding Problems and Scope for Future dipole moment on the shortest possible timescales, interesting
Progress in its own right but also a critical step towards separating solar
from geomagnetic variations in cosmogenic nuclide produc-
Over the past decade, there has been considerable progress in tion. In global and regional modeling efforts, it will be impor-
integrating millennial-scale geomagnetic field measurements tant to conduct a careful assessment of how to implement
to provide improved understanding of geomagnetic field more robust data processing tools and assess the quality and
behavior on timescales longer than the historical record. The temporal and spatial resolution of the resulting models.
current state of global paleomagnetic field modeling on mil- Future models could be used to address the following series
lennial scales can be summarized as follows: of outstanding geomagnetic questions:
• Time-varying geomagnetic field models back to 8000 BC (1) How dominant is the dipole on average? Most paleomag-
are feasible. netic data continue to point to the present field lying on the
• Accumulation of new data over the past decade has led to high side of the average (Tauxe, 2006; Tauxe and Staudigel,
improvement of earlier models. 2004), suggesting that features like the South Atlantic
Anomaly are probably a normal part of the field. New high- Contains archeomagnetic and lava intensity and directional
resolution models might be used to assess the longevity of data for the past 50 ky, including the option to produce model
the current feature and its temporal evolution. In the early predictions from the CALSxk.n models. An extension to
part of CALS10k.1b, the dipole moment is low compared include lake sediment records is in preparation.
with its current value, but the rate of change with time does
not seem to be anomalous. Improved temporal resolution • MagIC: http://earthref.org/MAGIC/
will help clarify this, while improved spatial resolution will Archive/database with even more metadata and for all time-
allow a better assessment of the dominant long wavelength scales of rock and paleomagnetic data. Incorporates all the
structure in the secular variation. older IAGA databases and GEOMAGIA50.
(2) Is the magnetic field and its secular variation persistently
lower in the Pacific than in the Atlantic hemisphere, as
Walker and Backus (1996) found for the present field? The Commonly Used Radiocarbon Age Calibration Programs
question of persistent hemispheric asymmetries in the • CALIB: http://calib.qub.ac.uk/calib/
millennial-scale field is under investigation as it is for
longer time intervals. Calibration program developed by M. Stuiver, P. J. Reimer,
(3) Is there observational evidence for wave motion in the and R. Reimer at Queen’s University Belfast.
geomagnetic field? On what timescales? What is its physical • OxCal: http://c14.arch.ox.ac.uk/oxcal.html
origin? As improved time-varying geomagnetic models have
become available, there has recently been substantial inter- Calibration program developed by C. B. Ramsey at the
est in understanding the role that might be played by tor- University of Oxford.
sional oscillations and other kinds of waves in the Earth’s
core (Bloxham, 2002; Finlay and Jackson, 2003; Zatman Regional Holocene Magnetic Field Models
and Bloxham, 1997). The torsional oscillations detected in
historical models are too short period to be detected in the • SCHA.DIF.3K: http://pc213fis.fis.ucm.es/scha.dif.3k/index.
CALS models, but the work by Dumberry and Bloxham html
(2006) suggests that millennial-scale field may be of use • SCHA.DIF.8K: http://pc213fis.fis.ucm.es/scha.dif.8k/index.
in developing new ideas about the observational evidence html
and physical understanding for longer-period waves. Both models are also available in the EarthRef Digital
Archive (see ERDA links in preceding text).
Appendix: Electronic Links

Global Holocene Magnetic Field Models
Several of the databases, radiocarbon age calibration tools, and • CALSxk.n: http://www.gfz-potsdam.de/magmodels
regional and global models are available electronically. Data-
bases and calibration tools are accessible on interactive web- The CALSxk.n models are also available in the EarthRef
sites. The model coefficients mostly come with tools to Digital Archive (see preceding text) and model predictions at
evaluate them and with animations showing the field evolu- the Earth’s surface for any location can additionally be
tion. A number of useful links are given in the succeeding text. obtained from the GEOMAGIA50 database (see preceding
text).
General
Acknowledgments
• http://earthref.org/
A general Earth science reference archive containing several This work was partly supported by NSF Grant EAR05-37986
linked databases. Related to this article, it hosts the MagIC and EAR0809709 and facilitated by funding from a Humboldt
paleomagnetic database with the PMAG search portal noted Award to CC. We thank Masaru Kono and Tadahiro Hata-
in the succeeding text. Additionally it contains keyama for their comprehensive and constructive reviews at
short notice.
○ http://earthref.org/ERR/ a reference database providing
links to published articles;
○ http://earthref.org/ERDA/ a general digital archive that con- References
tains, for example, the CALSxk.n global models and the
SCHA.DIF.xk regional models, software for their evaluation, Aitken M (1998) Introduction to Optical Dating. Oxford: Oxford University Press.
and some global animations of time-varying geomagnetic Aitken MJ, Alcock PA, Bussell GD, and Shaw CJ (1981) Archaeomagnetic
determination of the past geomagnetic intensity using ancient ceramics: Allowance
elements. Information can be found by entering relevant
for anisotropy. Archaeometry 23(1): 53–64.
key words into the search field. Amit H, Korte M, Aubert J, Constable C, and Hulot G (2011) The time-dependence of
intense archeomagnetic flux patches. Journal of Geophysical Research 116. http://
dx.doi.org/10.1029/2011JB008,538.
Recent Paleomagnetic Databases Arrighi S, Rosi M, Tanguy J-C, and Courtillot VV (2004) Recent eruptive history of
Stromboli (Aeolian Islands, Italy) determined from high-accuracy archeomagnetic
• GEOMAGIA50: http://geomagia.ucsd.edu/ dating. Geophysical Research Letters 31(19).
Arrighi S, Tanguy J-C, Courtillot V, and LeGoff M (2005) Reply to comment by Constable CG, Johnson CL, and Lund SP (2000) Global geomagnetic field models for
F. Speranza et al. on “recent eruptive history of Stromboli (Aeolian Islands, Italy) the past 3000 years: Transient or permanent flux lobes? Philosophical Transactions
determined from high-accuracy archeomagnetic dating” Geophysical Research of the Royal Society of London A 358: 991–1008.
Letters 32(23), L23, 305. Constable C and Korte M (2006) Is Earth’s magnetic field reversing? Earth and Planetary
Baag C, Helsley CE, Xu S-Z, and Lienert BR (1995) Deflection of paleomagnetic Science Letters 246: 1–6.
directions due to magnetization of the underlying terrain. Journal of Geophysical Constable CG and McElhinny MW (1985) Holocene geomagnetic secular variation
Research 100(B6): 10013–10027. records from north-eastern Australian lake sediments. Geophysical Journal of the
Bard E, Raisbeck G, Yiou F, and Jouzel J (2000) Solar irradiance during the last 1200 Royal Astronomical Society 81: 103–120.
years based on cosmogenic nuclides. Tellus 52B: 985–992. Constable C and Parker R (1991) Deconvolution of long-core palaeomagnetic
Barton CE and Burden FR (1979) Modifications to the Mackereth corer. Limnology measurements—Spline therapy for the linear problem. Geophysical Journal
Oceanography 24(5): 977–983. International 104: 453–468, iRMQ1.2.
Bauer LA (1895) On the secular motion of a free magnetic needle. II. Review of Physics Constable C and Tauxe L (1996) Towards absolute calibration of sedimentary
3: 34–48. paleointensity records. Earth and Planetary Science Letters 143: 269–274.
Blaauw M and Christen JA (2011) Flexible paleoclimate age-depth models using an Constable C, Tauxe L, and Parker R (1998) Analysis of 11 Myr of geomagnetic intensity
autoregressive gamma process. Bayesian Analysis 6: 457–474. variation. Journal of Geophysical Research 103: 17735–17748.
Bloxham J (2002) Time-independent and time-dependent behaviour of high-latitude Cromwell G, Tauxe L, Staudigel H, Constable C, Koppers A, and Pedersen R-B (2013)
flux bundles at the core–mantle boundary. Geophysical Research Letters 29: 1854. In search of long-term hemispheric asymmetry in the geomagnetic field: Results
http://dx.doi.org/10.1029/2001GL014,543. from high northern latitudes. Geochemistry, Geophysics, Geosystems. http://dx.doi.
Bloxham J and Gubbins D (1987) Thermal core–mantle interactions. Nature org/10.1002/ggge.20174.
325: 511–513. Daly L and Le Goff M (1996) An updated and homogeneous world secular variation data
Bloxham J and Jackson A (1992) Time-dependent mapping of the magnetic field at the base. 1. Smoothing of the archeomagnetic results. Physics of the Earth and
core-mantle boundary. Journal of Geophysical Research 97(B13): 19537–19563. Planetary Interiors 93: 159–190.
http://dx.doi.org/10.1029/92JB01591. Damon PE and Peristykh AN (2004) Solar and climatic implications of the centennial
Bloxham J, Zatman S, and Dumberry M (2002) The origin of geomagnetic jerks. Nature and millennial periodicities in atmospheric d14C variations. In: Pap JM, Fox P,
420(6911): 65–68. Frohlich C, Hudson HS, Kuhn J, McCormack J, North G, Sprigg W, and Wu ST
Bohnel H, Morales J, Caballero C, et al. (1997) Variation of rock magnetic parameters (eds.) Solar Variability and Its Effects on Climate, pp. 237–249. Wiley Online
and paleointensities over a single Holocene lava flow. Journal of Geomagnetism and Library.
Geoelectricity 49(4): 523–542. Dodson RE (1979) Counterclockwise precession of the geomagnetic field vector and
Bowles J, Gee JS, Kent DV, Bergmanis E, and Sinton J (2005) Cooling rate effects on westward drift of the non-dipole field. Journal of Geophysical Research: Solid Earth
paleointensity estimates in submarine basaltic glass and implications for dating (1978–2012) 84(B2): 637–644.
young flows. Geochemistry, Geophysics, Geosystems 6(7). Dodson MH and McClelland-Brown E (1980) Magnetic blocking temperatures of
Brachfeld SA, Domack EW, Kissel C, et al. (2003) Holocene history of the larsen ice single-domain grains during slow cooling. Journal of Geophysical Research
shelf constrained by geomagnetic paleointensity dating. Geology 31: 749–752. 85(B5): 2625–2637.
Braginskiy SI (1972) Analytical description of geomagnetic field of the past and spectral Donadini F, Korhonen K, Riisager P, and Pesonen LJ (2006) Database for Holocene
analysis of magnetic waves in the Earth’s core. Geomagnetism and Aeronomy geomagnetic intensity information. EoS, Transactions American Geophysical
12: 1092–1105. Society 87(14): 92–93.
Braginskiy SI (1974) Analytical description of geomagnetic field of the past and spectral Donadini F, Korte M, and Constable C (2009) Geomagnetic field for 0–3 ka: 1. New data
analysis of magnetic waves in the Earth’s core II. Geomagnetism and Aeronomy sets for global modeling. Geochemistry, Geophysics, Geosystems 10: Q06007.
14: 522529. http://dx.doi.org/10.1029/2008GC002,295.
Braginskiy SI and Burlatskaya SP (1979) Spherical analysis of the geomagnetic field Downey W (2011) Orientations of Minoan buildings on Crete may indicate the first
based on archaeomagnetic data. Izvestiya Physics of the Solid Earth 15: 891–895. recorded use of the magnetic compass. Mediterranean Archaeology and
Bucur I (1994) The direction of the terrestrial magnetic field in France during the last Archaeometry 11: 9–20.
21 centuries. Recent progress. Physics of the Earth and Planetary Interiors Dumberry M and Bloxham J (2006) Azimuthal flows in the Earth’s core and changes in
67: 95–109. length of day at millennial timescales. Geophysical Journal International
Bullard EC, Freedman C, Gellman H, and Nixon J (1950) The westward drift of the 165: 32–46.
Earth’s magnetic field. Philosophical Transactions of the Royal Society of London A: Dumberry M and Finlay CC (2007) Eastward and westward drift of the Earth’s
Mathematical, Physical and Engineering Sciences 243(859): 67–92. magnetic field for the last three millennia. Earth and Planetary Science Letters
Cande SC and Kent DV (1995) Revised calibration of the geomagnetic polarity timescale 254: 146–157.
for the Late Cretaceous and Cenozoic. Journal of Geophysical Research 100(B4): Dunai T (2001) Influence of secular variation of the geomagnetic field on production
6093–6095. rates of in situ produced cosmogenic nuclides. Earth and Planetary Science Letters
Castro J and Brown L (1987) Shallow paleomagnetic directions from historic lava flows, 193: 197–212.
Hawaii. Geophysical Research Letters 14(12): 1203–1206. Elsasser W, Ney E, and Winckler J (1956) Cosmic-ray intensity and geomagnetism.
Channell JET (1999) Geomagnetic paleointensity and directional secular variation at Nature 178: 1226–1227.
ocean drilling program (ODP) site 984 (Bjorn Drift) since 500 ka: comparisons with Finlay CC (2008) Historical variations of the geomagnetic axial dipole. Physics of the
ODP site 983 (Gardar Drift). Journal of Geophysical Research: Solid Earth Earth and Planetary Interiors 170: 1–14.
(1978–2012) 104(B10): 22937–22951. Finlay CC and Jackson A (2003) Equatorial dominated magnetic field change at Earth’s
Channell JET, Hodell DA, McManus J, and Lehman B (1998) Orbital modulation of the core surface. Science 300: 2084–2086.
Earth’s magnetic field intensity. Nature 394(6692): 464–468. Folgerhaiter G (1899) Sur les variations séculaire de l’inclination magnétique dans
Charvátová I, Klokoĉn/’ik J, Kolmas J, and Kostelecky J (2011) Chinese tombs oriented l’antiquité. Journal of Physics 8(3): 5–16.
by a compass: Evidence from paleomagnetic changes versus the age of tombs. Fox JMW and Aitken MJ (1980) Cooling-rate dependence of thermoremanent
Studia Geophysica et Geodetica 55: 159–174. magnetisation. Nature 283: 462–463.
Chauvin A, Garcia Y, Lanos P, and Laubenheimer F (2000) Paleointensity of the Gallet Y, Genevey A, and Courtillot V (2003) On the possible occurrence of
geomagnetic field recovered on archaeomagnetic sites from France. Physics of the ‘archaeomagnetic jerks’ in the geomagnetic field over the past three millennia. Earth
Earth and Planetary Interiors 120: 111–136. and Planetary Science Letters 214: 237–242.
Chevallier R (1925) L’aimantation des laves de l’Etna et l’orientation du champ terrestre Gallet Y, Genevey A, and Fluteau F (2005) Does Earth’s magnetic field secular variation
en Sicile du XIIe et XVIIe sieclé. Annales de Physique 4: 5–161. control centennial climate change? Earth and Planetary Science Letters
Clark RM (1975) A calibration curve for radiocarbon dates. Antiquity 49: 251–266. 236: 339–347.
Constable C and Johnson C (1999) Anisotropic paleosecular variation models: Gallet Y, Genevey A, and Goff ML (2002) Three millennia of directional variation of the
Implications for geomagnetic observables. Physics of the Earth and Planetary Earth’s magnetic field in western Europe as revealed by archaeological artefacts.
Interiors 115: 35–51. Physics of the Earth and Planetary Interiors 131: 81–89.
Constable C and Johnson C (2005) A paleomagnetic power spectrum. Physics Genevey A and Gallet Y (2002) Intensity of the geomagnetic field in western Europe over
of the Earth and Planetary Interiors 153: 61–73. http://dx.doi.org/10.1016/j. the past 2000 years: New data from ancient French pottery. Journal of Geophysical
pepi.2005.03.015. Research 107: 2285. http://dx.doi.org/10.1029/2001JB000,701.
Genevey A, Gallet Y, Constable C, Korte M, and Hulot G (2008) ArcheoInt: An upgraded Kono M and Roberts PH (2002) Recent geodynamo simulations and observations of the
compilation of geomagnetic field intensity data for the past ten millennia and its geomagnetic field. Reviews of Geophysics 40(4): 1–4.
application to the recovery of the past dipole moment. Geochemistry, Geophysics, Kono M, Sakuraba A, and Ishida M (2000) Dynamo simulation and palaeosecular
Geosystems 9: Q04038. http://dx.doi.org/10.1029/2007GC001,881. variation models. Philosophical Transactions of the Royal Society of London A
Genevey A, Gallet Y, and Margueron J-C (2003) Eight thousand years of geomagnetic 358(1768): 1123–1139.
field intensity variations in the eastern Mediterranean. Journal of Geophysical Korhonen K, Donadini F, Riisager P, and Pesonen L (2008) Geomagia50: An
Research 108. http://dx.doi.org/10.1029/2001JB001612. archeointensity database with PHP and MySQL. Geochemistry, Geophysics,
Goguitchaichvili A, Greco C, and Morales J (2011) Geomagnetic field intensity behavior Geosystems 9: Q04029. http://dx.doi.org/10.1029/2007GC001,893.
in South America between 400 AD and 1800 AD: first archeointensity results from Korte M and Constable CG (2003) Continuous global geomagnetic field models for the
Argentina. Physics of the Earth and Planetary Interiors 186: 191–197. past 3000 years. Physics of the Earth and Planetary Interiors 140: 73–89.
Gómez-Paccard M, Chauvin A, Lanos P, et al. (2006) First archaeomagnetic secular Korte M and Constable CG (2005a) Continuous geomagnetic field models for the past 7
variation curve for the Iberian Peninsula: Comparison with other data from western millennia: 2. CALS7K. Geochemistry, Geophysics, Geosystems 6: Q02H16. http://
Europe and with global geomagnetic field models. Geochemistry, Geophysics, dx.doi.org/10.1029/2004GC000,801.
Geosystems 7: Q12001. http://dx.doi.org/10.1029/2006GC001,476. Korte M and Constable CG (2005b) The geomagnetic dipole moment over the last 7000
Gosse JC and Phillips FM (2001) Terrestrial in situ cosmogenic nuclides: Theory and years—New results from a global model. Earth and Planetary Science Letters
application. Quaternary Science Reviews 20: 1475–1560. 236: 348–358.
Griffiths DG (1953) Remanent magnetism of varved clays form Sweden. Nature 172: 539. Korte M and Constable C (2006a) Centennial to millennial geomagnetic variation.
Gubbins D, Jones AL, and Finlay CC (2006) Fall in Earth’s magnetic field is erratic. Geophysical Journal International 167: 43–52.
Science 312: 900–902. Korte M and Constable CG (2006b) On the use of calibrated relative paleointensity
Guyodo Y and Valet JP (1999) Global changes in geomagnetic intensity during the past records to improve millennial-scale geomagnetic field models. Geochemistry,
800 thousand years. Nature 399: 249–252. Geophysics, Geosystems 7: Q09004. http://dx.doi.org/10.1029/2006GC001,368.
Hagstrum JT, Atwater BF, and Sherrod BL (2004) Paleomagnetic correlation of late Korte M and Constable CG (2008) Spatial and temporal resolution of millennial scale
Holocene earthquakes among estuaries in Washington and Oregon. Geochemistry, geomagnetic field models. Advances in Space Research 41: 57–69.
Geophysics, Geosystems 5(10). Korte M and Constable C (2011) Improving geomagnetic field reconstructions for
Hagstrum JT and Blinmann E (2010) Archaeomagnetic dating in western North America: 0–3 ka. Physics of the Earth and Planetary Interiors 188: 247–259.
An updated reference curve based on paleomagnetic and archeomagnetic data sets. Korte M, Constable C, Donadini F, and Holme R (2011) Reconstructing the Holocene
Geochemistry, Geophysics, Geosystems 11: Q06009. http://dx.doi.org/ geomagnetic field. Earth and Planetary Science Letters 312: 497–505.
10.1029/2009GC002,979. Korte M, Donadini F, and Constable C (2009) Geomagnetic field for 0–3 ka: 2. A new
Halgedahl SL, Day R, and Fuller M (1980) The effect of cooling rate on the intensity of series of time-varying global models. Geochemistry, Geophysics, Geosystems 10:
weak-field TRM in single-domain magnetite. Journal of Geophysical Research: Q06008. http://dx.doi.org/10.1029/2008GC002,297.
Solid Earth (1978–2012) 85(B7): 3690–3698. Korte M, Genevey A, Constable CG, Frank U, and Schnepp E (2005) Continuous
Hartmann GA, Genevey A, Gallet Y, et al. (2010) Archeointensity in Northeast Brazil over geomagnetic field models for the past 7 millennia: 1. A new global data compilation.
the past five centuries. Earth and Planetary Science Letters 296: 340–352. Geochemistry, Geophysics, Geosystems 6: Q02H15. http://dx.doi.org/
Hatakeyama T (2007) Effects of the site distribution and the prior information on the 10.1029/2004GC000,800.
inverted geomagnetic field model: A case study applying the abic method to the Korte M and Holme R (2010) On the persistence of geomagnetic flux lobes in global
synthetic datasets. Earth, Planets and Space 59: 703–709. field models. Physics of the Earth and Planetary Interiors 182: 179–186.
Heirtzler J (2002) The future of the South Atlantic Anomaly and implications for Korte M and Muscheler R (2012) Centennial to millennial geomagnetic field variations.
radiation damage in space. Journal of Atmospheric and Solar Terrestrial Physics Journal of Space Weather and Space Climate 2. http://dx.doi.org/10.1051/swsc/
64: 1701–1708. 2012,006.
Hongre L, Hulot G, and Khokhlov A (1998) An analysis of the geomagnetic field over the Kostadinova M, Jordanova N, Jordanova D, and Kovacheva M (2004) Preliminary study
past 2000 years. Physics of the Earth and Planetary Interiors 106: 311–335. on the effect of water glass impregnation on the rock-magnetic properties of baked
Hughen KA, Baillie MGL, Bard E, et al. (2004) Marine04 marine radiocarbon age clay. Studia Geophysica et Geodaetica 48: 1573–1626.
calibration, 0–26 cal kyr BP. Radiocarbon 46(3): 1059–1086. Kovacheva M, Jordanova N, and Karloukovski V (1998) Geomagnetic field variations as
Incoronato A, Angelino A, Romano R, et al. (2002) Retrieving geomagnetic secular determined from Bulgarian archeomagnetic data. Part II: The last 8000 years.
variations from lava flows: Evidence from Mounts Arso, Etna and Vesuvius Surveys in Geophysics 19: 431–460.
(southern Italy). Geophysical Journal International 149(3): 724–730. LaBelle JM and Eighmy JL (1997) Additional archaeomagnetic data on the south-west
Jackson M, Bowles JA, Lascu I, and Solheid P (2010) Deconvolution of u channel USA master geomagnetic pole curve. Archaeometry 39: 431–439.
magnetometer data: Experimental study of accuracy, resolution, and stability of Laj C, Kissel C, and Beer J (2004) High resolution global paleointensity stack since
different inversion methods. Geochemistry, Geophysics, Geosystems 11. http://dx. 75 kyr (GLOPIS-75) calibrated to absolute values. Timescales of the Paleomagnetic
doi.org/10.1029/2009GC002991. Field 255–265.
Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular Laj C, Kissel C, Mazaud A, Channell JET, and Beer J (2000) North Atlantic
variation from historical records. Philosophical Transactions of the Royal Society of palaeointensity stack since 75 ka (NAPIS 75) and the duration of the Laschamp
London A 358: 957–990. event. Philosophical Transactions of the Royal Society of London, Series A:
Johnson C and Constable C (1998) Persistently anomalous Pacific geomagnetic fields. Mathematical, Physical and Engineering Sciences 358(1768): 1009–1025.
Journal of Geophysical Research 25: 1011–1014. Laj C, Kissel C, Mazaud A, Michel E, Muscheler R, and Beer J (2002) Geomagnetic field
Johnson EA, Murphy T, and Torreson OW (1948) Pre-history of the Earth’s magnetic intensity, North Atlantic deep water circulation and atmospheric d14C during the last
field. Terrestrial Magnetism and Atmospheric Electricity 53: 349–372. 50 kyr. Earth and Planetary Science Letters 200(1): 177–190.
Kamen MD (1963) Early history of carbon-14. Science 140: 584–590. Lal D (1991) Cosmic ray labeling of erosion surfaces: In situ nuclide production rates
Key R (2001) Radiocarbon. In: Thorpe S and Turekian K (eds.) Encyclopedia of Ocean and erosion models. Earth and Planetary Science Letters 104: 424–439.
Sciences, pp. 2338–2353. Elsevier. http://dx.doi.org/10.1006/rwos.2001.0162. Lal D, Jull AJT, Pollard D, and Vacher L (2005) Evidence for large century time-scale
King JW, Banerjee SK, and Marvin J (1983) A new rock-magnetic approach to selecting changes in solar activity in the past 32 kyr, based on in-situ cosmogenic 14C in ice
sediments for geomagnetic paleointensity studies: Application to paleointensity for at Summit, Greenland. Earth and Planetary Science Letters 234(3): 335–349.
the last 4000 years. Journal of Geophysical Research 88: 5911–5921. Lal D and Peters B (1967) Cosmic ray produced radioactivity on the Earth. In: Flugge S
Klokoĉnı́k J (2011) Pyramids and ceremonial centers in mesoamerica and China: Were (ed.) Handbuch der Physik, pp. 551–612. Berlin: Springer.
they oriented using a magnetic compass? Procedia Environmental Sciences Lanos P (2004) Bayesian inference of calibration curves: Application to
10: 255–261. archaeomagnetism. In: Buck CE and Millard AR (eds.) Tools for Constructing
Klokoĉnı́k J, Kostelecky J, and Vı́tek F (2007) Pyramids and ceremonial centers in Chronologies, Crossing Disciplinary Boundaries. Lecture Notes in Statistics,
mesoamerica: Were they oriented using a magnetic compass? Studia Geophysica et vol. 177, pp. 43–82. London: Springer.
Geodetica 51: 515–533. Lanos P, Goff ML, Kovacheva M, and Schnepp E (2005) Hierarchical modelling of
Klovan JE (1964) Box-type sediment-coring device. Journal of Sedimentary Research archaeomagnetic data and curve estimation by moving average technique.
34(1): 185–189. Geophysical Journal International 160: 440–476.
Knudsen MF, Riisager P, Donadini F, et al. (2008) Variations in the geomagnetic dipole Lanza R, Meloni A, and Tema E (2005) Historical measurements of the Earth’s magnetic
moment during the Holocene and the past 50 kyr. Earth and Planetary Science field compared with remanence directions from lava flows in Italy over the last four
Letters 272: 319–329. centuries. Physics of the Earth and Planetary Interiors 148(1): 97–107.
Lawrence K, Tauxe L, Staudigel H, et al. (2009) Paleomagnetic field properties at high Morrison LV and Stephenson FR (2001) Historical eclipses and the variability of the
southern latitude. Geochemistry, Geophysics, Geosystems 10. http://dx.doi.org/ Earth’s rotation. Journal of Geodynamics 32(1): 247–265.
10.1029/2008GC002072. Muscheler R, Beer J, and Kubik PW (2004) Long-term solar variability and
LeGoff M (1975) Inductomètre à Rotation Continue Pour la Mesure des Faibles climate change based on radionuclide data from ice cores. In: Pap JM, Fox P,
Aimantations Rémanentes et Induites en Magnétisme des Roches. PhD Thesis, Frohlich C, Hudson HS, Kuhn J, McCormack J, North G, Sprigg W, and Wu ST
Môm., Diplôme d’Ingénieur CNAM, Paris. (eds.) Solar Variability and Its Effects on Climate, pp. 221–235. Wiley Online
LeGoff M, Gallet Y, Genevey A, and Warmé N (2002) On archeomagnetic secular Library.
variation curves and archeomagnetic dating. Physics of the Earth and Planetary Muscheler R, Beer J, Wagner G, et al. (2004) Changes in the carbon cycle during the
Interiors 134(3): 203–211. last deglaciation as indicated by the comparison of 10Be and 14C records. Earth and
Lengyel S (2010) The pre-AD 585 extension of the U.S. Southwest archaeomagnetic Planetary Science Letters 219(3): 325–340.
reference curve. Journal of Archaeological Science 37: 3081–3090. Muscheler R, Joos F, Muller SA, and Snowball I (2005) How unusual is today’s solar
Lengyel SN and Eighmy JL (2002) A revision to the U.S. Southwest archaeomagnetic activity? Nature 436: E3–E4.
master curve. Journal of Archaeological Science 29: 1423–1433. Neel L (1955) Some theoretical aspects of rock-magnetism. Advances in Physics 4(14):
Leonhardt R, Matzka J, Nichols ARL, and Dingwell DB (2006) Cooling rate correction of 191–243.
paleointensity determination for volcanic glasses by relaxation geospeedometry. Neukirch LP, Tarduno JA, Huffman TN, Watkeys MK, Scribner CA, and Cottrell RD
Earth and Planetary Science Letters 243(1): 282–292. (2012) An archeomagnetic analysis of burnt grain bin floors from ca. 1200 to
Levi S and Banerjee SK (1976) On the possibility of obtaining relative paleointensities 1250 AD Iron-Age South Africa. Physics of the Earth and Planetary Interiors
from lake sediments. Earth and Planetary Science Letters 29(1): 219–226. 190–191: 71–79.
Libby W (1955) Radiocarbon Dating, 2nd edn. Chicago: University of Chicago Press. Nilsson A, Muscheler R, and Snowball I (2011) Millennial scale cyclicity in the
Lifton N, Bieber JW, Clem JM, et al. (2005) Addressing solar modulation and long-term geodynamo inferred from a dipole tilt reconstruction. Earth and Planetary Science
uncertainties in scaling secondary cosmic rays for in situ cosmogenic nuclide Letters 311: 205–299.
applications. Earth and Planetary Science Letters 239(1): 140–161. Nilsson A, Snowball I, and Muscheler R (2010) Holocene geocentric dipole tilt model
Liritzis I (2011) Surface dating by luminescence: An overview. Geochronometria 38(3): constrained by sedimentary paleomagnetic data. Geochemistry, Geophysics,
292–302. Geosystems 11: Q08018. http://dx.doi.org/10.1029/2010GC003,118.
Livingstone DA (1955) A lightweight piston sampler for lake deposits. Ecology 36(1): Noller J, Sowers JM, and Lettis WR (eds.) (2000) In: Quaternary Geochronology:
137–139. Methods and Applications, vol. 4, p. 582. Washington, DC: American Geophysical
Lodge A and Holme R (2008) Towards a new approach to archaeomagnetic dating in Union.
Europe using geomagnetic field modelling. Archaeometry 50(3): 309–322. Oda H and Shibuya H (1996) Deconvolution of long-core paleomagnetic data of ocean
Love JJ and Constable CG (2003) Gaussian statistics for palaeomagnetic vectors. drilling program by Akaike’s bayesian information criterion minimization. Journal of
Geophysical Journal International 152(3): 515–565. Geophysical Research 101: 2815. http://dx.doi.org/10.1029/95JB02811.
Lund SP (1996) A comparison of Holocene paleomagnetic secular variation records Oguti T (1993) The auroral zone in historic times – The northern UK was in the
from North America. Journal of Geophysical Research 101(B4): 8007–8024. auroral zone 300 years ago. Journal of Geomagnetism and Geoelectricity
Mackereth FJH (1958) A portable core sampler for lake deposits. Limnology and 45: 231–242.
Oceanography 3: 181–191. Oguti T and Egeland A (1995) Auroral occurrences in Norwegian archives. Journal of
Mankinen E and Cox A (1988) Paleomagnetic investigation of some volcanic rocks from Geomagnetism and Geoelectricity 47: 353–359.
the McMurdo Volcanic Province, Antarctica. Journal of Geophysical Research Ohno M and Hamano Y (1992) Geomagnetic poles over the past 10,000 years.
93: 11599–11612. Geophysical Research Letters 19: 1715–1718.
Marco ED, Spatharas V, Gómez-Paccard M, Chauvin A, and Kondopoulou D (2008) Ohno M and Hamano Y (1993) Spherical harmonic analysis of paleomagnetic secular
New archaeointensity results from archaeological sites and variation of the variation curves. Central Core of the Earth 3: 205–212.
geomagnetic field intensity for the last 7 millennia in Greece. Physics and Chemistry Oishi Y, Tsunakawa H, Mochizuki N, Yamamoto Y, Wakabayashi K-I, and Shibuya H
of the Earth 33: 578–595. (2005) Validity of the LTD-DHT Shaw and Thellier palaeointensity methods: A case
Márton P (1970) Secular variation of the geomagnetic virtual dipole field during the last study of the Kilauea 1970 lava. Physics of the Earth and Planetary Interiors 149(3):
2000 yr as inferred from the spherical harmonic analysis of the available 243–257.
archeomagnetic data. Pure and Applied Geophysics 81: 163–176. Olson P and Christensen UR (2002) The time-averaged magnetic field in numerical
Márton P (2010) Two thousand years of geomagnetic field direction over central Europe dynamos with non-uniform boundary heat flow. Geophysical Journal International
revealed by indirect measurements. Geophysical Journal International 181: 261–268. 151(3): 809–823.
Masarik J and Beer J (1999) Simulation of particle fluxes and cosmogenic nuclide Panovska S (2012) Modelling Holocene Geomagnetic Field Evolution. PhD Thesis, ETH
production in the Earth’s atmosphere. Journal of Geophysical Research 104(12): Zurich.
12099–12111. Panovska S, Finlay CC, Donadini F, and Hirt AM (2012) Spline analysis of Holocene
McElhinny MW and Lock J (1996) IAGA paleomagnetic databases with Access. Surveys sediment magnetic records: Uncertainty estimates for field modeling. Journal of
in Geophysics 17: 575–591. Geophysical Research 117(B2). http://dx.doi.org/10.1029/2011JB008813.
McElhinny MW and Senanayake WE (1982) Variations in the geomagnetic dipole: I. The Parker RL (1994) Geophysical Inverse Theory. Princeton, NJ: Princeton University
past 50000 years. Journal of Geomagnetism and Geoelectricity 34: 39–51. Press.
McMillan DG and Constable CG (2006) Limitations in correlation of regional relative Parker RL (2000) Calibration of the pass-through magnetometer-I. Theory. Geophysical
geomagnetic paleointensity. Geochemistry, Geophysics, Geosystems 7(9). Journal International 142: 371–383. http://dx.doi.org/10.1046/
McMillan DG, Constable CG, and Parker RL (2002) Limitations on stratigraphic j.1365-246x.2000.00171.x.
analyses due to incomplete age control and their relevance to sedimentary Parker RL and Gee JS (2002) Calibration of the pass-through magnetometer. II.
paleomagnetism. Earth and Planetary Science Letters 201(3): 509–523. Application. Geophysical Journal International 150(1): 140–152.
McMillan DG, Constable CG, and Parker RRL (2004) Assessing the dipolar signal in Parnell A, Haslett J, Allen J, Buck C, and Huntley B (2008) A flexible approach to
stacked paleointensity records using a statistical error model and geodynamo assessing synchroneity of past events using Bayesian reconstruction of
simulations. Physics of the Earth and Planetary Interiors 145(1): 37–54. sedimentation history. Quaternary Science Reviews 27: 1827–1885.
McMillan DG, Constable CG, Parker RL, and Glatzmaier GA (2001) A statistical Pavón-Carrasco F, Osete M, and Torta J (2010) Regional modelling of the geomagnetic
appraisal of magnetic fields of geodynamo models. Geochemistry, Geophysics, field in Europe from 6000 to 1000 BC. Geochemistry, Geophysics, Geosystems 11:
Geosystems 2, 10.1029. Q11008. http://dx.doi.org/10.1029/2010GC003,197.
Merrill R and McElhinny M (1983) The Earth’s Magnetic Field: Its History, Origin and Pavón-Carrasco F, Osete M, Torta J, and Gaya-Piqué L (2008) A regional
Planetary Perspective. London: Academic Press, pp. 401. archaeomagnetic model for the palaeointensity in Europe for the last 2000 years and
Merrill RT, McElhinny MW, and McFadden PL (1996) The Magnetic Field of the Earth. its implications for climatic change. Pure and Applied Geophysics 165: 1209–1225.
London: Academic Press. Pavón-Carrasco F, Osete M, Torta J, and Gaya-Pique L (2009) A regional
Mitra R, Tauxe L, and Mcintosh SK (2013) Two thousand years of archeointensity from archeomagnetic model for Europe for the last 3000 years, SCHA.DIF.3 K;
West Africa. Earth and Planetary Science Letters 364: 123–133. Applications to archeomagnetic dating. Geochemistry, Geophysics, Geosystems 10:
Mochizuki N, Tsunakawa H, Oishi Y, Wakai S, Wakabayashi K-I, and Yamamoto Y Q03013. http://dx.doi.org/10.1029/2008GC002,244.
(2004) Palaeointensity study of the Oshima 1986 lava in Japan: Implications for the Pavón-Carrasco F, Osete M, Torta J, Gaya-Piqué L, and Lanos P (2008) Initial SCHA.
reliability of the Thellier and LTD-DHT Shaw methods. Physics of the Earth and DI.00 regional archaeomagnetic model for Europe for the last 2000 years. Physics
Planetary Interiors 146(3): 395–416. and Chemistry of the Earth 33: 596–608.
Peng L and King JW (1992) A late Quaternary geomagnetic secular variation record Snowball I and Sandgren P (2004) Geomagnetic field intensity changes in Sweden
from Lake Waiau, Hawaii, and the question of the Pacific nondipole low. Journal of between 9000 and 450 cal BP: Extending the record of “archaeomagnetic jerks” by
Geophysical Research 97: 4407–4424. means of lake sediments and the pseudo-Thellier technique. Earth and Planetary
Pick T and Tauxe L (1993) Holocene paleointensities: Thellier experiments on Science Letters 227: 361–376.
submarine basaltic glass from the East Pacific Rise. Journal of Geophysical Solanki S, Usoskin I, Kromer B, Schussler M, and Beer J (2004) Unusual activity of the
Research 98(B10): 17949–17964. sun during the recent decades compared to the previous 11,000 years. Nature
Pigati JS and Lifton NA (2004) Geomagnetic effects on time-integrated cosmogenic 431: 1084–1087.
nuclide production with emphasis on in situ 14C and 10Be. Earth and Planetary Speranza F, Sagnotti FL, and Meloni A (2005) Comment on “recent eruptive
Science Letters 226: 193–205. history of Stromboli (Aeolian Islands, Italy) determined from high-accuracy
Pinto O, Gonzalez WD, Pinto RCA, Gonzalez ALC, and Mendes O Jr. (1992) The South archeomagnetic dating” by S. Arrighi et al. Geophysical Research Letters 32(L23): 306.
Atlantic Magnetic Anomaly: Three decades of research. Journal of Atmospheric and Stephenson FR and Morrison LV (1995) Long-term fluctuations in the Earth’s rotation:
Terrestrial Physics 5: 1129–1134. 700 BC to AD 1990. Philosophical Transactions of the Royal Society of London A
Principe C, Tanguy J-C, Arrighi S, Paiotti A, Le Goff M, and Zoppi U (2004) Chronology 351(1695): 165–202.
of Vesuvius activity from AD 79 to 1631 based on archeomagnetism of lavas and Stephenson FR, Morrison LV, and Whitrow GJ (1984) Long-term changes in the
historical sources. Bulletin of Volcanology 66(8): 703–724. rotation of the Earth: 700 BC to AD 1980. Philosophical Transactions of the Royal
Ramsey CB (2008) Deposition models for chronological records. Quaternary Science Society of London A 313(1524): 47–70.
Reviews 27: 42–60. Stockhausen H (1998) Geomagnetic palaeosecular variation (0–13 000 yr BP) as
Reimer PJ, Baillie MGL, Bard E, et al. (2004) IntCal04 terrestrial radiocarbon age recorded in sediments from three maar lakes from the West Eifel (Germany).
calibration, 26–0 ka BP. Radiocarbon 46: 1029–1058. Geophysical Journal International 135: 898–910.
Reimer PJ, Bard E, Bayliss A, et al. (2013) IntCal13 and Marine13 Radiocarbon age Stoner JS, Channel JET, Hillaire-Marcel C, and Kissel C (2000) Geomagnetic
calibration curves 0–50,000 years cal BP. Radiocarbon 55(4): 1869–1887. http:// paleointensity and environmental record from Labrador Sea core MD95-2024:
dx.doi.org/10.2548/azu_js_rc55.16947. Global marine sediment and ice core chronostratigraphy for the last 110 kyr. Earth
Reimer P, et al. (2009) IntCal09 and Marine09 radiocarbon age calibration curves, and Planetary Science Letters 183: 161–177.
0–50,000 years cal BP. Radiocarbon 51: 1111–1150. Stoner JS, Laj C, Channell JET, and Kissel C (2002) South Atlantic and North Atlantic
Reimer PJ, Bard E, and Bayliss A (2013) IntCal13 and Marine13 Radiocarbon age geomagnetic paleointensity stacks (0–80 ka): Implications for inter-hemispheric
calibration curves 0–50,000 years cal BP. Radiocarbon 55(4): 1869–1887. http:// correlation. Quaternary Science Reviews 21: 1141–1151.
dx.doi.org/10.2548/azu_js_rc55.16947. St-Onge G, Mulder T, Piper DJW, Hillaire-Marcel C, and Stoner JS (2004) Earthquake
Rogers J, Fox JMW, and Aitken MJ (1979) Magnetic anisotropy in ancient pottery. and flood-induced turbidites in the Saguenay Fjord (Quebec): A Holocene
Nature 277: 644–646. paleoseismicity record. Quaternary Science Review 23: 283–294.
Rolph TC and Shaw J (1986) Variation of the geomagnetic field in Sicily. Journal of St-Onge G, Stoner JS, and Hillaire-Marcel C (2003) Holocene paleomagnetic records
Geomagnetism and Geoelectricity 38: 1269–1277. from the St. Lawrence Estuary, eastern Canada: Centennial- to millennial-scale
Runcorn S (1959) On the theory of geomagnetic secular variation. Annales de geomagnetic modulation of cosmogenic isotopes. Earth and Planetary Science
Geophysique 15: 87–92. Letters 209: 113–130.
Rymer L and Neale J (1981) Freeze coring as a method of collecting unconsolidated Stuiver M and Reimer PJ (1993) Extended 14C data base and revised CALIB 3.0 14C age
lake sediments. Australian Journal of Ecology 6(1): 123–126. calibration program. Radiocarbon 35: 215–230.
Sagnotti L, Budillon F, Dinarès-Turell J, Iorio M, and Macrı̀ P (2005) Evidence for a Stuiver M, Reimer PJ, Bard E, et al. (1998) INTCAL98 radiocarbon age calibration,
variable paleomagnetic lock-in depth in the Holocene sequence from the Salerno 24,000–0 cal BP. Radiocarbon 40: 1041–1083.
Gulf (Italy): Implications for “high-resolution” paleomagnetic dating. Geochemistry, Suttie N, Holme R, Hill MJ, and Shaw J (2011) Consistent treatment of errors in
Geophysics, Geosystems 6(11). archaeointensity implies rapid decay of the dipole. Earth and Planetary Science
Sakai H (1980) Spherical harmonic analysis of the geomagnetic field during the last Letters 304: 13–21.
2,000 years. Rock Magnetism and Paleogeophysics 7: 8–15. Tanguy JC, Goff ML, Chillemi V, et al. (1999) Secular variation of the geomagnetic field
Schnepp E and Lanos P (2005) Archaeomagnetic secular variation in Germany during direction recorded in lavas from Etna and Vesuvius during the last two millennia.
the past 2500 years. Geophysical Journal International 163: 479–490. Comptes Rendu de l’Academie de Sciences Series IIA: Sciences de la Terre et des
Schnepp E and Lanos P (2006) A preliminary secular variation curve for Planetes 329: 557–564.
archeomagnetic dating in Austria. Geophysical Journal International 166: 91–96. Tanguy JC, Goff ML, Principe C, et al. (2003) Archeomagnetic dating of Mediterranean
Schnepp E, Worm K, and Scholger R (2008) Improved sampling techniques for baked volcanics of the last 2100 years: Validity and limits. Earth and Planetary Science
clay and soft sediments. Physics and Chemistry of the Earth 33: 407–413. Letters 211: 111–124.
Selkin P and Tauxe L (2000) Long-term variations in palaeointensity. Philosophical Tanguy J-C and Le Goff M (2004) Distortion of the geomagnetic field in volcanic
Transactions of the Royal Society of London A 358: 1065–1088. terrains: An experimental study of the Mount Etna stratovolcano. Physics of the
Shea MA, Smart DF, and McCracken KG (1965) A study of vertical cutoff rigidities using Earth and Planetary Interiors 141(1): 59–70.
sixth degree simulations of the geomagnetic field. Journal of Geophysical Research Tauxe L (1993) Sedimentary records of relative paleointensity of the geomagnetic field:
70(17): 4117–4130. Theory and practice. Reviews of Geophysics 31(3): 319–354.
Shuey R, Cole E, and Mikulich M (1970) Geographic correction of archaeomagnetic Tauxe L (2006) Long-term trends in paleointensity: The contribution of DSDP/ODP
data. Journal of Geomagnetism and Geoelectricity 41: 485–489. submarine basaltic glass collections. Physics of the Earth and Planetary Interiors
Siegenthaler U, Heimann M, and Oeschger H (2006) 14C variations caused by changes 156(3): 223–241.
in the global carbon cycle. Radiocarbon 22(2): 177–191. Tauxe L and Kent D (2004) A simplified statistical model for the geomagnetic field and
Silverman S (1998) Early auroral observations. Journal of Atmospheric and Solar- the detection of shallow bias in paleomagnetic inclinations: Was the ancient
Terrestrial Physics 60: 997–1006. magnetic field dipolar? In: Time-Scales of the Paleomagnetic Field. Geophysical
Siscoe GL and Siebert KD (2002) Solar-terrestrial effects possibly stronger in biblical Monograph, vol. 145, pp. 101–115. Washington, DC: American Geophysical Union.
times. Journal of Atmospheric and Solar-Terrestrial Physics 64: 1905–1909. Tauxe L, LaBrecque JL, Dodson R, Fuller M, and Dematteo J (1983) U-channels—A
Siscoe GL, Silverman SM, and Siebert KD (2002) Ezekiel and the northern lights: new technique for paleomagnetic analysis of hydraulic piston cores. Eos,
Biblical aurora seems plausible. Eos, Transactions American Geophysical Union Transactions American Geophysical Union 64(18): 219.
83(16): 173–179. Tauxe L, Pick T, and Kok YYS (1995) Relative paleointensity in sediments:
Siscoe GL and Verosub KL (1983) High medieval auroral incidence over China and A pseudo-Thellier approach. Geophysical Research Letters 22(21): 2885–2888.
Japan: Implications for the medieval site of the geomagnetic pole. Geophysical Tauxe L and Staudigel H (2004) Strength of the geomagnetic field in the Cretaceous
Research Letters 10(4): 345–348. Normal Superchron: New data from submarine basaltic glass of the Troodos
Skiles DD (1970) A method of inferring the direction of drift of the geomagnetic field Ophiolite. Geochemistry, Geophysics, Geosystems 5(2).
from paleomagnetic data. Journal of Geomagnetism and Geoelectricity Tauxe L and Yamazaki T (2007) Paleointensities. In: Kono M (ed.) Geomagnetism.
22: 441–462. Treatise on Geophysics, 2nd edn., vol. 5, pp. 6506–6564. Oxford: Elsevier Ltd.
Snowball I and Sandgren P (2002) Geomagnetic field variations in northern Sweden Tema E, Hedley I, and Lanos P (2006) Archaeomagnetism in Italy: A compilation of data
during the Holocene quantified from varved lake sediments and their implications including new results and a preliminary Italian reference curve. Geophysical Journal
for cosmogenic nuclide production rates. The Holocene 12: 517–530. International 167: 1160–1171.
Thébault E (2008) A proposal for regional modelling at the Earth’s surface, R-SCHA2D. Walker AD and Backus GE (1996) On the difference between the average values of Br2 in
Geophysical Journal International 174: 118–134. the Atlantic and Pacific hemispheres. Geophysical Research Letters 23(15):
Thellier E (1938) Sur l’aimantation des terres cuites et ses applications géophysiques. 1965–1968.
An-nales Institut Physique du Globe de Paris 16: 157–302. Wardinski I and Korte M (2008) The evolution of the core-surface flow over the last
Thellier E (1981) Sur la direction du champ magnetique terrestre, en France, durant seven thousand years. Journal of Geophysical Research 113: B05101. http://dx.doi.
les deux derniers millenaires. Physics of the Earth and Planetary Interiors org/10.1029/2007JB005,024.
24: 89–132. Weeks R, Laj C, Endignoux L, et al. (1993) Improvements in long-core measurement
Thellier E and Thellier O (1959) Sur l’intensite du champ magnetique terrestre dans le techniques: Applications in palaeomagnetism and palaeoceanography. Geophysical
passe historique et geologique. Annals of Geophysics 15: 285–376. Journal International 114: 651–662.
Trapanese A, Batt C, and Schnepp E (2008) Sampling methods in archaeomagnetic Willis DM and Stephenson FR (2001) Solar and auroral evidence for an intense
dating: A comparison using case studies from Worterberg, Eisenerz and Gams recurrent geomagnetic storm during December in AD 1128. Annales Geophysicae
Valley (Austria). Physics and Chemistry of the Earth 33: 414–426. http://dx.doi.org/ 19: 289–302.
10.1016/j.pce.2008.02.035. Yamamoto Y, Tsunakawa H, and Shibuya H (2003) Palaeointensity study of the
Usoskin I, Korte M, and Kovaltsov G (2008) Role of centennial geomagnetic changes in Hawaiian 1960 lava: Implications for possible causes of erroneously high
local atmospheric ionization. Geophysical Research Letters 35: L05811. http://dx. intensities. Geophysical Journal International 153(1): 263–276.
doi.org/10.1029/2007GL033,040. Yang S, Odah H, and Shaw J (2000) Variations in the geomagnetic dipole moment over
Usoskin IG, Mironova IA, Korte M, and Kovaltsov GA (2010) Regional millennial trend the last 12000 years. Geophysical Journal International 140: 158–162.
in the cosmic ray induced ionization of the troposphere. Journal of Atmospheric and Yokoyama Y and Yamazaki T (2000) Geomagnetic paleointensity variation with a 100
Solar Terrestrial Physics 72: 19–25. kyr quasi-period. Earth and Planetary Science Letters 181(1): 7–14.
Usoskin IG, Solanki SK, and Korte M (2006) Solar activity reconstructed over the last Yoshihara A, Kondo A, Ohno M, and Hamano Y (2003) Secular variation of the
7000 years: The influence of geomagnetic field changes. Geophysical Research geomagnetic field intensity during the past 2000 years in Japan. Earth and Planetary
Letters 33: L08103. http://dx.doi.org/10.1029/2006GL025,921. Science Letters 210: 219–231.
Valet J-P (2003) Time variations in geomagnetic intensity. Reviews of Geophysics Yu Y, Doh S-J, Kim W, et al. (2010) Archeomagnetic secular variation from Korea:
41. http://dx.doi.org/10.1029/2001/RG000104. Implications for the occurrence of global archeomagnetic jerks. Earth and Planetary
Valet J-P, Herrero-Bervera E, LeMoüel J-L, and Plenier G (2008) Secular variation Science Letters 294: 173–181.
of the geomagnetic dipole during the past 2000 years. Geochemistry, Yukutake T and Tachinaka H (1969) Separation of the Earth’s magnetic field into the
Geophysics, Geosystems 9: Q01008. http://dx.doi.org/10.1029/ drifting and the standing parts. Bulletin of the Earthquake Research Institute 47: 65.
2007GC001,728. Zananiri I, Batt CM, Lanos P, Tarling DH, and Linford P (2007) Archaeomagnetic
Valet J-P and Meynadier L (1993) Geomagnetic field intensity and reversals during the secular variation in the UK during the past 4000 years and its application to
past four million years. Nature 366: 234–238. archaeomagnetic dating. Physics of the Earth and Planetary Interiors 160: 97–107.
Valet J-P and Soler V (1999) Magnetic anomalies of lava fields in the Canary islands. Zatman S and Bloxham J (1997) Torsional oscillations and the magnetic field within the
Possible consequences for paleomagnetic records. Physics of the Earth and Earth’s core. Nature 388(6644): 760–763.
Planetary Interiors 115(2): 109–118. Ziegler L, Constable C, and Johnson C (2008) Testing the robustness and limitations of
Veitch RJ, Hedley IG, and Wagner J-J (1984) An investigation of the intensity of the 0–1 Ma absolute paleointensity data. Physics of the Earth and Planetary Interiors
geomagnetic field during Roman times using magnetically anisotropic bricks and 170: 34–45. http://dx.doi.org/10.1016/j.pepi.2008.07.027.
tiles. Archives des Sciences Geneve 37(3): 359–373. Ziegler L, Constable C, Johnson C, and Tauxe L (2011) PADM2M: A penalized
Vernov SN, Gorchakov EV, Shavrin P, and Sharvina KN (1967) Radiation maximum likelihood model of the 0–2 Ma paleomagnetic axial dipole
belts in the region of the South Atlantic Anomaly. Space Science Reviews moment. Geophysical Journal International 184: 1069–1089. http://dx.doi.org/
7: 490–533. 10.1111/j.1365-246X.2010.04905.x.
5.10 Geomagnetic Excursions
C Laj, Unité Mixte CEA-CNRS-UVSQ, Gif-sur-Yvette Cedex, France; Ecole Normale Supérieure, Paris, France
JET Channell, University of Florida, Gainesville, FL, USA

5.10.1.1 History of the Polarity Timescale and Excursions 343
5.10.1.2 Nomenclature for Excursions and Polarity Intervals 345
5.10.2 Geomagnetic Excursions in the Brunhes Chron 345
5.10.2.1 The Laschamp Excursion 348
5.10.2.2 The Mono Lake Excursion 354
5.10.2.3 The Blake Excursion 356
5.10.2.4 The Iceland Basin Excursion 358
5.10.2.5 The Pringle Falls Excursion 360
5.10.2.6 Excursions in the Early Brunhes Chron 362
5.10.3 Geomagnetic Excursions in the Matuyama Chron 363
5.10.3.1 Background 363
5.10.3.2 Excursions Between the Gauss–Matuyama Boundary and the Réunion Subchron 364
5.10.3.3 Huckleberry Ridge 367
5.10.3.4 Gilsa 367
5.10.3.5 Gardar and Bjorn 367
5.10.3.6 Cobb Mountain 367
5.10.3.7 Punaruu 367
5.10.3.8 Intra-Jaramillo Excursion (1r.1n.1r) 367
5.10.3.9 Santa Rosa 367
5.10.3.10 Kamikatsura 368
5.10.4 Geomagnetic Excursions in Pre-Matuyama Time 368
5.10.4.1 C5n.2n (Late Miocene) 368
5.10.4.2 Other Miocene Excursions/Subchrons 369
5.10.4.3 Oligocene and Eocene 369
5.10.4.4 Middle Cretaceous 370
5.10.5 Duration of Geomagnetic Excursions 371
5.10.6 Excursional Field Geometry 372
5.10.7 Concluding Remarks 373
Acknowledgments 375
References 376
5.10.1 Introduction The modern era of paleomagnetic studies began with the
studies of Hospers (1951, 1953–1954) in Iceland and Roche
5.10.1.1 History of the Polarity Timescale and Excursions
(1950, 1951, 1956) in the Massif Central of France. The work
Brunhes (1906) was the first to measure magnetization direc- of Hospers on Icelandic lavas was augmented by Rutten and
tions in rocks that were approximately antiparallel to the pre- Wensink (1960) and Wensink (1964) who subdivided
sent Earth’s field. Brunhes (1906) recorded magnetizations in Pliocene–Pleistocene lavas on Iceland into three polarity
baked sedimentary rocks that were aligned with reverse mag- zones from young to old: N–R–N. Magnetic remanence mea-
netization directions in overlying Miocene lavas from central surements on basaltic lavas combined with K–Ar dating, pio-
France (Puy de Dôme). In so doing, Brunhes (1906) made first neered by Cox et al. (1963) and McDougall and Tarling
use of a field test for primary thermal remanent magnetization (1963a,b, 1964), resulted in the beginning of development of
that is now referred to as the ‘baked contact’ test (see Laj et al., the modern geomagnetic polarity timescale (GPTS). These
2002a for an account of Brunhes’ work). Matuyama (1929) studies, and those that followed in the mid-1960s, established
was the first to attribute reverse magnetizations in (volcanic) that rocks of the same age carry the same magnetization polar-
rocks from Japan and China to reversal of geomagnetic polarity ity, at least for the last few million years. The basalt sampling
and to differentiate mainly Pleistocene lavas from mainly sites were scattered over the globe. Polarity zones were linked
Pliocene lavas based on the polarity of the magnetization. In by their K–Ar ages and were usually not in stratigraphic super-
this respect, Matuyama (1929) was the first person to use position. Doell and Dalrymple (1966) designated the long
magnetic stratigraphy as a means of ordering rock sequences. intervals of geomagnetic polarity of the last 5 My as magnetic

344 Geomagnetic Excursions
epochs and named them after pioneers of geomagnetism 1967). Heirtzler et al. (1968) made the inspired choice of a
(Brunhes, Matuyama, Gauss, and Gilbert). The shorter polarity particular South Atlantic MMA record (V-20). Several itera-
intervals (events) were named after localities: for example, tions of the polarity timescale (e.g., LaBrecque et al., 1977)
Jaramillo (Doell and Dalrymple, 1966), Olduvai (Grommé culminated with the work of Cande and Kent (1992a) that
and Hay, 1963, 1971), Kaena and Réunion (McDougall and reevaluated the MMA record and used the South Atlantic as
Chamalaun, 1966), Mammoth (Cox et al., 1963), and the fundamental template with inserts from faster-spreading
Nunivak (Hoare et al., 1968). The nomenclature for excursions centers in the Indian and Pacific Oceans. They then interpo-
has continued this trend by naming excursions after the local- lated between nine numerical age estimates that could be
ities from where the earliest records were derived (e.g., linked to polarity chrons over the last 84 My. Eight of these
Laschamp in France). ages were based on ‘high-temperature’ radiometric ages (no
The fit of the land-derived polarity timescale, from paleo- glauconite ages were included), and the youngest age
magnetic and K–Ar studies of exposed basalts, with the polarity (2.60 Ma) was the astrochronological age estimate for the
record emerging from marine magnetic anomalies (MMAs) Gauss–Matuyama boundary from Shackleton et al. (1990).
(Heirtzler et al., 1968; Pitman and Heirtzler, 1966; Vine, In the subsequent version of their timescale, hereafter referred
1966; Vine and Matthews, 1963) resulted in a convincing to as CK95, Cande and Kent (1995) adopted astrochronolo-
argument for synchronous global geomagnetic polarity rever- gical age estimates for all Pliocene–Pleistocene polarity rever-
sals, thereby attributing them to the main axial dipole. The sals (Hilgen, 1991a,b; Shackleton et al., 1990) and modified
results relegated self-reversal, detected in the 1950s in the the age tie point at the Cretaceous–Tertiary boundary from
Haruna dacite from the Gunma Prefecture in Japan (Nagata, 66 to 65 Ma.
1952; Nagata et al., 1957; Uyeda, 1958), to a rock magnetic The astrochronological estimates for Pliocene–Pleistocene
curiosity rather than to a process generally applicable to volca- polarity chrons (Hilgen, 1991a,b; Shackleton et al., 1990),
nic rocks. incorporated in CK95, have not undergone major modifica-
The first magnetic stratigraphies in sedimentary rocks may tion in the last 15 years and indicated the way forward for
be attributed to Creer et al. (1954) and Irving and Runcorn timescale calibration further back in time. Subsequent to
(1957) who documented normal and reverse polarities at CK95, astrochronological estimates of polarity chron ages
13 locations of the Proterozoic Torridonian sandstones in have been extended into the Miocene, Paleogene, and
Scotland and in rocks of Devonian and Triassic age. For the Cretaceous (Abdul-Aziz and Langereis, 2004; Abdul-Aziz
Torridonian sandstones, normal and reverse magnetizations et al., 2000, 2003; Hilgen et al., 1995, 2000; Krijgsman et al.,
were observed from multiple outcrops, and an attempt was 1994, 1995). Many of these advances are included in the
made to arrange the observed polarity zones in stratigraphic timescale of Lourens et al. (2004). At present, more or less
sequence. Meanwhile, Khramov (1960) published magnetic continuous astrochronologies tied to polarity chrons are avail-
polarity stratigraphies in Pliocene–Pleistocene sediments able back to the Oligocene (e.g., Billups et al., 2004).
from western Turkmenia (Central Asia) and made chronos- Recognition of brief polarity excursions as an integral part
tratigraphic interpretations based on equal duration of polar- of the Earth’s paleomagnetic field behavior has developed
ity intervals. Early magnetostratigraphic studies were carried alongside astrochronological calibration of the polarity time-
out on Triassic red sandstones of the Chugwater Formation scale in the last 20 years. Although the first recognized polar-
(Picard, 1964), on the European Triassic Buntsandstein ity excursions (the Laschamp and Blake excursions) were
(Burek, 1967, 1970), and on the Lower Triassic Moenkopi documented in the late 1960s (Bonhommet and Babkine,
Formation (Helsley, 1969). The aforementioned studies were 1967; Smith and Foster, 1969), excursions were widely con-
conducted on poorly fossiliferous mainly continental (red) sidered to represent either spurious recording artifacts or, at
sandstones and siltstones; therefore, the correlations of polar- best, local anomalies of the geomagnetic field and of doubt-
ity zones did not have support from biostratigraphic correla- ful utility in stratigraphy. It was not until the late 1980s that
tions. It was the early magnetostratigraphic studies of the tide began to turn. As high-resolution sedimentary records
Pliocene–Pleistocene marine sediments, recovered by piston from the deep sea became available, it became accepted that
coring in high southern latitudes (Hays and Opdyke, 1967; excursions are frequent (with perhaps more than seven in the
Ninkovitch et al., 1966; Opdyke et al., 1966) and in the Brunhes Chron) and that they appear to be globally recorded
equatorial Pacific Ocean (Hays et al., 1969), that marked and not local anomalies of the geomagnetic field. Geomag-
the beginning of modern magnetic stratigraphy. These studies netic excursions were not recorded in the early days of mag-
combined magnetic polarity stratigraphy and biostratigraphy netic stratigraphy because the high sedimentation rate
and in so doing refined and extended the GPTS derived from sequences, required to record brief (a few kiloyears duration)
paleomagnetic studies of basaltic outcrops (e.g., Doell magnetic excursions, were generally not targeted during con-
et al., 1966). ventional piston coring expeditions. Indeed, at the time, low
Heirtzler et al. (1968) produced a polarity timescale for sedimentation rate sequences were often preferred as they
the last 80 My using South Atlantic MMA record (V-20) as the allowed the record to be pushed further back in time. The
polarity template. They assumed constant spreading rate and development of hydraulic piston coring (HPC) techniques,
extrapolated ages of polarity chrons using an age of 3.35 Ma first used in early 1979 during the Deep Sea Drilling Project
for the Gilbert/Gauss polarity chron boundary. This was a (DSDP) Leg 64, brought high sedimentation rate sequences
dramatic step forward that extended the GPTS to 80 Ma, within reach by increasing penetration by a factor of 15,
from 5 Ma that had been the limit based on magnetostrati- from 20 m for conventional piston coring to 300 m for
graphic records available at the time (e.g., Hays and Opdyke, the HPC.
Geomagnetic Excursions 345
5.10.1.2 Nomenclature for Excursions and Polarity Intervals low- or mid-latitude VGPs, rather than high-latitude reverse
VGPs, probably very often reflect inadequacy of the recording
The International Stratigraphic Commission has guided the
medium (Channell and Guyodo, 2004a,b; Roberts and
use of magnetostratigraphic units (polarity zones), their time
Winklhofer, 2004), inadequate rates of sediment accumulation,
equivalents (polarity chrons), and chronostratigraphic units
and/or inadequate sampling methods, rather than geomagnetic
(polarity chronozones) (see Anonymous, 1977; Opdyke and
characteristics. We therefore advocate the use of the term
Channell, 1996). The revelation in the last 20 years of numer-
microchron for brief polarity chrons with established duration
ous short-lived excursions within the Brunhes and Matuyama
less than 104 years (Table 1). The term ‘excursion’ would then be
Chrons, and probably throughout the history of the Earth’s
used only for features that represent departures from normal
magnetic field, requires an extension of terminology to include
secular variation, for which full polarity reversal has not been
excursions and brief polarity microchrons (Table 1). Terminol-
established. As these features become better documented, they
ogy based on duration of polarity intervals is obviously prob-
could then be elevated to the status of microchron, a term that
lematic at the low end of the duration spectrum. Records are
denotes a brief polarity chron. Under this nomenclature, the
always compromised by limitations of the recording medium
Laschamp and Iceland Basin ‘excursions’ would be elevated to
and, in addition, estimates of duration are limited by the
polarity microchron status as it has been established that these
availability of chronological tools of adequate precision. In
represent paired full polarity reversals defining a distinct polarity
the realm of MMAs, LaBrecque et al. (1977) applied the
interval.
term ‘tiny wiggles’ to minor, but lineated, MMAs. The time
equivalent of the ‘tiny wiggle’ was labeled ‘cryptochron’
(Cande and Kent, 1992a,b), expressing the uncertain origin
of ‘tiny wiggles’ as either short-lived polarity intervals or 5.10.2 Geomagnetic Excursions in the Brunhes Chron
paleointensity fluctuations. These authors placed the duration
separating ‘polarity chrons’ and ‘cryptochrons’ at 30 ky, repre- In the last few decades, numerous geomagnetic excursions
senting an estimate of the minimum duration of polarity have been discovered within the Brunhes Chron (Figure 1
intervals that can be resolved in MMA records. For magnetos- and Table 2). The Laschamp excursion, which is now known
tratigraphy, Krijgsman and Kent (2004) advocated a duration to have an age of 40 ka, was the first geomagnetic excursion
cutoff for separating ‘subchrons’ and ‘excursions,’ at 9–15 ky. to be recognized, in lavas from the French Massif Central
The drawback of such a scheme is that, in most cases, the (Bonhommet and Babkine, 1967). It is the most thoroughly
chronostratigraphic precision will be inadequate to establish studied excursion, and its existence is proved beyond doubt.
the distinction. An alternative is that an excursion be defined in This is not the case for other reported excursions in the late
terms of a ‘brief’ (<104 years) deviation of virtual geomagnetic Brunhes Chron, and some of these can be attributed to
poles (VGPs) from the geocentric axial dipole that lies outside sedimentologic and/or sampling artifacts: for example, the
the range of secular variation for a particular population of Starno event (Noel and Tarling, 1975), the Gothenburg ‘flip’ (),
VGPs. Some authors have adopted an arbitrary VGP cutoff and the Lake Mungo excursion (Barbetti and McElhinny,
(say a colatitude of 45 ) to define an excursion, although 1972). Other early papers that documented excursions have
Vandamme (1994) has advocated a method of calculating been ratified by later work: for example, the reported excursion
the cutoff for a specific VGP population based on VGPs lying (Blake excursion) in marine isotope stage (MIS) 5 from the
outside ‘normal’ secular variation. As higher fidelity records of Blake Outer Ridge (Smith and Foster, 1969). Wollin et al.
excursions have become available, it appears that the majority of (1971) documented three short intervals of reverse paleomag-
‘excursions’ is manifest as directional changes through 180 , netic inclination in cores from the Caribbean and the Eastern
followed by a return to the preexcursional directions within a Mediterranean. An approximate age model suggested that the
few thousand years. For example, the Laschamp and Iceland most recent of these was coeval with the Blake excursion, and
Basin excursions, although short-lived with durations of a few the ages of the other two were estimated to be around 180–210
kiloyears, represent paired reversals of the geomagnetic field as and 270 ka. Kawai et al. (1972) and Yaskawa et al. (1973)
VGPs reach high southerly latitudes for both excursions reported a paleomagnetic study of a 197 m core from Lake
(Channell, 1999; Laj et al., 2006). ‘Excursions’ displaying Biwa (Japan) that documented five short episodes of reverse
Table 1 Nomenclature for polarity intervals and excursions
Magnetostratigraphic polarity zone Geochronological (time) equivalent Chronostratigraphic equivalent Duration (year)
Polarity megazone Megachron Megachronozone 108–109

Polarity superzone Superchron Superchronozone 107–108
Polarity zone Chron Chronozone 106–107
Polarity subzone Subchron Subchronozone 105–106
Polarity microzone Microchron Microchronozone <104
Excursion zone Excursion Brief departure from normal secular
variation
Polarity cryptochron Cryptochron Cryptochronozone Uncertain existence
Age (ka)
(1) (2) (3) (4)
0
Mono Lake (28 ka) Mono Lake (24 ka)

3a Mono Lake (32 ka)
Laschamp (40–45 ka) Laschamp (42 ka) 3b (40 ka) Laschamp (41 ka)
Norwegian-
Greenland Sea
(70–80 ka) Fram Strait
(91–100 ka)
100
5a
Blake (110–120 ka) Blake (117 ka)
5b (125 ka) Blake (120 ka)
Albuquerque-
Fram Strait
(155–165 ka)
Biwa 1 (187–192 ka)

200 7a (195 ka) Iceland Basin (188 ka)
Jamaica-
Pringle Falls Pringle Falls (211 ka)
(205–215 ka)
7b (235 ka)
Fram Strait
Calabrian Ridge 0 8a (260 ka)
(255–265 ka) Biwa II (272 ka)
300 Calabrian
9a (310 ka)
Ridge 1
(315–325 ka) 9b (330 ka)
Levantine
(360–370 ka)
Biwa III (368–402 ka)
400
11a (410 ka)
Un-named
(400–420 ka)
Emperor (469 ka)
500
Calabrian 13a (510 ka)
Ridge 2/W. Eifel
(515–525 ka)
14a (535 ka)
Big lost (550–554 ka)

Emperor-
Big Lost- 15a (575 ka) Big Lost (560–580 ka)
Calabrian Ridge 3
(560–570 ka)
600 15b (605 ka)
Delta (660 ka) 17a (665 ka)
Stage 17 (690 ka)

700
Figure 1 Geomagnetic excursions in the Brunhes Chron according to various authors. Column 1: Langereis et al. (1997): asterisks mark ‘well-dated,
global’ excursions, others were deemed ‘restricted or not as certain’; column 2: Worm (1997); column 3: Lund et al. (2001a); column 4:
well-documented excursions with acceptable age control (this study).
Table 2 Excursions within the Brunhes Chron
Excursion name MIS Estimated Estimated Location or ODP Leg Principal references
age (ka) duration (ky)
Mono Lake 3 33 1 Mono Lake/Arctic/152 (1)

Laschamp 3 41 1.5 Atlantic realm/Arctic/177/172/152 (2)
Norwegian–Greenland Sea 60 Arctic (3)
Blake 5d/5e 120 5 North Atlantic/Med/172 (4)
Iceland Basin 6/7 188 3 Atlantic/Pacific/Baikal/152/162/ (5)
172/177/184
Pringle Falls 7 211 Western United States/New (6)
Zealand/172/152
Calabrian Ridge 0 8 260 Med/172 (7)
Calabrian Ridge I 9 325 Med/172 (8)
Unnamed 11 400 172 (9)
Calabrian Ridge II 13 525 Med/172 (10)
Big Lost 14/15 560–580 Yellowstone/172/162 (11)
Stage 17 17 670 Osaka Bay/172/162 (12)
Italics indicate poorly documented excursions; MIS, marine isotope stage.

(1) Denham and Cox (1971), Liddicoat and Coe (1979), Negrini et al. (1984, 2000), Lund et al. (2001a), Nowaczyk and Antonow (1997), Nowaczyk and Knies (2000), Channell (2006).
(2) Bonhommet and Babkine (1967), Bonhommet and Zahringer (1969), Levi et al. (1990), Laj et al. (2000), Lund et al. (2001a), Mazaud et al. (2002), Laj et al. (2006),
Channell (2006).
(3) Nowaczyk et al. (1994), Nowaczyk and Frederichs (1999).
(4) Smith and Foster (1969), Tric et al. (1991), Zhu et al. (1994), Thouveny et al. (2004).
(5) Channell et al. (1997), Weeks et al. (1995a,b), Roberts et al. (1997), Channell (1999), Oda et al. (2002), Stoner et al. (2003), Laj et al. (2006), Channell (2006).
(6) Herrero-Bervera et al. (1989), Herrero-Bervera et al. (1994), McWilliams (2001), Singer et al. (2005), Channell (2006).
(7) Kawai (1984), Langereis et al. (1997), Lund et al. (2001b).
(8) Lund et al. (2001b).
(9) Lund et al. (2001b).
(10) Langereis et al. (1997), Lund et al. (2001b).
(11) Champion et al. (1988), Lund et al. (2001b), Quidelleur et al. (1999), Singer et al. (2002), Channell et al. (2004).
(12) Biswas et al. (1999), Lund et al. (2001b), Channell and Raymo (2003), Channell et al. (2004).
polarity in the Brunhes Chron. On the basis of a tentative be the Blake Event, at about 105–114 ka. Four additional
correlation with the results of Wollin et al. (1971), the youn- episodes of low inclination were named a, b, g, and d. Kawai
gest of these episodes was correlated to the Blake excursion and (1984) suggested that the youngest of the Biwa episodes could
the other two were labeled Biwa I at about 176–186 ka and correspond to the Blake excursion, the a episode to Biwa I, the
Biwa II at about 292–298 ka. Later, Kawai (1984), on the basis b excursion to Biwa II, and the g episode to Biwa III. Poor age
of fission track ages on zircons from tephra layers interbedded control and poor definition of magnetization directions at
in the sediments, obtained ages of 100, 160, and 310 ka for the Gioia Tauro (and Lake Biwa) make these correlations
first three episodes and 380 ka for a fourth episode, which he equivocal.
called Biwa III. As with much of the evidence for directional Champion et al. (1981) found evidence for a brief reversal
excursions published in the 1960s and 1970s, it was based on in a sequence of basalt flows in the Snake River Plain (Idaho)
magnetization directions that were poorly defined (by modern to which they assigned an age of 465 50 ka, using K–Ar ages
standards) and on age control that provided plenty of flexibil- of bracketing normally magnetized flows. They correlated this
ity in their correlation to other supposed excursions. Apart event with the Emperor Event of Ryan (1972). Later, however,
from the Blake and the Laschamp excursions that have stood Champion et al. (1988) revised the age of these reverse polarity
the test of time, we advocate abandoning the labels such as lavas to 565 10 ka. The new age implies a new reverse epi-
Biwa that refer to excursions that are poorly defined with ages sode, which they named the Big Lost excursion. Based on the
that are poorly constrained. literature available at the time, Champion et al. (1988) pro-
In exposed sediments at Mono Lake (California), an excur- posed the existence of eight reverse polarity microchrons in the
sion apparently younger than the Laschamp excursion was Brunhes Chron (Figure 1).
documented by Denham and Cox (1971) and subsequently Langereis et al. (1997) reviewed evidence for excursions in
by Liddicoat and Coe (1979). This excursion was named the the Brunhes Chron and made the case for seven well-dated
Mono Lake excursion after its type locality. Although the excur- ‘global’ excursions and five ‘restricted’ less certain Brunhes
sion has now been observed elsewhere, the age of the excursion excursions, for a total of 12 excursions in the Brunhes Chron
at the type locality remains controversial. (Figure 1). From a central Mediterranean core, these authors
Creer et al. (1980) studied a drill-core section at Gioia added a series of excursions (Calabrian Ridge 0, 1, 2, 3) to the
Tauro in southern Italy and suggested that an episode of excursion vocabulary, the oldest (CR3) being correlative to the
reverse polarity found in the upper part of the section could Big Lost excursion. Worm (1997) suggested a link between
geomagnetic reversals/excursions and glaciations and listed ten Considerable effort has been dedicated to determination of
geomagnetic excursions in the Brunhes Chron that he consid- the age of the Laschamp excursion. The first determination,
ered to be adequately documented (Figure 1). In more recent using K–Ar methods on whole rock by Bonhommet and
papers attempting a synthesis, Lund et al. (2001a,b, 2006) Zahringer (1969), yielded an age between 8 and 20 ka. Subse-
proposed 17 excursions during the Brunhes Chron. The data- quent determinations did not confirm such a young age: Hall
base for geomagnetic excursions has clearly evolved rapidly in and York (1978) obtained whole rock K–Ar and 40Ar/39Ar
recent years, but with little consensus on the existence or age of dates of 47.4 1.9 and 45.4 2.5 ka. Condomines (1978),
Brunhes-aged excursions (Figure 1). using the 230Th/238U radioactive disequilibrium method,
In this section, we first describe evidence for the five best- obtained an age of 39 6 ka, which is more than double the
documented excursions in the Brunhes Chron: the Laschamp, age estimate of Bonhommet and Zahringer (1969). Gillot et al.
Mono Lake, Blake, Iceland Basin, and Pringle Falls excursions. (1979), using the unspiked (Cassignol) K–Ar method,
We then describe other less well-documented excursions obtained an age of 43 5 ka for the Laschamp flow and
recorded in the early Brunhes Chron (Table 2). 50 7.5 ka for the Olby flow. An early attempt to date the
Laschamp excursion using thermoluminescence (Wintle,
1973) failed because of anomalous fading of the signal.
5.10.2.1 The Laschamp Excursion
Huxtable et al. (1978) mitigated the problem by using sedi-
The Laschamp excursion was the first reported geomagnetic ments baked by the Olby flow and obtained an age of
excursion and is certainly the best-known excursion in the 25.8 1.7 ka, which is somewhat younger than the 35 5 ka
Brunhes Chron. Bonhommet and Babkine (1967) discovered age determined by Gillot et al. (1979) from quartz contained
this excursion and named it after the Puy de Laschamp, in the in a granitic inclusion found in the Laschamp flow. Gillot et al.
French Chaı̂ne des Puys (Massif Central, France). The flow (1979) also obtained an age of 38 6 ka from five quartz
carries an anomalous characteristic paleomagnetic direction, pebbles found in a baked paleosol beneath the Olby flow.
up to 160 away from the expected dipole field direction. The In a recent determination of the age of the excursion,
excursion was initially discovered in the flow and scoria in the Guillou et al. (2004) combined K–Ar and 40Ar/39Ar results
Puy de Laschamp, but the same anomalous direction has also from the two basaltic flows at Laschamp and Olby to better
been found in the nearby Olby flow (Figure 2). resolve the age of the Laschamp excursion. This was possible in
D 68
Col de la
Moréno
Laschamp
D
52 Laschamp-1
Laschamp-2
Clermont-
Ferrand
Puy Monchler
45°
Olby 45’
Col de la Moreno
Olby Laschamp
Flow Puy de la
Moreno
Flow
Puy de
Laschamp
1 km
2° 55’
Figure 2 Location map of the sites at Laschamp and Olby in the Chaı̂ne des Puys, central France. Reproduced from Guillou H, Singer BS, Laj C, Kissel
C, Scaillet S, and Jicha B (2004) On the age of the Laschamp geomagnetic event. Earth and Planetary Science Letters 227: 331–343.
part due to recent advances in 40Ar/39Ar incremental heating Mediterranean tephra layer (Ton-That et al., 2001) as well as
methods using a resistance furnace, which can yield ages for data from Greenland ice cores where the Laschamp excursion is
basaltic lava flows between 100 and 20 ka with a precision manifested as a sharp and short increase in 10Be and 36Cl flux
better than 5% at the 95% confidence level (Heizler et al., (Muscheler et al., 2004; Raisbeck et al., 2007; Svensson et al.,
1999; Singer et al., 2000). Similarly, the unspiked K–Ar dating 2006; Wagner et al., 2000) and data from sediments that
method, with the new calibration procedure of Charbit et al. indicate an age for the excursion of 40–41 ka and a duration
(1998), allows accurate and precise measurements of minute of 1 ky (Channell et al., 2012; Evans et al., 2007; Laj et al.,
quantities of radiogenic argon (Guillou et al., 1998). The 2000, 2004).
results of six new unspiked K–Ar and 13 40Ar/39Ar incremental Singer et al. (2009), using the same analytic techniques as
heating experiments on subsamples from three sites in the Guillou et al. (2004), reported results from a third type locality
Laschamp and Olby flows are concordant and yield a weighted for the Laschamp excursion, the Louchadière flow, and pro-
mean age of 40.4 1.1 ka. Consideration of the uncertainties posed an age of 40.7 0.95 ka (2s), virtually identical to the
in the 40K/40Ar decay constant led to 40.4 2.0 ka (2s analytic value given by Guillou et al. (2004). Singer et al. (2009) also
error plus decay constant uncertainties) as the most probable questioned the accuracy of both the K–Ar ages and the
age for the Laschamp excursion (Figure 3). long 6000 years duration for the Laschamp excursion pro-
There are, however, still controversies. Plenier et al. (2007) posed by Plenier et al. (2007), citing crucial assumptions
reported unspiked K–Ar ages from several lava flows in the related to the K–Ar method (Dalrymple and Lanphere, 1969;
Chaı̂ne des Puys. These include an age of 37.0 1.4 ka (2s) Sharp and Renne, 2005; Singer, 2007). Indeed, even small
for the Olby flow, which is inconsistent with that of Guillou deviations in isotopic composition of the initial argon from
et al. (2004). On the basis of this age and the ages 41.9 2.8 atmospheric composition can result in significant errors in the
and 33.4 1.4 ka obtained from normally and transitionally apparent age of very young lava flows containing very small
magnetized lavas from other flows in the Chaı̂ne des Puys, amounts of radiogenic argon (40Ar*), such as those having
Plenier et al. (2007) concluded that the Laschamp excursion recorded the Laschamp excursion. This stresses the importance,
began at about 39.6 ka and lasted about 6 ka until 33.3 ka. The in particular for periods concerned by very fast changes in the
conclusion was in conflict with K–Ar and 40Ar/39Ar results Earth’s magnetic field, of coupled K–Ar and 40Ar/39Ar dating,
from the same sites (Guillou et al., 2004) and from a because the latter allows validation of basic assumptions of the
K–Ar clock and consequently helps to validate the radiometric
ages. In parallel, Cassata et al. (2008) obtained a 40Ar/39Ar age
for the Laschamp excursion in New Zealand of 39.1 4.1 ka.
Over 10 years after the Laschamp excursion was first
reported in the literature, doubts were expressed about its
geomagnetic origin. Heller (1980) and Heller and Petersen
(1982) noted partial or complete self-reversal of the natural
remanent magnetization (NRM) of many Olby samples, and to
a lesser extent of Laschamp samples, during thermal laboratory
experiments. They suggested that the almost reverse magneti-
zation in the two flows could be attributed to self-reversal
properties, rather than to the geomagnetic field. This sugges-
tion was at odds with earlier work of Whitney et al. (1971) who
concluded that the excursional magnetization was carried by
single-domain magnetite grains, a magnetic mineralogy incon-
sistent with suggested mechanisms for self-reversal. Roperch
et al. (1988) pointed out that, when more than one magnetic
phase is present, the species with the lowest blocking temper-
ature usually experiences self-reversal by magnetostatic inter-
action, while the species with higher (>200 C) blocking
temperature invariably carries the primary reverse direction.
Furthermore, Roperch et al. (1988) found a reverse direction
in a sample of clay, which had been baked by the overlying
Olby flow. Although limited to one sample (due to the diffi-
culty of finding clays baked by the flow), this result provided
evidence for a geomagnetic origin of the Laschamp excursion.
Roperch et al. (1988) also used the Thellier–Thellier paleoin-
tensity method to obtain a value of 7.7 mT (i.e., less than one
sixth of the present field) for the reverse polarity flows at
Figure 3 Radioisotopic ages of the Laschamp and Olby flows. N is Laschamp and Olby. They argued that this low value is more
the number of individual age determinations. The broad characteristic of transitional geomagnetic field behavior and,
probability–density curve with a maximum at 46 ka represents therefore, that the paleomagnetic directions of the Laschamp
previously published K–Ar and 40Ar/39Ar ages, whereas the narrow and Olby flows were not acquired during a period of stable
envelope peaking at 40.5 ka is from Guillou et al. (2004). reverse polarity. In the Chaı̂ne des Puys, Barbetti and Flude
(1979) obtained low paleointensity values from sediments excursion was not found in sediments from Lac du Bouchet, a
baked by the lava flow at Royat. Here, paleointensity data maar lake only about 100 km from Laschamp floored by sedi-
imply field strength about 30% of its present value and a ments of comparable age (Creer et al., 1990). Nowaczyk and
40
Ar/39Ar age of around 40 ka (Hall et al., 1979), which is Antonow (1997) found evidence for the Laschamp excursion in
close to that of the Laschamp and Olby flows. In the Chaı̂ne four sediment cores from the Greenland Sea. Subsequently,
des Puys, Chauvin et al. (1989) obtained paleointensity results Nowaczyk and Knies (2000) reported directional and relative
from different flows on the Louchadière volcano, where K–Ar paleointensity (RPI) records of the Laschamp and Mono Lake
ages indicated approximate synchroneity with the Laschamp excursions from the Arctic Ocean. One of the cores, however,
flow. Chauvin et al. (1989) documented anomalous paleo- documented in this study (PS 2212-3 KAL) illustrates an
magnetic directions (although different from the Laschamp anomaly common to these high-latitude cores: The cumulative
directions) and determined a paleointensity of 12.9 3.3 mT, percentage thickness of reverse polarity zones in sediment accu-
which is about one-third the present field value, confirming mulated during the last few 100 ky is far greater than expected
the transitional character of the Louchadière flow. (50% for the top 4 m of the section representing the last
In Iceland, Kristjansson and Gudmundsson (1980) found 120 ky in PS2212-3 KAL). Fortuitous fluctuations in sedimenta-
evidence for an anomalous paleomagnetic direction from three tion rates, which are not resolvable in the age models, appear to
different localities in the Reykjanes Peninsula, which they called have ‘amplified’ these excursional records. The anomalous
the Skalamaelifell excursion. Levi et al. (1990) subsequently excursional records from the Arctic Ocean and the Norwegian–
identified the same excursion at four additional localities in the Greenland Sea may be explained by the unusual diagenetic
same region, where they obtained a K–Ar age of 42.9 7.8 ka, conditions, due to low levels of labile organic matter, that lead
thereby associating the Skalamaelifell excursion with the to partial maghemitization of detrital titanomagnetite (Channell
Laschamp excursion. Levi et al. (1990) also obtained a paleoin- and Xuan, 2009). Maghemitized rims on detrital titanomagne-
tensity determination of 4.2 0.2 mT, which is consistent with an tite can lead to partial self-reversal of remanence directions in
earlier result by Marshall et al. (1988) and similar to that deter- individual grain that can explain the anomalous occurrence of
mined by Roperch et al. (1988) on the Laschamp and Olby flows thick intervals of ‘excursional’ magnetization directions in
and almost an order of magnitude less than the present field sediments from this region (Xuan and Channell, 2010; Xuan
strength in Iceland. Ferk and Leonhardt (2009) had resampled et al., 2012).
these and new Icelandic flows and obtained both directional data In the North Atlantic paleointensity stack (NAPIS-75) of Laj
and paleointensity results using a Thellier-type technique. They et al. (2000), all but one of the six cores are characterized by
also proposed a tentative stratigraphic relationship between the changes in characteristic magnetization directions associated
25 flows prior to, during and after the excursion by comparing with a well-marked paleointensity low around 41 ka. In two
the data with VGPs of the Laschamp excursion from different of the cores, the inclinations reach negative values in excess
marine sedimentary records. of –30 , a third one reaching 90 . Correlation among the
The age of the Skalamaelifell excursion has, however, been NAPIS-75 cores was determined using MIS boundaries with a
questioned by Jicha et al. (2011). These authors have per- second step involving matching cycles observed in downcore
formed nine 40Ar/39Ar incremental heating experiments on profiles of the anhysteretic remanent magnetization (ARM)
four flows carrying transitional magnetizations. For each indi- (Kissel et al., 1999). The chronology of NAPIS-75 (Laj et al.,
vidual sample, the uncertainty in the determination of the age 2000) is based on correlation of the planktonic d18O record
is of the order of the age itself, due to the extremely low from core PS2644-5 located close to Greenland (Voelker et al.,
amount of potassium in these rocks. On the other hand, 1998) to the d18O record from the GISP2 ice core (Grootes and
when averaged, the results yield an age of 91 13 ka, which Stuiver, 1997). Light d18O values in core PS2644-5 corres-
is consistent with the age of 96 13 ka obtained from parallel pond to iceberg discharges coeval with the glacial stadials
238
U/230Th experiments on the same samples. The authors numbered from 3 to 17, allowing 14 ties from the ice-core
suggest that the Skalamaelifell excursion could correspond to chronology to the sedimentary sequence. Placed on this age
a post-Blake excursion at 94 ka (Thouveny et al., 2004), which model, it appears that the Laschamp excursion occurred at
also coincides with a globally observed low in paleointensity. the end of glacial interstadial 10 (Baumgartner et al., 1997; Laj
Although sedimentary records often document the character- et al., 2000; Wagner et al., 2000), at about 41 ka. This age
istic paleointensity minimum associated with the Laschamp remains unchanged in the more recent ice age model con-
excursion, a large proportion of records fail to document excur- structed by yearly based layer counting in the NorthGRIP ice
sional directions associated with the minimum. This is likely due core (model GICC05, Andersen et al., 2006; Svensson et al.,
to low sedimentation rates (low resolution of the sedimentary 2006) and is consistent with the radiometric dating of the
records) combined with the fact that the directional anomaly excursional lava flow at the Laschamp, Olby, and Louchadière
associated with the Laschamp excursion may be brief (less than localities (Guillou et al., 2004; Singer et al., 2009) and of
1 ky) compared to the duration of the associated paleointensity the latest Mediterranean tephra layer corresponding to the
minimum. Any delay in remanence acquisition due to biotur- Campanian ignimbrite (Ton-That et al., 2001). The precise age
bation and a progressive magnetization lock-in, combined with models for the NAPIS-75 cores place the minimum paleointen-
inefficient alignment of the detrital remanent magnetization sity at about 40.9 ky associated with a duration for the direc-
(DRM) at times of low field intensity, may lead to smearing of tional excursion of about 500 years. The duration of the
the NRM directions, thereby inhibiting the recording of brief paleointensity low associated with the excursion is about 1500
directional events (Channell and Guyodo, 2004a,b; Roberts and years (Figure 4). These estimations benefit from the direct
Winklhofer, 2004). In central France, for example, the Laschamp correlation of NAPIS-75 sediments and Greenland ice cores.
90
60
30
Inclinations
PS 2644-5
SU90-33
0
MD95-2009
10 MD95-2034
NAPIS-75
SU90-24 -30
8
-60
VADM (1022 Am2)
6 -90
10Be
12
and 36Cl related VADM (1022 Am2)

2 10
8
0
0
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
Time (GISP scale) (kyr)
Figure 4 The NAPIS-75 stack. Upper part: inclination records of the five cores used in NAPIS-75. For ages younger than 38 ka, only the inclination
record of PS2644-5 is shown. Middle part: records of the virtual axial dipole moments (VADMs) of the five cores and of the NAPIS-75 (black curve).
Bottom part: VADM records obtained from measurements of 36Cl (green) and 10Be (blue) records using the polynome of Masarik and Beer (1999).
Paleointensity profiles characterized by average sedimenta- reaching 80 . The directions have been interpreted to display
tion rates ranging between 7 and 34 cm ky1 have been com- oscillatory behavior, growing in amplitude during the decrease
piled to construct a paleointensity stack valid at global scale in paleointensity, with an estimated time constant of 1200
(GLOPIS-75; Laj et al., 2004 and references therein). A total of years based on a sedimentation rate assumed to be constant
24 records obtained from widely distributed sites, from the at 28 cm ky1 during the excursion. Lund et al. (2005) calcu-
Southern Hemisphere, Tropics, and Northern Hemisphere, lated the rates of intensity and directional changes during the
have been considered in the construction of this stack (see excursion, based on oxygen isotope stratigraphy. For paleoin-
Laj et al., 2004 for references). The individual age models, tensity, these rates are typically less than 50 nT year1 and
mainly based on oxygen isotope stratigraphy, were tuned to never more than 150 nT year1. Although these values are
NAPIS-75, and therefore, the resulting stack was placed on the averaged over some 100–130 years and are strongly dependent
Greenland ice-core age model. A well-marked intensity low on the assumed age models, they are not significantly different
associated with the Laschamp excursion was recorded in all from the mean annual rate of change of the historic field
profiles confirming the global character of the Laschamp (Peddie and Zunde, 1988). Similarly, the rates of directional
excursion. change are of the order of 30–70 arcmin year1, which com-
Lund et al. (2005) presented two records of the Laschamp pares with the maximum rate of 40 arcmin year1 observed
excursion from sediment cores from the Bermuda Rise and the historically.
Blake Outer Ridge in the Western North Atlantic Ocean. The In order to describe the morphological changes of the field
sediment accumulation rates are estimated to be around during geomagnetic excursions, Laj et al. (2006) compiled five
20–28 cm ky1 in the interval recording the Laschamp excur- records of the Laschamp excursion obtained from rapidly
sion. In these records, the excursional declinations display deposited sediments at widely separated sites (the Bermuda
almost a full reversal (120 change in direction), while the Rise, the Blake Outer Ridge, the Greenland Sea, the Orca
inclinations change from 49 to 49 with intermediate values Basin of the Gulf of Mexico, and the southern Indian Ocean)
(Figure 5). This compilation yields a coherent picture of excur- (Channell et al., 2012), also shown in Figures 5 and 6, is
sional field behavior during Laschamp excursion in which the consistent with these paths. The VGP path of the Laschamp
excursional VGPs trace a clockwise loop (Figure 6), moving excursion as recorded at ODP Site 919 in the Irminger Basin
southward over East Asian–West Pacific longitudes, reaching (off east Greenland) (Channell, 2006) also describes a clock-
high southern latitudes, followed by a northward-directed VGP wise loop, albeit on different longitudes, and the VGPs do not
path over Africa and Western Europe. The turning point where reach latitudes in excess of 20 S. Here, the mean sedimentation
the VGPs change from being southward-directed to northward- rate in the vicinity of the excursion exceeds 20 cm ky1 and,
directed coincides with the minimum in RPI. A subsequent according to the oxygen isotope age model, the age and dura-
record of the Laschamp excursion from ODP Site 1063 tion of the excursion is 40 ka and <2 ky, respectively.
ODP-984
ODP-983
PS2644-5
MD99-2242
ODP 1063
MD95-2034 Black Sea
Lake Baikal ODP-884 MD02-2551
CH89-9 MD99-2247
JPC-14 ODP-1146 MD02-2552
ODP-1145
MD94-103
Iceland Basin excursion
Laschamp excursion
Figure 5 Map of core locations that have yielded records of the Laschamp excursion (squares) and the Iceland Basin excursion (circles).
PS2644-5
ODP1063
MD02-2551
MD95-2034
CH89-9
JPC-14 MD02-2552
MD94-103
Figure 6 Transitional magnetization directions for the Laschamp excursion represented as virtual geomagnetic polar (VGP) paths. The large arrow
illustrates the sense of looping, which is consistently clockwise.
Based on the Laj et al. (2006) compilation, Leonhardt et al. Black Sea display a first short loop over NE America reaching
(2009) had performed a Bayesian inversion of several records South Florida before the main excursional path, which coincides
of the Laschamp excursion and had obtained a scenario in with the Laj et al. (2006) trends for the N ! R path but not for
which inverse magnetic flux patches emerge near the equator the R ! N return path, which is over the central Indian Ocean.
at the core–mantle boundary and move poleward, and con- The total duration of these directional variations, estimated at
trary to the scenario for the Matuyama–Brunhes boundary, 3 ky, generally exceeds that estimated for the duration of the
they do not cross the boundary of the inner-core tangent Laschamp directional excursion from other sedimentary records
cylinder, which could establish a clear distinction between (e.g., Channell et al., 2012; Laj et al., 2000) and the cosmogenic
reversals and excursions. This Bayesian inversion necessitates isotope peak in the Greenland ice core (Svensson et al., 2008).
that one of the sedimentary cores (in this case PS2644-5 used The record of Nowaczyk et al. (2012) describes a field recovery
in NAPIS) is calibrated against a nearby volcanic record of the during the time of fully reversed polarity directions during the
same excursion (Skalamaelifell/Laschamp excursion in Ice- Laschamp excursion, with two paleointensity lows. This double
land). If future work confirms that the Skalamaelifell excursion paleointensity structure is also present in one of the Bermuda
is different from the Laschamp excursion, the results of Rise cores (MD95-2034), one of the cores of the NAPIS.
Leonhardt et al. (2009) obviously need to be rescrutinized. It is now firmly established that the strength of the geomag-
The age of the Laschamp directional excursion has become netic field is the most important factor controlling cosmogenic
sufficiently well established such that its recording can now be radionuclide production (Lal and Peters, 1967; Masarik and
used to refine age models in the Eirik Drift (North Atlantic) Beer, 1999): The smaller the field intensity, the larger the
(Evans et al., 2007), in the western equatorial Pacific (Blanchet production rate. As shielding of cosmic rays by the geomag-
et al., 2006), South Atlantic (Mazaud et al., 2002), and Deep netic field occurs at distances of several Earth radii, only
Cape Basin of the Southern Ocean (Kissel et al., 2008). changes in the dipole field (i.e., global changes) are relevant
Recently, Nowaczyk et al. (2012) had reported new sedi- to the regulation of cosmogenic radionuclide production.
mentary records from the Black Sea. Age models are con- Cosmogenic nuclide flux therefore provides an independent
strained by 16 accelerator mass spectrometer (AMS) 14C ages way of establishing the global character of the intensity mini-
back in time to 39 ka and by the identification of the Cam- mum associated with the Laschamp excursion. Initial work
panian Ignimbrite, independently dated at 39.28 0.11 ka. Of along these lines demonstrated that 10Be flux and geomagnetic
the six cores, four show a hiatus between 20 ka and 39 ka but field intensity anticorrelate in a North Atlantic core (Robinson
have recorded the Laschamp excursion. One of the other cores et al., 1995). Subsequently, Frank et al. (1997) correlated
does not extend beyond 39 ka and does not record the a reconstruction of paleofield intensity, obtained from 10Be
Laschamp excursion but all the 14C dates were obtained from flux records in marine sediments, to long-term trends in
this core. For the older ages (including the Laschamp excursion the SINT-200 paleointensity stack (Guyodo and Valet, 1996).
time interval), the stratigraphy was obtained by tuning differ- The 96–25 ka record of 36Cl flux from the GRIP ice core
ent paleoclimatic characteristics of the Black Sea cores to the (Baumgartner et al., 1998) agrees reasonably well with a pro-
Greenland (NorthGRIP) oxygen isotope record (Andersen et al., duction rate calculation from a paleointensity stack from the
2006; Svensson et al., 2006, 2008), with the assumption of no Somali Basin (Meynadier et al., 1992). More recently, Wagner
time lag between the two. Finally, one core spans the interval 0– et al. (2000) compared the NAPIS-75 (Laj et al., 2000) to the
60 ka, with variable sedimentation rate of 12 cm ky1 in the field intensity record estimated from an improved, higher-
vicinity of the Laschamp excursion. Interpretation of the results resolution 36Cl record, assuming that the variations in the
36
of the Black Sea cores is complicated by the magnetic mineral- Cl flux are entirely due to modulation by the geomagnetic
ogy of the sediments. Greigite, pyrrhotite, and other sulfides field. The 36Cl-derived profile has been smoothed out using
appear to coexist with magnetite. The sulfides carry diagenetic a 3000-year window in order to filter the solar modulation.
magnetizations that may postdate deposition, and alternating The coincidence of a prominent maximum in 36Cl flux with
field (AF) demagnetization is not efficient in separating magne- the paleointensity profile through the Laschamp excursion
tite and greigite/pyrrhotite magnetization components. On the (Figure 7) leaves little doubt concerning the geomagnetic
assumption that sulfide-bearing samples are usually character- nature of the increase in cosmogenic production at around
ized by saturated isothermal remanent magnetization (SIRM)/ 40 ka and provides strong evidence for the global nature of
kLF ratios 10 kA m1 (where kLF is the low field volume mag- the dipolar intensity minimum associated with the Laschamp
netic susceptibility) and those carrying magnetite by SIRM/ excursion.
kLF < 5 kA m1, Nowaczyk et al. (2012) used an arbitrary cutoff Ménabréaz et al. (2011, 2012) had reported authigenic
value of 10 kA m1 to exclude samples containing sulfides. 10
Be/9Be measurements from Portuguese Margin (core
While this can be considered as a reasonable first-order selec- MD04-2811) and equatorial Pacific (core MD05-2920) in
tion, it is clear that sulfides are ubiquitous, as is apparent in the order to reconstruct variations in cosmogenic 10Be production
thermomagnetic curves where magnetization changes in the rates and infer related geomagnetic dipole field modulation. In
230–280 C temperature range are typical of sulfides. The pres- the Portuguese Margin core, a low in RPI at 740–810 cm depth
ence of sulfides, with magnetizations that postdate deposition, is associated with the Laschamp excursion, although no excur-
may create spurious directional/paleointensity fluctuations, par- sional directions were observed. The main feature of the
10
ticularly during periods of fast changes of the geomagnetic field Be/9Be record is a significant enhancement, with an increase
strength, that is, during the Laschamp excursion itself. In con- of nearly a factor of 2 with respect to the long-term average.
trast with the simple clockwise pattern compiled by Laj et al. This interval overlaps the depth interval recording the lowest
(2006) augmented in Figure 6, the VGPs recovered from the RPI values, but there is a 50 cm offset between the two peaks.
2 (California) in search of the same excursion. Based on radio-

carbon ages available at the time, this section was thought to
span the 13–30 ka time interval. Denham and Cox (1971)
detected an excursion at Wilson Creek (Mono Lake), at a
1.8 horizon with an estimated age of 24 ka. This excursion has
since been named the Mono Lake excursion. Denham (1974)
interpreted the excursion as arising from eastward drift of the
nondipole field and showed that the pattern of observed paleo-
1.6 magnetic directions could be reproduced assuming an eccen-
Relative units
tric radial dipole source first increasing and then decreasing

during its eastward movement.
Liddicoat and Coe (1979) confirmed the existence of a
1.4 geomagnetic excursion at Mono Lake, documented the direc-
tional record in more detail, and extended the record further
down section. There was initially widespread skepticism about
the validity of this excursion. Verosub (1977) did not find
1.2
evidence for it in sediments from Clear Lake (California) and
Palmer et al. (1979) failed to find it in gravity cores from Lake
Tahoe (California). Turner et al. (1982) obtained a distinctive
pattern of ‘normal’ paleosecular variation, but no excursion,
from an 18 m sedimentary sequence at Bessette Creek (British
1
30 32 34 36 38 40 42 44 Columbia), which was radiocarbon dated to 31.2–19.5 ka.
Age BP (kyr) Verosub et al. (1980) did not find this excursion in sediments
exposed on the shores of Pyramid Lake (Nevada), which covers
Figure 7 Comparison of the 36Cl production rate (dashed line)
calculated from paleomagnetic field data (modified NAPIS-75) and the
the time interval 25–36 ka and is only about 230 km from
decay-corrected 36Cl flux (solid line) from the GRIP ice core. The shaded Mono Lake. The absence of the excursion at both Pyramid
area corresponds to the 2s uncertainties in the 36Cl production rate Lake and Clear Lake led Verosub et al. (1980) to conclude
due to uncertainties in the relative paleointensity data. Both records are that the estimated age of the Mono Lake excursion was in error.
low-pass-filtered and normalized to the present field values. The interpretation of the excursion at Mono Lake has pro-
Reproduced from Wagner G, Beer J, and Laj C (2000) Chlorine-36 gressively evolved. Liddicoat et al. (1982) discovered the excur-
evidence for the Mono lake event in the Summit GRIP ice core. Earth and sion in sediments from Lake Lahontan exposed in Carson Sink
Planetary Science Letters 181: 1–6. (Nevada), some 60 km east of Mono Lake. Negrini et al. (1984,
2000) discovered the excursion in sediments from Summer
About 10 cm of this lag can be attributed to the residence Lake, Oregon, 550 km north of Mono Lake. The excursion
time with respect to scavenging of Be. The authors note that was apparently detected in sediments recovered by hydraulic
it is not possible to assign the remaining 40 cm to magne- piston core from the DSDP Site 480 in the Gulf of California
tization lock-in depth, but do not provide any explanation. (Levi and Karlin, 1989). Liddicoat (1992) resampled at Carson
The peak in the cosmogenic record lies 40 cm (equivalent Sink and Pyramid Lake (Nevada), at the same locality studied
to 2 ky) beneath the susceptibility peak corresponding to by Verosub et al. (1980), and showed that, while the older half
H4 (Heinrich layer 4), which is dated at 39 ka. This suggests of the excursion is not recorded, the younger half can be
that the Laschamp excursion occurred at 41.2 1.6 ka. In the recognized. Liddicoat (1996) also reported the excursion
equatorial Pacific core, the 10Be-based paleointensity recon- from the Lahontan Basin, Nevada.
struction has striking similarities with the GLOPIS-75 record As in the case for the Laschamp excursion, evidence for the
of Laj et al. (2004), except for a peak at 31 ka present in global character of the Mono Lake excursion has come from
GLOPIS but not in the sedimentary 10Be record. The peak in the study of marine cores. Nowaczyk and Knies (2000)
the 10Be/9Be record corresponding to the Laschamp excursion obtained a record from cores in the Arctic Ocean, in which
occurs at 41 ka and its maximum duration was estimated they identified directional and RPI changes that they associated
at 1500 years (values of 10Be/9Be above mean 2s). The with the Laschamp and to the Mono Lake excursions (dated at
authors also estimated that the collapse of the virtual dipole 35 and 27–25 ka, respectively). As noted earlier, the excur-
moment prior to the Laschamp excursion occurred at a rate sional records from the Arctic Ocean may be influenced by
(1.5 1022 Am2 ky1) of about half the value of the decay maghemitization of titanomagnetite under oxic diagenetic
over the last millennium ( 2.7 1022 Am2 ky1) estimated conditions (see Channell and Xuan, 2009).
by Gallet et al. (2009). In one of the cores (PS2644-5) included in the NAPIS-75
(Laj et al., 2000), an intensity low associated with a brief
directional change has been recorded between two light
5.10.2.2 The Mono Lake Excursion
peaks in d18O correlated to Dansgaard–Oeschger (D–O) events
About 35 years ago, the age of the Laschamp excursion 7 and 6 (Kissel et al., 1999; Voelker et al., 1998). In the other
was thought to be 20 ka from studies in central France NAPIS-75 cores, although not in Irminger Basin core (SU90-
(Bonhommet and Zahringer, 1969). Denham and Cox 24), only the RPI low is observed. In the Irminger Basin at ODP
(1971) sampled the Wilson Creek Formation at Mono Lake Site 919, Channell (2006) observed transitional directions
around 32–34 ky that were associated with the Mono Lake provided overestimates of the age of deposition. In addition,
excursion, at 32–34 ka, based on a planktic oxygen isotope Benson et al. (2003) took advantage of the unique chemical
age model augmented by the occurrence in the core of Ash composition of Ash #15, which allows it to be distinguished
Layer II at 55 ka. from other tephra layers outside the Mono Basin, and dated
Although the 10Be sedimentary record does not provide this layer in the Pyramid Lake Basin. The advantage of this
evidence for a significant change corresponding to the Mono strategy is that the total organic carbon fraction in the Pyramid
Lake excursion (Ménabréaz et al., 2011, 2012), increased cos- Lake Basin is mostly composed of algae that obtain their car-
mogenic isotope flux is observed in Greenland ice-core records. bon from dissolved CO2 3
3 and HCO . Therefore, while the
For example, in the 36Cl record obtained from the GRIP ice radiocarbon ages from the Pyramid Lake Basin still incorporate
core (Wagner et al., 2000), in addition to the peak in produc- reservoir effects, they are not likely to be seriously contami-
tion associated with the Laschamp excursion described in the nated by modern carbon. Based on a previous estimate of the
preceding text, a second distinct peak is present between D–O reservoir effect of around 600 years in the last 3000 years,
events 6 and 7, at approximately 33 ka in the GISP2 age model Benson et al. (2003) concluded that the age of Ash #15 is
and 34.25 0.63 ky in the most recent GICC5 age model 28 620 300 years (not corrected for the reservoir age).
(Svensson et al., 2008) (Figure 7). This production peak can Benson et al. (2003) therefore concluded that the Mono Lake
therefore be attributed to the geomagnetic field intensity min- excursion recorded at Wilson Creek occurred in the
imum associated with the Mono Lake excursion, providing 31.5–33.3 ka interval, which is close in age to the paleointen-
evidence for the global character of this paleointensity sity minimum in NAPIS-75 (Laj et al., 2000) and to the peak in
36
minimum. Cl flux observed in the GRIP ice-core record (Wagner et al.,
Kent et al. (2002) had obtained new radiocarbon data from 2000). On the other hand, the correlation of the Mono Lake
lacustrine carbonates (ostracods and tufa nodules) from RPI stack (Zimmerman et al., 2006) with the GLOPIS-75
11 stratigraphic horizons from the lower part of the Wilson record supports the attribution of the excursion recorded at
Creek (Mono Lake) section. These authors argued that previ- Mono Lake to the Laschamp excursion. The age of the excur-
ous age estimates for the excursion recorded in the Mono Lake sion reported at the Mono Lake type locality remains contro-
section, based on a series of 27 published radiocarbon mea- versial and unfortunately, the most recent 40Ar/39Ar
surements on tufa or ostracods (Benson et al., 1990, 1998; investigation of the ash layers did not solve the problem
Lund et al., 1988), did not take into account radiocarbon because of the absence of juvenile, eruptive crystals (Cassata
reservoir effects, modern carbon contamination effects, or et al., 2010), although after review of the radiocarbon ages and
radiocarbon production variations and should be viewed as RPI correlations, these authors favored an age of 32–34 ka for
minimum age constraints. Kent et al. (2002) also obtained new the excursion at Mono Lake. The controversy has been
40
Ar/39Ar age estimates from sanidine crystals imbedded in the addressed by Cox et al. (2012) who used the (U–Th)/He sys-
different ash layers in the Wilson Creek section. These ages are tem on allanite (epidote) to date the Wilson Creek Ash #15,
at the younger limit of the method and are further complicated which lies 20 cm above the excursion, to 38.7 1.2 ka,
by residence time in the magma chamber and other sources of thereby implying that the excursion recorded in this section
inherited (old) argon. As a consequence, the ash layers are is, indeed, the Laschamp excursion. Several ash layers at
usually characterized by a wide range of sanidine ages. In the Wilson Creek have also been dated by Vazquez and
middle of the Mono Lake excursion, 34 individual sanidine Lidzbarski (2012) using U–Th methods applied to allanite
samples from Ash #15 yielded ages between 49 and 108 ka, a and zircon rims. Their results yield a zircon isochron age of
range that far exceeds analytic precision. In such cases, Chen 40.8 1.9 ka for Ash #15, also implying that the excursion
et al. (1996) suggested that the youngest 40Ar/39Ar ages provide recorded at Wilson Creek is, in fact, the Laschamp excursion.
realistic estimates of eruption age, as they are generally closer Although the excursion recorded at Mono Lake may in fact
to accompanying radiocarbon ages. Using similar logic, Kent be the Laschamp excursion, the global manifestation of an
et al. (2002) considered 49 ka as the maximum depositional excursion at 32 ka (labeled here the Mono Lake excursion)
age for Ash #15. Using radiocarbon dates as minima and has come from volcanic formations, surprisingly given its short
40
Ar/39Ar dates as maxima, Kent et al. (2002) derived a new duration and the sporadic nature of volcanic eruptions. In the
age model for the Wilson Creek section and estimated the age Auckland volcanic field in New Zealand, outcrops yielded
of the lower part of the Wilson Creek section as > 46 ka. On similar anomalous paleomagnetic directions, indicating that
this basis, Kent et al. (2002) proposed that the excursion they may have recorded the same ‘snapshot’ of the geomag-
recorded at Mono Lake is, in reality, the Laschamp excursion. netic field (Cassidy and Hill, 2009; Mochizuki et al., 2004,
Indirect evidence for this interpretation is provided by the 2006, 2007; Shibuya et al., 1992). Cassata et al. (2008) radio-
absence of a second excursion that might otherwise be identi- metrically dated these outcrops, obtaining a pooled age of
fied with the Laschamp excursion in the lower part of the 31.6 1.8 ka. The anomalous directions were therefore attrib-
section. uted to the Mono Lake excursion. In the Canary Islands, a
Benson et al. (2003) argued against the conclusions of Kent study by Kissel et al. (2011) took advantage of a previous
et al. (2002). First, they noted that Kent et al. (2002) assumed a survey on the island of Tenerife by Carracedo et al. (2007).
reservoir age of 1000 years, while previous work by Benson Of the three different studied lava flows, two yielded very low
et al. (1990) had concluded that the carbonates deposited in paleointensities and the third one, in addition, had a very
the Mono Basin during the past 650 calendar years exhibit a anomalous paleomagnetic direction. Age estimates that com-
reservoir effect ranging from 1100 to 5300 years. The radiocar- bined 40Ar/39Ar and K–Ar data, from two of the flows, yielded
bon dates obtained by Kent et al. (2002) may therefore have values of 32.0 1.3 and 32.2 1.2 ka, identifying the Mono
Lake excursion. The discrepancy between these radiometric Blake excursion had been documented, but affecting only a
dates and the age implied for the Mono Lake excursion from small portion of the globe.
the 36Cl record in the Greenland ice core (34.5 0.65 ka) was Subsequently, Creer et al. (1980) reported evidence for the
attributed to inherent difficulties of radiometric dating of very Blake Event in clays from Gioia Tauro in southern Italy, where
young volcanic formations with low Ar contents (Kissel et al., the double structure of the episode was also apparent. Verosub
2011). Finally, the paleomagnetic results obtained from the (1982) pointed out that the record of the Blake excursion is
two long volcanic cores, SOH-1 and SOH-4, in Hawaii (Laj characterized by two reverse intervals separated by a short
et al., 2002b; Teanby et al., 2002) have been combined into a normal interval in records of the Blake excursion from Italy,
single composite record (Kissel and Laj, 2004; Laj et al., 2011). Japan (Lake Biwa record), and the North Atlantic Ocean,
A large group of 11 successive flows display very anomalous thereby providing evidence for the global character of this
inclination values (no declinations can be obtained from these complex directional characteristic.
cores), associated with very low paleointensities. These cores Tucholka et al. (1987) reported a paleomagnetic and oxy-
could not be radiometrically dated, given the very low radio- gen isotope study of five cores from the Eastern and central
metric argon contents. Their age model, obtained by successive Mediterranean that could be intercorrelated using sapropel
approximations using volcanic data and correlation to the occurrences. The quality of the results from cores from the
NAPIS-75 curve, associated these anomalous magnetization central Mediterranean (sampled several years after coring)
directions with the Mono Lake excursion. was not high, but the Eastern Mediterranean cores yielded
In summary, all the new volcanic results, combined with unambiguous results. In all three Eastern Mediterranean
previous cosmogenic and sedimentary results, imply global man- cores, a change to negative inclinations was observed, accom-
ifestation of the Mono Lake excursion as a geomagnetic phenom- panied by reverse declinations in two cores and a smaller
enon younger than the Laschamp excursion by 8 ky. The declination change in the third. The excursion was recorded
excursion recorded at Wilson Creek close to Mono Lake (CA), in sediment directly overlying sapropel S5. In two of the cores,
the type locality for the Mono Lake excursion, may not be synch- the record is probably incomplete, as no underlying normal
ronous with the supposed ‘Mono Lake’ excursion at 32 ka polarity directions were found above the sapropel layer. One
recorded elsewhere. The excursion recorded at Wilson Creek record, on the other hand, shows the bipartite structure that is
may, in fact, be a record of the Laschamp excursion based on characteristic of the Blake Event in the North Atlantic, Italy,
the recent U–Th and (U–Th)/He ages that yield ages close to and Japan. Detailed oxygen isotope analyses, made on four of
40 ka for Ash #15, which lies 20 cm above the excursion at this the cores, established the stratigraphic position of the Blake
locality. Event between MIS 5e and MIS 5d. Estimates of the duration of
the event, based on constant sediment accumulation rate
between tie points, varied in the 2.8–8.6 ky range, with a
5.10.2.3 The Blake Excursion
mean of 5.3 2.7 ky.
Smith and Foster (1969) first defined the Blake Event from a Tric et al. (1991) reported a detailed record of the Blake
paleomagnetic study of four deep-sea cores recovered from the Event from two Mediterranean cores, one from the Tyrrhenian
Blake Outer Ridge. Their study established the existence of a Sea and the other being one of the cores already studied by
short interval of reverse polarity in the later part of the Brunhes Tucholka et al. (1987) that was resampled at higher resolution.
Chron. The paleomagnetic declinations could not be reliably In the Tyrrhenian Sea core, only five excursional (intermediate)
determined in these piston cores, but the event was clearly directions were recorded possibly due to a lower sediment
revealed by anomalous inclinations, reaching 10 to 70 , accumulation rate synchronous with the Blake excursion. On
compared to an expected inclination at this location of 40 . the other hand, the Eastern Mediterranean core yielded more
On the basis of the position of the episode within the biostra- detailed results, with 70 intermediate and reverse polarity
tigraphy of Ericson et al. (1961), and the age estimate by directions, making this record of the Blake Event the most
Broecker et al. (1968), the boundaries of the Blake Event detailed obtained so far and the first one for which numerous
were placed at 108 and 114 ka. Subsequently, the oxygen intermediate VGPs were obtained (Figure 8). Although no
isotope data of Broecker and Van Donk (1970) indicated that record of field intensity was obtained, the directional varia-
the Blake excursion occurred within MIS 5. tions provide insight into the complex dynamics of the Blake
Denham and Cox (1971) and Denham (1976) provided Event. First, a sudden jump of inclination from positive to
corroborating evidence for the Blake Event from the Greater negative values was observed, accompanied by intermediate
Antilles Outer Ridge and the Blake–Bahama Outer Ridge. declinations and ending with a recovery of normal polarity.
These studies also provided the first evidence that the Blake The VGPs in this first phase, which lasted about 1100 years
Event might consist of two short intervals of almost reverse according to the age model, lie over North and South America
polarity separated by a short interval of almost normal polar- followed by an abrupt jump to Australia. The excursion in this
ity. The evidence for the Blake Event in these early studies came core is characterized by two intervals of reverse polarity direc-
from a limited geographic area and the reality of the Blake tions with an intervening normal polarity period. The upper
excursion as a global phenomenon remained questionable. bound of duration of the excursion, based on the isotope
Denham (1976) considered the possibility of a ‘local’ reversal. stratigraphy and tephrochronology, was estimated to be 4 ky.
He designed a model involving a small dipole source antipar- Two intervals of reverse inclinations in the 115–122 ka
allel to the main dipole and situated near the core–mantle interval were observed by Thouveny et al. (2004) in core
boundary, directly beneath the ‘Blake region.’ This model MD95-2042 from the Portuguese Margin. Oxygen isotope stra-
accounted for a reversal in the vicinity of the zone where the tigraphy in this core placed this record of the Blake excursion in
Figure 8 Transitional magnetization directions represented as VGP paths for the Blake excursion recorded in core MD84-627 from the Eastern
Mediterranean. Data from Tric E, Laj C, Valet J-P, Tucholka P, Paterne M, and Guichard F (1991) The Blake geomagnetic event: Transition geometry,
dynamical characteristics and geomagnetic significance. Earth and Planetary Science Letters 102: 1–13.
MIS 5c according to Table 2 of Thouveny et al. (2004). Zhu total duration of 5.5 ky, which is consistent with the estimate
et al. (1994) obtained a record of the Blake excursion from a of Tucholka et al. (1987) from the Mediterranean Sea.
loess section at Xining (Western China). The sequence of direc- Recently, Bourne et al. (2012) had reported a record of the
tional changes is contained in paleosol S1, which corresponds Blake excursion in sediments from ODP Site 1062 on the
to MIS 5 (An et al., 1991; Li et al., 1992). Zhu et al. (1994) Blake–Bahama Outer Ridge. These authors have determined
estimated the top and base of the studied section to be at the duration of the excursion by using 230Th excess measure-
72 and 130 ka and estimated ages of 117 and 111 ka for the ments to assess the variations in the sedimentation rate during
onset and demise of the Blake excursion, respectively, yielding the excursion. They propose that the excursion occurred
a duration estimate of 5.6 ky. The directional record of the between 129 and 122 ka, for a total duration of 6.5 ky. Only
excursion is characterized by stop-and-go behavior, reminis- a single negative polarity interval is observed, unlike the other
cent of the (Miocene) volcanic record at Steens Mountain records of the Blake excursion in which two (e.g., Fang et al.,
(Mankinen et al., 1985). The Xining record contains three 1997; Tric et al., 1991) or three (e.g., Zhu et al., 1994) reverse
clearly defined periods of reverse polarity separated by two polarity intervals are observed.
short intervals of normal polarity. The structure of this excur- The Blake excursion from ODP Site 1063 on the Bermuda
sional record is more complex than had been previously Rise only indicates a single reverse polarity interval at 115.5 ka
reported for this excursion. The documented nonuniform tim- coincident with the MIS 5e/5d boundary (Channell et al.,
ing of acquisition of magnetization for Chinese loess/paleosol 2012). The age derived from ODP Site 1062 (Bourne et al.,
sequences (e.g., Heslop et al., 2000; Spassov et al., 2003) may 2012) would place the Blake excursion firmly within MIS 5e. In
contribute to this apparent complexity and to the fact that addition, Bourne et al. (2012) gave a duration for the Blake
searches for the Blake Event in Chinese loess have not always excursion (6.5 ky), consistent with the estimate of Tric et al.
been successful (Parés et al., 2004). Although the Blake excur- (1991) but contrasting with the estimate of < 1 ky at Site 1063
sion has also been documented in the Junzhoutai loess of the (Channell et al., 2012).
excursion section at Lanzhou (Fang et al., 1997), the structure Osete et al. (2012) had studied the Blake excursion in a
is not the same as at Xining. At Lanzhou, it comprises two short dated speleothem collected in a cave in northern Spain. The
reverse polarity intervals separated by a short normal polarity age of the excursion, determined by uranium-series dating, was
interval, which is similar to some records of the excursion in estimated to be between 116.5 and 112.0 ka, for a total dura-
marine sediments. Based on thermoluminescence and astro- tion of the excursion of 4.5 ky. The speleothem records
nomically tuned cycles of magnetic susceptibility, the age of indicate one clear interval of negative (Southern Hemisphere)
the Blake excursion was bracketed between paleosol S1-c virtual geomagnetic pole (VGP) latitudes with an interval of
(equivalent to MIS 5e) and loess 2-2 (MIS 5d) (Fang et al., equatorial VGPs above, which may be the manifestation of a
1997). This corresponds to an age range of 120–115 ka and a second reverse polarity interval confirming the double
structure of the Blake excursion. The age of the Blake excursion Weeks et al. (1995a,b) reported a paleomagnetic study of
from the speleothem record is consistent with the 115 ka four piston cores from the North Atlantic Ocean. The age
(MIS 5d/5e boundary) age at ODP Site 1063 (Channell et al., model was based on oxygen isotope stratigraphy from planktic
2012); however, the estimated duration from the speleothem foraminifera. The paleomagnetic record indicates that a large
record appears more consistent with the estimate of Bourne swing in inclination to negative values and a marked decrease
et al. (2012) from ODP Site 1062. in RPI occurred around 180–200 ka. Lehman et al. (1996)
Observations from the highest-resolution marine cores, reported a paleomagnetic study of three marine cores in the
from loess sections, and from speleothems therefore indicate Azores area in the North Atlantic Ocean. For one of the cores, a
that the Blake excursion is a global geomagnetic feature, most detailed d18O record was obtained, and the two others were
probably with a characteristic structure comprising two short correlated to it using sediment grayscale reflectance data. The
periods (not always well resolved) of reverse polarity separated paleomagnetic record contains fluctuations in declination and
by a short period of almost normal polarity, with a total inclination and a marked drop in paleointensity at 190 ka,
duration that may approach 5 ky. but no clear shift in the inclination. Roberts et al. (1997)
documented an excursion in the North Pacific Ocean at ODP
Site 884 (Figure 4). Preservation of foraminifera was poor at
5.10.2.4 The Iceland Basin Excursion
this site, so the age model was obtained by transferring the
Over the last 10–15 years, many studies have provided evi- oxygen isotope stratigraphy from ODP Site 883 (Keigwin,
dence for geomagnetic excursions in the 180–220 ka interval. 1995), by correlation of magnetic susceptibility records. How-
Labels ‘Jamaica,’ ‘Pringle Falls,’ and ‘Biwa I’ have been used ever, the d18O record was not well defined at Site 883 in the
somewhat arbitrarily by different authors in referring to excur- vicinity of the excursion (MIS 6/7 boundary), so the age of the
sions with estimated ages in the 180–220 ka interval. As the excursion is relatively poorly constrained.
fidelity and age control of available records have improved, Channell et al. (1997) reported d18O and paleomagnetic
evidence has accumulated for two excursions in the (directions and paleointensity) data from rapidly deposited
180–220 ka interval. The younger one is labeled the Iceland (10–15 cm ky1) sediments recovered at ODP Site 983 on the
Basin excursion (185–190 ka), and the older one takes its Gardar Drift in the Iceland Basin (Figure 4). Paleomagnetic
name from Pringle Falls (211 ka). analyses revealed, in the 186–189 ka interval, a short-lived
As mentioned in the preceding text, paleomagnetic analyses excursion in which the VGPs move to high southern latitudes.
of cores from Lake Biwa provided early evidence of several The age of this excursion is constrained close to the MIS 6/7
excursions during the late Brunhes Chron (Kawai et al., 1972; boundary (188 ka) and appears to be distinct from the Pringle
Yaskawa et al., 1973). The authors suggested that the youngest Falls excursion (see succeeding text). A paleointensity mini-
could be the Blake excursion and two other episodes of reverse mum with an onset age of 218 ka at ODP Site 983 may be
inclination were tentatively correlated to intervals of anoma- coeval with the Pringle Falls excursion, although no directional
lous inclination in the record of Wollin et al. (1971) from the excursion is recognized in association with this paleointensity
North Pacific Ocean. From this correlation, an age between minimum at this site. Channell et al. (1997) named this excur-
176 and 186 ka was assigned to an event that they called the sion the Iceland Basin Event (now Iceland Basin excursion),
Biwa I event. These early records have remained problematic and this labeling is adopted here when referring to the geo-
due to inadequate paleomagnetic analysis by modern magnetic excursion at the MIS 6/7 boundary. The Iceland Basin
standards and poor age control. excursion appears to be coeval with the geomagnetic excur-
In the marine realm, records corresponding to this time sions recognized in the central North Atlantic Ocean by
interval obtained from Norwegian–Greenland Sea revealed Weeks et al. (1995a,b), as well as in the western equatorial
several intervals of negative inclination (Bleil and Gard, Pacific Ocean (Yamazaki and Yoka, 1994).
1989). These intervals have also been observed in a series of The record of the Iceland Basin excursion from ODP Sites
piston cores from further north in Fram Strait (Nowaczyk and 984 (Channell, 1999) and 980 (Channell and Raymo, 2003),
Baumann, 1992) and on the Yarmak Plateau (Nowaczyk et al., where mean sedimentation rates in the Brunhes Chron
1994). These early studies, however, lacked a well-defined exceeded 11 cm ky1, is similar to that obtained at Site 983.
d18O record, leading to uncertainties in assessing the precise At all three sites, inclinations are negative in the 180–195 ka
age of the intervals of negative inclination. Nowaczyk and interval, and the estimated duration of the excursion is 3 ky.
Antonow (1997) produced d18O and magnetostratigraphic The apparent discrepancy in age at Sites 980, 983, and 984 can
records from the Greenland Basin that indicate the presence be attributed to uncertainties in the isotopic age model, partic-
of negative inclinations with ages corresponding to the Mono ularly at Site 984 where the age model is not well defined. The
Lake (27–28 ka) and Laschamp (40 ka) excursions, with an VGP paths for the Iceland Basin excursion from Sites 983 and
additional excursion at around 188 ka. Nowaczyk and 984 are similar (Figure 9) and feature a large-scale counter-
Antonow (1997) referred to this latter excursion as the ‘Biwa clockwise loop that is located over Europe and Africa in the first
I/Jamaica’ excursion, thereby associating it with the excursions N/S part of the loop and returns to northern latitudes over the
documented in early papers by Wollin et al. (1971), Ryan West Pacific Ocean. The directional changes coincide with a
(1972), and Kawai et al. (1972). It has since been suggested prominent paleointensity minimum that is a characteristic
that these high-latitude cores from the Arctic and Norwegian– feature of sedimentary RPI records. Oda et al. (2002) obtained
Greenland Seas are affected by oxidation diagenesis that can a paleomagnetic record, including a paleointensity record,
affect the magnetic properties and yield excursional artifacts from core Ver 98-1 from Academician Ridge, Lake Baikal. The
(Xuan et al., 2012). age model was derived using sediment bulk density as a proxy
ODP 984 Lake Baikal

ODP 983
MD99-2242
MD99-2247
ODP 1146
ODP 1145
Figure 9 Transitional magnetization directions represented as VGP paths for the Iceland Basin excursion. The large arrow illustrates the sense of
looping of the VGPs, which is consistently counterclockwise.
for biogenic silica content. Interglacial intervals were then cor- isotope age model yielded an age of 189–191 ka and a dura-
related to maxima in silica (density). An age–depth plot was tion of 2 ky for this excursion.
then produced based on SPECMAP ages (Martinson et al., 1987) Laj et al. (2006) compiled seven records (directions and
for isotope substages back to MIS 7.3. The average sedimenta- RPIs) of the Iceland Basin excursion. Among which, core
tion rate was estimated to be 4.5 cm ky1. A directional MD99-2242 was from the Eirik Drift (south of Greenland),
excursion in the paleomagnetic record between 6.70 and and core MD99-2247 was from the western flank of the
6.96 m below the lake floor was estimated to have an age of Reykjanes Ridge (Figure 4). Two other records were obtained
177–183 ka. The excursion coincides with a marked minimum from the South China Sea, at ODP Sites 1145 and 1146 (ODP
in the RPI proxy that can be correlated with the record from Leg 184). Average sedimentation rates ranged from 7 to
ODP Site 983. Oda et al. (2002) also redated an excursion 10 cm ky1 for cores in the Atlantic Ocean and 15 cm ky1
from another core from Academician Ridge previously inferred for cores from the South China Sea. The three other cores were
to record the Blake excursion (Sakai et al., 1997). The new age is ODP Sites 983 and 984 (Channell, 1999; Channell et al.,
223 ka, which provides evidence for an excursion (possibly 1987) and Lake Baikal (Oda et al., 2002). The VGP paths are
Pringle Falls) preceding the Iceland Basin excursion in sedi- fairly uniform for each of the dispersed site locations
ments from Lake Baikal. (Figure 9). During the first part of the excursion, the VGPs
The general pattern of the VGP paths for the Lake Baikal move southward over Africa along a narrow band of longitudes
record is similar, albeit much more scattered than those before crossing the equator and reaching high southern lati-
observed at ODP Sites 983 and 984. After an initial swing tudes (fully reverse polarity directions), with a return path in
over the western Atlantic Ocean (which does not appear in East Asia.
the ODP records), the VGPs move southward over Africa reach- Based on this compilation, Lanci et al. (2008) had pre-
ing high southern latitudes. The S ! N return path is defined sented models of the excursional geomagnetic field at the
by only three points, all of which are in the Southern Hemi- Earth’s surface using two different approaches. First, a spherical
sphere, which consequently makes the return path ill-defined. harmonic analysis was performed, after synchronization of
The overall counterclockwise loop is, however, similar to that the records using their paleointensity profiles. Second, the
observed from the two North Atlantic ODP cores. Bayesian procedure of Leonhardt and Fabian (2007) was
Stoner et al. (2003) recorded the Iceland Basin excursion at used, calibrated using the single available volcanic record of
ODP Site 1089 in the sub-Antarctic South Atlantic. Mean sed- the Iceland Basin excursion. The two approaches show some
imentation rates are in the 15–20 cm ky1 range for a record similarities and some differences, partly arising from the lim-
that extends back to 580 ka. The RPI record from this site can ited geographic distribution of sites. Nevertheless, the general
be correlated to those from the North Atlantic and Lake Baikal picture is that of a dipole field undergoing a strong reduction in
that record the same excursion; the excursion occurs within the intensity but remaining higher than that of the nondipole field
same prominent paleointensity low. At this site, the oxygen during most of the excursion. The nondipole field does not
significantly decrease during the excursion. Figure 10 (adapted record and those compiled by Laj et al. (2006): either unrecog-
from Lanci et al., 2008) gives the results obtained using the two nized sediment deformation has affected parts of this record or
approaches. excursional records are more complex when fully recorded at
In the compilation of Iceland Basin excursion records of Laj some (high-latitude) locations.
et al. (2006), the record obtained by Roberts et al. (1997) from
the North Pacific was not considered due to poorly constrained
5.10.2.5 The Pringle Falls Excursion
age model and resemblance of VGP paths with the one of
Pringle Falls from Herrero-Bervera et al. (1994). More recently, A detailed record of a geomagnetic excursion obtained from a
Roberts (2008) reconsidered this excursion arguing on the sedimentary lacustrine sequence near Pringle Falls (Oregon)
basis of paleointensity correlations that the recorded excursion was initially thought to represent the Blake excursion (Herrero-
at ODP Sites 883 and 884 is the Iceland Basin excursion. Bervera et al., 1989). Subsequently, however, the age of this
Channell (2006) recorded the Iceland Basin excursion at episode was revised on the basis of 40Ar/39Ar dating of plagio-
ODP Site 919 in the Irminger Basin (off east Greenland). clase feldspars from an ash layer (Ash D) located close to the
According to the oxygen isotope age model, the excursion base of the excursion (Herrero-Bervera et al., 1994). The iso-
occurs in the 180–188 ka interval. Its manifestation at this chron age and the plateau age obtained from step heating were
site is considerably more complex than other records of the not significantly different, but the isochron age had a larger
same excursion. Although the sediments have all the attributes relative uncertainty, resulting from low radiogenic yield. The
for high-fidelity recording, and the excursion is recorded by authors, therefore, took the plateau age of 218 10 ka to be the
both discrete samples and u-channel samples, the excursion is best estimate of the age of the Ash D layer, which corresponds
characterized by a first VGP loop to high southern latitudes to the onset of the excursion. Herrero-Bervera et al. (1994)
characterized by an outward and return path over Africa, fol- associated this excursion with the Jamaica excursion of
lowed by a second VGP loop with an outward path over East Wollin et al. (1971) and Ryan (1972). By modern standards,
Asia and a return path over Africa. These two VGP loops are the Jamaica excursion was not adequately resolved either
followed by a complex group of VGPs at low latitudes over paleomagnetically or stratigraphically in these early publica-
western Africa. In view of the uniformity in magnetic hysteresis tions. Following the practice of naming geomagnetic excur-
properties through the excursion, the complexity of the VGP sions after the location where they have been unambiguously
path cannot be readily attributed to lithologic variability. There recorded, we advocate use of Pringle Falls as the name for this
are two feasible explanations for the discrepancy between this excursion.
AgeODP1146 (kyr) AgeODP1146 (kyr)

180 184 188 192 196 200 180 184 188 192 196 200
1.6 1600
1.4
Energy (arbitrary units)
1.2 1200
Dipole
Energy (µT2)
1
0.8 800 Dipole
0.6
0.4 400
Nondipole
0.2
Nondipole
(a) (b)
0 0
Gauss coefficients (mT)
-150
Gauss coefficients
-1.2 g1
(arbitrary units)
0
g1 -100
-0.8 0
-50
-0.4 g1
g1 1
1 0
0.0
h1 h1
1 50 1
0.4
180 184 188 192 196 200 180 184 188 192 196 200
(c) AgeODP1146 (kyr) (d) AgeODP1146 (kyr)
Figure 10 Top part: energy of dipolar and nondipole field components during the Iceland Basin excursion from (a) SHA and (b) IMIBEe approach.
The results from SHA model are reported in arbitrary units and normalized to have the preexcursional dipole moment Bo ¼ 1. Filled bands represent 95%
confidence limits are computed with jackknife statistics. Bottom part: first-order (dipolar) Gauss coefficient for the IBE as computed by (c) SHA and
(d) IMIBEe models. Gray bands in SHA model represent the 95% confidence levels; no confidence levels are available for IMIBEe model. Reproduced
from Lanci L, Kissel C, Leonardt R, and Laj C (2008) Morphology of the Iceland Basin Excursion from a spherical harmonics analysis and an
iterative Bayesian inversion procedure of sedimentary records. Earth and Planetary Science Letters 169: 131–139.
Herrero-Bervera et al. (1994) used the chemical and petro- transitional field was dominated by a dipolar field component
graphic characteristics, stratigraphic position, and available age (McWilliams, 2001).
data to correlate the tephra from Pringle Falls to tephra layers More recently, Singer et al. (2008a) reported 16 40Ar/39Ar
present in other sequences in western North America, includ- laser incremental heating ages from plagioclase crystals
ing Summer Lake, Mono Lake, and Long Valley. Coeval paleo- derived from Ash D (deposited during the onset of the excur-
magnetic records of excursions from each of these localities sion) at Pringle Falls that define an 40Ar/39Ar isochron of
had been previously published (Herrero-Bervera and Helsley, 211 13 ka. In the same paper, 40Ar/39Ar analyses in the
1993; Liddicoat, 1990; Liddicoat and Bailey, 1989; Negrini Albuquerque volcanics yielded an isochron age of 218 14 ka
et al., 1988, 1994), but only after this correlation of tephras (Singer et al., 2008a) for well-grouped excursional site mean
were they recognized as records of the same excursion. The magnetization directions (mean declination ¼ 101 , mean
paleomagnetic records from two sites at Pringle Falls (the inclination ¼ 36 ) obtained from 65 sites collected from the
second one being at a distance of 1.5 km from the original six major flows (Geissman et al., 1990). These two statistically
one) and from Long Valley, some 700 km away, are strikingly indistinguishable ages appear to indicate that the excursional
similar (Figure 11). The VGP paths lie over the Americas in magnetization directions in the Albuquerque volcanic field are
the first N ! S part of the excursion and then move to the recording the same excursion as recorded at Pringle Falls in
northwest Pacific for the S/N return path (Herrero-Bervera Oregon.
et al., 1994). At ODP Site 919, off east Greenland, Channell (2006)
Subsequently, a series of transitional paleomagnetic direc- recorded an excursion that predates the Iceland Basin excur-
tions were obtained from the Mamaku ignimbrite in the North sion and, according to the stable isotope age model, correlates
Island of New Zealand (Tanaka et al., 1996). Three new to the 206–224 ka age interval. The VGP path has a complex
40
Ar/39Ar ages from plagioclase from the Mamaku ignimbrite series of loops to equatorial latitudes over the Pacific and South
(Houghton et al., 1995) yielded a weighted mean of America, followed by a clockwise loop to high southern lati-
223 3 ka, which is statistically indistinguishable from the tudes. This last loop is in the same sense (clockwise) and
age of Ash D at Pringle Falls according to Herrero-Bervera follows a similar path as those recorded at Pringle Falls, Long
et al. (1994). McWilliams (2001) plotted 29 VGPs for the Valley, and New Zealand for supposedly the same excursion.
Mamaku ignimbrite together with those from the two sites at On the other hand, there is no trace in the Pringle Falls and
Pringle Falls and at Long Valley. The agreement in time and Long Valley records of the initial complex structure seen at
space is remarkable (Figure 11). The VGPs from volcanic rocks ODP Site 919, despite the high sediment accumulation rates
in New Zealand record only a fraction of the total excursion at these two localities (the excursion is recorded over strati-
and are not nearly as complete as the records from Pringle Falls graphic intervals of 100 cm at Long Valley and over 700 cm at
or Long Valley, as expected due to the episodic nature of Pringle Falls). A recent sedimentary record of an excursion
volcanic eruption. Nonetheless, this remarkable agreement supposed to represent the Pringle Falls excursion, at ODP Site
suggests that the excursion was manifest globally and that the 1063, yields an age of 238 ka according to an age model based
Pringle
Falls
Long
Valley
Mamaku
Figure 11 Transitional magnetization directions represented as VGP paths for the Pringle Falls excursion. Squares are individual VGPs for the event
recorded at Pringle Falls (two records), Long Valley, and New Zealand. Circles are sampling sites, color-coded to match respective points on the
VGP paths. Arrows indicate the progression of the VGP paths from older to younger. Reproduced from McWilliams M (2001) Global correlation of the
223 ka Pringle falls event. International Geological Reviews 43: 191–195.
on oxygen isotopes and RPI (Channell et al., 2012). In the complex redox conditions in the multicolored sediments of
PISO paleointensity stack (Channell et al., 2009), minima in core KC-01B (see Langereis et al., 1997), and the fact that the
RPI centered on 215 and 238 ka, separated by an RPI peak, CR excursions are defined by single samples, indicate that they
open the possibility that more than one excursion is present in require independent corroboration prior to their incorporation
the 210–240 ka interval. into the library of geomagnetic excursions.
Biswas et al. (1999) reported a magnetostratigraphy of
a 1700 m core from the Osaka Bay, southwestern Japan,
5.10.2.6 Excursions in the Early Brunhes Chron
spanning the last 3.2 My. Although the mean sedimentation
Although fewer studies have been conducted for the Brunhes rate in the Brunhes Chron (50 cm ky1) is far higher in these
Chron prior to 250 ka, there is evidence for geomagnetic excur- coastal sediments than in deep marine sequences, only one
sions in the early part of the Brunhes Chron. The use of excursion is recorded in the Brunhes Chron. This Brunhes
different names, combined with imprecise age control, has Chron excursion is recorded in lacustrine silty clays that are
led to confusion in labeling and correlation. immediately overlain by marine clays and is characterized
Ryan (1972) defined the so-called Emperor Event as a short by intensity fluctuations with steep inclinations (up to þ72
reverse polarity zone at the base of Caribbean core V12-22. and 82 ). It occurred during MIS 17, probably during sub-
Based on the work of Ericson et al. (1961) and Broecker and stages 17.4–17.3, based on a sea-level interpretation of the
Van Donk (1970), the age of this interval was estimated to be marine/terrestrial sequence. This would correspond to an
about 460–480 ka. The excursion was, however, defined by excursion age of 690 ka. The same age is obtained from the
only one sample that was only demagnetized in peak alternat- assumption of uniform sedimentation rate between the
ing fields of 5–15 mT. Nevertheless, some support for the Matuyama–Brunhes (M/B) boundary and a characteristic tuff
excursion was found in the study of axial MMA records from layer (Aira Tuff dated at 24.5 ka). Biswas et al. (1999) named
the Galapagos Spreading Center that suggested a brief reverse this excursion the ‘Stage 17 Event.’ They correlated it to the
polarity event at 490 50 ka (Wilson and Hey, 1981). This age d event reported by Creer et al. (1980) from Gioia Tauro. The
depends on the assumption of linear spreading rate of the age of the d event can be revised to 680 ka, using the modern
Galapagos Ridge and is therefore not robust. Interestingly, value of 780 ka for the M/B reversal. Interestingly, the Stage
the apparent occurrence of the Emperor Event on the 17 excursion is manifest in Osaka Bay as a double excursion
Galapagos Ridge MMA data provides the basis for the one/ with steep negative inclinations and an intervening interval
only ‘cryptochron’ within the Brunhes Chron in the timescale with positive inclinations. An excursion within MIS 17 has
of Cande and Kent (1992a). also been recognized at ODP Site 980 collected from the Feni
As mentioned in the preceding text, initial dating of a Drift, North Atlantic (Channell and Raymo, 2003). The oxygen
reverse polarity flow in Idaho at 490 50 ka (Champion isotope age model yields an excursional age of 687–696 ka
et al., 1981) provided land-based support for the existence of within MIS 17 and an excursional duration of 9 ky. Paleo-
the Emperor Event. Later, however, Champion et al. (1988) magnetic inclinations reach low negative values accompanied
revised the age of these reverse polarity lavas to 565 10 ka, by a 180 swing in declination.
which therefore corresponds to a different (older) reverse epi- Lund et al. (2001a,b) documented many apparent excur-
sode, which they named the Big Lost excursion. In their sem- sions in the Brunhes Chron in marine cores taken from sedi-
inal paper, Champion et al. (1988) reviewed evidence for ment drifts in the Western North Atlantic Ocean (Blake Outer
excursions from both volcanic and sedimentary sequences Ridge, Bahama Outer Ridge, and Bermuda Rise) during ODP
and concluded that at least eight excursions existed during Leg 172. From initial shipboard measurements, made on half-
the Brunhes Chron. In addition to the newly discovered cores with a low-resolution pass-through cryogenic magne-
Big Lost excursion and the well-known excursions discussed tometer, 14 geomagnetic episodes were identified as ‘plausible’
in the preceding text, they presented evidence for (1) the d Brunhes Chron magnetic field excursions (Lund et al., 1998).
excursion, first documented by Creer et al. (1980) at Gioia Laboratory measurements conducted on u-channels and dis-
Tauro in Southern Italy, dated at 640 ka; (2) the Emperor crete samples documented narrow intervals of relatively low or
excursion (470 ka) (Ryan, 1972); (3) the Biwa III excursion high inclination or westerly and easterly declinations, which
(390 ka); and the (4) Levantine excursion (or Biwa II) could be traced among the records from independent holes
(290 ka) (Kawai, 1984; Kawai et al., 1972; Ryan, 1972; separated by distances <1 km and were correlated using varia-
Yaskawa et al., 1973). tions in magnetic susceptibility. Shore-based measurements
Langereis et al. (1997) documented four short excursions led Lund et al. (2001a,b) to confirm 12 of the 14 originally
in piston core KC-01B collected from the Calabrian Ridge in defined excursions, one of which appears to be synchronous
the Ionian Sea (Mediterranean Sea). Astronomical calibration with the Stage 17 excursion. The large number of geomagnetic
of the age sapropels, recognized on the basis of rock-magnetic excursions in the Brunhes Chron at ODP Sites 1060–1063
and geochemical properties, allowed development of an age (ODP Leg 172) indicates that they are not rare, episodic dis-
model, according to which the ages of the four excursions are turbances of an otherwise stable geomagnetic field, but an
255–265 ka, 318 3 ka, 515 3 ka, and 560–570 ka. Langereis integral component of the field. In addition, Lund et al.
et al. (1997) labeled these excursions Calabrian Ridge 0 (CR0), (2001b) proposed that most of the excursions tend to occur
CR1, CR2, and CR3. The authors proposed that CR0 could in bundles of two or three close together separated by intervals
correspond to the Fram Strait excursion (Nowaczyk et al., of ‘regular’ secular variation. This observation, in addition to
1994), while the oldest one could be the Big Lost excursion highlighting an important characteristic of the geomagnetic
(which they associated with the Emperor excursion). The field, may also explain why there has been considerable
difficulty in identifying and distinguishing among near-coeval Quidelleur et al. (1999) recognized a new, significantly youn-
excursions (i.e., Iceland Basin–Pringle Falls) prior to the avail- ger episode and proposed the name La Palma excursion. In
ability of high-resolution stratigraphy. support of this claim, Quidelleur et al. (1999) noted that the
Lund et al. (2001b) stated, ‘for almost any previously iden- proposed excursion coincides with a marked minimum in the
tified excursion anywhere in the world, we can find an excur- SINT-800 paleointensity stack of Guyodo and Valet (1999).
sion record in Sites 1060–1063 that is not significantly Subsequently, Singer et al. (2002) obtained 40Ar/39Ar ages of
different in age.’ Here lies the problem: the ODP Leg 172 580.7 7.8 ka for the same transitional flows studied by
sites lack continuous oxygen isotope data that would aid pre- Quidelleur et al. (1999). This age is indistinguishable at the
cise age model construction. The most recent age models for 95% confidence level from the 40Ar/39Ar age (558 20 ka)
ODP Leg 172 sediments were determined by tuning filtered obtained by Lanphere (2000) for the Big Lost excursion in
records of carbonate percentage, derived from grayscale reflec- the Snake River Plain. The 500–600 ka interval incorporates
tance records calibrated with shipboard and postcruise carbon- site mean magnetization directions from lava flows in the
ate measurements, to the astronomical solution for precession West Eifel volcanics that yield mid-latitude Northern Hemi-
and obliquity (Grützner et al., 2002). The age models are more sphere VGPs (Böhnel et al., 1987; Schnepp and Hradetzky,
robust for the shallow-water sites (Sites 1055–1059) than for 1994; Singer et al., 2008b). The key question is whether these
the deeper-water sites (Sites 1060–1063). Precession-related intervals with low component inclinations denote high ampli-
cycles are weak, particularly in the MIS 6–7 interval and prior tude secular variation or inadequately recorded magnetic
to 0.5 Ma for the deeper-water sites (Grützner et al., 2002). The excursions.
tuning was performed on one ‘reference’ site from each group
(Sites 1058 and 1062), and the other sites in the group were
correlated to these reference sites using the filtered and unfil- 5.10.3 Geomagnetic Excursions in the Matuyama
tered carbonate records. Chron
The correlation and labeling of ODP Leg 172 excursions
5.10.3.1 Background
have recently been facilitated by age models based on oxygen
isotope data. Channell et al. (2012) provided an oxygen iso- The age and structure of the predominantly reverse polarity
tope age model for one of the sites (ODP Site 1063) studied by Matuyama Chron have been progressively refined since the
Lund et al. (2001a,b). The age model is based on tandem coupled K–Ar and paleomagnetic studies on basaltic lavas
correlation of complete oxygen isotope and RPI records for that began with the work of Cox et al. (1963) and McDougall
this site. New u-channel magnetic data, accompanied by and Tarling (1963a,b, 1964). The Olduvai Subchron takes its
complete AF demagnetization and calculation of component name from normal polarity lavas dated at 1.72 Ma from the
magnetization directions, yield evidence for eight putative Olduvai Gorge, Tanzania (Grommé and Hay, 1963). The Jar-
excursions at ODP Site 1063 (Table 3). amillo Subchron was first recognized by Doell and Dalrymple
Some claims for new excursions in the early Brunhes Chron (1966) and takes its name from Jaramillo Creek (New Mexico).
remain controversial, particularly those from lava flows where The Réunion Subchron originates from the work of
precise radiometric dating is often lacking and critical to ade- Chamalaun and McDougall (1966) who found both normal
quate correlation among excursions. For instance, Quidelleur and reverse magnetizations in basaltic rocks yielding K–Ar ages
and Valet (1996) conducted a paleomagnetic study in the close to 2.0 Ma from the island of La Réunion. These normal
Barranco de los Tilos, La Palma (Canary Islands), at a location polarity directions were, at that time, considered to be coeval
previously studied by Abdel-Monem et al. (1972). Initially, with those from Olduvai Gorge as documented by Grommé
they proposed that transitional magnetization directions and Hay (1963). McDougall and Watkins (1973) sampled two
from the southern side of the Barranco represented post- basaltic sections on La Réunion and documented a normal
transitional rebound associated with the M/B boundary polarity zone dated by K–Ar methods to the 1.95–2.04 Ma
reversal. Later, on the basis of three transitionally magnetized interval, which corresponds to 2.07 Ma using more modern
lavas that gave a mean unspiked K–Ar age of 602 24 ka, decay constants (Steiger and Jager, 1977). By the early 1970s, it
Table 3 Eight outative excursions in the Brunhes Chron

a
Excursion Cat mcd (midpoint) MIS Age (ka) Thickness (cm) Sed. rate (cm ky–1) Duration (ky)
Laschamp (3b) 1 18.47 3 40.8 20 44 0.5

Blake (5b) 1 36.32 5d/5e 115.5 10 11.5 0.87
Iceland Basin (7a) 1 54.25 6/7 189.7 20 14 1.4
Pringle Falls (7b) 1 59.92 7 238.8 10 11.3 1.1
‘Bermuda’ (11b) 2 90.09 11 412.4 10 15 1.5
‘318 ka’ (9a) 2 69.77 9 318.1 20 31.2 0.6
‘530 ka’ (13a) 2 111.25 14 529.6 40 23.6 1.7
‘557 ka’ 2 116.03 14/15 557.3 15 13.5 1.1
a
Terminology of Lund et al. (2001a,b) in parentheses; Cat, category.
Reproduced from Channell JET, Hodell DA, and Curtis JH (2012) ODP Site 1063 (Bermuda Rise) revisited: Oxygen isotopes, excursions and paleointensity in the Brunhes Chron.
Geochemistry, Geophysics, Geosystems 13: Q02001. http://dx.doi.org/10.1029/2011GC003897
was realized that the Réunion Event is significantly older than The sequence of reversals of the Matuyama Chron in the
the Olduvai Subchron. Grommé and Hay (1971) considered CK95 differs from that in the timescale of Mankinen and
that a bimodal distribution of K–Ar ages for normally mag- Dalrymple (1979). In CK95, one (as opposed to two) normal
netized lavas with ages of 2.00–2.14 Ma from a variety of polarity subchron comprises the Réunion Subchron. CK95
locations indicated the existence of two Réunion Events, adopted the astrochronological age estimates for Pliocene–
although there was, and continues to be, no evidence for two Pleistocene polarity reversals (Hilgen, 1991a,b; Shackleton
events within any single stratigraphic section (see review in et al., 1990). These original astrochronological age estimates
Channell et al., 2003a). In their compilation of the GPTS for have generally stood the test of further astrochronological
the last 5 My, Mankinen and Dalrymple (1979) adopted the dating of sediments during the last 15 years (see Table 4 and
two Réunion Events proposed by Grommé and Hay (1971) Figure 12). The exception is the age of the Réunion Subchron
and estimated their ages as 2.01–2.04 and 2.12–2.14 Ma, in CK95 (2.14–2.15 Ma with 10 ky duration, derived from
respectively. Hilgen, 1991a,b), which should now be amended to
Mankinen et al. (1978) detected normal polarities in 2.115–2.153 Ma (38 ky duration) based on data from North
1.1 Ma volcanics from Cobb Mountain (Coso Range, Califor- Atlantic high sedimentation rate drift sites (Channell et al.,
nia); however, Mankinen and Dalrymple (1979) were not 2003a).
sufficiently confident in a normal polarity subchron of this Soon after the K–Ar ages for Pliocene–Pleistocene polarity
age to include it in their timescale. Evidence for this normal reversals (Mankinen and Dalrymple, 1979) were superseded
polarity subchron was strengthened by further study of volca- by astrochronological determinations, a large number of
40
nics in the Coso Range by Mankinen and Grommé (1982). Ar/39Ar ages confirmed the astrochronological ages of polar-
Subsequent observation in high sedimentation rate sediment ity chrons. For example, 40Ar/39Ar age determinations for the
cores from the Caribbean (Kent and Spariosu, 1983) and M/B boundary and boundaries of the Jaramillo Subchron
North Atlantic (Clement and Kent, 1987; Clement and (Baksi et al., 1993; Izett and Obradovich, 1994; McDougall
Martinson, 1992) cemented the Cobb Mountain Subchron as et al., 1992; Spell and McDougall, 1992; Tauxe et al., 1992), for
a feature of the Matuyama Chron. the Cobb Mountain Subchron (Turrin et al., 1994), for the
The GPTS of Mankinen and Dalrymple (1979), based on Réunion Subchron (Baksi et al., 1993), and for the boundaries
coupled K–Ar and paleomagnetic studies of basaltic lavas of the Olduvai Subchron (Baksi, 1994; Walter et al., 1991)
(Figure 12 and Table 4), remained the reference for the were all close to the astrochronological estimates. The exercise
0–5 Ma GPTS for over 10 years. Beginning in the 1980s, astro- becomes somewhat academic in view of the suggestion of
chronologies from sedimentary sequences demonstrated that Renne et al. (1994) that the 40Ar/39Ar standard (Fish Canyon
the K–Ar ages for polarity reversals in volcanic rocks compiled sanidine, Mmhb-1) should be calibrated using the astrochro-
by Mankinen and Dalrymple (1979) were young by an average nological ages of polarity reversals.
of about 7% (due to argon loss). The first astrochronological In CK92/95, ‘tiny wiggles’ in MMA data, interpreted as
evidence that the generally accepted K–Ar age for the M/B either brief polarity chrons (with duration <30 ky) or paleoin-
boundary (0.73 Ma) was too young can be attributed to tensity fluctuations, were labeled ‘cryptochrons.’ Fifty-four
Johnson (1982) who gave an age of 0.79 Ma for the M/B ‘cryptochrons’ were recognized over the last 83 My, since the
reversal based on matching the orbital insolation curve to middle of the Late Cretaceous. In CK92/95, two ‘cryptochrons’
oxygen isotope records from two cores (V28-238/9) that record are listed within the Matuyama Chron at 1.201–1.212 Ma
the M/B reversal. It was, however, the study of ODP Site 677 in (Cobb Mountain Subchron) and 2.420–2.441 Ma (Anomaly
the equatorial Pacific Ocean (Shackleton et al., 1990) that X in Heirtzler et al., 1968). As outlined earlier, the Cobb
opened the door to astrochronological revision of the polarity Mountain Subchron is now a well-established feature within
timescale. Benthic oxygen isotope data at ODP Site 677 are the Matuyama Chron, whereas Anomaly X has yet to be well
dominated by orbital obliquity, and the planktic record is established in magnetostratigraphic records.
controlled by orbital precession. Because the precession signal The Jaramillo, Réunion, and Olduvai normal polarity sub-
is modulated by eccentricity, the planktic oxygen isotope data chrons were identified almost 40 years ago. Since then, starting
at ODP Site 677 can be matched to astronomical data with with the unequivocal identification of the Cobb Mountain
more confidence than the obliquity signal at this site or at Subchron in marine sediments (e.g., Clement and Kent,
DSDP Site 607 (Ruddiman et al., 1989) where the number of 1987), up to nine additional normal polarity excursions, and
obliquity cycles within the Brunhes Chron was underesti- one reverse polarity excursion within the Jaramillo Subchron,
mated by a single cycle, thereby inadvertently supporting the have now been identified within the Matuyama Chron
K–Ar M/B boundary age (0.73 Ma). Although ODP Site 677 (Figure 12 and Table 4).
played a pivotal role in revamping the GPTS for the Matuyama
Chron, the site itself did not yield a polarity stratigraphy.
5.10.3.2 Excursions Between the Gauss–Matuyama
The Cobb Mountain Subchron, the M/B and Gauss–Matuyama
Boundary and the Réunion Subchron
(G/M) boundaries, and the boundaries of the Jaramillo and
Olduvai Subchronozones were recorded at DSDP Site 607 At ODP Site 982, two intervals with anomalous magnetization
(Clement and Robinson, 1987), and their ages were deter- directions are observed in MIS 85/86 and MIS 95 (Figure 12
mined by transferring astrochronological ages from ODP Site and Table 4) (Channell and Guyodo, 2004a,b). The older of
677 to DSDP Site 607 using oxygen isotope correlations the two coincides in age (2420 ka) with Anomaly X from the
between the two sites (Shackleton et al., 1990) (Figure 12 MMA data of Heirtzler et al. (1968) and hence with the ‘cryp-
and Table 4). tochron’ of this age featured in CK92/95.
K–Ar ages ODP Site 677 Ar/Ar ages

(volcanics) DSDP Site 607 (volcanics) ODP Sites 980–984
Ma ka ka ka ka
0.70 Brunhes
730 Brunhes Brunhes Brunhes
780 779 774

0.80
850 Kamikatsura
886 Kamikatsura
0.90 900 922 Santa Rosa 932 Santa Rosa
Jaramillo
970
990 986 987
1.00
Jaramillo Jaramillo (1r.1n)
Jaramillo
1053 1048(1r.1n.1r)
1070 1068
1.10 1105 Punaruu
1115 Punaruu (1r.2r.1n)
1181 Cobb Mt.

1.20 1190 Cobb Mt. 1190–1215 Cobb Mt. (1r.2r.2n)
1255 Bjorn (1r.2r.3n)

1.30
1.40
1.50 1465–1485 Gardar (1r.2r.4n)
1567–1575 Gilsa (1r.2r.5n)

1.60
1670
1.70
Olduvai 1770
1.80 1778
Olduvai
1870 Olduvai (2n)
1.90
1950 1945
1977 (2r.1r.1n)
2.00 2010
2040
Reunion
2.10
2120 2115
2140
Reunion (2r.1n)
2.20 2153
2236 ?
2.30
2.40
2421 ?
2480
2.50
Gauss
2.60 2595
2600
Gauss (2An)
Gauss
Figure 12 The geomagnetic polarity timescale for the Matuyama Chron. From left: (a) coupled K–Ar and paleomagnetic studies summarized by
Mankinen and Dalrymple (1979); (b) ODP Site 677 and DSDP Site 607 (Shackleton et al., 1990) as adopted by Cande and Kent (1995); (c) 40Ar/39Ar
and paleomagnetic studies of volcanics (summarized by Singer et al., 2004); (d) ODP Sites 980–984 (Channell and Guyodo, 2004a,b; Channell and
Kleiven, 2000; Channell et al., 2002).
366
Geomagnetic Excursions
Table 4 Excursions within the Matuyama Chron
Excursion or subchron Label MIS Sites 607, MIS Site 659 (2) MIS Italy (3) MIS (Sites 980–984) (4) Age (ka) (Sites Duration (ky) (Sites Age (ka)
40
609, 677 (1) 980–984) (4) 980–984) (4) Ar/39Ar (5)
Base Brunhes Base 1n Base 19 19 774 791

Kamakitsura 21 850 ? 899
Santa Rosa Top 25 932 3 936
Top Jaramillo Top 1r.1n Mid 27 27 Base 27 985 8 1001
1r.1n.1r Base 30 1050 3
Base Jaramillo Base 1r.1n Mid 31 31 Base 31 1070 5 1069
Punaruu Mid 34 1115 5 1122
Top Cobb Mountain Base 35 Base 35 1190
Cobb Mountain 35 1194
Base Cobb Mountain Base 35 Top 37 1125
Bjorn Top 38 1255 3
Gardar 49 1465–1485 20
Gilsa 53 54 1567–1575 8
Top Olduvai Top 2n Base 63 64 64 63 1778 4 1775
Base Olduvai Base 2n Base 71 72 71 Base 71 1945 5 1922
Pre-Olduvai Top 73 1977 3
Huckleberry Ridge 75 2040 2086
Top Réunion Top 2r.1n 79 81 79 2115
Réunion 38 2137
Base Réunion Base 2r.1n 81 81 2153
Pre-Réunion 1 85/86 2236 ?
Pre-Réunion 2 95 2421 ?
Top Gauss 104 Base 103 2595
MIS: marine isotope stage.

(1) Ruddiman et al. (1989), Raymo et al. (1989), Shackleton et al. (1990).
(2) Tiedemann et al. (1994).
(3) Lourens et al. (1996).
(4) Channell and Kleiven (2000), Channell et al. (2002), Channell et al. (2003a), Channell and Guyodo (2004a,b).
(5) Summarized in Singer et al. (2004).
5.10.3.3 Huckleberry Ridge presumably due to their brief duration and paucity of deep-sea
cores of this age with sufficiently high sedimentation rates. The
Reynolds (1977) recorded anomalously shallow inclinations
names for these excursions are derived from the locations of
and southwestward-directed declinations in paleomagnetic
Sites 984 (Bjorn Drift) and 983 (Gardar Drift); however, both
data from 57 sites representing 23 separate localities in the
excursions are present at both sites.
Huckleberry Ridge Tuff (Yellowstone group). The anomalous
directions were considered by Reynolds (1977) to record a
polarity transition or an excursion, possibly associated with 5.10.3.6 Cobb Mountain
the Réunion Subchron. The tuff has recently been dated at
Mankinen et al. (1978) documented a normal polarity site in
2.06 Ma using 40Ar/39Ar methods (Lanphere et al., 2002),
the Alder Creek rhyolite at Cobb Mountain (California), which
indicating that the transitional directions are younger than
has yielded an 40Ar/39Ar age of 1186 ka (Turrin et al., 1994).
the Réunion Subchron for which the mean 40Ar/39Ar age is
Further work by Mankinen and Grommé (1982) in the Coso
2.14 Ma (Baksi and Hoffman, 2000; Baksi et al., 1993; Roger
Range (California) supported the existence of the Cobb Moun-
et al., 2000; Singer et al., 2004). At ODP Site 981, low inclina-
tain Event. DSDP Site 609 in the central North Atlantic Ocean
tions are recorded in MIS 75/76, stratigraphically above the
provided the first unequivocal documentation of a normal
Réunion Subchronozone, which occurs in MIS 79–81
polarity zone of similar age in deep-sea sediments (Clement
(Channell et al., 2003a). It remains to be confirmed whether
and Kent, 1987). At Site 609, the so-called Cobb Mountain
the anomalous directions within MIS 75/76 correlate to the
Subchron can be correlated to MIS 35/36 (Ruddiman et al.,
Huckleberry Ridge excursion as recorded in the Huckleberry
1989) (Table 4). This subchron has also been recognized at
Ridge Tuff.
ODP Hole 647B in the southern Labrador Sea (Clement and
Martinson, 1992), in the Celebes and Sulu Seas (Clement,
5.10.3.4 Gilsa 1992; Hsu et al., 1990), in the Lau Basin (Abrahamsen and
Sager, 1994), in New Zealand (Pillans et al., 1994), off the
The name Gilsa, in a geomagnetic context, owes its origin to California Margin (Guyodo et al., 1999; Hayashida et al.,
McDougall and Wensink (1966) who detected two normal 1999), in the West Philippine Sea (Horng et al., 2002, 2003),
polarity flows separated by a reverse polarity flow within the on the Bermuda Rise (ODP Leg 172) (Yang et al., 2001), and at
lower part of the Matuyama Chronozone at Jokuldalur ODP Sites 980 and 983/984 and at IODP Site U1308, where it
(Iceland). They assigned the older normal polarity flow to the correlates to MIS 35 at 1.2 Ma and has an estimated duration of
Olduvai Subchronozone and the younger normal polarity 35 ky (Channell et al., 2002, 2008; Channell and Raymo,
interval, dated at 1.60 Ma, was named the Gilsa Event. 2003).
Watkins et al. (1975) subsequently sampled the same sections
and measured normal polarity directions in superposed lava
flows that yielded ages of 1.58 and 1.67 Ma, but with no 5.10.3.7 Punaruu
intervening reverse polarity interval. More recently, Udagawa The Punaruu excursion originates from the Punaruu Valley
et al. (1999), from work on five separate sections in the (Tahiti) where normal polarity magnetizations were recorded
Jokuldalur region, confirmed the existence of a reverse polarity in basaltic lava flows stratigraphically below, and distinct from,
flow below the 1.60 Ma normal polarity flow and above the the Jaramillo Subchronozone (Chauvin et al., 1990). The
normal polarity flows associated with the Olduvai Chrono- Punaruu excursion in the type section on Tahiti has yielded
zone. This work supports the conclusion of McDougall and an age of 1105 ka using 40Ar/39Ar methods (Singer et al.,
Wensink (1966) that the Gilsa excursion (at 1.60 Ma) is an 1999). This excursion appears in the sediment record from
interval of normal polarity distinct from the underlying ODP Site 1021 (California Margin) where an age of 1.1 Ma is
Olduvai Chronozone. In view of the Watkins et al. (1975) based on assumed uniform sedimentation rate between the
study, Mankinen and Dalrymple (1979) were not sufficiently M/B boundary and the top of the Jaramillo Subchronozone
confident to include the Gilsa Event in their timescale, nor was (Guyodo et al., 1999). At both sites 983 and 984, the same
it included in CK95 as a ‘cryptochron’ as it is not evident in excursion lies within MIS 34, which corresponds to an age of
MMA records. The presence of a normal polarity interval 1115 ka (Channell et al., 2002).
at 1.55 Ma in two holes at DSDP Site 609, however, con-
firmed its existence (Clement and Kent, 1987). This was corrob-
orated at ODP Sites 983 and 984, where the correlative normal 5.10.3.8 Intra-Jaramillo Excursion (1r.1n.1r)
polarity excursion correlates to MIS 54 at 1567–1575 ka An intra-Jaramillo excursion has been detected in the Jingbian
(Table 4), implying duration of 8 ky (Channell et al., 2002). loess sequence from Northern China within loess 10 (Guo
et al., 2002), in marine cores from New Zealand (Pillans
et al., 1994), and at ODP sites 983 and 984 where it is corre-
5.10.3.5 Gardar and Bjorn
lated to MIS 30 at 1048 ka (Channell and Kleiven, 2000;
At ODP Sites 983 and 984, two normal polarity excursions are Channell et al., 2002).
present between the Gilsa excursion zone and the Cobb Moun-
tain Subchronozone (Channell et al., 2002). These two normal
5.10.3.9 Santa Rosa
polarity intervals, which occur in MIS 49 (Gardar excursion)
and MIS 38 (Bjorn excursion), with estimated durations of 8 ky Transitional magnetization directions from volcanic rocks at
(Gardar) and 3 ky (Bjorn), have not been detected elsewhere, Cerro Santa Rosa I dome in New Mexico (Doell and
Dalrymple, 1966; Doell et al., 1968) were originally inter- Kamikatsura excursion and paleointensity minimum immedi-
preted as recording the polarity transition at the end of the ately prior to the M/B boundary (Hartl and Tauxe, 1996; Kent
Jaramillo Subchron. 40Ar/39Ar ages (Izett and Obradovich, and Schneider, 1995) remains to be determined.
1994; Spell and McDougall, 1992) from this dome are signif-
icantly younger than the end of the Jaramillo Subchron. More
recently, anomalous (transitional) magnetization directions 5.10.4 Geomagnetic Excursions in
(mean: dec/inc: 103 / 63 , a95 ¼ 9.5 ) and a mean 40Ar/39Ar Pre-Matuyama Time
age of 936 ka (Singer and Brown, 2002) support the existence
of a Santa Rosa excursion. An excursion of similar age (932 ka) In the Gauss and Gilbert Chrons, there are no well-documented
has been observed at ODP Sites 983 and 984 at the top of MIS excursions in magnetostratigraphic records, nor are there ‘tiny
25 (Channell et al., 2002) and at the same position in the West wiggles’ in MMA data denoting ‘cryptochrons’ of this age.
Philippine Sea (Horng et al., 2002, 2003). Based on ‘tiny wiggles’ in MMA data, CK92/95 included four
‘cryptochrons’ in the Late Miocene, 18 in the Oligocene, three
in the Late Eocene, and 23 in the Paleocene and Early Eocene.
5.10.3.10 Kamikatsura
These ‘tiny wiggles’ can often be correlated between ship’s
The name ‘Kamikatsura excursion’ originates from the work of tracks, and they have been thought to represent fluctuations
Maenaka (1983) who documented excursional magnetization in geomagnetic field intensity (Cande and Kent, 1992b; Cande
directions in the Kamikatsura Tuff of the Osaka group (SW and LaBrecque, 1974) and/or short polarity intervals (e.g.,
Japan). The existence of this excursion is supported by a Blakely, 1974; Blakely and Cox, 1972).
0.83 Ma normal polarity flow from near Clear Lake From magnetostratigraphic studies of sedimentary
(Mankinen et al., 1981) and was promoted by Champion sequences, there is evidence for 14 polarity excursions in
et al. (1988) in their review of Matuyama–Brunhes excursions. the Miocene (Figure 14), four of which correlate with crypto-
More recently, Takatsugi and Hyodo (1995) documented chrons in CK92/95. The Miocene magnetostratigraphic record,
another excursion in marine clays about 10 m above the like that in the Brunhes Chron, implies that there are far more
Kamikatsura Tuff and 1 m above another tuff (Azuki Tuff) polarity excursions than are represented in the cryptochron
that has yielded a K–Ar age of 0.85 Ma. The Kamikatsura and record from MMA data. On the other hand, whereas 44 Paleo-
Azuki Tuffs are separated by about 10 m of pebbly sand in the cene to Oligocene (Paleogene) cryptochrons are given in
Osaka group (see Takatsugi and Hyodo, 1995), and rapid CK92/95, only three excursions have been documented in the
deposition of this facies may mean that the two excursional Paleogene, and only one of these appears to correlate with a
intervals record a single geomagnetic excursion. On the other cryptochron in CK92/95.
hand, two excursions with age estimates of 0.89 and 0.92 Ma In the Middle Cretaceous, at the base of the Cretaceous long
have been detected in loess 9 (L9), which corresponds to MIS normal interval, the existence of reverse polarity subchrons
22 according to the loess chronology of Heslop et al. (2000), in younger than CM0 has been advocated (e.g., Tarduno et al.,
the Baoji loess section in Southern China (Yang et al., 2004). 1992) although their existence remains controversial. Here, the
This observation follows earlier work in the Lishi and Luo- uncertainty is due to the lack of continuous sedimentary sec-
chuan regions (China) where an apparent excursion, also tions recording both CM0 (at the base of the Cretaceous long
within L9, has been recorded (Liu et al., 1985; Wang et al., normal) and the younger proposed subchrons and the inabil-
1990). Although excursional directions in loess deposits have ity of 40Ar/39Ar age dating to clearly distinguish these sub-
often been dismissed as remagnetizations in view of the prob- chrons in discontinuous volcanic sequences.
able delay of remanence acquisition of Chinese loess (see
Spassov et al., 2003), an interval of anomalously low VGP
5.10.4.1 C5n.2n (Late Miocene)
latitudes also occurs at ODP Site 983 in MIS 21 at about
850 ka (Channell and Kleiven, 2000). An interval of negative Late Miocene polarity subchron C5n.2n in the CK92/95 GPTS
(reverse) inclination in equatorial Pacific core KK78O30 includes three ‘cryptochrons’ first recorded as ‘tiny wiggles’ in
between the M/B boundary and the Jaramillo Subchronozone MMA profiles from the North Pacific Ocean (Blakely, 1974).
(Laj et al., 1996) was assigned to the Kamikatsura excursion. Its Over the last 25 years, ‘tiny wiggles’ in MMA records of C5n.2n
age (based on uniform sedimentation rate between the M/B have been represented as polarity subchrons in several versions
boundary and the top Jaramillo Subchronozone) is closer to of the GPTS notably those of Ness et al. (1980) and Harland
that of the Santa Rosa excursion, which was not recognized at et al. (1982, 1990). They were not included in the GPTS by
the time. Low VGP latitudes from volcanics on Maui and Tahiti Lowrie and Alvarez (1981) or Berggren et al. (1985) because of
associated with the Kamikatsura excursion yield mean ages of the lack of confirmation for polarity subchrons of this age from
866 ka (Singer et al., 1999). Coe et al. (2004) associated a 25 m magnetostratigraphy.
thick interval of anomalous declinations on Maui with the A single track of deep-tow magnetic anomaly data, covering
Kamikatsura excursion and gave a weighted mean 40Ar/39Ar polarity subchron C5n.2n, was collected in 1998 at 19S on
age of 900.3 4.7 ka for this interval. To add to the confusion the flanks of the East Pacific Rise (Bowers et al., 2001).
in this interval, a transitionally magnetized flow from La Palma The half-spreading rate at this site was estimated to have been
(Canary Islands) has yielded a K–Ar age of 821 ka (Quidelleur 9 cm year1 for the Late Miocene. This profile revealed more
et al., 2002). There is clearly much to be done to resolve the than twice as many ‘tiny wiggles’ as seen in profiles from the
behavior of the geomagnetic field in this interval immediately North Pacific region, where Blakely (1974) first identified
prior to the M/B boundary. The relationship of the the four short-wavelength anomalies within C5. Bowers et al.
(2001) identified a number of short-wavelength magnetic 5.10.4.2 Other Miocene Excursions/Subchrons

anomalies within C5n.2n, three of which were interpreted as
In sediments from the equatorial Pacific Ocean recovered dur-
correlative to the three cryptochrons in CK92/95. Bowles et al.
ing ODP Leg 138, a normal polarity subchron was recognized
(2003) addressed the question as to whether these tiny wiggles
within C4r.2r (Schneider, 1995) that correlates to polarity
within C5n.2n represent polarity subchrons/excursions or
subchrons recorded in sediments from ODP Site 1092 in the
paleointensity fluctuations. These authors compared the direc-
South Atlantic (Evans and Channell, 2003; Evans et al., 2004)
tional and paleointensity record from C5n.2n in the sedimen-
and from ODP Site 1095 off the Antarctic Peninsula (Acton
tary sequence at ODP Site 887 with the deep-tow record of
et al., 2006), as well as to a cryptochron in CK92/95 (Figure 1).
Bowers et al. (2001). The absence of directional excursions in
Independent verification from multiple sites argues strongly
the magnetostratigraphic record of C5n.2n led Bowles et al.
for an excursion or brief subchron within C4r.2r (Figure 13).
(2003) to conclude that the tiny wiggles in MMA records of
Three other subchrons or excursions within C4r.1r,
C5n.2n were produced by paleointensity fluctuations rather
C4Ar.1r, and C5r.2r (Figure 13) were apparent from the
than directional excursions or brief polarity subchrons.
study of ODP Leg 138 sediments (Schneider, 1995). All three
The magnetic polarity stratigraphy at ODP Site 884, on the
of these subchrons (or excursions) are apparently supported by
slopes of the Detroit Seamount (NW Pacific), was initially inter-
recent data from a resampling of DSDP Site 608 (Krijgsman
preted from the shipboard pass-through magnetometer measure-
and Kent, 2004). One of these subchrons/excursions, that
ments of split half-cores (Weeks et al., 1995a,b). The C5n part of
within C5r.2r, is also supported by the studies of Abdul-Aziz
the section was reinterpreted by Roberts and Lewin-Harris (2000)
and Langereis (2004) in continental sediments from NE Spain.
who recognized two short polarity subchrons and one excursion
Abdul-Aziz and Langereis (2004) documented a total of three
within a polarity zone correlative to C5n.2n (see Figure 3 in
subchrons within C5r (Figure 13). The oldest of the C5r excur-
Roberts and Lewin-Harris, 2000). The excursion and subchrons
sions (that within C5r.3r) may correlate with a subchron found
were estimated to have durations of 6, 23, and 28 ky, respectively,
within the same polarity chron at ODP Site 1092 (Evans and
assuming constant sedimentation rates in C5n.2n.
Channell, 2003). At ODP Site 1092, yet another newly recog-
At ODP Site 1092, in the sub-Antarctic South Atlantic, the
nized subchron has been observed within chron C5ACr (rein-
stratigraphic interval correlative to C5n.2n is 34 m thick with a
terpreted from the initial placement within C5AAr by Evans
mean sedimentation rate of 3.3 cm ky1. At this site, three
and Channell, 2003). The plethora of Late and Middle Mio-
polarity excursions are recognized within C5n.2n, all of which
cene subchrons recognized in stratigraphic sections clearly out-
have estimated durations <10 ky (Evans and Channell, 2003;
number the cryptochrons identified in CK92/95 derived from
Evans et al., 2004). The presence of directional excursions
MMA data (Figure 13). At least 13 subchrons that were not
within C5n.2n has recently been supported by measurements
included in the CK92/95 polarity chron template have been
of new discrete samples collected at DSDP Site 608 (Krijgsman
documented in stratigraphic section since the publication of
and Kent, 2004), although at this site, the excursions correlated
CK95.
to C5n.2n are represented by single samples with shallow neg-
Four of 13 newly recognized subchrons correlate with cryp-
ative inclinations.
tochrons in CK92/95.
Continental records of short polarity events in C5n.2n have
been reported from NE Spain (Garces et al., 1996), Bolivia
(Roperch et al., 1999), and Western China (Li et al., 1997).
5.10.4.3 Oligocene and Eocene
The section from the Pyrenees in NE Spain consists of a com-
posite section from two terrestrial sequences that cover the Three ‘cryptochrons’ appear in C13r in the CK92/95 GPTS
interval from 8.7 to 11.1 Ma (Garces et al., 1996). The mag- (Figure 14) based on ‘tiny wiggles’ in MMA data. Bice and
netic stratigraphy can be interpreted on the basis of polarity Montanari (1988) found one normal polarity sample in the
zone pattern fit to the GPTS aided by the first occurrence datum top third of C13r at Massignano, Italy, in the Eocene–
of Hipparion. The polarity zone correlative to C5n.2n has a Oligocene boundary stratotype section. Lowrie and Lanci
thickness of 175 m and includes a 5 m polarity zone in its (1994) resampled the Massignano section and concluded
upper part that was correlated with cryptochron C5n.2n 1 in that there were no normal polarity intervals within C13r. This
CK92/95. A 4.5 km thick composite section of red beds from conclusion appears to be corroborated by the study of cores
the Bolivian Altiplano produced a magnetic stratigraphy for the recovered from a borehole drilled at Massignano (Lanci et al.,
9–14 Ma interval (Roperch et al., 1999). The polarity zone 1996). At ODP Site 1090, in the sub-Antarctic South Atlantic
correlative to C5n.2n is 1 km thick in this section and contains Ocean, the high sedimentation rates in C13r appear to have
a single 5 m thick polarity zone that was correlated with cryp- facilitated the recording of a short subchron represented by a
tochron C5n.2n 1. The interpretation of the polarity stratig- 3 m thick normal polarity zone, implying a duration for the
raphy is supported by 40Ar/39Ar age determinations on two tuff C13r.1n subchron of 79 ky, which is comparable to the dura-
layers within the section, one of which falls within C5n.2n. An tion of the Jaramillo Subchron (Channell et al., 2003b). Apart
upper Cenozoic section in Western China recorded three short from the apparent subchron within C13r, two additional
reverse polarity zones within C5n.2n (Li et al., 1997). The polarity subchrons not included in the CK92/95 GPTS are
polarity interpretation at the site is fairly unambiguous and is observed at ODP Site 1090 (Channell et al., 2003b). One of
corroborated by the presence of vertebrate fossils. The events these (within C5Dr) is listed as a cryptochron in Table 7 of
are, however, only represented by single discrete samples. The CK92 and was included as a subchron in versions of the GPTS
authors correlate these events with the three cryptochrons in prior to CK92/95. This normal polarity subchron was origi-
CK92/95. nally identified in the North Pacific MMA stack (Blakely,
Crypto- 1974). Two other normal polarity subchrons have been docu-
Time Chrons Polarity chrons Epochs
(Ma) (line) mented in the Late Oligocene (Figure 14). One of these occurs
Excur.(*)
Pliocene
within C7Ar and is recorded at ODP Site 1090 (Channell et al.,
3r
2003b) and another occurs within C8n.1n at ODP Site 1218 in
n the equatorial Pacific Ocean (Lanci et al., 2005). Neither of
6 3An 3An.1 r
these subchrons has been recognized in other magnetostrati-
3An.2 n
3Ar graphic records, or in MMA records.
3Bn
7 1r/n
3Br 2r/n L
3r
4n.1 n
r
*(3)
a 5.10.4.4 Middle Cretaceous
4n
4n.2 n t
8 4r.1 r *(1,5) e The record of polarity subchrons in the Middle Cretaceous
n
4r begins with the so-called ISEA reversal that was originally
4r.2 r
*(1,3)
4An
recognized in a bed of reddish Aptian limestone along a road
9 r *(1,5) outcrop in the Umbrian Apennines (Italy) by VandenBerg et al.
4Ar.1
4Ar n
4Ar.2 r
n
(1978). Lowrie et al. (1980) resampled this anomalous lime-
4Ar.3 rn
5n.1 r stone bed and confirmed the reverse polarity magnetization
10 *(3,4,5) but were unable to substantiate the existence of this reverse
5n 5n.2 n *(3,4,5)
*(3,4,5)
polarity zone in other coeval Umbrian sections. Some 10 years
5r.1 r later, Tarduno (1990) documented reverse polarity magnetiza-
n
11 5r.2 *(1,4,5) tions in two samples within the Globigerinelloides algerianus
5r r *(4,5) M
n
*(4,5) i foraminiferal zone of the middle Aptian at DSDP Site 463.
5r.3 r
*(3) Documentation of this short polarity subchron is strengthened
12 5An.1 n
r
o
5An
5An.2 n *(5) M by two samples with reverse magnetization spanning a 43 cm
c
5Ar.1 r
n i interval at ODP Site 765 (Ogg et al., 1992). Reverse magneti-
5Ar r/n e
5Ar.2
5Ar.3 r d zations in basalts from the Tarim Basin (China) yielding
13 5AAn n 40
5AAr d Ar/39Ar whole rock and plagioclase fraction ages of 113 and
5ABn e l
5ABr 119 Ma, respectively (Sobel and Arnaud, 2000), have been
e
5ACn associated with ISEA (Gilder et al., 2003). According to some
14 5ACr *(3)
timescales (e.g., Channell et al., 1995a), an age of 119 Ma is
5ADn
less than 2 My younger than CM0 (the youngest reverse polar-
5ADr
15 5Bn.1 n
r
ity chron of the M-sequence) at the Aptian/Barremian bound-
5Bn
5Bn.2 n ary. On the other hand, in the timescale of Gradstein et al.
5Br (2004), an age span of 113–119 Ma would correspond to the
16 Late Aptian, with 119 Ma corresponding closely to the sup-
5Cn.1 n
5Cn 5Cn.2
r
n posed age of ISEA. An andesitic lava sequence from Liaoning
r
5Cn.3 n Province (China) also yields reverse magnetization directions
17 5Cr and 40Ar/39Ar ages of 116.8 and 3.0 Ma (Zhu et al., 2004), and
5Dn these authors associate the reverse magnetizations either to
ISEA or to CM0.
5Dr *(6)
18 Tarduno et al. (1992) documented seven reverse polarity
zones in the middle Albian of the Contessa section (Umbria,
5En Italy). The cyclostratigraphy of Herbert et al. (1995) applied to
19 5Er
E the Contessa section implies a duration of 800 ky for the
a
thickest of these polarity zones. As this duration is greater than
6n r
the estimated duration of CM0, the polarity chron at the base
l
20 of the Aptian that is easily recognized in MMA records, it is
y
6r unlikely the Albian polarity zones documented in the Contessa
n section represent reverse polarity chrons. As the reversely mag-
6An 6An.1
r
21 netized limestones in the Albian at Contessa are reddish, and
6An.2 n
the NRM is partly carried by hematite, the hematite magneti-
6Ar
zations may be recording a Late Cretaceous or younger mag-
22 6AAn
6AAr.1 r netization rather than an Albian one. On the other hand,
6AAr n
6AAr.2 r
n
reverse polarity magnetizations have been retrieved from sed-
3 n
6Bn
6Bn.1 r iments of Albian age recovered during ODP Leg 171 although
23 6Bn.2r n
6Br
6Cn.1 n
6Cn r
n
polarity subchrons/excursions according to various authors: (1)
6Cn.2
Schneider (1995), (2) Acton et al. (2006), (3) Evans and Channell (2003),
Figure 13 Miocene geomagnetic polarity timescale of Cande and Kent Evans et al. (2004), (4) Abdul-Aziz and Langereis (2004), (5)
(1995) with cryptochrons from Cande and Kent (1992a) and additional Krijgsman and Kent (2004), and (6) Channell et al. (2003b).
Time Chrons
Crypto-
Polarity chrons Epochs
5.10.5 Duration of Geomagnetic Excursions
(line)
(Ma) Excur,(*)
6Cn.1 n Precise evaluation of the duration of geomagnetic excursions
6Cn r
n
Miocene
24
6Cn.2 r has become a point of interest since the proposal of Gubbins
6Cn.3 n
(1999) that, during excursions, the geomagnetic field reverses
6Cr
polarity in the Earth’s liquid outer core but that this outer-core
7n.1
25 7n 7n.2 field does not persist for long enough in a reversed polarity
7r
L state for diffusion of the field into the solid inner core. Diffu-
7An
7Ar n
*(6) a sion times of 3 ky for the inner core, therefore, provide a
26 8n.1 r *(7)
O t prediction for excursion duration. It is not trivial to determine
8n 8n.2 n
l e excursion durations of a few thousand years in the geologic
8r i
27 record. Even in cases where astronomical tuning is possible to
g
9n o the level of orbital precession, we can only expect durations of
c the order of thousands of years to be adequately resolved.
28 9r e Interpolation, which necessarily assumes constant sedimenta-
10n.1 n n tion rates, among tie points matching an oxygen isotope record
10n r
10n.2 n e to an (astronomically tuned) isotope target curve is unlikely to
29 10r
provide duration estimates of sufficient precision. Radiometric
11n
11n.1 n
r (40Ar/39Ar or K–Ar) ages are also unlikely to have realistic
30 11n.2 n uncertainties within a few thousand years, limiting their utility
11r
E for estimating the duration of excursions.
12n a There are also differences in the way the different authors
31 r have determined the duration of an excursion. An excursion
l (as well as a reversal) starts with a significant decrease in the
y geomagnetic field intensity. The geomagnetic field is a vector,
12r
32
and therefore, the decrease in intensity is an essential charac-
teristic of the excursion and should consequently be consid-
33 ered when evaluating its duration. In most cases, only the
13n directional change is used to evaluate the duration of the
excursion. For the Laschamp excursion, the duration was deter-
34 13r *(6) mined from the NAPIS record as 1.5 ky (Laj et al., 2000)
based on correlation of marine cores to the GISP2
L
15n E a (Greenland) layer-counted ice-core chronology. The correla-
35 15r o t tion among the NAPIS-75 cores was achieved through correla-
16n.1 n c e tion of the susceptibility records (Kissel et al., 1999), and the
r
16n e oxygen isotope record from one of the cores provided a high-
36 16n.2 n
n resolution correlation to GISP2 (Voelker et al., 1998). A similar
16r e
duration estimate was obtained from each of the five NAPIS-75
17n.1
37 cores that record the excursion. As these cores are spread over a
Figure 14 Late Eocene–Oligocene geomagnetic polarity timescale of distance of about 5000 km, it is unlikely that the duration
Cande and Kent (1995) with cryptochrons from Cande and Kent (1992a) estimate is affected by local changes in the sediment deposition
and additional polarity subchrons/excursions according to (6) rate at the time of the excursion.
Channell et al. (2003b), and (7) Lanci et al. (2005). Both 10Be and sedimentary RPI records recently transferred
onto the layer counting-based age model of the NorthGRIP ice
these varicolored sediments indicate extensive iron mobiliza- core (GICC05) (Kissel et al., 2011; Svensson et al., 2006, 2008)
tion and probably indicate a (Late Cretaceous) remagnetiza- are consistent with the radiometric age and the 1.5 ky duration
tion (Ogg and Bardot, 2001). The template for the Late Jurassic estimate. Although the GICC05 age model may be wrong by
and Early Cretaceous polarity sequence (CM0–CM25) stems 1 ky at the time of the Laschamp excursion (Svensson et al.,
from the MMA records published by Larson and Hilde (1975). 2008), its relative accuracy for the interval of the excursion is
Correlation of this sequence of polarity chrons to biozona- better than 60 years, making the duration estimate from the ice
tions/stage boundaries and numerical ages has been reviewed core very robust.
by Channell et al. (1995a). The only modification of this Further evidence for the duration of the Laschamp excur-
template in the intervening 30 years since publication of sion was obtained from the record of flux of 36Cl in the GRIP
Larson and Hilde (1975) has been the identification of a ice core. Wagner et al. (2000) showed that variations in 36Cl
second reverse polarity subchron between CM11 and CM12. flux (assumed to be entirely due to modulation by the geo-
The two reverse polarity subchrons within CM12n have been magnetic field) are similar, in inverse sense, to changes in
recognized in MMA records (Tamaki and Larson, 1988) and in geomagnetic field intensity (e.g., Figure 7). A duration estimate
magnetostratigraphic sections in Italy (Channell et al., 1987, of 1.5 ky for the 36Cl anomaly, corresponding to the paleoin-
1995b). tensity minimum associated with the Laschamp excursion,
supports the excursion duration determined from sedimentary controls rather than geomagnetic behavior. Since then,
paleomagnetic records. The 36Cl flux ice-core record also indi- O’Regan et al. (2008) utilized apparent excursions recorded
cates that a second peak, associated with the Mono Lake excur- in these Pleistocene sediments to constrain age models. Further
sion, has approximately the same duration, about 1.5 ky (on investigation of the magnetic properties of Brunhes-age sedi-
the GRIP–GISP age model). ments from the Arctic Ocean has led to the hypothesis that
As discussed in the preceding text, sedimentary records of maghemitization of detrital titanomagnetite under oxic diage-
excursions are, unfortunately, not always associated with RPI netic conditions can account for some of the observed excur-
records, only the directional changes are reported. There is sional magnetization directions (Channell and Xuan, 2009;
generally good consistency in estimates of excursion duration, Xuan and Channell, 2010; Xuan et al., 2012).
even from cores deprived of astrochronological tuning to In summary, the duration estimates for the Laschamp,
orbital solutions. For example, Lund et al. (2001a, 2005) esti- Mono Lake, and Iceland Basin excursions (Table 2) and for
mated a duration of 2 ky for the Laschamp excursion using most Matuyama Chron excursions (Table 3) provide support
age models based on AMS radiocarbon ages from Keigwin and for the excursion mechanism of Gubbins (1999) in that the
Jones (1994). A similar duration is apparent for records of the excursion durations are estimated to be no more than 2–3 ky,
Laschamp excursion in the South Atlantic Ocean (Channell which is comparable with the magnetic diffusion time of the
et al., 2000) and the southern Indian Ocean (Mazaud et al., inner core. This result is consistent with the concept that excur-
2002). In the Irminger Basin, both the Mono Lake and sions can lead to a full polarity reversal and a subsequent
Laschamp excursions have apparent durations of 1 ky prolonged polarity interval only if the excursion (with direc-
(Channell, 2006). At ODP Site 1063, where the sedimentation tions approximately antiparallel to the preexcursional field) is
rate in the vicinity of the Laschamp excursion is estimated to be maintained for times exceeding the magnetic diffusion time of
44 cm ky1 based on the oxygen isotope age model, the dura- the inner core.
tion estimate for the Laschamp excursion is 0.5 ky (Channell
et al., 2012).
For records that have oxygen isotope age control, duration 5.10.6 Excursional Field Geometry
estimates for the Iceland Basin excursion range from 2 to 3 ky
(e.g., Channell, 1999; Channell et al., 1997; Laj et al., 2006; Determination of the field geometry during an excursion
Stoner et al., 2003) to 1.4 ky at ODP Site 1063 (Channell requires availability of multiple records from different and
et al., 2012). A longer duration (8 ky) for the Iceland Basin widely separated geographic locations. Only an incomplete
excursion recorded at ODP Site 919 (Channell, 2006) may be picture can be obtained, because only two excursions, the
attributable to inadequacy of the oxygen isotope age model. Laschamp and Iceland Basin excursions, have been studied at
The duration of 2 ky is implied by the record of the Iceland distant localities from rapidly deposited sediments that yield
Basin excursion from two cores from the Eirik Drift where the detailed VGP paths. For these two excursions, Laj et al. (2006)
sedimentation rate is 5–7 cm ky1 in the excursional interval had proposed a relatively simple behavior of the excursional
(Evans et al., 2007). field. For the Laschamp excursion, the initial southward part of
A somewhat longer duration has been documented for the the VGP paths during the excursion passes over East
Blake excursion, with values of 4–5 ky (Bourne et al., 2012; Asian–West Pacific longitudes and then reaches high southern
Osete et al., 2012; Tric et al., 1991), although a shorter dura- latitudes (Figure 6). The northward-directed part of the VGP
tion of 1 ky is estimated by Channell et al. (2012) at ODP paths proceeds over Africa and Western Europe, with a large
Site 1063. The longer duration estimates are about 2–3 times clockwise loop. For the Iceland Basin excursion (Figure 9),
larger than most of those observed for the Laschamp, Mono VGPs move first southward over Africa along a rather narrow
Lake, and Iceland Basin excursions, although the duration band of longitudes before crossing the equator and reaching
estimates are heavily dependent on the quality of the individ- high southern latitudes. The return paths to the Northern
ual age models. It should be noted that the structure of this Hemisphere are contained in a longitudinal band over East
excursion is more complicated than for the other Brunhes Asia. The overall picture is that of a large counterclockwise
Chron excursions. loop of the VGPs. For both excursions, the turning point,
According to Clement’s (2004) review, reversal duration at where the VGPs change from being southward to northward,
the Matuyama–Brunhes boundary, and at the boundaries of coincides with the minimum in RPI. There is no clear evidence
the Jaramillo and Olduvai Subchrons, varies with latitude and for recovery of paleointensity (except in one core for the
falls in the 2–10 ky range. This distribution of reversal dura- Laschamp excursion) within the excursion interval, although
tions with latitude is broadly consistent with a model in which a recovery in paleointensity may be filtered by the DRM acqui-
nondipole fields are allowed to persist, while the axial dipole sition process.
decays to zero and then builds in the opposite direction. The This simple geometry underlines similarities among the
apparent increase in excursion duration in high-latitude cores records obtained for the Laschamp and Iceland Basin excur-
(e.g., Nowaczyk and Antonow, 1997; Nowaczyk and Knies, sions, although the sense of looping of the VGP path is oppo-
2000) appears to be greater than can be accommodated by site for the Iceland Basin (counterclockwise) and Laschamp
such a model. King et al. (2005) reported a strong correlation (clockwise) excursions (Figures 6 and 9), the two sets of VGP
between rock magnetic variations, color changes, and physical paths pass over similar longitudinal bands. This suggests that a
properties in Pleistocene sediments collected during IODP Leg similar core–mantle boundary structure may have prevailed
302 on the Lomonosov Ridge (Arctic Ocean) and attributed during the two excursions. In addition, the repetitive structure
the observed Arctic paleomagnetic behavior to environmental of the VGP paths for dispersed site locations can be taken as an
indication of a simple, possibly dipolar, geometry for the excursion (McWilliams, 2001). As originally pointed out by
transitional geomagnetic field for both excursions. The VGP Valet and Meynadier (1993), the intensity of the geomagnetic
paths and the intensity records for the two excursions are field is substantially reduced during excursions. The apparent
consistent with a decrease in strength of the axial dipole, a dominance of nonaxial (equatorial) dipolar fields during
substantial transitional equatorial dipole, and a reduced non- excursions implies that the amplitudes of the nondipole com-
dipole field relative to the axial dipole. During the first N ! S ponents may have been relatively reduced during excursions.
part of the Iceland Basin VGP path, over Europe and Africa, the Remarkably, the transitional VGPs appear to follow similar
g11 term of the equatorial dipole appears to be preponderant, paths for the Laschamp and Iceland Basin excursions, albeit
while h11 appears to be dominant during the S ! N part of the with a different sense of motion, and are separated in time by
excursion. For the Laschamp excursion, the opposite is true, 140 ky. This is obviously longer than the time constant asso-
with the first part dominated by h11 and the second part dom- ciated with fluid motions in the outer core and therefore
inated by g11 (Laj et al., 2006). suggest a deep Earth (lower mantle) control on the excursional
The experimental results obtained in recent years partly field geometry (Laj et al., 2006).
support and partly contrast with the simple geometry for the The Blake excursion, however, situated in time between the
transitional field suggested by Laj et al. (2006). For the Laschamp and Iceland Basin excursions, does not appear to
Laschamp excursion, the record obtained by Channell et al. have the same transitional field geometry. For the Blake excur-
(2012) from ODP Site 1063 is consistent with the geometry of sion, there is only a single record (Tric et al., 1991) that
Laj et al. (2006) (Figure 6). In the study of Nowaczyk et al. incorporates enough transitional magnetization directions for
(2012) from the Black Sea, VGPs display a first short loop over the transitional VGP path to be mapped. The VGP path for this
NE America reaching South Florida before the main excur- record lies over the Americas during the N ! S part reaching
sional path that coincides with the trends for the N ! R path high southern latitudes (Figure 7). The return path, although
but not for the R ! N return path, which is here over central much less detailed, is situated over Australia and Southeast
Indian Ocean. Results from volcanic formations in New Asia. The VGP path, albeit based on a single record, is thus
Zealand (Cassata et al., 2008) and the French Massif Central different from that of the Laschamp and Iceland Basin excur-
(Plenier et al., 2007) indicate that most of the VGPs related to sions. For the Pringle Falls excursion, on the other hand, sev-
the Laschamp excursion do not lie close to the clockwise VGP eral records from different sites yield consistent results. As
loop of path. These results were interpreted by the authors to noted by McWilliams (2001), this is an indication that the
show that the transitional field was more complex than sug- transitional field had a large dipolar component during the
gested by Laj et al. (2006). Pringle Falls excursion. Curiously, the path for the Pringle Falls
For the Iceland Basin excursion (Figure 9), Roberts (2008) excursion is similar to that observed for the Blake excursion
suggested that if data from ODP Hole 884 were added, a more from the Mediterranean record of Tric et al. (1991).
complex excursional field behavior would appear. Laj et al.
(2006) did not include the ODP record in their compilation
because of its lack of precise age model and low sedimentation 5.10.7 Concluding Remarks
rate (5 cm ky1). The VGP path for the Iceland Basin excur-
sion obtained recently by Channell et al. (2012) differs from The paleomagnetic records for the two best-documented excur-
the VGP path obtained by Knudsen et al. (2006) from the same sions (Laschamp and Iceland Basin excursions) have started a
ODP Site 1063. This difference can, however, be accounted for debate on the nature of the geometry of the excursional field
by differences in the azimuthal correction of individual (unor- and its importance for understanding of the core–mantle pro-
iented) cores that is facilitated by the record from the study of cesses leading to excursions. Some excursions, notably the
the entire core (Channell et al., 2012), rather than just the Blake excursion (Tric et al., 1991; Zhu et al., 1994) and some
excursional interval (Knudsen et al., 2006). The southward records of the Iceland Basin and Pringle Falls excursions
motion of the VGP lies over Africa according to Channell (Channell, 2006), appear to be characterized by multiple
et al. (2012), similar to the compilation in Figure 9, but the VGP swings to high latitudes. Such field instability may be
poorly defined return path appears to lie over the Americas explained by the observation that the critical Reynolds number
rather than East Asia as in Figure 9. for the onset of core convection is very sensitive to the poloidal
Clearly, there is a need for additional records of these two, field and the strength of core convection varies wildly in
as well as other excursions, to improve the geographic distri- response to changes in magnetic field strength particularly
bution of sites, in particular with a better coverage in the during intensity minima (Zhang and Gubbins, 2000). As dis-
Southern Hemisphere. This better coverage in space and time cussed earlier, this will likely be one of the main aspects of
would test several hypothesis associated with deep Earth future studies of excursions.
dynamics and their role in geomagnetic excursions. For ex- The duration of excursions in the Brunhes Chron (Table 2),
ample, the results of Laj et al. (2006) do not support the view as well as excursions within the Matuyama Chron (Table 4),
that the geomagnetic intensity minima that coincide with based mainly on constraints from oxygen isotope stratigraphy,
directional excursions reflect the emergence of nondipole geo- is usually estimated to be <5 ky. This duration is comparable
magnetic components (e.g., Guyodo and Valet, 1999; Merrill with the 3 ky timescale for diffusive field changes in the
and McFadden, 1994). If nondipole fields were dominant, one Earth’s solid inner core, which must reverse polarity in order
should observe widely different VGP paths at the different sites, for a full geomagnetic reversal to be sustained. The fact that our
which is not the case here, or in an analysis of the Cobb estimate of excursion duration is comparable with the 3 ky
Mountain Subchron (Clement, 1992) or the Pringle Falls time constant for inner-core magnetic diffusion provides
support for the suggestion by Gubbins (1999) that there is a correspondence, we show the ODP Site 983 paleointensity
mechanistic distinction between polarity reversals, which record for the 750–1800 ka interval (Figure 16) with the posi-
define polarity chrons, and excursions. Nevertheless, the sim- tion of polarity reversals and excursions as recorded in the
ilarity of some VGP paths of excursions and reversal transitions same sediment sequence (Channell and Kleiven, 2000;
bounding polarity chrons suggests an inherent link between Channell et al., 2002).
the mechanisms that give rise to geomagnetic excursions and In the next few years, paleomagnetic studies of rapidly
reversals. For instance, the East Asian longitudinal bands iden- deposited sediments will be combined with astrochronological
tified in the excursional VGP paths coincide with one of the age models and coupled with 40Ar/39Ar dating and paleomag-
preferred longitudinal bands for transitional VGP paths during netic studies from volcanic rocks. This will hopefully converge
reversals (Clement, 1991; Laj et al., 1991) and are featured in toward a consensus on the short-term (transitional) behavior
VGP clusters that appear in volcanic and high-resolution sed- of the geomagnetic field. Needless to say, to attain this
imentary records of reversal transitions (e.g., Channell and consensus, a large experimental effort has to be made in future
Lehman, 1997; Channell et al., 2004; Hoffman, 1992). The years, not only to obtain new data but also to develop and
longitudinal band over Western Europe and Africa, on the follow criteria to judge on their reliability.
other hand, is less evident in compilations of sedimentary Criteria for reliability have already been suggested for general
VGP reversal paths as well as from volcanic rocks (Figure 15; paleomagnetic studies (Van der Voo, 1990) and for magnetos-
Valet and Herero-Bervera, 2003). tratigraphic data (Opdyke and Channell, 1996). Some of these
An increasing number of RPI records, based on normalized criteria naturally apply to the study of the excursions; however,
remanence data from marine sediments, are now available for sometimes, unique criteria have to be employed often linked to
the Matuyama Chron (Channell and Kleiven, 2000; Channell the very short duration of paleomagnetic excursions.
et al., 2002, 2008; Dinarès-Turell et al., 2002; Guyodo et al., Among the main points that paleomagnetic studies of
1999, 2001; Hayashida et al., 1999; Horng et al., 2003; Kok excursions must meet, common to other type of studies such
and Tauxe, 1999; Meynadier et al., 1995; Valet and Meynadier, as reversals, are the following:
1993). The sedimentary sequences that have yielded high-
quality RPI records only rarely capture the directional changes The age of the excursion has to be known with great
associated with excursions. Presumably, paleointensity min- accuracy.
ima are captured more readily because the paleointensity Multiple and precisely correctable, geographically distrib-
features are longer-lasting than the accompanying directional uted records must be obtained before any claim for a new
excursions and are perhaps less susceptible to magnetic excursion can be made.
overprinting and the filtering effect of a finite magnetization Complete thermal or alternating field demagnetization
lock-in zone during DRM acquisition. The short duration of should be performed and magnetization components
directional excursions (Tables 2 and 3) result in them being obtained with vector analysis.
rarely recorded in sediments with mean sedimentation rates of A sufficient number of transitional VGP positions should be
less than several centimeters per thousand years. It has often obtained if conclusions about the symmetry of the excur-
been noted that the ages of paleointensity minima in paleoin- sion are investigated.
tensity records correspond to the excursions found elsewhere Magnetic mineralogy should be carefully investigated. In
(e.g., Valet and Meynadier, 1993). As an example of this case of mixed mineralogy, the presence of sulfides with
VGPs
90
60
-30
-30
-60
-90
180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180
Figure 15 Polarity transition VGPs with latitudes lower than 60 obtained from volcanic sites with large global distribution. Adapted from Valet J-P and
Herero-Bervera E (2003) Some characteristics of geomagnetic reversals inferred from detailed volcanic records. Comptes Rendus Geoscience 335:
79–90.
ODP Site 983 (relative paleointensity)
Santa Rosa
Bjorn
Top Jaramillo
Top Olduvai
Kamikatsura
Gardar
Punaruu
MBB
Cobb Mt.
Base Jara.
Gilsa
800 1000 1200 1400 1600 1800
Age (ka)
Figure 16 ODP Site 983 relative paleointensity record (blue) with a 10 ky running mean (black) of the same record. Ages of principal polarity
reversals and excursions are indicated and correspond to paleointensity minima. Data from Channell JET and Kleiven HF (2000) Geomagnetic
palaeointensities and astrochronological ages for the Matuyama–Brunhes boundary and the boundaries of the Jaramillo Subchron: Palaeomagnetic and
oxygen isotope records from ODP Site 983. Philosophical Transactions of the Royal Society of London A 358: 1027–1047; Channell JET, Mazaud A,
Sullivan P, Turner S, and Raymo ME (2002) Geomagnetic excursions and paleointensities in the 0.9–2.15 Ma interval of the Matuyama Chron at
ODP Site 983 and 984 (Iceland Basin). Journal of Geophysical Research 107(B6), http://dx.doi.org/10.1029/2001JB000491.
magnetizations that postdate deposition must be inter- So the effects of translation speed on u-channel measure-
preted with great care, because these delays may create ments suggest that standardization of procedures should be
spurious directional and paleointensity fluctuations, partic- explored and in any case that translation speeds should be
ularly during periods of fast changes of the geomagnetic routinely reported when publishing u-channel data sets.
field, such as during excursions. At present, the majority of documented excursions are in
the Brunhes Chron (Figure 1), the Matuyama Chron
Apart from these general points, studies conducted on marine
(Figure 12), and the Late and Middle Miocene (7–14 Ma)
cores using u-channels beg for special attention that has been
interval (Figure 13). It remains to be seen whether these time
very often neglected. First of all, the bandwidth of the used
intervals are representative of excursion frequency in general or
instrument (linked to the geometry of the pickup coils and the
whether they remain exceptional, as the catalog of excursions
diameter of the access for the u-channels) has to be carefully
expands. It is interesting to note that the high frequency (and
investigated. The final resolution of the record cannot be better
short duration) of excursions over the last 1.5 My, if applied to
than the bandwidth, unless deconvolution programs are applied.
the entire Cenozoic GPTS, would lead to a distribution of
But the most serious point is linked to the speed at which
chron/excursion durations inconsistent with a Poisson distri-
u-channels move through the magnetometer. As shown by
bution (see Lowrie and Kent, 2004), implying that polarity
Brachfeld et al. (2004), the translation speed of the u-channels
excursions are mechanistically distinct from polarity chrons.
has a significant effect on the intensity and on the subsequent AF
The study of polarity excursions, today, is analogous to the
demagnetization. At slow translation speed, a u-channel sample
study of polarity chrons 50 years ago when the principal polarity
experiences a number of alternating field half-cycles comparable
chrons of the last 5 My were in the process of being resolved by
to that experienced by discrete samples in an ordinary stationary
coupled K–Ar and paleomagnetic studies in volcanic rocks and
experiment. For instance, ARM acquisition at 1 cm s1 gives
by studies of conventional piston cores from the oceans. The
the maximum remanence and hardest demagnetization spec-
challenge of resolving brief excursions, which have remained
trum. Faster translation speeds result in an incomplete ARM
obscure and poorly documented until very recently, has only
that will be artificially soft upon magnetization.
just begun.
These effects are nonnegligible and can complicate the use
of u-channel derived magnetic parameters as quantitative envi-
ronmental proxies. As an example, ratios ARM/SIRM and sus-
ceptibility of ARM normalized by magnetic susceptibility often Acknowledgments
used to infer particle size interpretations (Banerjee et al., 1981;
Channell et al., 2013; King et al., 1982; Stoner et al., 2007; The authors thank M. McWilliams and J.-P. Valet for providing
Yamazaki and Ikehara, 2012) may cause errors if not correctly Figures 10 and 14, respectively, and N. Nowaczyk and H. Oda
measured in u-channel studies. for providing numerical data.
We also thank C. Kissel for her invaluable help in the writing of Boundary in the Marche–Umbria Basin (Italy), vol. 4, pp. 111–117. Ancona, Italy:
this update of the manuscript. International Subcommission on Paleogene Stratigraphy (Special Publication II).
Billups K, Pälike H, Channell JET, Zachos J, and Shackleton NJ (2004) Astronomic
calibration of the late Oligocene through Early Miocene geomagnetic polarity time
scale. Earth and Planetary Science Letters 224: 33–44.
Biswas DK, Hyodo M, Taniguchi Y, et al. (1999) Magnetostratigraphy of Plio–
References Pleistocene sediments in a 1700-m core from Osaka Bay, southwestern Japan and
short geomagnetic events in the middle Matuyama and early Brunhes chrons.
Abdel-Monem A, Watkins ND, and Gast PW (1972) Potassium–argon ages, volcanic Palaeogeography, Palaeoclimatology, Palaeoecology 148: 233–248.
stratigraphy, and geomagnetic polarity history of the Canary Islands: Tenerife, La Blakely RJ (1974) Geomagnetic reversals and crustal spreading rates during the
Palma and Hierro. American Journal of Science 272: 805–825. Miocene. Journal of Geophysical Research 79: 2979–2985.
Abdul-Aziz H, Hilgen F, Krijgsman W, Sanz E, and Calvo JP (2000) Astronomical Blakely RJ and Cox A (1972) Evidence for short geomagnetic polarity intervals in the
forcing of sedimentary cycles in the middle to late Miocene continental Calatayud Early Cenozoic. Journal of Geophysical Research 35: 7065–7072.
Basin (NE Spain). Earth and Planetary Science Letters 177: 9–22. Blanchet C, Thouveny N, and de Garidel-Thoron T (2006) Evidence for multiple
Abdul-Aziz H, Krijgsman W, Hilgen FJ, Wilson DS, and Calvo JP (2003) An paleomagnetic intensity lows between 30 and 50 ka BP from a western Equatorial
astronomical polarity timescale for the late middle Miocene based on cyclic Pacific sedimentary sequence. Quaternary Science Reviews 25(9–10): 1039–1052.
continental sequences. Journal of Geophysical Research 108: 2159. http://dx.doi. Bleil U and Gard G (1989) Chronology and correlation of quaternary
org/10.1029/2002JB001818. magnetostratigraphy and nannofossil biostratigraphy in Norwegian–Greenland sea
Abdul-Aziz H and Langereis CG (2004) Astronomical tuning and duration of three new sediments. Geologische Rundschau 78: 1173–1187.
subchrons (C5r.2r-1n, C5r.2r-2n and C5r.3r-1n) recorded in a middle Miocene Böhnel H, Reismann N, Jager G, Haverkamp U, Negendank JFW, and Schminke HU
continental sequence in SE Spain. In: Channell JET, et al. (ed.) Timescales of the (1987) Paleomagnetic investigation of Quaternary West Eifel volcanics
Paleomagnetic Field. AGU Geophysical Monograph, vol. 145, pp. 141–160. (Germany)—Indication for increased volcanic activity during geomagnetic
Washington, DC: AGU. excursion event. Journal of Geophysics 62(1): 50–61.
Abrahamsen N and Sager W (1994) Cobb mountain geomagnetic polarity event and Bonhommet N and Babkine J (1967) Sur la présence d’aimentation inverse dans la
transitions in three deep-sea sediment cores from the Lau Basin. In: Hawkins J, Chaı̂ne des Puys. Comptes Rendus Hebdomadaires des Seances de l’Académie des
Parson L, and Allan J, et al. (eds.) Proceedings of ODP Science Results, 135, Sciences Series B 264: 92–94.
pp. 737–762. College Station, TX: Ocean Drilling Program. Bonhommet N and Zahringer J (1969) Paleomagnetism and potassium–argon
Acton G, Guyodo Y, and Brachfield S (2006) The nature of a cryptochron from a determinations of the Laschamp geomagnetic polarity event. Earth and Planetary
paleomagnetic study of chron C4r.2r recorded in sediments off the Antarctic Science Letters 6: 43–46.
Peninsula. Physics of the Earth and Planetary Interiors 156: 213–222. Bourne M, Mac Niocaill C, Thomas AL, Knudsen MF, and Henderson GM (2012) Rapid
An ZS, Kukla GJ, Porter SC, and Xiao J (1991) Magnetic susceptibility evidence of directional changes associated with a 6.5 kyr-long Blake geomagnetic excursion at
monsoon variation on the loess plateau of central China during the last the Blake–Bahama Outer Ridge. Earth and Planetary Science Letters 333: 31–34.
130,000 years. Quaternary Research 36: 29–36. Bowers NE, Cande S, Gee J, Hildebrand JA, and Parker RL (2001) Fluctuations of the
Andersen KK, Svensson A, and Johnsen SJ (2006) The Greenland ice core chronology paleomagnetic field during chron C5 as recorded in near bottom marine magnetic
2005, 15–42 ka. Part 1: Constructing the time scale. Quaternary Science Reviews anomaly data. Journal of Geophysical Research 106: 26379–26396.
25: 3246–3257. Bowles J, Tauxe L, Gee J, McMillan D, and Cande S (2003) Source of tiny wiggles in
Anonymous (1977) Magnetostratigraphic polarity units. A supplementary chapter of the Chron 5n: A comparison of sedimentary relative paleointensity and marine magnetic
international subcommission on stratigraphic classification international anomalies. Geochemistry, Geophysics, Geosystems 4: 6. http://dx.doi.org/
stratigraphic guide. Geology 7: 578–583. 10.1029/2002GC000489.
Baksi AK (1994) Concordant sea-floor spreading rates obtained from geochronology, Brachfeld S, Kissel C, Laj C, and Mazaud A (2004) Behavior of u-channels during
astrochronology and space geodesy. Geophysical Research Letters 21: 133–136. the acquisition and demagnetization of remanence: Implications for paleomagnetic
Baksi AK and Hoffman KA (2000) On the age and morphology of the Réunion event. and rock magnetic measurements. Physics of the Earth and Planetary Interiors
Geophysical Research Letters 27: 2297–3000. 145: 1–8.
Baksi AK, Hoffman KA, and McWilliams M (1993) Testing the accuracy of the Broecker WS, Thurber DL, Goddard J, Ku T, Matthews RK, and Mesolella KJ (1968)
geomagnetic polarity time-scale (GPTS) at 0 5 Ma utilizing 40Ar/39Ar Milankovitch hypothesis supported by precise dating of coral reefs and deep-sea
incremental heating data on wholerock basalts. Earth and Planetary Science Letters sediments. Science 159: 297–300.
118: 135–144. Broecker WS and Van Donk J (1970) Insolation changes, ice volumes, and the 18O
Banerjee SK, King JW, and Marvin JA (1981) A rapid method for magnetic granulometry record in deep-sea cores. Reviews of Geophysics and Space Physics 8: 169–197.
with applications to environmental studies. Geophysical Research Letters Brunhes B (1906) Recherches sur la direction de l’aimantation des roches volcaniques.
8: 333–336. Journal de Physique 5: 705–724.
Barbetti M and Flude K (1979) Paleomagnetic field strengths from sediments baked by Burek PJ (1967) Korrelation revers magnatisierter Gesteinfolgen in Oberen
lava flows of the Chaı̂ne des Puys, France. Nature 278: 153–156. Bandsandstein S.W. Deutschland. Geologisches Jahrbuch 84: 591–616.
Barbetti M and McElhinny MW (1972) Evidence of a geomagnetic excursion 30,000 yr Burek PJ (1970) Magnetic reversals: Their application to stratigraphic problems.
BP. Nature 239: 327–330. American Association of Petroleum Geologists Bulletin 54: 1120–1139.
Baumgartner S, Beer J, Masarik J, Wagner G, Meynadier L, and Synal H-A (1998) Cande SC and Kent DV (1992a) A new geomagnetic polarity timescale for the late
Geomagnetic modulation of the 36Cl flux in the GRIP Ice Core, Greenland. Science Cretaceous and Cenozoic. Journal of Geophysical Research 97: 13917–13951.
279: 1330–1332. Cande SC and Kent DV (1992b) Ultrahigh resolution marine magnetic anomaly profiles:
Baumgartner S, Beer J, Suter M, et al. (1997) 36Cl fallout in the Summit Greenland Ice A record of continuous paleointensity variations? Journal of Geophysical Research
Core Project ice core. Journal of Geophysical Research 102: 26659–26662. 97: 15075–15083.
Benson LV, Currey DR, Dorn RI, et al. (1990) Chronology of expansion and contraction Cande SC and Kent DV (1995) Revised calibration of the geomagnetic polarity timescale
of four Great Basin lake systems during the past 35,000 years. Palaeogeography, for the late Cretaceous and Cenozoic. Journal of Geophysical Research
Palaeoclimatology, Palaeoecology 78: 241–286. 100: 6093–6095.
Benson L, Liddicoat J, Smoot J, Sarna-Wojcicki A, Negrini R, and Lund S (2003) Age of Cande SC and LaBrecque JL (1974) Behaviour of the Earth’s paleomagnetic field from
the Mono Lake excursion and associated tephra. Quaternary Science Reviews small scale marine magnetic anomalies. Nature 247: 26–28.
22: 135–140. Carracedo JC, Rodrı́guez Badiola E, Guillou H, et al. (2007) Eruptive and structural
Benson LV, Lund SP, Burdett JW, et al. (1998) Correlation of late Pleistocene lake-level history of Teide volcano and rift zones of Tenerife, Canary Islands. Geological
oscillations in Mono Lake, California, with North Atlantic climate events. Quaternary Society of America Bulletin 119: 1027–1051.
Research 49: 1–10. Cassata WS, Singer BS, and Cassidy J (2008) Laschamp and Mono Lake geomagnetic
Berggren WA, Kent DV, Flynn JJ, and Vancouvering JA (1985) Cenozoic excursions recorded in New Zealand. Earth and Planetary Science Letters
geochronology. Geological Society of America Bulletin 96(11): 1407–1418. 268: 76–88.
Bice DM and Montanari A (1988) Magnetic stratigraphy of the Massignano section Cassata WS, Singer BS, Liddicoat JC, and Coe RS (2010) Reconciling discrepant
across the Eocene/Oligocene boundary in the Marche–Umbria basin (Italy). chronologies for the geomagnetic excursion in the Mono Basin, California: Insights
In: Premoli-Silva I, Coccioni I, and Montanari A (eds.) The Eocene/Oligocene from new 40Ar/39Ar dating experiments and a revised relative paleointensity
correlation. Quaternary Geochronology 5: 533–543. http://dx.doi.org/10.1016/j. Channell JET, Mazaud A, Sullivan P, Turner S, and Raymo ME (2002) Geomagnetic
quageo.2010.02.001. excursions and paleointensities in the 0.9–2.15 Ma interval of the Matuyama
Cassidy J and Hill MJ (2009) Absolute palaeointensity study of the Mono Lake Chron at ODP Site 983 and 984 (Iceland Basin). Journal of Geophysical Research
excursion recorded by New Zealand basalts. Physics of the Earth and Planetary 107(B6).
Interiors 172: 225–234. Channell JET and Raymo ME (2003) Paleomagnetic record at ODP Site 980 (Feni Drift,
Chamalaun FH and McDougall I (1966) Dating geomagnetic polarity episodes in Rockall) for the past 1.2 Myrs. Geochemistry, Geophysics, Geosystems 4(4): 1033.
Réunion. Nature 210: 1212–1214. http://dx.doi.org/10.1029/2002GC000440.
Champion DE, Dalrymple GB, and Kuntz MA (1981) Radiometric and paleomagnetic Channell JET, Stoner JS, Hoddell DA, and Charles C (2000) Geomagnetic
evidence for the emperor reversed polarity event at 0.46 0.005 Myr in basalt lava paleointensity for the last 100 kyr from the subantarctic south Atlantic: A tool for
flows from the eastern Snake River plain, Idaho. Geophysical Research Letters interhemispheric correlation. Earth and Planetary Science Letters 175: 145–160.
8: 1055–1058. Channell JET and Xuan C (2009) Self-reversal and apparent magnetic excursions in
Champion DE, Lanphere MA, and Kuntz MA (1988) Evidence for a new geomagnetic Arctic sediments. Earth and Planetary Science Letters 284: 124–131.
reversal from lava flows in Idaho: Discussion of short polarity reversals in the Channell JTT, Xuan C, and Hodell DA (2009) Stacking paleointensity and oxygen
Brunhes and late Matuyama polarity chrons. Journal of Geophysical Research isotope data for the last 1.5 Myr (PISO-1500). Earth and Planetary Science Letters
93: 11667–11680. 283: 14–23.
Channell JET (1999) Geomagnetic paleointensity and directional secular variation at Charbit S, Guillou H, and Turpin L (1998) Cross calibration of K–Ar standard
Ocean Drilling Program (ODP) Site 984 (Bjorn Drift) since 500 ka: Comparisons minerals using an unspiked Ar measurement technique. Chemical Geology
with ODP Site 983 (Gardar Drift). Journal of Geophysical Research 150: 147–159.
104: 22937–22951. Chauvin A, Duncan RA, Bonhommet N, and Levi S (1989) Paleointensity of the Earth’s
Channell JET (2006) Late Brunhes polarity excursions (Mono Lake, Laschamp, Iceland magnetic field and K–Ar dating of the Louchadière volcanic flow (central France):
Basin and Pringle Falls) recorded at ODP Site 919 (Irminger Basin). Earth and New evidence for the Laschamp excursion. Geophysical Research Letters
Channell JET, Bralower TJ, and Grandesso P (1987) Biostratigraphic correlation of Chauvin A, Roperch P, and Duncan RA (1990) Records of geomagnetic reversals from
Mesozoic polarity chrons (M1 to CM23) at Capriolo and Xausa (Southern Alps, volcanic islands of French Polynesia, 2-paleomagnetic study of a flow sequence
Italy). Earth and Planetary Science Letters 85: 203–321. (1.2 to 0.6 Ma) from the island of Tahiti and discussion of reversal models. Journal
Channell JET, Cecca F, and Erba E (1995a) Correlations of Hauterivian and Barremian of Geophysical Research 95: 2727–2752.
(early Cretaceous) stage boundaries to polarity chrons. Earth and Planetary Science Chen Y, Smith PE, Evensen NM, York D, and Lajoie KR (1996) The edge of time: Dating
Letters 134: 125–140. young volcanic ash layers with the 40Ar/39Ar laser probe. Science
Channell JET, Curtis JH, and Flower BP (2004) The Matuyama–Brunhes boundary 274: 1176–1178.
interval (500–900 ka) in north Atlantic drift sediments. Geophysical Journal Clement BM (1991) Geographical distribution of transitional V.G.P.’s: Evidence for non-
International 158: 489–505. zonal equatorial symmetry during the Matuyama–Brunhes geomagnetic reversal.
Channell JET, Erba E, Nakanishi M, and Tamaki K (1995b) A late Jurassic–early Earth and Planetary Science Letters 104: 48–58.
Cretaceous timescale and oceanic magnetic anomaly block models. In: Berggren W, Clement BM (1992) Evidence for dipolar fields during the Cobb mountain geomagnetic
Kent D, Aubry M, and Hardenbol J (eds.) Geochronology, Timescales, and polarity reversals. Nature 358: 405–407.
Stratigraphic Correlation, pp. 51–64. Tulsa, OK: Society of Economic Clement BM (2004) Dependence of the duration of geomagnetic polarity reversals on
Paleontologists and Mineralogists, SEPM Special Publication 54. site latitude. Nature 428: 637–640.
Channell JET, Galeotti S, Martin EE, Billups K, Scher H, and Stoner JS (2003a) Eocene Clement BM and Kent DV (1987) Short polarity intervals within the Matuyama:
to Miocene magnetostratigraphy, biostratigraphy, and chemostratigraphy at ODP Transition field records from hydraulic piston cored sediments from the North
Site 1090 (sub-Antarctic south Atlantic). Geological Society of America Bulletin Atlantic. Earth and Planetary Science Letters 81: 253–264.
115: 607–623. Clement BM and Martinson DG (1992) A quantitative comparison of two paleomagnetic
Channell JET and Guyodo Y (2004a) The Matuyama Chronozone at ODP Site 982 records of the Cobb Mountain subchron from North Atlantic deep-sea sediments.
(Rockall Bank): Evidence for decimeterscale magnetization lock-in depths. Journal of Geophysical Research 97: 1735–1752.
In: Channell JET, et al. (ed.) Timescales of the Geomagnetic Field. AGU Geophysical Clement BM and Robinson R (1987) The magnetostratigraphy of Leg 94 sediments.
Monograph Seminar, vol. 145, pp. 205–219. Washington, DC: AGU. In: Ruddiman W, Kidd RB, and Thomas E, et al. (eds.) Deep Sea Drilling Project
Channell JET and Guyodo Y (2004b) The Matuyama Chronozone at ODP site 982 Initial Reports, vol. 94, pp. 635–650. Washington: U.S. Government Printing Office.
(Rockall Bank): Evidence for decimeter-scale magnetization lock-in depths. Coe RS, Singer BS, Pringle MS, and Zhao X (2004) Matuyama–Brunhes reversal and
Timescales of the Geomagnetic Field. AGU Geophysical Monograph. vol. 145, pp. Kamikatsura event on Maui: Paleomagnetic directions, 40Ar/39Ar ages and
205–219. Washington DC: AGU. implications. Earth and Planetary Science Letters 222: 667–684.
Channell JET, Hodell DA, and Curtis JH (2012) ODP Site 1063 (Bermuda Rise) Condomines M (1978) Age of the Olby–Laschamp geomagnetic polarity event. Nature
revisited: Oxygen isotopes, excursions and paleointensity in the Brunhes Chron. 286: 697–699.
Geochemistry, Geophysics, Geosystems 13: Q02001. http://dx.doi.org/ Cox A, Doell RR, and Dalrymple GB (1963) Geomagnetic polarity epochs and
10.1029/2011GC003897. Pleistocene geochronometry. Nature 198: 1049–1051.
Channell JET, Hodell DA, and Lehman B (1997) Relative geomagnetic paleointensity Cox SE, Farley KA, and Hemming SR (2012) Insights into the age of the Mono Lake
and 18O at ODP Site 983 (Gardar Drift, north Atlantic) since 350 ka. Earth and excursion and magmatic crystal residence time from (U–Th)/He and Th-230 dating
Planetary Science Letters 153: 103–118. of volcanic allanite. Earth and Planetary Science Letters 319: 178–184.
Channell JET, Hodell DA, Margari V, Skinner LC, Tzedakis PC, and Kesler MS (2013) Creer KM, Irving E, and Runcorn SK (1954) The direction of the geomagnetic field in
Biogenic magnetite, detrital hematite, and relative paleointensity in Quaternary remote epochs in Great Britain. Journal of Geomagnetism and Geoelectricity
sediments from the Southwest Iberian Margin. Earth and Planetary Science Letters 6: 164–168.
376: 99–109. Creer KM, Readman PW, and Jacobs AM (1980) Paleomagnetic and paleontological
Channell JET, Hodell DA, Xuang C, Mazaud A, and Stoner JS (2008) Age calibrated dating of a section at Gioia Tauro, Italy: Identification of the Blake Event. Earth and
relative paleointensity for the last 1.5 Myr from IODP Site U1308 (North Atlantic). Planetary Science Letters 50: 289–300.
Earth and Planetary Science Letters 274: 59–71. http://dx.doi.org/10.1016/j. Creer KM, Thouveny N, and Blunk I (1990) Climatic and geomagnetic influences on the
epsl.2008.07.005. Lac du Bouchet palaeomagnetic SV record through the last 110 000 years. Physics
Channell JET and Kleiven HF (2000) Geomagnetic palaeointensities and of the Earth and Planetary Interiors 64: 314–341.
astrochronological ages for the Matuyama–Brunhes boundary and the boundaries Dalrymple GB and Lanphere MA (1969) Potassium–Argon Dating: Principles,
of the Jaramillo Subchron: Palaeomagnetic and oxygen isotope records from ODP Techniques and Applications to Geochronology. San Francisco, CA: W. H. Freeman
Site 983. Philosophical Transactions of the Royal Society of London A and Co, 258 pp.
358: 1027–1047. Denham CR (1974) Counterclockwise motion of paleomagnetic directions 24,000 years
Channell JET, Labs J, and Raymo ME (2003b) The Réunion subchronozone at ago at Mono Lake, California. Journal of Geomagnetism and Geoelectricity 26: 487.
ODP Site 981 (Feni Drift, North Atlantic). Earth and Planetary Science Letters Denham CR (1976) Blake polarity episode in two cores from the Greater Antilles outer
215: 1–12. ridge. Earth and Planetary Science Letters 29: 422–443.
Channell JET and Lehman B (1997) The last two geomagnetic polarity reversals Denham CR and Cox A (1971) Evidence that the Laschamp polarity event did not occur
recorded in high-deposition-rate sediment drifts. Nature 389: 712–715. 13,300–34,000 years ago. Earth and Planetary Science Letters 13: 181–190.
Dinarès-Turell J, Sagnotti L, and Roberts AP (2002) Relative geomagnetic paleointensity Guyodo Y, Richter C, and Valet J-P (1999) Paleointensity record from Pleistocene
from the Jaramillo Subchron to the Matuyama/Brunhes boundary as recorded in a sediments (1.4–0 Ma) off the California margin. Journal of Geophysical Research
Mediterranean piston core. Earth and Planetary Science Letters 194: 327–341. 104: 22953–22964.
Doell RR and Dalrymple GB (1966) Geomagnetic polarity epochs: A new polarity event Guyodo Y and Valet J-P (1996) Relative variations in geomagnetic intensity from
and the age of the Brunhes–Matuyama boundary. Science 152: 1060–1061. sedimentary records: The past 200,000 years. Earth and Planetary Science Letters
Doell RR, Dalrymple GB, and Cox A (1966) Geomagnetic polarity epochs—Sierra 143: 23–36.
Nevada data. Journal of Geophysical Research 71: 531–541. Guyodo Y and Valet J-P (1999) Global changes in intensity of the Earth’s magnetic field
Doell RR, Dalrymple GB, Smith RL, and Bailey RA (1968) Paleomagnetism potassium- during the past 800 kyr. Nature 399: 249–252.
argon ages and geology of rhyolites and associated rocks of the Valles caldera, New Hall CM and York D (1978) K–Ar and 40Ar/39Ar age of the Laschamp geomagnetic
Mexico. Geological Society of America Bulletin 116: 211–248. polarity reversal. Nature 274: 462–464.
Ericson DB, Ewing M, Wollin G, and Heezen BC (1961) Atlantic deep-sea sediment Hall CM, York D, and Bonhommet N (1979) 40Ar/39Ar age of the Laschamp event and
cores. Geological Society of America Bulletin 72: 193–206. associated volcanism in the Chaı̂ne des Puys. Eos, Transactions American
Evans HF and Channell JET (2003) Late Miocene magnetic stratigraphy from ODP Site Geophysical Union 60: 244.
1092 (sub-Antarctic south Atlantic): Recognition of ‘cryptochrons’ in C5n.2n. Harland WB, Armstrong R, Cox AV, Craig L, Smith AG, and Smith D (1990) A Geologic
Geophysical Journal International 153: 483–496. Time Scale 1989. Cambridge, NY: Cambridge University Press, 263 pp.
Evans HF, Channell JET, Stoner JS, et al. (2007) Paleointensity-assisted Harland WB, Cox AV, Llewellyn PG, Pickton CAG, Smith AG, and Walters RA (1982)
chronostratigraphy of detrital layers on the Eirik Drift (North Atlantic) since marine Geologic Time Scale. Cambridge, NY: Cambridge Univ. Press, 131 pp.
isotope stage 11. Geochemistry, Geophysics, Geosystems 8: Q11007. http://dx.doi. Hartl P and Tauxe L (1996) A precursor to the Matuyama/Brunhes transition-field
org/10.1029/2007GC001720. instability as recorded in pelagic sediments. Earth and Planetary Science Letters
Evans HF, Westerhold T, and Channell JET (2004) ODP Site 1092: Revised composite 138: 121–135.
depth section has implications for upper Miocene ‘cryptochrons’. Geophysical Hayashida A, Verosub KL, Heider F, and Leonhardt R (1999) Magnetostratigraphy and
Journal International 156: 195–199. relative paleointensity of late Neogene sediments at ODP Leg 167 Site 1010 off Baja
Fang X, Li J, Van der Voo R, et al. (1997) A record of the Blake event during the last California. Geophysical Journal International 139: 829–840.
interglacial paleosol in the western loess plateau of China. Earth and Planetary Hays JD and Opdyke ND (1967) Antarctic radiolaria, magnetic reversals, and climatic
Science Letters 146: 73–82. change. Science 158: 1001–1011.
Ferk A and Leonhardt R (2009) The Laschamp geomagnetic field excursion recorded in Hays JD, Saito T, Opdyke ND, and Burckle LH (1969) Pliocene and Pleistocene
Icelandic lavas. Physics of the Earth and Planetary Interiors 177(1–2): 19–30. sediments of the equatorial Pacific: Their paleomagnetic, biostratigraphic and
Frank M, Schwarz B, Baumann S, Kubik P, Suter M, and Mangini A (1997) A 200 kyr climatic record. Geological Society of America Bulletin 80: 1481–1514.
record of cosmogenic radionuclide production rate and geomagnetic field intensity Heirtzler JR, Dickson GO, Em H, Pittman WC III, and LePichon X (1968) Marine
from 10Be in globally stacked deep-sea sediments. Earth and Planetary Science magnetic anomalies, geomagnetic field reversal and motions of the ocean floor and
Letters 149: 121–129. continents. Journal of Geophysical Research 73: 2119–2136.
Gallet Y, Genevey A, Le Goff M, Warme N, Gran-Aymerich J, and Lefevre A (2009) On Heizler MT, Perry FV, Crowe BM, Peters L, and Appelt R (1999) The age of Lathrop
the use of archeology in geomagnetism, and vice-versa: Recent developments in Wells Volcanic Center: An 40Ar/39Ar dating investigation. Journal of Geophysical
archeomagnetism. Comptes Rendus Physique 10(7): 630–648. http://dx.doi.org/ Research 104: 767–804.
10.1016/j.crhy.2009.08.005. Heller F (1980) Self-reversal of natural remanent magnetization in the Olby–Laschamp
Garces M, Agusti J, Cabera L, and Pares JM (1996) Magnetostratigraphy of the lavas. Nature 284: 334–335.
Vallesian (late Miocene) in the Valles-Penedes Basin (northeast Spain). Earth and Heller F and Petersen N (1982) The Laschamp excursion. Philosophical Transactions of
Planetary Science Letters 142: 381–396. the Royal Society A 306: 169–177.
Geissman JW, Brown L, Turrin BD, McFadden LD, and Harlan SS (1990) Brunhes chron Helsley CE (1969) Magnetic reversal stratigraphy of the lower Triassic Moenkope
excursion/polarity episode recorded during the late Pleistocene, Albuquerque formation of western Colorado. Geological Society of America Bulletin
volcanoes, New Mexico, USA. Geophysical Journal International 102: 73–88. 80: 2431–2450.
Gilder S, Chen Y, Cogné J-P, et al. (2003) Paleomagnetism of upper Jurassic to lower Herbert TD, D’Hondt SL, Premoli-Silva I, Erba E, and Fisher AG (1995) Orbital
Cretaceous volcanic and sedimentary rocks from the western Tarim Basin and chronology of Cretaceous-Early Paleocene marine sediments. In: Berggren WA,
implications for inclination shallowing and absolute dating of the M-0 (ISEA) chron. Kent DV, Aubry M, and Hardenbol J (eds.) Geochronology, Time Scales and Global
Earth and Planetary Science Letters 206: 587–600. Stratigraphic Correlation, vol. 54, pp. 81–94. Tulsa: Society for Economic
Gillot PG, Labeyrie J, Laj C, et al. (1979) Age of the Laschamp paleomagnetic excursion Palaeontologists and Mineralogists (SEPM), Special Publication.
revisited. Earth and Planetary Science Letters 42: 444–450. Herrero-Bervera E and Helsley CE (1993) Global paleomagnetic correlation of the Blake
Gradstein F, Ogg JG, and Smith A (2004) A Geologic Time Scale. Cambridge, NY: geomagnetic polarity episode. In: Aissaoui DM, McNeill DF, and Hurley NF (eds.)
Cambridge University Press, 589 pp. Applications of Paleomagnetism to Sedimentary Geology, pp. 71–82. Tulsa: Society
Grommé CS and Hay RL (1963) Magnetization of basalt of Bed I, Olduvai Gorge. Nature of Economic Paleontologists and Mineralogists, SEPM Special Publication 49.
200: 560–561. Herrero-Bervera E, Helsley CE, Hammond SR, and Chitwood LA (1989) A possible
Grommé CS and Hay RL (1971) Geomagnetic polarity epochs: Age and duration of the lacustrine paleomagnetic record of the Blake episode from Pringle falls, Oregon,
Olduvai normal polarity event. Earth and Planetary Science Letters 18: 179–185. USA. Physics of the Earth and Planetary Interiors 56: 112–123.
Grootes PM and Stuiver M (1997) Oxygen 18/16 variability in Greenland snow and ice Herrero-Bervera E, Helsley CE, Sarna-Wojcicki AM, et al. (1994) Age and correlation of
with 103 to 105 year time resolution. Journal of Geophysical Research a paleomagnetic episode in the western United States by 40Ar/39Ar dating and
102: 26455–26470. tephrochronology: The Jamaica, Blake, or a new polarity episode? Journal of
Grützner J, Giosan L, Franz SO, et al. (2002) Astronomical age models for Pleistocene Geophysical Research 99: 24091–24103.
drift sediments from the western north Atlantic (ODP Sites 1055–1063). Marine Heslop D, Langereis CG, and Dekkers MJ (2000) A new astronomical timescale for the
Geology 189: 5–23. loess deposits of Northern China. Earth and Planetary Science Letters 184: 125–139.
Gubbins D (1999) The distinction between geomagnetic excursions and reversals. Hilgen FJ (1991a) Extension of the astronomically calibrated (polarity) time scale to the
Geophysical Journal International 137: F1–F3. Miocene/Pliocene boundary. Earth and Planetary Science Letters 107: 349–368.
Guillou H, Carracedo JC, and Day S (1998) Dating of the upper Pleistocene–Holocene Hilgen FJ (1991b) Astronomical calibration of Gauss to Matuyama sapropels in the
volcanic activity of La Palma using the unspiked K–Ar technique. Journal of Mediterranean and implications for the Geomagnetic Polarity Time Scale. Earth and
Volcanology and Geothermal Research 86: 137–149. Planetary Science Letters 104: 226–244.
Guillou H, Singer BS, Laj C, Kissel C, Scaillet S, and Jicha B (2004) On the age of Hilgen FJ, Krijgsman W, Langereis CG, Lourens LJ, Santarelli A, and Zachariasse WJ
the Laschamp geomagnetic event. Earth and Planetary Science Letters (1995) Extending the astronomical (polarity) time scale into the Miocene. Earth and
227: 331–343. Planetary Science Letters 136: 495–510.
Guo B, Zhu R, Florindo F, Ding Z, and Sun JA (2002) A short, reverse polarity interval Hilgen FJ, Krijgsman W, Raffi I, Turco E, and Zachariasse WJ (2000) Integrated
within the Jaramillo subchron: Evidence from the Jingbian section, northern China stratigraphy and astronomical calibration of the Serravallian/Tortonian boundary
Loess plateau. Journal of Geophysical Research 107: 10029–10040. section at Monte Gibliscemi (Sicily, Italy). Marine Micropaleontology 38: 181–211.
Guyodo Y, Acton GD, Brachfeld S, and Channell JET (2001) A sedimentary Hoare JM, Condon WH, Cox A, and Dalrymple GB (1968) Geology, paleomagnetism,
paleomagnetic record of the Matuyama chron from the western Antarctic margin. and potassium-argon ages of basalts from Nunivak Island, Alaska. Geological
Earth and Planetary Science Letters 191: 61–74. Society of America Memoirs 116: 377–413.
Hoffman KA (1992) Dipolar reversal states of the geomagnetic field and core–mantle Kissel C, Laj C, Piotrowski AM, Goldstein SL, and Hemming SR (2008) Millennial-scale
dynamics. Nature 359: 789–794. propagation of Atlantic deep waters to the glacial southern ocean. Paleoceanography
Horng CS, Lee MY, Pälike H, et al. (2002) Astronomically calibrated ages for 23: PA2102. http://dx.doi.org/10.1029/2008PA001624.
geomagnetic reversals within the Matuyama Chron. Earth, Planets and Space Knudsen MF, MacNiocaill C, and Henderson GM (2006) High resolution data of the
54: 679–690. Iceland Basin geomagnetic excursion from ODP sites 1063 and 983: Existence of
Horng CS, Roberts AP, and Liang WT (2003) Astronomically tuned record of relative intense flux patches during the excursion? Earth and Planetary Science Letters
geomagnetic paleointensity from the western Philippine Sea. Journal of Geophysical 251: 18–32.
Research 108: 2059. http://dx.doi.org/10.1029/2001JB001698. Kok YS and Tauxe L (1999) A relative geomagnetic paleointensity stack from Ontong–
Hospers J (1951) Remanent magnetization of rocks and the history of the geomagnetic Java plateau sediments for the Matuyama. Journal of Geophysical Research
field. Nature 168: 1111–1112. 104: 25401–25413.
Hospers J (1953–1954) Reversals of the main geomagnetic field I, II, and III. Krijgsman W, Hilgen FJ, Langereis CG, Santarelli A, and Zachariasse WJ (1995) Late
Proceedings of the Koninklijke Nederlandse Akademie van Wetenschappen-Series B Miocene magnetostratigraphy, biostratigraphy and cyclostratigraphy in the
56: 467–491, 57, 112–121. Mediterranean. Earth and Planetary Science Letters 136: 475–494.
Houghton BF, Wilson CJN, McWilliams MO, et al. (1995) Chronology and dynamics of Krijgsman W, Hilgen FJ, Langereis CG, and Zachariasse WJ (1994) The age of the
a large silicic magmatic system: Central Taupo volcanic zone, New Zealand. Tortonian/Messinian boundary. Earth and Planetary Science Letters 121: 533–547.
Geology 23: 13–16. Krijgsman W and Kent DV (2004) Non-uniform occurrence of short-term polarity
Hsu V, Merrill DL, and Shibuya H (1990) Paleomagnetic transition records of the Cobb fluctuation in the geomagnetic field? New results from middle to late Miocene
mountain event from sediment of the Celebes and Sulu seas. Geophysical Research sediments of the north Atlantic (DSDP Site 608). In: Channell JET, et al. (ed.)
Letters 17: 2069–2072. Timescales of the Geomagnetic Field. AGU Geophysical Monograph Series,
Huxtable J, Aitken MJ, and Bonhommet N (1978) Thermoluminescence dating of vol. 145, pp. 161–174. Washington, DC: AGU.
sediments baked by lava flows of the Chaı̂ne des Puys. Nature 275: 207–209. Kristjansson L and Gudmundsson A (1980) Geomagnetic excursions in late-glacial
Irving E and Runcorn SK (1957) Analysis of the palaeomagnetism of the Torridonian basalt outcrop in south-western Iceland. Geophysical Research Letters 7: 337–340.
sandstone series of north-west Scotland. Philosophical Transactions of the Royal LaBrecque JL, Kent DV, and Cande SC (1977) Revised magnetic polarity time-scale for
Society of London A 250: 83–99. the late Cretaceous and Cenozoic time. Geology 5: 330–335.
Izett GA and Obradovich JD (1994) 40Ar/39Ar age constraints for the Jaramillo normal Laj C, Kissel C, and Beer J (2004) High resolution global paleointensity stack since
subchron and the Matuyama–Brunhes geomagnetic boundary. Journal of 75 kyrs (GLOPIS-75) calibrated to absolute values. Timescales of the Geomagnetic
Geophysical Research 99: 2925–2934. Field. AGU Monograph, vol. 145, pp. 255–265. Washington, DC: AGU.
Jicha BR, Kristjansson L, Brown MC, Singer BS, Beard BL, and Johnson CM (2011) Laj C, Kissel C, Davies C, and Gubbins D (2011) Geomagnetic field intensity and
New age for the Skalamaelifell excursion and identification of a global geomagnetic inclination records from Hawaii and the Réunion Island: Geomagnetic implications.
event in the late Brunhes chron. Earth and Planetary Science Letters 310(3–4): Physics of the Earth and Planetary Interiors 187: 170–187.
509–517. http://dx.doi.org/10.1016/j.epsl.2011.08.007. Laj C, Kissel C, Garnier F, and Herrero-Bervera E (1996) Relative geomagnetic field
Johnson RG (1982) Brunhes–Matuyama magnetic reversal dated at 790,000 yr B.P. by intensity and reversals for the last 1.8My from a central Pacific core. Geophysical
marine–astronomical correlations. Quaternary Research 17: 135–147. Research Letters 23: 3393–3396.
Kawai N (1984) Paleomagnetic study of the Lake Biwa sediments. In: Horie S and Laj C, Kissel C, and Guillou H (2002a) Brunhes’ research revisited: Magnetization of
Junk W (eds.) Lake Biwa Dordrecht: Dr. W Junk Publisher. volcanic flows and baked clays. Eos, Transactions American Geophysical Union
Kawai N, Yaskawa K, Nakajima T, Torii M, and Horie S (1972) Oscillating geomagnetic 83(35), 381, 386–387.
field with a recurring reversal discovered from Lake Biwa. Proceedings of the Japan Laj C, Kissel C, Mazaud A, Channell JET, and Beer J (2000) North Atlantic
Academy 48: 186–190. paleointensity stack since 75 ka (NAPIS-75) and the duration of the Laschamp
Keigwin LD (1995) Oxygen isotope stratigraphy from Hole 883D, Leg 145. Proceedings event. Philosophical Transactions of the Royal Society of London 358: 1009–1025.
of the ODP Science Research 145: 257–264. Laj C, Kissel C, and Roberts AP (2006) Geomagnetic field behavior during the Icelandic
Keigwin L and Jones G (1994) Western North Atlantic evidence for millennial-scale basin and Laschamp geomagnetic excursions: A simple transitional field geometry?
changes in ocean circulation and climate. Journal of Geophysical Research Geochemistry, Geophysics, Geosystems 7: Q03004. http://dx.doi.org/
99: 12397–12410. 10.1029/2005GC001122.
Kent DV, Hemming SR, and Turrin BD (2002) Laschamp excursion at Mono Lake? Earth Laj C, Kissel C, Scao V, et al. (2002b) Geomagnetic intensity and inclination variations
and Planetary Science Letters 197: 151–164. at Hawaii for the past 98 kyr from core SOH-4 (Big Island): A new study and a
Kent DV and Schneider DA (1995) Correlation of the paleointensity variation records in comparison with existing contemporary data. Physics of the Earth and Planetary
the Brunhes/Matuyama polarity transition interval. Earth and Planetary Science Interiors 129: 205–243.
Letters 129: 135–144. Laj C, Mazaud A, Fuller M, and Herrero-Bervera E (1991) Geomagnetic reversal paths.
Kent DV and Spariosu DJ (1983) High resolution magnetostratigraphy of Caribbean Nature 351: 11–26.
Plio–Pleistocene sediments. Palaeogeography, Palaeoclimatology, Palaeoecology Lal D and Peters B (1967) Cosmic ray produced radioactivity on the Earth. In: Flügge S
42: 47–64. (ed.) Handbuch fur Physik, pp. 551–612. Berlin: Springer.
Khramov AN (1960) Palaeomagnetism and Stratigraphic Correlation. Canberra: Lanci L, Kissel C, Leonardt R, and Laj C (2008) Morphology of the Iceland Basin
Geophysics Department, A.N.U, Lojkine AJ (trans.), 408 pp. Excursion from a spherical harmonics analysis and an iterative Bayesian inversion
King JW, Banerjee SK, Marvin JA, and Özdemier Ö (1982) A comparison of different procedure of sedimentary records. Earth and Planetary Science Letters
magnetic methods for determining the relative grain size of magnétite in natural 169: 131–139.
materials: Some results from lake sediments. Earth and Planetary Science Letters Lanci L, Lowrie W, and Montanari A (1996) Magnetostratigraphy of the Eocene/
59: 404–419. Oligocene boundary in a short drill-core. Earth and Planetary Science Letters
King JW, Heil C, O’Regan M, et al. (2005) Paleomagnetic results from Pleistocene 143: 37–48.
sediments of Lomonosov ridge, central Arctic ocean, IODP Leg 302. Eos, Lanci L, Pares JM, Channell JET, and Kent DV (2005) Oligocene magnetostratigraphy
Transactions American Geophysical Union 85(47), Fall Meeting Supplement, from equatorial Pacific sediments (ODP Sites 1218 and 1219, Leg 199). Earth and
Abstract GP44A-04. Planetary Science Letters 237: 617–634.
Kissel C, Guillou H, Laj C, et al. (2011) The Mono Lake excursion recorded in phonolitic Langereis CG, Dekkers MJ, de Lange GJ, Paterne M, and Van Santvoort PJM (1997)
lavas from Tenerife (Canary Islands): Paleomagnetic analyses and coupled K/Ar and Magnetostratigraphy and astronomical calibration of the last 1.1 Myr from an
Ar/Ar dating. Physics of the Earth and Planetary Interiors 187: 232–244. http://dx. eastern Mediterranean piston core and dating of short events in the Brunhes.
doi.org/10.1016/j.pepi.2011.04.014. Geophysical Journal International 129: 75–94.
Kissel C and Laj C (2004) Improvements in procedure and selection criteria Lanphere MA (2000) Comparison of conventional K–Ar and 40Ar/39Ar dating of young
(PICRIT-03) for Thellier and Thellier paleointensity determinations: Application to mafic rocks. Quaternary Research 53: 294–301.
Hawaiian basaltic long cores. Physics of the Earth and Planetary Interiors Lanphere MA, Champion DE, Christiansen RL, Izett GA, and Obradovich JD (2002)
147: 155–169. Revised ages for tuffs of the Yellowstone plateau volcanic field: Assignment of the
Kissel C, Laj C, Labeyrie L, Dokken T, Voelker A, and Blamart D (1999) Rapid Huckleberry ridge tuff to a new geomagnetic polarity event. Geological Society of
climatic variations during marine isotopic stage 3: Magnetic analysis of sediments America Bulletin 114: 559–568.
from Nordic seas and north Atlantic. Earth and Planetary Science Letters Larson RL and Hilde TWC (1975) A revised time scale of magnetic reversals for the early
171: 489–502. Cretaceous and late Jurassic. Journal of Geophysical Research 80: 2586–2594.
Lehman B, Laj C, Kissel C, Mazaud A, Paterne M, and Labeyrie L (1996) Relative Lund SP, Liddicoat JC, Lajoie KL, Henyey TL, and Robinson S (1988) Paleomagnetic
changes of the geomagnetic field intensity during the last 280 kyear from evidence for long-term 104 year memory and periodic behavior in the Earth’s core
piston cores in the Azores area. Physics of the Earth and Planetary Interiors dynamo process. Geophysical Research Letters 15: 1101–1104.
93: 269–284. Lund SP, Schwartz M, Keigwin L, and Johnson T (2005) Deep-sea sediment records of
Leonhardt R and Fabian K (2007) Paleomagnetic reconstruction of the global the Laschamp geomagnetic field excursion (<41,000 calendar years before
geomagnetic field evolution during the Matuyama/Brunhes transition: Iterative present). Journal of Geophysical Research 110: Q12006. http://dx.doi.org/
Bayesian inversion and independent verification. Earth and Planetary Science Letters 10.1029/2005GC001036).
253: 172–195. Lund SP, Stoner JS, Channell JET, and Acton G (2006) A summary of Brunhes
Leonhardt R, Fabian K, Winklhofer M, Ferk A, Laj C, and Kissel C (2009) Geomagnetic paleomagnetic field variability recorded in ODP Cores. Physics of the Earth and
field evolution during the Laschamp excursion. Earth and Planetary Science Letters Planetary Interiors 156: 194–204.
278: 87–95. Maenaka K (1983) Magnetostratigraphic study of the Osaka group, with special
Levi S, Gudmunsson H, Duncan RA, Kristjansson L, Gillot PV, and Jacobsson SP reference to the existence of pre and post-Jaramillo episodes in the late Matuyama
(1990) Late Pleistocene geomagnetic excursion in Icelandic lavas: Confirmation of polarity Epoch. Memoirs of the Hanazono University 14: 1–65.
the Laschamp excursion. Earth and Planetary Science Letters 96: 443–457. Mankinen EA and Dalrymple GB (1979) Revised geomagnetic polarity time scale for the
Levi S and Karlin R (1989) A sixty thousand year paleomagnetic record from Gulf of interval 0–5 m.y. B.P. Journal of Geophysical Research 84: 615–626.
California sediments: Secular variation, late Quaternary excursions and geomagnetic Mankinen EA, Donnelly JM, and Grommé CS (1978) Geomagnetic polarity event
implications. Earth and Planetary Science Letters 92: 443–447. recorded at 1.1 m.y.b.p. on Cobb mountain, Clear Lake volcanic field, California.
Li J-J, Fang X-M, Van der Voo R, et al. (1997) Late Cenozoic magnetostratigraphy Geology 6: 653–656.
(11-0 Ma) of the Dongshanding and Wangjiashan sections in the Longzhong basin, Mankinen EA, Donnelly-Nolan JM, Grommé CS, and Hearn BC Jr (1981)
western China. Geologie en Mijnbouw 76: 121–134. Paleomagnetism of the Clear lake volcanics and new limits on the age of the
Li J-J, Zhu J-J, Kang J-C, et al. (1992) The comparison of Lanzhou loess profile with Jaramillo normal-polarity event. U.S. Geological Survey Professional Paper
Vostok ice core in Antarctica over the last glaciation cycle. Science in China 35(B): 1141: 67–82.
476–487. Mankinen EA and Grommé CS (1982) Paleomagnetic data from the Cosa Range,
Liddicoat JC (1990) Aborted reversal of the paleomagnetic field in Brunhes California, and current status of the Cobb mountain normal geomagnetic polarity
normal Chron in east-central California. Geophysical Journal International event. Geophysical Research Letters 9: 1239–1282.
102: 747–752. Mankinen EA, Prévot M, Grommé CS, and Coe RS (1985) The Steens mountain
Liddicoat JC (1992) Mono Lake excursion in Mono basin, California, and at (Oregon) geomagnetic polarity transition. 1. Directional history, duration of the
Carson sink and Pyramid lake, Nevada. Geophysical Journal International episodes, and rock magnetism. Journal of Geophysical Research 90: 393–416.
108: 442–452. Marshall M, Chauvin A, and Bonhommet N (1988) Preliminary paleointensity
Liddicoat JC (1996) Mono lake excursion in the Lahontan basin, Nevada. Geophysical measurements and detailed magnetic analyses of basalts from the
Journal International 125: 630–635. Skalamaelifell excursion, south-west Iceland. Journal of Geophysical Research
Liddicoat JC and Bailey RA (1989) Short reversal of the paleomagnetic field about 93: 11681–11698.
280,000 years ago at Long Valley, California. In: Lowes FJ, Collinson DW, Parry JH, Martinson DG, Pisias NG, Hays JD, Imbrie J, Moore TC Jr, and Shackleton NJ (1987)
Runcorn SK, Tozer DC, and Soward AM (eds.) Proceedings of the NATO Age dating and the orbital theory of the Ice Ages: Development of a high-resolution
Advanced Study Institute on Geomagnetism and Paleomagnetism. NATO Advanced 0 to 300,000-year chronostratigraphy. Quaternary Research 27: 1–29.
Study Institutes Series: Mathematical and Physical Sciences. pp. 137–153. Masarik J and Beer J (1999) Simulation of particle fluxes and cosmogenic nuclide
Dordrecht: Kluwer Academic Publ. production in the Earth’s atmosphere. Journal of Geophysical Research
Liddicoat JC and Coe RS (1979) Mono lake geomagnetic excursion. Journal of 104: 12099–12111.
Geophysical Research 84: 261–271. Matuyama M (1929) On the direction of magnetization of basalts in Japan, Tyosen and
Liddicoat JC, Lajoie KR, and Sarna-Wojcicki AM (1982) Detection and dating of the Manchuria. Proceedings of the Imperial Academy (Tokyo) 5: 203–205.
Mono lake excursion in the lake Lahontan Sehoo formation, Carson sink, Nevada. Mazaud A, Sicre MA, and Ezat U (2002) Geomagnetic-assisted stratigraphy and sea
Eos, Transactions American Geophysical Union 63: 920. surface temperature changes in core MD94-103 (southern Indian Ocean): Possible
Liu TS, An ZS, Yuan BY, and Han JM (1985) The loess-paleosol sequence in China and implications for north–south climatic relationships around H4. Earth and Planetary
climatic history. Episodes 8: 21–28. Science Letters 201: 159–170.
Lourens LJ, Antonarakou A, Hilgen FJ, van Hoof AAM, Vergnaud-Grazzini C, and McDougall F, Brown FH, Cerling TE, and Hillhouse JW (1992) A reappraisal of the
Zachariasse WJ (1996) Evaluation of the Plio-Pleistocene astronomical timescale. geomagnetic polarity time scale to 4 Ma using data from Turkana basin, east Africa.
Paleoceanography 11(4): 391–413. http://dx.doi.org/10.1029/96PA01125. Geophysical Research Letters 19: 2349–2352.
Lourens LJ, Hilgen FJ, Laskar J, Shackleton NJ, and Wilson D (2004) The Neogene McDougall I and Chamalaun FH (1966) Geomagnetic polarity scale of time. Nature
period. In: Gradstein FM, Ogg JG, and Smith AG (eds.) Geologic Time Scale 2004, 212: 1415–1418.
pp. 409–440. Cambridge, NY: Cambridge University Press. McDougall I and Tarling DH (1963a) Dating of reversals of the Earth’s magnetic field.
Lowrie W and Alvarez W (1981) One hundred million years of geomagnetic polarity Nature 198: 1012–1013.
history. Geology 9: 392–397. McDougall I and Tarling DH (1963b) Dating of polarity zones in the Hawaiian Islands.
Lowrie W, Alvarez W, Premoli Silva I, and Monechi W (1980) Lower Cretaceous Nature 200: 54–56.
magnetic stratigraphy in Umbrian pelagic carbonate rocks. Geophysical Journal of McDougall I and Tarling DH (1964) Dating geomagnetic polarity zones. Nature
the Royal Astronomical Society 60: 263–281. 202: 171–172.
Lowrie W and Kent DV (2004) Geomagnetic polarity timescales and reversal McDougall I and Watkins ND (1973) Age and duration of the Réunion geomagnetic
frequency regimes. In: Channell JET, et al. (ed.) Timescales of the Geomagnetic polarity event. Earth and Planetary Science Letters 19: 443–452.
Field. AGU Geophysical Monograph Series, vol. 145, pp. 117–129. Washington, McDougall I and Wensink J (1966) Paleomagnetism and geochronology of the
DC: AGU. Pliocene–Pleistocene lavas in Iceland. Earth and Planetary Science Letters
Lowrie W and Lanci L (1994) Magnetostratigraphy of Eocene–Oligocene boundary 1: 232–236.
sections in Italy: No evidence for short subchrons within chrons 12R and 13R. Earth McWilliams M (2001) Global correlation of the 223 ka Pringle Falls event. International
and Planetary Science Letters 126: 247–258. Geological Review 43: 191–195.
Lund SP, Acton GD, Clement B, Okada M, and Williams T (2001a) Brunhes chron Ménabréaz L, Bourlès DL, and Thouveny N (2012) Amplitude and timing of the
magnetic field excursions recovered from Leg 172 sediments. In: Keigwin LD, Rio D, Laschamp geomagnetic dipole low from the global atmospheric 10Be
Acton GD, and Arnold E (eds.) Proceedings of the ODP Science Research, vol. 172, overproduction: Contribution of authigenic 10Be/9Be ratios in west equatorial
pp. 1–18. Texas A&M University, Ocean Drilling Program. Pacific sediments. Journal of Geophysical Research 117: B11101. http://dx.doi.org/
Lund SP, Acton GD, Clement B, and OkadaM WT (2001b) Paleomagnetic records of 10.1029/2012JB009256.
Stage 3 excursions, Leg 172. In: Keigwin LD, Rio D, Acton GD, and Arnold E (eds.) Ménabréaz L, Thouveny N, Bourles DL, Deschamps P, Hamelin B, and Demory F (2011)
Proceedings of the ODP Science Research, vol. 172, pp. 1–20. Texas A&M The Laschamp geomagnetic dipole low expressed as a cosmogenic Be-10
University, Ocean Drilling Program. atmospheric overproduction at similar to 41 ka. Earth and Planetary Science Letters
Lund SP, Acton G, Clement B, et al. (1998) Geomagnetic field excursions occurred often 312(3–4): 305–317.
during the last million years. Eos, Transactions American Geophysical Union Merrill RT and McFadden PL (1994) Geomagnetic field stability: Reversal events and
79(14): 178–179. excursions. Earth and Planetary Science Letters 121: 57–69.
Meynadier L, Valet J-P, and Shackleton NJ (1995) Relative geomagnetic intensity Ogg JG and Bardot L (2001) Aptian through Eocene magnetostratigraphic correlation of
during the last 4 My from the equatorial Pacific. Proceedings of the ODP Science the Blake Nose Transect (Leg 171B), Florida continental margin. In: Kroon D,
Research 138: 779–793. Norris RD, and Klaus A (eds.) Proceedings of the ODP, Science Results,
Meynadier L, Valet J-P, Weeks R, Shackleton NJ, and Hagee VL (1992) Relative 171B: College Station, TX: Ocean Drilling Program (available online).
geomagnetic intensity of the field during the last 140 ka. Earth and Planetary Ogg JG, Karl SM, and Behl RJ (1992) Jurassic through early Cretaceous sedimentation
Science Letters 114: 39–57. history of the central equatorial Pacific and of Sites 800 and 801. Proceedings of
Mochizuki N, Tsunakawa H, Shibuya H, Cassidy J, and Smith IEM (2006) ODP, Science Research. 129. College Station, TX: Ocean Drilling Program,
Paleointensities of the Auckland geomagnetic excursions. Physics of the Earth and pp. 571–613.
Planetary Interiors 154: 168–179. Opdyke ND and Channell JET (1996) Magnetic Stratigraphy. San Diego, CA: Academic
Mochizuki N, Tsunakawa H, Shibuya H, Tagami T, Ozawa A, and Smith IEM (2007) Press, 346 pp.
Further K–Ar dating and paleomagnetic study of the Auckland geomagnetic Opdyke ND, Glass B, Hays JP, and Foster J (1966) Paleomagnetic study of Antarctic
excursions. Earth, Planets and Space 59: 755–761. deep-sea cores. Science 154: 349–357.
Mochizuki N, Tsunakawa H, Shibuya H, et al. (2004) K–Ar ages of the Auckland Osete ML, Martin-Chivelet J, Rossi C, et al. (2012) The Blake geomagnetic excursion
geomagnetic excursions. Earth, Planets and Space 56: 283–288. recorded in a radiometrically dated speleothem. Earth and Planetary Science Letters
Mörner NA and Lanser JP (1974) Gothenburg magnetic “flip”. Nature 251: 408–409. 353: 173–181. http://dx.doi.org/10.1016/j.epsl.2012.07.041.
Muscheler R, Beer J, Wagner G, et al. (2004) Changes in the carbon cycle during the Palmer DF, Henyey TL, and Dodson RE (1979) Paleomagnetic and sedimentological
last deglaciation as indicated by the comparison of 10Be and 14C records. Earth and studies at Lake Tahoe, California, Nevada. Earth and Planetary Science Letters
Nagata T (1952) Reverse thermal-remanent magnetism. Nature 169: 704. Parés JM, Van der Voo R, Yan M, and Fang X (2004) After the dust settles: Why is
Nagata T, Uyeda S, and Ozima M (1957) Magnetic interaction between ferromagnetic the Blake event imperfectly recorded in Chinese Loess? In: Channell JET, Kent DV,
minerals contained in rocks. Philosophical Magazine, Supplement Advanced Lowrie W, and Meert J (eds.) Timescales of the Paleomagnetic Field. AGU
Physics 6: 264–287. Geophysical Monograph Series, vol. 145, pp. 191–204. Washington, DC: AGU.
Negrini RM, Davis JO, and Verosub KL (1984) Mono lake geomagnetic excursion found Peddie N and Zunde A (1988) The magnetic field of the Earth: 1985. U.S. Geol. Surv.
at Summer Lake, Oregon. Geology 12: 643–646. Geophys. Invest. Map, GP-987-1.
Negrini RM, Erbes DB, Faber K, et al. (2000) A paleoclimate record for the past Picard MD (1964) Paleomagnetic correlation of units within Chugwater (Triassic)
250,000 years from Summer Lake, Oregon, USA: I. Chronology and magnetic formation, west-central Wyoming. American Association of Petroleum Geologists
proxies for lake level. Journal of Paleolimnology 24: 125–149. Bulletin 48: 269–291.
Negrini RM, Erbes DB, Roberts AP, Verosub KL, Sarna-Wojcicki AM, and Meyer CE Pillans BJ, Roberts AP, Wilson GS, Abbott ST, and Alloway BV (1994)
(1994) Repeating waveform initiated by a 180–190 ka geomagnetic excursion in Magnetostratigraphic, lithostratigraphic and tephrostratigraphic constraints on
eastern north America: Implications for field behavior during polarity transitions Lower and Middle Pleistocene sea-level changes, Wanganui Basin, New Zealand.
and subsequent secular variation. Journal of Geophysical Research Earth and Planetary Science Letters 121: 81–98.
99: 24105–24119. Pitman WC III and Heirtzler JR (1966) Magnetic anomalies over the Pacific–Antarctic
Negrini RM, Verosub KL, and Davis JO (1988) The middle to late Pleistocene ridge. Science 154: 1164–1171.
geomagnetic field recorded in fine-grained sediments from Summer Lake, Oregon, Plenier G, Valet J-P, Guérin G, Lefèvre J-C, LeGoff M, and Carter-Stiglitz B (2007)
and double hot springs, Nevada, U.S.A. Earth and Planetary Science Letters Origin and age of the directions recorded during the Laschamp event in the Chaı̂ne
87: 173–192. des Puys (France). Earth and Planetary Science Letters 259: 414–431.
Ness G, Levi S, and Couch R (1980) Marine magnetic anomaly timescales for the Quidelleur X, Carlut J, Gillot PY, and Soler V (2002) Evolution of the geomagnetic field
Cenozoic and Late Cretaceous: A precis, critique and synthesis. Reviews of prior to the Matuyama–Brunhes transition: Radiometric dating of a 820 ka excursion
Geophysics and Space Physics 18: 753–770. at La Palma. Geophysical Journal International 151: F6–F10.
Ninkovitch D, Opdyke ND, Heezen BC, and Foster JH (1966) Paleomagnetic Quidelleur X, Gillot PY, Carlut J, and Courtillot V (1999) Link between excursions and
stratigraphy, rates of deposition and tephrachronology in north Pacific deep-sea paleointensity inferred from abnormal field directions recorded at La Palma around
sediments. Earth and Planetary Science Letters 1: 476–492. 600 ka. Earth and Planetary Science Letters 168: 233–242.
Noel M and Tarling DH (1975) The Laschamp ‘event’. Nature 253: 705–707. Quidelleur X and Valet J-P (1996) Geomagnetic changes across the last reversal record
Nowaczyk NR and Antonow M (1997) High resolution magnetostratigraphy of four in lava flows from La Palma, Canary Islands. Journal of Geophysical Research
sediment cores from the Greenland Sea—I. Identification of the Mono Lake 101: 13755–13773.
excursion, Laschamp and Biwa I/Jamaica geomagnetic polarity events. Geophysical Raisbeck GM, Yiou F, Jouzel J, and Stocker TF (2007) Direct north-south
Journal International 131: 310–324. synchronization of abrupt climate change records in ice cores using beryllium 10.
Nowaczyk NR, Arz HW, Frank U, Kind J, and Plessen B (2012) Dynamics of Climate of the Past 3: 541–547.
the Laschamp geomagnetic excursion from Black Sea sediments. Earth and Raymo ME, Ruddiman WF, Backman J, Clement BM, and Martinson DG (1989) Late
Planetary Science Letters 351: 54–69. http://dx.doi.org/10.1016/j. Pliocene variation in northern hemisphere ice sheets and North Atlantic deep water
epsl.2012.06.050. circulation. Paleoceanography 4(4): 413–446. http://dx.doi.org/10.1029/
Nowaczyk NR and Baumann M (1992) Combined high-resolution magnetostratigraphy PA004i004p00413.
and nannofossil biostratigraphy for late Quaternary Arctic ocean sediments. Deep Renne PR, Deino PL, Walter RC, et al. (1994) Intercalibration of astronomical and
Sea Research 39: 567–601. radioisotopic time. Geology 22: 783–786.
Nowaczyk NR and Frederichs TW (1999) Geomagnetic events and relative Reynolds R (1977) Paleomagnetism of welded tuffs of the Yellowstone group. Journal of
palaeointensity variations during the past 300 ka as recorded in Kolbeinsey Ridge Geophysical Research 82: 3677–3693.
sediments, Iceland Sea: indication for a strongly variable geomagnetic field. Roberts AP (2008) Geomagnetic excursions: Knowns and unknowns. Geophysical
International Journal of Earth Sciences 88: 116–131. Research Letters 35: L17307. http://dx.doi.org/10.1029/2008GL034719.
Nowaczyk NR, Frederichs TW, Eisenhauer A, and Gard G (1994) Magnetostratigraphic Roberts AP, Lehman B, Weeks RJ, Verosub KL, and Laj C (1997) Relative
data from late Quaternary sediments from the Yermak plateau, Arctic Ocean: paleointensity of the geomagnetic field over the last 200,000 years from ODP
Evidence for four geomagnetic polarity events within the last 170 ka of the Brunhes Sites 883 and 884, north Pacific ocean. Earth and Planetary Science Letters
Chron. Geophysical Journal International 117: 453–471. 152: 11–23.
Nowaczyk NR and Knies J (2000) Magnetostratigraphic results from the eastern Arctic Roberts AP and Lewin-Harris JC (2000) Marine magnetic anomalies: Evidence that ‘tiny
ocean: AMS 14C ages and relative paleointensity data of the Mono Lake and wiggles’ represent short-period geomagnetic polarity intervals. Earth and Planetary
Laschamp geomagnetic reversal excursions. Geophysical Journal International Science Letters 183: 375–388.
140: 185–197. Roberts AP and Winklhofer M (2004) Why are geomagnetic excursions not always
O’Regan M, King J, Backman J, et al. (2008) Constraints on the Pleistocene chronology recorded in sediments? Constraints from post-depositional remanent magnetization
of sediments from the Lomonosov Ridge. Paleoceanography 23: PA1S19. http://dx. lock-in modeling. Earth and Planetary Science Letters 227: 345–359.
doi.org/10.1029/2007PA001551. Robinson C, Raisbeck GM, Yiou F, Lehman B, and Laj C (1995) The relationship
Oda H, Nakamura K, Ikehara K, Nakano T, Nishimura M, and Khlystov O (2002) between 10Be and geomagnetic field strength records in central North Atlantic
Paleomagnetic record from Academician ridge, Lake Baikal: A reversal excursion at sediments during the last 80 ka. Earth and Planetary Science Letters 136: 551–557.
the base of marine oxygen isotope stage 6. Earth and Planetary Science Letters Roche A (1950) Sur les caracteres magnétiques du système éruptif de Gergovie.
202: 117–132. Comptes Rendus de l’Académie des Sciences, Paris 230: 113–115.
Roche A (1951) Sur les inversions de l’aimentation remanente des roches volcaniques Singer BS, Jicha BR, Kirby BT, Zhang X, Geisman JW, and Herrero-Bervera E (2005) An
dans les monts d’Auvergne. Comptes Rendus de l’Académie des Sciences, Paris 40Ar/39Ar age for geomagnetic instability recorded at the Albuquerque volcanoes
223: 1132–1134. and Pringle Falls, Oregon. Eos, Transactions American Geophysical Union 86 (53).
Roche A (1956) Sur la date de la derniere inversion du champ magnetique terrestre. Fall Meeting Supplement Abstract GP21A-0019.
Comptes Rendus de l’Académie des Sciences, Paris 243: 812–814. Singer BS, Relle MK, Hoffman KA, et al. (2002) 40Ar/39Ar ages from transitionally
Roger S, Coulon C, Thouveny N, et al. (2000) 40Ar/39Ar dating of a tephra layer in the magnetized lavas on La Palma, Canary Islands, and the geomagnetic instability
Pliocene Seneze maar lacustrine sequence (French Massif central): Constraint on timescale. Journal of Geophysical Research 107: 2307. http://dx.doi.org/10.1029/
the age of the Réunion-Matuyama transition and implications on 2001JB001613.
paleoenvironmental archives. Earth and Planetary Science Letters 183: 431–440. Smith JD and Foster JH (1969) Geomagnetic reversal in the Brunhes normal polarity
Roperch P, Bonhommet N, and Levi S (1988) Paleointensity of the Earth’s magnetic field epoch. Science 163: 565–567.
during the Laschamp excursion and its geomagnetic implications. Earth and Sobel ER and Arnaud N (2000) Cretaceous–Paleogene basaltic rocks of the Tuyon
Planetary Science Letters 88: 209–219. Basin, NW China and the Kyrgyz Tian Shan: The trace of a small plume. Lithos
Roperch P, Herail G, and Fornari M (1999) Magnetostratigraphy of the Miocene 50: 191–215.
Corque basin, Bolivia: Implications for the geodynamic evolution of the Spassov S, Heller F, Kretzschmar R, Evans ME, Yue LP, and Nourgaliev DK (2003)
Altiplano during the late Tertiary. Journal of Geophysical Research Detrital and pedogenic magnetic mineral phases in the loess/palaeosol sequence at
104: 20415–20429. Lingtai (Central Chinese Loess Plateau). Physics of the Earth and Planetary Interiors
Ruddiman WF, Raymo ME, Martinson DG, Clement BM, and Backman J (1989) 140: 255–275.
Pleistocene evolution: Northern hemisphere ice sheet and north Atlantic ocean. Spell TL and McDougall I (1992) Revisions to the age of the Brunhes/Matuyama
Paleoceanography 4: 353–412. boundary and the Pleistocene geomagnetic polarity timescale. Geophysical
Rutten MG and Wensink H (1960) Paleomagnetic dating, glaciations and chronology of Research Letters 19: 1182–1184.
the Plio-Pleistocene in Iceland. In: International Geological Congress Session 21 IV, Steiger RH and Jager E (1977) Subcommission on geochronology: Convention on the
pp. 62–70. use of decay constants in geo- and cosmochronology. Earth and Planetary Science
Ryan WBF (1972) Stratigraphy of late Quaternary sediments in the eastern Letters 36: 359–362.
Mediterranean. In: Stanley DJ (ed.) The Mediterranean Sea, pp. 149–169. Stoner JS, Channell JET, Hodell DA, and Charles C (2003) A 580 kyr paleomagnetic
Stroudsberg, PA: Dowden, Hutchinson and Ross. record from the sub-Antarctic south Atlantic (ODP Site 1089). Journal of
Sakai H, Nakamura T, Horii M, et al. (1997) Paleomagnetic study with 14C dating Geophysical Research 108: 2244. http://dx.doi.org/10.1029/2001JB001390.
analysis on three short cores from Lake Baikal. Bulletin of the Nagoya University Stoner JS and St-Onge G (2007) Magnetic stratigraphy in paleoceanography: Reversals,
Furukawa Museum 13: 11–22. excursions, paleointensity, and secular variation. In Hillaire-Marcel C and De Vernal
Schneider DA (1995) Paleomagnetism of some Ocean Drilling Program Leg 138 A (eds.) Proxies in Late Cenozoic Paleoceanography. Developments in Marine
sediments: Detailing Miocene stratigraphy. In: Pisias NG, Mayer LA, Janecek TR, Geology, pp. 99–137. Amsterdam: Elsevier B.V.
Palmer-Julson A, and Van Andel TH (eds.) Proceedings of the ODP Science Svensson A, Andersen KK, Bigler M, et al. (2006) The Greenland Ice Core Chronology
Research, vol. 138, pp. 59–72. College Station, TX: ODP. 15–42 kyr. Part 2: Comparison to other records. Quaternary Science Reviews
Schnepp E and Hradetzky H (1994) Combined paleointensity and 40Ar/39Ar age 25: 3258–3267.
spectrum from volcanic rocks of the West Eifel field (Germany)—Evidence for an Svensson A, Anderson KK, Bigler M, et al. (2008) A 60,000 year Greenland stratigraphic
early Brunhes geomagnetic excursion. Journal of Geophysical Research 99(B5): ice core chronology. Climate of the Past 4: 47–57.
9061–9076. http://dx.doi.org/10.1029/93JB03365. Takatsugi KO and Hyodo M (1995) A geomagnetic excursion during the late Matuyama
Shackleton NJ, Berger A, and Peltier WR (1990) An alternative astronomical calibration Chron, the Osaka group, southwest Japan. Earth and Planetary Science Letters
of the lower Pleistocene timescale based on ODP Site 677. Transactions of the Royal 136: 511–524.
Society of Edinburgh: Earth Sciences 81: 251–261. Tamaki K and Larson RL (1988) The Mesozoic tectonic history of the Magellan microplate
Sharp WD and Renne PR (2005) The 40Ar/39Ar dating of core recovered by the Hawaii in the western central Pacific. Journal of Geophysical Research 93: 2857–2874.
Scientific Drilling Project (phase 2), Hilo, Hawaii. Geochemistry, Geophysics, Tanaka H, Turner GM, Houghton BF, Tachibana T, Kono M, and McWilliams MO (1996)
Geosystems 6(4): 18 pp. Paleomagnetism and chronology of the central Taupo Volcanic zone, New Zealand.
Shibuya H, Cassidy J, Smith IEM, and Itaya T (1992) A geomagnetic excursion in the Geophysical Journal International 124: 919–934.
Brunhes epoch recorded in New Zealand basalts. Earth and Planetary Science Tarduno JA (1990) Brief reversed polarity interval during the Cretaceous normal polarity
Letters 111: 41–48. superchron. Geology 18: 683–686.
Singer BS (2007) Polarity transitions: Radioisotopic dating. In: Gubbins D and Tarduno JA, Lowrie W, Sliter WV, Bralower TJ, and Heller F (1992) Reversed polarity
Bervera EH (eds.) Encyclopedia of Geomagnetism and Paleomagnetism, characteristics magnetizations in the Albian Contessa section, Umbrian Appenius,
pp. 834–839. Dordrecht, The Netherlands: Springer. Italy: Implications for the existence of a mid-Cretaceous mixed polarity interval.
Singer B and Brown LL (2002) The Santa Rosa Event: 40Ar/39Ar and paleomagnetic Journal of Geophysical Research 97: 241–271.
results from the Valles rhyolite near Jaramillo Creek, Jemez Mountains, New Tauxe L, Deino AL, Behrensmeyer AK, and Potts R (1992) Pinning down the Brunhes/
Mexico. Earth and Planetary Science Letters 197(1–2): 51–64. Matuyama and upper Jaramillo boundaries: A reconciliation of orbital and isotopic
Singer BS, Brown LL, Rabassa JO, and Guillou H (2004) 40Ar/39Ar chronology of late time scales. Earth and Planetary Science Letters 190: 561–572.
Pliocene and early Pleistocene geomagnetic and glacial events in southern Teanby N, Laj C, Gubbins D, and Pringle M (2002) A detailed palaeointensity and
Argentina. In: Channell JET, et al. (ed.) Timescales of the Paleomagnetic Field. inclination record from drill core SOH1 on Hawaii. Physics of the Earth and
AGU Geophysical Monograph Series, vol. 145, pp. 175–190. Planetary Interiors 131(2): 101–140. http://dx.doi.org/10.1016/S0031-9201(02)
Washington, DC: AGU. 00032-8.
Singer BS, Guillou H, Jicha BR, et al. (2009) 40Ar/39Ar, K–Ar and 230Th–238U dating of Thouveny N, Carcaillet J, Moreno E, Leduc G, and Nérini D (2004) Geomagnetic
the Laschamp excursion: A radioisotopic tie-point for ice core and climate moment variation and paleomagnetic excursions since 400 kyr BP: A stacked record
chronologies. Earth and Planetary Science Letters 286: 80–88. from sedimentary sequences of the Portuguese margin. Earth and Planetary Science
Singer B, Hildreth W, and Vincze Y (2000) 40Ar/39Ar evidence for early deglaciation of Letters 219: 377–396.
the central Chilean Andes. Geophysical Research Letters 27: 1663–1666. Tiedemann R, Sarnthein M, and Shackleton NJ (1994) Astronomic timescale for the
Singer BS, Hoffman KA, Chauvin A, Coe RS, and Pringle MS (1999) Dating Pliocene Atlantic d18O and dust flux records of Ocean Drilling Program site 659.
transitionally magnetized lavas of the late Matuyama Chron: Toward a new 40Ar/39Ar Paleoceanography 9(4): 619–638. http://dx.doi.org/10.1029/94PA00208.
timescale of reversals and events. Journal of Geophysical Research 104: 679–693. Ton-That T, Singer B, and Paterne M (2001) Ar-40/Ar-39 dating of latest Pleistocene
Singer BS, Hoffman KA, Schnepp E, and Guillou H (2008a) Multiple Brunhes (41 ka) marine tephra in the Mediterranean Sea: Implications for global climate
Chron excursions recorded in the West Eifel (Germany) volcanics: Support for records. Earth and Planetary Science Letters 184(3–4): 645–658. http://dx.doi.org/
long-held mantle control over the non-axial dipole field. Physics of the Earth 10.1016/S0012-821X(000)00358-7.
and Planetary Interiors 169(1–4): 28–40. http://dx.doi.org/10.1016/j. Tric E, Laj C, Valet J-P, Tucholka P, Paterne M, and Guichard F (1991) The Blake
pepi.2008.05.001. geomagnetic event: Transition geometry, dynamical characteristics and
Singer BS, Jicha BR, Kirby BT, Geissman JW, and Herrero-Bervera E (2008b) 40Ar/39Ar geomagnetic significance. Earth and Planetary Science Letters 102: 1–13.
dating links Albuquerque Volcanoes to the Pringle Falls excursion and the Tucholka P, Fontugne M, Guichard F, and Paterne M (1987) The Blake polarity episode
Geomagnetic Instability Time Scale. Earth and Planetary Science Letters 267(3–4): in cores from the Mediterranean Sea. Earth and Planetary Science Letters
584–595. http://dx.doi.org/10.1016/j.epsl.2007.12. 86: 320–326.
Turner GM, Evans ME, and Hussin IB (1982) A geomagnetic secular variation study field intensity or environmental signal? Physics of the Earth and Planetary Interiors
(31,000–19,000 yr BP) in western Canada. Geophysical Journal of the Royal 87: 213–229.
Astronomical Society 71: 159–171. Weeks RJ, Roberts AP, Verosub KL, Okada M, and Dubuisson GJ (1995b)
Turrin BD, Donnelly-Nolan JM, and Hearn BC Jr. (1994) 40Ar/39Ar ages from the Magnetostratigraphy of upper Cenozoic sediments from Leg 145, north Pacific ocean.
rhyolite of Alder Creek, California: Age of the Cobb mountain normal polarity In: Rea DK, Basov IA, Scholl DW, and Allan J (eds.) Proceedings of the ODP,
subchron revisited. Geology 22: 251–254. Science Research, vol. 145, pp. 209–302. College Station, TX: Ocean Drilling
Udagawa S, Kitagawa H, Gudmundsson A, et al. (1999) Age and magnetism of lavas in Program.
Jokuldalur area, eastern Iceland: Gilsa event revisited. Physics of the Earth and Wensink H (1964) Paleomagnetic stratigraphy of younger basalts and intercalated Plio–
Planetary Interiors 115: 147–171. Pleistocene tillites in Iceland. Geologische Rundschau 54: 364–384.
Uyeda S (1958) Thermoremanent magnetism as a medium of palaeomagnetism with Whitney J, Johnson HP, Levi S, and Evans BW (1971) Investigations of some magnetic
special reference to reverse thermoremanent magnetism, Japan. Journal of and mineralogical properties of the Laschamp and Olby flows, France. Quaternary
Geophysics 2: 1–123. Research 1: 511–521.
Valet J-P and Herero-Bervera E (2003) Some characteristics of geomagnetic reversals Wilson DS and Hey RN (1981) The Galapagos axial magnetic anomaly: Evidence for the
inferred from detailed volcanic records. Comptes Rendus Geoscience 335: 79–90. emperor reversal within the Brunhes and for a two-layer magnetic source. Eos,
Valet J-P and Meynadier L (1993) Geomagnetic field intensity and reversals during the Transactions American Geophysical Union 61: 943 (abstract).
last four million years. Nature 366: 234–238. Wintle AG (1973) Anomalous fading of thermoluminescence in mineral samples. Nature
Vandamme D (1994) A new method to determine paleosecular variation. Physics of the 245: 143–144.
Earth and Planetary Interiors 85: 131–142. Wollin G, Ericson DB, Ryan WBF, and Foster JH (1971) Magnetism of the earth and
VandenBerg J, Klootwijk CT, and Wonders AAH (1978) Late Mesozoic and Cenozoic climatic changes. Earth and Planetary Science Letters 12: 175–183.
movements of the Italian peninsula; further paleomagnetic data from the Umbrian Worm H-U (1997) A link between geomagnetic reversals and events and glaciations.
sequence. Geological Society of America Bulletin 89: 133–150. Earth and Planetary Science Letters 147: 55–67.
Van der Voo R (1990) The reliability of paleomagnetic data. Tectonophysics 184: 1–9. Xuan C and Channell JET (2010) Origin of apparent magnetic excursions in deep-sea
Vazquez JA and Lidzbarski MI (2012) High-resolution tephrochronology of the Wilson sediments from Mendeleev-Alpha Ridge, Arctic Ocean. Geochemistry, Geophysics,
Creek Formation (Mono Lake, California) and Laschamp event using U-238–Th- Geosystems 11: Q02003. http://dx.doi.org/10.1029/2009GC002879.
230 SIMS dating of accessory mineral rims. Earth and Planetary Science Letters Xuan C, Channell JET, Polyak L, and Darby DA (2012) Paleomagnetism of Quaternary
357: 54–67. http://dx.doi.org/10.1016/j.epsl.2012.09.013. sediments from Lomonosov Ridge and Yermak Plateau: Implications for age models
Verosub KL (1977) The absence of the Mono Lake geomagnetic excursion from the in the Arctic Ocean. Quaternary Science Reviews 32: 48–63.
paleomagnetic record at Clear Lake, California. Earth and Planetary Science Letters Yamazaki T and Ikehara M (2012) Origin of magnetic mineral concentration variation in
36: 219–230. the Southern Ocean. Paleoceanography 27: PA2206. http://dx.doi.org/
Verosub KL (1982) Geomagnetic excursions: A critical assessment of the evidence as 10.1029/2011PA002271.
recorded in sediments of the Brunhes epoch. Philosophical Transactions of the Yamazaki T and Yoka N (1994) Long-term secular variation of the geomagnetic field
Royal Society of London A 306: 161–168. during the last 200 kyr recorded in sediment cores from the western equatorial
Verosub KL, Davis JO, and Valastro S (1980) A paleomagnetic record from Pyramid Pacific. Earth and Planetary Science Letters 128: 527–544.
Lake, Nevada, and its implications for proposed geomagnetic excursions. Earth and Yang Z, Clement BM, Acton GD, Lund SP, Okada M, and Williams T (2001) Record of
Planetary Science Letters 49: 141–148. the Cobb mountain subchron from the Bermuda Rise (ODP Leg 172). Earth and
Vine FJ (1966) Spreading of the ocean floor: New evidence. Science 154: 1405–1415. Planetary Science Letters 193: 303–313.
Vine FJ and Matthew DH (1963) Magnetic anomalies over oceanic ridges. Nature Yang T, Hyodo M, Yang Z, and Fu J (2004) Evidence for the Kamikatsura and Santa
199: 947–949. Rosa excursions recorded in Eolian geomagnetic Excursions 415 Treatise on.
Voelker A, Sarnthein M, Grootes PM, et al. (1998) Correlation of marine 14C ages from Geophysics 5: 373–416.
the Nordic sea with GISP2 isotope record: Implication for 14C calibration beyond Yaskawa K, Nakajima T, Kawai N, Torii M, Natsuhara N, and Horie S (1973)
25 ka BP. Radiocarbon 40: 517–534. Paleomagnetism of a core from Lake Biwa (I). Journal of Geomagnetism and
Wagner G, Beer J, and Laj C (2000) Chlorine-36 evidence for the Mono lake event in the Geoelectricity 25: 447.
Summit GRIP ice core. Earth and Planetary Science Letters 181: 1–6. Zhang K and Gubbins D (2000) Is the geodynamo intrinsically unstable? Geophysical
Walter RC, Manega PC, Hay RL, Drake RE, and Curtis GH (1991) Laser fusion 40Ar/39Ar Journal International 140: F1–F4.
dating of Bed 1, Olduvai Gorge, Tanzania. Nature 354: 145–149. Zhu R, Hoffman KA, Nomade S, et al. (2004) Geomagnetic paleointensity and direct age
Wang Y, Evans ME, Rutter N, and Ding ZL (1990) Magnetic susceptibility of Chinese determination of the ISEA (M0r?) chron. Earth and Planetary Science Letters
loess and its bearing on paleoclimate. Geophysical Research Letters 217: 285–295.
17: 2449–2451. Zhu RX, Zhou LP, Laj C, Mazaud A, and Ding DL (1994) The Blake geomagnetic
Watkins ND, McDougall I, and Kristjansson L (1975) A detailed paleomagnetic survey of polarity episode recorded in Chinese loess. Geophysical Research Letters
the type location for the Gilsa geomagnetic polarity event. Earth and Planetary 21: 697–700.
Science Letters 27: 436–444. Zimmerman SH, Hemming SR, Kent DV, and Searle SY (2006) Revised chronology for
Weeks R, Laj C, Endignoux L, et al. (1995a) Normalized NRM intensity during the last late Pleistocene Mono Lake sediments based on paleointensity correlation to the
240,000 years in piston cores from the central north Atlantic ocean: Geomagnetic global reference curve. Earth and Planetary Science Letters 252: 94–106.
5.11 The Time-Averaged Field and Paleosecular Variation
CL Johnson, University of British Columbia, Vancouver, BC, Canada; Planetary Science Institute, Tucson, AZ, USA
P McFadden, Geoscience Australia, Canberra, ACT, Australia
Published by Elsevier B.V.

5.11.1.1 The GAD Approximation 387
5.11.1.2 Paleosecular Variation 387
5.11.1.3 The Time-Averaged Field: Departures from GAD? 388
5.11.2 Essential Concepts 389
5.11.2.1 Paleomagnetic Observations 389
5.11.2.2 Paleomagnetic Data Analysis 391
5.11.2.2.1 Comparing data from different locations 391
5.11.2.2.2 Measures of PSV and the TAF 391
5.11.2.3 Global Field Models: Spherical Harmonic Representation 393
5.11.3 Data Sets 393
5.11.3.1 Global Database: Paleosecular Variation from Lavas (PSVRL Database) 394
5.11.3.2 Other Global Lava Flow Data Sets 394
5.11.3.3 Recent Regional Lava Flow Compilations 396
5.11.3.4 Sedimentary Records 396
5.11.4 Paleosecular Variation 396
5.11.4.1 Early PSV Models 397
5.11.4.2 Giant Gaussian Process Models 398
5.11.5 The Time-Averaged Field 399
5.11.5.1 Early Studies 401
5.11.5.2 Longitudinal Structure in the TAF? 401
5.11.5.3 Joint Estimation of PSV and the TAF 403
5.11.6 Discussion 404
5.11.6.1 Issues in PSV and TAF Modeling 404
5.11.6.1.1 Data sets: Temporal sampling 404
5.11.6.1.2 Data sets: Spatial distribution 404
5.11.6.1.3 Data quality 405
5.11.6.1.4 Transitional data 407
5.11.6.2 Bias from Unit Vectors 407
5.11.6.2.1 Modeling approaches 408
5.11.6.3 Successes and Limitations of TAF and PSV Models 409
5.11.7 Future Directions 411
5.11.7.1 Toward New Global Data Sets 411
5.11.7.2 A New Generation of Paleomagnetic Field Modeling 412
5.11.8 Concluding Remarks 414
Acknowledgments 414
References 414
5.11.1 Introduction historically these have been the main contributing data sets.
A complete understanding of paleofield behavior requires mea-
The geomagnetic field measured at Earth’s surface includes con- surements of the full vector field – that is, both direction and
tributions from sources internal and external to the planet. The intensity – and we discuss paleointensity data in this context.
main field – that generated by magnetohydrodynamic processes (A detailed review of paleointensity data is provided in Chapter
in Earth’s liquid iron outer core – exhibits spatial and temporal 5.13.) We focus our attention on the time period 0–5 Ma. This
variations that can be observed directly today via surface, interval bridges the gap between the past few hundred (Chapter
aeromagnetic, and satellite measurements and indirectly over 5.05) or few thousand (Chapter 5.09) years for which global,
geologic timescales via remanent magnetization in crustal rocks. continuously time-varying field models can be constructed and
This chapter deals with the geometry of the geomagnetic field long paleo-timescales (tens of million years and longer, Chapter
and its temporal variability as recorded in volcanic and sedi- 5.14) for which data sets are sparse and geographic information
mentary rocks over the past few million years. The focus of the is limited. Paleomagnetic data for the past 5 Myr provide suffi-
chapter is the information provided by directional records, as cient temporal and spatial coverage to enable regional and

386 The Time-Averaged Field and Paleosecular Variation
global investigations, although we shall see that most of the physical and chemical properties over a large range of spatial
debates surrounding field behavior stem from the need for scales (e.g., Bower et al., 2011; Garnero and McNamara, 2008;
improvements and additions to current data sets. Hernlund and Jellinek, 2010; Jellinek and Manga, 2004; Lay
The characterization of magnetic field behavior on 105– and Garnero, 2011; To et al., 2011). Long-wavelength thermal
6
10 -year timescales is required to understand not only the anomalies, evidenced in seismic tomographic images of the
evolution of the geodynamo but also interactions among lower mantle, have been suggested to account for some of the
crust, mantle, and core processes. From an observational per- features observed in the historical magnetic field (e.g.,
spective, the knowledge of the temporal evolution of the mag- Bloxham et al., 1989), leading to the speculation that such
netic field globally provides essential context for features might persist over timescales consistent with mantle
understanding specific aspects of field behavior (e.g., Aubert convection. As we shall see, the question of whether spatial
et al., 2010). Examples include whether there are conditions variations in CMB conditions result in detectable signals in the
(e.g., low or variable intensity) under which magnetic reversals 0–5 Myr paleomagnetic record is an area of considerable
are more likely and how magnetic reversals relate to the overall discussion.
temporal spectrum of field variations. Paleomagnetic data play Over the past 15 years, enormous advances have been made
an important role in tectonics – from studies of local or in numerical simulations of the geodynamo (see reviews in
regional rotations to global plate reconstructions. Their useful- Busse, 2002; Kono and Roberts, 2002, Chapter 8.08; Roberts
ness depends critically on the validity of a very simple model and Glatzmaier, 2000). Such simulations have had remarkable
for the long-term average magnetic field direction. success in capturing the main qualitative features of the real field
Electromagnetic coupling between the inner core and outer (dipole-dominated, occurrence of reversals, and some aspects of
core, together with the very presence of the inner core, affects temporal variability), despite their current inability to operate in
the geometry of fluid flow in the outer core. The tangent the parameter regime applicable to the outer core. This success
cylinder – a hypothetical cylinder coaxial with Earth’s rotation has, in turn, led to the comparisons of dynamo simulations with
axis and tangent to the inner core at the equator – separates paleomagnetic observations (e.g., Aubert et al., 2010;
regions in the inner core and outer core in which the fluid flow Glatzmaier et al., 1999). Clearly, paleomagnetic data provide
and the resulting magnetic field are expected to be quite differ- important constraints for future simulations, in particular in
ent. The inner core has been suggested as the cause of some of enabling comparison of statistical aspects of real and simulated
features seen in historical field models (see, e.g., Bloxham global field behavior (Aubert et al., 2010; Hulot and Bouligand,
et al., 1989; Jackson et al., 2000) and as providing needed 2005; McMillan et al., 2001).
stability against reversals (Gubbins, 1999; Hollerbach and The temporal and spatial scales of variability in Earth’s
Jones, 1993a,b), although this is debated (Wicht, 2002). The magnetic field provide insight into the underlying magnetohy-
intersection of the tangent cylinder with the core–mantle drodynamics of the outer core and into how the dynamics of
boundary (CMB) occurs at latitudes of 70 , and an obvious that system are affected by its boundaries at the inner core and
question is whether there are observable differences in the the mantle. Determining those spatial and temporal variations
paleomagnetic field at latitudes above and below this at Earth’s requires geographically distributed data over a range of time-
surface, although sampling locations are clearly restricted. The scales. We deal here with paleomagnetic data on timescales of
presence, geometry, and strength of the field preserved in tens of thousands to millions of years; thus, we are examining
ancient (Archaean) rocks (Hale and Dunlop, 1984; Tarduno processes that have a signature at Earth’s surface over periods
et al., 2007, 2010) provide a leading-order constraint for the much longer than core overturn times. Our interest is in both
early thermal evolution of our planet, including perhaps key the time-averaged field geometry and the temporal variations
insight into the age and growth of the inner core (Labrosse, about that long-term average. The most basic approximate
2003; Lay et al., 2008; Olson and Deguen, 2012). Posing description of the field over time is that due to a geocentric
testable hypotheses regarding such earliest field behavior axial dipole (GAD): we examine temporal variations in the
requires proper characterization of ‘recent’ paleomagnetic field (PSV) and long-term time-averaged departures in geome-
field behavior (past few million years). For example, several try of the field from GAD (the time-averaged field or TAF).
studies have suggested that the onset of inner core growth The chapter is organized as follows: In the remainder of the
might be diagnosed geomagnetically via a low-intensity field introduction, we provide further background and motivation,
(Valet, 2003). However, what constitutes a ‘weak’ field requires specifically for studies of PSV and the TAF. We then outline
an understanding of the mean global field strength over a some of the essential concepts that are needed to navigate the
range timescales, and this is a topic of intense debate (Chapter material that follows and the relevant literature
5.13). (Section 5.11.2). In Section 5.11.3, we review global data
Over timescales of millions of years and longer, geomag- sets that have been used in global and regional TAF and PSV
netic field spatial and temporal variability may reflect not only studies. In Sections 5.11.4 and 5.11.5, we summarize global
the presence of the inner core but also the nature of CMB models for PSV and the TAF. The review is not exhaustive; we
thermal (Bloxham and Jackson, 1990; Driscoll and Olson, highlight contributions that exemplify the major viewpoints of
2011; Gubbins, 1988; Gubbins and Richards, 1986; Gubbins the community as they evolved. We analyze the important
et al., 2007; Merrill et al., 1990; Olson et al., 2010, 2012, 2013; issues related to data sets and modeling approaches that have
Sreenivasan and Gubbins, 2011; Willis et al., 2007) and elec- led to differing conclusions in the literature (Section 5.11.6).
tromagnetic (Clement, 1991; Costin and Buffett, 2005; Laj Several reviews of the TAF have been published previously
et al., 1991; Runcorn, 1992) coupling. There is growing evi- (McElhinny et al., 1996a; Merrill and McFadden, 2003;
dence that the lowermost mantle is laterally heterogeneous in Merrill et al., 1996), but the concluding remarks in these
The Time-Averaged Field and Paleosecular Variation 387
reviews represent one interpretation of existing models. We distinguish temporal variations during stable polarity periods
discuss the current status of both modeling and database from geomagnetic reversals.
efforts and, in conclusion (Section 5.11.7), offer some PSV is studied via time series of field variations or via
thoughts on future directions. statistical descriptions of paleomagnetic records with incom-
plete age control. These provide quite different measures of
temporal variations. Secular variation *
can be defined as the
5.11.1.1 The GAD Approximation
rate at which the field changes, @t B (Courtillot and Valet,
The geomagnetic field at Earth’s surface can be approximated 1995; Love, 2000), and time series of paleomagnetic observa-
by that due to a dipole at Earth’s center and aligned with the tions allow this to be assessed. Relative paleointensity and
rotation axis. The instantaneous field shows significant depar- direction from long sediment cores are candidates for time
tures from this GAD approximation (e.g., the present magnetic series analyses, especially the use of spectral estimation tech-
north is not colocated with the geographic north); however, niques. However, even when such data sets exist, the charac-
over geologic timescales, it might be expected that temporal terization of PSV is far from straightforward due to issues such
variability in the field means that such departures will average as age calibration, smoothing of the magnetic field during
to zero. Using the statistical methods developed by Fisher remanence acquisition, and the comparison of data from
(1953), Hospers (1954) showed that, averaged over several different geographic locations. In contrast to sedimentary
thousand years, GAD provides a good approximation to the records, lava flows provide geologically instantaneous record-
observed magnetic field. Opdyke and Henry (1969) used ings of the paleofield, but the data distribution is determined
observations of paleomagnetic directions from deep-sea piston by the occurrence of volcanism and the present-day accessi-
cores to demonstrate that the GAD approximation has held bility of flows. Radiometric ages are typically available
over the past 2.5 Myr (Figure 1). It is now well known that for only a small percentage of flows sampled for paleomag-
Earth’s field can be described to the first order by GAD, ori- netic purposes. Consequently, statistical methods must be
ented in either its current (normal) or reverse configuration. employed to assess PSV, via, for example, the variance in
The GAD approximation has been heavily exploited in tectonic directions and/or intensity. These summary statistics provide
studies and is central to models for plate reconstruction. a measure of PSV, but no rate information. (In cases where
sequences of flows are available, the correlations between
vectors from stratigraphically adjacent flows can provide
5.11.1.2 Paleosecular Variation
some rate information; Love, 2000.) Confusingly, the paleo-
The magnetic field generated in the outer core exhibits vari- magnetic literature often refers to analyses based on time
ability on all timescales. Paleosecular variation (PSV) describes series from sedimentary records as ‘PSV’ studies and analyses
the temporal variations in the paleomagnetic field, for which based on spot recordings from volcanic rocks as ‘PSVL’ studies
observations are obtained at Earth’s surface. While the upper (paleosecular variation from lavas).
bound on timescales associated with PSV is clearly the age of Here, we use the term PSV to describe temporal variations
the geomagnetic field itself, the term ‘PSV’ is often used in the in the field manifest by the magnetic field vector preserved in
context of characterizing field variations during stable polarity the rock record. We examine directional and intensity records
periods. Furthermore, the use of the term PSV usually carries (ideally in the full vector field at any location) from lavas or
the implicit assumption that only variations in direction will sediments. Of particular interest are geographic variations in
be discussed. These rather arbitrary (and limiting) separations PSV: Are differences recorded inside and outside the tangent
of timescale and components of the field vector arise from the cylinder, and are there longitudinal and latitudinal variations?
historical availability of specific data types and a desire to Also important to understanding PSV is the issue of how
–80⬚ –80⬚
Brunhes Matuyama
–60⬚ –60⬚
–40⬚ –40⬚
Mean inclination
Mean inclination
–20⬚ –20⬚
0⬚ 0⬚
20⬚ 20⬚
40⬚ 40⬚
60⬚ 60⬚
80⬚ 80⬚
–80⬚ –60⬚ –40⬚ –20⬚ 0⬚ 20⬚ 40⬚ 60⬚ 80⬚ –80⬚ –60⬚ –40⬚ –20⬚ 0⬚ 20⬚ 40⬚ 60⬚ 80⬚
Latitude Latitude
Figure 1 Inclination data from the deep-sea sediment cores of Opdyke and Henry (1969) (filled circles) for the Brunhes (left) and Matuyama (right)
epochs. Solid line is the inclination predicted by a geocentric axial dipole (GAD) field.
variations in direction are related to variations in intensity 180

(e.g., Love, 2000).
Analyses of the paleomagnetic power spectrum (Barton,
1982; Constable and Johnson, 2005; Ziegler et al., 2011) indi-
cate a continuum of temporal variations, but of interest histor-
ically has been the question of whether one can usefully
characterize PSV over a specified timescale. Barton (1982)
used sediment records to suggest that timescales of at least
103 years provided reasonable estimates of PSV; more recent
studies suggest at least 104–105 years (Carlut et al., 1999; ⬚N
80
Merrill and McFadden, 2003). One of the major difficulties
in understanding the temporal and spatial scales of PSV is the
⬚N
limited and uneven distribution of paleomagnetic data. 70
(a) 0
5.11.1.3 The Time-Averaged Field: Departures from GAD?

–80⬚
While the GAD approximation describes most of the structure
in the geomagnetic field, it has been known for some time –60⬚
(Wilson, 1970, 1971) that the paleomagnetic record displays –40⬚
Mean inclination
small but persistent departures from GAD. Records of
paleodirection from Europe and Asia were found to have –20⬚
time-averaged pole positions that showed an offset from the 0⬚
geographic north, specifically poles that were far-sided (on
the other side of the geographic north relative to the observa- 20⬚
tion location) and right-handed (eastward of the observation 40⬚
location for normal polarities). Far-sided pole positions result
from time-averaged values of inclination that have a small 60⬚
negative bias compared with the GAD prediction (Figure 2). 80⬚
Wilson (1970, 1971) interpreted the deviations from GAD as
–80⬚ –60⬚ –40⬚ –20⬚ 0⬚ 20⬚ 40⬚ 60⬚ 80⬚
due to a dipole offset northward along the rotation axis,
although this is of course a nonunique solution. (b) Latitude
Departures of the time-averaged geomagnetic field (TAF) Figure 2 (a) Stable polarity continental sediment and lava flow virtual
from GAD are a topic of this chapter. An issue that becomes geomagnetic poles (VGPs) from 72 sites for the past 25 Myr (data are the
immediately apparent is whether in fact there is a stable TAF alternating field demagnetized data of Wilson, 1971). Center of the
geometry, in other words whether the process by which the projection is the geographic pole; contours are VGP latitudes of 80 and
magnetic field is generated is statistically stationary. Even if this 70 . VGPs are plotted as if all the sites were on a common site longitude,
is not the case, the question remains as to over what intervals taken here as 0 . VGPs are seen to plot on the far side of geographic
north, compared with the common observation site longitude. The so-
the field should be averaged in order to provide useful
called right-handed effect is also seen – VGPs appear to plot on average
information.
eastward of 180 . (b) Upper tertiary inclination data from Wilson (1970)
Over the historical period, 1590–1990 AD, satellite, obser- (solid circles). Solid line is the predicted inclination as a function of
vatory, and survey measurements of the geomagnetic field have latitude from a GAD model. The dashed line is the predicted inclination for
permitted the construction of a spatially detailed, temporally a dipole with an offset r ¼ 306 km (Wilson, 1970). The observed
varying magnetic field model, GUFM1 ( Jackson et al., 2000; inclinations are systematically shallower than those predicted by GAD
and see also, e.g., Bloxham and Jackson, 1992; Bloxham et al., and are better matched by the offset dipole model.
1989). A representation of the TAF structure can be obtained by
averaging the model over its respective time interval. The radial
component of the magnetic field, Br, is shown in Figure 3 after flux patches that are pronounced in the time-varying version of
downward continuation to the surface of Earth’s core under the GUFM1 and in modern satellite models (e.g., Hulot et al.,
assumption that the mantle is an insulator. The field due to 2002) and that appear to propagate westward in the Atlantic
GAD is also shown and displays only latitudinal structure. hemisphere are attenuated in the 400-year temporal average.
GUFM1 has significant non-GAD structure, which has been Overturn times for fluid motions in the outer core are on the
extensively discussed elsewhere ( Jackson et al., 2000), and is order of a few hundred years, and thus, the persistence of
thought to be influenced by the presence of the inner core and features over 400 years in GUFM1 suggests that magnetic
by lateral heterogeneity in the lowermost mantle. Regions of field generation in the outer core may be influenced by the
increased radial flux at high latitudes, commonly referred to as inner core and outer core boundaries. Of interest in the context
flux lobes, have persisted in much the same locations for 400 of other areas of deep-Earth geophysics have been the sugges-
years. Low radial field over the North Pole has been interpreted tions of inner core/outer core electromagnetic and core–
as a manifestation of magnetic thermal winds and polar vorti- mantle thermal coupling (Bloxham et al., 1989; Jackson
ces within the tangent cylinder (Hulot et al., 2002; Olson and et al., 2000). If conditions at the CMB are important, then
Aurnou, 1999; Sreenivasan and Jones, 2005, 2006). Equatorial the timescales associated with these will be those of mantle
1590 – 1990 time averaged field
400
300
200
100
0
–100
–200
–300
(a)
–400
Field due to a geocentric axial dipole
400
300
200
100
0
–100
–200
–300
(b)
–400
Figure 3 Radial field at the core–mantle boundary (CMB) in millitesla; Hammer–Aitoff projection. (a) Model GUFM1 ( Jackson et al., 2000), (b) field due
to a GAD – note that this possesses only latitudinal structure in contrast to the time-averaged historical field in (a).
convection, and some of the steady features in GUFM1 might X north (geographic)
be expected to persist over the past few million years, albeit
damped by PSV.
D
5.11.2 Essential Concepts I Bh

Y
In this section, we cover some of the essential concepts needed East
to navigate the PSV and TAF literature. Some are soundly based Z
on mathematical formulation; others are (for better or worse) Down
conventions that have been adopted to assist the analysis of
paleomagnetic data. B
5.11.2.1 Paleomagnetic Observations Figure 4 Geomagnetic field elements – local orthogonal components
north, east and down, and paleomagnetic measurements of
Paleomagnetic observations comprise records of ancient field
declination, D and inclination, I.
direction and intensity recorded in sedimentary and igneous
rocks. Paleodirections are specified by declination, D, and !

inclination, I (Figure 4). Declination is the angle between the 1 Y 1 Z
field vectors projected onto the horizontal plane and geo- D ¼ tan , I ¼ tan 1=2 [1]
X X2 + Y 2
graphic north. Inclination is the dip of the magnetic field
vector from the local horizontal. Paleodirections are related Ideally, one would like a measure of the full vector field;
to the local north (X), east (Y), and down (Z) magnetic field however, a consequence of practical difficulties inherent to
elements as follows: obtaining intensity measurements (Chapter 5.13) is that current
ODP 983 record: (60.4⬚ N, 23.6 ⬚W)
Relative paleointensity
16
12
8
4
(b)
Declination (⬚)
100
–100
(c)
80
Inclination (⬚)
40
0
–40
–80
(d)
80
VGP latitude (⬚)
40
0
–40
–80
0 100 200 300
(a) (e) Age (ka)
Figure 5 Example of sediment record (a) core photo, (b)–(e) time series of observations. In (d), solid red line is the inclination predicted by GAD.
The large variability in declination compared with inclination is due to the high site latitude. Blue dashed line (e) is at a VGP latitude of 45 ; several places
in the time series indicate directions with VGP latitudes lower than 45 – whether these are excursions or part of typical stable polarity paleosecular
variation (PSV) is not immediately obvious. A reversed polarity direction is clearly recorded at 180 ka. The total VGP dispersion, ST (eqn [8]),
calculated for directions with VGP latitudes greater than 45 is ST ¼ 16.6 . Data provided courtesy of J. Channell and reported in Channell et al. (1997).
data sets spanning the past 5 Myr are dominated by paleodirec- carrier present. These relative paleointensity data are easier to
tions. Lava flow data are (D, I) pairs, but lack of azimuthal acquire, although a whole field of study surrounds the choice of
orientation information means that often only I measurements normalizer (see Chapter 5.13) and the issue of how best to
are available for deep-sea sedimentary cores. The former provide calibrate relative to absolute paleointensity has not yet been
intermittent or spot readings of the paleofield, while the latter satisfactorily explored.
provide continuous time series of observations. Figures 5 and 6 Details of paleomagnetic sampling techniques are covered
show typical representations of each of these data types; specific in Chapter 5.04. Typically, samples are drilled in the field
data sets are discussed in Section 5.11.3. from individual lava flow or sedimentary units. Each flow or
Néel theory (1949) suggests that under certain conditions, the sedimentary unit allows an estimate to be made of the paleo-
thermal remanent magnetization (TRM) acquired by lava flows field at a specific location at the time of cooling (lavas) or
can be reproduced in the laboratory, allowing measurements of compaction (sediments). This is known as a site. At a given
absolute paleointensity. In practice, such data are more difficult site, multiple samples are collected to allow averaging of
to acquire than paleodirections due to alteration of the sample directions to reduce the influence of measurement error, in
during the laboratory procedure. In sedimentary rocks, rema- particular orientation error. These samples are demagnetized
nence is acquired during deposition and compaction, and unlike in the laboratory to remove (unwanted) secondary rema-
the case for lava flows, no theoretical basis exists for absolute nence. Important issues for the use of paleomagnetic data in
paleointensity measurements. Instead, one measures the natural field modeling studies are whether site-level measurement
remanent magnetization of the sample and normalizes by a noise can be adequately assessed and whether overprints
proxy that can account for the amount and type of magnetic have been removed.
North
Figure 6 Lava sequence from the Southwest Rift zone of Kilauea Volcano, Hawaii. Paleomagnetic directions shown in the equal-area figure are
samples taken from sites surrounding Kilauea with ages from the mid-Pleistocene to the Quaternary. (Data are a subset of the compilation in
Lawrence et al., 2006.) The direction predicted by GAD is shown in red. Paleomagnetic directions are specified by declination (azimuth clockwise from
North) and inclination (radial distance, where the center of the figure is I ¼ 90 and the circumference is I ¼ 0 ). Photograph courtesy of Roi Granot.
5.11.2.2 Paleomagnetic Data Analysis reverse polarity data. Some directions are quite far from the
GAD prediction. The VGP equal-area projection gives VGP
5.11.2.2.1 Comparing data from different locations
longitude (azimuth) and latitude (distance from center, with
For a GAD field, inclination and intensity of the field will vary
center being 90 ). Several directions in this data set would be
with latitude (l), with inclination predicted by
regarded as transitional (VGP latitudes less than 45 ) and are
tan I ¼ 2tan l [2] also seen to be quite anomalous in the D0 , I0 representation.
Note that the shape of the distributions in all three figures
A standard approach in paleomagnetism is to calculate appears different because of the nonlinear mappings
the equivalent geocentric (but not axial) dipole that would involved in converting D, I to VGP or to D0 , I0 .
give rise to the observed site-level paleodirection. The pole
position of this equivalent dipole is known as a ‘virtual
geomagnetic pole’ or VGP and, under the assumption of a 5.11.2.2.2 Measures of PSV and the TAF
geocentric dipole, is independent of the site location. An Global studies of PSV and the TAF have used data sets for
alternative approach, proposed by Hoffman (1984), but which temporal control is incomplete, and so summary statis-
used less frequently is to compute the direction, D0 , I0 , rela- tics are needed to characterize the field. Given a collection of N
tive to the expected GAD direction at the site. An example of sites, which span a sufficiently long time period to enable an
a data set represented by D, I pairs, D0 , I0 pairs, and VGP estimate of PSV and the TAF, the TAF can be described via the
positions is shown in Figure 7. D, I pairs from 115 normal mean direction at a given location, and PSV via the variance
polarity and 60 reverse polarity lavas from Hawaii are shown (dispersion) in the field.
(there is no significance to the choice of data) on an equal- The paleodirection specified by a (D, I) pair can be
area projection. The closed (open) circles represent projec- expressed as direction cosines:
tions onto the lower (upper) hemisphere and for northern
hemisphere sites represent normal (reverse) polarities. The x ¼ cos D cos I; y ¼ sin D cos I; z ¼ sin I [3]
GAD-predicted direction at the mean site location is given
Given a set of N pairs of D and I measurements, the unit
for normal (reverse) polarities by the filled (open) red tri-
vector mean direction has direction cosines:
angles. Azimuth on the figure represents declination, mea-
sured clockwise from the north; distance from the center of 1X N
1X N
1X N
the figure gives inclination, where the center of the figure is X¼ xi ; Y ¼ yi ; Z ¼ zi [4]
R i¼1 R i¼1 R i¼1
I ¼ 90 . On average, the directions are shallower than GAD (a
negative inclination anomaly), most obvious visually in the R is the vector sum of the individual unit vectors, given by
North
D, I
VGP D’, I’
Figure 7 Equal area projections of a data set from Hawaii (see text for description) to show (D, I) projection (top figure) with GAD prediction shown in
red. Representation in terms of VGP positions (lower left), and representation in (D0 , I0 ) coordinates. Directions that have large angular deviations
from GAD can be seen in the top figure and are manifest as VGPs with latitudes less than 45 (dashed circle) and (D0 , I0 ) coordinates that plot
far from the center of the (D0 , I0 ) figure. Solid (open) circles represent normal (reverse) directions.
" !2 !2 !2 #1=2 !1=2

X
N X
N X
N 1 X N
R¼ xi + yi + zi [5] S¼ Y2 [7]
N 1 i¼1 i
i¼1 i¼1 i¼1
X, Y, and Z from eqn [4] can be used in eqn [1] to estimate where Yi is the angle between the ith direction and the sample
the mean declination, D, and inclination, I, for the set of N mean and again N is the number of sites. Approaches vary
directions. Interest in the TAF stems from the desire to quantify among studies: Sometimes, the scatter about the mean VGP
departures from GAD (if any). A TAF direction is usually (computed from the data themselves) is calculated; often, the
expressed in terms of its deviation from that predicted at the scatter is calculated assuming that the mean VGP coincides
site location by GAD, that is, in terms of the inclination anom- with the geographic north (the GAD hypothesis). This differ-
aly (DI) and declination anomaly (DD), where ence in approach is in fact significant and the choice is critical
in the development of models of PSV (McFadden et al., 1988).
DI ¼ I IGAD DD ¼ D [6]
The scatter, S, calculated as in the preceding text, can be
IGAD can be calculated from eqn [2] and DGAD ¼ 0. assessed at the site level in which case it measures within-site
The nonlinear mapping between the field direction (D,I) at error (SW). If the site represents a single instant in time, then SW
a particular site and the latitude, l, and longitude, f of the VGP should represent only our own measurement errors plus errors
position, means that the scatter caused by variations in the caused by imperfection in the original recording mechanism of
field manifests itself differently in the VGP frame of reference the rock; it should not include any variation to changes in the
from the (D,I) frame of reference (Figure 7). Consequently, a geomagnetic field itself. The scatter ST, again calculated as in
decision has to be made as to which frame of reference is to be the preceding text, but using the individual site means from
used. Historically, discussions of PSV have used the VGP frame several sites in a PSV study as the observations, then gives the
of reference. A mean VGP position can be calculated exactly as total scatter; this is a combination of the within-site scatter and
for the mean direction using eqns [1]–[5] and the procedure the scatter caused by the variation of the geomagnetic field
described in the preceding text. An angular standard deviation, from site to site, referred to as the between-site dispersion, SB.
S, can be defined as This between-site scatter is a measure of PSV and is given by
S2W mean values for gm m

l and hl over time intervals of interest – TAF
S2B ¼ S2T [8]
n models – and (ii) the specification of the variance in the
Gaussian coefficients a priori and forward modeling of sum-
where n is the average number of samples measured at each
mary statistics (e.g., SB) for PSV.
site. Equation [8] can be modified to account explicitly for
unequal numbers of samples at each site:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!
u 5.11.3 Data Sets
u 1 X N
S2Wi
SB ¼ t Yi 2
[9]
N 1 i¼1 ni In this section, we review paleodirectional data sets that were
designed to be suitable for global and/or regional TAF and PSV
SWi is the within-site dispersion determined from the ni sam- studies. We document compilations spanning at least the
ples at the ith site. Brunhes normal polarity chron (past 780 kyr) and as long as
Alternative statistics used to describe PSV include standard the past 5 Myr. We restrict the review to data sets that have been
deviation in inclination/declination (sI, sD, respectively) and used extensively in the global field modeling that is reviewed in
the root-mean-square angular deviation of paleodirections Sections 5.11.4 and 5.11.5. However, we note that substantial
(Sd). Site-level intensity measurements are usually converted efforts to improve the quality and distribution of such data
to the equivalent virtual axial dipole moment (VADM; see have been under way over the past 10–15 years and the collec-
Chapter 5.13), and variations in intensity are quantified by, tive results of these efforts are summarized in Section 5.11.7.
for example, the standard deviation in VADMs. As much of the discussion surrounding TAF and PSV modeling
is related to issues of data distribution and quality, we summa-
rize the selection criteria used in each study. We mention
5.11.2.3 Global Field Models: Spherical Harmonic
paleointensity data sets briefly – the reader is referred to
Representation
Chapter 5.13 for further details – but we note that good-
Models of the geomagnetic field can be constructed using a quality paleointensity data will provide important constraints
spherical harmonic representation. The magnetic scalar poten- in future PSV and TAF research.
tial in a source-free region due to an internal field obeys For paleomagnetic data to be useful for TAF or PSV studies,
Laplace’s equation and can be written as the following criteria need to be met:
X 1 X l l + 1 1. The sampling sites should not have been subjected to tec-

a
Cðr, y, f, t Þ ¼ a l ðt Þcosmf
gm tonic effects since the acquisition of magnetic remanence.
l¼1 m¼0
r
2. A sufficient number of temporally independent sites, N,
+ hm
l ðt ÞsinmfÞP l ðcos yÞ
m
[10] need to have been sampled, covering a time period of at
least 104 years (and preferably longer), to minimize bias in
where gm m
l (t) and hl (t) are the Schmidt partially normalized
estimates of PSV and the TAF.
Gaussian coefficients at a time t; a is the radius of the Earth; r,
3. Multiple samples per site (n) should have been taken to
y, and f are radius, colatitude, and longitude, respectively; and
allow assessment of within-site error.
Pm
l are the partially normalized Schmidt functions. l is the
4. The remanence should have been established as primary via
spherical harmonic degree and m is the spherical harmonic
! laboratory cleaning (demagnetization).
order. The magnetic field, B , is the negative gradient of the
potential C, so Because of the historical desire to examine PSV during
stable polarity periods as a distinct entity, separate from rever-
@C 1 @C 1 @C
Br ¼ By ¼ Bf ¼ [11] sals, transitional data (those that deviate greatly from GAD)
@r r @y rsin y @f
have been excluded from PSV and TAF studies. The measure
For paleomagnetic purposes (where we assume a spherical used to assess whether a paleodirection is transitional is the
reference surface), the relationship between the local coordi- VGP latitude; various choices for a VGP latitude that discrim-
nate system in Figure 4 and the spherical coordinate system inates between these two states have been proposed
is Br ¼ Z, By ¼ X, and Bf ¼ Y. (McElhinny and McFadden, 1997; Vandamme, 1994), but
A given field model is specified by the Gaussian coefficients whatever the choice, the distinction is artificial and purely for
gml (t) and hl (t). The m ¼ 0 terms correspond to spherical har-
m
convenience.
monic functions with no azimuthal structure – that is, they are We review data sets based on paleodirections from lavas
axially symmetrical or zonal. Such fields have predicted decli- and sediments separately. Lava flows offer the advantage of
nations of zero. g01, g02, and g03 are the Gaussian coefficients declination and inclination observations, but temporal infor-
representing the axial dipole, axial quadrupole, and axial octu- mation is poor in older collections (e.g., often, only the polar-
pole terms, respectively. ity chron is known, and there is little information on the time
Paleomagnetic observations, D, I, and intensity, are all non- interval spanned by a series of sites at a given location). Deep-
linearly related to the Gaussian coefficients, and because of sea sediment cores provide sampling in the ocean basins, but
limited temporal information, it is not currently possible to the lack of declination information means that in practice,
determine the variation of these coefficients as a function of only latitudinal variations in the TAF and PSV can be assessed
time. Instead, paleomagnetic field modeling to date has typi- with these data. Furthermore, the deep-sea sediment cores
cally involved two approaches: (i) the use of mean field direc- considered in the succeeding text are piston cores with no
tions along with parameter estimation or inversion to estimate independent orientation information. The advantage offered
by these data though is a better temporal average due to to all sites; this DMAG code varies between 1 (no cleaning) and
smoothing during remanence acquisition and the possibility 4 (modern lab methods, typically employed in most studies
of estimating the time interval spanned by a core of a given from the early 1990s onward). At a given location, five sites
length if sedimentation rates are known. (N ¼ 5) were considered sufficient to estimate PSV and the TAF,
and stable polarity data were defined as those with VGP latitudes
greater than 45 .
5.11.3.1 Global Database: Paleosecular Variation The geographic distributions of normal (0–5 Ma) and
from Lavas (PSVRL Database) reverse (0–5 Ma) data are shown in Figure 8. MM97 argue
To date, the most comprehensive lava flow database assembled that because modern magnetic cleaning methods are required
for PSV studies is that of McElhinny and McFadden (1997), to remove secondary overprints, only the DMAG ¼ 4 data are
hereafter MM97. It comprises data from 3719 lava flows and suitable for TAF and PSV modeling – these data comprise less
thin dikes covering the period 0–5 Ma. The restriction to lava than 12% of the database and are from only eight distinct
flows and dikes ensures that data are from igneous units that locations (Figure 8). As part of their assessment, MM97 listed
cooled quickly and for which the paleodirection is an instan- many previous studies that should be replaced by new data,
taneous recording of the field. subjected to current lab protocols for demagnetization; this list
This database, known as the PSVRL database, grew out of a has led to many of the recent paleomagnetic sampling efforts.
series of IAGA-sponsored databases (Lock and McElhinny,
1991; McElhinny and Lock, 1993, 1996) and the preceding
5.11.3.2 Other Global Lava Flow Data Sets
data sets, the earliest of which was that of McElhinny and
Merrill (1975). The goal of this database was to be as inclusive Several other lava flow data sets have been assembled for TAF
as possible and so only two samples per site (n ¼ 2) were and PSV modeling. We summarize two here, those of
required, and data subjected to any level of laboratory demag- Quidelleur et al. (1994) and Johnson and Constable (1996),
netization were accepted. A demagnetization code was assigned hereafter Q94 and JC96, respectively. Together with the PSVRL
Normal polarity data, GPMDB
Reverse polarity data, GPMDB
Figure 8 Geographic distribution of normal and reverse polarity sites in the PSVRL database (McElhinny and McFadden, 1997). All data (black
triangles), DMAG 4 data only (red stars). Note that there are very few distinct locations (eight normal polarity, six reverse polarity) with data that meet
current laboratory protocols.
database, these two data sets have formed the basis for most of n1
k¼ [12]
the TAF and PSV studies over the past 20 years. Both were built nR
on the compilation by Lee (1983), adding newer references
from the published literature. where n is the number of samples at the site and R is defined in
Q94 and JC96 contain many data in common; over 50% eqn [5]. The precision parameter is the analogue, for vectors on
of the data in JC96 are included in Q94. Q94 includes more a unit sphere, of the inverse of the variance of a two-
data than JC96, as criteria associated with the number of sam- dimensional normal distribution. Other studies use the size
ples per site, n, and the cutoff VGP latitude for transitional data of the 95% cone of confidence, a95:
were less stringent. The number of samples per site needed to
140
assess within-site error has been a subject of debate. In the a95 pffiffiffiffiffi [13]
absence of systematic noise, the uncertainty in the site mean kn
pffiffiffi
direction should decrease as 1= n, and the estimation of a We do not belabor differences among the Q94 and JC96
variance requires at least three samples per site. JC96 required data sets as the collective contribution of sampling efforts by
n ¼ 3, and this results in the exclusion of many sites, notably the paleomagnetic community over the past 20 years is now
large numbers of data from Iceland and other strategic loca- resulting in global data sets that supersede these existing com-
tions in the context of global data coverage. For a site with pilations in number and quality (see Section 5.11.6).
n > ncutoff, choices as to what constitutes large within-site error The data set of Q94 comprises 3179 lava flows, while that
also differ among studies. Measures of within-site error usually of JC96, 2187 records. The number of distinct locations
assume that paleofield directions at a given location can be reported in each study is 86 (Q94) and 104 (JC96), although
represented by a Fisher distribution. (The Fisher distribution is about 50 of those in JC96 differ by less than 1 spatially. Both
the analogue, for vectors on a unit sphere, of the two-dimen- data sets contain about twice as many normal polarity records
sional normal distribution; Fisher, 1953). Some studies define as reverse records, and the data sets are dominated by
a within-site error requirement based on k, the best estimate of Brunhes-age paleodirections. The spatial and temporal distri-
the Fisher precision parameter, k (Fisher, 1953): butions of JC96 are shown in Figures 9 and 10; those for Q94
Brunhes and 0–5 Ma normal polarity data, JC96
0–5 Ma reverse polarity data, JC96
Figure 9 Geographic data distribution of 0–5 Ma lava flow data set of Johnson and Constable (1996). Upper (lower) figure shows 0–5 Ma normal
(reverse) polarity sites (black triangles); blue stars are Brunhes-age normal polarity sites.
1200
80 B M
1000
Number sites
Number sites
60
800
600
40
400
20
200
B M Ga Gi
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Age (My) Age (My)
Figure 10 Age distribution of JC96 lava flow data (left) and SK90 sediment data (right). Polarity chrons are delineated by vertical dashed lines:
B, Brunhes; M, Matuyama; Ga, Gauss; Gi, Gilbert.
are not shown but are similar. Northern hemisphere data however, as shown in Johnson and Constable (1997), this
coverage of both JC96 and Q94 is reasonable, especially for does not reflect the strong regional variability in inclination
the 0–5 Ma normal polarity data combined. Reverse polarity estimates. For lava flow data, this variability might be caused
data provide poorer spatial coverage. The coverage of the by inadequate temporal sampling at one or more sites;
southern hemisphere is limited, because the data sets only however, sediment cores should provide a good TAF direction
include land-based observations. because the record is almost continuous and a large number of
sample directions are averaged to give an inclination anomaly
for each core. Regional variability in inclination anomalies
5.11.3.3 Recent Regional Lava Flow Compilations recorded by sediments suggests inconsistency among the
Several recent studies have combined new paleomagnetic data observations; this can be caused by inadequate demagnetiza-
with ‘legacy’ data (i.e., data sets already included in, e.g., the tion of some cores and by nonvertical coring. Cores from the
PSVRL database) to produce large regional data sets. Six com- North Pacific region appear to be particularly inconsistent
pilations published to date are for paleomagnetic directions ( Johnson and Constable, 1997) – these cores were some of
from the NW and SW United States (Tauxe et al., 2003, the first piston cores to be collected and were demagnetized in
2004a,b, respectively), Mexico (Lawrence et al., 2006; Mejia low alternating fields (5–15 mT compared with 15–40 mT for
et al., 2005), Hawaii, the South Pacific, and Reunion some later cores [SK90]) – and may contain viscous overprints.
(Lawrence et al., 2006). These data sets, combined with other Inclination records provided by more recent, very long drill
studies, will provide greatly improved lava flow data sets for cores (such as those of the Ocean Drilling Program (ODP) and
TAF and PSV studies, and we return to this in Section 5.11.7. Deep Sea Drilling Program) will contribute greatly to PSV and
TAF studies in the future (Section 5.11.7). For example,
15 cores worldwide contain data that span several kyr or
5.11.3.4 Sedimentary Records more in the period 0–2 Myr (Valet et al., 2005) provide not
only relative paleointensity records but also inclination data.
Deep-sea sediment cores fill a large gap in the geographic
In some cases, cores or core sections are sufficiently well ori-
distribution of PSV/TAF data. For deep-sea cores, the time
ented to provide absolute declination.
interval spanned by given core can be estimated via mean
sedimentation rates, and so the time averaging at each location
is better controlled than for lava flows. The sedimentation rate,
along with type of laboratory methods used (in particular 5.11.4 Paleosecular Variation
whether pass-through magnetometers are used), determines
the along-core sampling interval for which adjacent samples The study of PSV has been a major part of paleomagnetism over
can be considered temporally independent. the past four decades. In some cases, it is possible to examine
The first compilation of deep-sea sediment cores was the continuous sedimentary records using traditional time series
piston core data set used by Opdyke and Henry (1969) to analysis approaches (for an early paper see, e.g., Creer, 1983).
demonstrate the validity of the GAD approximation over the More commonly, analyses rely heavily on statistical approaches
past 2.5 Myr. This was expanded and improved by Schneider where the variance in the field is quantified. The two most
and Kent (1988, 1990), with the goal of investigating small, commonly used descriptions are the angular dispersion in
but long-lived departures from GAD. The data set of Schneider field directions, SD, or the angular dispersion in the VGPs, SB
and Kent (1990), hereafter SK90, comprises piston cores from (eqn [8]). Globally, VGP dispersion has been observed to
the low midlatitudes (Figure 11) and contains 176 cores with increase with latitude (see red symbols in Figure 12), and the
Brunhes data and 125 cores with Matuyama data. The standard form of this increase has motivated many of the PSV models to
error in inclination for each core is typically on the order of 1 ; date. Regionally, one of the earliest-noted signatures of PSV was
Brunhes polarity piston cores
Matuyama polarity piston cores
Figure 11 Geographic distribution of Brunhes and Matuyama piston cores in the Schneider and Kent (1990) data set.
low SB at Hawaii, as reported by Doell and Cox (1963, 1965). empirical observation that most of the power in the paleofield
This observation was interpreted as low variability in the non- observed at Earth’s surface can be accounted for by a geocentric
dipole contributions to the field and, combined with observa- dipolar field structure, while later models are based on statis-
tions from the historical record (Bloxham and Gubbins, 1985), tical properties of the field.
led to the suggestion of persistent low nondipole fields in the
Pacific: the so-called Pacific dipole window.
5.11.4.1 Early PSV Models
The statistical approach inherent in using SB, or some other
summary statistics for field variability, circumvents the lack of The earliest global PSV models considered dipole wobble
detailed age control and the absence of time series of observa- (model A: Irving and Ward, 1964 and model B: Creer, 1962;
tions that are inevitable in working with lava flows from dis- Creer et al., 1959) and were followed by a suite of models that
parate locations. An additional view, continued until recently, considered both dipole wobble and nondipole variations
was the idea of characterizing ‘typical’ PSV as distinct from (model C: Cox, 1962, model D: Cox, 1970, model E: Baag
reversals or excursions. and Helsley, 1974, model M: McElhinny and Merrill, 1975)
Early models for PSV attributed secular variation to three and its modification (Harrison, 1980, and model F: McFadden
sources: variations in the direction of the dipole (dipole wob- and McElhinny, 1984). For a thorough review of these studies,
ble), variations in the intensity of the dipole wobble, and the reader is referred to the discussion in Merrill et al. (1996).
variations in the nondipole field. Later statistical descriptions McFadden et al. (1988) proposed a different representation
have used present-day properties of the field to constrain PSV of PSV, in which they introduced the idea of separating contri-
models; for example, McFadden et al. (1988) used the present- butions to the variance in the field into two parts – the dipole
day field to establish the general form of a VGP dispersion and quadrupole families. This provided a direct link to the
curve, while Constable and Parker (1988) used the present- spherical harmonic description of the field as the dipole family
day power spectrum as a constraint on the paleo-power spec- comprises terms for which l–m is odd (i.e., asymmetrical about
trum. It is important to note that none of these models are the equator) and the quadrupole family comprises terms for
based on any physical theory: Earlier models were based on the which l–m is even (symmetrical about the equator). The VGP
Simulations from CP88

1.0 30
0.8 25
ST (degrees)
20
0.6
Cdf
15
0.4
10
0.2 5
0.0 0
0 10 20 30 –80 –40 0 40 80
(a) Virtual dipole moment Latitude (⬚)
Simulations from CJ98

1.0 30
0.8 25
ST (degrees)
20
0.6
Cdf
15
0.4
10
0.2 5
0.0 0
0 10 20 30 –80 –40 0 40 80
(b) Virtual dipole moment Latitude (⬚)
Simulations from CJ98.nz

1.0 30
0.8 25
ST (degrees)
20
0.6
Cdf
15
0.4
10
0.2 5
0.0 0
0 10 20 30 –80 –40 0 40 80
(c) Virtual dipole moment Latitude (⬚)
Figure 12 Comparisons of predictions of three statistical paleosecular variation models with 0–5 Ma paleointensity data of Tanaka et al. (1995) and the
0–5 Ma directional data JC96. (a) Model CP88, (b) Model CJ98, (c) Model CJ98.nz. The left panels show the cumulative distribution functions for
the paleointensity data (solid) and model (dashed), where the 100 simulations of the model are run with the same site distribution and number of
data per site as in the intensity data set. The right panel shows VGP dispersion, ST, as a function of latitude for the JC96 data set (red) and the
models (black). The data are averaged in latitude bands, and the mean dispersions along with the one standard deviation error bar are shown.
For the models, ten simulations at each data site are shown.
dispersion due to the quadrupole family is constant with lati- They proposed that the power spectrum of the present field be
tude, whereas that due to the dipole family varies linearly used as a guide in constructing models for paleosecular varia-
(from zero at the equator) up to a latitude of 70 . McFadden tion. In CP88, PSV is described by statistical variability of each
et al. (1988) used this form for VGP dispersion to establish the Gaussian coefficient in a spherical harmonic description of the
dipole and quadrupole family contributions to PSV for the geomagnetic field, with each coefficient treated as a normally
period 0–5 Ma. Their model G has survived remarkably well distributed random variable: The Gaussian coefficients of the
as a reasonable description of one measure (SB) of PSV. One of nondipole part of the field exhibit isotropic variability, and the
the important aspects of model G was its attempt to tie PSV variances are derived from the present field spatial power
statistics to dynamo theory, specifically mean field dynamos in spectrum. CP88 and its descendants are known collectively as
which the magnetic field solutions separate into the dipole and giant Gaussian process (GGP) models. The dipole terms have a
quadrupole families. We return to possible links between PSV special status in CP88, with a nonzero mean for the axial dipole
and dynamo models in Section 5.11.7. and lower variance than predicted from the spatial power
spectrum. All nondipole terms have zero mean except the
axial quadrupole. Isotropic variability means that the statistical
5.11.4.2 Giant Gaussian Process Models
variations of Gaussian coefficients about their mean value do
In the same year that model G was proposed, an alternative not depend on the orientation of the coordinate system in
approach was put forward by Constable and Parker (1988). which the Gaussian coefficients are defined. Apart from the
mean values, the statistical distributions for gm m

l and hl vary the particular model CJ98.nz is an end-member of a possible
only with spherical harmonic degree, l, not with the order m. choice of nonzonal models, it demonstrates several aspects of
The isotropic variations in the Gaussian coefficients for the PSV that could be tested with such models, given sufficiently
geomagnetic potential suggest a secular variation model in good data sets (Figure 13). Nonzonal (longitudinal) variations
which there is no preferred directional dependence, but this in paleosecular variation such as the high secular variation
isotropy does not extend to the actual physical measurements associated with the flux lobes seen in the historical magnetic
(X, Y, Z or D, I, F) of the local geomagnetic field. field are simulated by such nonzonal models. VGP dispersion is
One advantage of this model over earlier descriptions is that rather insensitive to longitudinal variations in structure of PSV,
it supplies a complete (albeit oversimplified) description of the whereas other measures such as inclination dispersion have the
model field, even though its only parameters are the means potential to be more informative. Finally, geographic variations
and variances of distributions of Gaussian coefficients. It in the frequency of occurrence of excursional directions are
allows direct calculation or simulation of the expected distri- predicted by nonzonal models and could eventually be tested
butions for any geomagnetic observable and these can be against observed distributions.
tested against the distributions of available observations. Tauxe and Kent (2004) suggested an alternative approach to
CP88 was found to match distributions of declination and modifying CP88 to fit both intensity and directional data.
inclination taken from the data set of Lee (1983). The most Rather than treating the dipole and l ¼ 2, m ¼ 1 variances as
notable shortcomings of CP88 are its inability to predict the special cases, they proposed a different variance for the dipole
observed form of SB with latitude and its underprediction of and quadrupole families (McFadden et al., 1988). The param-
the variance in paleointensity data (Figure 12). Kono and eter b in their model (TK03) describes the ratio of the dipole/
Tanaka (1995) showed that an isotropic model like CP88 cannot quadrupole family variance. They also set the mean value of
generate significant variations in VGP dispersion with latitude, the axial dipole term to be 50% of that used in CP88, CJ98, and
but that extra variance in Gaussian coefficients of degree 2 and CJ98.nz, reflecting their view of a lower global average value
order 1 might generate the observed variation in VGP dispersion for paleointensity (Selkin and Tauxe, 2000).
with latitude. This idea was developed further by Hulot and Hulot and Bouligand (2005) and Bouligand et al. (2005)
Gallet (1996) who suggested the dominance of order 1 terms extended the symmetry ideas introduced by Constable and
up to about spherical harmonic degree 4, emphasizing the need Johnson (1999) further, providing the mathematical constraints
for anisotropy in the field variations and/or cross correlations that a GGP model must obey, if it is to satisfy spherical,
among the Gaussian coefficients. Quidelleur and Courtillot axisymmetric, or equatorially symmetrical symmetry properties
(1996) carried out extensive simulations for available data sites (see Gubbins and Zhang (1993) for a discussion of symmetry
with a variety of different parameters for the means and vari- properties of the dynamo equations). They used the results
ances of the Gaussian coefficients and concluded that the of two numerical dynamo simulations – one with homoge-
simplest modification to CP88 compatible with observed neous CMB conditions and one with heterogeneous CMB con-
VGP dispersion required a standard deviation for the g12 and ditions (simulations described in Glatzmaier et al., 1999) – to
h12 terms about three times larger than for the rest of the demonstrate that the calculation of the mean and the covariance
quadrupole terms. matrix allowed the symmetry-breaking properties introduced by
Four parameters associated with CP88 can be adjusted to fit the CMB conditions to be identified correctly. This lays some
the observations: a determines the variance (sl)2 for l ¼ 2 groundwork for analyses of PSV data, assuming the ability to
through 12; additionally, there are the mean values of the axial estimate covariance among the Gaussian coefficients.
dipole and axial quadrupole contributions to the field and
the variance (sl)2 in the dipole contributions to the field.
These parameters reflect constraints on the model imposed by 5.11.5 The Time-Averaged Field
the observations. For CP88, the values of the average axial
quadrupole term and the dipole terms’ variance were chosen Following the observation that the GAD hypothesis provides a
to fit the paleodirectional data set of Lee (1983). g01 was arbi- first-order description of field behavior for 0–2 Ma (Opdyke
trarily set to 30 mT, and the variance of the nondipole terms and Henry, 1969), studies of the TAF geometry have addressed
was derived from the white spectral fit for the nondipole how good this approximation is and the nature of any depar-
present field. tures from GAD. Such signals, while second order in magni-
Constable and Johnson (1999) proposed variations on CP88 tude, may be profoundly important for understanding the
in which anisotropy in the variability in the Gaussian coeffi- interplay of deep-Earth processes. Several avenues of investiga-
cients was introduced. This was motivated by the desire to tion are thus required: (1) the detection of non-GAD structure,
examine models in which proxies for homogeneous versus het- (2) discrimination among possible sources (real field behavior
erogeneous CMB conditions could be introduced. The incorpo- vs., e.g., sampling bias or rock magnetic effects), and (3) the
ration of large variance in the axial dipole, and in the non-axial explanation of any non-GAD field structure. In this section, we
quadrupole Gaussian coefficients, g12 and h12, results in predic- review studies of the TAF, focusing on the inferred non-GAD
tions that provide an improved fit (over CP88) to the latitudinal structure. We return to the issues of discrimination and expla-
form of VGP dispersion and the distribution of paleointensity nation in Sections 5.11.6 and 5.11.7.
data in the Tanaka et al. (1995) data compilation (Figure 12). Models of the historical geomagnetic field (Figure 3) sug-
The resulting variance in paleomagnetic observables depends gest that the CMB region and inner core affect both the TAF and
only on latitude (zonal models), unless the variance in h12 is its temporal fluctuations (Bloxham et al., 1989). In the early
different from that in g12 (nonzonal models). Specific candidate 1990s, reports that magnetic records of reversals showed VGP
models discussed in the chapters are CJ98 and CJ98.nz. While paths that were confined to two longitude bands (Clement,
19 19
18 18
17 17
16 16
15 15
14 14
13 13
12 12
11 11
10 10
(a) CJ98 vgp dispersion (⬚) (b) CJ98.nz vgp dispersion (⬚)
15 15
14 14
13 13
12 12
11 11
10 10
9 9
8 8
7 7
(c) CJ98 sd inclination (⬚) (d) CJ98.nz sd inclination (⬚)
36 36
33 33
30 30
27 27
24 24
21 21
18 18
15 15
12 12
9 9
6 6
(e) CJ98 sd intensity (mT) (f) CJ98.nz sd intensity (mT)
10 20
9 18
8 16
7 14
6 12
5 10
4 8
3 6
2 4
1 2
(g) CJ98 % of vgp lats <45 (h) CJ98.nz % of vgp lats <45
Figure 13 Predictions for various summary statistics for PSV at Earth’s surface. Left column shows predictions of a zonal PSV model (CJ98),
right column predictions of a nonzonal model (CJ98.nz). Rows from top to bottom show VGP dispersion, standard deviation in inclination, standard
deviation in intensity, and percentage of excursional directions.
1991; Laj et al., 1991) caused much controversy in the paleo- during stable polarities to investigate the persistent non-GAD
magnetic community (McFadden et al., 1993; Prévot and structure in the field.
Camps, 1993; Valet et al., 1992). The apparent coincidence of Studies of the TAF have focused on the past 5 Myr due to the
these longitude bands with the static flux lobes seen in the availability of global data sets (Section 5.11.3), with sufficient
historical field models and with seismically high lower mantle data to investigate both normal and reverse polarity periods.
velocities in tomographic models available at the time was Departures of the TAF direction from GAD at any location are
interpreted as the influence of lower mantle thermal condi- small (on the order of a few degrees), and so restricting studies
tions on core flow and field generation during reversals. The to the past 5 Myr means that plate motion corrections can be
scarcity of records of field reversals and the debate about their performed with sufficient accuracy to distinguish magnetic
fidelity prompted a series of studies of the paleomagnetic TAF field behavior from geographic effects.
5.11.5.1 Early Studies calculated using a maximum likelihood technique (McFadden

and Reid, 1982). A best-fit zonal TAF model truncated at
Studies by Wilson (1970, 1971) noted small departures from
spherical harmonic degree 4 was calculated for Brunhes and
GAD in the TAF recorded by sediments and lavas. Inclination XN h i2
data appeared shallower than the predictions of GAD, with Matuyama separately, by minimizing i¼1
I i
obs I i
pred . As
correspondingly far-sided VGPS (Figure 2). Wilson’s interpre- a percentage of g10, the values of g20 and g30 were 2.6%
tation was that these data were best approximated by a dipole, and 2.9% for the Brunhes and 4.6% and 2.1% for the
offset northward by 150–300 km from Earth’s center, along Matuyama, again reflecting the larger reverse polarity
the rotation axis. In addition, he noted that the pole positions inclinations.
from volcanic rocks showed a ‘right-handed’ effect, corre- In summary, these early studies of the TAF detected a lati-
sponding to a persistent nonzero declination, with a magni- tudinal signal in inclination anomalies from lavas and sedi-
tude on the order of 3–5 . ments, best fit by two-parameter models. Models proposed
In a seminal contribution, Merrill and McElhinny (1977) included an axial quadrupole contribution of 2.5–5% of the
used time-averaged declination and inclination anomalies axial dipole term for normal polarities and about twice this for
(eqn [6]) from their global compilation of paleomagnetic reverse polarities. Axial octupole terms were generally on the
data to estimate time-averaged values of the Gaussian coeffi- order of 1.5% to 3.5% of the axial dipole term.
cients. Their data set comprised D, I pairs from 101 normal and
31 reverse Quaternary (0–2 Ma) and 70 normal and 64 reverse
Pliocene (2–5 Ma) records, to which they added I data from
50 normal and 50 reverse polarity piston cores of Opdyke and 5.11.5.2 Longitudinal Structure in the TAF?
Henry (1969). Because of the incorporation of sediment data, As motivated earlier, structure seen in historical field models,
the geographic distribution was quite good (see their Figure 2). along with the provocative suggestion of lower mantle control
An analysis of declination anomalies led the authors to con- on magnetic records of reversals, prompted interest in whether
clude that the right-handed effect is an artifact, resulting from there is persistent longitudinal structure in the field. Two inde-
unevenly distributed, and insufficient, data. Based on this con- pendent groups pursued this using the lava flow compilations
clusion, and on the desire to use as large a number of data as of Q94 and JC96. In contrast to the least squares fitting, and/or
possible in estimating the TAF direction, the authors binned grid search approaches of previous studies, the tools of inverse
their data into 10 latitude bands (separately for normal and theory (Gubbins, 2004; Parker, 1994) were used to construct
reverse polarities). Mean inclination anomalies (possibly an spherical harmonic models that fit the observations but that
arithmetic average of contributing sites, as there is no specific also possess minimum structure. The results of the two groups
mention of a vector mean), along with the standard error, were are documented in a series of papers – Gubbins and Kelly
calculated for each bin. Predominantly negative inclination (1993), Johnson and Constable (1995, 1997, 1998), and
anomalies were seen, with the greatest signal (5 for normal Kelly and Gubbins (1997) (hereafter GK93, JC95, JC97, JC98,
polarity, larger for reverse) at low latitudes (Figure 14). This KG97, respectively).
general observation has held for three decades. A zonal spher- GK93 modeled the normal polarity average field structure
ical model was visually matched to the observed inclination using the data set of Q94 for the past 2.5 Myr and a technique
anomalies. The predicted inclination for a zonal model, trun- similar to that applied to the historical data (Bloxham et al.,
cated at spherical harmonic degree lmax, is given by 1989). In a parallel study, JC95 used the JC96 data set and a
Xlmax different inversion algorithm to model the Brunhes, 0–5 Ma
ðl + 1Þg0l P l ðcos yÞ normal and 0–5 Ma reverse polarity data. The 0–2.5 Ma normal
tan I ¼ Xl¼1
lmax
[14]
l¼1
g0l P 0l ðcos yÞ polarity model of GK93, along with their preferred model in a
later paper [KG97], and the 0–5 Ma normal polarity model of
where Pl0 (cos B)¼ dPl(cos y)/dy (see Merrill et al., 1996, JC95 are remarkably similar, both displaying some features seen
appendix B5 for details). Merrill and McElhinny (1977) used in historical field, notably the presence of two northern hemi-
lmax ¼ 3. Values for the axial quadrupole and the axial octupole sphere flux lobes (Figure 15). The JC95 Brunhes polarity model
(g02, g03) terms as a percentage of g01 were 5% and 1.7% for normal contains less structure, interpreted as resulting from the smaller
polarity and 8.3% and 3.4% for reverse polarity. The difference data set. As in previous zonal models for the TAF, JC95 found
between the normal time-averaged signature and reverse time- larger deviations from GAD during reverse polarities: this is seen
averaged signature was shown to be statistically significant. Lateral in the raw inclination anomaly observations. Both groups fol-
temperature variations in the lower mantle (Dziewonski, 1984) lowed up these results with studies that include sediment data
were suggested as a possible cause of the long-term non-GAD TAF and in the case of KG97 also some intensity data (Tanaka et al.,
and the normal/reverse polarity asymmetry. 1995). Because only inclination data are available from sedi-
Schneider and Kent (1990) investigated the TAF using only ments, models that include these data are smoother than their
sediment cores because of the better time averaging provided lava flow-only counterparts (JC97 and KG97). Several tests of
by a core as compared with a limited number of lava flows of the robustness of the results were conducted: For example, JC97
unknown ages. They compiled and used an updated piston showed that purely zonal models could not fit the lava flow
core data set (Section 5.11.3.4) that was greater in number data, and they conducted bootstrap estimates of uncertainty to
than that used by Merrill and McElhinny (1977) and in which demonstrate resolvable features in their models. As the discus-
a larger percentage of cores had been subjected to magnetic sion of structure in these TAF models is helped by comparing
cleaning. For each core, an average inclination anomaly was maps of the radial field, Br, at the CMB, JC97 also showed how
0 – 5 Ma
6
Normal polarity
4
0
ΔI
–2
–4
–6
–8
80⬚ 60⬚ 40⬚ 20⬚ 0⬚ –20⬚ –40⬚ –60⬚ –80⬚
(a) Latitude
–2
–4
ΔI –6
–8
–10
–12
Reversed polarity
–14
80⬚ 60⬚ 40⬚ 20⬚ 0⬚ –20⬚ –40⬚ –60⬚ –80⬚
(b) Latitude
Figure 14 Average inclination anomaly (△I) and its standard error within 10 latitude bins for (a) normal and (b) reverse polarity during the past 5 Myr
(data from Merrill and McElhinny, 1977). The solid lines represent models of inclination anomaly calculated from eqn [14] and (a) g02/g01 ¼ 1/2,
g03/g01 ¼ 1/60, (b) g02/g01 ¼ 1/1, g03/g01 ¼ 1/20.
the geographic distribution of observations of declination and axial quadrupole, compatible with other studies and in the
inclination samples Br at the CMB. range of 3.5–5.5% of the axial dipole term, and concluded
In a series of subsequent papers, the longitudinal structure that model coefficients are not significantly different from
in TAF models was discussed at length. McElhinny et al. zero below a threshold of 300 nT, rendering only these two
(1996a) noted that the time-averaged declination anomalies terms robust. However, some of these conclusions result from
were not statistically distinguishable from zero and so argued the truncation level used in implementing the inversion algo-
that if any nonzonal features exist, they are too small to be rithm, rendering the resulting models more akin to those
reliably resolved with the then-available data. The authors derived from truncated least squares than smooth inversions.
advocated zonal model fits to latitudinally averaged inclina- Thus, regularized inversions of paleodirections from lava
tion anomalies, as in Merrill and McElhinny (1977). Carlut flows, and from lava flow and sediment data combined, result
and Courtillot (1998) used the JC95 code to investigate TAF in nonzonal models for the TAF, with longitudinal structure
models for both the Q94 and JC96 data sets, truncating their that bears some resemblance to that seen in the historical field.
inversions at degree and order 4. They found values for the Different views exist as to the robustness of this structure: we
Normal polarity field model, GK93
400
300
200
100
0
–100
–200
–300
(a) –400
0–5 Ma normal polarity field model, JC95
400
300
200
100
0
–100
–200
–300
(b) –400
0–5 Ma normal polarity field model, JC97
400
300
200
100
0
–100
–200
–300
(c)
–400
Figure 15 Normal polarity field models (a) GK93, based on lava flow data of Q94, (b) JC95, based on lava flow data of JC96, (c) JC97, based on JC96
lava flow compilation and SK90 sediment data set.
discuss some of the underlying issues in Section 5.11.5 and in Fisher distributions of directions locally (Kokhlov et al., 2001;
Section 5.11.6 offer some suggestions for future work that can Lawrence et al., 2006; Tauxe and Kent, 2004), as has been
address this question. observed in data sets (Lawrence et al., 2006; Tanaka, 1999).
Hatakeyama and Kono (2001, 2002) inverted for both the TAF
and PSV using the JC96 data set. They concluded that the
5.11.5.3 Joint Estimation of PSV and the TAF
nonzonal TAF structure in their resulting models is not robust
Recent global TAF studies have begun to assess another issue – and found an average axial quadrupole 4.3% of the axial
that of whether inversions of the TAF direction alone can result dipole term, similar to that found in other studies. They
in a bias due to the influence on the TAF of PSV. Statistical found a PSV model with large variance in the degree 2, order
models of the type described in Section 5.11.3.2 produce non- 1 terms, as proposed in several PSV models (see
Section 5.11.3.2). However, the absence of paleointensity data and Johnson, 2005; Ziegler et al., 2011) indicate that different
in the inversion results in their PSV model having insufficient estimates of PSV might be obtained by averaging over, say, 103
variance in the dipole terms. In Section 5.11.5, we discuss the versus 106 years. Whatever the case, significant improvements
issue of bias resulting from the absence of paleointensity data could be made by having data that sample the same, known
and the implications for TAF investigations. We return to joint time interval at different locations globally. Obtaining radio-
PSV/TAF models in Section 5.11.6, as the simultaneous esti- metric ages was not previously a routine part of paleomagnetic
mation of PSV and the TAF is likely to be an important aspect PSV/TAF data collection efforts; this has changed recently with
of future paleomagnetic field models. 20% of flows being dated on average in a given study
( Johnson et al., 2008).
Temporal independence is also problematic for lava flow
5.11.6 Discussion sites, since thick sequences can be erupted in very short time
intervals and paleodirections from one flow to the next are often
In this section, we first address technical issues that give rise to serially correlated. Unfortunately, the problem is worst at places
many of the discussions surrounding PSV and the TAF. Some where there are large amounts of data such as Hawaii and
are driven by the existing data sets, while others by differences Iceland. In a ‘catch-22,’ the more conservative viewpoint then
in modeling approaches. We then summarize the successes and attributes any non-GAD structure in regions with greatest data
limitations of current models with respect to understanding sampling to ‘too much of the same.’ Inadequate temporal sam-
paleofield behavior. pling will lead to biased mean directions and artificially low
estimates of PSV. This is the root of most of the discussion
surrounding the evidence for long-term anomalous TAF and
5.11.6.1 Issues in PSV and TAF Modeling
PSV at Hawaii (e.g., McElhinny et al., 1996b) and the resulting
The number, distribution, and quality of existing global paleo- Pacific/Atlantic hemisphere asymmetries in paleofield models.
magnetic data sets are the underlying cause of much of the It is then a matter of philosophy as to how to analyze the
debate surrounding regional variations in PSV and persistent data. Johnson and Constable (1997) advocated careful data
non-GAD structure in the 0–5 Ma field. Specific aspects of the selection (omitting the lava flow sequences that are the most
debate (e.g., the merits of one data set vs. another) have been problematic) and tested for bias by ‘thinning’ data sets from
discussed elsewhere (e.g., Carlut and Courtillot, 1998; lava flow sequences to reduce the effect of oversampling.
McElhinny and McFadden, 1997; Merrill et al., 1996); here, A conservative approach is to attribute all the longitudinal
we review the general issues and how they have led to different variation in the mean field direction, at a given latitude, to
conclusions in the literature. Some of these issues are being poor temporal sampling of the data and to small undetected
directly addressed in new data sampling and laboratory efforts. tectonic movements (e.g., McElhinny et al., 1996a), and to
It is important to remember that early (1960s and 1970s) advocate only TAF and PSV models that are a function of
paleomagnetic data were collected to investigate, for example, latitude.
magnetic polarity and the GAD hypothesis, not to pursue the
kinds of studies we wish to conduct now. Hence, the kind of
data coverage, and quality that we require from data sets today, 5.11.6.1.2 Data sets: Spatial distribution
simply does not exist in some of these earlier, but quite large The geographic distribution of paleomagnetic data spanning a
collections. prescribed time interval affects existing TAF and PSV models
In the succeeding text, we address issues that arise from four differently. Most global PSV models prescribe statistics that
sources: (1) data distribution – temporal and spatial – (2) data vary only as a function of latitude. For these models, the spatial
quality, (3) bias due to the lack of intensity data, and distribution of data is reasonable, although southern hemi-
(4) modeling approaches. sphere coverage is poor compared with its northern hemi-
sphere counterpart. TAF models, especially those that
5.11.6.1.1 Data sets: Temporal sampling investigate nonzonal structure discussed in Section 5.11.5.2,
This is perhaps the most thorny issue in studies to date and can are affected by differences in not only the spatial distribution
be subdivided into two problems at any location: (1) whether but also the type of contributing data sets.
paleodirections span a long enough time interval to provide a Equation [9] gives the solution to Laplace’s equation in
reliable estimate of PSV and the TAF and (2) whether site mean terms of spherical harmonic functions and for a particular set
paleodirections are temporally independent. For deep-sea sed- of spherical harmonic coefficients allows calculation of the
iment cores, neither of these problems is acute, since a time resulting potential or associated magnetic field at any point
series of observations is available for a given core. However, the outside the assumed source region in Earth’s core. An alterna-
uneven, and discrete, nature of lava flows means the situation tive way of writing the magnetic field at Earth’s surface is in
is quite different. Data sets from different locations globally terms of Green’s function for Br ð^sÞ, the radial magnetic field at
may all have data of Brunhes polarity age but may sample the CMB:
ð
! !
different, and quite short, parts of the polarity chron. While ! !
104 years has been loosely used as a time interval thought to be B r ¼ G r j^ s Br ð^
sÞd2 ^
s [15]
S
sufficient to characterize PSV (Carlut et al., 1999; Johnson and
Constable, 1996; Merrill et al., 1996), other studies (Merrill Green’s functions for the field elements X, Y, and Z have
and McFadden, 2003) suggest that at least 105 years are neces- been published several times in the geomagnetic literature (see,
sary. Studies of the paleomagnetic power spectrum (Constable e.g., Constable et al., 1993; Gubbins and Roberts, 1983).
Paleomagnetic observations – declination, inclination, and sediment data sample mainly low-latitude regions, and this
intensity, F – are nonlinear functionals of Br ð^ sÞ, so one can sampling dominates the combined lava plus sediment data
formulate linearized data kernels that describe how these set, explaining the smoother field models, and reduction in
observations respond to changes in Br ð^ sÞ ( Johnson and high-latitude structure in these joint data set inversions.
Constable, 1997). These data kernels can be used to indicate A somewhat different picture of CMB sampling is obtained if
the response of D, I, and F to departures from an axial dipole one sums the contributions from the sampling kernels for each
field configuration. They vary with geographic location site – this would be appropriate for inversions where site level
(Figure 16). Importantly, sampling of Br ð^ sÞ at locations other data, rather than time-averaged directions, are inverted.
than immediately below the observation site is obtained. Dec- Clearly, in such a picture, locations with many contributing
lination provides longitudinal information, preferentially sam- sites (such as Hawaii) would dominate CMB sampling.
pling Br ð^sÞ at locations east and west of the observation
location. Inclination observations at the equator preferentially
sample Br ð^sÞ, directly beneath the observation site, but at non- 5.11.6.1.3 Data quality
equatorial site locations, sampling of Br ð^ sÞ is biased toward Reliable PSV and TAF models require that at the site level,
lower latitudes. In contrast, intensity data at nonequatorial mean paleodirections must represent the primary remanence
sites provide maximum sampling of Br ð^ sÞ at latitudes higher and within-site error should be a minimum. As discussed
than the observation latitude. earlier, estimates of within-site error require several samples
The CMB sampling offered by a particular data set can be per site. Traditionally, two or three samples per site were taken
investigated by summing the magnitudes of the contributions in the field; much of the discussion in the literature has been
from the kernels for each sampling location. This is done about whether sites with fewer than three samples should be
by assuming that each location contributes an estimate of the used in field modeling. Today, 10 samples per site are typi-
TAF – in other words, one observation of the time-averaged cally taken in the field, and after laboratory cleaning, site mean
inclination and, for lavas, one of the time-averaged declina- directions are usually obtained for 5–10 samples. Simulations
tion. In Figure 17, we show this sampling for the JC96 normal from statistical PSV models also suggest that at least four or five
polarity lava flow distribution and the JC96 plus SK90 sedi- samples per site are desirable ( Johnson et al., 2008; Tauxe
ment data distribution. The longitudinal coverage provided by et al., 2003). Similarly, as stepwise demagnetization proce-
the declination information from lavas is apparent. The dures and the estimation of directions by principal component
Declination 60N Inclination 60N Intensity 60N
Declination 30N Inclination 30N Intensity 30N
Declination equator Inclination 0N Intensity 0N

Figure 16 Sampling kernels to show how observations of D, I, and F at the surface sample the radial field at the CMB. Color scale denotes
relative sampling: darker regions are sampled more heavily than lighter ones. Positive (negative) Br is shown in red (blue). Declination observations
provide the best longitudinal coverage; inclination (intensity) observations preferentially sample Br at the CMB at lower (higher) latitudes than the
observation site, except at the equator.
×10–5
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
(a)
×10–5
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
(b)
×10–5
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.50
1.25
1.00
0.75
0.50
(c)
Figure 17 Sampling of the CMB defined by the sampling function in Johnson and Constable (1997) and reported in the text here. Maps show
the relative sampling of Br at the CMB by measurements from (a) sediment cores (I only, SK90), (b) lava flows (I and D, JC96), and (c) sediments
and flows. Notice that the large number of sediment cores with only inclination data focuses sampling at low midlatitudes and accounts for
models based on lavas and sediments with less nonzonal and less high-latitude structure than models based on lava flow data alone.
analysis (Chapter 5.04) have become routine, concern about measures of data quality are the cone of 95% confidence about
data contaminated by overprints is decreasing. the mean direction, a95, and k (eqns [12] and [13]), and
Assuming sufficient samples per site and removal of over- choices for both have been used in assembling data sets.
prints, the question then is how and whether data should be Regional compilations have included legacy data that were
excluded on the basis of poor quality. As already noted, two subjected to less thorough lab procedures than are now
standard, and these compilations have typically used a cutoff " !2 !2 !2 #1=2
X
N X
N X
N
of k ¼ 100 to try to assure a robust estimate of the TAF and PSV R¼ bi xi + bi yi + bi zi [16]
(Johnson et al., 2008; Lawrence et al., 2006; Tauxe et al., i¼1 i¼1 i¼1
2004a,b). For modern sampling and laboratory studies, a cut-

off of k 50 may well be sufficient (Cromwell et al., 2012; where bi are the contributing individual intensities. PSV results
Cromwell, personal communication, 2013). Until now, TAF in a distribution of bi at any given location, so the unit vector
studies have typically discarded poor-quality data; an alterna- mean is not equal to the full vector mean. The bias in the mean
tive future approach is a more appropriate choice of data direction inferred from unit vectors depends on the statistics of
weights. the PSV, and in general, there is no closed-form analytic solu-
tion for this. Love and Constable (2003) provided an analytic
5.11.6.1.4 Transitional data solution for the bias in inclination, obtained by averaging only
Another selection criterion used in generating PSV and TAF measurements of inclination, when the underlying statistics of
models is to discard transitional data. For PSV models, this is PSV are given by a locally isotropic process (equal variances in
done on the basis that a few sites with low-latitude VGPs can Bx, By, and Bz).
significantly increase the estimate of dispersion and that field Here, we show the kind of bias that might be expected on
behavior during such times may be distinct from that during the basis of our current understanding of 0–5 Ma PSV. We first
stable polarities. This effect is shown in Figure 18, using sim- use the statistical model CJ98, but with a TAF of GAD, to
ulations from a modified version of the statistical model CJ98, simulate the apparent inclination anomalies obtained by aver-
in which a TAF of GAD was prescribed (instead of GAD plus an aging unit vectors, for a range of VGP cutoffs (Figure 19(a)).
axial quadrupole as in CJ98). Directions were simulated at 5 Inclination anomalies are well approximated by an apparent
latitude increments, with 10 000 sites at each latitude. VGP axial octupole signature as noted previously (Creer, 1983;
dispersion was calculated using sites with VGPs greater than a Merrill and McFadden, 2003), with a magnitude of 2–3% of
specified VGP cutoff, where the latter was varied from 55 to g01, depending on the choice of VGP latitude cutoff. For a VGP
90 . A VGP latitude cutoff of 90 includes all data, even latitude cutoff of 45 , declination anomalies are close to zero
those that may in fact be of opposite polarity, hence the large except at high (greater than 75 ) latitudes, and they show no
increase in dispersion. The VGP dispersion curves for VGP longitudinal dependence (Figure 19(b)).
latitude cutoffs of 0 , 45 , and 55 show an overall increase We examine the dependence of the bias on the partitioning
in dispersion as lower-latitude VGPs are permitted and also of variance in the statistical PSV models, by comparing CJ98
show that the increase itself is latitude-dependent. The use of a with TK03 (Figure 19(c)). For a VGP latitude cutoff of 45 ,
VGP latitude cutoff is not ideal, since the identification of two CJ98 results in bias approximated by that due to an axial
classes of field behavior –transitional versus stable polarity – is octupole term, g03 ¼ 2 % g01; TK03 suggests a larger (3%) axial
arbitrary. With more comprehensive and accurately dated data octupole term, although the shape of the apparent inclination
sets, one could explore the influence of temporal sampling and anomaly is less well described by only an axial octupole term.
what happens with the appropriate representative sampling of We also simulate the bias obtained by two modified versions of
transitions (see Section 5.11.7). TK03. In the first, the variance in the dipole family is increased,
at the expense of the variance in the quadrupole family. We set
the parameter a of TK03 ¼ 0.75 and b of TK03 ¼ 40.5; the
5.11.6.2 Bias from Unit Vectors choice of a and b is made so as to keep the total variance in
It has long been known (Creer, 1983; see also discussion in the spherical harmonic degree 1 terms constant and as pre-
Merrill and McFadden, 2003) that biased estimates of direction dicted by TK03. This results in little change in the bias com-
are obtained when unit vectors are averaged. If full vector pared with that obtained from TK03. In the second experiment,
measurements of the field are available, then the resultant we set b ¼ 1, and a ¼ 17.6, allowing equal variances in the
vector, R, of eqn [5] is modified as follows: dipole and quadrupole families. Note that this is similar to
model CP88, one that predicts VGP dispersion that is invariant
with latitude. This results in a larger magnitude bias at low
24 midlatitudes and a latitudinal dependence that is less well
VGP dispersion (⬚)
modeled by only a g03 signature. Finally, we show how the

20 magnitude of the bias could change if the overall level of PSV
16
were higher. We double values of a and b given in TK03. (This
results in values for VGP dispersion that range from 14 at the
12 equator to 22 near the poles, higher but not greatly so, than
those predicted by TK03 or CJ98 – see blue crosses in
–80 –40 0 40 80 Figure 18.) The resulting bias in inclination is shown in
Latitude Figure 19(d) for the values of VGP cutoff, used in Figure 19
(a). The higher PSV results in similar magnitude (on the order
Figure 18 Effect of VGP latitude cutoff on estimate of VGP
of 1 bias) for VGP cutoffs around 45 , compared with that
dispersion. Simulations use statistics prescribed in PSV model CJ98, but
with GAD as the time-averaged field (TAF). 10 000 sites at each latitude predicted by TK03, but the structure in the bias is no longer
were simulated. The total VGP dispersion (assuming zero within-site described by an axial octupole. VGP cutoffs of 90 result in a
error) was computed for VGP latitude cutoffs of 55 (black), 45 (blue), bias that is reasonably approximated by an axial octupole, but
0 (green), and 90 (brown). one that has a magnitude of 10% of g01. We note that this has
2 2
1 1
Δd apparent
Δi apparent
0 0
–1 –1
CJ98: Inclination CJ98: Declination

bias versus latitude bias versus latitude
–2 –2
–80 –40 0 40 80 –80 –40 0 40 80

(a) (b)
Latitude Latitude
3
6
2
4
1
2
Δi apparent
Δi apparent
0 0
–2
–1
Effect on bias of –4 Effect on bias of

–2 different descriptions increasing PSV
of PSV
–6
–3
–80 –40 0 40 80 –80 –40 0 40 80
(c) Latitude (d) Latitude
Figure 19 Simulations from PSV statistical models to show bias as a function of latitude incurred by averaging of unit vectors. (a) Apparent inclination
anomaly for CJ98, using a TAF of GAD and four choices of VGP latitude cutoff: 55 (red), 45 (blue), 0 (green), and 90 (brown). Predictions for
an axial dipole plus axial octupole field are shown by dashed curves, with g03 ¼ 2% g01 (smaller magnitude anomalies) or g03 ¼ 3% g01. (b) Apparent
declination anomaly computed along five lines of longitude: 0 (red), 45 (blue), 90 (green), 135 (brown), and 180 (orange). (c) Apparent inclination
anomaly using a VGP latitude cutoff of 45 for four prescriptions of PSV – CJ98 (blue triangles, as in (a)), TK03 (red triangles), modified TK03, with
a ¼ 17.6, b ¼ 1 (red stars), modified TK03 with a ¼ 0.75, b ¼ 40.5 (red crosses). (d) Apparent inclination anomaly, colors as in (a), for a modified
version of TK03 in which a ¼ 15 and b ¼ 7.6, that is, twice their values in TK03. Dashed lines as in (a); long dashed line (largest amplitude inclination
anomaly) is for g03 ¼ 10% g01.
interesting implications for paleofield investigations for older reasonably approximated by an axial octupole term,
times, for which plate motion corrections are not possible. In g03 ¼ 2 3 % of g01, when low-latitude VGP sites are excluded
such cases, it may not be possible to distinguish excursional or from analyses. However, this is not so for other PSV scenarios,
reverse polarity directions, and large axial octupole terms could in particular cases in which the overall level of PSV is increased.
erroneously be inferred.
In summary, the bias incurred by averaging of unit vectors 5.11.6.2.1 Modeling approaches
is zonal and thus cannot be invoked to argue against longitu- Some of the differences among existing studies are related to
dinal structure in TAF models. It is small, but of similar mag- modeling approaches. Studies that bin data in latitude bands
nitude at some latitudes to the observed zonal non-GAD can of course only examine zonal models for the TAF – that is,
signature. For the past 5 Myr, it is likely that the bias is ones lacking longitudinal structure. Another difference lies in
the philosophy as to how one should define ‘simple’ models. In 5.11.6.3 Successes and Limitations of TAF and PSV Models
the parameter estimation approach (least squares or grid
PSV and TAF models to date have been quite successful in
searches), models are found that have minimal parameters but
fitting summary statistics (the mean and measures of variance)
that fit the data (McElhinny, 2004). In the inverse theory
of observations of paleodirection and intensity. The need for
approach, ‘simple’ models are described via a smoothness or
such models has motivated the compilation of several global
regularization criterion (e.g., minimizing the power in the non-
data sets of directions recorded in lavas and sediments.
GAD coefficients). Models are found that fit the data to within a
However, several limitations and outstanding questions
specified tolerance but that also have minimal structure in the
remain; most result from the issues raised in Section 5.11.6.1.
specified sense. The regularization results in smooth models in
Many of these can now be addressed with new global data sets
regions where data coverage is sparse. To ensure that the mini-
under construction (see Section 5.11.7) that comprise only
mal structure model is found, the spherical harmonic truncation
modern, high-quality paleomagnetic measurements.
level must not be set too low. This difference in truncation levels
Still unanswered satisfactorily is the question of symmetries
is one of the causes of the differing interpretations of the same
in the TAF and PSV: do longitudinal and/or polarity asymme-
data set (JC96) by Johnson and Constable (1997) and Carlut
tries exist? This is important to resolve since there is increasing
and Courtillot (1998). Different philosophies as to how to
evidence from numerical simulations, as well as centennial- to
choose the Lagrange multiplier – the relative importance of the
millennial-scale field models, that lateral heterogeneities in the
regularization constraint – also exist (see discussions in Johnson
lowermost mantle influence magnetic field generation in the
and Constable, 1997, and Hatakeyama and Kono, 2002, along
core (Bloxham, 2000; Bloxham and Gubbins, 1987; Driscoll
with Kelly and Gubbins, 1997 for three different approaches).
and Olson, 2011; Gubbins et al., 2007; Korte and Holme,
Various ways of assessing model uncertainties have also been
2010; Olson and Glatzmaier, 1996; Olson et al., 2010, 2012,
used: for example, examining the covariance matrix
2013; Sreenivasan and Gubbins, 2011; Willis et al., 2007).
(Hatakeyama and Kono, 2002; Kelly and Gubbins, 1997),
Figures 20 and 21 summarize published time averages of the
choice of a resolvable threshold level for all spherical harmonic
global magnetic field on timescales of hundreds, thousands,
model coefficients (Carlut and Courtillot, 1998), and non-
and millions of years. Figure 20 shows the geomagnetic field
parametric approaches Johnson and Constable (1995, 1997).
350 350
200 200
50 50
–100 –100
–250 –250
–400 –400
(a) Model GUFM1 (c) Zonal field model
350 350
200 200
50 50
–100 –100
–250 –250
–400 –400
(b) Model CALS7K.2 (d) Model LSN1
350
200
50
–100
–250
–400
(e) Model LN1
Figure 20 Time-averaged radial magnetic field (Br) at the CMB, on different timescales. Units are microtesla. (a) Historical field: 1590–1990,
Model GUFM1 ( Jackson et al., 2000), (b) Archaeofield: 0–7 ka, Model CALS7K.2 (Korte and Constable, 2005), (c) Paleofield: 0–5 Ma, axial dipole plus
axial quadrupole field (see text), (d) Model LSN1 ( Johnson and Constable, 1997), (e) Model LN1 ( Johnson and Constable, 1995).
28 28
16 16
4 4
–8 –8
–20 –20
–32 –32
(a) GUFM declination anomaly (b) GUFM inclination anomaly
14 14
8 8
2 2
–4 –4
–10 –10
–16 –16
(c) CALS7K.2 declination anomaly (d) CALS7K.2 inclination anomaly
7 7
4 4
1 1
–2 –2
–5 –5
–8 –8
(e) LSN1 declination anomaly (f) LSN1 inclination anomaly

Figure 21 Declination (a, c, e) and inclination (b, d, f ) anomalies in degrees (deviations from GAD direction) at Earth’s surface predicted from models
for the three time intervals in Figure 18: (a, b) Model GUFM1, (c, d) CALS7K.2, (e, f ) Model LSN1. The scale bar for the historical field is twice that
for the archaeofield and four times that for paleofield anomalies.
averaged over three quite different time periods: the past 400 that seen in GUFM1. Note that the magnitude of the signal
years (model GUFM1 discussed earlier), the past 7 kyr (model decreases over longer timescales. From Figure 21(f), we see
CALS7K.2, Korte and Constable, 2005; and see also the that the average inclination anomaly in LSN1 is rather small,
updated 0–3 kyr model in Korte and Constable, 2011), and and we can expect the largest signal at equatorial latitudes. If
three models for the past 5 Myr. When averaged over 0–7 ka, this view of the TAF is approximately correct, then at mid-
CALS7K.2 shows longitudinal structure that suggests the pres- latitudes to high latitudes, it will be difficult to detect depar-
ence of flux lobes seen in the historical field. The radial mag- tures from GAD without large data sets that provide accurate
netic field is attenuated in CALS7K.2 compared with GUFM1: measures of DI.
The resolution and accuracy is clearly inferior, but averaging of Long-lived longitudinal variations in PSV are equally diffi-
millennial-scale secular variation also plays a major role in cult to assess with the data sets and modeling approaches
subduing the structure. Three quite different field models for discussed earlier. Low PSV in the Pacific, as suggested by
the past 5 Myr are shown, spanning the range of proposed Doell and Cox (1963), has been refuted by many on the
published geographic structure in the field. A purely zonal basis of oversampling of short (and hence unrepresentative)
model is shown in Figure 20(c), with a zonal quadrupole time intervals (e.g., McElhinny et al., 1996b). On historical
contribution, g02, of 5% of g01. Figure 20(d) and 20(e) shows timescales, hemispheric differences in Atlantic and Pacific
model LN1 (lava flows only) of JC95 and model LSN1 (lavas PSV exist, with much greater secular variation in the Atlantic
and sediments) of JC97. (Bloxham and Gubbins, 1985; Bloxham et al., 1989), and this
Figure 21 shows the signal expected at Earth’s surface for has been suggested to extend to the millennial paleomagnetic
the GUFM1, CALS7K.2, and LSN1 average field models, in the record (Gubbins, 2004; Johnson et al., 1998; Korte and
form of geographic variations in inclination anomaly and Constable, 2005). Recent data compilations for four regions
declination anomaly. The structure in the archaeo- and paleo- at 20 latitude indicate that VGP dispersion at Hawaii is lower,
field anomalies is rather similar (despite the very different data but not significantly so, than at Mexico or Reunion (Lawrence
distributions from which they are derived) and contrasts with et al., 2006). The paleomagnetic record of Pacific PSV is
complicated because data sets come primarily from Hawaii collection of paleodirections (and some intensities) from 894
and French Polynesia: The former have been argued to include lava flows at 17 locations and is summarized in Johnson et al.
oversampling of short stable polarity intervals, but the latter (2008). The TAFI study locations were chosen to improve the
include unrepresentatively high sampling of reversal records. geographic coverage of 0–5 Ma paleomagnetic directions at
The discussion in Section 5.11.4 along with Figure 13 indi- high latitudes (Spitsbergen, Aleutians, and Antarctica) and in
cates that detection of regional variations in PSV requires mea- the southern hemisphere (various South American locales,
sures other than VGP dispersion and clearly resolution of this Easter Island, and Australia). Several of the TAFI studies were
question requires the development of time-varying field conducted to replace previous, inadequately demagnetized
models. data. Samples in four studies – Aleutians (Coe et al., 2000;
The equations for the dynamo problem are such that if B is Stone and Layer, 2006), Antarctica (Tauxe et al., 2004a), and
a solution for the magnetic field, then so is B. This means that Easter Island (Brown, 2002) – were collected in the 1960s and
polarity asymmetries are not explained by dynamo theory, and 1970s, all other sites required new field work. New radiometric
so it has been suggested that, like longitudinal TAF structure, dates, along with 95% uncertainties, have been obtained for
they may result from long-term spatially heterogeneous CMB 226 of the TAFI sites ( Johnson et al., 2008).
conditions ( Johnson and Constable, 1995, 1997; Kelly and In addition, sampling and laboratory work by several other
Gubbins, 1997; Merrill and McElhinny, 1977). Others have groups worldwide, together with several regional data compila-
argued (see review in Merrill et al., 1996) that these result from tions (the SW United States, Tauxe et al., 2003, the NW United
overprints in the reverse polarity data. Viscous overprints from States, Tauxe et al., 2004b, and 20 latitude, Lawrence et al.,
the current normal polarity period should have a different 2006, Japan, and New Zealand, Johnson et al., 2008), are now
blocking temperature spectrum from the primary remanence allowing the construction of new global data sets comprising
and modern techniques are capable of removing these. Inter- only high-quality directional data from lava flows. One such data
estingly, many recent data sets still have reverse polarity direc- set, currently under preparation (Cromwell et al., 2012; Crom-
tions that are farther from GAD than their normal counterpart well, 2013, personal communication), spans the last 10 Myr and
( Johnson et al., 1998, 2008; Tauxe et al., 2004b), although it is includes only data meeting modern demagnetization standards,
often true that the 95% confidence limits of the two directions with at least four specimens per site, and a site mean direction
overlap. with a value of k of at least 50. Within this data set, 1920 sites are
Bias in existing TAF models can come from two sources as from the last Myr, and of these, most are from the Brunhes and
we have noted. The first is that due to averaging of unit vectors. Matuyama periods. The data set spans latitudes from 78 S to
Simulations using the statistical models of TK03 and CJ98 79 N, and Figure 22 shows that coverage in the southern hemi-
indicate that this bias is manifest as an inclination anomaly sphere is significantly improved over that in previous global data
as a function of latitude described by a g03 term with a magni- compilations (Figures 8 and 9; in particular note that only eight
tude of 2% of g01 (Johnson et al., 2007). The second source of locations in Figure 8 comprise high-quality data).
bias is that incurred by inverting an average field direction for a Figure 22 also shows the distribution of two other data
time-averaged spherical harmonic model (Hatakeyama and types: records of absolute paleointensity from the past 5 Myr
Kono, 2001, 2002; Khokhlov et al., 2001, 2006). and deep-sea sediment cores that span all or part of the past
Studies using large regional data sets indicate that none of 2 Myr for which directional and relative paleointensity records
the existing GGP models predict local directional distributions are available. The combination of these new lava flow direction
that fit large regional data sets (Lawrence et al., 2006). More and absolute paleointensity data and sediment inclination
generally, it is not yet clear whether Gaussian distributions for (sometimes declination) and relative paleointensity data
the spherical harmonic coefficients are appropriate. Compari- provides exciting new opportunities in global and regional
sons with numerical dynamo simulations (McMillan et al., field modeling.
2001) and millennial-scale time-varying field models (Korte A 0–5 Myr normal polarity TAF model based on the data set
and Constable, 2006) demonstrate the kinds of tests that could of Cromwell et al. (2012) is shown in Figure 23. The modeling
be performed, given a new generation of time-varying spherical approach is that of Johnson and Constable (1995, 1997) and is
harmonic paleofield models. based on 1301 site mean directions from 34 regions. All
regions have at least ten sites from which the TAF direction is
computed, with an average of 38 sites per region. The variance
5.11.7 Future Directions reduction over a GAD model is 75%, and the resulting TAF
model has structure in Br at the CMB (Figure 23(a)) that is
5.11.7.1 Toward New Global Data Sets
similar to that observed in previous models. Notably, the
Over the past decade, significant effort has been expended on regions of increased flux at high northern latitudes are seen.
the collection of 0–5 Ma paleomagnetic data from lavas. These The improved sampling of the CMB in the southern hemi-
new data differ in many ways from previous data sets: More sphere, especially over South America (Figure 23(b)), allows
samples per site are taken in the field, stepwise alternating field the detection of corresponding regions of increased flux at high
or thermal demagnetization is routine, and radiometric dating south latitudes. The predicted inclination and declination
of sampled flows is increasingly an integral component of the anomalies show similar structure to that seen in Figure 21.
study. It is not possible to summarize all the studies here, but Thus, a first look at TAF models derived using existing model-
we refer to one set of studies targeted specifically to address ing approaches, but based on new global data sets comprising
questions of global TAF or PSV behavior. The project known as only high-quality data, shows much of the longitudinal struc-
the ‘TAF investigations’ (TAFI) project has involved the ture hinted at by previous work.
Figure 22 Current status of global data sets for paleomagnetic field modeling for the past few Myr. Red circles denote studies with 0–5 Ma directional
data from lava flows, taken from a recent compilation – PSV10 – that spans the past 10 Myr (Cromwell et al., 2012, in prep.). The compilation
includes data collected as part of the TAFI project ( Johnson et al., 2008). Black squares are sediment cores included in either Sint800 or Sint2000.
Blue triangles are 0–5 Ma absolute paleointensity data sites where Thellier–Thellier measurements with pTRM checks were made on at least two
specimens.
5.11.7.2 A New Generation of Paleomagnetic Field

combinations of directions and intensities. A natural, but sub-
Modeling
stantive, extension of this work is to consider local anisotropy.
Future paleomagnetic field models will be greatly aided not Some related efforts are already under way: for example, the
only by the incorporation of the new, larger data sets but also work of Khokhlov and Hulot (2013) shows that 2-D probabil-
by the joint modeling of multiple types of data. This is non- ity uniformization can be used to both test data at the site level
trivial, but much work has already been started, for example, and assess GGP models of the TAF field and PSV against global
how best to calibrate absolute and relative paleointensity data data sets. At the very least, TAF modeling efforts should include
(Ziegler et al., 2011). The sampling kernels described in non-Gaussian uncertainties in all the contributing data types
Section 5.11.6 demonstrate that the inclusion of absolute and the use of robust statistics. In principle, future PSV models
paleointensity data will aid sampling of Br at the CMB, espe- could also include temporal covariance estimated from long
cially at high latitudes. The joint study of both intensity and time series of paleomagnetic observations, as done by Hulot
directional data will allow investigation of whether directional and Le Mouël (1994) for the historical field.
variability is indeed less during periods of higher field strength It is clear that the separation of PSV from the TAF is artificial
as has been suggested (e.g., Love, 2000) and is hinted at (via and our understanding of how to deal with reversal and transi-
lower PSV and smaller deviations from GAD in the TAF for tional data is poor. Detailed regional studies of field behavior can
the Brunhes compared with the rest of the past 5 Myr) in the examine the question of field behavior over different timescales,
20 latitude compilation of Lawrence et al. (2006). via a combination of different kinds of studies. Regional paleo-
While a good start on joint inversions of data for TAF and magnetic power spectra, along with estimates of PSV, and per-
PSV has been made, studies to date (Hatakeyama and Kono, centage of transitional/excursional data may offer better insight
2001, 2002) have solved iteratively for the TAF and the PSV. In than separate global analyses of the TAF, PSV, or reversals. Such
the algorithm used in these studies, an inversion for the time- regional studies are particularly needed to try to determine lon-
averaged Gaussian coefficients is performed, followed by an gitudinal variations in field behavior that might reflect CMB
inversion for the variance in these coefficients, and the proce- influences on field behavior and to characterize field behavior
dure repeated until convergence. Because the full inversion is at high latitudes to understand what if signatures of the influence
nonlinear, it is unclear to what extent this approximation is a of the inner core are observable in the paleofield.
satisfactory approach. The heavy reliance on the GGP of field The statistical models for PSV offer some potentially
modeling may in fact be unwarranted, as is suggested by the exciting avenues of investigation because they can link ana-
inability of current statistical models to match regional empir- lyses of paleomagnetic data themselves, as well as the output
ical directional distributions (Lawrence et al., 2006) and even from dynamo simulations. Some such studies have been
some global distributions (Khokhlov et al., 2001, 2006). In the conducted (Glatzmaier et al., 1999; McMillan et al., 2001),
case of the 20 latitude data set of Lawrence et al. (2006), but much remains to be done, subject of course to the
the empirical distributions cannot be fit by simply adjusting caveats inherent in interpreting current numerical dynamo
the relative variance contributions to the PSV. This result sug- simulations. For example, the partitioning of dipole to
gests that future work should consider inversions of regional quadrupole variance during reversals versus stable polarity
distributions of (D, I, and where possible F) for the local mean periods, as determined from dynamo simulations, might
and variance in the field. Love and Constable (2003) had provide an avenue to link or discriminate between reversal
already demonstrated how to formulate the inverse problem statistics and PSV statistics in the paleomagnetic record. One
for locally isotropic (Gaussian) fields, given mixed such study, confirming a prediction by McFadden et al.
350
250
150
50
–50
–150
–250
(a) –350
×10–3
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
(b)
–2
–4
(c) –6
12
8
4
0
–4
–8
(d) –12
Figure 23 (a) The TAF model based on 0–5 Ma lava flow D and I data from the data compilation shown in Figure 22. Figure shows Br at the CMB in
microtesla. Data locations shown by black triangles. (b) Relative sampling of Br at the CMB by the (D, I) data set in (a), using the sampling function in
Johnson and Constable (1997). (c) Inclination and (d) declination anomalies in degrees at Earth’s surface predicted by the model in (a).
(1988), is that of Coe and Glatzmaier (2006) in which an An obvious long-term goal is to incorporate all the disparate
inverse link between the stability and equatorial symmetry data types, along with their temporal information and invert for
of the simulated field is reported, suggesting that reversal time-varying paleomagnetic field models, as is now possible on
rates may have been low when the inner core was smaller millennial timescales (Korte and Constable, 2005, 2011). This
than presently. This kind of study, together with recent goal is still some way off since much work is required to cali-
paleomagnetic estimates (Ziegler et al., 2011) and dynamo brate and link the information provided by different data sets
simulations (Olson et al., 2012) of the power spectrum of and to develop the needed modeling tools. However, Figure 22
the geomagnetic field, exemplifies the need to consider field shows that the wealth of recently collected high-quality data
behavior over a variety of timescales. affords exciting avenues for future field modeling.
5.11.8 Concluding Remarks formal reviews that improved the manuscript. We also thank
Kristin Lawrence for help with figures and Geoff Cromwell for
The past two decades have seen an enormous transition in the providing the data set used in Figures 22 and 23. CLJ acknowl-
kind of information we require from paleomagnetic data. Ear- edges support from several NSF and NSERC grants.
lier studies needed data of sufficient quality to establish
whether the GAD approximation was a good first-order
description of the paleomagnetic field. Deviations from GAD
References
in the TAF direction are small, typically on the order of a few
degrees, and much better temporal control of data is required Aubert J, Johnson CL, and Tarduno JA (2010) Observations and models of the long-
to interpret them correctly. These more stringent data require- term evolution of Earth’s magnetic field. Space Science Reviews. http://dx.doi.org/
ments have necessitated the collection of many new data. This 10.1007/s11214-010-9684-5.
Baag C and Helsley CE (1974) Geomagnetic secular variation model E. Journal of
is a time-consuming and incremental process, but the ensem-
Geophysical Research 79: 4918–4922.
ble effort of the paleomagnetic community is just beginning to Barton CE (1982) Spectral analysis of palaeomagnetic time series and the geomagnetic
result in global data sets that outstrip their predecessors in spectrum. Philosophical Transactions of the Royal Society of London, Series A
number and quality and that can enable a new generation of 306: 203–209.
field models. Bloxham J (2000) The effect of thermal core–mantle interactions on the paleomagnetic
secular variation. Philosophical Transactions of the Royal Society of London, Series
Studies of the TAF have provided tantalizing suggestions for 358: 1171–1179.
the persistence of longitudinal structure in the field, and they Bloxham J and Gubbins D (1985) The secular variation of Earth’s magnetic field. Nature
continue to affirm early recognition of apparent asymmetries 317: 777–781.
in normal and reverse polarity fields. The current situation is Bloxham J and Gubbins D (1987) Thermal core–mantle interactions. Nature
325: 511–513.
unsatisfying, however, as the only agreed-upon non-GAD sig-
Bloxham J and Jackson A (1990) Lateral temperature variations at the core–mantle
nal in the TAF is a contribution that can be described by an boundary deduced from the magnetic field. Geophysical Research Letters
axial quadrupole term (g02) in global spherical harmonic 17: 1997–2000.
models that is 2–5% of the axial dipole term. The new direc- Bloxham J and Jackson A (1992) Time-dependent mapping of the magnetic field at the
tional data sets available from lava flows will allow better core–mantle boundary. Journal of Geophysical Research 97: 19537–19563.
Bloxham J, Gubbins D, and Jackson A (1989) Geomagnetic secular variation.
identification and separation of spatial and temporal field Philosophical Transactions of the Royal Society of London, Series A 329: 415–502.
variations, and mapping of the paleomagnetic field will be Bouligand CG, Hulot AK, and Glatzmaier GA (2005) Statistical paleomagnetic field
greatly improved, by, at the very least, inclusion of high- modelling and dynamo numerical simulation. Geophysical Journal International
quality, absolute paleointensity data. 161: 603–626.
Bower DJ, Wicks JK, Gurnis M, and Jackson JM (2011) A geodynamic and mineral
The GGP class of statistical models for PSV has enabled
physics model of a solid-state ultralow-velocity zone. Earth and Planetary Science
significant steps forward in PSV studies. Their major advantage Letters 303: 193202.
stems from the ability to predict distributions of any magnetic Brown L (2002) Paleosecular variation from Easter Island revisited: Modern
observable at the Earth’s surface, and their success has been to demagnetization of a 1970s data set. Physics of the Earth and Planetary Interiors
predict summary statistics of intensity and directional data sets. 133: 73–81.
Busse FH (2002) Convective flows in rapidly rotating spheres and their dynamo action.
Several models have been proposed that can predict the latitu- Physics of Fluids 14: 1301–1314.
dinal variation in VGP dispersion, and the empirical distribu- Carlut J and Courtillot V (1998) How complex is the time-averaged geomagnetic field
tion of VADMs calculated from intensity data globally, but over the last 5 million years? Geophysical Journal International 134: 527–544.
none are able to predict regional distributions of declination Carlut J, Courtillot V, and Hulot G (1999) Over how much time should the geomagnetic
field be averaged to obtain the mean-paleomagnetic field? Terra Nova 11: 239–243.
and inclination.
Channell JET, Hodell DA, and Lehman B (1997) Relative geomagnetic paleointensity
The development of new tools is needed to address many of and d18O at ODP Site 983 (Gardar Drift, North Atlantic) since 350 ka. Earth and
the outstanding questions regarding paleofield behavior. These Planetery Science Letters 153: 103–118.
are critical to understand the long-term role of boundary con- Christensen UR and Olson P (2003) Secular variation in numerical geodynamo models
ditions on core flow and magnetic field generation, to link with lateral variations of boundary heat flow. Physics of the Earth and Planetary
dynamo and paleofield models, and to propose sensible Clement BM (1991) Geographic distribution of transitional VGPs: Evidence for non-
hypothesis tests for examining field behavior further back in zonal equatorial symmetry during the Matuyama–Brunhes geomagnetic reversal.
time. Emerging data compilations indicate that the develop- Earth and Planetary Science Letters 104: 48–58.
ment of continuously time-varying regional and global field Coe RS and Glatzmaier GA (2006) Symmetry and stability of the geomagnetic field.
Geophysical Research Letters 33. http://dx.doi.org/10.1029/2006GL027903.
models spanning the past 2 Myr is a realistic goal for near-
Coe RS, Zhao X, Lyons JJ, Pluhar CJ, and Mankinen EA (2000) Revisiting the 1964
term advances and could afford great insight into interactions collection of Nunivak lava flows. Eos, Transactions American Geophysical Union
among deep-Earth processes. 81, Fall Meeting Supp. Abstract # GP62A-06.
Constable CG and Johnson CL (1999) Anisotropic paleosecular variation models:
Implications for geomagnetic observables. Physics of the Earth and Planetary
Constable C and Johnson C (2005) A paleomagnetic power spectrum. Physics of the
Acknowledgments Earth and Planetary Interiors 153: 61–73.
Constable CG and Parker RL (1988) Statistics of the geomagnetic secular variation for
Much of the work referred to in this chapter, and published by the past 5 My. Journal of Geophysical Research 93: 11569–11581.
its authors, would not have been possible without collabora- Constable CG, Parker RL, and Stark PB (1993) Geomagnetic field models incorporating
frozen flux constraints. Geophysical Journal International 113: 419–433.
tions. In particular, four individuals deserve special mention: Costin S and Buffett B (2005) Preferred reversal paths caused by a heterogeneous
Ron Merrill, Mike McElhinny, Cathy Constable, and Lisa conducting layer at the base of the mantle. Journal of Geophysical Research
Tauxe. Ron Merrill and an anonymous reviewer provided 109. http://dx.doi.org/10.1029/2003JB002853.
Courtillot V and Valet J-P (1995) Secular variation of the Earth’s magnetic field: From Hernlund JW and Jellinek AM (2010) Dynamics and structure of a stirred partially
jerks to reversals. Comptes rendus de l’Académie des Sciences, Paris molten ultralow-velocity zone. Earth and Planetary Science Letters 296: 18. http://
320: 903–922. dx.doi.org/10.1016/j.epsl.2010.04.02.
Cox A (1962) Analysis of the present geomagnetic field for comparison with Hoffman K (1984) A method for the display and analysis of transitional paleomagnetic
paleomagnetic results. Journal of Geomagnetism and Geoelectricity 13: 101–112. data. Journal of Geophysical Research 89: 6285–6292.
Cox A (1970) Latitude dependence of the angular dispersion of the geomagnetic field. Hollerbach R and Jones C (1993a) A geodynamo model incorporating a finitely
Geophysical Journal of Royal Astronomical Society 20: 253–260. conducting inner core. Physics of the Earth and Planetary Interiors 75: 317–327.
Creer KM (1962) The dispersion of the geomagnetic field due to secular variation and Hollerbach R and Jones C (1993b) Influence of Earth’s inner core on geomagnetic
its determination from remote times from paleomagnetic data. Journal of fluctuations and reversals. Nature 365: 541–543.
Geophysical Research 67: 3461–3476. Hospers J (1954) Reversals of the main geomagnetic field III. Proceedings of the
Creer KM (1983) Computer synthesis of geomagnetic paleosecular variations. Nature Koninklijke nederlandse Akademie van Wetenschappen B 57: 112–121.
304: 695–699. Hulot G and Bouligand C (2005) Statistical paleomagnetic field modelling and
Creer KM, Irving E, and Nairn AEM (1959) Paleomagnetism of the great Whin sill. symmetry considerations. Geophysical Journal International 161: 591–602.
Geophysical Journal of the Royal Astronomical Society 2: 306–323. Hulot G and Gallet Y (1996) On the interpretation of virtual geomagnetic pole (VGP)
Cromwell G, Tauxe L, Jarboe NA, Johnson CL, and Constable CG (2012) Paleosecular scatter curves. Physics of the Earth and Planetary Interiors 95: 37–53.
variation over the last 10 Ma from a new global dataset (PSV10): Evidence for long- Hulot G and Le Mouël JL (1994) A statistical approach to Earth’s main magnetic field.
term hemispheric asymmetry. In: Fall Meeting, 3–7 December 2012, San Francisco, Physics of the Earth and Planetary Interiors 82: 167–183.
CA: American Geophysical Union, Abstract GP11A-07. Hulot G, Eymin C, Langlais B, Mandea M, and Olsen N (2002) Small-scale structure of the
Doell RR and Cox AV (1963) The accuracy of the paleomagnetic method as geodynamo inferred from Oersted and Magsat satellite data. Nature 416: 620–623.
evaluated from historic Hawaiian lava flows. Journal of Geophysical Research Irving E and Ward MA (1964) A statistical model of the geomagnetic field. Pure and
68: 1997–2009. Applied Geophysics 57: 47–52.
Doell RR and Cox AV (1965) Paleomagnetism of Hawaiian lava flows. Journal of Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular
Geophysical Research 70: 3377–3405. variation from historical records. Philosophical Transactions of the Royal Society of
Driscoll P and Olson P (2011) Superchron cycles driven by variable core heat flow. London, Series A 358: 957–990.
Geophysical Research Letters 38. http://dx.doi.org/10.1029/2011GL046808. Jellinek AM and Manga M (2004) Links between long-lived hotspots, mantle plumes,
Dziewonski AM (1984) Mapping the lower mantle: Determination of lateral D00 and plate tectonics. Reviews of Geophysics 42. http://dx.doi.org/
heterogeneity in P velocity up to degree and order 6. Journal of Geophysical 10.1029/2003RG000144.
Research 89: 5929–5952. Johnson CL and Constable CG (1995) The time-averaged geomagnetic field as recorded by
Fisher RA (1953) Dispersion on a sphere. Proceeding of the Royal Society of London lava flows over the past 5 Myr. Geophysical Journal International 122: 489–519.
217: 295–305. Johnson CL and Constable CG (1996) Paleosecular variation recorded by lava flows
Garnero EJ and McNamara A (2008) Structure and dynamics of earths lower mantle. over the last 5 Myr. Philosophical Transactions of the Royal Society of London,
Science 320: 626–628. Series A 354: 89–141.
Glatzmaier GA, Coe RS, Hongre L, and Roberts PH (1999) The role of the Earth’s mantle Johnson CL and Constable CG (1997) The time-averaged geomagnetic field: Global and
in controlling the frequency of geomagnetic reversals. Nature 401: 885–890. regional biases for 0–5 Ma. Geophysical Journal International 131: 643–666.
Gubbins D (1988) Thermal core–mantle interactions and the time-averaged Johnson CL and Constable CG (1998) Persistently anomalous Pacific geomagnetic
paleomagnetic field. Journal of Geophysical Research 93: 3413–3420. fields. Geophysical Research Letters 25: 1011–1014.
Gubbins D (1994) Geomagnetic polarity reversals: A connection with secular variation Johnson CL, Wijbrans JR, Constable CG, et al. (1998) 40Ar/39Ar ages and
and core–mantle interaction? Reviews of Geophysics 32: 61–83. paleomagnetism of S eao Miguel lavas, Azores. Earth and Planetary Science Letters
Gubbins D (1999) The distinction between geomagnetic excursions and reversals. 160: 637–649.
Geophysical Journal International 137: F1–F3. Johnson CL, Constable C, Tauxe L, et al. (2008) Recent investigations of the
Gubbins D (2004) Time Series Analysis and Inverse Theory for Geophysicists. 0–5 Ma geomagnetic field recorded by lava flows. Geochemistry, Geophysics,
Academic Press. Geosystems 9. http://dx.doi.org/10.1029/2007GC001696.
Gubbins D and Bloxham J (1985) Geomagnetic field analysis—III. Magnetic fields on Kelly P and Gubbins D (1997) The geomagnetic field over the past 5 Myr. Geophysical
the core–mantle boundary. Geophysical Journal of Royal Astronomical Society Journal International 128: 315–330.
80: 695–713. Khokhlov A and Hulot G (2013) Probability uniformization and application to statistical
Gubbins D and Gibbons SJ (2003) Low Pacific secular variation. In: Channell JETC, paleomagnetic field models and directional data. Geophysical Journal International
Kent DV, Lowrie W, and Meert JG (eds.) Timescales of the Internal Geomagnetic 193: 110–121. http://dx.doi.org/10.1093/gji/ggs118.
Field. Geophysical Monograph Series, vol. 145, pp. 279–286. American Khokhlov A, Hulot G, and Bouligand C (2006) Testing statistical palaeomagnetic field
Geophysical Union, 10.1029/145GM07. models against directional data affected by measurement errors. Geophysical
Gubbins D and Kelly P (1993) Persistent patterns in the geomagnetic field over the past Journal International 167: 635–648.
2.5 Myr. Nature 365: 829–832. Kokhlov A, Hulot G, and Carlut J (2001) Towards a self-consistent approach to
Gubbins D and Richards M (1986) Coupling of the core dynamo and mantle: Thermal or palaeomagnetic field modelling. Geophysical Journal International 145: 157–171.
topographic? Geophysical Research Letters 13: 1521–1524. Kono M and Roberts PH (2002) Recent geodynamo simulations and observations of the
Gubbins D and Roberts N (1983) Use of the frozen-flux approximation in the geomagnetic field. Reviews of Geophysics 40. http://dx.doi.org/
interpretation of archaeomagnetic and palaeomagnetic data. Geophysical Journal of 10.1029/2000RG000102.
Royal Astronomical Society 73: 675–687. Kono M and Tanaka H (1995) Mapping the Gauss coefficients to the pole and the
Gubbins D and Zhang K (1993) Symmetry properties of the dynamo equations for models of paleosecular variation. Journal of Geomagnetism and Geoelectricity
paleomagnetism and geomagnetism. Physics of the Earth and Planetary Interiors 47: 115–130.
75: 225–241. Korte M and Constable CG (2005) Continuous geomagnetic models for the past 7
Gubbins D, Willis AP, and Sreenivasan B (2007) Correlation of Earth’s magnetic field millennia II: CALS7K. Geochemistry, Geophysics, Geosystems 6(2): Q02H16.
with lower mantle thermal and seismic structure. Physics of the Earth and Planetary http://dx.doi.org/10.1029/2004GC000801.
Interiors 162: 256–260. Korte M and Constable CG (2006) Centennial to millennial geomagnetic secular
Hale CJ and Dunlop DJ (1984) Evidence for an early Archean geomagnetic field: A variation. Geophysical Journal International 167: 43–52.
paleomagnetic study of the Komati Formation, Barberton greenstone belt, South Korte M and Constable CG (2011) Improving geomagnetic field reconstructions for
Africa. Geophysical Research Letters 11: 97–100. 0–3 ka. Physics of the Earth and Planetary Interiors 188: 247–259. http://dx.doi.
Harrison CGA (1980) Secular variation and excursions of the Earth’s magnetic field. org/10.1016/j.pepi.2011.06.017.
Journal of Geophysical Research 85: 3511–3522. Korte M and Holme R (2010) On the persistence of geomagnetic flux lobes in
Hatakeyama T and Kono M (2001) Shift of the mean magnetic field values: global Holocene field models. Physics of the Earth and Planetary Interiors
Effect of scatter due to secular variation and errors. Earth, Planets and Space 182: 179–186.
53: 31–44. Labrosse S (2003) Thermal and magnetic evolution of the Earth’s core. Physics of the
Hatakeyama T and Kono M (2002) Geomagnetic field models for the last 5 Myr: Time- Earth and Planetary Interiors 140: 127–143.
averaged field and secular variation. Physics of the Earth and Planetary Interiors Laj C, Mazaud A, Weeks R, Fuller M, and Herrero-Bervera E (1991) Geomagnetic
133: 181–215. reversal paths. Nature 351: 447.
Lawrence K, Constable CG, and Johnson CL (2006) Paleosecular variation and the Olson PL, Christensen UR, and Driscoll PE (2012) From superchrons to secular
average geomagnetic field at 20 latitude. Geochemistry, Geophysics, variation: A broadband dynamo frequency spectrum for the geomagnetic dipole.
Geosystems 7. http://dx.doi.org/10.1029/2005GC001181. Earth and Planetary Science Letters 319–320: 75–82.
Lay T and Garnero E (2011) Deep mantle seismic modeling and imaging. Annual Olson P, Deguen R, Hinnov LA, and Zhong S (2013) Controls on geomagnetic reversals
Review of Earth and Planetary Sciences 39: 91–123. and core evolution by mantle convection in the Phanerozoic. Physics of the Earth
Lay T, Hernlund J, and Buffett BA (2008) Core–mantle boundary heat flow. Nature and Planetary Interiors 214: 87–103.
Geoscience 1: 25–32. Olson P and Glatzmaier GA (1996) Magnetoconvection and thermal coupling of Earth’s
Lee S (1983) A study of the time-averaged paleomagnetic field for the past 195 million core and mantle. Philosophical Transactions of the Royal Society London
years. Phd Thesis, Australian National University. A354: 1413–1424.
Lock J and McElhinny M (1991) The global paleomagnetic database: Design, Opdyke ND and Henry KW (1969) A test of the dipole hypothesis. Earth and Planetary
installation, and use with ORACLE. Surveys in Geophysics 12: 317–491. Science Letters 6: 139–151.
Love JJ (2000) Paleomagnetic secular variation as a function of intensity. Philosophical Parker RL (1994) Geophysical Inverse Theory. Princeton University Press.
Transactions of the Royal Society of London, Series A 358: 1191–1223. Prévot M and Camps P (1993) Absence of preferred longitude sectors for poles from
Love JJ and Constable CG (2003) Gaussian statistics for paleomagnetic vectors. volcanic records of geomagnetic reversals. Nature 366: 53–56.
Geophysical Journal International 152: 515–565. Quidelleur X and Courtillot V (1996) On low degree spherical harmonic models of
Masters G, Johnson S, Laske G, and Bolton H (1996) A shear-wave velocity model of paleosecular variation. Physics of the Earth and Planetary Interiors 95: 55–77.
the mantle. Philosophical Transactions of the Royal Society of London, Series A Quidelleur X, Valet J-P, Courtillot V, and Hulot G (1994) Long-term geometry of the
354: 1385–1411. geomagnetic field for the last 5 million years; an updated secular variation database
McElhinny MW (2004) Geocentric axial dipole hypothesis: A least squares perspective. from volcanic sequences. Geophysical Research Letters 21: 1639–1642.
In: Channell JETC, Kent DV, Lowrie W, and Meert JG (eds.) Timescales of the Roberts PH and Glatzmaier GA (2000) Geodynamo theory and simulations. Reviews of
Internal Geomagnetic Field. Geophysical Monograph Series, vol. 145, pp. 101–115. Modern Physics 72: 1081–1123.
American Geophysical Union. http://dx.doi.org/10.1029/145GM07. Runcorn SK (1992) Polar path in geomagnetic reversals. Nature 356: 654–656.
McElhinny M and Lock J (1993) Global paleomagnetic database supplement number Schneider DA and Kent DV (1988) Inclination anomalies from Indian Ocean sediments
one, update to 1992. Surveys in Geophysics 14: 303–329. and the possibility of a standing non-dipole field. Journal of Geophysical Research
McElhinny M and Lock J (1996) IAGA paleomagnetic databases with Access. Surveys in 93: 11621–11630.
Geophysics 14: 303–329. Schneider DA and Kent DV (1990) The time-averaged paleomagnetic field. Reviews of
McElhinny MW and McFadden PL (1997) Palaeosecular variation over the past Geophysics 28: 71–96.
5 Myr based on a new generalized database. Geophysical Journal International Selkin P and Tauxe L (2000) Long term variations in paleointensity. Philosophical
131: 240–252. Transactions of the Royal Society of London, Series A 358: 1065–1088.
McElhinny MW and Merrill RT (1975) Geomagnetic secular variation over the past Sreenivasan B and Gubbins D (2011) On mantle-induced heat flow variations at the
5 My. Reviews of Geophysics and Space Physics 13: 687–708. inner core boundary. Physics of the Earth and Planetary Interiors 187: 336–341.
McElhinny MW, McFadden PL, and Merrill RT (1996a) The time averaged Sreenivasan B and Jones CA (2005) Structure and dynamics of the polar vortex in the
paleomagnetic field 0–5 Ma. Journal of Geophysical Research 101: 25007–25027. Earth’s core. Geophysical Research Letters 32, L20301.
McElhinny MW, McFadden PL, and Merrill RT (1996b) The myth of the Pacific dipole Sreenivasan B and Jones CA (2006) Azimuthal winds, convection and dynamo action in
window. Earth and Planetary Science Letters 143: 13–22. the polar regions of planetary cores. Geophysical and Astrophysical Fluid Dynamics
McFadden PL and McElhinny MW (1984) A physical model for paleosecular variation. 100: 319–339.
Geophysical Journal of Royal Astronomical Society 67: 19–33. Stone D and Layer P (2006) Paleosecular variation and GAD studies of 02 Ma flow
McFadden PL and Reid AB (1982) Analysis of paleomagnetic inclination data. sequences from the Aleutian Islands, Alaska. Geochemistry, Geophysics,
Geophysical Journal of Royal Astronomical Society 69: 307–319. Geosystems 7(4). http://dx.doi.org/10.1029/2005GC001007.
McFadden PL, Merrill RT, and McElhinny MW (1988) Dipole/quadrupole family Tanaka H (1999) Circular asymmetry of the paleomagnetic directions observed at low
modelling of paleosecular variation. Journal of Geophysical Research latitude volcanic sites. Earth, Planets and Space 51: 1279–1285.
93: 11588–11853. Tanaka H, Kono M, and Uchimura H (1995) Some global features of paleointensity in
McFadden PL, Barton CE, and Merrill RT (1993) Do virtual geomagnetic poles follow geological time. Geophysical Journal International 120: 97–102.
preferred paths during geomagnetic reversals? Nature 361: 344–346. Tarduno JA, Cottrell RD, Watkeys MK, and Bauch D (2007) Geomagnetic field strength
McMillan DG, Constable CG, Parker RL, and Glatzmaier GA (2001) A statistical 3.2 billion years ago recorded by single silicate crystals. Nature 446: 657–660.
appraisal geodynamo models. Geochemistry, Geophysics, Geosystems 2. http://dx. http://dx.doi.org/10.1038/nature05667.
doi.org/10.129/2000GC000130. Tarduno JA, Cottrell RD, Watkeys MK, et al. (2010) Geodynamo, solar wind, and
Mejia V, B€ohnel H, Opdyke ND, Otega-Rivera MA, Lee JKW, and Aranda-Gomez JJ magnetopause 3.4 to 3.45 billion years ago. Science 327: 1238–1240.
(2005) Paleosecular variation and time-averaged field recorded in late Pliocene– Tauxe L and Kent D (2004) A simplified statistical model for the geomagnetic field and
Holocene lava flows from Mexico. Geochemistry, Geophysics, Geosystems 6(7). the detection of shallow bias in paleomagnetic inclinations. Was the ancient
http://dx.doi.org/10.1029/2004GC000871. magnetic field dipolar? In: Channell JETC, Kent DV, Lowrie W, and Meert JG (eds.)
Merrill RT and McElhinny MW (1977) Anomalies in the time averaged paleomagnetic Timescales of the Internal Geomagnetic Field. Geophysical Monograph Series,
field and their implications for the lower mantle. Reviews of Geophysics and Space vol. 145, pp. 101–115. American Geophysical Union, 10.1029/145GM07.
Physics 15: 309–323. Tauxe L, Staudigel H, and Wijbrans J (2000) Paleomagnetism and 40Ar/39Ar ages from
Merrill R and McFadden P (2003) The geomagnetic axial dipole field assumption. La Palma in the Canary Islands. Geochemistry, Geophysics, Geosystems 1. ISSN:
Physics of the Earth and Planetary Interiors 139: 171–185. 1525-2027.
Merrill RT, McElhinny MW, and Stevenson DJ (1979) Evidence for long-term Tauxe L, Constable C, Johnson CL, Koppers AAP, Miller WR, and Staudigel H (2003)
asymmetries in the earth’s magnetic field and possible implications for dynamo Paleomagnetism of the Southwestern USA Recorded by 0–5 Ma igneous rocks.
theories. Physics of the Earth and Planetary Interiors 20: 75–82. Geochemistry, Geophysics, Geosystems 4(4): 8802. http://dx.doi.org/
Merrill RT, McFadden PL, and McElhinny MW (1990) Paleomagnetic tomography 10.1029/2002GC000343.
of the core–mantle boundary. Physics of the Earth and Planetary Interiors Tauxe L, Gans P, and Mankinen E (2004a) Paleomagnetism and 40Ar /39Ar ages from
64: 87–101. volcanics extruded during the Matuyama and Brunhes Chrons near McMurdo
Merrill RT, McElhinny MW, and McFadden PL (1996) The Magnetic Field of the Earth: Sound, Antarctica. Geochemistry, Geophysics, Geosystems 5(6). http://dx.doi.org/
Paleomagnetism, the Core and the Deep Mantle. Academic Press. 10.1029/2003GC000656.
Néel L (1949) Théorie du trainage magnétique des ferromagnétiques en grains fins avec Tauxe L, Luskin C, Selkin P, Gans P, and Calvert A (2004b) Paleomagnetic results from
applications aux terres cuites. Annals of Geophysics 5: 99–136. the Snake River Plain: Contribution to the time-averaged field global database.
Olson P and Aurnou J (1999) A polar vortex in the Earth’s core. Nature 402: 170–173. Geochemistry, Geophysics, Geosystems 5(8). http://dx.doi.org/
Olson P and Deguen R (2012) Lopsided inner core growth and eccentricity of the 10.1029/2003GC000661.
geomagnetic dipole. Nature Geoscience 5: 565–569. To A, Fukao Y, and Tsuboi S (2011) Evidence for a thick and localized ultra low shear
Olson PL, Coe RS, Driscoll PE, Glatzmaier GA, and Roberts PH (2010) Geodynamo velocity zone at the base of the mantle beneath the central Pacific. Physics of the
reversal frequency and heterogeneous core–mantle boundary heat flow. Physics of Earth and Planetary Interiors 184: 119–133.
the Earth and Planetary Interiors 180: 6679. http://dx.doi.org/10.1016/j. Valet J-P (2003) Time variations in geomagnetic intensity. Reviews of Geophysics 41. http://
pepi.2010.02.010. dx.doi.org/10.1029/2001RG000104.
Valet J-P, Meynadier L, and Guyodo Y (2005) Geomagnetic dipole strength and reversal Willis AP, Sreenivasan B, and Gubbins D (2007) Thermal core–mantle interaction:
rate over the past two million years. Nature 435: 802–805. Exploring regimes for ‘locked’ dynamo action. Physics of the Earth and Planetary
Valet JP, Tucholka P, Courtillot V, and Meynadier L (1992) Paleomagnetic Interiors 165: 83–92.
constraints on the geometry of the geomagnetic field during reversals. Nature Wilson RL (1970) Permanent aspects of the earth’s non-dipole magnetic field over
356: 400–407. Upper Tertiary times. Geophysical Journal of Royal Astronomical Society
Van der Voo R and Torsvik TH (2001) Evidence for late Paleozoic and Mesozoic non- 19: 417–439.
dipole fields provides an explanation for the Pangea reconstruction problems. Earth Wilson RL (1971) Dipole offset—The time-average palaeomagnetic filed over the past
and Planetary Science Letters 187: 71–81. 25 million years. Geophysical Journal of Royal Astronomical Society 22: 491–504.
Vandamme D (1994) A new method to determine paleosecular variation. Physics of the Ziegler LB, Constable CG, Johnson CL, and Tauxe L (2011) PADM2M: A penalized
Earth and Planetary Interiors 85: 131–142. maximum likelihood model of the 0–2 Ma paleomagnetic axial dipole moment.
Wicht J (2002) Inner-core conductivity in numerical dynamo simulations. Physics of the Geophysical Journal International 184: 1069–1089. http://dx.doi.org/10.1016/j.
Earth and Planetary Interiors 132: 281–302. pepi.2008.07.027.
5.12 Source of Oceanic Magnetic Anomalies and the Geomagnetic
Polarity Timescale
JS Gee, University of California, San Diego, CA, USA
DV Kent, Rutgers University, Piscataway, NJ, USA

5.12.2 Magnetic Anomalies as Records of Geomagnetic Field Behavior 421
5.12.2.1 Polarity Reversals and the GPTS 421
5.12.2.2 Geomagnetic Intensity Fluctuations 425
5.12.2.3 Anomaly Skewness and Nondipole Field 428
5.12.3 Magnetic Source Regions 430
5.12.3.1 Magnetization of Lavas 431
5.12.3.1.1 Initial grain size and composition 431
5.12.3.1.2 Low-temperature alteration 433
5.12.3.2 Magnetization of Dikes 439
5.12.3.3 Magnetization of Gabbros 440
5.12.3.4 Mantle-Derived Peridotites 445
5.12.4 Crustal Accretion and Structure of the Magnetic Source 447
5.12.4.1 Nonvertical Magnetic Boundaries 447
5.12.4.2 Rotations of the Magnetic Source Layer 448
5.12.5 Future Directions 451
Acknowledgments 455
References 455
5.12.1 Introduction superfast spreading refer to full-spreading rates of <40,

40–90, 90–140, and >140 km My1, respectively. This simple
As new seafloor is created at the ridge crest, it cools and model highlights some important aspects of magnetic anoma-
acquires a thermoremanence that captures a record of past lies measured at the sea surface. First, the depth of magnetiza-
geomagnetic field variations. The dominant geomagnetic sig- tion source (varying from about 2.6 km at the ridge crest to
nal recorded in the oceanic crust is the pattern of field reversals; 5.6 km for the oldest seafloor; Stein and Stein, 1994) exerts a
the juxtaposition of normal and reverse polarity crust consti- fundamental control on the wavelengths that can be resolved
tutes a large magnetization contrast that generates readily iden- at the sea surface. The depth and thickness of the magnetized
tifiable variations in the magnetic field measured at the sea layer act as a band-pass filter (Earth filter of Schouten and
surface. The lineated and globally correlatable character of McCamy, 1972, with the least attenuation at wavelengths of
these sea-surface magnetic anomalies has proven remarkably 8–15 km for typical seafloor depths). Polarity intervals corre-
useful in documenting the pattern of geomagnetic reversals sponding to these spatial scales ( 100–200 ky for fast-
since 160 Ma. This reversal pattern, calibrated with suitable spreading crust) are particularly well represented in the
numerical ages, provides the basis for the geomagnetic polarity sea-surface anomaly profile (e.g., Anomaly 2; 1.77–1.95 Ma).
timescale (GPTS) that has broad-reaching applications Significantly shorter polarity intervals (104 years) are also
throughout Earth sciences including mapping the age distribu- recognizable in high-resolution profiles from fast-spreading
tion of the ocean floor constituting more than half of the ridges, although such brief magnetization changes appear as
Earth’s surface area. broader, substantially attenuated anomalies at the sea surface.
The inferred pattern of geomagnetic reversals has changed Thus, the combination of rapid spreading and typical seafloor
relatively little since the pioneering work of Heirtzler et al. depths of 3–4 km is well suited to capture essentially all polar-
(1968), who used an assumption of constant seafloor spread- ity intervals with durations exceeding 104 years in sea-surface
ing in the South Atlantic to derive the first GPTS. The stability anomaly data. Indeed, independent magnetostratigraphic
of polarity interval widths in subsequent timescales testifies to studies corroborate that the GPTS can be viewed as essentially
the regularity of seafloor spreading, at least for carefully chosen complete at timescales greater than about 30 ky.
spreading corridors. The anomaly profile in Figure 1, calcu- Establishing the origin and significance of anomaly fluctu-
lated from a simple block model with vertical polarity bound- ations at scales both longer and shorter than the dominant
aries and for a full-spreading rate of 140 km My1, illustrates timescale of reversals (104 106 years) can potentially provide
the optimal resolution that is possible at the fastest spreading valuable additional insights into geomagnetic field behavior.
known from the present ocean basins. Note that the terms To the extent that the magnetization of the oceanic crust is a
slow spreading, intermediate spreading, fast spreading, and thermoremanence, the magnetization of the oceanic crust

420 Source of Oceanic Magnetic Anomalies and the Geomagnetic Polarity Timescale
C1n C3n C5bn C6Cn C13r
0 10 20 30 40
N/P
C21n C24r C29r C33n
40 50 60 70 80
P/K
C34n KQZ CM0r
80 90 100 110 120

500
nT
JQZ
CM0r CM16n CM25n 0
?
120 130 140 150 160
K/J
Age (Ma)
Figure 1 Geomagnetic polarity timescale from marine magnetic anomalies for 0–160 Ma. After Lowrie and Kent (2004); largely based on Cande and
Kent (1995) and Channell et al. (1995). Filled and open blocks represent intervals of normal and reverse geomagnetic field polarity (Table 1); key
chrons that were used as calibration tiepoints are identified above the bar graph (C1n, C3n, etc.) and correlated positions of geologic period boundaries
are indicated by ticks below the bar graph (N/P, Neogene/Paleogene; P/K, Paleogene/Cretaceous; K/J, Cretaceous/Jurassic). Idealized magnetic
anomaly profile uses this polarity pattern in a 1 km thick source layer that has a mean magnetization of 5 A m1 with Gaussian variation (s ¼ 42% of
mean) to mimic paleointensity variations and vertical magnetization boundaries. The anomaly is calculated at the pole (no skewness, magnetization,
and ambient field direction vertical) at a full-spreading rate of 140 km My1, with depth to the upper surface that increases with age0.5 (Ma)
(Stein and Stein, 1994). Note the amplitude modulation, which primarily reflects the sequence effect since source properties otherwise stay constant
other than modest monotonic change in water depth, except for magnetization ramp from JQZ into M-sequence (dashed line; factor of 5 increase
from 157 to 143 Ma) that is likely due to a systematic increase in geomagnetic intensity (Cande et al., 1978; McElhinny and Larson, 2003).
might be expected to preserve a broad spectrum of geomag- preserve higher-frequency variations that are attenuated in sea-
netic field variations, from brief (<103 years) excursions or surface anomaly data. The coherence among near-bottom
intensity fluctuations to longer-term variations on the scale of anomaly profiles and the similarity of these coherent fluctua-
superchrons (107–108 years). Documenting the relative impor- tions with independent records of field intensity from sediments
tance of geomagnetic field fluctuations in controlling magnetic suggest that a substantial geomagnetic intensity signal may be
anomaly amplitudes has proved difficult, however, since these preserved in the oceanic crustal magnetization.
anomalies are the end product of the recording of a paleofield Some longer-term variations in anomaly amplitudes may
signal (e.g., paleointensity variations, directional excursions, likewise have a geomagnetic origin although there has been
and polarity reversals), modulated by crustal accretionary pro- less progress in differentiating this from other causes of varia-
cesses (e.g., variations in geochemistry or the pattern of lava tion in source properties. For example, the gradual increase in
accumulation), and geometry of the source region(s). Despite anomaly amplitudes following the Jurassic quiet zone (JQZ;
the limitations in accounting for the recording medium, signif- crust older than 154 Ma) has been attributed to an increase
icant progress has been made in documenting short-wavelength in paleofield intensity (Cande et al., 1978), whereas the ampli-
anomaly fluctuations that may be attributed to geomagnetic tude envelope observed over some ridge axes has long been
field behavior. A key element in many of these studies is the interpreted as evidence of low-temperature alteration of the
collection of anomaly data nearer the magnetic source layer to source layer (e.g., Bleil and Petersen, 1983; Raymond and
Source of Oceanic Magnetic Anomalies and the Geomagnetic Polarity Timescale 421
LaBrecque, 1987). The particular pattern of polarity reversals reversals, making the age-calibrated geomagnetic polarity
and the relative widths of adjacent polarity intervals (some- sequence the basis for global correlations and geochronology
times termed the sequence effect) can modulate anomaly for the past 160 My (Figure 1). The relative widths of the
amplitudes, for example, the variations in amplitudes that are magnetic polarity intervals for the Late Jurassic to Recent were
apparent over the past 40 My in the simple model shown in determined from magnetic profiles initially by Heirtzler et al.
Figure 1, so recognition of long-term trends is not straightfor- (1968) for the Central Anomaly (Anomaly 1) to Anomaly
ward. There are also first-order geomagnetic questions about 32 (C-sequence) and by Larson and Pitman (1972) for Anom-
the marked contrast in the character of magnetic anomalies in alies M1–M22 (M-sequence). Larson and Pitman recognized
the Cretaceous quiet zone (KQZ; 83–120.6 Ma) and the JQZ, that the M-sequence anomalies were bracketed by long inter-
the former often associated with large amplitude anomalies vals with apparently few or no reversals: the KQZ between the
and the latter with small amplitude anomalies, which may be M-sequence and the ridge crest C-sequence of Heirtzler et al.
related to significant differences in mean field strength in the (to which Larson and Pitman added Anomalies 33 and 34) and
Jurassic and the Cretaceous. the JQZ prior to the M-sequence (which was soon extended to
Despite more than 40 years of study, many aspects of the M0 at the younger end and to M25 at the older end by Larson
magnetization source responsible for lineated marine magnetic and Hilde (1975) and to M29 by Cande et al. (1978)). Key
anomalies remain uncertain. Early studies (Atwater and Mudie, aspects of the geomagnetic interpretations were verified by
1973; Talwani et al., 1971) indicated that the source layer is magnetostratigraphic studies. For example, Helsley and
thin and dominated by extrusives, whereas compilations of Steiner (1969) documented an interval of constant normal
lava magnetizations suggest that additional deeper sources polarity (the Cretaceous normal polarity superchron (CNPS))
may be required, and indeed, in some cases (e.g., near Ocean corresponding to the KQZ observed in magnetic anomaly pro-
Drilling Program (ODP) Hole 735B; Dick et al., 1991), recog- files. Similarly, Lowrie and Alvarez (1981) verified the overall
nizable anomaly lineations are present where only the intru- correspondence between magnetostratigraphic polarity inter-
sive portion of the crust is preserved. Since the last review of vals and the 100 My anomaly record from Anomaly 6C to
crustal magnetization (Smith, 1990), there has been significant Anomaly M3 including the CNPS and KQZ, and Ogg and
progress in the sampling of the dikes, lower crustal gabbros, Lowrie (1986) provided magnetostratigraphic confirmation
and upper mantle material exposed in tectonic windows, and for the central part of the M-sequence. Correlations between
these studies indicate that dikes and gabbros are likely signif- anomaly interpretations and magnetostratigraphy are being
icant contributors to sea-surface magnetic anomalies. In addi- refined in land sections and marine cores (e.g., Billups et al.,
tion to the need for a fundamental characterization of possible 2004; Channell et al., 2003; Lanci et al., 2004, 2005; Speranza
magnetization sources within the crust, some understanding of et al., 2005) and provide the linkage to numerical ages based
the details of the accretionary process (e.g., width of the neo- on radioisotopic dates or astronomical cyclicity in stratigraphic
volcanic zone and off-axis volcanism) is also essential since sections for age calibration of the anomaly sequence in the
these provide an intrinsic limit on the fidelity of the crustal construction of GPTSs.
recording process. As our understanding of geomagnetic field Standard geomagnetic polarity chron nomenclature is
variations improves (particularly over timescales on the order based on the long-standing numbering schemes (sometimes
of 104 years over which much of the oceanic crust is con- with lettered additions) for prominent but irregularly spaced
structed), near-bottom magnetic anomaly data can be used to magnetic anomalies (Larson and Pitman, 1972; Pitman and
constrain aspects of crustal accretion. Such studies can, in turn, Heirtzler, 1966; Pitman et al., 1968). Polarity subdivisions for
be useful in evaluating the timescales at which geomagnetic the past 5 My based on compilations of radiometric dating of
information can be recovered from magnetic anomaly records. discrete lavas (Cox et al., 1964) were initially labeled after
In this chapter, we will focus primarily on the promise and prominent geomagneticians (Brunhes, Matuyama, Gauss, and
limitations of the oceanic crust as a recorder of geomagnetic field Gilbert for chrons, formerly called epochs; Anonymous, 1979)
variations, emphasizing the record of past geomagnetic field and type localities (Jaramillo, Olduvai, Mammoth, etc., for
variations recorded in anomalies (and therefore in source mag- subchrons, formerly called events), but this nomenclature sys-
netization) on timescales of 103 years (excursions) to 104–106 tem was impractical to be extended to the numerous older
(reversals) and 107–108 (superchrons). We review the origin of polarity intervals delineated by magnetic anomalies. By con-
the magnetization in the various crustal source layers responsible vention, identifiers usually refer to positive anomalies, which
for lineated magnetic anomalies and conclude by mentioning correspond to normal polarity for the C-sequence (Anomalies
some applications to deciphering how the oceanic crust formed 1–34) but mostly to reverse polarity intervals for the
and by speculating on future directions. The chapter is based M-sequence (M0–M29) because they formed in the Southern
mostly on published literature that appeared since the last major Hemisphere but are now in the Northern Hemisphere in the
review of oceanic crust magnetization by Smith (1990). Pacific where the M-sequence is well developed. A chron cor-
responds to the interval from the younger boundary of the
eponymous anomaly to the younger boundary of the preced-
5.12.2 Magnetic Anomalies as Records ing anomaly and has the prefix ‘C’ (e.g., Chron C3A or Chron
of Geomagnetic Field Behavior CM20). However, each of these chrons is usually subdivided
into the two constituent intervals of predominantly normal
5.12.2.1 Polarity Reversals and the GPTS
and reverse polarity, which are designated by adding to the
Marine magnetic anomalies (Vine, 1966; Vine and Matthews, name the suffix ‘n’ for normal polarity and ‘r’ for the preceding
1963) provide a complete record of geomagnetic polarity reverse polarity interval (e.g., Chron C3An and Chron C3Ar or
Chron CM20n and Chron CM20r). When these polarity Table 1 Geomagnetic polarity chrons from marine magnetic
chrons are further subdivided into shorter polarity intervals, anomalies
they are referred to as subchrons and identified by appending,
Reverse
from youngest to oldest, ‘.1,’ ‘.2,’ etc., to the polarity chron
Age range Normal polarity Age range polarity
name and adding an ‘n’ for normal polarity or an ‘r’ for reverse
(Ma) chrons (Ma) chrons
polarity (e.g., Chron C3An.1r or Chron CM20n.1r). Finally,
the designation ‘1,’ ‘2,’ etc., is used following a chron or 0.000–0.780 C1n 0.780–0.990 C1r.1r
subchron name to denote apparently brief geomagnetic fea- 0.990–1.070 C1r.1n 1.070–1.201 C1r.2r
tures corresponding to short-wavelength anomalies or ‘tiny 1.201–1.211 C1r.2r–1n 1.211–1.770 C1r.2r
wiggles,’ which, upon calibration, convert to durations less 1.770–1.950 C2n 1.950–2.140 C2r.1r
than 30 ky. In view of their uncertain origin, these globally 2.140–2.150 C2r.1n 2.150–2.600 C2r.2r
2.581–3.040 C2An.1n 3.040–3.110 C2An.1r
mapped geomagnetic features are referred to as cryptochrons
3.110–3.220 C2An.2n 3.220–3.330 C2An.2r
(Cande and Kent, 1992a); they can be elevated to subchron
3.330–3.580 C2An.3n 3.580–4.180 C2Ar
status by the addition of ‘n’ for normal polarity and ‘r’ for 4.180–4.290 C3n.1n 4.290–4.480 C3n.1r
reverse polarity if confirmed as polarity intervals in magnetos- 4.480–4.620 C3n.2n 4.620–4.800 C3n.2r
tratigraphic studies (e.g., Chron C1r.2r-1n). 4.800–4.890 C3n.3n 4.890–4.980 C3n.3r
Anomaly spacings for the Central Anomaly (Anomaly 1) to 4.980–5.230 C3n.4n 5.230–5.894 C3r
Anomaly 34 (younger end of KQZ) were comprehensively 5.894–6.137 C3An.1n 6.137–6.269 C3An.1r
refined by Cande and Kent (1992a) and for the M-sequence 6.269–6.567 C3An.2n 6.567–6.935 C3Ar
(M0–M29) by Channell et al. (1995). Cande and Kent (1992a) 6.935–7.091 C3Bn 7.091–7.135 C3Br.1r
used a combination of nine finite rotation poles to describe 7.135–7.170 C3Br.1n 7.170–7.341 C3Br.2r
7.341–7.375 C3Br.2n 7.375–7.432 C3Br.3r
seafloor spreading in the South Atlantic and 61 stacked profiles
7.432–7.562 C4n.1n 7.562–7.650 C4n.1r
distributed over the finite rotation pole intervals to develop a
7.650–8.072 C4n.2n 8.072–8.225 C4r.1r
continuous framework for the anomaly sequence; finer-scale 8.225–8.257 C4r.1n 8.257–8.606 C4r.2r*
information was derived from faster-spreading rate ridges in 8.606–8.664 C4r.2r–1n 8.664–8.699 C4r.2r*
the Pacific and Indian Oceans and inserted into the South 8.699–9.025 C4An 9.025–9.097 C4Ar.1r*
Atlantic pattern. For age calibration, Cande and Kent (1992a) 9.097–9.117 C4Ar.1r–1n 9.117–9.230 C4Ar.1r*
assumed that spreading rates in the South Atlantic were 9.230–9.308 C4Ar.1n 9.308–9.580 C4Ar.2r
smoothly varying and fit a cubic interpolation spline function 9.580–9.642 C4Ar.2n 9.642–9.740 C4Ar.3r
to a set of nine tiepoints that link radioisotopic ages with dis- 9.740–9.880 C5n.1n 9.880–9.920 C5n.1r
tances of correlative anomalies from the zero-age ridge axis to 9.920–10.949 C5n.2n 10.949–11.052 C5r.1r
11.052–11.099 C5r.1n 11.099–11.167 C5r.2r*
derive a GPTS for 0–83 Ma (Anomaly 34). The revised GPTS of
11.167–11.193 C5r.2r–1n 11.193–11.352 C5r.2r*
Cande and Kent (1995) (CK95) includes astrochronological
11.352–11.363 C5r.2r–2n 11.363–11.476 C5r.2r*
estimates for polarity reversals for the past 5.23 My (Hilgen, 11.476–11.531 C5r.2n 11.531–11.555 C5r.3r*
1991; Shackleton et al., 1990), which are negligibly different 11.555–11.584 C5r.3r–1n 11.584–11.935 C5r.3r*
(within 0.03 My) from the most recent retuning by Lourens 11.935–12.078 C5An.1n 12.078–12.184 C5An.1r
et al. (2004). For the M-sequence, Channell et al. (1995) com- 12.184–12.401 C5An.2n 12.401–12.678 C5Ar.1r
pared profiles from the Phoenix, Japanese, and Hawaiian linea- 12.678–12.708 C5Ar.1n 12.708–12.775 C5Ar.2r
tion sets and derived a representative anomaly sequence for M0 12.775–12.819 C5Ar.2n 12.819–12.991 C5Ar.3r
to M29 from a new block model for the Hawaiian lineations. 12.991–13.139 C5AAn 13.139–13.302 C5AAr
There are still relatively few age-diagnostic data available for 13.302–13.510 C5ABn 13.510–13.703 C5ABr
13.703–14.076 C5ACn 14.076–14.178 C5ACr
calibration of M-sequence anomalies – Channell et al. (1995)
14.178–14.612 C5ADn 14.612–14.800 C5ADr
used only three tiepoints that were regarded as reliable and
14.800–14.888 C5Bn.1n 14.888–15.034 C5Bn.1r
consistent with constant spreading on the new model of the 15.034–15.155 C5Bn.2n 15.155–16.014 C5Br
Hawaiian lineations to derive a GPTS (referred to by the authors 16.014–16.293 C5Cn.1n 16.293–16.327 C5Cn.1r
as CENT94) for 120.6 Ma (M0) to 157.53 Ma (M29). 16.327–16.488 C5Cn.2n 16.488–16.556 C5Cn.2r
The listing of polarity intervals given in Table 1 is the same as 16.556–16.726 C5Cn.3n 16.726–17.277 C5Cr
the one used for statistical analysis by Lowrie and Kent (2004) 17.277–17.615 C5Dn 17.615–17.793 C5Dr*
and is basically a combination of CK95 for the C-sequence and 17.793–17.854 C5Dr–1n 17.854–18.281 C5Dr*
CENT94 for the M-sequence. This polarity sequence for the past 18.281–18.781 C5En 18.781–19.048 C5Er
160 My is believed to be complete to a resolution of better than 19.048–20.131 C6n 20.131–20.518 C6r
20.518–20.725 C6An.1n 20.725–20.996 C6An.1r
30 ky and includes those relatively few cryptochrons that have
20.996–21.320 C6An.2n 21.320–21.768 C6Ar
been detected as short polarity intervals in high-resolution mag-
21.768–21.859 C6AAn 21.859–22.151 C6AAr.1r
netostratigraphic studies and elevated to subchrons, mostly in 22.151–22.248 C6AAr.1n 22.248–22.459 C6AAr.2r
the Cenozoic (see summary by Krijgsman and Kent, 2004). 22.459–22.493 C6AAr.2n 22.493–22.588 C6AAr.3r
Small-scale magnetic anomalies have been identified beyond 22.588–22.750 C6Bn.1n 22.750–22.804 C6Bn.1r
M29 (Handschumacher et al., 1988), most recently to M41 22.804–23.069 C6Bn.2n 23.069–23.353 C6Br
with an apparent age of 167 Ma (Sager et al., 1998; Tivey 23.353–23.535 C6Cn.1n 23.535–23.677 C6Cn.1r
et al., 2006). However, the character of the pre-M29 anomalies
resembles small-scale magnetic anomalies or tiny wiggles in the (Continued)
Table 1 (Continued) Table 1 (Continued)
Reverse Reverse
Age range Normal polarity Age range polarity Age range Normal polarity Age range polarity
(Ma) chrons (Ma) chrons (Ma) chrons (Ma) chrons
23.677–23.800 C6Cn.2n 23.800–23.999 C6Cn.2r 132.55–132.76 CM12n 132.76–133.51 CM12r.1r

23.999–24.118 C6Cn.3n 24.118–24.730 C6Cr 133.51–133.58 CM12r.1n 133.58–133.73 CM12r.2r
24.730–24.781 C7n.1n 24.781–24.835 C7n.1r 133.73–133.99 CM12An 133.99–134.08 CM12Ar
24.835–25.183 C7n.2n 25.183–25.496 C7r 134.08–134.27 CM13n 134.27–134.53 CM13r
25.496–25.648 C7An 25.648–25.678 C7Ar* 134.53–134.81 CM14n 134.81–135.57 CM14r
25.678–25.705 C7Ar–1n 25.705–25.823 C7Ar* 135.57–135.96 CM15n 135.96–136.49 CM15r
25.823–25.951 C8n.1n 25.951–25.992 C8n.1r 136.49–137.85 CM16n 137.85–138.50 CM16r
25.992–26.554 C8n.2n 26.554–27.027 C8r 138.50–138.89 CM17n 138.89–140.51 CM17r
27.027–27.972 C9n 27.972–28.283 C9r 140.51–141.22 CM18n 141.22–141.63 CM18r
28.283–28.512 C10n.1n 28.512–28.578 C10n.1r 141.63–141.78 CM19n.1n 141.78–141.88 CM19n.1r
28.578–28.745 C10n.2n 28.745–29.401 C10r 141.88–143.07 CM19n 143.07–143.36 CM19r
29.401–29.662 C11n.1n 29.662–29.765 C11n.1r 143.36–143.77 CM20n.1n 143.77–143.84 CM20n.1r
29.765–30.098 C11n.2n 30.098–30.479 C11r 143.84–144.70 CM20n.2n 144.70–145.52 CM20r
30.479–30.939 C12n 30.939–33.058 C12r 145.52–146.56 CM21n 146.56–147.06 CM21r
33.058–33.545 C13n 33.545–34.655 C13r 147.06–148.57 CM22n.1n 148.57–148.62 CM22n.1r
34.655–34.940 C15n 34.940–35.343 C15r 148.62–148.67 CM22n.2n 148.67–148.72 CM22n.2r
35.343–35.526 C16n.1n 35.526–35.685 C16n.1r 148.72–148.79 CM22n.3n 148.79–149.49 CM22r
35.685–36.341 C16n.2n 36.341–36.618 C16r 149.49–149.72 CM22An 149.72–150.04 CM22Ar
36.618–37.473 C17n.1n 37.473–37.604 C17n.1r 150.04–150.69 CM23n.1n 150.69–150.91 CM23n.1r
37.604–37.848 C17n.2n 37.848–37.920 C17n.2r 150.91–150.93 CM23n.2n 150.93–151.40 CM23r
37.920–38.113 C17n.3n 38.113–38.426 C17r 151.40–151.72 CM24n.1n 151.72–151.98 CM24n.1r
38.426–39.552 C18n.1n 39.552–39.631 C18n.1r 151.98–152.00 CM24n.2n 152.00–152.15 CM24r
39.631–40.130 C18n.2n 40.130–41.257 C18r 152.15–152.24 CM24An 152.24–152.43 CM24Ar
41.257–41.521 C19n 41.521–42.536 C19r 152.43–153.13 CM24Bn 153.13–153.43 CM24Br
42.536–43.789 C20n 43.789–46.264 C20r 153.43–154.00 CM25n 154.00–154.31 CM25r
46.264–47.906 C21n 47.906–49.037 C21r 154.31–155.32 CM26n 155.32–155.55 CM26r
49.037–49.714 C22n 49.714–50.778 C22r 155.55–155.80 CM27n 155.80–156.05 CM27r
50.778–50.946 C23n.1n 50.946–51.047 C23n.1r 156.05–156.19 CM28n 156.19–156.51 CM28r
51.047–51.743 C23n.2n 51.743–52.364 C23r 156.51–157.27 CM29n 157.27–157.53 CM29r
52.364–52.663 C24n.1n 52.663–52.757 C24n.1r
Geomagnetic polarity timescale tabulated from Cande and Kent (1995), with additional
52.757–52.801 C24n.2n 52.801–52.903 C24n.2r
short subchrons summarized by Lowrie and Kent (2004), for after the Cretaceous
52.903–53.347 C24n.3n 53.347–55.904 C24r
normal polarity superchron (CNPS), which corresponds to the Cretaceous quiet zone,
55.904–56.391 C25n 56.391–57.554 C25r
and from Channell et al. (1995) for before the CNPS. The names of chrons marked by
57.554–57.911 C26n 57.911–60.920 C26r
asterisks are repeated because the reverse polarity intervals contain one or two
60.920–61.276 C27n 61.276–62.499 C27r
cryptochrons that have been elevated to the status of new normal polarity subchrons.
62.499–63.634 C28n 63.634–63.976 C28r
63.976–64.745 C29n 64.745–65.578 C29r
65.578–67.610 C30n 67.610–67.735 C30r
67.735–68.737 C31n 68.737–71.071 C31r
71.071–71.338 C32n.1n 71.338–71.587 C32n.1r Cenozoic, which have generally been attributed to paleointen-
71.587–73.004 C32n.2n 73.004–73.291 C32r.1r sity variations rather than geomagnetic reversals (e.g., Cande
73.291–73.374 C32r.1n 73.374–73.619 C32r.2r and Kent, 1992b). Accordingly, the pre-M29 anomalies along
73.619–79.075 C33n 79.075–83.000 C33r with the cryptochrons in the Cenozoic that have not been
83.00–120.60 C34n (CNPS) 120.60–121.00 CM0r confirmed as polarity intervals are not included in Table 1 or
121.00–123.19 CM1n 123.19–123.55 CM1r shown in Figure 1.
123.55–124.05 CM2n 124.05–125.67 CM3r
A recently published timescale (GTS2004; Gradstein et al.,
125.67–126.57 CM4n 126.57–126.91 CM5r
2004) utilizes the distances to anomalies in the South Atlantic
126.91–127.11 CM6n 127.11–127.23 CM6r
127.23–127.49 CM7n 127.49–127.79 CM7r from Cande and Kent (1992a) and Cande and Kent (1995) but
127.79–128.07 CM8n 128.07–128.34 CM8r with a somewhat different array of age calibration tiepoints. The
128.34–128.62 CM9n 128.62–128.93 CM9r largest difference between CK95 and GTS2004 is less than 1 My
128.93–129.25 CM10n 129.25–129.63 CM10r and occurs at chron C6Cn.2n, which was set at 23.8 1 Ma in
129.63–129.91 CM10Nn.1n 129.91–129.95 CM10Nn.1r CK95 but has been recalibrated using astrochronology to
129.95–130.22 CM10Nn.2n 130.22–130.24 CM10Nn.2r 23.03 0.04 Ma (Lourens et al., 2004; Shackleton et al.,
130.24–130.49 CM10Nn.3n 130.49–130.84 CM10Nr 2000). CK95 also maintains age registry with the compilation
130.84–131.50 CM11n 131.50–131.71 CM11r.1r of Cenozoic cryptochrons by Cande and Kent (1992b), which
131.71–131.73 CM11r.1n 131.73–131.91 CM11r.2r
were not migrated to the chronology of GTS2004. For the
131.91–132.35 CM11An.1n 132.35–132.40 CM11An.1r
M-sequence, age calibration of M0 is a major difference:
132.40–132.47 CM11An.2n 132.47–132.55 CM11Ar
GTS2004 sets the age of Chron CM0 (base of Aptian) at about
125 Ma, which is 4 My older than in the CENT94 timescale of of the reversal frequency regimes. According to Lowrie and Kent
Channell et al. (1995) and needs to be corroborated because of (2004), mean polarity interval lengths are 0.25 My for C1n–
the profound implications for the rate of seafloor spreading in C12r, 0.75 My for C13n–C32r, and 0.415 My for M0–M29.
the KQZ (e.g., Cogné and Humler, 2006). Despite the large differences in mean polarity interval length
On a plot of the age of each reversal against the order of its and taking into account a finite reversal transition time of a few
occurrence (Figure 2), the CENT94 data define a straight line, thousand years, the distributions of chron lengths within each
whereas the CK95 C-sequence can be divided into two nearly of the regimes are not significantly different from a Poisson or
linear segments that intersect near Chron C12r at about 32 Ma. exponential distribution (Lowrie and Kent, 2004), which
The 37.6 My long CNPS is evident as an abrupt discontinuity implies that the reversal process is essentially free of memory
between the C-sequence and the M-sequence and is the longest (Cox, 1968). It has been suggested that the CNPS may represent
polarity interval in the entire 160 My long sequence. Polarity part of a continuous, long-term evolution of reversal rate
intervals C33n and C33r that immediately follow the CNPS are (McFadden and Merrill, 2000). However, the lack of precursory
the second and third longest chrons and may have a closer field behavior that might have heralded this superchron (Gallet
affinity to the CNPS than to the rest of the reversal sequence. and Hulot, 1997; Lowrie and Kent, 2004) suggests that the
Linear segments imply that the reversal process is stationary and CNPS may represent either an abrupt perturbation of the rever-
allow calculation of representative statistical parameters for each sal process or a separate (non)reversal regime.
160
CENT94 CM29
140
120 CM0 M-sequence
CNPS
100
Age (Ma)
C-sequence
80
C32r
60
40
C12n
CK95(2)
20
CK95(1)
0
0 50 100 150 200 250
6
CNPS =
37.6 My
5
CK95(1) CK95(2) CENT94
L = + 0.211 7 L = – 0.030 0 L = + 0.665 5
+ 0.000 6N + 0.005 0N – 0.001 1N
Chron length, L (My)
4 R = 0.095 9 R = 0.119 9 R = 0.069 8

Mean = 0.248 My Mean = 0.749 My Mean = 0.415 My
3
0
0 50 100 150 200 250
Reversal number, N
Figure 2 Top, ages of magnetic reversals (Ma), and bottom, polarity chron lengths (My) in the Cenozoic and Mesozoic oceanic anomaly sequences
plotted against their age order. Statistical properties are shown for two segments of the CK95 timescale and the CENT94 timescale in which linear
fits have slopes not distinct from zero. CNPS, Cretaceous normal polarity superchron. Reproduced from Lowrie W and Kent DV (2004) Geomagnetic
polarity timescales and reversal frequency regimes. In: Channell JET, Kent DV, Lowrie W, and Meert J (eds.) AGU Geophysical Monograph 145:
Timescales of the Paleomagnetic Field, pp. 117–129. Washington, DC: American Geophysical Union.
5.12.2.2 Geomagnetic Intensity Fluctuations magnetized layer, (2) variations in magnetization intensity as a
result of alteration, and (3) geomagnetic intensity fluctuations
Sea-surface magnetic anomaly profiles from fast-spreading
(Gee and Kent, 1994; Perram et al., 1990; Tivey and Johnson,
ridges allow the recognition of polarity intervals as short as
1987). Variations in the thickness of the extrusive magnetic
30 ky. Yet the geomagnetic field is known to fluctuate on
source layer might arise from an axial keel of less magnetic
significantly shorter timescales. For example, direct observa-
dikes or from elevated temperature that may exceed the Curie
tions indicate that the dipole field is decreasing at 15 nT year1
point of the Ti-rich titanomagnetite in the extrusives (Shah
so that g01 has decreased about 8% since 1832 when Gauss first
et al., 2003). The magnetization of seafloor basalts can be
estimated the field intensity from a spherical harmonic analy-
substantially reduced by alteration in localized hydrothermal
sis ( Jackson et al., 2000). Absolute paleointensities from arch-
upflow zones (e.g., Tivey and Johnson, 2002; Tivey et al.,
aeomagnetic materials suggest that field intensities at 1–3 ka
1996). As will be discussed further in Section 5.12.3.1, the
were approximately 40–50% higher than the present field
long-standing notion that low-temperature alteration of the
(Yang et al., 2000) although nondipolar field variations
magnetic source is the dominant process in controlling crustal
undoubtedly result in considerable spatial variability (Korte
magnetization has significantly influenced the interpretation
and Constable, 2005). On somewhat longer timescales, more
of magnetic anomaly data both in the near-ridge environment
sparse absolute intensity data from volcanic materials (see
(e.g., Schouten et al., 1999) and on much longer timescales
Chapter 1.14) provide evidence for extremely low field values
(e.g., Bleil and Petersen, 1983; Zhou et al., 2001).
(20% or less of the present value) at approximately 40 ka
Although a variety of mechanisms undoubtedly influence
associated with the Laschamp excursion (Levi et al., 1990;
magnetic anomaly amplitudes, near-bottom anomaly data and
Roperch et al., 1988).
associated absolute paleointensities from the superfast-
Short-term ( 102–103 years) fluctuations in geomagnetic
spreading southern East Pacific Rise (EPR) illustrate that geo-
intensity, even if quite substantial, will hardly be evident in
magnetic intensity variations are likely sufficient to account for
sea-surface anomaly data. Intensity variations on timescales of
much of the near-ridge anomaly signal (Figure 3). Models of
104 years may be detectable in sea-surface profiles from fast-
crustal alteration predict a monotonic decrease in
spreading ridges but will have greatly reduced amplitudes and
longer wavelengths as dictated by the Earth filter. In order to
document these higher-frequency anomaly variations (some of 12
which may reflect geomagnetic field behavior), it is useful to
make magnetic field observations nearer the magnetic source 10
VADM (1022 A m2)
(Larson and Spiess, 1969). One might well ask whether much
8 Present
additional information can be gained from these near-bottom field
records since even polarity intervals as short as 104 may be
6
recorded, albeit in filtered form, in sea-surface profiles from the
fastest-spreading ridges in the present ocean basins. The answer 4
is twofold. First, comparison of sea-surface and near-bottom
data can sometimes differentiate between short-wavelength 2
anomalies due to short reversals and those arising from inten-
Mag (A m–1)
sity fluctuations. For example, Bowers et al. (2001) had docu- 20

mented that many of the coherent tiny wiggles in sea-surface
profiles from Anomaly 5 evolve into more complex, but still
lineated, anomalies near the seafloor, suggesting these short- 0
wavelength anomalies represent geomagnetic intensity fluctu-
ations rather than discrete short polarity events (which should
be manifest by a higher-amplitude and simpler anomaly pat- –2.5
Depth (km)
Vehicle track
tern). A second benefit of near-bottom data is that two closely
–3.0
spaced short events (whether intensity variations or reversals)
can be resolved as distinct features but must be much more –3.5
widely separated to be differentiated in sea-surface
anomaly data.
–4 –2 0 2 4
Near-bottom magnetic anomaly data from near the axis of Distance from axis (km)
fast-spreading ridges are particularly useful for assessing any
geomagnetic signal, since they can provide high temporal res- Figure 3 Comparison of glass paleointensities (top) and inversion
olution over a time period where independent estimates of magnetization solution (middle) for near-bottom magnetic profile across
the East Pacific Rise at 19.75° S. For source-layer geometries (bottom),
geomagnetic intensity fluctuations are best known. Such
both a constant-thickness source (solid line) and a variable-thickness
anomaly data from fast-spreading ridges often show a short-
source (dashed line) approximating off-axis doubling of seismic layer 2A
wavelength (2–4 km) axial anomaly high with a superimposed thickness produce inversions indicating that a lower magnetization is
anomaly minimum or notch of even shorter wavelength (Gee required on-axis with flanking areas of high magnetization. Modified from
et al., 2000; Perram et al., 1990; Shah et al., 2003). Possible Gee JS, Cande SC, Hildebrand JA, Donnelly K, and Parker RL (2000)
mechanisms that have been advanced to explain this notch in Geomagnetic intensity variations over the past 780 kyr obtained from
the axial anomaly include (1) variations in the thickness of the near-seafloor magnetic anomalies. Nature 408: 827–832.
magnetization intensity away from the ridge axis. The initial
Anomaly (nT)
B/M
increase in anomaly amplitudes on the ridge flanks is thus
apparently at odds with low-temperature alteration being the 100
dominant process controlling crustal magnetization. Thinning
of the extrusive magnetic source layer at the ridge axis may also 0
(a)
contribute to the axial anomaly notch (Shah et al., 2003).
Anomaly (nT)
However, even a factor of 2 thinner extrusive source at the 500
ridge axis, as suggested by seismic data (e.g., Christeson et al.,
1992; Vera and Diebold, 1994), is apparently not sufficient to 0
account for the lower axial magnetization (Figure 3). Absolute
paleointensity data from surface glass samples near the ridge (b)
axis reveal a substantial variation in paleointensity (Bowles
et al., 2006; Gee et al., 2000; Mejia et al., 1996), with moderate
Mag. (A m–1)
paleointensity values at the ridge axis higher values 1–2 km off- 10
axis and much lower values farther from the ridge. This pattern
is remarkably similar to independent estimates of field inten-
0
sity variations established from archaeomagnetic and volcanic
materials, which document a substantial dipole moment (c)
increase from a low of 2 1022 A m2 at 40 ka to a peak of
VADM (1022 A m2)

12
11 1022 A m2, followed by a rapid decrease to the present
8
value of 7.9 1022 A m2 (e.g., Merrill et al., 1996). The simi-
larity of these paleointensity data, in both magnitude and 4
pattern, and the near-bottom anomaly data strongly suggests 0
that geomagnetic intensity fluctuations are an important, if not
the dominant, factor controlling anomaly amplitudes at many
ridges.
0 200 400 600 800
The comparison of geomagnetic intensity variations over the (d)
Age (ka)
past 800 ky from sedimentary records and near-bottom mag-
netic anomalies from the EPR at 19° S (Gee et al., 2000; Figure 4) Figure 4 Comparison of geomagnetic intensity variations over the past
also points to the potential importance of geomagnetic intensity 800 ky from sedimentary records and in sea-surface and near-bottom
variations in producing tiny wiggles. Absolute intensity magnetic anomalies from the East Pacific Rise at 19° S. (a) Stack of
sea-surface anomaly profiles coincident with near-bottom magnetic
information from well-dated archaeomagnetic and volcanic
anomaly (b) and inversion solution (c) stacks (see Gee et al., 2000, for
materials is too sparse prior to about 50 ka to allow the
details of inversion). Ages calculated assuming constant spreading
construction of a time series of field intensity fluctuations. How- rate and an age of 780 ky for Brunhes/Matuyama (B/M) boundary. Lower
ever, marine sediments can provide continuous, globally panel (d) shows Sint800 sedimentary relative paleointensity stack for
distributed records of relative paleointensity over longer time- 10–800 ky (Guyodo and Valet, 1999) combined with global arch-
scales. A stack of sedimentary relative paleointensity records aeomagnetic data for the past 10 ky (Merrill et al., 1996), all scaled
spanning the Brunhes (Guyodo and Valet, 1999) shows many as virtual axial dipole moment (VADM). Modified from Gee JS, Cande SC,
similarities with coherent anomaly (and magnetization) fluctu- Hildebrand JA, Donnelly K, and Parker RL (2000) Geomagnetic
ations on eight profiles from the southern EPR. Because these intensity variations over the past 780 kyr obtained from near-seafloor
anomaly profiles are separated by up to 60 km, it is unlikely that magnetic anomalies. Nature 408: 827–832.
crustal accretionary variables (variations in the source thickness
or geochemistry) would result in coherent anomaly variations.
Thus, variations in geomagnetic intensity are the most likely corresponding sea-surface anomaly data provide additional
cause of the coherent fluctuations in the near-bottom anomaly evidence that many short-wavelength anomalies reflect
profiles. paleointensity variations rather than short polarity reversals
The pattern of geomagnetic intensity fluctuations in the near- (Figure 6). Although the tiny wiggles in sea-surface anomaly
bottom data is also recognizable in many sea-surface anomaly profiles can conveniently be modeled as due to short reverse
profiles across the Central Anomaly. At the superfast-spreading polarity intervals (Blakely, 1974), these anomaly minima in
southern EPR, sea-surface anomalies reveal two broad minima the sea-surface data correspond to more complex but still
at 300 and 550 ky (Figure 4). A similar pattern is discernible lineated anomalies in the near-bottom data. For example, the
in many sea-surface profiles, particularly from intermediate- and sea-surface minima labeled 5.4 correspond to a local anomaly
fast-spreading ridges (Figure 5). The systematic variation of pro- maximum within a broader low in the near-bottom data
files of the Central Anomaly with spreading rate and their (Bowers et al., 2001). Numerous lineated magnetic lows were
similarity to progressively smoothed records of sediment- found in the 12 near-bottom profiles from the northeast
derived paleointensity suggest that the short-wavelength anom- Pacific. If modeled as short polarity reversals, the approxi-
alies within the Central Anomaly are global features represent- mately 1 My duration normal polarity interval would be
ing filtered variations in geomagnetic dipole intensity. interrupted by 26 reverse polarity intervals ranging from 1.2
Near-bottom anomaly profiles of Anomaly 5 in the north- to 19 ky in duration (Bowers et al., 2001). More than three-
east Pacific (Bowers et al., 2001) and comparison with the quarters of these would have durations less than 10 ky and so
Paleointensity model Paleointensity model

150 mm year–1 90 mm year–1
sotw02 15040042
pasc03 15040142
( – 49.22°, – 113.60°)
( – 29.87°, – 111.73°)
eel39
yaq7304 07010037 ( – 50.32°, 133.90°)
( – 31.16°, – 111.93°)
bm32b
dsdp92 15050092 ( – 50.29°, 133.49°)
( – 16.63°, – 113.07°)
Paleointensity model
Paleointensity model 60 mm year–1
120 mm year–1
Galapagos stack
risp1 15060029 ( – 85° to – 76°)
(6.82°, – 102.69°) Wilson and Hey (1981)
pasc03 15040142 tt–099

( – 26.46°, – 112.63°) (23.35°, – 108.52°)
80012103 l1281np 06050060

(11.07°, – 103.74°) (45.09°, – 130.18°)
pptu01 15040160 e9011 01210035

(13.14°, – 104.04°) ( – 26.31°, – 13.79°)
elt20 01020020 c2802 01010278

( – 44.49°, – 112.17°) ( – 26.69°, – 13.60°)
Paleointensity model Paleointensity model

90 mm year–1 30 mm year–1
(with bathymetry
of c2802)
– 60 – 40 – 20 0 20 40 60 – 20 0 20
Distance from axis (km) Distance from axis (km)
Figure 5 Sea-surface magnetic anomaly profiles illustrating the character of the Central Anomaly at different full-spreading rates and the similarity to
synthetic profiles (heavier lines) based on sedimentary paleointensity variations. All profiles have been reduced to the pole; actual profiles are
identified by the location of the ridge crossing and the NDGC designations where available. Modified from Gee J, Schneider DA, and Kent DV (1996)
Marine magnetic anomalies as recorders of geomagnetic intensity variations. Earth and Planetary Science Letters 144: 327–335.
would be comparable to or less than the time needed for two polarity subchrons (Evans et al., 2004). While the sedimentary
reversals (average reversal duration 7 ky; Clement, 2004). The data provide evidence for a small number of polarity fluctua-
implausibly large number and short duration of required tions, the bulk of the variations evident in near-bottom anom-
reversed polarity intervals suggest that most of the near-bottom aly records are likely to reflect intensity variations rather than
anomaly fluctuations reflect geomagnetic intensity fluctuations short polarity reversals. A linkage between directional excur-
rather than short polarity intervals. This interpretation is also sions and low intensities is also suggested by a statistical model
supported by the fact that some correlative sedimentary records of the geomagnetic field (Tauxe and Kent, 2004).
show no polarity reversals and yet have relative intensity vari- The geomagnetic intensity signal documented in the Central
ations that appear to correlate with the short-wavelength Anomaly and Anomaly 5 is likely to be a more general feature
anomaly features (Bowles et al., 2003). Other sedimentary of marine magnetic anomalies, since the thermoremanence in
records, however, reveal a small number of polarity fluctua- the oceanic crust should record geomagnetic intensity and
tions within Chron C5n (e.g., Evans and Channell, 2003; polarity information. Apparently, lineated, short-wavelength
Evans et al., 2004; Roberts and Lewin-Harris, 2000) though magnetic anomalies that most plausibly represent geomagnetic
the number, duration, and timing of these features are often intensity fluctuations have indeed been documented within
conflicting. In some cases, the inferred long duration (e.g., several portions of the C-sequence, for example, in sea-surface
23 and 28 ky events identified by Roberts and Lewin-Harris, anomaly profiles from Anomalies 12–13 (Cande and
2000) is incompatible with magnetic anomaly data (such long LaBrecque, 1974) and Anomalies 24–27 (Cande and Kent,
polarity intervals would be readily apparent in profiles from 1992b). They are mostly attributed to intensity fluctuations
fast-spreading ridges). In other cases, proposed events are suf- because attempts to find corresponding short polarity intervals
ficiently brief (5–11 ky) to not conflict with existing anomaly in sedimentary records have generally not been successful (e.g.,
data, though these events may represent excursions rather than Bowles et al., 2003; Evans and Channell, 2003; Hartl et al.,
proposed links between reversal rate and geomagnetic inten-

Northeast Pacific (84 km My–1)
sity; Tauxe and Hartl, 1997). Sea-surface magnetic anomalies
11 10 Ma
within the JQZ, arbitrarily delineated as older than anomaly
M25, typically have very low amplitudes, and there is a gradual
5.4 5.3 5.2 5.1 5.0 increase in amplitude, extending from at least M29 to around
M19, which may be field-related (Cande et al., 1978;
Sea surface
McElhinny and Larson, 2003; Figure 8(a)). Although the
lineated nature of even older M-sequence anomalies has been
43° 32′N documented over fast-spreading crust and used for tectonic
reconstructions (Handschumacher et al., 1988; Nakanishi
and Winterer, 1998; Nakanishi et al., 1989, 1992), magnetos-
43° 27′N tratigraphic evidence for polarity reversals over the time inter-
val represented by the JQZ, that is, prior to M25, is still
ambiguous (e.g., Channell et al., 1984; Juarez et al., 1994;
Steiner et al., 1985). These older anomalies resemble the kind
Stack of short-wavelength, low-amplitude anomalies observed in the
(n = 12) Cenozoic that have been attributed to paleointensity variations
(Cande and Kent, 1992b). In contrast, magnetic anomalies in
the KQZ can often be of high amplitude (e.g., Figure 8(b))
42° 40′N even though magnetostratigraphic studies and paleomagnetic
compilations provide strong evidence that the KQZ formed
during predominantly normal geomagnetic polarity (e.g.,
42° 38′N
Irving and Pullaiah, 1976; Lowrie and Alvarez, 1981). It is
not clear whether the anomalies in the KQZ are field-related
1000 (e.g., Cronin et al., 2001), but their large amplitude does
nT suggest that the field intensity during the KQZ may have been
10 km high (Tarduno et al., 2001; Tauxe and Staudigel, 2004), espe-
West Flank of East Pacific Rise (180 km My–1) cially compared with the JQZ, whose low-amplitude anomalies
11 10 Ma might be indicative of low average field intensity in the Jurassic
(McElhinny and Larson, 2003; Prévot et al., 1990).
5.4 5.3 5.2 5.1 5.0
125 Stack
nT 5.12.2.3 Anomaly Skewness and Nondipole Field
20 km In addition to the wealth of information on geomagnetic
Figure 6 Magnetic anomaly profiles for Anomaly 5. Top shows intensity and polarity reversals, the shape (skewness) of mag-
sea-surface stack compared to representative near-bottom profiles and netic anomalies can also provide an estimate of remanent
coincident near-bottom stack from intermediate-spreading (84 km My1 inclination that can, in turn, be used for tectonic or geomag-
full-rate) seafloor in the northeast Pacific; bottom shows sea-surface netic studies. The skewness of a magnetic anomaly is a function
stack from fast (180 km My1 full-rate)-spreading seafloor on the west of the effective inclinations (i.e., those projected in a plane
flank of the southern East Pacific Rise. Modified from Bowers NE,
perpendicular to the azimuth of the magnetized blocks) of
Cande SC, Gee JS, Hildebrand JA, and Parker RL (2001) Fluctuations
the ambient field and of the remanence of the source blocks
of the paleomagnetic field during chron C5 as recorded in near-bottom
marine magnetic anomaly data. Journal of Geophysical Research 106: (Schouten and Cande, 1976). Skewness amounts to a uniform
26379–26396. phase shift as a measure of the deviation in shape between the
observed magnetic anomalies and the ideal symmetrical pro-
file that would be produced in a vertical field by vertically
1993; Lanci and Lowrie, 1997; Schneider, 1995); these magne- oriented magnetizations separated by vertical source bound-
tostratigraphic data were summarized by Krijgsman and Kent aries. The determination of the skewness from anomaly pro-
(2004). Although the detailed pattern of pre-Brunhes intensity files, together with the known geometry of the spreading
variations is not known, the statistical properties of the recent lineation and ambient field, allows the remanent inclination
field (with 40% intensity fluctuations about the mean; Merrill to be estimated. For the Central Anomaly, Schneider (1988)
et al., 1996) provide a plausible model for generating short- applied this type of skewness analysis to 14 magnetic anomaly
wavelength anomaly variations (Cande and Kent, 1992b). The profiles from the Galápagos ridge to derive a precise estimate of
models shown in Figure 7 illustrate that coherent short-wave- the mean remanent inclination, which was found to differ by
length anomaly variations are likely to be present throughout several degrees from the expected field inclination, and indi-
the marine magnetic anomaly sequence. cated that there is a small but significant (5%) long-term
The amplitude and character of such short-wavelength nondipole (mainly axial quadrupole) contribution to the geo-
anomaly fluctuations, particularly in anomalous intervals centric axial dipole field (Figure 9). This result was confirmed
such as the KQZ and JQZ, may therefore provide important by an analysis of skewness in 203 profiles of the Central
clues about the origin of the geomagnetic field (e.g., by testing Anomaly from the global ridge system (Acton et al., 1996).
12
13 Chile
Rise
EL25
EL18
EL20
VI706
150
gamma
(a)
20 km
26 250
27 nT
0
82 mm yr–1
Southeast Pacific
25 km
54 mm yr–1 Indian Ocean
25 km
31 mm yr–1
North Pacific
25 km
(b)
Southeast
Pacific
82 mm yr–1
25 km
25 km
Indian Ocean
54 mm yr–1
200
nT
0
+1
Relative
dipole field
intensity
–1
(c)
Figure 7 Further examples of short-wavelength magnetic anomalies in sea-surface profiles (a) from between Anomalies 12 and 13 on Chile Rise and
(b) at various spreading rates from between Anomalies 26 and 27. A model for the source of the short-wavelength magnetic anomalies between
Anomalies 26 and 27 at different spreading rates is shown in (c) and assumes that the anomalies are due to random fluctuations in the dipole field
paleointensity whose statistical properties are compatible with observed paleointensity variations over the past 5 My. (a) Modified from Cande SC and
LaBrecque JL (1974) Behaviour of the Earth’s paleomagnetic field from small scale marine magnetic anomalies. Nature 247: 26–28. (b, c) Modified
from Cande SC and Kent DV (1992b) Ultrahigh resolution marine magnetic anomaly profiles: A record of continuous paleointensity variations. Journal of
Geophysical Research 97: 15075–15083.
In applications of the skewness method outside of the Cen- considered for interpretation of field behavior (Cande and
tral Anomaly (or Brunhes), where both isochronous bound- Kent, 1976). The skewness of magnetic anomalies that is not
aries of the source block contribute and thus simplify the accounted for by a simple block model is termed anomalous
interpretation of the anomaly shape, other aspects of the source skewness (Cande, 1976). While such anomalous skewness is a
that may not be symmetrically distributed to cancel out (e.g., complicating factor in extracting geomagnetic or tectonic infor-
sloping polarity boundaries and tectonic rotations) need to be mation from magnetic anomalies (e.g., Petronotis et al., 1992),
Jurassic Quiet Zone (JQZ)
M22 M25 M29

C1007
(Hawaiian)
50 km
V3214
(Phoenix)
50 km
V3214
500 (Japanese)
nT 50 km
Model
A m–1 50 km
10
Magnetization
–10
(a)
Cretaceous Quiet Zone (KQZ)
N (~95 Ma) 34y S

500
nT
Depth (km)
6
–400 –200 0 200 400
(b) Distance (km)
Figure 8 A comparison of sea-surface magnetic profiles from the Jurassic quiet zone (JQZ) in the western Pacific (top; after Cande et al., 1978) and the
Cretaceous quiet zone (KQZ) from the southwest Pacific (bottom; data from Pockalny et al., 2002). Representative anomaly profiles from the JQZ
from the Hawaiian, Phoenix, and Japanese lineation sets have been reduced to the pole. Note the pronounced decrease in amplitude from anomaly M22
to M29 and the low (<100 nT) amplitudes prior to M29. In contrast, many anomaly profiles within the KQZ show large amplitude variations that in
some cases are as large as known polarity reversals (e.g., Anomaly 34y – young end of Anomaly 34 (as for chrons in Table 1)).
the details of the anomaly shape provide strong constraints on was made on the slow-spreading Reykjanes ridge and yielded a
the geometry and tectonic deformation of the magnetization thickness of about 500 m, approximately the thickness of the
source that will be discussed further in Section 5.12.4. Asses- volcanic layer. Atwater and Mudie (1973) did a comparable
sing the fidelity of the geomagnetic field record in anomalies analysis using near-bottom magnetic profiles on the Gorda
requires an understanding of both these nonfield-related varia- Rise in the Pacific. They also found that a 500 m thick source
tions in source properties (e.g., transition zone width and layer could account for the seafloor-spreading magnetic anom-
geometry of polarity boundaries) and the intrinsic magnetic alies. Dredged and drilled samples of oceanic basalts have
properties of the various source layers. The nature and origin remanent magnetizations that are more or less compatible
of the remanent magnetization of these layers are the subject of with a relatively thin source layer but firm conclusions are
the following section. inhibited by the several orders of magnitude range in the
magnetization values (Lowrie, 1977).
Thicker and even deeper sources have also been invoked,
5.12.3 Magnetic Source Regions usually as a counterbalance to supposedly altered and dimin-
ished shallower sources to explain the apparent loss of fidelity
Analysis of sea-surface and near-bottom magnetic anomalies in earlier Cenozoic anomaly sequences (e.g., Blakely, 1976) or
suggests that most of the signal comes from the extrusive layer to produce anomalous skewness from sloping blocking
(Figure 10). For example, Talwani et al. (1971) determined the temperature isotherms (Cande and Kent, 1976). Although
mean magnetization of basement topography from sea-surface broader transition zone widths (and consequently some loss
magnetic surveys made along-strike, which allowed an esti- of short polarity events) might be expected if the slowly cooled
mate of the thickness of the layer responsible for the across- gabbroic layer is a substantial contributor, the still older
strike (seafloor-spreading) magnetic anomalies. The analysis (M-sequence) anomalies do not have markedly broader
sheeted dikes, and gabbros, plus serpentinized mantle

q = 220° peridotites – and show that sample remanence and anomaly
data can be reconciled by a three-layer model, with relatively
narrow transition zones in lavas and dikes and broader polarity
q =215° boundaries in the gabbros. Oceanic basement rocks acquire a
thermal remanent magnetization (TRM) during initial cooling.
The magnitude of the TRM not only is proportional to the
strength of ambient geomagnetic field but also depends on
q =210°
the geochemistry (e.g., iron content) of the rocks and magnetic
grain size. The rocks may alter with age, which may also modify
the TRM. The resultant magnetization measured on rock sam-
q =205°
ples in the laboratory is referred to as the natural remanent
magnetization (NRM). In order to facilitate the comparison of
magnetization values from different locations (dipole intensity
q =200°
varies by a factor of 2 from equator to pole), all magnetization
values are reported as equatorial values. Although the arith-
metic mean is most pertinent for comparison with magnetic
q =195° anomalies, most large magnetization datasets are characterized
by approximately lognormal distributions. Geometric mean
values are used where possible (many authors do not specify
q =190° how mean values were calculated) for comparisons since these
are less influenced by extreme values in smaller datasets that
are typical of many studies of oceanic rocks.
q =185°
5.12.3.1 Magnetization of Lavas

q =180°
Mid-ocean ridge basalts that constitute the oceanic extrusive
layer have a magnetization that is characteristically dominated
by a strong stable remanence compared to an induced compo-
2000 nT Original
nent, making these shallow seated rocks a traditional prime
100 Km source of seafloor-spreading magnetic anomalies. Titanium-
rich titanomagnetite (Fe3xTixO4, where x 0.6) that has
often undergone variable degrees of low-temperature oxida-
tion (maghemitization) is the principal magnetic mineral.
Model
However, it was recognized early on that the very wide range
of observed NRM intensities could not be explained by a mere
variation in low-temperature oxidation; instead, it is necessary
Figure 9 Representative sea-surface magnetic anomaly profile from
to invoke substantial variations in titanomagnetite composi-
the Galápagos Spreading Center illustrating the technique used to
determine skewness of the Central Anomaly. Arrow points to tion, concentration, and grain size (Lowrie, 1977).
approximate phase shift (y ¼ 207.5°) required to match the shape of the
Central Anomaly to the model profile formed at the pole (bottom). 5.12.3.1.1 Initial grain size and composition
Modified from Schneider DA (1988) An estimate of the long-term
The rapidly chilled margins of submarine lavas produce a large
non-dipole field from marine magnetic anomalies. Geophysical Research
gradient in cooling rate-dependent grain size and composi-
Letters 15: 1105–1108.
tional effects (Marshall and Cox, 1971). For example, detailed
sampling perpendicular to the chilled margin shows that the
transition zone widths (e.g., Cande et al., 1978), and magne- NRM and magnetic hysteresis parameters can vary systemati-
tostratigraphic data confirm that the reversal frequency in the cally by an order of magnitude on centimeter scales (Gee and
earlier Cenozoic was in fact lower (Lowrie and Alvarez, 1981; Kent, 1997, 1998, 1999; Kent and Gee, 1996; Marshall and
Lowrie and Kent, 2004). Similarly, the general absence of Cox, 1971; Figure 11). Near the chilled margin, hysteresis
anomalous skewness for Anomalies M0–M4 (Cande, 1978; parameters indicate grain sizes spanning the superparamag-
Larson, 1977; Larson and Chase, 1972) places limits on the netic/single-domain boundary (30 nm; Ozdemir € and
contribution from the gabbroic layer. These observations sug- O’Reilly, 1981). The presence of stable single-domain grains,
gest that contributions from deeper source layers are not nec- associated with a peak in NRM intensity, presumably accounts
essarily required, although they are also not precluded for much of the strong, stable remanence associated with the
(Harrison, 1987). Magnetic data from samples of these deeper extrusive layer. The pronounced grain-size variation, however,
layers suggest that they do indeed contribute significantly to presents a practical problem in characterizing the magnetic
magnetic anomalies. properties of oceanic basalts because random sampling of
Later in the text, we summarize the magnetization of the such large within-flow variations can seriously bias estimates
oceanic crust according to the main subdivisions – extrusives, of between-flow and between-site variations.
High-temperature
nonmagnetic
region
Lavas
Dikes
Gabbros
Figure 10 3-D perspective cartoon of oceanic crust, showing spreading centers separated by a transform fault. Crust generated during normal/reverse
geomagnetic polarity (shown as shaded/unshaded regions) can be inferred from magnetic anomalies, for example, as measured from ship-towed
magnetometers. Magnetization polarity boundaries are expected to be vertical in sheeted dikes and are likely to slope but in opposite directions
in extrusive basalt layer (due to successive emplacement of lavas) and in intrusive gabbroic layer (due to progressive cooling with depth of gabbros).
Reproduced from Johnson HP, Kent DV, Tivey MA, Gee JS, Larson RL, and Embley RW (1997) Conference on the magnetization of the oceanic
crust steers future research. Eos, Transactions, American Geophysical Union 78: 199–200.
Unblocking temperatures of NRM and Curie temperatures Johnson and Tivey (1995) found a poor correlation (R 0.2)
can also show a large systematic variation with depth in a between FeO* (total iron expressed as FeO) and NRM from the
flow (Grommé et al., 1979; Ryall and Ade-Hall, 1975), even Juan de Fuca Ridge. However, good linear correlations (R up to
in a zero-age pillow such as is illustrated in Figure 11. The 0.8) between NRM and FeO* were obtained from data from
within-flow variation in Curie temperatures in these and other the southern EPR (Gee and Kent, 1997, 1998; Figure 13(a)).
studies has usually been interpreted as reflecting progressive The greatly improved correlations can be attributed to detailed
low-temperature oxidation of an originally homogeneous tita- sampling that spanned the full range of cooling-related mag-
nomagnetite host (e.g., Kent and Gee, 1996). However, trans- netization changes within a flow, as well as to the young age of
mission electron microscopic observations (Zhou et al., 2000) the axial samples, which effectively minimized age-dependent
on a zero-age pillow fragment from the Juan de Fuca Ridge magnetization changes. An inversion of the axial magnetic
show no oxidation and that the titanium content of titano- anomaly profile shows a close correspondence between the
magnetite grains varies as a function of distance from the magnitude of the magnetization solution and the range of
chilled margin (Figure 12). The larger grains in the interior intensities observed in the axial samples (Figure 13(b)). The
have a more uniform composition of around x ¼ 0.6, whereas equivalent NRM values calculated from the mean FeO* con-
the smaller grains toward the glassy margin have more variable tent of each dredge using the linear regression also provide
compositions with a lower average x value of 0.45. Ultrafine confirmation of the predicted link between geochemistry and
magnetite (x 0) has been documented both in interstitial anomaly amplitude.
glass and in the chilled glassy margin (Pick and Tauxe, 1994; The geochemical dependence demonstrated on the south-
Zhou et al., 1999b), and these grains may remain unaltered for ern EPR indicates that the magnetization of oceanic basalts can
tens of millions of years (Zhou et al., 1999b). Kent and Gee vary by up to a factor of 4 as a function of their iron contents
(1996) also documented trace amounts of low-Ti magnetite in alone, from around 12 A m1 for 9% FeO* to more than
the crystalline interiors of very young flows. 50 A m1 for the highest FeO* values ( 15%; the separation
A more general geochemical factor, iron and titanium of early crystallizing phases causes FeO* to increase, but at
enrichment of basaltic melts, is expected to exert a fundamen- higher degrees of fractionation, FeO* decreases as FeTi oxide
tal control on the magnetization of oceanic basalts and has phenocrysts begin crystallizing). Geochemical variation is thus
been conveniently cast in terms of the magnetic telechemistry expected to exert a fundamental control on the magnetization
hypothesis (Vogt and Johnson, 1973). The iron content of of basalt and the source of magnetic anomalies, which should
basaltic melts increases during low-pressure fractionation vary more or less proportionately in amplitude as has been
(e.g., Juster et al., 1989) and this iron enrichment should be observed on the southern EPR. Ridge crest discontinuities,
accompanied by increased abundance of titanomagnetite, where enhanced fractionation is expected, are often accompa-
resulting in higher NRM and thus in enhanced magnetic nied by higher-amplitude magnetic anomalies (Bazin et al.,
anomaly amplitudes. Several attempts to test the magnetic 2001; Wilson and Hey, 1995). In addition, the average degree
telechemistry hypothesis produced uncertain results, largely of fractionation of lavas appears to vary with spreading rate
because of large scatter in the magnetization data. For example, (Sinton and Detrick, 1992). Slow-spreading ridges, which lack
a b c d e f g h i j k Depth from 0.3–0.7 cm, N = 59
NRM (10–3 Am2 kg–1), c (10–6 m3 kg–1) 25 1000 chilled margin

(cm)
Mrs and Ms (10–3Am2 kg–1)

Ms 1
20 800
0.8–1.2 cm, N = 32
15 NRM 600
1.3–1.7 cm, N = 33
10 Mrs 400
5 c 200
2 1.8–2.2 cm, N = 49
0 0
0 2 4 6 8 10
(a)
2.3–2.7 cm, N = 34
250
3 3.0–3.5 cm, N = 31
Temperature (°C)
200
Tmd
150 4.0–4.5 cm, N = 31

TC
a b c d e f g h i j k 4
100
0 2 4 6 8 10
5.0–7.0 cm, N = 84
(b) Depth from glassy margin (cm)
Figure 11 Within-lava flow variability in magnetic properties that
can be ascribed to cooling rate-dependent grain-size effects. (a) Variation
in NRM, susceptibility (w), saturation remanence (Mrs), and saturation
magnetization (Ms) as a function of distance from glassy chilled margin
in New Flow (erupted 1993) pillow fragment R-1 from the Juan de 6
Fuca Ridge. (b) Variation in Curie temperature (Tc) and median 0.2 0.4 0.6 0.8
demagnetizing temperature of NRM (Tmd) in the same samples. Similar x
cooling rate-dependent variations in magnetic properties in oceanic
Figure 12 Variation of composition parameter, x, of titanomagnetite as
basalts have been documented by Marshall and Cox (1971) and Gee and
a function of depth from the glassy chilled margin in a New Flow
Kent (1997) and others. Modified from Kent DV and Gee J (1996)
(erupted 1993) pillow fragment from the Juan de Fuca Ridge. Modified
Magnetic alteration of zero-age oceanic basalt. Geology 24: 703–706.
from Zhou W, Van der Voo R, Peacor DR, and Zhang Y (2000) Variable
Ti-content and grain size of titanomagnetite as a function of cooling rate in
steady magma chambers, have systematically less-evolved very young MORB. Earth and Planetary Science Letters 179: 9–20.
magmas with lower FeO* than faster-spreading ridges, differ-
ences that might be reflected in the overall values of basalt
remanence that is higher, by a factor of 4, than that of older
magnetizations and anomaly amplitudes.
lava samples. However, magnetic anomaly data and direct
determination of the degree of low-temperature oxidation sug-
5.12.3.1.2 Low-temperature alteration gest that this process is responsible for only about half of this
Low-temperature oxidation of stoichiometric titanium-rich decrease in magnetization, with the remaining discrepancy
titanomagnetite is widely considered to be the dominant pro- plausibly attributed to higher paleofield intensity in the axial
cess of magnetic alteration of oceanic basalts. The magnetic lavas. Moreover, the heterogeneity of alteration makes defini-
consequences of such low-temperature alteration include a tive resolution of the relevant timescales difficult.
fourfold reduction in the saturation magnetization (for com- Sea-surface magnetic profiles and inversion solutions
plete oxidation; O’Reilly, 1984) and a significant increase in from fast-spreading ridges like the Pacific–Antarctic Ridge
the Curie temperature (Xu et al., 1996), with the possible (Figure 14(a)) record a short (10 km) wavelength axial
acquisition of a chemical remanent magnetization (CRM) Central Anomaly magnetic high (CAMH) (Klitgord, 1976).
that might replace the initial TRM (Raymond and LaBrecque, Near-bottom anomaly data from fast-spreading ridges indicate
1987). Substantial (up to an order of magnitude) changes in that the width of this high magnetization zone is 2–3 km
the magnetization of the extrusive layer and the amplitude of (Figure 3). At slow-spreading ridges, the CAMH occupies a
the associated magnetic anomalies have been attributed to larger proportion of the Central Anomaly and consequently
low-temperature alteration (Bleil and Petersen, 1983; Irving, may not be readily distinguishable in sea-surface profiles
1970). As documented later in the text, zero-age lavas do have a from the Central Anomaly (Figure 14(b)). In concert with a
80
70
NRM = – 45.2 + 6.60 × FeO*
R = 0.81
60
Dredge NRM (A m–1)

n = 34
50
40
30
20
10
0
8 9 10 11 12 13 14 15 16
(a) Dredge FeO*
pasc2
Anomaly (nT)
200 aria2
0
– 200
(b)
60 16
Magnetization (A m–1)
50
14
40
FeO*
12
30
20 10
10
8
(c) 14 16 18 20 22
Latitude (°S)
Figure 13 Magnetic and geochemical variations along the axis of the southern East Pacific Rise. (a) Variation in NRM as a function of FeO* (total Fe as
FeO) for dredges from the southern East Pacific Rise. NRM values represent the arithmetic mean intensity (SD) of three or more flows/pillows,
each with multiple subspecimens (Gee and Kent, 1997). Dredge mean FeO* content (SD) calculated from microprobe analysis of glass chips (Sinton
et al., 1991) from samples used for magnetic study. (b) Composite axial magnetic anomaly (prior to reduction to pole). (c) Magnetization inferred
from profile inversion (heavy gray line) and grid inversion (dashed, both left scale) is compared to FeO* content (filled circles, with larger circles
indicating dredges for which sample magnetizations are available), with the scaling and offset between the axes determined by the regression result from
panel (a). Prior to inversion, the axial anomaly was reduced to pole using an effective source thickness of 375 m, corresponding to an extrusive
layer that thickens from 250 to 500 m within approximately 2.5 km of the ridge crest. A cosine-tapered band-pass filter was applied where wavelengths
<10 and >600 km were cut and wavelengths >20 and <300 km were passed unattenuated. Modified from Gee J and Kent DV (1998) Magnetic
telechemistry and magmatic segmentation on the southern East Pacific Rise. Earth and Planetary Science Letters 164: 379–385.
dominant Central Anomaly (e.g., Vine, 1966), the evidence for timescale to balance a presumed negative magnetic anomaly
a decay in basalt magnetizations has traditionally come from that would result from the thickening of layer 2A deduced from
the slow-spreading ridges in the Atlantic Ocean where oceanic seismic imaging (Figure 15).
basalts have been thought to suffer decay related to aging and Although originally interpreted as reflecting low-
alteration with a time constant on the order of 0.5 My (Irving, temperature alteration on rapid timescales (Gee and Kent,
1970; Johnson and Atwater, 1977). Alteration-induced magne- 1994), remanence data from near-ridge basalts at the EPR at
tization decay has also often been linked to the CAMH 12° N do not require a substantial decay in magnetization
(Klitgord, 1976), although this would require very different related to alteration. This is because low NRM intensities and
time constants for fast- and slow-spreading ridges. For exam- magnetic susceptibilities and high Curie temperatures, which
ple, profiles from the fast-spreading EPR at 12° N suggest that might be an indication of alteration, occur at the ridge axis
the magnetization contrast must occur over a much shorter (zero age) and off axis (Figure 16). Indeed, efforts to induce
Pacific–Antarctic Ridge 49° S 100

(full rate 100 km My–1)
500 Observed
nT
Anomaly (nT)
50
–500 Extrusives
J J
20 Dikes
0
Mag
Depth (km)
A m–1 Gabbros
2.5 –20
3.0 – 50
(a) 2500
Extrusive rocks
50 0 50 20 A m–1
(a) Distance (km)
3500
Depth (m)
Dikes
1.7 A m–1
Gorda Rise 41° N
4500 AMC
(full rate 24 km My–1) Gabbros 600
250 1.8 A m–1 °C
isot
herm
nT
5500
–250
–8 –4 0 4 8
2 2
(b) Distance from axis (km)
10
Mag
A m–1
–10
Depth (km)
2.0 3.5
Two way travel time (s)
Layer 2A
3.0
4.0
AMC
20 0 20
(b)
Distance (km)
4.5
Figure 14 Sea-surface magnetic profiles, inversion solutions, and (c)
bathymetric profiles from (a) the fast-spreading Pacific–Antarctic Ridge
and (b) the slow-spreading Gorda Rise illustrating the short (10 km) Figure 15 (a) Forward models illustrating the axial magnetic anomaly
wavelength axial Central Anomaly magnetic high (CAMH, vertical minimum generated by off-axis thickening of a uniformly magnetized
shading). The CAMH occupies a larger proportion of the Central Anomaly extrusive layer or cooling of deeper layers as shown in (b), compared to
at slow-spreading ridges where it is not readily distinguishable in the observed sea-surface profile at 19.5° S on the East Pacific Rise.
sea-surface profiles from the Central Anomaly because the wavelengths (b) Source models based on seismic data from near 19.5° S. (c) Wide
become similar. Modified from Klitgord KD (1976) Sea-floor aperture seismic profile from East Pacific Rise near 14° S (Detrick et al.,
spreading: The central anomaly magnetization high. Earth and Planetary 1993) that illustrates rapid thickening of layer 2A and the presence of
Science Letters 29: 201–209. axial magma chamber (AMC). Modified from Gee J and Kent DV (1994)
Variations in layer 2A thickness and the origin of the central anomaly
magnetic high. Geophysical Research Letters 21: 297–300.
alteration of young oceanic basalts in the laboratory have
generally failed (e.g., Kent and Gee, 1994). Instead, it now
appears that the high magnetization values on-axis producing processing (filtering and gridding) and the sequence effect in
the contrast that accounts for the CAMH on fast-spreading biasing the results, for example, in producing the apparent
ridges are mostly due to a paleointensity signal, which is also substantial (factor of 3–5) increase in magnetization with
suggested by the central notch in the CAMH in near-bottom increasing age beyond 10 Ma and into the KQZ. An influential
profiles over faster-spreading rate ridges (Gee et al., 2000; and more direct analysis in this regard was the compilation of
Perram et al., 1990; Figure 3). oceanic basalt magnetizations from Deep Sea Drilling Project
Evidence has also been sought for longer-term changes in (DSDP) sites by Bleil and Petersen (1983), who proposed that
oceanic basalt magnetic properties that could be attributed to the data showed an initial sharp decrease in NRM to minimum
the effects of alteration to explain an apparent envelope of values at around 20 Ma followed by a gradual increase to ages
decreasing anomaly amplitudes with distance from the ridge of around 120 Ma. Subsequent compilations (Furuta, 1993;
axis. Several magnetic anomaly inversion studies have sug- Johnson and Pariso, 1993; Figure 17) show a similar pattern
gested a decrease in magnetization by a factor of 2–3 within of variation with age although the data quality is highly vari-
about 10 My of the ridge axis (Sayanagi and Tamaki, 1992; able with more than one-third of the sites represented by 10 or
Wittpenn et al., 1989). However, it is difficult to evaluate the fewer samples (e.g., six sites have only one sample!). The
effects of various potential artifacts associated with data overall arithmetic mean NRM intensity for all 64 site means
80 EPR at 12° N characterized DSDP/ODP sites that have the deepest penetra-
tion and are represented by about 100 or more samples
70 (Figure 18). The NRM distributions are reasonably approxi-
60 mated by lognormal distributions. The zero-age southern EPR
samples have a geometric mean NRM of 18.9 A m1. This is
NRM (A m–1)
50 0.5 my
virtually identical to the geometric mean NRM intensity of
time constant
40 18 A m1 for the axial samples from 12° N on the EPR, which
were mostly taken within about 3 km of the axis and thus are
30
less than 50 ky old (Gee and Kent, 1994). Published data
20 from the youngest drill site with significant penetration (DSDP
0.02 my Site 482, which was sited on 0.4 Ma crust; Lewis et al., 1983)
10
time constant indicate NRM intensities of about 5 A m1, almost a factor of 4
0 less than the axial samples. Still older drill sites have geometric
(a)
0 2 4 6 8 10 12 mean NRM intensities that range from 2.3 A m1 for Site 395
Distance from axis (km)
(7.8 Ma) to 14.1 A m1 for Site 1256 (15.4 Ma) although the
data at this latter site may be biased to higher NRM values by a
16
particularly strong drilling remanence (Wilson et al., 2003).
100
Susceptibility (10–3 SI)
Nevertheless, there does not appear to be an obvious age-
Curie temperature (°C)

0.5 my dependent trend in the drill site-mean NRM values since the
12
time constant
150 oldest sites (Site 417/418 at 121 Ma and Hole 801C at 169 Ma)
have about the same average NRM intensities as some of the
8 200 youngest drill sites (e.g., Sites 482 and 483 at 2.3 Ma and Hole
504B at 6.8 Ma), which are all between about 4 and 6 A m1.
4 0.02 my 250 What does seem significant is the factor of almost 4 decrease in
time constant mean NRM intensity from the axial zone to older sites; if the
300 entire reduction is ascribed to alteration, the time constant
0 must be considerably shorter than about 0.4 My, the age of
0 2 4 6 8 10 12 the youngest off-axis site (Site 482) with reduced magnetiza-
Distance from axis (km)
tion. A similar pattern of magnetization change was observed
0 40 80 120 160 200 in the FAMOUS area (37° N on the slow-spreading Mid-
(b) Nominal age (ka) Atlantic Ridge) based on the magnetization of topography
derived from near-bottom observations (Macdonald, 1977).
Figure 16 (a) Natural remanent magnetization (NRM) and (b) magnetic
susceptibility for samples of dredged oceanic basalts from 12° N on the Direct observations of the degree of low-temperature alter-
East Pacific Rise as a function of distance from the ridge axis (or nominal ation suggest that the fourfold difference in the intensity of
age based on full-spreading rate of 98 km My1). Each data point (SD) is axial lavas and older drill sites cannot be attributed solely to
the average of 3–6 specimens collected within 5–8 cm of the glassy margin. maghemitization. The oxidation parameter (0 < z < 1, where 1
Exponential magnetization decay curves with time constants of 0.5 and indicates complete oxidation to titanomaghemite; O’Reilly,
0.02 My are shown for reference, but the large variability in the 1984) has now been estimated from lattice spacing and com-
magnetization data and in the previously unpublished Curie temperature position using transmission and analytic electron microscopy
data shown in (b) for these samples does not make a compelling case for on small titanomagnetite grains in fine-grained oceanic basalts
any simple decay scheme. Modified from Gee J and Kent DV (1994) over a broad age range (Wang et al., 2005, 2006; Zhou et al.,
Variations in layer 2A thickness and the origin of the central anomaly
2001). The development and application of microanalytical
magnetic high. Geophysical Research Letters 21: 297–300.
techniques (Zhou et al., 1999a) represent a major advance in
the study of oceanic basalts because they allow more direct
reduced to the equator is 4.4 A m1; high values of around characterization of the important fine-grained magnetic car-
10 A m1 occur both at very young ages and at around 120 Ma, riers that previously had to be inferred from larger magnetic
whereas values of 1 A m1 or less occur at around 40 Ma. There grains or bulk properties. Although there is considerable scatter
is considerable scatter throughout with only a handful of sites (Figure 19), the oxidation parameter data are consistent with
older than 40 Ma penetrating more than 50 m of basalt or an initial, very rapid increase in alteration such that z values up
represented by more than a nominal number (20–30) of sam- to 0.8 occur in basalts less than 1 My old, as suggested by Bleil
ples. Because of the typically small number of samples, as well and Petersen (1983). Much less convincing is any systematic
as the lack of control on geochemical variations, within-flow long-term trend because although values of z greater than 0.8
grain-size variations, or geomagnetic intensity fluctuations that do tend to occur in samples older than about 30 Ma, the oldest
strongly influence remanent intensity, suggestions of any tem- sample measured (Hole 801C, 169 Ma) has a z value of only
poral trends in magnetization intensity of seafloor lavas from 0.2. Thus, while low-temperature alteration evidently can
such plots should be regarded as tentative. occur rapidly, the extrusive layer is not uniformly oxidized
A more robust estimate of age-dependent changes can be even for the oldest seafloor.
obtained by comparing the distribution of NRM values for a Since complete oxidation would be required to account for
large collection of near-axis (zero-age) samples from the south- a factor of 4 reduction in saturation magnetization (O’Reilly,
ern EPR with histograms of NRM values for the best 1984), some additional parameter is apparently required to
Number
0 5 10 15
Log equatorial NRM (A m–1)

10 10
1
Equatorial NRM (A m–1)
1 1
0
> 50 m penetration
< 50 m penetration
0.1
0 40 80 120 160
1000
Samples
100
10
1
0 40 80 120 160
Age (Ma)
Figure 17 Compilation of equatorial NRM of oceanic basalts recovered from DSDP and ODP sites as a function of crustal age and sorted by depth of
basement penetration. Lower plot shows the number of sample measurements available for the calculation of site mean (from references listed in
Johnson and Pariso, 1993); histogram at the right is for all site-mean NRM values. NRM data are from the compilation of Johnson and Pariso (1993)
with some corrections (e.g., deletion of data for samples from sills). Ages are based on updated anomaly or biostratigraphic assignments and the
timescales of Berggren et al. (1995) and Channell et al. (1995).
explain the factor of 4 difference in magnetization of axial lavas trend in magnetic grain size with age from DSDP/ODP sites
and older drilled basalts. Paleofield changes might explain the (Gee and Kent, 1999).
difference if the present field is unusually high and it is worth Directional changes associated with low-temperature alter-
noting that magnetization variations over long timescales have ation might also contribute to reduction in seafloor basalt
been attributed by some authors to paleofield variations (e.g., magnetization with time if a CRM is acquired in a direction
Juarez et al., 1998; Wang et al., 2005). Even the traditional that opposes the initial thermoremanence. Several early studies
model of alteration-related variations in saturation magnetiza- (e.g., Marshall and Cox, 1972; Ryall and Ade-Hall, 1975) con-
tion (Ms) is still possible; for example, Matzka et al. (2003) cluded that the remanence direction was unaffected by low-
argued that the low NRM values for 10–40 Ma samples are temperature alteration since the oxidized titanomaghemite
primarily due to the reduction in saturation magnetization would inherit the original TRM direction. These results are
accompanying maghemitization, similar to the model of Bleil not conclusive, however, since these studies were conducted
and Petersen (1983). Some other alternative explanations are on young Brunhes-age samples where no large directional
described in the succeeding text. difference would be expected. More extensive studies on
Xu et al. (1997c) had suggested a model of preferential altered lavas from a range of ages (Beske-Diehl, 1990) were
dissolution with time of the finest-grained titanomagnetites also interpreted as reflecting no alteration-related directional
as the major process contributing to long-term temporal changes since the observed directional shifts could also be
changes in remanent intensity of mid-ocean ridge basalts. If attributed to viscous remanence acquisition. Although the
widespread, such dissolution could enhance the effect of the low Curie point of TM60 makes direct experimental investiga-
reduction in saturation magnetization accompanying low- tion of seafloor oxidation difficult, Kelso et al. (1991) found
temperature oxidation. However, the data in Figure 19 show that CRM in synthetic TM40 grains was acquired parallel to the
no obvious coarsening in magnetic grain size with age or applied field direction, with secondary magnetizations partic-
alteration, for example, some of the highest Mrs/Ms values ularly enhanced at lower pH values. If broadly applicable, such
(i.e., high ratios of saturation remanence to saturation magne- field-parallel CRM acquisition coupled with the pattern of
tization indicate the finest magnetic grain sizes) are associated geomagnetic polarity reversals could provide a mechanism
with the highest z values (greatest low-temperature oxidation) for long-term variations in NRM intensity as suggested by
in 30 My old samples. Instead, the results in Figure 19 are Raymond and Labrecque (1987).
consistent with the more voluminous hysteresis data shown in In addition, low-temperature oxidation has been suggested
Figure 20 that illustrate that substantial intrasample variability as a possible mechanism by which partial self-reversal might
in oceanic basalts precludes the recognition of any systematic occur, leading to the acquisition of an antipodal overprint that
Equatorial NRM (A m–1) 1

0.01 0.1 1 10 100
Mrs/Ms
0.8 >0.5
Oxidation parameter
300 0.4–0.5
SEPR (~0 Ma)
GM = 18.9 (0.38) 0.6 0.3–0.4
200 n = 2178 0.2–0.3
0.1–0.2
100 0.4 <0.1
0 5 ‘zero’ age
0.2
samples with
482 (~0.4 Ma) z=0
GM = 4.98 (0.30)
20 0
n = 133
0 (a) 0.01 0.1 1 10 100
483 (~2.3 Ma)

1
GM = 4.99 (0.40)
20
n = 89
0
0.8
Oxidation parameter
60 504B/896 (~6.8 Ma)
40 GM = 4.23 (0.42) 0.6
20 n = 430
Number of specimens
395 (~7.8 Ma)

0.4
GM = 2.23 (0.36)
20 n = 90
0 0.2
1256 (~15.4 Ma)

40 GM = 14.1 (0.35) 0
20 n = 184 0 50 100 150 200
0 (b)
Age (Ma)
100 Figure 19 Variation in oxidation parameter (and hysteresis ratio,
80 417/418 (~121 Ma)
GM = 6.15 (0.44) Mrs/Ms) of oceanic basalts samples as a function of crustal age (log scale
60
40 n = 878 in (a) and linear scale in (b)). Data are from Wang et al. (2005), Wang
20 et al. (2006), and Zhou et al. (2001). The plot shows no obvious evidence
0 of any systematic evolution of magnetic properties that could be
–2.0 –1.0 0 1.0 2.0 ascribed to low-temperature alteration, for example, the basalt samples
801C (~168 Ma) from Site 801C, which are the oldest at 169 Ma, have an oxidation
40 GM = 3.43 (0.45) parameter value of only 0.2.
20 n = 203
0
–2.0 –1.0 0 1.0 2.0
Log equatorial NRM (A m–1) although they argued that the ‘low-temperature’ oxidation
occurred at temperatures above the blocking temperatures of
Figure 18 Histograms of equatorial NRM for oceanic basalts for sites
the original titanomagnetite phase during initial cooling. Thus,
with largest datasets, including from the axial zone of the southern EPR
while such partial self-reversals may occur, the importance of
versus off-axis drill sites – Sites 482, 483, 504B/896, 395, 1256,
417/418, and 801C (see Table 2 for data sources). Geometric mean this process in generating long-term changes in seafloor mag-
(GM) is indicated by red arrow for each distribution. The green arrow for netization remains to be demonstrated.
Hole 1256D indicates average NRM for samples with initial inclinations A common motivation for seeking evidence of a time-
<30° and >30° that should be least affected by drilling remanence. dependent change in the magnetization of oceanic basalts is
the widespread perception of a decreasing amplitude envelope
for ridge crest anomalies. This concept stems largely from long-
could significantly reduce the NRM intensity (Doubrovine and standing observations that the Central Anomaly in the slow-
Tarduno, 2004). These authors found that the majority of spreading Atlantic and Indian Oceans tends to be anomalously
thermally demagnetized samples from Detroit Seamount had high (e.g., Vine, 1966). For example, Bleil and Petersen (1983)
two nearly antipodal remanence components. Some samples showed a magnetic profile across the North Atlantic as evi-
showed a maximum in the Ms(T) curves above room temper- dence of this phenomenon, which had also been described to
ature that might be indicative of N-type ferromagnetic behav- occur over some sectors of the Australian–Antarctic Ridge by
ior and partial self-reversal (Doubrovine and Tarduno, 2004). Weissel and Hayes (1972). However, the Central Anomaly is
A similar peak in Ms(T) and importantly a corresponding not such a pronounced feature in magnetic profiles from the
reversible peak in NRM above room temperature were reported faster-spreading sectors of the East Pacific, which are also typ-
for 10–40 Ma oceanic basalts (Matzka et al., 2003). Krasa et al. ically described as showing only a gradual decrease in ampli-
(2005) provided compelling documentation for partial self- tude on the ridge flanks (e.g., Pitman and Heirtzler, 1966;
reversal associated with titanomaghemite in subaerial lavas, Pitman et al., 1968). Indeed, the prominence of the Central
0.8 SEPR slabs Other DSDP/ODP

Phoenix/DSDP slabs Recent DSDP/ODP
0.7
0.6
0.5
Mrs /Ms
0.4
0.3
0.2
0.1
0
0.01 0.1 1 10 100
Age (Ma)
Figure 20 Variation of Mrs/Ms with age in oceanic basalt samples. Data acquired from systematic sampling relative to the chilled flow margin
(labeled slabs) include samples from the southern East Pacific Rise (SEPR), Phoenix dredge collection, and DSDP sites (see Gee and Kent, 1999, and
references therein). Other DSDP/ODP designates additional hysteresis data from isolated samples used in this same compilation. More recent
hysteresis data (recent DSDP/ODP) include data from Wang et al. (2005), Wang et al. (2006), and Zhou et al. (2001).
Anomaly (and the associated amplitude envelope) appears to The independent evidence for rapid changes in oxidation
be closely tied to spreading rate (Figure 21; Cande and parameter (Figure 19) favors a fast decay time constant,
Gee, 2001). although the entire observed decrease in NRM intensity may
The fact that neither a prominent Central Anomaly nor a not be entirely due to alteration. For example, the NRM of the
pronounced anomaly envelope is present on all ridge systems near-axis samples may be biased by high paleointensities over
places limits on systematic time-dependent processes that are the past few thousand years that were up to a factor of 2 higher
expected to affect the magnetization of all oceanic basalts, such than today’s field and perhaps a factor of 4 higher than the
as long-term changes in paleointensity and alteration. Sug- average long-term paleofield intensity (Selkin and Tauxe,
gested time constants of magnetic alteration, assuming an 2000). Absolute paleointensity data, which show a 3 change
exponential decay, have ranged over several orders of magni- near the ridge axis (Bowles et al., 2006; Gee et al., 2000), are a
tude, from 0.05 My (Gee and Kent, 1994), to 0.5 My powerful indication that a large part of that near-ridge signal is
( Johnson and Atwater, 1977; Macdonald, 1977), to 5 My field-related. A factor of 2 decrease in NRM intensity due to
or longer (Bleil and Petersen, 1983; Raymond and LaBrecque, alteration might be a reasonable estimate for oceanic basalts,
1987; Xu et al., 1997b; Zhou et al., 2001). We use these order with a possibly larger contrast in the slow-spreading ridges due
of magnitude increments in the estimated time constants to to more intense near-axis tectonics, for example.
model the expected effects on magnetic anomalies in Figure 22.
For fast-spreading ridges (Figure 22(a)), models using a factor
5.12.3.2 Magnetization of Dikes
of 4 decay in magnetization, corresponding to the nominal
decrease in NRM from the ridge axis to off-axis drill sites One of the most significant changes in our understanding of
(Figure 18), are virtually precluded over the various timescales. the magnetization of oceanic crust since the review by Smith
These models produce either a very prominent axial high at a (1990) has been in the role of dikes, which some earlier
short (0.05 My) decay constant or an obvious amplitude enve- compilations based on ophiolites and dredge metabasalt sam-
lope at a long (5 My) decay constant, but neither of these ples from fracture zones suggested were characterized by weak,
features is compatible with the representative Pacific–Antarctic rather unstable, remanences (e.g., Kent et al., 1978). The pen-
Ridge profile shown in Figure 21. This comparison suggests etration, albeit with limited recovery, of 1000 m section of
that the effective magnetization reduction from alteration is dike rocks at Site 504B (Figure 23) has allowed the documen-
likely much smaller than a factor of 4, perhaps reflecting the tation of the physical and magnetic properties of this impor-
heterogeneous nature of low-temperature oxidation in the tant constituent of oceanic crust in a more typical setting. An
crust. Indeed, models using a 2 factor of decay in magne- 100 m thick transition between pillow lavas and dikes (at
tization at a variety of decay constants for fast-spreading 1000 mbsf ) does have weak magnetizations, 0.1 A m1
ridges (Figure 22(a)) cannot be excluded as resembling the and less, although they have very high coercivities. Deeper
observed Pacific–Antarctic Ridge profile. Models for different levels in the sheeted dike complex have moderate NRM
decay factors and time constants for a slow-spreading ridge ( 2 A m1) and moderate median demagnetizing fields.
(Figure 22(b)), on the other hand, seem to favor a larger Moreover, analysis of logging data suggests that, due to void
decay (e.g., compare 4 models to observed Mid-Atlantic or spaces, the effective NRM of the lavas from Hole 504B may be
Southwest Indian Ridge profiles in Figure 21). only 3 A m1 rather than the 5 A m1 based on sample
crust exposed at Hess Deep provides a direct comparison of

Reykjanes Ridge
(18 kmMy–1) magnetic properties of extrusives, dikes, and gabbros
(Figure 24). Here, lavas and dikes have comparable NRM values
2000
nT (4 A m1). Dredged dike samples from 2 My crust exposed
along the Blanco Fracture Zone also have moderately strong
magnetizations ( 1.7 A m1 when reduced to equatorial
–100 100
Southwest Indian Ridge value; Johnson and Salem, 1994), although this is substantially
(18 kmMy–1) less than the magnetization of extrusives ( 6.4 A m1) inferred
from near-bottom magnetic anomaly surveys in this area (Tivey,
400
nT 1996; Tivey et al., 1998a). Although samples of the dike com-
plex are still limited, it appears that they may have a remanence
that is significant relative to that of the extrusive layer.
Central Atlantic A major difference from the pillows, however, is the mag-
(24 kmMy–1) netic mineralogy, which in pillows is variably maghemitized
titanomagnetite (Curie temperatures typically 200–400 °C),
400 whereas in the dikes, it is dominated by nearly pure magnetite
nT
(Curie temperatures >550 °C; Figures 23 and 24). The origin
of the magnetite is critical to assessing the role of dikes as a
Carlsberg Ridge source of seafloor-spreading magnetic anomalies. If magnetite
(28 kmMy–1) is primarily produced by alteration at temperatures below the
Curie point, as some have suggested, then the remanence may
400
nT be largely chemical in origin and may be acquired significantly
later than that of the extrusive layer (e.g., Hall and Muzzatti,
1999). Based on a transmission electron microscopy study of
South Atlantic dike samples from Hole 504B, Shau et al. (2000) documented
(36 kmMy–1)
multiple pathways for the formation of magnetite (Figure 25).
400 Initial oxyexsolution, at temperatures >600 °C, would have
nT produced only slightly Ti-depleted titanomagnetite, with a
Curie temperature too low to acquire a remanence at these
elevated temperatures. Further oxidation or true exsolution
Southeast Indian Ridge
and reduction, likely at temperatures of 500–400 °C, then
(68 kmMy–1)
produces the fine Ti-poor magnetite that acquires a thermal
800 and/or chemical remanence (Shau et al., 2000). While the
nT
remanence in dikes maybe, at least in part, chemical in origin,
the temperatures at which the alteration occurs imply that
Chile Rise remanence would be acquired very near the ridge axis. This
(80 kmMy–1) result is confirmed by the consistent (reverse) polarity in the
lavas and dikes inferred from magnetic logging data at Hole
400
nT
504B (Worm and Bach, 1996).
5.12.3.3 Magnetization of Gabbros

Pacific-Antarctic Ridge
(82 km My–1) Early estimates of the gabbroic contribution to marine mag-
netic anomalies relied on dredged samples and studies of
600 ophiolites, with the results from ophiolites yielding variable
nT and sometimes contradictory results (Banerjee, 1980). For
example, Swift and Johnson (1984) concluded from the low
–400 –200 0 200 400
remanent intensities and K€ onigsberger ratios (the ratio of rem-
Distance from Ridge Axis (km)
anent to induced magnetization) of unaltered gabbros from
Figure 21 Representative sea-surface anomaly profiles illustrating the the Bay of Islands ophiolite that only the altered intrusive
spreading rate dependence of the amplitude of the Central Anomaly. portion of the crust (characterized by higher magnetization
All profiles extend approximately to Anomaly 5 and have been reduced to values) would contribute to magnetic anomalies. In contrast,
the pole. The Central Anomaly has notably higher amplitude only at results from dredges and other ophiolites (Kent et al., 1978;
full-spreading rates less than 30 km My1.
Vine and Moores, 1972) indicated that gabbros had significant
magnetization (1 A m1) and high K€ onigsberger ratios and
measurements (Worm et al., 1996), which would further likely would contribute to magnetic anomalies.
reduce the magnetization contrast with the dike complex. At the time of the last review of oceanic crustal magnetiza-
Dike samples obtained by both dredges and submersible tion (Smith, 1990), relatively little direct information was
sampling also highlight the importance of dikes to crustal mag- available on the magnetic properties of oceanic gabbros. Initial
netization. Submersible sampling of 1.2 My old fast-spread results from the first significant drillcore penetration of
5.0 5.0
My My
4X Decay
1000 0.5 0.5 1000
nT My My nT
0.05 0.05
My My
5.0 5.0
My My
2X Decay
1000 0.5 0.5 1000
nT My My nT
0.05 0.05
My My
No No
decay decay
10 5 0 5 10 10 5 0 5 10
Age (Ma) Age (Ma)
–400 –200 0 200 400 –100 –50 0 50 100

Distance (km) Distance (km)
(a) (b)
Figure 22 Magnetic anomaly models illustrating the effect of alteration-related exponential decay in magnetization for (a) intermediate-spreading crust
(full rate of 80 km My1) and (b) slow-spreading crust (full rate of 20 km My1). For each spreading rate, anomaly profiles are shown for a
fourfold decrease in magnetization (20–5 A m1) and a twofold decrease in magnetization (10–5 A m1) for a range of decay time constants. The
magnetization model (block diagram) and model with constant remanence intensity are shown for comparison. In all models, the anomaly is calculated
for a 1 km thick source layer, with the depth to the upper surface that increases with age0.5 (Ma) (Stein and Stein, 1994). A Gaussian filter has
been applied to the magnetization (s ¼ 0.5 km and s ¼ 1.0 km for the intermediate- and slow-spreading ridges, respectively) to simulate the transition
zone width in the extrusive layer.
NRM (A m–1) Q Remanent inclination MDF (mT) Curie temperature (°C)

Depth Hole 504B 0.01 0.1 1 10 0.1 1 10 100 –60° 0° 60° 10 30 50 300 400 500 600
(mbsf)
Leg 69
Pillow lavas
400
Leg 70
800
Trans.
Leg 83
1200
137
Dikes
1600 111
Leg 140
2000
148
Figure 23 Depth variation in magnetic properties (NRM, Q calculated for 40 A m1 field, stable remanent inclination, median demagnetization field
(MDT), and Curie temperature) of samples of Miocene (6.8 Ma) oceanic basement at Site 504B on the Galápagos Spreading Center. Schematic
lithology and penetration during seven drilling legs shown at the left. Data compiled from Facey et al. (1985), Furuta (1983), Furuta and Levi (1983),
Kinoshita et al. (1985), Pariso and Johnson (1989), Pariso et al. (1995), Pechersky et al. (1983a), Smith and Banerjee (1985), and Smith and
Banerjee (1986).
10
Volcanics 20 Volcanics
Samples
3.56 A m–1 (0.40)
Samples
262 (138)
5 n = 61 10
0 0
–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 50 100 150 200 250 300 350 400 450 500 550 600
20
40
15
Dikes
Samples
30 Dikes
3.91 A m–1 (0.35)
Samples
10 499 (160)
n = 128 20
5 10
0 0
–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 50 100 150 200 250 300 350 400 450 500 550 600
10
20
Samples
Samples
Gabbros Gabbros
5 1.08 A m–1 (0.31) 564 (14)
10
n = 47
0 0
–0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 50 100 150 200 250 300 350 400 450 500 550 600
Log NRM (A m–1) Curie temperature (°C)
Figure 24 Histograms of NRM intensities (left) and Curie temperatures (right) for dive samples of volcanics, dikes, and gabbros from 1.2 Ma fast-spread
oceanic crust at Hess Deep (data from Gee et al., 1992; Varga et al., 2004). Note that the mean NRM intensity for dikes is higher than reported at Hole
504B and is more similar to the volcanics, suggesting that the dikes are an important source of magnetic anomalies at least in the Hess Deep area.
Primary titanomagnetite
(TM60-70)
(a)
Oxidation
(1) (2)
Exsolution and reduction (b) Oxidation
Ulvöspinel
Trellis-type lamellae // {100}Tmt Trellis-type lamellae // {111}Tmt
{ Ulvöspinel
Ti-bearing magnetite (c)
Ilmenite
{ Ilmenite
Ti-bearing magnetite
Hydrothermal alteration
Titanite + kassite Titanite + chlorite + Ca-zeolite

Magnetite (TM0) Magnetite (TM0)
(d)
Figure 25 Alteration pathways for primary titanomagnetite in sheeted dikes. Either oxyexsolution or true exsolution and reduction may produce
fine-grained magnetite (x 0) that is distinct from primary titanomagnetite in the extrusives. Reproduced from Shau Y-H, Torii M, Horng C-S,
and Peacor DR (2000) Subsolidus evolution and alteration of titanomagnetite in ocean ridge basalts from Deep Sea Drilling Project/Ocean Drilling
Program Hole 504B, Leg 83: Implications for the timing of magnetization. Journal of Geophysical Research 105: 23635–23649.
gabbros (500 m at Hole 735B; Robinson et al., 1989) indicated (Gillis et al., 1993), and Hole 735B has been deepened by an
highly variable NRM intensities, with some FeTi oxide gabbros additional kilometer (Dick et al., 1999) and supplemented by
having extremely high values. With only a single drillcore, additional drilling on the Atlantis Bank (Allerton and Tivey,
however, it was difficult to evaluate whether these magnetic 2001; Pettigrew et al., 1999). Although these sites all made use
(and petrologic) results were generally applicable. In the inter- of tectonic exposures, recent drilling in the Guatemala Basin
vening time, lower crustal gabbros have been sampled by (Wilson et al., 2006) penetrated lavas, dikes, and a small
drilling at three portions of the slow-spreading Mid-Atlantic amount of gabbro; continued drilling at this site could provide
Ridge (Mid-Atlantic Ridge near the Kane Fracture Zone or the first in situ gabbroic section from the oceanic crust.
MARK area at 23° N (Cannat et al., 1995), near the 15°200 These various drill sites into lower crustal gabbros provide a
fracture zone (Kelemen et al., 2004), and Atlantis Massif remarkably consistent average magnetization of 1–2 A m1
(Blackman et al., 2006)), at fast-spreading crust at Hess Deep (Table 2), suggesting that these rocks should constitute an
Table 2 Equatorial natural remanent magnetization values for oceanic crustal rocks
Site Sample type Age (Ma) n AM (A m1) std. (A m1) GM (A m1) std (log) References
Lavas
EPR Dredge 0 2178 24.68 15.54 18.94 0.38 1
Hess Deep Dive 1.2 61a 5.37 5.70 3.56 0.40 2, 3
Site 482 Drill 0.4 133 6.20 4.19 4.98 0.30 4, 5
Site 483 Drill 2.3 89 7.42 6.49 4.99 0.40 4, 5
Hole 504B/896b Drill 6.8 430 6.15 4.95 4.23 0.42 6–12
Site 395c Drill 7.8 90 2.95 2.01 2.23 0.35 15
Site 1256d Drill 15.4 184 18.36 12.22 14.06 0.35 16
Sites 417/418 Drill 121 878 9.22 8.03 6.15 0.44 17
Hole 801Ce Drill 168 203 4.92 3.89 3.43 0.45 18, 19
Deep Drillholes Drill – 7 7.89 5.01 4.92 0.25
Dikes
Hess Deep Dive 1.2 128a 5.27 4.50 3.91 0.35 2, 3
Hole 504Bf Drill 6.8 171 2.11 1.67 1.32 0.53 9–14
All – – 2 3.69 2.23 2.27 0.33
Gabbros
Hess Deep Dive 1.2 47a 1.40 1.17 1.08 0.31 2, 3
Hole 735B Drill 12 600 3.34 8.42 1.19 0.57 20, 21
Leg 209g Drill 0.5–2.0 61 6.15 6.23 2.36 0.78 22
Leg 153h Drill 1 252 1.19 1.74 0.62 0.53 23, 24
Hole 894Gi Drill 1 87 2.25 1.56 1.65 0.41 25
Exp. 304/305j Drill 1 472 3.11 6.28 0.56 0.97 26
All Drillholes Drill – 5 2.91 1.81 1.09 0.24
Peridotites
Oufi compilationk Drill – 234 5.84 5.05 3.15 0.62 27
Leg 209l Drill 0.5–2.0 99 2.75 3.57 0.79 0.99 22
All Drill – 333 4.92 4.86 2.09 0.79
AM, arithmetic mean; GM, geometric mean.

References: 1, Gee and Kent (1997); 2, Varga et al. (2004); 3, Gee et al. (1992); 4, Day et al. (1983); 5, Pechersky et al. (1983b); 6, Allerton et al. (1996); 7, Furuta and Levi (1983);
8, Pechersky et al. (1983a); 9, Kinoshita et al. (1985); 10, Facey et al. (1985); 11, Smith and Banerjee (1985); 12, Smith and Banerjee (1986); 13, Pariso and Johnson (1989);
14, Pariso et al. (1995); 15, Johnson (1978); 16, JANUS database for ODP Leg 206 (Wilson et al., 2003); 17, Levi (1980); 18, JANUS database for ODP Legs 129 (Lancelot
et al., 1990) and 185 (Plank et al., 2000); 19, Wallick and Steiner (1992); 20, Kikawa and Pariso (1991); 21, Dick et al. (1999); 22, Kelemen et al. (2004); 23, Gee et al. (1997);
24, Cannat et al. (1995); 25, Pariso et al. (1996a); 26, Blackman et al. (2006); 27, Oufi et al. (2002).
a
Each sample value based on three or more specimens.
b
Average based on Hole 504B lava samples above 898 mbsf (corresponding to low-temperature altered zone of Pariso and Johnson, 1991) and all lava samples from Site 896.
c
Excludes gabbro and peridotites recovered at this site.
d
Average based on shipboard data from 252–749 mbsf. Wilson et al. (2003) noted particularly large drilling-induced remanence.
e
Average based on samples from 551–933 mbsf (excluding younger alkalic lavas and hydrothermal deposits).
f
Average based on samples deeper than 1055 mbsf (dike/lava transition zone (878–1055 mbsf) excluded).
g
Average based on all gabbroic samples from Sites 1268–1275 (near 15°200 N fracture zone).
h
Average based on all gabbroic samples from Sites 921–924 (MARK area).
i
Minor recovery of gabbroic material in other drill sites at Hess Deep not included.
j
Includes gabbroic samples from Holes 1309B and 1309D (excluding diabase and ultramafic samples).
k
Compilation of peridotite samples from DSDP/ODP up to and including Leg 153.
l
Average based on all peridotite samples from Sites 1268–1274 (near 15°200 N fracture zone).
important magnetization source for magnetic anomalies. a substantial drilling-induced overprint and thus are not rep-
Although average magnetic properties are similar, gabbroic resentative of the in situ magnetization. Borehole magnetic
sections (particularly at slow-spreading ridges) exhibit substan- data from the upper 500 m of Hole 735B indicate an in situ
tial variability in both lithology and magnetic properties as equatorial remanence of 4.4 A m1 for oxide-rich gabbros and
exemplified by results from Hole 735B (Figure 26). FeTi 1.8 A m1 for other gabbroic rocks (Pariso and Johnson,
oxide gabbros with >2% (and occasionally >10%) oxide min- 1993a). Although NRM intensities show significant variability,
erals constitute a significant fraction of the core from Hole results from both drillcore and submersible samples (Table 2)
735B. Magnetic susceptibility data indicate the presence of suggest that an average intensity of 1 A m1 is typical for
several hundred oxide-rich zones, ranging from centimeter to oceanic gabbroic rocks.
several meters in thickness, throughout the core (Natland, Although less compelling than for lavas, there appears to be
2002). These oxide-rich intervals may be associated with a weak correlation between more fractionated (Fe-rich) gab-
extremely high NRM values (in some cases >103 A m1; bros and higher NRM intensities (Gee et al., 1997; Kikawa and
Robinson et al., 1989) though these high values likely reflect Ozawa, 1992). For a subset of gabbroic rocks from the MARK
Lithologic variability Magnetization (A m–1) Q Stable inclination MDF (mT)

0 20 40 60 80 100 0.1 1 10 100 1 10 0 30 60 90 20 60 100
0
500
Depth (mbsf)
1000
1500
Troctolitic gabbro Gabbronorite
Olivine gabbro Disseminated FeTi oxide gabbro
Gabbro FeTi oxide gabbronorite
FeTi oxide gabbro
Figure 26 Depth variation in lithology and magnetic properties (NRM, Q calculated for 30 A m1 field, stable remanent inclination and MDF, median
demagnetization field) for samples of gabbroic section in oceanic crust at ODP Hole 735B. Magnetic data compiled from Kikawa and Pariso (1991) and Dick
et al. (1999); lithology column from Dick et al. (2000). Heavy line in inclination plot is expected dipole value. Magnetization data have not been reduced
to equatorial values (open circles and + symbols represent the intensity after removal of the drilling-induced remanence and NRM, respectively).
area, Hess Deep, and Hole 735B (Figure 27(a)), the lowest fine-grained magnetite enclosed in silicate grains. Fine-grained
NRM values are exclusively associated with the least evolved oxides are present in olivine, pyroxene, and plagioclase in
gabbros (with high Mg# ¼ Mg2+/(Fe2++Mg2+)). Kikawa and nearly all troctolites and olivine gabbros from Hole 735B
Ozawa (1992) had noted a comparable broad increase in (Natland, 2002). Magnetite along cracks in olivine is attributed
remanent intensity from Hole 735B samples as fractionation to amphibolite–granulite facies alteration at temperatures of
proceeds from troctolite to olivine gabbro. These same authors 500–700 °C (Pariso and Johnson, 1993b). Crystallographi-
have suggested that gabbro magnetization values also may be cally oriented elongate magnetite within pyroxene is demon-
reduced by up to an order of magnitude as a result of alteration strably a high temperature ( 800 °C) ‘exsolution’ product in
(as measured by the percentage of secondary mafic minerals). some cases (Renne et al., 2002). A similar exsolution origin for
This trend is less obvious in a more qualitative comparison of crystallographically controlled magnetite in plagioclase has
NRM and shipboard estimates of alteration from MARK area also been suggested (Natland, 2002; Selkin et al., 2000; Xu
gabbros (Figure 27(b)). Nonetheless, it appears that earlier et al., 1997a). Silicate-hosted magnetite, at least in plagioclase
suggestions of increasing remanent intensity with degree of and pyroxene, appears to be a common feature in oceanic
alteration (Swift and Johnson, 1984) are not generally appli- gabbros (Gee and Meurer, 2002) and may be responsible for
cable to oceanic gabbros. much of the stable remanence in these rocks. Hysteresis and
Curie temperatures of gabbroic samples are uniformly near Curie temperature data from single plagioclase crystals from
580 °C (e.g., Gee et al., 1997; Pariso and Johnson, 1993b; gabbros at the MARK area and from Leg 209 (Figure 28)
Worm, 2001), indicating that Ti-poor magnetite is the domi- illustrate that these fine-grained magnetite grains may have
nant remanence carrying phase. Primary discrete magnetite substantial coercivities.
(associated with more abundant ilmenite) is abundant in With the exception of FeTi oxide gabbros, most oceanic
FeTi oxide gabbros but rare in the less-evolved rocks that gabbros have high stability magnetizations and K€ onigsberger
constitute the bulk of the gabbroic material. In these latter ratios (Q) > 1, indicating that they may constitute an ideal
samples, much of the stable remanence may be carried by magnetization source. For example, K€ onigsberger ratios for
gabbros from Hole 735B are essentially all >1 and exceed from Hole 735B are characterized by significantly higher MDF
10 for much of the lower kilometer of the section values (several 10s mT). Comparable MDF values (average
(Figure 26). The increase in Q downhole is paralleled by a 30 mT) have been reported for gabbros from Hess Deep
general increase in coercivity, as measured by the median (Pariso et al., 1996b) and for the 1400 m of gabbro sampled
destructive field (MDF). Oxide-rich gabbros are characterized from the Atlantis Massif (Blackman et al., 2006).
by lower stabilities, as might be expected from the abundance The variability of inclination values in gabbroic rocks
of coarse-grained magnetite grains that are particularly suscep- (Figure 26) is considerably less than for extrusive sections
tible to acquiring a drilling-induced remanence. The olivine and presumably reflects significant averaging of secular varia-
gabbros and gabbros that constitute the bulk of the material tion during slow cooling. For the lowermost 1000 m of Hole
735B sampled during Leg 176, the directional dispersion
10 parameter (kappa ¼ 59, n ¼ 339) is 1.6 times the value
expected from the paleosecular variation model of McFadden
et al. (1988). The higher value of kappa indicates tighter
grouping of the inclination data and the implied time averag-
NRM (A m–1)
1.0
ing during slow cooling is a necessary prerequisite for using the
average inclination data to infer tectonic tilts. Although the
inclinations from Hole 735B are offset from the expected
0.1 Leg 153 dipole value due to tilting, the uniform polarity and relatively
894G high magnetization of this 1500 m gabbroic section (where
735B
Hess Deep lavas and dikes have been tectonically removed) are sufficient
to account for the sea-surface anomaly amplitudes at the site
0.9 0.8 0.7 0.6 0.5 0.4 (Dick et al., 2000). While uniform polarity is likely the norm, it
(a) Mg # should be noted that complex multicomponent remanences
and dual polarities within a site have been reported from
100
gabbros from the MARK area and the Atlantis Massif
(Blackman et al., 2006; Cannat et al., 1995; Gee and Meurer,
2002). These complex remanence characteristics have been
attributed to crustal accretion and cooling spanning multiple
NRM (A m–1)
1 polarity intervals (Gee and Meurer, 2002) and, if more gener-

ally applicable, would serve to reduce the gabbroic contribu-
tion to the magnetic anomalies.
0.01
Leg 153 gabbros
735B olivine gabbro/troctolite 5.12.3.4 Mantle-Derived Peridotites
735B FeTi oxide gabbros
Although pristine ultramafic rocks are nonmagnetic (contain-
0 20 40 60 80 100 ing only paramagnetic silicates and spinels), serpentinization
(b)
Alteration (%) results in the production of magnetite (e.g., Toft et al., 1990)
Figure 27 Variation of NRM of oceanic gabbros as a function of Mg#
and magnetizations that are sufficient to constitute a potential
(whole rock Mg2+/(Mg2++Fe2+)) and alteration (after Gee et al., 1997, with significant source for marine magnetic anomalies (e.g.,
additional data from Kikawa and Ozawa, 1992). Qualitative alteration Nazarova, 1994). Well-documented exposures of serpenti-
data for Leg 153 gabbros are from shipboard descriptions; percent nized peridotite at slow-spreading ridges (Cannat, 1993;
alteration for Hole 735B gabbros (Kikawa and Ozawa, 1992) is based on Cannat et al., 1997) indicate that such upper mantle material
pyroxene alteration. NRM data have not been reduced to equatorial values. undoubtedly contributes, at least locally, to magnetic
4
1275D 30R-1, 31 cm 1275D 30R-1, 28 cm (plagioclase) 1.0 1275D 30R-1, 28 cm
Normalized magnetization
(plagioclase)
Moment (10–9 Am2)
2 0.8
0.6
0
0.4
–2 Hc = 36.8 mT
Hrc = 75.5 mT 0.2
Mr = 1.56 nAm2
Ms = 3.98 nAm2 0.0
–4
–400 0 400 200 400 600
Induction (mT) Temperature (°C)
Figure 28 Example of silicate-hosted magnetite in plagioclase from gabbro sampled during Leg 209, near 15°200 N fracture zone left, photomicrograph
(25, oil immersion) of plagioclase with elongate magnetite; center, hysteresis loop; right, Curie temperature curve for a single plagioclase crystal.
anomalies. Serpentinization of peridotites and lower crustal and Figure 30 shows that relatively little magnetite is produced
gabbros has also been suggested as the cause of higher anom- until the serpentinization exceeds 50%. This nonlinear behav-
alous skewness of anomalies at slow-spreading ridges (Dyment ior may be attributed to the FeO content of the silicate alter-
and Arkani-Hamed, 1995; Dyment et al., 1997). In addition, ation phases. For <75% serpentinization, early FeO-rich
the enhanced positive magnetization (for both normal polarity lizardite is formed, and with continued alteration, this lizardite
crust and reverse polarity crust) at the ends of slow-spreading is replaced by more FeO-poor chrysotile, with a concomitant
ridge segments has been attributed to induced magnetization release of iron and production of magnetite (Oufi et al., 2002).
of serpentinized peridotites and gabbros (Pariso et al., 1996b; Alteration assemblages that include even more FeO-rich
Pockalny et al., 1995; Tivey and Tucholke, 1998). brucite may further suppress magnetite production, as evident
As with sampling of the gabbroic portion of the oceanic in some of the lower magnetization samples from Sites 895
crust, drilling at tectonic exposures (Site 895 at Hess Deep and 670 (Figure 29). Most dredged peridotites and some drill-
(Gillis et al., 1993), Sites 670 and 920 in the MARK area core samples (particularly from breccias or fault zones) have
(Cannat et al., 1995; Detrick et al., 1988), and near the 15° experienced postserpentinization low-temperature oxidation,
20’ fracture zone (Kelemen et al., 2004)) has significantly as evidenced by the presence of maghemite (e.g., Nazarova
enhanced our understanding of the magnetic properties of et al., 2000). Although magnetic properties may vary consid-
oceanic ultramafic rocks. Minor amounts of serpentinized peri- erably, the geometric mean intensity (2.1 A m1 equatorial
dotite were also (unexpectedly) recovered from a number of value including results from Leg 209; Table 2) and average
Atlantic drill sites (Sites 334, 395, 556, 558, and 560) designed K€onigsberger ratio of 2 suggest that both remanent
to sample oceanic lavas. Oufi et al. (2002) provided a thorough magnetization and induced magnetization from oceanic ser-
analysis and review of magnetic studies on serpentinized peri- pentinized peridotites may contribute significantly to magnetic
dotites up to and including Leg 153 drilling in the MARK area. anomalies. The influence of serpentinized peridotites on the
These results and the more recent remanence data from Leg anomaly pattern, however, may largely be limited to fracture
209 (Kelemen et al., 2004) are summarized in Table 2. zones and slow-spreading ridges where mantle-derived perido-
Magnetizations of serpentinized peridotites, whether tite exposures are relatively common.
obtained by dredging or drilling, vary from <0.1 A m1 to Although many serpentinites may predominantly have a
values higher than 30 A m1 that are comparable to typical viscous remanence, the origin of the remanence in oceanic
values for older oceanic basalts (Figure 29; Table 2). The serpentinized peridotites is likely to include a chemical rema-
magnetite content, susceptibility, and magnetization of ultra- nence and perhaps also a partial thermoremanence as demon-
mafic rocks increase with the degree of serpentinization, strated by remanence directions that deviate substantially from
though this increase is not linear and depends on the particular the present field direction (e.g., Garces and Gee, 2007). Tem-
alteration phases produced (Toft et al., 1990). The degree of perature estimates for serpentinization based on oxygen iso-
serpentinization may be estimated from the sample density tope fractionation between magnetite and serpentine vary
(Miller and Christensen (1997), with a correction for the widely, though most serpentinites from Hess Deep and the
amount of substantially denser magnetite (Oufi et al., 2002)), MARK area yield temperatures conservatively estimated as
>350 °C (Agrinier and Cannat, 1997; Fr€ uh-Green et al.,
100 1996). Based on these relatively high temperatures and the
ODP 920
ODP 895
ODP 670 S%
DSDP 395
DSDP 556,558,560 100.0 89.2 76.4 63.7 50.9 38.2
10 8 b 8.7
7 7.6
NRM (A m–1)
6 6.5
10
Js (Am2 kg–1)
1 5 5.4
=
Q
m%
4 4.3
3 3.3
0.1 1.0
= 2 2.2
Q
1 1.1
0.1
=
Q 0 0
0.01 2.4 2.5 2.6 2.7 2.8 2.9 3
0.001 0.01 0.1 1 dc (g cm–3)
Susceptibility (SI)
Figure 30 Saturation magnetization (Js) as a function of corrected
Figure 29 Log–log plot of NRM and susceptibility of oceanic density (dc) of serpentinized peridotite samples from various DSDP
serpentinized peridotite samples from various DSDP and ODP sites (data and ODP sites, showing how percentage of magnetite (m%) increases with
compiled by Oufi et al., 2002). Distribution of values compared to lines of the degree of serpentinization (S%). Dashed curve is best-fitting
constant Q (K€onigsberger ratio of remanent to induced magnetization; relationship between Js and dc for a subset of samples. Reproduced from
calculated for field of 40 A m1) shows that remanence is greater than Oufi O, Cannat M, and Horen H (2002) Magnetic properties of variably
induced magnetization in most samples. NRM data have not been serpentinized abyssal peridotites. Journal of Geophysical Research
reduced to equatorial values. 107(B5). http://dx.doi.org/10.1029/2001JB000549.
simple demagnetization behavior of MARK area peridotites, accepted) pattern of nonvertical magnetic boundaries sepa-
Lawrence et al. (2002) suggested that much of remanence rately from the effects related to (more speculative) rotation
might be thermal in origin. While coarse-grained magnetite is of the magnetic source.
abundant, values of Mrs/Ms as high as 0.4 and direct grain-size
estimates from image analysis indicate that relatively fine-
grained magnetite is also present (Lawrence et al., 2002; Oufi 5.12.4.1 Nonvertical Magnetic Boundaries
et al., 2002). These finer particles are presumably responsible The pattern of magnetization boundaries (isochrons) in the
for the relatively stable remanence in the MARK area perido- oceanic crust is shown schematically in Figure 31. Isochrons
tites, which was used to infer that little tilting occurred during within the extrusive layer are expected to dip toward the ridge
uplift and exposure (Lawrence et al., 2002). (e.g., Kidd, 1977) as new lavas progressively cover preexisting
flows. The resulting sloping (sigmoid) magnetization bound-
ary in the extrusives can be conveniently modeled by a
5.12.4 Crustal Accretion and Structure Gaussian filter applied to the standard magnetization block
of the Magnetic Source model. The transition zone width, the width over which 90%
of the magnetization change occurs, is approximately four
Crustal accretionary processes exert a fundamental control on times the standard deviation (s) of the Gaussian filter
the spatial distribution of magnetization sources, limiting the (Atwater and Mudie, 1973). In a classic survey at 21° N on
temporal resolution of the geomagnetic signal resolvable from the EPR, Macdonald et al. (1983) documented both the valid-
marine magnetic anomalies and, in some cases, modulating ity of this general model and the spatial extent of lava spillover
their shape (skewness). In addition, the remanent magnetiza- onto preexisting flows by comparing the location of the rever-
tion vector may be rotated during accretion (e.g., tilt due to sal boundary inferred from near-bottom magnetic anomaly
lava loading) or by subsequent tectonic deformation. Any data and the polarity of individual volcanic features at the
tectonic rotation of the magnetic source layer about a ridge- surface from submersible observations. The surface polarity
parallel rotation axis will result in an equivalent phase shift of boundary was displaced (away from the ridge) by 250–500 m
the magnetic anomaly (Cande and Kent, 1985). Anomalous relative to the vertically averaged polarity boundary deter-
skewness, the phase shift not accounted for by a standard thin mined from the near-bottom anomaly data, suggesting a tran-
layer block model (Cande, 1976), therefore provides a strong sition zone width of 1.4–1.8 km at this intermediate-spreading
constraint on any tectonic tilt of the magnetization source, as rate (60 mm year1 full rate) (Macdonald et al., 1983). Com-
well as nonvertical polarity boundaries. In the discussion later parable transition zone widths (1.5–3.0 km) have been
in the text, we will address the consequences of the (generally determined from near-bottom anomaly data at intermediate-
Total
Lavas
200
Dikes
Anomaly (nT)
Gabbros
0
–200
Lavas
3
5.0 A m–1
4 } 2.3Dikes
Depth (km)
–1
Am
5
6 Gabbros
1.2 A m–1
7
8 100s meters
20 30 40 50 60
Distance from ridge axis (km)
Figure 31 Schematic illustration of magnetization boundaries in oceanic crust (bottom) and magnetic anomaly models for these magnetic source
bodies (top). The magnetization values of the three layers are based on the overall geometric mean values in Table 2. Anomalies are calculated for
vertical magnetization and ambient field for a full-spreading rate of 80 km My1. No Gaussian filter has been applied to the magnetization.
to superfast-spreading (up to 150 mm year1 full-rate) ridges ridge, but there is little detectable effect on the anomaly shape
(Bowers et al., 2001; Sempere et al., 1987). Less focused accre- or amplitude. Based on vertical magnetic profiling data from
tion at slow-spreading ridges likely results in broader transition intermediate-spreading crust exposed at the Blanco escarp-
zones (Sempere et al., 1987), though the spreading rate depen- ment, substantially broader ( 10 km) transition zone widths
dence has not been well established. have been suggested (Tivey, 1996; Tivey et al., 1998a). While
The narrow (1–3 km) transition zone widths (which repre- this novel profiling technique has considerable promise in
sent an upper bound due to the effects of minor extension mapping the distribution of polarities, the very broad transi-
and the 7 ky (Clement, 2004) necessary for the field to tions at the Blanco escarpment predict substantial changes
reverse) imply a relatively high-fidelity recording process at in anomaly amplitude and shape that are not generally
intermediate- and fast-spreading ridges, compatible with the apparent. The narrow polarity transition in the sheeted dikes
recognition of short polarity events (e.g., the 10 ky Cobb (sdikes sflows) is likewise not expected to significantly affect
Mountain or Reunion events; Table 1) in anomaly profiles at the anomaly shape or amplitude.
these spreading rates. These estimates are also consistent with In contrast, the gently sloping boundaries in the gabbroic
independent estimates of the width over which lavas accumu- lower crust should give rise to significant anomalous skewness
late. Submersible observations at intermediate- and fast- (Cande and Kent, 1976). Anomalies generated by such gently
spreading ridges suggest that most volcanic activity is concen- dipping polarity boundaries are illustrated in Figure 32 for a
trated within 1 km of the axis (Karson et al., 2002; Perfit and range of polarity interval lengths and spreading rates. The
Chadwick, 1998). Flows erupted off-axis or transported down- resulting anomalous skewness is most easily recognized for
slope may extend for a few kilometers from the ridge (Fornari broad polarity intervals and amounts to a phase shift of
et al., 1998; Perfit et al., 1994) and off-axis accumulation of 45°, in the same sense as global observations of anomalous
lavas can significantly affect the pattern of magnetic anomalies skewness. The distortion is not a pure phase shift, however, as
(Gee et al., 2000; Schouten et al., 1999). The off-axis thicken- evidenced by the asymmetrical shoulder at the older edge that
ing of seismic layer 2A (commonly interpreted as correspond- coincides with the magnetization contrast at the base of the
ing to the extrusive layer) within 1–3 km of the axis (e.g., gabbroic layer. Although the anomaly shapes for narrow inter-
Harding et al., 1993; Vera and Diebold, 1994) also suggests vals or at different spreading rates appear quite different, the
that lava accumulation occurs over a narrow zone comparable anomalous skewness introduced by a conductively cooled gab-
to the transition zone widths inferred from anomaly data. broic layer is essentially independent of spreading rate. A three-
The pattern of magnetization contrasts in the sheeted dikes layer source model (with subequal contributions from the
and gabbros is less well constrained. The zone of dike intru- lavas, dikes, and gabbros as indicated by the average rema-
sion, where mixed polarity dikes would occur (Figure 31), is nence of these layers and their thickness and depth; Table 2)
likely much narrower than for the lava flows. In order to match illustrates that the anomalous skewness imparted by the gab-
seismic constraints and the sharp extrusive/dike boundary broic layer results in an overall anomalous skewness of 10°
observed in ophiolites, Hooft et al. (1996) found that the (Figure 31). This value is well within the range (5–25° for a
zone of dike intrusion at the EPR must be less than a few full-spreading rate of 80 km My1) determined from global
hundred meters wide (1s of 10–50 m). To the extent that the compilations of anomalous skewness data (e.g., Petronotis
lower crust is conductively cooled, polarity boundaries in et al., 1992; Roest et al., 1992). Although rotations of the
the gabbroic portion of the crust should approximately follow magnetic remanence direction (which are not included in the
the shape of the 580 °C isotherm (Cande and Kent, 1976; model shown in Figure 31) may result in additional anoma-
Figure 32). Half-space conductive cooling models (Oxburgh lous skewness, such rotations are apparently not required to
and Turcotte, 1969; Sclater and Francheteau, 1970) predict that explain the modest anomalous skewness at intermediate- and
magnetization isochrons in the gabbroic layer should dip away faster-spreading rates.
from the ridge at 4–30° (for full-spreading rates from 150 to
16 km My1, respectively). Near-bottom magnetic anomaly
5.12.4.2 Rotations of the Magnetic Source Layer
data from tectonic exposures of lower crustal gabbro at Atlantis
Bank, together with inclination data from wireline rock drill Normal faulting is an integral part of extensional seafloor
samples and from the 1.5 km penetration at Hole 735B, have spreading, and therefore rotations, either toward or away
been interpreted as indicating the presence of a sloping iso- from the ridge, about ridge-parallel axes are not unexpected.
chron with a present-day dip of 25° (Allerton and Tivey, Inward (toward ridge)-dipping faults are dominant at all but
2001). However, the well-determined remanent inclination the fastest-spreading rates (Carbotte and Macdonald, 1990;
(71°; Figure 26) from Hole 735B suggests an outward tilt of Kriner et al., 2006), and even at superfast-spreading rates,
about 20°, so that the original dip of this isochron was approx- these faults accommodate most of the strain (Bohnenstiehl
imately 5°. While the dip direction is away from the ridge, as and Carbotte, 2001). These inward-dipping faults should
expected, the very shallow dip angle is difficult to reconcile result in outward tilt of the magnetic source layer, which will
with that expected (30°) from conductive cooling ( John produce a phase shift (1° for each 1° of tilt) of the anomalies
et al., 2004) at this slow-spreading ridge. that is compatible with the observed sense of anomalous skew-
The effect of nonvertical magnetization boundaries on the ness (Figure 32; Cande, 1978).
shape and amplitude of anomalies should be minor for lavas Tilting in response to lava loading has recently been sug-
and dikes but is significant for the gabbroic portion of the crust gested as a potential way of rotating the remanence vector of
(Figure 31). For 1–3 km wide transitions in the extrusive layer, seafloor lavas (Schouten and Denham, 2000; Schouten et al.,
the corresponding anomaly is shifted slightly away from the 1999; Tivey et al., 2005). In these models, the sloping
Nonvertical polarity boundaries Rotation of remanence in lavas

in lower crust
100
Anomaly ( nT )
No tectonic tilt
Anomaly ( nT )
50
0.1 My –100
–50
Normal polarity
Depth ( km)
Depth ( km)
5 3.0
6
3.2
7
8 3.4
0 100 200 300 40 80 120 160
16° Inward tilt

100
Anomaly ( nT )
(lava loading)
Anomaly ( nT )
50
–50 0.5 My –100

Normal polarity
Depth ( km)
5
Depth ( km)
3.0
6
3.2
7
8 3.4
0 100 200 300 40 80 120 160
40 km My–1 40 km My–1
80 km My–1 80 km My–1
140 km My–1 140 km My–1
100 16° Outward tilt

Anomaly ( nT )
(block faulting)
Anomaly ( nT )
50
–50 1.0 My –100

Normal polarity
Depth ( km)
Depth ( km)
5 3.0
6
3.2
7
8 3.4
0 100 200 300 40 80 120 160
Distance from ridge axis (km) Distance from ridge axis (km)
Figure 32 Magnetic anomaly models illustrating the effect of sloping cooling isotherms in gabbroic layer and various lava tilts. Left, conductively
cooled magnetization boundaries in the gabbroic layer are illustrated for three different length polarity intervals (0.1, 0.5, and 1.0 My) at three spreading
rates. This normal polarity interval is indicated by the three shaded regions (green, red, and blue for 40, 80, and 140 km My1 full-spreading rates,
respectively). The corresponding magnetic anomalies have an anomalous skewness of 45°. Anomalies are calculated for vertical magnetization and
ambient field. Right, the effect of tilt of the extrusive layer on anomalies from a 0.5 My normal polarity interval at the same three spreading rates.
The sigmoid magnetization boundary reflects lava accumulation over 2 km and introduces no detectable change in the anomaly shape
(upper plot). Ridgeward or outward tilting of 16° (the average tilt implied by lava accumulation over 2 km) results in an equivalent amount of
anomalous skewness.
isochrons within the extrusive layer are taken as also reflecting can result in significant scatter even when the entire lava sec-
the inward tilt of individual lava flows. The resulting downhole tion is sampled (e.g., 504B; Figure 23) and so average inclina-
distribution of inclinations (possibly modified by later block tion results from shallow penetration holes should be viewed
rotation) can then be used to infer the average pattern of lava with caution. Second, the lack of azimuthal orientation of the
accumulation. This type of model requires that the remanence drillcore samples precludes the recognition of a systematic
in relatively small intervals within the extrusive layer represent sense for ridges that strike nearly N–S, as does much of the
the time-averaged field direction, so that downhole trends in Pacific and Mid-Atlantic Ridge system. In principle, E–W strik-
the average inclination can be detected. If prevalent, such lava ing ridges might provide information on systematic tilt in the
loading would produce systematic inward tilt of the lavas and extrusives, though it is unclear whether sufficient data exist to
anomalous skewness in the opposite sense of that observed. conduct such an analysis.
The relatively thin extrusive layer and the generally narrow Magnetic anomaly shapes provide the most robust estimate
region over which flows accumulate limit the amount of tilting of the average tilt of the magnetization source. Comparison of
expected from lava loading. For accumulation of a 0.5 km thick drillcore data from three relatively deep penetration holes in
extrusive layer over a typical transition zone width of 2 km, the Cretaceous (M0 age) crust (Figure 33) and correlative
the average dip of the isochrons (and therefore inward tilt of anomaly shapes (Figure 34) illustrates how models of block
the remanence) is 16°, which is accompanied by a readily tilting and lava loading can be independently evaluated. The
observable anomalous skewness of 16° (Figure 32). mean inclinations in Holes 417A and 418A (where there is a
A significant number of drill sites sampling the extrusive polarity reversal at 500 mbsf ) are eminently compatible with
layer have inclinations that deviate from expected values, and expected inclinations from the Early Cretaceous paleopoles for
block rotations and/or lava loading undoubtedly contributes the North American Plate (Bosum and Scott, 1988; Levi,
to some of these deviations. In the most recent (although 1980). The inclinations from the upper portion of Hole
30 years ago) compilation of inclination results from drillcore 417D are steeper and have been attributed to tectonic tilting
data, Lowrie (1977) found that observed inclination values by some authors (Verosub and Moores, 1981), although Levi
were well correlated with expected values, though the results (1980) argued that the inclination data are compatible with
exhibit appreciable scatter about the ideal 1:1 line. Although paleosecular variation and do not demand any tectonic rota-
no significant bias toward either shallower or steeper inclination. Schouten (2002) interpreted the Hole 417A data (with
tions was found, two aspects of the drillcore data make it reverse polarity inclination very similar to the expected value)
unlikely that any systematic bias (whether from block rotations as reflecting two oppositely directed tectonic tilts: a large initial
or lava loading) would be evident. First, paleosecular variation tilt toward the ridge axis as function of lava loading (requiring
Hole 418A Hole 417D Hole 417A

inclination inclination inclination
–60° 0° 60° –60° 0° 60° –60° 0° 60°
300 300 200
400 400 300
500 500 400

Depth (mbsf)
600 600
700 700
Pillow
800 Massive flow
Dike
Breccia
900
Figure 33 Stable remanent inclinations (i.e., after partial demagnetization to remove spurious components) from discrete samples from seafloor lavas
at DSDP Holes 417A, 417D, and 418A. The heavy lines indicate the expected Cretaceous inclination values at the site. Data from Levi (1980).
417D 417A Relatively few observations are available that constrain pos-
500 m 5400 sible rotations of the intrusive portion of the crust. Based on
7.38 s
GC 031276 2210Z 7.375 s 5531 m submersible studies of tectonic exposures of fast-spread crust at
417D 5600 Hess Deep, much of the sheeted dike complex appears to have
102° 5686 m
Meters
417A 41°
a systematic dip (60°) away from the ridge (Karson et al.,
Fre 5800 5835 m
dM 2002) although areas with approximately vertical dikes are also
oor
e li
ne
25° 06.5' N 1 6000 present (Francheteau et al., 1992). As with the lavas, anoma-
68° 02.5' W
68° 03.0′ W
sp 3177 7.37 s lous skewness constraints suggest that such inward tilting of
6200 295° 115°
the sheeted dikes is unlikely to be a general feature, provided
–400 –200 0 200 400 that the dike remanence is a significant contributor to the
(a) (b) Meters magnetic anomalies as available data seem to now indicate.
Rotation Substantial rotations of the gabbroic layer have been docu-
models mented at slow-spreading ridges. For example, tectonic tilts of
20° have been well documented for gabbros at the Atlantis
= –60°
Bank (Dick et al., 1999) and even larger rotations (40–80°)
= –40° have been reported from gabbros (and associated serpenti-
nized peridotites) from the vicinity of the 15°200 N fracture
= –20°
zone in the Atlantic (Garces and Gee, 2007; Kelemen et al.,
2004). The outward rotation of the gabbroic sections, presum-
ably from flexural/isostatic adjustment in response to unload-
M4 M2 M0 ing, is compatible with the sense of anomalous skewness data.
250 Observed
nT
Although the magnitude of anomalous skewness appears to
reduced to the pole
0
increase at slower spreading rates (Roest et al., 1992), the very
large rotations noted from the 15°200 N area exceed the largest
No rotation
values of anomalous skewness and thus are unlikely to be
model representative of more typical slow-spread crust.
= 0°
0 25 km
5500
5.12.5 Future Directions
m Source
6000
There is increasing evidence that marine magnetic anomalies
(c) are capable of recording a broad spectrum of geomagnetic
Sites 417/418
field behavior, ranging from millennial-scale paleointensity
Figure 34 Tectonic rotation model for ODP Sites 417A and 417D. (a) and
variations, to polarity reversals, to apparent polar wander, to,
(b) location map; (c) diagrammatic cross section (reproduced from
Schouten H (2002) Paleomagnetic inclinations in DSDP Hole 417D
more speculatively, long-term changes in average field strength
reconsidered: Secular variation or variable tilting? Geophysical as suggested, for example, by the ramp in anomaly amplitudes
Research Letters 29(7). http://dx.doi.org/10.1029/2001GL013581), at the older end of the M-sequence. However, because of the
compared to observed and modeled magnetic anomaly data for a profile inherent ambiguity in determining source properties from
nearby Site 417A and 417D (reproduced from Cande SC and Kent DV potential field data, independent geomagnetic field estimates
(1985) Comment on ‘Tectonic rotations in extensional regimes and their will be needed to guide and calibrate inferences made from
paleomagnetic consequences for ocean basalts’ by K.L. Verosub and E.M. magnetic anomalies. Here, we highlight some general
Moores. Journal of Geophysical Research 90(B6): 4647–4651). Note that approaches – the use of autonomous vehicles, oriented sam-
the observed profile is matched best by a model with no tectonic rotation ples, absolute paleointensity of near-ridge lavas, and measure-
and is poorly represented by a model with 40° of rotation suggested
ments of the vector anomalous field – that are likely to
from inclination data at Site 417A and 417D by Schouten (2002).
significantly advance our understanding of the geomagnetic
signal recorded in the oceanic crust, as well as our ability to
the bulk of the lavas to accumulate within 300 m of the axis), utilize this information in addressing outstanding problems in
followed by 41° outward block tilting (Figure 34). Magnetic crustal accretion processes.
anomaly data across Anomaly M0, including from the imme- One general area that is ripe for further development and
diate vicinity of Sites 417 and 418 (Figure 34; Cande and Kent, applications is near-bottom observations, which are required
1985), indicate that the M0 age crust from both ridge flanks is to obtain high spatiotemporal resolution. Such near-bottom
characterized by little or no anomalous skewness and therefore observations from towed systems (often incorporating side-
little or no net tilt of the magnetic source. While such fortu- scan sonar or other instrument packages as well) have already
itous offsetting tilts could, of course, explain the lack of anom- provided seminal contributions in our understanding of the
alous skewness near Sites 417 and 418, this is unlikely to be fine structure of anomalies and source geometries (e.g.,
generally applicable. The magnitude and sense of anomalous Atwater and Mudie, 1973; Larson and Spiess, 1969). More
skewness indicate that substantial tilting (whether from lava recently, near-bottom data have been used to demonstrate
loading or block rotation) is not generally applicable, at least the lineated (i.e., geomagnetic) nature of small-scale magnetic
for fast- and intermediate-spreading ridges where anomalous anomalies, which have been shown to mainly reflect paleoin-
skewness magnitudes are generally small. tensity variations (Bowers et al., 2001; Bowles et al., 2003;
Gee et al., 2000) and to examine crustal accretionary processes The more detailed magnetic anomaly observations that are
(e.g., Hussenoeder et al., 1996; Smith et al., 1999; Tivey et al., possible with tethered vehicles, ABE, and manned submers-
2003). These towed packages are robust, proven marine geo- ibles are nicely complemented by the availability of high-
physics tools but tend to be used sparingly because such wire- resolution sampling techniques that allow the collection of
line deployments require a dedicated vessel and slow towing fully oriented samples. Block samples oriented with the Geo-
speeds. An exciting recently developed alternative is ABE Compass (which uses a compass and digital tilt meters to
(Autonomous Benthic Explorer), which can be operated simul- measure attitude) have been collected from submersible
taneously with other shipboard programs. ABE is capable of (Cogné et al., 1995; Hurst et al., 1994; Varga et al., 2004),
following preprogrammed, closely spaced track lines at low and this technique has also been successfully used with teth-
altitudes (5–40 m above the seafloor) to yield fine-scale ered remotely operated vehicles. Oriented samples from a wire-
bathymetry and high-resolution magnetic data. An example line rockdrill (Allerton and Tivey, 2001) further expand the
of an ABE survey at 17° S on the EPR (Shah et al., 2003) is range of seafloor outcrops that can be sampled to address
shown in Figure 35. In this study, a magnetic field low was tectonic problems. Indirect information on the remanence
found extending several kilometers along the axial trough, direction can also be obtained from borehole magnetometer
which was interpreted as reflecting the presence of a few- measurements with gyro-oriented logging (Bosum and Scott,
hundred-meter-wide region of weakly or nonmagnetic shallow 1988). Such borehole measurements have substantial advan-
dikes (Shah et al., 2003) but might also delineate lavas erupted tages in characterizing the magnetic source region, since they
during the recent (relative) geomagnetic intensity low can provide representative magnetization values by avoiding
(Figure 3). When combined with the submeter resolution drilling-induced remanence and directly account for void and
bathymetry, ABE magnetic anomaly data also provide a pow- porosity effects on assessments of discrete sample values.
erful technique to examine the relationship between individual Our understanding of the neovolcanic zone at fast-spreading
volcanic features and their corresponding anomaly signatures ridges, especially the EPR, has evolved and become increasingly
(e.g., Tivey et al., 1998b). Similar high-resolution anomaly sophisticated in recent years as higher-resolution observations,
data can also be obtained with the new generation of tethered such as with ABE, became available. Placing accurate age con-
vehicles that allow far more motion control than towed sys- straints to determine eruptive recurrence intervals on near-axis
tems and also may incorporate high-resolution swath mapping lava flows has become increasingly important given the struc-
and sampling capabilities. Such high-resolution mapping on tural and volcanic complexity of the neovolcanic zone. In this
the Endeavor segment of the Juan de Fuca Ridge has revealed regard, geomagnetic paleointensity of submarine basaltic glass
circular (100 m diameter) anomaly lows that correlate with (Pick and Tauxe, 1993) holds particular promise for placing
present-day and fossil hydrothermal upflow zones (Figure 36; quantitative age constraints on near-axis flows. For example,
Tivey and Johnson, 2002). multispecimen Thellier paleointensity results from four
113° 13.5 113° 13.0 113° 13.5 113° 13.0

–4800
Bathymetry Upward Survey
continued lines
–5200 field (nT)
17° 28.0
–5600
–6000
17° 28.5
–6400
–32 000
–6800
nT
–31 000
–7200
17° 29.0
–30 000
–7600
–1200 –800 –400 0 –1200 –800 –400 0

Figure 35 High-resolution bathymetric (left) and magnetic data (center) obtained by an Autonomous Benthic Explorer (ABE) survey (tracks on the
right) on the East Pacific Rise at 17° S. Note the presence of lower anomaly amplitudes along the axial summit trough (heavy lines in the center of the
figure). Modified from Shah AK, Cormier M-H, Ryan WBF, et al. (2003) Episodic dike swarms inferred from near-bottom magnetic anomaly maps
at the southern East Pacific Rise. Journal of Geophysical Research 108(B2): 2097. http://dx.doi.org/10.1029/2001JB000564.
5 311 000 New

A′
Field
Hulk-
Crypto
5 310 800
Dante-
Grotto
UTM meters north
Bastille
Relict
Field
5 310 200
0
Beach
600
nT
A
5 310 000
492 400 492 800 492 400 492 800 492 400 492 800
UTM meters east UTM meters east UTM meters east
2060 2140 2220 2300 m 0 1 2 3 4 5 6 7 A m–1

Figure 36 Near-bottom bathymetric and magnetic observations on the Endeavor segment of the Juan de Fuca Ridge showing hydrothermal burnholes.
Red areas indicate locations of active hydrothermal vents. Modified from Tivey MA and Johnson HP (2002) Crustal magnetization reveals
subsurface structure of Juan de Fuca Ridge hydrothermal vent fields. Geology 30: 979–982.
independent samples distributed over several kilometers from a Marine magnetic surveys have traditionally used total field
single (Animal Farm) flow on the EPR at 18° S were found to sensors that rely on well-established physical constants (e.g., the
be in excellent mutual agreement (35.6 1.0 mT), and by com- proton gyromagnetic ratio for proton precession magnetome-
parison to a geomagnetic reference curve projected to the site ter) and thus provide accurate (1 nT) field intensity data with
location, the Animal Farm paleointensity value indicated a fairly negligible drift. The utility of such total field measurements,
recent time of eruption (AD 1910 20) in general agreement however, is limited in certain geometries (N/S spreading ridges
with other observations (Carlut and Kent, 2000; Figure 37). An near the magnetic equator generate very low-amplitude total
integrated bathymetric, geochemical, and paleointensity study field anomalies), and external field variations, particularly at
of adjacent and contrasting ridge segments at 15° N on the low magnetic latitudes, may further inhibit the identification
EPR (Carlut et al., 2004) and one at 9–10° N that involved the of spreading-related anomalies. The use of total field magnetic
analysis of what may be the largest published dataset of absolute gradiometers, two sensors typically deployed with a horizontal
(Thellier) paleointensity determinations (551 accepted of 991 separation of >100 m, allows the recognition and removal of a
basalt glass samples analyzed; Bowles et al., 2006) further illus- significant portion of the external field variations (e.g., Roeser
trate the potential for dating lavas and, more generally, for et al., 2002). With the development of high-resolution
discriminating whether eruptions were synchronous. A cooling ( 0.01 nT) total field sensors, this technique shows consider-
rate bias in Thellier paleointensities was shown not to be as able promise for facilitating the recognition of low-amplitude
important as initially suspected in rapidly cooled submarine anomalies at equatorial latitudes. Measurement of the vector
basaltic glasses (Bowles et al., 2005). However, terrain effects components of the anomalous field also provides two distinct
may be an important source of uncertainty in absolute pal- advantages that may aid in the recognition of lineated (geomag-
eointensity values in some settings, such as the Juan de Fuca netic) anomalies. Perhaps, the most useful attribute of vector
Ridge where there are large amplitude magnetic anomalies anomaly data is the ability to characterize, with a single profile,
(Carlut and Kent, 2000). Ultimately, the reliability of paleoin- the degree to which the magnetic source is two-dimensional
tensities for dating depends on the reference field model, which (Blakely et al., 1973; Parker and O’Brien, 1997). Aeromagnetic
could be improved considerably with higher-quality (and better vector profiles, where the sensor is typically mounted on a gyro-
dated) Thellier data on archaeological and geologic materials. stabilized platform, have been used to estimate the spreading
This would also allow better calibration of sedimentary relative lineation direction and to map low-amplitude anomalies in the
paleointensity records, which are used for making comparisons equatorial Pacific (Horner-Johnson and Gordon, 2003; Parker
over time intervals beyond the radiocarbon-dated record (e.g., and O’Brien, 1997). Shipboard three-axis magnetometers have
Gee et al., 2000). also been used to determine the location and azimuth of
55.0, 58.9
18° 35.2 S 50 (off-scale)
Solid symbols = younger
Open symbols = older
35.4 North
45
Middle
35.6 South
Paleointensity (μT)
40
Latitude
35.8
36.0
35
36.2 1993 field = 31.3 μT

30
36.4
18° 36.6 s 25
30 32 34 36 38 40 1600 1700 1800 1900 2000
(a) Intensity (μT) (b) Calender date
Figure 37 Absolute paleointensity results from submarine basaltic glass illustrating the potential as a dating tool for young mid-ocean ridge lavas.
(a) Paleointensity results from multiple samples within a single flow (mapped by submersible) highlight the degree of reproducibility (1 mT) possible.
(b) Glass paleointensity results from the southern East Pacific Rise compared to global field model of Jackson et al. (2000) that illustrates how
paleointensity data can be used to infer eruption ages. The ages inferred from this comparison are consistent with all geologic data and radiometric dates from
these flows. (a) Modified from Carlut J, Cormier M-H, Kent DV, Donnelly KE, and Langmuir CH (2004) Timing of volcanism along the northern East Pacific Rise
based on paleointensity experiments on basaltic glasses. Journal of Geophysical Research 109: B04104. http://dx.doi.org/10.1029/2003JB002672.
(b) Modified from Bowles J, Gee JS, Kent DV, Bergmanis E, and Sinton J (2005) Cooling rate effects on paleointensity estimates in submarine basaltic glass and
implications for dating young flows. Geochemistry, Geophysics, Geosystems 6: Q07002. http://dx.doi.org/10.1029/2004GC000900.
23 24 25 26
PPM 1.2
40
MRU6
30 P-C
Aeromag
Inclination
Bx
20 MAR
MRU6 Bx P-N
seamount
Aeromag 10 1.5
Bz 2
400 5
20 10
MRU6 Bz nT 0
0 10 20 30 40
0 Declination–lineation azimuth
(a) (b)
0 100 km
Figure 38 Anomaly data from a surface-towed vector magnetometer and plot of amplitude ratio of vector components relative to total field anomalies
(Gee and Cande, 2002). (a) Comparison of vector anomalies measured by aircraft and with towed vector magnetometer. Upper panel shows comparison of
total field anomaly from proton precession magnetometer (PPM) and vector magnetometer (MRU6). Bz and Bx are the vertical and along-track vector
anomalies. (b) Ratio of amplitudes of vector components and total field anomaly (contours) for a perfectly two-dimensional source. This ratio depends on the
angular difference between the spreading lineation and the ambient field declination, as well as the remanent inclination. The boxes delineate the range
of these parameters for the equatorial Pacific–Nazca (P–N) Ridge, Pacific–Cocos (P–C) Ridge, and Mid-Atlantic Ridge (MAR) systems.
magnetization contrasts in several areas (e.g., Korenaga, 1995; Although many high-resolution magnetic studies are con-
Seama et al., 1993; Yamamoto et al., 2005). Towed vector ducted near the ridge crest, emerging technologies may also
magnetometer systems can effectively eliminate the ship effect allow the documentation of the seafloor geomagnetic signal
and resolve vector anomalies on the order of 30–50 nT throughout the ocean basins. A relevant new platform for
(Figure 38; Gee and Cande, 2002). vector and total field magnetic instruments would be on
unmanned aerial vehicles (UAVs) for surveying in the marine vol. 304/305. Integrated Ocean Drilling Program Management International, Inc., for
environment. Deployments of UAVs from oceanographic the Integrated Ocean Drilling Program. http://iodp.tamu.edu/publications/exp304_
305/30405title.htm (accessed on Jan 2007).
research vessels would allow multifold increases in magnetic
Blakely RJ (1974) Geomagnetic reversals and crustal spreading rates during the
coverage at spatial resolutions comparable to sea-surface data. Miocene. Journal of Geophysical Research 79: 2979–2985.
High-density magnetic data are necessary to quantitatively Blakely RJ (1976) An age dependent, two-layer model for marine magnetic anomalies.
evaluate the relative contributions from a lineated or coherent American Geophysical Union Monographs 19: 227–234.
geomagnetic source and nonlineated heterogeneities in the Blakely RJ, Cox A, and Lufer EJ (1973) Vector magnetic data for detecting short polarity
intervals in marine magnetic profiles. Journal of Geophysical Research
recording medium originating from crustal accretionary pro- 78: 6977–6983.
cesses. Of particular interest are regions like the 25% of the Bleil U and Petersen N (1983) Variations in magnetization intensity and low-temperature
ocean floor that is encompassed by the KQZ ( 84–122 Ma), titanomagnetite oxidation of ocean floor basalts. Nature 301: 384–388.
which remain virtually uncharted in terms of fundamental Bohnenstiehl DR and Carbotte SM (2001) Faulting patterns near 19°300 S on the East
Pacific Rise: Fault formation and growth at a superfast spreading center.
properties such as spreading history and the possible presence
Geochemistry, Geophysics, Geosystems 2(9). http://dx.doi.org/
of lineations that might be related to geomagnetic variations. 10.1029/2001GC000156.
There are also virtually no deep crustal drill holes in the KQZ to Bosum W and Scott JH (1988) Interpretation of magnetic logs in basalt, Hole 418A.
document the magnetization of oceanic crust that formed in Proceedings of the Ocean Drilling Program, Scientific Results 102: 77–95.
this unusual time interval of no geomagnetic reversals. Tools Bowers NE, Cande SC, Gee JS, Hildebrand JA, and Parker RL (2001) Fluctuations of the
paleomagnetic field during chron C5 as recorded in near-bottom marine magnetic
are thus available that should allow us to significantly increase anomaly data. Journal of Geophysical Research 106: 26379–26396.
our understanding of the source of oceanic magnetic anoma- Bowles J, Gee JS, Kent DV, Bergmanis E, and Sinton J (2005) Cooling rate effects on
lies and geomagnetic variations over timescales ranging from paleointensity estimates in submarine basaltic glass and implications for dating
103 to 108 years. young flows. Geochemistry, Geophysics, Geosystems 6: Q07002. http://dx.doi.org/
10.1029/2004GC000900.
Bowles J, Gee JS, Kent DV, Perfit M, Soule A, and Fornari D (2006) Paleointensity
results from 9°–10°N on the East Pacific Rise: Implications for timing and extent of
Acknowledgments eruptive activity. Geochemistry, Geophysics, Geosystems 7: Q06006. http://dx.doi.
org/10.1029/2005GC001141.
Bowles J, Tauxe L, Gee J, McMillan D, and Cande SC (2003) The source of tiny wiggles
The authors gratefully acknowledge the support of various NSF
in chron C5: A comparison of sedimentary relative intensity and marine magnetic
grants over more than a decade that allowed much of this work anomalies. Geochemistry, Geophysics, Geosystems 4(6): 1049. http://dx.doi.org/
to be completed. Lamont-Doherty Earth Observatory contri- 10.1029/2002GC000489.
bution 6992. Cande SC (1976) A paleomagnetic pole from Late Cretaceous marine magnetic
anomalies in the Pacific. Geophysical Journal of the Royal Astronomical Society
44: 547–566.
Cande SC (1978) Anomalous behavior of the paleomagnetic field inferred from the
References skewness of Anomalies 33 and 34. Earth and Planetary Science Letters
40: 275–286.
Acton GD, Petronotis KE, Cape CD, Ilg SR, Gordon RG, and Bryan PC (1996) A test of Cande SC and Gee JS (2001) Evidence for the rapid low-temperature alteration of
the geocentric axial dipole hypothesis from an analysis of the skewness of the oceanic crust on the Pacific–Antarctic Ridge from the amplitude envelope of
central marine magnetic anomaly. Earth and Planetary Science Letters magnetic anomalies. Eos, Transactions American Geophysical Union 82: F348.
144: 337–346. Cande SC and Kent DV (1976) Constraints imposed by the shape of marine magnetic
Agrinier P and Cannat M (1997) Oxygen-isotope constraints on serpentinization anomalies on the magnetic source. Journal of Geophysical Research
processes in ultramafic rocks from the Mid-Atlantic Ridge (23°N). Proceedings of 81: 4157–4162.
the Ocean Drilling Program, Scientific Results 153: 381–388. Cande SC and Kent DV (1985) Comment on “Tectonic rotations in extensional regimes
Allerton S and Tivey MA (2001) Magnetic polarity structure of the lower oceanic crust. and their paleomagnetic consequences for ocean basalts” by K.L. Verosub and E.M.
Geophysical Research Letters 28: 423–426. Moores. Journal of Geophysical Research 90(B6): 4647–4651.
Allerton S, Worm HU, and Stokking LB (1996) Paleomagnetic and rock magnetic Cande SC and Kent DV (1992a) A new geomagnetic polarity time scale for the Late
properties of Hole 896A. Proceedings of the Ocean Drilling Program, Scientific Cretaceous and Cenozoic. Journal of Geophysical Research 97: 13917–13951.
Results 148: 217–226. Cande SC and Kent DV (1992b) Ultrahigh resolution marine magnetic anomaly profiles:
Anonymous (1979) Magnetostratigraphic polarity units – A supplementary chapter of A record of continuous paleointensity variations. Journal of Geophysical Research
the ISSC International Stratigraphic Code. Geology 7: 578–583. 97: 15075–15083.
Atwater T and Mudie JD (1973) Detailed near-bottom geophysical study of the Gorda Cande SC and Kent DV (1995) Revised calibration of the geomagnetic polarity time
Rise. Journal of Geophysical Research 78: 8665–8686. scale for the Late Cretaceous and Cenozoic. Journal of Geophysical Research
Banerjee SK (1980) Magnetism of the oceanic crust: Evidence from ophiolite 100: 6093–6095.
complexes. Journal of Geophysical Research 85: 3557–3566. Cande SC and LaBrecque JL (1974) Behaviour of the Earth’s paleomagnetic field from
Bazin S, Harding AJ, Kent GM, et al. (2001) Three-dimensional shallow crustal small scale marine magnetic anomalies. Nature 247: 26–28.
emplacement at the 9 degrees 03 N overlapping spreading center on the East Pacific Cande SC, Larson RL, and LaBrecque JL (1978) Magnetic lineations in the Pacific
Rise: Correlations between magnetization and tomographic images. Journal of Jurassic quiet zone. Earth and Planetary Science Letters 41: 434–440.
Geophysical Research 106: 16101–16117. Cannat M (1993) Emplacement of mantle rocks in the seafloor at mid-ocean ridges.
Berggren WA, Kent DV, Swisher CC, and Aubry MP (1995) A revised Cenozoic Journal of Geophysical Research 98: 4163–4172.
geochronology and chronostratigraphy. In: Berggren WA, Kent DV, Aubry MP, and Cannat M, Karson JA, and Miller DJ (1995) Proceedings of the Ocean Drilling Program,
Hardenbol J (eds.) Geochronology Time Scales and Global Stratigraphic Initial Reports, vol. 153, p. 798. College Station, TX: Ocean Drilling Program.
Correlations (SEPM Special Publication), vol. 54, pp. 129–212. Tulsa, OK: SEPM. Cannat M, Lagabrielle Y, Bougault H, et al. (1997) Ultramafic and gabbroic exposures at
Beske-Diehl SJ (1990) Magnetization during low-temperature oxidation of sea-floor the Mid-Atlantic Ridge: Geological mapping in the 15°N region. Tectonophysics
basalts – No large scale chemical remagnetization. Journal of Geophysical Research 279: 193–213.
95: 21413–21432. Carbotte SM and Macdonald KC (1990) Causes of variation in fault-facing direction on
Billups K, Palike H, Channell JET, Zachos JC, and Shackleton NJ (2004) Astronomic the ocean floor. Geology 18: 749–752.
calibration of the Late Oligocene through Early Miocene geomagnetic polarity time Carlut J, Cormier M-H, Kent DV, Donnelly KE, and Langmuir CH (2004) Timing of
scales. Earth and Planetary Science Letters 224: 33–44. volcanism along the northern East Pacific Rise based on paleointensity experiments
Blackman DK, Ildefonse B, John BE, Ohara Y, Miller DJ, and MacLeod CJ (2006) on basaltic glasses. Journal of Geophysical Research 109: B04104. http://dx.doi.
Oceanic core complex formation, Atlantis Massif. Proceedings of the IODP-MI, org/10.1029/2003JB002672.
Carlut J and Kent DV (2000) Paleointensity record in zero-age submarine basalt Fr€uh-Green GL, Plas A, and Lecuyer C (1996) Petrologic and stable isotope constraints
glasses: Testing a new dating technique for recent MORBs. Earth and Planetary on hydrothermal alteration and serpentinization of the EPR shallow mantle at Hess
Science Letters 183: 389–401. Deep (Site 895). Proceedings of the Ocean Drilling Program, Scientific Results
Channell JET, Erba E, Nakanishi M, and Tamaki K (1995) Late Jurassic–Early 147: 255–288.
Cretaceous time scales and oceanic magnetic anomaly block models. Furuta T (1983) Magnetic properties of basalt samples from Holes 504B and 505B on
In: Berggren WA, Kent DV, Aubry M-P, and Hardenbol J (eds.) Geochronology, the Costa Rica Rift, Deep Sea Drilling Project Legs 69 and 70. Initial Reports of the
Time Scales and Global Stratigraphic Correlations (SEPM Special Publication), Deep Sea Drilling Project 69: 711–720.
vol. 54, pp. 51–63. Tulsa, OK: SEPM. Furuta T (1993) Magnetic properties and ferromagnetic mineralogy of oceanic basalts.
Channell JET, Galeotti S, Martin EE, Billups K, Scher HD, and Stoner JS (2003) Eocene Geophysical Journal International 113: 95–114.
to Miocene magnetostratigraphy, biostratigraphy, and chemostratigraphy at ODP Furuta T and Levi S (1983) Basement paleomagnetism of Hole 504B. Initial Reports of
Site 1090 (sub-Antarctic South Atlantic). Geological Society of America Bulletin the Deep Sea Drilling Project 69: 697–703.
115: 607–623. Gallet Y and Hulot G (1997) Stationary and nonstationary behaviour within the
Channell JET, Lowrie W, Pialli P, and Venturi F (1984) Jurassic magnetic stratigraphy geomagnetic polarity time scale. Geophysical Research Letters 24: 1875–1878.
from Umbrian (Italian) land sections. Earth and Planetary Science Letters Garces M and Gee JS (2007) Paleomagnetic evidence of large footwall rotations
68: 309–325. associated with low-angle faults at the Mid-Atlantic Ridge. Geology 35: 279–282.
Christeson GL, Purdy GM, and Fryer GJ (1992) Structure of young upper crust at the Gee JS and Cande SC (2002) A surface-towed vector magnetometer. Geophysical
East Pacific Rise near 9°300 N. Geophysical Research Letters 19: 1045–1048. Research Letters 29(14). http://dx.doi.org/10.1029/2002GL015245.
Clement BM (2004) Dependence of the duration of geomagnetic polarity reversals on Gee JS, Cande SC, Hildebrand JA, Donnelly K, and Parker RL (2000) Geomagnetic
site latitude. Nature 428: 637–640. intensity variations over the past 780 kyr obtained from near-seafloor magnetic
Cogné JP, Francheteau J, Courtillot V, et al. (1995) Large rotation of the Easter anomalies. Nature 408: 827–832.
microplate as evidenced by oriented paleomagnetic samples from the ocean floor. Gee J and Kent DV (1994) Variations in layer 2A thickness and the origin of the central
Earth and Planetary Science Letters 136: 213–222. anomaly magnetic high. Geophysical Research Letters 21: 297–300.
Cogné JP and Humler E (2006) Trends and rhythms in global seafloor generation rate. Gee J and Kent DV (1997) Magnetization of axial lavas from the southern East Pacific
Geochemistry, Geophysics, Geosystems 7: 17. Rise (14°–23°S): Geochemical controls on magnetic properties. Journal of
Cox A (1968) Lengths of geomagnetic polarity intervals. Journal of Geophysical Geophysical Research 102: 24873–24886.
Research 73: 3247–3260. Gee J and Kent DV (1998) Magnetic telechemistry and magmatic segmentation on the
Cox A, Doell RR, and Dalrymple GB (1964) Reversals of the Earth’s magnetic field. southern East Pacific Rise. Earth and Planetary Science Letters 164: 379–385.
Science 144: 1537–1543. Gee J and Kent DV (1999) Calibration of magnetic granulometric trends in oceanic
Cronin M, Tauxe L, Constable C, Selkin P, and Pick T (2001) Noise in the quiet zone. basalts. Earth and Planetary Science Letters 170: 377–390.
Earth and Planetary Science Letters 190: 13–30. Gee JS, Lawrence RM, and Hurst SD (1997) Remanence characteristics of gabbros from
Day R, Osterhoudt M, and Bleil U (1983) Rock magnetism of igneous rocks from Deep the MARK area: Implications for crustal magnetization. Proceedings of the Ocean
Sea Drilling Project Sites 482, 483, and 485. Initial Reports of the Deep Sea Drilling Drilling Program, Scientific Results 153: 429–436.
Project 65: 727–736. Gee J and Meurer WP (2002) Slow cooling of middle and lower oceanic crust inferred
Detrick RS, Harding AJ, Kent GM, Orcutt JA, Mutter JC, and Buhl P (1993) Seismic from multicomponent magnetizations of gabbroic rocks from the Mid-Atlantic Ridge
structure of the southern East Pacific Rise. Science 259: 499–503. south of the Kane fracture zone (MARK) area. Journal of Geophysical Research
Detrick R, Honnorez J, Bryan WB, and Juteau T (1988) Proceedings of the Ocean 107: 3–18.
Drilling Program, Initial Reports, vol. 106/109: p. 249. College Station, TX: Ocean Gee J, Natland JH, Hurst SD, and Nilsson K (1992) Magnetic properties of ocean crust
Drilling Program. samples from Hess Deep: Implications for marine magnetic anomalies. Eos,
Dick HJB, Natland JH, Alt JC, et al. (2000) A long in situ section of the lower ocean Transactions American Geophysical Union 73: 490.
crust: Results of ODP Leg 176 drilling at the Southwest Indian Ridge. Earth and Gee J, Schneider DA, and Kent DV (1996) Marine magnetic anomalies as recorders of
Planetary Science Letters 179: 31–51. geomagnetic intensity variations. Earth and Planetary Science Letters
Dick HJB, Natland JH, and Miller DJ (1999) Proceedings of the Ocean Drilling Program, 144: 327–335.
Initial Reports, vol. 176. College Station, TX: Ocean Drilling Program. Gillis K, Mevel C, and Allan J (1993) Proceedings of the Ocean Drilling Program, Initial
Dick HJB, Schouten H, Meyer PS, et al. (1991) Tectonic evolution of the Atlantis II Reports, vol. 147, p. 366. College Station, TX: Ocean Drilling Program.
Fracture Zone. Proceedings of the Ocean Drilling Program, Scientific Results Gradstein FM, Ogg JG, and Smith AG (2004) A Geologic Time Scale 2004, p. 589.
118: 359–398. Cambridge: Cambridge University Press.
Doubrovine PV and Tarduno JA (2004) Self-reversed magnetization carried by Grommé CS, Mankinen EA, Marshall M, and Coe RS (1979) Geomagnetic
titanomaghemite in oceanic basalts. Earth and Planetary Science Letters paleointensities by the Thelliers’ method from submarine pillow basalts: Effects of
222: 959–969. seafloor weathering. Journal of Geophysical Research 84: 3553–3575.
Dyment J and Arkani-Hamed J (1995) Spreading-rate-dependent magnetization of the Guyodo Y and Valet J-P (1999) Global changes in intensity of the Earth’s magnetic field
oceanic lithosphere inferred from the anomalous skewness of marine magnetic during the past 800 kyr. Nature 399: 249–252.
anomalies. Geophysical Journal International 121: 789–804. Hall JM and Muzzatti A (1999) Delayed magnetization of the deeper kilometer of oceanic
Dyment J, Arkani-Hamed J, and Ghods A (1997) Contribution of serpentinized crust at Ocean Drilling Project Site 504. Journal of Geophysical Research
ultramafics to marine magnetic anomalies at slow and intermediate spreading 104: 12843–12851.
centers: Insights from the shape of the anomalies. Geophysical Journal International Handschumacher DW, Sager WW, Hilde TWC, and Bracey DR (1988) Pre-Cretaceous
129: 691–701. tectonic evolution of the Pacific plate and extension of the geomagnetic polarity
Evans HF and Channell JET (2003) Upper Miocene magnetic stratigraphy at ODP Site reversal time scale with implications for the origin of the Jurassic ‘Quiet Zone’.
1092 (sub-Antarctic South Atlantic): Recognition of ‘cryptochrons’ in C5n.2n. Tectonophysics 155: 365–380.
Geophysical Journal International 153: 483–496. Harding AJ, Kent GM, and Orcutt JA (1993) A multichannel seismic investigation of
Evans HF, Westerhold T, and Channell JET (2004) ODP Site 1092: Revised composite upper crustal structure at 9°N on the East Pacific Rise: Implications for crustal
depth section has implications for Upper Miocene ‘cryptochrons’. Geophysical accretion. Journal of Geophysical Research 98: 13925–13944.
Journal International 156: 195–199. Harrison CGA (1987) Marine magnetic anomalies—The origin of the stripes. Annual
Facey D, Housden J, and O’Reilly W (1985) A magneto-petrological study of rocks from Review of Earth and Planetary Science Letters 15: 505–543.
Leg 83, Hole 504B, Deep Sea Drilling Project. Initial Reports of the Deep Sea Hartl P, Tauxe L, and Constable C (1993) Early Oligocene geomagnetic field behavior
Drilling Project 83: 339–346. from Deep Sea Drilling Project Site 522. Journal of Geophysical Research
Fornari DJ, Haymon RM, Perfit MR, Gregg TKP, and Edwards MH (1998) Axial summit 98: 19649–19665.
trough of the East Pacific Rise 9°–10°N: Geological characteristics and evolution of Heirtzler JR, Dickson GO, Herron EM, Pitman WC, and Le Pichon X (1968) Marine
the axial zone on fast spreading mid-ocean ridges. Journal of Geophysical Research magnetic anomalies, geomagnetic field reversals, and motions of the ocean floor
103: 9827–9855. and continents. Journal of Geophysical Research 73: 2119–2136.
Francheteau J, Armijo R, Cheminee JL, Hekinian R, Lonsdale P, and Blum N (1992) Helsley CE and Steiner MC (1969) Evidence for long intervals of normal polarity during
Dyke complex of the East Pacific Rise exposed in the walls of Hess Deep and the the Cretaceous period. Earth and Planetary Science Letters 5: 325–332.
structure of the upper oceanic crust. Earth and Planetary Science Letters Hilgen FJ (1991) Extension of the astronomically calibrated (polarity) time scale to the
111: 109–121. Miocene/Pliocene boundary. Earth and Planetary Science Letters 107: 349–368.
Hooft EEE, Schouten H, and Detrick RS (1996) Constraining crustal emplacement Kinoshita H, Furuta T, and Kawahata H (1985) Magnetic properties and alteration in
processes from the variation in seismic layer 2A thickness at the East Pacific Rise. basalt, Hole 504B, Deep Sea Drilling Project Leg 83. Initial Reports of the Deep Sea
Earth and Planetary Science Letters 142: 289–309. Drilling Project 83: 331–338.
Horner-Johnson BC and Gordon RG (2003) Equatorial Pacific magnetic anomalies Klitgord KD (1976) Sea-floor spreading: The central anomaly magnetization high. Earth
identified from vector aeromagnetic data. Geophysical Journal International and Planetary Science Letters 29: 201–209.
155: 547–556. Korenaga J (1995) Comprehensive analysis of marine magnetic vector anomalies.
Hurst SD, Karson JA, and Verosub KL (1994) Paleomagnetism of tilted dikes in fast Journal of Geophysical Research 100: 365–378.
spread oceanic crust exposed in the Hess Deep Rift: Implications for spreading and Korte M and Constable CG (2005) The geomagnetic dipole moment over the last 7000
rift propagation. Tectonics 13: 789–802. years – New results from a global model. Earth and Planetary Science Letters
Hussenoeder SA, Tivey MA, Schouten H, and Searle RC (1996) Near-bottom magnetic 236: 348–358.
survey of the Mid-Atlantic Ridge axis, 24°–24°400 N: Implications for crustal Krasa D, Shcherbakov VP, Kunzmann T, and Petersen N (2005) Self-reversal of
accretion at slow spreading ridges. Journal of Geophysical Research remanent magnetization in basalts due to partially oxidized titanomagnetites.
10: 22051–22069. Geophysical Journal International 162: 115–136.
Irving E (1970) The mid-Atlantic Ridge at 45°N. XIV. Oxidation and magnetic Krijgsman W and Kent DV (2004) Non-uniform occurrence of short-term polarity
properties of basalt; review and discussion. Canadian Journal of Earth Sciences fluctuations in the geomagnetic field? New results from Middle to Late Miocene
7: 1528–1538. sediments of the North Atlantic (DSDP Site 608). In: Channell JET, Kent DV,
Irving E and Pullaiah G (1976) Reversals of the geomagnetic field, magnetostratigraphy, Lowrie W, and Meert J (eds.) AGU Geophysical Monograph 145: Timescales of the
and relative magnitude of paleosecular variation in the Phanerozoic. Earth Science Paleomagnetic Field, pp. 161–174. Washington, DC: American Geophysical Union.
Reviews 12: 35–64. Kriner KA, Pockalny RA, and Larson RL (2006) Bathymetric gradients of lineated
Jackson A, Jonkers ART, and Walker MR (2000) Four centuries of geomagnetic secular abyssal hills: Inferring seafloor spreading vectors and a new model for hills formed
variation from historical records. Philosophical Transactions of the Royal Society of at ultra-fast rates. Earth and Planetary Science Letters 242: 98–110.
London A 358: 957–990. Lancelot YP, Larson RL, and Fisher A (1990) Proceedings of the Ocean Drilling
John BE, Foster DA, Murphy JM, et al. (2004) Determining the cooling history of in situ Program, Initial Reports, vol. 129. College Station, TX: Texas A&M University,
lower oceanic crust – Atlantis Bank, SW Indian Ridge. Earth and Planetary Science Ocean Drilling Program, p. 488.
Letters 222: 145–160. Lanci L and Lowrie W (1997) Magnetostratigraphic evidence that ‘tiny wiggles’ in the
Johnson HP (1978) Rock magnetic properties of igneous rock samples – Leg 45. Initial oceanic magnetic anomaly record represent geomagnetic paleointensity variations.
Reports of the Deep Sea Drilling Project 45: 397–406. Earth and Planetary Science Letters 148: 581–592.
Johnson HP and Atwater T (1977) Magnetic study of basalts from the Mid-Atlantic Lanci L, Pares JP, Channell JET, and Kent DV (2004) Miocene magnetostratigraphy
Ridge, lat 37°N. Geological Society of America Bulletin 88: 637–647. from equatorial Pacific sediments (ODP Site 1218, Leg 199). Earth and Planetary
Johnson HP, Kent DV, Tivey MA, Gee JS, Larson RL, and Embley RW (1997) Science Letters 226: 207–224.
Conference on the magnetization of the oceanic crust steers future research. Eos, Lanci L, Pares JP, Channell JET, and Kent DV (2005) Oligocene magnetostratigraphy
Transactions American Geophysical Union 78: 199–200. from equatorial Pacific sediments (ODP Sites 1218 and 1219, Leg 199). Earth and
Johnson HP and Pariso JE (1993) Variations in oceanic crustal magnetization: Planetary Science Letters 237: 617–634.
Systematic changes in the last 160 million years. Journal of Geophysical Research Larson RL (1977) Early Cretaceous breakup of Gondwanaland off western Australia.
98: 435–445. Geology 5: 57–60.
Johnson HP and Salem BL (1994) Magnetic properties of dikes from the oceanic upper Larson RL and Chase CG (1972) Late Mesozoic evolution of the western Pacific Ocean.
crustal section. Journal of Geophysical Research 99: 21733–21740. Geological Society of America Bulletin 83: 3627–3644.
Johnson HP and Tivey MA (1995) Magnetic properties of zero-age oceanic crust; a new Larson RL and Hilde TWC (1975) A revised time scale of magnetic reversals for the Early
submarine lava flow on the Juan de Fuca ridge. Geophysical Research Letters Cretaceous and Late Jurassic. Journal of Geophysical Research 80: 2586–2594.
22: 175–178. Larson RL and Pitman WC (1972) Worldwide correlation of Mesozoic magnetic
Juarez MT, Osete ML, Melendez G, Langereis CG, and Zijderveld JDA (1994) Oxfordian anomalies, and its implications. Geological Society of America Bulletin
magnetostratigraphy of the Aguilon and Tosos sections (Iberian Range, Spain) and 83: 3645–3662.
evidence of a pre-Oligocene overprint. Earth and Planetary Science Letters Larson RL and Spiess FN (1969) The East Pacific Rise crest: A near-bottom geophysical
85: 195–211. profile. Science 163: 68–71.
Juarez MT, Tauxe L, Gee JS, and Pick T (1998) The intensity of the Earth’s magnetic Lawrence RM, Gee JS, and Karson JA (2002) Magnetic anisotropy of serpentinized
field over the past 160 million years. Nature 394: 878–881. peridotites from the MARK area: Implications for the orientation of mesoscopic
Juster TC, Grove TL, and Perfit MR (1989) Experimental constraints on the generation structures and major fault zones. Journal of Geophysical Research 107: 14.
of FeTi basalts, andesites, and rhyodacites at the Galapagos spreading center, 85°W Levi S (1980) Paleomagnetism and some magnetic properties of basalts from the
and 95°W. Journal of Geophysical Research 94: 9251–9274. Bermuda Triangle. Initial Reports of the Deep Sea Drilling Project
Karson JA, Klein EM, Hurst SD, et al. (2002) Structure of uppermost fast-spread oceanic 51,52,53: 1363–1378.
crust exposed at the Hess Deep Rift: Implications for subaxial processes at the East Levi S, Audunsson H, Duncan RA, Kristjansson L, Gillot P-Y, and Jakobsson SP (1990)
Pacific Rise. Geochemistry, Geophysics, Geosystems 3(1): 4. http://dx.doi.org/ Late Pleistocene geomagnetic excursion in Icelandic lavas: Confirmation of the
10.129/2001GC000155. Laschamp excursion. Earth and Planetary Science Letters 96: 443–457.
Kelemen PB, Kikawa E, and Miller DJ (2004) Proceedings of the Ocean Drilling Lewis BTR, Snydsman WE, McClain JS, Holmes ML, and Lister CRB (1983) Site survey
Program, Initial Reports, vol. 209. College Station, TX: Ocean Drilling Program. results at the mouth of the Gulf of California, Leg 65, Deep Sea Drilling Project.
Kelso PR, Banerjee SK, and Worm HU (1991) The effect of low-temperature Initial Reports of the Deep Sea Drilling Project 65: 309–317.
hydrothermal alteration on the remanent magnetization of synthetic Lourens LJ, Hilgen FJ, Shackleton NJ, Laskar J, and Wilson D (2004) The Neogene
titanomagnetites; a case for acquisition of chemical remanent magnetization. period. In: Gradstein FM, Ogg JG, and Smith AG (eds.) A Geologic Time Scale
Journal of Geophysical Research 96: 19545–19553. 2004, pp. 409–440. Cambridge: Cambridge University Press.
Kent DV and Gee J (1994) Grain size-dependent alteration and the magnetization of Lowrie W (1977) Intensity and direction of magnetization in oceanic basalts. Journal of
oceanic basalts. Science 265: 1561–1563. the Geological Society of London 133: 61–82.
Kent DV and Gee J (1996) Magnetic alteration of zero-age oceanic basalt. Geology Lowrie W and Alvarez W (1981) One hundred million years of geomagnetic polarity
24: 703–706. history. Geology 9: 392–397.
Kent DV, Honnorez BM, Opdyke ND, and Fox PJ (1978) Magnetic properties of dredged Lowrie W and Kent DV (2004) Geomagnetic polarity timescales and reversal frequency
oceanic gabbros and the source of marine magnetic anomalies. Geophysical Journal regimes. In: Channell JET, Kent DV, Lowrie W, and Meert J (eds.) AGU Geophysical
of the Royal Astronomical Society 55: 513–537. Monograph 145: Timescales of the Paleomagnetic Field, pp. 117–129. Washington,
Kidd RGW (1977) The nature and shape of the sources of marine magnetic anomalies. DC: American Geophysical Union.
Earth and Planetary Science Letters 33: 310–320. Macdonald KC (1977) Near-bottom magnetic anomalies, asymmetric spreading,
Kikawa E and Ozawa K (1992) Contribution of oceanic gabbros to sea-floor spreading oblique spreading, and tectonics of the Mid-Atlantic Ridge near lat 37°N. Geological
magnetic anomalies. Science 258: 796–799. Society of America Bulletin 88: 541–555.
Kikawa E and Pariso JE (1991) Magnetic properties of gabbros from Hole 735B, Macdonald KC, Miller SP, Luyendyk BP, Atwater T, and Shure L (1983) Investigation of
Southwest Indian Ridge. Proceedings of the Ocean Drilling Program, Scientific a Vine–Matthews magnetic lineation from a submersible: The source and character
Results 118: 285–307. of marine magnetic anomalies. Journal of Geophysical Research 88: 3403–3418.
Marshall M and Cox A (1971) Magnetism of pillow basalts and their petrology. Pariso JE, Stokking LB, and Allerton S (1995) Rock magnetism and magnetic
Geological Society of America Bulletin 82: 537–552. mineralogy of a 1-km section of sheeted dikes, Hole 504B. Proceedings of the
Marshall M and Cox A (1972) Magnetic changes in pillow basalt due to sea-floor Ocean Drilling Program, Scientific Results 137(140): 253–262.
weathering. Journal of Geophysical Research 77: 6459–6469. Parker RL and O’Brien MS (1997) Spectral analysis of vector magnetic field profiles.
Matzka J, Krasa D, Kunzmann T, Schult A, and Petersen N (2003) Magnetic state of Journal of Geophysical Research 102: 24815–24824.
10–40 Ma old ocean basalts and its implications for natural remanent Pechersky DM, Tikhonov LV, and Pertsev NN (1983a) Magnetic properties of basalts,
magnetization. Earth and Planetary Science Letters 206: 541–553. Deep Sea Drilling Project Legs 69 and 70. Initial Reports of the Deep Sea Drilling
McElhinny MW and Larson RL (2003) Jurassic dipole low defined from land and sea Project 69: 705–710.
data. Eos, Transactions of the American Geophysical Union 84: 362–366. Pechersky DM, Tikhonov LV, and Zolotarev BP (1983b) Rock magnetism and
McFadden PL and Merrill RT (2000) Evolution of the geomagnetic reversal rate since paleomagnetism of basalts drilled during Deep Sea Drilling Project Leg 65. Initial
160 Ma: Is the process continuous. Journal of Geophysical Research Reports of the Deep Sea Drilling Project 65: 717–726.
105: 28455–28460. Perfit MR and Chadwick WW (1998) Magmatism at mid-ocean ridges: Constraints from
McFadden PL, Merrill RT, and McElhinny MW (1988) Dipole/quadrupole family volcanological and geochemical investigations. In: Buck WR (ed.). Faulting and
modeling of paleosecular variation. Journal of Geophysical Research Magmatism at Mid-Ocean Ridges Geophysical Monograph Series, vol. 106,
93: 11583–11588. pp. 59–115. Washington, DC: AGU.
Mejia V, Opdyke ND, and Perfit MR (1996) Paleomagnetic field intensity recorded in Perfit MR, Fornari DJ, Smith MC, Bender JF, Langmuir CH, and Haymon RM (1994)
submarine basaltic glass from the East Pacific Rise, the last 69 KA. Geophysical Small-scale spatial and temporal variations in mid-ocean ridge crest magmatic
Research Letters 23: 475–478. processes. Geology 22: 375–379.
Merrill RT, McElhinny MW, and McFadden PL (1996) The Magnetic Field of the Earth: Perram LJ, Macdonald KC, and Miller SP (1990) Deep-tow magnetics near 20°S on the
Paleomagnetism, the Core, and the Deep Mantle, p. 531. San Diego: Academic East Pacific Rise: A study of short wavelength anomalies at a very fast spreading
Press. center. Marine Geophysical Researches 12: 235–245.
Miller DJ and Christensen NI (1997) Seismic velocities of lower crustal and upper Petronotis KE, Gordon RG, and Acton GD (1992) Determining palaeomagnetic poles
mantle rocks from the slow-spreading Mid-Atlantic Ridge, south of the Kane and anomalous skewness from marine magnetic anomaly skewness data from a
Transform Zone (MARK). Proceedings of the Ocean Drilling Program, Scientific single plate. Geophysical Journal International 109: 209–224.
Results 153: 437–454. Pettigrew TJ, Casey JF, and Miller DJ (1999) Proceedings of the Ocean Drilling
Nakanishi M, Tamaki K, and Kobayashi K (1989) Mesozoic magnetic anomaly lineations Program, Initial Reports, vol. 179. College Station, TX: Ocean Drilling Program.
and seafloor spreading history of the Northwestern Pacific. Journal of Geophysical Pick T and Tauxe L (1993) Holocene paleointensities: Thellier experiments on
Research 94: 15437–15462. submarine basaltic glass from the East Pacific Rise. Journal of Geophysical
Nakanishi M, Tamaki K, and Kobayashi K (1992) A new Mesozoic isochron chart of the Research 98: 17949–17964.
Northwestern Pacific Ocean: Paleomagnetic and tectonic implications. Geophysical Pick T and Tauxe L (1994) Characteristics of magnetite in submarine basaltic glass.
Research Letters 19: 693–696. Geophysical Journal International 119: 116–128.
Nakanishi M and Winterer EL (1998) Tectonic history of the Pacific–Farallon–Phoenix Pitman WC and Heirtzler JR (1966) Magnetic anomalies over the Pacific–Antarctic
triple junction from Late Jurassic to Early Cretaceous: An abandoned Mesozoic ridge. Science 154: 1164–1171.
spreading system in the Central Pacific Basin. Journal of Geophysical Research Pitman WC, Herron EM, and Heirtzler JR (1968) Magnetic anomalies in the Pacific and
103: 12453–12468. sea floor spreading. Journal of Geophysical Research 73: 2069–2085.
Natland JH (2002) Magnetic susceptibility as an index of the lithology and composition Plank T, Ludden JN, and Escutia C (2000) Proceedings of the Ocean Drilling Program,
of gabbros, ODP Leg 176, Hole 735B, Southwest Indian Ridge. Proceedings of the Initial Reports, vol. 185. College Station, TX: Texas A&M University, Ocean Drilling
Ocean Drilling Program, Scientific Results 176. Program, http://www.odp.tamu.edu/publications/185IR/185TOC.HTM (accessed
Nazarova KA (1994) Serpentinized peridotites as a possible source for oceanic magnetic Jan 2007).
anomalies. Marine Geophysical Researches 16: 455–462. Pockalny RA, Larson RL, Viso RF, and Abrams LJ (2002) Bathymetry and gravity data
Nazarova KA, Wasilewski PJ, and Dick HJB (2000) Magnetic study of serpentinized across a mid-Cretaceous triple junction in the southwest Pacific basin. Geophysical
harzburgites from the Islas Orcadas Fracture Zone. Marine Geophysical Researches Research Letters 29: 1007.
21: 475–488. Pockalny RA, Smith A, and Gente P (1995) Spatial and temporal variability of crustal
O’Reilly W (1984) Rock and Mineral Magnetism. New York: Chapman and Hall. magnetization of a slowly spreading ridge: Mid-Atlantic Ridge (20°–24°N). Marine
Ogg JG and Lowrie W (1986) Magnetostratigraphy of the Jurassic/Cretaceous Geophysical Research 17: 301–320.
boundary. Geology 14: 547–550. Prévot M, Derder ME-M, McWilliams M, and Thompson J (1990) Intensity of the Earth’s
Oufi O, Cannat M, and Horen H (2002) Magnetic properties of variably serpentinized magnetic field: Evidence for a Mesozoic dipole low. Earth and Planetary Science
abyssal peridotites. Journal of Geophysical Research 107(B5). http://dx.doi.org/ Letters 97: 129–139.
10.1029/2001JB000549. Raymond CA and LaBrecque JL (1987) Magnetization of the oceanic crust:
Oxburgh ER and Turcotte DL (1969) Increased estimate for heat flow at oceanic ridges. Thermoremanent magnetization or chemical remanent magnetization? Journal of
Nature 223: 1354–1355. Geophysical Research 92: 8077–8088.
€
Ozdemir O and O’Reilly W (1981) High-temperature hysteresis and other magnetic Renne PR, Scott GR, Glen JMG, and Feinberg JM (2002) Oriented inclusions of
properties of synthetic monodomain titanomagnetites. Physics of the Earth and magnetite in clinopyroxene: Source of stable remanent magnetization in gabbros of
Planetary Interiors 25: 406–418. the Messum Complex, Namibia. Geochemistry, Geophysics, Geosystems 3(12):
Pariso JE and Johnson HP (1989) Magnetic properties and oxide petrography of the 1079. http://dx.doi.org/10.1029/2002GC000319.
sheeted dike complex in Hole 504B. Proceedings of the Ocean Drilling Program, Roberts AP and Lewin-Harris JC (2000) Marine magnetic anomalies: Evidence that ‘tiny
Scientific Results 111: 159–167. wiggles’ represent short-period geomagnetic polarity intervals. Earth and Planetary
Pariso JE and Johnson HP (1991) Alteration processes at Deep Sea Drilling Project Science Letters 183: 375–388.
Ocean Drilling Program Hole 504B at the Costa Rica Rift – Implications Robinson PT, Von Herzen RP, and Adamson AC (1989) Proceedings of the Ocean
for magnetization of oceanic crust. Journal of Geophysical Research Drilling Program, Initial Reports, vol. 118, p. 826. College Station, TX: Ocean
96: 11703–11722. Drilling Program.
Pariso JE and Johnson HP (1993a) Do Layer 3 rocks make a significant contribution to Roeser HA, Steiner C, Schreckenberger B, and Block M (2002) Structural development
marine magnetic anomalies? In situ magnetization of gabbros at Ocean Drilling of the Jurassic Magnetic Quiet Zone off Morocco and identification of Middle
Program Hole 735B. Journal of Geophysical Research 98: 16033–16052. Jurassic magnetic lineations. Journal of Geophysical Research 107(10). http://dx.
Pariso JE and Johnson HP (1993b) Do lower crustal rocks record reversals of the doi.org/10.1029/2000JB000094.
Earth’s magnetic field? Magnetic petrology of oceanic gabbros from Ocean Drilling Roest WR, Arkani-Hamed J, and Verhoef J (1992) The seafloor spreading rate
Program Hole 735B. Journal of Geophysical Research 98: 16013–16032. dependence of the anomalous skewness of marine magnetic anomalies.
Pariso JE, Kelso P, and Richter C (1996a) Paleomagnetism and rock magnetic Geophysical Journal International 109: 653–669.
properties of gabbro from Hole 894G, Hess Deep. Proceedings of the Ocean Drilling Roperch P, Bonhommet N, and Levi S (1988) Paleointensity of the earth’s magnetic field
Program, Scientific Results 147: 373–381. during the Laschamp excursion and its geomagnetic implications. Earth and
Pariso JE, Rommevaux C, and Sempere J-C (1996b) Three-dimensional inversion Planetary Science Letters 88: 209–219.
of marine magnetic anomalies: Implications for crustal accretion along the Ryall PJ and Ade-Hall JM (1975) Radial variation of magnetic properties in submarine
Mid-Atlantic Ridge (28°–31°300 N). Marine Geophysical Researches 18: 85–101. pillow basalt. Canadian Journal of Earth Sciences 12: 1959–1969.
Sager WW, Weiss MA, Tivey MA, and Johnson HP (1998) Geomagnetic polarity Speranza F, Satolli S, Mattioli E, and Calamita F (2005) Magnetic stratigraphy of
reversal model of deep-tow profiles from the Pacific Jurassic ‘Quiet Zone’. Journal of Kimmeridgian–Aptian sections from Umbria–Marche (Italy): New details on the
Geophysical Research 103: 5269–5286. M polarity sequence. Journal of Geophysical Research 110: B12109.
Sayanagi K and Tamaki K (1992) Long-term variations in magnetization intensity with Stein CA and Stein S (1994) Comparison of plate and asthenospheric flow models for
crustal age in the Northeast Pacific, Atlantic, and Southeast Indian Oceans. the thermal evolution of oceanic lithosphere. Geophysical Research Letters
Geophysical Research Letters 12: 2369–2372. 21: 709–712.
Schneider DA (1988) An estimate of the long-term non-dipole field from marine Steiner MB, Ogg JG, Melendez G, and Sequeiros L (1985) Jurassic
magnetic anomalies. Geophysical Research Letters 15: 1105–1108. magnetostratigraphy, 2. Middle–Late Oxfordian of Aguilon, Iberian cordillera,
Schneider DA (1995) Paleomagnetism of some Leg 138 sediments: Detailing Miocene northern Spain. Earth and Planetary Science Letters 76: 151–166.
magnetostratigraphy. Proceedings of the Ocean Drilling Program, Scientific Results Swift BA and Johnson HP (1984) Magnetic properties of the Bay of Islands ophiolite
138: 59–72. suite and implications for the magnetization of oceanic crust. Journal of Geophysical
Schouten H (2002) Paleomagnetic inclinations in DSDP Hole 417D reconsidered: Research 89: 3291–3308.
Secular variation or variable tilting? Geophysical Research Letters 29(7). http://dx. Talwani M, Windisch CC, and Langseth MG (1971) Reykjanes Ridge crest: A detailed
doi.org/10.1029/2001GL013581. geophysical study. Journal of Geophysical Research 76: 473–517.
Schouten H and Cande SC (1976) Paleomagnetic poles from marine magnetic Tarduno JA, Cottrell RD, and Smirnov AV (2001) High geomagnetic intensity during the
anomalies. Geophysical Journal of the Royal Astronomical Society 44: 567–575. Mid-Cretaceous from Thellier analyses of single plagioclase crystals. Science
Schouten H and Denham CR (2000) Comparison of volcanic construction in the 291: 1779–1782.
Troodos ophiolite and oceanic crust using paleomagnetic inclinations from Cyprus Tauxe L and Hartl P (1997) 11 million years of Oligocene geomagnetic field behaviour.
Crustal Study Project (CCSP) CY-1 and CY-1A and Ocean Drilling Program (ODP) Geophysical Journal International 128: 217–229.
504B drill cores. In: Dilek Y, Moores EM, Elthon D, and Nicolas A (eds.). Ophiolites Tauxe L and Kent DV (2004) A simplified statistical model for the geomagnetic field and
and Oceanic Crust: New Insights from Field Studies and the Ocean Drilling Program the detection of shallow bias in paleomagnetic inclinations: Was the ancient
Geological Society of America Special Paper, vol. 349, pp. 181–194. Boulder, CO: magnetic field dipolar. In: Channell JET, Kent DV, Lowrie W, and Meert J (eds.)
Geological Society of America. Geophysical Monograph 145: Timescales of the Paleomagnetic Field, pp. 101–115.
Schouten H and McCamy K (1972) Filtering marine magnetic anomalies. Journal of Washington, DC: AGU.
Geophysical Research 77: 7089–7099. Tauxe L and Staudigel H (2004) Strength of the geomagnetic field in the Cretaceous
Schouten H, Tivey MA, Fornari DJ, and Cochran JR (1999) Central anomaly Normal Superchron: New data from submarine basaltic glass of the Troodos
magnetization high: Constraints on the volcanic construction and architecture of Ophiolite. Geochemistry, Geophysics, Geosystems 5(2). http://dx.doi.org/
seismic layer 2A at a fast-spreading mid-ocean ridge, the EPR at 9°30–50 N. Earth 10.1029/2003GC000635.
and Planetary Science Letters 169: 37–50. Tivey M (1996) Vertical magnetic structure of ocean crust determined from near-bottom
Sclater JG and Francheteau J (1970) The implications of terrestrial heat flow magnetic field measurements. Journal of Geophysical Research 101: 20275–20296.
observations on current tectonic and geochemical models of the crust and upper Tivey MA and Johnson HP (1987) The Central Anomaly magnetic high: Implications for
mantle of the Earth. Geophysical Journal of the Royal Astronomical Society ocean crust construction and evolution. Journal of Geophysical Research
20: 509–542. 92: 12685–12694.
Seama N, Nogi Y, and Isezaki N (1993) A new method for precise determination of the Tivey MA and Johnson HP (2002) Crustal magnetization reveals subsurface structure of
position and strike of magnetic boundaries using vector data of the geomagnetic Juan de Fuca Ridge hydrothermal vent fields. Geology 30: 979–982.
anomaly field. Geophysical Journal International 113: 155–164. Tivey MA, Johnson HP, Bradley A, and Yoerger D (1998a) Thickness of a submarine
Selkin PA, Gee JS, Tauxe L, Meurer WP, and Newell AJ (2000) The effect of remanence lava flow determined from near-bottom magnetic field mapping by autonomous
anisotropy on paleointensity estimates: A case study from the Archean Stillwater underwater vehicle. Geophysical Research Letters 25: 805–808.
Complex. Earth and Planetary Science Letters 183: 403–416. Tivey M, Johnson HP, Fleutelot C, et al. (1998b) Direct measurement of magnetic
Selkin PA and Tauxe L (2000) Long-term variations in paleointensity. Philosophical reversal boundaries in a cross-section of oceanic crust. Geophysical Research
Transactions of the Royal Society of London A 358: 1065–1088. Letters 25: 3631–3634.
Sempere J-C, Macdonald KC, Miller SP, and Shure L (1987) Detailed study of the Tivey M, Kona PA, and Kleinrock MC (1996) Reduced crustal magnetization beneath
Brunhes/Matuyama reversal boundary on the East Pacific Rise at 19°300 S: relict hydrothermal mounds: TAG hydrothermal field, Mid-Atlantic Ridge, 26°N.
Implications for crustal emplacement processes at an ultra fast spreading center. Geophysical Research Letters 23: 3511–3514.
Marine Geophysical Researches 9: 1–23. Tivey M, Larson R, Schouten H, and Pockalny R (2005) Downhole magnetic
Shackleton NJ, Berger A, and Peltier WR (1990) An alternative astronomical calibration measurements of ODP Hole 801C: Implications for Pacific oceanic crust and
of the lower Pleistocene timescale based on ODP Site 677. Transactions of the Royal magnetic field behavior in the Middle Jurassic. Geochemistry, Geophysics,
Society of Edinburgh: Earth Sciences 81: 251–261. Geosystems 6: Q04008. http://dx.doi.org/10.1029/2004GC000754.
Shackleton NJ, Hall MA, Raffi I, Tauxe L, and Zachos J (2000) Astronomical calibration Tivey MA, Sager WW, Lee S-M, and Tominaga M (2006) Origin of the Pacific Jurassic
age for the Oligocene–Miocene boundary. Geology 28: 447–450. quiet zone. Geology 34: 789–792.
Shah AK, Cormier M-H, Ryan WBF, et al. (2003) Episodic dike swarms inferred from Tivey M, Schouten H, and Kleinrock MC (2003) A near-bottom magnetic survey of the
near-bottom magnetic anomaly maps at the southern East Pacific Rise. Journal of Mid-Atlantic Ridge axis at 26 N: Implications for the tectonic evolution of the TAG
Geophysical Research 108(B2): 2097. http://dx.doi.org/10.1029/2001JB000564. segment. Journal of Geophysical Research 108(B5): 2277. http://dx.doi.org/
Shau Y-H, Torii M, Horng C-S, and Peacor DR (2000) Subsolidus evolution and 10.1029/2002JB001967.
alteration of titanomagnetite in ocean ridge basalts from Deep Sea Drilling Project/ Tivey MA and Tucholke BE (1998) Magnetization of 0–29 Ma ocean crust on the
Ocean Drilling Program Hole 504B, Leg 83: Implications for the timing of Mid-Atlantic Ridge, 25°300 to 27°100 N. Journal of Geophysical Research
magnetization. Journal of Geophysical Research 105: 23635–23649. 103: 17807–17826.
Sinton JM and Detrick RS (1992) Mid-ocean ridge magma chambers. Journal of Toft PB, Arkani-Hamed J, and Haggerty SE (1990) The effects of serpentinization on
Geophysical Research 97: 197–216. density and magnetic susceptibility: A petrophysical model. Physics of the Earth and
Sinton JM, Smaglik SM, Mahoney JJ, and Macdonald KC (1991) Magmatic processes Planetary Interiors 65: 137–157.
at superfast spreading mid-ocean ridges: Glass compositional variations along the Varga RJ, Karson JA, and Gee JS (2004) Paleomagnetic constraints on deformation
East Pacific Rise 13°–23°S. Journal of Geophysical Research 96: 6133–6156. models for uppermost oceanic crust exposed at the Hess Deep Rift: Implications for
Smith GM (1990) The magnetic structure of the marine basement. Reviews in Aquatic axial processes at the East Pacific Rise. Journal of Geophysical Research 109:
Sciences 2: 205–227. B02104. http://dx.doi.org/10.1029/2003JB002486.
Smith GM and Banerjee SK (1985) Magnetic properties of basalts from Deep Sea Vera EE and Diebold JB (1994) Seismic imaging of oceanic layer 2A between 9°30’N
Drilling Project Leg 83: The origin of remanence and its relation to tectonic and and 10°N on the East Pacific Rise from two-ship wide-aperture profiles. Journal of
chemical evolution. Initial Reports of the Deep Sea Drilling Project 83: 347–357. Geophysical Research 99: 3031–3041.
Smith GM and Banerjee SK (1986) Magnetic structure of the upper kilometer of the Verosub KL and Moores EM (1981) Tectonic rotations in extensional regimes and their
marine crust at Deep Sea Drilling Project Hole 504B, Eastern Pacific Ocean. Journal paleomagnetic consequences for oceanic basalts. Journal of Geophysical Research
of Geophysical Research 91: 10337–10354. 86: 6335–6349.
Smith DK, Tivey MA, Schouten H, and Cann JR (1999) Locating the spreading axis Vine FJ (1966) Spreading of the ocean floor: New evidence. Science 154: 1405–1415.
along 80 km of the Mid-Atlantic Ridge south of the Atlantis Transform. Journal of Vine FJ and Matthews DH (1963) Magnetic anomalies over oceanic ridges. Nature
Geophysical Research 104: 7599–7612. 199: 947–949.
Vine FJ and Moores EM (1972) A model for the gross structure, petrology and magnetic properties of Banded Series rocks of the Stillwater Complex, Montana. Journal of
properties of oceanic crust. Geological Society of America Memoir 132: 195–205. Geophysical Research 102: 12139–12157.
Vogt PR and Johnson GL (1973) Magnetic telechemistry of oceanic crust? Nature Xu W, Peacor DR, Dollase W, Van der Voo R, and Beaubouef R (1997b) Transformation
245: 373–375. of titanomagnetite to titanomaghemite: A slow, two-step, oxidation-ordering process
Wallick BP and Steiner MB (1992) Paleomagnetic and rock magnetic properties of in nature. American Mineralogist 82: 1101–1110.
Jurassic quiet zone basalts, Hole 801C. Proceedings of the Ocean Drilling Program, Xu W, Peacor D, Van der Voo R, Dollase W, and Beaubouef R (1996) Modified lattice
Scientific Results 129: 455–470. parameter/Curie temperature diagrams for titanomagnetite/titanomaghemite within
Wang D, Van der Voo R, and Peacor DR (2005) Why is the remanent magnetic intensity the quadrilateral Fe3O4–Fe2TiO4–Fe2O3–Fe2TiO5. Geophysical Research Letters
of Cretaceous MORB so much higher than that of Mid- to Late Cenozoic MORB? 23(20): 2811–2814. http://dx.doi.org/10.1029/96GL01117.
Geosphere 1(3): 138–146. http://dx.doi.org/10.1130/GES00024.1. Xu W, Van der Voo R, Peacor DR, and Beaubouef RT (1997c) Alteration and dissolution
Wang D, Van der Voo R, and Peacor DR (2006) Low-temperature alteration and of fine-grained magnetite and its effects on magnetization of the ocean floor. Earth
magnetic changes of variably altered pillow basalts. Geophysical Journal and Planetary Science Letters 151: 279–288.
International 164: 25–35. Yamamoto M, Seama N, and Isezaki N (2005) Geomagnetic paleointensity over 1.2 Ma
Weissel JK and Hayes DE (1972) Magnetic anomalies in the Southeast Indian Ocean. from deep-tow vector magnetic data across the East Pacific Rise. Earth, Planets and
In: Hayes DE (ed.) Antarctic Research Series, 19: Antarctic Oceanology II: The Space 57: 465–470.
Australian–New Zealand Sector, pp. 165–196. Washington, DC: American Yang S, Odah H, and Shaw J (2000) Variations in the geomagnetic dipole moment over
Geophysical Union. the last 12 000 years. Geophysical Journal International 140: 158–162.
Wilson DS and Hey RN (1981) The Galapagos axial magnetic anomaly: Evidence for the Zhou W, Peacor DR, Van der Voo R, and Mansfield JF (1999a) Determination of lattice
Emperor event within the Brunhes and for a two-layer magnetic source. Geophysical parameter, oxidation state, and composition of individual titanomagnetite/
Research Letters 8: 167–188. titanomaghemite grains by transmission electron microscopy. Journal of
Wilson DS and Hey RN (1995) History of rift propagation and magnetization intensity for Geophysical Research 104: 17689–17702.
the Cocos–Nazca spreading center. Journal of Geophysical Research Zhou W, Van der Voo R, and Peacor DR (1999b) Preservation of pristine titanomagnetite
100: 10041–10056. in older ocean-floor basalts and its significance for paleointensity studies. Geology
Wilson DS, Teagle DAH, and Acton GD (2003) Proceedings of the Ocean Drilling 27: 1043–1046.
Program, Initial Reports, 2006. College Station, TX: Ocean Drilling Program. Zhou W, Van der Voo R, Peacor DR, Wang D, and Zhang Y (2001) Low-temperature
Wilson DS, Teagle DAH, Alt JC, et al. (2006) Drilling to gabbro in intact ocean crust. oxidation in MORB of titanomagnetite to titanomaghemite: A gradual process with
Science 312: 1016–1020. implications for marine magnetic anomaly amplitudes. Journal of Geophysical
Wittpenn NA, Harrison CGA, and Handschumacher DW (1989) Crustal magnetization in Research 106: 6409–6421.
the South Atlantic from inversion of magnetic anomalies. Journal of Geophysical Zhou W, Van der Voo R, Peacor DR, and Zhang Y (2000) Variable Ti-content and grain
Research 94: 15463–15480. size of titanomagnetite as a function of cooling rate in very young MORB. Earth and
Worm H-U (2001) Magnetic stability of oceanic gabbros from ODP Hole 735B. Earth Planetary Science Letters 179: 9–20.
and Planetary Science Letters 193: 287–302.
Worm H-U and Bach W (1996) Chemical remanent magnetization in oceanic sheeted
dikes. Geophysical Research Letters 23: 1123–1126.
Worm H-U, Bohm V, and Bosum W (1996) Implications for the sources of marine
magnetic anomalies derived from magnetic logging in Holes 504B and 896A. Relevant Websites
Proceedings of the Ocean Drilling Program, Scientific Results 148: 331–338.
Xu W, Geissman JW, Van der Voo R, and Peacor DR (1997a) Electron microscopy of Fe http://www.fugroairborne.com – Fugro Airborne Surveys is a versatile and
oxides and implications for the origin of magnetizations and rock magnetic technologically advanced airborne survey company.
5.13 Paleointensities
L Tauxe, University of California San Diego, San Diego, CA, USA
T Yamazaki, The University of Tokyo, Chiba, Japan

5.13.2 Theory of Paleointensity 462
5.13.3 Paleointensity with Thermal Remanence 462
5.13.3.1 Linearity Assumption 464
5.13.3.2 Alteration During Heating 466
5.13.3.2.1 KTT family of experiments 466
5.13.3.2.2 Shaw family of experiments 469
5.13.3.3 Methods that Minimize Alteration 470
5.13.3.3.1 Reduced number of heating steps 470
5.13.3.3.2 Use of controlled atmospheres to reduce alteration 471
5.13.3.3.3 Measurement at elevated temperature 472
5.13.3.3.4 Use of microwaves for thermal excitation 472
5.13.3.3.5 Using materials resistant to alteration 472
5.13.3.4 Methods for Non-Single Domain Remanences 473
5.13.3.5 Use of IRM Normalization 473
5.13.4 Paleointensity with Depositional Remanences 474
5.13.4.1 Physical Alignment of Magnetic Moments in Viscous Fluids 474
5.13.4.1.1 Nonflocculating environments 475
5.13.4.1.2 Flocculating environments 476
5.13.4.2 Postdepositional Processes 478
5.13.4.3 Note on Aeolian Deposits 480
5.13.4.4 Normalization 480
5.13.5 Remagnetization 481
5.13.5.1 Magnetic Viscosity 481
5.13.5.2 Chemical Alteration 481
5.13.6 Evaluating Paleointensity Data 481
5.13.6.1 Thermally Blocked Remanences 481
5.13.6.2 Depositional Remanences 481
5.13.7 Current State of the Paleointensity Data 482
5.13.7.1 Paleomagnetic Databases 482
5.13.7.2 Conversion to VDM 488
5.13.7.2.1 Absolute paleointensity data 488
5.13.7.2.2 Relative paleointensity data 488
5.13.8 Discussion 489
5.13.8.1 Selection Criteria from the PINT13 Database 489
5.13.8.2 What Is Long-Term Strength of the Geomagnetic Field? 491
5.13.8.3 Are There Any Trends? 491
5.13.8.3.1 Intensity versus polarity interval length 491
5.13.8.3.2 Source of scatter in the CNS 492
5.13.8.3.3 The oldest paleointensity records 492
5.13.8.3.4 The paleointensity ‘sawtooth’ 493
5.13.8.4 High-Resolution Temporal Correlation 495
5.13.8.4.1 Sediments 495
5.13.8.4.2 Ridge crest processes 497
5.13.8.4.3 Archaeomagnetic dating 497
5.13.8.5 Atmospheric Interaction 497
5.13.8.6 Frequency of Intensity Fluctuations and the Climatic Connection 499
Acknowledgments 502
References 502
Treatise on Geophysics, Second Edition http://dx.doi.org/10.1016/B978-0-444-53802-4.00107-X 461

462 Paleointensities
magnitudes of the ancient and laboratory fields, respectively.

5.13.1 Introduction
If alab and aanc are the same, we can divide the two equations
and rearrange terms to get
The geomagnetic field acts both as an umbrella, shielding us
from cosmic radiation, and as a window, offering one of the MNRM
Hanc ¼ Hlab
few glimpses of the inner workings of the Earth. Ancient Mlab
records of the geomagnetic field can inform us about the
In other words, if the laboratory remanence has the same
geodynamics of the early Earth and changes in boundary con-
proportionality constant with respect to the applied field as the
ditions through time. Thanks to its essentially dipolar nature,
ancient one, the remanences are linearly related to the applied
the geomagnetic field has acted as a guide, pointing to the axis
field, and the natural remanence (NRM) is composed solely of
of rotation, thereby providing latitudinal information for both
a single component, all one needs to do to get the ancient field
explorers and geologists. A complete understanding of the
is to measure MNRM, give the specimen a laboratory proxy
geomagnetic field requires not only a description of the direc-
remanence Mlab, and multiply the ratio by Hlab.
tion of field lines over the surface of the Earth but also infor-
In practice, paleointensity is not so simple. The remanence
mation about its strength. While directional information is
acquired in the laboratory may not have the same proportion-
relatively straightforward to obtain, intensity variations are
ality constant as the original remanence (e.g., the specimen has
much more difficult and are the subject of this chapter.
altered its capacity to acquire remanence or was acquired by a
In his treatise, De Magnete, published in 1600, William
mechanism not reproduced in the laboratory). The assump-
Gilbert described variations in field strength with latitude
tion of linearity between the remanence and the applied field
based on the sluggishness or rapidity with which a compass
may not hold true. Or the NRM may have multiple compo-
settled on the magnetic direction. Magnetic intensity was first
nents acquired at different times with different constants of
measured quantitatively in the late 1700s by French scientist
proportionality.
Robert de Paul, although all records were lost in a shipwreck.
In Sections 5.13.3 and 5.13.4, we will discuss the assump-
Systematic measurement of the geomagnetic field intensity
tions behind paleointensity estimates and outline various
began in 1830 (see, e.g., Stern, 2003 for a review). Despite
approaches for getting paleointensity data. We will start by
studies of the geomagnetic field that included some mention
considering thermal remanences and then address deposi-
of its strength, stretching back to at least the time of Gilbert,
tional ones. (To our knowledge, no one has deliberately
basic questions such as what is the average field strength and
attempted paleointensity using other remanence types such as
whether there are any predictable trends remain subject to
chemical or viscous remanences.) In Section 5.13.5, we will
debate. To study field intensity in the past requires us to use
briefly consider ways in which these remanences can be com-
‘accidental’ records; we rely on geologic or archaeological
promised by remagnetization processes. Section 5.13.6 con-
materials, which can reveal much about the behavior of the
siders how paleointensity data can be evaluated as to their
Earth’s magnetic field in ancient times.
reliability and Section 5.13.7 reviews the published data and
There have been several fine reviews of the field of paleointen-
database initiatives. We concentrate here on data prior to the
sity (see, e.g., Valet, 2003) and the subject is developing very
Holocene as the Holocene is the subject of a separate chapter in
rapidly. Paleointensity data derived from archaeological materials
this volume. Finally, Section 5.13.8 highlights some of the
will be considered elsewhere (Chapter 5.09). This chapter will
major issues posed by the paleointensity data.
review the theoretical basis for paleointensity experiments in igne-
ous and sedimentary environments especially with regard to
experimental design. We will then turn to new and updated exist-
ing databases. Finally, we will highlight current topics of interest.
5.13.3 Paleointensity with Thermal Remanence
It appears that Folgheraiter (1899) was the first to propose that

5.13.2 Theory of Paleointensity normalized thermal remanences of pottery be used to study the
ancient magnetic field, although K€ onigsberger and/or Thellier
In principle, it is possible to determine the intensity of ancient is most often given credit. K€ onigsberger (1936) described an
magnetic fields because the primary mechanisms by which experimental protocol for estimating the ratio of NRM to a
rocks become magnetized (e.g., thermal, chemical, and detrital laboratory-acquired TRM (Figure 1(a)) and assembled data
remanent magnetizations or TRM, CRM, and DRM, respec- from igneous and metamorphic rocks that spanned from the
tively) can be approximately linearly related to the ambient Precambrian to the present (Figure 1(b)). He noted that with
field for low fields such as the Earth’s. Thus, we have by few exceptions, the ratio MNRM/Mlab decreased with increasing
assumption age and discussed various possible explanations for the trend,
MNRM ’ aanc Hanc including changing geomagnetic field strength and shaking by
earthquakes. His preferred reason for the trend in normalized
and remanence, however, was that magnetized bodies lose their
magnetism over time, a phenomenon we now recognize as
Mlab ’ alab Hlab [1]
magnetic viscosity. In fact, K€ onigsberger believed that the
where alab and aanc are dimensionless constants of proportion- trend in normalized remanence could be used to date rocks
ality, MNRM and Mlab are natural and laboratory remanent (see also Koenigsberger, 1938a,b). It was Thellier (1938) who
magnetizations, respectively, and Hanc and Hlab are the argued strongly for the use of the thermal remanences of
Paleointensities 463
us
ro
s
ry
ou
ife
n
rna
sic
ian
nia
ce
ic
y
on
iar
ate
ss
eta
r as
rm
vo
rb
rt
ia
Qu
Ca
De
Pe
Te
Ju
Cr
Tr
2
10
8 1.0
4 6
2
0 Jrn + Jrt
H = +0.4 Oe
1 0.8
1
#13
3 Qnt = 0.63
#13
Scale of Qnt
0.6
J (10-3)
5 7
0
Jrn - Jrt
H = -0.4 Oe
0.4
-1
0.2
-2
200 400 100 200
(a) Temperature (⬚C) (b) Scale of years (Ma)
Figure 1 (a) Example of thermal normalization experiment of Koenigsberger (1938a,b). A specimen is heated to given temperature and cooled in a
field of +0.4 Oe (40 mT) (e.g., step #1). Then, the specimen is heated to same temperature and cooled in field of 0.4 Oe (e.g., step #2). The two curves
can be decomposed to give Mnrm and Mlab, the ratio of which was termed Qnt by K€onigsberger. (b) Qnt data for a number of specimens compiled by
K€onigsberger (1938b). The specimen from (a) is labeled #13. These data were interpreted by K€onigsberger to reflect the decay of magnetic remanence
with time.
moment may have enough thermal energy to flip the sense of

archaeological artifacts normalized by laboratory TRMs for
magnetization from one easy axis to another.
studying the past magnetic fields.
If we had a collection of magnetized particles with some
K€onigsberger’s approach was largely empirical; he knew
initial statistical alignment of moments giving a net remanence
that TRMs were proportional to the magnetic fields in which
Mo, the random flipping of magnetic moments from one easy
they cooled and that remanences tended to decay over time
axis to another over time will eventually lead to the case where
and he was well aware of the relationship between coercivity
there is no preferred direction and the net remanence will have
and thermal blocking. Nonetheless, he had very few tools at his
decayed to zero. The rate of approach to magnetic equilibrium
disposal to discriminate among the myriad possible explana-
is determined by the ‘relaxation time,’ which describes the
tions for his observed trend that the NRM/TRM ratio appeared
frequency of moments flipping from one easy axis to another.
to decay with increasing age. For example, he did not call on
Relaxation time according to the Néel theory is given by
apparent polar wander to explain deviant directions, relying
instead on the idea that parts of lava flows tend to cool below 1 ½anisotropy energy 1 ½Kv
t ¼ exp ¼ exp [2]
their Curie temperatures before they stop moving. C ½thermal energy C ½kT
The theoretical basis for how ancient magnetic fields might
be preserved was clarified with the work of Nobel Prize winner where C is a frequency factor with a value of something like
(Néel, 1949, 1955). Modern theory of TRM is discussed in 1010 s1, v is volume, and K is an ‘anisotropy constant.’
detail in Chapter 5.08 (see also Tauxe et al., 2010 and a recent Equation [2] is sometimes called the Néel equation.
review by Valet, 2003), but we review the essential ideas here. The energy barrier for magnetic particles to flip through a
Briefly, a magnetized rod in the absence of a magnetic field ‘hard direction’ into the direction of the applied field H (the
will tend to be magnetized in one of several (often two) ‘easy’ anisotropy energy) requires less energy than to flip the other
directions. In order to overcome the intervening energy barrier way, so relaxation time must be a function of the applied field.
and get from one easy direction to another, a magnetic particle The more general equation for relaxation time is given by
must have energy sufficient to leap through some intervening
1 ½Kv H 2
‘hard’ direction. According to the Boltzmann distribution law, t ¼ exp 1 [3]
C ½kT Hc
the probability of a given particle having an energy e is propor-
tional to ee/kT where k is the Boltzmann constant and T is the where H and Hc are the applied field and the field required to
temperature in kelvin (yielding thermal energy for the product overcome the anisotropy energy and change the moment of
kT ). Therefore, it may be that at a certain time, the magnetic the particle (known as the ‘coercivity’).
From eqn [2], we know that t is a strong function of population of these grains is in equilibrium with the applied
temperature. As described by Néel (1955), there is a very field and is called ‘superparamagnetic.’ Further cooling
sharply defined range of temperatures over which t increases increases the relaxation time such that the magnetization is
from geologically short to geologically long timescales (see effectively blocked and the rock acquires a thermal remanence.
€
Dunlop and Ozdemir (1997) and Tauxe et al. (2010) for Consider a lava flow that has just been extruded (see
more details). Taking reasonable values for magnetite, the Figure 3). First, the molten lava solidifies into rock. While
most common magnetic mineral, we can calculate the varia- the rock is above the Curie temperature, there is no remanent
tion of relaxation time as a function of temperature for a cubic magnetization; thermal energy dominates the system. As the
grain of width ¼ 25 nm as shown in Figure 2. At room temper- rock cools through the Curie temperature of its magnetic
ature, such a particle has a relaxation time of longer than the phase(s), exchange energy (the energy that encourages elec-
age of the Earth, while at a few hundred degrees centigrade, the tronic spins to align with each other) becomes more important
grain has a relaxation time that allows the magnetization to flip and the rock acquires a magnetization. The magnetization,
frequently between easy axes and can maintain an equilibrium however, is free to track the prevailing magnetic field because
with the external field. Such populations will have a slight anisotropy energy is still less important than the energy
statistical preference for the direction of the applied field encouraging alignment with the magnetic field (the magneto-
because of the small difference in relaxation time between static energy). At this high temperature, the magnetic moments
directions closer to the applied field direction from eqn [3]. in the lava flow are superparamagnetic and tend to flop from
The temperature at which t is equal to about 102–103 s is one easy direction to another, with a slight statistical bias
defined as the blocking temperature, Tb. At or above the block- toward the direction with the minimum angle to the applied
ing temperature, but below the Curie temperature (the temper- field (Figure 3(c)). The equilibrium magnetization of super-
ature at which all spontaneous magnetization is lost), a paramagnetic grains is only slightly aligned, and the degree of
alignment is a quasilinear function of the applied field for low
fields like the Earth’s. The magnetization approaches satura-
1021 tion at higher fields, depending on the details of the controls
on anisotropy energy like shape, size, and mineralogy.
1018 1 Billion years
1015
5.13.3.1 Linearity Assumption
Relaxation time (s)
1012 1 My
109 From theory, we expect thermal remanences of small single
domain particles to be approximately linearly related to the
106 1 Year
applied field for low fields like the Earth’s. However, as particle
103 size increases, TRMs can become quite nonlinear even at rela-
100 s tively low fields (see Figure 4). Predicted TRM curves with
100
respect to the applied field for randomly oriented populations
10-3 of single domain particles ranging in size from 20 to 100 nm
10-6 widths are plotted in Figure 4(a). We calculated these curves
assuming quasi-equidimensional grains (1.5:1) and highly
0 100 200 300 400 500 elongate grains (10:1). For the elongate grains, the TRM is
Temperature (⬚C) predicted to be distinctly nonlinear even for the 80 nm parti-
Figure 2 Variation of relaxation time versus temperature for a 25 nm cles. (The approximate range of the present Earth’s field is
width cube of magnetite. Reproduced from Tauxe L, Banerjee SK, shown as the shaded box.) Particles of magnetite larger than
Butler R, and van der Voo R (2010) Essentials of Paleomagnetism. about 90 nm will have more complicated remanent states
Berkeley, CA: University of California Press. (flower, vortex, and multidomain) and will not necessarily
(a) (b) (c)

Figure 3 (a) Picture of lava flow courtesy of Daniel Staudigel. (b) While the lava is still well above the Curie temperature, crystals start to form but
are nonmagnetic. (c) Below the Curie temperature but above the blocking temperature, certain minerals become magnetic, but their moments
continually flip among the easy axes with a statistical preference for the applied magnetic field. As the lava cools down, the moments become fixed,
preserving a thermal remanence. (b and c) Modified from animation of Genevieve Tauxe available at http://magician.ucsd.edu/Lab_tour/movs/TRM.mov.
Reproduced from Tauxe L, Banerjee SK, Butler R, and van der Voo R (2010) Essentials of Paleomagnetism. Berkeley, CA: University of California Press.
1.0
10:1:1 100 nm
80 nm 60 nm
0.8 1.5:1:1
40 nm
0.6 100 nm
TRM/sIRM
80 nm
60 nm
0.4
40 nm
0.2
20 nm
0.0
(a) 0 50 100 150 200
1.0
100 nm
80 nm 100 nm ± 50
0.8 80 nm ± 40
0.6
TRM/sIRM
0.4
All 10:1:1
0.2 Range of
present field
0.0
0 50 100 150 200
(b) mo H (mT)
Figure 4 Predicted thermal remanent magnetization (TRM) expressed as a fraction of saturation for various particle sizes and distributions of
magnetite. Note the nick point for which the linearity assumption fails is a strong function of particle size, but linearity holds true for equant particles in
fields less than a hundred mT. Strongly elongate particles will behave in a more nonlinear fashion.
follow the predicted curves, which are based on single domain distributed particle widths (all with 10:1 elongation) with mean
theory. widths of 80 and 100, respectively. The effect of the distribution
We note in passing that Kletetschka et al. (2006) had pos- of particle sizes is depression of the TRM below that for a uniform
tulated that multidomain particles have TRMs that are highly distribution because smaller particles have much lower TRMs at a
nonlinear at fields below some threshold value with linear given field strength that the effect is asymmetrical. (The difference
behavior at higher field values. This behavior was observed between 80 and 100 nm widths at, say, 100 mT is much less than
using a Schonstedt oven, which has very poor field control, the difference between 60 and 80 nm at the same field. Note also
and we were unable to reproduce the observations in the SIO that grains smaller than about 20 nm are superparamagnetic at
laboratory, which has excellent field control; linear behavior room temperature and do not contribute to the TRM at all, so
was observed in fields as low as 10 nT (Yongjae Yu, personal distributions that include small particles will have suppressed
communication). TRMs relative to their theoretical saturation remanences.)
Figure 4(b) shows the effect of having a distribution of grain Dunlop and Argyle (1997) discovered strongly nonlinear
sizes. We calculated curves for populations with normally TRM acquisition behavior in synthetic specimens with mean
grain sizes in the single domain grain range. Although their lab
fields were mostly much higher than those of the Earth’s field (up
to 9 mT!), the results should give practitioners of paleointensity
pause. Moreover, Selkin et al. (2007) had found nonlinear TRM
Thermal remanent magnetization

behavior in natural specimens with single domain behavior, and
the nonlinearity is distinct in fields as low as 50 mT.
The modeling exercises shown in Figure 4 and the experi-
mental results of Dunlop and Argyle (1997) and Selkin et al.
(2007) suggest that it would be a wise practice to incorporate a
test of TRM linearity into paleointensity experiments as a mat-
ter of routine. If the relationship between TRM and applied
field is known empirically, then biased results can be corrected pTRM
to the true ancient intensity. (500,400)
5.13.3.2 Alteration During Heating

The second assumption for absolute paleointensity determina-
tions is that the laboratory and ancient proportionality con- 0 100 200 300 400 500 600
stants are the same (i.e., alab ¼ aanc in eqn [1]). Simply Tc
Temperature (⬚C)
measuring the NRM and giving the specimen a total TRM
does nothing to test this assumption. For example, alteration Figure 5 Laws of independence and additivity. Partial thermal
of the specimen during heating could change the capacity to remanences (pTRMs) acquired by cooling between two temperature
steps are independent from one another and sum together to form the
acquire thermal remanence and give erroneous results with no
total TRM. Modified from McElhinny MW (1973) Paleomagnetism and
way of assessing their validity. Plate Tectonics. Cambridge: Cambridge University Press.
There are several ways of checking the ability of the speci-
men to acquire thermal remanence in paleointensity experi-
ments. The most commonly used are experiments that employ pTRMs (see Figure 5), an assumption called the ‘law of
stepwise replacement of the NRM with a laboratory thermal additivity.’
remanence (K€ oenigsberg/Thellier–Thellier, or KTT family of There are several possible ways to progressively replace the
experiments) and those that compare anhysteretic remanence NRM with a pTRM in the laboratory. In the original KTT
before and after heating (‘Shaw’ family of experiments). Other method (see, e.g., Figure 1(a)), the specimen is heated to
approaches attempt to prevent the alteration from occurring, some temperature (T1) and cooled in the laboratory field
for example, by heating in controlled atmospheres or vacuum Blab. After measurement of the combined remanence, what is
or by using microwaves to heat just the magnetic phases, left of the NRM plus the new laboratory pTRM is
leaving the rest of the specimen cool. Another approach is to M1 ¼ MNRM + MpTRM
find materials that are particularly resistant to alteration (e.g.,
submarine basaltic glass or single plagioclase crystals). Finally, Then, the specimen is heated a second time and cooled
some methods attempt to normalize the remanence with iso- upside down (in field Blab). The second remanence is
thermal remanent magnetization (IRM) and avoid heating therefore
altogether. We will briefly describe each of these in turn, begin-
M2 ¼ MNRM MpTRM
ning with the KTT family of experiments.
Simple vector subtraction allows the determination of the
5.13.3.2.1 KTT family of experiments NRM remaining at each temperature step and the pTRM gained
Detection of changes in the proportionality constant caused by (see Figure 6(a)). These are nowadays plotted against each other
alteration of the magnetic phases in the rock during heating in what is usually called an ‘Arai plot’ (Nagata et al., 1963) as in
has been a goal in paleointensity experiments since the earliest Figure 6(b). The KTT method implicitly assumes that a magne-
days. As we have already seen (Figure 1(a)), Koenigsberger tization acquired by cooling from a given temperature is entirely
(1936, 1938a,b) heated specimens up in stages, progressively replaced by reheating to the same temperature (i.e., Tb ¼ Tub), an
replacing the NRM with partial thermal remanences (pTRMs), assumption known as the ‘law of reciprocity.’
an experiment that was elaborated on by Thellier (1938) and As magnetic shielding improved, more sophisticated
Thellier and Thellier (1959). The so-called KTT approach is approaches were developed. In the most popular paleointen-
particularly powerful when lower-temperature steps are sity technique (usually attributed to Coe, 1967), we substitute
repeated, to verify directly that the ability to acquire a thermal cooling in zero field for the first heating step. This allows the
remanence has not changed. direct measurement of the NRM remaining at each step. The
The stepwise approach relies on the assumption that pTRMs two equations now are
acquired by cooling between any two temperature steps (e.g.,
500 and 400 C in Figure 5) are independent of those acquired M1 ¼ MNRM
between any other two temperature steps. This assumption is and
called the ‘law of independence’ of pTRMs. The approach also
assumes that the total TRM is the sum of all the independent M2 ¼ MNRM + MpTRM
9
8 100 ⬚C
NRM remaining (nAm2)

1.0 7
175
0.8 6
NRM
pTRM 5
M (nAm2)
0.6 4 250
0.4 3
300
2
0.2 350
1
0.0 0
0 100 200 300 400 500 0 1 2 3 4 5 6 7 8 9
(a) Temperature (⬚C) (b) pTRM gained (nAm2)
Figure 6 Illustration of the KTT method for determining absolute paleointensity. (a) Thermal demagnetization of NRM shown as filled circles and the
laboratory-acquired pTRM shown as open symbols and (b) plot of NRM component remaining versus pTRM acquired for each temperature step.
Reproduced from Tauxe L, Banerjee SK, Butler R, and van der Voo R (2010) Essentials of Paleomagnetism. Berkeley, CA: University of California Press.
The laboratory MpTRM in this ‘zero-field/infield’ (or ZI) responsible for ‘transient hysteresis’ (see Fabian, 2003; Yu
method is calculated by vector subtraction. Alternatively, the and Tauxe, 2005).
first heating and cooling can be done in the laboratory field One can imagine that something similar to transient hyster-
and the second in zero field (Aitken et al., 1988; see also Valet esis could occur if we cooled a particle from its Curie
et al., 1998), here called the ‘infield/zero-field’ or (IZ) method. temperature and then heated it back up again. Just below the
In all three of these protocols, lower-temperature infield Curie temperature, the particle would be in a saturated state
cooling steps can be repeated to determine whether the rema- (because the magnetization is quite low and the vortex structure
nence carrying capacity of the specimen has changed. These is just an attempt by the particle to reduce its external field). As
steps are called ‘pTRM checks.’ Differences between the first the specimen cools down, the magnetization grows. At some
and second MpTRMs at a given temperature indicate a change in temperature, a vortex structure may form. As the specimen is
capacity for acquiring thermal remanences and are grounds for heated back up again, the vortex may well remain stable to
suspicion or rejection of the data after the onset of such a higher temperatures than its formation temperature by analogy
change. Some have proposed that paleointensity data can be to the behavior in the simulated hysteresis experiment.
‘fixed’ even if the pTRM checks show significant alteration (e.g., If the particle is large enough to have domain walls in its
McClelland and Briden, 1996; Valet et al., 1996). The argu- remanent state, then the scenario is somewhat different and
ment is that if pTRM checks can be brought back in accordance not easily tractable by theory (see Chapter 5.08). At just below
with the original pTRM measurements using a correction fac- its Curie temperature, the particle is at saturation. As the par-
tor, then if that same correction factor is applied to all subse- ticle cools, domain walls will begin to form at some tempera-
quent pTRM measurements, the effect of the alteration has ture. The remanent state will have some net moment because
been accounted for and the data can be considered ‘reliable.’ the domain walls are distributed such that there is incomplete
We consider this correction to carry some risk and ‘corrected’ cancelation leaving a small net remanence, proportional to the
data should be clearly marked as such. applied field for moderate field strengths. As the temperature
Despite its huge popularity and wide spread use, the ramps up again, the walls ‘walk around’ within the particle
approach of progressively replacing the NRM with a thermal seeking to minimize the magnetostatic energy and are not
remanence has several drawbacks. Alteration of the ability to destroyed until temperatures very near the Curie temperature.
acquire a pTRM is not the only cause for failure of the assump- The fact that blocking and unblocking of remanence occur
tion of equality of alab and aanc. Both experiment (Bol’shakov at different temperatures in some particles means that a
and Shcherbakova, 1979; Shcherbakova et al., 2000) and the- pTRM blocked at a given temperature will remain stable to
ory (e.g., Dunlop and Xu, 1994; Xu and Dunlop, 1994) suggest higher temperatures; the unblocking temperature is not equal
that the essential assumption of equivalence of blocking and to the blocking temperature. This means that alab 6¼ aanc and
unblocking temperatures may break down for larger particles. the key assumptions of the KTT-type methods are not met.
Micromagnetic modeling of hysteresis behavior can shed The Arai plots may be curved (see Dunlop and Ozdemir, €
some light on what might be going on. In simulated hysteresis 1997, for a more complete discussion), and if any portion of
experiments, particles can be subjected to a large DC applied the NRM/TRM data are used instead of the entire temperature
magnetic field, sufficient to completely saturate them. As the spectrum, the result could be biased. For example, the lower-
field is lowered, certain particles form vortex structures at some temperature portion might be selected on the grounds that the
applied field strength (see Figure 7). These vortex structures are higher-temperature portion is affected by alteration or the
destroyed again as the field is ramped back up to saturation. higher-temperature portion might be selected on the grounds
However, the field at which the vortex is destroyed is higher that the lower-temperature portion is affected by viscous rem-
than the field at which it formed. This is the phenomenon anence. Both of these interpretations would be wrong.
Descending curve
1
(a)
0.8 Ascending curve
M/M (0.6 T)
0.6
(b)
0.4
0.2
0 40 80 120 160 200
moH (mT )
Descending curve Ascending curve
Figure 7 Example of irreversible behavior when particle is brought from a saturated state at high field to zero fields and back up again. A vortex
structure forms on the descending curve in the simulated hysteresis loop at the sharp drop in magnetization at about 40 mT (labeled ‘(a)’). A snapshot of
the micromagnetic state is shown to the lower left. This feature intensifies as the field drops to zero, resulting in a loss of magnetization. When the
field ramps back up again, the vortex remains stable well past its formation field (labeled ‘(b)’). A snapshot of the micromagnetic state on the ascending
curve is shown in the lower right. The vortex is not destroyed until a higher field, when the loop is closed. Reproduced from Yu and Tauxe (2004).
In order to detect inequality of blocking and unblocking There are several other violations of the fundamental
and the presence of unremoved portions of the pTRM known assumptions that require additional tests and/or corrections
as ‘high-temperature pTRM tails,’ several embellishments to in the paleointensity experiment besides alteration or failure
the KTT-type experiment have been proposed and more are of the law of reciprocity. For example, if the specimen is
on the way. In one modification, a second zero-field step is anisotropic with respect to the acquisition of thermal rema-
inserted after the infield step in the ZI method. This so-called nence, the anisotropy tensor must be determined and the
pTRM tail check (e.g., Riisager and Riisager, 2001) assesses intensity corrected (e.g., Fox and Aitken, 1980). The detection
whether the pTRM gained in the laboratory at a given temper- and correction for anisotropy can be very important in
ature is completely removed by reheating to the same temper- certain paleomagnetic (and archaeomagnetic) materials. The
ature. If not, the specimen is said to have a ‘pTRM tail,’ a correction involves determining the TRM (or the anhysteretic
consequence of an inequality of the unblocking temperature remanence (ARM) proxy) anisotropy tensor and matrix mul-
Tub and the original blocking temperature Tb in violation of the tiplication to recover the original magnetic vector (see Selkin
law of reciprocity and grounds for rejection. A second modifi- et al., 2000, for a more complete discussion). Moreover,
cation is to alternate between the IZ and ZI procedures (the because the approach to equilibrium is a function of time,
so-called ‘IZZI’ method first conceived with AgNés Genevey slower cooling results in a larger TRM; hence, differences in
and described by Yu et al., 2004). The IZZI method is also cooling rate between the original remanence acquisition and
extremely sensitive to the presence of pTRM tails and may that acquired in the laboratory will lead to erroneous results
obviate the need for the pTRM tail check step. An example of (e.g., Halgedahl et al., 1980). Compensating for differences in
a complete IZZI experiment is shown in Figure 8. cooling rate is relatively straight forward if the original
relationship between the coercivity spectrum of the MNRM

yVDS y -z (presumed to be a thermal remanence) and ARM prior to any
laboratory heating.
100
30 As with the KTT-type methods, MNRM is normalized by a
200
250 laboratory thermal remanence. But in the case of the Shaw-type
25 300 methods, the specimen is given a total TRM, ðMTRM1 Þ, which is
x
AF demagnetized as well. Finally, the specimen is given a
NRM (nAm2)
20 yo 350
400 second ARM ðMARM2 Þ and demagnetized for the last time.
d450
15 The general experiment is shown in Figure 9(a) and 9(b). If
450
10 the first and second ARMs do not have the same coercivity
ΔyT spectrum as in Figure 9(b), the coercivity of the specimen
5 500 has change and the NRM/TRM ratio is suspect.
D500 Rolph and Shaw (1985) suggested that the ratio
0
MARM1 =MARM2 at each demagnetizing step be used to ‘correct’
0 5 10 15 20 25 30 35 40
for the alteration bias of MTRM1 by
pTRM (nAm2)
MARM1
Figure 8 Data from an IZZI experiment. Circles are the pTRM gained at MTRM* ¼ MTRM1
a particular temperature step versus the NRM remaining. Solid
1 MARM2
symbols are those included in the slope calculation. Blue (darker) So doing can in some cases restore linearity between NRM
symbols are the infield–zero-field steps (IZ) and the brown (lighter)
and TRM as shown in Figure 9(c).
symbols are the zero-field–infield steps (ZI). The triangles are the pTRM
checks and the squares are the pTRM tail checks. The difference
Valet and Herrero-Bervera (2000) argued that only data
between the pTRM check and the original measurement is di as shown by requiring no correction and utilizing the entire coercivity spec-
the horizontal bar labeled d450. The difference between the first NRM trum should be used. They further pointed out that many
measurement and the repeated one (the pTRM tail check) is shown by the specimens are required to lend credibility to a paleointensity
vertical bar labeled D500. The vector difference sum (VDS) is the sum experiment. As the former requirement generally leaves very
of all the NRM components (tall vertical bar labeled VDS). The NRM few specimens, Valet and Herrero-Bervera reasoned that a
fraction is shown by the vertical dashed bar. The insets are the vector quicker experimental procedure would ultimately result in
components (x, y, z) of the zero-field steps. The solid symbols are (x, y) more acceptable data, hence a better overall outcome, even if
pairs and the open symbols are (x, z) pairs. The specimen was unoriented the results from many experiments are discarded. To speed up
with respect to geographic coordinates. The laboratory field was
the measurement process, they employed a truncated Shaw
applied along the z-axis in the infield steps. Redrawn from Tauxe L and
Staudigel H (2004) Strength of the geomagnetic field in the Cretaceous
method in which no ARMs are imparted, but both the NRM
Normal Superchron: New data from submarine basaltic glass of the and the laboratory TRM are completely demagnetized using
Troodos Ophiolite. Geochemistry, Geophysics, Geosystems 5(2): AFs. Linearity of the two when plotted as in Figure 9(a) is
Q02H06. http://dx.doi.org/10.1029/2003GC000635. taken as the sole criterion for acceptance.
Tsunakawa and Shaw (1994) suggested that a tendency for
chemical alteration could also be detected if the specimen
is heated to above the Curie temperature twice, each followed
cooling rate is known and the specimens behave according by AF demagnetization (see Figure 9(d)–9(f )). During the
to single domain theory. Alternatively, one could take an second heating step, the specimen is left at high temperature
empirical approach in which the rock is allowed to acquire a for a longer period of time than the first heating step to
pTRM under varying cooling rates (e.g., Genevey and Gallet, encourage alteration to continue so that it may be detected
2003), an approach useful for cooling rates of up to a day by the method. If the slope MTRM1 =MTRM* differs by more than
2
or two. experimental error, the experimental results are rejected.
The issue of contamination of the remanence by multi-
5.13.3.2.2 Shaw family of experiments domain particles has also been considered in the Shaw-type
The previous section was devoted to experiments in which methods. It has long been known (Ozima et al., 1964) that
detection of nonideal behavior is done by repeating various specimens can lose much of their remanence by cooling to
temperature steps. In this section, we will consider an alterna- temperatures below about 160 C and warming in zero field.
tive approach, long in use in paleointensity studies, which This behavior is generally attributed to magnetocrystalline-
employs the laboratory proxy ARM. The so-called Shaw dominated remanences cycling through the so-called Verwey
method (e.g., Shaw, 1974) is based on ideas first explored by transition at which the axis of magnetocrystalline anisotropy
van Zijl et al. (1962a,b). In its simplest form, we measure the changes, erasing the magnetic memory of these particles (see,
NRM and then progressively demagnetize the NRM with alter- €
e.g., Dunlop and Ozdemir, 1997). This behavior is frequently
nating fields (AFs) to establish the coercivity spectrum of the assumed to occur most readily in multidomain particles; hence,
specimen prior to heating. The specimen is then given an ARM their contribution could be minimized if specimens are pre-
ðMARM1 Þ by subjecting the specimen to progressively higher treated to low temperatures (low-temperature demagnetization,
peak AFs, which decay in the presence of a small bias field. or LTD) prior to measurement. Yamamoto et al. (2003) and
The use of ARM has been justified because it is in many ways Yamamoto and Tsunakawa (2005) argued that one of the major
analogous to the original TRM (see Chapter 5.08). MARM1 is causes of failure in paleointensity experiments is the effect of
then progressively demagnetized to establish the original multidomain particles, which violate the essential assumption
0 mT
0 mT 0 mT 0 mT
MARM1
MARM2
MTRM1
MNRM
AF
g
sin
ea
cr
In
(a) MTRM1 (b) MARM2 (d) MTRM2 (e) MARM3
0 mT
0 mT
MTRM1
MNRM
(c) MTRM*1 (f) MTRM*2
Figure 9 Shaw family of methods (see text). (a) Plot of pairs of NRM and the first TRM for each AF demagnetization step. (b) Plot of pairs of the first ARM
and the second ARM for each AF demagnetization step. (c) Plot of pairs of NRM and TRM adjusted by the ratio of ARM1/ARM2 for that AF step from
(b) TRM∗1. (d) Same as (a) but for the first and second TRMs. (e) Same as (a) but for the second and third ARMs. (f ) Same as (c) but for first and second
TRM where TRM∗2 is adjusted using ARM2/ARM3 ratio from (e). Redrawn from Yamamoto Y, Tsunakawa H, and Shibuya H (2003) Palaeointensity
study of the Hawaiian 1960 lava: Implications for possible causes of erroneously high intensities. Geophysical Journal International 153(1): 263–276.
that the original blocking temperature is the same as the labo- heating specimens in controlled atmospheres, reducing the
ratory unblocking temperature. They therefore treat specimens time at temperature by, for example, measuring the specimens
to LTD prior to AF demagnetization of each remanence. This at elevated temperature, and using microwaves to excite spin
‘LTD-DHT Shaw’ method gave improved results for the other- moments as opposed to direct thermal heating. Finally, there
wise disappointingly difficult Hawaiian 1960 lava flow (see, e.g., has been some effort put into finding materials that resist
Tanaka and Kono, 1991; Valet, 2003; Valet and Herrero-Bervera, alteration during the heating experiments.
2000).
The LTD-DHT Shaw experiment assumes that mainly, the
multidomain particles are affected by the LTD step. However,
5.13.3.3.1 Reduced number of heating steps
Kono and Ueno (1977) described in detail a single heating per
Carter-Stiglitz et al. (2002, 2003) found that single domain
temperature step method suggested by Kono (1974) whereby
magnetites can and do lose substantial remanence by LTD as
the specimen is heated in a laboratory field applied perpendic-
well. This behavior means that LTD treatment may demagne-
ular to the NRM. MpTRM is obtained by vector subtraction.
tize part of the desired as well as the undesired NRM. It is
Reducing the number of heatings can reduce the alteration to
possible that the SD remanence removed by LTD may be the
some extent. However, this method has only rarely been
low coercivity contribution and unimportant to the
applied because it can only be used for strictly univectorial
paleointensity.
NRMs (an assumption that is difficult to test with the data
The primary reasons stated for using the Shaw method are
generated by this method) and requires rather delicate posi-
that (1) it is faster and (2) because the specimen is only heated
tioning of specimens in the furnace or fancy coil systems that
once (albeit to a high temperature), alteration is minimized.
generally have a limited region of uniform field, reducing the
The first rationale is no longer persuasive because modern
number of specimens that can be analyzed in a single batch.
thermal ovens have high capacities and the KTT method is
While pTRM checks are possible with this method, they neces-
certainly not slower than the Shaw method on a per specimen
sitate additional heating steps and are not generally performed.
basis. This is particularly true for the LTD-DHT Shaw method
A second strategy for reducing the number of heating steps
as this experiment takes approximately 8 h to complete per
was proposed by Hoffman et al. (1989) and modified by
specimen. The second rationale may have some validity. The
Hoffman and Biggin (2005) (see also Dekkers and B€ ohnel,
key features of any good experiment are the built-in tests of the
2006). In the Hoffman–Biggin version, at least five specimens
important assumptions.
from a given cooling unit are sliced into four specimens each,
one of which is dedicated to rock magnetic analysis. The
remaining specimens (at least 15) are heated a total of five
5.13.3.3 Methods that Minimize Alteration
times giving remanence measurements M1 M5. (Please note
Several alternative approaches have been proposed in which that bold face parameters are vectors, while normal text vari-
instead of detecting nonideal behavior such as alteration, they ables are scalars, in this case the magnitudes.) In the first three
attempt to minimize it. These methods include reducing the heating steps, the specimens are treated to increasingly high
number of heating steps required (as in the Shaw methods), temperatures (T0, T1, and T2) and cooled in zero field. The first
heating step ostensibly removes any secondary overprint (e.g., Hoffman and Biggin (2005) suggested that the data, which
a VRM, viscous remanent magnetization) and M1 serves as the are heavily influenced by data from a single, low blocking
baseline for normalizing all subsequent steps so that data from temperature specimen (green symbols in Figure 10), could be
different specimens can be combined. After the three zero-field reweighted to remove the bias. Furthermore, they proposed
heating steps, the specimens are heated again to T2 and cooled averaging all the data by accepted specimen and including
with the laboratory field switched on between T2 and T0 after the y-intercept in the calculation. These modifications yielded
which it is switched off. This treatment step gives the pTRM a concordant result with the known field within error. Finally,
acquired between T2 and T0 by vector subtraction of M4 M3. they redefined many of the parameters typically used in
The fifth heating step is to T1 followed by a zero-field cooling. paleointensity experiments (see Section 5.13.6) for use with
This final step serves to supply both the pTRM acquired the multispecimen method.
between T2 and T1 by vector subtraction of M4 M5 and a The primary advantage of the multispecimen approach put
kind of ‘pseudo pTRM check’ step as explained later. forward by Hoffman and Biggin (2005) is the speed with
In interpreting results, there are two data points from each which measurements can be made, allowing many more spec-
specimen with estimates for NRM remaining versus pTRM imens to be analyzed. While the method may be fast, it loses
gained, denoted T1 and T2. The NRM remaining part of T1 multiple pTRM checks and any ability to assess the equivalence
and T2 are ratios M2/M1 and M3/M1, respectively. The pTRMs of blocking and unblocking. Moreover, the method strongly
gained at T1 and T2 are jM5 M4j/M1 and jM4 M3j/M1. emphasizes the lower blocking temperature portion of the
Because all remanences are normalized by the NRM remaining blocking temperature spectrum (especially in the moment cor-
after zero-field cooling from T0 (M1) measured for each spec- rected version). This means that the remanence is contami-
imen, we can combine data from the different specimens nated by viscous or multidomain remanences leading to a
together on a single Arai-like plot (see Figure 10). concave downward curve in the Arai plot; the multispecimen
Hoffman and Biggin (2005) had several criteria that help result will overestimate the true value of the paleointensity.
screen out ‘unreliable’ data. First, they require that the direc- Finally, it is experimentally very difficult to turn the laboratory
tions of the zero-field steps trend to the origin on an orthogo- field off precisely when the specimen’s internal temperature is
nal plot and have low scatter. This helps eliminate data for T0 because only the oven temperature is known and the spec-
which the characteristic remanence has not been isolated imen temperature lags behind that of the oven by variable and
(although three zero-field steps are not generally considered unknown amounts, depending on the exact disposition of the
sufficient for this purpose). Second, they require the y-intercept specimens in the oven. This bias will lead to scatter and con-
to be between 0.97 and 1.03 and that the correlation coeffi- tribute to a systematic bias (the field will always be turned off
cient must be 0.97. If the T1 data are displaced from the line at too high a temperature, thereby underestimating the pTRM
connecting the T2 point and a y-intercept of 1.0, then the gained).
specimen may have altered during laboratory heating (e.g., On the positive side, the presentation of all specimen data
open symbols in Figure 10) and can be rejected. on a single Arai diagram (also proposed by Chauvin et al.,
Noting that the results of the multispecimen procedure 2005) is an interesting modification of the traditional Arai
when applied to the 1971 Hawaiian flow (shown in Figure 10) diagram. Plotting all the KTT data from specimens from a
were significantly different than the known field (37 mT), given cooling unit on a single Arai diagram allows instant
assessment of the reproducibility of data and of course can be
done with traditional experimental results.
1.00
5.13.3.3.2 Use of controlled atmospheres to reduce
0.75 alteration
Alteration during heating is caused by oxidation (or reduction)
NRM/NRMo
of the magnetic minerals in the specimen. There have been

0.50 several strategies to reduce this effect with varying degrees of
success. Thellier (1938) tried using vacuum and nitrogen atmo-
spheres. Taylor (1979) developed a technique for paleointen-
0.25 sity determination that encapsulated specimens in silica glass.
By placing an oxygen ‘getter’ such as titanium in the evacuated
glass tube along with the specimen, he suggested that oxygen
0
0 0.25 0.50 0.75 1.00 fugacity could be maintained and alteration would be reduced.
TRM/ TRMo This technique was tested by Sugiura et al. (1979) who claimed
some improvement in experimental results on lunar glass spec-
Figure 10 Illustration of multispecimen approach for specimens from imens. Valet et al. (1998) performed paleointensity experi-
the 1971 Hawaiian lava flow. Different colors represent data from
ments by heating in argon atmospheres and cooling in
different specimens. Different symbols are different heating steps. Open
nitrogen atmospheres. They reported a significant improve-
symbols are from specimen that failed initial selection criteria. The
best-fit line represents a slope that predicts a ‘paleo’ field of 33 mT, ment in their argon results over those performed in air. More
whereas the actual field was 37 mT. Redrawn from Hoffman KA and recently, Kissel and Laj (2004) also used a furnace with a
Biggin AJ (2005) A rapid multi-sample approach to the determination of controlled argon atmosphere to minimize alteration, and
absolute paleointensity. Journal of Geophysical Research 110: B12108. Mochizuki et al. (2004) incorporated heating in a vacuum to
http://dx.doi.org/10.1029/2005JB003646. the LTD-DHT Shaw experiments, a practice that has been
adopted as routine for this method in subsequent investiga- particles can also be excited directly, inviting the possibility
tions and successfully recovered the historical field for samples of heating only the magnetic phases, leaving the matrix alone
from the 1986 Oshima lava flow in Japan. The difficulties of (e.g., Walton et al., 1993). The rationale for developing this
heating and cooling in vacuum and controlled atmospheres method is to reduce the degree of alteration experienced by the
are (1) difficulty in achieving a uniform and reproducible specimen because the matrix often remains relatively cool,
temperature in the oven and (2) unintended oxidation or while the ferromagnetic particles themselves get hot. (The
reduction reactions. It appears that reduction in alteration magnons get converted to phonons, thereby transferring the
can be achieved using these techniques, although the over- heat from the magnetic particle to the matrix encouraging
whelming majority of paleointensity experiments are done alteration, but there may be ways of reducing this tendency
in air. (see Walton, 2004).)
The same issues of nonlinearity, alteration, reciprocity,
5.13.3.3.3 Measurement at elevated temperature anisotropy, cooling rate differences, etc. arise in the microwave
Boyd (1986) suggested that measurements could be made approach as in the thermal approach. Ideally, the same exper-
more rapidly if they were measured at elevated temperatures imental protocol could be carried out with microwave ovens as
instead of cooling back to room temperature for measurement. with thermal ovens. In practice, however, it has proved quite
The idea was that alteration could be detected immediately and difficult to repeat the same internal temperature, making dou-
the experiment aborted, before wasting time finishing the ble (or even quadruple) heatings problematic although pro-
entire measurement sequence. A variant of the method using gress toward this end may have been made (e.g., B€ ohnel et al.,
an infield–infield approach was applied to the 1960 Hawaiian 2003). It is likely that the issues of reciprocity of blocking and
lava flow and the 2.2 ka Kotaki pyroclastic flow by Tanaka et al. unblocking in the original (thermally blocked) and the labo-
(1995a). In the case of the historical flow where the ‘ancient’ ratory (microwave unblocked) and differences in the rate of
field is known (36 mT), the method overestimated the true field blocking and unblocking will remain a problem for some time
by 13%. The idea of measuring at elevated temperatures was as they have for thermally blocked remanences.
warmed up by Le Goff and Gallet (2004) who developed a Concerns were raised by Valet (2003) and Le Goff and
vibrating sample magnetometer equipped with magnetic field Gallet (2004) that the theoretical equivalence between thermal
coils, which allow the specimen to be measured at temperature unblocking and microwave unblocking has not yet been
and in controlled fields, greatly speeding up the measurement explained. In fact, Walton (2005) pointed out that resonance
process and, one hopes, reducing the effects of cooling rate within the magnetic particles is wavelength-dependent. This
and specimen alteration. This method has been applied in raises the possibility that unblocking may occur in an entirely
archaeomagnetic studies with great success (e.g., Gallet and different manner in microwave processes than in thermal ones
Le Goff, 2006). (by chords instead of scales to use a musical metaphor) leading
to serious questions about the applicability of the method for
5.13.3.3.4 Use of microwaves for thermal excitation recovery of paleointensity estimates.
Until now, we have not concerned ourselves with HOW the With improvements in the ability to repeat internal temper-
magnetic moment of a particular grain flips its moment. Ear- ature steps, the microwave method has been adapted to a
lier, we mentioned ‘thermal energy’ and left it at that. But how variety of experimental protocols without a loss in experimen-
does thermal energy do the trick? tal quality (e.g., Biggin et al., 2007; Hill et al., 2002; Yamamoto
An external magnetic field generates a torque on the elec- and Shaw, 2008). The results of Yamamoto and Shaw (2008)
tronic spins, and in isolation, a magnetic moment will respond provide further encouragement as their investigations recov-
to the torque in a manner similar in some respects to the way a ered the laboratory field and only slightly overestimated the
spinning top responds to gravity: the magnetic moment will historical field value of the Hawaiian 1970 lava flow using a
precess about the applied field direction, spiraling in and microwave version of the LTD-DHT Shaw protocol.
come to a rest parallel to it. Because of the strong exchange or
superexchange coupling in magnetic phases, spins tend to be 5.13.3.3.5 Using materials resistant to alteration
aligned parallel (or antiparallel) to one another, and the spira- Another very important approach to the paleointensity prob-
ling is done in a coordinated fashion, with neighboring spins as lem has been to find and exploit materials that are themselves
parallel as possible to one another. This phenomenon is known resistant to alteration. There are an increasing variety of prom-
as a ‘spin wave.’ ising materials, ranging from quenched materials, to single
Raising the temperature of a body transmits energy (via crystals extracted from otherwise alteration-prone rocks, to
‘phonons’) to the electronic spins, increasing the amplitude very slowly cooled plutonic rocks (e.g., layered intrusions).
of the spin waves. This magnetic energy is quantized in Quenched materials include volcanic glasses (e.g., Pick and
‘magnons.’ In the traditional KTT experiment, the entire spec- Tauxe, 1993, 1994), metallurgical slag (e.g., Ben-Yosef et al.,
imen is heated, and the spin waves are excited to the point that 2008a,b, 2009; Shaar et al., 2010, 2011a,b), and welded tuffs
some may flip their moments as described in Section 5.13.3. (Gee and Bowles, 2010). Single crystals of plagioclase or other
As in most kitchens, there are two ways of heating things silicates extracted from igneous rocks (Tarduno et al., 2006,
up: the conventional oven and the microwave oven. In the 2007, 2010) can yield excellent results, while the lava flows
microwave oven, molecules with certain vibrational frequen- themselves may be prone to alteration or other nonideal
cies (e.g., water) are excited by microwaves. These heat up, behavior. Parts of layered intrusions (e.g., Selkin et al., 2000)
passing their heat on to the rest of the pizza (or whatever). If can also perform extremely well during the paleointensity
the right microwave frequency is chosen, ferromagnetic experiment.
While some articles have called the reliability of submarine having multidomain magnetic carriers and thought to circum-
basaltic glass results into question (e.g., Heller and Liu, 1982), vent the problem by imparting a second TRM and repeating the
Tauxe and Staudigel (2004), Bowles et al. (2005), Bowles et al. IZZI experiment in order to plot unblocking temperature
(2011), and Tauxe et al. (2013) addressed these concerns in against unblocking temperature. However, the second experi-
great detail, and the reader is referred to those papers and the ment yielded straight Arai plots and the initial curved behavior
references therein for a thorough treatment of the subject. could not be reproduced. Although there was no evidence for
There is no basis in experiment or theory for questioning the chemical alteration (all pTRM checks passed), the domain state
origin of remanence of submarine basaltic glass. of the specimen had clearly changed. Nonetheless, this type of
approach may be useful in some cases in which the curved Arai
plots can be reproduced. If there is a straight line between the
5.13.3.4 Methods for Non-Single Domain Remanences
natural and laboratory remanence unblocking spectra, pTRM
We have already noted several approaches intended to detect checks pass in both experiments, and the NRM trends directly
and/or remove the effect of pTRM tails (e.g., the pTRM tail to the origin, there may be a way to determine paleointensity
check of Riisager and Riisager, 2001, the IZZI method of Yu from multidomain remanences that are unbiased. Worryingly,
et al., 2004, and the LTD-DHT Shaw method of Yamamoto Shaar et al. (2011b) found that multidomain remanences tend
et al., 2003). Dekkers and B€ ohnel (2006) argued that their to underestimate the ancient field strength even if the total
multispecimen procedure (‘multispecimen parallel differential NRM is used, implying that the NRM decayed in a manner
pTRM method’), which employs a single heating/cooling step not predicted by the Néel theory (Néel, 1949) and that no
with the laboratory field oriented parallel to the NRM, can be paleointensity method can recover the ancient field strength
used on specimens of any domain state. The fundamental for MD remanences.
assumption of this method is the assumed linearity of pTRM Wang and Kent (2013) embellished the method of direct
with applied field, which the authors claim is independent of comparison of unblocking of the NRM with unblocking of the
domain state. As already discussed, this may not be true, par- laboratory TRM in a protocol they call ‘back-zero-forth’ or BZF.
ticularly for multidomain grains. Fabian and Leonhardt (2010) In this protocol, the NRM of the specimens is first treated to
experimentally showed that the original protocol of Dekkers infield cooling along one axis (B), followed by zero-field cool-
and B€ ohnel overestimates paleointensity for intermediate ing (Z) and then by a second infield cooling step with the field
pseudo single domain (PSD) and multi-domain (MD) grain applied along the opposite direction (F). This mimics the
sizes, and they proposed a modified protocol for correcting traditional infield–infield method of K€ onigsberger (1936)
potential domain state-dependent effects including pTRM tail. and embeds the IZ approach of Aitken et al. (1988) and a ZI
On the other hand, de Groot et al. (2012) observed underesti- similar to the Coe (1967) approach (except for the presence of
mates of paleointensity even when the modified protocol was pTRM tails generated by the infield step). After the initial BZF
used. They attributed this behavior to changes in magnetic treatment, the specimens are given a total TRM in a laboratory
domain state during heating. field and the BZF experiment is repeated. The unblocking of
Another approach for eliminating the effect of pTRM tails is the NRM is plotted against the unblocking of the laboratory
to compare unblocking temperature of the original NRM with TRM, similar to the Sbarbori et al. (2009) and Muxworthy
unblocking temperature of a laboratory-induced TRM. This (2010) approaches. This new BZF protocol enables compari-
method was tried first by Wilson (1961) and was recently son of the II, IZ, and ZI protocols. As in the much less labor-
rediscovered by Muxworthy (2010). The so-called Wilson intensive IZZI protocol, the influence of MD grains may be
method compares continuous measurement of the NRM detected and the second BZF experiment may allow the MD
while demagnetizing by heating with that of a full TRM effect to be corrected.
induced in a known field. In order to get around the problem
of pTRM tails, which persist to blocking temperatures higher
5.13.3.5 Use of IRM Normalization
than the temperature at which they were acquired (high-
temperature tails) or, more insidiously, become demagnetized Sometimes, it is difficult or impossible to heat specimens
at temperatures lower than they were acquired (low- because they will alter in the atmosphere of the lab or the
temperature tails), the Wilson method first demagnetizes the material is too precious to be subjected to heating experiments
NRM and then demagnetizes a laboratory remanence and (e.g., lunar samples and some archaeological artifacts). Looking
compares the two unblocking spectra. In this method, the again at Figure 4 suggests an alternative for order of magnitude
measurements are done as a continuous thermal demagnetiza- guesstimates for paleointensity without heating at all. TRM nor-
tion (measuring the remanence at high temperature), and so malized by a saturation remanence (IRM) is quasilinearly
this method is faster than the standard Thellier-type methods related to the applied field up to some value depending on
in which the samples must first be cooled and measured at mineralogy and grain-size population.
room temperature. The problem with this method is that alter- Several investigators (e.g., Cisowski and Fuller, 1986;
ation of the sample may occur in the first demagnetization and Gattacceca and Rochette, 2004; Kletetschka et al., 2004) have
can never be detected. advocated the use of IRM normalization of the NRMs of lunar
Sbarbori et al. (2009) took a different approach. They first and meteorite samples to estimate paleointensity. Cisowski
completed a set of IZZI experiments on a suite of samples from and Fuller (1986) argued that, especially when both rema-
Isla Socorro, Mexico. Some specimens had curved Arai plots nences were partially demagnetized using AF demagnetization,
but passed all the alteration checks and had magnetizations the NRM:IRM ratio gave order of magnitude constraints on
that trended straight to the origin. They interpreted these as absolute paleointensity and reasonable relative paleointensity
estimates. Their argument is based on monomineralic suites of d2 y dy

I ¼ l mBsin y [4]
rocks with uniform grain size. They further argued optimisti- dt 2 dt
cally that multidomain contributions can be eliminated by the
where l is the viscosity coefficient opposing the motion of the
AF demagnetization.
particle through the fluid and I is the moment of inertia.
As can be seen by examining Figure 4, at best, only order of
Nagata (1961) solved this equation by neglecting the inertial
magnitude estimates for absolute paleointensity are possible.
term (which is orders of magnitude less important that the
The monomineralic and uniform grain-size constraints make
other terms) as
even this unlikely. Finally, multidomain TRMs and IRMs do
not behave similarly under AF demagnetization, the former y yo
tan ¼ tan eðmBt=lÞ [5]
being much more stable than the latter. Nonetheless, if 2 2
magnetic uniformity can be established, it may in fact be
where yo is the initial angle between m and B. He further
useful for establishing relative paleointensity estimates as is
showed that by setting l ¼ 8pr3 where r is the particle radius
done routinely in sedimentary paleointensity studies (see
and is the viscosity of water (103 kg m1 s1), the time
Section 5.13.6). The caveats concerning single component
constant g of eqn [5] over which an initial yo is reduced to 1/e
remanences are still valid, and perhaps, complete AF demag-
of its value is
netization of the NRM would be better than a single ‘blanket’
demagnetization step. Moreover, we should bear in mind that l 6
for larger particles, TRM can be strongly nonlinear with applied g¼ ¼ [6]
mB MB
field at even relatively low fields (30 mT) according to the
experimental results of Dunlop and Argyle (1997) (see also where M is the volume normalized magnetization.
Figure 1(a) of Kletetschka et al., 2006). The problem with the Now we must choose values , M, and B. As noted by many
IRM normalization approach is that domain state, linearity of authors since Nagata himself (see discussion by Tauxe et al.,
TRM, and nature of the NRM cannot be assessed. The results 2006), plugging in reasonable values for , M, and B and
are therefore difficult to interpret in terms of ancient fields. assuming isolated magnetic particles, the time constant is
extremely short (ms). The simple theory of unconstrained rota-
tion of magnetic particles in water, therefore, predicts that
5.13.4 Paleointensity with Depositional Remanences sediments with isolated magnetic particles should have mag-
netic moments that are fully aligned and insensitive to changes
Sediments become magnetized in quite a different manner in magnetic field strength. Yet even from the earliest days of
than igneous bodies. Detrital grains are already magnetized, laboratory redeposition experiments (e.g., Johnson et al.,
unlike igneous rocks, which crystallize above their Curie 1948; see Figure 11(a)), we have known that DRM can have
temperatures. Magnetic particles that can rotate freely will a strong field dependence and that DRMs are generally far less
turn into the direction of the applied field, which can result than saturation magnetizations (0.1%). Much of the research
in a DRM. Sediments are also subject to postdepositional on DRM has focused on explaining the strong field depen-
modification through the action of organisms, compaction, dence observed for laboratory redepositional DRM.
diagenesis, and the acquisition of VRM, all of which will affect The observation that DRM is usually orders of magnitude
the magnetization and our ability to tease out the geomagnetic less than saturation and that it appears to be sensitive to
signal. In the following, we will consider the syndepositional changing geomagnetic field strengths implies that the time
processes of physical alignment of magnetic particles in viscous constant of alignment is much longer than predicted by eqn
fluids (giving rise to the primary DRM) and then touch on the [6]. To increase g, one can either increase viscosity or decrease
postdepositional processes important to paleointensity in sed- magnetization.
imentary systems. One can increase g by using the viscosity in the sediment
column (e.g., Denham and Chave, 1982) instead of the water
column. However, something must act to first disrupt the
5.13.4.1 Physical Alignment of Magnetic Moments
alignment of particles prior to burial, so calling on changes in
in Viscous Fluids
viscosity is at best an incomplete explanation.
The theoretical and experimental foundation for using DRM There are several ways of increasing g by reducing the value
for paleointensities is far less complete than for TRM. Tauxe of M hence inhibiting the alignment in the first place. For
(1993) reviewed the literature available through 1992 thor- example, one could use values for M much lower than the
oughly and the reader is referred to that paper for background saturation magnetizations of common magnetic minerals
(see also Valet, 2003). In the last decade, there have been (e.g., Collinson, 1965; Stacey, 1972). However, even using
important contributions to both theory and experiment and the magnetization of hematite, which is two orders of magni-
we will outline our current understanding here. tude lower than magnetite, results in a time constant of align-
Placing a magnetic moment m in an applied field B results ment that is still less than a second.
in a torque G on the particle G ¼ m B. The magnitude of the There are two mechanisms by which the time constant of
torque is given by G ¼ mB sin y, where y is the angle between alignment can be reduced, which account for experimental
the moment and the magnetic field vector. This torque is what results of laboratory redeposition experiments: Brownian
causes compasses to align themselves with the magnetic field. motion and flocculation. Collinson (1965) called on Brownian
The torque is opposed by the viscous drag and inertia and motion to disrupt the magnetic moments by analogy to para-
the equation of motion governing the approach to alignment is magnetic gases. Reasonable parameter assumptions suggest that
30
25 70 70
60
M (10-5 Am2 kg-1)

20 60
Intensity (mA m-1)

50
Inclination (°)
15
40 50
10 30
40
20
5
10 30
0 0
0 4 8 12 0 4 8
(a) B/Bo (b) Salinity (ppt)
Figure 11 (a) Depositional remanence versus applied field for redeposited glacial varves. Bo was the field in the lab. Data from Johnson et al. (1948).
Reproduced from Tauxe L (1993) Sedimentary records of relative paleointensity of the geomagnetic field: Theory and practice. Reviews of
Geophysics 31: 319–354. (b) Relationship of detrital remanent magnetization (DRM) intensity and salinity for synthetic sediment composed of a mixture
of kaolinite and maghemite. Data of Van Vreumingen (1993b). Reproduced from Roberts AP, Tauxe L, and Heslop D (2013) Magnetic paleointensity
stratigraphy and high resolution Quaternary geochronology: Successes and future challenges. Quaternary Science Reviews 61: 1–16.
particles smaller than about 100 nm will be affected by Brow- isolated (e.g., freshwater lakes; see Figure 12(a)) and ones in
nian motion suggesting a possible role in DRM of isolated which flocculation plays a role (e.g., marine environments; see
magnetite grains free to rotate in water. Furthermore, Yoshida Figure 12(b)). For the case of magnetite in freshwater, Brow-
and Katsura (1985) presented experiments on the magneti- nian motion may well be the dominant control on DRM
zation of suspensions in response to applied fields that were efficiency. In saline waters, the most important control on
entirely consistent with a Brownian motion model. Flocculation DRM is the size of the flocs in which the magnetic particles
was fingered by Shcherbakov and Shcherbakova (1983) (see are embedded. In the following, we briefly explore these two
also Katari and Bloxham, 2001) who noted that in saline envi- very different environments.
ronments, sedimentary particles tend to flocculate and that
isolated magnetic particles would be highly unlikely. When mag- 5.13.4.1.1 Nonflocculating environments
netic moments are attached to nonmagnetic ‘fluff,’ it is the net In freshwater, we expect to have isolated magnetic particles
magnetization of the floc that must be used in eqn [6], that is, whose magnetic moments would presumably be a saturation
much smaller than the magnetization of the magnetic mineral remanence. The overwhelming majority of laboratory redepo-
alone. sition experiments have been done in deionized water (e.g.,
The role of water chemistry (e.g., pH and salinity) has been Kent, 1973; Lovlie, 1974) and hence are in the nonflocculating
investigated by several authors since the early 1990s (Katari regime. However, only a few studies have attempted to model
et al., 2000; Lu et al., 1990; Tauxe et al., 2006; van Vreumingen, DRM using a quantitative theory based on Brownian motion
1993a,b). In Figure 11(b), we replot data from one of the van (e.g., Collinson, 1965; King and Rees, 1966; Stacey, 1972;
Vreumingen experiments. The data were obtained by deposit- Yoshida and Katsura, 1985). Here, we outline the theory to
ing a synthetic mixture of kaolinite, illite, and maghemite investigate the behavior of DRM that would be expected from a
under various conditions of salinity. There is an intriguing Brownian motion mechanism (henceforth a Brownian rema-
increase in intensity with small amounts of NaCl followed by nent magnetization or BRM).
a dramatic decrease in intensity, which stabilizes for salinities To estimate the size of particles affected by Brownian
greater than about 4 ppt. motion, Collinson used the equation
Both the increase and the decrease in intensity (solid line)
1 1
can be explained in terms of Brownian motion and flocculation, mBf2 ¼ kT [7]
which is encouraged by increasing salinity. The initial increase in 2 2
intensity with small amounts of NaCl could be the result of the where f is the Brownian deflection about the applied field
maghemite particles adhering to the clay particles, increasing direction (in radians), k is the Boltzmann constant
viscous drag, hence reducing the effect of Brownian motion. The (1.38 1023 J K1), and T is the temperature in kelvin. The
subsequent decrease in intensity with higher salinities could be effect of viscous drag on particles may also be important when
caused by building composite flocs with decreased net the magnetic moments of the particles are low (see Coffey et al.
moments, hence lowering the time constant of alignment. The (1996) for a complete derivation), for which we have
decrease in net moment with increasing flocculation was also
f2 kT
supported by the redeposition experiments of Lu et al. (1990), ¼
d 4pr 3
Katari et al. (2000), and Tauxe et al. (2006).
There are therefore two completely different systems when where d is the time span of observation (say, 1 s). According to
discussing DRM: ones in which magnetic particles remain this relationship, weakly magnetized particles smaller than
Nonflocculating environment (freshwater) Flocculating environment (marine)
Turbulent water Particle Input of magnetic minerals ( ), clays ( )

concentration
Flocculation/
pellitization
Still water or Export to bottom

B
laminar flow
Dilute
suspension
Sediment/water Resuspension/
interface reflocculation
Bioturbation
Gel zone Injestion/excretion
Consolidation Benthic boundary layer
‘Lock-in depth’ Consolidation
Consolidated
Compaction zone Historical layer Compaction
(a) (b)
Figure 12 (a) Schematic drawing of traditional view of the journey of magnetic particles from the water column to burial in a nonflocculating
(freshwater) environment. Magnetic particles are black. Redrawn from Tauxe L (1993) Sedimentary records of relative paleointensity of the geomagnetic
field: Theory and practice. Reviews of Geophysics 31: 319–354. (b) View of depositional remanence in a flocculating (marine) environment.
Redrawn from Tauxe L, Steindorf J, and Harris A (2006) Depositional remanent magnetization: Toward an improved theoretical and experimental
foundation. Earth and Planetary Science Letters 244: 515–529.
about a micron will be strongly affected by Brownian motion. and Bloxham (2001), and Tauxe et al. (2006). We summarize
Particles that have a substantial magnetic moment, however, the current state of the theory in the following.
will be partially stabilized (according to eqn [7]) and might Katari and Bloxham (2001) rearranged eqn [5] by replacing
remain unaffected by Brownian motion to smaller particle time with settling distance l, a parameter that is more easily
sizes (e.g., 0.1 mm). In the case of isolated particles of magne- measurable in the laboratory using the empirical relationship
tite, therefore, we should use eqn [7] and BRM should follow of settling velocity to radius of Gibbs (1985). They got
the Langevin equation for paramagnetic gases, that is,
y yo
tan ¼ tan exp mBl=8:8pr 3:78 [9]
BRM mB kT 2 2
¼ coth [8]
sIRM kT mB As in Nagata (1961), a magnetic moment m making an
To get an idea of how BRMs would behave, we first find m initial angle yo with the applied field B will begin to turn
from Mr (here, we use the results from micromagnetic model- toward the direction of the magnetic field. After time t (or
ing of Tauxe et al. (2002)). Then, we evaluate eqn [8] as a equivalently, settling distance, l), the moment will make an
function of B for a given particle size (see Figure 13(a)). We angle y with the field. Tauxe et al. (2006) showed that the
can also assume any distribution of particle sizes (e.g., that new coordinates of m (x0 , y0 , z0 ) are related to the initial values
shown as the inset to Figure 13(b)) and predict BRM/sIRM for (xo, yo, zo) by
the distribution (blue line in Figure 13(b)). It is interesting to sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 0 1 x2o z
note that BRMs are almost never linear with the applied field x ¼ cos y, y ¼ , and z0 ¼ y0 o [10]
unless the particle sizes are very small. 1 + z2o =y2o yo
BRMs would be fixed when the particles are no longer free to
From the preceding, we can make a simple numerical model
move. The fixing of this magnetization presumably occurs dur-
to predict the DRM for an initially randomly oriented assem-
ing consolidation, at a depth (known as the lock-in depth) where
blage of magnetic moments, after settling through l. For an initial
the porosity of the sediment reduces to the point that the parti-
set of simulations, Tauxe et al. (2006) followed Katari and Blox-
cles are pinned (see Figure 12(a)). Below that, the magnetization
ham, using the viscosity of water, m of 5 fAm2 (where femto
may be further affected by compaction (e.g., Deamer and
(f ) ¼ 10–15), and a settling length l of 0.2 m. In Figure 14(a) and
Kodama, 1990) and diagenesis (e.g., Roberts, 1995).
14(b), we show the predicted DRM curves as a function of
magnetic field and radius. We see that particles, in general, are
5.13.4.1.2 Flocculating environments either nearly aligned with the magnetic field or nearly random
DRM in flocculating environments (saline waters) has been with only a narrow band of radii in between the two states for a
studied in the laboratory by Lu et al. (1990), van Vreumingen given value of B. Increasing B increases the size for which parti-
(1993a,b), Katari et al. (2000), and Tauxe et al. (2006) and cles can rotate into the field, giving rise to the dependence of
theoretically by Shcherbakov and Shcherbakova (1983), Katari DRM intensity on applied field strength. Taking a given particle
1.0
0.9
70
0.8 60 0.8
50
0.7
40 20
BRM/sIRM
0.6
0.6
% Total
15
30
0.5 10
0.4
0.4 5
25 nm 0
0.2 0.3 0 80 180
0.2 Radius (nm)
0.0 0.1
0 40 80 120 0 40 80 120
(a) B (μT) (b) B (μT)
Figure 13 (a) Numerical simulations of Brownian remanent magnetization (BRM) for various sizes of magnetite. (b) BRM simulated for distribution of
particle sizes of magnetite shown in inset.
1.0 1.0
5
10
0.8 0.8
100 μT 15
DRM/sIRM
0.6 0.6
70 20
0.4 0.4
50
30
10 25 μm
0.2 0.2
1 μT
0.0 0.0
0 10 20 30 40 50 0 10 20 30 40 50
(a) Radius (μm) (b) B (μT)
40
9
1 ppt
35 2.2 ± 1.6
Magnetic moment (aAm2)
8
30
2.3 ± 1.6 7
DRM/sIRM
25
5 ppt
2.80 ± 1.6 6
20
5
15 2.90 ± 1.6
4
10
3
5
2
0
0 10 20 30 40 2.0 3.0
(c) B (μT) (d) Equivalent radius (μm)
Figure 14 (a) Results of numerical experiments of the flocculation model using the parameters: l ¼ 0.2 m and the viscosity of water. M/Mo is the DRM
expressed as a fraction of saturation, holding m constant and varying B. For a given field strength, particles are either at saturation or randomly
oriented, except for within a very narrow size range. (b) Same as (a) but plotted versus applied field (B). (c) Results of settling experiments as a function
of field (B) in a flocculating environment. The assumed mean and standard deviations of truncated log-normal distributions for floc radii are shown
in the legends and are indicated using the different line styles in the figure. (d) m versus equivalent radius for composite flocs as in inset. Line given by
polynomial fit m ¼ ar2 + br + c where a ¼ 3.61 107, b ¼ 1.2 1012, c ¼ 2.1 1019 is based on a fundamental floc of 1 mm with a measured
saturation remanence. Redrawn from Tauxe L, Steindorf J, and Harris A (2006) Depositional remanent magnetization: Toward an improved theoretical
and experimental foundation. Earth and Planetary Science Letters 244: 515–529.
size and predicting DRM as a function of the applied field respect to the applied field. The theories of Brownian motion,
(Figure 14(b)) determines the opposite behavior for DRM which predicts low DRM efficiency for the smallest particles
than the Brownian motion theory (Figure 13) in that the larger increasing to near saturation values for particles around 0.1 mm
the floc size, the weaker the DRM and also the more linear with and composite flocs theory, which predicts decreased
DRM efficiency for larger floc sizes, can therefore explain the horizontal)). Group H magnetic moments attempt to align
experimental data of van Vreumingen (1993a,b) shown in with the field, but the net moment is biased shallow. When
Figure 11(b). the flocs become too large to maintain equilibrium with the
The flocculation model of DRM makes specific predictions field, their net magnetization is essentially zero. Therefore, the
that can in principle be tested if the model parameters can be net magnetic declination of group H flocs tracks the field
estimated or controlled. Tauxe et al. (2006) tested the theory azimuth, but the net inclination is near zero (Figure 15).
by dispersing natural sediments in settling tubes to which The net magnetic moments of both groups of flocs contribute
varying amounts of NaCl had been introduced. Prior to dis- to the observed DRM (Figure 15). In reality, a floc is not
persal, each specimen of mud was given a saturation IRM. They expected to behave according to the simple scheme of particles
measured DRM as a function of floc size (increasing salinity aligning through action of a magnetic torque, as envisaged in
enhanced floc size) and the applied field (see Figure 14(c)). In the model, but it would instead follow a complicated trajectory
general, their results suggest the following: (1) The higher the under the simultaneous influence of magnetic and hydro-
NaCl concentration, the lower the net moment (confirming dynamic torques (Heslop, 2007). However, the average
previously published efforts); (2) the higher the salinity, the behavior of an ensemble of flocs can be approximated by this
faster the particles settled (a well-known phenomenon in simple conceptual model. Mitra and Tauxe (2009) used this
coastal environments; see, e.g., Winterwerp and Kestern, model to successfully simulate the laboratory results shown in
2004); (3) the higher the applied field, the higher the DRM, Figure 11(b).
although a saturation DRM (sDRM) appears to be nearly
achieved in the 1 ppt NaCl set of tubes by 30 mT (Figure 14(c));
5.13.4.2 Postdepositional Processes
and (4) the relationship of DRM to B was far from linear
with applied field in all cases. Moreover, in the Katari and It appears that by combining the effects of Brownian motion
Bloxham (2001) model of DRM, a single magnetic particle for nonflocculating environments and a composite floc model
is assumed to be embedded in each floc; hence, the mag- for flocculating environments, we are on the verge of a quan-
netization of the flocs is independent of floc size. In this titative physical theory that can account for the acquisition of
view, the sDRM should equal the sum of all the individual depositional remanence near the sediment/water interface. At
flocs, that is, sIRM in the case of these experiments. sDRM some point after deposition, this DRM will be fixed because no
was well below sIRM in all experiments (see, e.g., Figure 14(c)) further physical rotation of the magnetic particles in response
and no Katari–Bloxham-type model can account for the to the geomagnetic field is possible. The depth at which
results. moments are pinned is called the lock-in depth. If lock-in
Tauxe et al. (2006) modified the simple theory of Katari and depth is selective and some magnetic particles would be fixed
Bloxham (2001) by incorporating the understanding of floc- while others remain free, there will be some depth (time)
culation from the extensive literature on the subject. In nature, interval over which remanence is fixed, resulting in some tem-
flocs are formed by coalescing of ‘fundamental flocs’ into poral smoothing of the geomagnetic signal. Physical rotation
composite flocs. Each fundamental floc would have tiny mag- of particles in response to compaction can also change the
netic particles adhering to them and would have the sIRM magnetic remanence. Other processes not involving post-
imparted prior to settling. As the composite flocs grow by depositional physical rotation of magnetic particles including
chance encounters with other flocs, the net moment of the ‘viscous’ (in the sense of magnetic viscosity) remagnetization
composite floc will be the vector sum of the moments of the and diagenetic alteration resulting in a chemical remanence
fundamental flocs (see, e.g., inset to Figure 14(d)). They mod- may also modify the DRM. All of these processes influence the
eled the magnetization of flocs as a function of floc radius intensity of remanence and hamper our efforts to decipher the
(assuming a quasispherical shape) through Monte Carlo sim- original geomagnetic signal. We will briefly discuss the effects
ulation, an example of which is shown in Figure 14(d). By specific to sediments in the following; chemical alteration and
choosing reasonable log-normal distributions of flocs for set- viscous remagnetization effect both TRMs and DRMs and will
tling tube, their model predicts the curves shown in Figure 14 be addressed in Section 5.13.5.
(c), in excellent agreement with the redeposition data. The ‘standard model’ of depositional remanence (DRM)
A critical aspect of the work of Van Vreumingen (1993a,b) acquisition was articulated, for example, by Verosub (1977)
remained unexplained by the theoretical work of Tauxe et al. and Tauxe (1993). In this view, detrital remanence is acquired
(2006) and that was the dependence of the inclination on the by locking in different grains over a range of depths. This
salinity (dashed line in Figure 11). To explain this, Mitra and phased lock-in leads to both significant smoothing and an
Tauxe (2009) reprised the idea of ‘plates and spheres’ of King offset between the sediment/water interface and the fixing of
(1955) and incorporated it into the model of Tauxe et al. the DRM. Many practitioners of paleomagnetism still adhere to
(2006) using the numerical approximation of Heslop (2007). this concept of DRM, which stems from the early laboratory
Instead of using plates and spheres, however, Mitra and Tauxe redeposition experiments, which were carried out under non-
(2009) assumed a population of slightly elongate flocs with a flocculating conditions (see Section 5.13.4.1). Several studies
continuous size distribution (Figure 15). They separated the on natural marine sediments (e.g., deMenocal et al., 1990;
sizes into two groups: one was small enough to respond Kent and Schneider, 1995; Lund and Keigwin, 1994; see also
mainly to magnetic torques (group M) with a net moment Channell et al., 2004) are frequently cited, which suggest a
essentially parallel to the applied field and the other was con- high degree of mobility of magnetic particles after deposition
trolled by hydrodynamic torques (group H attaining hydrody- resulting in sedimentary smoothing and delayed remanence
namic stability while settling (with long axes on average acquisition.
H
H
DI H
NM = 21 NM = 12 NM = 7
DI DI
DI
M
NH = 9 NH = 13 M NH = 17
M
M M M
1.0
1400
2000 0.9
1.0
0.9
2000 1.0
0.9
0.8 0.8 0.8
Floc count
1500
0.7
1000 0.7
1500 0.7
0.6 0.6 0.6
0.5 0.5
H 0.4 H 0.4
300
H 0.5
1000 0.3 600 0.3

1000 0.4
0.3
150
0.2 0.2
0.2
0.1 0.1
500 0.0 0.0
500 0 15 30
0.1
0.0
200
0 20 40 60 80 100 0 20 40 60 80 100 120 0 200 400 600 800

(a) Floc size (μm) (b) Floc size (μm) (c) Floc size (μm)
Figure 15 Numerical simulation of how paleomagnetic intensity and inclination in sediments are affected by floc size. Lower: the solid blue and red
dashed lines represent floc distributions in simulated M and H groups, respectively, where M represents flocs that respond to magnetic torques and
H represents flocs that are dominated by hydrodynamic torques. A northward-directed applied field with 45 inclination and 45 mT intensity was
used for the simulations. In equal area projections of floc moments for the M (upper) and H (lower) groups, no distinction is made between hemispheres;
the lower plots are normalized by the maximum concentration of flocs. High and low concentrations are indicated by darker and lighter shading,
respectively, and homogeneity (or lack of alignment) of moment direction is indicated by midtone shading over the entire stereographic projection (as in
(c)). For the M group, small dark areas indicate that the majority of moments are well aligned with the applied field. Upper insets: schematic
representations of three cases. Left: blue dots and red squares in stereographic projections represent individual floc directions from the M and H groups,
respectively. The orange cross is the applied field direction. Right: blue and red arrows represent the recorded paleomagnetic inclinations for the
M and H groups. The black arrow is the resultant, and DI is the inclination flattening. (a) For small floc sizes, most flocs are in group M; few flocs are in
group H, and DI is small. (b) For larger floc sizes, more flocs are in group H and DI increases. (c) For the largest floc sizes, most flocs are in group H
and are so large that they are oriented randomly with respect to the field. The small number of group M flocs is sufficient for DI to become small.
The net magnetic moment decreases from (a) to (c) because of the less efficient alignment of increasingly larger flocs. Adapted from Roberts AP, Chang L,
Heslop D, Florindo F, and Larrasoaña J (2012) Searching for single domain magnetite in the “pseudo-single-domain” sedimentary haystack:
Implications of biogenic magnetite preservation for sediment magnetism and relative paleointensity determinations. Journal of Geophysical Research
117: B08104; Mitra R and Tauxe L (2009) Full vector model for magnetization in sediments. Earth and Planetary Science Letters 286: 535–545.
The work of deMenocal et al. (1990) called for a deep lock- The most quoted examples of significant smoothing in
in depth of up to 16 cm for marine sediments based on a natural sediments are those of Lund and Keigwin (1994) and
compilation of deep-sea sediment records with oxygen iso- Kent and Schneider (1995). On close examination, the evi-
topes and the Matuyama–Brunhes boundary (MBB). However, dence is weak. Lund and Keigwin (1994) postulated that the
Tauxe et al. (1996) updated the compilation with twice the paleosecular variation (PSV) record of the Bermuda Rise, west-
number of records and, using the same logic, concluded that, ern North Atlantic Ocean, was systematically subdued with
on average, the magnetization is recorded within the top few respect to the PSV recorded in Lake St. Croix stemming from
centimeters. This is supported by coincidence of 10Be produc- the observed difference in sediment accumulation rate, the
tion peak and relative paleointensity low during the Iceland Lake St. Croix record having been deposited at a rate several
Basin excursion in two Ocean Drilling Program (ODP) cores times that of the Bermuda Rise record. They suggested that
from the Atlantic Ocean (Knudsen et al., 2008). smoothing the Lake St. Croix data with a 10 or 20 cm moving
Several papers have revived the deep lock-in debate (e.g., average window reproduced the Bermuda Rise data with high-
Bleil and von Dobeneck, 1999; Channell et al., 2004). The frequency features smoothed out and the amplitude of varia-
former used a complicated lock-in model to explain results tion significantly reduced. However, they ignored the age con-
not observed anywhere else (substantial reversely magnetized straints available in the original St. Croix record. Tauxe et al.
intervals in apparently late Brunhes age equatorial sediments). (2006) showed that a substantially better fit of the Bermuda
The latter noted that in North Atlantic drift deposits, the mid- data could be achieved when the available age constraints are
point of the MBB is ‘younger’ isotopically than records with used and no smoothing was required by the data.
lower sedimentation rates, implying a deep lock-in. However, The study of Kent and Schneider (1995) showed three
drift deposits by nature collect sediments from a large catch- records of relative paleointensity across the MBB and inter-
ment area. A particular bit of plankton from the surface waters preted these in terms of sedimentary smoothing. These records
of the North Atlantic will be transported along the bottom for came from low and moderate sediment accumulation rates.
some time before it finds a permanent home in the drift. The Hartl and Tauxe (1996) augmented the database of relative
age offset between the isotopic (acquired at the surface) and paleointensity records spanning the MBB with additional ten
magnetic ages (acquired at the final point of deposition) obvi- records obtained from a wide range of sediment accumulation
ates the need for a deep lock-in depth. rates and showed that the single low sedimentation rate core of
Kent and Schneider (V16-58) most probably had a poorly 5.13.4.4 Normalization
constrained timescale. Once again, little, if any, smoothing of
Until now, we have considered only how magnetic moments
sedimentary paleointensity records is required.
behave when placed in a magnetic field and are allowed to
As sediments lose water and consolidate, compaction can
rotate freely. Paleointensity studies in sediments make the a
have a strong effect on DRM intensity (e.g., Anson and
priori assumption that DRM is quasilinear with the applied
Kodama, 1987). Consolidation is a continuous process starting
field (although as we have seen in the section on DRM theory,
from the sediment/water interface when sedimentary particles
that is only true under certain circumstances). However, we
first gel (see, e.g., Figure 12(b)) and continuing until the
have not yet considered the effect of changing the magnetic
sediment is completely compacted, perhaps as deep as hun-
content of the sediment, which of course will have a profound
dreds of meters. The effect on magnetic remanence depends on
effect on the intensity of the remanence. Such changes must
volume loss during compaction, which depends largely on clay
be compensated for through some sort of normalization pro-
content, so clay-rich sediments will have the largest effect.
cess (see, e.g., Kent, 1982). Methods of normalization were
Recently, Suganuma et al. (2010, 2011) revived again the
reviewed thoroughly by King et al. (1983) and Tauxe (1993),
issue of lock-in depth and smoothing of remanence in post-
but there have been a few contributions to the subject pub-
depositional processes. They found a lock-in depth of 15 cm
lished since. Here, we briefly summarize the most commonly
from within-core offsets of paleointensity minima at the
used methods of normalization.
Matuyama–Brunhes polarity transition between normalized
Most studies use some easily measured bulk magnetic
remanence intensity and paleointensity estimated from cosmo-
parameters such as saturation remanence ( Johnson et al.,
genic nuclide 10Be in three sediment cores from the Pacific
1948), magnetic susceptibility (w, Harrison, 1966), and ARM
Ocean with relatively low sedimentation rates, 20 m My1
( Johnson et al., 1975), which will compensate for changes in
or less. Furthermore, they suggested that a Gaussian lock-in
concentration of the magnetic minerals in a relatively crude
function, in which the rate of remanence lock-in increases in
way. Levi and Banerjee (1976) proposed a more sophisticated
the middle of the lock-in zone, can explain the observation
approach in which the natural remanence was partially demag-
that indicates relatively little smoothing of the geomagnetic
netized as was the anhysteretic remanent normalizer to ensure
signals. When using an exponential lock-in function, which
that the same coercivity fraction was used to normalize the
was often used for simulating remanence acquisition with
remanence as was carrying the NRM. Following up on this
sediment compaction and dewatering, the geomagnetic signals
line of reasoning, Tauxe et al. (1995) suggested that the NRM
are overly smoothed.
be normalized by ARM in a manner similar to the KTT exper-
The inevitable conclusion from this section is that, despite
iments for thermal remanences using a technique known as
the long history of research in the matter, much remains
‘pseudo-Thellier’ normalization. King et al. (1983) reminded
to be learned for remanent magnetization acquisition pro-
us that ARM itself is a strong function of concentration with
cesses of sediments. Nonetheless, sedimentary paleointensity
higher magnetite concentrations being less efficient at ARM
remains a useful tool in high-resolution Quaternary geochro-
acquisition than lower concentrations. As a result, zones with
nology, as summarized in a recent review by Roberts et al.
varying concentrations will be normalized differently (in effect,
(2013).
different a’s in eqn [1]) and violate the fundamental assump-
tions of the method. More recently, Brachfeld and Banerjee
(2000) proposed a secondary correction for normalized inten-
5.13.4.3 Note on Aeolian Deposits
sity that attempted to remove some of the nonlinear effects of
The theoretical and experimental foundations of relative the normalization process. Tauxe and Wu (1990) argued that if
paleointensity studies have all been done on waterborne the power spectrum of the normalizer was coherent with the
sedimentary deposits. Nonetheless, it is clear that aeolian normalized remanence, the normalization process was insuffi-
sediments, in particular, loess, can retain an NRM that cient. Constable et al. (1998) expanded on this idea, suggesting
appears to record the direction of the geomagnetic field that the normalizer most coherent with the remanence should
(e.g., Heller and Liu, 1982). Details of how the geomagnetic be used.
field is impressed on loess deposits are not well known, but One of the important implications of the composite
mechanisms must include viscous remanence, pedogenic floc model of DRM of Tauxe et al. (2006) described in
modification (chemical remanence), and perhaps also a Section 5.13.4.1 is that current methods of normalizing sedi-
remanence acquired at dry deposition and postdepositional mentary records for changes in magnetic grain size and concen-
wetting (Spassov et al., 2003; Wang and Lovlie, 2010; Zhao tration do not account for changes in floc size and hence will be
and Roberts, 2010). What controls the intensity of rema- only partially effective in isolating the geomagnetic contribution
nence acquired during deposition of windblown dust is to changes in DRM. This has practical implications in the role of
unknown, yet there have been several attempts to use the climate in influencing relative paleointensity records. For exam-
normalized remanence in loess as a proxy for geomagnetic ple, changes in the clay content could well lead to differences in
intensity variations (e.g., Liu et al., 2005; Pan et al., 2001; flocculation, which in turn could influence paleointensity with
Zhu et al., 1994). These studies rely heavily on the theoret- no observable change in the magnetic mineralogy apart from a
ical and experimental work developed for water-lain sedi- change in concentration. Other ‘stealth’ influences could be
ments (see also Spassov et al., 2003); theoretical and miniscule changes in salinity of lakes, which could result in
experimental efforts must be carried out for the mechanism profound changes in the paleointensity recorded, with no
involved in remanence acquisition in loess. means of detecting it. However, in stable environments with
only small changes in magnetic mineralogy and concentration, external field. The lower the value of t, the quicker the
we can only hope that the normalization procedures chosen will approach to equilibrium.
give records that are reasonably linear with the applied field. According to eqn [2], relaxation time varies with external
Recently, it was revealed that magnetic mineral assemblages factors such as temperature (as seen in Figure 2) and applied
of pelagic sediments are often dominated by biogenic magne- field B and with factors specific to the magnetic particle such as
tite (Roberts et al., 2012; Yamazaki and Ikehara, 2012). Varia- volume and its intrinsic resistance to changing external fields
tions in the relative proportion of biogenic and terrigenous reflected in its anisotropy constant K. In any natural substance,
magnetic mineral components emerge as variations of ARM/ there will be a range of values for t that could span from
SIRM ratio (Yamazaki and Ikehara, 2012), reflecting differ- seconds (or less) to billions of years. It is interesting to note
ences in grain-size distribution. Normalized intensities (with that a TRM is in effect the equilibrium magnetization (see, e.g.,
either ARM or IRM) sometimes correlate with ARM/SIRM Yu and Tauxe, 2006) and TRMs will only be subject to mag-
ratios (Hoffmann and Fabian, 2009; Xuan and Channell, netic viscosity if the field changes. DRMs, however, are typi-
2008a; Yamazaki et al., 2013), which suggests that neither cally one or two orders of magnitude less than the TRM that
ARM nor IRM can compensate well for variations in relative would be acquired in the same field and hence are almost
proportion of biogenic and terrigenous components and that never in equilibrium and therefore will nearly always be sub-
such sediments are not suitable for relative paleointensity esti- ject to viscous remagnetization, depending on the spectrum of
mation. Using a regional group of sediment cores, Hoffmann t values (see Kok and Tauxe, 1996a, for discussion).
and Fabian (2009) tried to correct for the influence of
lithologic variations including those appearing as ARM/SIRM
ratio variations. 5.13.5.2 Chemical Alteration
Geologic materials form in one environment (e.g., extruding
red hot from the mouth of a volcano!) and wind up in quite
5.13.5 Remagnetization different environments. Inevitably, they will break down as
part of the rock cycle. Magnetic minerals are no exception
Theoretical treatment of how rocks get magnetized and how and growth, alteration, and dissolution of magnetic minerals
that magnetization might be used for paleointensity studies change the original remanence. The magnetization that is fixed
assume that the remanence was blocked either thermally by growth or alteration of magnetic minerals is termed chem-
(Section 5.13.3) or depositionally (Section 5.13.4). Yet almost ical remanent magnetization, and while this too is controlled
no NRM remains completely unchanged for long. Thermody- in part by the external magnetic field, the theory of how to
namics teaches us that all substances out of equilibrium with normalize CRM to retrieve the geomagnetic signal has never
their environments will approach equilibrium as the energy been properly developed. In general, paleointensity studies
available permits. Magnetic particles out of equilibrium with strive to recognize CRMs and exclude such remanences from
the magnetic field in which they sit are subject to magnetic interpretation.
viscosity. If they are out of chemical equilibrium, they will alter
chemically. The former results in the acquisition of a viscous
remanence and the latter a chemical one. These are discussed 5.13.6 Evaluating Paleointensity Data
in more detail in Chapter 5.08. We will briefly describe their
importance to paleointensity in the following.
5.13.6.1 Thermally Blocked Remanences
A well-done paleointensity experiment allows us to test (1)
whether the NRM was a single component magnetization, (2)
5.13.5.1 Magnetic Viscosity whether alteration occurred during laboratory reheating,
Returning to Figure 2, we see that magnetic moments can (3) whether blocking and unblocking were reciprocal, and
respond to external fields even if the magnetic crystal itself is (4) whether the TRM is a linear function of the applied field.
fixed on timescales determined by the magnetic relaxation time Parameters can be calculated to provide measures of overall
t. When the relaxation time is short relative to the time span of quality (scatter about the best-fit line, distribution of tempera-
observation, the magnetization is in equilibrium with the ture steps, fraction of the NRM, etc.) of a given experiment.
external field and the particles are called ‘superparamagnetic.’ Some useful parameters are listed for convenience in Table 1.
This means that magnetic particles have sufficient thermal This subject has been debated rather intensely in the recent
energy to overcome intervening energy barriers and flip their literature (e.g., Biggin et al., 2003; Kissel and Laj, 2004;
magnetic moments from one easy direction to another. The Paterson et al., 2010, 2012; Shaar and Tauxe, 2013; Tauxe
energy barrier is in part controlled by the external field, with a and Staudigel, 2004), and it appears that as yet there is no
lower threshold into the direction of the applied field than out consensus on what constitutes a standard method for deter-
of it. Therefore, magnetic moments will tend to ‘pool’ in the mining reliability. This is a rapidly developing field, so stay
direction of the applied field. tuned.
The magnetization that is acquired in this isochemical,
isothermal fashion is termed viscous remanent magnetization.
5.13.6.2 Depositional Remanences
With time, more and more grains will have sufficient thermal
energy to overcome anisotropy energy barriers and flip their How can sedimentary relative paleointensity data be judged?
magnetizations to an angle more in alignment with the Here are some thoughts:
Table 1 Parameters
Parameter Name Definition/notes References
jbj Best-fit slope Slope of pTRM acquired versus NRM remaining Coe et al. (1978)
Banc Ancient field estimate jbj times the laboratory field Coe et al. (1978)
b Scatter parameter Standard error of the slope over jbj Coe et al. (1978)
Q Quality factor Combines several parameters Coe et al. (1978)
VDS Vector difference sum Sum of vector differences of sequential demagnetization steps Tauxe (1998)
Fvds Fraction of the total NRM Total NRM is VDS Tauxe and Staudigel
(2004)
di pTRM check Difference between pTRM and pTRM check step at Ti Tauxe and Staudigel
(2004)
Tmax Maximum blocking Highest step in calculation of jbj Tauxe and Staudigel
temperature P (2004)
DRATS Difference ratio sum di normalized by pTRM (Tmax) Tauxe and Staudigel
(2004)
NpTRM Number of pTRM checks Below Tmax Tauxe and Staudigel
(2004)
Di pTRM tail check Difference between NRMs remaining after the first and second zero-field Tauxe and Staudigel
steps (2004)
MD% Percent maximum difference 100 maximum value of Di/VDS Tauxe and Staudigel
(2004)
T Orientation matrix Matrix of sums of squares and products of demagnetization data Tauxe (1998)
ti Eigenvalues of T t1 > t2 > t3 Tauxe (1998)
Vi Eigenvectors of T Best-fit direction is V1 Tauxe (1998)
MAD Maximum angle of deviation tan1(√ (t22 + t23)/t1) Kirschvink (1980)
DANG Deviation angle Angle between origin and V1 Tauxe and Staudigel
(2004)
Source: Tauxe L (2006) Long term trends in paleointensity: The contribution of Deep Sea Drilling Program (DSDP)/Ocean Drilling Program (ODP) submarine basaltic glass
collections. Physics of the Earth and Planetary Interiors 244: 515–529.
1. The NRM must be carried by a detrital phase of high mag- both). Lake sediments are more difficult to date and rely
netic stability. Furthermore, the portion of the natural rem- for the most part on radiocarbon ages.
anent vector used for paleointensity should be a single,
well-defined component of magnetization. The nature of
the NRM can be checked with progressive demagnetization
using AF and thermal techniques. Supplementary informa- 5.13.7 Current State of the Paleointensity Data
tion from hysteresis and rock magnetic experiments can
5.13.7.1 Paleomagnetic Databases
also be useful.
2. The detrital remanence must be an excellent recorder of the There has been an enormous effort in collecting and preserving
geomagnetic field and exhibit no inclination error, and if paleomagnetic data since the early 1960s (e.g., Irving, 1964),
both polarities are present, the two populations should be but since the 1987 meeting of the IAGA in Vancouver, the
antipodal. The associated directional data should therefore effort has been more concerted with seven IAGA-sponsored
be plotted on equal area projections (or at least histograms databases. Absolute paleointensities have been assembled in
of inclination) whenever possible. a series of compilations by Tanaka and Kono (1994), Tanaka
3. Large changes in concentration (more than about an order et al. (1995b), Perrin and Shcherbakov (1997), Perrin et al.
of magnitude) and changes in magnetic mineralogy or (1998), Perrin and Schnepp (2004), Tauxe and Yamazaki
grain size should be minimized. These changes can be (2007), and Biggin (2010). In their assessment of the most
detected with the use of biplots of, for example, IRM versus recent release of the absolute paleointensity database, Perrin
w. Such biplots should be linear, with low scatter. and Schnepp (2004) stated:
4. The relative paleointensity estimates that are coherent with
bulk rock magnetic and lithologic parameters should be For the future, a harmonization or a combination of all IAGA
databases would be desirable. Furthermore, the input of raw data
treated with caution. Coherence can be assessed using stan-
at the specimen level would be useful in order to allow reinterpre-
dard spectral techniques. tation of data with more developed and sophisticated methods
5. Records from a given region should be coherent within the based on our increasing understanding of rock magnetism.
limits of a common timescale. Whenever possible, dupli-
cate records should be obtained and compared. In order to address this widely felt sentiment, the MagIC
6. For a relative paleointensity record to have the maximum database was created and is accessible at http://earthref.org/
utility, it should have an independent timescale. Many MAGIC/. This database has merged several of the existing
deep-sea sediment records are calibrated using oxygen iso- IAGA databases and allows for data ranging from original
topic curves or magnetostratigraphic age constraints (or magnetometer output (including magnetometer, hysteresis,
thermomagnetic, susceptibility, and other measurements) and included is the standard deviation of the replicate measure-
their interpretations. Detailed descriptions of the data are pos- ments from a given cooling unit. Because there are many
sible by using ‘method codes.’ The database is constantly useful reliability criteria for judging paleointensity data (see
updated to include data as they are published and existing Section 5.13.6), efforts should be made to update the contri-
entries are inspected for errors and corrected. The absolute butions in the MagIC database to include as many of these as
paleointensity database can be accessed through the MagIC are available.
website by searching on the method code ‘LP–PI–TRM’ (lab Perrin and Schnepp (2004) ably summarized the character-
protocol–paleointensity–thermal remanence). The search istics of the PINT03 database and we will not repeat their
for data for the period from 50 ka to 200 Ma can be retrieved analysis here. Nonetheless, it is useful to reiterate that most
at http://earthref.org/MAGIC/search#1382302391017. We of the data come from the last million years and are from the
will call this PINT13 in the following. northern hemisphere. The temporal bias is particularly egre-
The data in the databases include information on geo- gious when only the most ‘reliable’ data are used (i.e., that
graphic location (see map of data locations in Figure 16(a)), employed TRM normalization with pTRM checks).
rock type and age of the sampling sites, type of paleointensity There was no IAGA database for relative paleointensity data
experiment, the remanence vector (including direction if avail- (except those included in the TRANS database) prior to the
able), and summary statistics such as the standard deviation of compilation of Tauxe and Yamazaki (2007). As a step toward
replicate specimens from a given cooling unit. In many cases, rectifying this problem, they summarized the published litera-
the database includes even the measurement data. If no mea- ture with relative paleointensity data in Table 2. Locations of
surement data are available, we may know that, for example, records are shown in Figure 16(b). There are data from nearly
pTRM checks were performed (e.g., studies with pTRM listed 100 references and contributed them to the MagIC database
under alteration check), but we do not know whether they (obtainable individually through the original reference or col-
‘passed’ any particular criterion. The only reliability criterion lectively through this reference). We refer to this compilation
0⬚ 60⬚ 120⬚ 180⬚ 240⬚ 300⬚ 0⬚
60⬚ 60⬚
30⬚ 30⬚
0⬚ 0⬚
-30⬚ -30
-60⬚ -60
(a) 0⬚ 60⬚ 120⬚ 180⬚ 240⬚ 300⬚ 0⬚

0⬚ 60⬚ 120⬚ 180⬚ 240⬚ 300⬚ 0⬚
60⬚ 60⬚
30⬚ 30⬚
0⬚ 0⬚
-30⬚ -30⬚
-60⬚ -60⬚
(b) 0⬚ 60⬚ 120⬚ 180⬚ 240⬚ 300⬚ 0⬚
Figure 16 Locations of all paleointensity data in the (a) absolute (PINT13) and (b) relative (SEDPI06) databases compiled for this paper.
Table 2 Summary of relative paleointensity records
Record Latitude/ Age range Dating Comp. TS References

longitude methods
10-pc03a 47/6 23–115 ka RPI Stoner et al. (2002) Martinson et al. (1987) Channell and Kleiven (2000),
Stoner et al. (2002)
1010a 30/118 19–2036 ka POL, MS Guyodo and Valet Bassinot et al. (1994), Hayashida et al. (1999),
(2006) Cande and Kent Leonhardt et al. (1999)
(1995)
1021a,b 39/128 13–1562 ka POL Guyodo and Valet Cande and Kent Guyodo et al. (1999),
(1999) (1995), Bassinot Leonhardt et al. (1999)
et al. (1994)
1089a 41/10 20–578 ka d18O Stoner et al. (2002) Martinson et al. (1987) Stoner et al. (2003)
1092 46/7 5.9–3.5 Ma POL Cande and Kent (1995) Evans and Channell (2003)
1101a,b 64/70 706–1105 ka POL, Guyodo and Valet Cande and Kent Guyodo et al. (2001)
MS, (2006) (1995), Shackleton
d18O et al. (1990)
21-pc02a 41/8 0–81 ka d18O Stoner et al. (2002) Martinson et al. (1987) Channell and Kleiven (2000),
Stoner et al. (2002, 2003)
305-a5a 53/106 0–11 ka Martinson et al. (1987) Peck et al. (1996)
337-t2a 53/106 13–84 ka RPI Martinson et al. (1987) Peck et al. (1996)
522a 26/5 22.8–34.7 Ma POL Cande and Kent (1995) Tauxe and Hartl (1997)
606aa 37/323 773–792 ka POL Hartl and Tauxe Cande and Kent (1995) Clement and Kent (1986)
(1996)
609ba 50/336 777–825 ka POL Hartl and Tauxe Cande and Kent (1995) Hartl and Tauxe (1996)
(1996)
664da 0/336 670–807 ka POL Hartl and Tauxe Cande and Kent (1995) Valet et al. (1989
(1996)
665aa 3/340 770–817 ka POL Hartl and Tauxe Cande and Kent (1995) Valet et al. (1989
(1996)
767a,b 5/124 601–1518 ka POL Guyodo and Valet Cande and Kent (1995) Guyodo and Valet (2006),
(2006) Schneider et al. (1992)
767ba 5/124 759–829 ka POL Guyodo and Valet Cande and Kent (1995) Guyodo and Valet (2006),
(2006) Schneider et al. (1992)
768aa 8/121 5–94 ka d18O, 14C Guyodo and Valet Linsley and Thunnell Schneider and Mello (1996)
(1999) (1990), Tiedemann
et al. (1994)
768ba 8/121 9–130 ka d18O, 14C Guyodo and Valet Linsley and Thunnell Schneider and Mello (1996)
(1999) (1990), Tiedemann
et al. (1994)
769a 9/121 5–831 ka d18O, 14C Guyodo and Valet Linsley and Thunnell Schneider (1993), Schneider
(1999) (1990), Shackleton and Mello (1996)
et al. (1990),
Tiedemann et al.
(1994)
803aa,b 2/161 783–2178 ka POL Guyodo and Valet Cande and Kent (1995) Kok and Tauxe (1999)
(2006)
803ba,b 2/161 1487–2786 ka POL Guyodo and Valet Cande and Kent (1995) Kok and Tauxe (1999)
(2006)
804ca 1/161 1448–1470 ka POL Hartl and Tauxe Cande and Kent (1995) Hartl and Tauxe (1996)
(1996)
805ba 1/160 770–821 ka POL Hartl and Tauxe Cande and Kent (1995) Hartl and Tauxe (1996)
(1996)
848–851a 2/110 34–4035 ka POL Guyodo and Valet Cande and Kent (1995) Valet and Meynadier (1993)
(1999)
877a 54/148 9.4–11.3 Ma POL Cande and Kent (1995) Bowles et al. (2003)
882b 50/168 0–200 ka Shackleton et al. Okada (1995)
(1990)
(Continued)
Table 2 (Continued)

longitude methods
883 51/168 15–200 ka d18O Shackleton et al. Roberts et al. (1997)

(1990)
884 51/168 15–200 ka MS Shackleton et al. Roberts et al. (1997)
(1990)
884a,b 51/168 9.9–10.3 Ma POL Cande and Kent (1995) Roberts and Lewin-Harris
(2000)
983a,b 60/24.1 0–1889 ka POL, Guyodo and Valet Cande and Kent Channell (1999), Channell et al.
d18O (2006), Laj et al. (1995), Shackleton (1997, 1998, 2000, 2002,
(2000) et al. (1990) 2004)
984a,b 60.4/23.6 0–2151 ka POL, Cande and Kent Channell et al. (1998, 2002,
d18O (1995), Shackleton 2004)
et al. (1990)
14
Chewaucan 43/121 65–102 ka C, ThL Roberts et al. (1994)
Sed-17aKa 25/17 9–224 ka d18O 1984 Haag (2000)
ch88-10pa 30/73 10–70 ka Martinson et al. (1987) Schwartz et al. (1996)
ch88-11p 31/74 10–70 ka Martinson et al. (1987) Schwartz et al. (1996)
ch89-1p 31/75 12–71 ka Martinson et al. (1987) Schwartz et al. (1998)
con-01-603-2 54/109 10–200 ka POL, Demory et al. (2005)
14
C,
con-01-604-2 52/106 0–60 ka ARM Demory et al. (2005)
con-01-605-3 52/105 0–40 ka ARM Demory et al. (2005)
ded8707a 40/14 10–60 ka ash Guyodo and Valet 1984 Tric et al. (1992)
(1999)
ded8708 40/14 40–80 ka ash 1984 Tric et al. (1992)
e113pa,b 2/159 4–380 ka d18O Guyodo and Valet Bassinot et al. (1994) Tauxe and Wu (1990)
(2006)
hu90-013-012 59/47 10–110 ka RPI, Stoner et al. (1995)
d18O,
14
C
hu90-013-013 58/48 10–110 ka RPI, 14C Stoner et al. (1995)
hu91-045-094 50/45 10–110 ka RPI, Stoner et al. (1995)
d18O,
14
C
ket8251a 40/14 8–95 ka ash Guyodo and Valet 1984 Tric et al. (1992)
(1999)
kh73-4-7 3/165 0–2000 ka POL Sato and Kobayashi (1989)
kh73-4-8 2/168 0–2000 ka POL Sato and Kobayashi (1989)
kh90-3-5 4/160 32–1159 ka POL Cande and Kent (1995) Sato et al. (1998)
kk78-030a 19/161 601–1785 ka POL Guyodo and Valet Cande and Kent (1995) Laj et al. (1996a,b)
(2006)
kr9912-pc2a 11/163 1003–3000 ka POL, Cande and Kent Yamazaki and Oda (2005)
ARM (1995), Lisiecki and
Raymo (2005)
kr9912-pc4a 13/162 2002–2845 ka POL Cande and Kent Yamazaki and Oda (2005)
(1995), Lisiecki and
Raymo (2005)
kr9912-pc5a 9/163 1295–2118 ka POL Cande and Kent Yamazaki and Oda (2005)
(1995), Lisiecki and
Raymo (2005)
ks87-752a,b 38/38 311–1023 ka POL, MS Guyodo and Valet Cande and Kent (1995) Valet et al. (1994)
(1999)
lc07a 38/10 754–1033 ka POL, Guyodo and Valet Shackleton et al. Dinares-Turell et al. (2002)
(2006) (1990)
ldba 45/4 20–308 ka MS, 14C Dansgaard et al. Thouveny et al. (1994),
(1993) Williams et al. (1998)
Massicore 44/14 32–35 Ma POL Cande and Kent (1995) Lanci and Lowrie (1997)
md01-2440 38/11 2–400 ka RPI, MS Martinson et al. (1987) Thouveny et al. (2004)
md01-2441 38/11 30–54 ka RPI, MS Martinson et al. (1987) Thouveny et al. (2004)
(Continued)
Table 2 (Continued)

longitude methods
md84-528 42/53 15–80 ka d18O 1984 Tric et al. (1992)

md84-629 36/33 15–62 ka d18O 1984 Tric et al. (1992)
md85-668a 0/46 21–187 ka d18O Guyodo and Valet Martinson et al. (1987) Meynadier et al. (1992)
(1999)
md85-669a 2/47 20–138 ka RPI Guyodo and Valet Martinson et al. (1987) Meynadier et al. (1992)
(1999)
md85-674a 3/50 18–138 ka RPI Guyodo and Valet Martinson et al. (1987) Meynadier et al. (1992)
(1999)
md90-0940a 6/62 108–1954 ka POL, Guyodo and Valet Cande and Kent (1995) Meynadier et al. (1994)
MS, (1999)
fossils
md95-2009a 63/4 10–76 ka ARM Laj et al. (2000) Grootes and Stuiver Kissel et al. (1999), Laj et al.
(1997) (2000)
md95-2024a 50/46 1–117 ka MS, Stoner et al. (2000) Bender et al. (1994) Stoner et al. (2000)
d18O
md95-2034a 34/58 12–76 ka ARM Laj et al. (2000) Grootes and Stuiver Kissel et al. (1999), Laj et al.
(1997) (2000)
md95-2039 40/10 0–320 ka d18O, 14C Martinson et al. (1987) Thouveny et al. (2004)
md95-2042 40/10 32–160 ka d18O, 14C Martinson et al. (1987) Thouveny et al. (2004)
md97-2140a,b 2/142 568–1465 ka POL Cande and Kent (1995) Carcaillet et al. (2003)
md97-2143a,b 16/125 601–2226 ka POL, Guyodo and Valet Cande and Kent Horng et al. (2003)
d18O (2006) (1995), Laskar et al.
(1993)
md98-2181a 6/126 12–660 ka d18O, 14C Sowers et al. (1993) Stott et al. (2002)
md98-2183a,b 2/135 20–1193 ka POL, Yamazaki and Oda Cande and Kent Yamazaki and Oda (2005)
MS, (2005) (1995), Lisiecki and
ARM Raymo (2005)
md98-2185a,b 3/134 9–2256 ka POL, Guyodo and Valet Cande and Kent Yamazaki and Oda (2005)
MS, (2006) (1995), Lisiecki and
ARM Raymo (2005)
md98-2187a,b 4/135 51–3053 ka POL, Cande and Kent Yamazaki and Oda (2005)
MS, (1995), Lisiecki and
ARM Raymo (2005)
md99-2334 38/11 0–38 ka RPI Martinson et al. (1987) Thouveny et al. (2004)
ngc16a 2/135 2–191 ka MS Guyodo and Valet Martinson et al. (1987) Yamazaki and Ioka (1994)
(1999)
(1999)
(1999)
ngc36a 1/161 1–546 ka d18O Guyodo and Valet 1984, Bassinot et al. Yamazaki et al. (1995)
(1999) (1994)
ngc38a 15/175 9–406 ka d18O Guyodo and Valet 1984, Bassinot et al. Yamazaki et al. (1995)
(1999) (1994)
ngc65a 35/175 6–635 ka S Bassinot et al. (1994) Yamazaki (1999)
ngc69a,b 40/175 7–881 ka S Bassinot et al. (1994) Yamazaki (1999)
np35a 4/141 127–798 ka d18O Guyodo and Valet 1984, Bassinot et al. Yamazaki et al. (1995)
(1999) (1994)
np5a 1/137 8–196 ka d18O Guyodo and Valet Martinson et al. (1987) Yamazaki and Ioka (1994)
(1999)
np7a 2/138 6–199 ka MS Guyodo and Valet Martinson et al. (1987) Yamazaki and Ioka (1994)
(1999)
p012a 59/47 14–177 ka RPI Guyodo and Valet Guyodo and Valet Stoner et al. (1998)
(1999) (1999)
(1999) (1999)
(1999) (1999)
(Continued)
Table 2 (Continued)

longitude methods
p226a 3/170 41–780 ka POL Guyodo and Valet Cande and Kent (1995) Yamazaki et al. (1995)
(1999)
14
ps1535-10 79/2 0–100 ka C Nowaczyk et al. (2003)
14
14
ps1707-2 73/14 0–80 ka MS Martinson et al. (1987) Nowaczyk and Antonow (1997)
ps1852-2a 70/16 4–283 ka MS 1984, Martinson et al. Nowaczyk and Frederichs
(1987) (1999)
14
ps1878-3 73/10 0–45 ka d18O, 14C Martinson et al. (1987) Nowaczyk and Antonow (1997)
ps2138-1 82/31 10–75 ka d18O, 14C Nowaczyk and Knies (2000)
ps2644-5a 68/22 12–76 ka ARM Laj et al. (2000) Grootes and Stuiver Kissel et al. (1999), Laj et al.
(1997) (2000)
rc10-167a 33/150 11–781 ka POL Guyodo and Valet Cande and Kent (1995) Kent and Opdyke (1977)
(1999), Hartl and
Tauxe (1996)
rndb75pa 2/160 124–668 ka d18O Guyodo and Valet 1984, Shackleton et al. Tauxe and Shackleton (1994)
(1999) (1990)
su90-24a 63/37 11–76 ka ARM Laj et al. (2000) Grootes and Stuiver Kissel et al. (1999), Laj et al.
(1997) (2000)
su90-33a 60/22 12–76 ka ARM Laj et al. (2000) Grootes and Stuiver Kissel et al. (1999), Laj et al.
(1997) (2000)
su9003 41/32 10–240 ka Color Pisias et al. (1984) Weeks et al. (1995)
su9004 41/32 0–240 ka Color Pisias et al. (1984) Weeks et al. (1995)
su9008 44/30 10–180 ka d18O Pisias et al. (1984) Weeks et al. (1995)
su9039 52/22 0–240 ka d18O Pisias et al. (1984) Weeks et al. (1995)
su92-17a 39/27 4–280 ka Color Martinson et al. (1987) Lehman et al. (1996)
su92-18a 38/27 4–280 ka d18O Guyodo and Valet Martinson et al. (1987) Lehman et al. (1996)
(1999)
su92-19a 38/27 4–279 ka Color Guyodo and Valet Martinson et al. (1987) Lehman et al. (1996)
(1999)
v16-58a 46/30 767–770 ka POL Hartl and Tauxe Cande and Kent (1995) Kent and Schneider (1995)
(1996)
ver98-1-1 53/108 20–60 ka ARM Demory et al. (2005)
ver98-1-6a 54/108 65–235 ka Silica Martinson et al. (1987) Oda et al. (2002)
Kotsiana 36/24 – Laj et al. (1996b)
Lingtaia 35/107 10–73 ka MS, 14C, Pan et al. (2001)
ThL
Potamida 36/24 – Laj et al. (1996b)
lra,b 43/13 90–94.9 Ma POL, Stratigraphy (2004) Cronin et al. (2001)
fossils
WEGAstack 65/144 0–800 ka RPI, 14C, Macri et al. (2005)
fossils
MBstacka 3/162 32–1159 ka POL Guyodo and Valet Cande and Kent (1995) Sato et al. (1998)
(2006)
PMstacka 39/10 0–402 ka RPI, Martinson et al. (1987) Thouveny et al. (2004)
d18O,
14
C
NAstacka 45/25 10–250 ka d18O Pisias et al. (1984) Weeks et al. (1995)
a
Submitted to the MagIC database.
b
Ages recalculated.
Dating methods: RPI: relative paleointensity; POL: polarity stratigraphy; MS: correlation of magnetic susceptibility; ARM: correlation of ARM; carb.: correlation of calcium
carbonate; ash: tephrostratigraphy; color: correlation of color; d18O: oxygen isotopes; 14C: radiocarbon; : correlation of some unspecified wiggle; ThL: thermoluminescence; S:
correlation of high to low coercivity IRM; silica: correlation of silica variations; fossil: correlation based on fossils.
of relative paleointensity data as SEDPI06 in the following. 4pr 3 1=2

VDM ¼ Banc 1 + 3cos 2 ym [11]
Authors are encouraged to contribute or augment their own mo
data in the database. In Section 5.13.8, we will discuss the
The VDM calculation requires a good estimate for the incli-
highlights of the available paleointensity data from both the
nation, which is not always available, especially when unor-
PINT13 (absolute) and SEDPI06 (relative) compilations.
iented specimens are used. In such cases, it may be possible to
Before we discuss the global data set, we will first describe
use either the site (co)latitude or a paleo(co)latitude estimated
methods of converting to virtual dipole moment (VDM).
by a plate reconstruction in the place of magnetic colatitude in
eqn [11]. This moment is known as the virtual axial dipole
moment (Barbetti, 1977).
5.13.7.2 Conversion to VDM In order to compare the two forms of normalization, we
selected data from the PINT13 database for the last 200 My
5.13.7.2.1 Absolute paleointensity data
that (1) were obtained with thermal normalization and used
The intensity of the magnetic field varies by a factor of two
pTRM checks and (2) had multiple specimens that had stan-
from equator to pole simply as a result of a dipole source, so
dard deviations < 15% of the mean or were less than 5 mT. We
data from different latitudes must be normalized to take this
estimated paleolatitudes for the sampling sites using the global
inherent dipole variation into account. In the following, we
apparent polar wander paths of Besse and Courtillot (2002)
discuss methods for converting intensities to ‘virtual dipole
and used these to calculate VADMs for many sites. We show the
moments’ in both absolute and relative paleointensity
two estimates of dipole moment in Figure 17(c); the two are
data sets.
essentially equivalent representations.
There are several ways to calculate equivalent dipole
It is important to note here that neither VDMs nor VADMs
moments for paleointensity data. Early studies tended to pre-
actually represent the true dipole moment (see Korte and
sent a given intensity result as a ratio with some expected field.
Constable, 2005). They do not take into account the rather
For example, Thellier and Thellier (1959) normalized intensity
substantial effect of the nondipole field contributions and in
data to a reference inclination of 65 , using the paleomagnet-
fact overestimate the true dipole moment based on an evalua-
ically determined inclination and the relationship between
tion of data for the last 7000 years.
inclination and field strength expected from a magnetic dipole.
Most studies published over the last few decades, however,
express paleointensity in terms of the equivalent geocentric
dipole moment, which would have produced the observed
5.13.7.2.2 Relative paleointensity data
Sedimentary paleointensity data are at best ‘relative’ paleoin-
intensity at that (paleo)latitude. There are two ways in which
tensity. Nonetheless, several studies have attempted to cali-
this is done, the VDM (Figure 17(a)) and the virtual axial
brate relative paleointensity data into a quasiabsolute form
dipole moment (VADM; Figure 17(b)). The VDM (Smith,
and cast them as ‘VADMs’ in order to compare them with the
1967) is the moment of a geocentric dipole that would give
igneous data sets. There are different strategies for accomplish-
rise to the observed magnetic field vector at location P. (The
ing this conversion (see Figure 18): setting the ‘floor’ to some
piercing point on the surface of the globe of this moment is the
minimum value expected for the field (Constable and Tauxe,
virtual geomagnetic pole or VGP.) To get the VDM, we first
1996), setting parts of the sedimentary record to be equal to
calculate the magnetic (paleo)colatitude ym from the observed
coeval igneous records (e.g., Guyodo and Valet, 1999; Valet
inclination I and the so-called dipole formula (tan I ¼ 2cot ym).
et al., 2005), and setting the mean value to be some assumed
Then, assuming a centered (but not axial) magnetic dipole
value.
with moment VDM, we have
200
Virtual 180
geomagnetic 55 160
pole (lp,fp)
VADM (ZAm2)
140
50
P(lS,fS) 120
P(lS,fS)
45 100
B (μT)
40 80
Virtual Virtual 60
dipole axial dipole 35 40
moment moment
(Am2) 20
(Am2) 30
0
0 40 80 120 160 200
(a) (b)
(c) VDM (ZAm2)
Figure 17 (a) The virtual dipole moment (VDM) is the geocentric dipole that would give rise to the observed geomagnetic field vector at the location P.
ls, fs are the site latitude and longitude, respectively. (b) The virtual axial dipole moment is the geocentric axial dipole that would give rise to the
observed intensity at P. (c) Comparison of VDM and VADM for paleointensity data (see text).
120 Sint-800
Std. error
Volcanic data and error
100
VADM (ZAm2)
3.5 80
70
3.0 60
60
2.5 40
50
VADM (Am2)
B∗ (μT)
2.0 20
40
1.5 0
30 (b) 0 100 200 300 400 500 600 700 800
1.0 20 120
VADM (ZAm2)
0.5 10 100
80
0.0 0 60
−80 −40 0 40 80
40
(a) VGP latitude
20
0 500 1500 2000

(c) Age (ky)
Figure 18 Calibration of sedimentary relative paleointensity data to quasiabsolute values. (a) Approach of Constable and Tauxe (1996) on relative
paleointensity data from the Oligocene record at DSDP Site 522. Data associated with transitional fields (low virtual geomagnetic pole (VGP) latitudes)
are assumed to be on average 7.5 mT, the value of the present average non-GAD field. This converts the relative intensity data to mT. Taking a
paleolatitude of 32 , Tauxe and Staudigel (2004) converted these data to VADM using eqn [11]. (b) Stacked record of relative paleointensity data
spanning the last 800 ky (SINT-800) of Guyodo and Valet (1999). They used the VADM data from contemporaneous lava flows to calibrate the record to
VADM. (c) Similar to (b) but data span the last 2 My (SINT-2000) from Valet et al. (2005). They used averages of the PINT03 database (red dots) to
calibrate the SINT record to VADM.
In the Constable and Tauxe (1996) method, the nonaxial criteria) and averaged them in 100 000-year bins (red dots).
dipole field is assumed to be on average 7.5 mT as it is for the These were used to convert the SINT-2000 stack to VADM values.
present field. Reasoning that because the axial dipole must go The two calibrations are somewhat different, with the latter
through zero in a polarity transition, the average transitional version being higher on average. Ziegler et al. (2011) also com-
field should be about 7.5 mT. Figure 18(a) illustrates an appli- bined relative and absolute paleointensity data but in a different
cation of this method to calibrate the Oligocene relative way; they constructed the PADM2M time-varying axial dipole
paleointensity data from Deep Sea Drilling Program (DSDP) intensity model over the last 2 My based on a penalized maxi-
Site 522 of Tauxe and Hartl (1997; see Table 2) into VADM mum likelihood inversion procedure using a joint set of absolute
values. Setting the average value of the intensity in transitional paleointensity data and relative paleointensity time series.
records (red square in Figure 18(a)) to a value of 7.5 mT Because amplitudes of relative paleointensity records must
calibrates the entire record to B* in mT. Assuming a paleolati- be related to latitude, it is preferable to convert individual
tude of 32 S allows these B* values to be converted to VADM records to VADM prior to stacking, instead of stacking first
values using eqn [11]. The problem with this method is that it and then converting to VADM. However, the ‘floor setting’
is extremely sensitive to the choice of the nonaxial dipole field method of Constable and Tauxe (1996) required transitional
‘floor’ value, which is not known for ancient times. Small data, which are not always available, and has severe drawbacks
changes in the choice of floor result in large changes for the of its own as mentioned before.
calibrated record. Channell et al. (2009) followed Constable
and Tauxe (1996) for calibrating their PISO-1500 stack, which
covers for the last 1.5 My with sediment cores of relatively high 5.13.8 Discussion
sedimentation rates mainly from the North Atlantic Ocean.
A different approach was taken by Guyodo and Valet (1999) In the following, we will discuss some of the ‘hot topics’ in
who collected together many relative paleointensity records paleointensity. The issues for many of these are still under
spanning the last 800 ky (Guyodo and Valet, 1999). The ‘SINT- debate and conclusions are still tentative. Nonetheless, the
800’ stack (see Figure 18(b)) overlapped a sequence of absolute spirit of this volume is to present the ‘state of the field’ and
paleointensity data whose ages were well known (red dots). we will endeavor to do so.
These absolute data were used to calibrate the SINT-800 stack
into VADM. Valet et al. (2005) extended the relative paleointen-
5.13.8.1 Selection Criteria from the PINT13 Database
sity stack to span the last 2 My (see Figure 18(c)). In this latest
version, known as the SINT-2000 stack (a subset of the records For the purpose of this discussion, we plot all the available data
compiled in Guyodo and Valet (2006)), they took the global from the PINT13 database as white triangles and those that had
paleointensity data in the PINT03 database (with no selection standard deviations either 5 mT or 15% of the mean as green
300
250
200
VADM (ZAm2)
150
100
50
0
0 50 100 150 200
Age (Ma)
Figure 19 Summary of published data downloaded from the MagIC database for the last 200 Ma. Red dots are submarine basaltic glass data. Blue
squares are single crystal results. Triangles are all other data and the light green triangles meet the consistency criteria (<15% of mean or <5 mT);
VADMs calculated using the paleolatitudes indicated in the database (model_lat) for data points older than 5 Ma. Magenta dashed line is present
field and solid cyan line is long-term stable value for the last 140 Ma. Modified from Tauxe L, Gee JS, Steiner M, and Staudigel H (2013)
Paleointensity results from the Jurassic: New constraints from submarine basaltic glasses of ODP Site 801C. Geochemistry, Geophysics, Geosystems.
http://dx.doi.org/10.1002/ggge.20282.
triangles in Figure 19. Data that came from submarine basaltic

glass are marked as red circles, while those from single crystals 1.0
are blue squares. Following the treatment of Tauxe et al.
(2013), we calculate median values for 5 My bins for bins
Cumulative distribution
with at least ten reliable data points (green triangles); these 0.8
are shown as stars. The median value of all the median values is
42 ZAm2 (shown as the solid cyan line) and the present field
value is shown as the dashed magenta line. 0.6
There are a total of 4475 cooling unit averages with an
estimate for either VADM or VDM, of which 2887 meet the
0.4
cooling unit consistency test applied here. The median value of
all results from the last 140 Ma is 42 ZAm2 (‘Z’ stands for Zetta
(1021)). We have deliberately chosen a very loose standard for
0.2
acceptance as there is no agreement in the community as to
what constitutes a ‘reliable’ result and it is not clear that,
without the original measurement data, an intelligent selection 0.0
can be made. 0 50 100 150 200 250 300 350
As an example of the problem of data selection, there are a VADM (ZAm2)
total of 3328 cooling unit averages with checks of alteration Figure 20 Cumulative distributions of data that come from experiments
during the experiment (method code of ‘LP–PI–ALT’) and the with a check for alteration (heavy red line) versus those that did not
rest have no such test. The two data sets are plotted as cumu- (light blue line). The two data sets cannot be distinguished from one
lative distributions in Figure 20. One might expect a significant another at the 95% level of confidence, based on a two-tailed
difference between the two distributions, but, in fact, there is Kolmogorov–Smirnov test.
no statistical difference between the two. This does not mean
that experimental design makes no difference; the discussion
of theory makes it quite clear that many things can give an latitudinal bins, and the intensity at each latitude expected
erroneous result and these things should be tested for. The from the median dipole moment of 61 ZAm2 for the same
problem is that there is insufficient information in the data- interval is shown as the cyan line. The overall fit of the paleoin-
base to make a meaningful selection. tensity data shown in Figure 21 to a dipole field is poor as
The difficulty with the published data as a whole becomes pointed out by Lawrence et al. (2009). Reasons for the failure
apparent when we plot the intensity data from the last 5 My of the dipole hypothesis in the PINT13 data compilation
against latitude (Figure 21). Stars are median values in 10 include the following: (1) the data may be ‘no good,’ (2)
200 The lack of consensus on the ‘stable’ value of the field stems
in part from differing views on which data to include as well as
the explosive growth of paleointensity data available
(compare, e.g., Biggin et al., 2003; Goguitchaichvili et al.,
150
2004; Heller and Liu, 1982; Selkin and Tauxe, 2000; Tauxe
Intensity (μT)
et al., 2013). While Biggin et al. (2003) argued that because

such procedures as the so-called pTRM check designed to iden-
100 tify alteration during the paleointensity experiment cannot
guarantee the quality of a particular result, there is no need to
reject data that do not have pTRM checks; others (e.g., Riisager
50 and Riisager, 2001; Tauxe and Staudigel, 2004) have tried to
develop more rigorous experimental protocols to detect and
reject ‘bad’ data. Here, we take the broad view advocated by
Biggin et al. (2003), relying strongly on strict consistency tests
0
−100 −50 0 50 100 at the cooling unit level.
Latitude Tauxe et al. (2013) considered the subject anew and calcu-
Figure 21 Paleointensity versus latitude for data from the last 5 My. lated the median value of the field, as opposed to the
Stars are median values for 10 bins. arithmetic average. The arithmetic average is heavily influenced
by outliers, of which there are many in the database.
In their treatment, the median value of the field (shown as
there may be long-term nonaxial dipole field contributions to the cyan line in Figure 19) is 42 ZAm2 and is quite stable
the geomagnetic field, (3) the average may be nonstationary through time, with the only long-term departure from this
and the data in different bins in Figure 21 are from different value occurring in the Jurassic and early Cretaceous (the Meso-
ages and average fields, and (4) there may be long-term hemi- zoic dipole low).
spheric asymmetry (Cromwell et al., 2013). One clear result from the data in Figure 19 is that the
Although the PINT13 data set is by now quite large, the geomagnetic field is highly variable on both short and long
age distribution of paleointensity data for the last 200 Ma is timescales. Therefore, it is likely that unless the same time
still patchy with 24% of the data being younger than 1 Ma. period is considered in all latitudinal bins in figures like
By far, most data come from the northern hemisphere (see Figure 21, there will be scatter introduced from comparing
Figure 16(a)). So one point of agreement among all papers times with different average field strength. Some effort should
on the subject is that more and better data would be helpful be put in obtaining data for certain time slices as a function of
for defining the average paleofield intensity and its variation. paleolatitude.
5.13.8.3 Are There Any Trends?

5.13.8.2 What Is Long-Term Strength of the Geomagnetic
5.13.8.3.1 Intensity versus polarity interval length
Field?
Although there are no clear long-term trends in the paleoin-
The present Earth’s magnetic field is well approximated by a tensity data shown in Figure 19, there are times when the field
geocentric magnetic dipole with a moment of about 80 ZAm2. is stronger than others as noted previously. Cox (1968) sug-
But what is the average value of the dipole moment? A great gested that strong geomagnetic fields could inhibit reversals of
deal of effort has been put into assembling paleointensity the geomagnetic field and this makes sense with the observa-
databases over more than three decades, yet there remains little tion that geomagnetic fields are low when the field is reversing.
consensus on the answer to this most basic question. Tauxe and Hartl (1997) and Constable et al. (1998) dem-
Early studies suggested that the average field strength has onstrated a weak correlation between the length of a given
been either quite a bit lower than the present (e.g., Coe, 1967; polarity interval and the average paleointensity in the relative
Smith, 1967) or approximately equivalent to today’s field paleointensity data from DSDP Site 522 (but Yamazaki et al.
(Bol’shakov and Solodonikov, 1980; Kono, 1971; McFadden (2013) suggested that this data set may be influenced by sed-
and McElhinny, 1982). Some studies found no trend with age iment property changes). So one of the primary motivations
in VDMs (e.g., Bol’shakov and Solodonikov, 1980) over the for initiating the study of the DSDP/ODP submarine basaltic
last few hundred million years, while others found a significant glasses for paleointensity was to test the hypothesis that long
increase in dipole moment from the Mesozoic to the present intervals of stable polarity (like the Cretaceous Normal Super-
(e.g., Smith, 1967). Tanaka et al. (1995b) estimated the aver- chron or CNS in Figure 19) were associated with unusually
age dipole moment for the last 20 My to be approximately strong fields (see, e.g., Pick and Tauxe, 1993). It was therefore
84 ZAm2 with significantly lower values in the Mesozoic (the puzzling and a bit disappointing when Selkin and Tauxe
so-called Mesozoic dipole low of Prévot et al., 1990), a view (2000) compared paleofield strength with reversal rate and
also held by Perrin and Shcherbakov (1997) and reiterated by found no clear relationship. There were just too few data
Biggin et al. (2003). But a series of more recent papers have from the CNS to make a definitive statement.
argued for a lower average (Juarez and Tauxe, 2000; Juarez Now, there are many more data from the last 175 Ma with
et al., 1998; Selkin and Tauxe, 2000; Tauxe, 2006; Tauxe several data sets available from intervals whose polarity chrons
et al., 2013; Yamamoto and Tsunakawa, 2005). are known and the data can be associated with known polarity
interval length. In Figure 22, we also show the data from 5.13.8.3.2 Source of scatter in the CNS
Figure 19 associated with known polarity intervals. These Another prediction made from the relative paleointensity data
data are from SBG obtained from holes drilled on clearly from Site 522 was that the scatter in the data is proportional to
identifiable magnetic anomalies compiled by Tauxe (2006) the average value and strongly linked to polarity interval length
and data from lava flows in magnetostratigraphic sections (Constable et al., 1998). This observation also appears to be
correlated to the timescale (Herrero-Bervera and Valet, 2005; weakly supported by the absolute paleointensity data set. One
Riisager et al., 2003). It appears that the correlation suggested question that springs to mind, however, is whether the scatter
by Tauxe and Hartl (1997) and Constable et al. (1998) based is geomagnetic in origin. To address this issue, Granot et al.
on relative paleointensity in sediments is supported by the (2007) assembled a data set from the Troodos Ophiolite,
absolute paleointensity data set, although not strongly. which formed during the CNS. Their data set includes new
data from gabbros as well as the submarine basaltic glass data
of Tauxe and Staudigel (2004). Many of the gabbro data came
from a sequence of small plutons with a clear relationship to
200 the ancient spreading axis and their relative age relationships
Virtual (axial) dipole moment (ZAm2)
CNS
were therefore known. Tauxe and Staudigel (2004) had sam-
pled two transects through the entire oceanic extrusive layer,
150 separated by some 10 km. Data from these two transects are in
C26r stratigraphic order, so their age relationships are also known.
C13n,C27n In Figure 23, we show their plot of the three time sequences.
C2An.3n, M0r
100 The data exhibit remarkable serial correlation, which Granot
et al. (2007) used to argue that the scatter in the CNS data is
largely geomagnetic in origin. Magnetic anomaly data of
50 Granot et al. (2012) seem to support the geomagnetic origin
of the scatter; the amplitude envelope of magnetic contrasts
from inversion of a deep-tow magnetic anomaly profile from
0 the Central Atlantic Ocean spanning the entire CNS shows
0.1 1 10 100
some similarity with time variations of the paleointensity
Polarity interval length (My)
data scatter.
Figure 22 VADM data from the PINT13 database from polarity intervals
of known duration. Data from the CNS are from Cottrell and Tarduno
5.13.8.3.3 The oldest paleointensity records
(2000), Riisager et al. (2003), Sherwood et al. (1993), Tarduno et al.
(2001, 2002), Tauxe (2006), Tauxe and Staudigel (2004), Tanaka and Under the topic of ‘trends in paleointensity,’ one of the most
Kono (2002), Zhu et al. (2004), and Zhu et al. (2004); C26r and C27n interesting questions concerns the earliest records of paleoin-
from Riisager and Abrahmsen (2000); C12n, C13n, C32n are from Tauxe tensity. In Figure 24, we show published results satisfying
(2006); M0r are from Riisager et al. (2003), Tauxe (2006), and Zhu et al. minimum consistency constraints obtained from Archean-
(2001); and C2An.3n are from Herrero-Bervera and Valet (2005). aged rocks. Until the 1990s, there were very few studies that
Time (year)
Young Old
0 3–11×105 0 ~105 0 ~105
180 SBG
Gabbro
160
140
VADM (ZAm2)
120
100
80
60
40
~5
~11
×1
00
66
Half width
0
20
5
yye
of Solea
ye
axis
ar r
ar
0
0 2 4 6 8 10 0 5 10 5 10
Distance from the solea spreading axis center (km) Successive lava flows
Figure 23 Three time series of paleointensity data during the late CNS from the gabbros and glasses. Gray areas represent gaps in time estimated from
moderate spreading rate (full spreading rate of 20–75 mm year1). Results for the gabbro specimens are shown as individual points and are
corrected for cooling rate and anisotropy. Results for the SBG sites correspond to the average results from successive cooling units. Data are from
Granot et al. (2007) and Tauxe and Staudigel (2004). Modified from Granot R, Tauxe L, Gee JS, and Ron H (2007) A view into the Cretaceous
geomagnetic field from analysis of gabbros and submarine glasses. Earth and Planetary Science Letters 256: 1–11.
140
120
100
VADM (ZAm2)
80
60
40
20
0
2000 2200 2400 2600 2800 3000 3200 3400
Age (Ma)
Figure 24 Archean data from the database that have standard deviations of <20% of the mean or are <5 mT. Age uncertainties are indicated by
the horizontal bars. Triangles are data from KTT-type experiments with pTRM checks and squares are from other types of experiments.
The diamonds are from single crystal experiments. Data are from Morimoto et al. (1997), Yoshihara and Hamano (2000), Sumita et al. (2001),
Macouin et al. (2003), Bergh (1970), Hale (1987), McElhinnhy and Evans (1968), Schwartz and Simons (1969), Smirnov and Tarduno (2003),
Smirnov et al. (2003, 2005), and Selkin et al. (2000).
were based on experiments that used pTRM checks (triangles While all transitional records with paleointensity data have
and diamonds in Figure 24). Recent data meet the highest low field intensities associated with transitional directions, not
experimental standards and show that the field had a large all records display a long-term decreasing trend toward a rever-
range in intensity, similar to more recent times, although the sal. Arguments supporting the existence of the sawtooth pat-
highest values come from experiments not done with pTRM tern were presented in rapid succession. Valet et al. (1994)
checks (squares in Figure 24). The inescapable conclusion examined paleointensity records near the Matuyama–Brunhes
from these data is that the geomagnetic field was ‘alive and transition from the Atlantic, Indian, and Pacific Oceans and
well’ by 3 Ga. found the rapid intensity growth after the transition.
Meynadier et al. (1994) recognized the ‘sawtooth pattern’ in a
relative paleointensity record spanning the last 4 My obtained
5.13.8.3.4 The paleointensity ‘sawtooth’ from a core in the Indian Ocean. Verosub et al. (1996) pre-
Valet and Meynadier (1993) (detailed data set in Meynadier sented a record focused near the MBB and the Jaramillo sub-
and Valet, 1995) presented a relative paleointensity record for chron of a sediment core from the central north Pacific, which
the last 4 My using sediment cores of ODP Leg 138 taken from supported the ‘sawtooth pattern.’
the eastern equatorial Pacific (see Figure 25). They postulated Counterarguments to the sawtooth also began to appear.
an ‘asymmetrical sawtooth pattern’ of paleointensity varia- For example, Laj et al. (1996a, 1997 (for correction)) failed to
tions, that is, a rapid intensity growth just after a polarity see the ‘sawtooth pattern’ utilizing the same core as that of
transition and a gradual decrease since then toward the next Verosub et al. (1996) when examining a longer period of time
reversal (see Figure 25). They also suggested that the length of up to the Olduvai subchron. Laj et al. (1996b) reported that no
a polarity zone is proportional to the magnitude of the inten- rapid intensity increase after polarity transitions was observed
sity jump. The latter observation is consistent with the data in relative paleointensity records from two Late Miocene sec-
shown in Figure 22 whereby long polarity intervals appear to tions in Crete (Kotsiana and Potamida in Table 2).
have higher average fields. Nonetheless, the ‘sawtooth’ idea Arguments against the ‘sawtooth pattern’ were presented
became the subject of heated arguments. also from paleointensity estimations based on recording mech-
The sawtooth was originally epitomized by the long-term anisms different from sediments. Records of 10Be/9Be reflect
decreasing trends observed in the Matuyama, one toward the geomagnetic paleointensity through a control on the produc-
Brunhes and one toward the Olduvai, and in the Gauss, one tion rate of the cosmogenic nuclide (10Be). Raisbeck et al.
toward the Kaena and one toward the Matuyama boundaries. (1994) argued that a 10Be/9Be record at Site 851 of ODP Leg
The sawtooth envisioned by Valet et al. (2005) (see SINT-2000 138, which is the same site as of Valet and Meynadier (1993), is
in Figure 25), however, is a shorter trend immediately preced- inconsistent with the ‘sawtooth pattern.’ Westphal and
ing the reversal boundary. We will refer to this more restricted Munschy (1994) showed that the ‘sawtooth pattern’ cannot
view as the ‘short sawtooth’ hypothesis in the following. First, explain the shape of stacked magnetic anomaly profiles over
we review the debate about the sawtooth in the literature. the Southeast Indian Ridge, Juan de Fuca Ridge, and East
2.5
1.5
OJP-stack 1
Kok and Tauxe (1999)
0.5
2 0
Brunhes–Matuyama
1.5
0.5 EPAPIS-3Ma
Matuyama–Gauss
Jaramillo
Yamazaki and Oda (2005)
Olduvai
0
2
1.5
Sint-2000 1
Valet et al. (2005)
3 0.5
2.5 ODP Leg 138 0

Valet and Meynadier (1993)
2
Kaena
1.5
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Age (Ma)
Figure 25 The paleointensity record from ODP Leg 138 cores (Valet and Meynadier, 1993) showed the ‘asymmetrical sawtooth pattern,’ whereas
other records, which reached the Gauss–Matuyama boundary, do not show such pattern: Ontong Java Plateau stack of Kok and Tauxe (1999)
and equatorial Pacific paleointensity stack of Yamazaki and Oda (2005).
Pacific Rise. McFadden and Merrill (1997) presented an anal- cumulative viscous remanence model, a very narrow distribu-
ysis of the Cenozoic polarity reversal chronology that effective tion of the relaxation times is required. They also showed that
inhibition of a future reversal can only last for about 50 ky at the sawtooth pattern of the Site 851 sediments did not change
most, which contradicts the ‘sawtooth pattern’ requiring much by thermal demagnetization and AF demagnetization of stron-
longer inhibition. ger fields, which is inconsistent with the result of Kok and
The ‘sawtooth pattern’ was also questioned from remanent Tauxe (1996b) despite using the sediments from the same
magnetization acquisition processes. Kok and Tauxe (1996a) site. Kok and Tauxe (2000) on the comments to Meynadier
proposed a cumulative viscous remanence model for remanence et al. (1998) stressed the nonuniqueness of relaxation time
acquisition of sediments that can yield intensity variations like distribution explaining the ‘sawtooth pattern’ and pointed
the ‘sawtooth’ pattern. Then, Kok and Tauxe (1996b) reproduced out that the t distribution of Kok and Tauxe (1996b) is not
the ‘sawtooth pattern’ of the ODP Leg 138 sediments by the all relaxation times present in the sediments, but merely a part
cumulative viscous remanence model using the values of the of them that behave viscously. In the reply, Meynadier and
equilibrium magnetization constrained by results of a Thellier- Valet (1995) mentioned that the remanent magnetization with
type paleointensity experiments applied to the ODP Leg 138 blocking temperatures between 150 and 300 C, which may
sediments. Furthermore, they resampled the Site 851 sediments carry cumulative viscous remanence, is only a minor part of
near the Gauss–Gilbert boundary and showed that the ‘sawtooth NRM of the sediments. The two groups also argued about the
pattern’ disappeared by thermal demagnetization to 400 C. validity of thermal demagnetization on relative paleointensity
Meynadier et al. (1998) made a counterargument to the estimation from sediments (Kok and Tauxe, 1999; Kok and
cumulative viscous remanence model of Kok and Tauxe Ynsen, 2002; Valet and Meynadier, 2001).
(1996a,b). They suggested that to produce the sawtooth pat- Mazaud (1996) proposed another model of magnetization
tern and to preserve the magnetostratigraphy with the acquisition, which produces the ‘sawtooth pattern’: a large
fraction (say, two-thirds) of magnetic grains acquire NRM at (1993) has not been universally observed, as would be
deposition time, while the remaining grains reorientate or expected from the behavior of a dipole source. Recently, how-
acquire magnetization after deposition. Although Meynadier ever, Ziegler et al. (2011) showed asymmetry in growth and
and Valet (1996) considered this model unlikely from the decay of the geomagnetic dipole that is not restricted near
knowledge of pDRM acquisition processes that prevailed at polarity boundaries. They found that on the 25–150 ky
that time, recent studies suggest that this model may have timescale, growth rates are larger than decay rates in their
difficulties. As discussed in Section 5.13.4, pDRM acquisition PADM2M model of paleointensity variations (Ziegler et al.,
with compaction that can be expressed by an exponential 2011), in which the data sets of relative paleointensity men-
function is unlikely to occur, although there are still different tioned earlier are incorporated.
opinions for the amount of depth-lag. An experiment of Katari
et al. (2000) using natural undisturbed sediments suggests that
5.13.8.4 High-Resolution Temporal Correlation
pDRM (reorientation of magnetic particles) is a rare phenom-
enon, probably because of the effects of flocculation: magnetic 5.13.8.4.1 Sediments
minerals would be aggregated with clay. Oxygen isotope stratigraphy revolutionized paleoceanography
In the 851 record of Valet and Meynadier (1993), the by providing a global signal with a resolution on the order of
‘sawtooth pattern’ is most apparent in the early Matuyama 10 s of thousands of years (e.g., Hays et al., 1976). Yet oxygen
chron after the Gauss–Matuyama transition and during the isotope stratigraphy has its drawbacks. It cannot be applied in
Gauss chron. The number of paleointensity records reported sediments deeper than the carbonate compensation depth or
so far that reached the Gauss–Matuyama transition is still lakes. Oxygen isotopic data are often difficult to interpret in
small. However, available records from different groups do marginal seas, where records may not reflect global ice-volume
not support the ‘sawtooth pattern’: neither a stacked record changes. Finally, temporal resolution better than 104 years is
since 2.8 Ma from the Ontong Java Plateau (Kok and Tauxe, critical for assessing the global nature of climatic events and
1999; OJP-stack in Figure 25) nor a stacked record since their durations. The prospect of using relative paleointensity
3.0 Ma from the equatorial Pacific (Yamazaki and Oda, 2005; from sediments as a high-resolution correlation and dating
EPAPIS stack in Figure 25) shows variations like the ‘sawtooth tool has therefore been met with great enthusiasm (e.g., Stott
pattern’ after the Gauss–Matuyama transition. More recently, a et al., 2002) and was reviewed recently by Roberts et al. (2013).
record from between 2.1 and 2.75 Ma at IODP Site U1314 in Here, we examine the prospects and problems with the so-
the North Atlantic (Ohno et al., 2012) also fails to reproduce called paleointensity-assisted chronology or PAC.
the sawtooth. The importance of PAC is not simply as a substitution for
We do not yet understand well the rock magnetic processes, d18O. It can also be used to examine consistency of other
which produce the ‘sawtooth pattern’ only for some sediments. chronologies such as d18O and 14C, because it is quasi-
If the DRM acquisition model of Mazaud (1996) works in independent of them. Regional and global intercore correla-
general, all sedimentary paleointensity records should display tions tied by paleointensity variations revealed discrepancies of
‘sawtooth’-like changes. From the cumulative viscous rema- up to several thousand years between those based on 14C and
nence model of Kok and Tauxe (1996a), the sediments pro- d18O (Stoner et al., 1995) and between GISP2 and the d18O
ducing the ‘sawtooth pattern’ are expected to have a particular chronologies (Stoner et al., 1995, 2000) during the last
magnetic grain-size distribution favorable for the long-term c.100 ky. Moreover, paleointensity stratigraphy can have higher
viscous remanence acquisition, but this has not yet been fully resolution than d18O stratigraphy because the variations con-
tested. tain shorter wavelength components than those of d18O. Truly
Valet et al. (2005) used 10 of the 15 records of relative dipolar features of geomagnetic field variations have a potential
paleointensity data compiled by Guyodo and Valet (2006) for providing a time reference for an interhemispheric paleocli-
to create the so-called SINT-2000 stack (see Table 2 and matic relationship with unprecedented resolution.
Figure 25). Based on this subset of the data, Valet and On a more limited scale, there is a possibility that intercore
Herrero-Bervera (2000) argued for the ‘short sawtooth’ pattern correlation and age estimation can be performed using
in the 80 ky interval immediately prior to the four reversals paleointensity by correlating patterns among cores with a stan-
included in the stack. Of the four reversals in the SINT-2000 dard curve such as SINT-800 (Guyodo et al., 2001) and NAPIS-
stack (Figure 4 of Valet and Herrero-Bervera, 2000), only the 75 (Laj et al., 2000), which is exemplified by Stoner et al.
upper Jaramillo and the lower Olduvai show convincing short (1995) in the Labrador Sea, Demory et al. (2005) in Lake
sawtooth patterns. Baikal, and Macri (2005).
Relative paleointensity records spanning the last 2 My have We feel that while regional correlations can be achieved,
steadily been produced (see, e.g., Table 2). Because the SINT- much is lost by using PAC as a primary dating tool. These records
2000 is a stack that did not include many of the records in can no longer be used to constrain paleointensity models or
Table 2, we have plotted those records that span at least the global stacks, because the age information is not independent
period from 800 to 900 ky in Figure 26 for the interval includ- and features correlate by assumption. Such records have been
ing the Brunhes and Jaramillo. Considering all the records, it clearly labeled in Table 2 (RPI) and in the MagIC database.
appears that even the short sawtooth is only observed in a If we desire a global correlation tool, we require a dipolar
small subset of the records, although an intensity peak just signal. However, it is as yet unclear at which wavelength the
after the B/M boundary (Guyodo and Valet, 1999; Valet et al., dipole terms give way to nondipole terms. Korte and Constable
2005) does appear in all records. In general, the long-term (2005) cautioned us that variations in VADMs may not be global
sawtooth pattern originally observed by Valet and Meynadier and their variations need not be synchronous because they can
Brunhes Matuyama Jaramillo Matuyama
984 (60.4)
983 (60)
1021 (39)
lc07 (38)
1010 (30)
kk78-030 (19)
md97-2143 (16)
md90-0940 (6)
767 (5)
md98-2187 (4)
md98-2185 (3)
md98-2183 (2)
md97-2140 (2)
848-851 (2)
803a (2)
ks87-752 (-38)
1101 (-64)
700 800 900 1000 1100 1200

Age (ka)
Figure 26 Plot of all records spanning at least the period from 800 to 900 ka in the MagIC database (see Table 2) for the range 700–1200 ka. Records
are plotted in order of latitude (in parentheses to right of record name). The positions of the Matuyama–Brunhes and the upper and lower Jaramillo
transitions are indicated by heavy black lines.
be strongly influenced by nondipole field effects. Moreover, the (2002), who tied cores together between the North Atlantic
duration of a polarity reversal is known to be dependent on and the western equatorial Pacific (see Figure 27). The large-
latitudes, ranging from about 2 ky near the equator to 10 ky at scale features (labeled H1–H10 and L1–L8) correlate reason-
60 (Clement, 2004). Furthermore, the contribution of non- ably well and are consistent with the oxygen isotopic records
dipole effects may be higher when paleointensity is low, and from the two cores. These allow correlation with a resolution
both the duration and shape of a particular paleointensity low of 2–3 ky. The difficulty of identifying global features on a
will be site-dependent. Nonetheless, there is an astonishing cor- submillennial scale is made apparent by the rather uncon-
relation between the ‘paleointensity dipole stack’ of Ziegler et al. vincing correlation of features marked by the stars. Although
(2011) and the magnetic anomaly inversion of Gee et al. (2000) these features may well be synchronous, they do not resemble
as illustrated in Figure 28. each other very much in the two hemispheres, and without
Doubts about the global nature of paleointensity features the excellent and very detailed chronological control of the
notwithstanding, global intercore correlations on a millen- independent oxygen isotopic records, their identification
nial scale have been attempted. A series of papers by Channell would not have been possible. How much should these
et al. (2000), Stoner et al. (2000), and Mazaud et al. (2002) ‘millennial’ features look like each other and how synchro-
correlated records between the high latitudes of the North nous they are expected to be require much more detailed
Atlantic and the South Atlantic and Indian Ocean sectors. As knowledge of the process of secular variation, a topic of active
an example of the method, we show the records of Stott et al. research (see Chapter 5.09).
H1 H2 H4 ch88-10p (North Atlantic) H10

H3 H6 H9
Laschamp
H8
1.5
H7
H5
1.0
H10
L1
L2 H6 H7
L4 H9
H2 L3
0.5 H1 H3 L5
1.5
L8
H5 L7
L6
H8
1.0
L1
L2
L3 L4
L5
0.5
md98-2181 (West Pacific) L7
L8
L6
0.0
10 20 30 40 50 60 70
Age (ka)
Figure 27 Comparison of deep-sea sediment relative paleointensity records from MD98-2181 and the North Atlantic Ocean site CH88-10P. The two
records were dated by correlation of their individual oxygen isotopic stratigraphies to the GISP2 record. Distinctive paleointensity highs (H1–H10)
and lows (L1–L8) are identified in the two records. Less convincing features (stars) are marked as well. The ages of the named features are
synchronous to within an uncertainty of less than 500 years. The ‘Laschamp excursion’ correlates with a distinctive 36Cl excursion in the GISP2 record
and provides an independent tie point to ice cores. Redrawn from supplement to Stott L, Poulsen C, Lund S, and Thunell R (2002) Super ENSO and
global climate oscillations at millennial time scales. Science 297: 222–226.
5.13.8.4.2 Ridge crest processes

The potential of paleointensity for high-resolution dating is 1.5
not restricted to sediments. Thellier-type paleointensity data
can be used to estimate ages of basalts near mid-ocean ridges, 1.0

and it is expected that paleointensity will be useful for studying
crustal accretion processes at ridges (Bowles et al., 2006; Gee 0.5
et al., 2000; Ravilly et al., 2001). This work is discussed in more
detail in Chapter 5.12 (Figure 28). 0.0
5.13.8.4.3 Archaeomagnetic dating −0.5

A thorough review of the possibilities of using paleointensity
estimates to provide age constraints for archaeological mate- −1.0
rials is beyond the scope of this chapter (but see Chapter 5.09 0 100 200 300 400 500 600 700 800 900
of this volume). Nonetheless, it is worth mentioning that the Age (ka)
so-called archaeointensity results are increasingly used for this Figure 28 Comparison of predicted relative paleointensity from the
purpose (e.g., Ben-Yosef et al., 2008a,b; Pavón-Carrasco et al., magnetization inverted from a high-resolution marine magnetic anomaly
2011). stack (gray; data from Gee et al., 2000) and dipole moments from the
PADM2M paleointensity stack (red; Ziegler et al., 2011). Ages for the
magnetic anomaly record were rescaled to a common age for the
5.13.8.5 Atmospheric Interaction Matuyama–Brunhes boundary. Redrawn from Roberts AP, Tauxe L,
and Heslop D (2013) Magnetic paleointensity stratigraphy and high
Radioactive forms of carbon, beryllium, and chlorine are pro- resolution Quaternary geochronology: Successes and future challenges.
duced in the atmosphere by cosmic ray bombardment. The Quaternary Science Reviews 61: 1–16.
decay of these isotopes is used for dating purposes in a wide 50

variety of disciplines. There are large variations in ages pre-
dicted from tree ring, varve, ice layer counting, or U/Th dating
and those estimated by radiocarbon dating. An example of
40
such a comparison is shown in Figure 29(a), which shows
C age (ky BP)

the age based on 14C dating from a core in the Cariaco Basin
versus layer counting in the Greenland Ice Sheet Project 2 ice
30
core from a core in the Cariaco Basin (Hughen et al., 2004).
The correlation of the marine sediment core to the ice core was
14
based on tying dark sedimentary layers to interstadials in the 20
GISP2 core.
The difference between radiocarbon and other age esti-
mates in Figure 29(a) is used to calculate variations in initial
10
radiocarbon in the atmosphere relative to the concentration in (a) 10 20 30 40 50
the modern atmosphere (see atmospheric D14C plotted as dots
in Figure 29(b)). An excess of radiocarbon (positive D14C)
results in an underestimate of the age because there is ‘too 800
much’ radiocarbon in the sample for its age. Changes in D14C
Atmospheric D14C (%0)

have been attributed to differences in the production of radio- 600
carbon in the atmosphere by cosmic ray bombardment and
changes in the carbon balance between the atmosphere and
400 d
the deep ocean, which is a reservoir of old carbon (see, e.g.,
Bard et al., 1990). If, for example, the transfer of atmospheric
c
carbon into the ocean was less efficient in the past or the 200
b
release of old carbon from the deep ocean was less efficient a
(‘ventilation’ was slower), then there would be an excess of (b) 0
radiocarbon in the atmosphere relative to the modern atmo- H0 H1 H2 H3 H4
sphere, resulting in ‘too young’ 14C ages.
-200
Radiocarbon production is thought to be strongly con-
trolled by changes in magnetic field strength because the mag- 1000
netic field shields the atmosphere from cosmic rays (see
14C
Chapter 5.09). Changes in the intensity of the magnetic field 900
production (moles year-1)

should therefore result in changes in radiocarbon production
(among other things); hence, the variation in intensity is a key 800
parameter in deriving accurate age information. Hughen et al.
(2004) used a paleointensity stack from the North Atlantic (the 700
NAPIS stack of Laj et al., 2000) as a proxy for changes in the
dipole moment of the Earth’s magnetic field over the last 70 ky. 600
By using the Monte Carlo simulations of the relationship
ML LA 500
between geomagnetic field strength and radiocarbon produc-
tion of Masarik and Beer (1999), Hughen et al. (2004) pre-
dicted radiocarbon production for the past 50 ky (Figure 29 400
0 10 000 20 000 30 000 40 000 50 000
(c)). The different curves in Figure 29(a)–29(d) use the pre-
(c) GISP2 Age (year BP)
dicted radiocarbon production values as input into box models
using different ocean/atmosphere boundary conditions relat- Figure 29 (a) Radiocarbon calibration data from Cariaco ODP Leg 165,
ing to different models of deep-sea ventilation, resulting in Holes 1002D and 1002E (blue circles), plotted versus calendar age
different estimates for atmospheric D14C. assigned by correlation of detailed paleoclimate records to the Greenland
Ice Core GISP2. The thin black line is high-resolution radiocarbon
None of the curves in Figure 29(b) based on the model
calibration data from tree rings joined at 12 cal. ka BP to the varve
predictions using the NAPIS stack provide a satisfactory fit to counting chronology. Red squares are paired 14C-U/Th dates from
the observed variations. Hughen et al. (2004) suggested that corals. Light gray shading represents the uncertainties in the
either the model of Masarik and Beer (1999) for translating Cariaco calibration. The radiocarbon dates are ‘too young,’ falling
geomagnetic field intensity to radiocarbon production is incor- well below the dashed line of 1:1 correlation. (b) Compilation of
rect at low field strengths or our understanding of the global data interpreted as production rate changes in radiocarbon (D14C)
carbon cycle is insufficient. It is of course also possible that the versus calendar age (symbols same as in (a)). (c) Predicted
NAPIS model of North Atlantic relative paleointensity is a poor variation of D14C from the geomagnetic field intensity variations
proxy for the global paleomagnetic field intensity variations. from sediments of the North Atlantic (Laj et al., 2000) using the
Roberts et al. (2013) considered the problem but used the model of Masarik and Beer (1999). Modified from Hughen K,
Lehman S, Southon J, et al. (2004) C-14 activity and global
PADM2M model of Ziegler et al. (2011) with the variations
carbon cycle changes over the past 50,000 years. Science 303(5655):
predicted from the 10Be stack of Frank et al. (1997) 202–207.
(Figure 30). The agreement is remarkably good. Overall,
2.0 bulk magnetic properties and there is no coherence between

the relative intensity and the normalizer (IRM), percent car-
bonate, and a magnetic grain-size proxy at 40 ky. Their
1.5 paleointensity records showed significant power also at the

100 ky eccentricity frequency, but they rejected this because
this frequency was observed also in bulk magnetic proper-
1.0 ties. Yamazaki (1999) instead proposed the possible presence
of the 100 ky frequency in his relative paleointensity records
from the North Pacific based on the same logic as that of
0.5 Channell et al. (1998): occurrence of 100 ky power in relative
paleointensity but not in the normalizer on the power spectra.
Possible occurrence of 100 ky period in paleointensity
0.0
0 50 100 150 200 records would not be limited to the Brunhes chron: Yamazaki
Age (ky) and Oda (2005) (see Figure 31) found significant power
at 100 ky period in paleointensity records from 0.8 to
Figure 30 Comparison of relative paleointensity and the relative
3.0 Ma. Kok and Tauxe (1999) found a peak at 150 ky in
paleointensity (gray) predicted from normalized 10Be stacks (reproduced
from Frank M, Schwarz B, Baumann S, Kubik P, Suter M, and Mangini paleointensity records during the Matuyama chron from the
A (1997) A 200 kyr record of cosmogenic radionuclide production rate Ontong Java Plateau.
and geomagnetic field intensity from 10Be in globally stacked deep-sea Yokoyama and Yamazaki (2000) applied a wavelet analysis
sediments. Earth and Planetary Science Letters 149: 121–129) and to five paleointensity records from the Pacific Ocean reported
relative paleointensity (red) from the PADM2M model (Ziegler et al., in Yamazaki et al. (1995) and Yamazaki (1999) and found a
2011) for the last 200 ka. Redrawn from Roberts AP, Tauxe L, and Heslop quasiperiod of 100 ky. They considered the 100 ky period
D (2013) Magnetic paleointensity stratigraphy and high resolution inherent to the geomagnetic field, because of the good coinci-
Quaternary geochronology: Successes and future challenges. Quaternary dence of the relative intensity records in this scale despite
Science Reviews 61: 1–16.
significant phase differences in magnetic properties. Thouveny
et al. (2004) reported a 100 ky period in a paleointensity
variations in dipole intensity are not as well constrained as we record during the last 400 ky from Portuguese Margin sedi-
would like. As these variations are key to refining the radiocar- ments, North Atlantic. Yokoyama et al. (2007) reached the
bon calibration, more records with independent and accurate same conclusion using a similar approach with the data set of
age constraints are needed. Yamazaki and Oda (2002, 2005) and Yamazaki and
Kanamatsu (2007). Yokoyama et al. (2010) showed that vari-
ations in the geomagnetic vertical component were synchro-
5.13.8.6 Frequency of Intensity Fluctuations and the
nous with the first derivative of the eccentricity and suggested
Climatic Connection
that changes in the eccentricity may excite the geomagnetic
Arguments regarding the possible relationship between the variations. Saracco et al. (2009), using complex wavelet trans-
intensity of the geomagnetic field and climate have a long form, found orbital frequencies in paleointensity proxy records
history of more than 30 years (e.g., Wollin et al., 1971). Dis- based on production of cosmogenic nuclide (10Be/9Be) and
cussions in the 1970s were based on the remanent intensity magnetic anomalies of oceanic crust as well as in sedimentary
variations, which were not corrected for differences in magne- relative paleointensity records. Heslop (2007) and Xuan and
tizability of sediments. The relationship was convincingly Channell (2008a), on the other hand, showed in their cross
rejected by Kent (1982), who showed that the remanent inten- wavelet analyses that common power exists at orbital periods
sity variations of the relevant sediment cores were controlled in relative paleointensity records and orbital parameters, but
by climatically induced variations in carbonate contents. Kent they did not have a constant phase relationship, from which
and Opdyke (1977) were the first to suggest the presence of the they concluded that there is no direct causal relationship.
obliquity frequency, 43 ky in a normalized intensity record, However, this does not exclude a possibility for a nonlinear
but this idea was not seriously discussed at that time. interaction or a linear interaction in a different physical dimen-
Rapid progress in relative paleointensity studies revived the sion (Yokoyama et al., 2010).
orbital modulation issue in the late 1990s, and it has been Whether the orbital frequencies found in sedimentary
heatedly argued since then. Tauxe and Shackleton (1994) paleointensity records reflect geomagnetic field behavior or
found significant power in the power spectrum of a record not has been discussed mainly on the following three points:
from the Ontong Java Plateau in the 30–50 ky band but significance and stability of the orbital periodicities, error in
showed that the intensity fluctuations came in and out of age control, and lithologic contamination to paleointensity
phase of the associated oxygen isotopic record, arguing against records. Guyodo and Valet (1999) argued that there is no
a strong relationship between climate and paleointensity. In stable periodicity during the Brunhes chron by a spectrum
contrast, Channell et al. (1998) and Channell (1999) pro- analysis on the SINT-800 stack using sliding windows. How-
posed a 40 ky obliquity frequency from a power spectrum ever, the orbital modulation may be a nonstationary process.
analysis of their relative paleointensity records during the Sato et al. (1998) suggested that there is no constant period but
Brunhes chron obtained from ODP Sites 983 and 984 in the continuous shifts between 50 and 140 ky based on a stacked
North Atlantic. They interpreted it as geomagnetic field behav- paleointensity record from three cores in the western equato-
ior from the observations that no power exists at 40 ky in rial Pacific during the last 1.1 My. Horng et al. (2003) argued
100 ky
0 10
Paleointensity
95%
Power
1 lev
el
AR
1
0.1
1
0 0.01 0.02 0.03 0.04 0.05 0.06
ARM
10-2
Age (Ma)
Power 10-3
10-4
0 0.01 0.02 0.03 0.04 0.05 0.06
1
Coherency
Squared coherence
0.8
95% significance
0.6
0.4
0.2
3
0 0.1 0.1 0.2 0.03 0.05 0
Relative Normalizer 0 0.01 0.02 0.03 0.04 0.05 0.06
paleointensity ARM ARM/SIRM (d) Frequency (ky-1)
(NRM/ARM) (30 mT AFD, A/m)
(a) (b) (c)
Figure 31 (a) Relative paleointensity, (b) normalizer ARM, and (c) ARM/SIRM ratio of Core MD982187 in the western equatorial Pacific (Yamazaki and
Oda, 2005). This core is one of six cores that were used to construct the EPAPIS-3 Ma curve shown in Figure 26 and covers the longest period of
time among them. (d) Power spectra of relative paleointensity and the normalizer (ARM), and cross correlation between them (Yamazaki and Oda,
2005). The relative paleointensity variations contain quasiperiod of 100 ky but no significant correlation between relative paleointensity and the
normalizer. The statistical significance of the spectral peaks was tested against the red-noise background from a first-order autoregressive (AR1)
process, and the 95% confidence level is indicated.
that the orbital frequencies in paleointensity variations are not tens of thousand years observed from sedimentary paleointen-
statistically significant by applying a wavelet analysis on their sity records could be due to aliasing, an artifact of coarse
relative paleointensity record during the last 2.14 My from the sampling, and simulated using archaeointensity data with a
Western Philippine Sea. On the other hand, Teanby and 2 ky period how false longer periods appear by aliasing. How-
Gubbins (2000) proposed that the periodicities of several ever, orbital periodicities have been reported even from
sediment cores with significant variations in sedimentation site or a set of records from nearby sites that belong to a
rates, which is difficult to be explained by aliasing. Guyodo similar sedimentary environment, and thus, further studies
and Channell (2002) performed numerical simulation of are still required. For isolating true geomagnetic signals and
paleointensity records with various sedimentation rates and climatic contamination components from normalized inten-
variable quality of age control and showed how spectral infor- sity records, principal component analysis will be useful as
mation is lost with decreasing sedimentation rates: the power demonstrated by Valet et al. (2011).
spectra are reliable for periods as short as 4 ky in records with a On the possibility of orbital modulation of the geomag-
sedimentation rate of 15 cm ky1 with good age control, netic field, a relationship of excursions (see Chapter 5.10 in
whereas periods of only c.50 ky or longer are reliable in records the volume) and reversals with paleoclimate has also been
with a sedimentation rate of 1 cm ky1. McMillan et al. (2002, discussed since the 1970s. Rampino (1979, 1981) suggested
2004) evaluated effects of various sources of errors such as that excursions may have occurred at about 100 ky intervals
dating errors, misidentified tie points, changes in sedimenta- and the ages of the excursions seem to coincide with times of
tion rate, and the effect of nondipole components. They sim- peak eccentricity of the Earth’s orbit (but see Rampino and
ulated coherence of records among various sites and evaluated Kent, 1983). Worm (1997) revisited the problem and sug-
the accuracy of a stacked record suggesting that dipole varia- gested that excursions and reversals tend to have occurred
tions with periods of longer than 20 ky can be recovered (but during periods of global cooling or during cold stages. On
shorter ones would be problematic). the contrary, Kent and Carlut (2001) rejected the relation-
Strong arguments against orbital modulation of the geo- ship. They concluded that six excursions in the Brunhes chron
magnetic field come from possible lithologic contamination to and 21 reversals since 5.3 Ma have no tendency to occur at a
sedimentary paleointensity records. Because paleointensity consistent amplitude or phase of obliquity and eccentricity.
records during the last 200 ky look coherent with the oxygen The number of possible excursions during the Brunhes chron
isotope curve, Kok (1999) suspected that sedimentary paleoin- has increased significantly (up to ca. 20) (Lund et al., 2001).
tensity records including those derived from 10Be are con- Hence, the problem of the possible connection between
trolled by paleoclimate due to inadequate normalization. excursions and paleoclimate is not independent of the argu-
Guyodo et al. (2000) and Xuan and Channell (2008a) found ments on the orbital frequencies in relative paleointensity.
correlation between paleointensity and magnetic grain-size Fuller (2006) revived the problem by suggesting relationships
proxy (ARM/k or ARM/IRM) in orbital frequencies and con- between the obliquity signal and intensity, excursions, and
cluded that orbital frequencies in relative paleointensity reversals. His arguments include that reversals tend to occur
records are caused by lithologic contamination. As mentioned when average amplitude of the obliquity is low with a partic-
before in Section 5.13.4.4, changes in relative abundance of ular phase within the obliquity signal. This was supported by
biogenic and terrigenous magnetic mineral components Thouveny et al. (2008). However, a statistical test by Xuan
appear as variations in ARM/SIRM ratio (Yamazaki and and Channell (2008b) with a longer reversal sequence did not
Ikehara, 2012), and such changes may not be compensated support the conclusion. Besides paleointensity and excur-
well by the normalization with either ARM or SIRM, resulting sions, a discussion of orbital frequencies in paleomagnetic
in contamination to relative paleointensity. directions has been revitalized. Yamazaki and Oda (2002)
The possibility of orbital modulation of paleointensity reported a 100 ky periodicity in an inclination record dur-
cannot be excluded only from the existence of correlation ing the last 2.2 My from the western equatorial Pacific,
between records of relative paleointensity and magnetic prop- whereas Roberts et al. (2003) concluded that it is not statisti-
erties because the two can also have coherency if the orbital cally significant.
parameters affect both the geomagnetic field and depositional As a possible mechanism for relation between the geomag-
environment. To solve the problem, it is necessary to under- netic field and climate, orbital forcing on the geodynamo
stand quantitative relation between the magnitude of mag- through changes in the moment of inertia by ice-volume
netic property (e.g., magnetic grain-size and mineralogy) changes has often been invoked (e.g., Rampino, 1979;
changes and the magnitude of induced changes in normal- Worm, 1997), but the opposite way of the geomagnetic–
ized intensity. At present, we cannot even predict whether climate connection, that is, possibility of a geomagnetic con-
normalized intensity increases or decreases when magnetic trol on climate, has become a matter of debate (Courtillot
grain size increases in a certain grain-size range. It is also et al., 2007); changes on galactic cosmic ray flux, which is
important to examine phase relationships in coherency ana- controlled by the strength and morphology of the geomag-
lyses. Patterns of lithologic and magnetic property variations netic field, may affect cloud cover and then global tempera-
induced by paleoclimatic changes may vary place to place; for ture (Svensmark and Friis-Christensen, 1997). Kitaba et al.
example, magnetic grain size would increase in a certain (2013) reported climate cooling at paleointensity minima
period of time in some areas, but in other areas, it would during the Matuyama–Brunhes and lower Jaramillo polarity
decrease in the same period of time. Paleointensity, on the reversals.
other hand, should be globally synchronous. Thus, even As noted in Section 5.13.4, there can be a ‘stealth’ link
when paleointensity and magnetic property variations have between lithologic factors, like clay content, which are con-
coherency, lithologic contamination is suggested to be minor trolled by climate and the relative paleointensity records,
if the phase angles between the two from various places differ which would be difficult to detect using the standard methods
significantly. The conclusion of lithologic contamination by of normalization. To date, the significance and implications of
Guyodo et al. (2000) and Xuan and Channell (2008a) men- possible climatic controls on paleointensity have not yet been
tioned earlier was based on analyses of a record from a single adequately addressed.
5.13.9 Conclusions References

Aitken MJ, Allsop AL, Bussell GD, and Winter MB (1988) Determination of the intensity
Scientists have dreamed of analyzing the ancient magnetic
of the Earth’s magnetic field during archeological times: Reliability of the Thellier
field intensity for over four centuries. Since the first serious technique. Reviews of Geophysics 26: 3–12.
attempts to acquire paleointensity data in the 1930s, experi- Anson GL and Kodama KP (1987) Compaction-induced inclination shallowing of the
mental design has improved dramatically. The theoretical post-depositional remanent magnetization in a synthetic sediment. Geophysical
foundations for interpreting paleointensity data from both Journal of the Royal Astronomical Society 88: 673–692.
Barbetti M (1977) Measurements of recent geomagnetic secular variation in
thermal and depositional remanences are steadily southeastern Australia and the question of dipole wobble. Earth and Planetary
improving, and there has been explosive growth in publica- Science Letters 36: 207–218.
tions with paleointensity data. The data are slowly being Bard E, Hamelin B, Fairbanks RG, and Zindler A (1990) Calibration of the 14C timescale
contributed to the communal database whose scope has over the past 30,000 years using mass spectrometric U-Th ages from Barbados
corals. Nature 345: 405–410.
increased dramatically recently.
Bassinot FC, Labeyrie LD, Vincent E, Quidelleur X, Shackleton NJ, and Lancelot Y
We have highlighted some of the major topics involving (1994) The astronomical theory of climate and the age of the Brunhes–Matuyama
paleointensity data in this treatise. While these topics are still magnetic reversal. Earth and Planetary Science Letters 126: 91–108.
fresh and arguments abound, we can make the following state- Bender M, Sowers T, Dickson ML, et al. (1994) Climate correlations between
ments regarding paleointensity: Greenland and Antarctica during the past 100,000 years. Nature 372(6507):
663–666.
1. Everyone agrees that more and better data could resolve Ben-Yosef E, Ron H, Tauxe L, et al. (2008) Application of copper slag in geomagnetic
archaeointensity research. Journal of Geophysical Research 113. http://dx.doi.org/
many of the current debates. Experiments with built-in
10.1029/2007JB005235.
assessments of the fundamental assumptions of the Ben-Yosef E, Tauxe L, Ron H, et al. (2008) A new approach for geomagnetic
method and the ability to estimate reliability indices are archeointensity research: Insights on ancient metallurgy in the Southern Levant.
essential. Data sets are being contributed to the MagIC Journal of Archaeological Science 35: 2863–2879.
database, including measurements and full documentation Ben-Yosef E, Tauxe L, Levy TE, Shaar R, Ron H, and Najjar M (2009) Archaeomagnetic
intensity spike recorded in high resolution slag deposit from historical biblical
of methods and data processing. This new generation of a archaeology site in southern Jordan. Earth and Planetary Science Letters
database has the potential to go a long way toward settling 287: 529–539.
some of the major issues discussed in this chapter. Bergh HW (1970) Paleomagnetism of the Stillwater complex. Montana Paleogeophysics
2. In general, the ancient magnetic field has been highly 17: 143–158.
Besse J and Courtillot V (2002) Apparent and true polar wander and the geometry of the
variable on both short and long timescales. There have
geomagnetic field over the last 200 Myr. Journal of Geophysical Research
been extended periods of time with intensities lower 107. http://dx.doi.org/10.1029/2000JB000050.
than the present field, but there have also been intervals Biggin A (2010) Paleointensity database updated and upgraded. EOS, Transactions
with field strengths greater than the present field in the AGU 91: 15.
past. These periods of increased field strength may be Biggin AJ, Bohnel HN, and Zuniga FR (2003) How many paleointensity determinations
are required from a single lava flow to constitute a reliable average? Geophysical
related to the length of the polarity interval in which Research Letters 30(11).
they are found. Biggin A, Perrin M, and Shaw J (2007) A comparison of a quasi-perpendicular method
3. The geomagnetic field in the Archean appears to have been of absolute paleointensity determination with other thermal and microwave
‘alive and well.’ techniques. Earth and Planetary Science Letters 257: 564–581.
Bleil U and von Dobeneck T (1999) Geomagnetic Events and Relative Paleointensity
4. Arguments against the ‘asymmetrical sawtooth’ in paleoin-
Records: Clues to High-Resolution Paleomagnetic Chronostratigraphies of Late
tensity data associated with polarity reversals appear to be Quaternary Marine Sediments, pp. 635–654. Berlin: Springer-Verlag.
winning, while arguments about the coherence of paleoin- Bohnel H, Biggin AJ, Walton D, Shaw J, and Share JA (2003) Microwave
tensity with orbital frequencies and the ability to correlate palaeointensities from a recent Mexican lava flow, baked sediments and reheated
globally paleointensity features on a millennial scale pottery. Earth and Planetary Science Letters 214(1–2): 221–236.
Bol’shakov A and Shcherbakova V (1979) A thermomagnetic criterion for determining
remain unresolved. the domain structure of ferrimagnetics. Izvestiya, Physics of the Solid Earth
15: 111–117.
Bol’shakov A and Solodonikov GM (1980) Paleomagnetic data on the intensity
of the magnetic field of the Earth. Izvestiya, Physics of the Solid Earth
16: 602–614.
Acknowledgments Bowles J, Gee JS, Burgess K, and Cooper R (2011) Timing of magnetite formation in
basaltic glass: Insights from synthetic analogs and relevance for geomagnetic
We acknowledge fruitful discussion with A. Genevez, Y. paleointensity analysis. Geochemistry, Geophysics, Geosystems 12.
Bowles J, Gee JS, Kent D, Bergmanis E, and Sinton J (2005) Cooling rate effects on
Yamamoto, J. S. Gee, J.-P. Valet, C. Constable, A. Goguitchaitch-
paleointensity estimates in submarine basaltic glass and implications for dating
vili, K. Lawrence, Y. Yu, A. Biggin, J. Tarduno, A. P. Roberts, and young flows. Geochemistry, Geophysics, Geosystems 6: Q07002. http://dx.doi.org/
many others. The manuscript was significantly improved 10.1029/2004GC000900.
through reviews by J.-P. Valet, M. Kono, Y. Yamamoto, and Bowles J, Gee J, Kent DV, Perfit M, Soule A, and Fornari D (2006) Paleointensity
anonymous. We thank the many individuals who supplied applications to timing and extent of eruptive activity, 9 –10 N East Pacific rise.
Geochemistry, Geophysics, Geosystems 7: Q06006.
data for the MagIC database compilation: J. E. T. Channell, C. Bowles J, Tauxe L, Gee J, McMillan D, and Cande SC (2003) Source of tiny wiggles in
Laj, Y. Guyodo, M. Perrin, L. Schnepp, J. Stoner, N. Thouveny, Chron C5: A comparison of sedimentary relative intensity and marine magnetic
and J.-P. Valet. This work was in part supported by NSF Grants anomalies. Geochemistry, Geophysics, Geosystems 4(6): 1049. http://dx.doi.org/
EAR 0337399 and EAR 1013192 and the Grants-in-Aid for Sci- 10.1029/2002GC000489.
Boyd M (1986) A new method for measuring paleomagnetic intensities. Nature
entific Research ((A)(2) No. 16204034) of the Japan Society for
3196: 208–209.
the Promotion of Science to TY. Acknowledgment is also made Brachfeld SA and Banerjee SK (2000) A new high-resolution geomagnetic relative
to the Donors of the American Chemical Society Petroleum paleointensity record for the North American Holocene: A comparison of sedimentary
Research Fund, for partial support of this research. and absolute intensity data. Journal of Geophysical Research 105(B1): 821–834.
Cande SC and Kent DV (1995) Revised calibration of the geomagnetic polarity timescale Cromwell G, Tauxe L, Staudigel H, Constable C, Koppers A, and Pedersen R-B (2013)
for the late Cretaceous and Cenozoic. Journal of Geophysical Research Evidence for long-term hemispheric asymmetry in the geomagnetic field: Results
100: 6093–6095. from high northern latitudes. Geochemistry, Geophysics, Geosystems 14. http://dx.
Carcaillet JT, Thouveny N, and Bourles DL (2003) Geomagnetic moment instability doi.org/10.1002/ggge.20174.
between 0.6 and 1.3 Ma from cosmonuclide evidence. Geophysical Research Letters Cronin M, Tauxe L, Constable C, Selkin P, and Pick T (2001) Noise in the quiet zone.
30(15): 1792. Earth and Planetary Science Letters 190: 13–30.
Carter-Stiglitz B, Jackson M, and Moskowitz B (2002) Low-temperature remanence in Dansgaard W, Johnsen S, Clausen H, et al. (1993) Evidence for general instability of
stable single domain magnetite. Geophysical Research Letters 29. past climate from a 250-kyr ice core record. Nature 364: 218–220.
Carter-Stiglitz B, Moskowitz B, and Jackson M (2003) Correction to “Low-temperature de Groot L, Dekkers M, and Mullender T (2012) Exploring the potential of acquisition
remanence in stable single domain magnetite”. Geophysical Research Letters curves of the anhysteretic remanent magnetization as a tool to detect subtle
30(21): 2113. magnetic alteration induced by heating. Physics of the Earth and Planetary Interiors
Channell JET (1999) Geomagnetic paleointensity and directional secular variation at 194–195: 71–84.
Ocean Drilling Program (ODP) Site 984 (Bjorn Drift) since 500 ka: Comparisons Deamer GA and Kodama KP (1990) Compaction-induced inclination shallowing in
with ODP Site 983 (Gardar Drift). Journal of Geophysical Research: Solid Earth synthetic and natural clay-rich sediments. Journal of Geophysical Research
104(B10): 22937–22951. 95: 4511–4529.
Channell JET, Curtis JH, and Flower BP (2004) The Matuyama–Brunhes boundary Dekkers M and B€ohnel H (2006) Reliable absolute paleointensities
interval (500–900 ka) in North Atlantic drift sediments. Geophysical Journal independent of magnetic domain state. Earth and Planetary Science Letters
International 158(2): 489–505. 248: 508–517.
Channell JET, Hodell DA, and Lehman B (1997) Relative geomagnetic paleointensity deMenocal PB, Ruddiman WF, and Kent DV (1990) Depth of p-DRM acquisition in
and 18O at ODP Site 983 (Gardar Drift, North Atlantic) since 350 ka. Earth and deep-sea sediments – A case study of the B/M reversal and oxygen isotopic stage
Planetary Science Letters 153: 103–118. 19.1. Earth and Planetary Science Letters 99: 1–13.
Channell JET, Hodell DA, McManus J, and Lehman B (1998) Orbital modulation of the Demory F, Nowaczyk NR, Witt A, and Oberhansli H (2005) High-resolution
Earth’s magnetic field intensity. Nature 394(6692): 464–468. magnetostratigraphy of late Quaternary sediments from Lake Baikal, Siberia: Timing
Channell JET and Kleiven HF (2000) Geomagnetic palaeointensities and of intracontinental paleoclimatic responses. Global and Planetary Change 46(1–4):
astrochronological ages for the Matuyama-Brunhes boundary and the boundaries of 167–186.
the Jaramillo Subchron: Palaeomagnetic and oxygen isotope records from ODP Site Denham CR and Chave AD (1982) Detrital remanent magnetization: Viscosity theory of
983. Philosophical Transactions of the Royal Society of London Series A: the lock-in zone. Journal of Geophysical Research 87: 7126–7130.
Mathematical Physical and Engineering Sciences 358(1768): 1027–1047. Dinares-Turell J, Sagnotti L, and Roberts AP (2002) Relative geomagnetic
Channell JET, Mazaud A, Sullivan P, Turner S, and Raymo ME (2002) Geomagnetic paleointensity from the Jaramillo Subchron to the Matuyama/Brunhes boundary as
excursions and paleointensities in the Matuyama Chron at Ocean Drilling Program recorded in a Mediterranean piston core. Earth and Planetary Science Letters
Sites 983 and 984 (Iceland Basin). Journal of Geophysical Research: Solid Earth 194(3–4): 327–341.
107(B6). Dunlop D and Argyle K (1997) Thermoremanence, anhysteretic remanence and
Channell J, Stoner J, Hodell D, and Charles C (2000) Geomagnetic paleointensity for susceptibility of submicron magnetites: Nonlinear field dependence and variation
the last 100 kyr from the sub-antarctic South Atlantic: A tool for inter-hemispheric with grain size. Journal of Geophysical Research 102: 20199–20210.
correlation. Earth and Planetary Science Letters 175: 145–160. Dunlop D and Ozdemir O (1997) Rock Magnetism: Fundamentals and Frontiers,
Channell JET, Xuan C, and Hodell D (2009) Stacking paleointensity and oxygen isotope Cambridge Studies in Magnetism. Cambridge: Cambridge University Press.
data for the last 1.5 Myr (PISO-1500). Earth and Planetary Science Letters Dunlop DJ and Xu S (1994) Theory of partial thermoremanent magnetization in
283: 14–23. multidomain grains: 1. Repeated identical barriers to wall motion (single
Chauvin A, Roperch P, and Levi S (2005) Reliability of geomagnetic paleointensity data: microcoercivity). Journal of Geophysical Research 99: 9005–9023.
The effects of the NRM fraction and concave-up behavior on paleointensity Evans HF and Channell JET (2003) Upper Miocene magnetic stratigraphy at ODP site
determinations by the Thellier method. Physics of the Earth and Planetary Interiors 1092 (sub-Antarctic South Atlantic): Recognition of ‘cryptochrons’ in C5n.2n.
150(4): 265–286. Geophysical Journal International 153(2): 483–496.
Cisowski S and Fuller M (1986) Lunar Paleointensities via the IRM(s) Normalization Fabian K (2003) Some additional parameters to estimate domain state from isothermal
Method and the Early Magnetic History of the Moon, pp. 635–654. Houston, TX: magnetization measurements. Earth and Planetary Science Letters 213(3–4):
Lunar and Planetary Science Institute. 337–345.
Clement BM (2004) Dependence of the duration of geomagnetic polarity reversals on Fabian K and Leonhardt R (2010) Multiple-specimen absolute paleointensity
site latitude. Nature 428(6983): 637–640. determination: An optimal protocol including pTRM normalization, domain-state
Clement BM and Kent DV (1986) Short polarity intervals within the Matuyama: correction, and alteration test. Earth and Planetary Science Letters 297: 84–94.
Transitional field records from hydraulic piston cored sediments from the North Folgheraiter M (1899) Sur les variations séculaires de l’inclinaison magnétique dans
Atlantic. Earth and Planetary Science Letters 81: 253–264. l’antiquité. Journal de Physique 5: 660–667.
Coe RS (1967) The determination of paleo-intensities of the Earth’s magnetic Fox JMW and Aitken MJ (1980) Cooling-rate dependence of thermoremanent
field with emphasis on mechanisms which could cause non-ideal behavior magnetization. Nature 283: 462–463.
in Thellier’s method. Journal of Geomagnetism and Geoelectricity Frank M, Schwarz B, Baumann S, Kubik P, Suter M, and Mangini A (1997) A 200 kyr
19: 157–178. record of cosmogenic radionuclide production rate and geomagnetic field intensity
Coe RS, Grommé S, and Mankinen EA (1978) Geomagnetic paleointensities from from 10Be in globally stacked deep-sea sediments. Earth and Planetary Science
radiocarbon-dated lava flows on Hawaii and the question of the Pacific nondipole Letters 149: 121–129.
low. Journal of Geophysical Research 83: 1740–1756. Fuller M (2006) Geomagnetic field intensity, excursions, reversals and the 41,000-yr
Coffey W, Kalmykov Y, and Waldron J (1996) The Langevin Equation with Applications obliquity signal. Earth and Planetary Science Letters 245: 605–615.
in Physics, Chemistry and Electrical Engineering. Vol. 11 of World Scientific Series Gallet Y and Le Goff M (2006) High-temperature archeointensity measurements from
in Contemporary Chemical Physics. Singapore: World Scientific. Mesopotamia. Earth and Planetary Science Letters 241: 159–173.
Collinson DW (1965) DRM in sediments. Journal of Geophysical Research Gattacceca J and Rochette P (2004) Toward a robust normalized magnetic
70: 4663–4668. paleointensity method applied to meteorites. Earth and Planetary Science Letters
Constable C and Tauxe L (1996) Towards absolute calibration of sedimentary 227: 377–393.
paleointensity records. Earth and Planetary Science Letters 143: 269–274. Gee JS and Bowles J (2010) Paleointensity estimates from ignimbrites: An evaluation of
Constable CG, Tauxe L, and Parker RL (1998) Analysis of 11 Myr of geomagnetic the Bishop Tuff. Geochemistry, Geophysics, Geosystems 11: Q03010. http://dx.doi.
intensity variation. Journal of Geophysical Research: Solid Earth 103(B8): org/03010.01029/02009GC002834.
17735–17748. Gee J, Cande S, Hildebrand J, Donnelly K, and Parker R (2000) Geomagnetic intensity
Cottrell R and Tarduno J (2000) Late Cretaceous true polar wander: Not so fast. Science variations over the past 780 kyr obtained from near-seafloor magnetic anomalies.
288: 2283. Nature 408: 827–832.
Courtillot V, Gallet Y, Le Mouël J-L, Futeau F, and Genevey A (2007) Are there Genevey A and Gallet Y (2003) Eight thousand years of geomagnetic field intensity
connections between the Earth’s magnetic field and climate? Earth and Planetary variations in the eastern Mediterranean. Journal of Geophysical Research
Science Letters 253: 328–339. 108. http://dx.doi.org/10.1029/2001JB001612.
Cox AV (1968) Lengths of geomagnetic polarity intervals. Journal of Geophysical Gibbs R (1985) Estuarine flocs: Their size, settling velocity and density. Journal of
Research 73: 3247–3260. Geophysical Research 90: 3249–3251.
Goguitchaichvili A, Alva-Valdivia LM, Luis M, Rosas-Elguera J, Urrutia-Fucugauchi J, International Commission on Stratigraphy (2004) Geologic Time Scale 2004.
and Sole J (2004) Absolute geomagnetic paleointensity after the Cretaceous Normal Cambridge: Cambridge University Press.
Superchron and just prior to the Cretaceous-Tertiary transition. Journal of Irving E (1964) Paleomagnetism and Its Application to Geological and Geophysical
Geophysical Research 109: B01105. http://dx.doi.org/10.1029/2003JB002477. Problems. New York: Wiley.
Granot R, Dymont J, and Gallet Y (2012) Geomagnetic field variability during the Johnson HP, Lowrie W, and Kent DV (1975) Stability of anhysteretic remanent
Cretaceous Normal Superchron. Nature Geoscience 5: 220–223. magnetization in fine and coarse magnetite and maghemite particles. Geophysical
Granot R, Tauxe L, Gee JS, and Ron H (2007) A view into the Cretaceous geomagnetic Journal of the Royal Astronomical Society 41: 1–10.
field from analysis of gabbros and submarine glasses. Earth and Planetary Science Johnson EA, Murphy T, and Torreson OW (1948) Pre-history of the Earth’s magnetic
Letters 256: 1–11. field. Terrestrial Magnetism and Atmospheric Electricity 53: 349–372.
Grootes P and Stuiver M (1997) Oxygen 18/16 variability in Greenland snow and ice Juarez M and Tauxe L (2000) The intensity of the time averaged geomagnetic field: The
with 10–3 to 105 year time resolution. Journal of Geophysical Research last 5 m.y. Earth and Planetary Science Letters 175: 169–180.
102: 26455–26470. Juarez T, Tauxe L, Gee JS, and Pick T (1998) The intensity of the Earth’s magnetic field
Guyodo Y, Acton GD, Brachfeld S, and Channell JET (2001) A sedimentary over the past 160 million years. Nature 394: 878–881.
paleomagnetic record of the Matuyama chron from the Western Antarctic margin Katari K and Bloxham J (2001) Effects of sediment aggregate size on DRM intensity:
(ODP Site 1101). Earth and Planetary Science Letters 191(1–2): 61–74. A new theory. Earth and Planetary Science Letters 186(1): 113–122.
Guyodo Y and Channell JET (2002) Effects of variable sedimentation rates and age Katari K, Tauxe L, and King J (2000) A reassessment of post depositional remanent
errors on the resolution of sedimentary paleointensity records. Geochemistry, magnetism: Preliminary experiments with natural sediments. Earth and Planetary
Geophysics, Geosystems 3: 1–18, art. no. –1048. Science Letters 183: 147–160.
Guyodo Y, Gaillot P, and Channell JET (2000) Wavelet analysis of relative geomagnetic Kent DV (1973) Post-depositional remanent magnetization in deep-sea sediment.
paleointensity at ODP Site 983. Earth and Planetary Science Letters 184(1): Nature 246: 32–34.
109–123. Kent DV (1982) Apparent correlation of paleomagnetic intensity and climatic records in
Guyodo Y, Richter C, and Valet JP (1999) Paleointensity record from Pleistocene deep-sea sediments. Nature 299: 538–539.
sediments off the California Margin. Journal of Geophysics 104: 22953–22965. Kent DV and Carlut J (2001) A negative test of orbital control of geomagnetic reversals
Guyodo Y and Valet JP (1999) Global changes in intensity of the Earth’s magnetic field and excursions. Geophysical Research Letters 28(18): 3561–3564.
during the past 800 kyr. Nature 399(6733): 249–252. Kent DV and Opdyke ND (1977) Paleomagnetic field intensity variation recorded in a
Guyodo Y and Valet JP (2006) A comparison of relative paleointensity records of the Brunhes epoch deep-sea sediment core. Nature 266: 156–159.
Matuyama Chron for the period 0.75–1.25 Ma. Physics of the Earth and Planetary Kent DV and Schneider DA (1995) Correlation of paleointensity variation records in the
Interiors 156: 205–212. Brunhes/Matuyama polarity transition interval. Earth and Planetary Science Letters
Haag M (2000) Reliability of relative palaeointensities of a sediment core with 129: 135–144.
climatically-triggered strong magnetisation changes. Earth and Planetary Science King RF (1955) The remanent magnetism of artificially deposited sediments. Monthly
Letters 180(1–2): 49–59. Notices of the Royal Astronomical Society, Geophysical Supplement 7: 115–134.
Hale CJ (1987) Palaeomagnetic data suggest link between the Archean-Proterozoic King JW, Banerjee SK, and Marvin J (1983) A new rock magnetic approach to selecting
boundary and inner-core nucleation. Nature 329: 233–237. sediments for geomagnetic paleointensity studies: Application to paleointensity for
Halgedahl S, Day R, and Fuller M (1980) The effect of cooling rate on the intensity of the last 4000 years. Journal of Geophysical Research 88: 5911–5921.
weak-field TRM in single-domain magnetite. Journal of Geophysical Research King RF and Rees AI (1966) Detrital magnetism in sediments: An examination of some
95: 3690–3698. theoretical models. Journal of Geophysical Research 71: 561–571.
Harrison CGA (1966) The paleomagnetism of deep sea sediments. Journal of Kirschvink JL (1980) The least-squares line and plane and the analysis of
Geophysical Research 71: 3033–3043. paleomagnetic data. Geophysical Journal of the Royal Astronomical Society
Hartl P and Tauxe L (1996) A precursor to the Matuyama/Brunhes transition-field 62: 699–718.
instability as recorded in pelagic sediments. Earth and Planetary Science Letters Kissel C and Laj C (2004) Improvements in procedure and paleointensity selection
138: 121–135. criteria (PICRIT 03) for Thellier and Thellier determinations: Application to
Hayashida A, Verosub KL, Heider F, and Leonhardt R (1999) Magnetostratigraphy and Hawaiian basaltic long cores. Physics of the Earth and Planetary Interiors 147(2–3):
relative palaeointensity of late Neogene sediments at ODP Leg 167 Site 1010 off 155–169.
Baja California. Geophysical Journal International 139(3): 829–840. Kissel C, Laj C, Labeyrie L, Dokken T, Voelker A, and Blamart D (1999) Rapid climatic
Hays JD, Imbrie J, and Shackleton NJ (1976) Variations in the Earth’s orbit: Pacemaker variations during marine isotopic stage 3: Magnetic analysis of sediments from
of the ice ages. Science 194: 1121–1132. Nordic Seas and North Atlantic. Earth and Planetary Science Letters 171(3):
Heller F and Liu T (1982) Magnetostratigraphical dating of loess deposits in China. 489–502.
Nature 300: 431–433. Kitaba I, Hyodo M, Katoh S, Dettman D, and Sato H (2013) Midlatitude cooling caused
Herrero-Bervera E and Valet JP (2005) Absolute paleointensity and reversal records by geomagnetic field minimum during polarity reversal. Proceedings of the National
from the Waianae sequence (Oahu, Hawaii, USA). Earth and Planetary Science Academy of Sciences of the United States of America 110: 1215–1220.
Letters 234: 279–296. Kletetschka G, Acuna MH, Kohout T, Wasilewski PJ, and Connerney JEP (2004) An
Heslop D (2007) Are hydrodynamic shape effects important when modelling the empirical scaling law for acquisition of thermoremanent magnetization. Earth and
formation of depositional remanent magnetization? Geophysical Journal Planetary Science Letters 226(3–4): 521–528.
International 171: 1029–1035. Kletetschka G, Fuller M, Kohout T, et al. (2006) TRM in low magnetic fields: A minimum
Hill MJ, Gratton M, and Shaw J (2002) A comparison of thermal and microwave field that can be recorded by large multidomain grains. Physics of the Earth and
paleomagnetic techniques using lava containing laboratory induced remanence. Planetary Interiors 154: 290–298.
Geophysical Journal International 151: 157–163. Knudsen M, Henderson G, Frank M, Niocaill C, and Kubik P (2008) In-phase anomalies in
Hoffman KA and Biggin AJ (2005) A rapid multi-sample approach to the determination Beryllium-10 production and palaeomagnetic field behaviour during the Iceland
of absolute paleointensity. Journal of Geophysical Research 110: B12108. http://dx. Basin geomagnetic excursion. Earth and Planetary Science Letters 265: 588–599.
doi.org/10.1029/2005JB003646. Koenigsberger J (1936) Die abhaengigkeit der natuerlichen remanenten magnetisierung
Hoffman KA, Constantine VL, and Morse DL (1989) Determination of absolute bei eruptivgesteinen von deren alter und zusammensetzung. Beitrage Angewandte
palaeointensity using a multi-specimen procedure. Nature 339: 295–297. Geophysik 5: 193–246.
Hoffmann D and Fabian K (2009) Correcting relative paleointensity records for Koenigsberger J (1938a) Natural residual magnetism of eruptive rocks, Pt I. Terrestrial
variations in sediment composition: Results from a South Atlantic stratigraphic Magnetism and Atmospheric Electricity 43: 119–127.
network. Earth and Planetary Science Letters 284: 34–43. Koenigsberger J (1938b) Natural residual magnetism of eruptive rocks, Pt II. Terrestrial
Horng CS, Roberts AP, and Liang WT (2003) A 2.14-Myr astronomically tuned record Magnetism and Atmospheric Electricity 43: 299–320.
of relative geomagnetic paleointensity from the western Philippine Sea. Journal of Kok YS (1999) Climatic influence in NRM and Be-10-derived geomagnetic
Geophysical Research: Solid Earth 108(B1). paleointensity data. Earth and Planetary Science Letters 166(3–4): 105–119.
Hughen K, Lehman S, Southon J, et al. (2004) C-14 activity and global carbon cycle Kok YS and Tauxe L (1996a) Saw-toothed pattern of relative paleointensity records and
changes over the past 50,000 years. Science 303(5655): 202–207. cumulative viscous remanence. Nature 137: 95–99.
Imbrie J, Hays JD, Martinson DG, et al. (1984) The orbital theory of Pleistocene climate: Kok YS and Tauxe L (1996b) Saw-toothed pattern of sedimentary paleointensity records
Support from a revised chronology of the marine d18 O record. In: Part 1, explained by cumulative viscous remanence. Earth and Planetary Science Letters
Milankovitch and Climate. Hingham, MA: Reidel. 144: E9–E14.
Kok YS and Tauxe L (1999) A relative geomagnetic paleointensity stack from Ontong- Lund S, Williams T, Acton GD, Clement BM, and Okada M (2001) Brunhes Chron
Java Plateau sediments for the Matuyama. Journal of Geophysical Research: Solid magnetic field excursions recovered from Leg 172 sediments. Proceedings of the
Earth 104(B11): 25401–25413. ODP Science Results, vol. 172. College Station, TX: Ocean Drilling Program.
Kok Y and Tauxe L (2000) Comment on “Saw-toothed variations of relative Macouin M, Valet JP, Besse J, et al. (2003) Low paleointensities recorded in 1 to 2.4 Ga
paleointensity and cumulative viscous remanence: Testing the records and the Proterozoic dykes, Superior Province, Canada*1. Earth and Planetary Science
model”, by L. Meynadier, J.-P. Valet, Y. Guyodo, and C. Richter. Journal of Letters 213(1–2): 79–95.
Geophysical Research 105: 16609–16612. Macri P, Sagnotti L, Dinares-Turell J, and Caburlotto A (2005) A composite record of
Kok YS and Ynsen I (2002) Reply to comment by J.-P. Valet and L. Meynadier Late Pleistocene relative geomagnetic paleointensity from the Wilkes Land Basin
on “A relative geomagnetic paleointensity stack from Ontong-Java Plateau sediments (Antarctica). Physics of the Earth and Planetary Interiors 151(3–4): 223–242.
for the Matuyama” Journal of Geophysical Research 107(B3), EPM 3-1–EPM 3-2. Martinson DG, Pisias NG, Hays JD, Imbrie J Jr., Moore TC, and Shackleton NJ (1987)
Kono M (1971) Intensity of the earth’s magnetic field during the Pliocene and Age dating and the orbital theory of the Ice ages: Development of a high-resolution
Pleistocene in relation to the amplitude of mid-ocean ridge magnetic anomalies. 0–300,000 year chronostratigraphy. Quaternary Research 27: 1–29.
Earth and Planetary Science Letters 11: 10–17. Masarik J and Beer J (1999) Simulation of particle fluxes and cosmogenic nuclide
Kono M (1974) Intensities of the Earth’s magnetic field about 60 m.y. ago determined production in the Earth’s atmosphere. Journal of Geophysical Research 104(D10):
from the Deccan Trap basalts, India. Journal of Geophysical Research 12099–12111.
79: 1135–1141. Mazaud A (1996) ‘Sawtooth’ variation in magnetic intensity profiles and delayed
Kono M and Ueno N (1977) Paleointensity determination by a modified Thellier method. acquisition of magnetization in deep sea cores. Earth and Planetary Science Letters
Physics of the Earth and Planetary Interiors 13: 305–314. 139: 379–386.
Korte M and Constable C (2005) The geomagnetic dipole moment over the last 7000 Mazaud A, Sicre MA, Ezat U, et al. (2002) Geomagnetic-assisted stratigraphy and sea
years – New results from a global model. Earth and Planetary Science Letters surface temperature changes in core MD94–103 (Southern Indian Ocean): Possible
236: 348–358. implications for North–South climatic relationships around H4. Earth and Planetary
Laj C, Kissel C, and Garnier F (1996) Relative geomagnetic field intensity and reversals Science Letters 201(1): 159–170.
for the last 1.8 My from a central equatorial Pacific core. Geophysical Research McClelland E and Briden J (1996) An improved methodology for Thellier-type
Letters 23: 3393–3396. paleointensity determination in igneous rocks and its usefulness for verifying
Laj C, Kissel C, and Lefevre I (1996) Relative geomagnetic field intensity and reversals primary thermoremanence. Journal of Geophysical Research 101: 21995–22013.
from Upper Miocene sections in Crete. Earth and Planetary Science Letters McElhinnhy MW and Evans ME (1968) An investigation of the strength of the
141(1–4): 67–78. geomagnetic field in the early Precambrian. Physics of the Earth and Planetary
Laj C, Kissel C, Mazaud A, Channell JET, and Beer J (2000) North Atlantic Interiors 1: 485–497.
palaeointensity stack since 75 ka (NAPIS-75) and the duration of the Laschamp McElhinny MW (1973) Paleomagnetism and Plate Tectonics. Cambridge: Cambridge
event. Philosophical Transactions of the Royal Society 358(1768): 1009–1025. University Press.
Laj C, Rais A, Surmont J, et al. (1997) Changes of the geomagnetic field vector obtained McFadden PL and McElhinny MW (1982) Variations in the geomagnetic dipole 2:
from lava sequences on the island of Vulcano (Aeolian Islands, Sicily). Physics of Statistical analysis of VDM’s for the past 5 million years. Journal of Geomagnetism
the Earth and Planetary Interiors 99: 161–177. and Geoelectricity 34: 163.
Lanci L and Lowrie W (1997) Magnetostratigraphic evidence that “tiny wiggles” in the McFadden P and Merrill R (1997) Sawtooth paleointensity and reversals of the
oceanic magnetic anomaly record represent geomagnetic paleointensity variations. geomagnetic field. Physics of the Earth and Planetary Interiors 103: 247–252.
Earth and Planetary Science Letters 148(3–4): 581–592. McMillan DG, Constable CG, and Parker RL (2002) Limitations on stratigraphic
Laskar J, Joutel F, and Boudin F (1993) Orbital, precessional, and insolation quantities for analyses due to incomplete age control and their relevance to sedimentary
the Earth from 20 Myr to +10 Myr. Astronomy and Astrophysics 270(1–2): 522–533. paleomagnetism. Earth and Planetary Science Letters 201(3–4): 509–523.
Lawrence KP, Tauxe L, Staudigel H, et al. (2009) Paleomagnetic field properties near the McMillan DG, Constable C, and Parker RL (2004) Assessing the dipolar signal in
southern hemisphere tangent cylinder. Geochemistry, Geophysics, Geosystems stacked paleointensity records using a statistical error model and geodynamo
10: Q01005. http://dx.doi.org/10.1029/2008GC00207. simulations. Physics of the Earth and Planetary Interiors 145: 37–54.
Le Goff M and Gallet Y (2004) A new three-axis vibrating sample magnetometer for Meynadier L and Valet JP (1995) Relative geomagnetic intensity during the last 4 m.y.
continuous high-temperature magnetization measurements: Applications to paleo- from the equatorial Pacific. Proceedings of the Ocean Drilling Program, Scientific
and archeo-intensity determinations. Earth and Planetary Science Letters 229(1–2): Results, 138: 779–795.
31–43. Meynadier L and Valet JP (1996) Post-depositional realignment of magnetic grains and
Lehman B, Laj C, Kissel C, Mazaud A, Paterne M, and Labeyrie L (1996) asymmetrical saw-tooth patterns o magnetization intensity. Earth and Planetary
Relative changes of the geomagnetic field intensity during the last 280 kyr from Science Letters 126: 109–127.
piston cores in the Acores area. Physics of the Earth and Planetary Interiors Meynadier L, Valet JP, Bassinot F, Shackleton NJ, and Guyodo Y (1994)
93: 269–284. Asymmetrical saw-tooth pattern of the geomagnetic field intensity from equatorial
Leonhardt R, Heider F, and Hayashida A (1999) Relative geomagnetic field intensity sediments in the Pacific and Indian Oceans. Earth and Planetary Science Letters
across the Jaramillo subchron in sediments from the California margin: ODP Leg 126: 109–127.
167. Journal of Geophysical Research 104: 29133–29146. Meynadier L, Valet JP, Guyodo Y, and Richter C (1998) Saw-toothed variations of
Levi S and Banerjee SK (1976) On the possibility of obtaining relative paleointensities relative paleointensity and cumulative viscous remanence: Testing the records and
from lake sediments. Earth and Planetary Science Letters 29: 219–226. the model. Journal of Geophysical Research 103: 7095–7105.
Linsley B and Thunnell R (1990) The record of deglaciation in the Sulu Sea: Evidence Meynadier L, Valet JP, Weeks R, Shackleton NJ, and Hagee VL (1992) Relative
for the Younger Dryas event in the western tropical Pacific. Paleoceanography geomagnetic intensity of the field during the last 140 ka. Earth and Planetary
5: 1025–1039. Science Letters 114: 39–57.
Lisiecki L and Raymo ME (2005) A Pliocene-Pleistocene stack of 57 globally distributed Mitra R and Tauxe L (2009) Full vector model for magnetization in sediments. Earth and
benthic d18O records. Paleoceanography 20: PA1003. http://dx.doi.org/10.1029/ Planetary Science Letters 286: 535–545.
2004PA001071. Mochizuki N, Tsunakawa H, Oishi Y, Wakai S, Wakabayashi K, and Yamamoto Y (2004)
Liu Q, Banerjee SK, Jackson MJ, Deng C, Pan Y, and Zhu R (2005) Inter-profile Paleointensity study of the Oshima 1986 lava in Japan: Implications for the
correlation of the Chinese loess/paleosol sequences during Marine Oxygen Isotope reliability of the Thellier and LTD-DHT Shaw methods. Physics of the Earth and
Stage 5 and indications of pedogenesis. Quaternary Science Reviews 24(1–2): Planetary Interiors 146: 395–416.
195–210. Morimoto C, Otofuji Y, Miki M, Tanaka H, and Itaya T (1997) Preliminary palaeomagnetic
Lovlie R (1974) Post-depositional remanent magnetization in a re-deposited deep-sea results of an archaean dolerite dyke of west Greenland: Geomagnetic field intensity at
sediment. Earth and Planetary Science Letters 21: 315–320. 2.8 Ga. Geophysical Journal International 128: 583–585.
Lu R, Banerjee SK, and Marvin J (1990) Effects of clay mineralogy and the electrical Muxworthy A (2010) Revisiting a domain-state independent method of paleointensity
conductivity of water on the acquisition of depositional remanent magnetization in determination. Physics of the Earth and Planetary Interiors 179: 21–31.
sediments. Journal of Geophysical Research 95: 4531–4538. Nagata T (1961) Rock Magnetism. Tokyo: Maruzen.
Lund SP and Keigwin L (1994) Measurement of the degree of smoothing in sediment Nagata T, Arai Y, and Momose K (1963) Secular variation of the geomagnetic total force
paleomagnetic secular variation records: An example from late Quaternary deep-sea during the last 5000 years. Journal of Geophysical Research 68: 5277–5282.
sediments of the Bermuda Rise, western North Atlantic Ocean. Earth and Planetary Néel L (1949) Théorie du trainage magnétique des ferromagneétiques en grains fines
Science Letters 122: 317–330. avec applications aux terres cuites. Annales de Geophysique 5: 99–136.
Néel L (1955) Some theoretical aspects of rock-magnetism. Advances in Physics Ravilly M, Horen H, Perrin M, Dyment J, Gente P, and Guillou H (2001) NRM intensity
4: 191–243. of altered oceanic basalts across the MAR (21 degrees N, 0–1.5 Ma): A record of
Nowaczyk NR and Antonow M (1997) High-resolution magnetostratigraphy of four geomagnetic palaeointensity variations? Geophysical Journal International 145(2):
sediment cores from the Greenland Sea.1. Identification of the Mono Lake 401–422.
excursion, Laschamp and Biwa I Jamaica geomagnetic polarity events. Geophysical Riisager P and Abrahamsen N (2000) Palaeointensity of West Greenland Palaeocene
Journal International 131(2): 310–324. basalts: Asymmetric intensity around the C27n-C26r transition. Physics of the Earth
Nowaczyk NR, Antonow M, Knies J, and Spielhagen RF (2003) Further rock magnetic and Planetary Interiors 118(1–2): 53–64.
and chronostratigraphic results on reversal excursions during the last 50 ka as Riisager P and Riisager J (2001) Detecting multidomain magnetic grains in Thellier
derived from northern high latitudes and discrepancies in precise AMS C-14 dating. palaeointensity experiments. Physics of the Earth and Planetary Interiors 125(1–4):
Geophysical Journal International 155(3): 1065–1080. 111–117.
Nowaczyk NR and Frederichs TW (1999) Geomagnetic events and relative Riisager J, Riisager P, and Pedersen A (2003) The C27n-C26r geomagnetic polarity
palaeointensity variations during the past 300 ka as recorded in Kolbeinsey Ridge reversal recorded in the west Greenland flood basalt province: How complex is the
sediments, Iceland Sea: Indication for a strongly variable geomagnetic field. transitional field? Journal of Geophysical Research 108. http://dx.doi.org/10.1029/
International Journal of Earth Sciences 88(1): 116–131. 2002JB002124.
Nowaczyk NR and Knies J (2000) Magnetostratigraphic results from the eastern Arctic Roberts AP (1995) Magnetic properties of sedimentary greigite (Fe3S4). Earth and
Ocean: AMS 14C ages and relative palaeointensity data of the Mono Lake and Planetary Science Letters 134: 227–236.
Laschamp geomagnetic reversal excursions. Geophysical Journal International Roberts AP, Chang L, Heslop D, Florindo F, and Larrasoaña J (2012) Searching for
140(1): 185–197. single domain magnetite in the “pseudo-single-domain” sedimentary haystack:
Oda H, Nakamura K, Ikehara K, Nakano T, Nishimura M, and Khlystov O (2002) Implications of biogenic magnetite preservation for sediment magnetism and
Paleomagnetic record from Academician Ridge, Lake Baikal: A reversal excursion at relative paleointensity determinations. Journal of Geophysical Research
the base of marine oxygen isotope stage 6. Earth and Planetary Science Letters 117, B08104.
202(1): 117–132. Roberts A, Lehman B, Weeks R, Verosub K, and Laj C (1997) Relative paleointensity of
Ohno M, Hayashi T, Komatsu F, et al. (2012) A detailed paleomagnetic record between the geomagnetic field over the last 200,000 years from ODP Sites 883 and 884,
2.1 and 2.75 Ma at IODP Site U1314 in the North Atlantic: Geomagnetic excursions North Pacific Ocean. Earth and Planetary Science Letters 152: 11–23.
and the Gauss-Matuyama transition. Geochemistry, Geophysics, Geosystems Roberts AP and Lewin-Harris JC (2000) Marine magnetic anomalies: Evidence that ‘tiny
13: Q12Z39. http://dx.doi.org/10.1029/2012GC004080. wiggles’ represent short-period geomagnetic polarity intervals. Earth and Planetary
Okada M (1995) Detailed variation of geomagnetic field intensity during the late Science Letters 183(3–4): 375–388.
Pleistocene at Site 882. Proceedings of the Ocean Drilling Program, Scientific Roberts AP, Tauxe L, and Heslop D (2013) Magnetic paleointensity stratigraphy and
Results, vol. 145, pp. 469–474. College Station, TX: Ocean Drilling Program. high resolution Quaternary geochronology: Successes and future challenges.
Ozima M, Ozima M, and Akimoto S (1964) Low temperature characteristics of remanent Quaternary Science Reviews 61: 1–16.
magnetization of magnetite. Journal of Geomagnetism and Geoelectricity Roberts AP, Verosub KL, and Negrini RM (1994) Middle Late Pleistocene relative
16: 165–177. paleointensity of the geomagnetic-field from lacustrine sediments, Lake Chewaucan,
Pan Y, Zhu R, Shaw J, Liu Q, and Guo B (2001) Can relative paleointensities be Western United States. Geophysical Journal International 118(1): 101–110.
determined from the normalized magnetization of the wind-blown loess of China? Roberts AP, Winklhofer M, Liang WT, and Horng CS (2003) Testing the hypothesis of
Journal of Geophysical Research 106: 19221–19232. orbital (eccentricity) influence on Earth’s magnetic field. Earth and Planetary Science
Paterson G, Biggin A, Yamamoto Y, and Pan YX (2012) Towards the robust selection of Letters 216(1–2): 187–192.
Thellier-type paleointensity data: The influence of experimental noise. Rolph TC and Shaw J (1985) A new method of palaeofield magnitude correction for
Geochemistry, Geophysics, Geosystems 13: Q05Z43. http://dx.doi.org/10.1029/ thermally altered samples and its application to Lower Carboniferous lavas.
2012GC004046. Geophysical Journal of the Royal Astronomical Society 80: 773–781.
Paterson G, Heslop D, and Muxworthy A (2010) Deriving confidence in paleointensity Saracco G, Thouveny N, Bourlès D, and Carcaillet JT (2009) Extraction of non-
estimates. Geochemistry, Geophysics, Geosystems 11: Q07Z18. http://dx.doi.org/ continuous orbital frequencies from noisy insolation data and from palaeoproxy
10.1029/2010GC003071. records of geomagnetic intensity using the phase of continuous wavelet transform.
Pavón-Carrasco J, Rodrı́guez-González J, Osete M, and Torta J (2011) A Matlab tool for Geophysical Journal International 176: 767–781.
archaeomagnetic dating. Journal of Archaeological Science 38: 408–419. Sato T, Kikuchi H, Nakashizuka M, and Okada M (1998) Quaternary geomagnetic field
Peck J, King J, Colman S, and Kravchinsky V (1996) An 84-kyr paleomagnetic record intensity: Constant periodicity or variable period? Geophysical Research Letters
from the sediments of Lake Baikal, Siberia. Journal of Geophysical Research 25: 2221–2224.
101: 1365–1385. Sato T and Kobayashi K (1989) Long-period secular variations of the Earth’s magnetic
Perrin M and Schnepp E (2004) IAGA paleointensity database: Distribution and quality field revealed by Pacific deep-sea sediment cores. Journal of Geomagnetism and
of the data set. Physics of the Earth and Planetary Interiors 147(2–3): 255–267. Geoelectricity 41: 147–159.
Perrin M, Schnepp E, and Shcherbakov V (1998) Paleointensity database updated. EOS, Sbarbori E, Tauxe L, Gogichaishvili A, Urrutia-Fucugauchi J, and Bohrson W (2009)
Transactions AGU 79: 198. Paleomagnetic behavior of volcanic rocks from Isla Socorro, Mexico. Earth, Planets
Perrin M and Shcherbakov V (1997) Paleointensity of the Earth’s magnetic field for the and Space 61: 191–204.
past 400 Ma: Evidence for a dipole structure during the Mesozoic low. Journal of Schneider DA (1993) An estimate of Late Pleistocene geomagnetic intensity variation
Geomagnetism and Geoelectricity 49: 601–614. From Sulu Sea sediments. Earth and Planetary Science Letters 120(3–4): 301–310.
Pick T and Tauxe L (1993) Geomagnetic paleointensities during the Cretaceous normal Schneider DA, Kent DV, and Mello GA (1992) A detailed chronology of the Australasian
superchron measured using submarine basaltic glass. Nature 366: 238–242. impact event, the Brunhes-Matuyama geomagnetic polarity reversal and global
Pick T and Tauxe L (1994) Characteristics of magnetite in submarine basaltic glass. climate change. Earth and Planetary Science Letters 111: 395–405.
Geophysical Journal International 119: 116–128. Schneider D and Mello G (1996) A high-resolution marine sedimentary record of
Pisias NG, Martinson DG, Moore TC, et al. (1984) High-resolution stratigraphic geomagnetic intensity during the Brunhes Chron. Earth and Planetary Science
correlation of benthic oxygen isotopic records spanning the last 300,000 years. Letters 144: 297–314.
Marine Geology 56(1–4): 119–136. Schwartz E and Symons D (1969) Geomagnetic intensity between 100 and 2500 million
Prévot M, Derder MEM, McWilliams M, and Thompson J (1990) Intensity of the Earth’s years ago. Physics of the Earth and Planetary Interiors 2: 11–18.
magnetic field: Evidence for a Mesozoic dipole low. Earth and Planetary Science Schwartz M, Lund SP, and Johnson TC (1996) Environmental factors as complicating
Letters 97: 129–139. influences in the recovery of quantitative geomagnetic-field paleointensity estimates
Raisbeck G, Yiou F, and Zhou S (1994) Paleointensity puzzle. Nature 371: 207–208. from sediments. Geophysical Research Letters 23: 2693–2696.
Rampino M (1979) Possible relations between changes in global ice volume, Schwartz M, Lund SP, and Johnson TC (1998) Geomagnetic field intensity from 71 to
geomagnetic excursions, and the eccentricity of the Earth’s orbit. Geology 12 ka as recorded in deep-sea sediments of the Blake Outer Ridge, North Atlantic
7: 584–587. Ocean. Journal of Geophysical Research: Solid Earth 103(B12): 30407–30416.
Rampino MR (1981) Revised age estimates of Brunhes paleomagnetic events: Support Selkin P, Gee JS, and Tauxe L (2007) Nonlinear thermoremanence acquisition and
for a link between geomagnetism and eccentricity. Journal of Geophysical Research implications for paleointensity data. Earth and Planetary Science Letters 256: 81–89.
Letters 8: 1047–1050. Selkin P, Gee J, Tauxe L, Meurer W, and Newell A (2000) The effect of remanence
Rampino MR and Kent DV (1983) Geomagnetic excursions and climate change. Nature anisotropy on paleointensity estimates: A case study from the Archean Stillwater
302: 455. complex. Earth and Planetary Science Letters 182: 403–416.
Selkin P and Tauxe L (2000) Long-term variations in paleointensity. Philosophical Sugiura N, Strangway D, Pearce G, Wu Y, and Taylor L (1979) A new magnetic
Transactions of the Royal Society London 358: 1065–1088. paleointensity value for a ‘young lunar glass’. In: Proceedings of Lunar and
Shaar R, Ben Yosef E, Ron H, Tauxe L, Agnon A, and Kessel R (2011) Geomagnetic field Planetary Science Conference X, pp. 2189–2197.
intensity: How high can it get? How fast can it change? Constraints from Iron-Age Sumita I, Htakeyama T, Yoshihara A, and Hamano Y (2001) Paleomagnetism of late
copper-slag. Earth and Planetary Science Letters 301: 297–306. http://dx.doi.org/ Archean rocks of Hamersley Basin, Western Australia and the paleointensity at early
10.1016/j.epsl.2010.11.013. Proterozoic. Physics of the Earth and Planetary Interiors 128: 223–241.
Shaar R, Ron H, Tauxe L, Kessel R, and Agnon A (2011) Paleomagnetic field Svensmark H and Friis-Christensen E (1997) Variation of cosmic ray flux and global
intensity derived from non-SD: Testing the Thellier IZZI technique on MD cloud coverage – A missing link in solar-climate relationship. Journal of
slag and a new bootstrap procedure. Earth and Planetary Science Letters Atmospheric and Solar – Terrestrial Physics 59: 1225–1232.
310: 213–224. Tanaka H, Athanassopoulos J, Dunn J, and Fuller M (1995) Paleointensity
Shaar R, Ron H, Tauxe L, et al. (2010) Testing the accuracy of absolute intensity determinations with measurements at high temperature. Journal of Geomagnetism
estimates of the ancient geomagnetic field using copper slag material. Earth and and Geoelectricity 47: 103–113.
Planetary Science Letters 290: 201–213. Tanaka H and Kono M (1991) Preliminary results and reliability of palaeointensity
Shaar R and Tauxe L (2013) Thellier_GUI: An integrated tool for analyzing studies on historical and C14 dated Hawaiian lavas. Journal of Geomagnetism and
paleointensity data from Thellier-type experiments. Geochemistry, Geophysics, Geoelectricity 43: 375–388.
Geosystems 14: 677–692. Tanaka H and Kono M (1994) Paleointensity database provides new research. EOS,
Shackleton NJ, Berger A, and Peltier WR (1990) An alternative astronomical calibration Transactions AGU 75: 498.
of the lower Pleistocene timescale based on ODP Site 677. Transactions of the Royal Tanaka HF and Kono M (2002) Paleointensities from a Cretaceous basalt platform in
Society of Edinburgh: Earth Sciences 81: 251–261. Inner Mongolia, Northeastern China. Physics of the Earth and Planetary Interiors
Shaw J (1974) A new method of determining the magnitude of the paleomagnetic field 133(1–4): 147–157.
application to 5 historic lavas and five archeological samples. Geophysical Journal Tanaka H, Kono M, and Uchimura H (1995) Some global features of paleointensity in
of the Royal Astronomical Society 39: 133–141. geological time. Geophysical Journal International 120: 97–102.
Shcherbakov V and Shcherbakova V (1983) On the theory of depositional remanent Tarduno JA, Cottrell RD, and Smirnov AV (2001) High geomagnetic intensity during the
magnetization in sedimentary rocks. Geophysical Surveys 5: 369–380. mid-Cretaceous from Thellier analyses of single plagioclase crystals. Science
Shcherbakova V, Shcherbakov V, and Heider F (2000) Properties of partial 291(5509): 1779–1783.
thermoremanent magnetization in PSD and MD magnetite grains. Journal of Tarduno JA, Cottrell RD, and Smirnov AV (2002) The Cretaceous superchron
Geophysical Research 105: 767–782. geodynamo: Observations near the tangent cylinder. Proceedings of National
Sherwood G, Shaw J, Baer G, and Mallik S (1993) The strength of the geomagnetic field Academy of Sciences of the United States of America 99: 14020–14025.
during the Cretaceous Quiet Zone: Paleointensity results from Israeli and Indian Tarduno J, Cottrell R, and Smirnov A (2006) The paleomagnetism of single silicate
Lavas. Journal of Geomagnetism and Geoelectricity 45: 339–360. crystals: Recording geomagnetic field strength during mixed polarity intervals,
Smirnov AV and Tarduno JA (2005) Thermochemical remanent magnetization in superchrons, and inner core growth. Reviews of Geophysics 44. http://dx.doi.org/
Precambrian rocks: Are we sure the geomagnetic field was weak? Journal of 10.1029/2005RG000189.
Geophysical Research 110. http://dx.doi.org/10.1029/2004JB003445. Tarduno JA, Cottrell RD, Watkeys M, and Bauch D (2007) Geomagnetic field
Smirnov AV, Tarduno JA, and Pisakin BN (2003) Paleointensity of the early geodynamo strength 3.2 billion years ago recorded by single silicate crystals. Nature
(2.45 Ga) as recorded in Karelia: A single-crystal approach. Geology 31(5): 446: 657–660.
415–418. Tarduno JA, Cottrell RD, Watkeys M, et al. (2010) Geodynamo, solar wind, and
Smith PJ (1967) The intensity of the ancient geomagnetic field: A review and analysis. magnetopause 3.4 to 3.45 billion years ago. Science 327: 1238–1240.
Geophysical Journal of the Royal Astronomical Society 12: 321–362. Tauxe L (1993) Sedimentary records of relative paleointensity of the geomagnetic field:
Sowers T, Bender M, Labeyrie L, et al. (1993) A 135,000 year Vostok-Specmap Theory and practice. Reviews of Geophysics 31: 319–354.
common temporal framework. Paleoceanography 8: 737–766. Tauxe L (1998) Paleomagnetic Principles and Practice. Dordrecht: Kluwer Academic
Spassov S, Heller F, Evans ME, Yue LP, and von Dobeneck T (2003) A lock-in model for Publishers.
the complex Matuyama-Brunhes boundary record of the loess/palaeosol sequence Tauxe L (2006) Long term trends in paleointensity: The contribution of DSDP/ODP
at Lingtai (Central Chinese Loess Plateau). Geophysical Journal International submarine basaltic glass collections. Physics of the Earth and Planetary Interiors
155(2): 350–366. 244: 515–529.
Stacey FD (1972) On the role of Brownian motion in the control of detrital remanent Tauxe L, Banerjee SK, Butler R, and van der Voo R (2010) Essentials of
magnetization of sediments. Pure and Applied Geophysics 98: 139–145. Paleomagnetism. Berkeley, CA: University of California Press.
Stern DP (2003) A millennium of geomagnetism. Reviews of Geophysics 41(2): 1007. Tauxe L, Bertram H, and Seberino C (2002) Physical interpretation of hysteresis loops:
Stoner JS, Channell JET, and Hillaire-Marcel C (1995) Late Pleistocene relative Micro-magnetic modelling of fine particle magnetite. Geochemistry, Geophysics,
geomagnetic paleointensity from the deep Labrador Sea: Regional and global Geosystems 3: 1–22. http://dx.doi.org/10.1029/2001GC000280.
correlations. Earth and Planetary Science Letters 134: 237–252. Tauxe L, Gee JS, Steiner M, and Staudigel H (2013) Paleointensity results from
Stoner JS, Channell JET, and Hillaire-Marcel C (1998) A 200 ka geomagnetic the Jurassic: New constraints from submarine basaltic glasses of ODP Site
chronostratigraphy for the Labrador Sea: Indirect correlation of the sediment record 801C. Geochemistry, Geophysics, Geosystems. http://dx.doi.org/10.1002/
to SPECMAP. Earth and Planetary Science Letters 159(3–4): 165–181. ggge.20282.
Stoner J, Channell J, Hillaire-Marcel C, and Kissel C (2000) Geomagnetic Tauxe L and Hartl P (1997) 11 million years of Oligocene geomagnetic field behaviour.
paleointensity and environmental record from Labrador Sea core MD95–2024: Geophysical Journal International 128: 217–229.
Global marine sediment and ice core chronostratigraphy for the last 110 kyr. Earth Tauxe L, Herbert T, Shackleton NJ, and Kok YS (1996) Astronomical calibration of the
and Planetary Science Letters 5626: 1–17. Matuyama Brunhes Boundary: Consequences for magnetic remanence acquisition
Stoner JS, Channell JET, Hodell DA, and Charles CD (2003) A 580 kyr paleomagnetic in marine carbonates and the Asian loess sequences. Earth and Planetary Science
record from the sub-Antarctic South Atlantic (Ocean Drilling Program Site 1089). Letters 140: 133–146.
Journal of Geophysical Research 108(B5). Tauxe L, Pick T, and Kok YS (1995) Relative paleointensity in sediments: A pseudo-
Stoner JS, Laj C, Channell JET, and Kissel C (2002) South Atlantic and North Atlantic Thellier approach. Geophysical Research Letters 22: 2885–2888.
geomagnetic paleointensity stacks (0–80 ka): Implications for inter-hemispheric Tauxe L and Shackleton NJ (1994) Relative paleointensity records from the Ontong-Java
correlation. Quaternary Science Reviews 21(10): 1141–1151. Plateau. Geophysical Journal International 117: 769–782.
Stott L, Poulsen C, Lund S, and Thunell R (2002) Super ENSO and global climate Tauxe L and Staudigel H (2004) Strength of the geomagnetic field in the Cretaceous
oscillations at millennial time scales. Science 297: 222–226. Normal Superchron: New data from submarine basaltic glass of the Troodos
Suganuma Y, Okuno J, Heslop D, Roberts AP, Yamazaki T, and Yokoyama Y (2011) Ophiolite. Geochemistry, Geophysics, Geosystems 5(2): Q02H06. http://dx.doi.org/
Post-depositional remanent magnetization lock-in for marine sediments deduced 10.1029/2003GC000635.
from 10Be and paleomagnetic records through the Matuyama-Brunhes boundary. Tauxe L, Steindorf J, and Harris A (2006) Depositional remanent magnetization: Toward
Earth and Planetary Science Letters 311: 39–52. an improved theoretical and experimental foundation. Earth and Planetary Science
Suganuma Y, Yokoyama Y, Yamazaki T, Kawamura K, Horng C-S, and Matsuzaki H Letters 244: 515–529.
(2010) 10Be evidence for delayed acquisition of remanent magnetization in marine Tauxe L and Wu G (1990) Normalized remanence in sediments of the Western
sediments: Implication for a new age for the Matuyama-Brunhes boundary. Earth Equatorial Pacific: Relative paleointensity of the geomagnetic field? Journal of
and Planetary Science Letters 296: 443–450. Geophysical Research 95(B8): 12337–12350.
Tauxe L and Yamazaki T (2007) Paleointensities. Treatise on Geophysics, vol. 5, Verosub KL, Herrerobervera E, and Roberts AP (1996) Relative geomagnetic
pp. 509–563. New York: Elsevier. http://dx.doi.org/10.1016/B978-044452748-6/ paleointensity across the Jaramillo Subchron and the Matuyama/Brunhes boundary.
00098-5. Geophysical Research Letters 23(5): 467–470.
Taylor L (1979) An effective sample preparation technique for paleointensity Walton D (2004) Avoiding mineral alteration during microwave magnetization.
determinations at elevated temperatures. In: Proceedings of Lunar and Planetary Geophysical Research Letters 31: L03606. http://dx.doi.org/10.1029/
Science Conference X, pp. 1209–1211. 2003GL019011.
Teanby N and Gubbins D (2000) The effects of aliasing and lock-in processes on Walton D (2005) Isolating viscous overprints with microwaves. EOS Transactions, AGU
palaeosecular variation records from sediments. Geophysical Journal International Fall Meeting Supplement 86, GP12A-05.
142(2): 563–570. Walton D, Share J, Rolph TC, and Shaw J (1993) Microwave magnetisation.
Thellier E (1938) Sur l’aimantation des terres cuites et ses applications géophysique. Geophysical Research Letters 20: 109–111.
Annales de l’Institut de Physique du Globe de Paris 16: 157–302. Wang H and Kent D (2013) A paleointensity technique for multidomain igneous rocks.
Thellier E and Thellier O (1959) Sur l’intensité du champ magnétique terrestre dans le Geochemistry, Geophysics, Geosystems 14. http://dx.doi.org/10.1002/
passé historique et géologique. Annales de Geophysique 15: 285–378. ggge.20248.
Thouveny N, Bourlès D, Saracco G, Carcaillet JT, and Basssinot F (2008) Paleoclimatic Wang R and Lovlie R (2010) Subaerial and subaqueous deposition of loess:
context of geomagnetic dipole lows and excursions in the Brunhes, clue for an Experimental assessment of detrital remanent magnetization in Chinese loess. Earth
orbital influence on the geodynamo? Earth and Planetary Science Letters and Planetary Science Letters 298: 394–404.
275: 269–284. Weeks RJ, Laj C, Endignoux L, et al. (1995) Normalised natural remanent magnetisation
Thouveny N, Carcaillet J, Moreno E, Leduc G, and Nerini D (2004) Geomagnetic intensity during the last 240000 years in piston cores from the central North Atlantic
moment variation and paleomagnetic excursions since 400 kyr BP: A stacked record Ocean: Geomagnetic field intensity or environmental signal? Physics of the Earth
from sedimentary sequences of the Portuguese margin, Pt 1. Earth and Planetary and Planetary Interiors 87: 213–229.
Science Letters 219(3–4): 377–396. Westphal M and Munschy M (1994) Saw-tooth pattern of the Earth’s magnetic
Thouveny N, Debeaulieu JL, Bonifay E, et al. (1994) Climate variations in Europe over field tested by magnetic anomalies over oceanic spreading ridges. Comptes
the past 140-kyr deduced from rock magnetism. Nature 371(6497): 503–506. Rendus de l’Academie des Sciences Series IIA Earth and Planetary Science
Tiedemann R, Sarnthein M, and Shackleton NJ (1994) Astronomical timescale for the 329: 565–571.
Pliocene Atlantic delta18O and dust flux records of Ocean Drilling Program site 659. Williams T, Thouveny N, and Creer KM (1998) A normalised intensity record from Lac
Paleoceanography 9: 619–638. du Bouchet: Geomagnetic palaeointensity for the last 300 kyr? Earth and Planetary
Tric E, Valet JP, Tucholka P, et al. (1992) Paleointensity of the geomagnetic field during Science Letters 156(1–2): 33–46.
the last 80,000 years. Journal of Geophysical Research 97: 9337–9351. Wilson R (1961) Paleomagnetism in Northern Ireland, 1. The thermoremanent
Tsunakawa H and Shaw J (1994) The Shaw method of paleointensity determinations magnetization of natural magnetic moments in rocks. Geophysical Journal of the
and its application to recent volcanic rocks. Geophysical Journal International Royal Astronomical Society 5: 45–58.
118: 781–787. Winterwerp J and van Kesteren W (2004) Introduction to the Physics of Cohesive
Valet J-P (2003) Time variations in geomagnetic intensity. Reviews of Geophysics Sediment in the Marine Environment. Developments in Sedimentology, vol. 56.
41. http://dx.doi.org/10.1029/2001RG000104. Amsterdam: Elsevier.
Valet JP, Brassart J, Lemeur I, et al. (1996) Absolute paleointensity and Wollin G, Ericson D, and Ryan WBF (1971) Variations in magnetic intensity and climatic
magnetomineralogical changes. Journal of Geophysical Research 101(B11): changes. Nature 232: 549–550.
25029–25044. Worm H (1997) A link between geomagnetic reversals and events and glaciations. Earth
Valet JP and Herrero-Bervera E (2000) Paleointensity experiments using alternating and Planetary Science Letters 147: 55–67.
field demagnetization. Earth and Planetary Science Letters 177: 43–58. Xu S and Dunlop DJ (1994) Theory of partial thermoremanent magnetization in
Valet JP and Meynadier L (1993) Geomagnetic field intensity and reversals during the multidomain grains: 2. Effect of microcoercivity distribution and comparison with
past four million years. Nature 366: 234–238. experiment. Journal of Geophysical Research 99: 9025–9033.
Valet JP and Meynadier L (2001) Comment on “A relative geomagnetic paleointensity Xuan C and Channell JET (2008a) Origin of orbital periods in the sedimentary relative
stack from Ontong-Java plateau sediments for the Matuyama” by Yvo S. Kok and paleointensity records. Physics of the Earth and Planetary Interiors 169: 140–151.
Lisa Tauxe. Journal of Geophysical Research 106: 11013–11015. Xuan C and Channell JET (2008b) Testing the relationship between timing of
Valet JP, Meynadier L, Bassinot FC, and Garnier F (1994) Relative paleointensity across geomagnetic reversals/excursions and phase of orbital cycles using circular
the last geomagnetic reversal from sediments of the Atlantic, Indian and Pacific statistics and Monte Carlo simulations. Earth and Planetary Science Letters
Oceans. Geophysical Research Letters 21: 485–488. 268: 245–254.
Valet JP, Meynadier L, and Guyodo Y (2005) Geomagnetic dipole strength and reversal Yamamoto Y and Shaw J (2008) Development of the microwave LTD-DHT Shaw method
rate over the past two million years. Nature 435: 802–805. for absolute paleointensity determination. Physics of the Earth and Planetary
Valet JP, Moreno E, Basssinot F, et al. (2011) Isolating climatic and paleomagnetic Interiors 170: 15–23.
imbricated signals in two marine cores using principal component analysis. Yamamoto Y and Tsunakawa H (2005) Geomagnetic field intensity during the last
Geochemistry, Geophysics, Geosystems 12: Q08012. http://dx.doi.org/10.1029/ 5 Myr: LTD-DHT Shaw palaeointensities from volcanic rocks of the Society Islands,
2011GC003697. French Polynesia. Geophysical Journal International 162(1): 79–114.
Valet JP, Tauxe L, and Clement BM (1989) Equatorial and mid-latitude records of the Yamamoto Y, Tsunakawa H, and Shibuya H (2003) Palaeointensity study of the
last geomagnetic reversal from the Atlantic Ocean. Earth and Planetary Science Hawaiian 1960 lava: Implications for possible causes of erroneously high
Letters 94: 371–384. intensities. Geophysical Journal International 153(1): 263–276.
Valet JP, Tric E, Herrero-Bervera E, Meynadier L, and Lockwood JP (1998) Absolute Yamazaki T (1999) Relative paleointensity of the geomagnetic field during Brunhes
paleointensity from Hawaiian lavas younger than 35 ka. Earth and Planetary Science Chron recorded in North Pacific deep-sea sediment cores: Orbital influence? Earth
Letters 161: 19–32. and Planetary Science Letters 169(1–2): 23–35.
van Vreumingen M (1993a) The magnetization intensity of some artificial suspensions Yamazaki T and Ikehara M (2012) Origin of magnetic mineral concentration variation in
while flocculating in a magnetic field. Geophysical Journal International the Southern Ocean. Paleoceanography 27: PA2206. http://dx.doi.org/10.1029/
114: 601–606. 2011PA002271.
van Vreumingen M (1993b) The influence of salinity and flocculation upon the Yamazaki T and Ioka N (1994) Long-term secular variation of the geomagnetic field
acquisition of remanent magnetization in some artificial sediments. Geophysical during the last 200 kyr recorded in sediment cores from the western equatorial
Journal International 114: 607–614. Pacific. Earth and Planetary Science Letters 128: 527–544.
van Zijl JSU, Graham KWT, and Hales AL (1962a) The paleomagnetism of the Yamazaki T, Ioka N, and Eguchi N (1995) Relative paleointensity of the
Stormberg Lavas 1. Geophysical Journal of the Royal Astronomical Society geomagnetic field during the Brunhes Chron. Earth and Planetary Science Letters
7: 23–39. 136: 525–540.
van Zijl JSU, Graham KWT, and Hales AL (1962b) The paleomagnetism of the Yamazaki T and Kanamatsu T (2007) A relative paleointensity record of the
Stormberg Lavas, 2. The behavior of the Earth’s magnetic field during a reversal. geomagnetic field since 1.6 Ma from the North Pacific. Earth, Planets and Space
Geophysical Journal of the Royal Astronomical Society 7: 169–182. 59: 785–794.
Verosub KL (1977) Depositional and postdepositional processes in the magnetization of Yamazaki T and Oda H (2002) Orbital influence on Earth’s magnetic field: 100,000-year
sediments. Reviews of Geophysics and Space Physics 15: 129–143. periodicity in inclination. Science 295(5564): 2435–2438.
Yamazaki T and Oda H (2005) A geomagnetic paleointensity stack between 0.8 and Yu Y and Tauxe L (2005) On the use of magnetic transient hysteresis in paleomagnetism
3.0 Ma from equatorial Pacific sediment cores. Geochemistry, Geophysics, for granulometry. Geochemistry, Geophysics, Geosystems 6: Q01H14. http://dx.doi.
Geosystems 6. http://dx.doi.org/10.203-/2005GC001001. org/10.1029/2004GC000839.
Yamazaki T, Yamamoto Y, Acton G, Guidry E, and Richter C (2013) Rock-magnetic Yu Y and Tauxe L (2006) Acquisition of viscous remanent magnetization. Physics of the
artifacts on long-term relative paleointensity variations in sediments. Geochemistry, Earth and Planetary Interiors 159: 32–42.
Geophysics, Geosystems 14: 29–43. Yu Y, Tauxe L, and Genevey A (2004) Toward an optimal geomagnetic field intensity
Yokoyama Y and Yamazaki T (2000) Geomagnetic paleointensity variation with a 100 determination technique. Geochemistry, Geophysics, Geosystems 5(2): Q02H07.
kyr quasi-period. Earth and Planetary Science Letters 181(1–2): 7–14. http://dx.doi.org/10.1029/2003GC000630.
Yokoyama Y, Yamazaki T, and Oda H (2007) Geomagnetic 100-kyr variation extracted Zhao X and Roberts AP (2010) How does Chinese loess become magnetized? Earth and
from paleointensity records of the equatorial and North Pacific sediments. Earth, Planetary Science Letters 292: 112–122.
Planets and Space 59: 795–805. Zhu R, Laj C, and Mazaud A (1994) The Matuyama-Brunhes and upper Jaramillo
Yokoyama Y, Yamazaki T, and Oda H (2010) Geomagnetic 100-kyr variation excited by a transitions recorded in a loess section at Weinan, north-central China. Earth and
change in the Earths orbital eccentricity. Geophysical Research Letters 37: L11302. Planetary Science Letters 125: 143–158.
http://dx.doi.org/10.1029/2010GL042898. Zhu RX, Lo CH, Shi RP, Shi GH, Pan YX, and Shao J (2004) Palaeointensities
Yoshida S and Katsura I (1985) Characterization of fine magnetic grains in sediments by determined from the middle Cretaceous basalt in Liaoning Province,
the suspension method. Geophysical Journal of the Royal Astronomical Society northeastern China. Physics of the Earth and Planetary Interiors 142(1–2):
82: 301–317. 49–59.
Yoshihara A and Hamano Y (2000) Intensity of the Earth’s magnetic field in late Archean Zhu RX, Pan YX, Shaw J, Li DM, and Li Q (2001) Geomagnetic palaeointensity just prior
obtained from diabase dikes of the Slave Province, Canada. Physics of the Earth and to the Cretaceous normal superchron. Physics of the Earth and Planetary Interiors
Planetary Interiors 117: 295–307. 128(1–4): 207–222.
Yu Y and Tauxe L (2004) On the use of magnetic transient hysteresis in paleomagnetism Ziegler LB, Constable C, Johnson CL, and Tauxe L (2011) PADM2M: A penalized
for Granulometry. Geochemistry, Geophysics, Geosystems 6: Q01H14. http://dx.doi. maximum likelihood model of the 0–2 Ma paleomagnetic axial dipole moment.
org/10.1029/2004GC000839. Geophysical Journal International 184: 1069–1089.
5.14 True Polar Wander: Linking Deep and Shallow Geodynamics
to Hydro- and Biospheric Hypotheses
TD Raub, Yale University, New Haven, CT, USA
JL Kirschvink, California Institute of Technology, Pasadena, CA, USA
DAD Evans, Yale University, New Haven, CT, USA
5.14.1 Planetary Moment of Inertia and the Spin Axis 511

5.14.2 Apparent Polar Wander (APW) ¼ Plate Motion + TPW 511
5.14.2.1 Different Information in Different Reference Frames 511
5.14.2.2 ‘Type 0’ TPW: Mass Redistribution at Clock to Millennial Timescales of Inconsistent Sense 512
5.14.2.3 Type I TPW: Slow/Prolonged TPW 513
5.14.2.4 Type II TPW: Fast/Multiple/Oscillatory TPW, a Distinct Flavor of Inertial Interchange 513
5.14.2.5 Hypothesized Rapid or Prolonged TPW: Late Paleozoic–Mesozoic 514
5.14.2.6 Hypothesized Rapid or Prolonged TPW: ‘Cryogenian’–Ediacaran–Cambrian–Early Paleozoic 515
5.14.2.7 Hypothesized Rapid or Prolonged TPW: Archean to Mesoproterozoic 516
5.14.3 Geodynamic and Geologic Effects and Inferences 516
5.14.3.1 Precision of TPW Magnitude and Rate Estimation 516
5.14.3.2 Physical Oceanographic Effects: Sea Level and Circulation 517
5.14.3.3 Chemical Oceanographic Effects: Carbon Oxidation and Burial 519
5.14.4 Critical Testing of Cryogenian–Cambrian TPW 522
5.14.4.1 Ediacaran–Cambrian TPW: ‘Spinner Diagrams’ in the TPW Reference Frame 522
5.14.4.2 Proof of Concept: Independent Reconstruction of Gondwanaland Using Spinner Diagrams 524
5.14.5 Summary: Major Unresolved Issues and Future Work 526
Acknowledgments 527
References 527
5.14.1 Planetary Moment of Inertia and the Spin Axis Goldreich and Toomre (1969) demonstrated that this
hydrostatic bulge only exerts a stabilizing influence on the
Planets, as quasi-rigid, self-gravitating bodies in free space, orientation of the planetary spin vector on timescales <105
must spin about the axis of their principal moment of inertia years, with little or no influence on the bulk solid Earth over
(Gold, 1955). Net angular momentum (the L vector), a con- longer timescales (see also Richards et al., 1997; Steinberger
served quantity in the absence of external torques, is related to and O’Connell, 1997, 2002). For the purposes of this discus-
the spin vector (v) by the moment of inertia tensor (I) such sion, only the nonhydrostatic component of the Earth’s inertial
that L ¼ Iv. It is well known that the movement of masses tensor is geologically important, and inertial perturbations of
within or on the solid Earth, such as sinking slabs of oceanic sufficient scale should induce TPW in a sense that moves a new
lithosphere, rising or erupting plume heads, or growing or principal inertial axis (Imax) back toward the average rotation
waning ice sheets, can alter components of that inertial tensor vector, v.
and induce compensating changes in the v vector so that L is
conserved. Observable mass redistributions caused by postgla-
cial isostatic readjustment, by earthquakes, and even by 5.14.2 Apparent Polar Wander (APW) ¼ Plate Motion
weather systems cause detectible v changes, manifested as + TPW
shifts in the geographic location of the Earth’s daily rotation
axis and/or by fluctuations in the spin rate (‘length-of-day’ 5.14.2.1 Different Information in Different Reference
anomalies). Generally, such shift of the geographic rotation Frames
pole is called true polar wander (TPW). For mid-Mesozoic and younger time, seafloor magnetic linea-
Because its mantle and lithosphere are viscoelastic, the tions permit accurate paleogeographic reconstructions of con-
Earth’s daily spin distorts its shape slightly from that of a tinents separated by spreading ridges. For older times,
perfect sphere, producing an equatorial bulge of about 1 part paleogeographers must rely upon paleomagnetic data refer-
in 300 or an excess equatorial radius of 20 km relative to its enced to an assumed geocentric, spin-axial dipole magnetic
polar radius. Variations in the Earth’s spin vector (v) will cause field. There is ample evidence for dominance of this axial
this hydrostatic bulge to shift in response, with a characteristic dipole field configuration for most of the past 2 Gy (Evans
timescale of 105 years, governed by the relaxation time of the (2006) and references therein). While many authors have
mantle to long-wavelength loads (Ranalli, 1995; Steinberger used the unfiltered global paleomagnetic database to estimate
and O’connell, 1997). the maximum permissible contribution of nondipole terms in

512 True Polar Wander: Linking Deep and Shallow Geodynamics to Hydro- and Biospheric Hypotheses
the geomagnetic spherical harmonic expansion (e.g., Evans, Courtillot (2002)). Short intervals of nonnegligible TPW have
1976; Kent and Smethurst, 1998), the likelihood that conti- been hypothesized for various intervals of the Phanerozoic
nents are not distributed uniformly on the globe over time and are permitted by the sometimes-coarse resolution of the
(e.g., Evans, 2005) introduces a ready alternative explanation global paleomagnetic database (Besse and Courtillot (2002,
for the database-wide trends. 2003); and see subsequent sections). Irving (1988, 2005) fur-
Mesozoic–Cenozoic paleogeographic reconstructions usu- ther recounted the historical adoption of paleomagnetism to
ally rely on the collection of a time series of well-dated paleo- test hypotheses of continental drift.
magnetic poles (called apparent polar wander paths (APWPs)) In the wake of modern hypotheses suggesting that ancient
from distinct continental blocks. When combined with other TPW rates may sometimes match or exceed long-term plate
geologic or paleontological constraints, it is possible to match velocities (see subsequent sections), Evans (2003) enunciated
and rotate similarly shaped portions of those paths into over- the combinatorial possibilities by which TPW may either
lapping alignment (Irving, 1956; Runcorn, 1956), thereby pro- enhance or mask the plate-tectonic component of APW
viding, to first order, the absolute paleolatitude and relative (Figure 1). Most powerfully, TPW must be recognized in the
paleolongitude for constituent plate-tectonic blocks of ancient paleomagnetic record as an APW component common to all
supercontinents such as Pangaea. For times prior to the oldest plates in the celestial or geomagnetic reference frames. Plate-
marine magnetic lineations, matching exceptionally quick tectonic motion is expected to vary considerably and even
APWP segments that might correspond to rapid, sustained possibly sum to zero (no net rotation) (Gordon, 1987; Jurdy
TPW provides an alternative paleomagnetic method for con- and Van der Voo, 1974, 1975). Lack of kinematic information
straining paleolongitude, relative to the arbitrary meridian of a from ancient areas of oceanic lithosphere that have been lost
presumed-equatorial TPW axis (Kirschvink et al. (1997) and via subduction precludes rigorous estimation of this no-net-
see subsequent sections). rotation reference frame and thus hinders ancient TPW esti-
As discussed subsequently, accumulating evidence suggests mates using that method.
that Neoproterozoic and early Paleozoic time experienced
much larger and more rapid TPW episodes than apparent
during the Mesozoic and Cenozoic Eras, suggesting fundamen- 5.14.2.2 ‘Type 0’ TPW: Mass Redistribution at Clock
tal (and intellectually exciting) temporal changes in basic geo-
to Millennial Timescales of Inconsistent Sense
physical parameters that control the Earth’s moment of inertia Although not the principal subject of this review, measurable
tensor. amounts of small-magnitude TPW have been observed to fol-
Following theoretical work on ‘polar wandering’ in the late low major earthquakes that redistribute surficial mass nearly
nineteenth century, Wegener (1929) (English translation by instantaneously (e.g., Soldati et al., 2001). The conservation of
Biram (1966, pp. 158–163)) recognized that continents momentum response of this variety of TPW is often associated
might appear to drift not only at the behest of spatially varying with (and more popularly reported as) changes in the length of
internal forces but also by steady or punctuated whole-scale
shifting of the solid Earth’s reference frame with respect to the
ecliptic plane (i.e., the ‘celestial’ or ‘rotational’ reference TPW Continent I Continent II Continent III
frame): t1 t0
t1
. . .the geological driving force acts as before and shifts the principal t0
axis of inertia by the amount x in the same direction, and the process t0 t1
(a)
repeats itself indefinitely. Instead of a single displacement by the
amount x, we now have a progressive displacement, whose rate is set
by the size of the initial displacement x on the one hand, and by the t1 t0 t1
viscosity of the earth on the other; it does not come to rest until the t0 t1
t1 t0
geological driving force has lost its effect. For example, if this geo- t0
logical cause arose from the addition of a mass m somewhere in the (b)
middle latitudes, the axial shift can only cease when this mass
increment has reached the equator. . .. (Wegener’s, p. 158) t1 t0 t1
t0 t1 t0 t1 t0
Since the Earth’s geomagnetic field derives directly from its (c)
spin influence on convection cells in its liquid outer core,
which are sustained by growth of the solid inner core and by Figure 1 Cartoon showing contribution of TPW to APWP. For zero
plate tectonics-driven secular mantle cooling (Nimmo, 2002; (a), slow (b), or fast (c) TPW over time increment t0–t–i
Stevenson, 1983), TPW causes the Earth’s mantle and crust to (thin black vectors in the top, middle, and bottom rows, respectively),
slip on the solid/liquid interface at the core–mantle boundary, plate-tectonic motion (light gray vectors) may dominate – or be obscured
within – the net APW signal. The fidelity by which APW represents
while the Earth’s magnetic field most likely remains geocentric
plate motion will depend not only on the relative rates of TPW and plate
and average spin-axial.
motion but also on the relative directions in an assumed independent
Consequently, TPW was recognized as a possibly confound- (e.g., geomagnetic) reference frame. The observed coherence of
ing signal by early paleomagnetists (Irving, 1957; Runcorn, APW between continents with dissimilar plate motion vectors divulges
1955, 1956), although ultimately, TPW was assumed to be an increased TPW rate. Reprinted from Figure 2 in Evans DAD (2003)
negligible relative to rates of plate motion (Irving (1957), True polar wander and supercontinents. Tectonophysics 362(1–4):
DuBois (1957), and most comprehensively Besse and 303–320, with permission from Elsevier.
True Polar Wander: Linking Deep and Shallow Geodynamics to Hydro- and Biospheric Hypotheses 513
day, although specific mass redistributions may change the discontinuity to counteract the negative effect of its inherent
Earth’s spin rate though not move its inertial axes. On a neo- density deficiency. When rising through the upper mantle,
tectonic timescale, earthquake-induced TPW may have no net which is 10–30 times less viscous than the lower mantle,
effect on APW, since fault orientation is controlled by subplate- however, immediate dynamic topographic effects should be
scale stress regimes, which vary globally (although the long- diminished. In that case, the density deficiency of the rising
term effect of earthquake-induced TPW might also be nonzero) mantle plume, plus its greater radial distance, should effectuate
(see, e.g., Spada, 1997). a decrease in the Earth’s inertial moment along the radial axis
Very short timescale TPW is also driven by momentum of plume ascent, whereas only 10 My earlier, the same man-
transfer within and between the circulating ocean and atmo- tle plume could have produced a positive inertial anomaly
sphere systems and the Earth’s solid surface (e.g., Celaya et al., along the same axis.
1999; Fujita et al., 2002; Seitz and Schmidt, 2005). Such vari- Moreover, entrainment of surrounding material at the front
eties of subannual, itinerant TPW are superimposed upon the of and beside moving mantle density anomalies may also
larger-magnitude annual and Chandler wobbles (e.g., Stacey, either exaggerate or mitigate the effects of those parcels (e.g.,
1992) and 12- and c.14-month decametric oscillations of v Sleep, 1988; Zhong and Hager, 2003), depending on anomaly
(e.g., data from the US Naval Observatory, International Earth speed and surrounding mantle viscosity, as well as radial
Rotation and Reference Systems Service). position.
On a somewhat longer timescale, glacial ice loading and Finally, the Earth’s inertial tensor is also sensitive to the
postglacial isostatic rebound produce lithospheric mass redis- volumes of density anomalies. Considering both the size and
tribution that drives TPW at the same rate (centimeters per location of moving mass components, we might expect mantle
year) as continental plate motions (Mitrovica et al., 2001a,b, superswells to exhibit great, consistently positive effects on the
2005; Nakada, 2002; Nakiboglu and Lambeck, 1980; Sabadini, Earth’s net inertial tensor, whereas lesser and variable magni-
2002; Sabadini and Peltier, 1981; Sabadini et al., 2002; tude effects could be produced by individual mantle plumes or
Vermeersen and Sabadini, 1999). However, the 10 ky time- subducting slabs (Figure 2).
scale for dampening of isostatic dynamics and the cyclic nature Evans (2003) related these putative geodynamic drivers to a
of Quaternary ice sheets probably also relegate this TPW to TPW–supercontinent cycle. In the first stage of the cycle, individ-
insignificance when examining APW paths over million-year ual cratonic components of a future supercontinent amalgamate
timescales. at any particular latitude on Earth. During amalgamation, plate-
Following Evans (2003) and elaborated subsequently, we tectonic velocities of already assembled fragments will tend to
term all TPW of fleeting duration relative to plate-tectonic slow. Consequently, mantle beneath the centroid of a supercon-
timescales or hydrostatic geoid relaxation as ‘type 0.’ For tinent will tend to become buoyant, as a result of thermal insu-
more detailed treatment of the considerable literature examin- lation by the supercontinent and its circumferential, subducting
ing type 0 TPW, we point the reader to Spada et al. (2006). slabs and of upward return flow induced by those sinking slabs
(this argument adapts Anderson, 1982).
Elevated dynamic topography from the resulting superswell
5.14.2.3 Type I TPW: Slow/Prolonged TPW and the overlying supercontinent that created it will alter the
For TPW at timescales and magnitudes relevant to plate tecton- Earth’s inertial tensor so that slow, continuous TPW places the
ics, the effects of surficial and internal mass parcels that dom- supercontinent and superswell on the Earth’s equator (Anderson,
inate changes to the Earth’s net moment of inertia tensor will 1994; Richards and Engebretson, 1992). Evans (2003) termed this
vary in part with the square of radial distance. Other contrib- stage of the hypothesized TPW–supercontinent cycle ‘type I’ TPW.
uting factors to the relative effects of such anomalies are some- The Tharsis volcanic region on Mars is plausibly an example
what more complicated. of a type I TPW-shifted surface mass (Phillips et al., 2001). In
For instance, Hager et al. (1985) noted that while the contrast, the south polar water geyser on Saturn’s moon Ence-
Earth’s nonhydrostatic geoid is dominated by the signature of ladus is an apparent example of the eruption of a hot, buoyant
subducting slabs residing in the upper mantle, the residual plume associated with migration from midlatitudes toward a
geoid (nonhydrostatic geoid minus the modeled contribution spin axis (Nimmo and Pappalardo, 2006). Enceladus’ mantle
of those slabs) correlates with presumed lower-mantle thermal is presumably water ice while its core is (mostly nonmolten)
anomalies. Positive dynamic topography at viscosity disconti- silicate and/or iron (Porco et al., 2006). We therefore suggest
nuities (principally the core–mantle boundary, the crust– that the aforesummarized logic of Hager et al. (1985) restricts
mantle boundary, and possibly the 660 km discontinuity) partial silicate melting to Enceladus’ uppermost core; or else
could balance lower-mantle density deficiencies such as rising core diapirism would surely have overwhelmed inertial effects
plumes, while negative dynamic topography at the same dis- of mantle (ice) upwelling, forcing equatorward rather than
continuities could counteract upper-mantle density excesses poleward TPW.
due to slabs (Hager et al., 1985).
A buoyant, upper-mantle plume head illustrates the non-
5.14.2.4 Type II TPW: Fast/Multiple/Oscillatory TPW,
trivial compensatory effects of a density anomaly interacting
a Distinct Flavor of Inertial Interchange
with surrounding mantle of variable viscosity and structure, in
different radial positions. While in the lower mantle, this Once an equatorial-migrated supercontinent begins to frag-
plume head might have created sufficient positive dynamic ment, its superswell might remain essentially fixed in an
topography at the core–mantle boundary and 660 km equator-centered position in the angular momentum reference
Solar Arc w w
t
radiation Circ ic Obliquity
le = 23°
Tro
p
Can ic of
cer
Equ
ato
r
Tro
Cap pic of
rico
Ecliptic rn Inner
plane core
Outer core
(geodynamo)
Ant
arc
Circ tic
le
Mantle TPW
Figure 2 Cartoon of possible mantle phenomena driving TPW. In addition to surface loading, mass redistribution in the mantle also drives changes in
the Earth’s moment of inertia. Entrainment of flowing material along the edges of rising (light) and sinking (dark) mass anomalies will modify their
effective contributions. Dynamic topography at viscosity discontinuities leading and trailing the moving anomalies would similarly affect the net inertial
change. A dynamic planet (equatorial bulge exaggerated) spins stably and conserves momentum by shifting positive inertial anomalies toward the
equator and negative inertial anomalies toward the poles via TPW. Because TPW affects only the solid Earth, whereas the Earth’s geomagnetic field
arises from the vorticity of convection in its liquid outer core, continents will rotate through the geomagnetic and celestial reference frames, while those
reference frames remain, on average, fixed with respect to each other. Adapted from Figure 1 in Evans DAD (2003) True polar wander and
supercontinents. Tectonophysics 362(1–4): 303–320.
frame (disregarding orbital oscillations), while the constituent (Imin) axis over an interval while Imax Iint. The magnitude of
continental fragments disperse away from the superswell. any single event is unpredictable, as it would depend upon
The centroid of that relict superswell would likely define the the magnitude and duration of the transient changes in the
Earth’s minimum moment of inertial axis. If the Earth’s max- moment of inertia tensor, which in turn are a function of the
imum and intermediate inertial axes were of nearly equal speed and strength at which the driving anomalies are imposed
moment – dominated by the first harmonic geoid minimum and compensated.
to the superswell and by the uniformly dispersing fragment
plates – then incremental changes in one or the other of those
5.14.2.5 Hypothesized Rapid or Prolonged TPW: Late
approximately equivalent moments might reverse their relative
Paleozoic–Mesozoic
magnitudes, producing an ‘inertial interchange’ event (Evans,
2003; Evans et al., 1998; Kirschvink et al., 1997; see also Marcano et al. (1999) noted that the Pangaean APW path
Matsuyama et al., 2006). tracks 35 of equatorward arc from 295 to 205 Ma, indi-
For instance, individual mantle plume ascents and/or erup- cating a corresponding poleward shift of the lithosphere via
tions, orogenic root delaminations, initiation of new subduc- plate tectonics and/or TPW. We note that 25 of arc occurs
tion on the trailing edge of one dispersing supercontinent during the latter half of this interval, at an inferred rate
fragment but not another, or even changes in the relative of 0.45 My1. Because the leading edge of Pangaea in such
sizes and speeds of those fragments might enhance the inter- a plate kinematic reconstruction should have overrun a con-
mediate inertial moment and cause it to transiently equal or siderable sector of Panthalassic Ocean lithosphere, those
slightly exceed that of the Earth’s previously maximum (aver- authors suggest the subduction-depauperate geologic record
age daily spin) axis. of Pangaea’s Cordilleran–Arctic margin favors interpretation
In such a case, the Earth’s new, equatorial maximum of the long APW track as relatively fast, sustained TPW. Irving
moment of inertial axis would experience poleward drift in and Irving (1982) made a similar suggestion of anomalous
the celestial reference frame, until the Earth’s inertial tensor TPW near the Permian–Triassic boundary interval.
satisfied simple spin mechanics once more. We suggest that the apparently poleward motion of Pang-
If the mass anomaly of the supercontinental superswell aea over this interval might be due to TPW motion caused by
were large enough to pin Imin stably on the equator, ‘normal’ upper-mantle ascent of the Siberian Traps plume. For a plume-
plate-tectonic and mantle-dynamic events would continue to consistent model incorporating marine accommodation of
enhance one of the other two (nonminimum) inertial much early Siberian Traps stratigraphy, see Elkins-Tanton and
moments. Repeated, frequent, and sudden switches of the Hager (2005). Such a plume might have originated in the
maximum and intermediate inertial moments could produce lower mantle yet avoided triggering equatorward TPW as dis-
multiple episodes of TPW adjustment. Thus, the aforemen- cussed previously if Imax had significantly exceeded Iint during
tioned ‘type I’ TPW might prepare the planet for multiple lower-mantle plume ascent, though the two inertial moments
‘type II’ events that could occur much more frequently than converged prior to 251 Ma.
would be predicted by considering only randomly moving Perhaps the span of this hypothesized Permian–Triassic
internal masses (Anderson, 1990, p. 252; Fisher, 1974). Pangaea-poleward TPW (55 My) corresponds to the minimum
In summary, ‘type II’ TPW might produce back-and-forth timescale for the buildup of thermal insulation, reflected by later
rotation or repeated, same-sense tumbling about an equatorial equatorward APW of the North American Jurassic APW cusps
(Beck and Housen, 2003) and possible correlatives on other, events from somewhat longer (5–10 My) events of slower
less-constrained continents (Besse and Courtillot (2002), pace. Thus, while Prevot et al. (2000) favored a short interval
although see Hynes (1990) and subsequent discussion). It is of still-quicker TPW at 115 Ma, the conceptual introduction of
certainly worth testing this scenario with high-resolution paleo- oscillatory, type II TPW further complicates such recognition.
magnetic studies in the Permian–Triassic boundary interval, par- Besse and Courtillot (2002) favored conservative interpre-
ticularly from sedimentary sections that have preserved stable tation of the synthetic global APWP dataset and ascribed part
components of that age (e.g., Ward et al., 2005). or all of Prevot’s 115 Ma event to a small number of data
For the post-200 Ma global paleomagnetic record, in which a bracketing the interval and/or to inaccurate dating of suppos-
fixed hot-spot reference frame may be hypothesized in order to edly 118–114 Ma, petrologically dissimilar kimberlites in
model true plate-tectonic drift with respect to the mesosphere, South Africa.
Besse and Courtillot (2002, 2003) had presented the most Petronotis and Gordon (1999) and Sager and Koppers
recent thoroughly integrated database and discussion (e.g., (2000a) hypothesized very fast TPW between 80 and 70 Ma
Andrews et al., 2006; Besse and Courtillot, 1991; Camps et al., and near 84 Ma, respectively, possibly as the middle of an oscil-
2002; Cottrell and Tarduno, 2000; Dickman, 1979; Gordon, latory, triple event. Cottrell and Tarduno (2000) emphasized
1987, 1995; Gordon et al., 1984; Harrison and Lindh, 1982; those authors’ enumeration of potential ambiguities and pitfalls
Prevot et al., 2000; Schult and Gordon, 1984; Tarduno and of magnetic anomaly modeling and age dating of seamounts
Smirnov, 2001, 2002; Torsvik et al., 2002; Van Fossen and (but see Andrews et al. (2006) for a promising new treatment)
Kent, 1992). Besse and Courtillot (2002) critically discussed and suggested that hypothesized Campanian–Maastrichtian
hypothesized Cretaceous fast TPW and provided in-depth con- TPW events fail a global signal test when compared with magne-
sideration of tests and caveats that are only summarized here. tostratigraphic data from Italy (e.g., Alvarez and Lowrie, 1978;
In brief, while many authors have noted that hot spots may Alvarez et al., 1977). Sager and Koppers (2000b) questioned
form a significantly nonuniform reference frame (e.g., recently, whether that Italian magnetostratigraphic data are of sufficient
Cottrell and Tarduno, 2003; Riisager et al., 2003; Tarduno and precision to rule out all possible interpretations of the Pacific
Cottrell, 1997; Tarduno and Gee, 1995; Tarduno and Smirnov, seamount dataset, within its own uncertainty.
2001; Tarduno et al., 2003), Besse and Courtillot (2002) based The apparent dispersion of Pacific paleomagnetic data,
their synthesis upon Indo-Atlantic hot spots, which appear hypothetically accounted for by TPW, remains unresolved in
more fixed to each other than they are to Pacific hot spots still more recent studies (Sager, 2006). The younger (<85 Ma)
(Muller et al., 1993). Some Pacific hot-spot motion models interval of Cretaceous TPW ought to be readily testable by
(e.g., DiVenere and Kent, 1999; Petronotis and Gordon, 1999) new magnetostratigraphies from various continents to parallel
are similar to Besse and Courtillot’s (2002) Indo-Atlantic hot- the classic Italian sections. Detailed magnetostratigraphic
spot APWP, although Van Fossen and Kent (1992) argued for analysis of Paleocene–Eocene transitional sections exposed
incompatibility of the two oceanic reference frames. Ulti- on continents led Moreau et al. (2007) to hypothesize small-
mately, whether or not hot-spot motion represents coherent magnitude, fast, back-and-forth TPW during that time, a pos-
or independent drifting in a ‘mantle wind,’ TPW of the whole sibility that may also be supported by paleomagnetic data from
solid Earth in the geomagnetic/celestial reference frame should the North Atlantic Igneous Province (Riisager et al., 2002).
still be resolvable if it is rapid or long-lived, by discerning
common APW tracks from all continents.
5.14.2.6 Hypothesized Rapid or Prolonged TPW:
Hynes (1990) invoked a sort of TPW for Early Jurassic time,
‘Cryogenian’–Ediacaran–Cambrian–Early Paleozoic
c.180–150 Ma, but later compilations based on better-
constrained hot-spot age and position data mark the same Substantial evidence is accumulating that Earth experienced
period as a standstill (e.g., Besse and Courtillot, 2002). The large and rapid bursts of TPW during Neoproterozoic and
post-140 Ma interval, which Hynes (1990) marked as a early Paleozoic time, of a sort that perhaps has not been
standstill, appears strongest as prolonged, possibly slow TPW experienced since. This ‘type II’ TPW, invoked as a single
in those later models. Since the seafloor hot-spot record event through early to middle Cambrian time by Kirschvink
before 150 Ma is poor and continental hot-spot tracks may et al. (1997) and debated by Torsvik et al. (1998), Evans et al.
be subject to eruptive–tectonic complications, Hynes’ (1990) (1998), and Meert (1999), was originally termed inertial inter-
Jurassic TPW event is of questionable support. Recent studies change TPW (IITPW). We prefer Evans’ (2003) renaming of the
in lower Jurassic South American strata (Llanos et al., 2006) process principally because it deemphasizes the reference
and Upper Jurassic–Early Cretaceous successions of Adria frame of the pre-TPW inertial tensor.
(Satolli et al., 2007) appear to support possible bursts of fast Per Kirschvink et al. (1997), ‘interchange’ refers to the
TPW during this interval. beginning of such TPW, when the maximum and intermediate
Both Besse and Courtillot (2002) and Prevot et al. (2000), inertial moments of that tensor’s reference frame switch iden-
however, marked the middle Cretaceous as an interval over tities. One frequent misconception has been that, conse-
which TPW appears to have sustained a net faster, coherent quently, the TPW response of the solid Earth must be a
component of motion in the geomagnetic (celestial) reference before-to-after change of 90 . As noted in, for example, Evans
frame (<0.5 My1) than during TPW standstill intervals et al. (1998) and Matsuyama et al. (2006), that change may be
before and afterward. In both studies, the authors note that any magnitude of rotation at all < 90 and continuing until
even a global synthetic APWP lacks the time resolution (and new axes are definable and stable, orthogonal from each other
for some intervals, global pole coverage and/or paleomagnetic and from the (effectively stationary) equatorial, Imin. The end-
pole precision) necessary to discriminate very short, rapid TPW member case of a 90 shift would imply that the positive mass
anomaly causing the shift was imposed precisely along the spin succession deposited on Rodinia’s margin c.800 Ma (Maloof
axis (as might be the case with an upper-mantle plume head). et al., 2006). With primary remanence established by a positive
Off-axis eruption would lead to smaller magnitude events. synsedimentary fold test and by apparently correlatable polarity
We prefer renaming Kirschvink et al.’s (1997) ‘IITPW’ as reversals, this result is highly robust. Although the sequence is
‘type II’ TPW also to emphasize its hypothesized proclivity to undated except by stratigraphic and carbon isotopic correlations
multiple events and to differentiate it from other patterns, as to two Australian subbasins, inferred TPW events appear rapid on
per Sections 5.14.2.4 and 5.14.2.5. a sequence-stratigraphic timescale. Using global carbon isotopic
The hypothesized pan-Cambrian type II TPW event of correlation and thermal subsidence analysis, the TPW event-
Kirschvink et al. (1997) defines an approximate Imin axis (the recurrence timescale is estimated at 15 My.
‘TPW axis’). Evans (2003) noted that this TPW axis is essen- Li et al. (2004) invoked type I TPW to explain dramatic
tially identical to both a plausible centroid of the superconti- paleomagnetic difference between South China’s high paleola-
nent of Rodinia and that hypothesized in the reference frame titude, 802 10 Ma Xiaofeng dikes, and unconformably over-
of the largest early Paleozoic continent, Gondwanaland, by lying, gently tilted, low paleolatitude, 748 12 Ma Liantuo
Van der Voo (1994) (also see Veevers (2004) and later Piper glacial deposits (Evans et al., 2000; Piper and Zhang, 1999;
(2006) and Figure 3 here; and see subsequent sections). and see subsequent sections). Although the age of Li et al.’s
That coincidence of the Van der Voo (1994) and Kirschvink (2004) hypothesized TPW is imprecisely constrained and
et al. (1997) TPW axes was invoked as an example of the Maloof et al.’s (2006) hypothesized events are not dated
potential long-lived legacy of supercontinental superswell directly, the two studies appear to invoke different types of
geoid anomalies by Evans (2003), also citing, as a modern TPW for, essentially, a single phase of a supercontinent cycle
analog, Pangaea’s relict geoid persisting to the present at c.800 Ma. Large uncertainties in the absolute ages of the
(Anderson, 1982; Chase and Sprowl, 1983; Davies, 1984; Svalbard succession studied by Maloof et al. preclude precise
Richards and Engebretson, 1992). Whether relict from a large correlations to the South China poles; furthermore, uncer-
Ediacaran–Cambrian continent or from the earlier superconti- tainties in Rodinia paleogeography (e.g., Pisarevsky et al.,
nent Rodinia, Evans and Kirschvink (1999) and Evans (2003) 2003) leave considerable flexibility for reconstructing South
noted that, in fact, at least three distinct TPW pulses (of appar- China relative to Laurentia (including Svalbard). Identifying
ently alternating sense) are implied for both late Neoprotero- the type (I vs. II) of TPW at c.800 Ma and inferring its dynamic
zoic time and early Paleozoic time. cause must await better constraints in these two topics.
Oscillatory, rapid TPW about an equatorial axis has also been
invoked to explain stratigraphically systematic but 50 dispersed
5.14.2.7 Hypothesized Rapid or Prolonged TPW: Archean
paleomagnetic poles produced from Svalbard’s Akademikerbreen
to Mesoproterozoic
Evans (2003) suggested that long tracks and loops of Lauren-
tia’s Mesoproterozoic APWP might represent oscillatory TPW
Middle in the inertial legacy of Earth’s Paleoproterozoic superconti-
Cambrian nent, Nuna, although global paleomagnetic database synthesis
testing of such a hypothesis has yet to be presented robustly.
Mid-
Considering the oldest volcanic succession with detailed
Vendian
strata-bound paleomagnetic analyses, Strik et al. (2003) noted
Early that one of several unconformities punctuating the basalt
Carboniferous series of Pilbara’s late Geon 27 Fortescue Group separates
zones of distinct magnetization direction, but with similar pet-
rographic character, marked by a 27 inclination difference
across 3 My. These authors enumerate rapid TPW as one pos-
Early sible explanation, although they prefer alternative scenarios.
Vendian
Earliest
Cambrian
Silurian–
5.14.3 Geodynamic and Geologic Effects and
Imin Inferences
Devonian
5.14.3.1 Precision of TPW Magnitude and Rate Estimation
700(?)–350 Ma
The magnitude and rate of type 0 TPW are directly measured or
Figure 3 Stability of Imin. Van der Voo (1994) noted that derived quantities. Precision is often limited, in practice, by
Gondwanaland’s Cambrian–Carboniferous APWP includes dramatic technical capabilities of geodetic and satellite instruments. In
excursions; and Kirschvink et al. (1997) and Evans (2003) noted that late
principle, it is ultimately limited by uncertainties in fundamen-
Neoproterozoic–Cambrian paleopole dispersion marks swings with
tal constants like g and by observational design, which may not
approximately the same orientation. Evans (2003) suggested that all
those swings might represent type II TPW about a paleoequatorial readily quantify special relativistic effects. These uncertainties
minimum moment-of-inertia axis defined by a mantle superswell legacy are probably trivial for the purposes of any geologic applica-
of the Neoproterozoic supercontinent, Rodinia. Reproduced from tion of type 0 TPW data.
Figure 3 in Evans DAD (2003) True polar wander and supercontinents. The magnitude and rate of types I and II TPW are controlled
Tectonophysics 362(1–4): 303–320. by the same processes, although we expect type I TPW to be
slower than type II TPW. This is because the hypothesized TPW(h): instantaneously imposed mantle heterogeneity
forcing (long-wavelength mantle upwelling in context of initial 0
Imin Iint) is more slowly imposed. Both phenomena should
Depth axis (km)

be inherently rate-limited by the speed at which the Earth’s 660
viscoelastic daily rotational bulge can relax through the incre-
mentally shifting solid Earth. Both phenomena should be
resolved in magnitude and rate at the lowest limits of paleo-
magnetic and geochronological error and uncertainty.
Quantifying the viscoelastic rate control on types I and II 2900
TPW is nontrivial. The three controlling variables are probably 1020 1021 1022 1023
for ‘full’ (90°) TPW

Relative timespan
1x
forcing magnitude of inertial changes, timescale of inertial Mantle viscosity (Pas)
change, and effective mantle viscosity (Tsai and Stevenson
(2007); expanding on Munk and MacDonald (1960)). Most 3x
TPW numerical models assume a single viscosity or two-
layered viscosity for the solid Earth; at any rate, there is no 9x
consensus on the precise viscosity structure of the modern
mantle. We argue that these models might equally well be Figure 4 Viscosity structure in the Earth’s mantle affecting hypothetical
driven by empirical paleomagnetic data and vice versa. TPW dynamics. For a given controlling mantle viscosity, maximum
Simplified mantle modeling confirms that lesser effective TPW rate may be modeled (red curve). In the cartoon representation
mantle viscosity permits faster TPW and that more highly struc- shown here, present-day TPW controlled by mantle viscosity of
1023 Pa s would span a given arc of rotation 10 times slower than
tured mantle viscosity enhances sensitivity to a given forcing
TPW controlled by mantle viscosity of 2 1021 Pa s. The viscosity
(because of the dynamic topography and surrounding entrain-
structure of the Earth’s modern mantle is uncertain in sense and
ment effects). The balance between those influences, however, magnitude at many levels, particularly in the deep mantle. Mitrovica
is unclear. Given a mantle-viscosity structure (e.g., Mitrovica and Forte (2004), however, suggested a dual inverse-modeled mantle-
and Forte, 2004), simple TPW rate limitation calculations (e.g., viscosity structure as depicted in yellow (with viscosity uncertainties
Spada and Boschi, 2006; Spada et al., 2006; Steinberger and occupying horizontal space and mantle position and thickness varying
O’Connell, 1997) may be controlled by average mantle viscos- vertically). It is not clear to us whether a highly structured mantle, such
ity or perhaps maximum viscosity. Obviously, more experimen- as that inferred for modern Earth, would be effectively ‘controlled’ by
tal and theoretical work is needed on this question. its maximum-viscosity shell or whether it will be controlled by some
We project the possible shape of a TPW rate limitation integrated average viscosity, permitting faster net TPW. Projection of
ever-more refined TPW rate-response calculations and ever-more
versus controlling viscosity curve onto the Mitrovica and
sophisticated viscosity structure models is critical to understanding
Forte (2004) mantle-viscosity structure in such a way that
the putative TPW rate limitation on modern Earth. If ancient TPW is
suggests, today, full type I (i.e., pole to equator) TPW could demonstrated to occur at significantly faster rates, secular evolution of
occur over an order-of-magnitude different timescales, within controlling mantle viscosity (or development of viscosity structure)
the span of possible controlling mantle viscosity (Figure 4). could be inferred and inversely modeled. Adapted from Figure 4(b) in
Besse and Courtillot (2002) synthesized the global paleo- Mitrovica JX and Forte AM (2004) A new inference of mantle viscosity
magnetic record over the past 200 My and considered the most based upon joint inversion of convection and glacial isostatic adjustment
conservative solid Earth (e.g., Indo-Atlantic hot spot) reference data. Earth and Planetary Science Letters 225(1–2): 177–189.
frame in their analysis. TPW appears to be an episodically
measurable contributor to APW. Over ‘fast’ TPW intervals esti-
mated to last 20–40(?) My, TPW may occur at <0.5 My1. As delayed response of the solid Earth’s viscoelastic bulge com-
already discussed, those authors acknowledge that time resolu- pared with the immediate response of the ocean’s bulge to
tion in the global coverage of the paleomagnetic database limits changes in the spin vector, we again return to the prophetic
their analysis to moving 5 or 10 My windows at finest resolu- words of Alfred Wegener (Biram, 1966; translated from
tion. If faster TPW occurred over a shorter timescale, it would Wegener, 1929):
not be readily recognized using a time-averaged approach.
If ancient TPW were estimated – by APWP compilation in . . . Since the ocean follows immediately any reorientation of the
equatorial bulge, but the earth does not, then in the quadrant in
older times or by single-location paleomagnetic records of any front of the wandering pole increasing regression or formation of
age – to occur substantially quicker, then it should be possible dry land prevails; in the quadrant behind, increasing transgression
to use the TPW data to construct inverse models of the control- or inundation. (p. 159)
ling mantle viscosity. This approach might allow constraints to
be placed on mantle viscosity and related parameters as a The physics of this effect have been subsequently modeled
function of geologic time. with increasing sophistication (Gold, 1955; Mound and
Mitrovica, 1998; Mound et al., 1999, 2001, 2003; Sabadini
et al., 1990).
5.14.3.2 Physical Oceanographic Effects: Sea Level
Consider only the magnitude of the spin vector, v. A simple
and Circulation
decrease in magnitude of v (which is an increase in the length of
George Darwin (1877) was the first to note that changes in the day without any associated TPW) will reduce the size of the
Earth’s rotation would induce fluctuations in local sea level. In Earth’s equatorial bulge. However, because the solid Earth is
reviewing these early studies that specifically contrasted the viscoelastic, it requires 104 to 105 years to fully relax, whereas
the Earth’s ocean surface responds instantaneously. Hence, in Physical oceanography should also show second-order
equatorial regions, mean sea level will experience a transient responses to TPW. The Earth’s spin vector modulates a variety
relative fall with respect to the land surface, with a compensating of oceanographic parameters, including temperature, the pole-
rise at higher latitudes. Similarly, simply increasing v (shorten- to-equator atmospheric energy gradient, and the chemical
ing the day) will produce the opposite effect of equatorial trans- dynamics of the world ocean (e.g., mean and local salinity).
gressions and high-latitude regressions (Peltier, 1998). It also has a large influence on the tides, which depend on the
In contrast, altering the orientation of the spin vector rela- orientation and character of seafloor topography and conti-
tive to the equatorial bulge yields a quadrature pattern in rela- nental geography. The well-known poleward boundary cur-
tive sea level. As the v vector remains constant, dimensions of rents on east-facing continental margins and deepwater
the overall bulge also remain the same. A point moving toward upwellings on west-facing margins are fundamental outcomes
the equator, for example, will impinge onto the equatorial of the Earth’s sense of spin. Large TPW events will force changes
bulge and hence tend to increase its distance from the spin axis. in many of these parameters that could leave fingerprints in the
The solid Earth beneath this point, however, responds with geologic record (Figure 6).
the aforementioned viscoelastic time constant and will lag the During TPW events, we expect thermohaline circulation in
ocean’s instantaneous shift to the new, itinerant equilibrium the ocean to reorganize – possibly several times – at consider-
shape. That response difference generates a marine transgres- ably shorter timescales than the TPW events and possibly at
sion. Similarly, points on the Earth’s surface that move away shorter timescales than the resulting sea-level fluctuations. As
from the equator during TPW vacate the spin bulge and will originally west-facing margins near the equator rotate clockwise,
relax at the solid Earth’s timescale while the adjacent ocean upwelling will cease on those margins and may begin anew on
surface ‘falls’ faster, effecting a regression. The maximum rela- the west-rotating, originally southern margins of the continent
tive sea-level effect for an instantaneous (and impossible!) shift in consideration. Boundary currents driving basinal sediment
of exactly 90 would simply be the size of the Earth’s equatorial advection may reorganize, foundering or winnowing sand drifts,
bulge, or 10 km. In reality, sea-level excursions are con- as low-latitude east-facing continental margins migrate to polar
strained by the effective elastic lithosphere thickness and vis- regions, cross the equator, or rotate into zonality. Figure 6
cosity structure of the mantle. Mound et al. (1999) estimated illustrates some of the wide range of effects conceivable during
that, for reasonable estimates of present viscosity structure, a a TPW event for distinct continents occupying various original
90 TPW event acting over 10 and 30 My will generate >200 to geographic locations. Certainly, hypothesized ancient TPW
50 m excursions, respectively (Figure 5). TPW sea-level effects ought to leave dramatic sedimentological and geochemical sig-
are certainly of equivalent magnitude and perhaps larger than natures in the eu- and miogeoclinal records of most then-extant
those of standard glacioeustasy and should exert first-order passive margins (see also Section 5.14.3.3).
control of global sequence stratigraphy during intervals of At a dramatic extreme, Li et al. (2004) followed Schrag et al.’s
time for which TPW was significant, quick, and/or frequent. (2002) geochemical modeling and suggested that c.800–750 Ma
Estimates of the sea-level effects of the fastest possible TPW TPW may have initiated the ‘Snowball Earth’ glaciations of the
motions tend to flounder on the uncertainties of detailed Cryogenian period. Kirschvink and Raub (2003) invoked
mantle-viscosity structure, but excursions of approximately changes in eddy circulation and lateral sediment advection and
several kilometers in scale might be feasible for TPW events permafrost transgression during multiple Ediacaran–Cambrian
on the 1–3 My timescale (Kirschvink et al., 2005; Mound et al., TPW to destabilize methane clathrates, increase organic carbon
1999). Of course, sea level will relax to roughly its initial state remineralization, and effect ‘planetary thermal cycling’ on the
when a TPW event ends. biosphere. In turn, this stimulated the Cambrian explosion by
160 200
175
120
Relative sea level (m)
Relative sea level (m)
150
80 2 125
40 100
75
0 3 1 2 3
50
–40 25
1 0
–80
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 0 5 10 15 20 25 30 35
(a) Time (My) (b) Time (My)
Figure 5 Sea-level fluctuations as a function of TPW rate. (a) Modeled sea-level fluctuations over 25 My of TPW for three locations on the globe:
(1) a continent initially at midlatitude moving through the equatorial bulge to midlatitudes in the opposite hemisphere experiences moderate
transgression followed by significant regression, (2) an initially polar continent moving toward the equator experiences significant transgression
followed by moderate regression, and (3) a continent near the TPW axis experiences little sea-level fluctuation. The period and amplitude of each
anomaly in this model are most sensitive to input parameters of lithospheric thickness, upper-mantle viscosity, and TPW rate. (b) The amplitude of any
one continent’s sea-level anomaly increases nonlinearly with TPW rate; other variables held constant. Reproduced from Figure 2 in Mound JE, et al.
(1999) A sea-level test for inertial interchange true polar wander events. Geophysical Journal International 136(3): F5–F10.
Spin axis:
Iint or Imax
Cold
Loci of
deepwater
formation
will change
Evaporite Belt
Boundary
Upwelling
current
turns on
turns on
Boundary Warm Upwelling TPW axis:
current turns off
turns off Imin
Boundary Upwelling
current turns on
turns on Evaporite Belt
Cold
Figure 6 A wide range of first-order physical and chemical oceanographic changes may accompany sea-level fluctuations during TPW, depending
on the initial location of a continental margin with respect to the TPW spin axis. Sediment supply to margins initially girdled by eastern boundary currents
may founder; nutrient-rich upwellings on initially west-facing margins may cease and begin anew on initially zonal, later west-facing margins. In
general, TPW is expected to leave a legacy of eddy instability and thermohaline reorganization that proceeds episodically during and possibly after
an inertial shift.
alternately stressing species and forcing them to reduce genera- In Svalbard, this isotope shift is bracketed by two sequence
tion time, accumulating genetic mutations that would promote boundaries (sea-level erosional horizons). Although magnetiza-
morphological disparity and speciation. tion directions before and after the Bitter Springs event in the
Akademikerbreen Group are similar to each other, precisely at
these sequence boundaries, the mean magnetic declination
5.14.3.3 Chemical Oceanographic Effects: Carbon Oxidation
rotates >50 . Maloof et al. (2006) noted that this rotation is
and Burial
present in three separate stratigraphic sections separated by
As noted in Section 5.14.3.2, TPW is capable of producing a >100 km on a single craton, with congruent carbon isotope
variety of effects on global sea level, ranging from small, meter- signals. A pair of rapid TPW events separated by 10 My provide
scale regional fluctuations associated with isostatic effects from a single, unified explanation for two important observations.
glacial loading and unloading to possibly larger, even kilometer- First, TPW can account for the Bitter Springs isotope anom-
scale effects from rapid and ‘ultrarapid’ TPW. In turn, sea-level aly by producing a sudden shift in the fraction of global carbon
fluctuations can force shoreline migration, alter drainage pat- burial expressed as organic carbon deposition (forg), plausibly
terns, and cause geologic organic carbon to be remineralized. by changing the proportion of riverine sediment (and nutri-
Potentially, this remineralized carbon could come from bitu- ents) deposited at low-latitude versus at midlatitudes. Per gram
mens, kerogen, or pressure-destabilized methane clathrates. of sediment, low-latitude rivers are more effective at organic
As discussed previously, Maloof et al. (2006) provided a carbon burial than those at mid- or high-latitudes (Halverson
compelling case for a pair of rapid TPW events in middle et al., 2002; Maloof et al., 2006). Although paleomagnetic data
Neoproterozoic platform carbonates of East Svalbard, Norway. argue Svalbard was close to the Imin inertial axis, a 50 type II
Several profiles through the 650 m thick section record a TPW event could swing pan-hemispheric Rodinia, shifting oro-
clear step-function offset in the d13C signature of approxi- graphic precipitation loci, drainage patterns, and continental
mately 5% identified as the 800 Ma Bitter Springs event sediment delta locations, varying forg.
(Halverson et al., 2005), which lasted 10 My based on ther- Second, this pair of type II TPW events would produce
mal subsidence modeling and younger chronostratigraphic transient sea-level variations. Only two sequence boundaries
correlations (Figure 7). exist in this interval in Svalbard, and they bracket the isotope
802 ± 10
<746 ± 2
636 ± 1
663 ± 4
760 ± 2
735 ± 5
548
542
580
10 10
Keele Peak
Gaskiers
Trezona Anomaly
5 5
0 0
Bitter Springs Stage
–5 –5
Marinoan
–10 Sturtian Wonoka –10
Anomaly
800 750 700 650 600 550

(a) Age (Ma)
545 540 535 530 525 (Ma)

Ediac Nemakit–Daldyn Tommot Adtab Bot
8
Khorbusuonka
Turukhansk
6
SIBERIA
Kotuikan
Dvortsy
4 Bol’shaya Kuonamka
Isit
Zhurinsky
2 Anti-atlas, Morocco
–2
–4
–6
d 13C
(b)
Figure 7 TPW effects on the geologic carbon cycle. (a) One possible global inorganic carbon isotopic evolution time series through middle and late
Neoproterozoic time. At least three distinct glacial events punctuate this interval. While successions around the world are ever-better dated by high-
precision U–Pb ages on magmatic zircons in interbedded ashes and volcanic flows, ambiguities still exist in sequence-stratigraphic and
lithostratigraphic correlations between continents. (As discussed in the text and in captions to Figures 6 and this figure, global sequence-stratigraphic
correlations may be severely complicated by TPW.) Maloof et al. (2006) considered the long-lived negative carbon isotopic excursion dubbed ‘Bitter
Springs stage’ a consequence of type II TPW (see text). (b) Cambrian inorganic carbon isotopic evolution time series from well-dated successions
in Morocco and undated successions in Siberia. Kirschvink and Raub (2003) noted that organic carbon-based isotope curves over parts of the same time
interval from western United States and other areas share many similarly shaped and positioned excursions. Those authors hypothesize that some
or all of the negative carbon isotopic excursions in Cambrian time were partly driven by destabilization of seafloor and permafrost methane clathrates
during a legacy of eddy instability and thermohaline reorganization associated with Ediacaran–Cambrian type II TPW. Adapted from Figure 15(b) in
Halverson GP, et al. (2005) Toward a Neoproterozoic composite carbon-isotope record. Geological Society of America Bulletin 117(9–10): 1181–1207;
Part of Figure 8 in Maloof AC, et al. (2005) An expanded record of Early Cambrian carbon cycling from the anti-atlas margin, Morocco. Canadian
Journal of Earth Sciences 42(12): 2195–2216.
anomaly. As discussed previously, Li et al. (2004) had also equally well when applied to Ediacaran–Cambrian time; the
argued for rapid TPW in this general interval of time. geologic and physiographic conditions are distinctly different.
On a more speculative note, Kirschvink and Raub (2003) TPW-induced changes to the geologic carbon cycle are likely
suggested that a hypothesized interval of multiple, type II to be myriad, with potential positive feedbacks. Figure 8 elabo-
TPW may have been in part responsible for the production of rates the pervasive, systematic perturbations to the geologic
large and episodic oscillations in inorganic d13C during early
Cambrian time, the so-called Cambrian carbon cycles (Brasier
Relative time
et al., 1994; Kirschvink et al., 1991; Magaritz et al., 1986, 1991;
Maloof et al., 2005). The principal mechanism suggested – Continent ‘1’
Transgression
remineralization of organic carbon from exposed sediments
and/or destabilization of methane clathrate reservoirs along
continental margins and in permafrost – was patterned
after similar suggestions for methane release associated with
a marked inorganic carbon anomaly at the Initial Eocene
Thermal Maximum (IETM, formerly called the Paleocene–
Eocene Thermal Maximum, PETM) event (Dickens et al.,
T
1995, 1997).
Subsequent analyses have questioned this interpretation for ‘2’
the IETM event, based on revised estimates of the available R
methane reservoir stored in late Mesozoic and early Tertiary
Relative sea level

continental shelf and upper-slope sediments, methane clath-
rate residence times, and the short (decadal) residence time of T
R ‘3’
methane in the Earth’s present atmosphere relative to the
apparent 100 000+ interval of initial Eocene carbon isotopic T
‘4’
excursion (Buffett and Archer, 2004; Farley and Eltgroth, R
2003). In reviewing the controversy, Higgins and Schrag
(2006) argued for the sudden close-off of an epicontinental T
seaway through tectonic action to provide 5000 Gt of
organic carbon needed to produce the effect. ‘5’
We note here that Kirschvink and Raub’s (2003) TPW-based R
‘methane fuse hypothesis’ for the Cambrian anticipated many
of the arguments against methane summarized by Higgins
(2005). Several factors argue that the geologic context of the
Regression
multiple Cambrian carbon cycles is markedly different than

the singular end-Paleocene event, in our view making at least ‘6’
the partial involvement of a methane-derived light isotopic
reservoir more plausible.
First, Neoproterozoic time (including the Ediacaran Period)
is characterized by extended intervals during which the inor- Figure 8 Cartoon elaboration on systematic perturbations to the
ganic d13C remained markedly positive (+2 to +5%), most geologic carbon cycle caused by TPW-driven sea-level fluctuations. For
likely implying relatively high organic carbon burial fractions, schematic sea-level anomalies experienced by continents in six different
plausibly leading to order-of-magnitude larger methane reser- locations relative to a TPW axis, a different interval of each anomaly
voirs than extant. The long-term average inorganic carbon will be most susceptible to regression-induced methane clathrate
destabilization and oxidation of long-buried organic matter (green), to
isotopic value for most of the Phanerozoic Eon is 0%.
new burial of organic carbon on TPW-transgressed shelves (white) and
Second, Cambrian carbon cycles are individually no longer,
subsequent oxidation of part of the same organic carbon pool during
though collectively more prolonged, than Phanerozoic nega- later emergence (blue), and to renewed oxidation of ancient organic
tive carbon isotopic excursions, with at least 12 named cycles matter (yellow). Destabilization of methane clathrate in permafrost, not
up to middle Cambrian time (Brasier et al., 1994; Montanez explicated in this figure, is probably preferred at the very beginning of
et al., 2000), spanning an interval of 15–20 Ma. When such a cycle for appropriate continents. Depending on the initial sizes and
viewed at exceptionally high resolution (e.g., Maloof et al., mean isotopic values of each reservoir, the rate and duration of TPW,
2005), many individual Cambrian carbon cycles likely have and specific parameters including paleogeography and thermohaline
duration 100 ky–1 My, but with occasional sharp excursions circulation, it is possible to imagine (or quantitatively model) a wide
superimposed, reminiscent of the Early Eocene event super- range of global inorganic carbon reservoir isotopic responses. Continent
#1 moves from either of the Earth’s poles to its equator. Continent #2
imposed on its still-longer Early Eocene climate optimum
swings from midlatitudes on one side of the equator to midlatitudes in
oscillation. We suspect that the carbon isotopic oscillations
the opposite hemisphere. Continent #3, located near the TPW axis,
discovered recently in the Early Triassic (Payne et al., 2004) rotates slightly further away from the equator. Continent #4, located near
are individually of longer average duration, but occur over a the TPW axis, rotates slightly closer toward the equator. Continent #5
shorter overall chronological interval than those of the early swings from midlatitudes across the pole to midlatitudes in the
Cambrian. The reservoir arguments against methane contrib- same hemisphere, but on the opposite side of the globe. Continent #6
uting to the initial Eocene carbon excursion, then, do not apply moves from near the Earth’s equator to one of its poles.
carbon cycle expected during TPW-driven sea-level fluctuations (adopted by Kirschvink and Raub (2003)) noted that, in fact,
for continents in six distinct locations relative to a TPW axis. straightforward interpretation of the time series order of dis-
Some continents will be especially prone to regression-induced persed poles for relatively well-constrained continents like
methane clathrate destabilization and oxidation of long-buried Australia demands multiple TPW events.
organic matter, while others will dominantly accommodate new Although none of the Cryogenian–Cambrian paleomagnetic
burial of organic carbon on newly TPW-transgressed shelves. If poles for Australia is ‘absolutely’ dated, stratigraphic order
the continental location is appropriate and TPW sufficiently between those poles is clear; many come from a single sedimen-
rapid, then TPW transgression may flood continental highlands tary succession (South Australia’s Adelaide ‘Geosyncline’).
of exceptionally large area. That same organic carbon pool may We prefer a terminal Cryogenian–Ediacaran–Cambrian
oxidize during later emergence related to sea-level reequilibra- APWP for Australia composed of the poles in Table 1. While
tion; other continents may only remineralize ancient organic these poles are of varying quality, none can be demonstrated
matter for the first time in the TPW event near its end. persuasively to have been remagnetized or to have dramatically
Depending on the initial sizes and mean isotopic values of wrong age assignments without basing such an argument on
each reservoir, the rate and duration of TPW, and specific the inexplicability of the magnetization direction itself. Many
parameters including paleogeography and thermohaline circu- other paleomagnetic poles for Australia in this time interval
lation, it is possible to imagine (or quantitatively model) a have been excluded from our analysis, using data selection and
wide range of global inorganic carbon reservoir isotopic supercedence criteria we believe uncontroversial, but beyond
responses. If a global Ediacaran–Cambrian paleogeography the scope of this chapter.
were confidently reconstructed, box modeling of reservoirs The oldest group of poles, uppermost ‘Cryogenian’ in age,
over relative timescales as indicated in Figure 8 ought to are probably 635 Ma or slightly older, based on ‘Snowball
match the general shape and timescale of Ediacaran–Cambrian Earth’ lithological and chemostratigraphic correlations to well-
carbon isotopic excursions, if the carbon cycles are, in fact, dated successions in Namibia and South China. Four pub-
genetically linked to an oceanographic legacy of type II TPW. lished poles from the Elatina Formation and one from the
immediately underlying Yaltipena Formation, presumed
genetically related to the onset of Elatina glaciation or deglaci-
5.14.4 Critical Testing of Cryogenian–Cambrian TPW ation, cluster well, placing Australia very near the equator. (In
today’s geomagnetic reference frame, paleopole longitudes for
5.14.4.1 Ediacaran–Cambrian TPW: ‘Spinner Diagrams’
this cluster are near 330, 360 E and pole latitudes are in 40,
in the TPW Reference Frame
55 S.) A pole position from immediately overlying, earliest
While Kirschvink et al. (1997) suggested that a single TPW Ediacaran Brachina Formation siliciclastics, is similar though
event in the interval 530–508 Ma accounted for the anoma- slightly far-sided.
lous dispersion between late Neoproterozoic and late Cam- The only mid-Ediacaran paleomagnetic pole we consider is
brian poles for most continents, Evans and Kirschvink (1999) that of the undated Bunyeroo Formation, 200–400 m
Table 1 Paleomagnetic poles defining ‘Cryogenian’–Ediacaran–Cambrian APW oscillations for Australia
Pole Site Site Pole Pole Age order

# Pole ID latitude longitude latitude longitude dp dm (1 ¼ oldest) Reference
14 Hudson 17 129 18 19 13 13 9 Luck (1972)

13 Hugh River 23.8 133 11.2 37.2 4.9 9.4 8 Embleton (1972)
12 Billy Creeka 31.1 138.7 37.4 20.1 7.2 14.4 7 Klootwijk (1980)
11 Kangaroo Isl. A 35.6 137.6 33.8 15.1 6.2 12.3 7 Klootwijk (1980)
10 Todd River/Allua/Eninta 23.4 133.4 43.2 339.9 4.5 7.7 7 Kirschvink (1978)
9 Upper Arumbera 23.4 133.4 46.6 337.4 2.6 4.6 6 Kirschvink (1978)
8 Lower Arumbera/upper 23.4 133.4 44.3 341.9 7.7 13.6 5 Kirschvink (1978)
Pertatataka
7 Bunyeroo 31.6 138.6 18.1 16.3 6.5 11.8 4 Schmidt and Williams
(1996)
6 Brachina 30.5 139 33 328 12 20 3 McWilliams and
McElhinny (1980)
5 Elatina SKCB 31.3 138.7 39.6 1.7 3.3 6.4 2 Sohl et al. (1999)
4 Elatina SW 31.2 138.7 51.5 346.6 7.7 15.3 2 Schmidt and Williams
(1995)
3 Elatina SWE 32.4 138 54.3 326.9 0.9 1.8 2 Schmidt et al. (1991)
2 Elatina EW 32.4 138 51.1 337.1 1.7 3.4 2 Embleton and Williams
(1986)
1 Yaltipena 31.3 138.7 44.2 352.8 5.8 11.3 1 Sohl et al. (1999)
Best-fit great circle 114.3 32.2
a
Billy Creek result is excluded from swath-pole statistics by preference for possibly coeval ‘Kangaroo Island A’ result but listed here to weigh against oroclinal rotation between distant
Flinders Ranges sites as an alternative explanation for pole dispersion.
stratigraphically higher and separated from uppermost et al., 1997) or multiple events of unresolved duration and
Brachina Formation by at least one supersequence boundary magnitude summing to the same net rotation.
in those study areas of South Australia. Although the Bunyeroo The resulting Australian APWP (Figure 9(a)) does not
Formation is lithologically similar to the Brachina Formation, resemble a time-progressive ‘path’ so much as a pole ‘swath’
shares a similar burial and gentle-folding tectonic history, and (although in fact, time progression of that path would indicate
passes a reliability-by-comparison test with the Acraman repeated reversal of rotational motion for Australia in the
Impact melt rock aeromagnetic anomaly on the nearby Gawler celestial reference frame).
Craton (the Bunyeroo Formation hosts ejecta of that impact We suggest that a more insightful way to view such an APW
event), its pole position is significantly different, lying at swath is in the hypothesized TPW reference frame itself. We
16.3 E, 18.1 S (Schmidt and Williams, 1996) and implying calculate a best-fit great circle to Australia’s APW swath using
mostly counterclockwise rotation of Australia, in the celestial modified least-squares techniques; in so doing, we assume that
reference frame, during early Ediacaran time. (multiple) TPW over the time interval spanned by the constit-
Kirschvink’s (1978) middle–upper Ediacaran pole from uent poles occurred at a rate far greater than Australia’s
central Australia’s upper Pertatataka Formation and lower plate-tectonic motion in the deep-mantle reference frame.
Arumbera Formation returns to the vicinity of Elatina poles, This best-fitting great circle and its corresponding pole with
suggesting mid-Ediacaran clockwise rotation for Australia in 95% confidence are shown in Figure 9(b).
the celestial reference frame. Terminal Ediacaran magnetiza- By rotating Australia and its pole swath and best-fitting
tion from upper Arumbera Formation is broadly similar. great circle so that the pole to that circle is at the center of an
Lower Cambrian poles from South Australia once more equal-area projection, we place the hypothesized TPW axis at
return to the vicinity of 340, 015 E and 45 S, suggesting the ‘pole’ of the stereonet. If TPW were sufficiently fast and
terminal Ediacaran or earliest Cambrian counterclockwise since polarity choice of each paleopole is arbitrary, Australia
rotation of Australia and Cambro–Ordovician paleopoles may effectively occupy any rotationally identical position
from South, Central, and northern Australia all are still further about that TPW axis, except for precise ages where its paleola-
shifted northward in the modern reference frame, permitting titude is constrained by a corresponding pole. (At those ages,
one long Ediacaran–Cambrian TPW event (per Kirschvink Australia may occupy one of precisely two absolute
14
13
14
13
7
11
12
6 7
5
1 11
8 10
9 4 6
2 5 4
3 1
8,10 9
2 3
(a) (b)
Figure 9 Example of a ‘spinner diagram.’ (a) Late Cryogenian–Ediacaran–Cambrian APW swath for Australia in the modern geographic reference
frame. Fourteen critical poles, dated by biostratigraphy or relative stratigraphic position, appear widely dispersed, and the pole (with 95% confidence
trace) to a great circle best-fitting thirteen of those poles lies near the western margin of the Australian craton. (b) Rotation of Australia, its APW
swath, and the swath’s best-fitting great circle so that the pole to that swath lies at the center of an equal-area projection creating a ‘spinner diagram.’ If it
is supposed that the back-and-forth time series comprising a continent’s APW swath represents multiple (type II) TPW events at rates substantially
faster than (or parallel to) plate motion and if it is assumed, by reasoning outlined in the text and in Kirschvink et al. (1997) and Evans (2003), that the
TPW axis lies on Earth’s spin equator, then because of geomagnetic polarity ambiguity in Neoproterozoic time, Australia’s actual geographic location at
most times during the interval spanning all proposed TPW events may be any rotationally equivalent location that preserves coincidence of its APW
swath pole and the TPW axis. For the purposes of paleogeographic reconstruction, Australia’s geomagnetic (¼rotational) paleolatitude is constrained
only for the discrete times of direct paleomagnetic constraints as represented by the individual poles. If rapid type II TPW oscillations were occurring on
timescales of a few million years or less, then traditional comparisons of separate plates’ paleolatitudes would be meaningless except in the rare instance
of extremely precise agreement in pole ages from those plates. Each point on the Earth does, however, maintain a constant angular distance from a type
II TPW axis; thus, continents can be reconstructed relative to that reference frame. In practical terms from the figured example, Australia may be ‘spun’
around the (constantly equatorial) TPW axis shown in the center of the diagram. An alternative set of paleogeographic positions occupies the far
hemisphere about the antipole to the TPW axis.
paleogeographic positions, in the TPW reference frame, recog- ‘Africa’ (assuming Congo–Kalahari coherence and ignoring
nizing geomagnetic polarity ambiguity.) poleless São Francisco), and for all of ‘South America’ (assum-
At all other times, there is rotational paleolatitude ambigu- ing Amazonia–Rı́o de la Plata coherence). These swaths are
ity associated with rapid oscillations about the TPW reference depicted in Figure 10 using poles in Table 2, both in the
axis. We call this projection a ‘spinner’ diagram, because the present geomagnetic reference frame and as spinner diagrams
position of a continental block may be freely ‘spun’ about the in the inferred TPW reference frame.
TPW axis at the center of the spinner projection. Since all We are favorably impressed that, by superimposing each
continents must undergo identical TPW, all continental blocks’ continent’s APW swath pole at the center of a common
computed spinner diagrams may be overlapped; and each spinner projection, continents may be ‘spun’ so that a nearly
continent ‘spun’ to avoid overlaps, to produce a permissible Gondwanaland configuration is produced, with relatively
intercontinental paleogeography in the TPW reference frame. minor shifting of each continent’s APW swath pole within its
95% confidence limit (Figure 11).
Although we recognize that Gondwanaland was not yet
5.14.4.2 Proof of Concept: Independent Reconstruction
entirely formed during Ediacaran–Cambrian time (e.g., Boger
of Gondwanaland Using Spinner Diagrams
et al., 2001; Collins and Pisarevsky, 2005; Valeriano et al.,
Although Australia’s paleomagnetic APW swath is by far the 2004), the post-Ediacaran relative movements between its con-
best constrained of the Gondwanaland constituent continents, stituents should have been minor, and even Late Precambrian
all those Gondwanaland elements show dispersed paleopoles stitching of constituent African and South American Cratons
easily fit by great circles. We show those APW swaths for India was permissibly nonmobilistic on a tens of degrees spatial scale
and Antarctica (Mawsonland), for ‘Arabia(–Nubia),’ for all of (Veevers, 2004). We also recognize that our Arabian–Nubian
49 44 45 40 39
15 47
41
46 42 38
43
1716 48
18
19
21
20
(a) (d) (g)
26 28–37 46 47
15 1716
27 44
18 49
19
21 48
23, 24 20
22
25
(b) (e) (h)
40 39
42
51
43 52
23, 24 53
41 22
38
28–37 25
50
(c) (f) 26 (i)

27
Figure 10 Elaboration of proto-Gondwanaland’s APW commonality. Late Cryogenian–Ediacaran–Cambrian APW distributions, in the modern
geographic reference frame (but various views) for the Indian and Antarctic Cratons; for ‘Arabia,’ a simplification of the Arabian–Nubian Shield and
Neoproterozoic basement in modern Iran; for ‘South America,’ a simplification of the Neoproterozoic Rı́o de la Plata and Amazonia Cratons in late
Brasiliano time; and for ‘Africa,’ a simplification of the Neoproterozoic Kalahari and Congo–São Francisco Cratons (see text for tectonic reconstruction
caveats). All cratons except ‘Africa’ are also represented in spinner diagrams using poles to best-fitting great circles through critical-pole APW
distributions. Adapted from Figure 3 in Evans DA, et al. (1998) Polar wander and the Cambrian. Science 279: 9a–9e.
Table 2 Paleomagnetic poles defining ‘Cryogenian’–Ediacaran–Cambrian APW oscillations for other Gondwanaland constituents

India 89.3 38.3 13.8 13.8

21 Salt Range pseudomorph rot. 30c 32.7 73 7.2 192 9 11.8 6 Wenski (1972)
20 Purple sandstone rot. 30c 32.7 73 8.2 189.4 12.5 18.5 5 McElhinny (1970)
19 Khewra A rot. 30c 32.3 71 19.3 197 10 14 4 Klootwijk et al.
(1986a,b)
18 Bhander sandstone 23.7 79.6 31.5 199 3.1 5.9 3 Athavale et al. (1972)
17 Upper Vindhyan, Rewa–Bhander 27 77.5 51 217.8 5.6 11.2 2.5 McElhinny (1970)
16 Rewa 23.8 78.9 35 222 9.6 15.9 2 Athavale et al. (1972)
15 Mundwarra Complex 25 73 42 295 12 22 1 Athavale et al. (1963)
Antarctica (Mawsonland) 234.8 83.5
37 Teall Nunatak 74.9 162.8 11 31 3.6 7.2 8 Lanza and Tonarini
(1998)
36 Wright Valley granitics 77.5 161.6 5.5 18.4 4.1 8.2 8 Funaki (1984)
35 Lake Vanda feldspar porphyry 77.5 161.7 4 27 4.3 8.6 7 Grunow (1995)
dikes
34 Dry Valley mafic swarm 77.6 163.4 9.4 26.7 5.5 10.9 7 Manzoni and Nanni
(1977)
33 Killer Ridge, Mt. Loke diorites 77.4 162.4 7.1 21.2 6 12 6 Grunow and
Encarnacion (2000)
32 Lake Vanda Bonney pluton 77.5 161.7 8.3 27.8 6.6 13.1 6 Grunow (1995)
31 Granite Harbour pink granite 76.8 162.8 3.4 16.6 4.4 8.5 6 Grunow (1995)
30 Granite Harbour mafic dikes 77 162.5 0.9 15 4.7 9 6 Grunow (1995)
29 Granite Harbour gray granite 77 162.6 3.6 21.7 3.3 6.5 6 Grunow (1995)
28 Mirnyy Station charnockites 66.5 93 1.4 28.6 8 16 5 McQueen et al.
(1972)
27 Upper Heritage Group 79.2 274 4 296 7.1 12.5 5 Watts and Bramall
(1981)
26 Liberty Hills Formation 80.1 276.9 7.3 325.4 4.3 8.6 4 Randall et al. (2000)
25 Sør Rondane intrusions 72 24 28.4 9.5 9.9 5.7 3 Zijderveld (1968)
24 Briggs Hill Bonney pluton 77.8 163 2.7 41.4 7.8 15 2 Grunow (1995)
23 Wyatt and Ackerman Formations, 86.5 170 1.1 39.3 4 8 1 Grunow and
Mt. Paine tonalite Encarnacion (2000)
22 Zanuck granite 86.5 38.8 7.1 38.8 5.5 10.9 1 Grunow and
Encarnacion (2000)
Arabia 60 5.8
43 Hornblende diorite porphyry 21.7 43.7 25.8 332.2 9.1 14.7 Unknown Kellogg and
Beckmann (1982)
42 Red sandstone, Jordan 29.7 35.3 36.7 323.4 6.7 10 5 Burek (1968)
41 Jorden dikes 29.5 35.1 26 161 5.1 9.4 4 Sallomy and Krs
(1980)
40 Arfan/Jujuq 21.3 43.7 77.4 297.9 6.9 12.2 3 Kellogg and
Beckmann (1982)
39 Mt. Timna 29.8 34.9 83.6 223.2 3.3 5.4 2 Marco et al. (1993)
38 Mirbat Sandstone 17.1 54.8 23.3 141.8 3.9 7.5 1 Kempf et al. (2000)
‘South America’ (Rı́o de la Plata, 48.5 34.3
Amazonia, late Brasiliano)
49 Mirassol d’Oeste remagnetization 15 302 33.6 326.9 7.1 10 6 Trindade et al. (2003)
48 Upper Sierra de las Animas 34.7 304.7 5.9 338.1 19.6 26.7 5 Sanchez-Bettucci and
Rapalini (2002)
47 Campo Alegre 26.5 310.6 57 43 9 9 4 D’Agrella-Filho and
Pacca (1988)
46 Playa Hermosa 34.7 304.7 43 18.4 8.6 16 3 Sanchez-Bettucci and
Rapalini (2002)
45 Mirassol d’Oeste characteristica 15 302 82.6 112.6 5.6 9.3 2 Trindade et al. (2003)
44 Urucum 19 302 34 344 9 9 1 Creer (1965)
‘Africa’ (Congo and Kalahari) 61.5 21.4
53 Doornpoortb 23.7 17.2 21.8 64.9 6.8 11 4 Piper (1975)
52 Dedza Mtn. syenite 14.4 34.3 26 341 10 10 3 McElhinny et al.
(1968)
(Continued)
Table 2 (Continued)

51 Ntonya ring structure 15.5 35.3 27.8 344.9 1.4 2.3 2 Briden et al. (1993)
50 Sinyai 0.5 37.1 29 319 3 5 1 Meert and van der
Voo (1996)
a
The Mirassol ChRM pole falls considerably off of the great circle defined by the younger five South American poles. Following Valeriano et al. (2004), we suppose that the
Mirassol d’Oeste characteristic as an old pole located near a border of the mobile belt might be rotated. The ‘South American’ great circle including Mirassol d’Oeste
characteristic has a pole at 19.7, 242.9.
b
Doornport, a likely Cambrian remagnetization, is internal to the Damaride fold belt. It is far from the (admittedly underconstrained) ‘African’ great circle that it must either have
indicated an inappropriate lumping of date from the constituent cratons or have been rotated by Damaride structures. We acknowledge the relatively unsatisfying character of the
‘African’ and ‘South American’ databases, but leave full consideration for a future paper.
Neither ‘South America’ nor ‘Africa’ existed in any semblance of the modern, during ‘Cryogenian’–Ediacaran–Cambrian time. However, the ‘constituent’ Precambrian cratons of
each – Rı́o de la Plata and Amazonia for South America and Kalahari and Congo–São Francisco for Africa – might have remained nearly fixed with respect to each other during the
pan-African and Brasiliano mobilisms. Likewise, although the ‘Arabian–Nubian’ Shield incorporates significant juvenile arc material, we suppose it might have occupied a
common general area of the latest Neoproterozoic real estate, including basement terranes. In any case, it is intriguing that, as a ‘block,’ Arabia may be treated, essentially, as one of the
other cratons, by owning a dispersed APW swath that, when fit to a great circle, reconstructs nearly to Gondwanaland configuration independent of seafloor retrorotations.
We consider it almost certain that some of these paleomagnetic poles, unconstrained by direct field tests of magnetization fidelity, are in fact remagnetizations or tectonically rotated.
However, we have attempted to compile a list of possible primary magnetizations that could not be discounted except by concern for the actual direction obtained.
(a) (b)
Figure 11 Multiple type II TPW in late Cryogenian–Ediacaran–Cambrian time, proof of concept. (a) Using only spinner diagrams, spinning each
continent about its own APW swath-fitting great circle pole and shifting those poles about a common, inferred TPW axis within each 95% confidence
limit, it is possible to reconstruct Gondwanaland approximately. Because Gondwanaland formed during the same period of time as these
hypothesized multiple TPW events, with possibly minor reshuffling of the constituent Rodinia cratons, we expect its later-dispersed fragments to
share the essentially common APW swath shown in (b). This is equivalent to the earliest APW swings depicted by Evans (2003) (Figure 3) using the
same logic, and it expands upon the pole database of Kirschvink et al. (1997). Any alternative interpretation to the hypothesis of multiple, rapid
type II TPW events for late Neoproterozoic to Cambrian time must explain not only the individual, dispersed poles and continental APWPs but also why
all of the constituent cratons of Gondwanaland show APWPs that are of differing sense in the modern geographic reference frame, but which
restore to near-coincidence when viewed in the Gondwanaland reference frame.
‘craton’ includes amalgamated terranes and basement of vary- hypothesis is a viable, nontrivially discounted explanation for
ing age (Veevers, 2004); therefore, our tectonic definition of general dispersion of paleomagnetic poles of that age.
‘Gondwanaland,’ and its implied inheritance from the latest
Neoproterozoic Rodinia fragmentation, is first order only.
Nonetheless, this reconstruction restores Gondwanaland to 5.14.5 Summary: Major Unresolved Issues
a configuration resembling that calculated by backrotating and Future Work
post-Pangaean seafloor spreading ridges by an entirely inde-
pendent method, predicated only on the assumption that late Although little-metamorphosed Precambrian rocks exist and
Neoproterozoic–early Cambrian TPW was multiple and rapid may hold primary magnetization with equal fidelity to unme-
relative to the rates of plate motion. We consider this proof tamorphosed Phanerozoic rocks, the ancient paleomagnetic
of concept that the Ediacaran–Cambrian multiple TPW database is severely underconstrained, both spatially and
temporally. We consider it equally likely that better pan- Alvarez W, et al. (1977) Upper Cretaceous–Paleocene magnetic stratigraphy at Gubbio,
continental paleomagnetic coverage of TPW intervals already Italy. Part 5: Type Section for Late Cretaceous–Paleocene geomagnetic reversal time
scale. Geological Society of America Bulletin 88(3): 383–388.
hypothesized will offer non-TPW alternatives for those events
Anderson DL (1982) Hotspots, polar wander, Mesozoic convection and the geoid.
and that more detailed geographic and temporal APW deter- Nature 297: 391–393.
minations will recognize many new apparent TPW events. Anderson DL (1990) Theory of the Earth. Boston, MA: Blackwell Scientific Publications,
For continental reconstructions of any age, quick and/or 366 pp.
repeated, oscillatory type II TPW will be resolved better by detailed Anderson DL (1994) Superplumes or supercontinents. Geology 22(1): 39–42.
Andrews DL, et al. (2006) Uncertainties in plate reconstructions relative to the hotspots;
magnetostratigraphy of thick, continuous-deposited volcanic or Pacific-hotspot rotations and uncertainties for the past 68 million years.
sedimentary successions. Relatively few such magnetostratigra- Geophysical Journal International 166(2): 939–951.
phies exist, even through Paleozoic time and Mesozoic time. Athavale RN, et al. (1963) Paleomagnetism of some Indian Rocks. Geophysical Journal
Geodynamic modeling of TPW should address the predicted of the Royal Astronomical Society 7(3): 304–313.
Athavale RN, et al. (1972) Paleomagnetism and age of Bhader and Rewa sandstones
effects of multiple and/or quick (type II) TPW events, as well as
from India. Geophysical Journal of the Royal Astronomical Society 28(5): 499–509.
longer-duration, large-magnitude (type I) TPW events. We have Beck ME and Housen BA (2003) Absolute velocity of North America during the
not yet been able to satisfactorily fathom the dynamics of visco- Mesozoic from paleomagnetic data. Tectonophysics 377(1–2): 33–54.
elastic relaxation by a highly viscosity-structured mantle, if those Besse J and Courtillot V (1991) Revised and synthetic apparent polar wander paths of
dynamics in fact have been addressed in the geophysical literature. the African, Eurasian, North-American and Indian Plates, and true polar wander
since 200 Ma. Journal of Geophysical Research-Solid Earth and Planets 96(B3):
Although mantle-driven TPW seems to predict (and even 4029–4050.
partly rely upon) dynamic topography, we understand that it is Besse J and Courtillot V (2002) Apparent and true polar wander and the geometry of the
not clear that any modern-day surface physiographic feature geomagnetic field over the last 200 Myr. Journal of Geophysical Research: Solid
is unambiguously isostatically uncompensated. We hope that Earth 107(B11).
Besse J and Courtillot V (2003) Apparent and true polar wander and the geometry of the
this paradox will trigger additional work and debate in the near
geomagnetic field over the last 200 Myr (vol. 107, art no 2300, 2002). Journal of
future. Geophysical Research: Solid Earth 108(B10).
If multiple, fast Ediacaran–Cambrian TPW events are not Biram J (1966) A. Wegener: The Origin of Continents and Oceans, 4th edn. New York,
borne out by future, detailed magnetostratigraphic investiga- NY: Dover.
tion, we will be fascinated to discover the otherwise incredible Boger SD, et al. (2001) Early Paleozoic tectonism within the east Antarctic craton: The
final suture between east and west Gondwana? Geology 29(5): 463–466.
misfortune explaining the distribution of the dispersed APW Brasier MD, et al. (1994) Multiple delta-13C excursions spanning the Cambrian
swaths that today rotate those hypothesized TPW events into explosion to the Botomian crisis in Siberia. Geology 22(5): 455–458.
coincidence with the same parameters that reconstruct Gond- Buffett B and Archer D (2004) Global inventory of methane clathrate: Sensitivity to
wanaland in the Mesozoic era. changes in the deep ocean. Earth and Planetary Science Letters 227(3–4): 185–199.
Camps P, et al. (2002) Comment on ‘Stability of the Earth with respect to the spin axis
Sequence-stratigraphic compilations for times of hypothe-
for the last 130 million years’ by J.A. Tarduno and A.Y. Smirnov (2001) [Earth and
sized TPW should show global quadrature, rather than global Planetary Science Letters 184: 549–553]. Earth and Planetary Science Letters
constructive synchroneity, when paleolatitudes of those 198(3–4): 529–532.
deposits are considered in a spinner diagram-derived global Celaya MA, et al. (1999) Climate-driven polar motion. Journal of Geophysical Research:
paleogeography. Solid Earth 104(B6): 12813–12829.
Chase CG and Sprowl DR (1983) The modern geoid and ancient plate boundaries. Earth
Physical and chemical oceanographic changes predicted by and Planetary Science Letters 62(3): 314–320.
TPW should be observed in the rock record synchronous with Collins AS and Pisarevsky SA (2005) Amalgamating eastern Gondwana: The evolution
hypothesized or demonstrated TPW of sufficient magnitude; of the Circum-Indian orogens. Earth-Science Reviews 71(3–4): 229–270.
and some of these predictions (e.g., atmospheric temperature Cottrell RD and Tarduno JA (2000) Late Cretaceous true polar wander: Not so fast.
Science 288: 2283a.
anomaly resulting from methane clathrate destabilization)
Cottrell RD and Tarduno JA (2003) A Late Cretaceous pole for the Pacific plate:
should be testable by independent methods, where the rock Implications for apparent and true polar wander and the drift of hotspots.
record is amenable. Tectonophysics 362(1–4): 321–333.
Darwin G (1877) On the influence of geological changes on the Earth’s axis of rotation.
Philosophical Transactions of the Royal Society of London 167: 271–312.
Acknowledgments Davies GF (1984) Lagging mantle convection, the geoid and mantle structure. Earth and
Planetary Science Letters 69(1): 187–194.
Dickens GR, et al. (1995) Dissociation of oceanic methane hydrate as a cause of the
The authors are grateful for comments and criticisms from carbon-isotope excursion at the end of the Paleocene. Paleoceanography 10(6):
Dennis Kent, Bob Kopp, Adam Maloof, Ross Mitchell, and 965–971.
Will Sager. Adam Maloof developed spinner diagrams in con- Dickens GR, et al. (1997) A blast of gas in the latest Paleocene: Simulating first-order
cert with TDR and DADE. Jean-Pascal Cogné’s PaleoMac© pro- effects of massive dissociation of oceanic methane hydrate. Geology 25(3):
259–262.
gram (downloaded with password permission from J-PC) Dickman SR (1979) Continental-drift and true polar wandering. Geophysical Journal of
readily produces spinner diagrams. This work is supported in the Royal Astronomical Society 57(1): 41–50.
part by a 3-year NSF Graduate Fellowship to TDR; NSF grants DiVenere V and Kent DV (1999) Are the Pacific and Indo-Atlantic hotspots fixed?
9807741, 9814608, and 9725577; and funds from the NASA Testing the plate circuit through Antarctica. Earth and Planetary Science Letters
170(1–2): 105–117.
National Astrobiology Institute to JLK and a David and Lucile
DuBois PM (1957) Comparison of palaeomagnetic results from selected rocks of Great
Packard Foundation Award to DADE. Britain and North America. Philosophical Magazine Supplement Advances in
Physics 6: 177.
Elkins-Tanton LT and Hager BH (2005) Giant meteoroid impacts can cause volcanism.
References Earth and Planetary Science Letters 239(3–4): 219–232.
Evans ME (1976) Test of dipolar nature of geomagnetic field throughout Phanerozoic
Alvarez W and Lowrie W (1978) Upper Cretaceous paleomagnetic stratigraphy at Moria time. Nature 262(5570): 676–677.
(Umbrian Apennines, Italy) – Verification of Gubbio section. Geophysical Journal of Evans DAD (2003) True polar wander and supercontinents. Tectonophysics 362(1–4):
the Royal Astronomical Society 55(1): 1–17. 303–320.
Evans ME (2005) Testing the geomagnetic dipole hypothesis: Palaeolatitudes sampled Kirschvink JL (1978) Precambrian–Cambrian boundary problem – Paleomagnetic
by large continents. Geophysical Journal International 161(2): 266–267. directions from Amadeus Basin, Central Australia. Earth and Planetary Science
Evans DAD (2006) Proterozoic low orbital obliquity and axial-dipolar geomagnetic field Letters 40(1): 91–100.
from evaporite palaeolatitudes. Nature 444(7115): 51–55. Kirschvink JL and Raub TD (2003) A methane fuse for the Cambrian explosion: Carbon
Evans DAD and Kirschvink JL (1999) Multiple episodes of rapid true polar wander in cycles and true polar wander. Comptes Rendus Geoscience 335(1): 65–78.
Neoproterozoic–Cambrian time? In: Paper Presented at Geological Society of Kirschvink JL, et al. (1991) The Precambrian–Cambrian boundary: Magnetostratigraphy
America Annual Meeting, Boulder, CO: Geological Society of America (GSA). and carbon isotopes resolve correlation problems between Siberia, Morocco, and
Evans DA, et al. (1998) Polar wander and the Cambrian. Science 279, pp. 9a–9e. South China. GSA Today 1: 69–91.
Evans DAD, et al. (2000) A high-quality mid-Neoproterozoic paleomagnetic pole from Kirschvink JL, et al. (1997) Evidence for a large-scale Early Cambrian reorganization of
south China, with implications for ice ages and the breakup configuration of continental masses by inertial interchange true polar wander. Science
Rodinia. Precambrian Research 100(1–3): 313–334. 277: 541–545.
Farley KA and Eltgroth SF (2003) An alternative age model for the Paleocene–Eocene Kirschvink JL, et al. (2005) Rapid inertial-interchange true polar wander during
thermal maximum using extraterrestrial He-3. Earth and Planetary Science Letters ediacaran time induced by intrusion of the Sept-Îles intrusive suite, Quebec,
208(3–4): 135–148. Canada: A possible trigger for massive sea-level and carbon isotopic excursions.
Fisher D (1974) Some more remarks on polar wandering. Journal of Geophysical In: Paper Presented at Gondwana XII, Mendoza, Argentina.
Research 79(26): 4041–4045. Klootwijk CT, et al. (1986a) Paleomagnetic constraints on formation of the Mianwali
Fujita M, et al. (2002) Oceanic torques on solid Earth and their effects on Earth rotation. reentrant, Trans-Indus and Western Salt-Range, Pakistan. Earth and Planetary
Journal of Geophysical Research, Solid Earth 107(B8): ECV 1-1–ETG 5-18. Science Letters 80(3–4): 394–414.
Funaki M (1984) Investigation of the Paleomagnetism of the basement-complex of Klootwijk CT, et al. (1986b) Rotational overthrusting of the northwestern Himalaya –
Wright Valley, southern Victoria Land, Antarctica. Journal of Geomagnetism and Further paleomagnetic evidence from the Riasi thrust sheet, Jammu foothills, India.
Geoelectricity 36(11): 529–563. Earth and Planetary Science Letters 80(3–4): 375–393.
Gold T (1955) Instability of the Earth’s axis of rotation. Nature 175: 526–529. Lanza R and Tonarini S (1998) Palaeomagnetic and geochronological results from the
Goldreich P and Toomre A (1969) Some remarks on polar wandering. Journal of Cambro-Ordovician granite harbour intrusives inland of Terra Nova Bay (Victoria
Geophysical Research 74(10): 2555–2567. Land, Antarctica). Geophysical Journal International 135(3): 1019–1027.
Gordon RG (1987) Polar wandering and paleomagnetism. Annual Review of Earth and Li ZX, et al. (2004) A 90 degrees spin on Rodinia: Possible causal links between the
Planetary Sciences 15: 567–593. Neoproterozoic supercontinent, superplume, true polar wander and low-latitude
Gordon RG (1995) Plate motions, crustal and lithospheric mobility, and glaciation. Earth and Planetary Science Letters 220(3–4): 409–421.
paleomagnetism: Prospective viewpoint. Journal of Geophysical Research: Solid Llanos MPI, Riccardi AC, and Singer SE (2006) Palaeomagnetic study of Lower
Earth 100(B12): 24367–24392. Jurassic marine strata from the Neuquén Basin, Argentina: A new Jurassic apparent
Gordon RG, et al. (1984) Paleomagnetic Euler Poles and the apparent polar wander and polar wander path for South America. Earth and Planetary Science Letters 252(3–4):
absolute motion of North-America since the Carboniferous. Tectonics 3(5): 379–397.
499–537. Magaritz M, et al. (1986) Carbon-isotope events across the Precambrian–Cambrian
Grunow AM (1995) Implications for Gondwana of new Ordovician Paleomagnetic data boundary on the Siberian platform. Nature 320: 258–259.
from igneous rocks in Southern Victoria Land, East Antarctica. Journal of Magaritz M, et al. (1991) The Precambrian–Cambrian boundary problem: Carbon
Geophysical Research: Solid Earth 100(B7): 12589–12603. isotope correlations for Vendian and Tommotian time between Siberia and
Grunow AM and Encarnacion JP (2000) Cambro-Ordovician palaeomagnetic and Morocco. Geology 19: 847–850.
geochronologic data from southern Victoria Land, Antarctica: Revision of the Maloof AC, et al. (2005) An expanded record of Early Cambrian carbon cycling from the
Gondwana apparent polar wander path. Geophysical Journal International 141(2): anti-atlas margin, Morocco. Canadian Journal of Earth Sciences 42(12):
391–400. 2195–2216.
Hager BH, et al. (1985) Lower mantle heterogeneity, dynamic topography and the geoid. Maloof AC, et al. (2006) Combined paleomagnetic, isotopic, and stratigraphic evidence
Nature 313(6003): 541–546. for true polar wander from the Neoproterozoic Akademikerbreen Group, Svalbard,
Halverson GP, et al. (2002) A major perturbation of the carbon cycle before the Ghaub Norway. Geological Society of America Bulletin 118(9–10): 1099–1124.
glaciation (Neoproterozoic) in Namibia: Prelude to snowball Earth? Geochemistry, Manzoni M and Nanni T (1977) Paleomagnetism of Ordovician lamprophyres from
Geophysics and Geosystems 3(6): 1–24. Taylor Valley, Victoria Land, Antarctica. Pure and Applied Geophysics 115(4):
Halverson GP, et al. (2005) Toward a Neoproterozoic composite carbon-isotope record. 961–977.
Geological Society of America Bulletin 117(9–10): 1181–1207. Marcano MC, et al. (1999) True polar wander during the Permo-Triassic. Journal of
Harrison CGA and Lindh T (1982) Comparison between the hot spot and geomagnetic- Geodynamics 28(2–3): 75–95.
field reference frames. Nature 300(5889): 251–252. Matsuyama I, et al. (2006) Rotational stability of dynamic planets with elastic
Higgins MD (2005) A new model for the structure of the Sept Iles Intrusive suite, lithospheres. Journal of Geophysical Research: Planets 111(E2): E02003.
Canada. Lithos 83: 199–213. McQueen DM, et al. (1972) Cambro-Ordovician paleomagnetic pole position and
Higgins JA and Schrag DP (2006) Beyond methane: Towards a theory for the rubidium–strontium total rock isochron for charnockitic rocks from Mirnyy Station,
Paleocene–Eocene thermal maximum. Earth and Planetary Science Letters East Antarctica. Earth and Planetary Science Letters 16(3): 433–438.
245(3–4): 523–537. Meert JG (1999) A paleomagnetic analysis of Cambrian true polar wander. Earth and
Hynes A (1990) 2-Stage rifting of Pangea by 2 different mechanisms. Geology 18(4): Planetary Science Letters 168(1–2): 131–144.
323–326. Mitrovica JX and Forte AM (2004) A new inference of mantle viscosity based upon joint
Irving E (1956) Palaeomagnetic and paleoclimatological aspects of polar wandering. inversion of convection and glacial isostatic adjustment data. Earth and Planetary
Geofisica Pura et Applicata 33: 23–41. Science Letters 225(1–2): 177–189.
Irving E (1957) Rock magnetism: A new approach to some palaeogeographic problems. Mitrovica JX, et al. (2001a) Glacial isostatic adjustment on a rotating Earth. Geophysical
Advances in Physics 6(22): 194–218. Journal International 147(3): 562–578.
Irving E (1988) Paleomagnetic confirmation of Continental Drift. EOS, Transactions of Mitrovica JX, et al. (2001b) Recent mass balance of polar ice sheets inferred from
American Geophysical Union 69(44): 994–1014. patterns of global sea-level change. Nature 409(6823): 1026–1029.
Irving E (2005) The role of latitude in mobilism debates. Proceedings of the National Mitrovica JX, et al. (2005) The rotational stability of an ice-age Earth. Geophysical
Academy of Sciences of the United States of America 102(6): 1821–1828. Journal International 161(2): 491–506.
Irving E and Irving GA (1982) Apparent polar wander paths carboniferous through Montanez IP, et al. (2000) Evolution of the Sr and C isotope composition of Cambrian
cenozoic and the assembly of Gondwana. Geophysical Surveys 5(2): 141–188. oceans. GSA Today 10(5): 1–7.
Jurdy DM and Van der Voo R (1974) Method for separation of true polar wander and Moreau M-G, et al. (2007) A new global Paleocene–Eocene apparent polar wandering
continental-drift, including results for last 55 my. Journal of Geophysical Research path loop by ‘stacking’ magnetostratigraphies: Correlations with high latitude
79(20): 2945–2952. climatic data. Earth and Planetary Science Letters. http://dx.doi.org/10.1016/j.
Jurdy DM and Van der Voo R (1975) True polar wander since early cretaceous. Science epsl.2007.05.025.
187(4182): 1193–1196. Mound JE and Mitrovica JX (1998) True polar wander as a mechanism for second-order
Kent DV and Smethurst MA (1998) Shallow bias of paleomagnetic inclinations in the sea-level variations. Science 279(5350): 534–537.
Paleozoic and Precambrian. Earth and Planetary Science Letters 160(3–4): Mound JE, et al. (1999) A sea-level test for inertial interchange true polar wander events.
391–402. Geophysical Journal International 136(3): F5–F10.
Mound JE, et al. (2001) Sea-level change and true polar wander during the Late Satolli S, Besse J, Speranza F, and Calamita F (2007) The 125–150 Ma high-resolution
Cretaceous. Geophysical Research Letters 28(10): 2057–2060. Apparent Polar Wander Path for Adria from magnetostratigraphic sections in Umbria-
Mound JE, et al. (2003) The equilibrium form of a rotating Earth with an elastic shell. Marche (Northern Apennines, Italy): Timing and duration of the global Jurassic–
Geophysical Journal International 152(1): 237–241. Cretaceous hairpin turn. Earth and Planetary Science Letters 257(1–2): 329–342.
Muller RD, et al. (1993) Revised plate motions relative to the hotspots from combined Schmidt PW and Williams GE (1996) Palaeomagnetism of the ejecta-bearing Bunyeroo
Atlantic and Indian-Ocean hotspot tracks. Geology 21(3): 275–278. Formation, Late Neoproterozoic, Adelaide fold belt, and the age of the Acraman
Munk WH and MacDonald GJF (1960) The Rotation of the Earth. Cambridge: impact. Earth and Planetary Science Letters 144(3–4): 347–357.
Cambridge University Press. Schrag DP, et al. (2002) On the initiation of a snowball Earth. Geochemistry,
Nakada M (2002) Polar wander caused by the Quaternary glacial cycles and fluid Love Geophysics, Geosystems 3(6): 1–14. http://dx.doi.org/10.1029/2001GC000219.
number. Earth and Planetary Science Letters 200(1–2): 159–166. Schult FR and Gordon RG (1984) Root mean-square velocities of the continents with
Nakiboglu SM and Lambeck K (1980) Deglaciation effects on the rotation of the Earth. respect to the hot spots since the Early Jurassic. Journal of Geophysical Research
Geophysical Journal of the Royal Astronomical Society 62(1): 49–58. 89(NB3): 1789–1800.
Nimmo F (2002) Why does Venus lack a magnetic field? Geology 30(11): 987–990. Seitz F and Schmidt M (2005) Atmospheric and oceanic contributions to Chandler
Nimmo F and Pappalardo RT (2006) Diapir-induced reorientation of Saturn’s moon wobble excitation determined by wavelet filtering. Journal of Geophysical Research:
Enceladus. Nature 441(7093): 614–616. Solid Earth 110, B11406.
Payne JL, et al. (2004) Large perturbations of the carbon cycle during recovery from the Sleep NH (1988) Gradual entrainment of a chemical layer at the base of the mantle by
end-Permian extinction. Science 305(5683): 506–509. overlying convection. Geophysical Journal: Oxford 95(3): 437–447.
Peltier WR (1998) Postglacial variations in the level of the sea: Implications for Soldati G, et al. (2001) The effect of global seismicity on the polar motion of a viscoelastic
climate dynamics and solid-Earth geophysics. Reviews of Geophysics 36(4): Earth. Journal of Geophysical Research: Solid Earth 106(B4): 6761–6767.
603–689. Spada G (1997) Why are earthquakes nudging the pole towards 140 degrees E?
Petronotis KE and Gordon RG (1999) A Maastrichtian palaeomagnetic pole for the Geophysical Research Letters 24(5): 539–542.
Pacific plate from a skewness analysis of marine magnetic anomaly 32. Geophysical Spada G and Boschi L (2006) Using the post-Widder formula to compute the
Journal International 139(1): 227–247. Earth’s viscoelastic Love numbers. Geophysical Journal International 166(1):
Phillips RJ, et al. (2001) Ancient geodynamics and global-scale hydrology on Mars. 309–321.
Science 291: 2587–2591. Spada G, et al. (2006) Glacial isostatic adjustment and relative sea-level changes: The
Piper JDA (2006) A approximate to 90 degrees Late Silurian–Early Devonian apparent role of lithospheric and upper mantle heterogeneities in a 3-D spherical Earth.
polar wander loop: The latest inertial interchange of planet Earth? Earth and Geophysical Journal International 165(2): 692–702.
Planetary Science Letters 250(1–2): 345–357. Stacey F (1992) Physics of the Earth, 3rd edn Brisbane, QLD: Brookfield, 513 pp.
Piper JDA and Zhang QR (1999) Palaeomagnetic study of Neoproterozoic glacial rocks Steinberger B and O’Connell RJ (1997) Changes of the Earth’s rotation axis owing to
of the Yangzi Block: Palaeolatitude and configuration of South China in the Late advection of mantle density heterogeneities. Nature 387(6629): 169–173.
Proterozoic supercontinent. Precambrian Research 94(1–2): 7–10. Steinberger B and O’Connell RJ (2002) The convective mantle flow signal in rates of
Pisarevsky SA, et al. (2003) Models of Rodinia assembly and fragmentation. true polar wander. In: Mitrovica JX and Vermeersen LLA (eds.) Geodynamics Series
In: Yoshida M, Windley BF, and Dasgupta S (eds.) Proterozoic East Gondwana: 29: Ice Sheets, Sea Level and the Dynamic Earth, pp. 233–256. Washington, DC:
Supercontinent Assembly and Breakup, pp. 35–55. London: Geological Society of American Geophysical Union.
London. Stevenson DJ (1983) Planetary magnetic fields. Reports on Progress in Physics 46(5):
Porco CC, et al. (2006) Cassini observes the active south pole of Enceladus. Science 555–620.
311(5766): 1393–1401. Strik G, et al. (2003) Palaeomagnetism of flood basalts in the Pilbara Craton,
Prevot M, et al. (2000) Evidence for a 20 degrees tilting of the Earth’s rotation western Australia: Late Archaean continental drift and the oldest known
axis 110 million years ago. Earth and Planetary Science Letters 179(3–4): reversal of the geomagnetic field. Journal of Geophysical Research: Solid Earth
517–528. 108(B12): 2551.
Ranalli G (1995) Rheology of the Earth. London: Chapman and Hall, 413 pp. Tarduno JA and Cottrell RD (1997) Paleomagnetic evidence for motion of the Hawaiian
Randall DE, et al. (2000) A new late Middle Cambrian paleomagnetic pole for the hotspot during formation of the Emperor seamounts. Earth and Planetary Science
Ellsworth Mountains, Antarctica. Journal of Geology 108(4): 403–425. Letters 153(3–4): 171–180.
Richards MA and Engebretson DC (1992) Large-scale mantle convection and the Tarduno JA and Gee J (1995) Large-scale motion between Pacific and Atlantic hotspots.
history of subduction. Nature 355(6359): 437–440. Nature 378(6556): 477–480.
Richards MA, et al. (1997) An explanation for Earth’s long-term rotational stability. Tarduno JA and Smirnov AV (2001) Stability of the Earth with respect to the spin
Science 275(5298): 372–375. axis for the last 130 million years. Earth and Planetary Science Letters 184(2):
Riisager P, Riisager J, Abrahamsen N, and Waagskin R (2002) New Paleomagnetic pole 549–553.
and magnetostratigraphy of Faroe Islands flood volcanics, North Atlantic igneous Tarduno JA and Smirnov AV (2002) Response to comment on ‘Stability of the Earth with
province. Earth and Planetary Science Letters 201(2): 261–276. respect to the spin axis for the last 130 Million Years’ by P. Camps, M. Prevot, M.
Riisager P, et al. (2003) Paleomagnetic paleolatitude of Early Cretaceous Ontong Java Daignieres, and P. Machetel. Earth and Planetary Science Letters 198(3–4):
Plateau basalts: Implications for Pacific apparent and true polar wander. Earth and 533–539.
Planetary Science Letters 208(3–4): 235–252. Tarduno JA, et al. (2003) The Emperor Seamounts: Southward motion of the Hawaiian
Runcorn SK (1955) The Earth’s magnetism. Scientific American 193(3): 152–154. hotspot plume in Earth’s mantle. Science 301(5636): 1064–1069.
Runcorn SK (1956) Palaeomagnetic comparisons between Europe and North America. Torsvik TH, et al. (1998) Polar wander and the Cambrian. Science 279(9) (Correction,
Proceedings of the Geological Association of Canada 8(1): 77–85. p. 307).
Sabadini R (2002) Paleoclimate – Ice sheet collapse and sea level change. Science Torsvik TH, et al. (2002) Relative hotspot motions versus true polar wander. Earth and
295(5564): 2376–2377. Planetary Science Letters 202(2): 185–200.
Sabadini R and Peltier WR (1981) Pleistocene deglaciation and the Earth’s rotation – Tsai VC and Stevenson DJ (2007) Theoretical constraints on true polar wander.
Implications for mantle viscosity. Geophysical Journal of the Royal Astronomical Journal of Geophysical Research 112: B05415. http://dx.doi.org/
Society 66(3): 553–578. 10.1029/2005JB003923.
Sabadini R, et al. (1990) Eustatic sea level fluctuations induced by polar wander. Nature Valeriano CM, et al. (2004) U–Pb geochronology of the southern Brasilia belt
345: 708–710. (SE-Brazil): Sedimentary provenance, Neoproterozoic orogeny and assembly of west
Sabadini R, et al. (2002) Ice mass loss in Antarctica and stiff lower mantle viscosity Gondwana. Precambrian Research 130: 27–55.
inferred from the long wavelength time dependent gravity field. Geophysical Van der Voo R (1994) True polar wander during the Middle Paleozoic. Earth and
Research Letters 29(10): 1373. Planetary Science Letters 122(1–2): 239–243.
Sager WW (2006) Cretaceous paleomagnetic apparent polar wander path for the pacific Van Fossen MC and Kent DV (1992) Paleomagnetism of 122 Ma plutons in New
plate calculated from Deep Sea Drilling Project and Ocean Drilling Program basalt England and the Mid-Cretaceous paleomagnetic field in North America: True polar
cores. Physics of the Earth and Planetary Interiors 156(3–4): 329–349. wander or large-scale differential mantle motion? Journal of Geophysical Research
Sager WW and Koppers AAP (2000a) Late Cretaceous polar wander of the Pacific plate: 97: 19651–19661.
Evidence of a rapid true polar wander event. Science 287(5452): 455–459. Veevers JJ (2004) Gondwanaland from 650–500 Ma assembly through 320 Ma merger
Sager WW and Koppers AAP (2000b) Late Cretaceous true polar wander: Not so fast. in Pangea to 185–100 Ma breakup: Supercontinental tectonics via stratigraphy and
Science 288(5452): 2283a. radiometric dating. Earth-Science Reviews 68(1–2): 1–132.
Vermeersen LLA and Sabadini R (1999) Polar wander, sea-level variations and ice age Wegener A (1929) Die Entstehung der Kontinente und Ozeane. Vierte gaenzlich umgearbeitete
cycles. Surveys in Geophysics 20(5): 415–440. Auflage. In: Braunschweig, Germany: Druck and Verlag von Fridrich Vieweg und Sohn.
Ward PD, et al. (2005) Abrupt and gradual extinction among land vertebrates in the Zhong SJ and Hager BH (2003) Entrainment of a dense layer by thermal plumes.
Karoo basin, South Africa. Science 307: 709–714. Geophysical Journal International 154(3): 666–676.
Watts DR and Bramall AM (1981) Paleomagnetic evidence from the Ellsworth Zijderveld JD (1968) Natural remanent magnetizations of some intrusive rocks from Sor
Mountains supports microplate nature of western Antarctica. Geophysical Journal of Rondane Mountains Queen Maud Land Antarctica. Journal of Geophysical Research
the Royal Astronomical Society 65(1): 271. 73(12): 3773–3785.

Geomagnetism: An Introduction and Overview: M Kono, Tokyo Institute of Technology, Tokyo, Japan

Uploaded by

Copyright:

Available Formats

Geomagnetism: An Introduction and Overview: M Kono, Tokyo Institute of Technology, Tokyo, Japan

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Geomagnetism: An Introduction and Overview: M Kono, Tokyo Institute of Technology, Tokyo, Japan

Uploaded by

Copyright:

Available Formats

5.

01 Geomagnetism: An Introduction and Overview

5.01.1 Geomagnetism in Perspective 1

Treatise on Geophysics, Second Edition http://dx.doi.org/10.1016/B978-0-444-53802-4.00095-6 1

This is the earliest written record about the magnetic com-

若遇天景噎霾. 夜色瞑黒. 又不能辨方向. 則當縦老馬前行. 令識道

After that, how to make this fish is described.

置炭火中焼之. 候通赤. 以鐡鈐. 鈐魚首出火. 以尾正對子位. 蘸水盆

以芥子許蝋綴於針腰. 無風處懸之. 則針常指南. 以木刻龜子一个. 一如前法製造. 但於尾邊敲針入去. 用小板子. 上安

Nature of Things) written about 1190, he described the use of

5.01.1.1.4 Epistola of Petrus Peregrinus

Figure 6 Experiment of cutting lodestone to show the appearance of

to the east of true north, which shows that the declination in

Figure 8 The title page of De Magnete.

5.01.1.1.7 De Magnete of William Gilbert

5.01.1.2.3 Magnetic charts

W 60 30 N 30 60 E 120 150 S 150 120 W

Fushun Other sites

K-Ar age (106 years)

APRIL 11, 1964 McDOUGALL and TARLING

JUNE 26, 1964 COX, DOELL, and DALRYMPLE

JAN. 15, 1966 DOELL, DALRYMPLE, and COX

MAY 20, 1966 DOELL and DALRYMPLE

JUNE 18, 1966 CHAMALAUN and MCDOUGALL

JULY 1966 MCDOUGALL and WENSINK

SEPT. 1966 COX, HOPKINS, and DALRYMPLE

DEC. 15, 1966 MCDOUGALL , ALLSOPP, and CHAMALAUN

DEC. 24, 1966 MC DOUGALL and CHAMALAUN

MAY 15, 1967 COX and DALRYMPLE

MAY 1967 COX and DALRYMPLE

OCT. 31, 1988 COX, DOELL, and DALRYMPLE

JAN. 1969 COX

5.01.1.3.3 Seafloor spreading Juan de Fuca Ridge 46⬚ N

Chron 5 6 13 18 21 25 31 34 M0 M4M10N M16 M21 M25

Rn, (nT)2, Core–mantle boundary

Core: Rn = 1.349 × 109 (0.270)n (nT)2

where l is the hyperparameter indicating the relative impor-

radiation). With the advent of the space age, direct measure-

View from the Sun

Yearly averaged sunspot numbers 1610–2010

5.01.1.7.4 Excursions, reversals, and polar wander 5.01.1.8 Conclusions

the Earth’s magnetism as one of the important problems of References

Treatise on Geophysics, Second Edition http://dx.doi.org/10.1016/B978-0-444-53802-4.00096-8 33

where f and fs are the (geographic) longitudes (in degrees) of

Preparation for Ørsted

Satellite Operation Inclination Altitude Data

OGO-2 Oct 1965–Sep 1967 87 410–1510 km Scalar only

GPS POD antenna GPS limb sounding

Scalar Overhauser magnetometer (OVH)

8 m deployable boom Optical bench (vector field Absolute scalar

Electric field Thermal ion imager

5.02.2.4.2 Dst, an index of magnetospheric ring-current

Beggan et al., 2013). [17]

where the index R of the Gauss coefficients is here to recall 2n2 Xn  2  2

MF1 MF3 MF5 MF7

-200 -150 -100 -50 0 50 100 150 200

Core + Ind + Ext

Core + Ind + Ext

CM4 Langel and Estes (1985)

OGO-2 Oct 1965–Sep 1967 87 410–1510 km Scalar only

where the index R of the Gauss coefficients is here to recall 2n2 Xn 2 2

tðlÞ ¼ t1 lg [11]