Fundamentals of Weather and Climate

You have spread out the heavens like a tent

and built your home on the waters above.
You use the clouds as your chariot
and ride on the wings of the wind.
You use the winds as your messengers
and flashes of lightning as your servants.

PSALM 104, vv. 2-4

Good News Bible
Fundamentals of
Weather and Climate
ROBIN MciLVEEN
Environmental Science Division
Institute of Environmental and Biological Sciences
Lancaster University

[0 11[ SPRINGER-SCIENCE+llUSINESS MEDIA, B,V,

First published in 1986 as Basic Meteorology: A physical outline
This edition 1992
© 1986, 1992 J.F.R McIlveen
Originally published by Chapman & Hall in 1992
Typeset in lOIII pt Times Roman by Colset Private Ltd, Singapore
ISBN 978-0-442-31476-7 ISBN 978-1-4899-6892-0 (eBook)
DOI 10.1007/978-1-4899-6892-0
Apart from any fair dealing for the purposes of research or private study,
or criticism or review, as permitted under the UK Copyright Designs and
Patents Act, 1988, this publication may not be reproduced, stored, or
transmitted, in any form or by any means, without the prior permission in
writing of the publishers, or in the case of reprographic reproduction only
in accordance with the terms of the licences issued by the Copyright
Licensing Agency in the UK, or in accordance with the terms of licences
issued by the appropriate Reproduction Rights Organization outside the
UK. Enquiries concerning reproduction outside the terms stated here should
be sent to the publishers at the London address printed on this page.
The publisher makes no representation, express or implied, with regard to
the accuracy of the information contained in this book and cannot accept
any legal responsibility or liability for any errors or omissions that may be
made.
A catalogue record for this book is available from the British Library

Library of Congress Cataloging-in-Publication data

Mcilveen, J.F.R.
Fundamentals of weather and climate/Robin Mcilveen.
p. em.
Includes bibliographical references and index.
ISBN 978-0-442-31476-7
I. Weather-Juvenile literature. 2. Climate-Juvenile literature.
I. Title.
QC981.3.M43 1991
551.5-dc20 91-18238
C!P
AC
Contents

Preface to this edition X

Preface to the first edition xii

Acknowledgements XV

Useful information xvii

1 Introduction
1.1 The film of gas 1
1.2 The stratified atmosphere 3
1.3 The disturbed atmosphere 5
1.4 Physical laws 7
1.5 Cause and effect 8
1.6 The Sun and the atmospheric engine 9
1.7 Determinism and chaos 10
Problems 13

2 Observations 15
2.1 Care in observation 15
2.2 Temperature measurement 16
2.3 Humidity measurement 18
2.4 Wind measurement 20
2.5 Pressure measurement 21
2.6 Precipitation measurement 23
2.7 Other surface observations 26
2.8 Observation networks 28
2.9 The surface network 30
2.10 The upper air network 33
2.11 The satellite network 34
2.12 Epilogue 38
Appendix 2.1 Instrumental response times 38
2.2 The Beaufort scale 40 v
Contents 2.3 International cloud classification 40
2.4 Present and past weather 42
Problems 44
3 The constitution of the atmosphere 47
3.1 The well-stirred atmosphere 47
3.2 Dry air 51
3.3 Ozone 54
3.4 Carbon dioxide 57
3.5 Sulphur dioxide 59
3.6 Aerosol 61
3.7 Water 62
3.8 The evolution of the atmosphere 67
Appendix 3.1 Residence time 70
Problems 70
4 State and climate 73
4.1 The equation of state 73
4.2 The vertical profile of temperature 74
4.3 The vertical profile of pressure 78
4.4 Hydrostatic equilibrium 79
4.5 The vertical profile of air density 81
4.6 Equator to poles 82
4.7 The general circulation 86
4.8 Weather and climate 90
4.9 Climate change 92
4.10 Human society and climate change 97
Appendix 4.1 Ideal gases 101
Appendix 4.2 The hydrostatic equation 102
Appendix 4.3 Geopotential height 103
Appendix 4.4 Exponential decay 104
Appendix 4.5 Motion in a circle 106
Problems 107
5 Atmospheric thermodynamics 109
5.1 Introduction 109
5.2 The first law of thermodynamics 109
5.3 Isobaric heating and cooling 111
5.4 Measures of water vapour 112
5.5 Measurements of water vapour 114
5.6 Adiabatic reference processes 116
5. 7 The dry adiabatic reference process and potential temperature 117
5.8 Thermodynamic diagrams 118
5.9 The dry adiabatic lapse rate 121
5.10 The saturated adiabatic reference process 122
5.11 Equivalent and wet-bulb potential temperatures 125
5.12 Convective stability 127
5.13 The maintenance of near-neutral stability 130
Appendix 5.1 The first law of thermodynamics (meteorological form) 133
Appendix 5.2 Precipitable water content 134
Appendix 5.3 A skeleton tephigram 134
Appendix 5.4 Equivalent potential temperature 136
VI Problems 137
6 Cloud and precipitation 140 Contents

6.1 Cloud 140

6.2 Saturation 141
6.3 The sub-cloud layer 142
6.4 The well-mixed sub-cloud layer 144
6.5 The dew-point profile 146
6.6 Condensation observed 147
6. 7 Condensation modelled 150
6.8 Cloud droplet growth 153
6.9 Cloud development 154
6.10 Precipitation 157
6.11 Precipitation modelled 160
6.12 Atmospheric electricity 164
6.13 Practical applications 167
Appendix 6.1 Dew-point lapse rate 172
Appendix 6.2 Droplet growth by diffusion and condensation 173
Appendix 6.3 Terminal velocities of falling drops 174
Appendix 6.4 Evaporation of falling drops 175
Appendix 6.5 Growth by collision and coalescence 177
Problems 178

7 Atmospheric dynamics 181

7.1 The equation of motion 181
7.2 Forces 182
7.3 Relative and absolute accelerations 186
7.4 The Earth's rotation and apparent g 188
7.5 The Coriolis effect 189
7.6 The equations of motion 190
7. 7 Synoptic-scale motion 191
7.8 Geostrophic flow 195
7.9 Why geostrophic? 198
7.10 Thermal winds 200
7.11 Geostrophic departures 203
7.12 Small-scale motion 207
7.13 Compressing and deforming air 212
7.14 The dynamics of rotation 219
Appendix 7.1 Dimensions 227
Appendix 7.2 Pressure gradient force 229
Appendix 7.3 Viscous force 230
Appendix 7.4 Centrifugal pressure gradient 230
Appendix 7.5 Curvature terms 231
Appendix 7.6 The Coriolis effect 232
Appendix 7. 7 Geostrophic flow 233
Appendix 7.8 Pressure gradient and contour slope 234
Appendix 7.9 Thermal wind 235
Appendix 7.10 Gradient balance in anticyclones 236
Appendix 7.11 Buoyancy 237
Appendix 7.12 Continuity of mass 238
Appendix 7.13 Compressibility and divergence 239
Problems 240

vii
Contents 8 Radiation and global climate 244
8.1 The solar spectrum 244
8.2 The Earth's energy balance 246
8.3 Terrestrial radiation and the atmosphere 248
8.4 The terrestial radiation cascade 251
8.5 The greenhouse effect 253
8.6 Average radiant-energy budget 255
8.7 Meridional distribution of radiative fluxes 259
8.8 Seasonal variations of radiative fluxes 261
8.9 Diurnal variations of radiative fluxes 263
8.10 The necessity of air motion 264
8.11 Convective heat fluxes 267
8.12 Advective heat fluxes 269
8.13 Weather and ocean systems convecting and advecting heat 270
8.14 Wind speeds and heat fluxes 274
Appendix 8.1 Blackbody radiation 275
Appendix 8.2 Surface irradiance 278
Appendix 8.3 Meridional advection profiles 279
Appendix 8.4 Convective and advective heat fluxes 280
Appendix 8.5 Radiative forcing 281
Problems 282

9 Surface and boundary layer 286

9.1 Introduction 286
9.2 Surface shape and radiation 289
9.3 Surface heat input and output 292
9.4 Surface thermal response 294
9.5 Partitioning between surface and air 297
9.6 The laminar boundary layer 299
9.7 Turbulence 301
9.8 The origins and role of turbulence 303
9.9 The surface boundary layer 306
9.10 Above the surface boundary layer 313
9.11 Airflow over uneven surfaces 315
9.12 Surface microclimate 320
Appendix 9.1 Direct sunlight on projections 324
Appendix 9.2 Daily insolation 325
Appendix 9.3 Sunlight in water 326
Appendix 9.4 Turbulent momentum transport 326
Appendix 9.5 The log wind profile 327
Problems 327

10 Smaller-scale weather systems 330

10.1 Introduction 330
10.2 Cumulus 331
10.3 Cumulus development 335
10.4 Cumulonimbus 337
10.5 Clouds and their environment 342
10.6 Severe local storms 346
10.7 Convective systems 353
10.8 Atmospheric waves 358
viii
 Appendix 10.1 Evaporative loss of buoyancy 364 Contents
Appendix 10.2 Convective overshoot 366
Appendix 10.3 Cumulonimbus tops 367
Appendix 10.4 Oscillations in a stable atmosphere 368
Problems 370
11 Large-scale weather systems in mid-latitudes 372
11.1 Historical 372
11.2 Extratropical cyclones 374
11.3 The mature depression 380
11.4 Three-dimensional airflow 388
11. 5Anticyclones 392
11.6 Waves in the westerlies 395
11.7 Polar lows and heat lows 400
11.8 Middle-latitude climates 404
Appendix 11.1 Uplift and rainfall 408
Appendix 11.2 Variation off with latitude 409
Appendix 11.3 Radial winds and heat lows 410
12 Large-scale weather systems in low latitudes 412
12.1 Subtropical anticyclones 412
12.2 Monsoons 417
12.3 Tropical weather systems 423
12.4 Tropical cyclones 421
12.5 Low-latitude climates 432
Appendix 12.1 Drying by descent 435
Appendix 12.2 Convergence and conservation of angular momentum 436
Appendix 12.3 Thermal wind in cyclostrophic balance 436
Appendix 12.4 Low pressure and warm core 437
Problems 438
13 The atmospheric engine 440
13.1 Introduction 440
13.2 Forms of energy 442
13.3 Energy paths 448
13.4 Efficiency of the atmospheric engine 453
13.5 Epilogue 455
Appendix 13.1 The energy equation 456
Appendix 13.2 Potential and internal energies of an air column 457
Appendix 13.3 Available potential energy 459
Appendix 13.4 Frictional dissipation 460
Appendix 13.5 Human energy production 461
Appendix 13.6 Optimum mechanical efficiency of a heat engine 461
Problems 462

Bibliography, references and other sources 464

Glossary 469
Notes on selected problems 481
Appendix 486
Index 488
ix
Preface to this edition

I have taken the opportunity of a second edition to revise the original text from
beginning to end. There were some errors of course, and I hope I have spotted
most; and there were sections of text which were unclear when I read them again in
the cool light of reappraisal. And of course there were sections which cried out for
amendment (mainly extension) because of international and personal changes in
perceived importance. Once I had begun to consider these, I became increasingly
aware that the title of the book, and of many of its sections, reflected a traditional
perception of the scope of meteorology somewhat at odds with the actual contents,
and overtaken by the pace of work on climate, climate change and human inter-
actions with the atmosphere. Accordingly I have retitled the book and many of its
sections. The underlying rationale of the book (to provide a mechanistic introduc-
tion to the physical behaviour of the lower atmosphere) is unaltered, having been
justified by positive reaction to the first edition.
Alterations include the addition of separate sections on the concept of chaos,
stratospheric ozone, the long-term evolution of the atmosphere, human society
and climatic change, diurnal variations of radiative heat fluxes, and summaries of
mid-latitude of low-latitude climates. The particular problem of anthropogenic
carbon dioxide and its effect on the atmospheric greenhouse is treated in several
sections now, as befits its profound significance for the future of our planet.
Previously separate chapters on radiation and global climate are amalgamated,
and the large-scale weather systems in low and middle latitudes are now given
separate chapters. To help instructors with students still using the first edition,
there is a final appendix listing section equivalences and major errata. Smaller
changes are a simplified derivation of the Coriolis acceleration, and a new
appendix on geopotential height. Associated with all of this is the appearance of 40
new photographs and diagrams, all in addition to those used in the first edition. I
have tried to include more examples from outside my immediate maritime, mid-
latitude experience, but hope that the inevitable residual bias will be as interesting
to those who live in other climates as their accounts are to me.
I am very grateful to those who have provided valuable suggestions and
comments on the first edition, many of which have been assimilated in the second;
to the publishers who have tactfully galvanized me toward completion; to my
cartographer Jane Rushton who has once again translated impressionistic scribbles
into legible diagrams; and to my wife and daughters Annwen and Heather who
X cheerfully combed the text for all the alterations to figure and section numbers. I
hope the text continues to be interesting and useful, and will gladly receive Preface to this edition
comments from readers.
In the time between starting on the first edition (roughly 1982) and the present,
public perception of aspects of meteorology and climatology has changed
remarkably. To all of us involved in the study of our magnificent, and yet
bewildering atmosphere, it is exciting to see our subject centre stage. It is also a
little worrying, since however little else we know, we know how uneasily its
stubborn subtleties are likely to fit into the simplistic schemes of politicians and the
media. I will be well content with my several thousand hours of effort if I have
helped more people to appreciate the simplicities and the subtleties of our
atmosphere, and to begin to distinguish between them.

J.F.R. Mcilveen
Lancaster University

xi
Preface to the first
edition

I have written this book for undergraduates and serious amateurs seeking introduc-
tory information about the behaviour of the lower atmosphere (especially the
troposphere), and an outline of the physical mechanisms involved. There are many
good texts which outline behaviour without giving any coherent account of
mechanism; and there are several good texts which outline the theoretical frame-
work of meteorology, but assume the reader to be fairly familiar already with the
atmospheric behaviour being discussed. A few books bridge the gap, but I sense a
tendency in them to concentrate on the many essential nuts and bolts at the expense
of a more connected view of the vast and detailed atmospheric activity which daily
surrounds us.
The reasons for the presence of such a gap in the literature become obvious as
soon as you try to fill it. I have found it in practice very difficult indeed to balance
the need to outline a minimum of actual behaviour against the need to offer some
accompanying discussion of mechanism, without causing mental indigestion in the
reader and writer, and without falling into undue emphasis on either description or
mechanism. I know that I have succeeded only patchily, but I hope that my aim in
the process will become sufficiently clear to the sympathetic reader that he or she
will be able to read a personally more suitable rendering between the lines. The deli-
berately discursive style is adopted for the same reason, for although it expands the
sheer volume of the text, it reveals the personal attitude of the author to the matters
in hand, in ways which can be useful to the new student, anxious for clues about the
relative importance of different parts of the total mass of information. Since I
have also tried to maintain an open style of discussion, minimizing the diplo-
matic niceties which can so easily blunt the impact of basic ideas, I know that I
have probably rushed into several places where more experienced angels fear to
tread. I have never done so wittingly, though a patient critic has already pointed
out that I may have erred in overplaying the concept of the highest opaque layer in
Chapter 8.
The inclusion of both familiarization and mechanistic understanding in an intro-
ductory text raises the danger of saying a little about everything and not enough
about any one thing. Whenever I have been conscious of this danger I have deli-
berately tried to err on the side of being selective, and this being the case it might be
useful to read this book in conjunction with one from the first of the three groups
mentioned above, although I have tried to make the book largely self-contained.
xii The balance and level of the text has risen largely, but not exclusively, from my
experience of teaching introductory meteorology to students in the Department of Preface to the first edition
Environmental Science of the University of Lancaster over a period of years. In
that context it covers the material presented over at least the first two years of a
three-year undergraduate course which covers a wide range of aspects of the
environment. In the context of a more specialized degree, it probably covers the
outline of the first year of meteorological studies, and provides a mechanistic core
for later years.
A significant complication when planning courses or texts in meteorology is that
people come to the subject with a wide variety of academic backgrounds. I have
assumed that readers are familiar with the terminology and some of the elementary
manipulations of the differential calculus, going little beyond the coverage of the
Ordinary Level syllabuses of the English General Certificate of Education. This
standard of background would not suffice in Physics, however, but rather than
assume competence to the Advanced Level GCE standard (which anyway is
seriously deficient in elements of classical physics, especially thermodynamics, and
has virtually no coverage of fluids), I have chosen to outline the necessary basic
relationships as they arise in the text. Sometimes a physical concept seems to me to
be so fundamental to the meteorology in hand that I have chosen to describe it in
some detail; and I have done the same thing when experience has taught me that a
particular matter is difficult to grasp (the Coriolis effect is a notorious case in
point). In other places I have included only a brief exposure in the body of the text,
reserving formal treatments to appendices. I have always tried to outline only the
essential basis of mechanism, and have been interested to note that this seems to
involve less formal detail in some regions (such as thermodynamics and radiation)
than other authors have felt necessary, and more in others (such as dynamics, ener-
getics and surface behaviour). I have deliberately avoided vectors, partly because
they are not widely encountered before first-year University level (at least not to a
useful extent), and partly because their very success in pulling dynamical formula-
tions into very neat packages can actually obscure the physical realities being
described. It would be a useful step towards more advanced studies to transcribe
the expressions in Chapter 7 (the dynamics chapter) into vector form.
I have taken considerable pains to assemble problems which will test the reader's
familiarity and understanding of each chapter. They are graded into three levels as
follows: the first level consists of short questions testing knowledge of concepts
and nomenclature, often at a very basic level. The second consists of questions
which test the application of concepts, usually in the form of equations, to simple
problems. Where the procedure is at all complex or subtle it is clearly cued. The
third level tests a more assured understanding of the material, either by setting
simple questions without clues about procedure, or by setting more complex or
subtle questions, or by seeking a discursive summary in the style of a short scientific
essay.
On looking over the text and comparing it with the aims stated initially, I feel
that I have probably been most successful in those parts where I have managed to
recapture (at least internally in the process of writing) the spirit of enthusiastic and
almost innocent inquiry which I encountered as a graduate student in the Depart-
ment of Meteorology at Imperial College London, particularly at the hands of Dr
J .S.A. Green and the late Professor F .H. Ludlam. I began to recall the latter's
infectious enthusiasm for clouds at several points in the text, and hope that it has
helped to integrate a subject which is too often left to specialist texts. Further back I
feel a debt, especially when engaged in the deceptively simple business of trying to
convey my sense of the logical way to approach complex reality, to two former
physics masters at the Royal Belfast Academical Institution, W.L. Brown and
H.F. Henderson, both dead now but remembered with affection and respect by xiii
Preface to the first edition two generations of more or less dedicated pupils.
I wish to thank Dr R.S. Harwood of Edinburgh University for urgently and care-
fully reading about half the text, and making many wise and useful comments.
Grateful thanks also go to Jane Rushton of the Department of Environmental
Science, for working with skill and enthusiasm on the numerous line drawings. I
am grateful to colleagues who tolerated a considerable degree of distraction during
the period of maximum effort, and am very grateful to my wife and family for
putting up with considerable domestic disruption over the same period, usually
with gentle forbearance. My wife also helped crucially with the considerable
editorial load which accumulates in the later stages of a work of this scale. Finally I
wish to say to the youngest members of my family that I am sorry I could not begin
'Once upon a time', but there is a kind of 'happily ever after' at the end.

XlV
Acknowledgements

I am very grateful to the following people and institutions for their kind permission
to use copyright material, and in many cases for the supply of prints, etc. Those
entries including NML were traced through the National Meteorological Library,
Meteorological Office, London Road, Bracknell, Berks. All photographs not
acknowledged below are by the author.

Fig. 1.1 National Remote Sensing Centre, Royal Aircraft Establishment, Farn-
borough, Hants.
Fig. 1.2 National Aeronautical and Space Administration, Washington D.C.
Figs. 1.3 and 12.10 Hong Kong Royal Observatory, Nathan Road, Kowloon.
Fig. 2.6 Meteorological Office, Bracknell, Berks. Crown Copyright.
Fig. 2.9 From 'The cyclone problem'-Professor F.H. Ludlam's inaugural
lecture, Imperial College London, 1966. With permission from Mrs Ludlam.
Figs 2.11 and 10.23 Vaisala Oy, Helsinki, Finland.
Figs. 2.12 and 2.13 P .E. Bayliss. Department of Electrical Engineering and Elec-
tronics, University of Dundee, and the Natural Environmental Research
Council.
Fig. 3.2 Copyright Dr J .E. Harries, Rutherford Appleton Laboratory, Chilton,
Didcot, Oxfordshire, UK.
Fig. 3.8 From Dr J.N. Cape. Copyright Institute of Terrestrial Ecology,
Penicuik, Midlothian, Scotland.
Fig. 3.10 Dr D.O. McGarvey, Department of Earth Sciences, The Open
University, Milton Keynes, Bucks.
Fig. 3.11 Dr D.A.R. Simmons, 21 Dougalston Avenue, Milngavie, Glasgow.
Fig. 3.13 Dr R.I. Jones, Institute of Environmental and Biological Sciences,
Lancaster.
Figs. 4.3 and 6.9 Copyright Robert Nutbrown, University of Warwick, UK.
Fig. 4.16 Chief Engineer, Lancaster City Council.
Fig. 5.3 Copyright BP Educational Service.
Figs. 5.5, 5.6, 5.7, 5.11, 6.6 Tephigrams after Metform 2810, Meteorological
Office, Bracknell, Berkshire. Crown Copyright.
Fig. 6.14 Photograph by Sergeant AI Roberts, Royal Air Force. Copyright
British Aerospace, Warton, Lancashire, UK.
Fig. 6.22 We have been unable to trace the owner of the copyright to this picture
but would welcome any information relating to this matter. XV
Acknowledgements Fig. 6.25 British Aerospace, Pilton, Bristol.
Fig. 6.26 Dr K.A. Browning, from Reports on Progress in Physics, 41 (1978),
The Institute of Physics, London.
Fig. 6.27 J .W.F. Goddard, Radio Communications Research Unit, Rutherford
Appleton Laboratory, Chilton, Didcot, Oxfordshire.
Fig. 7.4 Copyright Paul Hodgkins.
Fig. 7.10 John Bowman, Department of Environmental Science, University of
Lancaster.
Fig. 7.22 Dr R. Hide, Meteorological Office, Bracknell, Berks. From [45], the
Royal Meteorological Society, James Glaisher House, Grenville Place,
Bracknell, Berks.
Fig. 7.24 We have been unable to trace the owner of the copyright to this picture
but would welcome any information relating to this matter.
Figs. 8.10 and 11.12 Rod Stainer, New Zealand Meteorological Service,
Wellington, New Zealand.
Fig. 8.13 Captain J.R.C. Young, Brackenhill, Cotswold Road, Cumnor Hill,
Oxford, in Weather, 20: 18 (1965). The Royal Meteorological Society, as in
Fig. 7.22.
Fig. 9.10 P. Larsson and M. Beadle, Department of Environmental Science,
University of Lancaster, Lancaster, UK.
Fig. 9.11 Dr N.E. Holmes, 25-134, Crimea Road, Marsfield, New South Wales,
Australia.
Fig. 9.24 Doug Allan, Underwater Images, 19 Armscroft Road, Gloucester.
Fig. 9.31 Taken in 1938 by Fit Lieut. G.R. Shepley (killed in action 1940). The
Dean and Chapter of Salisbury Cathedral.
Fig. 10.14 Copyright NOAA, Rockville, MD 20852, USA.
Fig. 10.21 We have been unable to trace the owner of the copyright to this picture
but would welcome any information relating to this matter.
Fig. 10.22 R. Cunningham, AFGL, Hanscom, Mass. 01731, USA. From
Weather, 33: 62 (1978). The Royal Meteorological Society.
Fig. 10.24 D.J. Tritton in Physical Fluid Dynamics, Van Nostrand Reinhold
(UK) Ltd, Molly Millars Lane, Wokingham, Berks.
Fig. 10.30 Copyright B. Goddard, via Ross Reynolds, Reading University,
Berkshire, UK.
Fig. 10.31 As Fig. 2.12.
Fig. 10.32 As on the figure caption
Fig. 11.1 S. Petterssen (1956) Weather Analysis and Forecasting, Vol. I, Motion
and Motion Systems. McGraw-Hill Book Company, New York.
Fig. 11.7 As Fig. 2.12.
Fig. 11.9 R.K. Pilsbury, Nyetimber, Uplands Road, Totland, Isle of Wight.
Fig. 11.13 Chief Engineer, Lancaster City Council.
Fig. 11.26 P .G. Wickham, in [85], Her Majesty's Stationery Office, London.
Crown Copyright.
Fig. 12.1 Daily Weather Report of the Meteorological Office, 1200Z 4 August
1972. Crown Copyright.
Fig. 12.4 Copyright Professor Bruce Atkinson, Queen Mary Westfield College,
University of London, UK.
Fig. 12.10 As Fig. 1.3.
Fig. 12.12 Dr Neil Frank, National Hurricane Center, Coral Gables, Florida.

xvi
Useful information

Commonly used prefixes for powers of ten

Name Symbol
nano n 10-9
micro fJ- 10-6
milli m 10-3
centi c 10-2
kilo k 103
mega M 1()6
gig a G 109

Spherical properties (radius R)

Diameter 2R
Circumference 21rR
Maximum cross-section area 1rR 2
Surface area 47rR 2
Surface area between latitudes 30° 21rR 2
Volume j1rR3

xvii
Usefu I information
SI units and common equivalents

PRIMARY

Quantity Name Units Equivalence

mass kilogram kg
tonne 103 kg
length metre m
angstrom A 10-w m
nautical mile 1852 m
time second s
temperature kelvin K
Celsius oc same unit size but
=
roc (T + 273.2) K
quantity of
pure element
or compound mole mol (formerly called gram molecule)
angle radian rad
degree 0
0.017 45 rad
(360° = 21r rad)
electrical
current ampere A

SOME SECONDARY UNITS USED IN METEOROLOGY

Quantity Name Units Equivalence

frequency hertz Hz one cycle per second

volume cubic metre m3
litre IQ-3 mJ
electric
charge coulomb Cor As
specific mass kilogram per
kilogram
gram per kilogram g kg-l IQ-3
parts per million ppm IQ-6
velocity m per s m s- 1
knot (nautical
mile per hour) kt 0.514ms- 1
angular velocity rad s- 1 57.3° s- 1
xviii temperature gradient K m- 1 or oc m- 1
 Useful information
Quantity Name Units Equivalence
velocity gradient, divergence, vorticity
s-t
density kg per cubic m kg m- 3
acceleration m per s per s m s- 2
momentum kg m s- 1
force newton Norkgms- 2
pressure pascal Pa or N m- 2
millibar mbar or hPa 100 Pa
pressure
gradient Pam- 1
mbar per 100 km I0- 3 Pa m- 1
angular momentum
(or moment of momentum) kg m2 s- 1
moment of force Nm
energy joule J or N m
gram calorie cal 4.18 J
power watt W or J s- 1
energy flux
density W m-2
langley (cal per em) per minute
Ly min-t 697 W m - 2 (radiation)
heat capacity J K- 1 or J oc- 1
specific heat capacity J kg- 1 K-l
cal g- 1 oc- 1 4180 J kg- 1 K- 1
(See Appendix 7.1 for further treatment of units)

Universal constants

Speed of light in vacuo c 3.00 x 108 m s- 1

Planck constant h 6.626 x w- 34 1 s
Avogadro number N 6.022 x 1023 moi- 1
(molecules per mole)
Universal gas constant R. 8.314 J K- 1 mol- 1
Gravitational constant G 6.672 X w-tt N m2 kg-2
Stefan-Boltzmann constant a 5.670 X I0- 8 wm- 2 K- 4

xix
Useful information
Useful values in units normally used

SOLAR AND TERRESTRIAL

Solar diameter 1.37 x 106 km

Solar-terrestrial separation 150 (±5) x 1Q6 km
Effective blackbody temperature
of solar surface 5800K
Solar constant 1380 W m- 2
Sidereal day (period of terrestrial
rotation) 86 040 s (23.93 hours)
Angular velocity of terrestrial
rotation 7.303 X 10-s rad s- 1
Coriolis parameter 1.0 x 10- 4 rad s- 1 (or s- 1) at latitude
43.2°
Solar day (mean period of
apparent terrestrial rotation
relative to the Sun) 86 400 s (24 hours)
Year (period of terrestrial orbit
of Sun) 3.156 x 107 s (365.3 days)
Calendar year 3.154 x 107 s (365 days)
Angle between equatorial and
ecliptic planes 23.4° (latitudes of tropics)
Effective blackbody temperature
of planet Earth 255 K
Earth's planetary albedo in solar
wavelengths about 0.33
Terrestrial mean radius 6370 km
Average gravitational acceleration at
mean sea level 9.81 m s- 2
Terrestrial surface area 5.10 x 10 14 m2
Meridional length subtending
latitude degree 111.2 km or 60 nautical miles
Elevation of highest surface above
mean sea level (Mt Everest) 8848 m
Depression of lowest surface
below mean sea level (Dead Sea
surface) 392m

ATMOSPHERIC

Standard temperature 0 oc J Standard temperature

XX Standard atmospheric pressure 1013.2 mbar and pressure (STP)
Effective molecular weight of Useful information
dry air 29.0
Specific gas constant for dry air 287 J kg- 1 K- 1
Specific heat capacity for dry air
at constant volume 717 J kg- 1 K- 1
Specific heat capacity for dry air
at constant pressure 1004 J kg- 1 K- 1
Density of dry air at STP 1.292 kg m- 3
Decadal scale height of isothermal
dry atmosphere at -20 °C 17.1 km
Velocity of sound in dry air at 0 oc 331.5 m s- 1
Thermal conductivity of dry air
at 0 oc 2.43 x 10- 2 W m- 1 K- 1
Dynamic viscosity of dry air at 0 oc 1.71 X 10-s kg m- 1 s- 1
Saturated vapour at 0 oc
density 4.85 x 10-3 kg m-3
pressure 6.11 mbar
Specific latent heat of vaporization
of water at 0 oc 2.50 MJ kg- 1
Specific latent heat of fusion of
ice at 0 oc 0.334 MJ kg- 1
Density of water at 0 oc 1000 kg m- 3
Density of ice at 0 oc 917 kg m- 3
Density of mercury at 0 oc 13 595 kg m-3

Selection from International Civil Aviation

Organization standard temperate atmosphere

Height 1m Pressure/mbar Temperature/K Density/kg m- 3

0 1013.3 288.2 1.23
5 000 540.5 255.7 0.736
10 000 265.0 223.2 0.414
15 000 121.1 216.7 0.195
20 000 55.3 216.7 0.0889

xxi
Useful information
Typical magnitudes

Solar horizontal irradiance at

Earth's surface 0-1000 W m - 2 low latitudes and mid-
latitude summer
0-200 W m - 2 mid-latitude winter
Average tropospheric wind speed IO m s- 1
Wind speed in jet core 50 m s- 1
Height of jet core above sea level 9 km polar front jet stream
I2 km subtropical jet stream
Height of tropopause above sea
level 10 km high latitudes
15 km low latitudes
Height of freezing level above sea
level 4 km low latitudes
0-3 km mid latitudes
Depth of sub-cloud layer Ikm
Depth of precipitable water 30mm
Average annual precipitation IOOO mm
Specific humidity 10 g kg- 1 low troposphere
I g kg- 1 mid troposphere
Sea-level pressure gradient
associated with gale force wind
in mid latitudes 5 mbar per IOO km
Equivalent tilt of isobaric
surface 40 m per 100 km

WEATHER SYSTEMS

Surface
Type Breadth Lifetime wind Updraught Rainfall

cumulonimbus 10 km I hr I5ms- 1 5 m s- 1 10 mm hr- 1

extratropical
cyclone IOOO km I week I5 m s- 1 5 em s- 1 1 mm hr- 1
(background)
hurricane 500km I week 40 m s- 1 10m s- 1 40 mm hr- 1
(active annulus around the eye)

xxii
Introduction CHAPTER

1.1 The film of gas

From our earliest childhood we are aware of the skies above us, but daily
familiarity blunts this awareness so much in most of us that it requires an especially
striking sunset, or a particularly massive and ominous cloud overhead, to rekindle
interest. Even then, the bias naturally introduced by the relatively tiny size of the
human frame, and its usually surface-based viewpoint, can produce very
misleading impressions of the scale of atmosphere at the bottom of which we spend
almost all of our lives. In fact the Earth's atmosphere has two very different scales.
Horizontally the atmosphere is enormously larger than we can perceive from a
point on the Earth's surface, just as the earth itself is larger. The order of
magnitude (a term used throughout this book to describe magnitudes to the nearest
power of ten) of the maximum horizontal scale of the atmosphere is of course that
of the Earth itself, which is tens of thousands of kilometres, written - 10 000 km.
Photographs such as Fig. 1.1 allow us to appreciate this relatively vast scale much
more directly than was possible before the advent of the meteorological satellite in
the 1960s.
By contrast the vertical scale of the atmosphere is very much smaller than the
radius of the Earth. In Fig. 1.2 you can barely see the thin limb of air illuminated
on the horizon by the hidden sun. In fact on the scale of the Earth, the atmosphere
is the merest film of gas clinging to the surface by gravitational attraction. As we
shall see, distances which a person could walk in a few hours in the horizontal are a
significant fraction of the vertical depth of the atmosphere. Our impression of the
great height of a towering thunder cloud for example is greatly enhanced by our
awareness of the difficulty we would encounter in climbing the cloud if it were a
solid mountain. But although the atmosphere is really quite modest in vertical
scale, its influence on the conditions in which we live is enormous, as will be
detailed at many points throughout this book.
Unlike the ocean, the atmosphere has no definite upper surface. For example its
density falls continuously with increasing height, from values close to 1 kg m- 3 at
the surface, to values more than ten thousand million (10 10) times smaller in the
tenuous interplanetary gas found more than a few hundred kilometres above the
surface. Despite this apparent lack of definition, the vertical extent of the
atmosphere can be usefully quantified by a so-called scale height, which is the
 Introduction

Fig. 1.1 The face of the Earth centred on the equator at 0° longitude as seen in visible light by
the European Space Agency geostationary satellite Meteostot 2. The picture was token at 1130 Z
on 25 September 1983 and shows Iorge cloud patches in low latitudes associated with tropical
weather systems in and around the Intertropical Convergence Zone (see section 4.7) . Cloudy
vortices and streaks at higher latitudes ore associated with extrotropicol cyclones and their
fronts and jet streams, and ore especially extensive in the southern hemisphere on this
occasion .

2
height interval in which a particular property, such as air density, changes by a 1 .2 The stratified
atmosphere
particular numerical factor. Consider for example the case of atmospheric
pressure. It is observed that throughout the lowest 100 km of the Earth's
atmosphere the decadal scale height for pressure (the interval separating levels with
pressures differing by a factor of 10) remains close to 16 km. Thus pressure at
mean sea level (MSL) and at 16, 32 and 48 km above MSL, are about 1000, 100, 10
and 1 mbar respectively. Since the pressure is proportional to the weight of the
overlying atmosphere, it follows that about 900Jo of the weight of the atmosphere is
concentrated in the lowest 16 km- a layer whose depth is only about 0.2% of the
Earth's radius. This concentration is produced in the easily compressible
atmosphere because the air is squeezed against the Earth's surface by its own
weight. Partly because of this concentration, but more importantly because the
weather-giving cloudy activity of the atmosphere is almost entirely confined within
the same layer, this shallow layer constitutes the bulk of the lower atmosphere
discussed in this book.

1.2 The stratified atmosphere

Since much of the material of the atmosphere is squeezed into such a shallow layer
overlying the surface, the distributions of temperature, humidity and indeed of
almost every observable property are strongly anisotropic, in the sense that their
gross vertical and horizontal distributions are very different - vertical structure
being usually much more closely packed and vertical gradients correspondingly
larger. For example, away from the complicating close proximity of the surface,
temperature decreases with height at a typical rate of about 6°C km - 1 up to heights
of between 10 and 15 km above MSL, whereas the strongest extensive horizontal
temperature gradients (those associated with vigorous fronts in middle latitudes)
rarely exceed 0.05°C km - 1 • The result is that isop/eths (lines or surfaces of equal
quantity, such as isobars and isotherms) are usually nearly horizontal, and vertical
sections such as Fig. 4.6 usually have a horizontally layered or stratified
appearance.
Corresponding to this marked stratification of the structure of the atmosphere,
there is an equally marked stratification of its behaviour. Immediately overlying
the Earth's surface is a region dominated by turbulent interaction with the surface
-the planetary boundary layer. Its depth varies greatly with time and location,
and even with the particular atmospheric property under discussion, but is often
-500 m. By contrast, the rest of the atmosphere (about 95% of the total mass) is
much less directly influenced by the surface and is accordingly termed the free
atmosphere. The planetary boundary layer together with the first 10-15 km of
the free atmosphere (the larger value applying to lower latitudes) comprises the
troposphere, so-called after the Greek word for turning because it is the site of the
commotion and overturning associated with so much significant surface weather.
The active and cloudy troposphere is separated from the relatively quiet and cloud-
free stratosphere above by the nearly horizontal and often very sharply defined
tropopause. This book deals mainly with the troposphere.
The anisotropy of the atmosphere is well marked only on scales which are at least
not insignificant in comparison with the decadal scale height of the atmosphere.
On much smaller scales, for example - 1 m, the atmosphere does not seem to 3
 introduction

Fig. 1.2 The Earth's limb, photographed manually by Gemini astronauts in visible light from the
setting sun . The thin atmospheric Ioyer on the horizon is visible because of light scattered from
air molecules, dust and some cloud. Molecular scattering produced a blue cost in the original
colour photograph, while the others gave no colour bios (section 8.6). The clouds on the left ore
anvils spreading from the tops of shower clouds over the Andes.

4
'know' that it is so flattened by gravity and compressibility, and disturbances are 1 .3 The disturbed
atmosphere
much more isotropic. Such small scales are particularly important in the planetary
boundary layer, which is heavily influenced by a hierarchy of turbulent eddies
ranging in spatial scale from - 500 m to - 5 mm. Though the free atmosphere too
is turbulent, from many points of view it is not essentially so. In fact it is dominated
by much larger-scale disturbances, many of which share the flattened configura-
tion of the lower atmosphere to such an extent that vertical sections have to be
drawn with grossly expanded vertical scales. Though this practice is essential for
clarity, it is useful to recall when examining diagrams such as Fig. 4.7 that the
lower atmosphere and many of its common larger-scale disturbances are in fact
two orders of magnitude 'flatter' than conventionally depicted.

1.3 The disturbed atmosphere

The atmosphere, and particularly the troposphere, is the site of incessant

commotion. Cumulus clouds as big as the mountains they resemble (the name
means hill in Latin) erupt and dissolve in periods of tens of minutes as the
updraughts within them wax and wane. In middle latitudes vortices of continental
scale (Fig. 1.1) grow and decay in periods of a few days. During their lifetime they
are streaked with great bands of cloud which produce rain, snow and locally hail,
and are associated with large areas of depressed atmospheric pressure at the
surface, from which they take their popular title. The technical name for such
depressions is extratropical cyclones - the prefix extratropical serving to
distinguish them from the tropical cyclones which are a family of disturbances
confined to low latitutes. The most intense of the latter are the hurricanes(Fig. 1.3)
which seasonally haunt the Caribbean and other similar regions (where they are
known locally by other names), and which constitute the most energetic type of
tropospheric disturbance. Each of these weather systems contains and interacts
with activity on relatively smaller scales: all types of cyclone contain cumulus, and
cumulus contain large turbulent elements which give the characteristically knobbly
appearance to their tops and sides (Fig. 1.4), and which can make flying in them
uncomfortably or even dangerously bumpy.
Superficially it is a scene of huge chaos, but the existence of characteristic types
suggests an underlying order. There is method in the apparent madness which it is
the object of atmospheric science to identify and describe in the most simple and
concise manner. According to its etymology, meteorology (which comes from the
Greek for lofty) should be the study of everything above the Earth's surface, but in
practice it has come to mean the study of the physical rather than the chemical
nature of the lower atmosphere, especially the troposphere. And its meaning has
been narrowed further to focus on the dynamic, changeable aspects of this nature,
leaving the term climatology to cover the typical and average aspects. In fact most
significant physical structures and behaviour of the lower atmosphere have their
meterological and climatological aspects, which often overlap. And the confusion
is compounded by the relatively recent explosion of interest in climate change,
which of course focuses on the changeable nature of climate.
Curiosity is the mainspring of meteorology, as it is of every other branch of
science, but so much atmospheric activity affects man so vitally, occasionally even
disastrously, that for over a century meteorology has been financed with the aim of 5
 Introduction

Fig. 1.3 low latitudes centred on longitude 140°E, seen in visible light from the Japanese
Meteorological Agency's geostationary satellite GMS-3 at 0300 Z 16 October 1985. Among the
substantial areas of cloud there ore two which hove the dense swirl appearance typical of
severe tropical storms: Typhoon Dot lies east of the Philippines and Typhoon Cecil lies over
Vietnam.

Fig. 1.4 Several swelling

cumulus showing flat bases,
knobbly sides and tops , and
the mountainous bulk typical
of rapidly growing convective
clouds. Notice the 'silver
lining' produced by strong
forward scatter of sunlight,
and the cloud shadow cast on
haze nearer the camera.

6
routinely forecasting atmospheric behaviour, especially that affecting surface 1..4 Physical laws
conditions, for as little as six hours ahead of time. In this period forecasting has
developed into a complex and highly specialized technique practised by nationally
organized networks of observers and analysts cooperating internationally. In the
last few years forecasting of climate change has become very important too,
attracting international effort. Though forecasting as such is considered only
incidentally in this book, its bias towards those types of tropospheric commotion
which strongly affect human activity naturally pervades meteorology and is
reflected here. For such reasons in this book we will largely ignore the many
beautiful but complex optical phenomena such as the rainbow, and will be highly
selective in treatment of the otherwise very important chemical behaviour of the
atmosphere, though each is a large subject in its own right. Atmospheric chemistry
in particular has attracted increasing attention with the recent realization that
man's industry is capable of altering natural balances substantially for the worse.
Much of the weather-producing activity of the atmosphere can be arranged in a
spectrum according to scale, as in Fig. 1.5. There is a range of over eight orders of
magnitude in space scales and a range of time scales which is only slightly smaller.
Space and time scales are related so that big events happen relatively slowly while
small events happen quickly. In fact the ratio of space to time scales is - 1 m s -I
for all the items tabulated. This value is a measure of the intensity of the activity of
the atmosphere, and can be compared with the much larger value for the much ·
more violently disturbed visible surface of the Sun, where granulations (a type of
solar cumulus) with diameters - 100 km appear and disappear within minutes.
The list in Fig. 1.5 is necessarily incomplete; known types of lesser importance are
omitted for simplicity, and no doubt others are missing because they have yet to be
identified among the welter of observations.

1.4 Physical laws

We believe that the observed structure and behaviour of the atmosphere can be
related to the operation of physical and chemical laws, and the forecasting bias of

Fig. 1.5 Types of weather-

related atmospheric
~~ disturbance displayed
~E.~EASON CLIMATIC according to lifetime and
Earth radius ~ '--~ CHANGE
z.o~ ~~/~ maximum space scale. Each
105
z§ 'bo
, type is discussed at some
length later in the text. Note
~~((;~ Key that logarithmic scales (equal
ff. ~~ CONVECTION - small cumulus steps along an axis
_% ~((,~ to big showers corresponding to equal
o~ _:?" MESO (scale systems) - shower multiples of value) cover wide
;:;;
"' 10 1 Ci ~
,§:f .f .iff
~z
patterns, sea breezes etc
ranges of value very

-' 1cf 1 z ;J Jf!f

!:f. .;) ff
SYN(optic scale systems) -
extra tropical and tropical cyclones etc
compactly, but require careful
interpolation.
HEM(ispheric) - long waves in
~~ff tropospheric flow
-3~~ff
10 ~
mn hr dy wk yr

Life time 7
 Introduction meteorology directs attention to the former in particular. Fortunately there is no
reason to suspect the operation of any fundamental physical law not already well
established by the development of physical science during the last three centuries.
In fact most of the laws which play important roles in the lower atmosphere appear
at least in outline in school science courses: the laws of statics and dynamics
(complicated in expression but not in essence by the fluid nature of the atmo-
sphere), matter, heat and radiation. However, it is emphasized repeatedly in this
book that even when considering some quite specific feature of the atmosphere, no
single law suffices; instead there is an interaction of many, though fortunately it is
often possible to pick out a few laws which between them dominate most of what is
observed in any particular instance.
Consider for example the forces acting on a small volume of air, curiously but
conventionally entitled an air parcel. In the lower atmosphere there are three such
forces, arising respectively from pressure gradients, gravity and friction. At a given
time and place there is a resultant F of these forces which is related to the
acceleration a of an air parcel of mass M by a form of Newton's second law of
motion.
F
-=a
M (1.1)
Note that F and a are written in bold type to show that they have direction as well as
magnitude, i.e. that they are vector rather than scalar quantities like mass, which
have magnitude only. In different zones of the atmosphere, atmospheric motion
(represented here by the acceleration a) differs considerably in type. For example,
in the planetary boundary layer there are quite large frictional forces associated
with the relatively intense turbulence there, especially near the surface, whereas in
the free atmosphere such friction is relatively unimportant in comparison with
other forces and can often be neglected without serious error. Indeed much of the
progress made in this century by detailed application of relations such as eqn ( 1.1)
to the atmosphere has depended on judicious separation of primary and secondary
effects, since only then are the relations mathematically simple enough to be
solved. The skeleton theory outlined in this book contains a number of simple
examples of such separation, but in more advanced work the process is often
elaborate and subtle. Note in passing that the eqn (1.1) (often called the equation of
motion) is insufficient on its own to describe any physical situation in the atmos-
phere at all comprehensively; formal expressions of other laws such as the
conservation of energy and of mass must be combined with eqn (1.1) before a
definite solution is possible even in principle.

1.5 Cause and effect

It is difficult to avoid considering F and a in eqn (1.1) as cause and effect

respectively, probably because in homes, factories and laboratories we arrange
forces to achieve desired effects; but it is important to realize that the natural laws
which man has identified and labelled indicate neither cause nor effect but simply
relationship. For example, even if we were to say that a certain pressure gradient
force caused air to accelerate, the terminology would be misleading since the
8 resulting redistribution of air could be said with as much justification to cause a
change in the pressure gradient force. This is a quite general point, but it is 1.6 The Sun and the
particularly relevant to the largely self-activating lower atmosphere, so let us atmospheric engine
consider it a little further in that context.
The temptation to identify cause and effect is strong because it satisfies our
innate desire for order, but it is probably enhanced in meteorology by the nature of
much weather-giving activity. Above the planetary boundary layer the dominant
types of disturbance are in the cumulus and larger scales (Fig. 1.5), and are such
that individual examples rest uneasily between statistical uniformity and individual
uniqueness. In their combination of individuality, regularity and transience,
cumulus, thunderstorms, depressions, hurricanes etc. resemble living things, and
you wiii detect in forecasts and books (including this one) a tendency to treat them
as such. This does not matter so long as it is recognized merely as a device to
simplify description, but unfortunately it reinforces the tendency to identify
imaginary causes and effects. The spurious animism implied by a statement such as
'the weakening of the anticyclone over the British Isles is allowing a belt of rain to
move into north-west Ireland' is descriptively very convenient, but it is quite mis-
leading if it is taken to imply any wiii, purpose or even priority in the action. In fact
the weakening no more allows the advance than the latter causes the weakening; the
two events are related. Such relationships are very much the concern of
meteorology, but long experience has shown that it is positively unhelpful to search
for 'chicken and egg' amidst the interminable flux of atmospheric events.
However, there is a resounding exception to this rule in the conspicuous action of
the Sun as prime mover of all atmospheric activity, as we now briefly consider.

1. 6 The Sun and the atmospheric engine

It is obvious on any sunny day that the Sun has a direct and important influence on
the atmosphere. Solar energy, much of it in the form of visible sunlight, pours con-
tinually onto the Earth, affecting both the surface and the overlying air. In fact the
role of the Sun in determining the condition and behaviour of the Earth's
atmosphere is even greater than our experience of its warmth, or observation of the
growth of cumulus clouds on sunny mornings, might lead us to suppose. The atmo-
sphere is quite literally solar-powered, and the Sun can be considered as the prime
mover of all atmospheric activity, and as a cause of atmospheric effects in the
restricted sense that it influences our atmosphere but is not itself significantly
influenced in return.
Sir Napier Shaw, a distinguished pioneer of modern meteorology, wrote in the
early years of the twentieth century that 'the weather is a series of incidents in the
working of a vast natural engine'. As in man-made heat engines, heat is taken in at
a heat source, and exhausted at a heat sink (Fig. 1.6) and mechanical energy
is generated, i.e. massive bodies are made to move. The source is provided by
the absorption of solar energy, mainly at the Earth's surface, and the sink by the
emission of infra-red terrestrial radiation to space, mainly from the upper tropo-
sphere. The motion and unstable distribution of the atmospheric mass is the
mechanical energy produced. Unlike man-made engines however, the atmospheric
engine is not constrained by a rigidly pre-ordained structure of turbines, gear trains
etc.: it generates and constantly maintains its own structure, which includes the
hierarchy of disturbances associated with surface weather. The atmospheric engine 9
 Introduction also controls its intake of solar energy by producing cloud masses which reflect a
very significant proportion of sunlight to space, and rather less obviously regulates
the output of terrestrial radiation to space. Between them these two effects act like
a governor which keeps the activity of the engine close to its normal level.
Obviously the behaviour of an engine which continually regenerates and regulates
itself is much more complex and subtle than that of man-made engines, but there is
no difference in underlying principle. The description of the atmosphere as a heat
engine is detailed at many points throughout this book, and is discussed at length in
the final chapter, but for the moment it is enough to accept that the sun is its
furnace.

Fig. 1.6 Schematic diagram Solar Terrestrial

heat heat
of the atmospheric heat Heat Heat
engine with heat source, sink source sink
in out
and mechanical energy flow.
The mechanical energy is
(Winds etc.) Mechanical
subsequently transformed as
I energy
discussed in Chapter 13. I

1. 7 Determinism and chaos [1, 2]

Although equations like (1.1) do not identify cause and effect, they do show that
present conditions are inescapably linked to all past and future conditions. Of
course from the outset, people have been more interested in predicting future
conditions; and it was quickly assumed that present conditions, together with the
laws of behaviour, somehow 'determined' all future conditions (neatly side-
stepping the question of what determines the laws). This assumption was stated
most confidently by the French natural philosopher and mathematician Laplace
when he was writing his Celestial Mechanics in the late eighteenth and early nine-
teenth centuries. Having recast Newton's laws into an extremely compact form for
application to the motion of the planets, he claimed that perfect knowledge of the
present state of everything in the universe, together with perfect understanding of
all its laws, would give total knowledge of the future. Such Laplacian determinism,
as it came to be known, underpinned the confidence of the scientific community as
it grew and developed during the nineteenth century, investigating subjects (like
thermodynamics and fluid mechanics) vastly more complex than Laplace's
billiard-ball model of the the solar system. And meteorologists too came to share
this confidence in the twentieth century, as they began to apply established physical
laws to understanding and forecasting atmospheric behaviour.
The first obvious breach of Laplacian determinism came with increasing under-
standing of atomic-scale behaviour in the 1920s, when quantum theory postulated
an observational uncertainty (named after Heisenberg) which is inescapable in
practice and principle. Indeed it is now accepted that individual events on the
atomic scale are blurred by an uncertainty which can only be reduced by
considering the statistics of large numbers of such events. Fortunately, atoms are
so small and numerous that this is exactly what happens in the study of events on
scales much larger than the atomic. Nevertheless many scientists were unsettled by
the apparent failure of a principle which had come to be seen as a watershed
10 separating modern science from primitive capricious superstition, and to the end of
 (a) 1.7 Determinism and chaos
Variable

Time

Fig. 1.7 Portraits of chaos.

(a) The butterfly effect: two
runs of the some lorenz
(b)
atmospheric model, starting
from minutely different initial
conditions. Resemblance
between behaviour of a
chosen variable
(displacement along the
vertical axis) virtually
disappears about two thirds
the way through the time
period (the horizontal) of the
otherwise identical
simulations.
(b) loose repetition with
unexpected switches: the line
representing the continuous
sequence of states of the
lorenz convection model
turns in upon itself as the
convection cell rolls about a
horizontal axis. However the
rolling changes slightly and
progressively until it switches
slowly varying rotation in the
x opposite direction. (After E.N .
lorenz in [1 ] and [2]).

his life Albert Einstein resisted the idea of a purely statistical basis to physical law in
his famous dictum 'God does not play dice'.
Despite this revolution on the atomic scale, larger-scale behaviour was still
assumed to be forecastable in the Laplacian sense, until the spread of the electronic
computer in the 1960s allowed people to begin to investigate numerically the full
range of behaviour described by equations representing established natural laws,
rather than the very limited range covered by analytical (e.g. algebraic) solutions.
Quite unexpected results emerged separately from a wide variety of scientific
studies: physical, chemical and biological. When brought together in the 1970s,
recognition of common ground triggered one of those profound reviews of
accepted wisdom which seem to agitate the scientific community every 50 years or
so. Interestingly, what is now recognized as one of the most revealing of those early
separate studies was made by the meteorologist Edward Lorenz in the early 1960s
[2].
Lorenz was using a primitive computer to model global climate by allowing a
numerical weather-forecasting model to run for long periods, simulating years of
behaviour from assumed initial conditions. He found by accident that the model
was so very sensitive that minuscule differences in some initial condition of the
model atmosphere could give rise to completely different global conditions some 11
 Introduction finite simulated time later. Lorenz guessed that this was a basic property of the real
atmosphere, rather than an unrealistic property of his simple model. To express the
implied limits to the scope of useful weather forecasting, he coined the graphic
image of 'the butterfly effect', in which the flapping (or not) of a butterfly's wings
at some place and time could lead to a completely different disposition of weather
systems over the whole Earth some months later.
A little later Lorenz found similarly extreme sensitivity in an even more crude
model of atmospheric convection; and subsequent work by others has found it in
even simpler systems, such as the simple pendulum and the annual growth of
animal populations. In every case the sensitivity arises from non-linear relations
between parts of a system (i.e. sensitivity greater than simple direct propor-
tionality), and the result is a type of irregular, effectively unpredictable behaviour
which has been named chaos. It now seems that virtually every natural system has a
capacity for such chaotic behaviour, and that the atmosphere in particular is
probably chaotic on most of the scales depicted in Fig. 1.5.
Chaos is not random, despite superficial appearances. The present state of any
chaotic system is still related to ('determines') all future states through the
operation of natural laws, but the future behaviour of a chaotic system is
predictable in only a limited sense, even in principle. Sensitivity to initial conditions
means that uncertainties there (arising from inaccurate or inadequate observation,
or even ultimately from atomic-scale uncertainty) grow more and more rapidly as
the system evolves in time, and the same applies to any model simulation. As a
result, the doubling of accuracy in the initial conditions of even the most perfect
weather-forecasting model (for example) produces much less than a doubling in its
useful forecasting period. Although weather-forecasting models are still imperfect
in physical principle and computational practice, there is a growing consensus that
it may never be possible to forecast weather with current confidence more than a
couple of weeks in advance (current practice manages one week), confirming
Lorenz's doubts of 20 years ago. Of course less specific seasonal forecasts may
become possible, as understanding of climatic systems improves (section 4.9), but
they too will be limited if those systems are chaotic, as seems likely.
Another property of chaotic systems is that they never repeat themselves
precisely, although they often behave in a loosely periodic fashion. Quite simply,
if they did repeat exactly they would not be chaotic, since the 'deterministic'
nature of natural law would ensure that they repeated the cycle for ever, or until
disturbed from the outside. This capacity for loose rather than precise repeti-
tion, with broadly similar things repeating, but always slightly differently, is of
course just what we see in the atmosphere: depressions form every few days in
preferred parts of the middle latitudes, but the time intervals vary consider-
ably, and the depressions are never precisely the same; seasonal variations are
broadly but never precisely the same; and no two cumulus, turbulent eddies or
snowflakes are ever identical. Indeed the combination of individuality, regularity
and transcience mentioned in section 1.5 is now seen to represent the fingerprints of
chaos.
More subtly, it is known from the study of simple chaotic systems that details of
typical behaviour are closely related to the precise nature of the non-linear relations
at work in them. Although the chaotic nature of such a large and complex system as
the atmosphere is only beginning to be examined in detail, it is likely that typical
atmospheric structure (for example the sizes and lifetimes of depressions) may be
explicable to this extent some day. The doddering Polonius saw dimly through
Hamlet's simulated insanity in one of Shakespeare's most seminal quotations:
'Though this be madness, yet there is method in 't'. Until the recognition of chaos
12 as a type of behaviour in its own right, meteorology and climatology (in common
 Problems
with many other studies of simple and complex systems) concentrated on the
method and shunned the madness: now it seems that the madness itself is subtly
structured, and likely to reward careful study with increased understanding of our
marvellous and yet bewildering atmospheric environment.

Problems

LEVEL 1

1. Find the order of magnitude of the depth/breadth ratio for the troposphere.
2. What is the atmospheric pressure two decadal scale heights above sea
level?
3. Place the following in order of ascending height: troposphere, planetary
boundary layer, stratosphere, free atmosphere, stratopause, tropopause. Deal
with overlap using brackets.
4. What is the ratio (in m s -I) of the space and time scales of solar granulations?
5. Divide the following atmospheric quantities into vectors and scalars: wind
velocity, temperature, gravitational force, wind speed, pressure. (Note that
experiment shows that the pressure at any point in a fluid is independent of
direction.)
6. Identify the effective heat source for the atmospheric engine: the upper tropo-
sphere, tops of clouds, the Earth's surface, the stratosphere.

LEVEL 2

7. Figure 12.11 represents a vertical section through a tropical cyclone. If the

vertical scale of the diagram were redrawn to be equal to the horizontal, what
would be the length of the new vertical axis?
8. On a certain weather map point A has a temperature of 23 oc, and point B has
a temperature of 15 °C. If isotherms are drawn at intervals of 2 oc for even
values, at least how many isotherms must there be between A and B?
9. Horizontal accelerations of air in large weather systems are observed to
be - 10- 4 m s - 2 ; what is the implied net force on an air parcel of mass
1 tonne?
10. Estimate the volume of the air parcel in problem 9 near sea level.

LEVEL 3

11. Sketch an example of a temperature field for problem 8 with more than the
minimum number of isotherms.
12. Actually the spuriously animistic statement in section 1.5 about the anti-
cyclone and the rain belt can be rewritten just as simply without implying
animism, cause or effect. Try it.
13. A Hollywood film director is threatening to make a disaster movie in which 13
 Introduction a hurricane develops from nothing to maturity in 12 hours. Sketch an argu-
ment demonstrating the unreality of the event. And what would you say
to another proposing to have a rogue tornado scour the countryside for several
days?
14. What ultimately happens to the mechanical energy generated by the
atmospheric engine?

14
Observations CHAPTER

2.1 Care in observation

At an early stage in his development, man must have learnt to scan the skies and to
consider his sensations of warmth, wind strength etc. for evidence of impending
change in the weather, and in this sense observational meteorology may be over one
million years old. The need for such awareness increased greatly with the onset of
the agricultural revolution about ten thousand years ago, since farmers are under-
standably very sensitive to weather and climate at certain phases of the crop cycle.
The rapid development of long-range ocean voyaging in the last millenium
encouraged an even more critical sensitivity among mariners, and it is hardly sur-
prising that modern meteorology originated largely in their need for reliable
warning of heavy weather, and in its early stages was often assisted by their
observations and perceptions. The development of aviation in the present century
has produced a demand for foreknowledge of winds, cloud and hazardous
conditions along flightpaths and at departure and destination airports.

OBSERVATIONAL AIDS

If sufficiently motivated, people can make themselves aware of a wide range of

meteorological conditions, but the often weak and ambiguous reaction of many of
the human senses severely limits the quality and quantity of our observations, even
when directed by an anxious and enquiring mind. It is true that our three-
dimensional hunter's vision is very acute, but this asset is largely wasted in the
absence of other similarly accurate observations, and in the absence of even
minimal understanding of what is observed. To resolve the apparent confusion of
atmospheric behaviour, data must be as unambiguous and definite as possible, and
therefore mechanical aids are needed to refine and extend our basic human
sensations. Most of the more detailed observational information mentioned in this
book was gleaned only after basic observational instruments had been developed.
You can gain some idea of the crucial role played by such artificial aids by
examining the inspired but erratic speculations about the atmosphere made by the 15
2 Observations Roman poet and natural philosopher Lucretius about 55 BC [3], who relied
completely on his unaided senses.
Given that mechanical aids to observation are needed, several general points
arise. First, although it may be clear that temperature, for example, should be
measured by thermometer rather than by bodily sensation of warmth, it may not be
clear initially exactly how the thermometer should be exposed. Is atmospheric
temperature more truly represented by a thermometer exposed to available
sunlight, whose warmth is such a significant component of what we perceive to be
the weather, or by a shaded thermometer? No doubt you know that the standard
practice now is to use a carefully shaded thermometer, but this was not generally
accepted for nearly a century after thermometers were first used meteorologically.
Nor was the delay due to stupidity; the best argument for using shaded thermo-
meters is that they give the most coherent patterns of temperature across time and
geographical space, but this emerged only after extensive trial and (as we now
know) error. There are theoretical justifications too, but they were unlikely to be
understood until correct observations began to build a body of meteorological
understanding in the nineteenth century.
The problem of how best to use an observing instrument is particularly acute
when it does not correspond to any existing human sense. The measurement of
atmospheric pressure by barometer is an early example which is discussed further
in section 2.5, but the present hectic pace of technical innovation produces new
examples frequently, the meteorological satellite being a conspicuous case. Novel
data are at once challenging and baffling, and there is normally a substantial time
period in which our understanding is struggling to assimilate the new data, and in
which data can easily be misused. The satellite may still be in this phase.
Last, it is important to bear in mind the simple fact that all measurements are
subject to limitation and error. The glossy trappings of modern instrumentation
may tend to inspire blind confidence, especially if the data emerge from a
computer, but this should be firmly resisted. Every meteorological observation is
less than perfect by an amount which should be known when it is being interpreted.
Imperfections in individual observations are compounded by the problems of
representativity and consistency which arise when data are assembled from widely
scattered observation points, as is normal in meteorology.
The following sections exemplify and enlarge on these points, beginning with a
consideration of a few basic and familiar instruments used for making observa-
tions near the surface. The outline is highly selective, and much more compre-
hensive descriptions are found in manuals dedicated to observing instruments and
techniques [4, 5]. The previous remarks about the time taken to assimilate new data
will apply more or less forcefully to you depending on your previous knowledge of
meteorology. It may not be at all clear to you why certain measurements are made
the way they are, or are made at all. Clarification should increase with knowledge
of the atmosphere, but this cannot be compressed into a single chapter, or indeed a
single book.

2.2 Temperature measurement

Air temperature is one of the most important of atmospheric properties, but

16 though our bodies are sensitive to it, they are quite unreliably so. We feel hot or
 2.2 Temperature
measurement

.
(
1 Fig. 2.1 Stevenson screen
with door open for
thermometer reading,
showing the normal dry- and
wet-bulb thermometers inside
(vertical left and right
respectively). The horizontal
thermometers ore maximum
and minimum dry-bulb
thermometers .

cold depending on whether our bodies are having to waste or conserve heat to
maintain their internal temperatures at about 37 °C, and which of these is
happening at any moment depends on exposure to sunlight, terrestrial radiation,
wind, humidity (which affects our ability to lose heat by sweating), health, and the
time, size and temperature of our latest meal, as well as on the temperature of the
ambient air [6]. Clearly a less ambiguous sensor is required.
The mercury-in-glass thermometer was perfected by Fahrenheit in the early
eighteenth century, but though it was used almost immediately to measure air
temperature, over a century elapsed before the need for careful exposure was
generally recognized. Such care is necessary because, like the human body, the
thermometer is sensitive to solar and terrestrial radiation, from which it must be
shielded to isolate the effect of air temperature. However, the shield must allow
sufficient ventilation for the temperature of the air in contact with the thermometer
to be at the same temperature as the air outside the shield. The familiar Stevenson
screen (Fig. 2.1) is a simple, robust and reasonably satisfactory compromise
solution. The white surfaces absorb little sunlight, and the thick wooden walls
insulate the interior from the warming effect of residual solar absorption, and
from the warming and cooling effects of terrestrial radiation. The Iouvred walls
and floor allow natural ventilation of the interior by available wind. The access
door is placed on the poleward side of the screen so that direct sunlight does not
enter when readings are taken. The ventilation is the least satisfactory part of the
compromise: on still, sunny days it is so poor that the temperature inside the screen
may rise a degree or more above the external air temperature. The ventilation can
be improved by using a fan to force a constant draught through the interior of the
screen, but the instrument then requires a source of power, and is consequently
more elaborate and prone to breakdown.
The mercury-in-glass thermometer is still in widespread use, though it has two
major disadvantages: it cannot be used at very low temperatures (such as occur at
high latitudes and altitudes), since mercury freezes at about -40 oc, and it cannot
give a continuous record of temperature automatically. Alcohol-in-glass thermo-
meters can register much lower temperatures, and are widely used in minimum
thermometers, in which the alcohol meniscus moves a metal index inside the bore 17
2 Observations of the thermometer, leaving it behind at the minimum value for subsequent
reading. Simple continuously recording thermometers (thermographs) contain a
bimetallic strip which moves a pen over a clockwork-powered chart drum; and in
recent years a wide range of thermometers has been developed making use of the
temperature sensitivity of electrical resistance, often to produce continuous
records.

INSTRUMENTAL LAG

No measurement can be made instantaneously: in principle each is an average over

a finite time interval, however short. The length of this interval is most simply
assessed by observing the reaction of the measuring instrument to an abrupt change
in the variable being measured (e.g. temperature). The lag of the instrument is con-
ventionally defined to be the time taken to register 63o/o of the change, this value
being appropriate to an exponential type of response (Appendix 2.1), such as is
observed in many instruments. The lag is clearly a measure of the sluggishness of
instrumental response, and of the degree of averaging therefore inherent in all its
data. As a rule of thumb, an instrument cannot reliably measure variations
occurring with time periods less than ten times the lag.
In thermometers the chief cause of lag is the thermal capacity of the actual
sensor, and the speed of heat transfer between it and the air. The heat capacity of
the mercury reservoir and glass bulb, and the slow conduction of heat through
the glass, are limiting factors in the mercury-in-glass thermometer, and large
specimens have in consequence lags of about a minute in a ventilation draught of
5 m s - 1• They are even more sluggish in slower draughts because heat flow between
bulb and air is reduced. Such lags are quite adequate for routine hourly measure-
ments near the Earth's surface, but they are much too long to allow measurement
of the rapid temperature fluctuations which accompany turbulence in the air, each
of which may last for only a fraction of a second. Thermometers for such purposes
must have very small thermal capacities, and are therefore physically very small,
having fine wires or beads as sensors.

2.3 Humidity measurement

The amount of water vapour in the air is measured by hygrometers, and is

expressed in several different but related ways.
When evaporation or condensation is the main interest, the most relevant
measure is the relative humidity, which is the ratio of the actual vapour density (or
pressure) to the value which would produce saturation at the same temperature,
usually expressed as a percentage. (The concept of saturation is examined more
closely in section 6.2.) Since the saturation vapour density (or pressure) depends
only on temperature - a dependence well-known from laboratory measurement
(Fig. 6.2) - simultaneous measurement of relative humidity and air temperature
18 determines the actual vapour density. Animal tissues, such as hair or skin, respond
directly to relative humidity, and this response is the basis for several simple hygro- 2.3 Humidity
meters. For example, in the common hair hygrometer, a hair is kept under slight measurement
tension so that the decrease in its length with increasing relative humidity is easily
registered. The length variation is sufficiently regular and repeatable to allow any
particular hair or bunch of hairs to be calibrated in different known relative
humidities (determined by other methods) to produce a quantitative instrument.
Hair and skin hygrometers have lags - 10 seconds at room temperature, but are
much more sluggish at lower temperature. Unfortunately they suffer so severely
from hysteresis and ageing that they are best used to register changes rather than
absolute values. However, their simplicity and ability to produce a continuous
record (in a hygrograph) make them very popular for semi-quantitative purposes.
Some hygrometers measure relative humidity through its effect on the electrical
resistance of a hygroscopic surface, and can be made somewhat more reliable than
the hair and skin types.
Psychrometers measure the humidity of the air from the cooling effect of water
evaporating into it, and their simplicity and reliability make them the most widely
used instruments not requiring calibration. The simplest types consist of a wet-bulb
thermometer mounted beside a normal thermometer (i.e. one with a dry bulb)
inside a Stevenson screen (Fig. 2.1). Natural ventilation allows water to evaporate
from the saturated wick of the wet bulb, cooling it below the temperature of the dry
bulb by an amount which varies with the humidity of the ventilating air. In dry air,
such as is found in desert regions, this wet-bulb depression is very large (- 10 °C},
whereas it is zero in saturated air, for example in fog. Given the wet and dry bulb
temperatures, a special book of hygrometric tables [7] or a special slide rule gives
the associated vapour pressure (or density, though pressure is conventionally
preferred), relative humidity and dew-point temperature (see below). When the
Stevenson screen is inadequately ventilated, the actual humidity of the air outside
the screen is overestimated somewhat, since the air inside the screen is significantly
moistened by evaporation from the wet bulb. Mechanical ventilation of both bulbs
gives more accuracy in this respect, and is provided in the aspirated type of
psychrometer. The wet bulb has much the same lag as the corresponding dry bulb,
and of course requires similar shielding to minimize radiation error (if it is not
already in a screen). Electrical psychrometers with tiny wet and dry bulbs can detect
the rapid humidity fluctuations associated with atmospheric turbulence.
A further measure of the humidity of the air is the dew-point temperature, or
simply dew point, which is defined to be the temperature of a chilled surface just
cold enough to collect dew from the adjacent moist air. In a dew-point meter a
polished metal surface is chilled progressively below the temperature of the
ambient air, and the surface temperature is noted at the moment when the first dew
appears. In this form it is an elaborate, sluggish but potentially very accurate
instrument which requires no calibration. Dew points can be converted to other
humidity measures by using the hygrometric tables mentioned previously. Very
roughly, the dew-point depression below the ambient air temperature is twice the
wet-bulb depression, ranging from tens of degrees in arid air to zero in saturated
air.
The theoretical relationships between vapour pressure, density, relative
humidity, wet-bulb temperature and dew-point temperature, and other measures
of humidity, are not straightforward. They are outlined in sections 5.4 and 5.5 and
discussed fully in advanced texts [8].

19
2 Observations
2.4 Wind measurement

Horizontal wind direction is indicated by vanes which are in principle the same as
those exposed on church steeples and old buildings. An electromechanical device
can be used to give a remote and recordable readout of the vane direction. The
standard meteorological vane has a lag of several seconds in normal winds but is
more sluggish and ultimately unresponsive in very light winds, which are anyway
very variable in direction. Small, light and aerodynamically optimized vanes are
needed to register the rapidly varying directions associated with turbulence.
The meteorological convention is to describe the direction of a horizontal wind
by the direction from which it is blowing, as in everyday speech. Thus a north-
westerly wind blows from the north-west, or equivalently from an angle of 315°
(the azimuth measured clockwise from zero north). This opposes the convention
used in other branches of physical science, where the direction of any motion is
considered to be the direction towards which the body of air is moving, but arises
naturally from the age-old practice of sailors and farmers of looking upwind to see
impending weather, which in turn reflects the fact that many weather systems
travel more or less with the wind near the surface (but not always; see section 10.4
for example). Because of its rapid variations, wind direction is seldom specified to
better than the nearest 5o.
Wind speed is measured by various types of anemometer, but because these are
surprisingly difficult to make usefully accurate and reliable, the earliest successful
procedure used instead a classification of wind strength in terms of observable
effects of wind on the sea and sailing ships. This famous Beaufort scale was soon
extended to include the effects of wind over land and is still used very widely to
describe the strength of wind near the surface. Subsequent measurement by
anemometer has associated a range of wind speeds with each Beaufort number
(appendix 2.2). Another maritime relic is the continued use of the knot (1 nautical
mile per hour) as a unit of wind speed in parallel with the proper SI unit, the metre
per second. The precise equivalence is given in the secondary table on p. xviii, but
for rough purposes the metre per second can be approximated to 2 knots (I m s- 1
= 2 kt).
The common cup anemometer (Fig. 2.2) is insensitive to the horizontal wind
direction or azimuth, unlike the propel/or type which has to be kept pointed into
the wind by a steering vane. The cups of the former type whirl continually because
the drag of the wind is greater when blowing into the mouth rather than into the
back of each cup, but this results in greater lags in dropping rather than in rising
gusts, with the result that wind speeds are significantly overestimated in gusty
conditions. With careful design the rate of cup rotation can be made almost
directly proportional to wind speed over a usefully wide range of speeds. In many
observations the run of wind is found by converting the number of cup axle
revolutions in a certain time period into a length of wind. Divided by the time
period used, which ranges from ten minutes to a day depending on the type of
observation, the run of wind yields the average wind speed in this period. Ten-
minute averages, together with maximum speeds in gusts, are used to define wind
speeds near the surface in routine observations. The instrumental lag associated
with the robust standard type of cup anemometer ensures that gusts are themselves
smoothed over periods of several seconds.
The only anemometer not requiring prior calibration in a wind tunnel (at least in
principle) is the pitot tube type, which measures the excess pressure developed in air
20 as it rams into the mouth of a small tube steered into the wind by a vane. Since the
 2.5 Pressure measurement

Fig. 2.2 A standard British

Meteorological Office pattern
cup anemometer. Its robust
construction enables it to
cope with stormy conditions
with minimum maintenance.
Revolutions are counted and
displayed on the instrument
(as here). or remotely by
electrical signal.

rammed air is more or less completely halted in the process, the air loses all its
kinetic energy, with the result that the excess pressure is proportional to the square
of the wind speed (by Bernoulli's equation) and the instrument is insensitive to low
wind speeds.
So far it has been assumed that the horizontal component of wind is the only one
of interest. This is largely true because the flattened shape of the atmosphere
ensures that the horizontal wind speeds are usually more than 100 times larger than
vertical ones, and almost always more than 10 times larger. When the vertical
component is to be included, additional vanes and anemometers are used, such as a
propellor on a vertical axle. Small-scale turbulence in particular has an important
vertical component which is comparable with the horizontal ones because the
turbulence is essentially isotropic. Updraughts in shower clouds are known to be
quite significant, but cannot be measured by fixed anemometers for obvious
reasons. They can however be measured from aircraft traversing through them by
using routine and specialized aircraft instrumentations. The very weak but
persistent updraughts which maintain the great sheets of frontal cloud
(nimbostratus) are too small to be directly measurable by direct observation,
though they can be inferred from other measurements, sometimes by way of quite
sophisticated theory (section 7 .13).

2.5 Pressure measurement

Man is insensitive to naturally occurring pressure variations below the audible

range (say 20 Hz), with the result that the existence of substantial but much slower
variations in atmospheric pressure was quite unsuspected before the development
of the barometer in the mid seventeenth century. However, the relation between
pressure variations and weather sequences was then almost immediately noted, and 21
2 Observations (b)

vs

Fig. 2.3 (o) A Fortin-type mercury barometer. The mercury reservoir in the soft bog B is
adjusted by the adjusting screw AS so that the exposed mercury surface just touches the point P
at the bottom of the fixed vertical scale S (visible through the window W). The Vernier slide V is
moved by VS until it appears to rest on top of the mercury meniscus M. (b) A barograph with
glass case removed. The aneroid capsule is visible on the right, and the irregular atmospheric
pressure trace on the left shows variations of about 10 mbor in 24 hours.
w

the barometer has occupied an important place among meteorological instruments

ever since. The relationship between pressure and weather is by no means as simple
as the legends on domestic barometers suggest, but it is nevertheless interesting and
important in ways which are apparent in Chapter 11.
The basic absolute instrument is still the mercury-in-glass barometer, working
on the same principle as the ones first made by Torricelli in 1644 (Fig. 2.3).
Atmospheric pressure acts on an exposed surface of mercury to maintain a column
of mercury in a vertical glass tube whose top end is sealed and evacuated. The
(o)
vertical height h of the top of the mercury column above the exposed surface is
related to the atmospheric pressure p by the barometric equation eqn (4.16)
p = gph (2.1)
where g is the gravitational acceleration and pis the density of mercury (see table on
p. 183). Such barometers are heavy and frail, but if they are read carefully and
corrected for variations in g and temperature (which affects p), they are accurate to
better than 0.1 mbar. Marine barometers are mounted in gimbals so that they
remain nearly vertical despite the movement of the ship.
The reading procedure imposes an effective lag of - 10 s on the mercury
barometer. However, most of the significant variations of atmospheric pressure
associated with weather systems occur over much longer time periods, and in fact
the standard hourly reading is usually quite adequate. Recently the more compact
and robust precision aneroid barometer has largely replaced the mercury-in-glass
type in the meteorological network. A partly evacuated (aneroid) flexible metal
capsule expands as the ambient air pressure decreases and vice versa. The move-
ment of the capsule is detected by a manually operated probe, and the instrument is
calibrated against a master mercury-in-glass barometer. The common domestic
22 aneroid barometer operates on the same basic principle, but the capsule movement
is displayed by mechanical linkages whose friction and looseness make the 2.6 Precipitation
indicated pressure liable to errors of several millibars. Even so this simple measurement
instrument can give a useful impression of the big slow pressure variations which
are associated with weather systems in middle and high latitudes.
A major shortcoming of these barometers is their inability to yield a continuous
record. This is overcome in the traditional barograph (Fig. 2.3), in which the
movement of the aneroid capsule is communicated mechanically to a pen writing
on a graph fixed to a drum rotating by clockwork. Even when very well made, the
problems associated with mechanical linkages are even worse than in the aneroid
barometer, but the pressure trends (the pressure tendencies) are graphically
depicted, and are very useful for weather analysis and forecasting. At the moment
electromechanical linkages are beginning to be produced which should give much
better accuracy, and instruments using the pressure sensitivity of electrical
properties of crystals may eventually give rise to barographs as accurate as the best
mercury-in-glass barometers.
Note that all such instruments measure the ambient atmospheric pressure of the
barometer, usually known as the station pressure. As discussed in section 2.9 these
may need substantial correction to produce values which are consistent across a
network of stations.

2.6 Precipitation measurement

Most precipitation reaches the surface as rain, but strictly the term includes all the
forms in which water and ice fall to the surface, which include the various types
of snow and hail, as well as rain and drizzle. Wherever possible, amounts of snow
and hail are expressed as rainfall equivalents. Rainfall measurements are of great
interest to hydrologists concerned with the management of rivers and reservoirs, as
well as to meteorologists.
The quantity of rainfall in an observation period is specified by the depth of
water collected on a horizontal area, but because this may be as little as 0.1 mm in a
typical collecting period (12 or 24 hours), rain gauges always use funnels to con-
centrate the precipitation after collection. The commonest standard gauge has a
circular mouth about 130 mm in diameter, and a much narrower glass measuring
cylinder with graduations adjusted to allow for the concentrating effect of the
funnel. Automatic gauges with wider mouths operate by registering electrically the
times taken to fill small cups beneath the spout of the collecting funnel (e.g. the
tipping bucket gauge in Fig. 2.4).
The ability of the gauge to resolve temporal variations of rainfall usually
depends on the sampling rate rather than on instrumental lag. For example,
manual gauges are read once or twice per day, which is adequate for many purposes
but obliterates all the interesting variations occurring in the intervals between
successive readings. These are revealed by automatic gauges, many of which
register the collection of 0.05 mm of rain. The time charts of rainfall revealed by
automatic gauges are often extremely spiky, and show that large fractions of total
rainfall occur in relatively short bursts of heavier rain (Fig. 2.5).
The great weakness of all these types of gauge is that their spatial representativity
is always in doubt: X mm of rain falls at gauge A in a certain storm, but what would
have been collected by identical gauges nearby if they had been in operation? From 23
2 Observations a
turf wall p it

Fig. 2.4 (a) An idealized

vertical section through rain
gauges set inside a turf wall
and in a pit, to minimize
interference by airflow over
b y
funnel

the gauges. (b) A vertical

section through the
mechanism of a tipping t ipping b ucket
mecha ni sm
bucket type of automatic
gauge. The bucket see-saw
tilts into the dashed position
when the filling bucket is full.

Fig. 2.5 Rate of rainfall

recorded on a Jardi-type
gauge. The continuous rain
(with embedded heavier
bursts around 1800 Z) fell
from a passing active
occluded front. The isolated
spikes early next day
correspond to showers in the
following colder north-
westerly flow.

experimental work with dense networks of gauges it is known that measured rain-
fall can be very strongly and closely patterned in space - the patterning arising
from details of gauge exposure, topographical influence and small-scale structure
inherent in precipitating weather systems.

1. Gauge exposure. The efficiency of collection and therefore the accuracy of a

gauge is quite sensitive to the way in which the collecting mouth is exposed.
Until recently gauge mouths were always raised above the surrounding surface
to avoid flooding or splash-in. But this deforms the flow of air over the gauge
and may significantly reduce the amounts collected, especially in the higher
winds which often accompany substantial rainfalls. Placing the gauge mouth
about 30 em above the surface has been found to be an adequate compromise
over land. Recently flush-mounted gauges surrounded by a dummy splash-free
surface have begun to come into increasing use (Fig. 2.4), and where this is not
desirable the aerodynamic 'waisting' of the body of the normally elevated
gauge has been found to reduce error.
Problems of exposure are especially severe in ship-borne gauges, where spray
and the effects of the ship's large bulk and motion are effectively unavoidable.
Some might be reduced by mounting gauges on carefully designed unmanned
observation buoys which are kept on station automatically, but these are
sophisticated devices whose design is still being optimized. So far the
24 distribution of rainfall over the seas is much more uncertain than it is over land,
 and is probably the most inadequately observed major element of the world's 2.6 Precipitation
measurement
weather.
2. Topographical influence. Rainfall can be very strongly enhanced over the
windward side of high ground and reduced in its lee (forming a rain shadow).
Since such patterns may have an important effect on agriculture and water
supply, for example, and yet be too finely detailed to be properly represented by
any feasible gauge network, quite subtle techniques have been devised by
hydrologists to interpolate between gauges as realistically as possible [9].
3. Inherent rainfall patterns. These range in scale all the way from the fine streaks
apparent in curtains of rain backlit by the sun, through showers, to the
extensive swathes associated with major weather systems such as fronts.
Showers in particular are difficult to allow for since they often contribute signi-
ficantly to local totals, but yet are so small that they tend to slip between the
inevitably sizeable gaps in the gauge network. This results in overall under-
estimation of the total amount of precipitation (since there is almost no chance
of the few showers coinciding well enough with the few gauges to produce over-
estimation) and a wide variability between adjacent gauges.

Since about 1950 increasing use has been made of radar to observe the
distribution and intensity of precipitation over wide areas (Fig. 2.6). The technique

Fig. 2.6 A composite display

from six meteorological
radars in the southern British
Isles at 1200 Z on 4 August
1985. The echo strengths
correspond to rainfall rates of
between about 0.5 and
8 mm hr- 1 , and the rainy
areas clearly outline the
warm front (centre right) and
the cold front (trailing left)
associated with a small but
vigorous depression centred
west of Ireland. The rain over
the Irish Sea (top centre) is
associated with o relatively
weak occluded front bending
westwards toward the low·
pressure centre.

25
2 Observations originated in the Second World War when it was discovered that heavy precipita-
tion produced radar echoes whose strengths were afterwards found to increase with
the average size of the raindrops, snowflakes and hailstones, and with the numbers
of these per unit volume of air. Since these in turn are fairly directly related to the
rate of precipitation toward the underlying surface, carefully calibrated radars can
be used to measure precipitation rates in a way which avoids many of the snags of
conventional point measurement. Although echo strength varies somewhat
depending on whether the precipitation is rain, hail or snow, the radar echo also
improves on the conventional gauge in its ability to assess snowfall, which is very
difficult to measure accurately otherwise because of crippling exposure problems
and the choking of gauges, even when heated to measure equivalent rainfall.

2. 7 Other surface observations

Several of the routine types of observation not mentioned so far are made by the
unaided human senses. These are ones which have been found by long experience
to be good indicators of the current state of the atmosphere from the point of view
of forecasting, and are complex in ways which make them difficult to instrument.
For example the amounts, types and heights of clouds are judged by eye, using
internationally agreed categories of cloud which have been perfected over the last
ISO years and are now enshrined in the International Cloud Atlas and derived
handbooks [I 0]. The categories of type and height have been chosen so that cloud
reports emphasize factors which are especially telling evidence of the presence of
particular weather systems: for example, the high clouds which are characteristic
forerunners of a warm front (Fig. 2. 7). Further examples appear in Chapters 6, 10,
II and I2, and the Cloud Atlas types and names are listed in Appendix 2.3 .

Fig. 2.7 Cirrus plumes (cirrus

uncinus) ahead of a warm
front . Patches of ice cloud in
the high troposphere are
producing trails of larger
crystals which are drawn out
as they fall through the
strong wind shear there. Note
that the wind shears in
direction as well as speed.

26
 In a similar way the current or present weather is described by selecting one from 2.7 Other surface
a series of 99 types arranged in order of increasing forecasting significance, only observations
the most significant type present being reported on any occasion. For example, all
types of fog are less significant than rain, light rain is less significant than moderate
rain, and the most significant weather type of all is a heavy hail shower with
thunder (indicating extreme convective instability). Past weather, meaning
weather in the period (often 6 hours) immediately before the time of observation, is
similarly described using a list of 10 types. Present and past weather types are
briefly summarized in appendix 2.4.
Of the many other types of observation which could be mentioned, the most
important concern solar and terrestrial radiation. For nearly a century, periods of
bright sunshine have been recorded by the elegant and simple Campbell-Stokes
sunshine recorder (Fig. 2.8). Indeed its very success makes it difficult to replace by
a comparable instrument with an output more convenient than its scorched paper
strip. In recent decades various types of solarimeter have been devised to measure
the intensity of direct and diffuse solar radiation by registering the electrical effects
of its absorption. The simplest types use a thermopile to measure the warming of a

(a) (b)
Fig. 2.8 (a) A Campbell-Stokes sunshine recorder. The gloss boll focuses the sunlight on the
recording card on the polar side of the mounting. The image of the bright sun chars a trace from
left to right (in the northern hemisphere) across the printed hour scale. A new cord is inserted
doi ly. (b) A solorimeter. Sunlight diffuses through the small gloss dome (diameter 7 em) and
worms the blackened top end of a thermopile whose shielded lower end stays at ambient
temperature. The white collar provides a rodiotively uniform surround. 27
2 Observations blackened surface beneath a glass dome (Fig. 2.8), and these can be made sensitive
to terrestrial radiation as well by using a shield which is also transparent to infra-
red radiation. Since glass is opaque to terrestrial radiation, like most terrestrial
solids and liquids, the dome formerly had to be made of quartz, but fortunately
polythene now provides a cheaper transparent alternative. A pair of these, one
facing up and the other down, with their electrical outputs arranged to give the
algebraic sum of the downward fluxes, constitutes a net radiometer, directly
measuring the resultant of all solar and terrestrial radiant fluxes passing downward
(conventionally) through a horizontal surface. Values of the net flux measured
near the Earth's surface are of great importance in quantifying the factors deter-
mining local climate.
Temperatures at and beneath the ground surface are particularly useful in agri-
culture. The daily minimum surface temperature is especially significant when it is
near 0 °C, and is usually measured by an alcohol-in-glass minimum thermometer
exposed just above short grass. This grass minimum temperature is often used as a
measure of the degree of ground frost, though minimum temperatures actually
vary considerably with the type of surface (section 9.4). Thermometers are also
buried at various shallow depths in soil to measure the temperatures experienced by
the roots of crops, etc.

2.8 Observation networks

By the early nineteenth century it was realized that weather in middle latitudes was
organized in moving patterns with horizontal scales of 1000 km or more. Given
their speed of movement and development, any analysis of such patterns could
only be retrospective while the maximum speed of messages was that of a galloping
horse. The development of the electric telegraph by Morse in 1844 transformed the
situation. A network of meteorological observing stations connected by telegraph
to a centre for analysis and forecasting was established in France in 1863, and the
USA, Britain and other technically developed countries quickly followed suit.
Thus began the network which now covers the land areas of the globe, albeit very
unevenly (coverage increasing with local material wealth), and which is formally
organized by the World Meteorological Organization (WMO) acting to coordinate
the many national networks. The complete observing, communicating, and
analyzing and forecasting system is now entitled World Weather Watch (WWW).
We will consider only the observing part of WWW, under the separate headings of
surface, upper air and satellite networks. First consider some quite general points.
WWW is geared to record the present state of the atmosphere with just enough
detail and accuracy to permit useful forecasting. Initially, when only the surface
network existed, the forecasting was primarily for shipping near coasts, but it is
now aimed at all sea and air transport, and land transport in hazardous conditions,
as well as agriculture and industry, and the public at large. In some parts of the
globe, special hazards such as hurricanes or tornadoes receive special attention.
Just how much detail and accuracy will suffice for forecasting is a crucial
question which is answerable only by experience. Given the obvious complexity of
the weather and our limited understanding of it, it might seem that we should try to
observe as much as possible, everywhere and all the time. However, since personne'
28 and finance are obviously severely limited in practice, the question is always 'wh t
 2.8 O bservation networks
H.W. BRAN DES:
Cadc synopliquc
rlu GMaPs 1183 .

Fig. 2.9 An early synoptic

chart reconstructed by the
pioneer Brandes some
decodes after the event. The
0 isobars (of pressure
difference from normal. to
allow for the effects of
variable station altitude)
clearly reveal the south-
eastern quadrant of a
depression centred between
England and Holland. with
cyclonic surface winds angled
u toward low pressure by
friction (section 7.11 ).

is the minimum network of people and instruments which will provide data good
enough for useful forecasting?' Even this question does not have a definitive
answer, since the amount of coverage considered adequate depends on the level of
our understanding of the systems to be forecast (in principle, better understanding
permits more efficient observing), and with our needs. Up to now, forecasters have
been content with networks only just good enough to define large-scale weather
patterns, such as extratropical cyclones and their somewhat smaller-scale frontal
structures, so that the networks of surface and upper air stations making synoptic
(simultaneous) observations have evolved to fit this role, to the extent that this
scale of weather phenomena is now called synoptic scale.
Even though the density of a reasonable compromise network is widely agreed
by forecasters, the fact remains that this has been implemented by only the most
wealthy nations: others have networks which are more or less inadequate, and the
oceans are effectively data deserts. This patchiness of observational coverage in the
face of the need for homogeneous, minimal coverage, is the fundamental problem
facing WWW, and will require decades to solve.
Even if the synoptic network were to be completed overnight, it would still be
much too open to permit more than random observations of sub-synoptic-scale 29
2 Observations systems such as shower clouds and the quite substantial clusters which they
sometimes form. Forecasting such events therefore is and will probably remain
statistical, and predictions such as 'heavy showers are likely near the west coast'
will no doubt remain normal. Weather radar can give more specific warning of all
substantial showers within range, but showers are so short-lived that individual
forecasts could be given only by interrupting local radio or television broadcasts.
This is precisely what is done in parts of the United States in the tornado season
(tornadoes being associated with very heavy showers), but it is worthwhile only
because tornadoes are so damaging and dangerous. Continuously broadcast
televised 'pages' of local forecasts could no doubt be updated frequently to provide
similar though less arresting advice.
Lastly, in considering networks of surface and upper air observations in
particular, note the need for standardization of observational practice: unless great
care is taken, data from different stations may well differ more on account of dif-
ferences in observational practice than atmospheric variations. This is obvious in
principle, but its effective implementation requires surprising rigour in even the
smallest matters.

2.9 The surface network

Ideally, observations of wind, temperature, humidity, pressure, visibility, and

cloud and weather types are made at each station of the network every hour just
before the hour (Greenwich Mean Time or Universal Time, usually denoted by Z),
so that it is possible in principle to have a synopsis of world-wide data at any hour.
In regions where such regularity cannot be relied on, efforts are made to ensure that
the observations at 0000, 0600, 1200 and 1800 Z are made and reported. The
treatment of precipitation totals is rather different, since manual gauges are tradi-
tionally read at 0900 and 2100 local time to reveal the systematic differences
between night and day which are so conspicuous over land. Obviously all instru-
ments are designed to conform to international standards, and are operated by
staff trained to use them in a consistent and proper manner.
As far as is possible the observations are made in sites which are standard in
construction and exposure, but yet are representative of the surrounding terrain.
Full details, even down to the length of grass in the fenced enclosure, are specified
in handbooks such as [5]. However, sites vary considerably in altitude, and since
the atmospheric pressure decreases by about 1 mbar (0.1 OJo) for every 10 m rise,
substantial corrections must be made to station pressures to prevent the pressure
variations associated with differences in station altitude from swamping those
arising from weather systems. These corrections are made by estimating the
additional pressure which would have arisen had the atmosphere continued down
to mean sea level (the agreed datum level) at the station. The correction p is
calculated from a form of eqn (2.1), where h is the height of the station above MSL
and pis the fictional air density, which is equal to or closely related to the air density
at the station. This is a satisfactory procedure for the many stations which lie
within a few hundred metres of MSL, but it gives trouble at mountain stations
where a kilometre or more of fictional atmosphere may have to be invoked.
No attempt is made to correct air temperatures for station altitude, but the
30 height of the Stevenson screen above the local ground surface must be carefully
chosen. Daily temperature variations decrease sharply with increasing height 2.9 The surface network
above the ground surface, because heating and cooling are concentrated at the
surface. Ideally we might consider measuring air temperature above the surface
layer, but since the screen would then have to be at least several tens of metres
above the surface, there would be severe practical problems and the data would be
largely irrelevant to the immediate human environment on many occasions. The
compromise solution is to place all Stevenson screens in the network at the same
height above the local surface, so that they are subject to similar daily cycles. The
standard height of the thermometer bulbs (often known as screen level because it is
in the middle of the screen interior) is 1.5 m, with slight national variations, and is
obviously related to normal human stature.
Wind speed increases with height above the ground surface, as well as with
distance from upwind obstructions. The latter effect is minimized by judicious
siting, but the former again requires an arbitrarily chosen measurement level, since
the frictional drag of the surface on the air only really disappears some 500 m
above the surface. The actual level chosen is 10m, which is about the highest level
which can be reached using a small mast. The wind speed there is typically about
three times the speed at 1 m height. At many secondary observation stations even
this modest height is unattainable, and so the anemometer has to be sited at a lower
level and its data multiplied by a standard factor to give even a reasonable
approximation for the 10 m wind speed. Wind direction is measured at the same
height as speed.
The distribution of surface synoptic stations throughout the Northern
Hemisphere and in the British Isles in particular is shown in Fig. 2.10. With an
average separation of 50 km, this is typical of the surface networks of wealthy,
densely populated countries. In poorer and more sparsely populated areas the
networks are much more open. Fully automatic stations, requiring manning only
for maintenance and repair, are being developed to plug gaps and reduce high
staffing costs. Over the sea there is increasing but still very inadequate coverage
through the voluntary participation of merchant shipping. This is extremely
valuable, though data from shipping suffer from the effects of non-standard
exposure (since there is no such thing as a standard ship) and crowding along
popular shipping lanes at the expense of intervening expanses. Because of their
great capital and running cost, there are very few special weather ships, and so
unmanned instrumented buoys with radio transmitters are being designed and
tested for eventual inclusion in WWW. The worldwide total number of reporting
land-based synoptic surface stations is static at about 2500, whereas the average
number of ships reporting at the important synoptic hours has trebled in the ten
years since 1975, and is approaching 1500.
In addition to the standard synoptic network, there is in Britain, for example, a
much denser network of voluntarily manned stations, many of them at schools,
where limited standard observations are made each day at 0900 Z. These data are
not available to forecasters, but they are collected, scrutinized and published in
monthly compilations by the British Meteorological Office (the national fore-
casting service); these compilations serve to define the British climate and its
variations on a much finer scale than is possible with the synoptic network. Other
countries have similar non-standard systems.
Networks of weather radars are as yet very incomplete, and confined to the
wealthiest nations. There are quite large numbers of radars in North America, not
all coordinated effectively; there is a small coordinated network in the southern
British Isles, and a fairly comprehensive network is envisaged for Europe. The
large useful range of radars means that adjacent sets can be separated by over
100 km in non-mountainous terrain. Their data are analysed, improved (for 31
2 Observations ,________....
100 km
Lerwlck

(a)
Key o Surface • Upper air

Fig. 2.10 (a) Map of British

Isles surface and upper air
stations. Weather ships M, L
and R are far beyond the map
edges in the direction shown.
(b) Radiosonde stations in the
Northern Hemisphere
supplying upper air data on
21 October 1979. Note that
the British Isles network
shown in (a) is one of the
densest parts of the network.
The Southern Hemisphere is
relatively empty being largely
oceanic.

32 (b)
example by the removal of ground echo), amalgamated and displayed in real time 2.10 The upper air network
largely by computer.

2.10 The upper air network

Observations of wind, temperature, relative humidity and pressure are made by

radiosondes - free-flying balloons released from the upper air stations of the
synoptic network (Figs. 2.10 and 2.11). The sondes are released at 0000 and 1200 Z
daily, and climb at about 5 m s _, until they burst between 20 and 30 km above
MSL, whereupon the instrument package returns to the surface by parachute for
possible re-use. While in flight the temperature, humidity and pressure data are
sent by radio to the ground station, and the sonde's position is monitored by auto-
matically tracking radar. The sonde has so little inertia that it moves with the

(a) (b)

Fig. 2.11 (a) A radiosonde at launch . After launch the reel at the base of the balloon (now in
the launcher's right hand) pays out a 30 m cable to keep the instrument package (now in his left
hand) clear of local balloon effects. This particular sonde (a Vaisalo RS-80) can be tracked for
wind finding by Nevoid systems, radio theodolite or radar (with a reflector immediately below
the balloon). (b) Close-up of the instrument package of the British Meterological Office's Mark 3
radiosonde, sitting as it hangs in flight. The plastic ring on top supports the very fine wire of the
electrical resistance thermometer; the open shield just beneath c.ontoins the skin hygrometer,
and the main body houses the aneroid pressure capsule, electrical battery, converting , switching
and timing devices, and radio transmitter. 33
2 Observations horizontal wind of each layer it passes through. Successive horizontal positions of
the sonde (for example at minute intervals) can therefore be made to yield a vertical
profile of horizontal wind. Since the analysis of the radio and radar data proceeds
while the sonde is in flight, the staff of the ground station can assemble and send to
the local analysis centre within a few minutes of the end of the flight the profiles of
wind, temperature and pressure up to the bursting level. Humidity data are usually
ignored above the 10 km level because sensors are unreliable in the very low
temperatures prevailing at these and higher levels. Strictly speaking, radiosondes
do not measure a vertical profile, since the sonde's flight path can be a rather
gradual slope in strong winds, but in the grossly flattened configuration of all
synoptic scale systems, even a slope of 1:10 is effectively vertical. We therefore
make no serious error in assuming the profiles to be vertical in most data analysis.
Midway in time between the full radiosonde ascents (i.e. at 0600 and 1800 Z),
balloons bearing only a radar reflector are flown and tracked to give profiles of
wind only. The alternation of these wind sondes with the radiosondes at six-hourly
intervals provides sufficient resolution in time to define the structure of the tropo-
sphere and low stratosphere associated with synoptic-scale weather and northern
hemisphere systems. Figure 2.10 shows the network of upper air stations in the
British Isles, and this again is typical of such networks in wealthy countries. In
other regions networks are much more open, and indeed they hardly exist over
substantial land areas. Upper air data from sea areas are obtained only from the
few weather ships remaining in the North Atlantic and North Pacific after
increasing cost enforced a sharp decline in numbers in the last twenty years. These
provide an invaluable but totally inadequate sampling of the 700Jo of the lower
atmosphere which lies over the oceans, and it is a major task of WWW to improve
the synoptic upper air network, or devise complementary systems. A direct
improvement is envisaged through the development of completely automatic
systems for launching radiosondes and telemetering data from suitable merchant
ships. A parallel approach which is already in extensive use is to fit large civil air-
craft with automatic sensors which will transmit pressure, temperature and wind
data while in flight. However, the temperature and wind data are not very reliable,
and are heavily concentrated at about the 300 mbar level along the major air
routes. There are currently about 500 upper air stations reporting regularly, and
the daily total of aircraft reports is well in excess of 2000.
Note that sampling intervals in space and time are much greater in the upper air
network than in the surface network. You may wonder therefore why the upper air
network is not totally inadequate given that the much finer mesh of the surface
network is only just adequate. The answer is that atmospheric structure is much
smoother and larger in scale aloft than it is near the surface, where surface inhomo-
geneities generate significantly smaller and more transient structures, especially
over land. A much more open network is therefore adequate aloft, which is just as
well given the much greater cost of making regular observations there.

2.11 The satellite network

The first meteorological satellite was launched in 1960 and immediately provided
extremely interesting and useful data. First, the huge panoramic views of the
atmosphere directly revealed and confirmed the structures of large cloudy weather
34 systems which had previously emerged only after painstaking assembly and
 2.11 The satellite network

Fig. 2.12 Visible image from polar orbiting Tiros N, showing a vast panorama of sea, snow and
cloud in the north Atlantic at 1600 Z, 11 March 1980. Cellular patterns of shower clouds spread
across a westerly flow , north and west of dense frontal cloud associated with a depression
centred near Iceland, whose north-west tip is just visible upper centre. Oblique sunlight
highlights a profusion of cloud structure on the small and mesoscales.

analysis of synoptic data. I suspect all meteorologists who were active before I 960
still have a sense of relief and awe when they see the work of the pre-satellite
generations of analysts so vividly confirmed in pictures such as Fig. I. I. Secondly
the relatively high resolution of the satellite pictures (a few kilometres compared
with the tens of kilometres of the synoptic surface network and the much coarser
scale of the upper air network) revealed an almost bewildering range of sub-
synoptic-scale structures in cloudy systems, many of which have yet to be under-
stood and incorporated in observational models. For example, the quasi-regular
patterning of shower clouds in vigorously convecting zones (Fig. 2.12) is quite
obvious on satellite pictures but is completely lost in the wide gaps between stations
in the synoptic network, and has yet to be understood in useful detail. Thirdly the 35
2 Observations satellite data were, even from the outset, much more uniformly spread over the
Earth's surface, covering the enormous oceanic and low latitude land areas which
had virtually no synoptic network.
Meteorological satellites have multiplied and developed considerably since 1960,
and the broad outline of a permanent network of satellites is probably now
established, although details may continue changing for years to come.
Basically, meteorological satellites are platforms for electromagnetic scanning
of the atmosphere from above - top-side observation in the vivid jargon. The
scanning is passive in the sense that satellites merely make use of existing radiation
emitted or reflected from the atmosphere, without adding to it as radar does -
though special radars are being tried on account of their ability to penetrate cloud.
The radiometers used are sensitive to one or more wavelength bands in the visible
and infra-red ranges. The field of view from the satellite is scanned line by line,
either by using the equivalent of a television camera, or by physically sweeping a
very narrow-field radiometer across the view. In either case the data are sent in
sequence to a receiving station on Earth for reconstitution of the whole picture.
The width of a scan line on the Earth's surface fixes the ultimate limit of resolution
since details less than a few lines across are unresolved. Most satellites operate so
that this limit is a few kilometres at a point vertically beneath the satellite; it
obviously increases with increasing obliqueness of view.
If visible wavelengths are used, the reflected sunlight picks out cloud vistas
which are slightly blurred versions of those photographed conventionally by
astronauts, and may be in full colour if at least three distinct bands of wavelengths
are used and synthesized. A major disadvantage is the inevitable blindness on the
night side of the Earth.
Radiometers sensitive to wavelengths in the far infra-red (i.e. > 3 11m) are
independent of solar radiation and respond instead to the terrestrial radiation
emitted continually by the Earth's surface and atmosphere. This increases in
intensity as the temperature of the emitting materials rises (Chapter 8) so that the
resulting picture has a brightness scale which corresponds to a temperature scale in
the original panorama. Sometimes the brightness scale is replaced by an arbitrary
colour scale for ease of subsequent analysis by eye. Because the cloud-free atmo-
sphere is nearly opaque in some parts of the infra-red spectrum, and nearly
transparent in others (Fig. 8.4), quite different pictures emerge in different wave-
lengths. For example, wavelengths to which air is transparent yield pictures of the
temperatures of the top surfaces of clouds (which are quite opaque in the infra-
red), or of the ground or sea surface if unshielded by cloud. Monochrome cloud
vjstas produced in this way resemble visible pictures if the brightness-temperature
correspondence is suitably chosen (Fig. 2.13). By contrast, pictures taken in wave-
lengths in which the air is opaque show the distribution of air temperatures in the
middle and upper troposphere. Careful selection of wavelengths covering a range
of opaqueness, together with subsequent complex analysis, can produce vertical
temperature profiles of cloud-free air with low but nevertheless useful vertical
resolution. These are potentially very valuable additions to the high resolution (in
the vertical) but otherwise inadequate traditional radiosonde measurements, and
are currently under active development.
Two distinct types of satellite orbit are in use. In the sun-synchronous type, the
satellite orbits about 860 km (i.e. one seventh of an Earth radius) above the
surface, passing near the poles but making an angle to the meridians which is just
enough to allow the orbit to remain effectively fixed relative to the sun (Fig. 2.14).
The satellite takes about 102 minutes between successive passes near one pole,
while its radiometers scan the swath of planet passing continually below it. The
36 radiometer data can be received in real time by any suitably equipped station as it is
 2.11 The satellite network

Fig. 2.13 Infra-red image of

frontal clouds over the north-
eastern Atlantic (from polar
orbiting NOAA-7 at 0853 Z, 18
October 1979). Though at this
time only the eastern part of
the panorama is sunlit, the
image is unaffected since its
brightness does not depend
on visible light. The false
brightness scale makes the
coldest parts (the high cloud
tops) look brilliant white, the
low clouds and rather cool
land surfaces grey, and the
warmest water black, much
as they would in visible light.
A narrow ridge of high
pressure over Britain is
associated with clear skies
there. An occluded front is
overflown, and any particular geographical location is most nearly overflown once approaching from the west,
every twelve hours at predictably varying clock times. In addition, data are while a wave develops on an
old cold front to the east, the
accumulated over twelve-hour periods and transmitted very rapidly to a master
two systems being connected
station to provide a complete data set. Radiometers using visible wavelengths of by a loop of low or middle-
course provide only half the coverage because of data black-out on the night side of level cloud associated with a
the planet. weak cold front south of
In the geosynchronous or geostationary type of orbit, the satellite orbits about England . There are patches of
vigorous convection east of
36 000 km (over five Earth radii) above the equator, moving in the same direction England where cold air is
as the rotating Earth. Since the orbital period at this distance is exactly one sidereal pushing eastward south of
day, the satellite hangs vertically above a fixed point on the equator (Fig. 2.14). the developing wave.
The field of view is very nearly a full hemisphere, though of course the view of the
polar regions and the western and eastern limbs is very oblique (see Fig. 1.1). A
very high-resolution radiometer scans the view like a very slow television set,
completing a scanning cycle in about 20 minutes and providing resolution which is
nearly as good as in the very much lower altitude sun-synchronous type. The data
are transmitted to one or more stations on the surface far below. In visible wave-
lengths the view disappears during the night, so that infra-red radiometers are
always included. For World Weather Watch the global coverage is made usefully
complete by having a network of five geosynchronous satellites equally spaced
along the equator. The bias of good coverage toward low latitudes is welcomed
because of the very poor synoptic network in those vast areas (remember half of the

I
I
Fig. 2.14 Satellites in
tl;lr geosynchronous and sun-
synchronous orbits. For ease
of portrayal, the
geosynchronous orbit has
been drown at o reduced
scale relative to the earth.
Note that the retrograde sun-
synchronous orbit misses the
poles by about 10° latitude.

37
2 Observations surface area of the Earth lies between 30° Nand 30° S). Geosynchronous satellites
are very well placed to detect the appearance and movement of tropical cyclones,
since these form over the warm oceans of low latitudes. Before the advent of the
meteorological satellite, even the most vigorous of these systems (the hurricanes-
section 12.4) were often detected only through chance encounter with ships,
making estimation of storm paths and landfall very unreliable.

2.12 Epilogue

A very large scale of individual and collective effort is needed to maintain World
Weather Watch, even in its present incomplete state. Shortly before every major
synoptic hour (0000, 0600, 1200, 1800 Z) tens of thousands of observers all over the
Earth are making their routine observations from the surface, and smaller numbers
are launching and tracking radiosondes. Many more people are active around the
clock in the global networks of associated telecommunication and data collection,
and at the numerous forecasting offices. The completely international directorate
and administration of the World Meteorological Organization coordinates the
activities of the national networks and deals with matters ranging from the stan-
dardization of equipment and procedure to the planning and monitoring of
improvements to WWW.
In the latter part of a century which has seen much of the optimism of our fore-
fathers overwhelmed by confusion and strife, and a comprehensive approach to
the human predicament frustrated by unenlightened sectional interest, it is heart-
warming to see what people can do globally when they really put their minds to it.
Of course the cooperation arises in the face of a fairly well-defined problem -the
need to forecast the short-term future behaviour of the lower atmosphere - and
the atmosphere is inescapably a global entity which is mercifully independent of
Man's random subdivisions of the Earth's surface. Nevertheless we should give
credit where credit is due, if only because the future of civilization depends on this
kind of low-key, practical and sincere cooperation rather than on the strident
rhetoric and divisive mayhem which is now so widely current.

Appendix 2.1 Instrumental response times

Suppose that an accurate and properly exposed thermometer has been exposed to
air of a certain steady temperature for so long that the thermometer reading 8 is
steady at the same value. Now suppose that the air temperature suddenly increases
to a different steady value T (Fig. 2.15a). The reading 8 will begin to move toward
the new value as heat flows into the thermometer. According to Newton's law of
cooling (and heating), the rate of warming reduces in proportion to the
38 temperature difference .:l between air and thermometer. Formally
 dd = -k d Appendix 2.1
(2.2)
dt
where k is the constant of proportionality which describes the speed of response of
this particular thermometer, and d is T- (}. Equation (2.2) is the equation for
exponential decay, whose solution is given in all basic mathematical texts.
(2.3)
where tis the time elapsed since the sudden change d 0 in air temperature. This expo-
nential response is graphed in Fig. 2.15a.
The lag coefficient A., or simply lag, of the thermometer is defined to be the time
taken for d to fall from d 0 to d 0 exp( -1), which is 0.3679 d 0 • At this stage the
thermometer has registered just over 630Jo of the initial jump in temperature. By
inspection of eqn (2.3) it follows that A. is equal to Ilk. The time taken for a larger
or smaller response can also be found from eqn (2.3). For example, the time ford
to fall to d/10 is 2.3026 A.. Such relationships are common to all exponential
behaviour, however it arises. Appendix 4.4 contains further examples.
The response of a thermometer to an oscillating air temperature is particularly
relevant to meteorology. The analysis begins with eqn (2.2) as before, but is com-
plicated by the continuous variation of T. Figure 2.15(b) depicts the response to a
regular wavy oscillation in T. The sluggishness of the thermometer has two effects:
the oscillation of the thermometer reading lags behind that of the air temperature,
and its amplitude is reduced. The magnitudes of these effects depend on the ratio of
A. to the time period r of the air temperature oscillation. For example, detailed
analysis of a pure sinusoidal oscillation shows that the ratio of amplitudes

rJ
(Fig. 2.15(b)) is given by

~~ = [~ + ( 2:)\ -Y, (2.4)

It follows that when r is equal to lOA. the amplitude ratio is 0.85, which means that
the thermometer underestimates the true amplitude by 15%. The corresponding
phase lag <I> (Fig. 2.15b) is about r/10. The distortion of the thermometer's
temperature pattern increases sharply with A./r, so that when this ratio is unity the
amplitude ratio is 0.16 and the phase lag is 0.22r.

Fig. 2.15 Thermometer

response to (a) a sudden rise

---
in ambient temperature, and
T (b) a regular sinusoidal

•
~
.......
" _,.""'
,.,.,..

Thermometer oscillation af ambient

I; 6o ,/'"e temperature.
.,E
Q.
/
//
....

{a) ,_
Time I

••" '
~ AT

.,
Q.
E
....
-T
(b) Time 1 39
2 Observations Although we have considered only thermometers so far, the concepts and
equations we have used apply to any instrument which responds exponentially to
changes in the quantity it measures. Even when the response is not strictly expo-
nential, when for example the instrument responds at first more quickly and then
more slowly than the exponential curve allows, the lag is usually still defined as the
time for 6307o response. The lags of anemometers and wind vanes tend to decrease
as average wind speed increases, since changes in speed and direction are commu-
nicated more efficiently by higher winds, and so it is customary to express their lags
in terms of the runs of wind needed to produce 63% response. For example, the
standard surface anemometer shown in Fig. 2.2 requires about 15 m of air to pass
its cups in order to register 63% of a sudden change in wind speed. At the lowest
wind speed to which this anemometer responds (about 2 m s - 1), this run
corresponds to a lag in time of about 8 s, whereas the lag in response to gusts in a
gale is roughly 1 s.

Appendix 2.2 The Beaufort scale

By about 1800 AD it was becoming obvious that it would be useful to categorize

wind strength in terms of its effects on the sea and sailing ships. This was done so
effectively in 1806 by Francis Beaufort while surveying for the British Royal Navy
that his scale became the standard for all wind observations until 1946, and is
still used in weather forecasts for shipping. The scale of effects in the Beaufort
scale (Table 2.1) was capable of being used so consistently by trained seamen that
assessment differences between observers of the same sea were believed to be
substantially less than one Beaufort number. Beaufort also devised a concise code
of Beaufort letters for weather description, which is still useful for weather diaries.
He is recognized as a pioneer of the movement to make meteorological observa-
tions as consistent and concise as possible.
Table 2.1 lists simplified descriptions of effects on fairly open land surfaces,
together with the equivalent ten-minutes average wind speeds measured by anemo-
meter 10 m above the land surface - an equivalence established by anemometry
long after the scale was initially devised [5].

Appendix 2. 3 International cloud

classification [ 10]

The basic division is into distinguishable types or genera, each with a latin name
and standard abbreviation. Each genus occurs so often in one of three bands of
heights above sea level, that it is assigned to that band, even though exceptions are
quite common. The larger types of cumuliform cloud, in particular the cumulo-
nimbus, are vertically so extensive that they extend to medium and high levels from
40 their low-level origins and bases, but are ascribed to low levels for simplicity.
Table 2.1 Beaufort scale of wind force Appendix 2.3

Force Specifications Equivalent mean wind

for use on land speed 10 m above ground

0 Calm; smoke rises vertically. 0 kt 0 m s- 1

Light air; wind direction shown by smoke drift, not by 2 0.8
vanes.
2 Light breeze; wind felt on face; leaves rustle; vanes 5 2.4
move.
3 Gentle breeze; leaves and small twigs moving; light 9 4.3
flags lift.
4 Moderate breeze; dust and loose paper lift; small 13 6.7
branches move.
5 Fresh breeze; small leafy trees sway; crested 19 9.3
wavelets on lakes.
6 Strong breeze; large branches sway; telegraph wires 24 12.3
whistle; umbrellas difficult to use.
7 Near gale; whole trees move; inconvenient to walk 30 15.5
against.
8 Gale; small twigs break off; impedes all walking. 37 18.9
9 Strong gale; slight structural damage. 44 22.6
10 Storm; seldom experienced on land; considerable 52 26.4
structural damage; trees uprooted.
11 Violent storm; rarely experienced; widespread 60 30.5
damage.
12 Hurricane; at sea, visibility is badly affected by >64 >32.7
driving foam and spray; sea surface completely white.

Nimbostratus often extend to both low and high levels from their middle-level
centres of gravity.

LOW-LEVEL CLOUDS (0-2 km above sea level)

Stratus (St)- Low, structureless, extensive sheet of cloud, often just deep enough
to produce drizzle or snow grains at the surface.
Stratocumulus (Sc) - Similarly shallow cloud sheet, broken into more or less
regular cumuliform masses, giving drizzle or snow grains at the surface if cloud
base is not too high.
Cumulus (Cu) - Detached hill-shaped clouds with flat bases which may be at a
uniform level across any particular population.
Cumulonimbus (Cb) - Large cumulus, usually showering. Their considerable
vertical extent ensures that bases are quite dark and upper parts often reach to high
levels. The lower and middle levels of any particular Cb look like a conglomeration
of cumulus, but the high levels often consist largely of ice cloud and as a result look
like dense cirrus (see below).

MEDIUM-LEVEL CLOUDS (2-4, 7, 8 km in polar, temperate and tropical regions

respectively

Altocumulus (Ac) - Shallow sheet broken into more or less regular, rounded
patches or rolls, sometimes with virga (see below). Distinguished from Sc by 41
2 Observations smaller apparent size ( < 5), and from cirrocumulus (see below) by the presence of
shading.
Altostratus (As) - Largely featureless extensive sheet of watery looking cloud,
often thick enough to make the Sun appear as if seen through thick ground glass.
Nimbostratus (Ns) - Extensive sheet of cloud, usually precipitating. The con-
siderable depth blots out the sun, moon and stars, making days dull and nights
pitch black.

HIGH-LEVEL CLOUDS (3-8 km in polar, 5-14 in temperate and 6-18 in tropical

regions)

Cirrus (Ci)- Detached, white fibrous clouds, often with a silky sheen.
Cirrocumulus (Cc)- Shallow patch or sheet, broken into more or less regular,
unshaded and apparently very small blobs or ripples, which are usually fibrous in
places.
Cirrostratus (Cs) - Shallow extensive sheet of white cloud, thin enough to be
largely transparent and produce haloes round the sun and moon. Cs may look
fibrous or smooth or both.

The genera may be amplified by adding a description name quite misleadingly

called a species, exemplified by the following selection.
Fractus (fra)- Broken or ragged (of Stand Cu).
Lenticularis (len)- Elements shaped like a lens or almond (of Sc, Ac and Cc).
Humilis (hum)- Of only slight vertical extent (of Cu).
Congestus (con) - Growing markedly, often by bulging like the heart of a
cauliflower (of Cu).
Capillatus (cap)- Distinctly fibrous upper parts (of Cb).
The addition of a list of varieties, such as radiatus (banded), supplementary
features such as virga (fibrous precipitation trails tapering down from cloud base
to extinction above the surface), accessory clouds such as pileus (thin, smooth webs
stretched over the tops of Cu con or Cb), and the new category of mother-cloud
(producing other clouds also present) completes a scheme some of whose details
suggest an unhappy marriage of observable reality and Linnaean taxonomy.

Appendix 2.4 Present and past weather

Table 2.2lists the categories of present weather used internationally by observers.

Each is given a number, and in any observation only the highest applicable number
is reported. The descriptions have been simplified for present purposes, since the
full version is rather legalistically worded to avoid ambiguity. Note that 'freezing'
means freezing on impact with solid surfaces.
Table 2.3 similarly lists the categories of past weather, which by definition refers
to the 6-hour period before each of the principal synoptic hours: 0000, 0600, 1200
42 and 1800 Z.
Table 2.2 Present weather categories Appendix 2.4

00-19 No precipitation at station at time of observation

00 Cloud development unobserved during past hour
01 Clouds generally dissolving during past hour
02 Clouds generally unchanged during past hour
03 Clouds generally developing during past hour
04 Visibility reduced by smoke
05 Haze
06 Widespread dust
07 Dust or sand raised by local winds, but not by dust or sandstorm or whirls
08 Dust or sand whirls seen in last hour, but no dust or sandstorm
09 Dust or sandstorm seen at or near station in past hour
10 Mist
11 Shallow patchy fog or ice fog
12 Shallow continuous fog or ice fog
13 Lightning but no thunder
14 Precipitation within sight not reaching surface
15 Precipitation reaching the surface in distance
16 Precipitation reaching surface near but not at station
17 Thunderstorm but no observed precipitation
18 Squalls seen at time of observation or during past hour
19 Funnel cloud(s) seen at time of observation or during past hour

20-29 precipitation, fog or thunderstorm at station in post hour but not at time of observation
21 Drizzle (not freezing) or snow grains not in showers
22 Rain (not freezing) not in showers
23 Rain and snow or ice pellets, not in showers
24 Freezing drizzle or freezing rain
25 Rain showers
26 Rain and snow showers or snow showers
27 Hail and rain showers or hail showers
28 Fog or ice fog
29 Thunderstorm

30-39 duststorms, sandstorms, drifting or blowing snow

40-49 fog or ice fog at time of observation

S0-59 drizzle at time of observation

50 Not freezing, intermittent, slight at time of observation
51 As 50 but continuous
52 As 50 but moderate at time of observation
53 As 52 but continuous
54 As 50 but heavy at time of observation
55 As 54 but continuous
56 Slight freezing drizzle
57 Moderate or heavy freezing drizzle
58 Slight drizzle and rain
59 Moderate or heavy drizzle and rain

60-69 rain at time of observation: as S0-59 but with drizzle replaced by rain, and rain (in 58 and
59) replaced by snow

70-79 solid precipitation not in showers

70 Snow flakes, intermittent, slight at time of observation
71-75 Inclusive: pattern is in 51-55
76 Ice prisms with or without fog
77 Snow grains with or without fog
78 Isolated starlike snow crystals with or without fog
79 Ice pellets

80-90 showery precipitation

80 Rain showers, slight 43
2 Observations Table 2.2 continued

81 Rain showers, moderate or heavy

82 Rain showers, violent
83 Rain and snow showers, slight
84 Rain and snow showers, moderate or heavy
85 Snow showers, slight
86 Snow showers, moderate or heavy
87 Slight showers of snow pellets, whether or not encased in ice, with or without showers of
rain or rain and snow
88 As 87 but moderate or heavy
89 Slight hail showers, with or without rain or rain ond snow, without thunder
90 As 89 but moderate or heavy

91-94 current precipitation with thunderstorm in post hour

91 Slight rain
92 Moderate or heavy rain
93 Slight snow, or rain and snow, or hoil
94 Moderate or heavy snow, or rain ond snow, or hoil

95-99 current precipitation and thunderstorm

95 Slight or moderate storm without hail but with rain and/or snow
96 Slight or moderate storm with hail
97 As 95 but heavy storm
98 Storm with sandstorm or duststorm
99 Heavy storm with hail

Table 2.3 Post weather categories

0 cloud covering half or less of the sky throughout the period

1 cloud covering more than half of the sky for port of the period ond less thon this for the rest
2 cloud covering more than half of the sky throughout the period
3 sandstorm, duststorm or blowing snow
4 visibility less than 1 km because of fog, ice fog or thick hoze
5 drizzle
6 rain
7 snow or mixed rain and snow
8 showers
9 thunderstorm with or without precipitation

Problems

LEVEL 1

1. A manufacturer seeks your advice on plans to build a transparent Stevenson

screen. Advise him rationally and politely.
2. Why should a Stevenson screen have a floor?
3. The vapour content of a certain room full of air is nsmg, though its air
temperature and pressure are constant. Consider in turn whether each of the
following should be rising or falling: vapour density, relative humidity, wet-
bulb depression, dew-point depression.
4. What are the azimuths of the following winds: E, SE, S, NNW (i.e. half-way
44 between NW and N)?
 Problems

Fig. 2.16 Altocumulus-

visually distinguishable from
stratocumulus because its
greater height makes the
patchiness (caused by
overturning within the layer)
look smaller.

5. If you read a mercury barometer without correcting for a too high mercury
temperature, will you over- or underestimate the atmospheric pressure?
Mercury density decreases with increasing temperature.
6. Use appendices 2.2, 2.3 and 2.4 to assess current wind strength, cloud types
and present and past weather at some convenient daytime and place.
7. Ideally how high should be the mast of a fully instrumented meteorological
ocean buoy?
8. List all the ways of getting information about distributions of temperature,
cloud and precipitation throughout the troposphere.

LEVEL 2

9. Suppose that a pitot tube anemometer gives an output of 1 unit in a wind speed
of 5 m s- 1 • What maximum output should be allowed for in its design to
enable it to register the presence of a hurricane?
10. Careful work with a water-filled barometer shows that atmospheric pressure
falls by about 0.8 em of water for each 10 m rise near sea level. Use realistic
values for g and p in eqn (2.1) to show that the pressure fall is about 78 Pa
(0.8 mbar).
11. The rain collected by a 127 mm (5 in) diameter rain gauge on a certain occasion
filled the 25.4 mm diameter measuring cylinder to an actual depth of 300 mm.
Find the true depth of rainfall, and the consequent rise in water level, if half of
this rain falling on a 100 km2 catchment is collected by a 500 m square
reservoir with vertical sides.
12. The surface of the Dead Sea lies 392 m below MSL. If the air density is
assumed to be 1.2 kg m- 3 , what correction must be applied to the station
pressure at Dead Sea level to find the associated MSL value?
13. What must be the time of launch of a radiosonde so that its flight to the 20 km
level (from MSL) is completed by 1200 Z? Given that for the first half of this
period it is borne by a 20 m s -I westerly wind, and for the second half it is
borne by an 80 m s -I southwesterly, find the horizontal distance and direction
of the final sonde position from its launching site. Such extreme translations 45
2 Observations give the Japanese meteorological service problems when the subtropical jet
stream is blowing strongly overhead.
14. The lag of a certain moderately fast response thermometer is 3 s. What is the
error of the thermometer 10 s after an instantaneous ambient temperature rise
of 3 °C?

LEVEL 3

15. In strong sunlight, sunshine often seems more intense to the human senses
inside a greenhouse than in the open air. Actually the reverse must be the case.
Explain.
16. Reconsider problem 10 to show that the density of water is 1250 times the
implied density of air, regardless of the value of g. Consider what would be
observed if the same experiment were conducted inside a station on the Moon
filled with air at normal terrestrial pressure and temperature. Assume the
station to be on the lunar surface where the local g is 0.17 of the terrestrial
value.
17. In cool windy weather, large rain drops on the outside of a singly glazed
window are often accompanied by mist patches on the warm inside of the
window. Explain.
18. In the absence of a dense gauge network, the spatial variation of rainfall round
small objects such as hedges and houses can only be guessed. However, typical
patterns of snowfall should provide some evidence. Consider factors which
support and others which reduce the value of such evidence.
19. In section 2.3 it is suggested that a small thermometer responds rapidly merely
because of its small heat capacity. This is too simple, since its small size must
tend to slow the response by limiting the heat fluxes between the thermometer
and the ambient air. Argue the case more thoroughly for the simple case of a
spherical thermometer to show that the speed of response is likely to be
inversely proportional to the bulb diameter.

46
CHAPTER
The constitution of the
3 atmosphere

3.1 The well-stirred atmosphere

It is found that up to about 100 km above sea level the atmosphere consists largely
of a mixture of gases in remarkably uniform proportions, together with a small and
variable quantity of water vapour concentrated almost entirely in the troposphere.
We will call the uniform mixture dry air to distinguish it from moist air which
includes water vapour as well. In addition to these gases there are small quantities
of water and ice in the form of clouds and precipitation, and a population of even
tinier particles known as aerosol particles (strictly the term aerosol refers to the air
and the particles), all heavily concentrated in the troposphere. And there are traces
of many other gases, of interest mainly to atmospheric chemists. Figure 3.1
represents the vertical distributions of several of the most important of the
materials just mentioned, together with other distributions and processes to be
mentioned in this chapter.
Over 99.90Jo of the mass of dry air in the atmosphere consists of a mixture of
molecular nitrogen, oxygen and argon in the proportions shown in Table 3.1 -
proportions which vary only minutely up to the 100 km level. Of course the
absolute amount of any constituent (in kilograms per cubic metre for example)
diminishes very rapidly as we ascend through this large height range, but so does
the density of the air (see section 4.5), with the result that the ratio of the two is
nearly uniform. Let us call this ratio the specific mass of the constituent, and
examine the significance of the uniformity of specific mass of some of the most
important constituents.

DIFFUSIVE AND CONVECTIVE EQUILIBRIA

The atmosphere is not a static fluid whose properties are to be explained in terms of
some dynamic origin long since past; it is incessantly dynamic and active, and yet
continually maintaining a nearly steady state. Therefore its properties, including
the uniform distributions now in question, must arise from a balance between
continuing competing processes. Let us distinguish between processes promoting 47
3 The constitution of the
atmosphere

Thermosphere

100
1()6 0.001
......
...
"E as
so. 10- 5 0.01 .c
"'E
~
'"' I E
Qj
>
0.1
...:::l
Q)

~ Mesosphere (/J
as 60 (/J
Q)
(/J ...a.
Q)

Q)
Stratopause -~ 10- 3
> 4: c:: 0
·;:
0 Q)
.c 40 "0 Q)
Stratosphere ._o
Q)"'
as .t:
CD -
~ 10-2 a.
Q) ..c::"'
Q.Z 10 (/J
Fig. 3.1 Schematic vertical "0 0
VJE
... E
:::l
distributions of atmospheric .&e;
temperature, pressure,
density and atmospheric
:!:::
c(
20
Tropopause
...:!::
:II<:
~;:, 10- 1 100
<
material. Note the large
seasonal variation in the Troposp~e~~-r-
temperature of the high- 0 1000
latitude mesopause, induced 0 20
-80 -60 -40 -20
by variation of direct input of
solar hard UV. Temperature (OC)

uniformity and those promoting variability. Clearly the former must be dominant
in the cases of the uniformly distributed constituents of dry air.
It is fairly obvious that mixing tends to promote uniformity. For example an
uneven distribution of ink in a beaker of water is replaced by a uniform distribution
almost as soon as stirring begins, whether by a stirring rod or convection from a
heated base; and the uniformity is maintained by the stirring for as long as it
continues. Uniformity would also result from the intermingling of ink and water by
molecular diffusion, even if there were no stirring, but the equilibrium state would
take months rather than seconds to achieve, because molecular diffusion is so very
much slower than bulk stirring.
In domestic circumstance, diffusion in gases is much faster than diffusion in
liquids, because of the much freer molecular motion, but it is still easily swamped
by stirring if the dimensions of the gas volume exceed a few millimetres.
Atmospheric dimensions are enormously larger than this, with the result that
diffusion is relatively ineffective by comparison with stirring on all but the smallest
scales. At least this is so in the lower atmosphere; at great heights the air is so
tenuous that molecules can move considerable distances between consecutive
collisions, and since they move at about the speed of sound, the speed of diffusion
can be very high indeed.
In a room-sized volume of air, mixing by either stirring or diffusion tends to even
out all distributions, just as in the case of ink in water. However, if the volume of
air is at least 1 km tall, gravity will produce a significant lapse of air density from
bottom to top, and the two types of mixing tend to produce different vertical
distributions.
In the case of diffusion in a tall air column, each molecular species settles into an
equilibrium distribution in which the downward drift of its molecules under the
48 pull of gravity is balanced by the upward net diffusion maintained by the upward
Table 3.1 Composition of the turbosphere 3.1 The well-stirred
atmosphere
Gases
Specific Molecular Specific gas
Component mass weight constant

Dry air
N2 0.755 28.02 297
02 0.231 32.00 260
Ar 0.013 39.95 208
C0 2 0.0005 44.01 189
plus hundreds of trace gases

Moist air
Dry air 0.997 28.96 287
H20 vapour "'0.003 overall 18.02 461
( -0.03 maximum)

Particles and droplets

Cloud -5 x 10- 5 overall Water and ice
( -0.004 maximum)
Precipitation -3 x 10- 6 overall Water and ice
( -0.004 maximum)
Aerosol 10- 8 overall Various solids or concentrated
( -10- 7 maximum) solutions in water

lapse in numbers of those particular molecules per unit volume (number density),
regardless of any other molecular species present. Heavier molecules diffuse
toward an equilibrium with a steeper lapse of number density than do lighter
molecules, and in fact each molecular constituent achieves a distribution defined
by its own particular scale height, which is inversely proportional to its molecular
weight (section 4.3). In such diffusive equilibrium under gravity, therefore, the
specific masses of the heavier molecular species are greater at low altitudes, whilst
those of the lighter molecular species are greater at high altitudes. This is essentially
the state of things in the atmosphere above the IOO km level, where diffusion is
rapid in the tenuous air and stirring is relatively weak.
In the case of a tall column of air in which mechanical or convective stirring
predominates over diffusion, the different molecular species are being inter-
mingled much faster by the random movement and mixing of relatively large eddies
than they are being separated by molecular diffusion under gravity. Consider the
life cycle of an eddy from the time of its formation from the debris of precursor
eddies to the time of its destruction by mixing. In that brief time (- I 0 sin the case
of small-scale turbulence) the parcel may move - 10 m, conserving its original
mixture of molecular species. The combination of conservative motion and
sharing-through-mixing, endlessly repeated throughout the stirred volume for as
long as stirring persists, leads to the uniformity of specific masses which is charac-
teristic of convective equilibrium - the term applied to stirred equilibrium
maintained by mechanical or convective stirring or both. Since the eddy parcel
expands on differential rising through the ambient pressure lapse, and is
compressed on sinking, the vertical layering of molecular species in diffusive
equilibrium is replaced in convective equilibrium by the vertical layering of the
density (section 4.5) of the otherwise uniform mixture.
Clearly the uniformity of the specific masses in Table 3.I strongly suggests that
the great bulk of the atmosphere is in convective rather than diffusive equilibrium.
The region in which this dominance prevails is known as the turbosphere, and it
extends from the Earth's surface to an ill-defined turbopause at about the 100 km 49
3 The constitution of the level, above which diffusive equilibrium prevails. Stirring in the turbosphere is
atmosphere maintained by atmospheric motion of all types and scales, including the great
weather systems of the troposphere, as well as the small-scale turbulence cited
above.
So far we have considered convective equilibrium distributions of gaseous
constituents only. However, much the same balance between gravitational
separation and turbulent mixing tends to apply to dust particles etc. whose fall
speeds through still air are much smaller than typical vertical speeds of parcels in
the stirring air. Particles with larger fall speeds tend to concentrate more at lower
levels, the gradients of their specific masses obviously increasing with particle fall
speed and decreasing with increasing vigour of atmospheric stirring. In fact the
situation rather resembles the distribution of gaseous species in diffusive
equilibrium, even though the condition may be one of convective equilibrium with
respect to gas molecules. This kind of quasi-diffusive distribution is observed in
some of the range of atmospheric particles discussed in sections 3.6 and 3. 7.
The distributions discussed so far result from the homogenizing effects of
diffusive and convective mixing and the selective effects of gravity acting on the
various constituents. There are of course other selective processes at work in the
atmosphere, including most obviously a great range of chemical reactions. In many
cases, such as most of those in Table 3 .I, the selective effects are not strong enough
to compete with atmospheric mixing in the short term, but in some cases they are
able to compete at least marginally, and in many they have profound effects in the
long term. In the rest of this chapter we consider some important examples of the
distributions of atmospheric gases and particles (including droplets), and of
the ways in which these arise from competition between mixing on the one hand,
and gravity and other selective effects on the other.

NEAR AND ABOVE THE TURBOPAUSE

Above about 100 km, diffusive equilibrium produces marked stratification of

constituent gases according to molecular weight, with the heaviest components
concentrating at the lowest levels. Lighter gases therefore predominate aloft,
though the concentrations of hydrogen and helium (the two lightest and therefore
the two with the fastest-moving molecules) have been reduced far below their
cosmic abundances by persistent leakage to space throughout the Earth's history.
As height increases there is increasing photodissociation of molecules into their
constituent atoms, and this considerably affects the composition of air above
about the 80 km level. For example, above about 120 km the reaction
0 2 + photon (solar UV) --> 20
maintains more than half of the population of oxygen atoms and molecules in the
form of atomic oxygen (0), and in so doing strongly absorbs solar ultraviolet with
wavelengths between 0.1 and 0.2 micrometres (J.tm), thereby contributing to the
warming which maintains the thermosphere (Fig. 3.1). Incidentally the atomic
oxygen acts like a gas with half the molecular weight and therefore half the density
of molecular oxygen (0 2), so that this widespread photodissociation significantly
reduces the density of air at these levels.
Photoionization also increases with altitude. For example the reaction
0 + photon (solar UV) --> Q+ + electron
50 maintains a population of ionized oxygen atoms and free electrons, absorbs solar
ultraviolet with wavelengths less than about 0.1 p.m, and helps warm the ambient 3.2 Dry air
air over a wide height range. Actually there are sufficient electrons maintained by
photoionization in the upper half of the turbosphere to justify its description as the
lowest part of the ionosphere. Its electrical conductivity is sufficient to reflect radio
waves emitted by radio transmitters on the Earth's surface, which is very important
for telecommunications since radio propagation around the spherical Earth would
be impossible otherwise. The specific masses of ions of all types increases sharply
with height above the turbopause, and at altitudes greater than about 150 km they
significantly affect the movement of air in the presence of the Earth's magnetic
field. The atmosphere at these and higher levels is a diffuse plasma with properties
which are very different from those of the electrically nearly neutral lower
atmosphere [11].

3.2 Dry air

Because of the rapid fall of air density with increasing height only one millionth of
the mass of the atmosphere lies above the turbopause; the rest lies in the well-stirred
turbosphere and consists almost entirely of dry air, and in particular of molecular
nitrogen (N2), molecular oxygen (OJ and Argon (40Ar), as shown in Table 3.1. Let
us consider these major gases in order of increasing chemical activity.
Argon is chemically inert and is therefore the simplest of the major gases in its
chemical behaviour. Though stirred ceaselessly through the turbosphere it is not
otherwise affected in the short term, so that the uniformity of its specific mass
throughout the turbosphere, despite its having the heaviest molecules of the major
atmospheric gases, is a measure of the predominance of convective over diffusive
equilibrium prevailing in the turbosphere.
In the long term the picture is less simple. The present very considerable
abundance of argon consists almost entirely of the isotope 40Ar, which has
accumulated throughout Earth's history by radioactive decay of potassium (40K)
within the solid bulk of the planet and subsequent gradual diffusion to the surface
(section 3.8). However, the rate of accumulation of 40Ar is so slow in comparison
with the rate of its dispersion throughout the turbosphere by mixing, that any
resulting excess of specific mass near the surface would be infinitesimal. The half-
life for radioactive decay of 40K is 1.3 x 109 years - nearly one third of the age of
the Earth.
Nitrogen in its molecular form accounts for nearly three quarters of the mass of
the atmosphere. Although chemically not very active in terrestrial conditions it is
by no means inert, and is more heavily involved in the biosphere than its early name
Azote (lifeless) suggests. Atmospheric N2 is fixed (chemically combined with other
substances) and removed from the atmosphere in three main ways whose
magnitudes have been estimated in detailed studies over recent decades [12].
Expressed in billionths (10- 9) of the mass of the atmospheric reservoir of N2 per
annum, these removals were as follows in 1968:
1. fixation by biological micro-organisms, mainly on land 14
2. fixation by lightning and other ionizing processes in the atmosphere 2
3. industrial fixation, mainly to make artificial fertilizers 8
Before man's intervention, the annual fixation was presumably balanced by the 51
3 The constitution of the
atmosphere

Fig. 3.2 A high-atmosphere

research balloon soon after
launch. The instrument
package contains special
radiometers to detect water
vapour and other trace gases
in the stratosphere and
mesosphere, and hongs about
15 m below the bose of the
slack balloon envelope. As
the balloon ascends, the
bubble of helium expands,
fully filling the spherical
envelope at the flotation
altitude of about 50 km.

annual denitrification (return to N2) by anaerobic bacteria. However, the industrial

fixation and the four units of the biological fixation which arise from agricultural
cropping of legumes, both of which have developed largely in the present century,
are believed to be making current annual denitrification rates fall about 2 units
short of the total fixation rate.
Consider the consequences of annual fixation and denitrification of about 24
billionths of the atmospheric reservoir of N2 , ignoring for the moment the probable
slight imbalance between the two . On average a particular N2 molecule must spend
about 42 million years in the atmosphere between one denitrification and the next
fixation. This is the residence time for nitrogen in the atmospheric branch of the
52 nitrogen cycle, found by dividing the mass of the reservoir by the rate of influx or
efflux (Appendix 3.1). Although any particular molecule may move through the 3.2 Dry air
atmosphere very much more quickly or slowly than this, the residence time is a
useful concept in this and many other studies of nearly balanced cyclic systems.
In the present discussion, although fixation and denitrification rates are very
impressive when expressed in tonnes per annum, the residence time for
atmospheric N 2 is so long that stirring of N2 throughout the turbosphere is almost
instantaneous by comparison. We should therefore expect the specific mass of N2
to be very uniform throughout the turbosphere, as indeed it is. The distributions of
N 2 and 40Ar are therefore similar in this respect.
Note that if the current slight excess of fixation over denitrification were
maintained for 250 million years, half the N2 would be removed from the atmo-
sphere. However, long before this stage could be reached, the relatively small
reservoir of nitrogen in organic debris would be grossly inflated, producing oxygen
depletion and blooms of algae in rivers and seas. In fact the problem is much more
immediate than this would suggest. Industrial fixation may well treble in the last
thirty years of the twentieth century, and there have already been many localized
blooms in rivers and lakes affected by run-off from intensively fertilized land
nearby.
Oxygen is chemically the most reactive of the major atmospheric gases. In fact it
is so reactive that at first glance it is surprising that molecular oxygen (OJ should
comprise nearly one quarter of the mass of the atmosphere. However, virtually all
inorganic terrestrial materials are already fully oxidized, so that the residual
reactions are mainly with the continuously forming organic materials.
0 2 is consumed by many types of respiration through which living organisms
produce energy by highly regulated oxidation of food, and by decay, in which
bacteria similarly act on the food stored in the corpses of dead organisms. Detailed
estimates [13] suggest that the annual consumption is about 200 units, where a unit
is about one millionth (10- 6) of the mass of 0 2 in the atmosphere. This is nearly
balanced by the release of 0 2 as a by-product of photosynthesis in the presence of
sunlight by green plants on land and by phytoplankton in water. Photosynthesis is
the process by which these organisms build themselves from carbon dioxide and
water, using some of the energy in sunlight

C0 2 + H 20 + photon (solar visible)--+ CH20 + 0 2

where CH 20 represents the sugars which are the basic foods for the photo-
synthesizing organisms.
0 2 is also consumed by combustion of all sorts, from forest and bush fires to the
coal and oil-fired furnaces of industry, but the total consumption is currently
about 10 units annually, which is only about 1120 of the consumption by
respiration. Clearly the residence time of 0 2 in the atmosphere is dominated by
photosynthesis, respiration and decay, and the quoted values for reservoir and
throughput imply a residence time of about 5000 years. This is much shorter than
the residence time for N2 (proving oxygen's much greater reacivity), but it is still
very long compared with the speed of stirring of the atmosphere, so that we should
expect the specific mass of 0 2 to be very uniform throughout the turbosphere,
which is almost always the case. A couple of exceptions prove the rule: in the
immediate vicinity of a large bush fire, or a runaway fire storm in a bombed city,
the rate of consumption of 0 2 by combustion may be so high as to cause temporary
local depletion.
On account of its chemical reactivity, the present large amounts of atmospheric
0 2 cannot have survived the Earth's convulsive origin; and for the same reason no
0 2 is to be found in gases vented by volcanoes. Almost all atmospheric 0 2 is 53
3 The constitution of the believed to have arisen instead by means of a slight excess of photosynthesis over
atmosphere
respiration and decay which has persisted throughout much of the long develop-
ment of life on Earth, and which has been especially marked in the almost explosive
development of life in the last few hundred million years (section 3.8). This excess
has probably been maintained by the burial of organic debris before completion of
decay. Radioactive tracing shows that of the two oxygen atoms produced in each
photosynthetic reaction, one comes from water input and the other comes from the
carbon dioxide. There are very large reservoirs of each on the Earth, both largely in
the oceans (which contain over 50 times more carbon dioxide than the atmosphere,
in dissolved form), though of course the small quantities of each in the atmosphere
are crucial for the land-based part of the biosphere. Using these reservoirs and the
power of sunlight, life maintains the reservoir of free oxygen on which it now
depends.
The current phase of man's wholesale combustion of fossil fuels, which cannot
last more than a few hundred years even in the case of the most abundant reserves
(coal), represents the very sudden completion of decay left incomplete by
organisms which began life hundreds of millions of years ago. The oxygen freed
very gradually by the action of ancient sunlight is now being suddenly reclaimed,
with consequences which are barely understood, except that they will probably be
felt more immediately in the reservoir of free C0 2 in the atmosphere and oceans
than in the much larger reservoir of 0 2 (section 3.4).

3.3 Ozone

A minute fraction of total atmospheric oxygen is maintained in the form of ozone

(0 3) by photochemical reactions involving solar ultraviolet. A maximum of
specific mass of 0 3 is maintained in the stratosphere between about the 20 and
40 km levels by a series of photochemical and chemical reactions:
02 + photon (solar UV) ..... 20
02 + 0 +M --> 03+ M
03 + photon (solar UV) ..... 0 2 + 0
0 + 03 ..... 202
The first is the one already mentioned in section 3.1 as being very important at even
higher altitudes. It maintains a proportion of oxygen in the form of atomic oxygen
which is very small in the upper stratosphere and almost zero at lower levels. In the
second reaction atomic and molecular oxygen combine to form ozone. The third
body M of the triple collision can be any other gas molecule, and is needed to take
away surplus energy and so prevent the newly formed ozone from flying apart
immediately because of its initial excitement. Once formed, the ozone is a powerful
absorber of soft solar ultraviolet (wavelengths between about 0.2 and 0.3 ~m)
through the photodissociation described in the third reaction. Some of the atomic
oxygen produced in this way combines with ozone to reform molecular oxygen in
the fourth reaction.
This series of reactions maintains a maximum number density of ozone
molecules (i.e. number per unit volume) between the 20 and 30 km levels, and
54 maximum specific masses - 10 km higher. The maximum is maintained in a
 3.3 Ozone

a;
>
.!!
.. 40
.,
Q)
0

'
Q)
>
0
\
~ 30
:E
\03
I
·;' ·../ / . 3 body collision
Fig. 3.3 Schematic vertical
J: 20 ..J.__...:.......:.:;:;_ _ _ _ _ __ profile of ozone concentration
and mechanisms maintaining
Con centra tion/frequenc y its stratospheric maximum.

region where the first and fourth reactions balance to maintain a small but not
negligible proportion of oxygen in the form of odd oxygen (0 and/ or 0 3) as distinct
from molecular oxygen (02). The second and third reactions then maintain a
dynamic equilibrium between 0 3 and 0, which favours 0 3 in the height region
specified because it is high enough to have significant 0 and yet low enough for the
triple collision to be reasonably frequent (Fig. 3.3).
The 0 3 maximum maintained in this somewhat tortuous way is only a very small
proportion of the atmospheric mass at these levels, maximum specific masses being
- I0- 5 • And yet this apparently remote and slight proportion is extremely
important to most living things, because the soft ultraviolet which it very effec-
tively absorbs is potentially very damaging to exposed living tissue. Although
individual photons of the soft ultraviolet are less energetic than those of the hard
ultraviolet, they are much more numerous than the latter, being nearer in wave-
length to the peak of the solar spectrum (Fig. 8.1). If the apparently insignificant
veil of ozone were to be removed, then all living things not shielded by several
metres of water or its equivalent would die almost immediately of catastrophic
sunburn as their weak molecular bonds were disrupted by the destructive deluge of
soft ultraviolet. The development of this literally vital filter is obviously related to
the growth of atmospheric 0 2 , and through this to the development of the
biosphere which it largely protects, and an understanding of these complex
relationships is central to current attempts to trace the origin and development of
life on Earth (section 3.8).
The maintenance of the 0 3 maximum in the stratosphere demonstrates the
presence of selective reactions which can act quickly enough to compete with the
moderately vigorous ·stirring at those levels. The competition is quite nicely
balanced, however, and ozone would be strongly concentrated in the low-latitude
stratosphere, especially in winter, if meridional air currents did not carry it
poleward from the low-latitude formation zones.
In recent years, observations over most of Antarctica have shown a sharp
temporary fall in the amount of stratospheric ozone in late winter and spring
(September to December- Fig. 3.4). Although the detailed chemistry is difficult,
it is now widely accepted that this ozone 'hole' is being produced by domestic and
industrial CFCs (chlorofluorocarbons used in refrigerators and some aerosol
cans). These gases are vented mainly from the heavily populated northern
hemisphere, and are gradually mixed throughout the troposphere and into the
stratosphere. Although CFCs are manufactured because they are chemically stable
in the situations where they are used, they can break down in the stratosphere to
release chlorine, which is potentially very active chemically. The release of chlorine
is encouraged in the low Antarctic stratosphere in winter, because a thin haze of
ice-cloud particles forms in the unusually cold conditions there. As winter begins, 55
3 The constitution of the
atmosphere
1987
10
25
E
.:.:.
_,,----'' 20 ..J
:; rJ)

.0 J' :::!!
E { 13 OCT Q)
>
I 0
....
Q) I 15 .0
100 \
'
::J
V>
V>
-......'•
\
<II
.....
\ .s:::.
a:
Q)
0)
10 ·a;
J:
Fig. 3.4 Observed profiles of
0 3 content above Holley 5
{75° S, 28° W) Antarctica in
1987, during the heavy
springtime depletion {13 1000
October) and after summer 0 100 200
recovery {15 August). {After
[14]) Ozone partial pressure (nbarl

the low stratosphere settles into a huge closed westerly circumpolar vortex,
spinning and cooling in relative isolation from the surrounding atmosphere.
Chemical reactions occur on the myriads of tiny ice surfaces and release relatively
large quantities of molecular chlorine (CI2) which is photodissociated to the very
active atomic chlorine (CI) on the return of weak sunlight in spring. Ozone is then
consumed at several percent per day by a series of fast catalytic reactions. The
destruction ceases when the cold circumpolar vortex breaks down in early summer,
as intrusions of air from lower latitudes bring heat and ozone from lower latitudes.
In view of its importance for life on Earth (section 3.8), there is great concern
over any reduction of the Earth's ozone shield. Though Antarctica is virtually
uninhabited, and the loss there is only seasonal, it represents a net loss of global
stratospheric ozone, and there are some signs of a smaller seasonal loss beginning
to appear in the North polar regions, though the winter circumpolar vortex is less
isolated there. In principle too, losses could escalate as local solar absorption
reduces with the ozone, perpetuating the cold conditions on which the loss
depends. It is therefore quite proper that the discovery of the ozone hole has
triggered international efforts to stop the manufacture of CFCs, and has played a
major role in bringing environmental matters up the political agenda nationally
and internationally.
Note that ozone and CFCs are significant greenhouse gases (section 8.5), and
that ozone is produced sporadically in the low troposphere in photochemical smogs
(the Los Angeles type - section 11.5), where its action as a strong oxidization
agent makes it damaging to plant and animal tissues.

56
3.4 Carbon dioxide [15] 3.4 Carbon dioxide

Carbon dioxide (C0 2) interacts with the biosphere in ways which complement the
0 2 reactions, being produced by combustion and respiration and consumed by
photosynthesis. Indeed the mass fluxes of C02 exceed the fluxes of 0 2 specified in
the previous section in precisely the ratio of their molecular weights (44/32).
However, the mass of C02 in the atmosphere is so much smaller than the mass of 0 2
(Table 3.1) that the residence time of C0 2 in the atmosphere between successive
involvements with the biosphere is only about five years. This time is so short that
we might expect to observe temporary localized gradients of the specific mass of
atmospheric C02 , as is indeed the case. The situation is complicated somewhat by
continual exchange with the relatively very large reservoir of C02 dissolved in the
oceans , which takes place at a considerably slower rate than the exchange with the
biosphere.
Photosynthesis can proceed so rapidly in sunlit vegetation, especially when the
vegetation is dense and well-watered, that C02 levels in the immediate vicinity may
fall 200Jo below average values. At night, respiration may raise C02 levels as far
above the average. The large diurnal variation from maximum depletion to
maximum production of C0 2 overcomes the mixing capacity of the local air, which
is considerably reduced in dense vegetation, to produce substantial but highly
localized diurnal variations in C02 concentrations.
A similarly cyclic imbalance produces the substantial seasonal variations in C0 2
levels in the troposphere (Fig. 3.5). The burst of photosynthesis by land plants in
middle latitudes in spring and early summer consumes more C0 2 than is released by
respiration; and the reverse happens in autumn and early winter when respiration
and decay exceed photosynthesis. The effect is particularly pronounced in the
northern hemisphere because of the concentration of land masses there with large
seasonal climatic range.
The weak oscillation observed in Antarctica with a six-month phase shift
(Fig. 3.5) gives a clue to the speed of mixing of C02 across the globe in the low
troposphere- a speed which must apply to all unreactive and moderately reactive
gases since mixing rates are independent of the type of gases. On the one hand the
speed of mixing is not so great that the dominant northern hemisphere maximum
appears even in the Antarctic. But on the other hand the speed is sufficient to carry
the southern hemisphere oscillation from its source in middle southern latitudes to
the biological desert of Antarctica. This suggests that in the troposphere,
atmospheric mixing throughout a hemisphere is largely complete in less than a
year, whereas a considerably longer period is needed to complete mixing across the
equator - inferences which are consistent with the restricted air flow between the
hemispheres apparent in the wind pattern discussed in section 4. 7.

c
.2 Fig. 3.5 Concentrations of
Mauna Loa (Hawaii)
~ 330 carbon dioxide monitored
c., since 1958 in sites remote
0
c 320 from major industrial zones.
0
0
(After [B) and C.D. Keeling)
"'e
~<>
K £>
o~

'T5
c
-e.
0

0
57
3 The constitution of the 16r-------------. .----~--~---.
atmosphere Stratosphere

12 Tropopause

e~ 8
Troposphere

E
Fig. 3.6 Schematic vertical 0>
September
profiles of autumn and spring "
J: 4
variations in carbon dioxide
content of the northern
hemisphere (north of latitude Boundary layer
30°) troposphere and low o~~-L-L--L-L-J_-L-J~L-~LL~

stratosphere. (After B. Bolin -12 -8 -4 0 4 8 12

in [15]) Departure from annual mean C02 concentration (ppm)

The vertical distribution of the seasonal oscillation of C02 levels (Fig. 3 .6)
confirms and extends this picture. The seasonal oscillation is largest in the
planetary boundary layer, whose extremely well-mixed volume extends about 1 km
above the surface, where all sources and sinks of C02 are localized. The oscillation
is smaller, though still quite large, throughout the troposphere, which extends to a
little over 10 km in the middle and high latitudes depicted, and is noticeably smaller
in the high troposphere than in the low - consistent with the picture outlined in
later chapters of tropospheric weather systems pumping air sporadically from the
boundary layer up through the troposphere, reaching highest levels least often.
Only a small fraction of the seasonal oscillation survives in the stratosphere, where
stirring and connections with the troposphere are quite feeble.
The steady rise in the seasonally averaged levels of C02 apparent in Fig. 3.5 is a
consequence of the rapid combustion of fossil fuels which began on its present
scale with the industrial revolution. Rapid though it is, the increase in C0 2 is much
slower than its speed of mixing through the atmosphere, with the result that it is
observed as clearly in the most remote locations as it is in the most industrial
(Fig. 3.5). It is quite another matter in the oceans, where only the surface layers are
in even sluggish equilibrium with the atmosphere, and the nearly still, voluminous
depths will not be reached for centuries.
Figure 3. 7 sets the recent rise in atmospheric C02 levels in the context of the large
natural variations associated with climate change in the last 160 000 years. Levels
have shot up by over 200Jo since pre-industrial times, and are now well above the
values reached in the previous, rather warmer interglacial, 120 000 years ago. In
fact C02 levels are now probably higher than at any time since the global climate
slid into the current series of ice ages several million years ago (section 4.9). On the
time scale of Fig. 3. 7 the current rate of rise (about 100Jo per decade) is too fast to be
resolved by the thickness of the printed line. The rise in the atmospheric reservoir
probably accounts for less than half of the anthropogenic total, the other half
being dissolved in the oceans, which will eventually (hundreds of years, as above)
take a much larger fraction. All this represents a very large and rapid shift
away from the natural C02 economy, and there is great concern that it may be
inducing significant changes in global climate, because of the important role of
58 C02 in the greenhouse effect (sections 4.10 and 8.5). As with the case of ozone
 (a) Date IADI 3.5 Sulphur dioxide

10
!\ C02 la\1I 300

I I1 ''
I< I \ /
'v v I I E
I I C.
\ f'! 250 .9-
1 1 .e
Fig. 3.7 Variations in
V.V\ ltv\~~\.\
1/ \ l \I \
/
1
f
~~c
::_
atmospheric C02 in the last
\}v\,\ ~\! o 160 000 years.
\ I I\ f..J \ I 0
(a) High correlation between
\./ v\
v
/ \"'\ /
'I
200 Q
0 C0 2 concentrations and
global mean surface
temperatures as deduced
from 0 18 abundance in ice
150 cores from Vostack
(Antarctica). After the end of
the last glaciation C0 2 levels
-10 approached values prevalent
in the pre-glacial optimum
L,-------.----,-----.---- -+ 100
160 120 80 40 NOW 140 000 years ago.
(b) Recent Antarctic data and
Time before present (thousands of years)
direct measurements from
Hawaii since 1958 show a
(b) 350-r--------------, large and accelerating rise in
C0 2 levels since the onset of
Ea. 330 industrialization 200 years
.9- ago. Current levels are
c
.e believed to be well above the
~ 310
highest values occurring since
<D
g the onset of recent
8
--- ---
290 glaciations several million
B years ago. (After [17] and
0 270+-.-----~~,_,-,--,---,-r-r-r~ [18])
1720 1760 1800 1840 1880 1920 1960 2000

depletion, this has attracted international concern, but reduction of anthropogenic

C02 is a much more difficult matter than reduction of CFCs, which play only a
marginal role in human development. Global industrialization currently depends
on continued and increasing consumption of fossil fuels and consequent increasing
production of C02 • With continuing public concern about the safety of power pro-
duction by nuclear fission, and in the absence of useful progress in power produc-
tion by nuclear fusion, the only way out of this impasse is to encourage energy
conservation and the sustainable exploitation of alternative energy resources. In
the context of this section, there is considerable scope for the burning of fast-
growing trees, which simply returns to the atmosphere the C02 which was extracted
during the growth of the tree over preceding seasons.

3.5 Sulphur dioxide [19]

Sulphur dioxide (S0 2) is one of the most biologically destructive of the common
atmospheric trace gases. Background concentrations of - I0- 4 ppm (specific mass 59
3 The constitution of the
atmosphere

Fig. 3.8 Norway spruce in

Bavaria showing 'classic'
crown thinning attributed to
acid rain and associated air
pollution.

- I0- 10) are maintained by natural oxidation of reduced sulphur compounds such
as hydrogen sulphide (H2 S) and dimethyl sulphide ((CH 3) 2 S), which are themselves
emitted by living and decaying organisms. However, artificial combustion of
sulphur-rich coal and some oils, and smelting of sulphide ores, add greatly to local
and total production of S02 •
Irritation and damage to plant and animal tissues are caused when concentra-
tions are raised significantly above the natural background, though complex
details of this sensitivity are still being established. If sources are too strong or
inadequately ventilated, local concentrations may rise dramatically, and in the past
there have been examples of total destruction of vegetation around potent
industrial sources. Even after the implementation of stringent air pollution
controls in recent decades, concentrations of about I ppm are still recorded in
some modern cities, with the result that anthropogenic S02 is one of the biggest
causes of serious lung damage after cigarette smoking. The situation is worse in
countries where governments have not been able or willing to meet the considerable
cost of implementing pollution controls (e.g. in Eastern Europe and much of the
Third World). The most serious pollution episodes occur in calm conditions or
light winds, especially when there is a stable 'lid' only a few hundred metres above
the surface source region (sections 9.12 and 11.5).
Continuing oxidation of S02 tends to produce S03 which is hydrated to H 2S04 in
cloud droplets, producing acidity far in excess of the natural pH limit of about 5.6
maintained by solution of atmospheric carbon dioxide. In this way the vast
megalopoli spawned this century in the industrial heartlands of Europe, North
America and Japan now maintain zones of seriously acidified rain and snow in
broad, persistent, downwind swathes, with consequent biological damage to soils,
vegetation (especially trees (Fig. 3.8)), rivers and lakes, over areas which can
approach continental scale. The complex mechanisms which underlie the effects of
such acid rain are still only partly understood, and it seems likely that oxides of
nitrogen and other substances and factors are involved, but the fundamental point
in respect of all such pollution is that raised concentrations result when local
60 production or deposition outruns the atmosphere's dispersive capacity [19].
3.6 Aerosol [20] 3.6 Aerosol

An aerosol is a suspension in air of solid and liquid particles. The term is conven-
tionally applied to the very large numbers of solid and liquid particles which are
found in the lower atmosphere, and which are too small and slow-falling to be
regarded as precipitation. Especially highly concentrated in the low troposphere,
they range in size from aggregates of only a few hundred molecules (- 10- 3 J.Lm
across) to particles about ten thousand times larger (- 10 J.Lm). The latter are com-
parable in size with true cloud droplets but are distinguished from them by a much
larger proportion of solid or dissolved material.
Particles with radii less than 0.1 J.Lm are called Aitken nuclei because they make
up the great majority of particles able to nucleate cloud droplets in a cloud chamber
- a method used by the Scottish pioneer Aitken to isolate and count them. They
dominate the total aerosol population numerically, but because of their small indi-
vidual masses comprise only about one fifth of the total aerosol particulate mass.
They are produced by natural and man-made combustion, by gas-to-particle con-
version of gases produced by the metabolism and decay of living organisms, and by
volcanoes. Since most of them therefore enter the atmosphere at or near land
surfaces, their concentrations are particularly high in continental air, often
exceeding 106 per litre. Specific masses fall somewhat above the planetary
boundary layer, but are then fairly uniform through the middle and upper tropo-
sphere. The fall speeds of such tiny particles are - 10 em per day at most
(Fig. 6.10), which is slower than the random Brownian motion imparted by
impacts with air molecules and is quite negligible in comparison with even the
weakest updraughts. Once carried away from their source regions, the Aitken
population reduces quite rapidly by Brownian collision and coagulation of
adjacent Aitken nuclei to form larger particles. Aitken nuclei are also removed by
being dragged by water vapour diffusing onto the surfaces of growing cloud
droplets. For all these reasons maritime air masses have relatively few Aitken
nuclei. Above the tropopause there is an even smaller and fairly uniform
distribution of specific mass which is consistent with an input of micro-meteorites
from space.
Particles with radii between 0.1 and 1.0 J.Lm are called large nuclei. They are
usually at least ten times less numerous than the Aitken nuclei but comprise nearly
half of the total particulate mass. Fall speeds are still much smaller than
updraughts, but even so there is a very marked decrease in population above typical
cloud base levels in the low troposphere (never more than a few kilometres above
sea level). This is associated with the efficient removal of large nuclei during the
process of cloud formation above these levels, a process in which they play a very
important part on account of their ability to nucleate cloud droplets at very low
humidity supersaturations (section 6.9). Large nuclei are produced directly by the
same processes which also produce Aitken nuclei, by coagulation of Aitken nuclei,
and by the injection of small salt particles when bubbles burst on the breaking sea
surface. Though they are more numerous in continental air than they are in
maritime air, the difference is less than in the case of Aitken nuclei. Dense
populations of large nuclei are effective in scattering sunlight, producing the dry
haze which can reduce visibility quite considerably in anticyclones. In fact the haze
is largely confined to the convecting boundary layer which in these conditions is
confined below a pronounced temperature inversion at a height which rarely
exceeds one kilometre (Fig. 12.3).
Particles with radii larger than 1.0 J.Lm are called giant nuclei. Though they are 61
3 The constitution of the less numerous again than large nuclei, their numbers rarely exceeding 103 per litre
atmosphere
even in the very low troposphere, they comprise nearly half of the total particulate
mass. Giant nuclei are produced and lost in the same ways as are the large nuclei,
except that they are not formed by Brownian coagulation since this is inhibited by
their large masses, but they are produced in significant quantities in the form of
fine dust lifted by wind off arid land surfaces. They too are very effective nuclei for
cloud-droplet formation, but being too large to swerve round falling raindrops
they can be efficiently swept to the underlying surface by rain and showers, causing
dramatic reductions in haze beneath cloud base. Fall speeds of giant nuclei are
sufficiently large (Fig. 6.10) to ensure that the giant population reduces signi-
ficantly by gravitational settling when updraughts are low - an effect which
increases sharply with particle size. For the same reason the reduction in the giant
population above the very low troposphere is even more marked than it is in the
case of large nuclei.
The total mass of aerosol particles suspended in a column of air resting on a
square metre of the Earth's surface amounts to only a few tenths of a gram- more
than ten million times smaller than the mass of the air column itself. But as well as
playing a vital role in cloud formation, and considerably affecting visibility, this
minute fraction of the atmosphere significantly influences some chemical
processes in the troposphere. Indeed it seems probable that periods of enhanced
volcanism can influence global climate quite significantly by injecting aerosol
particles into the stratosphere in sufficient quantities to interfere with incoming
solar radiation (section 4.9). As in the case of ozone (section 3.3), it is remarkable
how, in a complex system like the atmosphere, minority constituents can have an
importance out of all proportion to their abundance.
The present discussion has emphasized how the behaviour of aerosol exempli-
fies competition between uniform dispersion by stirring and localization by
selective influences - of which the most obvious is differential gravitational
settling.

3.7 Water

From previous sections it might seem that the variability of atmospheric

constituents is in inverse proportion to their abundance: all the major constituents
have essentially uniform specific masses; only minor variations occur in carbon
dioxide; and large percentage variations occur only in trace materials such as ozone
and aerosol. There is however an extremely important exception to this pattern
of behaviour: the atmospheric water substance (water in all its states and forms
- vapour, cloud and precipitation) is very unevenly distributed through the
atmosphere, being almost entirely confined to the troposphere, indeed con-
centrated heavily in the warm, low troposphere in particular (Fig. 3.1). For
example, the overall specific humidity (the specific mass of water vapour) of
the atmosphere is only about 0.30Jo, whereas the specific humidity of the warmest
parts of the troposphere often exceeds 3%. And it is observed that cloud,
and of course therefore precipitation, are almost entirely confined to the
troposphere.
About 97% of the hydrosphere (all the water substance on and close to the
62 Earth's surface) is currently in the oceans, comprising a mass of water which would
 3.7 Water

8
Volcani c
effusion

'rr."
Advect i on

~-') •1 Convect i on •I
p I I
i, ·, I I

Sea
Key
Vapour
Water
Fig. 3.9 Schematic outline of
E Evaporat i on
P Precipi tati on the hydrologic cycle.
/ Subducti on emphasizing the atmospheric
branch.

cover the Earth's surface to a depth of about 2.8 km if evenly distributed. Of the
remainder about 3/4 is in the ice-caps of Greenland and Antarctica, l I 4 is in lakes
and groundwater, and only l / 3000 is in the atmosphere, almost all of it in the form
of vapour. In fact the total vapour content of the atmosphere would produce a
rainfall of only 3 em if it were suddenly and completely precipitated (this being
therefore the precipitable water content of the atmosphere (appendix 5 .2)), which
shows by comparison with the equivalent depth of the oceans that only IO- s of the
total hydrosphere is in the atmosphere.
The hydrosphere is believed to have been formed by outgassing of steam from
volcanoes during the early life of the Earth, the vapour condensing on the cool
exterior to form the oceans (section 3.8). The mass of the current hydrosphere is
effectively constant on all meteorologically relevant timescales, except perhaps
those related to very slow climatic change. However, within the hydrosphere, a
much more rapid exchange takes place continually as the water substance moves
through the hydrologic cycle (Fig. 3.9)- evaporating from the oceans and moist
land surfaces, moving extensively but briefly as the vapour content of moist air,
condensing to form cloud, and being precipitated back to the surface to be avail-
able for recycling either immediately or after movement through groundwater,
rivers and lakes. In the ice ages which have dominated substantial periods of
Earth's history, including most of the last few million years, polar regions have
been cold enough to support ice caps, fed by falling snow and drained by glaciers.
But although these have diverted as much as 50Jo (over twice the present amount) of
the hydrosphere from the oceans, and have affected some parts of the Earth
directly by burial under ice and many coastal regions indirectly by alteration of sea
level, they represent the diversion of only a small part of the global hydrologic cycle
[2I].
The flux of water through the hydrologic cycle is very rapid indeed. Globally
about I metre of rainfall equivalent falls annually, most of it in the form of rain.
Since the total mass of the hydrosphere is fixed on meteorological timescales, this
precipitation must be balanced on average by evaporation. Indeed the precipitable
water content of the atmosphere is so much smaller than the annual throughput of
water vapour that the balance between precipitation and evaporation on a global or
hemispheric scale must be nearly as close instantaneously as it is on average.
Comparing throughputs and capacities, we see that the residence time of water
substance in the oceans is about 2800 years, but that it is only II days in the
atmosphere. 63
3 The constitution of the
atmosphere

Fig. 3.10 Cloud forming as

moist gases vent from hot
springs in a volcanically
active region. Chilling is
probably by mixing rather
than expansion.

64
 Water is sent on its way along the atmospheric branch of the hydrologic cycle 3.7 Water
when it is evaporated from the surface of the sea or moist land, or when it sublimes
from an ice surface. All of this conversion to vapour is maintained by solar heating,
either directly as when moist land surfaces are dried quickly in strong sunlight, or
less directly as when a drying wind (air warmed by the sun elsewhere) does the
same. As discussed in Chapter 8 this evaporation accounts for a substantial part of
the solar input to the Earth's surface, which would otherwise be considerably
warmer in sunlight and colder at night.
Water vapour moves rapidly on the winds and updraughts and downdraughts,
and is obviously closely involved in some of the most vigorous motion and activity
of the troposphere. But since the vapour remains highly concentrated (in terms of
specific humidity) in the warm, low troposphere, it is clear that there must be
selective processes working quickly to counter the homogenizing effects of mixing.
These are the rapid cooling and cloud formation which take place in rising air, and
the equally effective precipitation which returns water and ice to the surface from
all but the smallest clouds. This behaviour is considered in some detail in Chapters
5 and 6, but a simple outline now will serve to complete our summary of the con-
stitution of the atmosphere and act as a preface to much of the rest of this
book.

CLOUD AND PRECIPITATION

As mentioned in the previous section, many aerosol particles and droplets act as
condensation nuclei, encouraging cloud formation so readily in supersaturated air
that the vapour content of any air parcel is effectively limited by saturation at the
parcel temperature. This has two important and related consequences.
1. In the presence of the sharp temperature fall with increasing height imposed by
nearly adiabatic cooling of rising air and warming of sinking air (section 5.6),
the saturation limit ensures that the great bulk of water vapour is concentrated
in the warm, low troposphere. In fact specific humidities at saturation in
temperatures typical of the high troposphere ( -40° to -60 oq are so low that
this region acts as a very effective barrier to the passage of vapour into the
stratosphere and beyond.
2. As moist air rises from the surface layers in convection and other updraughts, it
is brought to saturation as the air expands and cools (though just why there
should almost always be an unsaturated layer between the surface reservoirs
and the lowest clouds layer is surprisingly difficult to explain - section 6.3).
Cloud forms and thickens as the cloudy air ascends still further, and does it so
rapidly that the air is only marginally supersaturated in even the most vigorous
updraughts. When clouds are more than a few hundreds of metres deep some of
the cloud droplets or crystals are able to grow large enough to fall through the
rising air and reach the surface as precipitation. The efficiency of precipitation
varies with the surprisingly complex microscale and cloud-scale mechanisms
involved (sections 6.11 and 10.4), but generally it increases with the vertical
extent of the precipitating cloud. Deep frontal clouds and shower clouds quite
quickly return a very considerable fraction of their cloud water and ice to the
surface.
In these ways cloud and precipitation as well as water vapour are largely con-
centrated in the lower troposphere. There is, however, an important difference 65
3 The constitution of the between the horizontal distribution of precipitation as compared with cloud. A
atmosphere
large fraction of the vapour entering the atmosphere rises little more than a kilo-
metre or two before condensing to form the clouds which usually cover more than
half of the Earth's surface at any instant. However, much of this cloud is so
shallow that it precipitates little if at all before re-evaporating as the cloudy air
sinks or mixes with dryer surroundings. At any instant only the relatively small
fraction of the surface which is covered by deeper clouds (extending into the middle
and upper troposphere) receives precipitation at the moderate and heavy rates
which dominate the total input in almost all locations (Fig. 2.5). The instantaneous
distribution of precipitation is therefore relatively much less extensive horizontally
than is the distribution of cloud. The thin and sometimes extensive veils of ice cloud
often seen in the upper troposphere can give a quite misleading impression of the
concentration of cloud there; they are usually very tenuous and are important
mainly for their ability to veil sunlight and partly close the atmospheric window to
terrestrial radiative emission to space (section 8.3).
There is almost no cloud above the tropopause. Only the strongest updraughts in
vigorous showers may overshoot the tropopause a little way before falling back by
negative buoyancy and spreading out as anvils just beneath the tropospause
(section 10.4). The nacreous or mother-of-pearl clouds occasionally seen in the
lower middle stratosphere (about 20-30 km above sea level) are believed to be ice
cloud forming in the crests of waves of air flowing over hills far below (section
10.8). In the very cold Antarctic winter stratosphere, such clouds are believed to
play an important role in ozone depletion (section 3.3). Noctilucent clouds
(Fig. 3.11) are occasionally observed well after dusk (hence their name) at altitudes
of about 80 km, but their constitution is not well established. Each of these is a
rarity occurring only in high latitudes and providing very minor exceptions to the
rule that clouds are confined to the troposphere.
Though cloud is extremely widespread, its observed transience on many
occasions, and the speed with which shower clouds in particular are seen to form

Fig. 3.11 Noctilucent clouds

over Scotland, looking north
from near Glasgow (56° N) at
2315 Z, 22 June, 1981 . At this
time and date the nearest
daylight at sea level is over
10° of latitude further north.

66
and precipitate themselves out of existence, suggest that residence times of the 3.8 The evolution of the
atmospheric water substance in the form of cloud and precipitation are very much atmosphere
shorter than the eleven days found for water vapour. It follows that the atmo-
spheric water substance exists in the form of water vapour most of the time, or
equivalently that at any instant most of the water substance is in the form of
vapour.
Although the water substance is actually only a very minor constituent of the
atmosphere in terms of specific mass, it is so heavily involved in several important
aspects of atmospheric behaviour that it must be regarded as a major constituent in
effect. Many of its roles have been mentioned in this section, and are dealt with in
detail later, but it is obvious even at this stage that the confinement of the water
substance to a rather shallow layer overlying the Earth's surface must have
important consequences, the most obvious of which is the localization of weather
as we know it in the same shallow layer.

3. 8 The evolution of the atmosphere

ORIGINS

The Earth is believed to have formed in the aftermath of a supernova (the

catastrophic explosion of a heavy star) which left a nebula of debris from which the
solar system emerged by gravitational agglomeration of gas and dust about 4.6 Gy
BP (4600 million years ago). It is likely that the very young Earth had a primordial
atmosphere consistent with the make-up of the parent nebula, but that this was
swept away by violent activity in the very active young solar system. Certainly the
current amount of 36Ar (not the abundant 40Ar mentioned in section 3.2) is about
1()6 times less than the value for the solar system as a whole, and there seems to be
no other way of losing this very inert, heavy gas. It is therefore very likely that our
present atmosphere is a secondary atmosphere produced by outgassing from the
Earth's interior as it warmed by self-compression and radioactive decay. Regard-
less of its mode of formation it is clear that this paleoatmosphere has since been
greatly altered by terrestrial processes, including the evolution of life.
Current evidence [22] suggests that by about 4.4 Gy BP, initial fast outgassing of
water vapour, hydrogen, carbon monoxide and dioxide, hydrogen chloride and
molecular nitrogen (N2) was largely complete (unlike the slow outgassing of 40Ar).
The water vapour probably condensed quite quickly to form deep archaean
oceans, and the very light hydrogen molecules escaped from the Earth's gravita-
tional field, leaving the atmosphere dominated by C02 and N2 • The large amounts
of atmospheric C02 (probably hundreds oftimes present values) may have kept the
Earth's surface a little warmer than at present despite the young Sun's heat output
being about 300Jo below current values -the mechanism being a greenhouse effect
several times larger than today's (section 8.5).
This paleoatmosphere was very different from today's, in which carbon dioxide
is reduced to a trace, and molecular oxygen (02) is the most abundant gas after N2
and these differences must be explained by any account of the subsequent
evolution of the Earth's atmosphere. The current range of differing models shows
how little we confidently understand of events so removed from the familiar
present by the abyss of time, but the following summary represents what seems to
be an emerging consensus. 67
3 The constitution of the EVOLUTION
atmosphere
It seems likely that much of the C0 2 was removed by weathering of exposed or
submarine rock, or by hydrothermal pumping of sea water through fissures in
submarine mid-ocean ridges, becoming locked in a variety of carbonates in the
Earth's crust and upper mantle. Indeed there is speculation that such removal may
have prevented a runaway greenhouse effect like that suffered by our neighbouring
planet Venus as the Sun slowly increased its heat output - which would have
abruptly ended the prospects of life on Earth. Instead, it seems that by about 3.8
Gy BP, atmospheric C02 had been reduced to an important minor component,
able to maintain surface temperatures near modern values by a modest greenhouse
effect involving water vapour and cloud, much as now (section 8.5).
There is no fossil evidence of life at this early stage, and it is known that in the
absence of photosynthetic production, atmospheric 0 2 levels must have been
limited to - 10- 9 present values, maintained by photolysis of H 20 and C02 by
solar UV. The subsequent increase in atmospheric 0 2 is closely related to the
evolution of life on Earth (Fig. 3 .12). From about 3. 5 Gy BP there is increasingly
unambiguous fossil evidence of single-celled bacteria capable of feeble, anaerobic
(in the absence of oxygen) photosynthesis (section 3.4), and by 2.7 Gy BP large
populations of blue-green algae (Fig. 3.13) were laying down beds of limestone in
shallow seas, achieving net production of atmospheric 0 2 as their numbers
increased. Such early life was aquatic, since several metres' depth of water were
needed to filter the damaging UV from the necessary sunlight. Oxygen levels in the
atmosphere probably rose only very slowly, but as they did, so did the much
smaller concentrations of ozone, maintained as outlined in section 3.3, which in
turn increased the protective UV filter and therefore the size and range of aquatic
habitats suitable for primitive life.
When atmospheric 0 2 reached about 1OJo of present values, a new and much
more efficient type of aerobic photosynthesizing organism developed, raising
oxygen levels and widening habitats much more rapidly. There is fossil evidence of
such organisms from about 1.2 Gy BP, distinguished by the appearance of a
nucleus in each single cell. Even more efficient multicelled organisms followed, so
that by about 0.6 Gy (600 My) BP there were marine organisms with shells, which
are believed to have required dissolved oxygen in equilibrium with atmospheric 0 2
levels of at least lOOJo of present values. As they died they formed fossil-rich beds of
calcium carbonate in the shallow seas, further raising 0 2 levels and 0 3 levels to the
point where organisms no longer needed a shield of water against solar UV.
Whether or not it followed immediately, the result was the vast Cambrian
explosion of life, and the colonization of land surfaces. Widespread intense photo-
synthesis raised 0 2 levels to present levels and, in the Carboniferous period, laid
down the rich beds of hydrocarbon debris which are currently supporting
mankind's brief spree of hectic combustion. It is possible that this led to a
temporary depletion of C0 2 which reduced the greenhouse effect and encouraged
the great Permian ice ages (280 My BP). All this time, 40K was decaying radio-
actively to 40Ar which slowly diffused through the crust to produce the third major
component of the present atmosphere, albeit a biologically and chemically inactive
one.
This brief survey shows how far the present atmosphere has evolved from its
paleoatmospheric origins; how a variable greenhouse effect has maintained
surface temperatures in the narrow range needed to allow the accelerating
evolution of life, and how that evolving life has in turn built up the reservoir of free
oxygen on which it now depends. Such obvious self-regulation has recently
68 encouraged the view that the terrestrial biosphere should be regarded as a coherent
 3.8 The evolution of the
-3 -2
10 10 atmosphere
4000
-- --- .......... ......
oklett Mchmenlarwo roc. 1

OJ',
2000 ' blue- grH n at goo
\
\
cello wllh nuc

'\
1000 mulll-coll.cl orgon 11111
\
PRECAMBRIAN \
\

50 0 CAMBRIAN
""'11 1-coll.cl orgon tms
.
Wllh "' . ...

"1\,..lo
and IM'III

IMCI

CARBONIFEROUS
250 ............
I
TRIASSIC

125 JURASSIC u ung planll

CRETACEOUS
\
Fig. 3.12 Probable co-
- I
10 evolution of atmospheric 0 2 ,
0 3 and life on Earth. (After
Surface 0 2 10 3 total relative to current values [23])

Fig. 3.13 Anabaena: a

modern blue-green alga,
descendent of a group of
organisms first apparent in
the fossil record of more than
2000 million years ago. The
filament is about 200 p.m long
and contains heterocysts
capable of fixing atmospheric
N2.

69
3 The constitution of the superorganism, Gaia [24], whose intrinsic purpose is to regulate its environment to
atmosphere
its own advantage. Whatever the merits of such a view, it is impossible not to be
impressed by (and grateful for!) the continued cooperation of the Earth's
biosphere and atmosphere. It would be a sad irony if the only part of the biosphere
capable of being consciously aware of such cooperation should choose to ignore its
obvious lessons.

Appendix 3 .1 Residence time

Consider atmospheric N2 as an example of any reservoir maintained in dynamic

equilibrium by balanced production and loss. Let Mbe the mass of atmospheric N2
in kg, and let Fbe the rate of denitrification (production of atmospheric N2) in kg
per year, which must also be the rate of fixation (loss of atmospheric N2), sinceMis
constant. If the individual molecules of N2 entered and left the atmosphere in the
same chronological order, each would spend time M/Fyears in the atmosphere,
which is by definition the residence time Tof N2 in the atmosphere. Since of course
the molecules do not enter and leave in such an orderly manner, some individuals
leave very soon after entry and others remain unfixed for very long periods of time.
Tis therefore an average period of residence.
T= MIF
The concept of residence time is potentially useful in all systems which are in or
near dynamic equilibrium. The quantity M is usually a mass or a measure of mass,
as for example in the treatment of the hydrologic cycle in section 3. 7, but it may be
any conservative quantity, such as in the discussion of atmospheric energy in
section 13.3. Quantities in M and F may be expressed in any convenient way
provided the same units are used in each. For example in section 3. 7 the quantities
of water substance are expressed effectively in volumes of water, which is satis-
factory because water is so incompressible that volume is a measure of mass. The
residence time Twill be given in whatever time units are used in F- years in the
case of the annual throughput of nitrogen cited above.
If the quantity passes through successive reservoirs in series without gain or loss,
the flux F is conserved along the sequence, and the ratio of quantity to residence
time must be the same for each reservoir.
MI
-=-=-
Mz M3
Tl Tz T3
This chain rule applies nicely to the sequence of vapour, cloud and precipitation in
the atmospheric branch of the hydrologic cycle (section 3. 7).

Problems

LEVEL 1

70 I. Uniformity of specific mass of a component of a compressible mixture under

 gravity indicates which type of equilibrium? Does this follow also in the Problems
absence of gravity (e.g. in an orbiting space station)?
2. List the order of component layering from the top down in the diffusive
equilibrium of a planetary atmosphere consisting of moist air, with some
photodissociation of molecular oxygen and water vapour.
3. Briefly define the following terms: denitrification, residence time, respiration,
photosynthesis.
4. Why is the altitude of maximum specific mass of ozone (or any other
component with a level of maximum concentration) greater than the altitude
of maximum ozone density?
5. List the effects producing variations of the specific mass of C0 2 in space or
time, indicating in each case whether the tendency is upward or downward.
6. Why does the number density of large nuclei fall rapidly in the vicinity of
cloud-base level?
7. Starting from evaporation, sketch and label a likely sequence of events
experienced by a water molecule which is eventually precipitated onto land.
Follow the sequence back to the sea. Add branches describing the cases of
evaporation from land and precipitation onto the sea.

LEVEL 2

8. Assuming a reasonable value of 4 1 for accessible capacity of a pair of human

lungs, find the maximum mass of contained oxygen assuming an air density of
I.2 kg m- 3 • Why is the actual value likely to be a little lower?
9. Suppose that the maximum specific mass of ozone is 10- 5 at a height of 32 km
above sea level. Using a reasonable value for air density there (about 2 decadal
scale heights above the surface), find the density and number density of ozone
in quantities per litre.
10. The reservoir of carbon in the living and dead terrestrial biosphere (ignoring
fossilized reservoirs like coal and oil) is believed to be about 2400 units, and the
annual exchange with the atmosphere is about 110 units [14, 2I], where a unit
is I0 12 kg. Find the implied residence time for carbon in the terrestrial
biosphere. Since the exchange is almost entirely in the form of C0 2 , find the
annual exchange of C02 in the same units.
1I. Find the specific mass of an aerosol dominated by giant nuclei of radius I J.Lm
and number density 103 per litre, where the air density is I kg m- 3, and
compare with Table 3.1.
I2. Assuming realistically that 971Jfo of the Earth's water substance is in the form
of oceans which would cover the planet uniformly to a depth of 2.8 km, find
the average ocean depth given that they actually cover only 70% of the total
surface. If all of the present ice caps were to melt and run into the oceans, what
would be the maximum rise in sea level?

LEVEL 3

I3. If convective equilibrium always tends to encourage uniformity of

distribution, why does it maintain the observed sharp vertical lapse of air
density rather than a uniform profile?
I4. Speculate on possible short- and long-term effects on the total mass of 71
3 The constitution of the atmospheric carbon dioxide of a brief large-scale nuclear war followed by a
atmosphere
very prolonged winter induced by resulting smoke and dust.
15. Discuss critically but sympathetically the possible physical origins of the story
of the universal deluge which is widespread in the early culture of the Middle
East.
16. The answer to problem 12 is the maximum possible rise in sea level. Why
should the actual rise be less than this? And consider the apparent rise in sea
level relative to a coastal landmark in a region initially under the edge of an ice
cap, remembering that the apparently solid land surface is substantially
depressed by an ice burden.
17. Assuming that the typical precipitation particle or droplet falls at about
4 m s -I from about 4 km above the surface, equate the implied fall time with
the residence time of water in the form of precipitation, and hence find the
global fraction of the atmospheric water substance which is in the form of
precipitation at any instant. You can assume a reasonable value for the
residence time of water in the form of vapour.

72
State and climate CHAPTER

4.1 The equation of state

We have seen that all but a minute fraction of the mass of the atmosphere consists
of the mixture of gases called air. Let us look closely at how the condition or state
of an air parcel depends on its temperature and pressure. Since a gas has neither
rigidity nor shape the only remaining simple property is its volume. So let us look
for a relationship between temperature, pressure and volume.
In terrestrial conditions the major atmospheric gases, and many of the minor
ones too, behave very nearly as ideal gases in that they obey the gas laws discovered
by Boyle, Charles and others in the early stages of the scientific revolution. It is
shown in Appendix 4.1 that these laws can be combined in a single equation ofstate
applicable to all ideal gases. For one mole of ideal gas (i.e. a quantity whose mass in
grams is numerically equal to the molecular weight of the gas) the equation of state
is
p V = R. T (4.1)
where p, Vand Tare respectively the pressure, volume and absolute temperature of
the gas, and R. is the universal gas constant whose value is 8.314 J K- 1 moi- 1 •
It is useful to change the format of eqn (4.1), since the presence of the parcel
volume Vis inappropriate to the case of an unconfined gas like atmospheric air. It
is shown in Appendix 4.1 that it is relatively easy to find an equivalent equation
involving the density of the air rather than its volume V. This is the meteorological
form of the equation of state
p = pRT (4.2)
where R is the specific gas constant for the gas, so called because its value varies
from gas to gas, unlike R •. Since R is given by 1()3 R.lMit follows that the densities
of ideal gases at the same pressures and temperatures are inversely proportional to
their molecular weights.
Equations (4.1) and (4.2) apply to single ideal gases. To deal with a mixture like
air we must combine the equations of state for the constituent gases - nitrogen,
oxygen etc. It is shown in Appendix 4.1 that a mixture of gases, each with
molecular weight M; and specific mass X;, behaves as a single ideal gas with specific
gas constant R given by 73
4 State and climate (4.3)

and that the equivalent molecular weight of the mixture is

M = 11~ x;IM; (4.4)

Substituting values for X; and M; from Table 3.1, we find that dry air behaves as a
single ideal gas with specific gas constant 287 J K -I kg- 1 and molecular weight
29.0 (which means that 29.0 g of dry air obey eqn (4.1)).
It is apparent from eqn (4.3) that a mixture of gases with component specific
masses uniform throughout a certain volume (e.g. dry air in the turbosphere) has a
uniform specific gas constant. This means that the mixture can be treated as a
single gas for purposes depending on the modified equation of state (eqn (4.2)).
Though the value of 287 J K- 1 kg- 1 for R is strictly applicable to dry air only, it
can be used for moist or even cloudy air in all but the most precise calculations. In
principle, however, the presence of water vapour with its relatively low molecular
weight reduces the density of moist air below that of dry air at the same
temperature and pressure, and this can be significant when considering the very
small buoyancies typical of cumulus convection (Appendix 4.1 and section
7.12).
The meteorological form of the equation of state (eqn (4.2)) is used mainly to
determine the remaining variable when the other two (out of pressure, temperature
and density) are known. This is often done to put equations into their most
convenient form, as for example in deriving the pressure-height relationship in
section 4.3. There, as in several places throughout this book, the equation of state
is used to express air density in terms of pressure and temperature. This matches
laboratory practice, where direct measurement of air density is so slow and clumsy
(for example by weighing a rigid container of known volume before and after
evacuation of air) that instead density is calculated from simultaneous
measurement of pressure and temperature. As an exercise in this procedure let us
find the approximate value of air density in the low troposphere using typical
values of pressure and temperature. Pressure is always close to 1000 mbar (105 Pa)
near sea level, and let us suppose that the air temperature is 20 oc (293.2 K). Using
the value for the specific gas constant for dry air, the air density is found to be
1.19 kg m- 3 • Since this value is not very sensitive to realistic variations in pressure
and temperature in the low troposphere, we can generalize roughly that air
densities near sea level are just a little greater than 1 kg m - 3 • Though this is only
one thousandth of the density of water, it is far from being negligible, as related
properties such as atmospheric pressure and the momentum of a high wind clearly
show.

4.2 The vertical profile of temperature

Figure 3.1 contains a greatly simplified picture of the major features of the vertical
profile of temperature observed over most of the globe - only the upper turbo-
sphere in middle and high latitudes in winter being systematically different on
74 account of the prolonged absence of solar radiation there. Let us consider these
 Wind (0 kt) Height (m) 4.2 The vertical profile of
temperature

Fig. 4.1 A typical winter

sounding in the vicinity of an
Atlantic depression, obtained
by routine radiosonde from
the synoptic upper air station
/<>0 in north.west Scotland. It is
displayed on a tephigram
ou u~~~~~-\~~~~--~~500
(mbar) (section 5.8), with
temperature, height and wind
~~~-?L-1-~--~~--71600
azimuth and air speed plotted
or listed against pressure.
The tropopause is sharply
defined a little above the
200 mbar level on this
occasion, being surmounted
by a temperature inversion in
the base of the stratosphere.

features with the aim of identifying physical mechanisms at work in the different
height zones.
The first 10-15 km above sea level stand out as being a region in which tempera-
ture falls sharply and fairly consistently with increasing height. This is the tropo-
sphere, whose characteristically large temperature lapse rate is closely related to the
widespread occurrence of convection there, both cloudy and cloudless. Values of
this lapse rate vary somewhat with time and geographical location when averaged
between the surface and the tropopause, but generally lie fairly close to 6 oc km - 1•
For example the value on the occasion of the radiosonde flight displayed in Fig. 4.1
was 6.5 oc km- 1 •
Consider how the steep lapse rate of temperature in the troposphere is related to
that region's incessant commotion, and in particular to the associated vigorous
vertical movements of air. As an air parcel rises into regions of lower ambient air
pressure, it expands and cools. The cooling occurs as air molecules in the
expanding parcel collide with their retreating neighbours and use up some of their
molecular kinetic energy in the process. When there are no complicating gains or
losses of heat by any kind of exchange between the parcel and its near or distant
environment (by conduction or radiation for example), the process is said to be
adiabatic, and the degree of cooling is fairly readily given by theory, as discussed in
section 5.6 and following sections. Sinking parcels undergo a corresponding
warming.
Since vertical motion is so widespread and vigorous in the troposphere we might
expect there to be a convective equilibrium temperature profile with corresponding
lapse rate(s), just as there is a convective equilibrium distribution of specific masses
of the constituents of dry air. This expectation is confirmed by further discussion
and observation (sections 5.9 and 5.10), which show that there are different
equilibrium profiles depending on whether the air is cloudless or cloudy. In cloud-
less air there is a single well-defined temperature lapse rate of very nearly
10 oc km -I, whereas in cloudy air the lapse rate ranges continuously from around
5 o C km - I in high temperatures to the cloudless limit at the lowest temperatures.
On the very cloudy occasion of Fig. 4.1, the cloud-free zone extended only about
250 m above the surface, and was therefore too shallow to appear on the rather 75
4 State and climate gross vertical scale of the diagram. However, in the deep cloudy layer filling the rest
of the troposphere, there is a clear tendency for the lapse rate to increase with
increasing height (i.e. with decreasing temperature), in qualitative agreement with
the convective equilibrium model for cloudy air.
To avoid oversimplifying this picture of the relationship between convection and
tropospheric temperature profile we must consider two major complicating
factors.

1. Although convective air movements in the troposphere are nearly adiabatic,

they are not exactly so. Air parcels actually gain or lose heat by direct inter-
action with solar and terrestrial radiation at rates which produce significant
temperature changes over periods of a day or more, as discussed in section 8.10.
Though this has relatively little effect on air rising rapidly in a shower cloud or a
front, it can considerably influence the great bulk of tropospheric air which
rises and falls much more slowly than this.
2. Two apparently insignificant temperature inversions visible in the low tropo-
sphere in Fig. 4.1 arise from the second important complication. Although
small in appearance, a detailed examination of their effects on the buoyancy of
air parcels trying to rise through them would show that they can inhibit all but
the most violent convection. In fact there was no showery convection on the
occasion of Fig. 4.1. The deep and horizontally extensive layers of pre-
cipitating cloud which were present were produced instead by the completely
different type of vertical motion associated with a complex system of vigorous
fronts then moving north-east across the British Isles. In such fronts the large
sheets of cloud are produced and maintained by ascent which is very much
slower but more persistent and extensive than that associated with shower
clouds (section 12.4). In such slow ascent air parcels usually take several days to
rise from the low to the high troposphere, in which time they may well move
1000 or more kilometres horizontally in the relatively strong winds typical of
frontal regions (notice the high wind speeds in the upper troposphere in
Fig. 4.1). It follows that such air is ascending along horizontally extensive,
slightly sloping paths (gradients - 1: 100), and not along the nearly vertical
paths of either radiosondes or air parcels in showery convection. In fact the
temperature profile recorded by a radiosonde in or near such fronts, which are
quite commonplace in middle and high latitudes, is a nearly vertical section
through a complex sandwich of nearly horizontal strata, each layer of which
may be related to a geographically distinct surface source region (Fig. 11.14).
In such circumstances we cannot expect to be able to interpret details of the
temperature profile in terms of processes, such as showery convection, which
are confined to the profile itself.

TROPOPAUSE AND LOW STRATOSPHERE

The characteristically steep lapse of temperature ends at the tropopause, often

abruptly as in Fig. 4.1. Above this there may or may not be a temperature inversion
below the deep, nearly isothermal layer which makes up most of the low strato-
sphere. Clearly the marked contrast between tropospheric and stratospheric
temperature profiles must be associated with a significant difference in
atmospheric behaviour. The confinement of clouds below the tropopause suggests
76 that updraughts are confined similarly. In fact it is observed that atmospheric
motion in all directions is much more subdued in the stratosphere than it is in the 4.2 The vertical profile of
troposphere, though quite high wind speeds are found up to several kilometres temperature
above the tropopause in the vicinity of fronts and their associated jet streams (as in
Fig. 4.1). In the absence of vertical motion comparable with that found in the
troposphere, we must expect the non-adiabatic (or diabatic) effects of solar and
terrestrial radiation on the temperature profile to be more pronounced. Indeed the
largely isothermal stratification of the low stratosphere depends much more on the
radiative transfer of heat than it does on convection, which would otherwise
naturally tend to maintain the steep lapse rate associated with cloud-free
convection.
The upper stratosphere is even more obviously dominated by the diabatic effects
of radiation. The steep and deep temperature inversion extending up to the strata-
pause (Fig. 3.1), where air temperatures approach those of the Earth's surface
about 50 km below, is maintained by the absorption of soft ultraviolet radiation by
ozone, as discussed in Section 3.3.
Such mild conditions contrast strongly with the abysmal cold which would
prevail if thermal equilibrium were maintained solely by convection from the
surface to those altitudes. In fact temperatures are maintained by a combination of
diabatic heating and cooling, and weak vertical and horizontal movements of air
from hotter and cooler regions. For example, in middle and high latitudes,
temperatures throughout the stratosphere fall sharply in winter, as solar warming
is reduced and cooling by net infra-red emission continues (both through inter-
actions with 0 3 - section 8.6). However, the fall is smaller than would be expected
by these radiative processes alone, because air currents import heat from warmer
regions. In some Northern hemisphere winters, there is a dramatic sudden warming
of the low stratosphere as air moves and sinks rapidly when some dynamic
instability of the stratosphere is triggered by patterns of gross airflow in the tropo-
sphere. In high latitudes the average temperature of the high stratosphere varies
from about +20 oc in summer to -40 oc in winter.
Above the stratopause the temperature falls to values at the mesopause (Fig. 3.1)
which can reach -100 oc in winter in high latitudes -the lowest values in the
atmosphere. No doubt some of this lapse is maintained adiabatically by weak
vertical air motions, but observed average lapse rates of about 2.8 oc km -I are
much smaller than the value of 9.8 to be expected in cloud-free convective
equilibrium (sections 5.9 and 5.12), showing that air currents are carrying
significant amounts of heat, as in the stratosphere. In summer, continuous local
warming by photodissociation and photoionization of air molecules (section 3.1)
raises mesopause temperatures to -30 °C.
Above the mesopause the temperature rises sharply with height in the thermo-
sphere because increasing absorption of the harder components of solar ultraviolet
swamps the opposing convective effect (see section 3.1).
The variations of temperature along a vertical profile through the turbosphere
discussed above may seem impressive on the Celsius scale, especially in relation to
the narrow temperature tolerances of most living organisms, but it is important to
note that they are quite modest on the absolute scale. It is apparent from Fig. 3.1
that temperatures in most of the atmosphere up to the 100 km level lie within 50° of
250 K. In fact to an approximation which is surprisingly good for most purposes
which explicitly involve the absolute temperature scale, the whole turbosphere can
be considered to be isothermal, with a temperature of 250 K ( -23 °C}.

77
4 State and climate
4.3 The vertical profile of pressure

By contrast with temperature, atmospheric pressure varies enormously with

height, decreasing by a factor of about a million between the surface and the turbo-
pause. As mentioned in section 1.1, and as apparent in Fig. 3.1, this huge variation
in pressure is very regularly distributed in the vertical, having a nearly uniform
scale height throughout the full depth of the layer. In fact the decadal scale height
(the height interval in which pressure drops to one tenth of any initial value)
approximates quite closely to 16 km throughout. More detailed observations in
Fig. 4.2 show that the pressure-height relation is not quite so simple, but the
discrepancies are so small that for the moment they serve to emphasize how closely
actual soundings approximate to the simplicity of Fig. 3.1. Let us consider in
particular the significance of the smoothness and regularity of the vertical
distribution of atmospheric pressure.
Smoothness. Atmospheric pressure varies very smoothly with height; there are
no signs of any fine structure in the vertical pressure profile corresponding to those
in the associated temperature profile (Fig. 4.1). Horizontal pressure variations are
very much smaller, as will be seen in section 4.6, and are also very smoothly
distributed; and pressure variations with time at a fixed position are very modest
and well ordered. In fact pressure is by far the most regularly and smoothly
distributed of the readily observable atmospheric variables. To explain such
smoothness from the viewpoint of Chapter 3, we must look for a mechanism
capable of smoothing the atmospheric pressure field much more quickly than it is
disturbed by normal atmospheric commotion. Let us try to clarify the issue by con-
sidering what would happen if a substantial pressure variation were suddenly intro-
duced in some small part of the atmosphere.
Suppose that a toy balloon is blown up and burst. A water manometer will show
that the excess pressure inside the balloon just before bursting is a few tens of
millibars. In the typical low troposphere this is equivalent to the pressure difference
over several hundreds of metres of altitude, over hundreds of kilometres in the
horizontal, or over a day or more at a fixed position. It is therefore a very
substantial pressure difference which on this occasion is concentrated across a
fraction of a millimetre of stretched rubber. On bursting, the rubber is almost
instantaneously removed, leaving a large, localized pressure excess in the air. This
excess spreads outward behind a three-dimensional shock wave expanding at about
the speed of sound, so that within a fraction of a second it is distributed over a
volume five or six orders of magnitude larger than that of the balloon, and is

Fig. 4.2 Log pressure from pressure (mbar)

Fig. 4.1 plotted to show its
100 200 300 500 700 1000
near linearity with height.
The logarithmic isobar e 16
spacing in Fig. 4.1 (and all .=Oi 14
such diagrams) therefore > 12
.2!
corresponds to a nearly linear
height axis.
.,
"' 10
.,
U) 8
~ 6
.0
4
:.::" 2
·;;;"'
I 0~----------~---------~
2 2.5 3
78 log 10 pressure
diluted accordingly. The dilution is so rapid that within a second of the burst we 4.4 Hydrostatic equilibrium
should expect no trace of the originally large pressure excess to be detectable by
even the most dense and extensive network of sensitive pressure gauges. And since
not even the most vigorous atmospheric disturbances contain air moving at more
than a small fraction of the speed of sound, the smoothness of the observed
pressure fields is explained at least qualitatively.
Regularity. The foregoing does not argue that atmospheric pressure should be
uniform. The smoothness which is explained is instead the absence of significant
deviation from the pressure distribution to be expected when the air is fully
adjusted to the prevailing force, which is gravity. Assuming that such adjustment
implies that the air is stationary, let us examine the equilibrium of air at rest under
its own weight, looking for an explanation for the marked regularity of vertical
distribution apparent in Fig. 3.1. The assumption of an absolutely static atmo-
sphere is obviously unrealistic in principle, but we will see in section 7.12 that
deviations from static equilibrium are very slight in the cases of most natural air
motions.
In a static atmosphere the upward and downward forces on any air parcel must
balance. Gravitational attraction between the parcel and the Earth produces a
downward force which by definition is the weight of the parcel. Starting from some
imaginary unbalanced state, we can envisage the air settling under gravity and
raising the pressure at lower altitudes to the point where the net upward pressure
force on any parcel (i.e. the excess of the upward pressure on its base over the
downward pressure on its top) exactly balances its weight. This balance is known as
hydrostatic equilibrium, and it can obviously explain the observed upward lapse of
pressure throughout the atmosphere at least qualitatively. To show that it does so
quantitatively as well, we need a formal statement of hydrostatic equilibrium- the
hydrostatic equation.

4.4 Hydrostatic equilibrium

Several forms of the hydrostatic equation are derived in Appendix 4.2, the simplest
of which relates the difference in atmospheric pressure between two heights to the
weight of the intervening layer of air. Choosing an air column with unit horizontal
area, the pressure difference is simply equal to the weight M g of the air column
(mass M), i.e.
PI- Pz = Mg (4.5)
where PI is the pressure at height zi and p 2 is the smaller pressure at the greater
height z2 •
The appearance of the gravitational acceleration g as a separate factor in eqn
(4.5) implies that it has the same value at all heights, whereas it actually diminishes
with increasing height according to the inverse square of the distance from the
centre of the Earth, and also varies slightly with latitude (section 7 .4). However,
the depth of the turbosphere, especially the troposphere, is such a small fraction of
an Earth radius that g can be treated as a constant in all but the most precise meteo-
rological calculations; and even there (e.g. in measuring the heights of standard
isobaric surfaces from radiosondes) the variation in g can be accommodated by
replacing actual height by the very slightly different geopotential height (Appendix
4.3).
By setting p 2 equal to zero in eqn (4.5), it can be seen that the atmospheric 79
4 State and climate

Fig. 4.3 Many geologically

young mountains like this
reach above the 500 mbor
level, and therefore hove less
than 50% of the atmospheric
moss above their peaks. The
percentage of atmospheric
water substance above peak
level is vey much smaller for
reasons discussed in section
3.7.

pressure at any level is proportional to the total mass of the atmosphere in a vertical
column above that level. For example, since a typical value of atmospheric pressure
at sea level is 1010 mbar, you can substitute the familiar value of g into eqn (4.5) to
find that about 10.3 tonnes of atmosphere (almost all of it air) rest on each hori-
zontal square metre of the sea surface. And since it is apparent from Fig. 4.2 that
the pressure about 5.5 km above sea level is only 500 mbar, it follows that only
about half of the atmospheric mass lies above this height.
Regarding the vertical distribution of atmospheric pressure (Figs. 3.1, 4.1 and
4.2) as a measure of the distribution of atmospheric mass, we see that the mass is
heavily concentrated at low altitudes and is more and more thinly distributed at
greater and greater altitudes. Since the concentration of mass is expressed by
density, we can reformulate eqn (4.5) as shown in appendix 4.2 to produce

-
op = -gp (4.6)
oz
The term op/oz in eqn (4.6) is the instantaneous vertical gradient of pressure, and p
is the atmospheric density at the same location. The minus sign ensures that
pressure decreases with increasing height, as it obviously must.
Equation (4.6) does not only apply to a pressure profile like that in Fig. 4.4. In
fact it is the most generally useful form of the hydrostatic equation, describing the
hydrostatic equilibrium of any fluid, whether liquid or gas. In an almost incom-
pressible and homogeneous liquid like the sea, the near uniformity of water density
is associated with a nearly uniform pressure gradient according to eqn (4.6), and
hence with a nearly linear pressure profile (Fig. 4.4). However, in the case of the
Fig. 4.4 A pure exponential atmosphere, the compressibility of the air results in a highly nonlinear profile
decoy of pressure with (Fig. 4.4 again). Although apparently very irregular, this profile has an underlying
height. The straight-line
order which we can reveal by replacing air density using the equation of state (eqn
construction is discussed in
Appendix 4.4. 4.2). It follows that

-
op = -p/H (4.7)
80 OZ e
where He = RT!g and is termed the exponential scale height for reasons which 4.5 The vertical profile of
become clear when eqn (4.7) is integrated to relate pressures at two substantially air density
different heights z1 and z2 • This integration becomes very simple when the air
temperature T does not vary with height, giving an exponential relationship
between pressure and height.
(4.8)
According to eqn (4.8), in an isothermal layer of thickness He, the pressure
at the upper limit of the layer is 11e times the pressure at the lower limit,
where e is the base of natural logarithms (value 2.7183 -see Appendix 4.4). It
can then be shown that the height interval in which the pressure falls to one
tenth of the value at the lower level (the decadal scale height H 10) is given by
2.3026He.
Equation (4.8) describes an exponential decay of pressure with increasing
height which resembles the observed pressure distribution both in its shape
(Fig. 4.4) and in the uniformity of its scale heights (Fig. 3.1 ). To complete the
identity of theory and observation we should consider why He should be so
nearly uniform in the real atmosphere, and check that its numerical value is
realistic.
Of the three factors which determine He, the gravitational acceleration g is
effectively constant in the shallow height zone relevant to meteorology, and the
uniformity of the specific gas constant R throughout the turbosphere has been
established in section 4.1 and Chapter 3. The remaining factor is the absolute
temperature T, whose percentage variability has been shown in section 4.2 to be
surprisingly small. Since the temperature of most of the turbosphere lies within
200Jo of 250 K, it follows that we should expect the scale heights for pressure to be
constant to the same degree, as is indeed the case. The observed profiles of pressure
against height are nearly but not quite exponential, as shown by the nearly linear
height profile of the logarithm of pressure (Fig. 4.2). In comparison with a purely
exponential profile, pressure falls slightly less sharply in the warm, low tropo-
sphere than it does in the relatively cold high troposphere, where the scale height is
somewhat smaller. However, these are relatively minor amendments to the
underlying exponential profile.
Lastly, let us insert realistic values to check the realism of the theoretical scale
height values. Substituting values for g and R, and assuming a typical mid-
tropospheric 250 K for air temperature T, we find a value of 16.8 km for the
decadal scale height, in good agreement with observation (Fig. 3.1).
As a general conclusion, it seems that the vertical profile of atmospheric pressure
is determined essentially by the material constitution of the atmosphere, its
temperature profile, and the prevalence of a very close approximation to hydro-
static equilibrium - the last in spite of the commotion which keeps it so well
mixed.

4.5 The vertical profile of air density

The vertical distribution of air density in the atmosphere follows from the distribu-
tions of pressure and temperature, using the equation of state (eqn (4.2)). Indeed
since pressure varies so strongly in the vertical, whereas temperature variations are 81
4 State and climate quite modest on the absolute scale, the vertical profiles of air density and pressure
must be very similar. In fact the density profile shows a nearly exponential decay of
density with increasing height and a decadal scale height which is never very far
from 16 km throughout the turbosphere.
Since we live most of our lives within a few hundred metres of sea level, the air we
breathe is very nearly the most dense in the atmosphere, with values exceeding
1 kg m -J as demonstrated in section 4.1. Only 10 km above sea level the air density
is usually a little less than one third of sea-level values and is too low for sustained
human breathing. The peak of Mount Everest is at about this level, so that a
climber there, inhaling in a normal fashion, obtains less than one third of the mass
of air to which he is accustomed at sea level,.and the uniformity of specific masses
ensures that oxygen intake is reduced in the same proportion. Such a reduced
intake cannot sustain the life of even the fittest individual for more than a few
weeks, nor allow hard exertion for more than a few hours at a stretch. Air-
breathing aircraft engines, such as the common turbojet, are similarly constrained,
though they are of course expressly designed to operate at such altitudes to take
advantage of the reduced airflow drag on the airframe there.
At greater altitudes air is quite useless for human breathing, and as its density
falls away to the extremely low values found in the upper stratosphere and meso-
sphere, it even becomes ineffective in conducting or convecting heat to or from an
immersed body. For example, high-altitude research balloons floating at an
altitude of 50 km above sea level become extremely cold at night despite being
immersed in the relatively warm air there (Fig. 3.1), because radiative cooling
swamps the feeble conductive and convective warming powers of the very thin air.
However, at an altitude of 80 km, the air density is still sufficient to vaporize
meteorites and returning artificial satellites by frictional heating of these very fast-
moving bodies, even though the density is five orders of magnitude smaller than at
sea level.
Many other important consequences flow from the nearly exponential vertical
profile of air density. Given the tendency of the vigorous stirring of the turbo-
sphere to maintain spatially uniform specific masses of atmospheric constituents,
it follows that these too tend to have exponential vertical profiles of absolute
density. This in turn governs the vertical distribution of density-related phenomena
such as chemical reaction rates and interactions with sunlight. For example, those
solar wavelengths which are almost completely absorbed before reaching sea level
may be relatively abundant only a few kilometres higher up, because so much
absorption is concentrated in the relatively dense surface layers of the atmosphere.
This is the case for wavelengths at and just beyond the violet end of the visible
spectrum, which explains why people on mountain holidays experience such rapid
suntan and sunburn. For similar reasons the familiar blue of the sky, resulting
from preferential scattering of the blue end of the visible spectrum by air
molecules, deepens rapidly towards black with increasing altitude, and the blue
atmosphere seen from space is confined to a relatively shallow layer overlying the
Earth's surface (Fig. 1.2).

4.6 Equator to poles

In previous sections we have examined typical vertical distributions of temperature

82 and pressure. When we begin to examine horizontal distributions of these and
other meteorological variables, especially annual or seasonal averages, we enter the 4.6 Equator to poles
conventional domain of climatology. In fact the vertical distributions too are
climatologically very important, since they are closely related to prevailing weather
types, but this is often left implicit in introductory climatological treatments, and is
regrettably ignored in the more superficial treatments of climatological air masses.
We will examine horizontal and vertical distributions of meteorological variables
and their relationship with the state and behaviour of the atmosphere at many
points throughout the rest of the book, but here we introduce the topic with a
review of average distributions between equator and poles.

PRESSURE

Because of the very squashed shape of the atmosphere, horizontal gradients of

temperature, pressure, etc., are very much smaller than their respective vertical
gradients on all but the very smallest scales, but they are nevertheless very
important. The largest and most persistent horizontal gradients are those in meri-
dional directions (i.e. along a meridian), and we must expect some of these to be
directly or indirectly associated with the strong meridional gradient of solar input
to the planet. Significant gradients do occur in other horizontal directions, but they
tend to be much less pronounced in annual means since they are mostly associated
with seasonal land-sea contrasts and transient weather systems.
Consider first the meridional distributions of monthly mean pressure at the base
and in the middle of the troposphere (Fig. 4.5). The data used have in addition
been zonally averaged (i.e. along lines of latitude) to remove the differences which
occur between different longitudes. Data are from the northern hemisphere only,
since observations are much more complete there than in the southern hemisphere.
However, we should expect the latter to look broadly similar after allowing for the
six-month shift in seasons.
In Fig. 4.5 the distribution of pressure at the base of the troposphere is repre-
sented by the heights of the 1000 mbar isobar above mean sea level (MSL). On
average this is between 100 and 200m, which corresponds to MSL pressures of
between 1008 and 1016 mbar, after allowing for the lapse of about 1.2 mbar per
10 m which is typical of the layer. The global annual average pressure at MSL is
1013.2 mbar. Note that it is standard meteorological practice to represent the
horizontal distribution of pressure by means of the elevation of isobaric surfaces in
this way. Strictly speaking, the method compounds the effects of vertical and

Fig. 4.5 Monthly and zonally

] 5800 averaged height contours of
the 1000 and 500 mbar
.r: surfaces in the northern
Cl
·a; 5600 hemisphere in January and
.r:
July. The vertical height
~

~
0
"
0
E scales are chosen to enhance
gu 5400 0
u contour gradients rather than
absolute heights. (Data from
..,:;; 300
E
~ US Department of Commerce.
E 5200 200 0 Weather Bureau)
0
.,
0
0 100
0

Latitude ( 0 ) 83
4 State and climate horizontal pressure gradients, but the former are so much stronger that isobaric
slopes are very small and there is no confusion. Indeed it is shown in appendix 7.8
that the resulting expression for the horizontal pressure gradient force is nicely
simplified by the use of isobaric surfaces.
The elevation of the 1000 mbar surface varies little with latitude or season in
Fig. 4.5, but there is a slight downward slope toward low latitudes, a broad plateau
in the subtropics and an equally broad depression in middle latitudes. In fact these
slight pressure differences are associated with quite dramatic differences in
weather and climate, for example between the calm, dry subtropics and the
disturbed and wet mid-latitudes. However, the mechanisms involved in these asso-
ciations are quite subtle, depending as they do on the nature of the dominant
weather systems (Chapters 11 and 12). For the present we can simply say that the
1000 mbar surface is nearly horizontal from low to high latitudes, i.e. that the
average pressure at MSL is nearly uniform.
The pressure distribution in the middle troposphere is represented by the
500 mbar isobar, which lies between 5 and 6 km above MSL and differs
considerably from the 1000 mbar isobar in meridional profile. Figure 4.5 shows
that the 500 mbar surface slopes downward consistently from low to high
latitudes, and that there is a considerable fall in altitude in middle and high lati-
tudes in passing from summer to winter. Actually these variations are large only in
a relative sense; a slope of 600 min nearly 60 degrees of latitude (over 6000 km),
and a fall of 400 m in six months, are each very modest, and are only clearly
observable because the radiosonde network is operated so meticulously, for
example by allowing for variations of g with latitude (Appendix 4.3 and section
7 .3).
Combining all parts of Fig. 4.5, we see that the level of the 500 mbar surface
falls by about 15 OJo between low and high latitudes in winter, and by about half this
amount in summer, whereas the 1000 mbar surface is practically level. It follows
that the vertical depth of the lower half of the atmosphere (by mass) varies by the
same amounts as the 500 mbar level, as must its volume per unit horizontal area,
and hence its average density. Figure 4.6 shows the rather obvious origin of such
variations: the mean temperature of this layer falls consistently and significantly
from low to high latitudes, the fall being steepest in winter because high latitudes
are coolest then.

Fig. 4.6 Monthly and zonally

averaged isotherms in
January and July. Average 200
tropopause locations are 200
300 ~==-300
shown by thick lines, and .j--....1.....-~

areas above 0 oc are shaded.

500
500 mbar
Note the consistent vertical
700 700
temperature lapse throughout
the troposphere, the
concentration of tropospheric
baroclinity in mid-latitudes,
especially in winter, and the
reversal of normal baroclinity
about the 200 mbar level in
200 200
summer. (Redrawn from [27])
300 300
500 .,_____ __ 500 mba r
700 700
1 ooo ~~~~~~~~~~~~~~uz~
90°
Northern hemi sphere Southern hemisphere
84 J uly
 The relationship between layer depth (or thickness as it is conventionally called) 4.6 Equator to poles
and layer mean temperature can be quantified by considering the expression for the
appropriate scale height. The binary scale height H 2 is relevant in the case of
Fig. 4.5 since the bounding isobaric surfaces differ in pressure by a factor of two,
and according to Appendix 4.4 the expression for H 2 is
H 2 = 0.693RT/g.
H 2 has a value of nearly 5.5 km when the layer mean temperatureis273.2 K (0 °C}.
The crucial point about this, and all atmospheric scale heights, is that the layer
thickness is proportional to the layer mean temperature on the absolute scale. The
meridional gradients of the thickness and layer mean temperatures are therefore in
direct proportion, as are their seasonal variations.
In view of the sizeable and extensive horizontal pressure gradients apparent in
the middle troposphere in Fig. 4.5, it is perhaps surprising that sea-level pressure is
so nearly uniform. Such uniformity clearly shows that the atmosphere tends to
distribute its total mass quite evenly despite significant concentrations and
depletions localized in some layers.

TEMPERATURE

Figure 4.6 depicts the meridional distribution of air temperature up to the I 00 mbar
level in summer and winter. Note that this layer contains 90o/o of the mass of the
atmosphere and that its thickness is therefore one decadal scale height.
The distribution of the tropopause in Fig. 4.6 indicates that in low latitudes the
layer coincides very closely with the full depth of the deep troposphere there,
whereas in middle and high latitudes it includes a substantial part of the low strato-
sphere above the much lower tropopause. The troposphere is deepest in low
latitudes fairly obviously because of the deep and vigorous convection maintained
by the strong solar heating there. But although the surface layers in low latitudes
are amongst the warmest anywhere on Earth, the great depth of the troposphere
there, together with the typically steep tropospheric lapse rate of temperature,
produces the rather surprising result that temperatures at the 100 mbar level are
lower there than at any other latitude. In fact the poleward lapse of temperature,
which dwellers in middle and high latitudes regard as the natural order of things,
reverses when we consider levels which lie sufficiently above the mid-latitude
tropopause. In summer, when the low stratosphere in middle and high latitudes is
relatively warm because of the long hours of daylight, the poleward temperature
lapse reverses above the 200 mbar level, and the equatorward lapse exceeds 30 oc
at 100 mbar.
Recalling the connection between thickness and layer mean temperature, it is
clear that the poleward downslope of isobaric surfaces discussed previously in
relation to the 500 mbar surface must persist and increase up to somewhere
between the 300 and 200 mbar levels. Above this the increasing depth of relatively
warm stratosphere in middle and high latitudes progressively reduces the isobaric
slope, reversing it in much of the summer hemisphere.
Comparing Figs. 4.5 and 4.6, it is apparent that the patterns of isotherms and
isobars are similar in the troposphere, each tilting downward from the subtropics
toward the pole. However, the slopes of the isotherms are much larger than those
of the isobars, often by an order of magnitude. It follows that isotherms and
isobars must intersect throughout the region, the intersections being most
numerous in middle latitudes in winter. An atmosphere with intersecting isotherms 85
4 State and climate and isobars is said to be baroclinic, whereas it is called barotropic when the
isotherms and isobars are parallel. It is apparent that the troposphere as depicted in
Figs. 4.5 and 4.6 is nearly barotropic between the tropics but definitely baroclinic
in higher latitudes, especially in the winter hemisphere. As will be mentioned at
many points in the rest of this book, the distinction between the near barotropy of
low latitudes and the strong baroclinity of middle latitudes is basic to the quite
different dynamic behaviour in these regions, and the consequent differences in
weather and climate. Note that baroclinity is most generally defined by the inter-
section of isobars and isopycnals (lines or surfaces of uniform density) since
density is a more fundamental factor dynamically than temperature. However, the
behaviour of air as an ideal gas means that the above discussion of baroclinity and
barotropy can be repeated with isopycnals replacing isotherms at every stage.

4. 7 The general circulation

In many geographical regions even quite casual observation indicates that winds
have a preferred direction which either persists throughout the year or varies
seasonally. This implies that within these periods the incessant variations of
direction associated with weather systems do not cancel out on averaging, but leave
a bias towards some particular directions. For example, surface winds in middle
latitudes show a pronounced westerly bias, in that winds with azimuths lying
between 181 and 359° are more common and stronger than winds with a
component from the east. Such observations reveal the presence of mean motion in
the troposphere which is amply confirmed by quantitative measurement: when
flow at a particular location is averaged over time periods much longer than the
longest-lived weather system (which in practice means averaging over at least a
month, and preferably a season), a residual speed and direction often remain. The
global distribution of these residuals defines the mean flow over the month, season
or year, according to the averaging period.
Let us consider the steady component of the general circulation in and just above
the troposphere. Mean flow is everywhere nearly horizontal. Clearly this is not true
close to steeply sloping land surfaces, but only a tiny proportion of the total tropo-
spheric volume is affected by such slopes. Everywhere else mean wind speeds are
three or more orders of magnitude larger in the horizontal than they are in the
vertical; indeed the latter are so small that they often have to be inferred rather than
measured directly. Such asymmetry is of course yet another consequence of the
highly squashed configuration of the atmosphere.
Many important features of mean horizontal flow appear on the type of vertical
meridional section already used to display pressure and temperature. Consider first
the zonal components of mean flow, i.e. components of flow perpendicular to
meridional sections (Fig. 4.7).

WESTERLY COMPONENTS

Notice that throughout the troposphere in middle and high latitudes westerly com-
86 ponents predominate strongly over easterly components. Surface winds are
 Summer Winter
4.7 The general circulation

Fig. 4.7 Seasonally and

zonally averaged zonal wind
speeds in the northern
hemisphere in winter and
summer, with isotachs
labelled in m s - 1 and regions
with easterly components
shaded - the rest being
westerly . The local verticals
Pressure Pressure
represent pressure labelled in
(dbar) Northern hemisphere (dbar) decibars (1 dbar ~
100 mbar) . (After [25])
westerly in middle latitudes, in agreement with everyday experience, but even more
extensive and intense westerlies are found in the upper troposphere, with maximum
wind speeds occurring between the 200 and 300 mbar levels. In winter there is a
single very strong maximum in the high troposphere somewhat poleward of the
tropics, where seasonally and zonally averaged wind speeds exceed 40 m s -I
(nearly 80 knots) in the northern hemisphere, and are even higher in the southern
winter hemisphere. In summer, the maximum is considerably weaker and lies
substantially further poleward- a seasonal displacement which is much smaller in
the southern hemisphere. In both seasons and hemispheres a flange of strong
westerly flow extends poleward over middle latitudes in the middle and upper
troposphere. The scene being described is really very impressive: in each hemi-
sphere a vast belt of the troposphere is moving from west to east at speeds which
even after seasonal and zonal averaging would be equivalent to gale force at the
10 m level in the summer and far more than that in winter. Each belt is entitled a
circumpolar vortex, a name which nicely points to its continuity around the
hemisphere and truly enormous scale.
Note that in middle and high latitudes the maximum winds occur in the upper
parts of the baroclinic zones mentioned in the previous section, and that the
circumpolar vortices are rotating in the same direction as the Earth itself, though
faster. Each of these features is related by dynamics involving the rotation of the
Earth in ways which can be quite subtle. We will soon outline the simplest of these
mechanisms, but the more subtle must await a fuller discussion of dynamical
principles (Chapter 7) and later descriptions of the behaviour and energetics of
large-scale weather systems (Chapters 11, 12 and 13).
Above the high troposphere, mean westerly wind speeds fall away quite sharply,
at least in the summer hemisphere, where mean winds become easterly everywhere
above the lower middle stratosphere (i.e. above the 25 km level). In the winter
hemisphere mean winds remain westerly at all levels poleward of the tropics and
increase again with height above a minimum between the 20 and 30 km levels. As in
the mid-latitude troposphere there is a conspicuous association between
baroclinity and zonal flow: westerly flow increases as we rise through a layer with
poleward temperature lapse, and decreases when the temperature gradient
reverses.

EASTERLY COMPONENTS

In Fig. 4.7 easterly components are apparent at low latitudes in both the low and
high troposphere, and at high latitudes in the low troposphere. Though the high
latitude easterlies are locally significant, they cover only about 50Jo of the total 87
4 State and climate surface area of the Earth, as compared with the 4007o or more of the surface covered
by low-latitude easterlies.
The surface and low troposphere easterlies of low latitudes are the zonal
components of the trade winds, so called because their day-to-day persistence is so
high that they were relied upon in the days of commercial sailing ships. Their weak-
ness and confinement to very low altitudes make them largely irrelevant to modern
airborne trade. Unlike the westerlies of high latitudes and levels, the trades are not
true zonal winds, having substantial and systematic meridional components to be
described shortly. The marked weakening of the zonal component of the trades in
summer is confined to the northern hemisphere and is a consequence of the gross
distortion of atmospheric flow patterns by the southwesterly summer monsoon of
southern Asia (section 12.2). The summer easterlies in the high troposphere in low
latitudes are a downward extension of the easterlies of the summer stratosphere,
which in turn are associated with the meridional temperature contrast between the
cold winter and the warm summer hemispheres at these altitudes.

Fig. 4.8 Schematic plan view

of airflow in the low MERIDIONAL COMPONENTS
troposphere in the Hadley
circulation. The trade winds As depicted in Fig. 4.8, the trade winds converge towards equatorial latitudes, i.e.
{TW) blow from the
subtropical high-pressure
they have an equatorward meridional component of motion as well as an easterly
zones {STH) to the component. Figure 4.9 shows that seasonal mean equatorward speeds of flow in
intertropical convergence these latitudes exceed 2 m s -t throughout the year, and 3 m s- 1 in the winter hemi-
zone (ITCZ), where they feed sphere. Another pronounced meridional flow appears in the high troposphere in
the cloudy updraughts of
tropical weather systems. The
very symmetrical flow shown
is found when the ITCZ lies
near the equator; when it
does not, the cross-equatorial
flow is deflected to become
roughly south-westerly in the
northern hemisphere {for
example).

Fig. 4.9 Seasonally and W inter

zonally averaged meridional
wind speeds in the northern
hemisphere in winter and '2
"' 3
summer. Axes and labelling .0

are as in Fig. 4.7, with .,

.::; 5
shaded areas distinguishing
~ 7
northerly {i.e. from the north)
from southerly components.
.."
Notice that speeds are much ~ 1 0 +-----~--,--_u~~~~~~~~
90°
lower than in Fig. 4.7, and
that patterning is strongest in
the region dominated by the Summer
Hadley circulation. (Redrawn
from [27])

88 Lat i tude
the same low latitudes, directed polewards. These equatorward and poleward 4.7 The general circulation
flows are the horizontal branches of an enormous pair of meridional circulations
(one in each hemisphere, though not quite symmetrical about the Equator) shown
schematically in Fig. 4.8 and named after George Hadley whose discussion of 1735
was one of the earliest successful mechanistic descriptions of atmospheric motion.
Each Hadley cell is completed by a rising branch in the rainy equatorial zone, and a
descending branch in the arid subtropical zones (see section 8.13 and Fig. 8.14).
The zone in which the opposing trade winds converge is known as the intertropical
convergence zone or ITCZ, and this convergence clearly feeds air into the rising
branches. Similarly the diverging flow near the surface in the subtropics removes
the air which is descending there. The descent and divergence are associated with
the zones of elevated sea-level pressure often called the subtropical high-pressure
zones, and these coincide with the great warm and arid zones which virtually girdle
the Earth in the subtropics of each hemisphere. Although the consequences of
vertical motion are immediately obvious in the association of ascent with the cloud
and rain of the ITCZ, and of descent with the arid subtropical highs, the horizontal
areas over which the ascent and descent are spread are so large that mean vertical
speeds are too small to be observed directly, and have to be deduced by various
indirect methods.
In Fig. 4.8 it is apparent that the low-level equatorward flow of the Hadley
circulation is twisted westward. Indeed, since the mean meridional and zonal
components are comparable in strength, the resultant trade winds must be roughly
north-easterly in direction in the northern hemisphere, and south-easterly in the
southern hemisphere, directions which are frequently used to name the trades in
their respective hemispheres. As in the case of the intertropical convergence zone,
the lack of symmetry about the equator is associated with the summer monsoon in
southern Asia (section 12.2). In the high troposphere too, the meridional motion is
accompanied by zonal motion, though here the eastward flow is so much stronger
than the poleward flow, particularly in the poleward half of each Hadley cell, that
the result is a poleward spiral in which air parcels encircle the Earth at least once
between their equatorial origin and subtropical end.
Throughout the rest of the troposphere, mean meridional components of
motion are very small. There is some evidence in Fig. 4.9 of a relatively weak
meridional circulation poleward of the Hadley cell but in the opposite sense. This is
called the Ferrel cell, but though it seems to be quite significant energetically
(section 13.3), it is so weak that motion at these and higher latitudes is almost
completely dominated by the sporadic, intense air movements associated with
middle-latitude weather systems. Even in low latitudes, the relevance of the Hadley
circulation to day-to-day patterns of air flow and weather conditions must not be
exaggerated. Here too, though less obviously than in higher latitudes, the instan-
taneous distribution of rain and sunshine, and gentle and strong winds, is
explicable largely in terms of the distribution of a range of transient weather
systems, each of which may be associated with large deviations of local conditions
from the mean (Chapters 11 and 12). The Hadley circulation is very important, but
it is so largely because it exerts a controlling influence over the distribution of these
weather systems.

ZONAL AND MERIDIONAL COMPONENTS

The association of westerlies with poleward motion and easterlies with equator-
ward motion is quite evident in the Hadley circulation, and arises because of the 89
4 State and climate rotation of the Earth, and the movement of air across its spherical shape. Air at low
latitudes at rest on a weather map is actually moving very quickly eastward because
of the Earth's rotation - at over 460 m s- 1 on the equator (appendix 4.5). At
higher latitudes the eastward motion is smaller because points on or near the
Earth's surface are closer to its axis of rotation. For example, the eastward speed at
latitude 30° is only 402 m s - 1• It follows that air moving from equatorial to higher
latitudes will tend to preserve its initial eastward momentum (unless it is acted on
by a significant zonal force) and hence develop a westerly component of flow
relative to the weather map. In fact the effect is even more pronounced than the
58 m s- 1 implied by the above speeds, since a fuller discussion of rotational
dynamics shows that the quantity preserved is the product of the tangential
momentum and the length of the normal to the axis of rotation - the angular
momentum. When this is included (section 7 .14), air starting at rest on
the equator will arrive at latitude 30° as a westerly wind (relative to the weather
map) with speed 134m s- 1 • In fact air seldom moves so far poleward on a grand
scale, and there is some drag exerted on the air in the upper troposphere by types of
friction arising from air flow over mountains, and from deep convection. Never-
theless, the observed subtropical maximum in mean zonal wind speeds in the upper
troposphere is maintained essentially by the tendency toward conservation of
angular momentum in the poleward-moving branch of the Hadley circulation- as
described more fully in section 7 .14.
The above argument implies equally that air moving from rest at higher latitudes
should develop easterly winds as it moves to lower latitudes. Air descending near
the bases of the subtropical high-pressure zones usually has little zonal motion
relative to a weather map, having lost its earlier eastward momentum during the
long descent in the high-pressure systems. As it flows equatorward it therefore
develops an easterly component of flow. However, this is kept in check by intense
turbulent friction with the adjacent underlying surface, to the extent that
meridional and zonal speeds remain comparable. The modest easterly components
of the trade winds are the result.

4.8 Weather and climate

The relation between weather and climate is effectively that between instantaneous
and mean conditions of the atmosphere. In considering this relation we must avoid
the common tendency to consider average conditions, such as those described in
the previous sections, to be the continuous background on which individual
disturbances are superimposed. Such a distinction may be useful in the case of
disturbances whose associated deviations are small and symmetrical about the
mean. But although this is at least partly the case with the small, rapid fluctuations
which constitute much of the turbulence endemic in the atmospheric boundary
layer, it is of little value when considering the relationship between means and
deviations on space and time scales applicable to the synoptic and larger scales. For
such purposes we must be aware that mean conditions already contain the effects
of a large range of important disturbances. To make this point explicitly and draw
attention to some important aspects of tropospheric disturbances, let us re-
examine several features of the mean conditions outlined in the previous three
90 sections.
 The most striking feature of the temperature distribution in Fig. 4.6 is the 4.8 Weather and climate
strongly baroclinic zone in middle latitudes. However, on any particular meridian
and at any instant, this baroclinity is much less evenly distributed than it is on a
seasonal, zonal mean. In fact baroclinity is usually concentrated in one or more
narrow fronts separating broad, nearly horizontally uniform (barotropic) air
masses. A well-marked front is always associated with an extratropical cyclone, of
which it forms a vital part. The behaviour of extratropical cyclones and their
associated fronts is quite complex, though some aspects have become familiar to
the general public through forecasts in television and newspapers. A fuller
description is given in Chapter 11, where it is shown that a sharply defined front
persists for several days in the form of an intensely baroclinic zone -2000 km long
and 200 km broad, having almost any orientation and moving obliquely at speeds
- 10 m s _,. The transience, orientation and movement of fronts result in their
intense baroclinity being smeared out and greatly weakened by zonal and seasonal
averaging. The broad baroclinic zone of Fig. 4.6 is therefore not to be regarded as a
persistent feature to be found at all times and meridians in middle and high
latitudes, and only incidentally disturbed by the sporadic appearances of extra-
tropical cyclones and their associated fronts; instead it is the mean result of the
sporadic appearance of these disturbances over a wide latitude range.
Similar remarks apply to the meridional distributions of mean surface pressure
apparent in Fig. 4.5. Inhabitants of middle and high latitudes are familiar with the
incessant variations of surface pressure associated with the development and
passage of large-scale areas of low and high pressure associated with great weather
systems. Any one such low-pressure system would appear on an instantaneous
section along a particular meridian as a very sharp and deep dip in the 1000 mbar
isobaric profile. In fact, comparing it with the relatively smooth profile on
Fig. 4.5, the sharp dip would be -300 m deep (corresponding to about 30 mbar on
a horizontal surface) and 10° of latitude in meridional extent. The continual
development and decay of several such extratropical cyclones (the low-pressure
systems) and anticyclones (the high-pressure systems) around each latitude zone is
further smoothed over the seasonal averaging period, to give the broad, shallow
depression evident in Fig. 4.5.
The situation is somewhat different at lower latitudes, however. The high
average pressures apparent in the subtropics in Fig. 4.5 are actually considerably
more nearly representative of individual instantaneous meridional profiles than
the previous remarks might seem to imply, because they result from the presence of
unusually stable and persistent weather systems - the large subtropical anti-
cyclones which effectively girdle the Earth at these latitudes.
The pattern of mean westerly winds in Fig. 4. 7 is similarly related to patterns on
instantaneous meridional sections. The core of maximum wind speeds at about the
200 mbar level in the subtropics is associated with the nearly continual presence of
the powerful subtropical jet stream there, which represents a concentration of the
westerly flow maintained by the conservation of angular momentum as described
in the previous section. Though the intensity and latitude of this impressive stream
of air vary somewhat with time and longitude, the variations are sufficiently small
for the mean pattern to be only moderately weakened and spread compared with
the instantaneous local patterns. The subtropical jet stream is at its most intense in
the winter months, moving poleward and weakening considerably in the summer
months, in obvious association with variations in the Hadley circulation. The sub-
tropical jet was the last of the major features of tropospheric circulation to be
established by observation, becoming well known only during the Second World
War, when military aircraft flying westward at high altitude in the vicinity of Japan
consistently reported speeds over the ground dramatically less than their indicated 91
4 State and climate air speeds. It is now known that the subtropical jet stream is especially intense in
that region because of the influence of the Asian continent in general, and the
Himalayan and Tibetan massifs in particular, the last two obstructing flow in the
middle and high troposphere over an extensive region and producing a pronounced
confluence in their wake.
The poleward extension of the westerly maximum across middle latitudes is
similarly associated with the presence there of the polar front jet stream. Like the
fronts with which it is intimately connected, the polar front jet is only partly con-
tinuous, and is narrow and highly mobile, with the result that the mean section
gives no useful impression of the appearance of individual instantaneous sections.
In the latter the polar front jet appears as a relatively narrow stream of fast-moving
air centred at about the 300 mbar level and closely parallel to the associated fronts,
whose wave-like sinuosity it closely matches (Fig. 11.10). Indeed it will become
apparent later (section 7 .10) that the westerly wind maximum in mid-latitudes is
related dynamically to the baroclinic zone in those regions, and that this relation-
ship applies both to averaged and instantaneous sections.
Obviously it is potentially very significant that the subtropical and polar front jet
streams appear in the high troposphere in the height zones used by modern civil air-
craft. The minimize the fuel wastage in flying against such fast airflows, routes are
planned to avoid their forecast positions either by lateral or vertical displacement.
In addition, the strong vertical wind shears in the vicinity of the jet core are some-
times associated with a rather unpredictable kind of turbulence which can be
uncomfortable or even dangerous (section 10.8). Only supersonic civil aircraft and
some military aircraft normally fly well above the levels of the jet cores.

4. 9 Climate change

VARIABILITY

So far we have assumed that atmospheric variability ranges in period from the brief
fluctuations associated with small-scale turbulence to the week-long fluctuations
associated with synoptic-scale weather systems such as extratropical cyclones, with
only the relatively regular seasonal variations occurring on longer time scales. But
if this scheme accounted for all the variability, we should expect seasonal averages
to be almost identical from year to year in any one location, on the apparently
reasonable assumption that a season must contain so many randomly occurring
weather systems that averages of temperature, rainfall etc. will be well defined.
Observation and experience show that this is not the case. For example, Table 4.1
lists summer rainfall totals for ten consecutive years at a station in north-west
England. Clearly there are large year-to-year variations which common experience
suggests occur also in seasonally averaged temperature, sunshine, etc.
Such variability indicates the presence of non-seasonal periodicities in
atmospheric behaviour which are considerably longer than a week. If you live in
middle latitudes you can see at least one way in which such longer periods arise
if you follow the daily sequence of weather charts. Occasionally an anticyclone
becomes established in your particular region and persists there for several weeks,
replacing the normal sequence of eastward-moving weather systems by an unevent-
92 fullull which may be warmer or cooler than the usual seasonal average depending
Table 4.1 Summer rainfall totals at a coastal station in northwest England (Hazelrigg 54.0 °N, 4. 9 Climate change
2.8 °W, 92 m above MSL)
Data are collected over June, July and August in each year

Year Total in mm Year Toto/ in mm

1975 223 1980 295

1976 99 1981 254
1977 218 1982 276
1978 305 1983 152
1979 229 1984 165
10-year average 222mm
maximum year to year] + 83 i.e. + 37%
deviations - 123 - 55
Standard deviation u. 63 28

on season, the region in question, and the location of the centre of the anticyclone
relative to local geography (section 11.5). Such anticyclones are known as blocking
anticyclones or blocks because of their apparent role in blocking the movement of
other weather systems - but beware of the spurious implication of cause and
effect.
If events such as blocks, lasting for half a season at most, were the longest-lived
of the range of transient irregular weather systems found in the atmosphere, we
should expect annual averages to be considerably more stable than seasonal
averages, and averages over a sufficient number of consecutive similar seasons (for
example ten consecutive summers) to be virtually invariant. Averages over ten con-
secutive full years should be even more stable. Though annual averages are a little
more stable than seasonal averages, they are still quite significantly variable
(Fig. 4.10). For example, the winter of 1963, which was the coldest for 200 years in
the British Isles because of the persistence of a winter block (section 11.5), was
followed by a relatively cool summer, with the result that the annual average
temperatures for 1963 as a whole were quite unusually low. It seems that the atmo-
sphere must be prone to disturbances more persistent than anything associated
with the longest-lived weather systems identifiable on conventional weather maps
(the blocks). And since such hidden disturbances can apparently influence
conditions over several years, it is natural to ask if there is evidence of others acting

18
Fig. 4.10 Annual and
17 summer averages of screen
level temperatures in central
16
... England since 1940. (After
15 [28])

0 14 June, July , Augu s t

e. average

"' 13
"
~ 12
Q)
a.
E 11
~
c • Annua I a verage
10
e
Q)

0 9 ..
"' 8
1940 1950 1960 1970
Year
93
4 State and climate over even longer periods. Examination of climatological records confirm that there
is such evidence, and that there is no obvious upper limit to the lengths of
associated time periods.

CENTRAL ENGLAND TEMPERATURES

In Fig. 4.11 temperatures in central England are extended backward in time to the
first sustained meteorological use of thermometers in England (1659). This record
is one of the longest series of direct meteorological measurements anywhere in the
world, and of course pre-dates the beginning of the synoptic meteorological
networks by nearly two centuries. Early measurements were particularly prone to
error in instrumental performance and exposure, and it is fortunate that in recent
times people have been prepared to devote painstaking, unglamorous effort to the
collation, comparison, quality assessment, and even judicious correction of such
data, since without such care the data would be much less reliable [28].

Fig. 4.11 Ten-year running 11 11

means of annually averaged
scr-n temperature in central T<:~f 1~t;:?\:::::::~~o
England since 1659. Data are 1700 1800 1900
most reliable since the Decade ending
institution of the British
synoptic observation system
about 1860 and are least
reliable before 1700, because To smooth out fluctuations lasting only a year or two, the data have been
of poor observation practice. subjected to a ten-year running mean, and yet there is still very considerable
(Redrawn from [28]) variablity. For example there was a pronounced cold period centred around AD
1700 in the British Isles, and a prolonged warm period peaking around 1950.
Despite the obscurity of their origins, and the apparent smallness of their magni-
tudes, such variations can have pronounced social effects through agricultural
production and in other ways. For example, the very cold period at the end of the
seventeenth century was the culmination of decades of cooling so widespread and
intense that it has become known as the little ice age, though it does not begin to
compare in scale and intensity with the true ice ages of recent geological epochs
(Fig. 4.12). Nevertheless in Scotland the harshness of the climate caused such
distress, directly and indirectly by associated famine, that there was large-scale
emigration in search of better conditions. A major flow of Scots was directed to
north-east Ireland, where conditions were slightly easier and good agricultural land
was made available by forcible expulsion of the native Irish, with social and
political consequences which are distressingly evident to this day [29]. On a lighter
note, the short cool period centred around AD 1820 coincided with the most
impressionable phase of the childhood of Charles Dickens, and it is very probable
that his early memories of the several harsh winters which occurred then
encouraged his subsequent colourful evocations of winter and Christmas, and at
another remove gave rise to the Christmas card industry of today.

CLIMATE CHANGE AND PREHISTORY

Direct meteorological measurements hardly existed before the mid-seventeenth

94 century, but current interest in climate change (the name given loosely to all
 4.9 Climate change
16

12

6
0
~
1!:I 4
;
Q; 0
c.
.,
E
.... -4 Fig. 4.12 Near·surface July
temperatures in lowland
-8 Britain in the last 120 000
years as estimated by G.R.
-12 Coope from evidence of
20 16 12 8 4 0 insect life. (Redrawn from
Thousands of years Before Present [28])

c Fig. 4.13 Near-surface

~ 20 .!1
' II -;;; temperatures in central
~
1!:I
'II
I
.
c."
England in the last 20 000
years, estimated largely from
-;;; I
" 'I E pollen analyses. (Redrawn

.
I
:;; I
I I
,, I I
:I

."
c. \ I .§ from [29])
________ .... /
\ / II I
E '' '
! 10 ,_ I
I
\ ::!i
>- '
:;
...,
120 100 80 60 40 20 0

Thousands of years Before Present

climatic variations lasting more than a few years) is so strong that an immense
programme of indirect detective work is now under way. Old social records are
combed, and a wide variety of physical, chemical and biological clues are sifted for
information about past climate. For example, the analysis of pollen grains
embedded in buried soil and rock strata shows what flora were growing when those
strata were at the surface, and knowledge of the climatic tolerances of the flora
then indicates the climate prevailing. Figure 4.12 includes summer and winter
temperatures in central England over the last 20 000 years estimated by this
method, and shows clearly the rapid rise of temperature associated with the ending
of the last glaciation, one of whose ice sheets extended roughly to central England.
The plotted temperatures are believed to be the same as would have been logged by
thermometers subjected to a 500-year running mean. Figure 4.13 depicts estimates
of temperature in lowland Britain derived from analysis of insect fauna over the
last 120 000 years, and shows that the present warm spell is one of a very few
separated by long periods of extreme cold and associated widespread glaciation.
This period of repeated glaciation of high latitudes extends back several million
years, before which there was a prolonged warm period in which there was little or
no permanent ice even at the poles. Further back still, there is fossil evidence of a
long period of ice ages in the Permian (section 3.8 and Fig. 3.12), and again in the
Precambrian.
It is apparent from all these data that climatic conditions are variable on almost
any timescale we choose to examine. Since only the very shortest of these
correspond to the ranges which are the special concern of traditional meteorology, 95
4 State and climate it follows that there is a huge range oflonger-lasting events demanding more or less
unorthodox explanation. In recent years many people have responded to this
challenge, with the result that the field of climate change abounds in speculation
and theory. Since much of this is complex and incomplete, and since the subject is
in any case too large and specialized to be treated at all thoroughly in the present
text, we will consider only a selective outline of some current ideas. Several good
specialist introductions are now available [30].

MECHANISMS OF CLIMATE CHANGE

Broadly speaking two types of mechanism for climatic change are envisaged -
external and internal. The search for external causes obviously focuses attention on
the Sun as the ultimate powerhouse of the atmospheric engine. How variable is the
solar output, and do any of its variations correlate clearly with recorded climatic
variations? Unfortunately the solar output has been measured with useful accuracy
only since the early years of the twentieth century. Variations are small (- 1OJo) and
seem to correlate with sunspot numbers. These in turn have been recorded for
several centuries by astronomers and show a pronounced quasi-regular eleven-year
cycle with longer-term fluctuations superimposed. There is some evidence to
suggest that sunspot minima (probably corresponding to slightly lower solar
output) have been associated with lower global temperatures in that period. In
particular the temperature minimum of the little ice age corresponds with a period
between 1650 and 1700 AD when there seem to have been very few sunspots indeed.
Variations in the movement of the Earth relative to the sun significantly affect
the amount and global distribution of solar input to the Earth (Fig. 4.14). The
ellipticity of the Earth's orbit is quite small at present, but nevertheless produces a
60Jo range in solar input over the yearly cycle. However, the ellipticity itself goes
through a cycle lasting about 100 000 years, in which it reaches a maximum cor-
responding to a 300Jo annual range of solar input. Secondly, the Earth's axis of
rotation moves, varying the season of closest approach to the sun (perihelion). For
example perihelion now occurs in the northern hemisphere in midwinter, since it
coincides with the time when the axis through the northern hemisphere is tilted
Fig. 4.14 The precession of
furthest away from the sun. But just over 10 000 years ago perihelion occurred in
the equinoxes. Just as the the northern hemisphere in midsummer. Following this motion of the Earth's axis
non-vertical axis of a spinning back in time through several 21 000-year cycles, there must have been a period
top precesses (swings round) when a particularly close perihelion coincided with the northern hemisphere
about its foot, so the Earth's summer (and aphelion with winter) to produce much larger seasonal temperature
axis of spin precesses about
the perpendicular to the
swings than occur now. Thirdly, every 40 000 years the tilt of the Earth's axis of
plane of its orbit round the spin relative to the plane of its orbit round the sun completes a small regular cycle
sun. As a result, the location of variation about the present value of nearly 23.5°. Clearly the meshing of these
of the equinoxes (where the three independent cycles must produce a complex variation of solar input involving
Earth is tilted neither towards
many rhythms ranging in period from a few thousand to a million years or more, as
nor away from the sun)
moves round the elliptical well as the seasonal cycle itself. These variations are named after Milankovitch,
Earth orbit round the sun,
completing a circuit in 21 000
years. Other seasonal
positions, such as the position
of the northern hemisphere
midwinter (NMW- shown in
its current position) move in
step. __ e_tl_ · - - · - ~----
NMW S

96
who first discussed their possible climatological significance (in 1930). Evidence is 4.10 Human society and
now quite strong that the Milankovitch mechanisms play an important role at least climate change
in triggering ice ages and intervening interglacials.
Some possible internal causes of climatic variation are easily identified, at least
in principle. Since the Earth's atmosphere filters the incoming sunlight before it
reaches the surface and so enters the atmospheric engine, it is clear that anything
significantly affecting that filtering could be important climatically. Volcanoes do
this sporadically, and sometimes very dramatically, by spewing fine, reflective dust
into the stratosphere, where it drifts over the globe for months or years, reducing
the solar input to the troposphere and surface beneath. For example, a high veil of
dust was observed in many parts of the globe for several years after the cataclysmic
eruption of Krakatoa (Indonesia) in 1883, and the especially vivid sunsets of the
painter Turner come from a period in the early nineteenth century when there was a
great deal of volcanic activity in several widely separated areas. There is some
evidence that increases in volcanic dust in the atmosphere have been associated
with transient coolings of the atmosphere. This connects the atmospheric engine
with the independent and much more ponderous engine working in the Earth's
interior and producing varying amounts of volcanism as a kind of incidental
surface 'spray'. It seems probable that there have been periods in the past when
volcanism has been much more vigorous than at present and climatic effects may
have been correspondingly large.
More subtle internal mechanisms involve inherent chaotic (section 1. 7)
instability in the atmosphere/surface/ocean system. These may operate
sporadically, just as the baroclinic instability of the troposphere erupts from an
inherent instability, though much more sluggishly. For example, it is possible that
the opposing extremes between glaciation (very large polar ice caps) and no glacia-
tion at all, may represent two possible conditions of near equilibrium for the atmo-
sphere/ocean/ice cap system, like a dented metal tray which may contain a ball
resting in either of two shallow depressions. The system may move between these
extremes in response to variations in volcanism and to the Milankovitch
mechanism, and moreover may be especially prone to maximum glaciation when a
major continent is at one of the poles, as Antarctica has been for the last 30 million
years; and of course the continents themselves move significantly in time intervals
greater than about ten million years, again in response to the Earth's internal
engine. In terms of the ball-and-tray analogy, continental movement has the effect
of warping the tray. Another example of inherent instability involves interplay
between the atmospheric and oceanic systems. The large heat capacity of the
oceans must tend to make any variations much more sluggish than those of the
atmosphere alone, and indeed there is evidence that climatic variations lasting a
few years are associated with anomalous temperatures over extensive parts of
ocean surfaces [31].

4.10 Human society and climate change

GLOBAL WARMING

As mentioned in section 3.4, human industry is probably affecting global climate

by increasing the abundance of C0 2 and other gases controlling the temperature of 97
4 State and climate - +0.5 (aJ
~
Q)
_:;
as-
:><II
<::~
<=G)
0.0
ala.
E
"'-
-Q)

.0
Fig. 4.15 (a) Global annual
a ~-o.s
-
QQ)

average surface
temperatures since 1854 ~
:::>
expressed as o difference Ul
60 80 1900 20 40 60 80 2000
from the 1950-79 average.
Note how the ten-year
running mean smooths the 60
Iorge inter-annual variations E
(.)
and highlights longer periods. 50
Q)
(After (32]) (b) Highest, Ul
·;:
lowest and best estimates of
global sea-level rise in the
a; 40
>
next 60 years. About half .!!!
comes from melting ice and <II
Q) 30
Ul
half from the expansion of
<::
warming sea water, both of <II
which ore so sluggish that
Q) 20
E
even if global warming a;
stopped at AD 2050, the sea .0
10
a
0
would continue rising for a
further ? centuries, doubling
or trebling the rises shown. 0
(After [ 18]) 1990 2000 2010 2020 2030 2040 2050

the Earth's surface and atmosphere through the greenhouse effect (section 8.5). So
far the effect is too small to be clearly identified amongst the natural variations of
climate over fairly short periods (10-100 years). Figure 4.15(b) shows a series of
global temperatures which highlights the problem. Is the rising trend apparent
from the late nineteenth century something new, or is it a recent stage in the
continuing recovery from the little ice age? And note the problem of constructing a
single series from a global distribution of temperatures, from areas with different
climates and climate changes. Some have argued that a spurious rise in Fig. 4.15
may come from the proximity of many contributing observations to cities which
have expanded greatly in this time period, spreading their heat island effects
(though this has not been supported by subsequent detailed analysis).
Sophisticated theoretical models, similar to those used for routine five-day
weather forecasting, but run for very much longer simulated time periods, now
strongly suggest that an anthropogenic increase in the greenhouse effect will soon
produce clearly identifiable global warming. These models predict temperature
increases in the next century ranging from about I to 6 °C, depending on the model
[33]. Differences between models arise from uncertainties in modelling the wide
range of physical and chemical mechanisms at work, in particular those giving
strong feedback (e.g. those involving water vapour and cloud - section 8.5).
However, even the smallest predicted warming (ignoring a very few models which
predict no change or even a cooling) would be very significant on a global scale, and
the largest would take the Earth to its warmest condition since the onset of repeated
glaciations of the northern hemisphere over 3 million years ago.

98
LOCALIZED EFFECTS 4.10 Human society and
climate change
However large the global climate change, some localized effects will be much
larger. All models of the changing general circulation during global warming show
maximum warming at high latitudes, as poleward heat fluxes in mid-latitude
weather systems (section 8.12) increase in the warmer, moister atmosphere. The
warming predicted there is so large that melting of polar ice caps will contribute
substantially to a global rise of sea levels, already rising through the expansion of
the warming seas (Fig. 4.15). Again, the magnitude of the rise varies greatly with
the climatological model used, and local rises are very sensitive to local adjust-
ments of land height. Indeed some land masses, like Scandinavia, are still rising so
rapidly after losing the burden of their last ice cap, that sea levels along their coasts
will continue to fall for some time.
Pronounced meridional variation of climate change arises also from the shifting
of the great climate zones (sections 11.8 and 12.5). There is clear evidence that
during the most recent glaciation, the relatively slight global cooling and the
marked cooling in middle and high latitudes evident in Fig. 4.12 were associated
with substantial poleward movement and expansion of the subtropical arid zones
associated with the descending branch of the Hadley circulation (Fig. 8.14). For
example, ice cores from the Greenland ice cap show that desert dust was carried
there in much larger quantities than is now the case. It is very probable that these
trends would be reversed by gross global warming, which could be good news for
those people currently struggling in marginal conditions in North Africa, for
example.
Wind-driven ocean currents like the Gulf Stream are known to have changed
substantially in the great climatic events: the north-eastern Atlantic is believed to
have been about 10 °C cooler during the last glaciation because the Gulf Stream
and North Atlantic Drift lay well to the south of their present locations, which
encouraged the persistence of the great ice sheets on Britain and Europe. Such
air-sea interactions are potentially so important in high latitudes and low (see
section 12.5) that climate models have to cover the behaviour of the oceans as well
as the atmosphere- a major complication. And evidence of very rapid changes in
Greenland's climate during the last glaciation suggest that ocean currents can
suddenly shift as some threshold is crossed [34].

SOCIAL CONSEQUENCES

Human sensitivity to climate change can hardly be overemphasized. A general rise

in sea level will threaten low-lying coasts with increasing flooding, extending on a
smaller scale the huge inundations which drowned the North Sea and Hudson Bay
as the great ice caps melted after the last glaciation. Even quite small rises (for
example about 10 em) will force vulnerable populations to improve their coastal
defence to avoid increasing disruption during winter storms (Figs 4.16 and 11.13).
The human population is currently growing so rapidly that food production is
already inadequate in many areas, and though modern agriculture has helped
greatly in some regions, production limits imposed by land availability and social
efficiency may well be reached over much of the Earth within the next century.
Imagine the possible effects of rapid, large-scale climatic change on this tense
situation: established agriculturally productive zones (like the North American
'granary of the world' could be seriously damaged; and in the confusion there
might well be serious delay in realizing the increased potential of other zones. And 99
4 State and climate

Fig. 4.16 Modern and old-

fashioned coastal defences at
Morecambe, on the north-
east edge of the Irish Sea. A
broad, shallow bay
encourages 10 m tides and
short. steep waves which can
threaten the town, especially
when combined with a storm
surge (section 11.3). (a)
Modern defences use 'rock
armour' to diffuse and absorb
the wave energy which often
pounded older sea walls (b)
to destruction by coherent
impact and reflection
(Fig. 11.13).

recent experience in marginal lands like the African Sahel shows how easily
climatic deterioration can combine with social change to produce devastation:
fifteen years of sporadic drought following a wetter period in which the human and
animal populations boomed forced whole societies to mortgage the longer-term
future of their agriculture for the sake of short-term survival, leaving them
increasingly vulnerable to further climatic anomalies.
100 Time is the crucial factor in all this. If climate change occurs over many human
generations - for example 1000 years for major warming or descent toward Appendix 4. 1
another glaciation - then adjustment will hardly be noticed amidst all the other
changes we seem to wish on ourselves. If however there is substantial climate
change in a period as short as one or two centuries, as suggested by the more pessi-
mistic models of global warming, then adjustment will almost certainly be painful.
Faced with our evident ignorance and these possibilities, it is hardly surprising that
the study of climate change has recently become very popular after long years in the
scientific wilderness.

Appendix 4.1 Ideal gases

Consider a fixed mass of a certain gas at pressure p, absolute temperature T, and

filling a known volume V. Then the gas laws established in the seventeenth and
eighteenth centuries can be written as follows:
p ex 1/V at constant T (Boyle's law)
Vex T at constant p (Charles' law)
pexT at constant V
These can be combined in a single equation of state

p; = constant (4.9)

Through the work of Avogadro and others it became clear at the beginning of the
nineteenth century that. the constant in eqn (4.9) is proportional to the number of
gas molecules in the volume but is independent of the type of gas. In fact if there are
n moles of gas (i.e. n x M grams where M is the molecular weight of the gas), the
equation of state takes the form
pV = nR.T (4.10)
where R. is the universal gas constant, so called because the value 8.314 1 K- 1
mol- 1 applies to nearly all gases in a wide range of conditions. A gas which obeys
eqn (4.10) is by definition an ideal gas.
By using the definition of the mole we can rewrite eqn (4.10) to involve the mass
m of gas filling the volume V.
pV= 103mR.T (4.11)
M
The factor 103 is needed because the mole is defined in terms of grams whereas the
SI unit of mass is of course the kilogram. To remove the volume V, simply note that
m!V is by definition the gas density p, so that eqn (4.11) can be rewritten in the
meteorological form
p = pRT (4.12)
where R is the specific gas constant of the particular gas in question as defined by
R = 103 R.IM
According to Dalton's law of partial pressures, the total pressure produced when
a volume is filled by a mixture of ideal gases is equal to the sum of the partial 101
4 State and climate pressures which each would produce if they separately occupied the same volume at
the same temperature. Since the individual components of the mixture are ideal
gases, each must obey a form of eqn (4.12)
P; = P;R;T
where P; is the partial pressure of the component i, and P; and R; are respectively its
density and specific gas constant. Dalton's law can then be expressed in the form

p = ~ P; = ~ P;R;T (4.13)
i

p = ~ P;
i

where p and pare the total pressure and density of the mixture. Rearranging these

p =P ( :f P;PR; ) T

=p ( :f X; R; ) T (4.14)

where X; is the specific mass (p/p) of the component i. By comparing eqns (4.12)
and (4.14) we see that the mixture is behaving as a single ideal gas with specific gas
constant

R = ~ X;R;
i

With a little rearrangement it is apparent that the equivalent single molecular

weight of the mixture is

M=1~~~
; M;
The values of Rand Mare therefore known provided the specific masses and mole-
cular weights are known. The values for dry air (287 J K-I kg-I and 29.0
respectively) follow from the data in Table 3.I. Simple arithmetic shows that the
values for moist air with specific humidity 10- 2 (10 g per kg) are 289 J kg-I K-I and
28.8 respectively, showing that such air is about 0.50Jo less dense than dry air at the
same temperature and pressure.

Appendix 4.2 The hydrostatic equation

Figure 4.17 shows a vertical column of air with unit horizontal cross-section area,
which contains mass M of air between heights zi and z2 • Because air pressure acts
equally in all directions at any point, and because of the choice of cross-section
area, the upward force on the base of the segment is simply given by the pressure PI
there, and the smaller downward force on the top of the segment is p 2 • The net
upward force on the segment is therefore PI- p 2 , and in hydrostatic equilibrium
this must be balanced by the downward force of gravity Mg on the column segment
102 (i.e. its weight).
 l Vertical
up

HeightZ2
,.---
Pressure P2
Appendix 4.3

Forces
!
Net pressure
Weight

Fig. 4.17 Forces on a

vertical column of air in
Unit area hydrostatic equilibrium.

P1-P2 = Mg (4.15)
Since the segment volume is (z2 - z1), the segment mass M is given by
M = p (z2-Zt)
if the density p is uniform, so that
P1- P2 = pg(z2-zt) (4.16)
If the segment is very shallow
P2- Pt = ap
z2- z1 - az
where the right-hand side is by definition the vertical gradient of pressure, and the
approximation improves as the segment becomes shallower. In the infinitesimal
limit
ap
- = -gp (4.17)
az
which is the most generally useful form of the hydrostatic equation, because
density is no longer assumed uniform over a finite range of height. The minus sign
ensures that p falls with increasing z, as is always the case with the upward-pointing
z coordinate used in meteorology.

Appendix 4.3 Geopotential height

The work done against gravity in raising unit mass through a height interval dz is
given by g dz, and is by definition the increase in its gravitational potential energy
(geopotential) d!/1. That is
d!/1 = g dz
where g is the local apparent gravitational acceleration. Since a truly horizontal
surface is one with uniform geopotential rather than uniform height (given the
slight horizontal variations in g arising from the rotation and oblate shape of the
Earth - section 7 .4), it is useful to define a unit of geopotential (the geopotential
metre) in such a way that geopotentials relative to MSL are numerically very nearly
equal to heights in metres. This is done using 103
4 State and climate
<I>= 1/9.8 r
0
gdz

where <I> is called the geopotential height of the local surface with height z. Con-
versions between actual and geopotential heights can be calculated for any
geographical location (such as a particular radiosonde station), knowing the local
vertical profile of g values. Once this is done, actual heights may be replaced by
geopotential heights without changing the form of any equation apart from
replacing the slightly variable g by the constant 9.8. For example eqn (4.6) becomes
oplo<I> = -9.8p
Note that geopotential height is energy per unit mass, despite its name, and that one
geopotential metre is 1/9.8 m2 s- 2 •

Appendix 4.4 Exponential decay

As shown in section 4.4, the hydrostatic equation for an ideal gas under gravity can
be written in the form

-op
OZ = -p/H
e
(4.18)

If H. is a constant, this shows that the pressure gradient is directly proportional to

pressure. Equation (4.18) is then a differential equation describing an exponential
relationship betweenp and z, and the minus sign ensures that it is exponential decay
rather than growth. Equation (4.18) is identical in form to eqn (2.2), and to many
relations which arise in describing the natural world, including radioactive decay of
unstable nuclei. By integration and judicious scaling and labelling of axes we can
show that each equation has a solution which can be fitted to a common curve
which has the same shape as Fig. 4.4, except that in most examples the dependent
variable (the equivalent of p) is plotted on the vertical axis rather than on the
horizontal one (Fig. 4.18). The equation of the common curve is of the form
F=F0 exp(-x!L.) (4.19)
where F 0 is the Fvalue at zero x and

Fig. 4.18 Exponential decay X

in direct and logarithmic 0 2 La 3

0~----~----,-----~
presentation.

-1

f
Fo

-3

2 3
X
104 Le
 exp (1) = e = 2.7183 Appendix 4.4

is the base of natural logarithms. It follows that at x = Le

F = F0 exp ( -1) = F01e = 0.3679F0
The constant Leis the exponential scale length which describes the rate of decay of
the curve towards zero. In the case of the vertical profile of atmospheric pressure it
is the exponential scale height, and in the case of instrumental response (Appendix
2.1) it is the lag of the instrument in time. It follows that the reduction is 11e2
(0.1353) if the x increase is 2Le, and 1I en if the increase is nLe.
An exponential relation can always be written in an equivalent logarithmic form.
The equivalent form of eqn (4.19) is
ln(FIF0) = -xiLe (4.20)
which shows that a graph of the logarithm of F against x is a straight line with
gradient -1 I Le (Fig. 4.18), which is equivalent to - H 10 in Fig. 4.2. Figure 4.2
shows that the graph of the logarithm of pressure against height for the occasion of
the sounding in Fig. 4.1 is not quite linear, which means that the pressure-height
curve is not quite perfectly exponential - a fairly small complication which arises
from the vertical lapse of temperature in the troposphere.
We can find an expression for the decadal scale length of any exponential decay
by rearranging eqn (4.20), and using the fact that ln F = 2.3026log 10 F to produce
log 10 (FIF0 ) = -xi(2.3026Le)
Since by definition the decadal scale length L 10 is the x increase which reduces F to
F0110, setting FIF0 to 1110 shows us that
L 10 = 2.3026Le
In the same way we can show that the binary scale length L 2 is given by
L2 = 0.6931Le
In the case of radioactive decay with time, the equivalent of the binary scale length
is the half-life of the decaying species.
Although all scale lengths are equivalent, the exponential scale length has a
natural significance not shared by the others. Returning to the case of the vertical
profile of atmospheric pressure, we see from eqn (4.18) that the vertical pressure
gradient has magnitude p 01He at the level where pressure is p 0 • If this gradient were
to persist throughout the overlying layer (Figs 4.4 and 4.18), instead of dying away
along the actual exponential profile, the pressure would fall to zero in a height
interval exactly equal to He. Applying this to sea-level observations in the
atmosphere, we see that barometric pressure would be as observed if the actual
atmosphere were replaced by an incompressible 'sea' of air with depth He and
uniform density equal to sea-level air density. Realistic values for He are about
8.5 km in the atmosphere, which is just over 0.1 OJo of an Earth radius, emphasizing
again the extreme shallowness of the atmosphere. In a number of analyses of atmo-
spheric dynamics, it helps and clarifies to replace the actual compressible
atmosphere by its incompressible equivalent.
Apart from the unusually pure case of radioactive decay, natural decay profiles
in meteorology and elsewhere are seldom perfectly exponential. However, if a
profile is only slightly imperfect, the exponential solution is still very useful and can
be applied to actual measurements by allowing Le to vary slightly. This is the case
with the vertical profile of atmospheric pressure, where He varies with the absolute
air temperature (section 4.4). 105
4 State and climate
Appendix 4.5 Motion in a circle

Consider a point describing a circle of radius R in time T, at a steady tangential

speed V. Since it travels round the circumference exactly once in time T we have
V=27rR
T
Now by definition of radian measure the angle subtended by one circumference is
21r radians, so that we can regard 21r!T as the angular velocity Q of the orbiting
point in radians per second. It follows that
V= DR
which is often used to find V when Q is known, or vice versa. As an example of the
former, consider the tangential velocity of a point on the Earth's surface arising
from the Earth's rotation. The angular velocity of the Earth is
27r/(24x60x60) = 7.27 x 10~ 5 rads~'

and since its radius is nearly 6370 km we find that the tangential speed of a point on
the surface at the equator is 463.2 m s~'. At latitude cf> the radius of the circle
described by a point on the surface is R cos cf> (see Fig. 4.19), so that the tangential
speed is QR cos¢, which is simply cos cf> times the equatorial value. At latitude 30°
therefore the tangential speed is 401 m s ~ 1•

Fig. 4.19 A quadrant of the

earth showing how the radius
of rotation of a point on its
surface varies with latitude.

~'~· \
~~ R
Equator

We have just seen that the rotation of the position vector R about 0 at angular
velocity Q is associated with a velocity V which is instantaneously perpendicular to
R on the side favoured by the direction of rotation, and which has magnitude QR
(Fig. 4.20). By an identical examination of the rotation of the resulting velocity
vector Vabout 0, also at angular velocity n, it is apparent that the rate of change of
velocity is similarly instantaneously perpendicular to V. As shown in Fig. 4.20, this
direction is always toward 0 on a location (rather than velocity) diagram. By direct
analogy with the result for tangential velocity derived above, the magnitude of this
centripetal acceleration A is Q V, though either of the equivalent expressions V2! R
or Q2R is much more widely used. We have the very important result that a body
describing a circle at a uniform speed is continuously accelerating toward the centre
of the circle at a rate proportional to the square of either the speed or the angular
velocity. From previous values we find that a fixed point on the equator has a
centrifugal acceleration (i.e. toward the Earth's centre) of 0.034 m s~ 2 , which is
106 0.30Jo of g.
 Position Velocity Problems

Fig. 4.20 Position, velocity

and acceleration of a body
moving steadily around a
circle. The position diagram
describes path and velocity in
terms of a swinging radius,
whereas the velocity diagram
describes velocity and
acceleration in terms of a
swinging velocity vector.

Problems

LEVEL 1

1. If the pressure of a rising air parcel halves while temperature remains constant,
what variation do you expect in its density? Since in fact parcel temperature
always falls during significant ascent, will your first answer be an over- or
underestimate?
2. A certain very large, high-altitude research balloon measures 100 m from the
top of the envelope to the base of the instrument package. Using Fig. 3.1 as
typical, estimate how the temperatures of the base and top should compare as
it ascends from the surface to the stratopause.
3. State the conditions for a perfect exponential decay of atmospheric pressure
with increasing height.
4. Deep anticyclones have a characteristically warm core throughout all but the
highest parts of the troposphere. Qualitatively sketch the expected shape of the
1000, 500 and 300 mbar surfaces on a vertical section across the middle of such
a system. (Consider the distribution of thickness.)
5. Sketch isotherms on a vertical section through two atmospheric zones, only
one of which is baroclinic. Assume horizontal isobars, and label warm and
cold parts of each zone. Name the zone with horizontal isotherms.
6. Using Fig. 4. 7 as evidence, brief a party of inexperienced mountaineers about
likely wind conditions on Everest.
7. In middle latitudes especially, mean conditions are not to be considered to be a
steady background which is disturbed by occasional weather systems. Justify
this statement from a typical barograph trace.
8. Currently the northern hemisphere summer coincides with aphelion.
Speculate on how the seasons there would compare with the present when
aphelion coincides similarly in a time of maximum ellipticity of terrestrial
orbit round the sun.

LEVEL 2

9. Use the equation of state for dry air to calculate the air density at the summit of
Mt Everest on an occasion where the pressure and temperature there were 313
mbar and -38.5 oc respectively.
10. In a certain isothermal atmosphere in which the exponential scale height is 107
4 State and climate 7 km, find the height intervals between the following standard pressure levels
(levels for which radiosonde data is always extracted): 1000, 700, 500,
300 mbar.
11. Use data from Fig. 4.5 to estimate the mean temperatures of the 1000 to
500 mbar layer at latitudes 30 and 70° in winter, assuming an isothermal
atmosphere in each case. Use the expression for binary scale height.
12. Inspect Figs 4. 7 and 4.9 and estimate the time taken for air to travel (a) around
the Earth at constant latitude in the vicinity of the core of the subtropical jet
stream, and (b) from latitude 30° to latitude 10° in the winter trades.
13. Estimate the different travel times for aircraft shuttling with an air speed of
250 m s - 1 between airports A and B which are 1000 km apart, when there is a
jet stream flowing from A to Bat 75 m s - 1 at cruising altitude. (Ignore climbs
and descents.)
14. Express a typical polar front jet core speed as a fraction of the tangential speed
of a point fixed on latitude 50°. Find the centripetal acceleration of a fixed
point on the equator and express it as a fraction of g.

LEVEL 3

15. Find the density of moist air with specific humidity 15 g kg - 1, temperature
15 °C and pressure 1010 mbar, and compare it with the density of dry air at
the same temperature and total pressure. Find the temperature at which dry air
would have the same density as the moist air at the same total pressure. Note
that this last temperature is known as the virtual temperature of the moist air,
which is very useful in considering the buoyancy of cumulus etc.
16. Given that the atmosphere of Mars consists very largely of carbon dioxide, and
that the gravitational acceleration near the Martian surface is 3.8 m s- 2 , find
the decadal scale height of the Martian atmosphere and compare it with the
terrestrial value. (Martian atmospheric temperature - 220 K.)
17. Consider possible changes in low latitudes in Fig. 4.6 if deep convection there
were to be considerably reduced. And similarly consider how the appearance
of Fig. 4. 7 might differ in low latitudes if the Earth were not rotating. In fact
convection is likely to increase in some parts of low latitudes in the absence of
rotation. Where and why?
18. Assuming that the glaciated and unglaciated states of the Earth represent two
different equilibrium states between which the surface and atmospheric
system of the Earth vacillates in response to random or external factors, con-
sider how a small increase in the total ice-cap extent might lead to a further
increase and so to full glaciation. Given that low-latitude conditions alter
little, consider how the increased baroclinity implied by the onset of
glaciations might on the one hand tend to increase snow and ice cover, and on
the other tend to decrease it.
19. Show that the effect of ignoring the meridional variation of g is underestimate
the heights of horizontal surfaces at low latitude as compared with high by
about 50 min 10 000.

108
Atmospheric CHAPTER

thermodynamics 5

5 .1 Introduction

Heat enters and leaves the atmosphere in many different ways, and the air is usually
warmed or cooled as a result. For example, solar radiation warms the atmosphere,
and cooling is usually the net result of the absorption and emission of terrestrial
radiation (Chapter 8). In a different type of process, buoyant air rises and expands
against the falling pressure of ambient air, and as it expands it cools. Sinking air is
warmed by the same process in reverse. And in yet another type of process, large
amounts of heat are evolved or absorbed within air parcels as their contained water
substance changes state. For example, heat is evolved during cloud formation, the
latent heat of vaporization (latent since it was absorbed when the water was last
evaporated) being released and shared with the air between the swelling droplets.
In the present chapter we consider in some detail how the temperature changes
produced in these ways are related to the physical processes at work. In doing so we
make use of some of the techniques and results of the science of thermodynamics, a
subject which has grown in the last two centuries from a number of unrelated
speculations about the nature of temperature and heat to a powerful and
sophisticated part of physical science, indeed one of its cornerstones. During its
development, thermodynamics has evolved a subtle, precise and rigorously logical
approach which is fully revealed in texts such as [35] but which is not reproduced
here. In this book, certain basic thermodynamic concepts are applied to atmo-
spheric behaviour in just sufficient detail to clarify the physical processes at work,
and to justify some useful relationships.

5.2 The first law of thermodynamics

To describe the thermal behaviour of a typical parcel of air, we must be able to

allow for its response to the injection and removal of heat, as well as to its
expansion and compression. We need a relationship between exchanges of heat and 109
5 Atmospheric changes of temperature and volume, and this is provided by a form of the first law
thermodynamics of thermodynamics which is particularly suitable for describing the behaviour of
unit mass of an ideal gas:

dQ = c.dT+ pdVol (5.I)

where dQ is a very small heat input (from sunlight for example), dT is a very
small rise in temperature, and d Vol is a very small increase in the parcel volume
Vol (Fig. 5.I).
The first term on the right-hand side of eqn (5 .1) represents the amount of heat
p input which is used to increase the internal energy of the air parcel, i.e. the kinetic
energy of its air molecules' random motion, of which temperature is a measure.
This term is proportional to the temperature rise, and the constant of propor-
tionality c. is by definition the specific heat capacity of the air at constant volume,
dQ as can be seen in the special case where d Vol is zero. The value of c. for dry air
Fig. 5.1 Heat dO entering air throughout the turbosphere is close to 7I7 J K- 1 kg- 1, which means that this
trapped at constant pressure number of joules of heat input are required to warm I kg of air by I oc when the
p in a cylinder, warms the volume of the air is kept constant.
trapped air and pushes the
light, frictionless piston
The second term on the right-hand side of eqn (5.I) represents the amount of the
outward through a distance heat input which is used in doing work against the surrounding air as the parcel
dx, doing work p A dx expands. It is proportional to the expansion and to the internal pressure p of the
( =p dVol), where A is the parcel, which in terrestrial conditions is effectively equal to the surrounding
area of the piston face and pressure, for the reasons discussed in section 4.3.
dVol is the increase in the
volume of the trapped air.
The layout of eqn (5 .1) is a reminder that the heat input to the air parcel is shared
between the two processes in question: warming and expansion. The equality
expressed is an example of the principle of conservation of energy, which is
believed to hold under all circumstances. However, any particular example of
it, such as eqn (5.1), is valid only in so far as all sources, sinks and transforma-
tions of energy are accounted for. Fortunately the very simple scheme repre-
sented by eqn (5.1) accounts for several very important processes with useful
accuracy.
As in the treatment of the equation of state for air (section 4.I ), the parcel
volume Vol is inappropriate for dealing with an unconfined gas like air, and so it is
replaced by the reciprocal of the air density, to which it is numerically equal since
we are considering unit mass of air. When this is done, and the resulting density
variations dp are replaced using the equation of state again (Appendix 5.I), we find
1
dQ = Cp dT- -dp
p
(5.2)

Consideration of the special case where the little pressure change dp is zero shows
that CP is the specific heat capacity of the air at constant pressure. According to
Appendix 5.I, CP is larger than c. by the specific gas constant R,
cp =c.+ R
and it follows from the values of the terms on the right-hand side that the value of
CP for dry air in the turbosphere is about 1004 J K- 1 kg- 1: a value which can
usually be rounded to 1000. Strictly speaking, the different heat capacity of water
vapour makes the CP value for moist air differ slightly from the value for dry air,
but the difference is so small that for most purposes the dry value can be applied
throughout the turbosphere, regardless of humidity. As will appear in several of
the following sections, CP applies much more widely than might seem proper at
first glance, and is much more generally useful than Cv.
110
5.3 Isobaric heating and cooling 5.3 Isobaric heating and
cooling

Much heating and cooling of the atmosphere occurs while air pressure is steady or
nearly so. For example, the warming of the planetary boundary layer by day and its
cooling by night occur at pressures determined by the weight of the overlying
atmosphere, and this usually changes very little in percentage terms in the few
hours for which the warming or cooling persists. And because of the unconfined
nature of the atmosphere the heating or cooling does not in itself change the air
pressure in the way that it would for example if the air were trapped in a rigid box.
Of course heating may produce convection, with associated significant changes in
the pressures of air parcels as they rise or fall through the surrounding air, but even
so the individual pressure changes can be ignored if we choose a 'super parcel' large
enough to encompass the full depth of the convecting layer.
Equation (5.2) becomes suitable for treating isobaric processes when the term
involving dp is zero. The first term on the right-hand side (CP dT) then completely
determines the relationship between heat exchange and temperature change, and
can be used to estimate temperature changes from known gains or losses of heat, or
vice versa:
dQ = CPdT
This applies also to a finite slab of air embedded under the rest of the atmosphere,
even though there might seem to be a complication as the slab expands and uses
some of the energy input to raise its own centre of gravity. A full treatment shows
that the complication is illusory and that the simple relation still holds (appendix
13.2).
To consider the isobaric warming or cooling of a mass M of air, rather than unit
mass, we must use
(5.3)
Let us work through an example which shows the meteorological relevance of
this very simple relationship. On a sunny morning overland, it is often observed
Fig. 5.2 Morning warming
that the air temperature near the ground rises by a couple of degrees per hour for and deepening of a
several hours in response to solar heating. Observation in depth shows that the convecting layer over
warming layer is often about 300m deep (Fig. 5.2). What rate of heat input is extensive grassland in
required to account for this rate of warming of this amount of air? Australia. Some of the
temperature wriggles arise
Assuming the density of air to be 1.2 kg m -J and considering the warming of a
because measurements by
traversing aircraft cannot be
closely synchronized at
different heights (potential
temperatures should be
uniform along vertical
profiles in an ideal well-
1000
mixed, cloudless layer,
according to section 5.7). The
E situation is similar to that
shown in Fig. 5_9, except that
:;::: potential temperature is used
Ol 500
'iii in place of ordinary
~
temperature, and the
overlying stable layer is
maintained by anticyclonic
subsidence (section 12.1 ).
(After [36])

potential temperature (°Cl 111

5 Atmospheric column resting on one horizontal square metre of ground, the mass of a 300 m
thermodynami{:S column is 360 kg. Inserting this value for Min eqn (5.3), together with 2 oc for dT
and the established value for CP, we find that the required heat input per hour is 7.2
x 105 J m- 2 , which corresponds to a rate of input of200 W m- 2 • In fact the total
solar input is likely to be larger than this because some heat is used to evaporate dew
and soil moisture, and some heat is absorbed by the warming ground (section 9.5).
Values of this approximate magnitude are known to be realistic from direct
measurements by radiometer.
The heat in this example is entering the air largely by means of convection from
the sun-warmed ground surface: warm, buoyant parcels rise up, while cool ones
sink down, and the continual mixing distributes the net heat input throughout the
convecting layer. Such heat is called sensible heat because it can be sensed directly
by thermometers as the warm and cool air parcels pass by. However, in all but the
most arid lands, an even larger amount of latent heat is carried aloft by the same
convection, because air parcels rising from the moist surfaces contain more water
vapour than do their sinking neighbours. This latent heat flux has no warming
effect until the latent heat is released by condensation of the water vapour, almost
always in the course of cloud formation (dew represents a relatively very small sink
of atmospheric water vapour). The heat which then becomes sensible is the heat
which was absorbed from the surface and adjacent air when the water evaporated
in the first place. Since vapour, evaporation and condensation play very important
roles in the events discussed in this and succeeding chapters, we must first consider
carefully how vapour content is measured and quantified.

5.4 Measures of water vapour

Absolute vapour content is measured in terms of vapour density or pressure. Since

vapour behaves as an ideal gas in terrestrial conditions, we can use the equation of
state (eqn (4.2)) to relate the vapour pressure e to vapour density Pv and absolute
temperature T
e = PvRv T, (5.4)
where Rv is the specific gas constant for water vapour, which has value ~62 J K - 1
kg- 1 according to eqn (4.2) and following text. Note that e is the partial pressure of
the water vapour (appendix 4.1).
In many circumstances it is more useful to know the proportion of air which is
water vapour than it is to know the absolute vapour content, especially since the
very well-mixed state of the lower troposphere in particular makes proportional
measures much more uniform than absolute measures. We have already used one
proportional measure, the specific humidity q, which is the proportion by mass of
water vapour in the total moist mass (vapour plus dry air).

q = -mv = --"'---
mv (5.5)
m mv + md
An alternative and similar measure is the proportion of vapour to the mass of dry
air with which it is mixed - the humidity mixing ratio r.

(5.6)
112
 5.4 Measures of water
vapour

Fig. 5.3 Backlighting

highlights dew forming
thickly on a spider's web in
very slightly supersaturated
air. Because the
supersaturated zone is
confined to within - mm of
the slightly cool web there is
no fog in the background.

In fact the proportions of water vapour in air are so small, even in the warmest and
moistest parts of the troposphere, that there is no significant difference between q
and r values for any realistic air parcel. A typical value in the low troposphere is
10- 2 , quoted as 10 g kg- 1 to avoid exponents, and values in the high troposphere
are one or two orders of magnitude smaller. For simplicity we will always use q in
this book.
Since the moist air and water vapour occupy the same total volume (the volume
of the air parcel in question), q is also the ratio of water vapour density to moist air
density.
q = pjp (5.7)
Replacing each density using the appropriate equation of state we find
q = eRI(pR.)
where R is strictly speaking the specific gas constant for moist air, but can be
assumed to be the dry air value without serious error, and pis the total air pressure.
The ratio of the gas constants RIR. is simply the inverse ratio of their effective
molecular weights, which has the value 0.622, often represented by e. Hence
q = ee/p (5.8)
Vapour pressure is normally quoted in millibars, and can be used in eqn (5.8)
provided p is expressed similarly, which it usually is. The vapour pressure
corresponding to q = 10 g kg- 1 andp = 1000 mbar is 16 mbar, which is typical of
the low troposphere.
The total amount of vapour in a vertical column extending through the
atmosphere can be expressed as the depth of rainfall produced if it were all to be
rained out. This precipitable water content can be readily calculated from
radiosonde profiles of humidity and temperature against pressure (appendix 5.2).
It is found by laboratory study that the pressure and density of a saturated
vapour (a vapour in dynamic equilibrium with a plane surface of pure water - 113
5 Atmospheric section 6.2) depend only on the temperature of the vapour (Fig. 6.2). At any
thermodynamics
temperature T there is therefore a definite value of the corresponding saturated
vapour content, which means that it is possible to express the actual vapour content
in any situation as a fraction or percentage of the content needed to produce satura-
tion at that temperature. This is the relative humidity (RH) of the air, and it can be
expressed in terms of either vapour pressure or density,
RH = lOOe!e, = lOOp/ p, (5.9)

where the saturation values are denoted by subscripts. Values of RH range from
lOOOJo to about 10% in the troposphere, and are usually in excess of 500Jo near all
but the most arid land surfaces. In growing clouds they may exceed 100% (section
6.9), but the excess is so slight that saturation is the normal limit to vapour content,
and therefore 100% is the normal limit to RH.

5.5 Measurements of water vapour

Although direct measurement of relative humidity has long been normal in

radiosondes for reasons of instrumental simplicity, the most accurate measure-
ments of vapour content, and the most useful measures of vapour content for some
important meteorological purposes, are the wet-bulb and dew-point temperatures
(section 2.3).
In the case of a wet-bulb thermometer, water evaporates into the air from a
saturated wick enveloping the thermometer bulb (Fig. 5.4), and the excess of this
flux over the opposing flux, condensing onto the wick from the surrounding
vapour, cools the wick, bulb and surrounding air by net loss of latent heat. Some
movement of air over the wick is needed to ensure that the air in contact with it is
replaced quickly enough to inhibit the formation of a cocoon of saturated air which
would otherwise insulate the thermometer from effective contact with the ambient
air. However, the actual rates of net evaporation from the wick are so high in all
subsaturated atmospheres that the air in contact with the wick is fully chilled and
saturated before it is swept away by the air motion. (The efficiency of net
evaporation has another important consequence discussed in section 6.2.) The
temperature of the wet bulb is therefore the temperature at which the measured air
is saturated by evaporating water into it from the wet bulb. In practice the detailed
behaviour of the wet bulb is complex, and the reading is significantly affected by
heat conducted from the dry parts of the thermometer (which are warmer in all sub-
saturated conditions), by absorption of solar and terrestrial radiation, and by dif-
ferences between the rates of diffusion of water vapour and heat through air. For
this reason a simpler but less realistic process is envisaged to define a thermo-

Fig. 5.4 Evaporation from a

wet-bulb thermometer into
ambient air-flow.

114
dynamic wet-bulb temperature Tw, which is used as the ideal wet-bulb temperature 5.5 Measurements of water
in all meteorological discussion. vapour
The thermodynamic wet-bulb temperature of an air parcel is by definition the
lowest temperature to which the parcel can be cooled by evaporating water into it
isobarically and adiabatically (i.e. at constant pressure and in thermal isolation
from its surroundings). As water evaporates into the parcel, it is chilled and its
vapour content is increased, and the lowest temperature will be reached when the
air in the parcel reaches saturation by the combination of these two tendencies. If
the evaporation of a mass mv of water is needed to cool an air parcel of mass m from
its original temperature T to its wet-bulb temperature Tw, then the latent heat
required for the evaporation (Lmv) must equal the heat lost in cooling the air, since
there is no other source or sink of heat. (L is the specific latent heat of vaporization
of water - the quantity of heat needed to evaporate unit mass of water. In terre-
strial conditions the value of Lis always close to 2.5 MJ kg- 1, and its great size has
many important consequences for meteorological situations involving evaporation
or condensation.) By eqn (5.3) we therefore have
Lmv = mCP (T- Tw)
Now according to the process just described, the ratio mvlm is the difference
between the initial specific humidity q(T) and the final saturated specific humidity
at Tw, q,(Tw). (This ignores the change in total mass of moist air in comparison with
the change in the proportion of vapour, which is always justifiable in atmospheric
conditions.) Hence we find a version of the psychrometric equation:

q(T) = q,(Tw) -
cl (T- Tw) (5.10)

Note that the relatively small heat capacity of the vapour has been ignored
throughout. Replacing the q terms by eqn (5.7) or (5.8) allows either the vapour
density or the vapour pressure of the air to be calculated from the corresponding
values for saturation at the wet-bulb temperature, values which are tabulated, or
graphed as in Fig. 6.2.
Similar but slightly different relationships apply to actual wet-bulb thermo-
meters, but by a happy accident of opposing deviations the temperature registered
by a suitably aspirated wet-bulb thermometer (ventilation draught 4 m s - 1 or
more) approximates to the thermodynamic wet-bulb temperature so closely that
the error can be ignored. Wet-bulb thermometers relying on natural ventilation, in
a Stevenson screen for example, may differ more significantly from Tw, and require
a modified form of eqn (5.10) for accurate calculation of water content.
In the case of a dew-point meter, the air in contact with the cooling face of the
meter is cooled isobarically until it saturates and begins to deposit dew on the face.
The temperature of the face and the adjacent air when this happens is by definition
the dew-point temperature of the air. Since the proportion of vapour in the air is
unaltered by this process, at least until dew begins to form, and since the total air
pressure is fixed at its ambient value, eqn (5.8) shows that the vapour pressure of
the air before chilling (i.e. at the initial temperature T) is unaltered by the chilling
and is therefore the same as the saturation vapour pressure at the dew-point:
e(T)=e,(Td) (5.11)
Measurement of Td together with tables of saturation vapour pressure therefore
gives the vapour pressure of the ambient air, and hence all the other measures of
humidity through eqns (5. 7) to (5 .1 0). For example, the relative humidity RH is
given by
RH = 100e,(Td)le,(T) 115
5 Atmospheric When it is not zero, the dew-point depression T- Tct is significantly larger than
thermodynumics the wet-bulb depression T- Tw (roughly twice as large), because the saturated state
at Td is achieved purely by chilling, whereas saturation at Tw is reached by chilling
and net evaporation. Because of the complexity of the thermodynamics involved,
no simple relation exists between numerical values of T, Tw and Tct, and thermo-
dynamic tables or diagrams have to be used.
The multiplicity of measures and measurements of atmospheric humidity may
seem confusing, but in fact each is particularly useful for dealing with one or more
type of meteorological problem. The specific humidity has been used already, and
most of the others will be seen in action in this and the following chapter. Although
the dew-point meter is occasionally used for slow but accurate humidity measure-
ment, the dew point is used more as a measure of humidity than as a measurable
property- one which is very useful in discussions of cloud formation in chilling air
(Fig. 6.9), and for displaying humidity on thermodynamic diagrams (Fig. 6.6).

5.6 Adiabatic reference processes

It has already been mentioned (section 4.2) that in a number of important types of
tropospheric motion, air rises or falls so rapidly that the cooling caused by
expansion, or the warming caused by compression, considerably outweighs
temperature changes caused by other factors, such as absorption or emission of
radiation. In formal terms the air motion is nearly adiabatic, and it is very useful to
define strictly adiabatic reference processes with which to compare the inevitably
more complex reality of atmospheric behaviour.
The simpler of the two common reference processes is the dry adiabatic reference
process, in which there is neither condensation nor evaporation of water substance.
Strictly speaking it is relevant to the behaviour of all unsaturated air, in addition to
the unrealistically completely dry air implied by its title, but the term 'dry' has
become standard. The dry adiabatic reference process is defined by eqn (5.2) with
the left-hand side (dQ) set to zero, and it follows that the terms on the right-hand
side must balance exactly.
In many circumstances of course there is condensation or evaporation of water,
whose effects are to supply or remove heat to or from the air (almost always cloudy
air), and which corresponds to non-zero dQ in eqn (5.2). Since almost all tropo-
spheric cloud condenses or evaporates in air which is very close to saturation (held
there by the surprisingly complex mechanisms considered in the next chapter), it is
useful to define a saturated adiabatic reference process.
Detailed comparison with observation shows that the dry and saturated
reference processes do very well - surprisingly well in view of the deliberately
drastic simplifications assumed. In fact a full discussion of why they work so well is
beyond the scope of this book and involves matters which are the subject of current
research. Using models which work well for reasons which are not clear is of course
a symptom of imperfect understanding, but it is a very useful procedure even so,
and often tends to highlight particular matters needing further attention. We will
now look in some detail at properties of the adiabatic reference processes, knowing
that though they are over-simple they nevertheless provide useful keys for
116 understanding the real atmosphere.
5. 7 The dry adiabatic reference process and 5.7 The dry adiabatic
reference process
potential temperature

In this process no heat enters or leaves the air parcel, and none is evolved or
absorbed within it by condensation or evaporation of water. It follows that dQ is
zero in eqn (5.2) and that with rearrangement

(5.12)

Equation (5.12) is clearly consistent with common experience (with bicycle pumps
for example), that compression (positive dp) is associated with warming (positive
dT), and that decompression is associated with cooling. The relationship is made
more obvious when air density pis replaced using the equation of state (eqn (4.2)).
After rearrangement we find
dT R dp
-- (5 .13)
r cp P
where Tis the absolute temperature of the air parcel, and pis its total pressure.
Equation (5 .13) shows that small changes in temperature and pressure are in direct
proportion, and that the constant of proportionality is RICP when each is
expressed as a fraction or percentage of its absolute value. The value of R/CP for
dry air is always very close to 0.285 in the turbosphere, and it differs so little in
realistic humidities that the same value can be used for moist air in all but the most
accurate descriptions. It follows that for example a 30Jo drop in air pressure (about
30 mbar in the low troposphere) is associated with a fall of just under 1% in air
temperature (about 2.6 oc in 300 K).
When percentage changes inp and Tare no longer small, eqn (5.13) must be inte-
grated to relate initial and final value (with subscripts 1 and 2 respectively), with
useful accuracy. Standard integration then produces Poisson's equation,

T2 = ( P2 )RIC, (5.14)
T1 P1
which can be used to find any of the continuous sequence of intermediate tempera-
tures experienced by an air parcel undergoing compression or decompression. For
example, an air parcel in the deep, dry convecting layer over the Sahara desert in
the mid-afternoon may rise and expand and cool dry adiabatically from the
1000 mbar level (near sea level) to the 500 mbar level over 5 km higher. If such air
starts rising with a temperature of 305.2 K (32 oq at the lower level, eqn (5.14)
shows that it will reach the upper level with a temperature of 250.5 K (- 22.7 °C}.
The initial, final and all intermediate states are said to be on the same dry adiabat
- the continuous sequence of states, each defined by a pressure and temperature,
which is connected by this particular dry adiabatic process. Dry adiabats are
specified in thermodynamic tables such as [37], and are plotted on several types of
thermodynamic diagram used by meteorologists, one of the most common of
which (the tephigram) is used in Fig. 5.5. These diagrams are used to produce
graphical displays of the nearly vertical profiles of temperature, humidity and
pressure measured by radiosondes.
As outlined in the next section, the dry adiabats on a tephigram are straight lines
running diagonally from bottom right to top left. Any particular part of the tem-
perature/pressure profile from a radiosonde ascent which corresponds to a dry 117
5 Atmospheric adiabat will therefore be a straight line on a tephigram, lying parallel to the series of
thermodynamics standard dry adiabats printed on it. The Saharan example shows this clearly when
plotted on Fig. 5.5, as does a sounding more typical of vigorous convection in
middle latitudes, though the dry adiabatic layer is cooler and much shallower in the
latter.
It is useful to label each dry adiabat with a numerical value which is the same
everywhere along the line, and which is therefore conserved during that particular
dry adiabatic process. The convention is to use as a label the temperature at which
the dry adiabat crosses the 1000 mbar isobar. This temperature is known as the
potential temperature of the air at any point along that particular dry adiabat, and
it corresponds physically to the temperature which would be reached by an air
parcel if it were taken dry adiabatically from its initial temperature and pressure to
1000 mbar. It follows from this definition and eqn (5.14) that the potential
temperature() of air with absolute temperature Tand pressure p (expressed in milli-
bars) is given by

( -- 1000) RICp
O=T (5 .15)
p
Equation (5 .15) shows that potential temperature() is greater than actual tempera-
ture Twhen pressure is less than I 000 mbar, and vice versa, and is equal to T only
when pressure is exactly 1000 mbar. It follows that potential temperatures are
greater than actual temperatures throughout all but the very lowest parts of the
atmosphere, and will be greater even there if the surface pressure is below
1000 mbar, as in a vigorous depression, for example. In the Saharan case in
Fig. 5.5, where the potential temperature is 32 oc everywhere along the dry
adiabat, the actual temperature at 500 mbar is -22.7 °C. If the dry adiabat in this
case extended down to a rather high surface pressure of 1020 mbar (likely only in
those parts which are not far above sea level and in the absence of the heat low
which is so common in the Sahara in summer- see section 11. 7), then the actual
temperature at the surface would be 33.7 °C.

5. 8 Thermodynamic diagrams

It is very useful to be able to portray thermodynamic processes, such as the

meteorological reference processes, on a diagram which will graphically
summarize matters otherwise requiring equations or abstract text. A point on such
a diagram must correspond to a particular thermodynamic state of an air parcel,
i.e. to a definite temperature and pressure, so that a line will represent the series of
states corresponding to any particular process being portrayed. It does not follow,
however, that we must use temperature and pressure as axes. Indeed there is a long
tradition that the most useful axes are those which ensure that equal areas any-
where on such a diagram should represent equal energies- energy gained or lost in
going round the cycle of processes represented by the boundary of the area. There
are many different diagrams in use which fulfil this criterion, but a detailed dis-
cussion of their construction and relative merits involves more than the basic
thermodynamics covered in this book. It is sufficient to say that as well as
118 observing the thermodynamic niceties, any particular diagram must provide a clear
 Wind ( 0 kt)
5.8 Thermodynamic
diagrams

500
Fig. 5.5 Vertical soundings
of pressure, temperature and
humidity plotted on a
tephigram (sections 5.7 and
600
5.8). with temperatures and
dew points represented by
continuous and dashed lines
700 respectively. The main
sounding is typical of many
situations in middle latitudes,
while the layer of much
800
warmer air (further to the
right at the same pressure)
represents the hypothetical
900
Saharan example quoted in
section 5.7. (After Metform
1000 2810, Meteorological Office,
1050 Bracknell, Berks. Crown
0 copyright)

display of the processes at work, and if possible distinguish clearly between ones
which are felt to be significantly different.
The tephigram is the most widely used thermodynamic diagram in meteorology.
It was introduced by the English meteorologist Sir Napier Shaw, who was one of a
number of very distinguished meteorologists from a variety of cultures who helped
establish meteorology as a quantitative science at the beginning of the present
century. The axes are absolute temperature T(usually labelled in degrees Celsius),
and the logarithm of the potential temperature 0 labelled likewise. In a certain
highly idealized type of thermodynamic process (a reversible one), the potential
temperature axis can be regarded as an entropy axis: hence the name T¢ (for
entropy) gram. It can be shown that these axes meet the energy criterion cited
above, and this together with the many features of this versatile diagram are dis-
cussed in more advanced meteorology texts [8]. The basic construction of the
tephigram is probably best appreciated by making a simple outline of one, as
described in appendix 5.3. It then emerges that in the limited ranges of T and 0
needed to represent conditions in and just above the troposphere, the basic layout ll9
5 Atmospheric
thermodynamics

Fig. 5.6 Outline tephigrar.~

with axes and isopleths, os
discussed in section 5.8.
(After Metform 2810. Crown
Temperature (°C)
copyright)

is as sketched in Fig. 5.6. With the temperature axis horizontal and the log(} axis
vertical we find the following.
1. By definition, isotherms are vertical straight lines, and isotherms of potential
temperature (dry adiabats) are horizontal straight lines.
2. Isobars are slightly curved lines nearly bisecting the right angle between the
axes, and are spaced logarithmically, so that a line perpendicular to them is
nearly linear in height (because height and log pressure are nearly linear -
section 4.4). Highest pressures are to the bottom right.
3. lsopleths of saturation specific humidity are nearly straight lines making fairly
small angles with the isotherms.
4. The spacing between dry adiabats decreases slowly as(} increases.
5. Saturated adiabats (section 5.10) make angles of about 45° to the dry adiabats
. in the high pressures and temperatures typical of the low troposphere, but
become more and more nearly asymptotic to them as upper tropospheric
conditions are approached.
The diagram is usually displayed as in Fig. 5.5, which is Fig. 5.6 rotated
clockwise by about 45°. The isobars now lie nearly horizontally across the
diagram, and the undrawn vertical is regarded as a height axis, which can be
labelled in actual heights for any particular sounding after calculation of thick-
nesses. Isotherms now run diagonally upward to the right, with highest tempera-
tures to the bottom right, and dry adiabats run diagonally upward to the left, with
highest values to the top right. Saturated adiabats rise nearly vertically in the low
troposphere, and turn anticlockwise to join the dry adiabats in the high
troposphere.
120 To illustrate the usefulness of a tephigram, consider the middle latitude
sounding plotted in Fig. 5.5. The dry adiabatic layer in the low troposphere shows 5.9 The dry adiabatic lapse
clearly as a straight line parallel to the dry adiabats. Above this there is a substantial rate
region in which the profile crosses the dry adiabats toward higher potential
temperatures, sometimes running nearly parallel to saturated adiabats, and some-
times running across these toward even higher potential temperatures. At the
tropopause the profile turns sharply toward the top right, portraying the nearly
isothermal profile so typical of the low stratosphere. The sharp graphical
distinction between these different segments of the sounding is a very useful feature
of the tephigram, since each corresponds to a quite different regime of meteo-
rological behaviour, as will be made clear in subsequent sections. The dashed line
on Fig. 5.5 portrays the profile of dew point which is used in conjunction with the
temperature profile to describe the humidity structure of the middle latitude
sounding, as outlined in section 6.5.
Other diagrams used to display the thermodynamic state of the upper air are the
skew T log p and Stuve diagrams [38]. The former looks very like a tephigram,
except that the isobars are straight and the dry adiabats are slightly curved. The
Stuve diagram has a long history in meteorology, but is not so convenient in use
and does not satisfy the energy: area equivalence.

5. 9 The dry adiabatic lapse rate

So far we have considered the dry adiabatic variation of temperature with pressure
(eqns (5.12)-(5.14)). In many circumstances, however, pressure variation arises
almost entirely because of height variation, so that we can make use of the nearly
perfectly hydrostatic relation between pressure and height (section 4.4), to find the
equivalent relation between temperature and height for the dry adiabatic process.
By combining eqns (4.6) and (5.12), we find that when an air parcel of density p
moves dry adiabatically and vertically through ambient air which is in hydrostatic
equilibrium and has local air density p', small changes in parcel temperature dT
and height dz are related by
dT = _ _L !!..._
(5.16)
dz cp P
Note that as usual we assume negligible pressure difference between the air parcel
and its surroundings. The minus sign corresponds to the fact that temperature falls
(lapses) as height increases.
The appearance of both parcel and ambient air densities in eqn (5.16) threatens
to complicate matters. However, the equality of parcel and ambient pressures,
together with the behaviour of air as an ideal gas, means that the density ratio p' I p
is equal to the inverse ratio of the absolute air temperature TIT'. In all but the most
vigorous types of convection, buoyant air parcels are observed to be only very
slightly warmer than their surroundings - about 1 K being a typical upper limit.
And even in the most vigorous types of convection (those powerful cumulonimbus
associated with severe weather such as torrential precipitation, gale force winds
and intense electrical activity), temperature excesses in the rising air seldom exceed
5 K. It follows that the density ratio in eqn (5.16) is almost always very close to
unity, the deviation usually being less than 0.50Jo. To the extent that the ratio is
unity, eqn (5.16) simplifies to 121
5 Atmospheric dT (5.17)
thermodynamics
dz cp
The term on the right-hand side of the equation is effectively constant throughout
the lower atmosphere since g variations are - 0.50/o, and CP is essentially uniform
thanks to very efficient mixing. When values for g and CP are inserted, the
magnitude of the vertical temperature gradient is almost exactly 0.0098 oc m -I, or
9.8 oc km - 1 • This is known as the dry adiabatic lapse rate, and is usually denoted
by the symbol r ct· Equation (5 .17) now becomes simply
dT _ -r (5.18)
dz - ct

According to eqn (5 .18), if an air parcel were to rise dry adiabatically through the
full depth of the troposphere, it would cool by about 100° in the shallow tropo-
sphere (about 10 km deep) of high latitudes, and by about 150° in the deep tropical
troposphere. Such very large temperature lapses are never observed because
saturation always intervenes, introducing significant warming which we must now
consider.

5.10 The saturated adiabatic reference process

Air in most parts of the low troposphere is so moist that it becomes saturated after
rising quite modest distances - often less than 1 km above the local surface
(Fig. 1.4). If the air continues rising, the processes of cloud formation and con-
tinuing development outlined in the next chapter hold the air very close to satura-
tion and evolve large quantities of latent heat as the excess vapour is condensed. In
such conditions the most useful reference process is the saturated adiabatic one, in
which we assume that the chilling moist air between the cloud droplets is held
precisely at saturation by continual condensation of excess vapour, and that the
heat evolved is confined to the air parcel and warms it. The saturated adiabatic
process is reversible if we assume that, in descending air, evaporation from
available condensed water exactly maintains saturation as the air is compressed
and warmed. The condensate is assumed to be water rather than ice, even at
temperatures far below the freezing point, and this is reasonably realistic at
temperatures down to about -25 oc because of the widespread observed tendency
for cloud droplets to remain supercooled down to such temperatures (section 6.11).
At lower temperatures a more realistic reference process would allow for the
liberation of latent heat from the freezing of cloud droplets, but the rather variable
efficiency of observed freezing would make the added complication a doubtful
asset.
Though the ascending branch of the saturated adiabatic process is more
obviously relevant to actual atmospheric behaviour than the descending one,
something close to saturated adiabatic descent occurs in cumulonimbus down-
draughts, where air often descends in the presence of evaporating precipitation.
Somewhat surprisingly, the descending branch also relates to the behaviour of wet-
bulb thermometers at fixed altitudes, as outlined in the discussion of Normand's
theorem later in this section.
122 The evolution of latent heat in rising saturated air and its absorption in sinking
saturated air reduces the temperature lapse rate below the dry adiabatic value, since 5.10 The saturated
in each case the heat exchange partly offsets the temperature change expected adiabatic reference process
according to the dry adiabatic process. To investigate the saturated adiabatic
process quantitatively we must include all three of the terms in eqn (5.2).
Suppose that a saturated parcel rises a small distance. It cools slightly by
adiabatic expansion and would become supersaturated, because of the sharp
decrease of saturation vapour density with decreasing temperature (Fig. 6.2), if the
excess vapour were not almost immediately condensed in the form of cloud. Since
the excess does condense, and there is no other vapour source or sink, the vapour
content of the air between the growing droplets must fall to balance the rise of
droplet mass, and this fall must be just what is needed to maintain saturation at the
slightly reduced temperature. If the small change in saturation specific humidity is
dq,, then the amount of latent heat evolved in unit mass of moist air is - L dq,,
where L is the specific latent heat of vaporization. The minus sign ensures that a
decrease in vapour content (negative dq,) corresponds to a liberation of latent heat.
The first law of thermodynamics (eqn (5.2)) can now be used to relate the liberation
of latent heat to the changes of temperature and pressure:
dp
CPdT= -Ldq, + P (5.19)

The opposing signs on the right-hand side confirm the expectation that condensa-
tion (or evaporation) offsets the temperature changes induced by adiabatic
pressure change.
An important property of the saturated adiabatic process follows from the slope
of the graph in Fig. 6.2 relating saturation vapour density (and therefore satura-
tion specific humidity) with temperature. Since the slope increases with increasing
temperature, it follows that emissions or absorptions oflatent heat are much larger
at high temperatures than they are at low, where there is very little vapour left to
condense, even at saturation. The contrast between the saturated and dry adiabatic
processes is therefore much more marked at high temperatures than it is at low. In
fact when the associated saturated adiabatic lapse rates (i.e. values of- dT/dz) are
calculated, values in the warm, low troposphere may be as low as 5 oc km _,,which
is only about half of the dry adiabatic value, whereas values in the cold, high tropo-
sphere are very close to the dry adiabatic value.
The saturated adiabatic process is most directly compared with the dry adiabatic
one when it is expressed in terms of changes in potential temperature 8, which by
definition are zero in a dry adiabatic process. By rearranging the combination of
eqns (5.2) and (5.15) as shown in appendix 5.4, we find

d{) = LO
---dq (5.20)
cp T '
This shows that the effect of condensation (negative dq,) is to increase potential
temperature, whereas evaporation reduces it.
Beginning with any particular state of saturated air (i.e. any particular point on
Fig. 5.6 or 5.7) eqn (5.20) could be used step by infinitesimal step to trace out a
saturated adiabat. At high temperatures, for the reason mentioned, the saturated
adiabats cross the dry adiabats at relatively large angles, whereas at lower and
lower temperatures the angles reduce more and more slowly toward zero. In fact
saturated adiabats are characteristically curved in conditions corresponding to the
low troposphere, but are nearly straight and parallel to dry adiabats in the high
troposphere (Fig. 5.7). An associated effect is that adjacent saturated adiabats
diverge as they pass from the low to the high troposphere (say 800-300 mbar), with
those starting from warmer conditions in the low troposphere cooling less than 123
5 Atmospheric
thermodynamics

Fig. 5.7 Tephigram depicting

wet-bulb potential
temperature and equivalent
potential temperature. Inset:
Enlargement of the base of
the main diagram to depict
lifting condensation level and 1000
Normand's theorem, using ~~~~--~~L-~---L~~~~--~-L~~---L~_j 1 050
part of Metform 2810. (Crown
copyright)

those starting in cooler conditions, because of the greater vapour content to be con-
densed. The saturated adiabatic process is therefore much less simple than the dry
adiabatic, in the sense that there is no single saturated adiabatic lapse rate, even at a
particular altitude (i.e. pressure).
Actual soundings through the troposphere often reveal profiles of temperature
(or potential temperature) which are similar in outline to saturated adiabats.
Above the ubiquitous dry adiabatic layer (ignoring nocturnal or seasonal stable
layers near the surface) the potential temperature rises with altitude, usually quite
sharply at first, but then more and more slowly as the cold upper troposphere is
reached (Fig. 5. 7a and b), in qualitative agreement with the shapes of saturated
adiabats. However, on close inspection it is usually apparent that the agreement is
imperfect: actual profiles may deviate considerably from saturated adiabats either
over a substantial depth (Fig. 12.3) or over a limited depth in an angular profile
(Fig. 4.1). In many cases such deviations can be associated with details of the
structure or behaviour of weather systems, but this requires specialized discussion
which is best deferred until later (Chapters 11 and 12).
Both the dry and the saturated adiabatic reference processes become completely
irrelevant above the tropopause, where a nearly isothermal layer is often observed
extending many kilometres into the low stratosphere, and the corresponding
potential temperature profile shows a sharp and continuing rise with increasing
height (Fig. 5.5). These profiles point to the virtual absence of vertical or nearly
vertical motion in the low stratosphere, which of course is consistent with the
124 absence of cloud there.
 Finally, consider a point of principle which greatly simplifies the saturated adia- 5.11 Equivalent and wet-
batic model without significantly reducing its utility. In the strictly adiabatic model bulb potential temperatures
we must assume that all liquid water is retained in the air parcel after condensation.
The cloud burden in any strictly adiabatically moving parcel therefore varies
depending on the parcel's previous history, and the resulting slight variations in the
total heat capacity of the parcel and contents ensures that there is an infinite
number of slightly different saturated adiabats passing through any point on
Fig. 5.6. To remove this serious complication it is assumed that cloud water is
eliminated from rising parcels as soon as it appears, and that water is made
available in descending parcels just in time to maintain saturation by immediate
evaporation. This simplified process is called the saturated pseudo-adiabatic
process, but the differences between it and its strictly adiabatic counterpart are
usually unobservably small for realistic concentrations of cloud water and water
vapour. Because of its simplicity, the pseudo-adiabatic process is used in the
preparation of thermodynamic tables and diagrams. We will use the pseudo-
adiabatic process henceforth without further reference to its 'pseudo' nature. To
the same degree of approximation we will continue using the specific humidity q to
specify vapour content, rather than the humidity-mixing ratio (see section 5.4),
even though the latter is apparently more directly relevant to the behaviour of an
air parcel having a fixed mass of dry air and a variable mass of vapour.

5.11 Equivalent and wet-bulb potential

temperatures

It is useful to have a single temperature with which to label each saturated adiabat,
just as each dry adiabat can be labelled by its potential temperature. Figure 5. 7
shows two ways in which this can be done.
At high altitudes (i.e. low pressures), temperatures and vapour contents are so
low, even at saturation, that any particular saturated adiabat is asymptotic to a
particular dry adiabat, whose potential temperature is known as the equivalent
potential temperature (}e of the saturated adiabat. In physical terms (}e can be
considered to be the temperature which would be reached by a parcel after
saturated adiabatic decompression to nearly zero pressure and dry adiabatic
recompression to I 000 mbar.
An alternative procedure is to label each saturated adiabat by the actual
temperature at which it intersects the 1000 mbar isobar. Let us call this tempera-
ture (}wand examine the relationship between the two labels for saturated adiabats,
(}e and (}w·
The equivalent potential temperature (}e is larger than (}w (Fig. 5.7) because the
condensation of vapour releases latent heat within the air parcel as it rises in a
saturated adiabatic process from the 1000 mbar level to the great heights at which
the vapour is virtually all condensed (the 200 mbar level, which is about 12 km
above sea level, serves in most cases). The size of this temperature excess can be
found very accurately from thermodynamic tables or diagrams, but it can also be
estimated quite usefully from an integrated and simplified form of eqn (5 .20)
(appendix 5.4):
(}e - (}w = 2.5q, (5.21) 125
5 Atmospheric Here q. is the specific humidity of the saturated air at the 1000 mbar level expressed
thermodynamics in grams of water vapour per kilogram of moist air, as it always is in practice. A
typical value for q. in temperate latitudes is 10 g kg- 1, and the corresponding value
foro.- ()w is 25 oc according to eqn (5.21). This agrees to within a fraction of a
degree with the value found more rigorously from thermodynamic tables and
diagrams, and agrees with the value given by Fig. 5. 7 to well within the accuracy of
reading of even the large-format original used for plotting soundings. Note that the
combination of the variation of saturation vapour content with temperature
(Fig. 6.2) and eqn (5.21) is consistent with the curvature and divergence of
saturated adiabats discussed in section 5.I 0, both of these being a maximum in the
warm, low troposphere of low latitudes.
The physical meaning of ()w is obvious in the case of saturated air: it is the
temperature reached by saturated adiabatic descent to I 000 mbar. However, there
is a surprisingly direct relevance to unsaturated conditions which is best appre-
ciated by referring to the inset part of Fig. 5. 7. Suppose that an unsaturated air
parcel with temperature T has wet-bulb temperature Tw and pressure p. The wet-
bulb temperature Tw is less than T by an amount which is a measure of the sub-
saturation of the air, and is plotted on the same isobar as Tin Fig. 5.7 because it is
reached by isobaric cooling by evaporation (section 5.5). According to a thermo-
dynamic theorem named after the meteorologist Normand, the dry adiabat
through T and the saturated adiabat through Tw intersect at the lifting condensa-
tion level of the initially unsaturated parcel (LCL on Fig. 5. 7), which is the level at
which the air would become saturated if lifted dry adiabatically from its original
state. Since the theorem applies equally well at all levels (i.e. pressures) below LCL
it follows that the saturated adiabat traces out the actual wet-bulb temperature of
the adiabatically rising or falling parcel of unsaturated air. If it is traced down to
the 1000 mbar level, the temperature reached is known as the wet-bulb potential
temperature of the air parcel, which is the wet-bulb temperature which the air
would have if it were taken adiabatically to the 1000 mbar level. (Normand's
theorem is completed in section 6.5.)
Notice that ()w is conserved in any adiabatic process, whether dry or saturated,
and whether isobaric or not. If the process is dry adiabatic, then Normand's
theorem applies as already presented. If the process is saturated adiabatic, then ()w
is conserved by definition. If the process involves adiabatic evaporation of water at
constant pressure, then it is all or part of the wet-bulb thermometer process
outlined in section 5.5, which by definition does not alter the wet-bulb temperature
of the air, and therefore does not alter its wet-bulb potential temperature. Since
any actual adiabatic process can be constructed out of a sequence of tiny adiabatic
changes - dry, saturated, isobaric or not - the conservation of ()w in a parcel
undergoing any adiabatic process has been established. All this makes the wet-bulb
potential temperature the most widely conservative thermodynamic marker used in
meteorology. To illustrate this, consider two adiabatic processes as they might
realistically operate in the vicinity of a cumulonimbus.
Suppose that the cloud whose base is at the lifting condensation level LCL in
Fig. 5. 7 begins to produce thick, fine rain, consisting of a dense population of
small droplets. These have a large total droplet surface area from which there is
copious evaporation into the slightly unsaturated air below cloud base. Each
droplet chills itself and its surroundings like a wet bulb, so that the net effect is to
cool the whole rain-filled parcel to the wet-bulb temperature of the originally
unsaturated air at the level in question. The air may then tend to sink with the
falling rain, desaturating by descent and resaturating by evaporation so that the air
parcel traces out the saturated adiabat extending downward from cloud base. If
126 this continues to the 1000 mbar level, the saturated rain-filled air reaches there with
a temperature T equal to Ow. This sort of process accounts for the downdraughts of 5.12 Convective stability
cool, moist air observed near the surface just before the arrival of a shower.
Evaporation may not keep pace with desaturation by downdraught compression,
particularly if the rain drops are large, therefore presenting a much smaller total
area of evaporating surface, and in that case the downdraught arrives with the
same wet-bulb temperature as before but a higher dry-bulb temperature T.
If, instead, air were to sink completely dry adiabatically below cloud base LCL,
as happens in the unsaturated subsidence between adjacent cumulus or cumulo-
nimbus, then it would warm so that its temperature T traced out the dry adiabat
through LCL. By the corollary of Normand's theorem, the wet-bulb temperature
of the sinking, desaturating air traces out the saturated adiabat through LCL. If
such descent should continue to the 1000 mbar surface, the air arrives there with
temperature equal to the potential temperature of the subsiding parcel, and the
much lower wet-bulb temperature equal to Ow. The wet-bulb potential temperature
is again conserved. Note that the value of Ow in the air entering the undraught is
generally a little higher than the value in the downdraught, the difference
indicating the degree of instability which is driving the cumulonimbus convection.
The equivalent potential temperature o. is obviously conserved in any process
conserving the wet-bulb potential temperature, since it is simply an alternative
labelling system for all saturated adiabatic processes. Though clearly relevant to
the life cycle of air ascending in very deep convection, with or without subsequent
dry adiabatic descent, o. lacks the direct identification with conditions near the
surface (at the 1000 mbar level, strictly) which is the special virtue of Ow.

5.12 Convective stability

In previous sections we have considered the sequences of states experienced by

individual air parcels undergoing dry or saturated vertical motion. Before begin-
ning to assess the realism of this behaviour, it is important to realize that
radiosondes, or the satellite-borne radiometers which are becoming viable alter-
natives, measure something quite different. They provide snapshots of the vertical
structure of the atmosphere - profiles of temperature, humidity, pressure and
wind observed at instants of time. It is true that the time taken by a radiosonde to
rise to bursting level is about one hour, and that the atmosphere can change some-
what in that time interval, but a great deal of observation shows that such changes
are rarely significant, at least in respect of synoptic scale structure. On such scales
the radiosonde snapshots are therefore only slightly blurred.
The distinction between the parcel viewpoint of adiabatic theory and the snap-
shot or environmental viewpoint of observation is very basic, and might be
expected to make theory and observation too different for useful comparison. But
although the viewpoints are quite different, it should be clear from observations
cited in previous sections that there is often close agreement between parcel theory
and environmental observation. For example the dry adiabatic temperature lapse
rate (eqn (5.17)) is derived from parcel theory but is observed from an environ-
mental viewpoint throughout extensive layers of the low troposphere. Assuming
that the dry adiabatic parcel theory is as relevant as has been claimed, there must be
additional factors making its predictions immediately observable. To uncover
these we begin by looking more closely at the connection between an individual
parcel and its immediate environment. 127
5 Atmospheric Suppose that an air parcel is embedded in air with a certain environmental
thermodynamics vertical temperature gradient aTI az. The partial differential aI az represents rate of
change with height alone, ignoring possible variations with horizontal position or
time; it corresponds therefore to the vertical gradient (of temperature in this case,
but the symbolism is quite general) in an instantaneous picture of the environment
of the air parcel. Now suppose that this air parcel moves up or down through its
environment, undergoing a parcel process which is dry adiabatic. The parcel's
temperature sequence traces out a portion of a temperature profile with vertical
gradient given by eqn (5 .18):
dT _ -r
dz - d

where rd is the dry adiabatic lapse rate, value 9.8 oc km- 1• The difference in
symbolism between aT!az and dTidz corresponds to the difference between the
parcel and environmental viewpoints.
The buoyant behaviour of the parcel depends critically on the value of aT/az. If
the environmental lapse rate- arlaz is superadiabatic (meaning greater than the
dry adiabatic valuer d), i.e.
ar ~
- < - rd or equivalently <0 (5.22)
az az
we have the state of affairs depicted in Fig. 5.8(a). Upward movement of the air
parcel through its environment in the course of convection or turbulence then leads
to the parcel becoming warmer and therefore less dense than its surroundings. The
resulting buoyancy tends to drive the parcel even further from its original position,
generating even more buoyancy, and so on. However, if the parcel moves down-
wards from its original position, it becomes cooler and therefore more dense than
its environment, and the resulting negative buoyancy tends to drive it even further
down. (Buoyancy is discussed further in section 7 .12.)
It seems that the air parcel can continue in its original position only if it is
completely undisturbed. It is therefore in an unstable state in just the same way as is

Fig. 5.8 (a, b) Parcel and (a) Unsaturated unstable (b) Unsaturated stable
environmental temperature
profiles representing

.:::)~::m•"'
adiabatic convective stability 'Parcel
ond instability in an .<: ' .<:
~nment "'
.!:!> ·a;
unsaturated atmospheric
layer; (c, d) the same stability "'
J: J:

' ' 'I'd

regimes for saturated
adiabatic convection in a
'-rd
saturated layer.
Temperature Temperature

'\
(c) Saturated unstable (d) Saturated stable

' 'Parcel
' ':,~•o•m•"'
Parcel' .

.<: Environment
' \
.<:
"' "'
"'
J: \
\
"'
J:

\r.s 'r.s

128 Temperature Temperature

a ballbearing perched on the top of a smooth dome, and just as the ballbearing 5.12 Convective stability
inevitably rolls down off the dome in response to tiny movements in the dome or
surrounding air, so the air parcel in Fig. 5.8(a) rises or falls in response to the
inevitable stirrings of the surrounding air. In fact every other parcel making up this
superadiabatic temperature profile is unstable in just the same way, with the result
that the whole layer is said to be convectively unstable. The term statically
unstable is often used instead, and distinguishes this instability from other
types arising dynamically, such as the baroclinic instability which underlies
many large-scale weather systems, and the shearing instability discussed in
section 10.8.
Now consider the same parcel (obeying a dry adiabatic parcel process) embedded
in a layer with a subadiabatic lapse rate. We have

aT
-az> -r d or equivalently (5.23)

Figure 5.8(b) and the same type of analysis as before show that this layer of air is
convectively stable. For example, when the individual air parcels are displaced
upwards through their environment they are driven back to their original positions
by negative buoyancy, and when they are displaced downwards they are driven
back up by positive buoyancy. The situation is now analogous to a ball-bearing in
the hollow of an upturned bowl, which will return to its initial position in the lowest
point of the hollow after any displacement.
We can show in just the same way that a layer of air with a precisely dry adiabatic
environmental lapse rate, i.e.

or equivalently (5.24)

is in a state of neutral convective stability, meaning that any constituent parcel

tends to remain in its new position after it has been displaced, rather than move
further from or back to its original position, like a ball bearing on a flat horizontal
plate.
So far we have considered only the dry adiabatic process. Exactly the same kind
of analysis can be used to find the criteria for convective stability when the
constituent air parcels are following a saturated adiabatic process. Assuming that
there is always enough water to keep descending air saturated we can proceed as
before. Figure 5.8(c) looks much the same as Fig. 5.8(a) except that the straight dry
adiabats are replaced by the slightly curved saturated adiabats, whose lapse rates r,
are always less than the dry adiabatic value, and are quite significantly so in the low
troposphere.
The criteria for convective instability, neutral stability and stability are
assembled in Table 5.1 for both the dry and saturated adiabatic parcel processes. In
each case neutral convective stability occurs when the environmental temperature
lapse rate exactly equals the lapse rate defined by the appropriate adiabatic parcel
process, with instability or stability resulting when the environmental lapse rate is
larger or smaller respectively. Because there are two quite different parcel
processes possible (the dry and saturated ones), there is a range of environmental
lapse rates which represent convective stability or instability depending respec-
tively on whether the constituent air parcels are following dry or saturated
adiabatic processes, which means effectively whether the layer is cloud-free or
cloud-filled. Such a layer of air is said to be conditionally unstable, the condition
being whether or not the layer is saturated. Conditional instability can play an 129
5 Atmospheric Table 5.1 Criteria for convective stability
thermodynamics
Environmental temperature gradient Adiabatic parcel process
in T in 0 unsaturated saturated

ao
-<0
az u u
ar ~=0
az az N u

ar conditional
-rd < Tz < -r. 0< :: < (~!). s instability
u
ar -r. ao do
az a;-= (dz ). s N
ar ao do
- > -r.
az az- >(dz ). s s
Key U unstable
N neutrally stable
S stable

especially important role in the middle and low troposphere, since the difference
between rd and r, may be very considerable there (section 11.6). A typical sequence
is that conditional instability is realized and released in convection when a layer
with a supersaturated-adiabatic lapse rate is raised to saturation by slow ascent in a
large-scale weather system.
Equivalent diagrams, very like Fig. 5.8, can be drawn on tephigrams, etc., and
analysed similarly. For example, the shallow segment at the botton of the Long
Kesh sounding in Fig. 5.5 is strongly superadiabatic and therefore convectively
unstable (such instability being typical of shallow layers immediately overlying
warmed land surfaces- sections 5.13 and 9.9). The next layer, extending nearly to
the 800 mbar level, is slightly but definitely subadiabatic and therefore stable if
unsaturated (as suggested by the dew point depressions- but see below). The rest
of the troposphere has a temperature profile close to the saturated adiabatic, which
suggests neutral stability for saturated ascent and descent if the air is saturated. The
dew-point profile shows considerable subsaturation, but this may be spurious,
since radiosonde humidity sensors are potentially quite inaccurate and especially
prone to under-reading [39]. The low stratosphere, above the tropopause at
320 mbar in Fig. 5.5, typically contains a deep inversion which is extremely stable
for both saturated and unsaturated convection.

5.13 The maintenance of near-neutral stability

The concept of convective stability associates environmental temperature profiles

with thermodynamic processes believed to be at work in the populations of
constituent air parcels. But we have yet to trace in practical detail the way in which
any particular temperature profile is brought about or maintained. Done
130 thoroughly, this is an elaborate task involving many aspects of atmospheric
 5.13 The maintenance of
near-neutral stability

Fig. 5.9 A convectively

stable layer formed by
cooling of the surface, with a
temperature inversion
I throughout layer /. Numbered

l dashed lines represent

successive stages in the
Temperature evolution of the temperature
profile during subsequent
solar heating.
behaviour, some of which are still quite poorly understood, but it is worth trying in
outline for the dry adiabatic case, which is important in itself and suggests the kind
of mechanisms at work in less simple cases.
A nearly dry adiabatic lapse rate is often observed extending through most of the
sub-cloud layer in the low troposphere. However, it is not observed very near a land
surface when nocturnal cooling is in progress or has only just ended. The situation
then is typically as depicted in Fig. 5.9, which represents conditions close to a land
surface toward the end of a cloudless night. The ground and the air close to it have
cooled during the night by net emission of infra-red (terrestrial) radiation,
producing thereby a convectively stable layer which may extend a hundred or more
metres above the surface. The cooling may often be strong enough to produce a
temperature inversion in some part of the stable layer, as shown in Fig. 5.9. Now
let us consider how such a strongly stable layer is converted in a few hours into a
nearly neutrally stable layer which is then maintained throughout the day until the
next night's cooling begins to reintroduce stability near the surface.
When the sun comes up, the surface and immediately overlying air are warmed
by absorption of solar radiation, the warming being quite rapid initially because of
the small heat capacity of the initially very shallow layers of ground and air
affected (sections 5.3 and 9.4). In the air close to the warming surface, the environ-
mentallapse rate becomes positive (assuming it was negative initially, i.e. that there
was a temperature inversion) and approaches the dry adiabatic value. As soon as
this is exceeded the shallow layer begins to convect, since it is slightly super-
adiabatic and therefore convectively unstable. The convection has two effects
which are worth considering separately in some detail. First, it maintains the lapse
rate in the convecting layer at the dry adiabatic value, or more strictly very slightly
in excess of it. Second, it very efficiently pumps sensible heat (and water vapour
provided the surface is moist) into the convecting layer, thereby warming it
(Figs 5.2 and 5.9).
The dry adiabatic lapse rate is maintained because the incessant mixing of the air
parcels as they rise and fall in convection holds the vertical temperature profile of
the convecting layer very close to the dry adiabatic profile traced out by each
moving parcel before it is mixed out of existence. It is true that rising parcels are
usually warmer than falling ones (that is why they rise) but the small size and brief
lifetimes of the parcels ensure that such temperature differences are small. In
essence the environmental lapse rate stays very close to the individual parcel lapse
rate because the environment itself consists of an incessant flux of such parcels,
each of which moves nearly dry adiabatically in its brief life.
The second major effect of convection, returning to the case of the morning
removal of the nocturnal stable layer, is the gradual warming of the convecting
layer (Figs 5.2 and 5.9) which comes about because on average the air parcels rising
into the layer are warmer than the average temperature of the layer. As these
warmer parcels mix out of recognizable existence, their slight temperature excess is 131
5 Atmospheric given to their surroundings, which therefore gradually warm. The result is that
thermodynamics although the instantaneous environmental lapse rate remains close to the dry
adiabatic value, or some consistently superadiabatic value if the sun is strong, the
whole layer warms gradually but persistently. In Fig. 5.9 the temperature profile
moves bodily to the right as time passes, without change of shape or slope in the
convecting zone. However, the presence aloft of some remnant of the convectively
stable temperature. profile developed overnight ensures that the depth of the
convecting layer increases as it warms. Since this means that a progressively larger
mass of air is being warmed as time goes by, it follows that the rate of warming
should tend to decrease. This decrease may be partly offset by an increasing rate of
solar input as the sun rises toward the noon maximum, but the net result is that the
observed maximum rate of warming occurs earlier rather than later in the morning.
In fact the warming tends to become very small indeed if the deepening convecting
layer connects up with the upper part of the nearly dry adiabatic layer left over
from the previous day's convection, as often happens in the late morning just
before cumulus appear.
During the afternoon the solar input declines but usually remains strong enough
to maintain convection in the sub-cloud layer, and with it the characteristically dry
adiabatic environmental lapse rate. However some little time before sunset the net
loss of heat by infra-red radiation from the ground begins to exceed the net input
from the sun, and the lapse rate close to the surface becomes subadiabatic.
Convection there dies almost at once, and with it all the thermally driven con-
vection through the layers above. Air sufficiently far above the surface may still be
stirred by mechanical convection, and even if it is not, may retain a nearly dry adia-
batic lapse rate because it is largely disconnected from the cooling surface.
However, once formed the ground-based stable layer tends to deepen throughout
the night, reproducing by dawn next day another version of the temperature profile
with which we began this section.
A complication can arise in the very lowest layers when the sun is fairly strong. It
seems that the inevitably very small-scale convection which dominates at such low
levels does not transport sensible heat as efficiently as does the larger-scale con-
vection aloft, with the result that it may be unable to cope with the strong solar
input. When this happens the surface and overlying air become warmer, and the
resulting distinctly superadiabatic lapse rate restores thermal balance by driving
the convection more vigorously. The superadiabatic effect may be quite marked in
the lowest few metres of air, but diminishes so sharply with increasing height that it
is virtually unobservable more than a few tens of metres above all but the most
intensely heated surfaces (such as subtropical deserts in summer). This behaviour is
examined further in section 9.9. At greater heights, that is throughout most of the
convecting sub-cloud layer, temperature lapse rates are observed to lie within a few
percent of the dry adiabatic value.
Note that quite vigorous stirring is often observed in the absence of significant
surface warming by the sun- for example at night and over substantial depths of
water. Provided that a surface is not being strongly cooled by terrestrial radiation,
a turbulent tumbling motion is often maintained by moderate or stronger winds
near the surface, as shown in Fig. 5.10. This is termed mechanical convection to
distinguish it from the more obvious thermally driven convection already men-
tioned. Over many ocean surfaces the sub-cloud layer is more or less continually
stirred by a combination of mechanical and thermal convection. In summary it is
the vertical movements of stirred air, whether the stirring is driven by wind or
surface heating, which together with the dry adiabatic parcel process hold the
environmental lapse rates in the sub-cloud layer so close to the dry adiabatic value
132 for so much of the time.
 Thermal Mechanical Appendix 5. 1

_ 3_ __
------------. Mean

--------~

--- _..,..
Wind
Profile

Fig. 5.10 Thermal and

mechanical origins of
Horizontal turbulence.

Now that we have examined the particular case of the sub-cloud layer, we can
consider the maintenance of neutral convective stability more generally. In a
cloudy layer the individual parcels tend to move according to the saturated
adiabatic process, and convection therefore tends to maintain a similar tempera-
ture profile. The convection may originate in a sub-cloud layer, or in the cloudy
layer itself by release of latent heat from some triggering uplift, for example, or it
may be mechanical in origin. Though the principle established in the cloud-free
case clearly still applies, the behaviour in cloudy air may be considerably
complicated in practical detail, especially if the layer is only partly full of cloud, as
is often the case.
The maintenance of neutral convective stability may be seen as an example of the
operation of an extremely general law of mixing which can be stated as follows: the
distribution of each component of a continually stirred mixture tends toward a
dynamic equilibrium which is unaffected by further mixing. The mixing of ink and
water mentioned in section 3.1 is a simple example; only during the initial stirring is
there any visible change in the appearance of the mixture, which quickly settles into
a uniform shade apparently unaffected by further stirring. The uniformity of
specific masses of gases in the turbosphere (section 3.1) exemplifies the law again,
and shows that the result need not be as obvious as in the case of ink and water. And
in the present case the result is even less obvious. However, the law is still clearly
applicable: if the parcel process is dry adiabatic and the stirring has a component of
vertical exchange, the only vertical temperature profile which is unaffected by
further stirring is one with a dry adiabatic lapse rate, and this is the one so widely
observed.

Appendix 5.1 The first law of thermodynamics

(meteorological form)

Since the volume of unit mass of material is inversely proportional to its density,
the first law of thermodynamics (eqn (5.1)) can be written

dQ = Cv dT + p d( ~ )

= Cv dT + d( ~ ) - ~ dp
1
= Cv dT + R dT- - dp
p 133
5 Atmospheric
thermodynamics since p =R T (equation of state- eqn (4.2))
p
1
dQ = (C. + R) dT - - dp
p

By considering the case where pressure is kept constant (dp = 0) we see that c.+ R is
the specific heat capacity at constant pressure CP.
1
Hence dQ = CP d T - - dp
p

Appendix 5.2 Precipitable water content

Consider a vertical column of moist air, with air density p and specific humidity q.
The mass m of water vapour in a slice of depth (z 2 - z,) and unit horizontal area is
therefore given by
q P (z2 -z,)
which, according to eqn (4.16) (a form of the hydrostatic equation), can be
rewritten as

!!_{p,-Pz)
g
where p 1 and p 2 are the air pressures at the base and top of the slice respectively.
If this vapour were to be precipitated into a rain-gauge at the foot of the column,
it would yield a depth
h=mlpw
where Pw is the density of the precipitated water. It follows that
h = [q{p,- Pz))/g Pw
The term in square brackets can be readily calculated for any slice of an upper air
sounding in which humidity is reported, and the total precipitable water content f.h
found by adding the contributions from all moist slices. Note that f.h is given in
millimetres of equivalent rainfall by
w-zr, q; AP;
where q; is in g kg_, and AP; is in mbar. Values can be read directly off a plotted
tephigram or its equivalent.

Appendix 5.3 A skeleton tephigram

Let us outline a tephigram in the temperature (T) range -60 octo +60 oc and the
134 potential temperature (0) range -60 octo + 100 °C. Convert 0 values to degrees
kelvin and find their logarithms to the base ten. Draw the log() axis in the vertical Appendix 5.3
and the T axis in the horizontal, and draw and label the isotherms and potential
isotherms (dry adiabats) at 20 °C intervals including 0 °C.
1. Draw and labelthe 1000 mbar isobar as the locus of points where()= T. Prepare
to draw and label the 500 and 250 mbar isobars by taking the logarithm of eqn
(5.15).

log() = log T + !i.. log ( 1000 )

cp P
Since the first two terms represent the 1000 mbar isobar, any other isobar (p)
can be found by sliding the 1000 mbar isobar up the log() axis by a distance
given by the numerical value of

..B._ log ( 1000 )

cp P
We can therefore find the 500, 250 and 125 mbar isobars by sliding up multiples
of the distance 0.0861 (i.e. (R/CP) log 2).
2. Prepare to draw and label the 10, 5 and 2.5 g kg- 1 isopleths of saturation
specific humidity qs as follows. Convert the qs values into straight fractions
(the largest being 10- 2), and use these in eqn (5.8) to find the corresponding
saturation vapour pressures on the 1000 mbar isobar. The values will agree very
closely with some of the following table, and the corresponding temperatures
(i.e. the temperatures at which these are the saturation vapour pressures) can be
used to locate the intersections of the 1000 mbar isobar with the three q
isopleths. Repeat for the other three isobars, and connect the intersections to
form the q isopleths.

e/mbar 16 8 4 2 1
TfOC 14 3.7 -5.6 -14.7 -24

3. After reading section 5.1 0, sketch and label the saturated adiabat which passes
through the intersection of the 10 g kg-• isopleth of qs with the 1000 mbar
isobar, using the following approximate method, which simulates the con-
tinuous release of latent heat in a rising air parcel by a three-step process.

(a) Warm air by isobaric release of latent heat by condensing 5 g kg-• of

vapour (note that Tincreases by almost exactly 2.5 °C when qs falls by 1 g
kg-•- eqn (5.19) with dp zero applies). Cool this air dry adiabatically until
it saturates again by moving it along a dry adiabat until you reach the 5 g
kg -• isopleth for qs. (The vapour content of the air has now of course fallen
to 5 g kg- 1.) This is the intersection of the chosen saturated adiabat with
the 5 g kg -• qs isopleth.
(b) Repeat process (a) starting at its end point and condensing half of the
remaining vapour to reach the intersection of the saturated adiabat with qs
2.5 g kg-•.
(c) Find the asymptotic dry adiabat at the cold end of the saturated adiabat
(the equivalent potential temperature 8.) by using eqn (5.21) and noting
that ()w is simply the() value at the 1000 mbar isobar (the beginning of the
saturated adiabat).
(d) Connect up the intersections and the asymptote by a smooth curve which is
your representation of this particular saturated adiabat (()w 13.7 °C). The
simulation could be improved and made much more realistic in execution 135
5 Atmospheric by using many small alternately isobaric and dry adiabatic steps instead of
thermodynamics three very gross ones, but there would be no change in principle or in the
essential shape of the resulting saturated adiabat.
Some parts of your completed diagram represent conditions not observed in the
atmosphere. Pressures very seldom exceed 1050 mbar; dry adiabats extending up
from hot dry surfaces seldom exceed a potential temperature of 45 °C; deep
saturated convection (the agent of tropospheric warming in depth) seldom rises
from surface layers with specific humidity more than 25 g kg_,. Since this
saturates air at a temperature of just over 28 oc at 1000 mbar, it follows that the
maximum equivalent potential temperature is just over 90 °C, and this is realized
at about the 125 mbar level. A rough saturated adiabat can be sketched to indicate
the limiting conditions. Hatch the areas excluded by all of these limits, and
compare with Fig. 5.6.

Appendix 5.4 Equivalent potential temperature

Taking the logarithm of the equation defining potential temperature (eqn (5 .15)
and appendix 5 .3) and then differentiating to find the relationship between very
small changes in potential temperature (dO), temperature (dT) and pressure (dp),
we finally obtain
dO dT _ R dp
0 T
which shows that 0 increases with isobaric warming (positive dTand zero dp) and
decreases with isothermal pressurization (positive dp and zero dT).
If we now rearrange the meteorological form of the first law of thermodynamics
(eqn (5.2)) by replacing the air density using the equation of state (eqn (4.2)), and
dividing across by CP T, we find
dQ = dT _ R dp
CPT T Cpp
Since the two equations now derived have the same right-hand side, we have
dO dQ
(5.25)
0 CPT
which shows that the potential temperature increase dO is proportional to the heat
input dQ divided by the absolute temperature Tat which this occurs. This quotient
is in fact the increase in entropy of the system if the heat input occurs reversibly
(i.e. in a manner so close to thermal equilibrium that an infinitesimally small
change in conditions could reverse its direction). Entropy occupies a fundamental
place in the more advanced analysis of systems close to thermodynamic
equilibrium [35], and though its relevance to natural and man-made systems (all of
which are quite significantly irreversible) is complex and still unclear, it has a
general qualitative significance as a measure of the degree of disorder of complex
systems. The dry adiabatic reference process is often termed isentropic since by
definition it is reversible and assumes zero dQ.
136 In the saturated adiabatic reference process, the heat input during a small
adiabatic ascent is given by - L dq. (eqn (5.19) and preceding discussion), so that Problems
eqn (5.25) becomes
dO
(5.26)
0
which becomes eqn (5.20) after trivial rearrangement. In fact L dq. is not a heat
input from outside the parcel - it is energy released to one part of the parcel (the
air) by a substantial decrease in the molecular disorder in another part (the water
substance which has condensed from vapour to liquid). The saturated adiabatic
process is therefore also isentropic. The analysis of air motions following dry or
saturated adiabatic processes is therefore generally termed isentropic analysis
(section 11.4).
To find a relation between the equivalent and wet-bulb potential temperatures
(section 5.11) we follow a saturated adiabat from 1000 mbar to a pressure so low
that effectively all vapour has been condensed. By definition the initial 0 is the wet-
bulb potential temperature Ow and the final 0 is the equivalent potential temperature
Oe. We need to integrate eqn (5.20) (eqn (5.26) multiplied by 0) from initial to zero
q •. The simplest approach is to notice that most of the condensation occurs in the
warmest parts of the saturated adiabat, where of course 0 and Tare nearly equal
because pressures are close to 1000 mbar. To the extent that they are equal the
right-hand side of eqn (5.20) can be integrated immediately to give
Lq.
(5 .27)
cp
which is eqn (5.21). The ratio LICP simplifies to 2.5 when q. is expressed in grams
per kilogram. Equation (5 .27) is a special case of a general relationship between rise
in potential temperature and condensation of water vapour (fall in q.). More
thorough analysis [8] shows that the degree of approximation is even better than
suggested above, as is confirmed by close comparison with thermodynamic tables
and diagrams.

Problems

LEVEL 1

I. By applying the first law of thermodynamics qualitatively to a parcel of air

first in a rigid container and second in a slack plastic bag, argue why the
specific heat at constant pressure (Cp) should be larger than c ..
2. Examine the following list and decide which processes are likely to be approxi-
mately isobaric: daytime warming of the full depth of the atmosphere; air in a
shower cloud updraught; air flowing horizontally through the core of a jet
stream; air spiralling horizontally into the core of a tornado; the atmospheric
boundary layer during nocturnal cooling.
3. The humidity mixing ratio of adjacent air tends to rise when there is evapora-
tion from a land surface. What surface activity tends to make it fall? Can this
happen at ocean surfaces? Can it happen at frozen surfaces? Briefly describe
what happens in these last two cases. 137
5 Atmospheric 4. Air in the upper branches of the Hadley circulation may take 10 days to move
thermodynomics from the equatorial to the subtropical zones. Consider the applicability of the
adiabatic assumption to this process.
5. If meteorology had been pioneered on the Tibetan Plateau (about 4 km above
mean sea level), the reference level for potential temperature might well have
been chosen to be 600 rather than 1000 mbar. In this case, consider the follow-
ing pressure levels and decide in each case whether actual temperatures would
be greater, equal to or less than Tibetan potential temperatures: 1000, 700,
600, 300, 100 mbar.
6. Which of the following vertical temperature profiles is likely to be spurious?
Well clear of the surface: lapse of 1 oc in 200m, in 80 m; rise of 2° in 300m.
Close to the surface: lapse of 2° in 50 m, rise of 2° in 50 m.
7. On a certain saturated adiabat, place the following in order of ascending
numerical value: temperature at 500 mbar, wet-bulb potential temperature,
equivalent potential temperature, potential temperature at 500 mbar.

LEVEL 2

8. On a certain cloudless night, an infra-red radiometer showed that the net rate
of loss of radiant energy from a ground surface averages 50 W m- 2 for 8
hours. If the heat lost is drawn from 30 mbar 'depth' of air, find the resulting
fall of air temperature, assuming it to be uniformly distributed through the
layer.
9. The saturation vapour pressures at 15 and 10 oc are respectively 17.0 and 12.3
mbar. Find the associated saturation specific humidities when the atmospheric
pressure is 1000 mbar. Also find the relative humidity of air which has a
temperature of 15 oc and a dewpoint of 10 °C.
10. On a certain summer's day in mid-latitudes the air temperature at screen level
is observed to be 25 °C. If there is no significantly superadiabatic layer near
the surface, what is the minimum height of the 0 oc level above the surface?
11. Given that air at 1000 mbar pressure and 15 oc temperature has a wet-bulb
temperature of 10 °C, find its specific humidity and relative humidity,
assuming the measured wet-bulb temperature is identical to the thermo-
dynamic wet-bulb temperature.
12. In a mishap which was fortunately not fatal, an airliner depressurized
suddenly by losing a cargo door at altitude. If the internal and external
pressures were 750 and 400 mbar respectively just before depressurization, and
the internal air temperature was 22 °C, estimate the internal temperature just
after depressurization, assuming dry adiabatic cooling. No wonder passengers
complained of the cold!
13. In middle latitudes, air at the 300 mbar level is often observed to have
temperatures as high as -40 oc in the vicinity of jet-stream cores. Find the
potential temperature of such air. Use Fig. 5.5 to find its wet-bulb potential
temperature and the equivalent potential temperature.

LEVEL 3

14. Recalculate the temperature fall in problem 8 if the heat loss is shared with a
138 20 em deep layer of ground surface in addition to the air layer. Assume ground
 Problems

,,-' •/

' '
"' " /

MO
' /
/.

-,.,o...,,
oo-v-
"
~·
,.,;.., Fig. 5.11 Tephigram section
00
far the law troposphere
1000..,

.
(problem 5.15). Axes and
..'• .
10-

.
20

- isopleths are as in Fig. 5.5

'• ...
m

' ' 0
' (Crown copyright)

density 2000 kg m- 3 and specific heat capacity 2000 J kg- 1 K- 1• In which

direction would the answer be changed if the cooling produced fog?
15. Given that air at 1000 mbar and 20 oc has relative humidity 600Jo, find values
for all possible humidity parameters (including the lifting condensation level),
using the low-level section of the tephigram in Fig. 5 .11. Estimate the
equivalent potential temperature by using Fig. 5.5.
16. In the event described in problem 12 passengers reported a temporarily dense
fog in the cabin just after depressurization. Given that the cabin wet-bulb
temperature was 17 oc before the event, use Fig. 5.5 to make a more realistic
estimate of the cabin temperature just after depressurization. Why was the fog
only temporary?
17. Discuss the realism of dry adiabatic ascent from the surface to the jet core in
problem 13. If instead there was saturated adiabatic ascent, what global
position for the surface source is suggested by the data?
18. Consider a situation in which nocturnal cooling under clear skies establishes a
convectively stable layer in the first 300 m of the atmosphere, with a tempera-
ture inversion in the lowest 100 m. Sketch an example on the tephigram
fragment above, and on an equivalent graph of potential temperature versus
height. Now suppose that an approaching depression produces cloudy over-
cast which stops the cooling and rising winds which encourage mechanical
convection through the lowest 200 m. Sketch the altered temperature profile
on the same two diagrams.
19. Estimate the precipitable water content of the Long Kesh sounding (Fig. 5.5),
assuming the humidity profile to be accurate, and compare it with values from
saturated cold and warm atmospheres (0 oc and 20 oc saturated adiabats).

139
CHAPTER
Cloud and
6 precipitation

6.1 Cloud

The Earth's lower atmosphere is typically cloudy. At any instant about half the
planet's surface is overlain by cloud varying in thickness from a few tens of metres
to the full depth of the troposphere (the site of all but a minute fraction of
terrestrial cloud). All this cloud is extremely variable in both time and space, and
individual cloud elements are usually quite ephemeral. For example a cumulo-
nimbus may develop so rapidly that within half an hour of its first appearance as a
little cloudlet it effectively fills the troposphere in a region ten or more kilometres
across; and within a further hour it may vanish again, leaving no trace of its
majestic maturity beyond its legacy of precipitation on the underlying surface.
Lifetimes of constituent cloud droplets and crystals may be considerably shorter.
If you watch the outer parts of a cumulonimbus for a few minutes, you will
become aware that everything is in a state of ceaseless commotion, with bumps
and filaments emerging from the main cloud mass and evaporating to extinction
within a few tens of seconds. Simple calculations based on observed rates of
precipitation and estimates of cloud water masses suggest that even in the cloudy
interiors, individual droplet and crystal lifetimes are often just a few tens of
minutes.
The great swaths of cloud associated with middle-latitude fronts (nimbostratus)
are apparently much more persistent, often lasting for a week or more. But careful
studies show that air flows through them so quickly that any particular air parcel
remains within its parent cloud mass for only a day or two before emerging and
evaporating its cloud content.
Because cloud is so prevalent and yet so transient, its formation and destruction
must be both widespread and rapid, and the mechanisms involved commonplace.
In this chapter we are going to examine these mechanisms in some detail, but first
we must consider carefully the state of saturation which is observed to be essential
for the formation and maintenance of all significant cloud.

140
6.2 Saturation 6.2 Saturation

Consider the state of affairs in a very shallow layer of air in contact with an exposed
water surface (Fig. 6.1). As well as dry air, it contains a vapour consisting of a gas
of water molecules which is continually being fed by evaporation from the water
surface and drained by recondensation there. In terrestrial conditions, if we
consider a layer extending only a millimetre or so from the water surface, the
feeding and draining fluxes are so nearly equal that the layer is effectively in a state
of dynamic equilibrium which is termed saturation. As the name implies,
saturation is a state of maximum vapour content, since any tendency to exceed the
saturated state, by migration of more vapour from outside the layer, for example,
would be offset by increased recondensation at the surface. The analogy with a . ..Jj. Vapour

-~Water
saturated sponge is obvious, but it is unfortunate that it has become conventional
to speak of the air as being saturated, since it is clearly the vapour rather than the "" / or ice
air which is saturated; indeed the state of saturation is quite unaffected by the
presence of the dry air. However, the convention will be followed in the rest of this
Fig. 6.1 A saturated vapour
book, with the tacit understanding that saturated air means air whose vapour is overlying a plane surface of
saturated. pure ice or water .
The dynamic equilibrium between a water surface and its adjacent saturated air
ensures that both surface and air (including both vapour and dry air) have the same
temperature. When the water is pure and its surface is plane, it is found by observa-
tion that both the density and pressure of the saturated vapour depend only on their
common temperature. (Note that when the vapour is mixed with dry air we must
use the partial pressure or the density of the vapour.) Both vapour density and
pressure increase with temperature in the nearly exponential manner shown in
Fig. 6.2, the rising gradient corresponding to the increasing ease with which water
molecules escape from the surface as its temperature and their associated thermal
agitation increase. Across the atmospheric temperature range the increases are very
large - a fact which proves to be very important for cloud formation and
development.
Note that Fig. 6.2 also includes values for saturation close to a plane surface of
pure ice. Remember that ice can evaporate (sublime) without melting first, as is
evidenced by the disappearance of snow patches at sub-freezing temperatures. The
saturation vapour pressures and densities for an ice surface are less than those for a
supercooled water surface at the same temperature because the stronger inter-
molecular bonding in ice inhibits evaporation.
Careful observation shows that saturation occurs very close to exposed surfaces
of water or ice even when precise equilibrium between evaporation and recon-
densation does not apply. The disappearing snow is a case in point, as is the often
very rapid disappearance of water puddles in the presence of sun and wind - both
observations indicating an excess of evaporation over recondensation. The
formation of dew or hoar frost indicates an imbalance in the opposite direction.
Such disequilibria are normal in terrestrial conditions, but on detailed enquiry it
appears that in almost all such cases the difference between evaporating and con-
densing fluxes is so much smaller than either of the fluxes separately that the
adjacent vapours differ insignificantly from the saturated ideal. We will therefore
assume from now on that a saturated vapour always exists within a millimetre or so
of an exposed plane surface of pure water or ice. At greater distances saturation
may or may not be present, depending on factors such as the mixing of the moist air
with drier air from other sources, just as the temperature of more distant air may
differ significantly from that of the surface and closely adjacent moist air.
141
6 Cloud and precipitation
40+------1------~------+------+------4------R

30-1---l---+-----1-~/
2Q.+------4------~------+------+----~~~
v
/
[ ,"-1~-~~v
~ 10 +------+------~------~--~~
~------~----~

~ suPERC~~
~ 0~==~~====~~------+------+------~------1
.2
;o
:; i'ss
~ 6 +-'
~---+------~------~-----+------+-------+30

Fig. 6.2 Variation of

saturation vapour pressure
with temperature. Below 0 •c
" '\
.\
• +------+~-----+-
'\
. .l' /
/ /
/~--+------+------~------+ 20
....•'/
SS • 100 c svp iater - 1\
values for supercooled water X
svp ce '.J %
exceed those for ice, and the "J>~ /
chain line shows the
2+------~~~~~
~~~-----+------+------+------+10
supersaturation (SS in % )
~~~·
",
'\
relative to ice of a vapour
which is saturated relative to ~;
supercooled water. Find
0 ~------L----~-----·'~----~------~-----L O
corresponding vapour -30 - 20 - 10 0 10 20 30
densities by using the
equation of state for vapour. Temperature (°Cl

Finally note that saturation has been defined in terms of plane surfaces of either
pure water or pure ice. In fact dynamic equilibria are as readily set up over curved
surfaces and impure water or ice, but they may then differ markedly from those for
plane, pure surfaces. Indeed the differences are crucial to the formation of all
terrestrial cloud and will be examined in some detail in succeeding sections. To
avoid confusion we will reserve the term 'saturation' for the case of plane, pure
water or ice.

6.3 The sub-cloud layer

Clouds in the lower troposphere are normally separated from the Earth's surface
by a cloud-free layer whose depth may range from less than a hundred metres to
several kilometres, depending on weather conditions. Indeed the appearance of
clouds, or a cloud layer, above a substantial sub-cloud layer is so normal (Fig. 6.3)
that we might assume its explanation to be trivial. In this, as in so many meteo-
142 rological matters, we would be mistaken. However, the almost universal presence
 6.3 The sub-cloud layer

Fig. 6.3 Panorama of small

cumulus over low hills in
north·west England, with
cloud bases at a fairly
consistent level throughout
the field of view despite the
irregular topography.

of the sub-cloud layer in the low troposphere is so important for understanding

cloud formation there that we must try to explain it, if only in outline.
Seventy per cent of the Earth's surface is water, and much of the rest is moist
ground or ice. According to the argument of the previous section the air very close
to all but the most arid surfaces must therefore be saturated at surface tempera-
tures (the effect of impurities is not significant in this context). In the simplest view
this would lead us to expect to find a deep layer of saturated air covering most of
the surface and maintained by continual evaporation from it. Given the resulting
upward extension of the saturated layer, and the efficiency of the cloud-forming
processes to be described presently, we should expect much of the Earth to be
shrouded in fog. In fact very little fog is observed, which suggests that there are
mechanisms at work which are removing water from the atmosphere relatively
efficiently. These are of course the precipitation processes, which ensure that some
cloud droplets and crystals grow big enough to fall back to the surface under
gravity. Though the details of the precipitation processes are not considered until
later, we must now look at them from a very general viewpoint.
The least efficient type of precipitation is observed in deep fog layers, which con-
tinually lose part of their cloud water through a very fine drizzle of droplets inter-
mediate in size between the tiny cloud droplets and normal-sized drizzle droplets -
barely perceptible except when they touch our faces or car windscreens. This loss is
usually insufficient to clear the fog by precipitation, though some schemes for fog
clearance envisage its artificial enhancement.
More efficient types of precipitation depend on strong or persistent updraughts
and much deeper layers of cloud. When these are present, the rate of precipitation
onto any given surface area greatly exceeds the rate of evaporation from that area.
The situation is then as shown in Fig. 6.4: only a fraction of the Earth's total

Fig. 6.4 Vertical pathways of

vapour and precipitation in
the atmosphere.

143
6 Cloud and precipitation surface is covered by efficiently precipitating cloud, but this cloud and its precipita-
tion are maintained by evaporation from virtually the whole terrestrial surface, the
arid regions being relatively small. Between the cloudy updraughts there is sinking
air, desaturated by dry adiabatic descent after prior precipitation. As this air
approaches the surface it mixes with the saturated air continually produced by
evaporation there, thereby maintaining a layer of subsaturated and therefore
cloud-free air overlying the thin film of saturated air clinging to the surface.
Without going into details it should be clear that the depth of the subsaturated
layer, like the fraction of the total surface area overlain by precipitating cloud, is
controlled by the efficiency of the precipitating processes at work in the clouds. If
these were all as feeble as those in fog, for example, then our initial expectation of a
largely fog-bound planet would be fulfilled.
According to Fig. 6.4 the sub-cloud layer is continually fed by subsaturated air
from aloft and saturated air from beneath. In and just above the shallow saturated
layer, atmospheric mixing is relatively very feeble, with the result that there is a
shallow zone overlying the surface in which there is usually a significant vertical
lapse of vapour content, as sketched in Fig. 6.5. This zone is seldom more than a
few metres deep, and is bounded above by a deep layer, occupying most of the sub-
cloud layer, in which the vigorous stirring typical of the atmospheric boundary
layer in most conditions maintains a much more uniform distribution of water
vapour. This relatively deep and well-mixed layer plays such a basic role in the
formation of cloud in the low troposphere that we must consider it in some detail in
the following sections.

Specific humidity
_______ __________________
1
I
1:
I
Cloud base

I I
: I
I I
I I
I I
I I
I I
: I
I
.<:;
I
I
0> I

:r"
I
lq RH
I
I
I
Fig. 6.5 Idealized profiles of I
I I
specific humidity q and I
:
I
relative humidity RH in a 1 ~ Screen level
well-mixed sub-cloud Ioyer :----...:''""-----===-'. Wet surface
over a moist surface. 0 Relative humidity 1110

6.4 The well-mixed sub-cloud layer

Consider the vertical distribution of water vapour to be expected in the well-mixed

layer extending upward from the shallow transition zone (Fig. 6.5) immediately
overlying the surface. Ignoring relatively small volumes dampened by evaporation
from falling precipitation, there are no sources or sinks of vapour between the
lower and upper limits (the transition zone and cloud base respectively), so that the
equilibrium established by continual, efficient mixing must maintain uniformity of
144 specific humidity q, where by definition
 q = mass of contained vapour/mass of containing moist air 6.4 The well-mixed sub-
cloud Ioyer
=p/p (6.1)
and Pv and p are respectively the densities of vapour and moist air. Since air density
lapses with increasing height, the vapour density must lapse in proportion, to
maintain q. A typical depth of the sub-cloud layer is 1 km, in which both the
pressure and density of air fall by about 100Jo, and it follows that vapour pressure
and density must do likewise.
The stirring of the sub-cloud layer also establishes an equilibrium vertical
distribution of temperature, which is the dry adiabatic profile as discussed in
section 5.13. This temperature profile can be converted into a profile of equivalent
saturation vapour density by making use of Fig. 6.2 or thermodynamic tables. The
combination of the steepness of the dry adiabatic lapse rate and the rapidity of
the variation of saturation vapour density with temperature (i.e. the steepness
of the slope of Fig. 6.2) results in very large lapses of saturation vapour density
in the well-mixed layer. For example the 9.8 °C temperature lapse expected in a
kilometric deep layer is associated with a lapse of saturation vapour density
from 12.8 to 6.8 g m- 3 , assuming that the air at the bottom of the well-mixed
layer has temperature 15 °C. In warmer layers the lapse of saturation vapour
density is even larger because of the increasing rapidity of variation of saturation
vapour density with temperature (i.e. the increasing steepness of the slope of
Fig. 6.2).
We have now established that the lapse of saturation vapour density p, in the
well-mixed layer far exceeds the lapse of actual vapour density Pv· Since the relative
humidity (RH) is defined to be the ratio of the two densities expressed as a
percentage, that is
RH = 100pvlp, (6.2)
it follows that RH must increase with increasing height from a minimum value at
the base of the well-mixed layer, as shown in Fig. 6.5. This increase means that at
some level, higher or lower depending on the degree of subsaturation of the air
at the base of the well-mixed layer, the value of RH approaches 100%. Since this
value corresponds to saturation, it follows that any further ascent by an air parcel
would produce supersaturation and consequent cloud formation by condensation
of excess vapour onto cloud droplets. The level of first saturation therefore
corresponds to cloud base and the upper limit of the sub-cloud layer. In fact to the
extent that the atmosphere behaves like our simple adiabatic and well-mixed model
it coincides with the lifting condensation level defined in section 5.11.
Note that we have now produced an explanation for a fact which is so common-
place that we might have supposed no explanation was ever needed in the first place
- namely that saturation and cloud develop in rising rather than in sinking air. But
if for example the variation of saturation vapour density with temperature had
been very much smaller than is apparent in Fig. 6.2, so that the vertical lapse of
saturation vapour density in the well-mixed layer was less than the lapse of actual
vapour density, then the opposite would have been the case: rising air would have
desaturated, giving no possibility of an elevated saturated layer. And without the
associated precipitation, the Earth would be as fog-bound as naively assumed at
the beginning of this section, with far-reaching consequences for the environment
in which life has developed. It is clear that the apparently innocuous balance of
factors leading to the production of cloud in rising rather than sinking air is of
profound importance to us all.

145
6 Cloud and precipitation
6.5 The dew-point profile

Humidity profiles in the sub-cloud layer can be depicted most graphically using
dew-point temperatures. Consider air at the base of the well-mixed layer (Fig. 6.6).
It is significantly unsaturated and so its dew-point Td is significantly below its
temperature T. It follows from the mode of operation of a dew-point meter
(section 2.3) that the dew point of this air can be found on Fig. 6.6 by sliding iso-
barically to the left until saturation is reached, i.e. until the saturation specific
humidity q. (whose isopleths are the faint dashed lines obliquely crossing the
isotherms) is equal to the specific humidity q of the air. The temperature at which
this happens is the dew-point Td.

Fig. 6.6 Tephigram of the

lower troposphere showing
idealized profiles of
temperature T, wet-bulb
temperature Tw and dew-
point Td in a well-mixed, dry
adiabatic sub-cloud layer,
with lifting condensation level
LCL. This is the completed
version of Fig. 5.7(b).
Average temperatures and
pressures in the sub-cloud
layer are such that 1 mbar
pressure lapse corresponds to
about 9 m elevation.

Now according to our simple model of the well-mixed layer, the specific
humidity is the same at all levels up to the lifting condensation level. Therefore the
dew-point profile in Fig. 6.6lies along a line of constant q •. The detailed thermo-
dynamics underlying the tephigram ensure that q. isopleths are nearly straight lines
which are almost perpendicular to the dry adiabats, and so it follows that the
temperature and dew-point profiles in the well-mixed layer are effectively straight
lines which converge sharply and meet at the lifting condensation level LCL
(equality of Tand Td of course implying saturation). In fact we can now complete
the full expression of Normand's theorem (section 5.11) by stating that the dry
adiabat through the temperature, the saturated adiabat through the wet-bulb
temperature and the saturation specific humidity through the dew point all meet at
the lifting condensation level.
It can be shown from detailed thermodynamics (appendix 6.1) that the con-
vergence of the T and Td profiles in typical low troposphere conditions is such that
the dew-point depression T- Td falls by about 0.80 oc for each 100 m ascent
through the well-mixed sub-cloud layer - a value which can be roughly checked
using the height scale added to Fig. 6.6. It follows that if the dew-point depression
at the base of the layer is .6. T °C, then the height of the lifting condensation level
LCL above the base is approximately 125 x .6. Tmetres. This rule of thumb, or the
slightly more accurate equivalent construction on a tephigram, is often used to
calculate cloud-base height from Stevenson screen measurements of Tand Tw (con-
verted to T and Td). However, heights calculated in this way are often observed to
undershoot the true cloud base by lOOJo or more, and a brief discussion of the
reasons for such systematic error will usefully qualify the too simple model of the
146 well-stirred layer assumed so far.
 The parcel processes assumed in the dry adiabatic and specific humidity con- 6.6 Condensation observed
serving model are self-consistent under detailed scrutiny, so that for example
Normand's theorem is fully justified; however, they do not exactly match atmo-
spheric behaviour. In particular the actual mixing process falls short of the 1OOOJo
efficiency assumed in the model, and seems to do this in systematic ways. This has
been admitted from the outset in the case of the transition zone between the shallow
saturated surface layer and the well-mixed layer, but it also applies to the well-
mixed layer itself, though much more subtly. The details are not fully understood,
but their upshot seems to be that the actual environmental temperature lapse rate
often exceeds the dry adiabatic value by a few per cent in the first few hundred
metres above warm ground, and undershoots it to a similar extent in the rest of the
layer from there up to cloud base [40]. These effects, together with the observation
that the base of the well-mixed layer often lies somewhat above Stevenson screen
level (remember that this is only 1.5 m above the ground surface) result in the
underestimation of cloud base height mentioned previously. The simple model
may be good but it is not perfect.
Finally, note that the simple model discussed in this and preceding sections
applies only to a well-stirred layer, i.e. one in which air parcels are continually
moving up and down because of thermal or mechanical convection or both.
Though such stirring is very widespread in the sub-cloud layer, it is not universal,
being either ineffective or absent in the convectively stable layers which often
develop in the first 100-200 m over a land surface on a cloudless night. In such
conditions there is no simple behaviour observable in the profiles of temperature
and humidity in the stable layer, though at higher altitudes the profiles of the
previous day's deep convection may still remain. Falling temperatures and rising
relative humidities are of course observed near the surface as nocturnal cooling
progresses, and saturation and fog may result, but no simple quantitative relations
apply to their profiles to compare with those in well-mixed layers.

6.6 Condensation observed

You can watch cloud forming in rising air most easily when weather conditions are
encouraging fair weather cumulus (Fig. 6.3). As an updraught begins to reach its
lifting condensation level, the air changes its appearance within a few tens of
seconds from the slight haziness typical of the upper parts of the sub-cloud layer to
a state in which there is an embryonic patch of cloud veined and clotted with
thickening elements which quickly merge to form a small cumulus. The nearly
horizontal base, representing the lifting condensation level, and the cauliflower-
surfaced upper parts, are familiar enough from everyday observation (Fig. 6.7),
but the speed and dynamism of the formation and development of such clouds are
seldom noticed by the sporadic skyward glances allowed by modern suburban and

Fig. 6.7 Outline of o typical

Lifting small cumulus forming above
-Condensation the lifting condensation level.
Level
+ + +
Sub-cloud layer 147
6 Cloud and precipitation urban life. A few minutes spent continuously watching the sky on a bright
morning, just as the cumulus are beginning to appear, will offset the quite
misleading impression of the static drift of clouds which most people have on the
basis of their snapshot observations, and which unfortunately is reinforced by
Wordsworth's uncharacteristically inaccurate image:
I wandered lonely as a cloud
that floats on high o'er vales and hills.
Careful measurement by sensors mounted on aircraft shows that air throughout
the interiors of many clouds is almost exactly saturated with respect to liquid water,
having relative humidities of between 100 and 101 OJo (i.e. supersaturations of
between 0 and 1%). This, and other aspects of behaviour which will prove to play
important roles in the cloudy atmosphere, can be observed in a very simple cloud
chamber which can either be set up as described in the following, or merely
accepted as a descriptive framework for the discussion of cloudy behaviour.
Find a large glass bottle (the five-litre capacity types used for home winemaking
are ideal), place a small quantity of water inside and fit the mouth with a bung
pierced by a narrow glass tube tipped by a length of rubber tubing soft enough to
form an air-tight seal under finger pressure (Fig. 6.8). Illuminate the interior of the
bottle strongly with, for example, a slide projector placed close by. A useful extra is
a water manometer tall enough to measure an interior pressure excess of 30 mbar
(i.e. a 30 em difference in water level - see appendix 4.2).
After shaking the bottle to ensure that it and its contents are all at the same
temperature and that the interior air is saturated, blow into the rubber tube and
maintain the excess pressure for about 30 s by pinching. Although in principle the
blowing must introduce some warm, moist air, its main effect is to raise the interior
pressure and hence raise the interior air temperature by compressional warming. A
little further shaking should ensure that the interior air is saturated once again,
having been temporarily desaturated by the warming. If the rubber tube is now
released the excess interior pressure falls to zero, chilling the saturated air by
decompression and producing an apparently tenuous cloud. In fact the cloud
seems tenuous only because its volume is small: if a room were to be filled with
cloud of this concentration it would be judged to be a very dense fog (Fig. 6.9).
This demonstration shows that cloud can form extremely quickly in air which a
little calculation shows is only modestly supersaturated. If for example pressure
falls by 30 mbar on depressurization, according to eqn (5.13) the temperature
should fall by about 2.5 °C. The actual chilling must be less than this, since the
formation of cloud tends to make it a saturated adiabatic rather than a dry
adiabatic process, and is probably nearer 1.5 °C. According to Fig. 6.2, air

Fig. 6.8 A simple cloud

chamber, showing bottle,
inlet and venting tube
(hanging right) and water
manometer.

148
 6.6 Condensation observed

Fig. 6.9 A thin morning fog

(visibility - 300 m) in winter
in England, probably
produced by inland advection
of warm, maritime air and
overnight radiative chilling .

saturated at the initial temperature (i.e. just before depressurization) must be

about 50Jo supersaturated just after depressurization. In fact more sophisticated
observation and theory suggest that cloud droplets can grow so quickly at the
expense of the supersaturating vapour, that the maximum supersaturation is
limited to I% or less. This certainly seems to be the case in almost all atmospheric
cloud, where rates of depressurization are very much smaller than in the cloud
chamber.
If you look carefully into the bottle just after decompression, you will see a
moving graininess in the strongly illuminated cloud. This is all that can be seen of
the individual cloud droplets without magnification, but we can estimate their size
very roughly by gauging their fall speed after the initial swirls have died away. The
grainy texture is then seen to be sinking at a rate of millimetres per second. From a

Fig. 6.10 Graph of fall speed

versus radius for spherical
102
bodies ranging in size from
small cloud droplets to giant
hail. Speeds shown apply to
the low troposphere. Speeds
10

'
at higher altitudes are
Drop s h atter somewhat larger on account

"
.5
of the lower air density there.

.,..
(Unpublished survey by the
'0
author, 1968)
Q

"' 10"1
"'
LL

Hydrome t eor radius r

10 mm
149
6 Cloud and precipitation well-established relationship between droplet size and fall speed (Fig. 6.10) it
follows that droplets falling at such speeds must have radii of the order of micro-
meters. This compares reasonably with droplet sizes observed in atmospheric
cloud: a variety of techniques indicate average radii of about 5 JLm.
A further demonstration using the cloud chamber reveals an unexpected com-
plication. Several minutes after depressurization most cloud droplets have sunk to
the bottom of the bottle. If the bottle is again pressurized and depressurized the
resulting cloud is noticeably more sparse. Subsequent repetitions separated by
similar waiting periods produce less and less cloud even though the degree of
pressurization and subsequent cooling is maintained at a constant level. This
increasing reluctance to form cloud is dramatically reversed when a small amount
of smoke, from a snuffed match for example, is introduced into the bottle. The
merest wisp is enough to produce a dense cloud at the next depressurization, whose
removal will require many further cycles of depressurization and settling.
It is clear from this demonstration that cloud forms much less readily in clean air
than it does in smoky air. In fact as mentioned in section 3.6, smokes are just a part
of the considerable range of atmospheric aerosol particles which have been shown
by work with cloud chambers to provide effective condensation nuclei, i.e. they are
able to encourage the formation of cloud droplets at low supersaturations. Aitken
and Coulier used cloud chambers in the late nineteenth century to count the
particles in atmospheric aerosol by forming cloud on them and then allowing the
cloud droplets to settle onto a counting plate. In so doing of course they isolated
the condensation nuclei in particular. In industrial regions the connection between
smoke and cloud formation is sometimes quite obvious: Fig. 6.11 shows a cumulus
developing where smoke from a strong source rises to the lifting condensation
level. To examine the mechanism by which condensation nuclei encourage cloud
formation we must look closely at the relationship between small droplets and their
immediate surroundings.

Fig. 6.11 Small cumulus

forming out of on industrial
smoke plume over London in
Summer.

6. 7 Condensation modelled [41]

Saturation has been defined as the state of a volume of water vapour in equilibrium
150 with an exposed plane surface of pure water or ice. However, plane surfaces are not
relevant to cloud formation more than a few millimetres above the Earth's surface. 6.7 Condensation modelled
Instead there is an aerosol containing roughly spherical particles and solution
droplets mostly with radii of less than one micrometre, often much less (section
3.6). Equilibria in such populations are vitally affected by the tight curvatures of
the available surfaces, and by the fact that many of the droplets are quite strong
solutions. Dissolved impurity is commonplace because many aerosol droplets
originate as particles of hygroscopic (water-attracting) material, such as sodium
chloride, which form droplets of concentrated solution when the relative humidity
rises far enough.
Consider first the effect of tight surface curvature on the equilibrium between
water (or ice) and the adjacent vapour. A water molecule comprising part of a
convex surface of small radius of curvature r is less tightly bound to that surface
than it would be to a plane surface in otherwise identical conditions because there
are fewer attracting neighbouring molecules within any chosen range (Fig. 6.12).
Evaporation therefore proceeds more rapidly, and the equilibrium vapour density
p, is higher than the equivalent value for a plane surface in the same conditions
(which is by definition the saturation vapour density pJ. This is confirmed by
detailed observation and theory, which shows that

..fr.. = exp (~) (6.3)

p, rT
where A is a constant for the droplet liquid (water in the present case) and Tis the
absolute temperature of the droplet and its immediate surroundings. The effect is
named after Thomson (Lord Kelvin) who first derived eqn (6.3).

Pla ne
Fig. 6.12 Molecular
Spherical
environment of a molecule at
a ··o·--.---,- - o the tightly curved surface of a
.<f'ti' --- ~ 0 0
,,6 0 0 0 0 c._ 0 0 0
0
0
0
0 liquid or solid body, in
,o 0 0 6.
0 0 0 0
b 0 ooo 0 d 0 0 0 0 0
comparison with a plane
surface.

According to eqn (6.3) the ratio p/ p, increases as r decreases (i.e. as the droplet
becomes smaller) and would become infinite as r vanishes. However, the finite size
of molecules places a lower limit on r, as does the finite probability of multiple
collisions between vapour molecules in a sufficiently dense vapour. When these
effects are included, and realistic values for A and Tare used, it can be shown that
such homogeneous nucleation of embryonic droplets occurs relatively frequently
only when p/p, exceeds a value of about four, i.e. when ambient relative humidities
are about 4000Jo. At smaller values the homogeneous nucleation is ineffective
because embryonic droplets form less frequently and evaporate again almost
immediately, the ambient vapour density p being less than the equilibrium value p,
for the tiny droplet. It follows that we should not expect clouds to form in clean,
moist air until it has a supersaturation of about 3000Jo. As already mentioned such
large values are never observed in the atmosphere, fairly obviously because it is not
clean. The role of condensation nuclei now becomes apparent.
Consider a volume of air containing particles much larger than the embryonic
droplets formed by homogeneous nucleation. According to eqn (6.3) the ambient
vapour density needed to exceed the equilibrium vapour density for such relatively
large particles, and thereby initiate condensation onto their surfaces, is much
smaller than the value required for effective homogeneous nucleation. In fact the
exponential decrease of equilibrium vapour density with increasing particle radius
is so sharp that a supersaturation of only 1OJo is enough to enable a particle of radius 151
6 Cloud and precipitation 0.15 J.tm to grow by condensation. (Strictly speaking, to apply eqn (6.3) with the A
value for water we must assume that it is already coated with a layer of water
molecules. Despite some complicating surface effects if this is not the case, the
argument as presented still stands in principle and applies quite well numerically.)
Such a low value confirms the importance of condensation nuclei in promoting
cloud formation; without them relative humidities would have to rise to about four
times the saturation value before cloud could begin to appear. In the simple
cloud-chamber experiment, the repeated scavenging of condensation nuclei by
cloud formation and settling produced air so artificially clean that further cloud
formation was inhibited until smoke was introduced. In the atmosphere the several
sources of aerosol mentioned in section 3.6 ensure that there is always an adequate
supply of condensation nuclei.
The role of condensation nuclei in encouraging cloud formation is further
enhanced when we consider that many of them consist of hygroscopic solids which
readily form droplets of quite concentrated solution. According to Raoult's law
the presence of solute reduces the equilibrium vapour density over a plane surface
of solution below the value for a plane surface of pure water (i.e. below the satura-
tion value) by an amount which increases with the solute concentration, and we
must expect a similar effect in the case of droplets. If we consider a mass m of solute
dissolved in a droplet of radius r, then the Raoult effect is represented by a negative
term of form Bmlr\ where the value of B varies only with the type of solute, for
any given solvent (water). If we focus on conditions when equilibrium and satura-
tion vapour densities are nearly equal, then the Thomson effect simplifies to the
first two terms of the series expansion of the exponential function, and the
resultant of the Thomson and Raoult effects is
A = 1 + __:!_ - Bm (6.4)
Ps rT r3
The value of B varies only with the type of solute, for any given solvent, and the
other symbols are as for eqn (6.3). The first two terms on the right-hand side of eqn
(6.4) represent the Thomson effect and the third represents the Raourt effect. Note
that the dependence of the latter on solute concentration introduces great
sensitivity to the value of the droplet radius.
Figure 6.13 represents the combined Thomson and Raoult effects for several
values of m typical of sodium chloride in atmospheric aerosol. Each Kohler curve
(named after a Swedish pioneer meteorologist) represents the variation of equili-

Fig. 6.13 Kohler curves for

droplets containing the
indicated moss of NaCI. The .~ 100.2
scale of the RH axis expands
·e
"0

above 100% to clarify the

structure there, introducing "
.<:
100.1
Q)
kinks in the curves which >
would otherwise be smooth. ;
100
(After (41]) ~
E
"
~ 90
·;
0"
w

10" 2 10.1
152 Droplet radius (~m)
 6.8 Cloud droplet growth
brium relative humidity (1 00 p/p,) with droplet radius r for a fixed mass m of
sodium chloride in water. When the salty droplet is small enough the salt con-
centration is so high that the Raoult effect dominates and the equilibrium RH is
reduced below the saturation value (IOOOJo) despite the opposing Thomson effect.
It is therefore possible for small droplets of sodium chloride solution (and other
solutions too- all Kohler curves have the same general shape) to be in equilibrium
with a subsaturated vapour, as was observed by Aitken. At much larger radii the
Thomson effect predominates, reducing the equilibrium supersaturation towards
zero as the radius increases. Between these extremes there is a critical droplet radius
rc for each particular mass of solute at which the equilibrium supersaturation is a
maximum. The existence of such a maximum vitally affects the behaviour of
growing droplets, as is outlined in the next section.

6. 8 Cloud droplet growth

Consider a hygroscopic aerosol particle in air whose relative humidity is rising

because of adiabatic expansion and cooling. At some RH value well below 100%
the particle becomes a tiny droplet of concentrated solution whose equilibrium
relative humidity is equal to the ambient value. As the ambient RH rises further
(because the fall of temperature implies a fall in saturated vapour density which is
much larger than the fall in actual vapour density- section 6.4) the droplet grows
by diffusion and condensation of vapour, its equilibrium RH rising to match the
ambient value so rapidly that the difference between the two is negligible in the
modest rates of rise of ambient RH associated with even the most vigorous
atmospheric updraughts. Hence the Kohler curve (Fig. 6.13) appropriate to the
mass and type of solute can be used to describe the growth of the droplet with
ambient RH. Strictly speaking we should allow for the effect of falling temperature
on the shape of the Kohler curve, but this is relatively small, although the effect on
the saturation vapour density is of course very large and important for the reasons
discussed in section 6.4. Note that the droplet and ambient vapour are in a state of
neutrally stable equilibrium in the sense that any change in the ambient vapour
content produces rapid growth (or shrinkage) to a new equilibrium size. In the
realistic case of a droplet containing I0- 18 kg of NaCl, the equilibrium radius
increases from 0.1 ~-tm at 87% ambient RHto 0.2 ~-tm at 99%.
Such neutrally stable equilibrium persists until the droplet reaches its critical
radius, which occurs when the ambient vapour is supersaturated to some slight
degree (0.13% in the case of I0- 18 kg NaCl). lfthe ambient RH rises further the
droplet enters a phase of runaway growth because the falling equilibrium RH can
no longer match the ambient value. The droplet therefore grows at a rate which is
limited only by the ability of the diffusive properties of the air to deliver vapour to
the swelling droplet and remove the latent heat liberated by condensation there.
Despite the typically small excess of ambient over equilibrium vapour densities the
droplet radius grows ten or twentyfold beyond its critical radius in a very short time
period, accounting for the almost instantaneous appearance of micron-scale cloud
droplets in the simple cloud-chamber experiment, and in even faster cloud
production (Fig. 6.14). Assuming that ambient vapour pressure remains constant,
droplet growth continues indefinitely, but at a slower and slower pace as measured
by droplet radius, essentially because the very rapid increase of droplet mass with 153
6 Cloud and precipitation

Fig. 6.14 Cloud forming

quickly in wing-tip vortices.
Net rotation of air round the
wings (onticlockwise round
the near wing) spins off into
trailing vortices whose core
pressures are low enough to
produce saturation and hence
cloud. In more humid
conditions the cloud may
envelope all upper surfaces
of the airframe. Assuming an
air speed of 150m s -I and a
cloud formation fetch of
< 1 m, cloud is forming in
<7 ms, much faster than in
any cloudy weather system.

radius outruns the ability of vapour to diffuse sufficiently rapidly down the
gradient of vapour density.
When a growing droplet exceeds its crucial radius, the condensation nucleus is
said to be activated- a term which recognizes both the crucial change in behaviour
as it grows from a state of neutral stability to one of almost explosively unstable
growth, and the significance of this change for the conversion of embryonic
droplets to fully grown cloud droplets. Note that if the ambient humidity falls .
below the value associated with the critical radius the droplet deactivates by rapid
shrinkage to an equilibrium radius below the critical radius, shooting leftward over
the maximum on the Kohler curve (Fig. 6. 13).
A non-hygroscopic condensation nucleus has no neutrally stable growth regime
corresponding to a hygroscopically nucleated droplet below its critical size. Instead
it remains essentially dry until the ambient supersaturation exceeds the value cor-
responding to its radius of curvature. Vapour then begins to condense and the
droplet grows rapidly, like an activated hygroscopic nucleus.

6.9 Cloud development

Previous sections have outlined mechanisms at work on individual droplets during

formation and subsequent growth. Let us bring these together to describe the
production of a small cloud of droplets just above a well-mixed cloud-free layer. In
the first instance we can suppose that this represents the formation of fair-weather
cumulus (Fig. 6.3), but the same mechanism applies to air rising into the base of an
established cumulus congestus or cumulonimbus. Indeed it applies to all
atmospheric cloud formation, from very slow, static cooling in fog, to rapid
smooth uplift in lenticular cloud (Fig. 10.30), and from slow persistent ascent in
154 fronts, to rapid horizontal decompression in tornadoes (though size and time-
scales, and the degree of isolation from other clouds, can enforce very different 6. 9 Cloud development
subsequent development).
We begin with air rising toward the top of the sub-cloud layer, its temperature
falling and its relative humidity rising because of dry adiabatic expansion. The
population of condensation nuclei in the humidifying air is clearly very important,
both in terms of the total numbers and their distribution across the available
spectra of sizes and materials. Numbers and sizes in particular vary very widely
with position and time, but as a rough guide note that one litre of air from the low
troposphere at an inland site may contain 107 Aitken nuclei (radii between 5 x 10- 3
and 0.1 ~-tm), 1OS large nuclei (radii between 0.1 and 1.0 ~-tm) and 103 giant nuclei
(radii exceeding 1.0 ~-tm). No nuclei smaller than the largest Aitkens become
activated in natural clouds, because as we shall see maximum supersaturations are
held sufficiently close to zero by the simultaneous growth of droplets on all larger
nuclei. The numbers of the 'large' category can therefore be used as an order of
magnitude for the total numbers of nuclei available for activation in the rising air.
Over land these are believed to consist mainly of sulphates and sulphuric acid
(derived from the combustion product S02), fine soil particles and sea salt.
Obviously numbers are larger closer to localized sources such as active volcanoes
and natural and man-made fires. Over the oceans numbers may be down by a
factor of ten or more and the proportion of sea salt much larger.
When the relative humidity of the rising air exceeds about 800Jo many hygro-
scopic nuclei begin to swell by condensation, and their growth as RH exceeds 90%
may be sufficient to reduce visibility quite noticeably by a wet haze, especially when
the line of sight is nearly horizontal. Observers on mountains or aircraft often see a
marked haziness extending tens or even hundreds of metres below cumulus cloud
base. In heavily industrial regions this elevated haziness may be visible even from
the oblique viewpoints of ground-level observers. From Fig. 6.13 and equivalent
data it transpires that the radii of haze droplets forming on all but giant nuclei are
less than 1 ~-tm. The relatively weak interactions of such small droplets with visible
light make them effectively invisible in bulk in small volumes such as cloud
chambers.
While the rising air remains unsaturated, all haze droplets remain in equilibrium
with the ambient RH, swelling as it rises, but not yet activated. But almost as soon
as the air begins to supersaturate, some droplets exceed their critical radius,
become activated and so begin to grow very rapidly. At first only the most favoured
are activated - those forming on giant hygroscopic nuclei - but as the super-
saturation of the rising air increases, increasing numbers of less favoured droplets
are activated, as are some of the largest non-hygroscopic nuclei. In a relatively
short space of time so. many droplets are growing so rapidly that they begin to
deplete the ambient vapour quite substantially; the growth of supersaturation
checks and then reverses. Once the supersaturation starts to fall no further droplets
are activated, and those still struggling towards activation begin to shrink again.
However, the supersaturation close to the surfaces of the activated droplets is so
nearly zero (imagine the Kohler curves extended far to the right of the critical
radius) that they continue their rapid growth for as long as the ambient air remains
supersaturated, which means in effect for as long as the rising air remains within
the body of the cloud.
The way in which a population of droplets becomes activated has three impor-
tant consequences.

1. The transition from wet haze to cloud tends to be quite rapid, since most of the
droplets are activated at supersaturations of below 1%. The activated droplets
grow to micron scale in a few seconds, at which size they become very much 155
6 Cloud and precipitation more effective at scattering sunlight than they were as submicron-scale haze
droplets. This explains the dramatic visual transformation apparent in the
atmosphere and in cloud chambers. Together with the well-mixed nature of the
sub-cloud layer this also explains the familiar sharpness and uniformity of
cumulus cloud-base layer (Fig. 6.3).
2. The number of cloud droplets formed is equal to the number of droplets and
particles activated, which in turn is determined by the aerosol population and
the maximum supersaturation reached. Although cloud droplets may
coagulate and divide subsequently, to a large extent the cloud population is
established at least in respect of its numbers by the conditions in which it is first
formed. In the case of cumulus these are the conditions prevailing in the first
few tens of metres above cloud base.
3. With the numbers of cloud droplets firmly established, their ultimate size is
determined by the amount of vapour available for condensation. For example,
consider the case of a litre of air which contains 105 activated condensation
nuclei and 10 mg of vapour (corresponding to a specific humidity of 10 g per kg
- quite normal for the low troposphere). If half of the available vapour is
condensed, as might well happen if the air ascends through an appreciable
fraction of the troposphere, you can easily check that each cloud droplet must
grow to a radius of about 30 J.tm. In fact growth to such a size by diffusion and
condensation is relatively slow- it is rapid only in the initial stages, as droplets
activate and race to radii of a few microns - and is usually short-circuited by
mechanisms we must consider later to explain the observed ease and speed with
which clouds precipitate.

Deactivation of cloudy air occurs most dramatically at the edges of cumulus

clouds of all sizes. The cauliflower-like upper parts are in a state of continual and
rapid change as parcels of cloudy air erupt from the interior and are turbulently
mixed with the clear, subsaturated ambient air. As the cloudy parcels become sub-
saturated the cloud droplets shrink very rapidly below their critical radii. In fact the

Fig. 6.15 A decaying

cumulus, dull compared with
the developing cloud behind.

156
dryness of the ambient air usually takes them so far below their critical size that 6.10 Precipitation
there is little or no wet haze. On the sunlit side of the cumulus the sharp contrast
between the clear ambient air and the strong backscattering of sunlight by the cloud
gives rise to its familiar brilliant appearance (Fig. 10.9), and on the shaded side of
the strong forward scattering from the edge zone forms the 'silver lining' round the
dark body of the cloud (dark because so much light has been backscattered at the
far side) (Fig. 1.4). Often deactivation occurs more gradually throughout a larger
volume when an updraught weakens and dies. As the supersaturation gradually
falls to zero by mixing with somewhat drier air the population of cloud droplets is
thinned by selective deactivation, and the cloud loses its brilliant, sharp-edged
appearance in favour of a diffuse dullness which finally decays to nothing through
a transient haziness (Fig. 6.15).

6.10 Precipitation

So far we have concentrated on the mechanisms at work maintaining the cloudy

content of the troposphere. In the long run these are precisely balanced by the pre-
cipitation processes which return the condensed water and ice to the surface.
Evaporation of cloud and precipitation in the air is widespread but merely repre-
sents a redistribution of the water substance within the atmosphere.
Over most of the Earth it is a matter of common observation that water falls to
the surface in a variety of forms summarized in Table 6.1, and selectively depicted
in Fig. 6.16. Water droplets reach the surface with diameters ranging from about
100 ,urn to about 3 mm. The smallest of these are felt rather than seen in fog.
Normal drizzle (diameters about 200 ,urn) reaches the surface from low cloud such
as stratus, which can be considered to be a fog layer lifted clear of the surface by
stirring and slight desaturation of the surface layers. Raindrops cover the rest of
the size range and account for the great bulk of precipitation reaching the surface in
all but the coldest conditions. Fine rain, verging on drizzle, is observed only when

Table 6.1 Properties of selected precipitation droplets and particles

Diameter Fall speed

Name (mm) Shape (m s- 1)

Water
Drizzle 0.2 Sphere 0.8
Rain 0.5 Sphere 4.0
Rain 5.0 Unstable cap shape 10.0
Ice
Snow crystals 0.2 {Prism, plate,} 0.3
5.0 {Star, needle} 0.7
Snowflakes 1.0 Irregular aggregates 0.5
of from 2 to lOOs
20.0 of crystals (often 1.0
stars)
Groupe! 0.5 Conical 0.5
(soft hail) 5.0 2.5
Hail 3.0 Roughly spherical 8.0
Giant hail 20.0 Spherical with knobs 20.0
157
6 Cloud and precipitation Size <pml 100 1000
(em) 0.1 10

Droplets drizzle rain

0 0 (''--/
-'") ~ shattered
:E by airflow

Ice crystals snow crystals snowflakes

Q thin plate
::?~
~ by airflow

*star
irregular conglomerations
of snow crystals
~ column

Ice pellets (soft) graupel

0
Q
Ice pellets (hard) ice pellets

0
Fig. 6.16 (a) A selection of
precipitation particles and
~:~:~0~~~---------------•?
drops. Note the hexagonal @ layered@
shape of all the ice crystals,
reflecting the basic lattice
structure of ice. Snow crystals lobed{j
grow by developing dendrites
on the six corners, and
snowflakes by the amorphous Size (JLm) 100 1000
aggregation of crystals.
(em) 0.1 10

cloud bases are low, whereas larger raindrops (of the order of millimetres) require
several kilometres depth of cloud above cloud base - such as occurs in many
frontal and shower clouds. The largest raindrops fall from the heavy shower clouds
which are endemic in equatorial regions and are common in mid-latitudes in
summer.
Ice is precipitated in an even greater range of sizes and in literally innumerable
shapes (Fig. 6.16). Snow crystals of much the same size as water droplets fall to the
surface from stratiform clouds when the low level air is more than a few degrees
below 0 oc and the air aloft is colder still, as it normally is. (If unusually a cold low
layer is overlain by air above 0 °C, freezing rain may result: rain drops, super-
cooled by falling through the cold layer, freeze on impact with the surface.) When
the low-level air is close to the freezing point, the snow crystals aggregate by
collision to form snow flakes which in calm conditions may be several centimetres
across, though only millimetres thick.
Snow crystals and flakes are also observed to fall from shower clouds when the
low-level air is cold enough and the air aloft is colder. However, the central core of
the showers generally contains some hail. In mid-latitudes in winter the hail
158 particles are often millimetre-scale cones of white, low-density ice, known as
graupel or soft hail, the whiteness and low density arising because air is trapped 6.10 Precipitation
between the constituent ice grains of the particles. Over land in summer in mid-
latitudes larger hail may fall, often including layers of clear high-density ice alter-
nating concentrically with layers of white ice to produce an onion-like appearance.
There seems to be no clear upper limit to the sizes of such hailstones, and a few
comparable in size with cricket or baseballs fall from the most vigorous showers,
often having a knobbly exterior like that in Fig. 6.16.
The minimum sizes of precipitation droplets and particles are quite readily
explained. As soon as one falls out of its parent cloud into subsaturated air it begins
to evaporate, and it can be shown (appendix 6.4) that the sensitivity of both
evaporation rate and fall speed to droplet size combine to make the distance fallen
before total evaporation increase very rapidly with initial droplet size. In fact the
simplified model in appendix 6.4 suggests that the distance fallen increases with the
fourth power of the radius, and that a droplet must have an initial radius of at least
100 t-tm to survive a fall of a few hundred metres in air subsaturated by only a few
per cent. Cloud droplets, being ten times smaller in radius, survive for only a few
centimetres, which is consistent with the sharpness of cloud outline already
mentioned, whereas raindrops can fall tens of kilometres.
It follows that drizzle-sized droplets and crystals are the smallest which can
survive falling through even quite shallow sub-cloud layers. When cloud bases are
higher, the drizzle droplets evaporate on the way down with the result thatdroplets
reaching the surface are usually at least millimetric in size, corresponding to
normal sizes for rain and snow. Occasionally solar backlighting of precipitation
from a solitary cloud shows up the progressive evaporation of the smaller droplets
or flakes with increasing distance below cloud base, the tapering veil being called
virga (Fig. 6.17).
From the quoted sizes of precipitation droplets and particles, it is clear that all
are very much larger than typical cloud droplets. In terms of radius, drizzle and
raindrops are ten and one hundred times larger respectively, and in terms of
volume these ratios become 103 and 106. The latter values draw attention to the

Fig. 6.17 Virga trailing from

a small cumulus congestus.
possibly enhanced on this
winter occasion by the
melting of snow behind cloud
base.

_,
. -~..,

'.,.
159
6 Cloud and precipitation enormous growth which has to be explained by any mechanism for precipitation:
for example an ordinary raindrop has about one million times the water content of
an ordinary cloud droplet.

6.11 Precipitation modelled

In the previous sections we have followed the development of cloud droplets from
their hazy origins to the phase of rapid growth by condensation which follows
activation. Once the complex Kohler curves around the critical sizes are negotiated,
the growth of a droplet by condensation is controlled by the supersaturation of the
ambient vapour (the equilibrium vapour density at the droplet surface being almost
exactly equal to the saturation value for a plane, pure surface). As outlined in
Appendix 6.2 droplet growth by diffusion and condensation in the presence of a
constant supersaturation is inversely proportional to droplet radius. Although the
model ignores the complications introduced by the warming of the growing droplet
by release of latent heat, the predicted lengths of time for growth are known from
more elaborate models to be quite realistic [41]. They show that whereas a droplet
may grow to a radius of 5 J.Lm in about 20 s in the presence of 0.20Jo super-
saturation, it would take another 20 minutes to grow to a radius of 30 J.Lm and
several days to reach the size of a small raindrop. But simple observation shows
that a shower of raindrops can be produced by a cloud only half an hour after the
cloud's first appearance, and that more generally almost any cloud will precipitate
if it lasts more than a few tens of minutes and is at least a few hundred metres from
base to top. Clearly this familiar and vitally important behaviour cannot be
explained by diffusion and condensation alone.
Note that it is quite impossible for all the activated droplets in a typical cloud to
grow to the size of raindrops: this would require a million times more vapour than
is available in the cloud-forming air and it would produce an utterly unrealistically
catastrophic deluge. It seems that precipitation depends on the favoured growth of
a small fraction of the total population of cloud droplets. In looking for precipita-
tion mechanisms we are therefore looking for mechanisms allowing rapid, selective
growth.
Let us begin by noting that a cloud must contain a spectrum of droplet sizes.
During initial formation giant hygroscopic nuclei will have been activated some-
what earlier than less favoured nuclei, so that there will be a size spectrum in the
population of rapidly growing droplets. Now the terminal velocity (the equilibrium
fall speed at which weight is balanced by drag) of droplets in this size range
increases with the square of the droplet radius, as outlined in appendix 6.3 and
depicted in Fig. 6.10. It follows that in the developing cloud the larger droplets
must be falling a good deal faster than the smaller ones, up to ten times as fast in
fact. This must produce collisions, which in turn may lead to the growth of the
larger, faster-falling droplets by coalescence with smaller droplets. The process is
entitled growth by collision and coalescence, to distinguish it from the growth by
diffusion and condensation considered earlier. A very important property of
growth by collision and coalescence emerges if we consider the volume of air swept
out by a falling collector drop of radius rand terminal velocity V (Fig. 6.18). The
volume swept per second is proportional to r 2 V, which in turn is proportional to r4 ,
160
given the dependence of Von r 2 already noted. Assuming that the mass of smaller 6.11 Precipitation modelled
droplets collected is proportional to the volume swept, then it follows that the rate
of growth of the radius of the collecting drop drldt increases as r 2 (Appendix 6.5).
Such accelerating growth contrasts with the decelerating growth produced by
diffusion and condensation.
Observations in the atmosphere and very special cloud chambers, together with
computer simulation of the growth of a model cloud population, seem to confirm
that collision and coalescence is at work and effective in atmospheric clouds.
However, the mechanism is a good deal more complex than has been suggested
so far. For example, very small drops tend to flow round the would-be collector
drop in the disturbed air flow, and so avoid collision. Even if they do collide,
not all colliding droplets coalesce with the collector- significant numbers bounce Swept
off in some situations. There are in fact efficiencies of both collision and coale- volume

scence which are observed to vary with the sizes of both the collecting and col-
lected droplets in quite complex ways. Observation requires sophisticated and 0
painstaking laboratory work, but its results and theory agree in indicating that
collector droplets with radii less than 20 ~-tm are inefficient, and that droplets more
than 800Jo and less than 20% of the radius of the collector are not efficiently Fig. 6.18 A large cloud or
precipitation droplet falling
collected.
through smaller neighbours
From all this it seems that precipitation develops in two stages. First, diffusion and tending to collect those
and condensation produces a population of cloud droplets which includes some within the swept path
with radii exceeding 20 ~-tm - though it is still a little difficult to envisage how the (enveloped dashed). Dotted
largest of these are produced in realistically short times. Subsequently these large lines are paths of small
bodies relative to the falling
droplets grow by collision and coalescence with their smaller neighbours, at a rate collector.
which increases with size, to produce millimetric-sized drops in a few tens of
minutes. This two-stage process is believed to explain precipitation from all
relatively warm clouds, i.e. those which do not contain temperatures below about
-10 °C.
In fact much of the troposphere is colder than -10 oc, some of it very much
colder. Since clouds are at best only marginally warmer than the ambient air, much
cloud must be similarly cold. In low latitudes this will apply to the upper parts of
tall shower clouds, but in middle and high latitudes it may apply to all but the bases
of clouds, or even to the entire clouds, depending on season and geography. We
must consider whether the prevalence of cold cloud has any bearing on the develop-
ment of cloud and especially precipitation.
Everyday experience with water suggests that it freezes as soon as the water
temperature falls to 0 oc, or a little below this if the water is contaminated by salt
or some other impurity. However, it is observed that large amounts of atmospheric
cloud persist in the form of supercooled droplets down to -30 oc or so. It seems
that as in the case of equilibrium between vapour and liquid or ice, tiny volumes of
water behave very differently from much larger volumes.
Much careful work in the laboratory and atmosphere shows that water freezes
by a process of nucleation - the crystal lattice of ice spreading rapidly through
the water volume from a very small nucleus. At about -40 oc the thermal
movement of molecules in liquid water is so sluggish that clumps of them settle
into the lattice formation by chance and hence freeze the entire volume. This
is known as homogeneous nucleation, and the identity of name reflects the
strong similarity with homogeneous nucleation of cloud droplets in a highly
supersaturated vapour. At much higher temperatures homogeneous nuclea-
tion is so rare in the small volumes of typical cloud droplets that their freezing
requires the intervention of a tiny quantity of a different substance which
initiates freezing because its lattice structure is similar to that of ice. These
161
6 Cloud and precipitation are known as freezing nuclei and their action is termed heterogeneous
nucleation.
A freezing nucleus may be present from the outset within a supercooled droplet,
or it may make contact with its outer surface, but in either case they become
increasingly likely to be effective (i.e. initiate freezing) in any observation period as
the droplet cools, until they are swamped by homogeneous nucleation. A slight
practical complication is that most experimental work on ice clouds concentrates
on the resultant ice crystals, even though some of these may be produced by the
action of sublimation nuclei triggering the formation of an ice crystal directly from
the vapour. Sublimation nuclei are probably much less important than freezing
nuclei, but to avoid being too specific we will refer to all nuclei encouraging the
production of ice crystals in clouds as ice nuclei.
It is conventional to ascribe the increasing likelihood of heterogeneous
nucleation with decreasing temperature to an increasing concentration of ice
nuclei, even though it is their effectiveness which is increasing- their numbers pre-
sumably being fixed in the absence of a local source. Although there are still discre-
pancies between observations it is certain that concentrations of ice nuclei are very
much smaller than those of condensation nuclei, being about one per litre at
-10 °C.
Ice nucleation processes ensure that the ice crystal population in clouds increases
with decreasing temperature. Above -10 oc there is almost no ice; between
-10 oc and -20 °C there is an increasingly significant minority of ice crystals
whichmayconverttoamajoritybetween -20 °Cand -30 °C;andbelow -30 oc
most clouds are predominantly ice. No supercooled water can persist below
-40 °C. It follows that the concentration of ice crystals in clouds increases
upwards, that clouds in the middle troposphere in middle latitudes contain both ice
and supercooled water (i.e. they are mixed clouds), and that ice clouds pre-
dominate in the high troposphere (Fig. 6.19). This last statement corresponds to
simple observation: ice cloud has a characteristically fibrous appearance (hence the
term cirrus, being Latin for hair) which arises because of its slowness to evaporate
in the presence of unsaturated air and which can be seen frequently in the high
troposphere (Fig. 2. 7). The tops of vigorously growing shower clouds maintain
their sharp-edged cauliflower appearance until they reach the upper troposphere,
but then their edges can be seen to soften and diffuse into a fibrous anvil as the
cloud of supercooled water glaciates quickly and extensively in the very low
temperatures there. The Norwegian Bergeron, observing the development of
cumulus in clear westerly air flows reaching Norway from the Atlantic, noted in the
1930s that the clouds began to precipitate very shortly after their tops began to
glaciate, and he proposed the following connecting mechanism.
It is apparent in Fig. 6.2 that air which is saturated with respect to water is
considerably supersaturated with respect to ice, at least when the temperature is
significantly below 0 °C. In fact at -15 oc the supersaturation is over 600Jo. If an
ice crystal is produced in a cloud which consists mainly of supercooled water, it
follows that the vapour will be grossly supersaturated relative to the ice crystal,
which will therefore grow very much more rapidly by diffusion and sublimation
than will its supercooled neighbours by diffusion and condensation- hundreds of
times faster according to the ratio of the relevant supersaturations. Indeed as more
ice crystals grow more and more rapidly, they may reduce the ambient vapour
below saturation relative to super-cooled water and so cause wholesale transfer of
water substance from droplets to crystals. This mechanism is therefore powerfully
selective and has been named the Bergeron-Findeisen mechanism after Bergeron
and the German Findeisen who pioneered the use of very large cloud chambers to
162 study this and other cloudy processes. It is believed to be important in the mixed
 6.11 Precipitation modelled

Fig. 6.19 Shower clouds

over the Irish Seo in winter.
Since the upper ports show
the fibrous appearance
characteristic of ice cloud,
whereas the low ports are
pure water (tempel"otures -
0 °(), the middle ports ore
mixed ice and water cloud .

clouds present in cumulonimbus and nimbostratus, and therefore to contribute to

precipitation from all cold or partly cold clouds.
Crystals growing by diffusion and sublimation preserve the hexagonal shape of
the ice crystal lattice, but in different conditions produce different variations on
the hexagonal theme (Fig. 6.16). Fall speeds are often considerably less than those
of droplets of the same mass, with the result that it is at least possible that some fine
crystals falling out of fairly shallow but cold clouds may have had time to grow by
this mechanism alone. This could then explain some drizzle from cold strato-
cumulus, where the crystals fall into air warmer than 0 °C before reaching the
surface.
It is much more likely, however, that the Bergeron-Findeisen mechanism
operates in conjunction with the equivalent of the collision and coalescence
mechanism, though the latter is complicated in detail by the nature of the collector
crystal. An ice crystal falling through smaller supercooled water droplets grows by
accretion, the droplets freezing on impact with what amounts to a giant freezing
nucleus to produce a rapidly growing hailstone. If the droplets are only slightly
supercooled they may form a layer of water which freezes subsequently to form
clear, dense ice. If the droplets are heavily supercooled they freeze instantaneously
and trap the intervening air to form white, spongy, low-density graupel. Hail
containing alternate concentric rings of white and clear ice has probably oscillated
between higher and lower temperature before finally falling out of the parent
cloud: air flow in very vigorous showers is known to be capable of allowing a
substantial vertical oscillation of the growing hail (section 10.6).
If a large ice crystal is falling through smaller ice crystals it tends to grow by
aggregation, the smaller crystals colliding with and sticking to the collector crystal.
Not all collisions lead to aggregation however: it is observed that significant
aggregation occurs only when temperatures are above -10 °C, presumably
because some of the ice surfaces are coated with a slightly sticky liquid film. Aggre-
gation of snow crystals (where diffusion and sublimation has produced pro-
nounced hexagonal spikes - Fig. 6.16) is enhanced by interlocking of adjacent
spikes or dendrites, producing very large snowflakes in calm conditions.
Lastly, note that because so much of the troposphere is so cold, even in
equatorial regions, a great deal of precipitation begins in the form of ice crystals
and develops into snow crystals and flakes. However, if these fall into air which is
warmer than about 2 oc they melt and arrive at the surface as rain. In middle 163
6 Cloud and precipitation

Fig. 6.20 A rainbow

segment beneath melted
snow. A heavy shower of
snow is melting to give rain in
the lowest few hundreds of
metres. The sun behind the
camera produces a rainbow
by internal reflection in the
nearly spherical drops, but
produces unstructured
backscatter from the snow.

latitudes almost all rain is melted snow, even in the height of summer (Fig. 6.20).
Only the larger hailstones from summer thunder showers reach the surface largely
unmelted because of their much higher fall speeds. In winter the melting level for
snow often fluctuates between sea level and a kilometre above, as the synoptic scale
weather pattern changes, giving the variable snow cover on hills and mountains
and the occasional blanketing of lowlands so typical of maritime climates such as
Washington State's and Britain's.

6.12 Atmospheric electricity

Showers are often associated with thunder and lightning, and very heavy showers
always are. Indeed the very heaviest showers may contain nearly continuous
thunder and lightning which, together with their torrential rain, large hail and
damaging gusts of wind, earns them the technical title severe local storms. Thunder
and lightning also occur in some fronts, especially vigorous cold fronts, and in
rather special weather systems (called squall lines in the USA) intermediate in
character between fronts and lines of severe storms. All tropical cyclones produce
vigorous thunder and lightning. It seems that at any instant there are several
thousand thunderstorms active in the troposphere, concentrated in low latitudes
and over daylit land.
The loudness of thunder and the vividness of lightning, often accentuated by the
brooding darkness of the clouds in which they occur, have impressed man from his
earliest times and teased his powers of explanation. Gods like Zeus and Thor have
been invoked, and the latter has left his mark in the name thunder and in the use of
the term anvil to describe the shape of the glaciated upper parts of a mature
cumulonimbus - thunder being presumably the sound of Thor beating on his
gigantic anvil. Lucretius in 55 BC supposed that thunder was the noise of great
clouds crashing together, which at least recognizes the fact that large clouds are
164 always involved. In AD 1752 Benjamin Franklin and d' Ali bard proved inde-
 Approximate
6.12 Atmospheric electricity
Height
(km)

Fig. 6.21 Typical distribution

of electrical charge within on
active thunderstorm in middle
latitudes, as deduced by
remote observation of
changes in electrical field
strength accompanying
lightning strokes. Quantities
of separated charge ore given
in coulombs . (See [8] and
(41])

pendently that at least part of the activity of thunderstorms was electrical in nature,
by the potentially lethal procedure of flying kites nearby on conducting tethers.
Franklin in particular was able to determine that the electrical charges involved
were sometimes positive and sometimes negative. In the 1920s C . T. R. Wilson
established the typical distribution of charge within a thunderstorm (Fig. 6.21) by
analysing measurements of electrical field strength made on the ground at a range
of distances from the storm. In the next decade this work was confirmed and
amplified by G. Simpson and others who flew field strength meters into thunder-
storms on balloons. Meanwhile very fast-response cameras had been used to
examine the complex structure of lightning - the giant sparks jumping within
clouds, and from cloud to ground.
The typical charge distribution shown in Fig. 6.21 implies that positive and
negative charges are being separated by the thundercloud. Lightning within the
cloud temporarily neutralizes the main separations of charge by shorting them
together by a conducting channel, but the observations show that reseparation
requires only 20 s or so. In recent years much painstaking work in the laboratory
and by instrumented aircraft in and near thunderstorms has been directed at
finding out what th!;! charge-separating mechanisms are. Results so far are incon-
clusive but suggest that several distinct mechanisms may be important. Their
common feature is that small particles or droplets tend to become positively
charged while larger particles become negatively charged. The greater fall speeds of
the larger particles (many of them precipitation elements) enable gravity to drive
the required charge separation. For example, according to one proposed
mechanism, a small cold ice crystal collides with a hailstone warmed considerably
by release of latent heat from the freezing of impacting supercooled droplets.
During their brief contact the very mobile positive ions migrate toward the colder
end of the crystal (i.e. away from the hailstone) leaving the warmer end and the
hailstone negatively charged (Fig. 6.23). On separation the stone carries its
negative charge downwards and the crystal its positive charge upwards. The charge
separation at each collision is very small, but there are so many collisions in the
cubic kilometres of active precipitating cloud that the net result is at least poten-
tially consistent with the observed charging rate.
Lightning is a giant spark produced when the electrical insulation of the air,
cloud and precipitation mixture breaks down. This is believed to occur when the
electric field strength reaches about 1 MV m- 1• From basic electrostatics the field
strength at the edge of a 2 km radius sphere containing 24 C of electrical charge is 165
6 Cloud and precipitation

Fl1. 6.22 Lightning in and

around a large cumulonimbus
at night. Several unrelated
strokes have occurred during
the opening of the camera
aperture. Those within the
cloud are seen indistinctly,
while the directly visible
external strokes are seen to
have the familiar jagged
forked structure.

only about one tenth of this critical value, and it is supposed that the critical value is
exceeded because of some local concentration of charge. Once the spark has begun
it is able to propagate quickly across regions where the electric field is considerably
weaker. For obvious reasons cloud-to-ground lightning has been studied much
more than lightning within cloud, and it seems likely that many of these ground
strikes begin in the strong field between the small positive charge just below the
melting level and the large negative charge higher up. Once begun a so-called
stepped leader makes its way from the negative charge toward the surface, its
jagged steps outlining the familiar forked lightning shape. As it nears the surface
the electric fields there, which may already be high enough to support a flickering
corona discharge around sharp points such as the twigs of trees, cause a spark to
jump to meet the stepped leader. The main surge of free electrons toward the earth

O
then occurs in the established ionized channel, but in such a way that the surge pro-
pagates upward in a return stroke which is the main part of the lightning flash. The
current surge in a channel of a few centimetres across may be about 10 000 A, and
t I
the resultant heating of the air produces the incandescent flash and the explosive
I expansion whose shock waves are heard as thunder. As the negatively charged
Warm :
hail J region of the cloud may be only partly discharged by this process, there are often
surface '
I several return strokes at intervals of about one twentieth of a second, each

\::::
preceded by a downward dart leader reactivation of the ionized channel. The
succession of several brilliant return strokes is just perceptible to the eye as the
c r ys t al flicker of a clearly visible lightning flash. Discharge being now complete, the
recharging processes get under way again, preparing for another flash a few tens of
seconds later, which occurs in a fresh location, since the ionized channel will have
Fl1. 6.23 Charge separation
during collision between an long since dispersed. Storms producing much more frequent flashes to ground
ice crystal and a warmer must have several separate charging zones at work simultaneously.
hailstone. In everyday speech it is customary to distinguish between fork lightning and
sheet lightning. Sheet lightning may be the loom of a distant fork lightning when
seen through an intervening cloud, or it may be a lightning flash of the main type-
166 between the positive and negative charges accumulating within the cloud. The
latter appears to differ somewhat from the cloud-to-ground discharges discussed 6.13 Practical applications
already, for example being longer lasting and duller, although their location makes
detailed comparison difficult.
Thunder is the noise of the shock waves of sound rippling out from the
incandescent lightning flash. Since this spreads at the speed of sound compared
with the effectively instantaneous transmission of the visible flash, a distant flash
produces audible thunder after a time delay which is directly proportional to the
distance, the delay being about 3 seconds per kilometre. Thunder is seldom audible
more than about 10 km away, but lightning is clearly visible for many tens of kilo-
metres at night if there are no intervening clouds. Such silent flashes, indistinct
because of great range, are sometimes called sheet lightning, and sometimes heat-
lightning, because they are often seen on hot summer nights in middle latitudes.
The prolonged booming peals of thunder are rarely caused, as popularly
supposed, by echoes from clouds or mountains. Lightning flashes may be several
kilometres in length so that a ground strike near the observer may rumble for 10 s
or more as the noise from the more distant parts of the flash follows after the
hissing, tearing noise of the nearest part of the flash. And the forked path may
produce booming concentrations in the long rumble.
Thunderstorms play an important role in the atmospheric electrical cycle. The
cloud-to-surface flashes are continually pumping electrons earthward at a rate of
several thousands of amperes world-wide, and maintaining the high troposphere
and higher atmosphere at an electrical potential of about half a million volts above
earth potential. (Individual storms produce potentials ten or more times larger
than this in their upper parts, but much of this is wasted by cloud-to-cloud lightn-
ing and other leakages.) This maintains the fair- weather electric field as depicted in
Fig. 6.24, the field strength being a maximum near the surface where it is about
100 V m -t. In this field the small concentration of ions produced by cosmic
radiation and natural radioactivity in the surface and atmosphere maintains a
gentle, persistent current which world-wide balances the concentrated spasmodic
currents driven by thunderstorms.
The dissipations of energy in lightning flashes are very spectacular because they
are brief and highly localized. When averaged throughout the volume and lifetime
of a thunderstorm they represent only a small fraction of the storm's total energy
budget, the largest components of which are the buoyant potential energy of the air
which is about to rise and fall, and the kinetic energy of the rising and falling air
(Chapter 10). In effect thunder and lightning are spectacular sideshows in the wings
of the main drama, in which millions of tons of air, water and ice are in great
commotion. The noises which Lucretius thought were cloudy collisions are more

30

Fig. 6.24 Vertical profile of

electrical potential in fair
weather. The profile gradient
is the strength of the vertical
component of the electrical
field.
0&..::=--------
0 300
Electrical potential (kV) 167
6 Cloud and precipitation properly the relatively slight sounds of clouds tearing themselves apart under
gravity and buoyancy.

6.13 Practical applications

CLOUD SEEDING

Speculation about the importance of the Bergeron-Findeisen process in initiating

precipitation led to extensive work in artificial stimulation of precipitation
beginning in the 1940s and continuing to the present day. The ability of ice nuclei to
trigger precipitation from suitable clouds and the belief that such nuclei might be in
short supply in some conditions led people to believe that the artificial introduction
of ice nuclei might enhance the production of precipitation. This is clearly poten-
tially valuable to agricultural production in marginally arid areas. Initially some
early spectacular claims were made for the effectiveness of such so-called cloud
seeding, but although it can occasionally change the appearance of clouds quite
dramatically, years of exhaustive investigation have shown that the effects are
quite slight, probably not increasing precipitation more than 200Jo above the
natural level. Because precipitation is naturally so variable, the detection of such
small increases requires prolonged experimental study and careful statistical
analysis. Nevertheless such increases could be economically valuable in some
circumstances, and work has continued in some countries for several decades now,
though sometimes sustained more by hope than objective results. Silver iodide has
been used quite widely in this work, because laboratory studies have shown it to be
one of the most efficient freezing nuclei, beginning to be effective at -4 °C. Its
crystal structure is very similar to that of ice. Myriads of tiny crystals of Agi are
produced from small burners which can be flown through clouds or sited on the
ground to feed updraughts rising into clouds aloft. Note that seeding must be
carefully controlled. If a cloud is swamped by freezing nuclei the effect may well be
to produce larger numbers of smaller ice crystals than would have been produced
without seeding, which is the opposite of the selective mechanisms which underlie
all precipitation.

AIRCRAFT ICING

As soon as aircraft began to fly in supercooled water clouds, the icing-up of

exposed surfaces, especially the leading edges of wings, began to be a serious and
potentially fatal hazard. Surfaces at ambient sub-freezing temperatures act like
giant hailstones and collect masses of ice and freezing rain by the impact-freezing
of supercooled cloud droplets (Fig. 6.25). In the worst conditions these can accu-
mulate so quickly that their weight becomes a serious problem and their distortion
of the aerofoil shape seriously reduces the lifting capacity of the wing. Many air-
craft have been lost in this way and others have survived only by the last-minute
melting and loosening of the burden of ice as they fell out of control into the warm
low troposphere. Although flying lower or higher (where the collection efficiency
of the more numerous ice crystals is so low as to be no problem) does reduce the
problem, almost all modern aircraft have systems for dealing with icing, involving
168 heating or even flexing the leading edges of wings and other crucial parts.
 6.13 Practical applications

-
,

Fig. 6.25 Underside port

) wing of Concorde showing ice

(pole on the deliberately
blackened wing)
accumulating during f light
trials to determine heating
rates needed to prevent icing
in service.

RADAR

Radars emitting wavelengths of about 10 em receive strong echoes from precipita-

tion particles and droplets, which can be used to measure precipitation over wide
areas as mentioned in section 2.6. Radar has also been used to investigate the
invisible and fairly inaccessible interiors of cumulonimbus (shower clouds),
nimbostratus (precipitating frontal clouds), and the active inner zones of severe
tropical cyclones (Fig. 12. 12). Indeed some of the major features of cumulo-
nimbus outlined in sections 10.4 and 10.6 have been clarified very considerably
with the help of radar, including types which scan vertically to provide a vertical
section of echo (Fig. 6.26) rather than the more common plan view. Many large 169
6 Cloud and precipitation

Fig. 6.26 Radar echo from a

large cumulonimbus, showing
vertical shafts of heavy
precipitation. The vertically
scanning radar has been 73
programmed to produce a
37
display of echo strength Range (km)
(brightness) against range
and height (RHI display).

civil aircraft now carry small radars which enable them to detect and avoid regions
of heavy precipitation, mainly because they can be unpleasantly and even
dangerously turbulent because of the associated strong updraughts and down-
draughts. Nimbostratus often maintains a horizontal band of strongly enhanced
echo where snowflakes become coated with water as they melt, reflecting radar
waves much more efficiently than either the dry snow and ice above or the rain
beneath. Such melting bands (Fig. 6.27) confirm that most surface rain in
middle and high latitudes is formed by the melting of snow falling from higher
levels.

LIGHTNING CONDUCTORS

In what has become a classic example of the connection between pure research and
useful application, Benjamin Franklin followed his confirmation of the electrical
nature of lightning by proposing that buildings could be protected from lightning
damage by connecting a stout electrical conductor from just above the highest part
to a grounded metal stake. This he supposed would offer the lightning an easy
route to earth - so easy that little damaging heat would be generated en route.
(Buildings and trees are damaged by the heat generated as the current surges
through their substantial natural resistance, explosive boiling of interior moisture
or sap being particularly destructive.) His suppositions were amply confirmed by
170 those who followed his advice, and damage to church spires and other very exposed
 6.13 Practical applications

, Fig. 6.27 Radar echo

50•e strength from a section of
10 frontal cloud over southern
40 ~ England, showing the

30 ~ " horizontal melting or 'bright'

bond of enhanced echo at
altitude 3.2 km separating
-20 N the snow above from the rain
beneath. The upright stripes
of stronger echo neural small
scale cells of heavier rain
Distance, km which ore tilted by the strong
wind shear in the lowest
1.5 km.

buildings was dramatically reduced after protection. However, there was

resistance from traditionalists and cynics, and many decades were to pass before
the lightning conductor became universally accepted, as now.
Sailing ships had long suffered damage and injury to crew as a result of lightning
strikes to masts, spars and rigging. Injuries and deaths occurred because crew were
often aloft at the time of the lightning strike, adjusting the sails in response to the
squalls and shifts of wind associated with thunderstorms. The British Royal Navy
then maintained a large number of sailing ships in all seas and weather and suffered
from lightning accordingly. However, the Admiralty did not take kindly to advice
from an outsider, especially after the American revolution cast Franklin in the role
of rebel statesman, and so British sailors continued dying in significant numbers.
Eventually as a compromise their Lordships fitted temporary lightning con-
ductors, to be placed aloft when danger threatened, but otherwise to be coiled away
below. After a period in which more British sailors were killed hauling lightning
conductors aloft, permanent copper strips were attached to the masts and damage
and injury was reduced far below their previous levels.
Every year a few thousand people world-wide are killed or injured through inad-
vertently becoming lightning conductors. Local news reports often note that the
unfortunate person was struck on a metal cap badge or something similar, and this
is indeed often the case. However, it is probably quite wrong to suppose that the
metal object contributed to the accident or that its removal might have prevented
it. As the stepped leader reaches down to within about 100 m of the surface the
electric fields there increase dramatically, especially in the vicinity of any fairly
sharp upward projection from the surface. An upright human form will serve and
so localize the upward spark and the subsequent return stroke. If there is a sharp
protruding badge or golf club no doubt this will be selected, but the result is the
same. The only sensible precaution to take if you are caught in the open by a
thunderstorm is to get down on the ground and stay there until the lightning strikes
are at least a mile off. Standing under a solitary tree is no protection since a strike
on the tree will easily produce lethal fields and currents within many metres of the
base of the trunk, and the tree is of course a prime target. Nor is it sensible to
console yourself that only a few people are struck by lightning; only a few people
are caught in the open by thunderstorms, but a significant fraction of them are
struck.

171
6 Cloud ond precipitotion
Appendix 6.1 Dew-point lapse rate

According to more advanced texts [8] the variation of saturation vapour pressure e,
with absolute temperature Tis given by the Clausius-Clapeyron equation
de,
-=-----
L
dT T(a 2 -al)
where a 2 and a 1 are respectively the specific volumes of vapour and water. Since a 2
is very much larger than a 1, the latter can be ignored. Replacing vapour density
using the equation of state for water vapour, we find
de, L e,
-=--
dT RvT2
where Rv is the specific gas constant for water vapour. A solution of this equation
gives a very close approximation to Fig. 6.2. Since the vapour pressure e of an
unsaturated parcel is by definition equal to the saturated vapour pressure at the
dew-point Td of the parcel, we can reinterpret this equation as a relationship
between vapour pressure and dew point:
de Le
--=---
dTd RvT/
Jn the particular case where specific humidity is conserved, the vapour pressures
and total atmospheric pressures are directly proportional
e = qp (eqn (5.8))
€

so that vapour pressure may be replaced by atmospheric pressure to produce an

identical relationship between atmospheric pressure and dew point. Assuming
hydrostatic equilibrium (eqn (4.6)), this becomes a relationship between dew point
and height, which after inversion and rearrangement using the equation of state for
air (eqn (4.2)) simplifies to
dTd g Rv Tl
--= - - - - -
dz L R T
Since dew point and temperature differ only slightly on the absolute scale, this
simplifies further to
dTd _ _L ~
-- = L R
Td
dz
At a typical low tropospheric temperature of 283 K the right-hand side has
magnitude 1.8 x 10- 3 K m- 1, which means that dew point lapses by 0.18 K per
100m rise through the well-mixed layer, as can be checked from the crossing of
isotherms and q. isopleths on a tephigram. Since ordinary temperature lapses by
0.98 K per 100 m (the dry adiabatic lapse rate), we see that dew point and tempera-
ture converge at a rate of about 0.8 K per 100 min a typical well-mixed layer in the
low troposphere - which in other words is the lapse rate of the dew-point
depression.

172
Appendix 6.2 Droplet growth by diffusion and Appendix 6.2

condensation

The mass flux F of vapour diffusing toward a spherical droplet is given by Fick's
law:

where Dis the diffusion coefficient for water vapour in air and dpldn is the radial
density gradient of vapour density p at radius n from the centre of the droplet
(Fig. 6.28). In a steady state, F must be independent of n and so we can integrate
the diffusion equation to find a relation between F and the vapour density
difference !::,.p between the droplet surface (n = r) and the distant environment
(n = oo).
+

L
p b.p

F L ~~ = 47r D dp

which gives
F = 47rrD !::,.p
As expected we see that F is positive if !::,.pis also (i.e. the flux is toward the droplet
if the ambient vapour density exceeds the vapour density at the droplet surface),
and also that Frx !::,.p.
This vapour flux is condensed at the droplet surface and adds to the mass of the
droplet.

F = p
w
~(.±
dt 3
1rr 3 ) = 47rp r 2
w
~
dt
where Pw is the density of the droplet water (or ice if it is a crystal). Equating the
expressions for F we find
dr D!::,.p
r-=-- (6.5)
dt Pw

Notice that if the vapour density difference is constant, the rate of growth of
droplet radius is inversely proportional to the radius, with the result that the
droplet grows quickly at first and then more and more slowly.
We can integrate eqn (6.5) to find the time taken by the droplet to grow from one
radius to another. First let us interpret the vapour density difference l:,.p.
!::,.p = Pa - Ps
= __f!y__ ( 1OOpa - 100) = __f!y__ SS (6.6)
100 p, 100
where SS is the supersaturation (RH- 100) of the ambient air. Note that the vapour
density at the droplet surface has been assumed to be the saturation value p, at the
common temperature of droplet and environment. This ignores both the difference
between equilibrium and saturation vapour densities discussed in section 6. 7
(which can be important when r is small), and the temperature excess maintained
by continual release of latent heat at the droplet surface. However, neither of these 173
6 Cloud and precipitation Vapour flux

Fig. 6.28 Vapour diffusing

radially inwards towards the
surface of a spherical droplet
in response to greater
environmental vapour
density. F

effects has a substantial effect on the following results. Using eqn (6.6) in (6.5) we
find
dr SS
r-=-- (6.7)
dt X
whereX= 100pw/(Dp,),whichhasvalue4.5 x 10 11 s m- 2 whensaturationvapour
density is 10 g m- 3 , corresponding to a temperature of 11 o C.
Integrate eqn (6. 7) to find the timet to grow from radius r1 to r2 •
X
t = - - ( r22 -r 2 )
2SS I

Starting from radius 0.5 p.m, the times taken to grow to various larger sizes are as
follows at 11 °C and SS 0.2%:
r 1 2 4 8 16 32 p.m
t 0.9 4.3 18 72 291 1163 s
(1.2 4.8 19.4 min)
The unreality of assuming saturation over the surface of a small droplet clearly
cannot alter the conclusion that it takes a long time for a droplet to grow from a
small to a large cloud droplet. These results apply equally to the shrinkage of a
droplet by evaporation: SS is then the subsaturation (100- RH) and the table is
read from right to left.

Appendix 6.3 Terminal velocities of

falling drops [41]

When a spherical body of radius rand density Pw is immersed in air of density p, the
net downward force on it is its weight less the weight of the displaced air (by
Archimedes' principle)

N = 34 1r r3 g (pw-p)
Since the density of water or ice is - 103 times the density of air, the latter may be
neglected in the bracketed term. As the body accelerates downward under this
174 force, the air resistance to its fall increases until downward weight and upward drag
balance at a steady fall speed which is known as its terminal velocity or fall speed. Appendix 6.4
Since the weight of the body is proportional to its volume, while the drag is propor-
tional to its surface area, the terminal velocity increases with the size (radius, for
example) of the body, though not necessarily in direct proportion.
When the body is very small, the terminal velocity Vis so slow that the Reynolds
number Vrlv of the flow round the falling body is less than 0.25. The flow is then
completely dominated by the viscosity of the air, and theory and observation agree
that the drag X is proportional to the product Vr, being given by
X= 61rpvVr
At terminal velocity the values of Nand X must be equal, and it follows that

V = [ 2 gpw] r2
9 Pp

This is the Stokes regime of fall speeds (Fig. 6.10), and it is a matter of arithmetic to
show that it applies to spheres of water or ice up to radii of about 30 ~-tm, which
includes all cloud and aerosol, with modification for non-spherical shapes. The
proportionality to the square of the radius means that fall speeds increase rapidly
with size. In conditions typical of the low troposphere, the bracketed terms have
value 1.2 x 108 s- 1 m- 1, which is consistent with Fig. 6.10. Calling this Ywe have
simply
V= Yr 2 (6.8)
For droplets with radii between about 30 and 1000 ~-tm (i.e. large cloud to rain-
drop sizes), the flow around the falling drop is turbulent. The drag increases more
sharply with droplet size than in the Stokes regime, being proportional to the
product Vr 2 • It follows that the fall speed is almost exactly in direct proportion to
droplet radius. In the low troposphere in SI units
V = 8.0 x 103r (6.9)
which is the middle section of Fig. 6.10.
For larger bodies (large raindrops and virtually all hail) the flow is so strongly
turbulent that the drag is proportional to the product V2r 2 , which gives a terminal
fall speed proportional to the square root of the radius. In SI units we have
V= 250rYI
The turbulence is so great that raindrops shatter soon after reaching radii of about
2 mm. No such limit applies to hailstones, which as a result grow to diameters of as
much as 10 em on rare occasions.

Appendix 6.4 Evaporation of falling drops

Consider a droplet falling at speed V. In a little time interval dt it will fall a short
distance dz (inverting our height coordinate for this problem) given by
dz = Vdt

175
6 Cloud and precipitation If the droplet is evaporating, its radius r is decreasing with time at a rate given by
drldt. In fact the time interval dtcan be regarded as the time taken by the droplet to
change its radius by dr, where
dr ( dt)
dt = (drldt) = dr dr

Hence dz= V~dr (6.10)

r
dr

and z= r
v~
dr
dr = - J'
0
V~dr
dr
Note that the minus sign ensures that a positive downward distance is fallen when
the drop is shrinking (i.e. drldt is negative).
In the case of droplets less than about 30 ~-tm in radius, we already have
expressions in Appendices 6.2 and 6.3 for Vand drldt in terms of r, and these can
be used to evaluate the integral and find the distance z fallen during the total
evaporation of a droplet of any initial radius in a subsaturated environment. We
use eqn (6. 7), with SS interpreted as the degree of subsaturation (100- RH) of the
environment of the falling, evaporating drop, together with eqn (6.8) to find
X r
= J Yr 2 - - d r
r
z
0 ss
If SS is uniform along the path of the falling droplet, standard integration gives
z = X Yr 4
4SS
Using the quoted realistic values for X and Y, and assuming an ambient sub-
saturation of 5 (RH95o/o), we can find the distance fallen during the time taken to
evaporate completely from chosen initial radii
r 1 10 100 1000 ~-till
z 2.7 ~-tm 2.7 em 270m 2700 km
{240m 240 km}
These values exemplify the great sensitivity of z to r which arises from the r 4
dependence, and show clearly that cloud droplets (r< 30 !lm) cannot survive a fall
of more than a few centimetres through even mildly subsaturated air. The z values
for larger droplets are too large because these droplets are outside the size range in
which momentum and water vapour transfer is caused entirely by the molecular
diffusion assumed above, and actual magnitudes of V and dr!dt are respectively
reduced and enhanced by turbulence around the droplet. A thorough treatment is
complicated, because it involves not only the turbulently enhanced drag and
evaporation, but also the wet-bulb chilling of the rapidly evaporating drop below
the ambient air temperature. The bracketed z values above follow from combining
the V oc: r fall-speed regime, which includes the effects of turbulence on drag
(eqn (6.9)), with the above purely diffusive expression for drldt. Though the
approach is internally inconsistent, and on the simplest view should overestimate z
by underestimating dr/dt, it produces values of at least the right order of
magnitude in comparison with more thorough studies [42], presumably because
evaporation rates are depressed by wet-bulb chilling. At any rate they confirm that
drizzle can fall a few hundred metres in moist air, and that rain could fall further
than is ever available in the limited depth of the troposphere.
176
Appendix 6.5 Growth by collision and Appendix 6.5

coalescence

Consider a collector droplet of radius r, falling at speed Vand therefore sweeping

volume at the rate 1rr 2V (Fig. 6.18). If there is a population of much smaller
droplets with specific mass m, their mass per unit volume is mp, where p is the
aggregate density of the air plus cloud. If the small droplets have negligible fall
speed and all in the swept volume are collected, then the rate of collection of mass is
F = 1rr2 Vmp
As in appendix 6.2 this produces a rate of increase of collector mass

F = p
w
~(_i
dt 3
1rr3 )

where Pw is the density of the collected water or ice. Equating the expressions for F,
we find
dr = mp V (6.11)
d/ 4pw
Up to 30 ~m radius we have (appendix 6.3) V = Yr 2 , so that
dr 1
- = - r2 (6.12)
dt A
where A = 4pwl(mpY) = 3.3 x 10- 2 m sifmis 1 g kg- 1, andothervaluesareas in
appendix 6.3. The time taken for the collector to grow from radius r1 to r2 is given
by the integral of eqn (6.12)

t =A (-1r1 - _1r2 )
With m = 1 g kg - 1, it takes nearly 14 minutes for the collector to grow from 20 to
40 ~min radius, and this time halves as m doubles and so on. Growth by collision
and coalescence can therefore be faster than growth by diffusion and condensation
in this important size regime if there is a good supply of droplets suitable for
efficient collection (neither too small nor too large).
Collector droplets with radii between 30 and 1000 ~m have fall speeds given by
eqn (6.9). When substituted in eqn (6.11) it gives
dr 1
-=-r
dt B
where B = 4pwl(mp8 x 103) = 500 s for m = 1 g kg- 1 and previously quoted
values. The time taken for the collector to grow from radius r 1 to r 2 is now given by
t = BIn (r2/r1)
With m = 1 g kg- 1, growth from 30 to 300 ~m (a crucial transition from large
cloud to small precipitation) takes 19 minutes, and this time halves as m doubles
and so on. Clearly growth by collision and coalescence has the capacity to produce
precipitation in the observed short time intervals if conditions are favourable.

·177
6 Cloud and precipitation
Problems

LEVEL 1

1. Which of the following has no direct effect on the density of a vapour in equili-
brium with a water surface: water purity, air pressure, air purity, water
temperature, air temperature, wind speed, water surface shape, air?
2. Define a saturated vapour.
3. A friend gazes at rain beating on the window and exclaims in exasperation that
if only there could be less rain there would be more sunshine. Gently explain
his inconsistency.
4. Rain leaves a hilly road equally wet at two locations separated by 500 m in
altitude. If the ensuing sunshine and wind plays equally on each, give two
reasons why you would expect the lower level to dry faster.
5. List the deductions about cloud formation which can be drawn from the
simple cloud-chamber experiments described in the text.
6. Given the large number densities of aerosol, why are the number densities of
cloud droplets so comparatively small?
7. Explain why, when the upper parts of a mixed (ice and water) cloud are
dissolving, the last vestiges are always ice cloud.
8. Although rain drops with radii larger than about 2 mm are not observed
(appendix 6.3), much larger drops are observed beneath trees in wet weather.
Explain.
9. It is a common fallacy to suppose that sheet lightning is not associated with
thunder. Explain what actually happens.
10. Figure 6.27 shows a melting band on a range-height radar display. Consider its
appearance on the more normal plan-position display (like Fig. 2.6), in which
the radar beam continually scans around at a small angle above the horizontal.
Remember the curvature of the Earth's surface.

LEVEL 2

11. Use Fig. 6.2 to estimate vapour densities and pressures at the following
surfaces: an ice cap in summer ( -10 °C) and the equatorial ocean surface
(25 °C). Why might a similar estimate for the Sahara be very misleading?
12. In a certain subtropical desert, dew is observed to form each night when the
surface temperature falls to 5 °C. Assuming constant atmospheric pressure
and vapour content, estimate the relative humidity when the surface tempera-
ture reaches 50 °C during the heat of the day, using Fig. 6.2 and given that the
s.v.p. at 50 °C is 123 mbar. Assuming that the air temperature at screen level
is then 45 °C and that this is at the base of the well-mixed layer, use the simple
rule of thumb for lapse of dew-point depression to find the implied height of
cloud base above the surface.
13. Show from the approximate form of the relationship between equilibrium and
saturation vapour densities (eqn (6.4)) that on any particular Kohler curve the
radius r at which these densities are equal is given by
,.z = BmT
A
178
 Given that A = 3.16 X I0- 7 m K for water at 0 °C, and that for NaCl Problems
B = 1.47 x I0- 4 m 3 kg- 1 K- 1 , find r values corresponding to salt particle
masses of w-JS and I0- 16 kg.
14. Use the final equation of appendix 6.2 to calculate the time taken for a cloud
droplet of radius 5 JLID to evaporate completely in an environment with
ambient relative humidity 900Jo when the temperature of droplet and environ-
ment is 11 °C. Repeat the process at -20 °C, given that the saturation vapour
density is then 1.07 x I0- 3 kg m - 3 , and note the relevance of the result to the
persistence of ice cloud in the upper troposphere.
15. Find the time taken for a droplet to grow from radius 15 to 500 ~-tiD by collision
and coalescence in the low troposphere with cloud water of specific mass
3 g kg- 1 • Allow for the change in terminal velocity regime by working out in
two stages: the first to cover growth to 30 ~-tiD and the second to cover further
growth. How much longer is needed to grow to radius 1 mm?
16. Lightning strikes the surface 200m away from an observer. Given that the
flash extends to a maximum height 2 km vertically above the observer, and has
a nearly horizontal section 1 km above the observer, describe the observed
peal of thunder, given that the speed of sound in air is 330 m s-t.

LEVEL 3

17. Consider the realistic development of a layer of dew 1 mm thick in 1 hour to

find the implied flux of condensing water in molecules per second per unit
area. Compare this with the number of impacts (X) of molecules of water
vapour calculated from the following expression derived from the kinetic
theory of gases [35].
103 A e
X=-----
2.J3 RTM
where A is Avogadro's number, R is the universal gas constant, e is the vapour
pressure, M and T are respectively the molecular weight and absolute
temperature of the vapour, and the units of X are again molecules per second
per unit area. Notice that the result shows that the deviation from perfect
dynamic equilibrium is very small.
18. By analyzing the shape of the Kohler curve described by eqn (6.4), show that
radius rmat which the equilibrium vapour density is a maximum is .J3 times the
radius r at which the equilibrium vapour density is equal to the saturation
value (i.e . .J3 times the radius found in Question 13). Show further that the
maximum supersaturation SSm implied is given by 200A/(3.J3rT), where
terms are as defined in problem 13. Calculate numerical values for rm and SSm
for the salt particles cited in problem 13.
19. Discuss the role of equilibrium and runaway diffusive growth and evaporation
in ensuring that clouds have both sharp bases and edges.
20. Consider the growth by collision and coalescence of a droplet falling through
cloud with cloud water specific mass m, and show that droplet growth dr
increases in proportion to distance fallen dz (assuming no updraught in the
cloud) according to

dr=mpdz
4pw
179
6 Cloud and precipitation where p and Pw are respectively the density of air and water. Hence find the
depth of very wet cloud (m = 4 g kg-', typical of warm, low clouds over hills in
wet weather) which could produce an increase of 1 mm in droplet radius. Such
rapid growth partly accounts for the familiar increase of precipitation over
hills.

180
Atmospheric CHAPTER

dynamics 7

7.1 The equation of motion

When a force Facts on a body of mass M, then according to Newton's second law
of motion, which is the law underpinning all dynamics, we have

F = __E_ (MV) (7 .1)

dt
where Vis the velocity of the body and MVis called its momentum. The right-hand
side of eqn (7 .1) represents the rate of increase of the body's momentum with time
as a result of the action of the force. This rate of change can arise because of
changing mass or changing velocity or both, but in the common case of constant
mass we have Pole

F= M dV
dt
or
a=FIM (7.2)
where a(=dV/dt) is by definition the acceleration of the body. The form of
Newton's second law of motion apparent in eqn (7 .2) is used so widely in meteo- Equator

rology that it is known simply as the equation of motion, but it is important to Fig. 7.1 Northern
remember that it assumes constancy of air parcel mass and is therefore inapplicable hemisphere quadrant
to the dynamics of cumulus, for example, where air parcels grow rapidly by showing conventional
entrainment of ambient air (sections 7.15 and 11.2). meteorological axes with
origin 0 on the surface.
The vector notation of eqn (7 .2) allows for the fact that both acceleration and
force have magnitude as well as direction, and the simple form of the equation
means that their directions are the same: acceleration takes place in the direction of
the acting force. Being a vector equation it can be broken down into three scalar
(i.e. having magnitude only) equations in mutually perpendicular directions or
axes, and it is these component equations which are often most convenient for
examination and solution. The convention in meteorology is to use the axes shown
in Fig. 7.1, with origin 0 at a convenient point on the Earth's surface, and with the
x, y and z axes pointing horizontally eastward and northward and vertically 181
7 Atmospheric dynamics upward respectively. The wind speeds in these respective directions are written as u,
v and w by the same convention. Since the axes are at right angles to each other, the
component equations can be considered quite independently, although of course
all three may be needed for any particular problem. For example, the vertical
component of the equation of motion is as follows, regardless of the horizontal
components

dw = FIM (7 .3)
dt z

where Fz is the vertical component of the net force or forces acting on the parcel,
and dw!dt is the parcel's vertical acceleration. Notice that the right-hand side of
the equation is the force per unit mass of air parcel, as is always the case in the
equation of motion. If the force component is known, the acceleration is
determined, and successive integrations give its vertical speed and displacement. If,
however, the three-dimensional parcel motion is required, then the horizontal
components are needed also.
Although the notation and equations may seem rather abstract and forbidding,
it is important to realize that they are used precisely because they provide concise
and unambiguous descriptions of events, as well as the hope of manipulation and
solution. Solutions are minimal in an introductory treatment such as this, but state-
ments and manipulations involving mathematics are very useful nevertheless. Even
in the extensive equations which abound in advanced texts, a vital first stage in any
study of dynamics is to identify the physical forces and other factors represented by
the various symbols, and to consider the physical reality of the balance represented
by the equations combining them. When examined carefully in this way, even the
relatively simple equations quoted in this chapter will be seen to summarize
relationships and behaviour which would each require many paragraphs of purely
verbal description.

7.2 Forces

If the gradient of air pressure at a given location is steepest along a certain axis,
then an air parcel there is subjected to a force along that axis toward low pressure,
because the pressure on the up-gradient side of the parcel is greater than on the
opposite side. In appendix 7.2 it is shown that the force per unit mass of air parcel is
- (11p)/EJp!EJn, where EJp!EJn is the instantaneous pressure gradient in the direction
of increasing n (distance along axis), pis the density of the air in the parcel, and the
minus sign indicates that the force acts in the direction of decreasingp. The partial
derivative (EJ/EJn) has already been used in section 4.4 but now that we will be con-
sidering variations with time as well as position, it is important to be clear that the
partial derivative describes variation (of pressure pin this example) with only one
independent variable (n in this example) out of several. In dynamics we are usually
concerned with four independent variables - x, y, z and t - and they are
independent in the sense that we can readily imagine any one of them varying while
the others are held constant. For example, EJp/ ax represents the eastward gradient
of pressure at fixedy, z and t, and EJp/EJt represents the rate of variation of pressure
with time at fixed position (x, yand z)- the pressure tendency as measured from a
182 fixed barograph.
 The pressure gradients along the x, y and z axes are found in the case of any 7.2 Forces
particular axis n by taking components in the usual way (Fig. 7.2). For example, if
the angle between then and z axes is a, then the vertical pressure gradient oploz is
given by
ap op
-=-cos a
oz on
On scales much larger than those of small-scale turbulence and the perturbations of z
flow around buildings et_c., the atmospheric pressure gradient is very nearly
vertical, i.e. a is very small in the above and op/ oz is very much larger than either n
oplox or oploy. Since pressure falls with increasing height, it follows that by far the
largest component of the pressure gradient force per unit mass on these and larger
scales acts vertically upwards, tornado funnels being rare and highly localized
exceptions. For example, on the synoptic scale, horizontal pressure gradients X
associated with extratropical cyclones rarely exceed 5 mbar per 100 km, in
comparison with the vertical pressure gradient of 1 mbar per 10m which is typical Fig. 7.2 Axes x, y and z
of the low troposphere. However, thanks to the independence of the vertical and related to a generalized axis
n.
horizontal components of the equation of motion, the small horizontal pressure
gradient forces are nevertheless very important, as we shall see.

GRAVITATIONAL FORCE

The mutual gravitational attraction between an air parcel and the Earth produces
the downward force on the parcel which we call its weight, and an equal and
opposite force on the Earth which ordinarily has no observable effect. Measure-
ments on calibrated weighing machines at rest on the Earth's surface show that the
weight of any body of mass M is given by Mg, where g has an average value of
9.81 m s- 2 , and varies slightly with latitude cfJ and altitude z above sea level as
shown in Table 7 .1. (We assume that we already have an unambiguous definition
and measure of mass, though in fact Newtonian dynamics largely evades this point
by circular argument and an appeal to an intuitive sense of the quantity of matter.
It is simplest to accept that mass is already defined and measurable.) The factor g is
known as the gravitational acceleration because it is equal to the downward
acceleration which the body would undergo in the absence of any other force (i.e.
falling freely in a vacuum). In the meteorological context it is important to notice
that in either case (weighing machine or free fall) we make observations from a
laboratory fixed to the rotating Earth's surface, so that the measured g is apparent
g, rather than the absolute value arising purely from gravity which would be
measured from a non-rotating laboratory. The distinction will be discussed later
(section 7.4). Note that the gravitational force per unit mass is simply g in the
negative z direction, and that for many meteorological purposes g can be assumed
to have a uniform value of 9.8 m s- 2 throughout the troposphere. Indeed for
rough estimates the value of 10m s- 2 is often convenient and accurate enough.

Table7.1 Apparent gravitational acceleration

Altitude Latitude
lnkm oo 45° 900

0 9.780 9.806 9.832

20 9.719 9.745 9.770
40 9.538 9.684 9.709
183
7 Atmospheric dynamics FRICTIONAL FORCES

Friction arises when bodies in contact move with different velocities, or tend to do
so. Although we may think of it as a property of solids in contact, friction occurs in
all normal fluids and plays an important role in the incessant commotion of the
atmosphere by tending to reduce all velocity differences induced by other means.
Such differences appear in the form of shears (gradients) of air flow, and some of
the most significant shears, apart from the localized, transient ones associated with
turbulence, are shears of horizontal wind. A strong concentration of vertical shear
of horizontal wind is almost universal in the planetary boundary layer, particularly
in its lowest levels. This is depicted in Fig. 7.3 as an example for visualization and
discussion, but the principles outlined are quite general. For the moment, nothing
is lost by assuming the shear to be two-dimensional, though in fact it is usually
three-dimensional (like most atmospheric shears) in the sense that both wind speed
and direction vary with height.

______,
Wind vectors
I
/

.. ------o;/
~ /
Fig. 7.3 Two-dimensional ·;; ----~,/
_.... , /
-- --
picture of mean horizontal J:

wind near the Earth's surface,

showing vertical shear. Wind speed in x direction (example)

The effect of friction on the sheared flow in Fig. 7.3 is to tend to speed up the
slower and slow the faster air, and ultimately to make both move at the same speed
as the surface. Over land this would mean the air eventually coming to a halt, as
happens occasionally on very quiet nights, and over the sea it would mean the air
moving at the very low speeds (- I m s -I) typical of the wind-driven sea surface.
Neither of these conditions arises often, because the air flow is usually driven from
above by some large-scale weather system, such as an extratropical cyclone or the
trade winds. The shear between flowing troposphere and static or nearly static
surface is then maintained, and the effect of friction in the sheared layer is to
transfer horizontal momentum downwards from the weather system to the
surface. The air aloft and the surface beneath experience equal and opposite drags
(Fig. 7.4).
Momentum is transferred across wind shears because air itself is transferred, and
this occurs in two very different ways in the atmosphere. On the molecular scale,
molecules intermingle in the course of their incessant, random thermal motion,
and the resulting diffusion of momentum is called viscosity. On the scale of
turbulent eddies (- mm to 100m or more), individual eddies intermingle in the
course of turbulent motion and the resulting very rapid diffusion of momentum is
often ascribed for convenience to eddy viscosity. Let us consider the molecular
scale first because it is much simpler in principle, and because its formalism can
then be extended to eddy viscosity.
If rvc is the viscous drag per unit area in the x direction experienced by the
horizontal upper surface of an air parcel because of the relative motion of the air
above (so that rvc is a tangential stress to compare with the perpendicular stress
exerted by pressure), then according to eqn (7 .1) this drag must tend to increase the
momentum of the air parcel in the x direction at just this rate, Noting the presence
184 of the unit area in the definition of stress, it should become clear after a little
 7.2 Forces

Fig. 7.4 A small tree on an

exposed part of the South
Downs near Brighton, Sussex.
The branches have developed
systematically on the side
sheltered from the prevailing
wind giving the whole crown
an asymmetrical , streamlined
appearance.

thought that r zx can be regarded as the downward flux of eastward momentum

through unit horizontal area. It is called the x component of the horizontal shear-
ing stress because of its obvious connection with the shear au/az - the shear
depicted in Fig. 7.3.
According to Newton's empirical law of viscosity, the shearing stress and shear
are directly proportional, the constant of proportionality being the dynamic coeffi-
cient of viscosity p., which has a well-defined value for any given thermodynamic
state of the air.
au (7.4)
Tzx = ~-'-az

According to appendix 7.3 the net viscous force per unit mass eastward on a thin
horizontal slab of air embedded in the shear aulaz is given by
1 a
--T
p az zx
(7.5)

Note that as in the case of a pressure gradient, the net force depends on the gradient
of a stress, though in this case the stress and gradient are perpendicular rather than
parallel as they are in the case of pressure. Since in realistic atmospheric conditions
gradients of p. are proportionately much smaller than gradients of shear, we can
combine eqns (7 .4) and (7 .5) to obtain
1 a a2u
--r = v-
p az zx az 2 (7.6)

where v is the kinematic coefficient of viscosity (p.l p) of the air. The value of v
depends mainly on temperature in typical atmospheric conditions and is about
14 x IQ- 6 m2 s- 1 in the low troposphere. Note that the net viscous force increases
with the curvature a2 u/ az 2 of the velocity profile (the shear of the shear), which is
consistent with the intuitively reasonable tendency for viscosity to remove humps
and hollows in the velocity profile and smooth it as much as possible.
Because of basic similarities between molecular and eddy viscosity, we might
expect there to be the turbulent equivalent of eqn (7 .4) linking eddy stress to the
shear of the mean wind au/az (where the wind component is averaged to remove
the turbulent gustiness) 185
7 Atmospheric dynamics
Tzx-
_ K
az
au (7.7)

In fact the statistics of turbulence are observed to be so much more elaborate than
those of molecular motion that the coefficient of eddy viscosity K defined by eqn
(7. 7) is not related in any simply predictable way to the thermodynamic state of the
air, and indeed is not fully determined by any statistical description of the
turbulence discovered so far. This severely limits the usefulness of the concept of
eddy viscosity, but its attractive simplicity nevertheless keeps it in widespread use,
especially in experimental studies of the connection between eddy stress and mean
shear in the enormous range of turbulent conditions encountered in the atmo-
sphere, and in relations which can be derived from these. Values of K derived from
field measurements and eqn (7. 7) range across many orders of magnitude, but
more than a few centimetres above the surface they are almost always at least four
orders of magnitude larger than v. Despite uncertainties of detail the general
picture is clear: in almost all of the lower atmosphere eddy viscosity overwhelms
molecular viscosity as an agent of momentum diffusion to the extent that we can
assume that the motion of an air parcel is affected only by eddy viscosity.

7.3 Relative and absolute accelerations

Newton's second law of motion (eqn (7 .1)) relates resultant force and acceleration
as measured from an unaccelerated observation platform (an inertial reference
frame). The need to use an inertial frame should be clear from the example of a
man unwise enough to weigh himself on spring scales in an accelerating lift.
Though his acceleration relative to the lift is zero if he is standing steadily, it does
not follow that the downward force exerted on his feet by the scales (and registered
on their display) will be equal to his weight; in fact if the lift is accelerating upward
the scales will register more than his true weight, and if it is accelerating downward
they will register less than his weight. This illustrates the very fundamental point
than an unrecognized acceleration in one direction appears as an inexplicable force
in the opposite direction - mysterious weight or buoyancy in this case.
Now atmospheric motions, including accelerations, are measured from a
reference frame fixed to the Earth (Fig. 7 .1) and therefore rotating with it. The
meteorological reference frame is rotating and therefore accelerating continually,
and this acceleration must be allowed for when using the equation of motion,
otherwise mysterious forces will seem to invalidate Newton's second law of motion
and any predictions based on it.
Because the three-dimensional shape of the Earth complicates the issue without
altering basic principles, we will consider the effects of rotation first in relation to a
flat turntable. Figure 7.5 represents a model train T running at speed V around a
circular track of radius R on a turntable which is rotating with angular velocity !1
about the centre of the track 0. Let us use the basic kinematic rule that a body
whirling at speed v round a circle of radius r experiences a continual centripetal
acceleration v 2 /r toward the centre of the circle (Appendix 4.5). It follows that the
centripetal acceleration of the train toward 0 is
I. V2 I R as measured by an observer rotating with the track and turntable; but that
186 it is
2. ( V + OR)2/ R as measured by a non-rotating observer, for example one sitting 7.3 Relative and absolute
beside the turntable, where OR = ~ is the tangential speed of any part of the accelerations
track on account of its rotation (Appendix 4.5 again).
The difference between expressions (2) and (1) represents the centripetal
acceleration which is ignored when observations are made from the turntable
rather than a non-rotating frame, and which would therefore seem to the observer
on the turntable to be an outward (centrifugal) force acting on the train. Such
forces are often called apparent forces in the sense that they appear only when
observing from an accelerating reference frame, but of course in that frame they
are quite real. For example in the present case, the toy train would tip outward off
its rails if the turntable rotation and train speed were too high for the train's lateral
stability, and its toy-town passengers would be justifiably mystified if the derail-
ment were attributed to an unreal force!
Multiplying out the bracket of the absolute centripetal acceleration (2) and
subtracting the relative centripetal acceleration (1), we find that the difference
comprises the two terms
0 2R + 20V (7.8)
The first of these is the familiar centripetal acceleration of any fixed point on the
track (Appendix 4.5), which obviously must be one of the terms lost in making
observations from the turntable itself. The second term combines the motions of
the track (0) and of the train relative to the track ( V), and is known as the Co rio/is
acceleration, after the French engineer who drew attention to its importance in
rotational dynamics. It is a less obvious casualty of using a rotating reference
frame, but it is important to see that no principle is involved beyond what is implied
in the simplest treatment of centripetal acceleration. The Coriolis acceleration is a
component of the centripetal acceleration of a moving body (the train in this case)
which becomes identifiably distinct because we choose to observe the body from a
rotating frame (the turntable in this case) in which the body is not at rest. Note that
in the present example the Coriolis acceleration acts perpendicularly to the left of
the instantaneous velocity of the body relative to the rotating frame - the left
being the side of the train's direction of relative motion which is obviously
favoured by the particular direction of rotation of the turntable (Fig. 7 .5). It will
transpire that this is not just a result of this particularly simple example: like the
centripetal accelerations of which they are a part, all Coriolis accelerations are per-
pendicular to the relative velocity in the direction favoured by frame rotation
(section 7 .5).

Fig. 7.5 Toy train T travelling

v round a circular track on a

n::\
at T rotating turntable. The inset
shows the magnitude and

v
direction of the Corialis
acceleration at T.

Coriolis
accelera !ion
21lV
187
7 Atmospheric dynamics
7.4 The Earth's rotation and apparent g

If the Earth were a perfect sphere with concentric distribution of mass, the true
gravitational force on a body on the surface would everywhere act with uniform
strength toward the Earth's centre. However, if the gravitational force is measured
relative to a fixed point on the rotating Earth's surface, as it normally is of course,
true g will be offset by a 'mysterious' centrifugal force at all latitudes except the
poles, with the result that measured (apparent) g will be less than true g, and
directed slightly away from the Earth's centre unless the point is on the equator
(Fig. 7 .6). There should by now be no mystery about the centrifugal force: it is a
consequence of ignoring the centripetal acceleration of the Earth-bound reference
frame, and if we were unwise enough to measure g from a frame moving over the
Earth's surface we would have a Coriolis effect as well! The discrepancy between
true and apparent g values will be greatest at the equator because the centrifugal
effect is greatest there. As shown in appendix 4.5 fFR has value 0.034 m s - 2 , which
is of the same order of magnitude as the variations in apparent g in Table 7 .1.
However, more significantly the angling of apparent g away from the local radius
would imply a horizontal component of apparent g acting toward the equator (i.e.
one parallel to the local surface) with a maximum value at latitude 45°.
Consider the consequences of having an equatorward horizontal component of
apparent g. In the absence of any opposing force, the atmosphere and oceans
would move toward the equator and accumulate there, adding to the equatorial
girth of the fluid planet by denuding the polar regions. If the oceans were frozen
and unable to move, the atmosphere would redistribute its mass until the
accumulation of air in low latitudes produced a poleward pressure-gradient force
everywhere exactly balancing the equatorward apparent g force. A simple estimate
in Appendix 7.4 suggests that this would produce permanent horizontal pressure
gradients in middle latitudes fully comparable with those found in deep depres-
sions, and isobaric surfaces rising over 10 km from pole to equator. If the oceans
were free to move they would flow until mean sea level was similarly tilted, in the
process doing very strange things to the relative levels of sea surface, sea-bed and
land. No such relative displacement is observed because the 'solid' Earth itself is
nearly fluid on geological timescales and has bulged at the equator to adjust to the
centrifugal effect of its own rotation in essentially the same way as we have been
imagining for the atmosphere and oceans. In fact it has reached the equilibrium
distribution in which there is no longer any component of apparent g parallel to the
Earth's surface, so that the local vertical, as defined by a plumb line, is perpen-
dicular to the local horizontal as defined by mean sea level. The equilibrium is com-
plicated by the onion-like zoning of increasing density toward the Earth's centre,
and the distorting effect of the bulged mass on its true gravitational field, but the

Fig. 7.6 True and apparent

accelerations (g 1, g 0 ) on the
surface of a spherical Earth.
Key,
The dashed line represents
the actual spherical Earth. 9a apparent
gravitational
acceleration

9t ~rr~~itational
acceleration

188
upshot is that the Earth's equatorial radius exceeds its polar radius by about 21 km 7.5 The Corio! is effect
and the strength of apparent g varies with latitudes as shown in Table 7 .1. This
slight oblateness of the Earth is of no meteorological significance, and the meri-
dional and vertical variations of apparent g have to be considered only in the most
accurate representations of large-scale fields of atmospheric pressure. Because of
the long-term fluidity of the Earth itself therefore, the static distributions of the
atmosphere and oceans over the slightly oblate spinning globe are virtually what
would be expected on a non-rotating sphere.

7.5 The Coriolis effect

The static centrifugal bulge of the Earth outlined in the last section accommodates
by far the largest effects of the Earth's rotation, and leaves the resting atmosphere
and ocean effectively dynamically unaffected by that rotation provided we treat
mean sea level as our horizontal datum, which of course we do. But once these
fluids begin to move relative to the spinning Earth the smaller but still significant
Coriolis effect comes into play, producing terms which are proportional to the
wind and current speeds themselves. Because of this proportionality there is no
single distribution of fluid mass and flow which will accommodate the Coriolis
effects; there are temporary accommodations which change as the atmosphere and
oceans ceaselessly shift and vary. We need a simple but comprehensive description
of the Coriolis effect to apply to the many different situations in meteorology (and
oceanography) in which it is important.
Consider the model train again (section 7 .3), but this time moving steadily along
a straight track placed at random on the same rotating turntable (Fig. 7. 7). As the
turntable and track rotate, the train is forced to accelerate laterally, and it is shown
in Appendix 7.6 that this lateral acceleration is simply 20 V, which is identical to the
Coriolis term in eqn (7 .8), even though this time there has been no simple arrange-
ment of track relative to axis of rotation. The train may be moving straight towards Fig. 7.7 Coriolis acceleration
for relative motion on a
the turntable axis, or away from it, or along any intermediate line, but in every case counterclockwise rotating
the Coriolis acceleration is perpendicular to the track (i.e. perpendicular to the turntable.
train's line of motion relative to the turntable), to the left or right depending on the
direction of turntable rotation (Fig. 7. 7). Of course there is also an 0 2R type term
which does vary with position on the turntable and is always directed toward the
axis of rotation, but this is the equivalent of the static centripetal term, which the
previous section has shown we can ignore almost entirely.
The Corio lis effect is complicated in detail but not in principle by the fact that the
Earth is a rotating sphere rather than a flat turntable. In Fig. 7.8 it is apparent that
at any point Pat latitude cf> on the Earth's surface in the northern hemisphere, the
Earth's angular velocity vector (magnitude 0 and directed along the axis of
rotation in the sense of a right-handed screw) can be resolved into two components
-one of magnitude 0 sin ct> along the local vertical at P (the z axis), and the other of Equator
magnitude 0 cos ct> along the poleward-pointing horizontal (they axis). We can Fig. 7.8 Components of
therefore represent the Earth's rotation by means of two turntables, one rotating Earth's rotation about the
about the z axis and the other rotating about they axis, and can apply the turntable local vertical axis z and about
the local northward axis y, at
form of the Coriolis effect to each to produce all the Coriolis terms for the actually latitude</>.
nearly spherical Earth.
The turntables are shown in Fig. 7.9 as seen looking at P down the z andy axes 189
7 Atmospheric dynamics (a) ~ !1 sinlll
(b) ~ !l coslll
about z about y
y

r
axis axis z

Fig. 7.9 Components of ftCy. u u. cx¢=rw

-c.
Coriolis acceleration in (a) the ~-Cz
local horizontal plane, and p
X X
p
(b) the vertical plane
perpendicular to the local Cy = 2 !l sin !II u Cx = 2 !l cos !II w
northward axis. Cx =-2 n sin,ll v Cz =-2 !l cos Ill u

separately, with the sense of rotation as for the northern hemisphere. Considering
the components of wind speed u, v and win turn and comparing with Fig. 7. 7, we
find the following components of the Coriolis acceleration in the directions of the
conventional x, y and z coordinates axes (Fig. 7.1):
x - 20v sin 4> + 20w cos 4>]
y 20u sin 4> (7.9)
z -20u cos 4>
If for example we are considering a northerly wind (i.e. negative v and zero u and
w) we see that the only Coriolis component is a horizontal acceleration toward the
east (i.e. in the x direction). Westerly winds and updraughts are each associated
with two Coriolis components, and a wind with significant components in all three
directions is associated with a similarly three-dimensional Coriolis acceleration.
Notice that all Coriolis terms have the same form - a product of relative velocity
component, the Earth's angular velocity, and the tell-tale factor 2.
In many realistic meteorological situations not all of these terms are equally
important, and some are often negligible, but discrimination is possible only when
we have assembled all significant terms of each component of the equation of
motion.

7.6 The equations of motion

The various contributions to the equation of motion (eqn (7 .2)) may now be
summarized symbolically as a balance between accelerations (A terms) and forces
(Fterms), each of which is a vector quantity.
RA + FA = PGF + GF + FF (7.10)
The relative acceleration RA is what is observed from the meteorological
reference frame. For example, on the large scale, air on a chart of the upper tropo-
sphere may be seen accelerating into the core of a jet stream- a linear acceleration
in the sense that it is parallel to the wind velocity. Or in the low troposphere, air
may be seen moving in a horizontal circular path around a centre of low pressure,
which implies a centripetal acceleration toward the centre of the circle. On smaller
scales there may be transient intense accelerations as buoyant air rises into vigorous
shower clouds, and on even smaller scales air in the turbulent boundary layer is
continually accelerating and decelerating in the chaos of transient eddies. Each of
these is an example of the acceleration of air parcels relative to the Earth-bound
190 reference frame.
 The frame acceleration FA contains only the Coriolis acceleration CA because 7. 7 Synoptic-scale motion
the large fixed component of centripetal acceleration has been accommodated by
the fluid deformation of the rotating Earth and the use of apparent g instead of true
g (section 7 .4). It is customary to deal with the Coriolis effect as if it were a force on
the right-hand side of the equation of motion rather than an acceleration on the
left-hand side. As discussed in section 7.3 this is a perfectly satisfactory procedure
provided the apparent Coriolis force CF is equal and opposite to the Coriolis
acceleration CA. Equation (7 .10) then becomes
RA = CF+ PGF+ GF+ FF (7.11)
The remaining terms on the right-hand side are the pressure gradient force PGF,
the (apparent) gravitational force GF whose magnitude is defined by Table 7.1,
and the frictional force FF. Of the last two, GF always acts vertically downward
and FF in the low troposphere usually acts to oppose the air motion.
This vector form of the equation of motion is very compact, even when each
symbolic term is fully written out, but it is not particularly suitable for detailed
analysis or solution. These require that the equation of motion is broken into its x,
y and z components, producing in effect three equations of motion which are
independent except in so far as they share common terms. We can write down these
component equations by collecting the detailed terms discussed in the last few
sections.

xcomponent ~ = 20 sin c/J v - 20 cos c/J w - _!__ op +F (7.12a)

M p ~ X

y
dv
- = -20smc/Ju-
.
-
1 ap
-+F (7.12b)
dt p ()y y

z dw = 20 cos c/J u - 1 op - g + F (7 .12c)

dt p az '
These equations form an elaborate set whose complete solution, together with any
thermodynamic or other constraints which may apply, is far beyond the currently
very considerable capabilities of applied mathematicians. Indeed the friction terms
are not even reasonably defined by current theory and observation, though they
can be written much more specifically for many purposes (e.g. section 9.8).
Fortunately it is possible to gain very useful insights into some of the types of
behaviour they describe by means of judicious simplification of the full set,
removing terms which are believed for good reasons (ultimately observational) to
be relatively unimportant. A first step in this process is to examine the relative
magnitudes of the various terms in each component by substitution of observed
values, but since the relative sizes of these values often vary enormously depending
on the time and space scales of the types of motion being examined, we must first
specify these scales. This simple but extremely useful procedure is known as scale
analysis. Let us use it first to examine the relative importance of the various terms
of the equation of motion as they apply to synoptic-scale behaviour.

7. 7 Synoptic-scale motion

Typical values observed in the synoptic-scale weather systems which dominate the
troposphere in middle latitudes, and in modified form are still important in low
latitudes, are as follows. 191
7 Atmospheric dynamics Horizontal scale length L IOOO km 106 m
Vertical scale length H IO km 104 m
Timescale t I day 105 s
Horizontal pressure changes !:::,.p 10 mbar I03 Pa
Vertical pressure change p 1000 mbar 105 Pa
Air density p I kg m- 3
Earth's angular velocity o I0- 4 rad s- 1
Apparent g g 10m s- 2
Although the meaning of any particular scale is deliberately vague, and the
values are mostly simple orders of magnitude (such is the variability of atmospheric
behaviour), the following indicates the way in which we can interpret them. Lis the
horizontal distance in which there is usually a substantial change in any observable
property of the system, such as pressure or wind field. Hand t mean the same thing
in relation to the vertical scale and the timescale. !:::,.p is the typical horizontal
pressure variation observed in such weather systems, whether spatially from centre
to edge, or temporally as observed by a fixed barograph as a system travels past.
The variation in pressure in the vertical is dominated by the large background
pressure lapse, and its value is comparable with the typical pressure surface
pressure p because most large weather systems fill the full depth of the troposphere
and hence encompass more than a scale height of pressure. And p is a typical
average air density.
Values of other measurable properties, and terms in the component equations of
motion, can be derived from the above set. For example:
Horizontal wind speed U- Lit IO m s- 1
Vertical wind speed w - Hit 10-1m s-1
Horizontal acceleration U/t (or U 2 /L) 10-4m s-2
Vertical acceleration wit (or w 2 / H) 10-6m s-2
Major Coriolis acceleration OU 10-3m s-z
Horizontal pressure gradient I::::,. pi L 10- 3 Pa m- 1
Note that the bracketed alternatives remove the timescale in a way which will
prove very useful later, and that omission of sin cjJ and cos cjJ from the Corio lis terms
is unimportant provided we are not very near the equator.
Ignoring the friction terms for the moment, we can now give an order of
magnitude for each term in the eqns (7.12). The units are m s- 2 (acceleration or
force per unit mass) throughout.

.
20 smcpv ap
-du = - 20 cos cjJ w (7.I3a)
dt p ax
10-4 10-3 10-5 10-3
dv ap (7.13b)
-20 sin cjJ u
dt p ay
10-4 10-3 10-3
dw ap (7.I3c)
= 20cos cjJ u -g
-
dt p az
10-6 10-5 10 10

Several important simplifications now emerge.

I. The second Corio lis term in the x component equation (7 .I3a) is two orders of
magnitude smaller than the first because synoptic-scale updraughts are so much
I92 weaker than horizontal winds, and may therefore be ignored, leaving
 the two horizontal-component equations (7.13a) and (b) looking much more 7.7 Synoptic-scale motion
alike.
2. The relative acceleration terms (duldt, dvldt) in the horizontal-component
equations are smaller than the major Coriolis and pressure-gradient terms, but
by only one order of magnitude. This suggests that though the acceleration
terms are often relatively unimportant they are not always so. It will appear in
the next section that the relatively very simple situation which arises when the
relative accelerations really are negligible corresponds to geostrophic balance,
and that much of the lower atmosphere, but not all of it, is in a state of quasi-
geostrophic balance on the large scale.
3. In the vertical-component equation (7 .13c) only the gravitational and vertical
pressure-gradient terms are significant. The balance between these two cor-
responds to the hydrostatic equilibrium discussed in detail in Chapter 4. The
relatively complete unimportance of the other two terms shows how good the
hydrostatic approximation really is on the synoptic scale. But despite this, note
the absurdity of assuming from this that vertical accelerations are strictly zero:
there could then be no updraughts, other than over active volcanoes, no cloud
and no precipitation!
Let us now consider the possible importance of the frictional terms which have
been deliberately omitted from eqn (7 .13). If the friction were due solely to
molecular viscosity, then the frictional terms in the equation of motion would
include v (J2ul az 2and similar terms (section 7 .2). In fact this particular term and the
other terms involving the curvature of the vertical wind profile are always likely to
be the largest frictional terms, especially because of the concentration of strong
wind shears in the planetary boundary layer. To estimate such terms by scale
analysis, note that the second derivative a2u!az 2is the gradient of a gradient, i.e. a
velocity difference divided by a length and divided further by a length. If the length
is assumed to be the typical depth h of the planetary boundary layer then
a2u a2v U 1Q-Io -2
(7.14)
p az2 ' p az2 - p Ji2 - ms
since v = 10-s m2 s- 1 and h - 1()3 m.
The small size of the largest viscous friction terms compared with other terms in
eqn (7 .13) confirms the earlier contention (section 7 .2) that viscosity is dynamically
unimportant even in the strongly sheared boundary layer. In fact it is very much
less important even than the very weak synoptic accelerations du!dt and dvldt.
The comparison between acceleration and viscous terms has been found to be a
crucial one across the whole range of fluid dynamics for distinguishing between
flows which are significantly controlled by viscosity and flows which are not. It is
conveniently expressed by the ratio acceleration/(viscous term), which is called the
Reynolds number (Re) after the pioneer fluid dynamicist Osborne Reynolds.
Normally the same scale is assumed in both the acceleration and viscous terms,
with the result that
Re = UL (7 .15)
p

It has been observed in an enormously wide range of flows, from water in narrow
pipes to planetary atmospheres, that when Re is less than about 1()3 the flow is
always dominated by viscosity, whereas it is usually turbulent when Re is larger
(Fig. 7.1 0). In the present case we see that Re for synoptic-scale atmospheric flow is
about 1012 , or 1()6 if we compare synoptic-scale horizontal accelerations with the
boundary-layer-scale viscous terms in eqn (7 .14). In either case the critical value of 193
7 Atmospheric dynamics

(a)

Fig. 7.10 Water flow in o

gloss pipe with dye injected
centrally at the right-hand
end. In laminar flow (o) , the
dye plume's downstream
motion and divergence ore
imperceptible. In foster
turbulent flow (b), the plume
diffuses quickly to the pipe
walls while swirling
downstream (to the left). (b)

103 for the Reynolds number is greatly exceeded and the flow is potentially
turbulent.
Synoptic-scale flow is certainly unsteady, with large weather systems
developing, maturing and dying like enormous eddies. But there is such a gulf of
scale between these huge 'eddies' and the much smaller and unambiguously
turbulent eddies which are so common in the the planetary boundary layer that it is
difficult to say dogmatically that atmospheric flow is truly turbulent on the
synoptic scale. It might be quasi-laminar, as the smoothness of air flow on weather
maps seems to suggest, especially in the middle and upper troposphere: compare
the smooth jet-stream flow in Fig. 7.13 with the clearly turbulent-matter flow in
Fig. 7.10.
There is no doubt, however, that flow on the scale of the planetary boundary
layer itself is enormously turbulent. The gustiness of surface winds and the
looping, ragged dispersal of smoke plumes are ample evidence (Fig. 9.11), and the
value of Re = U h / v - 109 confirms the potential for turbulence. It was noted in
section 7.2 that turbulence in the boundary layer is similar to molecular viscosity in
some important ways, and that it maintains an eddy viscosity which overwhelms
molecular viscosity in almost all realistic atmospheric conditions. It follows that we
194 should use the kinematic coefficient of eddy viscosity K instead of " when
estimating the friction terms omitted from eqn (7.13). Values of K are extremely 7.8 Geostrophic flow
variable but - 10 m2 s - 1 is a reasonably typical value in the presence of large,
efficient eddies. Using this in eqn (7.14) with the original value for iJ2ulaz 2 , we find
that the eddy viscosity term has magnitude I0- 4 m s- 1, which is just one order of
magnitude down on the largest terms in eqns (7 .13a) and (b). This result suggests
that it is at least possible that the turbulent friction terms can be quite significant in
the horizontal components of the equation, as is amply confirmed by observation
(section 9.9).
Note that the Reynolds number is one of several dimensionless numbers we shall
use to describe the importance of terms on the right-hand side of the equation of
motion in comparison with the relative acceleration on the left-hand side. These
and many other dimensionless numbers have been found to be very useful in the
study of fluids, because similar values are found to correspond to dynamically
similar behaviour despite other quite dramatic differences of particular situation.
For example, the flow of glycerol round a spinning immersed ball, and the flow of
air round an identical ball spinning in air, are observed to be similarly laminar
when the Reynolds numbers are similarly small (say about 100), which occurs when
the ball in the glycerine is spinning about 500 times faster, since that is the ratio of
the kinematic viscosities of glycerol and air. These flows are said to be dynamically
similar in this respect, and the existence of a critical value of Re which distinguishes
between necessarily laminar and possibly turbulent flow is obviously a special
example of the more general use of Re as a criterion of dynamic similarity. Other
dimensionless numbers act as criteria for other types of flow.
Such numbers are called dimensionless because in each case the physical
dimensions of mass, length and time (and temperature when it is involved) cancel
out when they are inserted into the expression defining the number in question,
leaving a pure number whose value in any particular situation is therefore
independent of the system of units (SI, CGS, etc.) used. The concept and
elementary use of dimensions is outlined in appendix 7 .1.

7. 8 Geostrophic flow

In the free atmosphere (i.e. above the planetary boundary layer), a great deal of
observational evidence suggests that the friction terms in the equation of motion
are relatively small for synoptic-scale flow. The scale analysis of the last section
therefore suggests that horizontal flow is determined by the following equations:
du ap 1
-=fv---
dt ax p
dv 1 ap (7.16)
- -fu---
dt P ay
where f = 20 sin cJ> is called the Coria/is parameter, and represents the Coriolis
effect arising from the component of the Earth's rotation about the local vertical.
The scale analysis further suggests that the relative accelerations duldt and dvldt
may be an order of magnitude smaller than the Coriolis and pressure-gradient
terms.
We can examine the relative magnitudes of the relative accelerations and
Coriolis terms by setting up the ratio 195
7 Atmospheric dynamics relative acceleration
Coriolis acceleration
which is the second important dimensionless number we consider, and is called the
Rossby number (Ro), after the Swedish meteorologist who pioneered ways of
dealing with the meteorological implications of the Earth's rotation.
Ro = relative acceleration _ U2f L = _!!_
Coriolis acceleration OU OL
When Ro << 1 the Coriolis terms predominate over the relative acceleration terms,
and observations of real and model atmospheres show types of flow characterized
by large, flat vortices about the local vertical. Using the characteristic scales for
synoptic scale behaviour listed in section 7. 7, Ro - 0.1 for synoptic-scale flow,
suggesting that the Coriolis terms tend to predominate but not totally.
Consider what would happen if the Rossby number were zero. Equation 7.16
would then simplify to
fv = _!_ ap fu =---
1 ap (7 .17)
p ax p ay
By definition these equations define what is called geostrophic flow, some of whose
properties we will now consider. The reason for doing this is of course that the
actual behaviour of the Earth's atmosphere is approximately geostrophic in
important respects. The approximation is not nearly so good as the hydrostatic
approximation which dominates the vertical component of the equation of motion
(hence the descriptive term quasi-geostrophic), but it is good enough to be
extremely useful.
In the special case of a purely westerly wind (i.e. positive u and zero v) eqns
(7 .17) simplify to
1 ap ap
-- = -fu -=0
p ay ax
These show that the pressure gradient must be parallel to they axis, with pressure
increasing in the negative y direction (Fig. 7.11 ). Since isobars are perpendicular to
the pressure gradient (just as in maps of terrain height contours are perpendicular
to terrain slope) it follows that the isobars in this case must lie east-west, which is
parallel to the assumed westerly air flow, as shown in Fig. 7 .11. In addition, the
y
fact that the pressure gradient is directly proportional to wind speed means that
lowerp isobar spacing must increase with decreasing wind speed and vice versa, the
___ u constant of proportionality varying with air density and latitude (through f).
The properties of isobaric parallelism and spacing in relation to air flow are in
fact quite independent of the direction of air flow. It is readily shown (appendix
7. 7) that the horizontal wind speed v and the horizontal pressure gradient ap/ an
Higher p
L------X
are connected in magnitude by
Fig. 7.11 Easterly
ap v

r :~ rr
geostrophic flow in a
_!_ = f (7.18)
north-south pressure
gradient (northern
p an
hemisphere).
where :~ = [( :~ + (
and
and that the directional relationship in the northern hemisphere is as shown in
Fig. 7 .12. The latter is summarized verbally in a law named after the Dutch pioneer
196 meteorologist of the 19th century Buys-Ballot which is usually stated in a form such
 Horizontal 7.7 Geostrophic flow
Lowerp
plan
.,,...,,., Isobars
/"'
,..,··"/" Fig. 7.12 Generalized
geostrophic flow (northern
CF Higherp
hemisphere).

as 'low pressure in the northern hemisphere is on the left hand when facing down
wind'. (It is on the right in the southern hemisphere). The law was recognized
empirically, particularly from observations of surface winds blowing round extra-
tropical and tropical cyclones, before it was formally linked to the Earth's rotation
through the equation of motion.
It is customary to rearrange eqn (7.18) in the form

v= _1_ op
pf on
and to use this to define the geostrophic wind V8 as being that which satisfies this
balance in magnitude and direction. By definition of V8 therefore

v8 = - 1- op (7.19)
pf on
It should be clear from this equation and Fig. 7.12 that geostrophic flow can be
regarded as equilibrium between pressure-gradient force and Coriolis force (which
acts perpendicularly to the right of the direction of motion in the northern
hemisphere).
As mentioned in section 4.6 it is conventional to describe horizontal pressure
fields at levels above sea level by plotting contours, in metres above sea level, of a
convenient isobaric surface. Since the vertical distortion of such isobaric surfaces
from the horizontal is always quite small in the lower atmosphere, no significant
confusion of vertical and horizontal gradients arises from this practice. It follows
that the contours of an isobaric surface are almost exactly parallel to the isobars on
a closely adjacent horizontal surface, and that Buys-Ballot's law holds when lower
contours are substituted for lower pressures. It is shown in appendix 7.5 that the
replacement of op!on and oZP/on (the contour slope on thep isobaric surface) very
conveniently removes the variable p from eqn (7.19}, giving
g ozp
v =-- (7.20)
8 f on
The removal of p and the near uniformity of g means that at any particular latitude
(and hence value of/) the same contour slope corresponds to the same geostrophic
wind speed, regardless of whether we are dealing with observations near sea level or
in the stratosphere where air density may be 20 times smaller.
Let us use eqns (7 .20) and (7 .19) to estimate isobaric slopes and horizontal
pressure gradients geostrophically associated with typical actual winds in middle
latitudes, where/= 10- 4 rad s- 1• For a wind speed of 10m s- 1 the isobaric slope
is about 1 in 104, and slopes seldom exceed 8 in 104 even in the most powerful jet
stream. The pressure gradient corresponding to 10m s- 1 in the low troposphere
(where air density is about 1 kg m - 3} is about 1 mbar per 100 km, but the pressure
gradient corresponding to the jet stream is only about double this value because the
increased contour gradient is largely offset by the reduced air density in the high
troposphere. The greater simplicity of using isobaric contours is obvious.
The accuracy of the geostrophic approximation in any particular real situation is 197
7 Atmospheric dynamics

Key
660/ C ontour height above MSL
Fig. 7.13 Observed quasi- / in decametres
geostraphic flow at the
300 mbar level over the 1...-o 15knot Westerly
British Isles and western ~ 65 kno t No rthea sterly
Europe at 0000 Z, I January
1982.

apparent from the extent to which actual winds resemble geostrophic winds in both
strength (eqn (7.20)) and direction (Fig. 7.12). Typical correspondence between
actual and geostrophic winds can be judged from Fig. 7 .13, which shows isobaric
contours and actual winds in the vicinity of a polar-front jet stream over the British
Isles. The data are taken from synoptic radiosonde ascents on the occasion, and the
contours sketched by interpolation between the point measurements of the height
of the 300 mbar surface. The wind vectors, measured from the horizontal drift of
the radiosondes as they rose through the 300 mbar surface, are parallel to the
isobaric contours to within 20° everywhere and 10° over most of the area. The
actual wind speeds agree with the geostrophic speeds calculated from the contour
slopes using eqn (7 .20) to within 200Jo everywhere and 10% in most places. This is
typical of the extent to which actual winds are geostrophic provided we avoid the
planetary boundary layer, and the equator and its immediate vicinity.

7.9 Why geostrophic?

Before we consider any further aspects of geostrophic equilibrium, note what

198 strange behaviour it embodies. Air, instead of flowing horizontally down the
pressure gradients in response to the horizontal pressure gradient force, flows at 7. 9 Why geostrophic?
right angles to this apparently obvious direction. In the vicinity of a pressure
minimum like those defining the centres of depressions, the air, instead of flowing
directly inwards to fill up the low pressure, flows round the low, forming huge flat-
tened vortices turning anticlockwise on the weather maps of the northern hemi-
sphere, and clockwise in the southern hemisphere. Indeed this majestic wheel-like
rotation gave rise to the generic name cyclone, from the Greek for wheel.
Note too that we have not in any meaningful sense explained why actual winds
should be quasi-geostrophic. The scale analysis of the equations of motion, and the
estimation of the Ross by number, are consistent in showing that the geostrophic
balance follows from the comparative insignificance of the relative accelerations
and frictional terms, but they do not begin to explain why these terms should be so
small.
Rewriting the Ross by number in terms of time scale t we find
Ro- 1/flt
from which it follows that Ro is less than 0.1 provided t exceeds about one day, and
that this length of time is set by the Earth's rate of rotation. This is indeed the
observed timescale of synoptic-scale weather systems, but it is very hard to say why
this should be so. It is obviously associated with the relative quietness of the
atmosphere on the synoptic scale: depressions take days rather than hours to form
and die, and jet streams though strong are very definitely subsonic in speed.
Airflow therefore has time to settle into geostrophic equilibrium with pressure
fields and vice versa. And it seems intuitively reasonable to link this quietness in
turn with the gentleness of the solar input: if this input were ten times larger than it
is, then the atmospheric motion would be much more hectic, and the tendency
toward geostrophic equilibrium would be continually frustrated by the eruption of
tremendous cumulonimbus and the like.
You can consider the onset of geostrophic flow for an individual air parcel by
envisaging it starting very unrealistically from rest in the presence of a synoptic-
scale horizontal pressure field (Fig. 7.14). After initial motion toward low
pressure, it will veer to the right (in the northern hemisphere) in response to the
growing Coriolis force, or equivalently in response to the hidden anticlockwise
rotation of the weather map. If growing friction critically damps the parcel's
tendency to overshoot and undershoot, it will reach geostrophic equilibrium as
shown, with the parcel moving parallel to the isobars (or isobatic contours) at the
speed required by eqns (7 .19) or (7 .20). The parcel will obviously take a finite time
to reach this state - a time which is related to the magnitude of the Coriolis
parameter and hence to the Earth's rate of rotation. If we consider a situation with
a much shorter intrinsic timescale - for example air flowing at 10 m s -t in the
vicinity of a hill of horizontal scale 10 km (t - 103 s)- then it is clear that there is
not enough time to develop geostrophic equilibrium between the flow and the
pressure field associated with the obstructing effect of the hill, so that flow round
and over the hill cannot be expected to be even quasi-geostrophic, as is confirmed
by substituting the time or space scales for the flow in the equivalent versions of the
Rossby number. This points to the important fact that the geostrophic tendency

-----------P CJ rotation Fig. 7.14 Idealized approach

about local to geostrophic equilibrium
---~--P+5 vertical
from rest (northern
hemisphere).

_ _[__ _ _ _ _ _ _ _ p+10 199

7 Atmospheric dynamics appears only on the large mesoscale and on synoptic and larger scales of atmo-
spheric flow.
Such discussion is suggestive rather than persuasive, since the arguments are in
fact largely circular. A sophisticated review of atmospheric geostrophy in a
technical journal concluded that 'it seems unlikely that the near geostrophic
balance in our atmosphere can be accounted for in any simple manner' [43]. This is
an intriguing state of affairs, given the widespread and important consequences of
the geostrophic tendency, one of the most important of which we will now
consider.

7.10 Thermal winds

We have seen that geostrophic wind speed is proportional to isobaric contour

slope, and that geostrophic wind direction is parallel to isobaric contours on a hori-
zontal plan. It follows that if the contour slope varies consistently with altitude, so
must the geostrophic wind. Figure 7.15 depicts a vertical section through a part of
the troposphere where there is a consistent increase in contour slope with altitude
and an associated increase in geostrophic wind speed. For simplicity the situation is
two dimensional, in the sense that there is no change of the direction of contour
slope and wind with height slope and wind direction with height, but all relation-
ships discussed in the following have quite obvious vector equivalents when
directions vary with height.
Applying eqn (7 .20) to the isobaric slopes at pressures PI and p 2 (the higher
pressure PI corresponding to the lower altitude), we find the difference in the geo-
strophic wind speed directly by subtracting the equations for the two surfaces.

vg2 -
v =
gi
!_[
f an
azp, _ azp,
an
J
=
g
f
a [Zp, -
an ZP,J (7.21)

The right-hand side of eqn (7.21) contains the thickness gradient

ana [Zp, - zp,]

which is simply the horizontal gradient of the vertical separation of the PI and p 2
isobaric surfaces. In Fig. 7.15 the thickness of the layer between the 1000 and

-~-&r-~.- "'
Fig. 7.15 Vertical cross·
section illustrating the Q)

thermal wind relation. Tails ::>

of geostrophic wind arrows .c

01
"'"'
Q)
)>
are enclosed by circles whose
--------®--- s: 0.
Q)
.t:: () 500
diameters increase with wind JJ 01
01
c: 0 c:
speed (northern hemisphere).
"' 0 "'
:'"':' r :'"':'
()
()
Q)
~ 1000 Q
200
300 mbar surfaces increases sharply to the right, and the magnitude of its gradient 7.10 Thermal winds
could be found by subtracting the gradient of the lower surface from the gradient
of the upper.
The thickness of a layer bounded by any two isobaric surfaces is proportional to
the mean temperature of the layer (section 4.6) and it follows that the difference in
geostrophic wind speed across such a layer is proportional to the horizontal
gradient of the layer mean temperature. The presence of a horizontal gradient of
layer mean temperature in an isobarically bounded layer means that the layer must
be baroclinic, and the strength of the gradient is a measure of the degree of baro-
clinity. The thermal wind relation is therefore a relation between geostrophic wind
shear and baroclinity, and hence highly relevant to the great baroclinic systems of
the lower atmosphere, especially the troposphere.
Bearing in mind the smallness of realistic isobaric slopes in comparison with the
slopes of isotherms (the ratio can be as large as 100 in a frontal zone, and is usually
at least 10 - sections 4.6 and 11.2) the temperature gradient of an isobarically
bounded layer is effectively a temperature gradient of a horizontal layer. Equation
(7 .21) therefore associates a vertical shear of (horizontal) geostrophic wind speed
through a layer with a nearly horizontal temperature gradient across the layer.
Figure 7.15 shows that the temperature gradient lies along an axis which is perpen-
dicular to the wind shear - perpendicularity being the hallmark of the Coriolis
effect. It is apparent in Fig. 7.15 that an extension of Buys-Ballot's law connects
the directions of vertical shear and nearly horizontal temperature gradient: in the
northern hemisphere low mean temperature is on the left when standing with back
to the geostrophic wind shear. The very important relationship described by eqn
(7.21) and Fig. 7.15 is known as the thermal wind relation because it clearly
connects the distributions of temperature and geostrophic wind.
It follows from the presence of hydrostatic equilibrium that the connection
between layer thickness and layer mean temperature is such (appendix 7 .9) that
vg, - vg, _g_ aTP (7.22)
Zp,- Zp, fT an
which is the finite difference form of the thermal wind equation. The differential
form follows when we consider a very thin layer of air (appendix 7.9 again)
avg = _g_ aTP (7.23a)
az fT an
Since the thermal wind relation is simply a consequence of geostrophic and
hydrostatic equilibria, it will be observed in actual wind and temperature fields to
the extent that these equilibria apply. Hydrostatic equilibrium holds so very
accurately on the synoptic scale (section 7. 7) that we should expect the real atmo-
sphere to obey the thermal-wind relation to the same extent that it is geostrophic,
and this is observed to be the case on scales ranging from those of fronts to the
hemispheres. Details of this behaviour are given at many points in the book, so
only two of the most important examples are very simply outlined in Fig. 7 .16.

APPLICATIONS

In Fig. 7 .16(a) there is a section of the circumpolar vortex of westerly winds

which dominates the troposphere in middle and high latitudes (section 4. 7). The
thermal-wind relationship appears in the conjunction of the increase of westerly 201
7 Atmospheric dynamics

·Fig. 7.16 (a) The circumpolar

vortex in section, with
westerly winds represented
by arrow heads and tails, and
temperature gradients

.-r------,
(b)

t9\
between low and high
latitudes by exaggerated
thickness profiles (northern
hemisphere). -~
'e:J
''
~ cool ....._ warm
(b) A polar front in section,
with jet-stream arrow tail, Horizontal
zone of strongest horizontal
temperature gradient (dashed
line), and tropopauses (heavy
lines)(northern hemisphere).
lcl
" K~+
I!.Vg
(c) A close-up of the foot of
the cold front in (b). showing T
" "-,
-...
J).z " '-
'-
T+ J).T

the zone's finite horizontal "- C.n

cool Vg '- warm
breadth <ln and vertical
thickness <lz.

winds with height and the meridional temperature gradient imposed by the sun's
unequal input. The level of maximum westerly winds is associated through the
thermal-wind relation with the top of the layer with consistent poleward lapse of
temperature. The westerly sense of the shear is consistent with the thermal
extension of Buys-Ballot's law, just as the easterly shear (westerlies decreasing with
increasing height) above is associated with an equatorward temperature lapse.
In Fig. 7 .16(b) there is a section through an active front in middle latitudes - a
type of structure described in some detail in Chapter 11. The conjunction of the
wind shear up to the high winds of the jet core and the sharp lateral temperature
gradient which is the basic feature of a well-defined front is obviously consistent
with the thermal-wind relation. The thermal extension of Buys-Ballot's law
requires that in the northern hemisphere cold air lies to the left of the jet core,
looking down both wind and wind shear, as is always the case. In fronts there is
often a significant change of wind direction between the low and high troposphere.
According to the vector form of the thermal-wind relation which is derived in more
advanced texts, the horizontal temperature gradient is perpendicular to the vector
difference between the low- and high-level winds.
Equation 7 .23a can be applied in finite-difference form to a simple model of a
sloping frontal zone (Fig. 7 .16c) with horizontal width lin
liVg _ g liT
..:lz - fT lin (7.23b)

and rearranged to give an expression for the frontal slope liz/ lin:
..:lz
-=----"-
fT li vg
(7 .23c)
lin g liT
This shows that a certain increase in geostrophic wind speed across the frontal zone
is needed to counterbalance the tendency of the cold air increasingly to undercut
the warm air, and hence maintain the frontal slope. Inserting realistic values of
30 m s - t for li Vg, 5 oc for liT and typical values for /and T, we find the associated
frontal slope to be 1:60 in middle latitudes, tending to zero at the Equator, where as
usual geostrophic balance is impossible.
202 It should be clear by now that the importance of the thermal-wind relation for
the state of atmosphere can hardly be overestimated. However abstract eqns 7.11 Geostrophic
(7.21)-(7.23) may appear, they describe behaviour which dominates much of the departures
large-scale behaviour of the troposphere. As final reinforcement of this fact, it
appears from advanced dynamical analysis [44] that it is the combination of
vertical shear of wind and lateral temperature gradient which is baroclinically
unstable and which gives rise to the greater weather systems of middle latitudes
outlined in Chapter 11.

7.11 Geostrophic departures

Differences between actual and geostrophic winds are often apparently systematic
on the synoptic scale. Observations of winds in the entrance region of the core of
polar-front jet streams show that there is a systematic tendency for winds to be
supergeostrophic (slightly faster than geostrophic) and to be angled somewhat
across the isobaric contours towards lower levels. The opposite tendency appears
in jet exit regions. Such tendencies appear in other cases where synoptic-scale flow
is similarly subject to linear acceleration (i.e. acceleration effectively along the
wind direction). The dynamical sketch in Fig. 7.17 depicts the reason for this
behaviour. To provide for the linear acceleration towards the jet maximum, the
Coriolis force must exceed the pressure gradient force and be angled forward.
From the vector relation between Coriolis force and wind velocity it follows that
the wind speed must exceed the geostrophic wind speed (i.e. the speed which will
exactly balance the contour gradient through geostrophic equilibrium), and be
directed so that there is a component down the contour slope.
Fig. 7.17 Plan view of jet
entrance region with isobaric
contours, pressure gradient
and Coriolis forces, and
typical ageostrophic flow
(northern hemisphere).

Significant geostrophic departures also arise from lateral accelerations (i.e.

accelerations effectively perpendicular to the wind direction) such as those
associated with cyclonic and anticyclonic rotation of air. Figure 7.18 contains a
dynamic sketch of cyclonic flow of air round a horizontal circular path of radius R

Cyclonic flow Fig. 7.18 Plan views of

cyclonic and anticyclonic flow,
with pressure gradient and
Coriolis forces (northern
hemisphere).

Anticyclonic
flow

203
7 Atmospheric dynamics in the northern hemisphere. It is clear that to maintain the centripetal acceleration
toward the cyclonic centre, the wind must have speed V such that the pressure
gradient force exceeds the Coriolis force by V 2 /R. It is apparent that this can be
maintained only if the Corio lis force is less than the pressure gradient force, which
means that the wind must be subgeostrophic. Examination of the balance of radial
forces and acceleration shows

V2 = _!__ ap - J v (7.24)
R p aR

Since the pressure gradient can always be replaced by the Coriolis term involving
the equivalent geostrophic wind,/ Vg, we have
V2
R =f(Vg- V) (7 .25)

confirming that the balancing wind speed is subgeostrophic. Winds satisfying this
equilibrium are called gradient winds. Expressed as a fraction of the gradient wind
speed this geostrophic departure is
v-
g
v= v (7.26)
V Rf
which is a Ross by number for the laterally accelerating flow. Observations suggest
that deficits of actual winds below geostrophic winds often exceed 2007o in vigorous
extratropical cyclones.
Figure 7.18 also depicts anticyclonic flow in the northern hemisphere, and it is
apparent by the same line of reasoning as above that wind speeds in these cases
should be supergeostrophic, as is usually observed. In fact it is shown in appendix
7.10 that if the equivalent of eqn (7 .24) is solved to find an expression for the
gradient wind speed V corresponding to any particular geostrophic wind speed Vg
(and therefore to any particular pressure gradient), then Vg must be no greater than
Rfl 4, which implies that for any latitude and radius of trajectory curvature R there
is a maximum pressure gradient which is proportional to R. Although in fact
trajectory and isobaric curvatures are not identical (see next paragraph) there
clearly must be a strong connection, and the existence of such an upper limit to the
pressure gradient therefore suggests that pressure gradients should die away
toward the centre of an anticyclone, as is observed. No similar constraint applies in
theory or observation to any kind of cyclonic flow.
Although this simple analysis is instructive and beloved of textbooks, it is
actually very difficult to check by observation how close actual winds are to
gradient equilibrium values. The reason for this is that it is very difficult to observe
actual paths (trajectories) traced out by moving air parcels. The streamlines of flow
which we can produce by careful analysis of observed winds on weather maps may
be quite misleading in this respect for at least two important reasons. First, the flow
is usually fairly unsteady, so that streamlines are incessantly wriggling. Even so,
trajectories could in principle still be established by connecting up a succession of
streamline segments from a series of maps at consecutive observation times, if it
were not for the intervention of a second reason. Flow almost always has a
component of vertical motion which is strong enough to ensure that the actual
trajectory curvature of the rising or falling air can differ quite significantly from
the curvature deduced by assuming horizontal or isobaric motion. The same uncer-
tainty prevents accurate assessment of the balance of linear accelerations and geo-
strophic departures in situations such as Fig. 7.13: the actual linear acceleration of
204 a parcel rising through the strong wind shear typical of a jet entrance may be signi-
 7.11 Geostrophic
departures

Fig. 7.19 Synoptic surface

winds and mean sea-level
pressures over the British
Isles at 0000 Z, 10 January
1969, showing winds tending
to blow obl iquely towards low
pressure, most obviously in
lightest winds . The map
shows south-easterly flow
east of the occluded front of
an old low, with a weak
stagnant front between
northern Britain and
Key Scandinavia - a typical non-
MSL pressure in mbar @ calm standard situation! (After the
S 'ly 2kt b N ' ly 5kt (' SW'Iy 15kt Daily Weather Report of the
British Meteorological Office
for that date)

ficantly different from what appears on a standard isobaric chart. These are part of
a very general problem of analysis of air motion discussed in section 11.4.
Another important source of geostrophic departure arises from the enhanced
turbulent friction in the planetary boundary layer. Figure 7.19 represents a typical
state of affairs near the surface: winds observed at the standard observing height
(10m above the surface) are consistently angled at between 10° and 30° across the
isobars, varying with wind speed and local geography, and may have speeds which
are less than 30% of geostrophic values. The situation is outlined in some detail in
Chapter 9, but for the moment the role of frictional drag can be seen by considering
the three-way equilibrium between Coriolis, pressure-gradient and frictional
forces (Fig. 7 .20). The angular deviation and reduced strength of actual winds near
the surface are consistent with the expected opposition of friction to motion over
the surface, the rate of working by the pressure gradient on the parcel as it moves
toward lower surface pressures being equal to the rate of working against frictional
drag.
PGF
1000 mbar
for example Fig. 7.20 Horizontal balance
of pressure gradient, Coriolis
IT Horizonta l
F(f'~ plan and friction forces in the

'~
'''J.. .. CF
atmospheric boundary layer,
showing ageostrophic flow
1002 across the isobar s (northern
hemi sphere) .

The consistent motion of air across isobars has obvious implications for systems
having closed isobars (Fig. 7 .21). Friction obviously induces a tendency for low-
pressure centres to fill and high-pressure centres to empty - each of which is a 205
 00
7 Atmospheric dynamics

Fig. 7.21 Frictionally

induced convergence and
divergence in the boundary
layers of cyclones and
anticyclones (northern
hemisphere).

significant aspect in the decay of synoptic scale weather systems (Chapter 11 ). The
effect is quite strong near the surface, but is limited largely to the planetary
boundary layer and is seldom significant more than about 500 m above the surface.
Geostrophic equilibrium as defined by eqns (7 .16) is clearly impossible at the
equator since the Corio lis parameter f is zero there. Very close to the equator, but
not at it, the Coriolis terms are observed to be so small at realistic wind speeds that
large-scale flow is not even quasi-geostrophic. This has important practical con-
sequences for weather forecasters in the equatorial regions: there is little point in
using the horizontal pressure (or isobaric contour) fields to consolidate the wind
fields, as is standard over the rest of the globe, since the two are not usefully geo-
strophically related. Synoptic-scale flow is therefore analysed by drawing stream-
lines through the observed winds. In fact away from the occasional unusually
vigorous weather system the pressure gradients apparent on weather maps in those
regions are much weaker than at higher latitudes, and are often dominated by a
diurnal tide driven by the Sun's heating. As latitude increases, the local tendency to
geostrophy increases, so that for example in the weather systems which thrive on
the border between the intertropical convergence zone and the trade winds, there is
a fairly obvious relation between pressure and wind fields (section 12.3). At least
this is so in the northern hemisphere where this border is some distance from the
equator.

Fig. 7.22 Circumpolar flows of water in a dish pan operating in Rossby-number regimes similar
to those in the upper troposphere of (a) middle and (b) low latitudes in the northern hemisphere.
206 (From [45])
 More fundamentally, the increasing irrelevance of geostrophic equilibrium as 7.12 Small-scale motron
the equator is approached, apparent in the increasing Rossby number Ul Lf
(expressed in a form which recognizes the special importance of the component of
0 about the local vertical), should lead us to expect atmospheric flow observed in
low latitudes to differ in type from that prevailing in middle and high latitudes.
Although the detailed dynamical reasoning is far beyond the introductory level, it
should seem at least plausible that the increase in this Ross by number with latitude
is closely related to the striking difference between the Hadley circulation in low
latitudes and the continually contorting circumpolar vortex of middle and high
latitudes. This is nicely confirmed by experiments with a laboratory-scale analogue
of the atmosphere rather incongruously called the dish-pan. An annulus of water is
trapped between a cold central pole and a warm equator (Fig. 7 .22) in an open-
topped pan rotating about an axis through the pole. This represents a flat Earth
with only a single value for the Coriolis parameter (which remember is twice the
component of rotation about the local vertical). Experiments with relatively slow
rates of rotation therefore tend to simulate terrestrial behaviour at low latitudes,
provided a special form of the Ross by number is arranged to have a realistic value,
and produce Hadley circulations, whereas very realistically waved and slowly
eddying circumpolar flow is observed at faster rotation rates.

7 .12 Small-scale motion

SCALE ANALYSIS

Consider air flowing in the vicinity of a moderate-sized hill or shower cloud. The
horizontal scale Lis - 10 km, which is effectively the same as the vertical scale set
by the depth of the troposphere. This symmetry of scale is very different from the
extremely flattened configuration of flow on larger scales. Horizontal wind speeds
are by and large set by the synoptic-scale situation, so that 10 m s -I is again typical
for U. Vertical wind speeds are about an order of magnitude smaller, so that there
is some asymmetry but very much less than on larger scales. Timescales for air
being in the influence of such a system range from - 103 s for air passing hori-
zontally through the flow over a hill, to as much as an order of magnitude larger
for air rising through a shower cloud (cumulonimbus). Horizontal pressure
variations are observed by especially dense networks of sensitive barographs to
be - 1 mbar.
A list of basic and secondary scales is as follows, others being the same as in the
list for synoptic scales (section 7. 7):
Horizontal and
vertical scale length L 10 km 10" m
Minimum timescale t 103 s
Horizontal pressure change p 1 mbar 102 Pa
Horizontal wind speed u IOms-'
Vertical wind speed w 5 m s- 1
Horizontal acceleration U/t 10-2m s-2
Vertical acceleration wit 10- 3 m s- 2
Horizontal pressure gradient p!L 10- 2 Pa m- 1 207
7 Atmospheric dynamics Although the turbulent friction is difficult to scale even roughly, we can get some
notion of its maximum magnitude by using KUIV together with a K value com-
parable with what was used previously for the planetary boundary layer ( l 0 m2 s -I
-see section 7.7), but increased by an order of magnitude to offset the probable
error in using L rather than some smaller subscale on which friction is especially
effective (like the depth of the turbulent boundary layer). This probably over-
estimates friction in flow in the free atmosphere around a hill, but may not be too
high for the interior of a cumulonimbus, since this is kept very turbulent by
convection. The resulting value for the frictional term (see below) shows that
friction is relatively unimportant on the chosen scale, but becomes rapidly more
important with decreasing scale.
The magnitudes of terms in the components of the equation of motion
(eqn(7 .12)) are therefore as follows:

~ = 20 sin¢ v - 20 cos ct> w - -

1
-
ap +F (7.26a)
dt P ax X

ro- 2 ro- 3 s x ro- 4 ro- 2 10 5

dv
-20 sin ct> u - __!__ ap + F (7.26b)
dt P ay y
ro-2 ro- 3 w-z w-5
dw __!__ ap - F
- = 20 cos ct> u - g -
az (7 .26c)
dt p z
ro- 3 ro- 3 w w ro- 5

HORIZONTAL MOTION

In the horizontal, the balance is mainly between acceleration and pressure

gradient, with a marginally significant Coriolis effect. There is therefore little or
none of the geostrophic tendency which so dominates flow on larger scales, as is
nicely indicated by the much larger value of the Rossby number

Ro = -u- or
1
---10
OL Ot

which is about two orders of magnitude too large for quasi-geostrophic balance.
The time period for air parcels in the influence of such small systems is so short that
the Earth's rotation is largely irrelevant.
Local accelerations and pressure gradients are opposed, as shown by the
opposing signs of the dominant terms in eqns (7.26a) and (b), so that air under-
going a linear acceleration tends to experience falling pressure, and decelerating air
experiences rising pressure. For example, as air impinges on the bluff side of a hill
and decelerates into a stagnation zone on the upwind side of the hill (Fig. 7 .23), its
pressure rises even if it moves absolutely horizontally, with the result that the
pressure of the whole zone is raised slightly. Pressure is also raised a little in the lee
of the hill (where the pressure gradient force acting away from the hill is associated
Fig. 7.23 Air impinging on o
hill and inducing raised
pressures in the stagnation
zones immediately to High Pressure High Pressure
windward and leeward.
'
/

-\ - - - t . , - 1 -
Hill profile ------/
208
with the acceleration of air out of the stagnation zone there), but usually by less 7.12 Small-scale motion
than in the upwind stagnation zone because of the complex influence of friction,
with the result that there is a net downwind drag of the wind on the mountain, and
of course an equal and opposite drag of the mountain on the wind.
Sometimes, as on larger scales, there are local accelerations which are per-
pendicular to the local wind, as air flows round tightly curved trajectories (round
the sides of an isolated hill for example), and there is then a near balance between
lateral pressure gradient force and centripetal acceleration. Such flow is termed
cyclostrophic, and it applies to a wide range of curved flows centred on our chosen
I 0 km scale: on a smaller scale it applies to air in the vicinity of tornadoes (section
10.6 and Fig. 7 .24) and dust devils, and on a larger scale it applies to air in the
active annulus of high winds in a hurricane (section I2.4). The equation for cyclo-
strophic balance in air moving at tangential speed V around a circular path of
radius R is simply
vz ap
R p aR
Air near the funnel of a tornado often blows at speeds well in excess of 50 m s- 1
around paths with radii of curvature of about IOO m, which indicate very large
centripetal accelerations (- 25 m s- 2). Corresponding pressure gradients are
therefore - 25 mbar per I 00 m at least, and pressures in the funnel cores are
depressed by many tens of millibars below ambient pressures. As mentioned in
Section 10.6, this heavily distorted pressure field has consequences which are at
once visually striking and potentially very hazardous.
Conditions in most systems in the small-scale range are of course much less
extreme than in tornadoes, but there are nevertheless tendencies toward linear and
lateral accelerations with nearly balancing pressure gradients which account for the
dimpling of the small-scale pressure field which is observed by special arrays of
sensitive barographs.
Note that the cos ¢ Corio lis term in eqn (7 .26a), which was negligible on the

Fig. 7.24 A dust devil- a

dusty vortex induced by
intense local horizontal
convergence at the base of a
powerful thermal. They are
especially common in dry
seasons in tropical regions
during the heat of the day.

209
7 Atmospheric dynamics synoptic scale, is now nearly comparable with the familiar sin (jJ term, because of
the near equality of vertical and horizontal wind speeds. It is not clear whether or
not the additional term plays a significant role in cumulonimbus, for example,
because it is extremely difficult to make detailed observations of winds inside their
turbulent, cloudy and precipitation-streaked interiors, but the form of eqn (7 .26a)
would suggest that there should be an eastward tendency in updraughts and a west-
ward one in downdraughts.

VERTICAL MOTION

Vertical motion on the small scale is still dominated by hydrostatic balance between
the vertical pressure gradient and gravity, their terms being four orders of
magnitude larger than the next largest terms (vertical acceleration and Coriolis).
However, much more rapid vertical accelerations are known to occur at the
bottoms and tops of vigorous cumulonimbus, for example, in which the length and
timescales of accelerating and decelerating air are locally each an order of
magnitude smaller than the listed values. For example cold air pours downwards in
shafts of heavy precipitation in cumulonimbus (sections 10.4 and 10.6), and may
impinge on the underlying surface so vigorously that the downdraught (- w)
reduces from 20 m s -I to zero in the kilometre between cloud base and the surface.
The corresponding value for dwldt is - I0- 2 m s- 2 , which is 0.1o/o of g and is
associated with a corresponding distortion of the hydrostatically expected value of
surface pressure (a rise - 0.1 mbar). Transient pressure 'footprints' of this and
slightly larger orders of magnitude are observed by arrays of micro barographs as
big cumulonimbus pass over, though their explanation is complicated by the simul-
taneous presence of adjacent masses of thermally contrasting air.
This is a rather extreme example of deviation from hydrostatic equilibrium; in
convection in particular it is much more common to have an almost immeasurably
small and very localized deviation which would be seem to be almost trivial except
that it is associated with the very obvious and important production ofupdraughts
and cloud.
The ambient air in which the convecting parcel or parcels are about to move can
be considered to be in hydrostatic equilibrium, with only the convecting parcel out
of balance. The ambient vertical pressure gradient is therefore related to the
ambient air density by
iJp= -gp '
-
iJz
Substitution in the pressure gradient force term of eqn (7 .26c) gives gp 'I p for the
vertical pressure gradient force acting on the convecting parcel, where p is the
density of the convecting parcel, which is slightly different from the density p' of
the environment. The upward pressure gradient force and the downward gravita-
tional force on the convecting parcel can be put together to produce the net upward
force.
p' (p' - p)
g-- g =g = gB (7.27)
p p

This deviation from hydrostatic balance is more familiarly known as the buoyant
force on the parcel, and in fact eqn (7 .27) can be derived very easily from
210 Archimedes' principle (appendix 7.11). The term in brackets contains the crucial
density difference between the parcel and its environment which is the source of the 7.12 Small-scale motion
net buoyant force, and is known as the buoyancy B. When the parcel is less dense
than the surrounding air, then the buoyant force is positive and tends to produce an
upward acceleration. In the atmosphere the density deficit of a buoyant parcel
arises most often because it is warmer than its surroundings, though extra humidity
can assist significantly in some circumstances, and extra cloud burden can detract
somewhat (section 4.1).
The familiar dynamic effects of buoyancy are as described above, but in any
particular situation the direction and magnitude of the associated acceleration
depends on what other forces are acting in addition, and eqn (7 .26) shows that
these can be the Coriolis and frictional forces. To see whether acceleration and
buoyancy are closely matched, or whether there are other significant items in the
dynamic balance, we can compare vertical acceleration with net buoyant force by
examining the dimensionless ratio known as the Froude number Fr:

Fr = acceleration = dw l(gB) _ ~
buoyant term dt gBL
Strictly speaking this is the internal Froude number, to distinguish it from the
original number used by the pioneer fluid dynamicist Froude in modelling ship
wakes and waves. It is another important dimensionless number to rank with the
Reynolds and Rossby numbers, and describes the relative importance of gravity in
the dynamical balance.
When Fr is much less than unity in the present context, the buoyancy is much
larger than the vertical acceleration, showing that other factors as well as buoyancy
must be significant in determining acceleration. Values for B in cumulus are
surprisingly small (- 113oo} on account of the very small temperature excesses in
rising air, and those for cumulonimbus are only a little larger (section 10.4). It
follows from such values and observed strengths and dimensions of updraughts
that Fr values in atmospheric convection cover a considerable range centred on
about 0.2. It is clear therefore that there are forces at work significantly offsetting
the expected effects of buoyancy.
Convection is visibly full of turbulent friction, however difficult it may be to
devise a detailed mechanistic framework (Fig. 1.4). Cumulus of all sizes clearly
must encounter considerable resistance as they push through their cloudless
surroundings. If the buoyant parcels were rigid, the drag they encounter could be
modelled in wind tunnels or water flows provided we arranged the details to ensure
dynamical similarity with the original, i.e. similar values for the Reynolds and
Froude numbers. Strain gauges could be used to measure the resulting drag. But
the buoyant masses are not rigid of course; they are barely distinguishable from
their surroundings, and they deform and tumble as they rise, in ways which must
considerably alter the drag from the equivalent rigid body value and make it
impossible to use strain gauges. Some laboratory and theoretical models have been
used to show that the drag is very substantial, but the work is incomplete and
beyond the scope of an introductory text [46].
Simple observation of cumulus also shows that there is another dynamically
significant process at work in convection. Cloudy air expands very noticeably as it
rises, showing that the buoyant, cloudy air is incorporating air from its immediate
environment at a very considerable rate (Fig. 7 .25). Such entrainment is a basic
property of all vertical convection, from small thermals to large cumulonimbus
(section 10.3), and its presence means that we must modify the equation of motion
to allow for variation of air-parcel mass with time. This is done by returning to the
original, general form of Newton's second law of motion (eqn (7 .1)), whose
vertical component can be written simply as 211
7 Atmospheric dynamics

Fig. 7.25 A cloudy turret

rising out of a parent cumulus
congestus, dissipating quickly
as a resu It of rapid
entrainment of unsaturated
air.

d
-(Mw) = NF (7 .28)
dt z

where NFz represents the resultant of all vertical components of force on the parcel.
The left-hand side of eqn (7 .28) can be expanded to make explicit the effects of
variations in parcel mass M:
dw+ w dM
M- ---_ NF (7.29)
dt dt z

The first term is the familiar product of mass and acceleration. The second term
represents the rate of increase of momentum which is accounted for in the rate of
increase of the parcel mass rather than acceleration. If in any example we know the
strength of the net force NFzand observe only the upward acceleration of the parcel
(not its growth by entrainment) we will miss the production of momentum as large
masses of the initially static environment are set in motion through the process of
entrainment. In fact as usual an effect ignored on the left-hand side of the equation
of motion will seem like a mysterious force on the right-hand side, in this case a
force logically called entrainment drag. Observations and some theory suggest that
the obvious frictional drag and the less obvious entrainment drag are of
comparable importance in convective dynamics, and that they offset the buoyant
forces to such an extent that actual updraught speeds are only a tiny fraction of
what they would be in a frictionless and non-entraining thermal. Only large
cumulonimbus seem to have the capacity to operate considerably more freely than
this (section 10.6).

7.13 Compressing and deforming air

CONTINUITY

Air, being fluid, deforms readily when it impinges on other bodies. Sometimes
212 these are rigid material, such as mountains, or nearly rigid, such as the sea surface,
but most often they are other bodies of air. Although the emphasis so far in this 7.13 Compressing and
deforming air
chapter has been on the air parcel considered as an isolated, almost solid packet, we
must now come to terms with the essential fluidity of the air, and recognize also the
compressibility which is so much in evidence in the vertical distribution of its mass.
The expansion of air when rising and its contraction when sinking are large even
in the relatively shallow depth of the troposphere, because air is easily compressed.
Compressibility may suggest elasticity, but it is important to note that although
sound waves and the continual small pressure adjustments which travel at the speed
of sound (section 4.3) depend crucially on the elasticity of air (its ability to spring
outwards after compression), all other atmospheric dynamics related to weather
operate as if the air had no elasticity, even though it is quite obviously highly com-
pressible and deformable. Elasticity is unimportant because the flux of solar
energy through the atmosphere is not so high as to encourage near sonic
updraughts and jet streams, and this is the same moderation which allows the
atmosphere to be so nearly hydrostatic in its vertical distribution of mass. For
example air sinking towards the surface in the downdraught of a large cumulo-
nimbus (section 10.4) compresses substantially as its ambient pressure rises, but it
nevertheless impinges on the surface so gently that there is no perceptible tendency
to compress further and rebound upward by re-expansion. Let us look more closely
at the compression and deformation of air involved in synoptic and small-scale
weather systems, as defined in sections 7. 7 and 7 .12.
We can derive an expression which is very useful for this discussion by assuming
that the initial mass of any body of air is conserved no matter how it may subse-
quently deform, expand, combine or divide. The equation expressing the conserva-
tion of mass in flowing air is called the continuity equation because it is most easily
derived by considering the continuity of mass flux as air blows through a volume
fixed in space, and its derivation is outlined in Appendix 7.12:

-op = - [- a (pu) + -a (pv) + -a J

(pw) (7.30)
at ax ay az
The left-hand side of this equation is the rate of change of density at a fixed
position, known as the local rate of change. The terms in brackets on the right-
hand side describe the net flux of mass out of the fixed volume in question arising
from longitudinal gradients of mass flux. There is a component of this mass flux
divergence arising from each coordinate axis, as depicted in Fig. 7 .26. A
component (arising from the x axis for example) is positive if the mass flux in the x
direction increases with x, because in that case the flux into the near side of a per-
pendicular cube fact is less than the flux out of the opposite face. Such divergent
mass flux contributes to a fall of air mass in the cube and hence to a decrease in air
density, as ensured by the minus sign in eqn (7 .30). Fig. 7.26 Flows of oir mass
through opposite faces of a
fixed volume. When unequal.
they contribute to mass flux
convergence or divergence.
HORIZONTAL AND VERTICAL BALANCE

Detailed investigations of atmospheric mass fluxes on scales ranging from the

synoptic to the small show that the horizontal mass-flux divergence (the first two
terms on the right-hand side of eqn (7 .30)) is always very nearly balanced by an
opposing vertical mass-flux divergence o(pw)loz, so that the left-hand side is a rela-
tively small residual. In fact the observed magnitudes of the opposing terms are
such that if each were not opposed by the other, the local ch~nge of air density
would often approach lOOOJo per day on synoptic scales and per few minutes on
small scales. Nothing remotely like this is observed: local percentage variations in 213
7 Atmospheric dynamics
-:-:--:--:-:-
Fig. 7.27 Horizontal
convergence and vertical
~ -----..,----..,-
~ -.----..- ..........
-.., ,-,-
-.- .
..........
....,___
Horizontal
divergence of mass flux on a
vertical section through the
lower troposphere, with
consistent uplift in the middle air density are effectively equal in magnitude to local percentage variations in
troposphere. absolute temperature (local pressure variations being very much smaller than local
temperature variations when expressed as fractions) and seldom exceed a very
small fraction of the huge figures quoted above for the respective scales. To this
extent therefore

-
a (pu) + -
a (pv) +
a
-(pw) =0 (7 .31)
ax ay az
It seems that the atmosphere arranges its flow to minimize local concentrations
and depletions of mass: when mass is being gathered horizontally by converging
mass flow, as in the lower troposphere in an extratropical cyclone or a cumulo-
nimbus for example, it is stretched (diverged) vertically so that the resulting con-
gestion of atmospheric mass is small. Since the vertical mass flux is zero at the
Earth's surface, a vertical divergence of mass flux implies a vertical mass flux
which increases with height above the surface throughout the horizontally con-
verging layer, and hence a significant rate of uplift at its upper limit (the middle
troposphere in Fig. 7 .27), which of course is almost always associated with cloud
production. We can investigate the magnitude of this uplift by integrating the
terms of eqn (7 .31) over the height range from the surface to the level of interest, as
sketched in Fig. 7 .28. The integration of the first two terms simply gives the net
horizontal influx of mass through the sides of the imaginary column from surface
to chosen level. The integral of the balancing third term

J
l a
- (pw) dz = [pw]~ = (pW) 1
b az
simplifies to become the vertical mass flux out of the top of the column (i.e. at the
Top
chosen level) because there is of course no mass flux through the bottom surface. It
follows from the near balance expressed by eqn (7.31) that

= - -1 Jt [ a
-- (pu) a
+ -(pv) J dz (7.32)
ax ay
W1
Pt b

Values for the right-hand side have been measured from careful analysis of
synoptic-scale observations, and they show that the horizontal convergence of
Bottom
mass in the lower troposphere is consistent with rates of uplift - 10 em s - 1 in the
Fig. 7.28 Net convergent middle troposphere, which are known from other evidence (rates of steady rainfall
horizontal mass flux in an for example - section 11.4) to be quite typical. Mass convergence is difficult to
atmospheric column, with measure accurately on the scale of a cumulonimbus, but the inference from known
associated uplift at its top. updraughts in middle levels (- 5 m s- 1) that the convergences are 50 times larger
than synoptic scale values is quite reasonable.

COMPENSATION

The reluctance of the atmosphere to concentrate in a confined volume shows itself

214 also in a reluctance to concentrate in a confined column. The horizontal flux of
mass into the column shown in Fig. 7.28 adds to the total mass of the column 7.13 Compressing and
deforming air
extending up to the top of the atmosphere and hence to the pressure at the bottom
of the column (i.e. at the surface). This follows from the hydrostatic assumption,
which we can make as usual, and from the fact that the rising air does not of course
become weightless just because it rises above the level of our earlier interest. In fact
the rate of rise of surface pressure (the pressure tendency as measured by a baro-
graph) is simply g times the mass input, as we can see by assuming that the
imaginary column has unit horizontal area.

ap = g Jt -aM = - g Jt [ - a (pu) + - a (pv) J dz

at at ax ay
_s
b b

Since the integral on the right-hand side is given by the compensating rate of rise at
middle levels (pw at the top of the converging layer) we can quickly estimate the
pressure tendency associated with uplifts of magnitudes quoted above. In the
synoptic case it amounts to an amazing 500 mbar per day (assuming air density to
be 0.5 kg m - 3 at middle levels), and in the cumulonimbus case the same enormous
value occurs in about half an hour. Again nothing like this is observed, even though
tendencies measured at the bottom of the real atmosphere can obviously be inten-
sified in rate (and shortened in time) by the bodily translation of horizontal
pressure gradients embedded in moving weather systems. On the synoptic scale,
pressure tendencies seldom exceed 20 mbar per day, and on the scale of cumulo-
nimbus they seldom exceed a few millibars in the hour-long life of the cloud. The
only possible explanation is that almost all of the input of mass to the lower half of
the tropospheric column is offset by an output from the upper half, showing that
there is horizontal divergence of mass flux there. The simplest picture is presented
in Fig. 7.29. The horizontal convergence of mass extends up the middle tropo-
sphere and is then replaced by horizontal divergence of mass extending up to about
the tropopause. The balancing vertical mass flux increases with height up to the
level at which horizontal convergence is replaced by divergence, and then
diminishes steadily to about zero at the tropopause. The horizontal convergence
and divergence compensate (nearly balance), and the slight imbalance accounts for
the rate of change of pressure at the bottom of the column. The falling surface
minimum pressure in the formative stages of an extra tropical cyclone, for example,
indicates that divergence aloft slightly but persistently exceeds the convergence
beneath. More elaborate schemes with several layers of alternating convergence
and divergence are of course possible, and could equally well explain the small
pressure changes observed at the surface, but observations suggest that the two-
layer scheme accounts for most synoptic-scale and cumulonimbus-scale weather
systems. On the synoptic scale, for example, an anticyclone is like an inverted
cyclone, with convergence overlying divergence and maximum downward mass
flux between. This widespread and persistent subsidence is associated with many of
the most striking properties of anticyclones (section 12.1).
We have now seen that the atmosphere arranges its flow on the synoptic and
small scales so as to minimize the congestion and depletion of air mass in localized
volumes and columns, and that each of these tendencies is associated with

Fig. 7.29 Schematic flow in a

D cloudy weather system.
Horizontal convergence
ttt t
Vertical
below and divergence aloft
are linked by complementary
c uplift in middle levels.

Horizontal 215
7 Atmospheric dynamics extremely basic and important aspects of weather behaviour. Though the analysis
has therefore been useful, it is important to note that we have not made use of any
dynamics (forces and accelerations) in the process - the basic physical property
assumed is simply the conservation of mass. We have not in any sense explained
why the atmosphere arranges its flow in these ways, nor even discussed the
mechanisms involved; but the procedure has been valuable even so, and has
pointed to quite basic types of behaviour which must be explained by properly
dynamic theory. Such theory exists now in respect of synoptic scale motion (the
theory of baroclinic instability which is briefly mentioned in section 11.6) but is far
beyond the level of an introductory text; it barely exists at all in the case of
cumulonimbus.

SYNOPTIC SCALE

Though the basic form of the continuity equation has proved directly applicable, it
is useful to develop it a little further for the particular case of synoptic-scale
motion, to try to see how compression and decompression can be made explicit in
the formal description of deformability begun in eqn (7.31).
A serious snag appears when we consider the horizontal mass convergence in
geostrophic flow. We can do this by taking the geostrophic expressions for pu and
pv from eqn (7.17) and substituting them in the first two terms of eqn (7.31).
Ignoring the variation of fwith y (i.e. with latitude) which is relatively very small,
we get

_!_ [ - __!p_ +
J axay
__!p_]
ay ax
which is zero because the terms in the brackets cancel exactly, showing that a pure
geostrophic flow has zero horizontal mass divergence. Physically this arises from
the inverse relation between wind speed and pressure gradient which is basic to geo-
strophy. Figure 7.30 depicts a situation in which the geostrophic flow is increasing
to the east, which corresponds to positive iJ(pu)/ax. The associated narrowing of
the north-south separation of the contours corresponds to a negative iJ(pv)!iJy
which exactly balances iJ(pu)/iJx, giving zero net horizontal mass divergence. This
is disappointing, because it shows that the so far very useful geostrophic
approximation completely fails to allow for the convergence and uplift of air which
we know is the facet of atmospheric flow which is essential in producing cloud, pre-
Fig. 7.30 Plan view of cipitation, and pressure variations. It appears that geostrophic balance is too well
balancing longitudinal
velocity divergence and
balanced for the real atmosphere. It is still useful as described earlier, but it must
lateral convergence in not be used in ways which allow it to eliminate vertical motion.
confluent geostrophic flow. We can make less drastic and more useful simplifications by expanding each of
the first two terms of eqn (7 .31) into a term in the gradient of wind speed and
another one in the gradient of density. Combining the latter from the two horizon-
tal directions we have
ap ap
u-+v-
ax ay

which represents the contributions to density variation arising from motion in the
presence of instantaneous horizontal density gradients. Since air density tends to
move with the air itself, these terms can be regarded as representing the horizontal
216 advection of existing density gradients by the flow. Detailed observations of
synoptic-scale flow indicate that horizontal advection of density is about an order 7.13 Compressing and
of magnitude smaller than the other terms produced by the expansion of the deforming air
original mass flux divergence, which are

p (~ + ~) (7 .33)
ax ay

representing the contributions arising from longitudinal stretching of flow. The

relative insignificance of the advection terms is consistent with the tendency to
minimize density concentrations (and therefore density gradients) noted already,
and at least in part reflects the tendency of the thermal-wind relationship to align
horizontal winds and isopycnals and hence minimize density (and temperature)
advection.

DIVERGENCE

The bracketed expression in eqn (7.33) is called the horizontal divergence of

velocity, or simply the divergence, and is usually denoted by D. The approximate
form of the continuity equation (eqn (7.31)) can now be written in the form
1
D + - -
a (pw) == 0 (7 .34)
p az
Velocity divergence is similar to the divergence of mass flux discussed above, but it Fig. 7.31 Plan view of flow
divergent in both horizontal
is considerably simpler to interpret, depict and quantify. Figure 7.31 shows an directions.
example of flow with positive divergence D; if a patch of this air were marked with
smoke at some initial time, the horizontal area A of the patch would increase with
time for as long as D remained positive. The patch would also move bodily (north-
eastward in the example shown), but this is irrelevant to D which describes only the
relative motion of the edges of the patch. It is shown in more formal expositions
[44] that Dis equal to the fractional rate of change of horizontal area of the patch.

D=-1 dA (7.35)
A dt
From a variety of evidence it is known that D values - - IQ- 5 s - 1 are typical in the
zone of horizontal convergence which virtually fills the lower troposphere in an
extratropical cyclone. If divided equally between the two terms of D it follows that
each of the stretching terms au/ax, av!ay is similarly- I0- 5 s- 1 • This amounts to a
longitudinal gradient of wind speed of about 1 m s - 1 in 100 km, which is barely
measurable from standard synoptic observations, but corresponds to extremely
significant convergence on the synoptic scale. It is a major problem in meteorology
that such significant features are only marginally resolvable by direct observation,
and one which is compounded by the irrelevance of the geostrophic approximation
in this context. Considerable effort has gone into finding alternative ways of
estimating divergence from available observations; fortunately there is a useful
direct connection with large-scale rotation, as indicated in the next section.
In a region such as the lower troposphere in an extratropical cyclone, according
to eqn (7 .34) the convergence (negative D) in the lower troposphere is balanced at
every level by

p aza (pw) (7.36)

217
7 Atmospheric dynamics If the atmosphere were incompressible and uniform this would simplify to awI az, a
vertical stretching term, and the continuity equation would become simply
D + aw!az =0 (7 .37)
showing that at every level vertical stretching balances horizontal convergence, or
vertical squeezing balances horizontal divergence. If a D value of -10- 5 s- 1
prevails throughout the first 5 km of the troposphere, and is balanced at every level
by positive aw!az, then the cumulative uplift speed w at the 5 km level, which is
found by integrating the vertical stretching term up to this level, becomes very
simply

w = J: ~; dz = - zD
which is 5 em s- 1 in this example - a fairly typical value for rates of uplift
associated with nimbostratus in extratropical cyclones (see section 11.4). The
behaviour described by eqn (7.37) is that of a slack plastic bag containing a fixed
volume of water: when squeezed in the horizontal it expands in the vertical and vice
versa- an example of simple incompressible deformability.

COMPRESSIBILITY

Because the atmosphere is in fact highly compressible, its behaviour is more

complex than is allowed by eqn (7.37). However, the observational studies
mentioned above show that it is only the vertical compressibility and non-
uniformity which have significant effects, at least on the synoptic scale. If we try to
adapt eqn (7 .34) to the form of eqn (7 .37), we find
aw
D + - = Xw (7 .38)
az
where appendix 7.13 shows that X is nearly constant with a value of about 10- 4
m-I. The deviation from simple incompressible behaviour is significant but not
dramatic so long as no single layer of consistent convergence is as deep as the
scaling thickness II X, which is the case in practice since II X is about 10 km and
most layers of consistent divergence or convergence are no more than half this
depth. For example, in the case of the 5 km layer with divergence -10- 5 s- 1
quoted above, the balancing rate of uplift at the top of the converging layer is
increased from 5 to 6.5 em s -I by the inclusion of the right-hand side in eqn (7 .38)
(appendix 7 .13). An increase is to be expected since density falls with increasing
height and ultimately it is its product with w which provides the balancing mass
flux.
In advanced texts it is shown that the complication of the extra term on the right-
hand side of eqn (7 .38) can be avoided if the vertical coordinate z is replaced by
pressure p. In fact all the complicating minor terms in eqn (7 .30) vanish and the
continuity equation becomes
au+~+~= 0
ax ay ap
where w is dpldt, the rate of change of pressure of the moving air parcel, and the
horizontal stretching terms are actually evaluated on an isobaric surface. The x, y,
218 p coordinate system is used very widely in theoretical and computational work
because of such simplifications, but elegance is achieved at the cost of conceptual 7.14 The dynamics of
subtlety which makes it less helpful for introductory discussion. rotation

FRICTIONAL CONVERGENCE

Let us finish this section by applying the concept of virtually incompressible

deformability to the case of frictional convergence in the planetary boundary layer
of an extratropical cyclone. Because the wind vectors near the surface are con-
sistently angled somewhat across the isobars, there is an inward component of air
flow which may typically average - 2m s- 1 in the first 500 m. According to
Fig. 7.32 this produces a value of about - 10- 5 s- 1 if it persists across a 1000 km-
wide region such as the central zone of a mature extratropical cyclone. This must be
balanced by the vertical stretching aw/az, which is therefore 10- 5 s- 1• Integration
through the depth of the boundary layer shows that air at the top of the converging
layer (i.e. at the 500 m level) is rising at 0.5 em s- 1• This may seem a very small
value, but it is a significant event bearing in mind that it is lifting the overlying
atmosphere over an area of 106 km 2 • This lifting by frictional convergence
influences a much greater depth of air than can be reached directly by surface
friction, though the effects aloft have been transformed from the simple horizontal
drag near the surface.

Radial winds 2 ms-1

Diameter 1000 k m

-4
au
dX "'
-2u
X ;as
u -u 2Jv _ -2v -4
()y- X ;os Fig. 7.32 Frictional
convergence in the base of an
extratropical cyclone, with
wind components
perpendicular to circular
isobars.

7.14 The dynamics of rotation

ANGULAR MOMENTUM

It should be clear by now that the Earth's rotation influences atmospheric

behaviour in several important ways, especially on the synoptic and larger scales
with their conspicuous quasi-geostrophic balance. We can approach the basis of
the dynamics of rotation by examining the extension of Newton's second law of
motion to the case of a body whirling about a perpendicular axis of rotation under
the influence of a force which exerts a twisting leverage or torque about the axis. In
the simple case shown in Fig. 7.33 a body of mass M is whirling at a radial distance
r about an axis with tangential speed ~. while acted on by a tangential force F 1 •
Then the appropriate form of the second law is shown in basic dynamics texts to be
d
-(rM ~) (7.39)
dt 219
7 Atmospheric dynamics

Angular momentum of p
Fig. 7.33 Angular Vr r MV1
momentum of body P rotating
about axis X.

where the term in brackets on the right-hand side is the angular momentum (some-
times more logically called the moment of momentum) of the body in its rotation
around the axis. It follows from this equation that angular momentum is conserved
if there is no torque (no rF1 product) about the axis. In such a case
r M V1 = constant (7.40)
Conservation of angular momentum is most usefully applicable to atmospheric
motion on the synoptic and larger scales, because the absence of anything cor-
responding to the entrainment and mixing which so dominates small-scale flow
means that the mass M of a large air parcel or mass is usefully constant, and means
also that there is relatively little frictional torque outside the planetary boundary
layer. Equation (7 .40) then becomes simply
r ~ = constant (7 .41)
A surprising consequence of eqn (7 .41) is that if the perpendicular distance r
between the rotating body and the axis of rotation is reduced, then the tangential
speed ~must increase and vice versa. In the absence of any torque this may smack
of magic, but there are several quite commonplace observations which confirm
such a tendency, one of the most dramatic of which involves the spinning
manoeuvre beloved of figure skaters. Firstly they generate angular momentum by
carrying their initially linear momentum into a spiral trajectory which winds into a
spin on an effectively fixed vertical axis. Having done this with whirling arms out-
stretched, the arms are dropped close to the body, obviously reducing the radial
distance of their hands and arms from the axis of spin. In terms of eqn (7.41) r is
reduced and the resulting sharp increase in ~produces the intended visual effect of
a spinning blur. To slow down before attempting to set off over the ice again, the
arms are stretched out to reduce V1 to manageable proportions by the reverse of the
initial process.

ANGULAR MOMENTUM BUDGET

The tendency to conserve angular momentum shows up quite clearly in the Hadley
circulation, as already mentioned in section 4. 7. Let us reconsider the Hadley
circulation with possible conservation of angular momentum particularly in mind.
Air rises to the high troposphere in the intertropical convergence zone, usually
fairly near the equator, and then moves polewards in the high-altitude branch of
the circulation. At these heights there is little friction with the surface; and there
can be no overall east-west (zonal) component of pressure gradient force if we
consider a zonal ring of air encircling the Earth at any particular latitude, since
pressure at any altitude must form a continuous profile round the ring and hence
220 exert as much torque westwards in some longitudes as it does eastwards in others.
To this extent there should be little zonal torque acting on the ring of air, so that as 7.14 The dynamics of
it moves polewards it must tend to conserve the angular momentum it had when it rotation
first rose out of effective frictional contact with the surface in the intertropical
convergence zone.
If for simplicity the air starts its poleward migration with zero zonal wind speed
(relative to the Earth) at the equator, then its initial angular momentum per unit
mass is simply OR 2 , of which OR is the eastward tangential speed of the Earth's
surface at the equator (Fig. 7 .34). If this is to be conserved, then the eastward wind
speed Uq, (relative to the Earth's local surface of course) at some higher latitude <t>
must be such as to maintain the product of absolute eastward speed and perpen-
dicular distance from the axis of rotation (Fig. 7 .34)
(Uq, + OR cos </>)R cos <I>
at its initial value. It follows that

Uq, = OR ( -1- - cos</> ) (7.42)

cos <I>
Equator
As shown in Fig. 7.35, the implied increase of Uq, with latitude is quite rapid, with
Uq, values exceeding 55 m s - 1 at latitude 20°, and 130 m s - 1 at latitude 30°. When Fig. 7.34 Radius of rotation
of a point on the Earth's
averaged zonally, observed speeds seldom exceed 6007o of these ideal values, but surface at latitude</>.
nevertheless the combination of poleward drift of air in the high troposphere and
the sharp and consistent increase with latitude of upper tropospheric westerly wind
speeds on the equatorial flank of the subtropical jet stream in each hemisphere
strongly suggests that conservation of angular momentum in the poleward drift is a
major part of the mechanism maintaining these jet streams.

200 I
I
I

150
-:,
E
"0

""
Q
100
"'
"0
c:
~

?: 50
"
"' Fig. 7.35 Westerly wind
:;::" speed of air conserving
0 angular momentum from a
0 10 20 30 40 state of zero wind at the
equator.
Latitude (O)

The lack of complete agreement with predicted values shows that angular
momentum is not fully conserved, and it is useful to consider briefly why this
should be. Firstly there is still some frictional connection with the surface: cumulo-
nimbus locally connect the bottom and top of the troposphere, exchanging
momentum amongst other things. Secondly the argument about continuity of
zonal pressure profile breaks down at levels which are pierced by mountains. As
sketched in Fig. 7 .36, the observed tendency for lee pressures to be a little lower
than windward pressures means that there is a net torque opposing the zonal
motion of the air. Air flowing over the mountains maintains a pattern of gravity 221
 - -
7 Atmospheric dynamics

Fig. 7.36 Large-scale drag

~~-_
~ ---~~-------~--
arising from slight asymmetry ': Isobar __
of upwind-downwind
pressure fields around hills. Downwind horizontal

waves up to very great heights (section 10.8), one of whose effects is to extend some
of this torque to heights far above the mountain tops. The third effect requires a
hemispheric view of atmospheric angular momentum (Fig. 7 .37). The atmosphere

Fig. 7.37 Vertical ond

meridional fluxes of westerly
zonal momentum.

is continually gaining westerly angular momentum in low latitudes and losing it at

high latitudes, in each case by momentum exchange with the underlying surface.
Large masses of air take up the westerly momentum of the local surface through
boundary-layer friction and the mechanisms described above. They may then
move away to another region as they become involved in a large weather system,
where they meet and conform their flow to other masses from other source regions.
In so doing they exchange momentum, even though they do not mix in the way that
small parcels do. The general result of such exchanges is that relative westerly
momentum is transferred to high latitudes and relative easterly momentum is
transferred to low latitudes. The westerlies of mid-latitudes are maintained in this
way, largely by the momentum exchanges associated with synoptic and larger-scale
weather systems. In fact there is a continual conversion from a large-scale
poleward eddy flux of angular momentum to the zonal westerlies which is closely
related to a similar conversion of kinetic energy discussed in section 13.3. In this
context the upper branch of the Hadley circulation is like a flange of equatorial
angular momentum moving polewards on nearly frictionless bearings near the
equator, but being impeded more and more as it reaches toward the subtropics by
momentum exchange with air reaching down from higher latitudes. There is there-
fore a transition from a smooth poleward flux of zonal angular momentum, in
which zonal angular momentum is nearly conserved, to an eddy flux in which no
conservation is obvious in the zonal averages (though in principle angular
momentum is conserved throughout the eddying process in just the same way as is
heat), and the core of the subtropical jet stream meanders in the transition zone.
Conservation of angular momentum lies at the root of the observed tendency of
all poleward-moving air to increase its relative westerly motion, or reduce its east-
ward motion, and of all equatorward-moving air to do the opposite. Such tenden-
cies. are widespread in the atmosphere and are often conspicuous in synoptic-scale
weather systems. They are apparent in the great cloud swirls of Fig. 1.1, in
particular in the eastward curve of the swathes of frontal cloud moving polewards
in middle latitudes, known from detailed studies to be associated with huge injec-
tions of warm, moist air upwards and polewards from the polar flanks of the sub-
222 tropical high-pressure systems (section 11.4).
 The tendency of air to veer to the right of its current direction of motion (in the 7.14 The dynamics of
northern hemisphere) should remind you of the Coriolis effect. In fact the two rotation
approaches are entirely equivalent, the conservation of zonal angular momentum
in poleward-moving air being identical to the integrated effect of the eastward
Coriolis force associated with poleward motion. The Coriolis formulation is more
useful for examining the shape of individual flows, while zonal angular momentum
is more useful for looking at large-scale budgets and statistics.

VORTICITY

The large, flattened air masses of synoptic-scale weather systems also tend to
conserve angular momentum as they rotate about locally vertical axes in their
centres. Indeed the dynamics of this rotational motion is a very important part of
the overall dynamics of large-scale motion systems, and is treated at length and in
somewhat forbidding detail in advanced texts. However, one of the most funda-
mental results of such treatments emerges in the following simplified approach.
Consider a narrow horizontal circular ring of air rotating on a weather map as
part of a synoptic-scale cyclonic or anticyclonic system (Fig. 7 .38). If its angular
velocity about a locally vertical axis through the centre of the ring is w and its radius
is r, then conservation of angular momentum about the axis requires (from eqn
(7 .41)) that
r 2w = constant
d
or - ( r 2w) =0
dt
dr
tdt

-u _ dr i1U 2(dr/ dt) 11V

- dt- 11x = 2r 11y

D = 11u + 11v = 2 dr
/1x 11y r dt
Fig. 7.38 Expansion of a
dr horizontal ring of air in
dt symmetrically divergent flow.

We can expand the latter to find an expression for the rate of change of angular
velocity of the ring (dw/dt) when it is expanding or contracting in response to
synoptic scale divergence or convergence (section 7 .13).
dw d
r2 - + w - (r 2) =0
dt dt
Rearranging to isolate the required rate of change of angular velocity, we find
dw 2w dr
---
dt r dt
But it is apparent from Fig. 7.38 that the horizontal divergence D of the circular
plate of air bounded by the thin ring is equal to (2/r)dr/dt so that
dw
-= -wD (7 .43)
dt 223
7 Atmospheric dynamics So far we have considered the rotation of the air relative to the map, but since the
map is actually rotating with angular velocity !1 sin¢ about the local vertical, where
!] and¢ are the Earth's angular velocity and latitude (Fig. 7.34 again}, !1 sin¢
should be added to w in eqn (7 .43) to produce

~ (w + !1 sin¢) = - (w + !1 sin ¢)D (7 .44)

dt
It is conventional when dealing with rotation in fluids to describe localized rotation
by means of vorticity, which is defined to be twice the instantaneous angular
velocity of the parcel. The usual sign convention means that cyclonic rotation is
positive. The component of relative vorticity (2w) about the local vertical is always
symbolized by r in meteorology. Since the doubled angular velocity of the map
about the local vertical (2!1 sin¢) is the familiar Coriolis parameter f, it follows that
eqn (7 .44) can be rewritten

~<r+J) = -<r+f)D (7.45)

dt
The Corio lis parameter is sometimes called the planetary vorticity on this account,
and the total (r+ f) is known as the absolute vorticity of the air.
A more thorough derivation would have included torque terms. One arises from
friction, which fairly obviously exerts a torque tending to reduce relative vorticity.
Another very significantly arises from the baroclinity of the atmosphere, since an
extended form of the buoyancy force always tends to align intersecting isobars and
isopycnals and hence impose rotation. Other terms allow for the twisting of x andy
components of vorticity into the local vertical, in particular the substantial
component about the local horizontal which is associated with any thermal wind
there may be, since shear always implies vorticity about an axis perpendicular to
the plane in which the shear is profiled. However, the very simple eqn (7 .45) applies
quite usefully in many conditions to the evolution of synoptic-scale weather
systems, and even more reliably to events on larger scales.

EXAMPLES

As an important example, consider the development of cyclonic rotation in air

which is originally at rest on the weather map. Viewed from the weather map the
rotation seems to be generated from nothing, but in fact the initially apparently
static air has the planetary vorticity {f) of the map itself. In the presence of the
crucial ingredient of synoptic-scale convergence (negative D) in the lower tropo-
sphere, eqn (7 .45) shows that positive relative vorticity will begin to grow there
according to
dr = -JD
dt
If D remains constant and none of the complicating terms becomes significant, the
solution of eqn (7 .45) (Fig. 7 .39) shows that the absolute vorticity r +f increases
exponentially with a doubling time of In 2/D. Assuming a typical value of
-10- 5 s- 1 forD, we find that the absolute vorticity doubles (which is to say that
the relative vorticity increases from zero to f) in about 20 hours, which is observed
to be quite typical of a developing extratropical cyclone.
In the same way, eqn (7 .45) helps to explain how negative relative vorticity
224 develops in diverging flow. Starting from zero relative vorticity again, divergence
 I 7.14 The dynamics of
4 I 3 rotation

....... ...
....
- ,.. 3 ('0

"
" 2
...
.
u
~

.!! -1> "~

o"'
.
0

.. ..
0 >
> 0
2
~
Fig. 7.39 Exponential
...
>
2 'b'" growth with time of absolute
..
~

-
0 vorticity in a steadily
.a a: convergent part of a weather
< 1 0
system, and exponential
decay in a steadily divergent
part. The initial relative
vorticity is zero in each case,
0 -1
1.0 1.5 2.0 and axes have been
0.5
normalized by f and 1/D for
Divergence x time generality.

produces negative relative vorticity, which corresponds to anticyclonic rotation. If

the value of D remains constant, and no other terms interfere, the absolute
vorticity decays exponentially towards zero, which is to say that relative vorticity
decays to - f (Fig. 7.39). Such a limit to the intensity of anticyclonic vorticity is
amply confirmed by real synoptic-scale behaviour, and distinguishes anticyclones
from cyclonic systems which show no such restraint, and which often produce
relative vorticities several times as strong as the planetary vorticity f. To the extent
that all synoptic-scale rotation about the local vertical consists of concentrations of
the Earth's rotation, rotation in the opposite direction is impossible. This result is
consistent with a detailed examination of gradient balance in anticyclonic flow
(appendix 7.10).
The association of divergence with anticyclonic vorticity and convergence with
cyclonic vorticity is often quite obvious in a single weather system. In terms of the
profile of an extratropical cyclone sketched in Fig. 7.29, we should expect to find
cyclonic vorticity in the lower troposphere as already discussed, and anticyclonic
vorticity in the upper troposphere. The latter is usually much less obvious, but is
evident nevertheless after a little scrutiny as outlined in section 11.2.
In the previous section it was shown that synoptic-scale uplift and subsidence are
so sensitive to convergence and divergence that routine observations of wind are
barely able to detect convergent conditions capable of maintaining widespread,
vigorous nimbostratus. We have now seen that synoptic-scale rotation about the
local vertical is similarly sensitive. So although it is surprisingly difficult to measure
the associated horizontal convergences and divergences of mass, even a casual
glance at a satellite picture such as Fig. 1.1 shows striking evidence of both uplift
and rotation. In fact the sensitivity of rotation to uplift and subsidence can be used
to good effect in theoretical and computational procedures.
By contrast with its inability to cope with convergence or divergence, the geo-
strophic approximation remains well able to represent synoptic-scale rotation.
More advanced texts show that the relative vorticity about the local vertical is given
in terms of the usual horizontal wind components by

s= ~-~
ax oy
Substituting the geostrophically associated pressure gradients according to eqn 225
7 Atmospheric dynamics (7 .17), and ignoring gradients of p and fin comparison with gradients of wind
speed, we find

tg =
1 ( a2
pf
p + ayz
axz
ap)
2

The terms in the bracket do not cancel, as they did in the case of geostrophic
divergence. Each of the pair corresponds to the curvature of the horizontal
pressure profile across a horizontal (the actual curvature of the equivalent isobaric
surface), and a little thought will show that each is positive if the profile is concave
upwards (Fig. 7 .40). In other words a pressure minimum is associated geo-
strophically with positive vorticity, and a pressure maximum with anticyclonic
rotation. In fact a synoptic-scale map analysed with the help of the geostrophic
approximation provides a usefully accurate representation of vorticity
distribution.

I
(I l "
y 'g
"'
u I
I
Fig. 7.40 Vertical radial
isobaric profile across a "'
> I
~
depression with cyclonic I

geostrophic vorticity
(northern hemisphere}. Horizontal

If we describe the pressure field in terms of the slopes of isobaric surfaces (eqn
(7 .20)), the expression for Sg is obviously equivalent, and very importantly does not
require that we assume uniform air density.
g( azz azz )
sg= f ~+#
Note lastly that the relationship between convergence and rotation also applies
to very intense circulations on the small scale, the most extreme being the tornado.
Although detailed understanding of the formation of tornadoes is still very incom-
plete, it is clear that their intense rotation is associated with intense convergence,
and is described at least qualitatively by eqn (7 .44). As discussed in section 11.6,
tornadoes form beneath the bases of very vigorous cumulonimbus, where there is
intense convergence feeding the powerful updraughts aloft. This convergence con-
centrates and intensifies background vorticity to produce the intense and localized
vortex which is the tornado. However, in such tumultuous conditions it is not clear
which of the many complicating terms missing from eqn (7 .44) may also be signi-
ficant. Most tornadoes are observed to rotate cyclonically, but since conditions in a
tornado (section 7 .12) imply vorticities - 104 f, it would be very rash to conclude
that they are therefore simply extreme concentrations of planetary vorticity. As in
the hoary legend of the wash-basin vortex changing direction in ships crossing the
equator, there are many local factors which may be much more significant, though
these should probably not include the tendency to produce cyclonic vorticity by
road traffic driving on the right, as suggested recently by some American meteo-
rologists in playful mood.

226
Appendix 7.1 Dimensions Appendix 7. 1

All measurable quantities, however complex, can be regarded as a combination of

very basic measurements, the most common of which are mass M, length L and
time T. For example, every velocity consists of a length divided by a time, as
is obvious from the units in which values are expressed. Velocity is therefore
said to have dimensions of length divided by time - a statement which is often
written
[velocity] = [L] [T]- 1
Although the dimensions of a quantity are obviously consistent with the units in
which it is expressed, they are quite general and independent of the particular unit
system used. Such generality supports the axiom that all equations validly relating
observable quantities should be dimensionally consistent, as is justified by
experience. For example simple observations and quite elaborate theory show that
gravity waves propagate across the surface of a layer of water of depth hat a speed
c given by
c = ..jg h (7 .46)
provided the wavelength is considerably larger than h. The dimensions of g are
those of an acceleration, which are [velocity] [T]- 1 and consequently [L] [T]- 2 , and
the dimensions of hare simply [L]. The dimensions of the right-hand side of eqn
(7 .46) are therefore
{[L] [T]- 2 [L] }Y2 = [L] [T]- 1
in agreement with the obvious dimensions of the left-hand side.
Dimensional consistency is a powerful and useful constraint. If incorrect theory
or inaccurate measurement had suggested
c = g..Jh
its dimensional inconsistency would have shown at once that it could not be valid.
Inverting the procedure, if we had guessed that wave speed should depend only on g
and h in the form
(7 .47)
then dimensional consistency requires that a and b must each be Yz. This follows at
once if we substitute dimensions in eqn (7 .47) and require that each must separately
balance
[L T- 1] = [L T- 2] 0 [L]b
Equating powers we have
[L] I = a +b
and [T] - I = - 2a
therefore a= b = Yz, and the relation must be of the form
c = ..Jg h
Note that there is no constraint on any pure numbers, such as 1r, which may be in
the full relationship, since they are dimensionless. For example the equivalent of
eqn (7 .46) for waves whose wavelength "A is smaller than h is
c = {g"A/27r}Y' 227
7 Atmospheric dynamics This is dimensionally consistent of course, being like (7 .46) with the length A
replacing the length h, but its consistency is quite independent of the numerical
factor 271", which can be justified only by detailed theory or observation.
The dimensional consistency of valid equations enables us to find the dimensions
of less simple quantities. For example, from the most familiar form of Newton's
second law of motion
force = mass x acceleration
it follows that
[force] = [M] [L] [T]- 2 •
And from the definition of pressure as force per unit area
[pressure] = [M] [L]- 1 [T]- 2 •
As an important worked example, consider the dimensions of the exponential
scale height of the pressure profile of an ideal, isothermal and hydrostatic
atmosphere. In so doing we need to recognize another basic dimension - that of
temperature [8].
H = R T/g (7.48)
The dimensions of the specific gas constant R can be found from the equation of
state for air (eqn (4.2)):
[R] = [L]2 [T]- 1 [8]- 1
When the dimensions of the quantities remaining on the right-hand side of
eqn (7 .48) are substituted, we find
[RHS] = [L]
which of course is consistent with its interpretation as a height.
Relations are at their most general when they are expressed in the most
dimensionless form, the pressure/height relation in an isothermal atmosphere
being a case in point.
P = Po exp ( - z/ He)
If this is rewritten by normalizing or non-dimensionalizing pressure by expressing
it in units of pressure Po at some datum level, and normalizing height by expressing
it in units of the exponential scale height:
P. = plpo
and z = z/He
then p: = exp ( -z)
and the curve is the general one of pure exponential decay. A single graph with
suitably normalized axes could contain all such pressure profiles, and indeed all
other quite unrelated processes involving pure exponential decay (such as radio-
active decay of a single species of unstable nucleus). Very sophisticated use is made
of such normalization in analysing possible relationships in a field such as
turbulence, where observations cannot yet be brought together in a comprehensive
theoretical framework, as mentioned in section 9.9 (eqn (10.22)).
Dimensionless numbers such as the Reynolds number Re owe their generality to
judicious choice of dynamically significant factors whose ratio is dimensionless,
i.e. whose various dimensions cancel out, leaving a parameter which is a pure ratio,
independent of whatever system of consistent units is used for expressing the
various terms. As an exercise, check that Re is dimensionless.
228
Appendix 7.2 Pressure gradient force Appendix 7. 2

Consider a block of air in the presence of a component of pressure gradient ap/ax

along the x axis. Because of this gradient, slightly different forces must act on the
faces perpendicular to the x axis at slightly different positions x and x+ dx
(Fig. 7 .41a). If they and z dimensions of the block are dy and dz, then the force on
the block face at position xis

Fx = Pxdydz
in the positive x direction, where Px is the pressure on this face. The corresponding
force on the block face at position x+ dx is

Fx+dx = Px+dx dy dz

in the negative x direction.

Now the pressures on the two faces differ slightly because of the pressure
gradient, and by the formal definition of the gradient

PX+dx = p X + ap
ax
dx

for vanishingly small dx- (Fig. 7.41b). Hence the net force NF in the positive x
direction can be written in terms of the pressure gradient

NF = Fx - Fx+dx
= (px-Px+dx)dydz
- ap dxdydz
ax

Since the volume of the block is its mass divided by air density

dm
dvol = dxdy dz =-
p

we have
dm ap
NF
p ax
and consequently that the net pressure gradient force per unit mass is
1 ap
---
p ax

(a) (b) Fig. 7.41 {a) Net force on a

pair of opposite sides of an

J ~X+dx-Px
air parcel embedded in a
pressure gradient.
(b) Pressure difference and

lL_L
pressure gradient in the x
direction.

X
229
7 Atmospheric dynamics This is the particular case for the component in the x axis. Identical procedure
finds (11p) ap!ay and (11 p) ap!az for they and z components.

Appendix 7.3 Viscous force

Figure 7.3 depicts in profile a flow in the x direction which is sheared in the z
direction (i.e. a vertically sheared eastward flow). Consider the frictional forces
arising from viscosity and acting on a layer of air of unit horizontal area bounded
by heights z and z+dz. At level z+dz the faster-moving air above drags the top
surface of the layer in the positive x direction with a force numerically equal to the
tangential shearing stress rz+dz.x· At level z the slower-moving air below drags the
bottom surface of the layer in the negative x direction with a force which is
numerically equal to the tangential shearing stress Tz.x· Just as in appendix 7 .2, the
small difference in stress across the narrow gap can be written in terms of the
gradient of stress there:

ND =
a dz
-T
az z.x

and the volume of air in the layer (numerically equal to dz because of choosing unit
horizontal area) can be replaced by mass/ density, with the result that the net force
per unit mass is given by
a
p az Tz.x
1

If the horizontal shearing stress r,,x at any level obeys eqn (7 .4), then the net
viscous force per unit mass is given by
_!__( p. a2u +au ap.)
p az oz
az2
and if gradients in p. are proportionally much smaller than those in u, we have eqn
(7.6).

Appendix 7.4 Centrifugal pressure gradient

As shown in Fig. 7 .42, the equator ward component of centrifugal force per unit
mass at latitude ¢ is
0 2R cos cp sin cp or Y20 2R sin 2¢
If balanced by an equatorward horizontal pressure gradient

- _!_ iJp = Y2 0 2R sin 2¢ (7.49)

230 p ay
 Appendix 7.5
Pole

1-'-'-.::..::..:=---=-~r---,_,,.11 2 R cos IS Fig. 7.42 Centrifugal force

ll"2Rcos/IS sin r1 on unit mass on the Earth's
surface, and its equatorward
Equator component.

At latitude 45° with p 1.2 kg m - 3 we have

ap = -YlfJ/pR =: -2x I0- 2 Pa m- 1

ay
which means that near sea-level, pressure is increasing toward the equator at about
20 mbar per 100 km.
To integrate this balancing gradient from equator to pole, notice that dy = R d¢

I. a
if latitude ¢ is expressed in radians.

p
p
- p
e
= Ia
.J!__ dy
ay
= R
0
n
.J!__ d¢
ay
Substituting eqn (7 .49) for the balancing ap! ay we find
rr/2
- 1hpf!ZR2 J0 sin 2¢ d¢

1hpfl2R2
Ylp V2
where Vis the tangential speed of a point on the equator. Using values for V
(appendix 4.4) and sea-level air density we find
Pe - Pp = 129 000 Pa = 1290 mbar
which is nearly 30o/o larger than mean sea-level pressure.

Appendix 7. 5 Curvature terms

The conventional meteorological reference frame (Fig. 7 .1) gives rise to hidden
accelerations not only because it is rotating, but also because two of its axes (x and
y) are curved, and the z axis tilts as we move about over the Earth's surface.
Consider a westerly wind of speed u at latitude¢. Because the x axis at latitude¢ is
actually a latitude circle of radius R cos ¢, a body moving along it actually has
centripetal acceleration u 21R cos ¢ toward the Earth's axis of the rotation
(Fig. 7.43). Similar terms arise from other wind components. Except near the
poles, where obviously the conventional meteorological reference frame is incon-
venient and should not be used, all curvature terms arising from wind speed V are
of order V21 R. Assuming realistic values for V, the kind of comparison of magni-
tudes exemplified in this chapter shows that these curvature terms are almost
always very much smaller than the dominant terms in each component of the
231
7 Atmospheric dynamics

Fig. 7.43 Acceleration

implicit in motion along the
curved x axis (eastwards).

equation of motion, and can therefore be ignored. In the vicinity of jet streams,
where wind speeds occasionally approach 100 m s- 1, they are no longer negligible,
and they are retained in some of the more elaborate dynamical and computational
models of the atmosphere, but we will incur little error by ignoring them in the rest
of this book.

Appendix 7.6 The Corio lis effect

Figure 7.44 shows a body moving across a rotating turntable with uniform velocity
V relative to the turntable, which is rotating with angular velocity n about a
perpendicular axis. This means that the body is moving at a steady speed Valong a
straight line which rotates with the turntable. Since we are interested only in the
Coriolis component of the resulting centripetal acceleration, the only rotational
effect we need consider is the rotation of AB to A'B' (the obvious centripetal
acceleration toward the axis of the turntable is independent of V, and is therefore
not a Coriolis component). The body describes an arc AXB' of a circle when
viewed from a non-rotating frame, and therefore suffers a lateral Coriolis
acceleration toward the centre 0 of the circle.
Notice that the position of 0 is quite unrelated to the undefined point about
which the turntable is rotating; it simply lies on a perpendicular to the arc on the
side favoured by the direction of rotation of the turntable. The oversimplicity of

0
Fig. 7.44 As the body moves
along the rotating straight
line AB, the rotation takes B
to B' and the body describes
the arc AXB' in non-rotating
space.

A
s'

A'
232 B
Fig. 7.5 was therefore unnecessary, at least for the purpose of considering Appendix 7. 7
Coriolis.
We can find the magnitude of the Coriolis acceleration by finding the radius R of
the circular arc in terms of Vand 0. By symmetry the arc must make equal angles
0 dt/2 with the chords AX, A'X, XB' and XB wherever it touches them, where
0 dt is the angle turned by the turntable in the small time dt taken for the body
to move from A to B'. By noting that the internal angles of the triangle AOX
must total180°, it appears that the angle AOX is 0 dt. However, in radian measure
this angle is YIR, where Yis the length of arc AX. But from the body's motion,
Y = V dt/2, where the approximation improves as 0 dt diminishes. To this appro-
ximation therefore
Y Vdt
Ddt=-=--
R 2R
so that

R=-
v
20
The Corio lis acceleration of the body toward 0 as it moves along the arc at speed V
is V2/ R which becomes 2 0 V, as found in the simple concentric case in section 7 .3.
The other components of the total centripetal acceleration arising from the turn-
table rotation cannot be simply displayed on Fig. 7 .44, since they depend on the
position of the axis of turntable rotation. However it is easily shown that if Lis the
perpendicular distance between the body and this axis, then the centripetal
components are 0 2 L and v2/ L, where v is the tangential component of motion
relative to the turntable.
The treatment of the kinematics of rotation in two or three dimensions is very
elegantly simplified by using vectors, as shown in more advanced texts [44];
however the gain in algebraic simplicity is made at some cost in ease of perception
of physical significance.

Appendix 7. 7 Geostrophic flow

By definition, pure geostrophic flow is one in which the components of pressure

gradient and horizontal flow obey eqn (7 .17):
fv = _!_ op ju = op
p ax p oy
Squaring and adding we find
J v = _!_ op
p on
where

ap = [( ap )2 ( ap )2]v'
and an ax + ay
As shown in Fig. 7.45, if the resultant wind (speed V) makes an angle a with the 233
7 Atmospheric dynamics ;
I
I
v
\ :v
. I
Fig. 7.45 Horizontal wind
\I
~---~----+
components and pressure
gradient in geostrophic
\
balance (northern " Pressure gradient
hemisphere). (toward increasing pressure)

x axis, then tan a = vlu. If the resultant pressure gradient makes an angle {3 with
the x axis, then according to eqn (7 .17),

tan {3 = ap ; ap = - u/v = -!/tan a

ay ax
that is
{3 =a - 90°
This means that the pressure gradient is perpendicular to the geostrophic wind in
the sense depicted in Fig. 7.45. Since pressure gradient is positive toward higher
pressure, it follows that lower pressure lies perpendicularly to the left when
standing with your back to the geostrophic wind in the northern hemisphere, in
agreement with Buys-Ballot's law.

Appendix 7.8 Pressure gradient and contour

slope

Figure 7.46 shows a vertical section through a region with a horizontal pressure
gradient directed to the right, as indicated by the upward slope of the isobars in that
direction. Two isobars are shown, with pressures p and p + dp. The horizontal
pressure difference dp between them is of course identical to the vertical difference,
which can be related to the vertical separation dz through the hydrostatic relation
dp = gpdz. It follows that
dp dz
dn = gp dn
As dn becomes vanishingly small, dpldn becomes the horizontal pressure gradient
ap!an and dz/dn becomes the gradient of the p isobar, which we will write as
az/an. Dividing across by the air density we have

Fig. 7.46 Section through

tilted isobaric surfaces, with
vertical axis z and horizontal
axis n.

234
 1 op
-=g-
azp Appendix 7.9
p on on
where the left-hand side is the horizontal pressure gradient force per unit mass in
the direction of the n coordinate (appendix 7 .2). The right-hand side is therefore
the equivalent expression in terms of isobaric gradient.

Appendix 7. 9 Thermal wind

According to section 7.10 and eqn (7.21) the difference in geostrophic wind speed
between two isobaric levels is directly related to the horizontal gradient of the
thickness of the bounded layer

(7.51)

Since the atmosphere is so nearly hydrostatic we can relate the thickness (the
vertical depth) of a column to its mean density and therefore temperature, so that
the right-hand side depends on the horizontal gradient of column mean tempera-
ture. Let us examine this dependence in detail.
Begin with appendix 4.4 and eqn (4.18), which results from the vertical
integration of the hydrostatic equation:

= RT In (!!..!.) (7 .52)
g P2
where the column mean temperature Tis an average weighted by the logarithm of
pressure

T = JP' T d lnp
p, ln(p/p2)
Since everything on the right-hand side of eqn (7 .52) is constant except T
(remember the isobars are fixed once chosen), substitution in the right-hand side of
eqn (7.51) gives

v - v
g, g,
= ..!!._
f
In(!!..!.)
P2 on
aT

We can use eqn (7.52) again to replace the terms before oT!on on the right-hand
side, and rearrange to find
ar
g
(7 .53)
JT on
If we consider an extremely thin layer then the left-hand side of eqn (7 .53) tends
to a V/oZ, which is the vertical gradient of shear of the geostrophic wind. On the
right-hand side T becomes the temperature of the isobaric surface in the middle of
the thin layer. 235
7 Atmospheric dynamics
Appendix 7.10 Gradient balance in anticyclones

Figure 7 .18(b) represents the gradient balance in anticyclonic circular flow. Being
entirely radial the balance can be written as a single scalar equation
PGF- CF = CA
using the symbolism of Fig. 7 .18(a). The pressure gradient force PGF can be
replaced by the Coriolis force on the equivalent geostrophic wind fVg, and the
Coriolis force CF and the centripetal acceleration CA replaced by their standard
expressions to give:

Rearranging in the standard form of a quadratic equation we find

V2-RfV+RfVg=O
which has the standard solution
R f ± ..J(R 2J2- 4R j Vg)
v = ----------------~- (7 .54)
2
The term under the square root must be positive to avoid physically meaningless
imaginary terms, so that

or
V < Rf
g 4

which shows that there is a maximum geostrophic wind speed R f I 4 which cannot
be exceeded, and whose value increases with R and f.
If we substitute this maximum geostrophic wind in eqn (7 .54) we find

V = Rf
2
= 2V
g
(7 .55)

The linear increase of wind speed with radius means that the air is rotating about
the centre of the anticyclone like a solid wheel with uniform angular velocity

w= - - - = - -
v f
R 2
Sincef/2 is the component of the Earth's angular velocity about the local vertical,
we see that the limiting anticyclone is one which is rotating at a rate which exactly
cancels the rotation of the local surface and is consequently non-rotating when
viewed from an external inertial reference frame. As discussed in section 7.14 this is
expected to be the limiting result of pronounced horizontal divergence of flow with
conservation of angular momentum.
Consider the consequences of the restriction on the geostrophic wind speed.
Since at any latitude every value of Vg corresponds to a particular isobaric slope
through the geostrophic relation, we see that the maximum value of Vg implies a
236 maximum magnitude of isobaric slope
 aze = Rj2 Appendix 7. 11

aR 4g
where the minus sign means that the isobaric surface slopes downward away from
the centre of the anticyclone. When the air density is p, the equivalent expression
for maximum pressure gradient is
ap pRf2
----
an 4
The radial profile of pressure in such a pressure field can be found by integrating
the pressure gradient outward to an arbitrary outermost radius R (1500 km in this
case) at which pressure is assumed to be typical of the region outside the anti-
cyclone. The central elevation of pressure !:::.pis given by
pR2f2
!:::.p=--
8
and radial profiles of pressure at latitudes 15, 30 and 45° are sketched in Fig. 7 .47,
from which it is apparent that substantial elevations of surface pressure are hardly
possible at latitudes much lower than the subtropics, at least provided the flow is
dominated by the gradient balance.

35 Using p values
near sea level
30

25
!lp
(mbar) 20

15

10 Fig. 7.47 Horizontal radial

profiles of pressure elevation
5
in the low troposphere in the
0 steepest anticyclones allowed
0 300 600 900 1200 1500 by gradient balance at
R(km) latitudes 15, 30, and 45°.

Appendix 7.11 Buoyancy

Consider a parcel of air (or any body) of density p totally immersed in air (or any
fluid) of density p '. According to Archimedes' principle the immersed parcel
experiences an upward thrust which is equal to the weight of the air displaced. Since
the parcel experiences a downward force equal to its own weight, the net upward
force F is given by the difference
F = weight of displaced air - weight of parcel
If the volume of the parcel is Vol, this expression can be rewritten
F = g(p'-p)Vol
Dividing through by the mass of the parcel we find the net upward force per unit
mass 237
7 Atmospheric dynamics
_f_ = g(p'- p) = g B
M P
which is identical to the expression (7 .27) derived from the vertical component of
the equation of motion.
Note that Archimedes' principle, though still elegant and extremely useful, is no
longer regarded as a fundamental principle. The upward force arises because the
hydrostatic vertical gradient of pressure in the surrounding fluid, together with the
basic property that fluid pressure acts equally in all directions at a point, ensures
that the upward pressure force on the lower parts of the body exceeds the down-
ward pressure force on its upper parts. The same thing tends to happen in the
presence of any pressure gradient, vertical or horizontal, whatever its cause. In the
vortex of a tornado (section 7 .12) the horizontal pressure gradient may exceed the
vertical by a factor of two or more. The net inward force on an immersed body will
maintain a centripetal acceleration exactly matching that of the surrounding air
only if the body has the same density as the air. The dense debris picked from the
surface and demolished buildings etc. is centrifuged outward (actually tangen-
tially) in a potentially lethal spray which scours the zone at and around the base of
the vortex and its cloud funnel.

Appendix 7.12 Continuity of mass

Figure 7.26 shows a very small frame fixed in space with air blowing through the
open faces. Considering only air flow in the x direction, the mass flux into the face
perpendicular to the x axis at xis given by (pu)x dy dz, where the subscript x implies
that the air density and speed vary with x. A similar expression describes the mass
flux out of the face at x + dx, and the net influx is given by their difference
net mass influx = (pu)x dy dz - (pu)x+dx dy dz
As in appendix 7.2 we can replace the difference in the variable term by the product
of its gradient and the small separation dx, to find

net mass influx = - _g__ (pu) dx dy dz

ax
Equivalent expressions can be derived for the components of net mass influx in the
directions of they and z coordinates, and the resultant net flux of mass into the
volume is the algebraic sum of the three components:
net mass influx =
- [_g__(pu) + _g__(pv) + _g__(pw)] dx dy dz (7 .56)
ax ay az
Since mass is conserved, the net rate of mass influx must equal the rate of
increase of mass in the volume bounded by the frame:
a ap
-(mass) = -a dx dy dz (7 .57)
at r
since the volume bounded by the fixed frame is conserved. Combining eqns (7.56)
238 and (7 .57) and simplifying, we find
 Appendix 7.13
~
at
= - [_j_(pu) + _j_(pv) + _j_(pw)]
ax ay az
which is one of the standard forms of the continuity equation.

Appendix 7.13 Compressibility and divergence

Although somewhat beyond the general level of this book, the following derivation
is included because it does not appear explicitly in texts seen by the author. It is
given largely without commentary, but you will see the usual appeals to the
equation of state and the hydrostatic equation during manipulation.
The second term of eqn (7 .34) can be rewritten
.!_ _j_(pw) = aw + w ~
p az az P az
Since = _p_
P RT
ap I ap _ _ P_ aT
p az pRT az pRT2 az
_!_ ap __ I aT
p az T az
ap -gp
Now
az
aT
and
az
where Sis the environmental lapse rate expressed as a fraction of the dry adiabatic
lapse rate.
gp gS
Hence --+--
P az p CPT

__
g (.£e._- s)
CPT R
The value of C/ R is almost exactly 3.5. Outside relatively shallow inversions or
isothermal layers, S values range from about 0.4 (rather more stable than the
warmest saturated adiabat) to the limiting value of I (a dry adiabat). If for
simplicity we assume an S value of 0.5, the error in the bracket is relatively small
even in the extreme cases, and less than 50Jo most of the time.

where X has value""' 10- 4 m- 1 •

Finally
a
-1 -(pw) =--
aw Xw
p az az 239
7 Atmospheric dynamics so that eqn (7 .34) can be written in the form
ow
D + - =Xw (7.58)
oz
Let us solve eqn (7.58) for the simple two-layer case depicted in Fig. 7.48. This
includes uniform convergence C from zero z to z = h and uniform divergence - C
from there to z =2h; and w is zero at zero z. The solution is simplest if the equation
is used additionally in the form o2 wloz 2 = X ow/oz. The following solution can be
found by standard methods, or simply checked by substitution.

Up to z=h w
c
= X (ex'-1)

at z = h w = wh = X (exh -1)
c

up to z= 2h w = wh - S:... (eXz'- 1)
X
where z' = z - h
at z = 2h w=O
A particular case of this solution is plotted in Fig. 7 .48, where C = w-s s -I and
h = 5000 m, in obvious relevance to an extratropical cyclone. The above solution
can be compared with the solution to the equivalent incompressible atmosphere
undergoing the same combination of convergence and divergence (the straight
dashed lines), and it is clear that the difference is small because h is considerably
less than the scaling depth 1I X, the maximum updraught in the middle troposphere
being enhanced by a little over lOOJo by the inclusion of compressibility.
Divergence in 1 o- 5 s"1
-1 0 +1
10

w
8 ~
7 ~w
6 >
i5
E 5
-" w
4 0
Fig. 7.48 Vertical profiles of .£:. z
.,
Cl 3 w
updrought associated with a
simple two-layer model of
:I:
2
li!w
>
typical synoptic-scale z
convergence and divergence 0
0
in on incompressible
atmosphere (dashed line) and 2 3 4 5 6
0 7
a realistically compressible
one (solid line). Updraught (em s-1)

Problems

LEVEL 1

1. If pressure increases both northwards and eastwards at 5 mbar per 100 km,
240 show that the maximum rate of increase is 7.1 mbar per I 00 km towards the
 north-east. What contribution does this make to the pressure gradient in the Problems
local vertical?
2. Sketch the sheared planetary boundary layer, including the directions of the
shearing stresses, on the top and bottom surfaces of an embedded horizontal
layer and the direction of the associated momentum flux.
3. A person is standing on spring weighing scales resting on a rotating turntable.
In what way is the reading of the scales affected by the rotation if (a) the scales
are horizontal, and (b) they are tilted somewhat toward the axis of rotation?
4. In the case of Fig. 7.5, sketch the directions of the familiar centripetal and
Coriolis accelerations when the train and turntable are moving in opposite
directions.
5. From the scale analysis in section 7. 7, find the basic scales which determine the
relative sizes of the horizontal and vertical accelerations on the synoptic scale.
Likewise find the basic factors determining the ratio of horizontal to Coriolis
accelerations.
6. What would be the effect on the accuracy of the geostrophic approximation
near the Earth's surface if all turbulence there were to disappear?
7. In trying to set up a 1: 10 scale water model of low tropospheric flow (over
buildings for example) at about 10m s- 1, what water speed should be used to
ensure that the model is dynamically similar to the atmosphere in respect of
turbulence? Note that the kinematic viscosity of water is close to 10- 6 m2 s - 1•
8. Consider the directions of geostrophic flow (clockwise, anticlockwise) round
lows and highs in each hemisphere.
9; Assuming zero geostrophic wind speed in the very low troposphere, use the
thermal wind relation to find the wind direction in the high troposphere in the
following situations in the northern hemisphere: colder air to the north-west;
warmer air to the west; colder air to the north; warmer air to the south-west.
10. Why is there no familiar Coriolis effect at the equator? Consider why this
should have no effect on the occurrence of tornadoes there. (In fact they are
rare there, but for reasons most apparent in section 10.6.)
11. Consider the advisability of installing a convector heater in the interior of a
spacecraft.
12. What is the fundamental significance of the cyclonic direction of rotation?
(Hint: comparison with clock rotation is not helpful, as reversal between the
two hemispheres should suggest.)
13. What directions of rotation should be enhanced in the low troposphere of a
cyclone, the low troposphere of an anticyclone and the high troposphere of an
anticyclone?
14. If air has a tangential speed of 30m s- 1 at a distance of 300m from the centre
of a tornado, what tangential speed should it attain if it spirals in to a radius of
30m while conserving its angular momentum?

LEVEL 2

15. A toy train is running along a straight track resting on a rotating turntable and
passing through its axis of rotation. Given that a lateral force of 1 N per unit
train mass is sufficient to tip the train off the track, find the maximum safe
speed of the train when the turntable is completing a revolution every 3 s.
16. Carry out a scale analysis of the horizontal equation of motion for air moving
through a jet stream, in which individual air parcels pass through the jet core
from entrance to exit (1000 km apart) in 6 hours, and other factors are as 241
7 Atmospheric dynamics specified in section 7. 7. Look in particular at the implications for geostrophic
balance.
17. In the low troposphere, where air density is 1 kg m - 3 , what are the geostrophic
wind speeds associated with a horizontal pressure gradient of 5 mbar per
100 km at latitudes 60°, 30° and 5°?
18. Repeat the calculations of problem 17 in the high troposphere where air
density is only 0.25 kg m - 3 • Find the equivalent slopes of isobaric surfaces in
all these cases (problems 17 and 18).
19. Apply the thermal wind relation to the case of a well-defined front with an
isobaric (effectively horizontal) temperature gradient of 5 oc per 100 km
throughout a 1 km deep slab in which the mean temperature is 250 K. At a
latitude where/is 10- 4 s - 1 ' and simplifying g to 10 m s - 2 ' find the increase in
geostrophic wind speed along the front from bottom to top of the slab.
20. Consider an air parcel with absolute temperature T embedded in a slightly
cooler environment (temperature T'). Assuming the parcel to have the same
pressure as its surroundings, use the equation of state for air to show that the
buoyant force per unit mass is given by g(T- T')/T'.
21. Using the expression derived in problem 20, consider the vertical motion of a
parcel of constant temperature 300 K through air with temperature 299 K.
Find the buoyant force per unit mass and calculate the rate of rise of the parcel
after ascending 1 km from rest in the absence of other forces. (Use the
standard relation between distance and time for uniform acceleration:
w2 = 2az.) Why is this answer unrealistically large?
22. A cumulus cloud is observed to be increasing its volume by 1OJo per second
while rising at 1 m s - 1 • Find its entrainment drag per unit buoyant mass and
compare it with the typical buoyant force per unit mass estimated in problem
21 (- 1ho m s- 2).
23. A typical value for horizontal convergence in the lower troposphere of an
extratropical cyclone is 10- 5 s- 1• Supposing this to remain constant, find the
change in horizontal area of a large plate of embedded air. Calculate its area at
three consecutive 24-hour intervals after an initial state in which it was
104 km 2 •
24. In the case of problem 23, find the associated rate of uplift at heights 1, 3 and
5 km above the surface, assuming that the divergence is uniform throughout
the height range and that the atmosphere behaves like a simple incompressible
fluid.
25. In the case of problem 24, find the absolute and relative vorticities after 1, 2
and 3 days, assuming zero relative vorticity initially and a Coriolis parameter
of value I0- 4 s- 1•

LEVEL 3

26. Discuss the roles of drag of all types on synoptic and small-scale dynamics,
including cyclonic weather systems and the cumulus family.
27. Sketch the directions of vertical fluxes of easterly momentum in those parts of
the trade winds where the easterly flow is surmounted by westerlies in the
upper troposphere, and outline the types of mechanism carrying these fluxes
through the base and top of the layer of easterlies.
28. Summarize and discuss all points at which atmospheric dynamics is influenced
by the flatness of the atmosphere (the gross asymmetry between vertical and
horizontal scales). Speculate on how things might differ if the atmosphere
242 were much deeper.
29. Outline the influence of the Earth's rotation on atmospheric motion, Problems
discussing reasons for the difference between large- and small-scale motion
systems in this respect. Speculate briefly on how things might differ if the
Earth rotated (a) much faster than at present (as it did in the distant past
before being slowed by marine tidal friction), and (b) much more slowly than
at present (as it will do in the distant future for the same reason).
30. There is a widespread popular belief that the direction of water flowing out of
handbasins should tend to be anticlockwise in the northern hemisphere and
clockwise in the southern hemisphere. Critically examine the likely validity of
this belief by identifying the principle which is apparently assumed and
assessing the relative importance of other factors involved.
31. The increase of wind speed with height apparent in problem 19 is very sharp.
Check this by extrapolating to the case of a layer which fills the mid-latitude
troposphere. Such wind speeds are not observed. Consult Fig. 7.I6 and a
detailed description of the thermal structure of fronts (section II.3) to explain
why it is associated with lesser speeds, and outline the consequences for the
shape of associated jet streams as seen in vertical sections across the flow.
32. Assume cyclostrophic balance in a tornado, with air whirling round a radius of
I 00 m at I 00 m s-t. Find the associated slope of an isobaric surface and
consider its implications for the shape of the funnel cloud.

243
CHAPTER
Radiation and global
8 climate

8.1 The solar spectrum

The Sun is a powerful emitter of electromagnetic radiation in wavelengths ranging

from the ultraviolet to the infra-red (Fig. 8.1). Human vision has evolved to make
use of most of the shorter-wavelength half of the total range, apart from the rela-
tively small ultraviolet component, so that the longer-wavelength half is by
definition in the infra-red.
Almost all solar radiation comes from the photosphere, a relatively shallow shell
of incandescent gases in the outer parts of the sun. The photosphere has an outer
radius of nearly 700 000 km (over 100 Earth radii) and consists mainly of hydrogen
and helium at temperatures of nearly 6000 K and substantial pressures arising from
the weight of overlying, less conspicuous gases. In normal laboratory studies,
incandescent gast:s emit line spectra, but at the higher temperatures and very much
higher pressures prevailing in the photosphere, the radiating atoms collide with one
another so frequently and so violently that the individual lines are smeared out into
the relatively smooth hump shown in Fig. 8.1.
The visible spectrum is centred on wavelengths of about 0.5 JLm, not far from the
wavelength of maximum solar emission per unit wavelength range. On their own
these wavelengths would appear green to the human eye, but the substantial range
of wavelengths in visible sunlight, together with the variation in human visual
sensitivity across it, combine with the effects of the atmosphere on sunlight to
produce the familiar golden appearance of the sun when well above the horizon.
Although easily seen when the sun is viewed through a spectroscope, the dark
Fraunhofer lines produced by selective absorption of certain wavelengths by
relatively cool solar gases outside the photosphere would appear as insignificant
notches if they were included in Fig. 8.1. The much larger indentations which are
apparent there are produced in the Earth's rather than the sun's atmosphere.
The Earth's cloudless atmosphere modifies the solar spectrum quite significantly
by selective absorption and scattering, the effect of a standard cloudless atmo-
sphere being included in Fig. 8.1. Clouds are much less wavelength-selective, but
can of course reduce the total transmitted power very considerably when thick
enough. This is caused almost entirely by back-scattering, and can reduce the solar
power reaching the underlying surface by as much as 900Jo when there are thick
244 cloud layers. Data for Fig. 8.1 have been obtained by careful radiometry at the
 8.1 The solar spectrum
Key / , Curves I and II
Fig. 8.1 Power spectrum of
/ Curve ID solar irradiance at the Earth
/
2 - the distribution of power
B ounded area
e
=>.
equal to solar
constan t
(vertical axis) across the
range of solar wavelengths
N 1. 38 kW m-2
I (horizontal axis) . Curve I is
E
irradiance just outside the
3: Earth's atmosphere - the
~
total area under the curve

."
u
c being the solar constant .
Curve II is typical of the
""' irradiance at sea level with
- the Sun in the zenith of a
clear sky. Notches marked o,
"' c, v and z indicate
~
"Cl atmospheric absorption by
"' oxygen, carbon dioxide,
water vapour and ozone
0 ------ ~-- respectively. Curve Ill is the
0 I I 1.0 2 .0 3.0 power spectrum for a black-
:VIBGYOR: body at 5900 K reduced to
UV Visible IR Wavelength ()Jm)
allow for inverse-square
attenuation between Sun and
Earth. (After P.R. Gast (46])

Earth's surface, and at various altitudes up to tens of kilometres above sea level by
special balloons, and more recently of course by artificial satellites above the
atmosphere.
Observation and theory agree that the solar photosphere radiates with nearly the
maximum efficiency possible at its temperature. Perfectly efficient radiators are
known as blackbodies because, being also perfectly efficient absorbers, they reflect
no electromagnetic radiation; they and the blackbody radiation they emit have
been studied closely as an important part of thermodynamics. Some properties are
briefly outlined in appendix 8.1. The emission spectrum of a blackbody at
temperature 5900 K is included in Fig. 8.1 for comparison. After scaling to allow
for the attenuating effects of the Earth's distance from the sun, as outlined below,
the agreement with the solar spectrum observed outside the Earth's atmosphere is
obviously good. It is not perfect, particularly at the peak and on the ultraviolet
flank of the spectrum, but it is good enough to allow us to regard the sun as a black-
body at the temperature of the photosphere for many purposes. We can therefore
use Stefan's law (appendix 8.1) to estimate the power radiated from each square
metre of the photosphere (its emittance) to be about 70 MW m- 2 • Multiplying by
the surface area of the photosphere, we find the total solar output to be about 4.2
x 102° MW. Though this is a vast figure by terrestrial standards, using Einstein's
equation (£ = mc 2), we find it represents a rate of mass loss which would account
for only 0.1 OJo of the present solar mass if it persisted throughout the life of the Sun
to date. This output is maintained by thermonuclear reactions in the solar core, and
may have changed little during the life of the Earth, though this is obviously
difficult to confirm and is subject to debate.
Solar radiation floods outward through the solar system, its irradiance (the rate
of flow of radiant energy through unit area perpendicular to the solar beam at any
location) falling with increasing distance from the Sun. Since only trivial fractions
of the total are absorbed by the planets and interplanetary dust and gas, the same
total radiant flux must pass through any spherical surface concentric with the Sun.
Simple geometry then shows that the irradiance I at any distance R from the sun is
related to the photospheric emittance I, by 245
8 Radiation and global I= I,(R,IR) 2 (8.1)
climate
where R, is the radius of the photosphere. (Note the inverse-square relationship
between I and R.) Since the Earth orbits the Sun at a distance of about 200 solar
radii, it follows that the solar irradiance at the Earth's orbit is only about (1/ 2oo) 2 of
the photospheric emittance. The Earth's orbit around the Sun is slightly elliptical,
the semi-axis increasing from perihelion to aphelion by about 307o. The R
dependence produces a corresponding 6% decrease in the irradiation of the Earth,
with the maximum occurring in the northern hemisphere winter in the present state
of the Earth's orbit. Though the effects of this variation on seasonal climatic
variations are currently masked by the very large effects of the 23.5° tilt of the
Earth's equatorial plane relative to the plane of its orbit around the Sun (discussed
in section 8. 7), they may have been large enough at various times in the past to have
contributed to major climatic change (section 4.9).
The annual average solar irradiance just outside the Earth's atmosphere is the
solar constant, whose value is of the greatest importance for the state and
behaviour of the atmosphere, since it largely determines the power supply for the
atmospheric engine and thereby controls many very important physical conditions
prevailing at the Earth's surface, including several such as temperature which are
crucial to the evolution and maintenance of life. The solar constant has been
measured repeatedly with increasing sophistication throughout the twentieth
century, but only those measurements made from artificial satellites in recent
decades have been completely free from atmospheric interference and consequent
error. As mentioned in section 4. 9, there is evidence of small variations linked with
sun-spot numbers, and astronomical studies of the lifecycles of stars like our Sun
strongly suggest that the output from the young Sun was about 30% less than now.
Current values centre on 1.38 kW m - 2 , with variations of about 1%.
Note that the argument in this section reverses actual observational procedure.
Photospheric temperature is actually deduced from measurements of the solar
constant, using Stefan's law for blackbody emission and assuming zero inter-
planetary absorption. However, such estimates agree very well with others made by
different means, for example by applying Wien's law (appendix 8.1) to measure-
ments of the wavelength of maximum solar emission, and with theoretical models
of the Sun, and it is the consistency of all such results which confirms the validity of
the model presented, and with it the value of the photospheric temperature quoted
at the beginning of this section.

8.2 The Earth's energy balance

As mentioned earlier (section 4.9), the Earth's composite system of atmosphere,

ocean and land surface undergoes irregular vacillations between warm and glacial
epochs on timescales ranging from 10" to 108 years. But despite their dramatic
effects on life in middle and high latitudes, such temperature variations are small
distortions of the mean temperatures established by planetary equilibrium, and the
associated imbalances between energy fluxes to and from the planet Earth are very
much smaller than the constituent influxes and effluxes separately. Indeed it is
quite likely that the climatic effects of any such imbalances are considerably
246 enhanced by internal instabilities in the composite terrestrial system (section 4.9).
On a much shorter timescale, the big seasonal variations in localized heat fluxes are 8.2 The Earth's energy
reproduced year after year with relatively little variation, at least when viewed on balance
the hemispheric scale. To a very good approximation, therefore, the atmosphere,
ocean and land surfaces are in a nearly steady thermal state when viewed over one
or more full years, and their collective energy budget must be very nearly balanced
when its components are averaged over the same periods.
Solar radiation is the obvious energy input to this composite system, and in fact
is effectively the only one, since the next largest is the geothermal flux from the
Earth's hot interior (heat generated in situ by early gravitational collapse and
subsequent radioactive decay), which is nearly four orders of magnitude smaller
than the solar constant, being about 0.05 W m - 2 when averaged over the whole
Earth's surface. Starlight and the gravitational connections with the Moon and
Sun through the Earth's ocean tides contribute even less.
Since the Earth's energy budget is closely balanced, there must be an output to
balance the continuing solar input, and it too must be in the form of electro-
magnetic radiation to cross the near vacuum of interplanetary space. The simplest
assumption is that the Earth emits terrestrial radiation as a uniform spherical black
body whose size and surface temperature completely determine the total power
output in the form of terrestrial radiation. Let us consider the balance between
solar input and this terrestrial output.
Since the Earth and Sun subtend small angles when viewed from each other, the
Earth intercepts solar radiation almost exactly as a disc with radius equal to the
Earth's radius Re. The rate of interception must therefore be 1rR~ S, where Sis
the solar constant (Fig. 8.2). However, not all intercepted radiation is absorbed:
significant amounts are scattered back to space and play no part in the Earth's
energy balance. The fraction lost in this way is called the planetary albedo and is
denoted by the symbol a. The rate of absorption of solar energy is therefore given
by

Fig. 8.2 Nearly parallel solar

rays impinging on the Earth,

.___" 1/
~~~----~:~S-ol_a_r____" and terrestrial rays diverging
from it .

~ :: radiation

/ t ~ ~
Terrestrial radiation

(l-a)1rR~S (8.2)
which is the input to be balanced by the terrestrial output. Note that the non-zero
albedo of the Earth in solar wavelengths indicates a deviation from pure blackbody
behaviour which, according to Kirchhoff's law (appendix 8.1), must have con-
sequences for terrestrial emission of those same wavelengths. However, at
terrestrial temperatures, emission of wavelengths as short as solar wavelengths is so
extremely small (the Earth does not glow in the dark, after all!) that the shortfall in
emission of these wavelengths is quite unimportant. This contrasts sharply with the
great importance of the shortfall in absorption of solar radiation which arises from
the Earth's non-zero albedo. 247
8 Radiation and global Assuming that the Earth emits terrestrial radiation (which we will see shortly is
climate much longer in wavelength than solar radiation) as a spherical blackbody of radius
Rand absolute temperature TE, Stefan's law (appendix 8.1 and Fig. 8.2) requires
the total radiant power output to be
(8.3)
where a is the universal Stefan-Boltzman constant, with value 5.67 x
w-s W m-2 K-4.
The terrestrial energy balance can now be stated by equating expressions (8.2)
and (8.3). After rearrangement to isolate TE we have

Ts = [(1-a) 4as]v. (8.4)

Note that the Earth's radius does not appear in eqn (8.4), showing that the equili-
brium equivalent blackbody temperature TE of the Earth is independent of the
Earth's size and would apply as well to an artificial satellite of radius 1 m as to a
planet the size of Jupiter. In fact the only planetary factor determining Ts is the
albedo a; beyond that we have only the solar constantS, which is part of what we
might call the astronomical context of the planet, and the universal constant a. The
albedo is therefore a highly significant property of the planet. It is observed
(section 8.6) that the Earth's albedo is determined mainly by clouds, air molecules
and dust, as well as by the solid and liquid surface, but that clouds are by far the
most important. We might expect this from simple observations of their brightness
in sunlight, and indeed it is immediately obvious on all satellite pictures taken in
visible wavelengths. Since clouds are produced by the very activity of the
atmosphere which we know must be profoundly affected by the balance repre-
sented by eqn (8.4), it is clear that the atmosphere is significantly self-regulating at
the most fundamental level.
Let us substitute values in the right-hand side of eqn (8.4) to find a value for TE
consistent with radiative equilibrium of the planet. Satellite radiometry shows that
the Earth's albedo is about 0.3 in solar wavelengths, meaning that 300Jo of incident
solar energy is returned to space and therefore not absorbed by the planet. If we use
1380 W m - 2 for S and the established value for a, we find TE to be 255 K
(- 18 °C). This value is well below the observed average surface temperature of the
Earth (288 K or 15 °C), so that at first glance it would seem that the model we are
using for the irradiated and radiating Earth is too simple to be useful. In fact it will
soon appear that the model is unrealistic in only one important respect, whose
correction effectively explains the 33 oc anomaly, and is otherwise extremely
good. However, to remove the model's single major flaw, we need to know more
about the nature of terrestrial radiation. In particular we need to know about the
interaction of terrestrial radiation with the Earth's atmosphere and surface.

8.3 Terrestrial radiation and the atmosphere

So far the model suggests that terrestrial radiation is electromagnetic radiation

emitted by a blackbody with temperature 255 K. Although not quite true, this is
nearly so, and we will obtain a useful first approximation to actual terrestrial
248 radiation if we consider the power spectrum of radiation emitted by a blackbody at
this temperature (Fig. 8.3). This shows significant emission spread across wave- 8.3 Terrestrial radiation
and t he atmosphere
lengths ranging from about 4 to 100 ~tm, with maximum emission per unit wave-
length centred around 12 ~tm, which is consistent with Wien's law (appendix 8.1).
Virtually all of this terrestrial radiation is beyond the infra-red end of the solar
spectrum, and the lack of significant overlap between solar and terrestrial
radiation very conveniently allows the two fluxes to be considered separately. Note
from the axes of Figs 8.1 and 8.3 that the terrestrial radiation spectrum is very
much flatter and more spread out in wavelength than the spectrum of solar
irradiance.

Key
Fig. 8.3 Power spectrum of
terrestrial radiation.
Balloon - born e
interferometer measurements Measurements A and B were
25

o;-E
I \t
N:ack-body
288 K
w
o
Type A
Type B
mode by balloon-borne
spectrometers about 16 km
::>. / ' \ Black - body curves above sea level (above most
20
"'e Bounded area e qual
of the atmosphere active in
:1: to ave r age Earth the for infra-red) .
em i ttance 235 wm - 2 Observations on either side
..""'
c 15 of the peak loosely fit the
blackbody curve for 250 K
·e (bounded area 221 W m - 2).
.."' 10 Observations at the peak fall
short of the level expected
""'
C>
for radiation coming through
<f)
5 the window from a 288 K
surface (see blackbody peak).
consistent with reports of
0
65"/o cloud cover at altitude
0 10 20 30 40 7 km at the time . Allowing
50
for reasonable variability, the
Wavelength (~m)
area under the observed
spectrum agrees with the
235 W m - 2 needed for
As implied already, the radiative properties of many terrestrial materials differ planetary equilibrium . (After
quite markedly between solar and terrestrial wavelengths. Virtually all common J.S. Goring [47])
solids and liquids, including water and all but the most tenuous water and ice
clouds, act as almost perfect blackbodies for terrestrial radiation, absorbing these
wavelengths completely (and therefore allowing no transmission or reflection),
and emitting them as a blackbody would at the same temperature. Further,
although the two main atmospheric gases (nitrogen and oxygen) are effectively
transparent to terrestrial radiation, and are therefore very poor emitters of it,
according to Kirchhoff's law, the minor components carbon dioxide and water
vapour absorb it so strongly that the atmosphere is almost completely opaque in
substantial parts of the spectrum of terrestrial radiation. The details of this
complex behaviour are known from laboratory and theoretical investigations of
the radiative behaviour of the gas molecules in question. A greatly simplified
picture is sketched in Fig. 8.4, which shows that there are two main features: firstly
there is strong absorption by water vapour between wavelengths 5 and 8 ~tm, and
again beyond 14 ~tm by water vapour and carbon dioxide; and secondly there is
near transparency in a window centred at about I 0 ~tm. In the window there is some
absorption by the much rarer gas ozone, which is unimportant for the atmosphere
as a whole, but is quite important for the stratosphere on account of its relatively
small heat capacity (sections 3.3 and 8.6). The presence of the window is signi-
ficant, particularly for conditions at the Earth's surface (Chapter 9), and especially
because it nearly coincides with the peak in the spectrum of terrestrial radiation,
but it is the opaqueness of the atmosphere to wavelengths on either side of the 249
8 Radiation and global window which dominates the overall conditions of the atmosphere in the presence
climate
of terrestrial radiation.
The atmosphere is really very opaque indeed to the wavelengths strongly
absorbed by water vapour and carbon dioxide. The simplified absorption spectrum
in Fig. 8.4(b) shows that a layer of air deep enough to contain 300 g of water
vapour in a vertical column resting on a square metre of horizontal surface
completely absorbs wavelengths between about 5 and 7 ttm, and strongly absorbs
wavelengths in another micrometre range on either side. In conditions typical of
the low troposphere (air density - 1 kg m - 3 , specific humidity - 10 g kg- 1), such
a layer is only about 30 m deep. A similar estimation shows that the same layer
contains enough carbon dioxide and water vapour to absorb all terrestrial
radiation with wavelengths greater than 14 ttm.

Fig. 8.4 (a) Smoothed (a)

spectrum of atmospheric
absorption of solar and Solar Terrestrial
terrestrial radiation. 100
Percentages refer to typical
absorptions by the full depth
of a cloudless atmosphere. c:
0 50
Gaseous absorbers are
labelled where they are most e-
o
effective. (After [ 48]) .c"'
(b) Simplified profile of < 0 +--..........,"---i~">-!r'T'-..L--"="+.......¥-L....:."'--..L-'-i
atmospheric absorption in 0.1 Visible 1.0 5 10 50
and around the window, used Wavelength (IJm)
by Simpson in 1928 in the first
detailed examination of the (b)
Earth's radiation balance-
Terrestrial
an approach not bettered
100
until ways of accounting for
the microstructure of
absorption bands and lines
were developed. Absorption c:
0 50
values refer to a layer C. C02
containing 300 g of vapour o H20
per horizontal square metre. "'
.c
<

5 10 15
Wavelength (IJm)

Consider a package of radiant energy emitted from the Earth's surface in these
heavily absorbed wavelengths. It will be completely absorbed by the first 30 m of
air, warming first the molecules of water vapour and carbon dioxide, and then
almost immediately sharing this heat with the surrounding air molecules. But by
Kirchhoff's law, water vapour and carbon dioxide must emit these same wave-
lengths with the same efficiency as they absorb them; in fact a layer deep enough to
be opaque to these wavelengths must emit as a blackbody at the layer temperature.
(This applies only to the absorbed wavelengths: the same layer will absorb and emit
almost nothing in the wavelength range of the window.) If the air and surface were
at the same temperature, their emissions would be equally intense in the wave-
lengths in question. Unlike the surface however, the layer of air emits both
upwards and downwards.
Things are becoming complex, so we will pause to present an overall picture of
250 the unfolding situation.
8.4 The terrestrial radiation cascade 8.4 The terrestrial radiation
cascade

The picture emerging is that the Earth's solid and liquid surface radiates as a
blackbody and that all of this terrestrial radiation with wavelengths outside the
transparent window is completely absorbed by layers of air no more than a few tens
of metres deep overlying the surface. By contrast a considerable proportion of all
solar radiation passes straight through the atmosphere to the surface, which it
warms on absorption, with only a modest proportion being absorbed directly by
the atmosphere. Is it possible to imagine the Earth's surface, together with all the
overlying atmosphere, being maintained at constant temperatures solely by the
absorption of solar radiation and the absorption and emission of terrestrial
radiation? In other words can the surface and atmosphere achieve a purely
radiative equilibrium? For simplicity assume that the atmosphere consists of a
number of discrete layers each of which is just thick enough to be opaque to
terrestrial radiation with wavelengths outside the window. Such layers are about
30 m deep in the low troposphere, but are considerably deeper at higher levels
where the absorbing gases are much more tenuous. To account for observed totals
of C02 and water vapour there should be over one hundred of these opaque layers
in a deep atmospheric sandwich, but the principle of the argument can be estab-
lished by considering just a few, in fact only three in Fig. 8.5.

Terrestrial Net Solar Fig. 8.5 Schematic picture of

Radiation terrestrial radiation vertical fluxes of terrestrial
radiation in a three-layer
Space model atmosphere in purely
radiative equilibrium.

Layer 3

Layer 2

Layer 1

~~ti0~~ Surface

For simplicity we will initially ignore all direct absorption of solar radiation by
the atmosphere, leaving the surface as the only absorber, and ignore also the
transparent window in terrestrial wavelengths, leaving the atmosphere entirely
transparent to solar radiation and opaque to terrestrial radiation.
From Fig. 8.5 we see that the surface receives and absorbs solar radiation, emits
terrestrial radiation upwards, and absorbs terrestrial radiation coming downwards
from the overlying opaque atmospheric layer. Each opaque atmospheric layer
emits terrestrial radiation upwards and downwards, and absorbs terrestrial
radiation emitted by the overlying and underlying adjacent layers. The only
exception is the highest opaque layer, which absorbs terrestrial radiation coming
up from beneath, and emits terrestrial radiation directly to space, but receives no
terrestrial radiation from above because it is the highest layer containing signi-
ficant quantities of carbon dioxide and water vapour and there are no terrestrial
bodies close by.
If the surface is to be in purely radiant thermal equilibrium, then it must lose as
much energy by net loss of terrestrial radiation (emission less absorption) as it gains 251
8 Radiation and global by absorption of solar radiation. And if each atmospheric layer is to be in radiant
climate equilibrium, it must lose as much energy by emission of terrestrial radiation as it
gains by absorption of the same. In addition, the net upward flux of terrestrial
radiation between every adjacent pair of atmospheric layers must be the same, and
be equal both to the net upward flux of terrestrial radiation between the surface
and the first atmospheric layer, and to the upward flux to space from the highest
opaque layer. The end result of all this would be to keep all surface and
atmospheric components separately in radiant equilibrium, as well as the planet as
a whole, while maintaining a continual conversion from solar to terrestrial
radiation at the Earth's surface.
To sustain the net upward flux of terrestrial radiation through the opaque
atmosphere, the emissions of terrestrial radiation by the opaque atmospheric
layers must decrease with increasing altitude as depicted in Fig. 8.5, and since
decreasing emissions require decreasing temperature (Stefan's law again), it
follows that there must be a continuing lapse of temperature from a maximum
value at the surface to a minimum value at the highest opaque layer.
We can now modify the unrealistically simple initial assumptions by reinstating
some direct atmospheric absorption of solar radiation, and allowing for the
presence of an atmospheric window for terrestrial radiation.
Since there is in fact significant absorption of solar radiation in the lower layers
of the atmosphere (see section 8.6), the net upward flux of terrestrial radiation
must decrease with increasing altitude, reaching the value found in the initial model
only at levels above the highest absorbing layer. This means that the temperature
lapse between adjacent layers must increase upwards when compared with the
initial model, beginning at a value smaller than the corresponding value in the
initial model and increasing upwards throughout the layer with significant solar
absorption. The effect is therefore to reduce the temperature lapse required for
thermal equilibrium, at least in the lower part of the radiatively active atmosphere,
holding it a little below what would be required if all solar absorption were con-
centrated at the surface.
The inclusion of the transparent window centred on terrestrial radiation wave-
lengths simply allows some of the net upward flux of terrestrial radiation from the
surface to be carried out to space independently of the cascade of atmospheric
absorption and emission. Again the result is to reduce the temperature lapse
required for thermal equilibrium by comparison with the initial model, especially
between the surface and atmosphere. Since detailed study (section 8.6) shows that
the window flux accounts for less than lOOJo of the Earth's radiative output, the
effect is not large.
We can summarize the overall picture by saying that the surface and atmosphere
could maintain thermal equilibrium by absorption of solar radiation and an
inverted cascade of absorption and emission of terrestrial radiation, if there were
an adequate temperature lapse between each atmospheric layer and between the
surface and the lowest layer. However, to estimate values for these temperature
lapses, thereby detailing the vertical profile of temperature needed for purely
radiative equilibrium, we must have a thoroughly quantitative version of the model
depicted so simply in Fig. 8.5. This was first attempted by Emden in 1913 and has
been repeated since with increasing refinement. All such studies indicate that
the temperature profile required for purely radiative thermal equilibrium of the
surface and stratified atmosphere is so steep as to be convectively unstable in the
lower atmosphere (section 5. 9). Since the air is much too fluid to sustain convective
instability without breaking into vigorous convection, with consequent significant
vertical transport of heat by that convection, it follows that the answer to the
252 question posed at the beginning of this section is 'no': it is impossible for the atmo-
sphere with its observed distribution of radiatively active and inactive gases to 8.5 The greenhouse effect
maintain thermal equilibrium purely by absorption of solar radiation and the
emission and absorption of terrestrial radiation. Thermal equilibrium requires
convection as well as radiation, and it follows that the Earth's atmosphere is
necessarily rather than incidentally dynamic.

8.5 The greenhouse effect

We are now in a position to explain the apparent anomaly highlighted at the end of
section 8.2 - the difference of over 30 oc between the average surface
temperature of the Earth and the effective blackbody temperature of the planet
as a whole. Contrary to the original assumption, the planet does not act as a
blackbody with a single emitting surface situated at the land and sea surfaces:
the terrestrial radiation emitted to space comes mainly from the atmosphere and
only in small part from the solid and liquid surfaces. In fact the latter is significant
only when the atmosphere is cloudless; from the half of the globe on average
covered by cloud, even the window radiation comes from the top of the highest
cloud layer.
The details summarized in the next section show that about 900Jo of the terrestrial
radiation emitted to space comes from the atmosphere (10% by window radiation
from cloud tops and 80% by radiation outside the window emitted by water vapour
and carbon dioxide in the highest opaque layer of the atmosphere), with only 10%
coming through the window from the underlying land and sea. Detailed analysis
also shows that the highest opaque layer is in the upper troposphere, where the
temperature is usually 40° or more below the temperature of the underlying
surface. Cloud tops are often in the same height and temperature range. The
output of terrestrial radiation is therefore dominated by the output from the upper
troposphere, and this, together with the temperature lapse imposed by the inverted
cascade of absorption and emission of terrestrial radiation through the tropo-
sphere (constrained by the associated convection), means that the effective
temperature TE of the terrestrial blackbody is well below the temperatures familiar
to us at the bottom of the atmosphere. In fact detailed analysis confirms that TE is
very close to the 255 K estimated in section 8.2.
The effect of the atmosphere's near transparency to solar radiation and near
opacity to terrestrial radiation is therefore to raise the temperatures of land and sea
surfaces well above those which would prevail in the absence of the atmosphere.
This elevation of surface temperatures is called the greenhouse effect because the
glass of a greenhouse is similarly transparent to solar and opaque to terrestrial
radiation. Actually the name is quite misleading because the interior of a green-
house stays warm primarily because the glass inhibits convective heat loss to the
surrounding air - a fact which was established at a fairly early stage by studying
the behaviour of a very special quartz greenhouse, and is now confirmed by the
growing popularity of polythene greenhouse (quartz and polythene being trans-
parent to both solar and terrestrial radiation). However, since in spite of this the
term' greenhouse effect' has become conventional, it will be used conventionally in
this book, with the understanding that it refers to the behaviour of atmospheres
rather than greenhouses. 253
8 Radiation and global VENUS
climate
The planet Venus has an atmosphere about 100 times the mass of the Earth's,
consisting almost entirely of C0 2 and a thick shroud of sulphuric acid cloud, both
of which contribute to a very large greenhouse effect. The highly reflective cloud
blanket has an albedo of about 0.7 (making Venus such a brilliant object in the
night sky), which results in an effective blackbody temperature of only 245 K,
compared with the Earth's 255 K with less than half the solar constant (section 8.2
and problems 1 and 11). Strongly protected landing probes have measured surface
temperatures of about 730 K, implying a greenhouse effect of nearly 400 K, over
ten times the terrestrial value. Estimates of total amounts of carbon in the surface
layers (atmosphere, ocean and crust) of Earth and Venus show that they are
similar, but on Earth it is mostly chemically locked up in carbonate rocks, while on
Venus it largely degassed into the atmosphere as C0 2 during an early runaway
greenhouse effect (section 3 .8).

RADIATIVE FORCING

As noted in sections 3.8 and 4.10, the Earth's greenhouse effect is believed to have
changed considerably over geological time, and to be increasing now in response to
wholesale injection of carbon dioxide into the atmosphere by artificial combus-
tion. Other anthropogenic gases are known to add to the greenhouse effect by
absorbing terrestrial radiation and detailed estimates have been made of their
present and future effectiveness, assuming realistic future production rates. The
simplest way to assess this greenhouse effectiveness is to calculate the radiative
forcing of the terrestrial energy balance - the resulting increase in the trapping of
outgoing terrestrial radiation.
Table 8.1 presents recent results for two overlapping periods, each starting from
the dawn of the industrial revolution, and the second projecting into the middle of
the twenty-first century, by which time the atmospheric C0 2 level is expected to
have doubled. Values represent the increases in the rate of absorption of terrestrial
radiation by gases in the periods, and are calculated from detailed knowledge of
absorption spectra and likely distributions through the troposphere. The results
show significant contributions from increases in C0 2 and a number of other
greenhouse gases, differing sharply in efficiency per molecule - the CFCs being
effective out of all proportion to their very small concentrations. The calculations
suggest that the increase in total forcing since pre-industrial times is likely to be
over 8 W m - 2 by the middle of the twenty-first century, if human society continues
growing and industrializing.

Table 8.1 Changes in radiative forcing from 1765 assuming the

following IPCC scenarios for industrial and population growth [33]:(a)
business as usual; (b) accelerated control policies (C0 2 emission half of
1985 value, stringent controls in developed countries, moderate controls
elsewhere)

Period (AD) C0 2 CH4 N 20 CFCs Total (all

W m- 2 )

1765-2000 (a) 1.85 0.69 0.12 0.29 2.95

(b) 1.75 0.59 0.11 0.29 2.74
1765-2065 (a) 5.49 1.37 0.40 1.03 8.28
(b) 2.77 0.52 0.24 0.68 4.22
254
 8.6 Average radiant-energy
GLOBAL WARMING budget

By agreed definition, radiative forcing is the reduction in net upward flux of

terrestrial radiation at tropopause levels produced by increasing absorption by
greenhouse gases. Assuming this flux to be about 240 W m - 2 initially (equivalent
to an effective terrestrial blackbody temperature of 255 K - section 8.2 and
Fig. 8.6), this would be reduced to 232 W m - 2 by the above forcing. The Earth
would then warm by excess of solar input over reduced terrestrial output, until the
flux had been restored to its original value. The global distribution of such
warming depends crucially on details of how the atmosphere and surface layers
react to forcing, which is probably very complex (see below), so the simplest
approach is to find what rise in the equivalent blackbody temperature of the Earth
would recover the balancing outward flux of terrestrial radiation. According to
appendix 8.5, a forcing of 8 W m - 2 requires a rise of just over 2 K at around
255 K. This is a very substantial global warming which, depending on its distribu-
tion at the Earth's surface, could make the Earth warmer than at any time since the
onset of the present ice ages, and fulfil the worst fears expressed in section 4.10.
The distribution of the warming depends strongly on the complicating role of the
water vapour and cloud.
Table 8.1 omits the most effective greenhouse gas of all (water vapour) because it
is not directly anthropogenic. It has an important indirect effect however, because
any warming of the low troposphere will increase the vapour content there by
increasing evaporation, and the resulting increased cloud will blanket more of the
atmospheric window. Both these effects would lead to even greater warming, but
the increase in global cloud cover will also have a cooling effect, being very
effective in cutting down the solar input to the Earth (see next section). Details of
this complex of positive and negative feedbacks are still poorly understood,
because we need to be able to predict the vertical and horizontal distributions of
increased cloud cover, and indeed changes in many other features of the whole
climate system. Most current estimates of average surface warming range from
about 2 to 5 oc for a doubling of atmospheric carbon dioxide (with further
additions for other greenhouse gases), but the confidence limits on precise values
are not high [33]. It is very unfortunate, to say the least, that our understanding of
such global consequences of our modern way of life remains so poor, but the
problem is exceptionally severe and is likely to tax our pooled scientific resources
for decades before the situation becomes clear, for good or ill.

8.6 Average radiant-energy budget

Simplified models of interactions between the atmosphere and solar and terrestrial
radiation typified by Fig. 8.4 are very useful for indicating basic aspects of
behaviour, but they are inadequate for more detailed studies. In the last 50 years,
people have developed ways of dealing with the complex fine structure of lines and
bands underlying the gross absorption spectrum sketched in Fig. 8.4(a), with the
result that interactions between the real atmosphere and the radiation streams
crossing it are now understood in some detail. (Methods are discussed in more
advanced texts [8].) It has therefore been possible for some time to use observed
distributions of all the radiatively active materials to calculate all significant items 255
8 Radiation and global in budgets of radiant energy for the Earth's atmosphere and surface. An immense
climate amount of data is required, including appropriately averaged distributions of
temperature, pressure, water vapour, carbon dioxide, ozone, cloud and surface
albedo. Since the early 1960s, data from meteorological satellites have been used to
improve some of the data and to check some of the predicted fluxes.
A classic study of the radiant energy budget on the global scale was made by
London in 1957 and has been refined since [49]; results are summarized in Fig. 8.6.
Input data were averaged over many years from a large number of geographical
locations. The data were further averaged over ten degree zones of latitude before
being used in the extensive calculations needed to estimate radiant fluxes. The
results reveal the extremely important variation with latitude which is discussed in
the next section. For the global picture, the data were averaged further to produce
what is in effect an annual global average of the budget of radiant energy
(Fig. 8.6). Since all horizontal variations have been removed by averaging, the
picture represents the vertical distribution of radiant fluxes. For arithmetical
simplicity all values in Fig. 8.6 are quoted as p~rcentages of the annual global
average rate of input of solar energy per unit horizontal area at the top of the
atmosphere. Note that the latter is one quarter of the solar constant (i.e.
345 W m - 2) since the surface area of the spherical Earth is four times its cross-
sectional area (Fig. 8.2). Each unit in the budget in Fig. 8.6 is therefore very nearly
3.5 W m- 2 •

Fig. 8.6 Global annual

average flux densities of
solar, terrestrial and net
Terrestrial
f.. -10 ~m
Solar
f..- 0.5 ~m
=345 wm-
1 OOunits
2

radiation in the stratosphere,

- t-
troposphere and at the

-----------r---
67 Albedo 33
Earth's surface. On the solar Input
radiation side, dotted lines 59 Net 100 4.5 26 2.5
5 3

-~-
represent reflection and
scattering, and solid lines -- -r--
I
represent transmission and C02 I -3 0
absorption. On the terrestrial Oa I

radiation side, solid lines

represent emission, (Window) I
I
transmission and absorption. I
The central column contains I
H20 C02 I
the net radiative input to Clou I
I

~~[-~-
each part of the system.
:-49 -30
(Values are from [49])
I
I
I
I
I
-----r---
I
98 113 1-15 +30
I

THE ENERGY BUDGET OF THE EARTH'S SURFACE

After traversing the atmosphere, solar radiation reaches the surface as both direct
256 and diffuse radiation. The major diffuser is cloud, though air molecules and haze
are also important. The scattering of light by air molecules was first described 8.6 Average radiant-energy
theoretically in the nineteenth century by Rayleigh, who showed that shorter wave- budget
lengths are much more efficiently scattered than longer wavelengths when the
scattering bodies are much smaller than the scattered wavelengths (molecules -
I0- 4 J.Lm as compared with wavelengths - 0.5 J.Lm). This accounts for the blue
appearance of cloud-free sky away from the Sun's disc, since the scattered light we
receive is enriched in the short (blue) wavelengths by the scattering from other
initial directions. It also accounts for the reddening of the solar disc when observed
directly through the atmosphere, since the preferred scattering of the blue out of
the line of sight leaves the direct sunlight reddened. The reddening is particularly
noticeable when the sun is near the horizon because the long path length through
the atmosphere enhances the blue-biased molecular scattering, and the unbiased
scattering by much larger droplets and particles (see below) makes the solar disc
dim enough to view in comfort.
By contrast clouds, mists and most hazes do not alter the colour of sunlight
scattered by their droplets and particles, and scattering theory shows that this is
because the size of the scatterers is comparable with or larger than the scattered
wavelengths. Clouds therefore appear nearly white in normal sunlight, and are red
near dawn and dusk only because they are illuminated by reddened direct sunlight.
The chromatically uniform scattering of all available wavelengths is called diffuse
reflection, and Fig. 8.6 shows that this accounts for 21 out of 47.5 units of solar
radiation reaching the surface, compared with 7 units scattered by air molecules
and dust. Collectively labelling these as diffuse radiation we see that considerably
more than one half of the solar radiation reaching the surface is diffuse rather than
direct.
Although the distinction between the largely directionless diffuse component of
solar radiation and the highly directional direct component is visually obvious, and
can be important for the local climatology of hilly terrain (section 9.2) for example,
it is largely irrelevant to the energy budget of the Earth's surface as a whole, where
the total input is all that matters. Over the whole Earth almost all solar radiation
incident on the surface is absorbed there, if not at once then ultimately (there is
multiple scattering between the surface and the low atmosphere which is not
depicted in Fig. 8.6), only 2.5 units being scattered back to space. The effective
albedo of the Earth's surface is therefore very small, as is apparent in satellite
panoramas such as Fig. 1.1, where the surface is very dark in comparison with the
highly reflective clouds. Of course in localized areas surface albedos may be much
larger, and can exceed 700Jo in snow-covered areas, but the global average value is
low because of the preponderance of dark surfaces. Some values for surface
albedoes are given in Table 9.1.
Just above the surface, terrestrial radiation consists oflarge and opposing fluxes
which have a relatively small upward resultant. The upward flux in Fig. 8.6 is
equivalent to the output from a blackbody with temperature 288 K (15 oq -
essentially the average temperature of the Earth's surface. Cloud, carbon dioxide
and water vapour radiate strongly downwards, but do not quite match the upward
radiation from the surface because of the absence of any downward flux in the
window when there is no cloud, and because the lowest opaque atmospheric layers
are somewhat cooler than the surface almost everywhere. Although each of the
opposing fluxes is larger than the solar input to the planet (another aspect of the
greenhouse effect), the net upward flux is much smaller, and is significantly smaller
even than the net solar input to the surface.
Considering solar and terrestrial radiation together, it is apparent from Fig. 8.6
that there is a net input of radiant energy to the surface which proceeds at an
average rate of 30 units- very nearly 100 W m- 2 • 257
8 Radiation and global THE ATMOSPHERIC ENERGY BUDGET
climate
Consider the stratosphere first. Three units of solar input are absorbed, mainly in
the form of selective absorption of soft ultraviolet between altitudes of 25 and
45 km as described in section 3.3. This may seem a relatively small input, but the air
at those levels is so insubstantial that it would warm very rapidly if the input were
not balanced by the small net output of terrestrial radiation from the carbon
dioxide and ozone there. (There are no significant amounts of water vapour at
these levels.) The stratosphere as a whole is therefore in radiative equilibrium. This
does not mean that there are no convective heat fluxes either within the strato-
sphere or between stratosphere and troposphere, but it does indicate that
convection is not essential to the stratosphere in the way that we have seen it to be in
the troposphere.
Of the 97 units of solar radiation entering the top of the troposphere (Fig. 8.6),
16 units are absorbed by aerosol particles and water vapour, 30.5 units are
scattered back out to space, mainly by cloud, 3 units are absorbed by cloud, and
47.5 units pass through to the surface either directly or after the scattering
discussed above. Note the curious behaviour of cloud: on the one hand its diffuse
reflection both upwards and downwards dominates the solar radiation received at
the surface and returned to space, but on the other hand it contributes only slightly
to tropospheric absorption of solar radiation. Total tropospheric absorption is far
from trivial, accounting for 19 units, or about one third of the total solar
absorption by the planet. We are misled by our visual impression of the near-
transparency of the cloudless atmosphere, because the absorption is concentrated
in the infra-red half of the solar spectrum (Fig. 8.4).
Because of the distributions of temperature, water vapour, carbon dioxide and
cloud, the output of terrestrial radiation from the troposphere to space is 59 units,
much of it from the high troposphere. (By definition the flux in the atmospheric
window plays no role in the tropospheric budget unless it is intercepted by cloud.).
This large loss is only partly offset by the net gain of 10 units by terrestrial exchange
with the surface, so that the troposphere suffers a net loss of 49 units by terrestrial
radiation.
Including both solar and terrestrial radiation, it is apparent that the troposphere
suffers a net loss of 30 units of radiant energy, which is equivalent to 100 W m - 2,
and is exactly equal to the net radiative gain by the surface.

THE SURFACE AND TROPOSPHERE TOGETHER

These therefore comprise a system which is in radiative equilibrium overall, like the
stratosphere and like the Earth as a whole, but which has a large radiant energy
imbalance between its two components, amounting to 100 W m - 2 on a global
annual average. Since there is no appreciable warming of the surface, or cooling of
the troposphere, from year to year, it is clear that this imbalance must be almost
exactly offset by a non-radiant heat flux of 100 W m - 2 from the surface to the
troposphere. This is the convective heat flux associated with the whole range of
weather systems from the smallest buoyant thermals to the mighty Hadley
circulation. The radiative balance of the stratosphere strongly suggests that this
convective flux is confined to the troposphere, and this is confirmed by a variety of
observational evidence, including the virtual confinement of cloud to the tropo-
sphere, and the widespread appearance of convectively very stable layers just
258 above the tropopause.
8. 7 Meridional distribution of radiative fluxes 8.7 Meridional distribution
of radiative fluxes

The global energy budget introduced in section 8.6 includes as one of its most
important parts a calculation of the distribution of radiative fluxes with latitude.
Annually averaged values are displayed in Fig. 8.7(a) and (b) and agree well with
direct measurements by satellite-borne radiometers. Values should be compared
with one quarter of the solar constant as mentioned at the beginning of the last
section. The latitude scale has been deliberately chosen so that equal lengths repre-
sent zonal rings of equal surface area, and it follows that equal areas under the
curves in Fig. 8.7 represent equal energy flows to or from the Earth. This would
not be so if the latitude scale were linear, since surface areas and fluxes would be
exaggerated at high latitudes, just as in the familiar Mercator map projection.
Figure 8. 7(a) depicts the meridional distribution of absorbed solar inputS to the
surface and atmosphere and the terrestrial output T to space. The solar input is
concentrated in low latitudes because the sun passes near the zenith there each day,
whereas at higher latitudes it misses the zenith by an angle which increases with
latitude, and is equal to the angle of latitude at the equinoxes (Fig. 8.8). The
increase of obliqueness of illumination with increasing latitude has the effect of
spreading the solar beam over greater surface areas at higher latitudes, and of
increasing its path length through the atmosphere (Appendix 8.2). Both these
factors enhance the loss from the direct solar beam by scattering and diffuse
reflection, and increase the relative proportion of solar energy which is absorbed
by the atmosphere rather than the surface. This proportion is enlarged further by
the widespread tendency for surface albedo to increase with zenith angle, as
graphically evidenced by the brilliance of surface glitter beneath the rising or
setting sun.

(a) Fig. 8.7 (a) Meridional

profiles of zonally averaged

=
fluxes of solar and terrestrial
300 Key radiation absorbed and
N
I
"
260 ~
S Absorbed solar
power
emitted by the Earth. The
E vertical axis shows the
3: "' T Emitted terrestrial
250 :;;
radiant flux density per unit
200 c. power
240 E
;= 230 !
Satellite area of Earth surface.
measurements
"'c 220 Appendix 8.3 justifies the
"0
Q)
100 210 >
200:;::
" cosine weighting of the
X <.> horizontal axis. (b) Fluxes in
u.. " "
w
(a) analysed into: (i) the net
radiant flux to the planet; (ii)
0
the net radiant flux to the
90 60 30 0 30 60 90
surface; and (iii) the net
ON Latitude OS
radiant flux to the
(b)
200 atmosphere. (After [ 49])
N
I
Key
E
3: i Net flux to the planet
(S-T in (a))
ii Net flux to the surface
"'c
Q)
iii Net flux to the
atmosphere
"0
X

u.."

259
8 Radiation and global
climate ~ Noon sun
I
I _..........June solstice

December solstice

Fig. 8.8 Angles subtended

by the Sun at noon at P at
equinoxes and solstices,
using the Earth's equatorial
plane as reference plane (so
that the Sun seems to rise
and fall by 23.5° seasonally).

In annual averages, the resulting concentration of solar input Sin low latitudes
would be even more marked than it is, if it were not for the 23.5° angle between the
earth's solar orbital and equatorial planes (Fig. 8.8), because of which the
midwinter noon zenith angle is increased by 23.5° in middle and high latitudes, and
the midsummer value is similarly decreased. The seasonal meridional migration of
maximum solar input has the effect of blunting and spreading the low-latitude
maximum and the high-latitude minima in the annual averages. This is enhanced
by the persistently non-uniform distribution of cloud, with concentrations in
equatorial areas corresponding to the intertropical convergence zone, and relative
scarcity corresponding to the semi-permanent anticyclones concentrated in the
subtropics. The slight bias of S away from the Antarctic is due to the presence of
the permanent ice cap there, which scatters more sunlight to space than does the
seasonally broken sea ice of the Arctic ocean. In principle there must also be a slight
bias of S toward the southern hemisphere because perihelion occurs in the southern
summer, but the expected 60Jo effect (section 8.1) is swamped by those already
mentioned.
The meridional variation of absorbed solar inputS is very large and is obviously
related to the familiar zoning of climate. By comparison, the meridional variation
of the emission of terrestrial radiation Tis very much smaller. This emission comes
mainly from the upper troposphere, as discussed in section 8.5, much of it from the
water vapour there. Since continuing convection keeps most parts of the upper
troposphere close to saturation, the vapour density there is effectively determined
by air temperature alone (Fig. 6.2). And since the highest opaque layer must have a
well-defined vapour density, to be only just opaque, it follows that its temperature
and its emittance of terrestrial radiation must be similarly well defined. We should
therefore expect a relatively uniform distribution of terrestrial output T over the
globe. In fact the presence of carbon dioxide, cloud and the atmospheric window
ensure that there is some falling away of Tin high latitudes, but not very much.
Looking at the very different meridional distributions of Sand Tin Fig. 8. 7(a),
it is clear that there is a gross imbalance between low latitudes, which gain much
more radiant energy than they lose, and high latitudes which lose much more than
they gain. (There is of course no net gain or loss by the planet as a whole, as careful
checking of the areas between the Sand Tcurves will confirm.) If these unbalanced
260 fluxes of radiant energy were not offset by other types of energy exchange, then
high latitudes would cool and low latitudes would warm, until presumably their 8.8 Seasonal variations of
radiative fluxes
terrestrial output achieved the same steep meridional gradient as their solar input.
Some of the offsetting poleward fluxes must consist of an exchange of warm and
cool air, called advection rather than convection because the exchange is horizontal
rather than vertical. Clearly the familiar experience in middle latitudes of relatively
warm air moving polewards and relatively cool air moving equatorwards (cool
northerlies and warm southerlies in the northern hemisphere) represents just such
poleward advection of heat, as does the great poleward flux of relatively warm sea
water in the Gulf Stream of the North Atlantic, though the contrasting cool
equatorward drift is confined to the opposite side of the ocean. The role of the
atmosphere and oceans in advecting heat polewards is examined further in the next
chapter.
The solar and terrestrial fluxes in Fig. 8.7(a) are regrouped in Fig. 8.7(b) to show
the meridional profiles of flux densities of net radiative input to the planet (i), to
the surface (ii), and to the atmosphere (iii).
The first of these (i) is S- Tin Fig. 8. 7(a) and clearly portrays the meridional
imbalance already discussed. The net radiative flux to the Earth's surface (ii) has a
similar shape to (i), but is everywhere about 100 W m- 2 larger. It exceeds
100 W m - 2 (the global average net gain by the surface- see Fig. 8.6 and section
8.6) over the half of the Earth's surface area lying between latitudes 30°, and falls
below 50 W m - 2 only at latitudes greater than about 60°. By contrast the net
radiative input to the atmosphere (iii) is negative and remarkably uniform,
remaining close to - 100 W m - 2 at all latitudes, which obviously therefore agrees
closely with the global average tropospheric net loss apparent in Fig. 8.6.
The combination of the steep meridional profile of the net radiative input to the
surface ((ii) in Fig. 8. 7b) and the near uniformity of the net radiative loss by the
atmosphere (iii) means that in low latitudes the net radiative gain by the surface
exceeds the net radiative loss by the atmosphere, whereas the position is reversed in
high latitudes. To accommodate this strong meridional variation in the radiative
imbalance between surface and atmosphere requires more than the simple vertical
convection of heat envisaged in the previous section and the poleward advection
outlined above: there must be a consistent transport of heat both upward and
poleward, taking heat from the radiatively warming surface of low latitudes to the
radiatively cooling upper troposphere of high latitudes. Such transport could in
principle be accomplished by a combination of convection and advection, but
poleward-moving air is generally warmer than the surface it overruns, which tends
to suppress convection from the surface. In fact a significant fraction of the
required transport is effected by large-scale, slightly tilted flow or slope
convection, which is particularly associated with the development of extratropical
cyclones. Some of these tilted flows become convectively unstable through the
interleaving of flows from different surface sources (section 11.4), and through
destabilization by vertical stretching associated with large-scale horizontal
convergence (section 10.3), encouraging additional vertical transport. In addition,
the heat capacities of land and sea surface, especially the latter, allow heat carried
polewards in one flow to be pumped aloft by convection in the following
equatorward flow.

8.8 Seasonal variations of radiative fluxes

In middle latitudes the seasonal rhythm of solar input is one of the most prominent
features of weather and climate, even in the cloudy and relatively temperate 261
8 Radiation and global 400
climate
so
300

200
N'
I
E
::: 100
Fig. 8.9 Meridional profiles ,.,
of zonally averaged solar,
terrestrial and net radiant "'c:
Q)
0
input (S, T and N respectively)
to the Earth, averaged
">:
:::>
through the seasons u.
beginning in June and -100
December (J and D
respectively). For example
the curve NJ is net radiation -200
in the season June, July and
August. (Data from [50]) Latitude

western margins of continents. Figure 8.9 shows that the difference between the
winter and summer values of the solar input to the Earth's surface and atmosphere
exceeds 100 W m - 2 poleward of about latitude 25 o, and is about 150 W m- 2 over a
considerable part of the middle latitudes. This represents a variation of lOOOJo at the
lower-latitude limit and much more at higher latitudes, and is a direct consequence
of the march of the Sun into first one hemisphere and then the other, modified
somewhat by the uneven distribution of land and cloud. There is obvious
asymmetry between the Arctic and Antarctic summers, the solar absorption in the
latter being considerably reduced by the highly reflective ice cap there, as
mentioned in the previous section.
The seasonal variation in the output of terrestrial radiation is by contrast
extremely small, being significant only in the northern hemisphere because of the
concentration of land masses there. The relatively clear skies over the continental
interiors in middle and high latitudes inhibit the greenhouse effect somewhat,
enhancing the seasonal swing in surface temperatures and encouraging a modest
seasonal variation in the output of terrestrial radiation through the relatively large
atmospheric window.
The combination of the relatively small seasonal variation of terrestrial output
and the very large variations in solar input means that there is a very substantial
shift in the net radiative balance between summer and winter, the latitude zone of
maximum net input shifting from about 15 °N in the northern hemisphere summer
to 25 os six months later. Note that the southern summer maximum of net
radiative input is significantly higher than the northern one, and is centred further
from the equator. Figure 8.9 shows that this results partly from the enhancement of
solar input in the southern summer and partly from the slight reduction in
terrestrial output. The concentration of land in the northern hemisphere, the
December maximum of insolation because of the perihelion then, and the complex
response of the atmosphere, especially its cloud distribution, all play their parts no
doubt, but there is no obvious way of estimating their relative importance.
Apart from the latitudinal position and magnitude of the maximum of net
radiative input, the other major feature of Fig. 8.9 is the extensive meridional
gradient of net radiative input stretching from the maximum in the summer
262 hemisphere to the minimum at or near the winter pole. The algebraic difference
between the positive maximum and the strongly negative minimum is nearly 8.9 Diurnal variations of
500 W m - 2 , and the steep and extensive gradient between them is a major factor in radiative fluxes
maintaining the relatively intense weather activity typical of middle and high
latitudes, especially in the winter. We can trace the connection as follows: the
gradient of net radiation extends and steepens the meridional temperature gradient
of the troposphere in the winter hemisphere, especially in middle and high latitudes
(as noted in Fig. 4.6), and with it the baroclinity which drives much of the large-
scale weather activity there, including of course the extratropical cyclones. In the
summer hemisphere the difference between maximum and minimum net radiation
is smaller and much less extensive, being largely confined to latitudes greater than
30°. To judge from observed behaviour, this smaller and less strongly baroclinic
zone is considerably less effective in maintaining synoptic and larger-scale weather
activity, with the result that weather systems such as extratropical cyclones are less
numerous and vigorous than in the winter hemisphere.

8.9 Diurnal variations of radiative fluxes

The average surface budget in Fig. 8.6 conceals large diurnal variations. The solar
input, averaging 45 units overall, is obviously zero at night, and must therefore
average 90 units in daylight, assuming a 12-hour day and no variation with latitude.
Further assuming a semi-sinusoidal profile of solar influx from dawn to dusk
(appendix 9.2), the noon maximum is 141 units (486 W m - 2), which is clearly
consistent with values estimated from solar elevation and atmospheric
transmissivity (appendix 8.2). Maximum noon values (under a zenith sun and a
clear sky), can reach about 320 units (1100 W m - 2), and may be reduced to about
one tenth of this under thick overcast. Maximum values fall with decreasing solar
elevation (increasing latitude at a fixed time of year) as outlined in appendix 8.2.
Figure 8.6 also shows that on average the Earth's surface loses 15 units by net
output of terrestrial radiation. Since five of these are lost direct to space through
the atmospheric window, which is closed about 500Jo of the time by cloudy over-
cast, the net loss by the surface in cloudless conditions must be about 10 units
(nearly 35 W m - 2), with somewhat larger values in areas with less atmospheric
water vapour and therefore a larger atmospheric window, and smaller values
elsewhere. (These values are consistent with values derived differently in section
9.3, and with radiometer measurements.) Corresponding values are roughly halved
under cloud. There is little clear diurnal variation because the tendency to greater
emission by the hot sunlit surfaces is offset by the greater vapour content and cloud
cover maintained by convection.
It follows from all these values that there can be very large diurnal variations
about the average value of 100 W m - 2 quoted in Fig. 8.6 for the net solar and
terrestrial radiative input to the Earth's surface. On cloudless days at low latitudes
a maximum net input of about 1000 W m - 2 at noon may give way to a net output
of about 70 W m- 2 during the ensuing cloudless nights. In thick overcast
conditions, the maximum net daytime input could be reduced to about 100 W m - 2 ,
with zero nocturnal loss under a warm, low cloud base. At higher latitudes the
effect of the reduction in noon solar input is modulated by the increasing seasonal
variability of daylength (section 11.8 and [51]), and net losses of about 70 W m - 2
continue throughout the polar nights. 263
8 Radiation and global The impact of these diurnal variations on surface climate is outlined in sections
climate 9.3 and 9.4, where it is shown that the small effective heat capacities of land
surfaces allow the large swings of temperature from early afternoon maxima to
dawn minima which are such a familiar feature of cloudless conditions. Tempera-
ture variations at water surfaces, however, are much smaller on account of their
much greater heat capacities. In addition to their relevance to surface climates,
these diurnal variations are obviously consistent with the observed tendency for
overland convection to peak in the afternoon and die away in the evening.

8.10 The necessity of air motion

We have seen that interactions between solar and terrestrial radiation and the
Earth's surface and atmosphere maintain apparent energy imbalances between the
surface and troposphere, and between low and high latitudes. When these
imbalances were mentioned earlier, it was simply assumed that they must be
balanced by convective and advective heat fluxes because the surface is observed to
be in a thermally nearly steady state in yearly averages. Actually, before this
assumption can be safely made, we need to consider the magnitudes of the
imbalances, to assess their significance and to see if they really do require balancing
convection or advection of heat, rather than conduction.
We can demonstrate the significance of the imbalances most directly by
estimating the rates of change of temperature which would occur if there were no
balancing non-radiative fluxes. Consider for example the vertical imbalance
between surface and troposphere, by which the troposphere is losing radiant
energy at an average rate of 100 W m - 2 (Fig. 8.6). We can repeat the calculation in
section 5.3 to find the associated rate of cooling by estimating the mass and hence
the heat capacity of the air in a column of the troposphere resting on unit area of
horizontal surface. Since the troposphere is in hydrostatic equilibrium, we can use
eqn (4.6) to calculate its mass per unit area from the pressure difference between
the bottom and top of a typical tropospheric column. Such columns are con-
sistently taller in terms of pressure as well as height in lower latitudes (Fig. 4.6), so
we will choose an intermediate example, bounded below and above by the 1000 and
250 mbar isobaric surfaces. This involves a pressure difference of 750 mbar or
7.5 x 1()4 Pa, and it follows from section 5.3 that a vertical tropospheric column
resting on a horizontal square metre contains about 7.5 tonnes of air and has a heat
capacity of about 7.5 MJ oc- 1 •
In response to a net rate of loss of heat of 100 W, this column of air would cool at
about 1.3 x 10-s oc per second, which corresponds to about 1.1 oc per day. This
is obviously a very significant rate of cooling, being much faster than even the most
rapid seasonal cooling. Indeed, if it persisted for only a few weeks it would produce
catastrophic climatic change. Comparable and even faster rates of cooling are
observed in middle and high latitudes, but these do not arise primarily by radiative
processes, being associated with the passage of contrasting air flows embedded in
extratropical cyclones; as such they are localized and last for only a few days before
being as quickly reversed.
A similar calculation related to seasonal variations in net radiant energy budget
of the surface and atmosphere in middle and high latitudes (Fig. 8.9) would seem
264 to indicate tropospheric cooling of about 100 oc in a three-month winter season at
latitude 45° (where the net rate of loss is again about 100 W m - 2}. However, this 8.10 The necessity of air
cooling ignores the heat capacity ofthe Earth's surface layers, which are relatively motion
large in oceanic zones. For example it is easy to show that a 30m deep column of
water beneath 1 m2 of water surface has a heat capacity of about 126 MJ oc- 1,
which is over 16 times larger than the tropospheric heat capacity estimated above.
Though this additional capacity must help to smooth out what would otherwise be
much larger seasonal temperature variations in middle and high latitudes, and play
a major role in climatic amelioration over oceans and western continental margins
there, it would merely serve to slow the response to any persistent radiative loss of
heat in these latitudes, such as the net winter cooling apparent in Fig. 8.9. In fact
increasing the heat capacity twentyfold to allow for the oceanic surface layers as
well as the troposphere, and dropping the rate of heat loss to 50 W m - 2 to be
consistent with the longer-term averages apparent in Fig. 8. 7(b), would mean that
the 100 oc chilling would then take ten years rather than three months. On the
scale of the Earth's life span this is still an extremely fast rate of cooling, and is
hundreds of times larger than even the most rapid temperature changes associated
with short-term variations in climate.
Now that we have confirmed that the imbalances in the radiant energy budget are
very significant, we should check to see if they can be significantly offset by heat
conduction. Such conduction is a statistical consequence of the random thermal
motion of the myriads of molecules which make up even millimetric-sized
quantities of matter at terrestrial densities. If this thermal motion is unevenly dis-
tributed at some initial time (i.e. if there are temperature gradients to begin with),
then the systematically more energetic molecular jostling in the warmer zones
spreads their excess temperature to the cooler zones. Although there is individual

il) ~r=11
Fig. 8.10 A panorama of
southern-hemisphere
GMS IR 0233-0256 UTC 20-NOIJ-90 subtropics and middle
latitudes as observed in the
infra-red by the Japanese
geostationary satellite GMS
at about 0240 Z (1400 local
sun time). 20 November 1990.
A pronounced wave on the
cold front east of New
Zealand is associated with
cyclonic circulation (clockwise
in the southern hemisphere)
around a surface low
pressure centred at 38 °5
169 °E. The satellite
radiometer grey scale shows
that the brightest clouds are
below - 50 °C (the high
troposphere). whereas the
darker clouds are at about
- 20 °C (the lower middle
troposphere) , and the
Australian interior is at about
+ 30 °C in the early afternoon
sun .

265
8 Radiation and global molecular movement, heat conduction is effected without gross motion of the
climate conducting material, and occurs in solids, as well as in liquids and gases. Thermal
conduction is therefore a diffusion process, with many points in common with
molecular diffusion (for example of water vapour through air) and momentum
diffusion (viscosity). If conduction can account for some or all of the Earth's
imbalances of radiant energy, then to that extent gross motion such as convection
or advection of the atmosphere and oceans is not required.
From a wealth of laboratory work it is known that heat is always conducted
down the prevailing temperature gradient and at a rate (measured by the heat flux
density H - the rate of flow of heat through unit area perpendicular to the
direction of conduction) which is proportional to the temperature gradient oT!on.
Symbolically

H = - k aT (9.1)
an
where k is the thermal conductivity of the conducting material. The minus sign is
necessary because of the convention that the distance n is measured towards higher
temperatures, which is up the temperature gradient and in opposition to the
conducted heat flow. There is an obvious similarity with the comparable
expression for diffusion of momentum (eqn (7 .4)).
Values fork are well established for a wide range of materials and conditions.
Gases in terrestrial conditions are less efficient heat conductors than are liquids and
solids, largely because of their lower densities. In air in conditions typical of the
low troposphere, values are about 25 X 10- 3 w m - 2 oc- 1' which means that in
the presence of a temperature gradient of 1 oc m - 1, the conducted heat flux is
25 mW m- 2 • To account for an upward flux of 100 W m- 2 , such as is implied by
Fig. 8.6 in the very low troposphere, there would have to be a temperature lapse
rate of about 4000 oc m - 1 • This is thousands of times larger than the largest
temperature gradients ever observed in the atmosphere on any human scale (those
found just above dry, brightly sunlit land surfaces), and is 4 x 105 times larger than
the dry adiabatic lapse rate, observed to be the limiting lapse rate for all substantial
layers of the free atmosphere. It is therefore clear that thermal conduction can play
no significant role in offsetting the gross imbalances of the Earth's radiant-energy
budget. Conduction is significant in the laminar boundary layer, the very shallow
layer of air which coats all parts and details of the Earth's surface (discussed in
section 9.6), and also serves to smooth away the extremely sharp, localized
temperature gradients which would otherwise arise from the turbulent juxtaposi-
tion of contrasting air parcels, but on more than millimetric space scales it is almost
entirely ineffective, as can be justified by a scale analysis similar to that used in eqn
(7 .14) to demonstrate the ineffectiveness of viscosity.
We have now tried to justify the conclusions drawn rather prematurely earlier in
the chapter: the gross imbalances in the Earth's budget of radiant energy are highly
significant and must be offset by convection and advection of heat; gross atmo-
spheric (and oceanic) motion is therefore required to account for the maintenance
of the present thermal state of the Earth's surface and troposphere. Unfortunately
it is tempting to overestimate the extent to which we have explained such a funda-
mental aspect of the behaviour of our atmosphere, because the discussion began by
assuming the present distribution of the radiatively active materials such as water
vapour and carbon dioxide. In fact their distributions, particularly in the vertical,
are maintained largely by the convection and advection now in question, as
discussed in section 3. 7 for example. A full discussion of the role of convection in
the economy of the atmosphere should therefore include the evolution of atmo-
266 spheric structure and behaviour from some initial state (and what should we chose
for that?) - an awesome problem which is becoming approachable only now that 8.11 Convective heat fluxes
meteorologists and climatologists are getting access to very large computers. What
we have done is simply to show that convection and advection play an essential
rather than an incidental role in the atmosphere as it is - a much less basic study,
which comes close to confusing chicken and egg in the manner forewarned in
Chapter I, but which nevertheless highlights the fundamental importance of the
winds and up and down currents which continually trouble our atmosphere.

8.11 Convective heat fluxes

Convection carries the excess heat from the Earth's surface and distributes it
through the depth of the troposphere, offsetting the imbalance apparent in the
budget of radiant energy fluxes. On a global annual average the convective flux
must be 100 W m - 2 at the Earth's surface and zero at and above the tropopause to
be consistent with Fig. 8.6, and it is reasonable to suppose that it decreases upward
throughout most of the intervening troposphere as indicated in Fig. 8.11. Such a
decrease is consistent with observations of the cumulus family of clouds (an impor-
tant and obvious type of convection) which show that their prevalence is greatest in
the low troposphere and diminishes upwards, with the upper troposphere being
reached by a relatively few vigorous cumulonimbus.
If the Earth were arid, all this convective flux would have to be borne as sensible
heat, warmer air parcels rising and cooler ones sinking to effect a net upward
transport of heat. But as mentioned in the outline of the hydrological cycle
(sections 3. 7 and 6.3), the very widespread presence of water, ice or moist ground at
the surface means that the air there is often nearly saturated, and that rising air is
consistently moister than sinking air. Convection is therefore usually associated

Fig. 8.11 Global annual

Terrestri al Solar overages of convective and
radiative energy fluxes in the
64 atmosphere. Values ore from
Fig . 8.6 and other ports of
(49].

--, ', Troposphere

' ..... '

.......... , \
I

' I
45
''\1
......
---- ....,,' I
' Surface
15 11 19
_____________.
100 Wm- 2

Key 100 unils 345Wm - 2

Energy fluxes ~Radian!
--Sensible H ea t
- - - - - La l ent heat
267
8 Radiation and global with an upward transport of water vapour, which can be regarded as a transport of
climate latent heat, since there is inevitably an extraction of heat by evaporative cooling at
the location of the evaporation (the surface), and a delivery of this heat to the
location where the vapour condenses to form cloud (usually in the lower tropo-
sphere). Indeed we can convert a vapour mass flux into its associated flux of latent
heat simply by multiplying by the coefficient of latent heat L.
The apportionment of the convective heat flux between sensible and latent forms
has long been recognized as a very important measure of climatic type, and the
ratio of sensible to latent heat fluxes is known as the Bowen ratio {3. In arid zones
such as the subtropical deserts, {3 values are much larger than unity (approaching
10), while in warm, humid zones they are much smaller than unity (approaching
0.1). The global annual average value of {3 consistent with the surface fluxes in
Fig. 8.11 is nearly 0.6.
The division between sensible and latent heat fluxes at the surface (in Fig. 8.11)
is taken from London's study [49]. It is very difficult to make direct measurements
of these fluxes on a large scale, though it has been done in a few detailed studies in
various locations, and in fact London's estimates are based in part on worldwide
measurements of precipitation. These show that the global annual average rainfall
equivalent is about 800 mm. From the density of water, this implies a mass flux of
precipitated water of 800 kg m - 2 y- 1 • Since the global hydrologic cycle is nearly
steady over time periods of a couple of months or more (over five times the
residence time for water in the form of vapour in the atmosphere - see section
3. 7), and since dew and frost are relatively unimportant sinks of atmospheric water
vapour, this rate of precipitation to the surface must be very nearly equal to the
evaporation rate from the surface. Multiplying the mass flux density by the value
for the latent heat of vaporization, we therefore find the latent heat flux density to
be 2000 MJ m- 2 per year, which is 63 W m- 2 or 19 budget units to the nearest on
Fig. 8.6. Since the total convective flux is defined by the radient energy budget, the
sensible heat flux is found by deficit to be 37 W m - 2 or 11 budget units. The
accuracy of the estimation is not high, since it depends on assessment of global
precipitation and hence on the unreliable measurements over the oceans (section
2.6), but it is probably not in error by more than 2007o.
Figure 8.11 depicts a progressive handing over from latent to sensible heat flux,
which begins in the low troposphere and is largely complete in the middle and upper
troposphere. This corresponds to the widespread appearance of cloud, and
associated conversion from latent to sensible heat, in the first few kilometres above
the cloud-free layer. Of course there are many tall cumulonimbus, especially in the
showery outbreaks of middle latitudes and the intertropical zone of low latitudes,
but even there much of the cloud in the upper troposphere is initially formed at or
below the middle troposphere before being carried aloft by embedded buoyant
updraughts. Latent heat fluxes play a considerable role in increasing the efficiency
of convective transport of heat, and are especially effective in the lower tropo-
sphere in all but the most arid regions.
Lastly and most fundamentally, Fig. 8.11 depicts the progressive handing over
from the sensible heat flux to the cascade of terrestrial radiation, which is effec-
tively complete at levels corresponding to the high troposphere. These are in the
vicinity of the highest layer which is largely opaque to terrestrial radiation, and
above these levels the thermal equlibrium of the atmosphere is dominated by
radiation, with comparatively little scope for convection or advection.

268
8.12 Advective heat fluxes 8.12 Advective heat fluxes

It is obvious in Fig. 8.7(b) that the annual average fluxes of solar and terrestrial
radiation produce a net heat gain for the planet between latitudes about 32 °N and
38 as, and a net loss at higher latitudes. The observed nearly steady thermal state
of the planet therefore requires a balancing advection of heat from low to high lati-
tudes which must be the net result of meridional exchange of air and water -
warmer material moving polewards and cooler material moving equatorwards.
It will appear in Chapter 13 that the thermal expansion of air allows energy to be
stored and transported in the form of the gravitational potential energy of
vertically expanded air columns, in addition to the more obvious forms of warmer
and cooler air at comparable levels. In fact both these forms of energy, together
with the flows of latent heat implicit in net fluxes of vapour, are needed in any
comprehensive study of the heat/energy budget of the atmosphere and oceans.
However, for the present it will be much simpler, and still quite rewarding, to
concentrate on the obvious fluxes of sensible and latent heat.
The observed steadiness of thermal state means that the annual average
advection of heat across any latitude must equal the net loss of radiant energy from
all higher latitudes in that hemisphere, which can be calculated from curve (i) in
Fig. 8. 7b. We can make successive calculations to produce the meridional profile
of poleward advection of heat depicted in Fig. 8.12. Beginning at a pole and
choosing successively lower latitudes, the calculated poleward advection of heat
increases progressively until we reach the latitude at which the net radiative flux
changes from negative to positive, beyond which the advected flux begins to fall
away as the net input of radiant energy in each latitude zone lessens the need for
heat advection from lower latitudes. In fact the calculated poleward advection
reaches zero effectively at the equator, showing that the annual radiant energy
budget for the hemisphere is nearly balanced and does not depend on significant
net advection to or from the other hemisphere. Calculation of the meridional
profiles of poleward advection in Fig. 8.12 is quite simple in principle, and is
detailed in appendix 8.3.
For easier appreciation of the numerical values, the advective fluxes in Fig. 8.12
have been divided by the lengths of the local latitude circles to produce average
fluxes per unit length of latitude circle. Such values represent the average advection
of heat between gigantic imaginary goal-posts set one metre apart on a latitude
circle, and reaching up to the top of the effective atmospheric layer and down to the
bottom of the effective oceanic layer. In fact this would mean posts extending to
the tropopause and down to a depth of about one kilometre below sea level. Of
course 'average' is the operative word: at any particular time and place the instan-
taneous advection between the goal-posts might be poleward or equatorward

Fig. 8.12 Meridional profiles

of annually averaged
poleward heat fluxes
advected by the atmosphere
and oceans, expressed per
unit length of latitude circle.
(After (49])

60 30 0 30 60 90

Latitude os 269
8 Radiation and global depending on the local air flow. Only after averaging around a latitude circle and
climate for a substantial time period is there a systematic poleward flow with a reasonably
definite value.
One effect of displaying fluxes per unit length of latitude circle is to displace the
maxima poleward from the latitudes of reversing radiant balance. The maximum
values apparent in Fig. 8.12 are very large indeed, approaching 200 MW m- 1,
which means that the average advection between goal-posts 10 m apart is equal to
the generating capacity of a large power station.
There is significant seasonal variation in the advected fluxes, but it is much
smaller than we might expect from the very large seasonal variations in net
radiative balance (Fig. 8.9), because of the smoothing influence of the large heat
capacities of the oceanic surface layers. For example, the middle latitudes of
summer hemispheres would seem to require no poleward heat advection according
to Fig. 8.9, but in fact in midsummer the ocean surfaces there are still warming
toward their temperature maxima (to be reached in late summer or early autumn)
and in the process are significantly offsetting the radiative gain. Despite such
smoothing, winter advection in middle latitudes is noticeably stronger than
summer advection, as could be guessed for example from the greater frequency and
intensity of extratropical cyclones in winter.

8.13 Weather and ocean systems convecting and

advecting heat

On the simplest view there are two main types of weather system at work in the
troposphere: the nearly vertical cumuliform convection associated with the whole
cumulus family, from the smallest fair-weather cumulus to the largest cumulo-
nimbus; and the nearly horizontal but systematically slightly tilted slope
convection associated in particular with extratropical cyclones. Each of these types
contributes to vertical convection of heat, and the slope convection in addition
plays an important role in heat advection. Although cumuliform convection in
particular is extremely widespread, it is especially concentrated in parts of large-
scale systems, such as equatorward flows associated with extratropical cyclones
and the intertropical convergence zone of the Hadley circulation. Let us consider
convection and advection separately, outlining the properties of cumuliform and
slope convection which relates to their ability to transport heat.
Updraughts in cumuliform convection are strong (- 1 m s - 1 in small clouds and
- 10m s- 1 in large cumulonimbus) and highly localized. They occur typically in
concentrated populations (Fig. 6.3), in which, however, only a small fraction of
the total horizontal area is covered by updraught. Individual updraughts have hori-
zontal areas ranging from - 1()3 m2 to - 10 km2 and are separated by much larger
areas of gently sinking, cloud-free air. There is therefore a marked asymmetry
between strong localized ascent and weak diffuse descent, which of course is
directly responsible for the intermittent nature of the showers produced by clouds
big enough to precipitate. The inverse relation between strength and area of the up
and downdraughts means that the net vertical mass flux of air in any particular
population of cumulus is very much smaller than the constituent upward and
downward fluxes. In fact it is often insignificant. However, because the
270 updraughts are consistently warmer and moister than the downdraughts, there are
almost always significant net upward fluxes of sensible and latent heat. 8.13 Weather and ocean
Temperature excesses in updraughts are usually quite small, seldom exceeding systems
0.5 oc in small cumulus, and exceeding 5 oc only in large cumulonimbus.
Localized downward fluxes of sensible heat sometimes occur when updraughts
overshoot into a convectively stable layer, such as the subsidence inversion of an
anticyclone (section 11.5), or the base of the stratosphere (section 10.4), but these
always surmount layers in which sensible heat is directed in the normal upward
direction.
Updraughts in slope convection are much more extensive and much weaker than
those in cumuliform convection, and give rise to the large areas of stratiform cloud
and persistent precipitation which are so typical of the fronts associated with extra-
tropical cyclones. Updraught speeds are too small to be measured directly, but
from estimates based on precipitation rates or large-scale mass balances for air
(sections 7.13 and 11.4) it is known that on many occasions they are- 10 em s- 1 •
Updraughts often exceed 105 km2 in horizontal area. Corresponding down-
draughts are generally even weaker and more extensive, and are displaced from the
updraughts by long horizontal distances. The large areas of subsiding, cloud-free
air in anticyclones and ridges of high pressure are obvious examples. Though there
is some asymmetry between updraughts and downdraughts, it is much less pro-
nounced than in the case of cumuliform convection.
In middle latitudes both cumuliform and slope convection are widespread, con-
tributing significantly to the convection of heat from the surface to most levels of
the troposphere. As discussed in sections 11.2 and 11.3, the two types often
cooperate: cumuli form convection is often quite vigorous in cold fronts and in the
extensive equatorward flows of air which follow them; in other areas small cumuli-
form convection pumps heat into the lower troposphere, preparing the air there for
subsequent slope convection to the high troposphere in a front.
In low latitudes, cumuliform convection predominates. The intertropical con-
vergence zone (ITCZ) can be regarded as a ragged belt of very tall cumulonimbus
pumping heat into the upper troposphere, and the enormous areas covered by the
trade winds are populated by smaller cumulus and cumulonimbus which are
pumping water vapour (i.e. latent heat) into the dry air spreading out from the

Fig. 8.13 Aerial view of

typical trade-wind cumulus
and small cumulonimbus.
Clouds lean backwards
relative to their direction of
drift because easterly
components weaken and
even reverse aloft (section
10.2).

271
8 Radiation and global
climate

Fig. 8.14 A vertical

meridional section of the
Hadley cell in one 15km
hemisphere, with Earth
curvature ignored for
simplicity. In onticlockwise
order from the equator, the
main labelled features STH TW ITCZ
represent the intertropical Subtropics Equator
convergence zone, the
3000km ... " ...
subtropical jet stream, the
·. ::~

subtropical high-pressure Key / ' Air mo t i on , ...- - Hydro l ogic cycle ; Radiation
systems, and the trade winds .

flanks of subtropical high-pressure zones (Figs 8.13 and 8 .14). It is true that some
of these populations often seem to be organized on larger scales, producing
synoptic-scale tropical weather systems which can occasionally be as large as extra-
tropical cyclones, but it seems that the cumulonimbus is the essential building
block. For example, the severe tropical cyclone (hurricane, typhoon or cyclone
depending on geographical zone- section 12.4) can be regarded as being made up
of powerful cumulonimbus arranged in a dense circle round a relatively quiet eye,
with arms of aligned cumulonimbus spiralling out for several hundreds of kilo-
metres. Such organized populations are often much more efficient at pumping
latent heat into the upper troposphere than are more homogeneous distributions,
as their heavier precipitation shows.
Vertical transport of heat in the oceans is effected by wind-induced stirring of the
surface layers, which carries summer heat downwards for seasonal storage. During
the next cooling season, the warmed layer will feed heat back to the surface . In low
latitudes the heat is probably stored for shorter periods, smoothing the feeble but
nevertheless significant variations in air temperature associated with day and night
and passing weather systems. However, all such upward and downward heat
transport in the oceans virtually cancels out over time periods of a year at most.
In high latitudes there is persistent downward convection of water which has
been chilled at the sea surface, which profoundly affects the temperature distribu-
tion in the ocean deeps by maintaining temperatures close to freezing at all latitudes
including the equator. This downward flux of cold water in high latitudes (which is
of course an upward flux of heat) is balanced by a downward flux of heat which is
especially concentrated in the subtropics (Fig. 8.15). There the great semi-
permanent atmospheric anticyclones are able, by a subtle mechanism involving
horizontal wind drag and the Coriolis effect [52], to maintain large-scale con-
vergence of the warmed surface layers, which pushes the warm water slowly down
to greater depths than could be reached by simple stirring. The sinking warm water
and the cold water continually filling the deeps from high latitudes maintain a
dynamic equilibrium in a region of strong vertical temperature gradient, known as
the permanent thermocline, which is found at depths ranging from 300 to 700 m
below sea level there. But although these processes are vast and continuous, the
associated motion is so gentle that it transports heat at a rate which is quite negli-
gible in comparison with atmospheric convection or advection.
Advection of heat occurs in both the atmosphere and the surface layers of the
oceans, and usually has a poleward component. The ocean surface layers, with
their huge heat capacity, would seem to have the potential to play the dominant
272 role, but despite the sound and fury of an angry sea, the oceans are intrinsically
 8.13 Weather and ocean
systems

Fig. 8.15 An idealized

"E 2
=
.s::.
vertical meridional section
through the Atlantic, showing
c.., cold water sinking into the
0 4 deeps at high latitudes, and
the permanent thermocline
(centred on the 8 °C
isotherm) separating the
30 0 30 warm and cold water
L atitude spheres.

much less active than the atmosphere; indeed both their waves and their great
surface currents are essentially spin-offs from the much more vigorous activity of
the atmosphere. Regarded as heat engines, the oceans are relatively inactive
because both their low-latitude heat sources and their high-latitude heat sinks are
concentrated at the same level, the sea surface. This is in complete contrast to the
atmosphere, where the source is at the surface and the sink is in the high tropo-
sphere (section 13.4). It is intuitively reasonable from observation of domestic
saucepans of water convecting as their bases are heated, and it can be confirmed
theoretically and experimentally, that much more vigorous fluid motion ensues
when the heat source lies well below the heat sink, as in the atmosphere. Some of
the wind-driven ocean currents do advect heat quite effectively (the Gulf Stream is
an example, though the mechanism is curiously indirect) but only in limited
regions; and since such currents are maintained by atmospheric motion in the first
place, we should expect their contribution to the total heat advection to be con-
siderably less than that of the atmosphere. Detailed analysis of all significant ocean
currents suggests that together they carry only about one quarter of the total pole-
ward flux at the mid-latitude maximum, leaving about 150 MW m- 1 to be carried
by the atmosphere, effectively the troposphere. In lower latitudes, ocean currents
carry a larger proportion of the smaller total advected flux, averaging about 4007o
at 30° latitude.
In middle and high latitudes the poleward advection of heat in the troposphere is
effected largely by extratropical cyclones and anticyclones. Indeed the sequence of
conditions experienced at a particular surface location, as a depression passes by a
little to poleward, is obviously consistent with a poleward heat flux: warm moist air
in the warm sector moves with a poleward component and is followed after the
passage of the cold front by cool, less humid air moving with an equatorward
component (section 11.3). In the middle and upper troposphere the flow pattern is
different and less familiar, but observation and analysis show that an essential
feature of any extratropical cyclone is the broad swath of warm, cloudy air which
flows eastward and poleward in the vicinity of the cold front, rising gently from
near the surface at lower latitudes to the middle and upper troposphere in higher
latitudes (section 11.4). On the western sides of extratropical cyclones cool air
flows toward the equator, sinking gently despite containing populations of
localized updraughts in cumulonimbus.
In low latitudes the meridional advection of heat is directly related to the Hadley
circulation (Fig. 8.14). Air warmed in the great cumulonimbus populations of the
ITCZ drifts poleward in the high troposphere. The potential temperature of this air
is very high indeed, being little short of the equivalent potential temperature of the
saturated adiabats in many cases, because the cloud tops are so high (section 5.11).
The air cools somewhat by net radiation as it drifts poleward, and completes its 273
8 Radiation and global cooling and descent in the subtropical high-pressure zone. As it flows equatorward
climate again at low levels in the trade winds, it has a much lower potential temperature
than it had when flowing poleward in the high troposphere, and it is clear that the
net result of the whole cycle of ascent, descent and advection has been a strong
poleward transport of sensible heat.
Much of the heat evolved in the ITCZ and spread poleward in this way has
actually been gathered in the form of latent heat, by evaporation from the huge
surface area covered by the trade winds. The smallish cumulus and cumulonimbus
characteristic of the trades (Fig. 8.13) pump vapour into the great air streams
converging on the ITCZ, so that there is a powerful equatorward flux of latent heat
which nearly balances the net flux of sensible heat over a substantial latitude range.
However, poleward heat advection predominates where the dry poleward flow
aloft extends beyond the moist equatorward low beneath, as happens in the
subtropics. Some of this advected heat is leaked to space as terrestrial radiation
from the upper parts of the high-pressure systems there, but part descends with the
subsiding air and spills out on their poleward flanks, to be drawn into the very
different flow regime advecting heat to middle and high latitudes.
Sometimes direct heat advection from low to middle latitudes occurs when
weather systems develop in gaps in the chain of subtropical high-pressure systems
girdling the hemispheres, and the seasonal disruptions associated with the
monsoons of the northern hemisphere land masses in particular (section 12.2) are
clearly potentially very significant. Further study shows that symmetry of the
advected heat flows about the equator is far from perfect: the concentration of
land in the northern hemisphere apparently maintains the mean position of the
ITCZ several degrees of latitude north of the equator, after averaging out the
seasonal migrations associated with the monsoons. The flow patterns of Fig. 8.14
then ensure a substantial northward flux of latent heat across the equator which is
largely balanced by a southward net flux of sensible heat. The precise situation in
low latitudes is still obscured by the shortage of reliable data there, and a proper
discussion of the relative importance of the deviations from the simple model
depicted in Fig. 8.12 and 8.14 is beyond the scope of this book.

8.14 Wind speeds and heat fluxes

We have assumed that heat is carried upwards and polewards in the atmosphere by
cumuliform and slope convection, i.e. by weather systems associated with
cumulus, cumulonimbus and frontal stratiform clouds. Now such cloud systems
have typical properties which are reasonably well defined by observation, and it is
interesting to try to make an extremely simplified check that these properties are
consistent with the magnitudes of the heat fluxes which we know have to be
maintained.
Consider the situation at latitude 45° where the poleward heat flux per unit
length of latitude circle is about 200 MW m -I and the vertical heat flux density is
100 W m- 2 , as it is nearly everywhere over the globe. Assume that one quarter of
the poleward heat flux is carried by ocean currents, leaving 150 MW m- 1 to the
atmosphere, and that one half of the vertical and poleward heat fluxes in the atmo-
sphere are in the form of sensible heat. Then the flux densities of sensible heat
which must be maintained by atmospheric weather systems are 75 MW m -I pole-
wards and 50 W m- 2 upwards. Such equal partitioning between latent and sensible
274 heat is apparently inconsistent with the surface data in Fig. 8.11, where latent heat
predominates, but we are interested at present in transport in the middle and lower Appendix 8.1
troposphere, where vertical and slope convection are most readily identified with
visual observation of cloud, and where some of the latent heat flux at the surface
has been transformed to sensible heat.
Now we will make further rather sweeping simplifications of the properties of
vertical and slope convection and estimate the vertical and horizontal wind speeds
required to maintain the required fluxes of sensible heat. The realism of these
speeds will then be a measure of the realism, or at least the consistency, of the whole
approach.
Consider first the r:tet poleward advection of sensible heat. In our model this
must be effected entirely by slope convection. Assume that on average the extra-
tropical cyclones and anticyclones which embody the slope convection effectively
occupy one quarter of the 45 o latitude circle. And assume that these quadrants are
entirely covered by equal and opposite meridional air currents which differ in
temperature by 10 oc and have wind speed V. Then according to appendix 8.4, the
meridional wind speed V must have value 8 m s -I to provide the flux density of
300 MW m- 1 required in the active quadrant. This is an encouragingly realistic
value, bearing in mind that horizontal wind speeds in the more active parts of the
troposphere are - 10 m s- 1 •
In considering the vertical transport of sensible heat, we assume that this is borne
equally by cumuliform and slope convection, and that cumuliform convection
occupies another quadrant of the latitude circle. Unlike the case of slope convec-
tion, the opposing air currents in cumuliform convection are highly asymmetric, so
we assume that updraughts occupy only one tenth of the active horizontal area, the
remainder being occupied by the much weaker downdraughts. In line with aircraft
observation of clouds, we assume that updraughts are only 0.5 oc warmer than
downdraughts. Then according to the calculations outlined in appendix 8.4,
cumuliform and slope convection will each produce vertical heat fluxes with
density 100 W m - 2 in their active quadrants (giving the required 50 W m - 2 for the
zonal average of their sum) if the updraughts have speeds - 2 m s -I and -
2 em s- 1 respectively. Again these values are reasonably realistic. Updraughts in
cumuliform systems are known to range from about 1 m s -I in small cumulus to
over 10m s- 1 in large cumulonimbus. And updraughts in fronts are known from
indirect evidence to range from a few centimetres per second in weak systems to
several times this in vigorous ones. Notice that the ratio of vertical to horizontal
wind speeds in slope convection inferred from required heat fluxes is consistent
with air currents having slopes of about 1 in 400, which is known to be realistic
from the type of analysis described in section 11.4.
Such oversimplified calculations are not meant to be conclusive; the realistic
results for wind speeds and updraughts clearly depend on judicious selection of the
various values assumed at several· points in the procedure. Nevertheless these
values are in themselves quite plausible (for example the fraction of a latitude circle
occupied by cumulus systems, and the fraction of such areas covered by
updraught), and the consistency of the whole analysis suggests that cumuliform
and slope convection can account for the heat fluxes known to be advected and
convected through the middle latitude troposphere.

Appendix 8.1 Blackbody radiation

The emission of electromagnetic radiation from the molecules of a body as they
jostle in thermal agitation is analogous to evaporation from a wet surface, and just 275
8 Radiation and global as in the case of a saturated vapour, it is useful to consider dynamic equilibrium
climate between the 'vapour' of photons (packets of radiant energy) and the emitting
surface. The laws of thermodynamics can be applied to this equilibrium and a
fundamental law named after Kirchhoff (1859) can be derived [35] and expressed
as follows:
E>- = aJ(A, T)
where
1. E>- is the spectral radiant emittance of the surface at wavelength (the power
radiated by unit area of surface per unit wavelength range centred on A).
2. a>- is the spectral absorptivity of the surface (the fraction of incident radiation of
wavelength A absorbed by the surface). It is also known as the spectral
emissivity e>- of the emitter because it describes the efficiency of emission as
outlined below. When collisions no longer greatly outnumber emissions and
absorptions of photons, Kirchhoff's law fails and absorptivity and emissivity
are no longer identical. Kirchhoff's law holds good throughout the terrestrial
surface and lower atmosphere.
3. j(A, T) is a universal function which varies only with wavelength and surface
temperature, and is quite independent of the surface material.
In essence Kirchhoff's law states that the efficiency of absorption of radiation of
a certain wavelength by a surface is exactly matched by its efficiency of emission of
that same wavelength, where perfectly efficient emission is defined by the universal
function j(A, T). It follows that the most efficient emitter of electromagnetic
radiation at a given temperature is also a perfect absorber of all wavelengths and
therefore a reflector of none. Such an ideal surface is called a blackbody, and its
emission spectrum is given by the universal functionj(A, T), named after Planck
who first explained its shape in 1900, and in so doing began the quantum revolution
in physics. For a blackbody
E>- =/(A, T)

Planck functions are sketched in Fig. 8.16 for three absolute temperatures T3 ,
T2 and T 1 in the ratio 6:5:4.
Real materials approach the blackbody ideal if their molecules are sufficiently
energetic and close together to interfere fairly continually with each others'
radiatively active atomic energy levels as they jostle and collide, so that their
individual line spectra are smoothed into a Planck function hump. In terrestrial
conditions solids and liquids are nearly all effectively 'black' in the near and far
infra-red, whereas gases are not. Water vapour and carbon dioxide have very
complex lines and bands of lines in their far infra-red emission and absorption
spectra, which arise from quantized states of molecular rotation and vibration,
whereas they are almost completely ineffective as emitters and absorbers in the
visible and near infra-red. However, Kirchhoff's law applies throughout. In the
visible, a>- is very small and so is E>- because E>- = axf(A, T). In a thick cluster of
lines in the far infra-red, if there is enough vapour or carbon dioxide to make the
lines merge in effect, a>- is nearly 1 and E>- as nearly approaches the blackbody ideal
/(A, T) in those wavelengths.
It is shown by observation, and can be confirmed by manipulating the Planck
functions, that the total emittance of a blackbody across all wavelengths is a simple

c
function of absolute temperature T:

E = /(A, T) dA = aT4
276
 Appendix 8. 1

Fig. 8.16 Blackbody

emission from bodies with
absolute temperatures in the
ratios 6:5:4.

This is Stejan,s law, and the universal constant a is the Stefan-Boltzmann

constant, with value 5.67 x IQ- 8 W m- 2 K- 4 • Stefan's law gives the emittance of
the blackbody surface, which corresponds to the total area under the Planck curve
for that temperature (Fig. 8.16). It clearly shows that blackbody emission increases
very sharply with absolute temperature.
Observation and manipulation of the Planck function also show that the wave-
length Am of maximum spectral emittance of a black body (the peak of the relevant
Planck curve) decreases with increasing absolute temperature according to Wien ,s
displacement law
TAm= Y
where Y is a universal constant which has value 2897 ~-tm K when A is expressed in
micrometres. The reddening glow of a metal bar as it cools from white heat is a
classic example of Wien's law in action.
Absorption by a greybody is uniformly spread across all wavelengths so that
absorptivity (and emissivity) are independent of A. The wavelength of maximum
spectral emittance is unaltered by comparison with a blackbody at the same
temperature, and the emittance is given simply by
E = aaT4 = EaT4
Real solids and liquids are often greybodies across substantial wavelength ranges.
For example water and ice are very pale grey (small aJ in the visible range, and are
very dark grey (a~. only slightly less than 1) in the far infra-red.
Consider two simple but important examples which involve such radiative
behaviour.

I. For reasons mentioned in section 8.1 and the beginning of this section, the solar
photosphere behaves as a blackbody even though it is a gas. From Wien's law a
blackbody with temperature 5900 K has maximum ·spectral emittance at wave-
length 0.49 ~-tm. near the centre of the visible spectrum, which is consistent with
spectroscopic observations of the sun. From Stefan's law the emittance of the
solar photosphere is very close to 69 MW m - 2 , which is consistent with
measurements of the solar constant after allowing for inverse-square
attenuation by distance.
2. The solid and liquid surfaces of the Earth behave as blackbodies at the local
temperature. At temperature 288 K (the mean surface temperature) the wave-
length of maximum spectral emittance is almost exactly 10 ~-tm, which lies well
into the far infra-red, as confirmed by observation. At the same temperature,
Stefan's law gives blackbody emittance 390 W m - 2 , which is exactly consistent
with the 113 units on Fig. 8.6. Using Stefan's law in reverse, the terrestrial
output of 67 units to space (231 W m - 2) corresponds to the emittance of a
blackbody at 253 K, which is fully consistent with values quoted in Sections 8.1
and 8.2. 277
8 Radiation and global
climate Appendix 8.2 Surface irradiance

Consider a beam of direct sunlight impinging on a horizontal surface at an

elevation angle of {3 (a zenith angle of 90- {3). According to the geometry of
Fig. 8.17, the radiant flux density Fh on a horizontal surface (the horizontal
irradiance) is less than on a surface normal to the beam {Fn) as follows:
(8.5)
This confirms that the irradiance of a horizontal surface is a maximum when the
sun is in the zenith, and tends to zero as the sun approaches the horizon.
Another important factor which varies with {3 is the path length of the solar beam
through the atmosphere (Fig. 8.17), and with it varies the associated attenuation of
the beam by scattering and absorption. The details of the interaction between solar
radiation and the atmosphere are complex and are dealt with in more advanced
texts [8], but an important result emerges from quite simple examination. The
small reduction dF in irradiance arising from the presence of small mass dm' of
attenuating material in a section of sloping beam is proportional to both F and dm'
dF = -kdm' F
where the minus sign ensures that dF is a reduction and k is an attenuation
coefficient. (In fact k normally varies strongly with wavelength (Fig. 8.1), intro-
ducing considerable complication which we will ignore.) In many circumstances
attenuating material is horizontally stratified, like much of the atmosphere, and
the simple geometry of Fig. 8.17 shows that dm' is simply related to the element of
mass dm in a vertical column bounded by the same horizontals.
dm' = dm cosec {3
Hence

FdF = - k dm cosec {3
Integrating along the sloping path from the top of the atmosphere, where F 0 is
effectively the solar constant, to the surface where F" is the surface flux density
normal to the beam
F
-" = exp ( - km cosec {3)
Fo
where m is the total mass of the active material in a vertical column through the
atmosphere. This is a form of Beer's law, and the product km is known as the

Fig. 8.17 Oblique solar rays

passing through an absorbing I dm' =p dl' r--- --,-----
atmosphere and illuminating = p dl d/' dl i d/ 1 , /
a horizontal surface.
L_~_d~_cd~sec
I /
..i __ _L'_ __

Air
,/

278 Surface
optical depth of the atmosphere. The left-hand side of eqn (8.5) is called the trans- Appendix 8.3
missivity r13 of the angled beam- the fraction of radiation incident on the top of
the atmosphere which penetrates to the surface (in the direct beam). By definition
the zenith transmissivity r90 is given by
r 90 = exp ( -km)
so that we have
(8.6)
It is usual to describe the transparency of an atmosphere by its zenith transmissivity
even if it is at a latitude at which the sun can never reach the zenith. Thus a clear sky
in the British Isles may have a zenith transmissivity of as much as 0.8. It follows
that when the sun is 15 o above the horizon on the same occasion, the transmissivity
of the oblique beam is only 0.42.
The horizontal irradiance at the foot of an oblique sunbeam is found by
combining eqns (8.5) and (8.6):
(8.7)
Figure 8.18 is a graph of horizontal surface irradiance against solar elevation angle
for a clear summer's day in Britain.
Obviously this treatment does not try to cover atmospheric transmission when
attenuating material is not horizontally stratified (for example when the sky is
dotted with cumulus), nor does it deal with diffuse radiation, where photons
scattered out of the direct beam may nevertheless reach the surface, sometimes
after further scattering.

1000 Fig. 8.18 lrrodionce of a

N' horizontal surface near sea
I 800
E level in central England on a
;;: clear summer's day (zenith
600
....s: 400
transmissivity 0.8). The
horizontal axis is solar
elevation, and maximum
200
elevation corresponds to
noon at the summer solstice.
0 10 20 30 40 50 60 70
~0

Appendix 8.3 Meridional advection profiles

According to the global geometry outlined in Fig. 8.19, the surface area of a
narrow zone bounded by latitudes cf> and cf> + de/> is given by 27rR cos cf> R de/>. The
surface area 1A 2 of a wider zone bounded by distinctly different latitudes c/> 1 and c/> 2
is found by integrating the narrow zone expression between these limits.
1A2 = 21rR 2 (sin c/> 1 - sin c/> 2 )
Substituting values for c/> 1 and c/> 2 quickly confirms that the area of a hemisphere is
21rR 2 , and shows that half the surface area of a hemisphere lies between the equator 279
8 Radiation and global and latitude 30°, while only 13.407o lies poleward of latitude 60°. This exposes the
climate grossly misleading nature of the Mercator map projection and justifies the cos¢
weighting used in Figs 8.7 and 8.9.
Suppose that we divide the hemisphere into latitude zones each 10° wide
(obviously a more uniform division could be made as above), and that within each
zone we can define a zonal average net influx F of radiant energy per unit hori-
zontal area from the data summarized in section 8. 7. Then the total influx into a
zone is the product of the zonal area and the flux density. To be in a steady thermal
state this heat influx must be exactly balanced by the net efflux of heat, which can
only be borne by meridional advection. If advection is described by flux densities
M (average flux across unit length of latitude circle), then the advected flux across
any latitude circle is M27rR cos¢, and the expression for the thermal equilibrium of
a zonal strip bounded by latitudes ¢ 1 and ¢ 2 is easily shown to be
M 2 cos ¢ 2 - M 1 cos ¢ 1 = FR (sin ¢ 2 - sin ¢ 1)
SinceM at the poles is necessarily zero (there are no higher latitudes to flow to!), we
Fig. 8.19 Oblique view of
can work zone by zone to lower latitudes using observed values for Fto calculate
Earth, with zonal ring at
latitude¢. Mvalues.

Appendix 8.4 Convective and advective

heat fluxes

ADVECTION

Consider a vertical plane erected on a line of latitude which is crossed along a

fraction N of its length L by a poleward flow of air with speed V, temperature Tand
surface pressure p. If the remainder of the line is crossed by an equatorward flow
with speed V', temperature T' and surface pressure p', then the net poleward flux
of sensible heat is

N L C .!!__V T- (I - N) L C
p g p
!!_V'
g
T'

which is justified by noting that pig is a mass of air per unit surface area, that
(pi g) Vis therefore a poleward mass flux per unit length of latitude line, and that
this is converted into a flux of sensible heat by multiplying by temperature, specific
heat capacity and relevant line length (N L). The specific heat at constant pressure
CP is used for the reasons mentioned in section 5.3, and actually allows for the
presence of heat in the form of thermally extended air columns as well as directly
sensible heat (section 13 .2).
We assume there is zero net poleward mass flux of air, so that

N L _!!_v = (1 - N) L
g
Lv
g
Substituting this in the expression for the net heat flux and dividing by L to find the
net flux per unit length of latitude line, we obtain

NC l!_V(T- T') (8.8)

280 p g
which shows that the crucial factors are the temperature excess and speed of the Appendix 8.5
poleward flow, together with the fraction of the latitude line it occupies.
In section 8.14 it is assumed that the advective flux density in the active zone is
300 MW m - 1, that N is Y2 and that T- T' is 10 °C. To allow for the fact that only
the troposphere (i.e. the layer from about 1000 to 250 mbar) is effective in
advecting heat, we assume p to be 750 mbar. Together with g = 10 m s - 2 , we have
pig 7500 kg m- 2 , and with CP 1000 J kg- 1 K- 1 we deduce that Vmust be 8 m s- 1
to carry the required flux.

CONVECTION

We can treat vertical fluxes of sensible heat in an almost identical way. Suppose
that either cumuliform or slope convection is active in a certain region with
horizontal area A, so that its associated updraughts fill a fraction N of the
horizontal area with air of density p, temperature Tand speed w, the remainder of
the area being filled with air of density p' and temperature T', sinking at speed w'.
Then a similar procedure to before shows that the net upward flux of sensible heat
is given by
NAp w Cp T- (1 - N)A p' w' CPT'
Again assuming that there is no net upward mass flux, we have
NAp w = (1 - N) A p' w'
Using this and dividing by A we find the net upward sensible heat flux per unit area
to be
(8.9)
which is clearly equivalent to expression (8.8) in all important respects.
In section 8.14 we assume that both slope and cumuliform convection maintain
sensible heat flux densities of 100 W m - 2 in their respective active areas. A realistic
value for pin the low troposphere is 1 kg m -J. In slope convection Nand T - T'
are assumed to have the same values assumed above in the advection estimate,
because it is the same tilted flows which effect both advection and convection. It
follows that an updraught of 2 em s - 1 is needed to maintain an upward flux of
100 W m- 2 • In cumuliform convection, N is 1/w and T - T' is 0.5 °C, and it
requires an updraught of 2m s- 1 to maintain 100 W m- 2 • Note that this
asymmetry of area implies a compensating asymmetry of updraught and down-
draught speeds to maintain zero net mass flux (density differences being relatively
unimportant).

Appendix 8.5 Radiative forcing

The sensitivity of blackbody emission to changes in temperature is found by

differentiating Stefan's Law (appendix 8.1).
dE
- = 4 a T3
dT 281
8 Radiation and global This can be rewritten in finite-difference form and rearranged to find the
climate blackbody temperature rise !J.T corresponding to any rise in blackbody emittance
t::.E.
!J.T = t::.E/(4 a P)
Applied to radiative forcing of the Earth's atmospheric greenhouse (section 8.5),
this expression gives the first-order temperature rise !J. T corresponding to a forcing
of magnitude t::.E. Assuming T= 255 K, the temperature rise per unit forcing is
0.27 K per W m - 2 •

Problems

LEVEL 1

1. Some stars are much redder in appearance than others. What does this tell you
about their photospheric temperatures?
2. The orbital radius of Venus round the sun is about two thirds of the orbital
radius of the Earth round the sun. Find the ratio of the solar constants of the
two planets.
3. A black-and-white cat is lying in strong sunlight. Describe its appearance when
viewed by a radiometer sensitive to the far infra-red, whose display renders
strong radiators white and weak radiators black.
4. How would the shape of the terrestrial radiative spectrum observed in Fig. 8.3
have changed if at the time of observation there had been a complete overcast
of clouds with tops at temperature 250 K?
5. Which of the following alterations would tend to enhance or reduce the
atmospheric greenhouse effect: widening the atmospheric window; making
the flanks of the window more opaque to terrestrial radiation; having a much
hotter but smaller sun with the same solar constant; increasing the direct
absorption of solar radiation by the atmosphere?
6. Describe everyday observations which suggest that limitation of convection is
crucial to the performance of greenhouses?
7. Assuming that the fluxes of terrestrial radiation in Fig. 8.6 are otherwise
unchanged, what rate of heat loss do you expect at the surface at night (a) when
the sky is clear, and (b) when the sky is overcast by low cloud?
8. Check by squared paper or geometry that curve (i) of Fig. 8.7(b) indicates
global radiative balance. Redraw with a linear latitude scale to see the apparent
exaggeration of high-latitude losses.
9. Find the angle of solar elevation at noon at latitude 60° at the winter solstice.
10. In Fig. 8.11 the net upward flux of terrestrial radiant energy increases with
height through the troposphere, so that every parcel of finite depth loses more
through its top than it gains through its base. How is this sustained without
steady cooling of the air?
11. Consider the surface wind directions and air temperatures on the western and
eastern flanks of an anticyclone which straddles a western continental margin
in winter, and find the likely direction of resulting net advection of sensible
heat in the low troposphere.
282 12. The same weather system in summer would effect an equatorward flow of
 sensible heat. Consider the directions of net fluxes of latent heat in both Problems
situations.

In Fig. 8.14:

13. what happens to the latent heat released by condensation in the intertropical
convergence zone;
14. what are the main differences in the middle troposphere between air ascending
in the ITCZ and descending in the STH;
15. and what is the advantage for the Hadley circulation that air diverging from
the base of the subtropical high pressure systems is unusually dry?

LEVEL 2

16. Venus, Earth and Mars orbit the Sun at average distances of 108, 150 and 228
million kilometres respectively. Given that the sun is a spherical black body of
radius 700 000 km and temperature 5800 K, find the solar constants for each
planet.
17. Show that the equilibrium radiative temperature TE for the Earth (radius RE) is
related to the radiative temperature T 8 of the solar photosphere (radius R,) by

TE
= (1-a)Yi
...fi
(_!S_)y, T,
2 R
where a is the Earth's planetary albedo. Hence find the equilibrium tempera-
ture for Venus, assuming its planetary albedo to be 0.7 (higher than Earth's on
account of its thick cloud blanket).
18. Find the radiative emittance of the following surfaces, all except one of which
are blackbodies: water at 25 °C, sand at 40 oc (emissivity 0.85), fog at 5 °C,
cloud at -20 oc and at -40 °C.
19. Meteorological satellites usually carry two radiometers, one using visible
radiation and the other one or more bands of infra-red radiation. The latter
have displays in which whiteness of image increases with decreasing radiant
intensity. To see that this produces infra-red images which resemble visible
ones (in daylight), sketch the infra-red images you would expect to find in a
view containing stratocumulus, large cumulonimbus with tops in the middle
and high troposphere, and a front with cloud decks at various levels in the
middle and high troposphere.
20. Assuming that in Fig. 8.6 the direct solar inputs to the stratosphere and
troposphere are spread though the layers 16-48 km above sea level and
0-16 km respectively, find the ratio of the heat inputs per unit mass and hence
the relative speeds of implied temperature response.
21. In a certain clear sky the zenith transmissivity for solar radiation is 0. 75. Find
the irradiance at sea level of normal and horizontal surfaces when the sun is
25° above the horizon. Assume the normal solar irradiance above the
atmosphere is equal to the solar constant.

LEVEL 2'

22. Consider the energy loss represented by 100 W m -2 output over a three-month
season, and express it in terms of the cooling of a 10 m layer of water, the 283
8 Radiation and global cooling of the atmosphere, and the depth of ice formed by freezing of the
climate ocean surface at 0 °C.
23. If the maximum poleward heat advection across the 45° line of latitude were to
be converted with 1OJo efficiency into useful electrical power, find the power
available per head of world population.
24. Use the approach outlined in appendix 8.3 to show that if the net radiative
input per unit surface area is F when averaged over all latitudes from the
equator to 45°, and the average meridional heat flux to higher latitudes isM
per unit length of latitude line, then
M=RF
where R is the radius of the Earth. Use this expression to estimate a value for F
from Fig. 8.12 and compare your result with Fig. 8.7(a).

LEVEL 3

25. In another 5 x 109 years or so, our Sun will probably become a red giant, with
its photospheric temperature dropping to 4000 K and radius swelling to 25
times the present value. Find out all you can about the radiative environment
of the Earth which will then apply.
26. Discuss the significance for atmospheric conditions and behaviour of the
atmosphere's transparency to solar radiation and opaqueness to terrestrial
radiation. Speculate on how things might differ if the atmosphere were opaque
to solar radiation and transparent to terrestrial radiation.
27. Consider the following very simple model of the atmosphere in purely
radiative equilibrium. It consists of a single thin sheet of air at absolute
temperature Ta which is transparent to solar radiation but acts as a grey body
with absorptivity and emissivity a in the far infra-red. The underlying surface
acts as a blackbody with temperature Tg. Find the relation between Tg and the
solar input and between Ta and Tg, and calculate numerical values for Ta and
Tg using a realistic value for solar input and a = 0.8.
28. Despite using the display mode for infra-red imagery as described in problem
13, some differences remain between visible and infra-red images. Describe
and explain the differences to be expected in the following cases: an area of sea
fog; snow-covered low ground showing through high cloud (not in Antarctica
or Siberia); strongly heated pale desert; a plume of diffuse anvils (section 10.4)
from a cluster of shower clouds.
29. Check on the equatorial value for the solar input in Fig. 8. 7(a) by considering
the input there at an equinox. (Hint: in this situation solar radiation is
collected continually on a diameter of the Earth's disc, but is spread around
the Earth's circumference by the earth's rotation.)
30. Discuss the climatic significance of meridional and seasonal variations in solar
input and the relative invariance of terrestrial output. In most latitudes,
diurnal variations in solar input are even larger; why are their consequences
relatively small? Note that this is examined in some detail in sections 9.3 to 9.5.

LEVEL 3'

31. Consider how temperatures on the surface and in the atmosphere might alter if
284 all convection and advection were to cease.
32. The heat capacity of the Arctic ocean obviously tends to limit the seasonal Problems
swings of surface temperature there. How is this tendency affected by the
seasonal presence, freezing and melting of sea ice there?
33. In a popular science fiction story by a distinguished cosmologist [53], the
Earth is supposed to face catastrophe as an opaque cloud of interplanetary
dust threatens to cut off solar input and produce 'over 250 degrees of frost
within a month'. Persuade the writer that he is being unduly pessimistic.
34. Modify the method for calculating heat advection in appendix 8.4 to analyse
the following simplified picture of narrow poleward ocean currents such as the
Gulf Stream and Kuroshio: fraction of latitude circle at 40° occupied by
currents 1/200; poleward speed 1 m s -t; depth of current 500 m; temperature
excess of warmer flow 3 °C. Compare your result with Fig. 8.12.

285
CHAPTER
Surface and boundary
9 layer

9.1 Introduction

The lower boundary of the atmosphere is the solid or liquid surface of the Earth,
which is often either complex or dynamic or both. The sea surface is relatively
simple when there is little broken water, but it is highly mobile on account of the
presence of surface waves and currents. Land surfaces are often more static,
though the wavy motion of tall grasses, forests and (on a longer timescale) sand
dunes, are conspicuous exceptions, but they are usually highly irregular on space
scales ranging from those of sand grains to mountains. Dense vegetation is par-
ticularly complex and can significantly blur the boundary between surface and
atmosphere by spreading the sites of heat and momentum exchange significantly in
the vertical. The importance of the large heat capacities of surface layers of water
has already been stressed (section 8.10). By contrast, land surfaces have a very
different thermal behaviour which varies widely with surface type and condition.
Changes in ground cover by vegetation or snow can produce large changes in
thermal and other behaviour of the resultant surface- changes which can signifi-
cantly affect the overlying air.
In all sorts of natural and artificial fluid flows, it is observed that there is one or
more zone, adjacent to the boundary, in which the flow is significantly affected by
the nearby presence of the boundary, to the extent that the affected zones are called
boundary layers to distinguish them from the relatively free fluid beyond. There
are several distinguishable boundary layers at the base of the atmosphere, where
the presence of the nearby surface has a profound effect on atmospheric
behaviour. Figure 9 .I represents the most important of these diagrammatically -
layers which together make up the atmospheric or planetary boundary layer.
Dimensions are only notional since the various layers vary tremendously in
depth.
Within a few millimetres of any surface, no matter how irregular, air motion is
so strongly restrained by friction with the surface, and by the intermolecular
friction (viscosity) of the air itself, that gross motion is nearly non-existent.
Momentum, heat and matter such as water vapour are transported mainly by
molecular diffusion. This is the laminar boundary layer- so-called because, apart
from random molecular motion, the only significant movement of air molecules
286 resembles the sliding of a sheaf of thin plates (laminae) each of which is able to
 Zone Activity 9.1 Introduction
Height ( m > , - - - - - - - - - - - - , - - - - - - - - - - - - - - - ,
Synoptic, meso and
cumulus scale
Free atmosphere
weather systems
500 / / / / / / / / / / / / / / / / / / ( / / / / .

50
Ekman layer 1
(')as
Vabove
1\ as
vbelow

Q;
5 Surface >- Turbulence
~
boundary >- Fig. 9.1 The structure and
;;; Eddy transport
activity of the atmospheric
0.5 layer '0
c: boundary layer displayed

v. 6
0"
against a notional height
Scm 0.05
"'.~ scale. Arrow widths increase
' Q;
Transition zone :' .c.,c. as B., with relative importance. A, B
I
I
as below G.l:,! and C represent small,
0
SmmO.OOS /////////////////~ E bove · ~ ~;
;;; medium and large surface
:
I
Laminar Molecular motion -:; '"E
~ ., roughness elements
boundary layer I
I
I
diffusion A en ~a;
respectively.

move only very slowly between its neighbours (Fig. 7 .3). Although occupying only
a minute fraction of the total volume of the troposphere, the laminar boundary
layer significantly affects large-scale exchanges between the surface and the bulk of
the troposphere, by acting as a kind of separating membrane which is only
sluggishly permeable by fluxes of heat and matter.
As depicted in Fig. 9.1, another boundary layer known variously as the turbulent
or surface boundary layer extends upwards from the laminar boundary layer for a
highly variable and rather poorly defined distance. Just as the laminar boundary
layer is the region strongly influenced by the surface through molecular diffusion,
so the surface boundary layer is the region which the surface influences strongly by
eddy diffusion (see section 7 .2). In fact it is the most conspicuously turbulent part
of the atmosphere, and is the only part in which the turbulence often seems to tend
towards a relatively well-ordered dynamic equilibrium. Even so, the turbulent
transport of momentum, heat and matter through it is poorly understood in detail,
despite its significance for large-scale exchanges between the surface and tropo-
sphere, and despite the fact that the surface boundary layer is the atmospheric
environment of Man.
Above the surface boundary layer there is a relatively extensive transition zone in
which the direct influence of the surface weakens with increasing height, to the
level where it becomes the rather indirect, though still quite strong, influence which
the surface exerts on the rest of the troposphere. At this altitude the geostrophic
departure induced by turbulent friction (section 7 .11) has become unobservably
small, having decayed in a curious fashion first explained by Ekman and often
named after him. The depth of the whole nest of boundary layers (laminar, surface
and Ekman) conventionally comprising the atmospheric boundary layer, is very
variable but is often - 500 m. Above this the atmosphere is considered to be so
free of direct surface influence that it is called the free atmosphere. In fact this is
not a very rational title, and much of this book contains strong evidence that the
troposphere itself is yet another boundary layer - one influenced by the surface
through the action of the common weather systems. Nevertheless we will follow
convention and concentrate in this chapter on the atmospheric boundary layer as
conventionally defined and as depicted in Fig. 9 .1.
287
9 Surface and boundary
Ioyer

Fig. 9.2 Dendrites of hoar

frost growing from the edges
of small leaves.

288
9.2 Surface shape and radiation 9.2 Surface shape and
radiation

The surface of the Earth can be regarded as a horizontal plane with projections and
indentations ranging in size from the microscopic to the mountainous. These affect
the overlying boundary layers in a number of ways which are mentioned at various
points in this chapter, the most important of which are the redistribution of heat
sources and sinks arising from radiation fluxes, and the local disturbance of air
flow. In this section we concentrate on the effects of radiation. Notice that there
are many other effects of surface shape which are important or interesting in some
contexts; for example the environment of a mountain peak is greatly influenced by
the prevalent lapses of temperature and density in the ambient air, while if projec-
tions are as small as blades of grass, with correspondingly sharp edges, the diffu-
sion of water vapour is concentrated at the sharp edges, as is often apparent in the
distribution of hoar frost (Fig. 9.2).

Fig. 9.3 A beam of direct

sunlight falling on a ridge
with triangular cross-section.

Consider first the effects of surface shape on direct solar input. A simple model
of a solitary projection (Fig. 9.3) establishes some important effects. First, solar
radiation is concentrated on the sun-facing side at the expense of the other, and the
concentration is total if the other side is in shadow. In middle latitudes the typical
slopes of steep hills and mountains are such that sunlight often falls nearly
normally on sun-facing slopes and very obliquely on others. Since surface
irradiance is proportional to the sine of the angle between the plane of the
illuminated surface and the line of the incident sunbeam (Appendix 8.2), all
normally illuminated slopes are receiving the maximum flux density possible in any
given sunlight. It is shown in appendix 9.1 that the ratio of sloping surface
irradiance to horizontal irradiance is given by sin (a + E)/sinE, where a is the
angle of slope and E is the angle of solar elevation above the horizon (Fig. 9.3).
Values for this ratio are plotted in Fig. 9.4 for a slight slope, a steep slope, and a
nearly vertical slope typical of a cliff or tree. This shows that steep slopes are at a
great relative advantage when illuminated by the sun at low angles of elevation.
Although absolute values of irradiance are then quite small because of strong
attenuation of the oblique sunbeams in their long paths through the atmosphere
(Fig. 9.4 and appendix 9.1), the advantage can have quite marked effects on the
distribution of cumulus convection over hilly country. Since such convection is
driven by very small temperature excesses, it can be concentrated similarly over
sun-facing slopes, producing lines or streets of cumulus drifting downwind from
the nursery slopes (Fig. 9.5). In the examples plotted in Fig. 9.4, the absolute value 289
9 Surface and boundary 1. 0 10
layer 70

0 .8

0 .6
.AI RI
Fig. 9.4 Direct-beam solar s 5
irradiance on surfaces
making angles of 0, 15, 45, 0 .4
and 70° with the horizontal.
Curves Rl represent the ratio
of slope irradiance to 0 .2
horizontal irradiance. Curves
AI / S represent absolute slope
irradiance under a clear sky 0 0
with zenith transmissivity 0.8 0 30 60 90
(appendix 9.1), normalized by
Solar e l eva1ion E
the solar constant S.

of insolation is greatest for all but the slightest slopes when the sun is moderately
high in the sky- near the midday maximum elevation in summer in mid-latitudes.
The very considerable preferential warming of sun-facing slopes is then quite
obvious to anyone walking in hilly terrain. In the case of small projections such as
buildings or vegetation, the concentrations of input are too localized to have any
significant effects on larger scales, but they can be very important for conditions on
the scale of the projections themselves, or smaller, as is often apparent in the dis-
tribution of flora and small fauna. The effects of window location in relation to the
sun on conditions inside buildings are very familiar, although still hardly acknow-
ledged by much building design.
Another obvious effect of surface shape, already implicit above, is simply that if
the projection is steep enough in relation to the solar elevation angle, the sunlit face
concentrates insolation which would have been distributed over a much larger area
(i.e. larger than the base area of the projection) if the projection had not been there.
This is simply to say that a significant area is shadowed by the projection. In the
case of the simple projection in Fig. 9. 3, shadowing extends beyond the base when
the solar elevation is less than the angle of slope. It is easily shown that a steep
mountain illuminated by a fairly low sun can intercept sunlight which would other-
wise have been spread over a horizontal area twice as large as the base of the
mountain, and a very steep solitary projection such as a tree can do several times
better than this.
When there are several steep projections close together, each one may shade its
neighbour to some extent, so that the solar input is concentrated on the upper parts
of each. This can occur in hilly terrain when the sun is low, but is at its most obvious
in dense stands of vegetation such as a field of cereal, or a forest. Indeed the
infiltration of sunlight through vegetation is more akin to sunlight entering water
and being increasingly attenuated with depth below the surface (Fig. 9.6), the
leaves playing the same role as the absorbing and reflecting material in the water.
This is an example of how a complicated surface shape blurs the normally sharp
interface between atmosphere and surface, the radiatively active layer being
concentrated in this case in the upper parts of the surface projections.
In the simplest picture the presence of projections redistributes the insolation but
does not alter the total input. However, if the projections are so steep and close
290 together that light suffers repeated scattering from adjacent facing surfaces,
 9.2 Surface shape and
radiation

Fig. 9.5 A street of cumulus

drifting downwind from o
sunlit slope .

repeated absorption ensures that the total capture by the surface is enhanced. This
happens markedly when sunlight is steeply incident on a deeply indented surface
such as a forest, in which light passing below the treetops is repeatedly reflected
between adjacent trees and leaves, with further absorption at each reflection. As
aerial photographs often show, coniferous forests have a very low albedo as a
result, being amongst the darkest of all common land surfaces.
All of the above has assumed direct sunlight with geometrically edged zones of
illumination and shadow. In many regions, more than half of all surface irradiance
is diffuse, having been scattered from the incident direct beam by cloud, haze and
air molecules. All of the above effects becomes much less marked in diffuse
sunlight because, although not necessarily fully isotropic, diffuse sunlight is very
much less directional than direct sunlight. 291
9 Surface and boundary Surface
Solar flux density
layer
.<::
c.
Q)
"0

Q)

"'
~

Fig. 9.6 Exponential decay

l ___ Level of 90%
attenuation
of solar irradiance with depth
in water.

Output of terrestrial radiation is complicated by surface shape in ways which are

rather similar to those applying to the reception of solar radiation, although there
is no equivalent of the direct solar beam and the ways in which it is shaded, since
terrestrial radiation is always diffuse. In essence the net rate of loss by emission and
absorption of terrestrial radiation is governed by how much sky can be seen by a
fish-eye lens at the surface in question. A plane horizontal surface sees the
maximum amount of sky (a hemisphere), whereas at the bottom of a narrow cleft
between two projections, only a small amount of sky is 'seen' between the projec-
tion walls. Since the effective blackbody temperatures of the walls are usually
much higher than those of the sky, the net loss is greatly reduced in the cleft. Hoar
frost distribution on roads and short grass often demonstrates the radiative shelter-
ing effects of walls and trees very clearly. As with solar input, the terrestrial output
from a dense stand of vegetation is concentrated in the upper parts of the stand,
though not necessarily in precisely the same height zone as that in which the solar
input is concentrated.

9.3 Surface heat input and output

Let us consider the radiative and convective processes injecting and removing heat
from representative parts of the Earth's surface, so that we can define typical rates
of gain or loss of heat.
Consider first the absorption of solar radiation. The albedos of terrestrial
surfaces range very widely from a few OJo for damp, dark soils to over 90% for fresh
snow (Table 9.1). Snow values are localized in high altitudes and latitudes, and the
relatively pale land deserts are localized in the subtropics. Over the rest of the globe
values are dominated and held low by the presence of water and vegetation. Over
land albedos are mostly between 5 and 20%; and over the sea they are somewhat
lower, except where the direct beam of sunlight approaches grazing incidence.
Rough surfaces generally have lower albedos than equivalent smooth surfaces
because of the multiple absorption effect mentioned in the previous section.
Albedos in diffuse sunlight are marginally larger than in fairly high elevation direct
sunlight because of the inevitable presence of some grazing input in the diffuse
light.
292 Albedo values quoted in Table 9.1 are averaged over the whole solar spectrum,
Table 9.1 Radiative properties of surface materials 9.3 Surface heat input and
output
Surface Solar Terrestrial
material albedo emissivity

Soils 0.05-0.40 0.90-0.98

Grasses 0.16-0.26 0.90-0.95
Forests 0.05-0.20 0.97-0.99
Desert 0.20-0.45 0.84-0.91
Snow 0.40-0.95 0.82-0.99
Water
low solar elevation 0.10-1.0 0.92-0.97
high solar elevation 0.03-0.10 0.92-0.97

and may be applied directly to total solar fluxes to estimate inputs to irradiated
surfaces. The strong coloration of many terrestrial surfaces of course shows that
reflectivity varies considerably with wavelength in many cases, but though this may
sometimes be important (as in the photosynthesis process in plants), it is averaged
out in dealing with the heat economies of all surfaces on the human and larger
scales.
To establish typical values for solar input, consider a horizontal surface with
150Jo albedo irradiated by the sun from an angle of elevation of 50° through a
relatively clear sky which is transmitting 70% of the solar beam. Then using the
value for the solar constant, and the expression for oblique surface illumination
(appendix 8.2), we find that solar energy is being absorbed by the surface at a rate
of nearly 630 W m - 2 , a value which is typical of daytime values under clear skies in
middle latitudes. Towards midwinter, or at dawn or dusk, values are of course very
much smaller, as they are too under thick overcast (which may reduce atmospheric
transmission to less than 10% of its clear sky value). In the course of a day the total
amount of solar energy absorbed by an element of horizontal surface in middle
latitudes may range from as much as 30 MJ m - 2 in summer to less than 1 MJ m - 2
under a winter overcast. Daily totals can be measured directly by recording
solarimeter, or estimated from day length and midday maxima using the
approximation in appendix 9.2.
As mentioned earlier (section 8.3) terrestrial surfaces behave very nearly as
blackbodies in the wavelengths which make up terrestrial radiation. The closeness
to blackbody behaviour is apparent in the nearness of emissivity (Appendix 8.1) to
the value 1. Table 9.1 shows that most surfaces have emissivity values of between
0.9 and 0.98. Apart from old snow, only deserts have emissivities less than 0.9, and
that is because of the presence of quartz sand which is partly transparent in terres-
trial wavelengths. Note that water is almost completely opaque to terrestrial radia-
tion (which means that its emissivity is close to 1.0) although it is nearly transparent
to solar radiation - a striking example of how radiative properties can differ in
different wavelength ranges. Terrestrial radiation from the sea therefore comes
from the topmost millimetres of water.
For present purposes we will simply assume that terrestrial surfaces emit and
absorb terrestrial radiation as if they were blackbodies. We can therefore use
Stefan's law (Appendix 8.1) together with typical values of surface temperature
to show that radiant output from most terrestrial surface lies within 90 W m - 2 of
390 W m - 2 - the value corresponding to the global annual mean surface tempera-
ture (388 K - see Fig. 8.6). Much of this output is offset by absorption of terres-
trial radiation coming downwards from the sky, which has an average value of just
over 340 W m- 2 according to Fig. 8.6. This too varies considerably, exceeding 293
9 Surface and boundary 390 W m - 2 under warm, low clouds, and falling below 300 W m - 2 under cold,
layer
clear skies. Such variations can have important effects on the net radiative balance,
but for the moment let us assume that as a net result of the emission and absorption
of terrestrial radiation, a typical portion of the Earth's surface loses energy at a rate
of 50 W m- 2 or about 4.3 MJ m- 2 per day.
Combining solar gains with terrestrial losses we see that the net result is that on a
typical summer day in middle latitudes a net input of about 20 MJ m -z in daylight
is followed by a net output of about 2 MJ m - 2 during the night. In winter there
may be a continuous net output since the weak daylight is more than offset by the
continuing net terrestrial output, and the daily loss total may be as much as
10 MJ m- 2 • In summer in particular, convection tends to offset the daylight input
very significantly by carrying sensible and latent heat into the atmosphere, and on
all but the stillest nights mechanical stirring of the atmospheric boundary layer
feeds some sensible heat from the air back to the chilling ground. (Dew and frost
formation represent the return of some latent heat, but the amounts are usually
relatively small.) The net result of convective and radiative exchanges is that many
surfaces suffer a daily alternation between heat gains and losses, each of which is
- 5 MJ m- 2 • In middle and high latitudes there is also a large annual alternation
between persistent daily gains in summer, when daytime gains outweigh nighttime
losses, and persistent daily losses in winter, when night-time losses exceed daytime
gains. Allowing for convective and advective offsets, the summer gains and winter
losses may each be - 500 MJ m- 2 per season.

9.4 Surface thermal response

We have seen that there are daily and annual rhythms of heat gain and loss by any
part of the Earth's surface, and have estimated their orders of magnitude. We
know that surfaces respond by heating and cooling, and know also that their
responses vary widely. In this section we look at the mechanisms involved in such
responses.

WATER

Since water is fairly transparent to solar wavelengths, especially in the visible

range, sunlight is attenuated progressively as it travels downwards below the
surface- a small fraction of the incident energy usually reaching depths of tens of
metres. Most of the attenuation ultimately leads to absorption and heating of the
water, but of course there is some backscatter to the surface, contributing a very
little to the oceanic albedo and allowing us to admire submarine sunbeams. The
simple fact of the considerable transparency of water to sunlight, compared with
the almost total opaqueness of even the shallowest layer of ground, accounts for
much of the enormous difference in the thermal responses of land and sea, the heat
capacity of a typical layer of illuminated ocean being very much larger than that of
a land surface, and its temperature response correspondingly very much smaller.
Despite the fact that sea water absorbs different solar wavelengths with different
294 efficiencies, extinguishing the near infra-red more rapidly than the visible, the
depth profile of overall solar flux density often follows the ideal exponential shape 9.4 Surface thermal
response
expected in a homogeneous absorber (Fig. 9.6 and appendix 9.3), with depths for
90<1/o attenuation varying from only a few metres in turbid water, usually found
near coasts, to as much as 50 m in the relatively clear waters often found in the
open oceans (though plankton may contribute to turbidity there). Absorption in
turbid waters is principally by suspended sediment and organic matter, and some
associated dissolved substances, whereas in clear water it is principally by the water
itself.
As mentioned in section 8.13 turbulent mixing in ocean surface layers increases
the depth of water effectively influenced by sunlight, enhancing their effective heat
capacity. Observations suggest that depths subject to daily and annual cycles of
warming and cooling are - 10 and - 50 m respectively, with corresponding heat
capacities - 40 and - 200 MJ oc- 1 m- 2 • If the warming and cooling were dis-
tributed uniformly throughout these depths, then daily gains and losses of
5 MJ m- 2 would produce temperature rises and falls of 0.125 °C, and seasonal
gains and losses of 500 MJ m - 2 would produce temperature changes of 2.5 oc
(Fig. 9.9). Because temperature changes are largest at the surface, despite the
effects of mixing, actual temperature changes at the surface are often larger than
these estimates, but not very much. Marine temperature cycles are therefore very
much smaller than those familiar to us on land.
Some of the mixing in the surface layers of the oceans is effected by mechanical
turbulence extending down from the troubled surface, but some is convective, and
in fact convective stability exerts an overall control on vertical exchange just as it
does in the atmosphere. For example the depth of the layer affected by summer
warming is limited by that warming itself: warmed surface water is less dense than
the cooler water beneath (the intervening layer with the vertical temperature
gradient being called a thermocline), and the convective stability of the set-up
inhibits vertical exchange of water just as a stable layer does in the atmosphere. The
difference is that in the oceans surface warming enhances stability while in the
atmosphere it diminishes it.
Cooling in the oceans is localized at the surface in the first instance, since terres-
trial radiation, evaporation and wind chill are concentrated there. By increasing
the density of water at the surface, cooling encourages convective instability and
hence the downward extension of the cooling by vertical mixing. In very sheltered
waters such as lakes, the destruction of the summer thermocline in the autumn can
be quite sudden, playing an important role in the annual biological cycle by redis-
tribution of nutrients etc., but in the oceans it is masked by constant commotion.
However, after the removal of the summer thermocline, further cooling is dis-
tributed through a much deeper layer of water than before, which may even extend
to the sea bed kilometres below.

LAND

The response of a complex, vegetated land surface to surface heating and cooling is
very difficult to analyse, but some important features can be seen by considering
the relatively simple case of a homogeneous layer of rock or soil as it responds to a
regular cycle of warming and cooling at the surface. This is a standard problem in
the theory of heat conduction [53], with a well-known and tested solution which we
will now examine and compare with observation (Fig. 9.8).
We should expect that at any depth below the surface, the rate of rise of tempera-
ture after the onset of surface heating will increase with the thermal conductivity k 295
9 Surface and boundary of the intervening layer and decrease with its heat capacity. The full theory shows
Ioyer that the net effect is determined by the thermal diffusivity
K = k/p C (9.1)
where p and C are respectively the density and specific heat capacity of the layer
material. The larger the value of K, the more freely is a change in surface tempera-
ture conducted downwards, and the deeper is the layer affected. Values of K for
typical surface materials are listed in Table 9.2, and range from 0.1 for dry peat to
0.75 for wet sand (units being I0- 6 m2 s- 1). Trapped air in dry soils is such a poor
heat conductor that it ensures low K values despite smaller values of p and C.
The depth of material affected also depends on the time period of the surface
warming and cooling cycle. If this period is short, then heat which has not pene-
trated far enough during the surface-warming cycle is conducted back to the
surface and lost during the succeeding cooling cycle. If the period is long, then a
greater depth of soil is significantly warmed and cooled.
The detailed solution shows that if the surface heating and cooling cycle has
period P, then there is an important depth parameter D which is known as the
damping depth and is given by
D = (KPhr)llz (9.2)
The damping depth defines the depth of the soil which is significantly affected by
the surface heating and cooling in several different but related ways.
I. If the amplitude of the surface temperature wave is !:::. T0 (Fig. 9.8), then the
amplitude !:::. Tz at depth z below the surface is given by
!:::.Tz = !:::.T0 exp(- z/D) (9.3)
The temperature wave therefore dies away exponentially with increasing depth,
its amplitude falling to 370Jo of the surface value at depthD, and to 10% at 2.3D
(Appendix 4.3).
2. As the temperature wave is conducted down into the layer, it lags further and
further behind the surface wave and is exactly in antiphase (minima coinciding
with maxima and vice versa) at depth 1rD (Fig. 9.8).

Table 9.2 Thermal properties of surface materials

Material Dry peat Dry sand Wet sand Water Air units

Density 300 1600 2000 1000 1.2 kg m- 3

Specific heat
capacity 1920 800 1480 4180 1000 J kg-1 K-1
Thermal
conductivity 0.06 0.3 2.2 0.57 0.025 W m- 1 K- 1
4000 6000 )
1at' x thermal
diffusivity 0.10 0.23 0.74 (- 103 - 5 X 106 m2 s-1
Damping depth
diurnal 5.2 em 7.9 em 14 em 5 m 300m
annual 1.0 1.5 2.7 100 7 X 103 m
Heat capacity
diurnal 0.04 0.14 0.60 31 0.63 ) MJ K- 1 m- 2
annual 0.81 2.75 11.4 590 - 12
eddy values

296
3. The amplitude b. T0 of the temperature wave at the surface is related to the heat 9.5 Partitioning between
H entering unit area of surface during the heating phase by surface and air

b. T0 = H/(-J2 pCD) (9.4)

from which it appears that the effective heat capacity of the affected layer (per
unit surface area) is -J2 pCD, and that the depth of the layer which heat input H
would warm uniformly by b. T0 is -J2 D.
It is clear from the values of D included in Table 9.2 that even the relatively pene-
trating annual temperature wave affects a soil layer to a depth of no more than a
few metres. Even in a responsive wet soil, a worm living at a depth of about 6 m Fig. 9.7 Net radiative heat
would experience only lOOfo of the annual temperature range experienced by its input H to a land surface
cousin living at the surface, and their temperature calendars would differ by 6 being partitioned between
convected heat H 0 to the
months. The daily temperature wave penetrates only tens of centimetres. The near
atmosphere and conducted
constancy of temperature at quite shallow depths has long been exploited in the heat H5 to the subsurface.
construction and use of cellars.
Values for the effective heat capacity of typical ground layers are included in
Table 9.2 and show that even relatively responsive wet soils have heat capacities of
no more than a few per cent of the values quoted earlier for the sea surface, all of
which underlines the fact that heat conduction through terrestrial surfaces is so
poor that only very shallow layers are significantly affected by daily and yearly
surface heating and cooling. The simple argument presented at the beginning of the
subsection on water still stands.

O'Neill, Nebraska

Surface
0
',
)~~
E 10
' ' \
u

0 20 6
"'c 26
.c
c.
Q)
30
--------- --- -------- 25 -
/
/
Fig. 9.8 Depth-time section
0 40 with isotherms labelled in °C,
showing downward progress
0600 1200 1800 0000 0600
of the diurnal temperature
wave under a prairie. {After
CST August 31 1953 Sept 1 [54] and [55])

9.5 Partitioning between surface and air

Equation (9 .4) represents an incomplete view of the situation at the surface because
it ignores the finite effective heat capacity of the air. Using subscripts s and a for
surface and air respectively, the total heat input H during the warming phase of the
surface temperature cycle must be partitioned according to
Ha = {pCD)a
H, (pCD), (9.5)

where the symbols are as defined in the previous section for heat conduction. In
fact heat transfer in the atmospheric boundary layer is by convection (and some
terrestrial radiation) rather than by conduction, and in the surface layers of the seas 297
9 Surface and boundary it is by solar radiation and turbulent mixing, but for present purposes all such
layer transport is assumed to be effectively equivalent to conduction - very rapid
conduction in the cases of turbulent and radiative transport. Note that the relevant
His the excess of net radiant warming over evaporative cooling, since this is the
sensible heat to be redistributed by conduction, turbulence and radiation.
By an obvious extension of eqn (9.4) we can now relate the amplitude of the
0.5 -40 surface temperature wave to the heat input H by
-0.2
H
D. To = ----- ---- (9.6)
~2[(pCD), + (pCD)al
Fig. 9.9 Typical heat
capacities (underlined, in
Consider now the extremely different cases of heat input to desert and ocean
MJ 0 ( - 1 m- 2 ) and associated surfaces, to assess the different partitioning of input between surface and
diurnal temperature ranges atmosphere.
(- in °C) for subtropical As the sun's rays begin to strike the dry soil of the desert in the early morning, the
desert and ocean.
surface warms rapidly because the heat conducts downward very inefficiently
through the air-filled soil. According to Table 9.2, the damping depth for the daily
heating cycle is only a few centimetres, and the effective heat capacity is
- 0.1 MJ oc- 1 m- 2 • The gradual elimination of the nocturnal inversion in the
overlying atmospheric boundary layer (section 5.13) means that the heat transfer
there is complex and changeable, but observations indicate that turbulent transfer
often operates as if the eddy thermal diffusivity (the turbulent equivalent of K) was
within an order of magnitude of 5 m2 s- 1• This is equivalent to an effective heat
capacity of about 0.6 MJ oc- 1 m - 2 for the warming layer of air. It follows that the
great bulk of the heat input to the surface passes into the air rather than into the soil
beneath, and that the total effective heat capacity of the ground and the air
together is - I MJ oc- 1 m- 2 • In a subtropical desert, despite the relatively high
albedo and net loss by terrestrial radiation (clouds being scarce), the heat input H
to the surface during the warming phase can easily exceed 20 MJ m- 2 , giving a
temperature cycle at the surface with amplitude exceeding 20°, which means a
range from maximum to minimum of double this value (Fig. 9.9). In fact these
estimates are quite conservative, and surface conditions may be even more
extreme. Such violent swings of temperature between blistering daytime heat and
freezing in the hours before dawn, shatter rocks by repeated violent expansions and
contractions, and impose severe conditions on any organisms living there.
However, the excellent thermal insulation of the dry soil, which is one of the
factors contributing to the extreme conditions at the surface, offers a refuge which
is taken up by many burrowing animals and dormant plants.
Similar estimates for the annual heating cycle in deserts show similar parti-
tioning between air and ground, but at 15 MJ oc- 1 m- 2 the total effective heat
capacity is much larger because of the greater penetration of heating and cooling
into both air and ground. However, the annual cycle of heat input and output in the
subtropics is quite muted because the sun is still quite high in the sky even in mid-
winter, with the result that the seasonal temperature cycle is often smaller than the
diurnal cycle, though still quite large. In middle- and high-latitude deserts, many of
which lie in the rain shadows of mountain ranges, the much larger annual range of
H values, together with the small effective heat capacities typical of all deserts, give
rise to very large seasonal swings of temperature from searing summer heat to
sustained sub-zero temperatures in winter. These too make great demands of land-
scape and life, and for example severely tested the pioneers opening up the North
American Midwest in the nineteenth century.
At the other extreme we have the case of the ocean-air interface. The estimates
298 of ocean heat capacity made in the previous section will do well enough for the
effective surface-heat capacity in eqns (9.5) and (9.6), i.e. 40 MJ oc- 1 m - 2 for the 9.6 The laminar boundary
daily heating cycle. The effective heat capacity of the atmospheric boundary layer layer
will be assumed to be the same as in the case of the desert (0.6 MJ oc- 1 m- 2),
although reduction in convective mixing may not be completely offset by increased
stirring by the freer ocean winds. Regardless of detail, the heat input to the surface
is obviously overwhelmingly taken into the sea rather than into the air, and the
total effective heat capacity of the sea and air together is several tens of times larger
than in the case of the desert. Since in addition the sensible heat input to the surface
is significantly smaller than in the case of the desert (the large evaporative cooling
more than offsets the effect of the much lower albedo), the result confirms the
explanation for the observed very small oceanic surface temperature range given in
the previous section.

9.6 The laminar boundary layer

Air molecules in contact with a liquid or solid surface are so effectively stuck to it
by molecular attraction that there is no relative motion. At a very little distance
from the surface, molecules can move, but only very smoothly and sluggishly on
account of viscous friction, and this laminar flow is always parallel to the surface.
With increasing distance from the surface, the speed of flow increases, as do the
accelerations and decelerations of air in response to disturbing factors such as
surface shape, the influence of turbulent jostling of air beyond the laminar
boundary layer etc. At some small critical distance o from the surface, such
unsteadiness overcomes the smoothing and damping effects of viscosity, and the
air flow ceases to be laminar. The value of the thickness oof the laminar boundary
layer in any particular set-up is very important, since resistance to the diffusion of
momentum, heat and water vapour through the laminar boundary layer increases
with its thickness.
The outer limit of the laminar boundary layer is associated with a critical value of
the appropriate Reynolds number. As outlined in section 7. 7, a Reynolds number
consists of UL/v, where U and L are respectively the characteristic speeds and
dimensions of the air flow in the laminar boundary layer, and vis the kinematic
viscosity of the air. The dimension in question is obviously o, but the choice of U is
less obvious since there is a large shear in the flow in the laminar boundary layer. As
will become clearer in the next two sections, there is a very important velocity para-
meter in the turbulent air beyond the laminar boundary layer which is called the
shear velocity or friction velocity u. and which is defined by
u = (Tip)Vz (9.7)
where r is the flux of horizontal momentum toward the surface, which is the
tangential stress on an embedded imaginary horizontal surface. This flux is borne
by turbulence outside the laminar boundary layer and by viscosity within it, though
the latter statement has to be qualified when the surface is rough (see later in this
section). Checking by the method of dimensions (Appendix 7.I) confirms that the
right-hand side of eqn (9. 7) has indeed the dimensions of velocity. Observations in
the turbulent air beyond the laminar boundary layer show that u is comparable in
magnitude with the root mean square of the wind-speed fluctua.tions there.
Observations in all parts of the atmospheric boundary layer indicate the impor-
tance of the friction velocity. In the case of the laminar boundary layer, observa-
tions in many different situations, including wind tunnels and water channels as 299
9 Surface and boundary well as in the atmosphere, show that its depth ocorresponds to a critical value - 10
layer
for the Reynolds number
u. Ofp (9.8)
Values of r can be measured directly by attaching delicate strain gauges
tangentially to horizontal plates, and indirectly as mentioned in section 9. 7. Values
vary very widely with atmospheric conditions, and when combined with reasonable
values of air density pin eqn (9. 7) form a wide range of values of u centred on -
0.3 m s - 1• It follows from typical values for v that the associated ·thicknesses of
laminar boundary layers at the base of the troposphere are - 0.5 mm, being
thicker when the overlying air is less turbulent and vice versa. The layer in which
transport is dominated by molecular rather than turbulent diffusion is therefore
extremely thin, being in effect no more than a viscous membrane of air adhering to
all solid and liquid surfaces.
Although very thin, the laminar boundary layer covers all exposed surfaces, and
is only significantly bypassed by momentum (section 9.9). All the other fluxes
passing to or from the Earth's surface (heat, water vapour, carbon dioxide etc.) are
forced to pass through it, and as a result of the considerable resistances encoun-
tered there are often quite sharp gradients of concentrations and associated
meteorological quantities. For example, sensible heat diffusing from the surface
obeys the equation for heat conduction (eqn (8.5)), which can be written in simple
difference form
H=ki:::.Tio (9.9)
where His the sensible heat flux per unit area and D. Tis the temperature drop
across the laminar boundary, depth o. A value of 100 W m - 2 for His quite typical
when fairly dry land is being heated strongly by the sun, and it follows from eqn
(9.9) that a fall of over 2 oc is needed to drive sensible heat at this rate through a
1 mm thick laminar boundary layer (using a realistic value for the thermal conduc-
tivity k). Notice that this temperature gradient is five orders of magnitude greater
than the dry adiabatic lapse rate, which of course applies only when air parcels are
able to exchange freely in the vertical - a condition not met in the laminar
boundary layer, and not met either, as we shall see, in the lower parts of the over-
lying turbulent layer. Though such highly localized temperature gradients are not
usually perceptible to animals as tall as ourselves, they probably do explain some of
the extra warmth we feel on lying down to sunbathe in a sheltered position. They
must however be very important to small insects, and contribute to the substantial
overestimates of air temperature by normal-sized thermometers exposed to direct
sunlight.
The laminar boundary layer diffuses tangential (most usually horizontal)
momentum from the moving atmosphere to the surface, and it would seem to
follow from the appropriate form of eqn (7.4) that
r = p, ~:::. v1o (9.10)
where D. Vis the difference (i.e. the shear) in tangential wind speed across the
shallow layer. Assuming a value of - I0- 2 N m - 2 for r (consistent with the value
assumed earlier for u ) it follows that D. V - I m s - 1 - a very substantial shear
across such a shallow ~one. But although shears of this intensity must occur above
some surfaces, it is quite misleading to imagine that a jump to I m s - 1 in horizontal
wind speed could be widespread only 1 mm above land and sea surfaces. The
unreality of such a situation is apparent when we remember that almost all natural
surfaces, including all but the calmest water, have roughness elements which are
300 much larger than the depth of the laminar boundary layer. For example each blade
 9.7 Turbulence
of grass in a grassy surface is enclosed in its laminar boundary layer as expected,
but the air flowing over and through the grass is impeded by the physical obstruc-
tion of the protruding grass stems rather than by the viscous layers, with the result
that the vertical profile of horizontal wind speed amongst the grass is determined
principally by the distribution, size and shape of the stems. Such a surface is said to
be aerodynamically rough, and its roughness is influential to levels well above the
tallest roughness element, as we shall see in the next section. Virtually all natural
surfaces are aerodynamically rough, the only exception being water in almost zero
wind and with no waves coming from elsewhere.

9. 7 Turbulence

Above the very shallow laminar boundary layer, the atmospheric boundary layer is
chronically turbulent. The effects of turbulence are many and varied: the gustiness
of all but the lightest winds is familiar (Fig. 9.10), as is the unsteady dispersion of
smoke plumes (Fig. 9.11), but other properties are less obvious, though no less sig-
nificant. Though turbulence in the atmosphere and other fluids has been studied
for over a century, and particularly intensively in recent decades, it is surprisingly
difficult to define precisely what we mean by turbulence. Sophisticated experi-
mental and theoretical work has served mainly to confirm the subtlety of the topic,
so that even an elementary treatment is beyond the scope of this book [54]. The
following is therefore a selective and mainly descriptive outline of some important
properties.
l. Turbulent flow is irregular in both space and time. For example the wind speed
in Fig. 9.10 varies continually and widely but shows no trace of regular oscilla-
tions such as might be associated with waves. (In fact recent studies of the
nocturnal atmospheric boundary layer in particular show that waves are often
present at some height above the surface. However, these are quite distinct
from turbulence, though they sometimes give rise to turbulence by breaking.)
Virtually every meteorological measurable varies in this way in the presence of
turbulence, though wind shows by far the largest percentage range. Variations
in temperature and humidity content are relatively small but are especially
important because of their contribution to heat and vapour fluxes, as we shall
see.
2. The unsteadiness and irregularity typical of turbulent flow represents a large-
amplitude, highly chaotic response to inherent instability, which prevents

Fig. 9.10 Anemograph for

port of a typically breezy day
in north-west England. (P.
Lursson and M . Beadle.
De portment of Environmental
Scie nce, University of
Lancaster).

16
Time <hours GMTI 301
9 Surface and boundary
Ioyer

Fig. 9.11 A smoke plume

undulating and spreading in
the turbulent daytime
boundary layer of air flowing
along a steep-sided valley.

useful prediction of the precise pattern of flow at any future time. Quantitative
description and prediction therefore involve statistics ranging from the simple
arithmetic mean and root mean square deviation to the sophisticated power
spectrum (which shows the contribution of different periodicities to the total
amount of variability) and beyond . In qualitative description we can still make
use of the concept of temporarily distinguishable eddies, though with the tacit
assumption that they are too numerous and transient to be usefully observable
in practice_
3. Turbulence is intrinsically three-dimensional and cannot be adequately
described in fewer dimensions. This handicaps pictorial description and mental
conception (which is strongly two dimensional), though of course flat pictures
like Fig_ 9_12 have to be used for limited description. If there is a dominant
I
I describable type of motion it is a continual stretching and folding which seems
,
I
to be common to topological properties of chaos [2]_ And although a consider-
1 - - - - --+i'
able range of atmospheric turbulence is isotropic (without preferred direction

;
£: I

or axis), there is often a difference between longitudinal structure (horizontally

~;?j' across the flow) and lateral structure (across the flow).
4_ Turbulence is hierarchicaL Inherent instability produces large eddies which in
turn produce smaller eddies, and so on down a cascade of diminishing scales
Mean wi nd speed V
until the eddies are small enough to be literally rubbed out by viscosity.
Fig. 9.12 Air tumbling in According to a famous jingle by a pioneer in the study of atmospheric
strong wind shear near the turbulence, LF. Richardson:
surface.
Big whirls have little whirls which feed on their velocity,
And little whirls have smaller whirls, and so on to viscosity.
The largest eddies have scales comparable with that of the atmospheric boundary
layer itself (i.e. - 100 m or somewhat larger), and the scale of the smallest is -
I mm for the same reason that this is the depth scale of the laminar boundary layer_
At the large end of the scale it is sometimes difficult to distinguish between large
eddies and small structured systems, the ' smoke ring' vortex believed to be
302 common in buoyant convection being a case in point.
 The fact that the largest recognizable turbulent eddies influencing conditions 9.8 The origins and role of
turbulence
near the surface are - 100 m carries the implication that most turbulent variations
(considering a variation to be a cycle of overshoot and undershoot about a notional
mean) last less than - 100 s at a fixed position, assuming a typical horizontal wind
speed. If averages are taken over time periods several times this length, the
resultant values are reasonably stable. The meteorological synoptic convention is
to take ten-minute averages, and it is these values which are recorded in the
standard hourly observations, variability being very asymmetrically assessed by
recording the highest wind speed (the strongest gust) in the observation period.
There are of course variations on scales longer than a few minutes, and in fact there
is a continuous range or spectrum up to time periods corresponding to the passage
of major weather systems such as depressions, but the ten-minute average is espe-
cially useful because there is something of a gap in the spectrum separating
turbulence proper from longer period variations. In the following sections a ten-
minute averaging period is assumed for all averages unless otherwise stated.

9.8 The origins and role of turbulence

Turbulence persists because air flow is continually disturbed on scales which are
much too large to be smoothed away by viscosity. If disturbances of scale 100 m
are being introduced into an air flow of average speed 1 m s -I (a very modest value
in the atmospheric boundary layer), the very large value (- 106) of the associated
Reynolds number indicates that viscosity is quite incapable of maintaining laminar
flow. The disturbed air flow therefore continues jostling in an irregular fashion, so
that turbulence continues even if the disturbing factor operates only intermittently.
The single most important source of disturbance is the wind shear which is
endemic near the surface in all but absolutely calm conditions (section 7.2). Air
parcels embedded in the wind shear tend to roll forward about a lateral horizontal
axis, temporarily becoming the rotating parcel which is the archetypal eddy. Rota-
tion is not confined to the longitudinal section portrayed in Fig. 9.12, presumably
because as a parcel begins to roll about a lateral axis, it tends to generate by friction
opposing rolls about longitudinal axes on either side (Fig. 9.13).
All rolling parcels have descending air on one side and ascending air on the other,

Fig. 9.13 An air parcel

rolling about a lateral axis
with secondary vortices
rotating in both senses about
longitudinal axes.

' 'lateral
/ ' 303
9 Surface and boundary but if the air is convectively stable, as is often the case near the ground at night,
Ioyer
vertical motion is inhibited by the effects of reversing buoyancy (section 5.12). It
seems that the maintenance of turbulence by shear in these circumstances must
depend on a balance between the strengths of the shear (given by aVI az, where Vis
the average horizontal wind speed) and of the buoyancy, given by (g/0) ao;az
(derived from eqn (7 .27), where 0 is the average potential temperature. Note that
ao/az is positive when the air is convectively stable. This balance was analysed by
Richardson who showed theoretically that turbulence should tend to break out
when the magnitude of the dimensionless number (subsequently called the
Richardson number and denoted by Ri)

Ri = ..¥.._
o
~~(~)
az az
2 (9.11)

falls below 0.25. Many observations of the atmosphere confirm that critical values
of Ri are indeed of this order. Substitution of realistic values for g and 0 (in degrees
Kelvin) show that very considerable stability is needed to prevent generation of
turbulence by typical observed values of wind shear (a V/az) near the surface. This
is consistent with the observed persistence of turbulence there even on nights on
which there is pronounced cooling, provided the wind speed is kept up by the large-
scale weather situation. If, however, surface cooling is sufficiently strong to raise
the value of ao;az to the point where Ri exceeds the critical value, turbulence is
damped out by the static stability. Then turbulence no longer brings down V
momentum from air at higher levels, and friction with the ground slows the air near
the ground to a virtual standstill. Since surface cooling (assuming this to be a fairly
clear night) is no longer partly offset by the flux of sensible heat brought down by
turbulence from the warmer air aloft, the cooling and stability increase, and the air
remains still until the next day's sun or a large-scale change in the weather
intervenes.
Thermally-driven convection also generates turbulence, provided heating from
beneath exceeds the ability of the air to transfer heat by molecular conduction, and
provided the motion generated is too vigorous for viscosity to maintain laminar
flow. In a layer of air of depth l::,.z and potential temperature lapse 6.0, this balance
is described by the dimensionless Rayleigh number (Ra)

R = g/:::,.0 (/:::,. )3 (9.12)

a kvO Z

height In all terrestrial conditions in which a relatively warm surface maintains a lapse
<m> 6.0, Ra far exceeds the known critical value for the onset of turbulent convection
(- 50 000). Although very common, the resulting convection is not well under-
stood. Very near the surface the convection is quite small-scale, and the turmoil of
0
centimetric and metric-scale eddies rising and falling produces the shimmering
effects so noticeable when looking at distant objects just above a strongly heated
Fig. 9.14 Outline of surface. At greater heights it seems that these small movements cooperate to
probable transition from produce larger-scale thermals which rise up through the atmospheric boundary
metric-scale 'shimmer' just layer, leaving a turbulent wake as they go. The latter stage corresponds well to
above a strongly heated land Richardson's jingle, but the former suggests a reverse sequence of building from
surface to hectometric-scale
thermals aloft.
smaller to larger scales (Fig. 9.14). Regardless of detail the effect is to maintain
turbulence on a wide range of scales in the convecting layer.
As mentioned already, turbulence transports momentum, heat and material
such as water vapour and aerosol, and this transport occurs whenever air parcels
(eddies) moving in different directions carry systematically different quantities of
304 momentum, heat, etc. Like the molecular diffusion which it overwhelms,
turbulence always carries momentum etc. from regions of excess to regions of 9.8 The origins and role of
turbulence
deficit, i.e. down-gradient. In many cases the gradients are effectively vertical and
so therefore are the transports. As a result momentum is always transported down
to the surface, and water vapour is almost always transported up. Over land
sensible heat is often transported up by day and down by night, though the
nocturnal fluxes are usually much smaller.
Consider for example the downward transport by turbulence of V momentum
(i.e. horizontal downwind momentum) through horizontal unit area, bounded by a
wire frame for ease of visualization. Since most of the effective eddies are larger
than the frame itself, even in the SI system where the unit frame is 1 m square, an
average downward flux arises because in the sequence of up- and downdraughts
successively passing through the frame, the downdraughts systematically carry
more V momentum as they must do coming down from the faster-moving air aloft
(Fig. 9.12). The instantaneous downward flux of Vmomentum is given by- pwV,
where the minus sign ensures that downdraughts (negative w) contribute positively
to the flux. Note that - pw is the downward mass flux. The average downward flux
of momentum is therefore given by the average of all the instantaneous values of
the triple product. As mentioned in section 7.2, this flux is identical with the
average horizontal shearing stress
7 = - pwV (9.13)

Although the average value of the vertical mass flux is zero near a horizontal
surface (because air does not form condensations or rarefactions so long as it flows
more slowly than the speed of sound), the momentum flux can have quite substan-
tial values when variations in w and V are negatively correlated, as they are in
Fig. 9.15. We can look into this further by examining the right-hand side of eqn
(10.13) more closely.
Contributions from density variations are so small relatively that p can be treated
as a constant and removed from under the over bar. It is shown in appendix 9.4 that
eqn (9.13) simplifies to
7 = - pwV' (9.14)
where V' is the instantaneous deviation of V from its average value, and w is iden-
tical with w' because average updraught is zero. The term w V' is the covariance of
vertical and horizontal winds, and is non-zero when their fluctuations are
correlated (Fig. 9.15). Combining eqns (9.14) and (9.7) we see that
u. 2 = - wV' (9.15)
which stresses the interpretation of u as a velocity related to the gustiness of winds.
These results show that if we des~ribe air flow in terms of average rather than
instantaneous values, we must allow for correlations between variations occurring
within the averaging period otherwise we will miss potentially significant

Fig. 9.15 Idealized

anemographs of vertical and
, horizontal wind speeds over a
Q)
Q)
period of a few minutes in
c. relatively steady conditions,
,"'c: showing the negative
correlation which is normal
3:
near the surface.

305
9 Surface and boundary momentum fluxes, which is to say that we may ignore important forces. The equa-
layer
tions of motion (eqn (7 .12)) are always written in terms of average values, since
it would be useless to apply them to instantaneous observations of quantities
varying so rapidly and considerably as winds, and hence they need correction. A
thorough treatment of the equations of motion in terms of averages and instan-
taneous deviations [56] shows that terms such as a(wu')laz must be included along
with viscosity terms such as va 2ul az2 to represent friction terms such as Fx in full in
eqn (7.12). The extra terms are the gradients of terms such as pwu' which are
known as Reynolds stresses, and their inclusion completes the treatment of the
equations of motion begun in Chapter 7.
The Reynolds stress represented by eqn (9.14) is often the largest of the many
possible components because av;az is usually the largest shear, being the vertical
gradient of horizontal wind. It can be measured by recording simultaneously the
outputs from a conventional anemometer (measuring V) and a propellor on a
vertical shaft (measuring w) and calculating their covariance - easily done now
using a dedicated microchip. Each anemometer must of course be able to respond
accurately to the smallest eddies contributing significantly to the stress, and this
requires instruments with much faster responses than the normal robust and there-
fore rather sluggish instrument used for synoptic observation.
A very similar treatment shows that the vertical turbulent flux of sensible heatH
is given by
H = p Cp wT' (9.16)
where wT' is the covariance of updraught and temperature, and the minus sign is
absent this time because it is conventional to consider an upward heat flux to be
positive. At night, turbulence persisting above a cooling surface maintains a nega-
tive covariance of updraught and temperature and hence a downward heat flux,
whereas by day vigorous turbulence over strongly heated ground maintains a
strong positive covariance and hence an upward flux.
The corresponding expression for the turbulent flux of water vapour is
F = p wq' (9.17)
where q is the specific humidity. Multiplied by the coefficient of latent heat L, this
vapour flux becomes the turbulent flux of latent heat.

9.9 The surface boundary layer

Turbulence as outlined in the previous sections dominates the distributions and

vertical transports of momentum, heat and matter such as water vapour and
carbon dioxide throughout most of the atmospheric boundary layer (the laminar
boundary layer being a spatially very small exception). In the lower parts of this
turbulent layer the turbulence is so dominant that some striking regularities
emerge. For example the horizontal components of the equations of motion (eqns
(7 .12a) and (b)) give rise to a simple balance along the flow between the gradients of
the Reynolds stresses and the longitudinal component of the pressure gradient
force

(9.18)
306
where n is a horizontal axis pointing upwind. The pressure gradient op/on is 9.9 The surface boundary
layer
usually dictated by the synoptic-scale pressure field, and consequently is of order
10- 3 Pa m -I (section 7. 7). Ignoring vertical variations of air density p, the change
!:::. T in T to be expected in a vertical depth !:::.z is given by a simple form of eqn
(10.18):

!:::.7 =/:::.z op - 10- 3 !:::.z

on

1fT has a value of0.1 Pa (typical offairly quiet conditions), then the change !:::.Twill
be less than 100Jo of T if !:::.z is less than 10 m. In fact the uniformity of average
momentum flux T along a vertical from heights - em (clear of most small rough-
ness elements) to - 10 m is well confirmed by observation, and is one of the most
striking regularities of the surface boundary layer, otherwise termed the constant
flux layer because of the vertical uniformity of the vertical fluxes of heat and
matter as well as of momentum. Complications arise because the upper limit of the
surface boundary layer varies very considerably with time and position, especially
overland. Using the 10% criterion again, the layeris only - 1 m deep in nearly still
nocturnal conditions when turbulence and T have died away to low levels, but it
may be - 100 m deep when strong wind shear or convection encourage very
vigorous turbulence. But though the change from one extreme to the other can be
quite rapid, as from dawn to mid-morning for example, it is usually sufficiently
slow for T to be well-defined in ten-minute averages, and it is in these that the
vertical uniformity of profiles is observed.
The relative simplicity of the surface boundary layer, and the fact that it is the
part of the atmosphere in which we spend most of our outdoor lives, has attracted
great attention since the study of the atmospheric boundary layer began in earnest
early this century. In the absence of an approach from the first principles of the
theory of turbulence, people looked for simple relationships involving readily
measureable properties, in particular ten-minute average values. In the case of
wind, the vertical profile of average wind consistently shows a characteristic shape
with very strong wind shear oVI oz near the surface and smaller values at increasing
heights (Fig. 9.12). The simplest dimensionally consistent expression for the shear
oVloz is of the form
av Au
oz z
where A is a pure (i.e. dimensionless) number and U is a velocity characteristic of
air flow in the layer. A great deal of experimental evidence in the atmosphere, and
in wind tunnels and water flows, confirms that this simple relationship actually
holds in the form
av ~ (9.19)
oz kz
where u is the friction velocity introduced in section 9.6 and k is an apparently
univers.{l constant with value close to 0.4 (experimental difficulties frustrate very
accurate measurement) named after von Karm(m. It is obviously reasonable that
the velocity parameter on the right-hand side of eqn (9 .19) should be u since this
reflects the turbulent activity which dominates the layer. The further app~arance of
u is another example of its great significance.
·Notice that eqn (9 .19) is consistent with the shape of the mean wind profile, with
shear increasing with decreasing height (Fig. 9.12). In fact infinite shear would be
implied at zero height, but the surface boundary gives way to the laminar boundary 307
9 Surface and boundary layer (or to the layer penetrated by roughness elements such as grass stems) before
layer this unrealistic extreme is approached. Nevertheless the presence of very strong
shear at very low heights, together with the role of wind shear in maintaining
turbulence (eqn (9.11) and discussion) is consistent with the observed presence of
vigorous small-scale turbulence near the surface for so much of the time. Over land
this is quenched by the strong temperature inversions which develop on still, clear
nights, but over the sea these are inhibited by the great heat capacity of the surface
layer of water, so that strong shear and turbulence are present nearly all of the time,
each playing a role in raising and maintaining sea waves [58].
If we assume that 7 and p are uniform along a vertical axis through the surface
boundary layer at any particular location and time (realistic, though the uniformity
of average wind stress 7 is never very precise) then it follows that u too is uniform,
and we can easily integrate eqn (9 .19) to obtain an expression· for the vertical
distribution of average wind speed V:
u.
V = klnz + B (9.20)

where B is a constant. Such a simple linear relationship between average wind speed
and the logarithm of height is so widely observed, at least to a reasonable
approximation (Fig. 9.16), that it is known simply as the log wind profile. It implies
(Appendix 9.5) that wind speed increases in equal steps for equal multiples of
height increase, the speed increment for a doubling of height being (u In 2)/ k, i.e.
almost exactly 1. 73 u•. Because of the straight-line relationship in Fig.· 9.16, extra-

Fig. 9.16 A nearly pure log

wind profile (time 0835 CST)
of mean horizontal wind over
a prairie, including O'Neill Nebraska September 7 1953
aerodynamic roughness
length (5.4 mm) found by
15
extrapolation to zero wind Local times (CST)
speed (dashed line through
several cycles of the verticol
scale). Curved lines show 10
slight deviations observed in 9
adjacent convectively stable 8
(time 0435- before dawn)
and unstable (time 1035) '" 7

situations. (Data from [56])

"'
u
6
"'
Cl Zo
0 5

E 4

"''"
Cl

J:
3

I
I
I

4 6 8 10

308 Mean wind speed (m s-'J

polation suggests that there is a non-zero height z0 at which Vis zero. This is called 9.9 The surface boundary
layer
the aerodynamic roughness length, and it has a fairly well-defined value for any
particular situation in which the surface does not deform much with increasing
wind speed. It appears as the lower height limit in the particular form of eqn (9.20):

V = _!!:_ In __£ (9.21)

k Zo

According to eqn (9.21) the average wind speed at any height in the surface
boundary layer is completely determined by the characteristic length z0 and the
characteristic velocity u . A trivial rearrangement produces a general relationship
between normalized wilid speed Vlu. and normalized height z/z0 :

~ = - 1 In __£ (9.22)
u. k Zo

In recent studies of all parts of the atmospheric boundary layer, sophisticated use is
made of such general, normalized relationships, most of them much more complex
than eqn (9.22). It is possible to conduct quite general and subtle discussions in
such terms, and hence derive relationships to be checked against observation. For
example the logarithmic function on the right-hand side of eqn (9.22) is a particular
case of the more general function F(z/ zJ which applies to the surface boundary
layer in a much wider range of atmospheric conditions.
There is no simple relation between the value of Zo and the average height I of the
surface roughness elements (e.g. lengths of grass stems in a meadow); this is fairly
obviously because shape and distribution must also be important, but as a very
rough guide zufl - 1/w in many cases. Because Zo is typically very much smaller
than I it follows that the actual shape of the wind profile must deviate from the
idealized log profile well above a height of z0 , since we can hardly expect flow
among the roughness elements to be similar to flow clear of their tops. It follows
that the only way to measure z0 for a surface is to observe the profile of average
wind above it, using an array of anemometers on a mast, and to find z0 by extra-
polation as on Fig. 9.16. Since air flowing from above one surface to above another
with a very different z0 value adjusts only very slowly to the change in underlying
surface (as a rule of thumb the depth of the fully adjusted layer is one hundredth of
the fetch - the horizontal downwind distance from the boundary between the
surfaces - Fig. 9.17), it is important that the mast be positioned with adequate
fetch. In small-scale terrain like much of the British Isles, the shortness of homo-
geneous fetches means that the surface boundary layer is seldom fully adjusted,
and that the log wind pr9file holds only approximately.
When roughness elements are tall and densely packed, as they are in mature
cereal crops and forests, for example, heights must be measured relative to a datum

Unadjusted _.._.. ..... - - - -

-- --- Fig. 9.17 Sketch of
downwind growth of
_,,.""'""' adjustment of a surface
_,"" Partly adjusted boundary layer to a change in
Air flow
~ /
/"' "' .........
----------------
surface roughness at C. The
vertical is greatly enlarged,
"-----y' / ---- but is only notional since the
,(./~..,. Fully adjusted partly adjusted layer should
!lllm~\II\II!IIII,I,!!!Jl!luttlllj!Uiu!/1,,,,,], 1 111111J!I!II!i!!.IJ,,,,!!l!l!!!!lll!di!J!,HII!!!l,J,I!!IJ!l!h!''·'''''IU!!I,t!,!t!it,, be - 10 times deeper than
the fully adjusted Ioyer.
c
-----Fetch---- 309
9 Surface and boundary
layer
significantly raised above the ground to find a satisfactory fit with the log wind
type of profile. If the required zero plane displacement is d, then
15
V = ~In (z- d) (9.23)
k Zo
E

.c
fits observed data quite well down to levels just above the crop or canopy. Wind
.51'
Cl>
flow within the crop or forest follows no general profile, being controlled by the
:I: vertical distribution within the particular vegetation in question (Fig. 9 .18).
Forests with dense crowns and nearly bare trunks below can even have a secondary
wind speed maximum at bare trunk level.
Wind speed (m s-1)
Rearranging eqn (9.21) to find an expression for the tangential stress 7, we see
Fig. 9.18 Profile of hourly that this stress is proportional to the square of the average wind speed Vat any
mean wind speed in a forest particular level.
of Sitka spruce in the middle
of a sunny day. Curvature 7 = pC0 V2 (9.24)
represents profiles over open
ground above the level of
maximum foliage density.
The constant of proportionality C0 is the drag coefficient, and depends only on the
(After [54] and [59]) normalized height z/z0 at which Vis measured. The sensitivity of stress 7 to V
apparent in eqn (9.24) points to the importance of rare windy extremes in imposing
high stresses on the surface, and thereby causing damage. Forests can bear scars for
decades which were caused by only an hour or so of unusually high average winds
with their associated even briefer gusts.
Strictly speaking the log wind profile is to be expected only when the air is
neutrally stable for convection, otherwise a buoyancy term should appear in eqn
(9 .19) and its derivatives. In convectively stable situations there are slight but
systematic deviations from the log wind profile, because the damped turbulence
allows a greater wind shear aloft. In convective instability the opposite occurs, the
wind shear aloft being reduced by the enhanced turbulent friction. However, these
deviations from the pure log wind profile are quite small and are detected only by
careful observation. A much more obvious result of the reduction of turbulence
which occurs on most nights, and its increase during most daylit periods, is that at
any particular location (i.e. any particular height above any particular surface) the
average wind speed increases with the turbulence. This is apparent in eqn (9.21),
where Vis directly proportional to friction velocity u. for fixed z and z0 , and u. in
turn increases as the square root of the turbulent momentum flux 7 (which is the
horizontal shearing stress). Clearly when turbulence is vigorous, as it is when
enhanced by thermally driven convection, momentum from faster-moving layers
aloft is transferred down to the surface boundary layer, speeding up the average
wind there. At night this downward transfer may die away almost completely, if
the surface cooling and stabilizing is strong enough, so that the surface boundary
layer is brought nearly to a halt by friction with the underlying surface. The
resulting diurnal variation in wind speed near the surface is one of the most familiar
features of climate over land surfaces (Fig. 9.19). Over the sea, the diurnal rhythm
in thermally driven convection is greatly muted by the large thermal capacity of the
sea, so that the effect is weaker and much less closely tied to day and night. The
extra momentum delivered to the surface boundary layer daily over land is of
course brought down from higher levels, where the average wind speed is corres-
pondingly reduced. Figure 9.19 shows this happening quite clearly at the 300m
level, well above the surface boundary layer. At such levels the result of enhanced
turbulence in daylight is to extend upwards the effects of the frictional drag at the
surface.
By multiplying each side of eqn (9.19) by u and rearranging, we find another
310 expression for the horizontal shearing stress: •
 25 9.9 The surface boundary
O'Neill, Nebraska
layer

•., 20
E

.,
"'0

~ 15
"'
"'0
c:
"i
.
~ 10

.
E

:I
c:
E 5

Fig. 9.19 Typical diurnal

variation in mean wind speed
OLL-----L----~L------L----~~----~----~
0400 0800 1200 1600 2000 0000 0400 at heights of 12 and 300 m
above a prairie site. (Data
Eastern Standard Time September 7 1953 II Sept 8 from [56])

T = p (k u z)-
av (9.25)
• az
Comparing this with the equivalent expression for wind stress and shear in laminar
flow (eqn (7 .4)), it is apparent that k u z plays the same role as the kinematic
coefficient of eddy viscosity v. Howeve~. unlike its molecular counterpart this
coefficient of eddy viscosity K is not simply determined by the thermodynamic
state of the air; it depends on both the turbulent intensity (through u ) and the
height z, and it is the increase of K with z which maintains the invers'e relation
between wind shear aV/az and height z which is so characteristic of the surface
boundary layer.
Similar analysis leads to an equivalent expression for the turbulent vertical flux
of sensible heat

H = - p Cp (k u, z)
ao
az (9.26)

where His the sensible heat flux density in watts per square metre, and the minus
sign is included because it is conventional to regard an upward heat flux as positive.
Positive H occurs when there is a lapse of potential temperature with increasing
height, i.e. a superadiabatic lapse rate (section 5.12). In this simple type of analysis
the eddy coefficient for heat transport (the turbulent equivalent of the thermal
diffusivity) is assumed to be the same as the eddy coefficient for momentum trans-
port, k u z. More detailed analysis and observation suggest that this is not strictly
the case,· and that heat is transported more efficiently than momentum when the
layer is convecting, while the opposite is true when the layer is convectively stable.

HEAT AND VA POUR FLUXES

Just as T, the average vertical flux of horizontal momentum, is nearly uniform

throughout the depth of the surface boundary layer, so is the average vertical flux 311
9 Surface and boundary of sensible heatH, provided there is no stratifying mechanism at work, such as a
Ioyer shallow fog or haze layer, interfering with net radiant and convective heat fluxes
near the surface. If His nearly uniform along a vertical axis, then according to eqn
(9 .26) the vertical gradient of potential temperature must be inversely proportional
to height z, which is consistent with the commonly observed concentration of
strongly superadiabatic lapse rates close to strongly heated land surfaces (section
5.13), and a similar concentration of subadiabatic lapse rates close to cooled
surfaces. We can estimate the lapse b.() of potential temperature between heights z1
and z2 by integrating eqn (10.26) to obtain

b.()= H In ( - z2 ) (9.27)
p CPku. Zt

Substituting 100 W m- 2 for H (corresponding to fairly strong surface heating),

and 0.3 m s -t for u , it follows that the potential temperature lapse between the
base of the surface boundary layer and screen level (let us say heights 1 em and
1.5 m respectively) is about 5 °C. Although eqn (9.27) is probably not very
accurate in these conditions, because of the presence of significant buoyancy, the
indication that temperature lapses of this magnitude can occur is well supported by
observation. The temperature lapse switches over to a temperature inversion in the
presence of significant surface cooling. This behaviour, together with the similar
behaviour of the laminar boundary layer overlying the surface, indicates the
marked increase of diurnal temperature variations between the modest values
measured at screen level and the much larger values observed much closer to an
exposed land surface (Fig. 9.20)- an increase which has a marked influence on
the flora and fauna living there.
Similar analysis leads to the equivalent expression for the average vertical
transport of water vapour by turbulence:
dpv
F = - (ku z)-- (9.28)
• az
where F is the vapour flux density and Pv is the vapour density, and there are
reservations about applicability and accuracy similar to those associated with eqns
(9.26) and (9.25). Multiplied by the coefficient of latent heat L, this vapour flux
becomes the flux of latent heat
dpv
H 1 = - L (k u• z) --a-
z (9.29)

Notice that to the extent that the eddy diffusion coefficients are identical in eqns
(9.26) and (9.29) (though in principle buoyancy must enforce some differences

Fig. 9.20 Daytime and night- 150

time profiles near the
surface. The Iorge diurnal
range is typical of continental 100
mid-latitudes in summer E
(Paisley, Ontario, 19 July
1962). Detail within 2m of the .,
:c
."'
I
50 1400
surface is lost by the balloon
observing technique. (After
[55])
16 18 20 22 24 26
312 Temperature (°C)
which are ignored in this simple treatment), they cancel in the expression for the 9.10 Above the surface
Bowen ratio boundary Ioyer

f3 = HIH
1
= pCP (
L
!!__
az
j ap. )
az (9.30)

with the result that this very important climatological parameter can be estimated
simply by comparing vertical profiles of mean temperature and vapour density.
Although this must be severely complicated by the prevalence of large diurnal
variations in buoyancy effects, especially over land, it is a measure of the difficulty
of finding better methods that this approach has been used extensively to partition
sensible and latent heat fluxes in major studies of the atmospheric energy budget,
such as those discussed in Chapter 8.

9.10 Above the surface boundary layer

Rising above the surface boundary layer, turbulence gradually weakens in the
diminishing shear, and eddies become larger and in some cases rather too struc-
tured to be considered turbulence proper (for example the vortex-ring type of
thermal). The average horizontal stress r arising from the Reynolds stresses reduces
gradually from the value in the surface boundary layer, and eventually becomes
insignificant in comparison with horizontal pressure gradient force. At such levels
therefore the remaining Reynolds stresses have negligible effect on the balance
between pressure gradient and Coriolis forces, and the quasi-geostrophic balance
discussed in section 7.8 prevails. The lowest level at which this happens is known as
the gradient level, and though seldom sharply defined by observation, the gradient
level is considered to be the upper limit of the atmospheric boundary layer and the
lower limit of the free atmosphere.
Rather subtle arguments suggest that the height of the gradient level above the
local surface should be a modest fraction of the space scale u.ff, where u. is the
friction velocity in the local surface boundary layer andjis the Corio lis parameter. I
Observations in the atmospheric boundary layer bear this out reasonably well, and I
I
show that the gradient level is - 500 m above the surface on many occasions, being I
I
much lower when the turbulence in the surface boundary layer is relatively weak (as I
I
on clear, still nights over land), and considerably higher when the surface boundary "' T I

layer is unusually turbulent. -§,

I
1
v
Consider the forces acting on a layer of air somewhere between the surface I;;; 1
I
boundary layer and the gradient level. They will produce a net force opposing the I
motion of the layer because the turbulent drag forward maintained by the slightly
faster air just above the layer in question is somewhat less than the turbulent drag 0 -+-=----- '----
backward maintained by the slower motion just below the layer (remember the
shearing stress r diminishes upward, as in Fig. 9.21 ). As discussed in section 7.11, Mean wmd speed V
Horizontal sheanng stress T
this net drag pulls the air flow away from its geostrophic alignment with the isobars
so that it has a component toward low pressure and is subgeostrophic in speed. This Fig. 9.21 Idealized vertical
profiles of mean wind speed
sort of behaviour was first analysed by Ekman in 1905 (actually in relation to the and shearing stress in and
behaviour of surface layers of the ocean in the first instance), and his theory just above the surface
outlines the elegant exponential spiral of wind vectors depicted in Fig. 9.22. The boundary layer.
cross-isobar angle a increases with decreasing height above the surface, giving rise
to winds which back with decreasing height in the northern hemisphere (i.e. their 313
9 Surface and boundary /

layer / / Ekman ideal / A n actual case

Fig. 9.22 Ekman spiral

(dashed) of mean horizontal
wind vectors (arrows)
between the top T of the ""
Iii
surface boundary layer, and c0
the gradient level G. N

Numbers are heights in sixths 0

I
of the level at which the wind
is first parallel to the
geostrophic wind (the
gradient wind strictly). An
actual profile is plotted for
comparison (solid line). (After Hori zontal x
[8])

direction rotates anticlockwise when seen in plan) and veer with decreasing height
(rotate clockwise) in the southern hemisphere. The Ekman spiral is confirmed in
outline by observations, but the maximum value of a never reaches the 45° pre-
dicted on the strength of his over-simple treatment of turbulent diffusion, and
seldom exceeds 30°.
The maximum value of the cross-isobar angle is reached at the top of the surface
boundary layer and is maintained from there down to the surface because the rela-
tively small differences between the large values of T at different heights in the
surface boundary layer (T being large and nearly uniform throughout the depth of
the surface boundary layer) are able to balance the pressure gradient and Coriolis
forces without requiring significant turning of the wind vector with height. Of
course the wind speed in the surface boundary layer falls away more and more
rapidly with decreasing height (Fig. 9.21), in accordance with the log wind profile
outlined in the previous section. In fact observation and theory indicate that at the
10 m level wind speeds are typically about 40o/o of the geostrophic value, the per-
centage increasing with increasing cross-isobar angle and decreasing surface
roughness.
Although turbulence weakens above the surface boundary layer, the larger
eddies aloft seem to be well able to maintain the upward fluxes of sensible and
latent heat which power the great cloudy weather systems of the free atmosphere.
When the whole atmospheric boundary layer is being stirred by buoyant con-
vection, its upper parts are the site of rising, structured eddies often loosely entitled
thermals. The precise nature of such thermals has not been determined yet by
observation, since they are too large and mobile (- 100 m and - 1 m s _, respec-
tively) to be investigated by instruments on masts, and too small to be traversed by
more than one instrumented aircraft at a time. Indirect evidence from birds and
glider pilots who use their gentle updraughts to stay aloft with minimum effort can
be interpreted either in terms of plumes fountaining from the strongly super-
adiabatic surface layer, or in terms of discrete ring vortices (like invisible giant
smoke rings) floating up after disconnection from their surface origins.
However, the whole layer above the surface boundary layer is often layered to an
extent which makes unrealistic both the maintenance of the Ekman spiral, and con-
vective fluxes which are even remotely uniform in the vertical, especially over land
where the diurnal changes can be so large. For example, during the destruction of
the nocturnal inversion discussed in section 5.13, the deepening convecting layer is
bounded above by the very stable remains of the inversion, which may reach down
314 to within a few metres of the surface in the early morning. The resulting layering of
turbulence is bound to lead to sharp vertical gradients of vertical fluxes of heat and 9.11 Airflow over uneven
momentum and consequent shaping of the profiles of temperature and wind speed. surfaces
Layering is also produced by the differential flow associated with sea breezes and
slope winds, both of which are important examples of processes which influence
surface and boundary layer conditions very strongly over land on medium and
small scales.

9.11 Airflow over uneven surfaces

A number of different processes give rise to patterns of atmospheric behaviour

near the surface which have horizontal scales ranging from - 10 m to as much as
100 km. There are mobile patterns which are associated with shower clouds and
systems of shower clouds which will be dealt with later (section 10.7), and there are
patterns which are locked onto terrain features even though they may be quite
short-lived. Nearly all locked patterns occur over or near land rather than water
surfaces, although land-locked water often has its own distinctive effects. Consider
a few examples of land-based effects.

SLOPE WINDS

When a sun-facing slope is warmed, a proportion of the buoyant forces generated

can cooperate on the scale of the sloping surface itself to produce an overall up-
slope (anabatic) wind. This may arise in conditions which are otherwise calm, or it
may act to modify a larger-scale flow associated with prevailing weather systems.
Anabatic winds are often quite gentle (- 2m s- 1) because the overall buoyant
force is largely offset by the turbulent friction generated by the winds and by the
convection which typically penetrates the anabatic layer. However, in extreme
terrain much stronger winds can be generated.
By the same mechanism in reverse, nocturnal cooling on a sloping surface can
generate down-slope (katabatic) winds which tend to be rather stronger and
shallower than anabatic winds because of weaker turbulent friction with the over-
lying air (Fig. 9.23) and with the underlying surface. Violent katabatic winds can
sweep down snow-covered coastal slopes, triggered no doubt by some shift in the

100 Fig. 9.23 Vertical profile of

wind speed in gentle
katabatic flow down a 40°
80 slope on the Nordkette. near
lnnsbruck. (After Defant in
[60})
60
E
;;:; 40
"'
·a;
I

20

0 2 3

Wind speed lm s11 315

9 Surface and boundary
layer

Fig. 9.24 Arctic sea smoke

near the Ross Ice Shelf,
Antarctica. The shelf edge in
the background is about 25 m
high.

larger-scale air flow, and threaten small boats coasting round Greenland and
Iceland for example; and the climatology of the coastal fringe of Antarctica is
dominated by the katabatic gales which sweep off the central plateau in all but the
summer season. The diaries of pioneer explorers of Antarctica are full of descrip-
tions of the resulting blizzards, one of which cruelly trapped Scott and his south
polar party only ten miles from the relative safety of a large supply dump.
The gentler katabatic flows of small-scale hilly country can produce ponding of
cool air in hollows so that by dawn there may be quite sharp temperature contrasts
between the lower and upper parts of the slopes (kept relatively warm by the con-
tinual shedding of chilled air during the course of the night). Relative humidities
are highest in the coldest air, even if there is no stream or lake to contribute more by
evaporation, and fog will therefore form first in the hollows and valleys. If there is
a lake in the hollow, the ponding cold air may well become considerably colder
than the water surface (because of the latter's much larger heat capacity- section
9 .4), giving rise to steam fog as the warm surface air convects into the cooler air
ponding just above it and is chilled to sporadic saturation by the resulting mixing.
Katabatic flows onto the sea from glaciers and ice sheets in high latitudes have a
similar effect, but the greater temperature contrast makes the resulting arctic sea
smoke (Fig. 9.24) much deeper (- 30m) and more visually obvious than most
steam fog.
In more extreme terrain and in the presence of very strong nocturnal or even
seasonal cooling, the ponding of cold air can lead to frost hollows whose minimum
temperatures may be tens of degrees below those of the surrounding area. On a
larger scale, some of the lowest screen temperatures on Earth have been recorded
in winter in the Siberian town of Verkhoyansk (- 51.9 oc monthly mean in
December 1904), which lies in a valley just inside the Arctic circle surrounded by a
horseshoe of mountains rising extensively above the 1000 m level.

SEA BREEZES

When coastal land is warmed by the morning sun, it can quickly become con-
316 siderably warmer than the adjacent sea, especially if there is no strong overall wind
 9.11 Airflow over uneven
surfaces

Fig. 9.25 The nose of a

sharp fast-moving sea breeze
front 45 km inland from the
south coast of England, with
isopleths of specific humidity
in g kg_,_ The dry air aloft is
subsiding gently in an
anticyclone. (Redrawn from
[61])

to minimize the contrast by rapid advection. In addition to the normal convection

over land, there is in such conditions a tendency for the air there to rise relative to
the adjacent cooler and denser air over the sea, in a weak but persistent large-scale
buoyancy effect. In fact the uplift may occur by an enhancement of the convective
updraughts compared with downdraughts over the coastal margins, encouraging
enhanced development of cumulus and even cumulonimbus there. The rising air is
replaced by the cool, moist air which is drawn inland over the coast as a sea breeze.
The flow is seldom more than a few hundred metres deep, above which there is
often a more gentle and diffuse seaward flow. The boundary between the sea and
land air as the former flows inland can be very sharp, with marked contrasts in
temperature, humidity and haziness being ascribed to the presence of a sea breeze
front (Fig. 9.25). Sometimes there is a definite line of enhanced convection along
the front which is welcomed by glider pilots, and lifts insects and the birds which
feed on them.
In middle latitudes the most favourable conditions for well-marked sea breezes
seem to occur in summer when there is a light synoptic-scale air flow along the coast
with the sea on the left hand (in the northern hemisphere). The sea-breeze front
may move tens of kilometres inland in such conditions before the waning of the sun
in the late afternoon begins to remove the driving agency of unequal heating. Even
so, a well-developed sea-breeze front may continue rolling inland under the
influence of its established density contrast until late evening. When a sea breeze
persists for more than a couple of hours there is usually a rotation of the breeze
vector in the direction to be expected from the operation of the Coriolis force,
which produces a veering in the northern hemisphere.
Even though well-marked sea-breeze fronts may occur on only a few days per
year in any particular coastal region, the much more frequent sea-breeze tendency
may have a very considerable effect on local climate, lowering temperature and
reducing cloud and showery rainfall along the coastal margins as compared with
more than a few kilometres inland. Indeed the resulting enhanced freshness and
brightness of coastal climates has contributed greatly to the rise of the seaside
resort in Britain and similar countries.
At night the sea-breeze effect tends to reverse, with the production of cooler,
denser air over land leading to a seaward flow (a land breeze) which is known and
used by coastal sailing craft.

HILL AND MOUNTAIN WAVES

When an airflow impinges on a hill or mountain, air is diverted horizontally and

vertically by the obstruction. In the typical slight vertical stability (convective 317
9 Surface and boundary
layer

Fig. 9.26 Strong air flow

over a ridge which is lang
across the flow. Highest and
lowest wind speeds occur in Horizontal
positions H and L
respectively. Precise positions
vary with ridge shape and
profiles of wind speed and
stability) of the lower troposphere, the forcible elevation of air passing over a hill,
temperature to considerable and of many higher layers of air, sets up vertical oscillations of the air about its
heights. undisturbed level, which continue downwind of the obstruction, often for sub-
stantial distances. The consequences for the free atmosphere are mentioned again
in section 10.8, but at and near the surface the most important result is that wind
speeds can be greatly enhanced or reduced at several zones on and downwind of a
hill. A typical situation is depicted in Fig. 9.26, in which relatively high winds are
induced by congestion of air flow near the crest of the hill, and again several kilo-
metres downwind, with lower wind speeds where air flow is reduced by stagnation
(position 1 on Fig. 9.26) or separation (position 2). In very windy weather the
enhanced winds can become damaging: in a famous westerly gale over northern
England in 1962, enormous damage was done to forests and to the city of Sheffield,
in which over two thirds of all buildings were officially registered as having been
damaged. All this damage occurred in regions in the lee of the Pennines where
subsequent detailed analysis of the airflow in the gale indicated that wind speeds
were enhanced by being at the base of a lee wave [62].
In some cases the influence of topography on airflow is just as might be
expected: a valley which is nearly parallel to the prevailing wind has a channelling
effect which raises wind speeds above values found in otherwise identical valleys
lying across the prevailing wind. (Particular parts of a crosswind valley may of
course be especially windy because of the presence of lee waves, but not the whole
valley.)

CLOUD AND PRECIPITATION

Consider the flow of moist air over a hill or mountain. The enforced rise will be
rapid (the updraught is the product of the surface slope and the horizontal wind
speed) and therefore nearly adiabatic. Unsaturated air therefore tends to rise, pre-
serving its lifting condensation level, and produce cloud at this level if the total rise
is large enough.
In the particular case of a layer which is well stirred by mechanical or thermal
convection from the surface up to cloud base and beyond, the level of cloud base
over the surrounding low ground should be maintained over the hill, though there
may well be more and thicker cloud cover over the hill in response to larger and
stronger areas of uplift there (Fig. 9.27a). This is easily observed to be the case in
hilly country, where the nearly uniform level of nearby and more distant cloud
bases is particularly obvious when the observer has climbed close to cloud-base
level, and is a nice confirmation that the combination of turbulent mixing and
adiabatic eddy motion produces the same effect as wholesale adiabatic elevation.
In the presence of prolonged rainfall, however, it is normal to see low cloud
blanketing the upwind slopes of hills at levels far below the base of the nimbo-
stratus producing the general rain (Fig. 9.27b). It seems that air in the lower parts
of what is the subcloud layer over the low ground is significantly moistened by
318 evaporation from the wet surface and the falling precipitation, and has in conse-
 lal 9.11 Airflow over uneven
surfaces

Fig. 9.27 Cloud

(b) enhancement over hills. (a)
Cumulus enhanced by
buoyant or forced updraught,
but cloud base level is
uniform throughout the well-
mixed subcloud layer.
(b) Thicker nimbostratus over
the hill, with a collar of low
cloud (hill fag) where
dampened surface air is
forced to rise.

quence a lower lifting condensation level which however does not become apparent
until the air is forced to rise over the hill (though it may contribute to scraps of scud
over the lower ground- section 10.2).
At higher levels, moist air may be raised to condensation giving humps of cloud
outlining the distorted airflow, and in the middle and upper troposphere these may
be particularly smooth and lens-like (lenticular- see section 10.8). If the moist
layer is already cloudy, the cloud layer will be deepened by the enforced rise, as
shown in the case of a shallow layer in Fig. 9.27b; and if there is a deep layer of
nimbostratus, then it is deepened and its rate of precipitation enhanced over and
slightly upwind of the hill in ways examined in section 10.8. There is often a corres-
ponding reduction in cloud on the downwind side of the hill, and a reduction in
precipitation which may produce a climatological rain shadow if the orography is
extensive enough and the paths of prevailing weather systems well enough aligned.
A pronounced zone of sinking flow on the steep lee of a hill can suppress low cloud
which otherwise blankets the sky.
Note that the three examples of patterns mentioned above are patterns of flow.
Clearly the patterns of radiant flux discussed in section 9.2 can have effects on
small-scale climate which are equally important or even more so. Flow and
radiation can interlock when the undulations of air in the low troposphere produce
patterns of cloud which persist for hours or longer, shrouding some areas under
cloud in the crests of waves over or downwind of hills, while leaving other areas in
bright sun beaming through holes in the cloud cover maintained in the unsaturated
troughs of the undulating airflow. The presence of cloud holes is most marked in
the case of otherwise complete overcasts of low cloud such as stratocumulus or
stratus. In idle boyhood moments the author noticed that his part of the city of
Belfast was often bathed in sun amidst surrounding gloom as low cloud fell slowly
down the steep lee slopes of hills to the west, dissolving and not reforming until the
air rose suddenly and permanently above the condensation level again several kilo-
metres downwind. Such local effects are commonplace in non-uniform terrain,
and their influence on local deviations from overall weather patterns is often con-
siderable, as necessarily observant locals such as farmers and coastal fishermen are
usually aware. 319
9 Surface and boundary The best known of all such local effects are the john winds of the European Alps,
layer and the corresponding chinook winds on the eastern flanks of the Rocky
Mountains of North America. They appear suddenly as unusually warm, dry and
(a) strong winds blowing down the lee slopes of the respective mountain massifs and
continuing for as much as 100 km on to leeward. Once established they can
continue for days, until a change in the large-scale pattern of weather and flow cuts
them off. In winter the sudden warming can melt snow so rapidly that there is local
flooding. In summer the desiccating winds can dry the surface and vegetation so
much that there is serious risk of bush or forest fires. In the Italian Lake District in
(b)
summer, the troublesome haze of the Po valley is swept away in a trice, replaced by
unusually clear air, which persists throughout the fohn. Temperature rises range
from a few oc to over 20 oc in some almost legendary chinooks. Relative
humidities usually fall below 5007o and have gone below 10076 on occasion. Winds
often exceed gale force, and can be magnified by local topography to damaging
strengths.
Fig. 9.28 Contrasting air Fohn and chinook events are conventionally and very commonly explained in
flows associated with fohn
conditions in the lee of
terms of forced ascent and descent of air flowing over a mountain range which is
mountains. In (a) there is reasonably high, and extensive both along and across the impinging airflow - an
saturated ascent upwind, explanation first advanced by Hann in the infancy of meteorology (1866). The
whereas in (b) air slides over dryness and warmth of the descending flow is explained by invoking cloud and
dammed air upwind of the precipitation on the upwind side, and dry descent on the downwind side
mountain. In each case the
descending air tends to warm
(Fig. 9.28a). In the simplest picture this would impose saturated adiabatic ascent
dry-adiabatically. and dry adiabatic descent on the air stream crossing the mountains, which could
produce a resultant temperature rise of about 4 oc for every kilometre of inter-
vening mountain height. Detailed case study confirms that this certainly happens
on some occasions. Indeed the famous clarity of the notherly fohn in the Italian
Lake District shows that much industrial and natural haze has been washed out of
the air during cloudy ascent of the northern slopes of the Alps. However, there are
other occasions when a fohn or chinook occurs when there is no precipitation on
the windward side of the mountains. It seems that on such occasions the impinging
airflow is largely dammed below the top of the mountain barrier, leaving the air
above to slide over the top and descend on the lee side (Fig. 9.28b). The tempera-
ture rise on the lee side is then a measure of the stability (the increase in potential
temperature with height) of the air to windward. Neither of these models accounts
specifically for the considerable winds which are common in fohn and chinook,
and for these we must look to the dynamics of lee flow as mentioned earlier in this
section and again in section 10.8. Active research is continuing on these and other
aspects of these interesting and often quite dramatic examples of the interaction of
topography and meteorology [60].

9.12 Surface microclimate

VEGETATION

Conditions near the ground surface are strongly influenced by the presence of
vegetation.
Short grass establishes the value of the aerodynamic roughness length z0 and
320 therefore scales the log wind profile in the overlying surface boundary layer. The
presence of reasonably well-watered grass also ensures that the latent heat flux 9. 12 Surface microclimate
from the surface is relatively large (i.e. that the Bowen ratio is relatively small) and
that the daytime temperature rise is relatively modest. The binding of the soil by
grass roots prevents the production of dust storms, which are such a feature of arid
lands where they temporarily but drastically transform conditions by decimating
solar input. It seems possible that before the colonization of Earth's land areas by
plants (a relatively recent event on the timescale of Earth history), dust storms were
a major climatic feature, possibly even to the extent that they are still on the planet
Mars, where hemisphere dust storms shroud the surface for weeks on end.
Long grass and dense shrubbery interfere with fluxes of solar and terrestrial
radiation to such an extent that the radiatively effective surface is raised above the
ground surface and spread over a significant height range, the spread being greater
for more open vegetation. This means for example that it is the upper parts of the
vegetation which are warmed during the day and cooled at night, and that these
gains and losses of heat are only subsequently communicated downwards by
turbulence, radiation and even by the falling of rainwater from wetted leaves. The
upward communication to the overlying atmosphere is much the same as for an
unvegetated surface after allowing for differences in the effective heat capacity of
the active vegetated surface and the typical presence of a substantial flux of water
vapour with its associated evaporative cooling. The effect on daytime and night-
time temperature profiles in a stand of vegetation is outlined in Fig. 9.29 for the
case of a cereal crop, but effects in other types of vegetation are similar in principle.
The basic result of the elevation of the radiatively active layer above the ground 10 15 20
surface is to make the microclimate of the layer of the region between the active Temperature
layer and the ground much less extreme, elevating nocturnal temperature minima
Fig. 9.29 Idealized daytime
and depressing daytime maxima to produce a much smaller diurnal variation there. and night-time temperature
This moderate climate then encourages colonization by plants and animals which profiles in a barley crop.
could not tolerate the more extreme conditions prevailing before vegetation. (After [51])
The greatest effects of this type occur in forests, where the radiatively active
layer may lie 10 m above the ground in temperate forests and several times this
height in the great tropical rain forests which lie mainly in equatorial regions and
are watered by precipitation from the intertropical convergence zone. The micro-
climate of the surface region beneath the dense, lofty canopies of tropical rain
forests is familiar from the many films of wildlife there: the humid gloom, with its
near absence of wind and diurnal temperature variation, contrasts sharply with
conditions at and above the lofty canopy, and is inhabited by a whole range of
living things which are irreversibly eliminated when the trees are felled on any
substantial scale. Even the ground surface is then unable to cope with the change of
conditions imposed by sudden exposure, being alternately baked by the sun and
scoured by the deluges of rain from great equatorial cumulonimbus. Bearing in
mind the considerable role played by this lush vegetation in the oxygen and carbon
cycles of the Earth's atmosphere and surface (sections 3.2 and 3.4), the wholesale
destruction of large amounts of tropical rain forest in the present century seems
likely to be remembered as one of the classic examples of unenlightened self-
interest.

FOG

Fog forms when air close to a surface becomes slightly supersaturated and
produces a layer of cloud in contact with the surface. The steam fog already
mentioned is the only type of fog which is formed over a relatively warm surface;
all other true fogs are formed over relatively cool surfaces. (Hill jog is not a true 321
9 Surface and boundary fog, being simply cloud which is low enough to envelope the upper parts of hills or
layer
mountains).
Radiation fog forms in shallow layers, usually only a few metres deep, when
radiative cooling of a ground surface cools the overlying air below its dew point. It
often appears shortly after dusk on a relatively windless evening, under clear or
nearly clear skies, over land which is fairly moist. In these conditions the air in the
first few metres above the surface is almost still, and the fog when it forms outlines
what is virtually a huge laminar boundary layer (Fig. 9.30). Local inhomogeneities
may stratify the humidity and give rise to horizontal leaves or fingers of denser fog
which enhance the laminar appearance. Over uneven surfaces radiation fog tends
to form in hollows, as they fill with dense, moist air from neighbouring slopes.
Once formed, the fog layers tend to persist by direct cooling of their top surfaces by
net long-wave radiation, and thicken gradually, but unless they grow to a sub-
stantial depth (10 m or more) they are easily removed by an increase of wind, or by
solar heating next day. Fog formation shields the underlying surface somewhat
from further radiative cooling, though very gentle convection begins to spread the
top surface cooling throughout the thicker fog layers.
Advection fog occurs when relatively warm air is chilled to saturation by over-
running a sufficiently cool surface. If the air is close to saturation to start with, then
only a little further cooling is necessary to form a fog layer which may be tens of
metres thick. Pulses of relatively warm, moist air often move over land in autumn,
winter and spring in mid-latitudes, as part of air movements in weak, slow-moving
synoptic-scale weather systems. In autumn and winter particularly, the land is
usually much cooler than the sea, so that the invading air is cooled quite sharply.
Even if this is not sufficient, radiative cooling under clear night skies often
completes the production of saturation and fog. Advection fog over land is often
produced jointly by advection and radiation in this way. Fog layers produced
wholly or in part by advection are usually substantially thicker than those produced
purely by radiation, and this has important consequences.
The top surface of a substantial fog layer replaces the shrouded underlying
surface as the new radiatively effective surface. This does not affect the output of

Fig. 9.30 Radiation fog

about 2 m deep forming in a
meadow about dusk.

322
terrestrial radiation much, since thick fog behaves like a blackbody in much the 9 .12 Surface microclimate
same way as any terrestrial surface. However, it transforms the response to any
available solar radiation, since fog behaves like any cloud in back-scattering
strongly and absorbing very weakly. On a clear night therefore the top of the fog
layer loses heat rapidly by net emission of terrestrial radiation, but this cooling is
spread throughout the fog layer by the gentle convection which it encourages, so
that the sharply concentrated surface cooling typical of an exposed land surface is
spread out vertically, and somewhat reduced by the addition of the heat capacity of
the foggy air (a 20 m layer of air having much the same heat capacity as a centi-
metre layer of soil). The sharp temperature inversion which would be observed
over an exposed land surface is therefore lifted to the fog top and weakened sig-
nificantly. In late autumn and winter, the weak sun when it rises may be incapable
of removing the fog layer by warming, the direct warming being minimal and the
indirect warming of the underlying surface being very small if the fog layer is deep
enough. In fact all the daylight may do is reveal the gently convecting fog layer to
an overflying observer (Fig. 9.31). Once a substantial fog layer is established in
these conditions it tends to maintain itself and the cool layer of air in which it
persists, until the synoptic situation allows increasing wind to disperse the fog by
shear-driven turbulent mixing with the overlying air.
In industrial or urban regions with significant numbers of short chimneys, the
smoke and gases emptied into the fog layer (which is usually bounded above by a
weak inversion maintained by radiative cooling of the fog top) can rapidly produce
unpleasant and even dangerous concentrations of pollutants. London has been
prone to such smoky fogs (smogs) for centuries, in fact since the burning of coal in
domestic open fires became widespread in the sixteenth century. The old local
name for them is pea soupers, which nicely describes the greeny-brown obscurity in
which they enveloped buildings and people. The dirtying and corroding effects on
buildings have long been obvious (St Paul's Cathedral was already blackened
before its completion in the early eighteenth century), but the effects on people

Fig. 9.31 Aerial view of a

convecting fog blanket
produced by advection and
radiative coaling. The spire of
Salisbury Cathedral is 123m
high.

323
9 Surface and boundary were largely ignored until it became clear from hospital records that a spectacular
Ioyer
but not untypical smog which lasted for several days in the winter of 1953 had killed
several thousand people made vulnerable by age or chest infection. The ensuing
outcry led to the introduction of controls on the production of smoke by inefficient
combustion of fuels which have transformed the appearance of British cities in
subsequent decades. The numbers of foggy days in cities have also dropped,
showing that fog formation was being positively encouraged by the availability of
condensation nuclei in smoky air.
Advection fog also forms over the sea. The fogs of the Grand Banks fishing
grounds off Newfoundland (graphically described by Rudyard Kipling in the novel
Captains Courageous) are formed by the advection of warm, moist air from the
Gulf Stream over the cold, fish-laden waters of the Labrador Current. Fogs form
in summer in the coastal waters round the British Isles when warm, moist air from
nearby land is cooled as it flows over the cool sea. This sea fog or har plagues the
eastern coasts of Britain, particularly in fine weather, when light easterlies asso-
ciated with an anticyclone maintain large sheets of fog over the North Sea, some of
which drift onto the British coasts. The fog is dispersed by flowing over a few kilo-
metres of sunlit land, but the coastal strip may remain shrouded and chilled for
days while the rest of the country is basking in the sun.
If the overall wind speed in and above a deep fog layer begins to increase, the fog-
top inversion may survive and rise while the fog layer is deepened by stirring. In
these circumstances, the lowest part of the layer may become slightly unsaturated,
leaving a layer of stratus cloud, which in the context is simply lifted fog, overlying a
shallow subcloud layer. Something like this can occur in the poleward extremity of
the warm sector of an extratropical cyclone (section 11.3).

Appendix 9.1 Direct sunlight on projections

Figure 9.3 shows a beam of direct sunlight with solar elevation angle E impinging
on a symmetrical projection with base angle a. From the construction it is apparent
that the beam makes an angle a + E with the plane of the slope. Since the beam
makes an angle E with a horizontal (by definition of solar elevation angle), the ratio
RI of the irradiance of the sloping and horizontal surfaces is given by the sine rule
for oblique irradiance (appendix 8.2)
RI = sin (a + E)
sinE
Note that for any chosen E, this expression confirms that the maximum slope
irradiance occurs when a + E = 90°, ie. when the beam is normal to the slope.
Values of RI are plotted in Fig. 9.4 for three representative slopes and a range of
solar elevations.
To find an expression for the absolute irradiance AI of the slope, we can use eqn
(8. 7) (appendix 8.2) for the horizontal irradiance HI at the base of an atmosphere
with zenith transmissivity r 90 , together with the identity
AI =HI x RI
324 = s sin (a + E) T90cosec£
where Sis the solar constant. For simplicity, values plotted on Fig. 9.4 are for Appendix 9. 2
AilS. All are calculated for the case where zenith transmissivity is 0.8, which
corresponds to a very clear sky (appendix 8.2).

Appendix 9.2 Daily insolation

For many purposes in climatology and botany it is useful to have the daily total
solar input. If the zenith transmissivity is steady throughout the day, or if, more
realistically, several more variably sunny days are averaged, the profile of solar
horizontal irradiance from dawn to dusk lies close to half of a sinusoidal curve, and
the daily total insolation can be found by standard integration. Small fillets at
dawn and dusk are ignored in the process, but the error is small compared with
others, chiefly those arising from variations in zenith transmissivity.
Figure 9.32 shows half a sinusoid fitted between dawn and dusk to represent the
daily profile of horizontal irradiance I . The formal expression for such an ideal/ is

I= Im sin ( ~ )

where Im is the maximum irradiance (at noon), tis time measured from dawn, and
Nis the time interval between dawn and dusk. This can be justified by checking the
I values it gives at dawn, midday and dusk. The daily insolation DI is found by
integrating the expression for I between dawn and dusk.

DI = [ Idt

= lm J>in ( ~ ) dt

2Nim
7r

In the case of a sunny summer day in Britain, Nmay be 6 x 1()4 s (over 16 hours)
and Im 650 W m- 2 • It follows that DI is 24.8 MJ m- 2 • Note that the daylight
average irradiance (DUN) is 2!1r times the noon maximum.

Fig. 9.32 Idealized

symmetrical diurnal time
"cu sequence (dashed) of surface
insolation. The solid line is a
"'
:0 fitted half-sinusoid.
-~

Time of day 325

9 Surface and boundary
layer Appendix 9.3 Sunlight in water

Adapting the method of appendix 8.2 to the penetration of solar irradiance I into
water of increasing depth z, we have an expression for the change di produced as
sunlight passes a little further into the depths:
di=- kdmi
where k is the absorption coefficient (in a less simple model it varies with wave-
length), dm is the mass of water per unit horizontal area in the little range of depth
containing dm, and the minus sign ensures that I reduces with increasing depth.
Replacing dm by p dz, we find a relationship between di and dz which can be inte-
grated in the simple case where k and pare uniform to give an exponential decay of
irradiance with increasing depth.
I = I exp ( - k p z)

where I is the irradiance at the water surface.

According to appendix 4.4, the exponential and decadal scale depths are respec-
tively (k p)- 1 and 2.3026/(k p).

Appendix 9.4 Turbulent momentum transport

According to eqn (9.13), the average vertical downward flux of horizontal

momentum through horizontal unit area is given by
r = - pwV (9.31)
where the overbar represents averaging over a chosen time period (often 10
minutes) and the group of terms under it represents the instantaneous momentum
flux density. Since variations in density are proportionately very much smaller than
those in w and V, we can treat it as a constant and remove it from under the
overbar. The instantaneous horizontal downwind speed can be written as the sum
of the average speed and the instantaneous deviation from it:
V=V'+V
The vertical speed can be analysed in the same way, but w is virtually zero in most
realistic conditions in the surface boundary layer, so that
w = w'
Substituting these in eqn (9.31) we find
r = - p w' (V' + V)
= - p [w' V' + w' V]
Since Vis constant in the averaging period by definition, the second pair of terms in
the bracket includes w', the average of the deviations of w from its average,
which is zero by definition. Hence
r = - pwV' (9.32)
326 dropping the unnecessary prime from the updraught speed.
Appendix 9. 5 The log wind profile Problems

According to eqn (9.20) in the neutrally buoyant surface boundary layer we have
u
V = -j-lnz + B

If we choose two heights with ratio X

z2 = Xz 1
then it follows from the log wind profile that the difference in wind speed is
u
-"lnX (9.33)
k
Since k is a universal constant, and u is constant for a particular profile, we see that
the difference in mean wind speed on any particular profile depends only on the
choice of X. If we choose a series of heights in geometrical progression then it
follows that the wind speeds will be in arithmetical progression. As an example,
consider the difference between wind speeds at 2 m (head height) and 10 m
(standard anemometer height). According to eqn (9.33)

V 10 - V2 = 1.61 Tu
showing that in a gale (17 m s - 1 at 10 m height) with friction velocity u = 2 m s - 1
(a reasonable value), the wind speed at head height (2 m) is only 9 m "s - 1• Notice
that in the same conditions the wind speed at 50 m (near the masthead of a large
sailing ship) is 25m s- 1 (49 knots).

Problems

LEVEL 1

1. Why is the laminar boundary very significant to an elephant despite the gross
discrepancy in their volumes?
2. If the solar elevation is 40°, what can you say about sunlight falling on a slope
of 45° (a) facing towards and (b) facing away from the sun?
3. If a typical incident solar ray makes three successive impacts on the interior of
a stand of vegetation before escaping back to the atmosphere, and each
element of vegetation surface has an albedo of 0.5, what is the gross albedo of
the stand?
4. Why should there be less hoar-frost round the bole of the leafless tree than in
the middle of a field on the same occasion?
7. Which of the following usually contributes most to the very large heat
capacities of lakes and seas: the large specific heat capacity of water; its
turbulent mixing; its transparency to sunlight?
8. While illuminated by the sun, ground passes less sensible heat into the atmo- 327
9 Surface and boundary
Ioyer sphere when it is saturated than when it is dry. Suggest two reasons for this.
9. Why would it not be helpful for synoptic observers to report winds averaged
over 10 s instead of 10 mins?
10. A complex computer model of the nocturnal boundary level at an open site in
winter is able to forecast values for the Richardson number near the surface on
the basis of conditions at the previous dusk. What values would you look for
to give a warning of ground frost?
11. In a pure log wind profile, how do you expect (a) the wind speed and (b) the
wind shear to vary at equal multiples of height?
12. In a simple model of the temperature profile over heated ground, how should
the temperature gradient at a fixed level close to the ground vary with (a) con-
vective heat flux and (b) variance of wind speed?
13. Because of the Ekman spiral, the wind direction at 10 m needs to be altered in a
consistent sense to represent the direction of the gradient wind. What is this
sense in the northern hemisphere?
14. In sloping terrain, why are katabatic winds more common by night than by
day?
15. People living near hills often remark that in overcast conditions the sky is espe-
cially threatening in the direction of the hills. Why should this be so?
16. Why should fog often gather in hollows and valleys, leaving even very low hills
protruding into the clear air?

LEVEL 2

17. A beam of direct sunlight with flux density 1000 W m- 2 falls on several
surfaces. Given that the angle of solar elevation is 35°, find the insolation on
(a) a surface tilted toward the sun at an angle of 30° to the horizontal, (b) a
surface tilted away from the sun at the same angle, and (c) a horizontal
surface.
18. Calculate the direct-beam insolation values for the surfaces and solar elevation
of problem 17 on an occasion when the sky was cloudless but rather hazy,
having a zenith solar transmissivity of 0.65.
19. Given that problems 17 and 18 represent conditions at noon on a day in which
the clarity of the atmosphere remained uniform throughout the full 9 hours of
daylight, find the total insolation on a horizontal surface for the day.
20. Find the angles of solar elevation at noon in midwinter and midsummer at the
Arctic (or Antarctic) circle, and at latitudes 55° and 25°.
21. Using a finite-difference form of the Richardson number, estimate the
difference in wind speed across the layer between 2 and 3 m above the surface,
which would be just sufficient to initiate turbulence despite a temperature
inversion of 2 oc across the same layer.
22. In a very simple model of wind variations in the surface boundary layer,
horizontal wind speeds V are 3 m s- 1 when there is a downdraught (of
1 m s -'), and they fall to 1 m s _, when there is an updraught w (again of
1 m s -'). The up- and downdraughts alternate regularly and account for all
the time in the observation period. Sketch Vand won a common graph against
time, and calculate their covariance and hence the friction velocity and average
downward eddy flux of momentum (assuming air density to be 1.2 kg m - 3).
23. The average wind speeds at 8, 4 and 2m height on a certain occasion are 7, 6
and 5 m s -'. Find the implied roughness length of the underlying surface, the
328 wind speed at 10m, and the friction velocity.
 Problems
24. In a certain fohn event, the air descending the lee slopes is observed to be up to
5 oc warmer than air at the same level on the upwind slopes. Find the loss of
vapour content (specific humidity) implied by an adiabatic model of the ascent
and descent (Hint: consider eqn (5.21)).
25. In problem 24, given that the air at the base of the upwind slope was as speci-
fied in problem 5.15, find the implied atmosphere pressure at the crest of the
ridge, and convert this to a ridge height above sea level (1010 mbar) in kilo-
metres using the rule of thumb that pressure lapses by 1 mbar per 10m.

LEVEL 3

26. Discuss the factors which contribute to the large seasonal range of climate
observed in a continental site at latitude 40° in the rain shadow of a range of
hills.
27. A clear calm night has established a marked temperature inversion near the
surface. Now the wind begins to increase in response to an approaching
synoptic-scale weather system. Describe what you would observe with simul-
taneous measurements by fast-response thermometer and vertical-propellor
anemometer before and after turbulence breaks out. What will happen to the
average air temperature at the same time?
28. Estimate the solar, terrestrial and net radiative balances for midsummer's day
on the Arctic circle, when the surface solar albedo is 0.2, the zenith trans-
missivity for solar radiation is 0. 7, the surface temperature remains constant at
10 °C, and the surface long-wave emissivity is 0.95.
29. Compare and contrast the climatic environment of a small crawling insect, an
Irish wolfhound and a man, each living in a mid-latitude continental prairie.
Compare the experiences of the man and dog, given that they have the same
body weight.
30. Find the correction factor to be applied to wind speeds measured at 2 m to
bring them to the equivalent value at 10 m, assuming a log wind profile and an
aerodynamic roughness of 1 em over the surrounding area. Note that this is a
useful procedure to estimate synoptic standard wind speeds when there is no
anemometer at 10m. Consider the likely sources of error in such estimates.
31. The relationship between potential temperature lapse and sensible heat flux
density H in eqn (9.27) rather unhelpfully contains the friction velocity u .
However, u can be replaced using the log wind profile. Do this and derive the
useful relati"onship
H = pCP C0 6.0 b. V
where C0 = [k/ln (z2 /z 1)]2,which is the drag coefficient C0 in eqn (9.24).
Derive a similar expression for the latent heat flux, and use the two to show
that the expression for the Bowen ratio is
(3 = pCP ~
L b.pv
which is the most widely used expression for the Bowen ratio.
32. Give an account of ways in which local meteorological conditions in hilly
terrain can be influenced by the terrain.

329
CHAPTER
Smaller-scale weather
10 systems

10.1 Introduction

We have seen how the structure of the atmosphere both controls and is maintained
by atmospheric behaviour, and that such interaction is especially vigorous in the
troposphere. Much of this behaviour appears to us to be a series of events related to
types of surface weather: cumulus clouds blossom briefly and fade, showering the
surface with precipitation if they are big enough; depressions develop and trail
their attendant fronts across thousands of kilometers of middle latitudes, while
neighbouring areas are covered by gently moving and developing anticyclones;
tropical cyclones form over warm oceans and drive westward, scourging the
surface with tremendous winds and rains before turning polewards and subsiding
to more normal intensities. Since such weather-related events have quite charac-
teristic patterns of behaviour and structure, we will regard them as systems, to be
examined descriptively, much as living organisms are described and catalogued,
and also mechanistically, to see if we can identify their functional parts and how
they operate in terms of the basic physical and meteorological processes which we
believe underlie all such behaviour.
There is a wide range of tropospheric weather systems, differing in scale and
type. In this chapter we will concentrate on those which are not large enough to
appear unambiguously on a weather map, although they may individually and
quite strongly influence conditions at one or two adjacent stations on any
particular map. Such systems are called sub-synoptic in scale since they do not
appear individually on synoptic-scale maps, although populations of individual
systems may well dominate large areas of any particular map. The shower cloud
(cumulonimbus) is a typical example. The sub-synoptic range is so wide that it is
often subdivided into micro-scale, small-scale and meso-scale. Micro-scale refers
typically to systems no larger than large individual eddies in the atmospheric
boundary layer, say - 100 m. These affect surface conditions mainly through the
turbulent variability described in the previous chapter and will not be discussed
further. Small-scale refers usually to systems no larger than a single large cumulo-
nimbus, say 10 km in both horizontal and vertical dimensions (though the anvil of
a mature cumulonimbus may stretch up to 100 km downwind of the main body in
the presence of strong wind shear), and has been used in this way in Chapter 7.
330 Meso-scale usually refers to systems or patterns of systems intermediate in scale
between the small and synoptic scales, which means between - 10 km and - 10.2 Cumulus
100 km in the horizontal according to my summary. However, none of these sub-
ranges of scale is well defined in meteorological usage, and the meso scale in
particular is used especially loosely.
The following sections describe and simply analyse some of the most important
types of small and meso-scale tropospheric weather system.

10.2 Cumulus

One of the most common types of cumulus is the small cumulus which often begins
to appear in previously clear skies over land in mid-morning, after the stable layer
left by the previous night's cooling has been eliminated by solar heating (section
5.13). This is often called fair-weather cumulus because it is typical of fine summer
weather, though it is a misleading title on those quite common occasions when the
clouds subsequently develop enough to give a showery afternoon. Some features of
the appearance and behaviour of cumulus humilis have been mentioned already in
sections 6.4 and 6.9 in discussions about the sub-cloud layer and about droplet-
scale mechanisms. In the present section we will summarize and discuss these and
other aspects of cumulus, mainly as they are apparent to the unaided ground-based
observer.
The individual cumulus has a very characteristic shape, and is usually just a few
tens of metres across when it first becomes clearly established (Figs 6.3, 6.11 and
I 0.1 ). The nearly flat base corresponds reasonably to the lifting condensation level
of the well-mixed sub-cloud layer when observations are available (section 5.11),
and the quite obvious consistency in base level across an extensive field of such
cumulus (Fig. 10.1) supports this identification. As mentioned previously (section

Fig. 10.1 Cumulus with

characteristic flat bases and
humped upper parts.

331
10 Smaller-scale weather
systems

Fig. 10.2 A small cumulus

seen at close range, showing
its ragged base tilted slightly
upwards in the direction of
drift.

6.5) the cloud base is often a little higher than the lifting condensation level
deduced from screen level measurements, but the discrepancy is so consistent on
any,·particular occasion that it seems likely to arise from a too-simple model of the
lowe~t part of the well-mixed layer, rather than from inaccurate measurement.
Closet inspection (Fig. 10.2) usually reveals a raggedness of cloud base on a scale
of metres, and a slight but consistent upward tilt in the ambient downwind
direction. The first clearly reflects the turbulent inhomogeneity of the air in and
below the cloud, but the second seems to indicate a mechanism on the scale of the
cloud itself which is still unknown.
The sides and top of the individual cumulus are characteristically domed and
knobbly. The domed shape suggested the generic name 'cumulus' (Latin for hill) to
the pioneers who were naming cloud types in the early nineteenth century, in the
wake of the latinate Linnaean classification of living organisms, and is clearly con-
sistent with the rising of an extensive, rounded parcel of buoyant air above its
lifting condensation level. The knobbly appearance, which often closely resembles
the heart of a ripe cauliflower, again reflects the presence of turbulence on scales
much smaller than the cloud itself.
If an individual cumulus is observed for even a little longer than the usual casual
glance, it becomes apparent that it is very dynamic: the dome swells visibly in the
space of a minute, and the knobs protrude and retract even more rapidly - both
clearly consistent with the operation of mechanisms already mentioned.
The rate of rise of the top of the cloud is somewhat less than the maximum
updraught speed inside the cloud, because the rising mass of air continually tends
to turn itself inside out like a rising smoke ring (technically a ring vortex).
Updraught speeds seldom exceed 2 m s - t, because the weak buoyancy is almost
exactly offset by drag. Drag arises obviously as the rising air forces its way through
its surroundings, and less obviously but equally importantly because of the
entrainment (section 7 .12) of previously static ambient air.
The seething detail of the knobs in particular is suggestive of intense mixing
between the swelling cloud and its cloudless environment, and careful observation
confirms that the net effect of such mixing is to extend the cloudy volume quite
332 rapidly by entrainment and saturation of previously cloud-free air. This entrain-
ment is in addition to the tendency of the rising air simply to displace the sur- 10.2 Cumulus
rounding air, and is a crucial feature of the whole cumulus family (including
cumulonimbus). The sharpness of the boundary between cloudy and cloud-free air
shows that cloud droplets form and disappear so rapidly that any individual micro-
scale eddy is either densely clouded or quite clear depending on whether it is
saturated or not. The implied high speed of response (of activation and deactiva-
tion- see section 6.9) is made possible by the very small size of the droplets: -
10 ~m.
A particular cumulus usually continues growing horizontally and vertically for
five minutes or so, behaving as described and drifting in the ambient wind. It may
then be over 100m broad and somewhat less deep, and may have drifted several
kilometres downwind of its point of origin. After another few minutes it will often
seem to lose vitality (notice how the correspondences with living organisms keep
appearing), stop growing, lose sharpness of outline, and begin a process of dis-
solution which is completed in only a few more minutes, leaving the air as clear as it
was initially. During dissolution the cloud usually becomes less obviously white,
and its edges less distinct, which is consistent with selective evaporation of the large
numbers of smaller cloud droplets as the marginally saturated cloud continues to
entrain unsaturated air, despite its waning vigour. The largest droplets last longest,
so that toward the end of the cloud's recognizable existence it becomes a dull (the
few larger droplets scatter sunlight relatively poorly), diffuse mass, heading for
extinction by a process which could be called terminal detrainment.
Two factors may affect the shape of the cloud quite considerably at maturity. If
the effective lifting condensation level lies only a few tens of metres below the base
of a convectively stable layer (a temperature inversion is an extreme example), then
the cumulus will be vertically stunted by the loss of buoyancy in air rising into the
base of the stable layer, and will show a pancake shape throughout its life. This is
typical of the shape of cumulus forming beneath the base of the subsidence inver-
sion of an anticyclone (see section 11.5). Indeed the continual formation of such
flattened cumulus in anticyclones often maintains large areas of stratocumulus, in
which the individual clouds are so closely packed that they form a sheet patterned
on a scale of- 100m by narrow gaps between adjacent clouds, and the underlying

Fig. 10.3 Stratocumulus,

forming nearly total overcast.

333
10 Smaller-scale weather surface is very significantly shaded from the sun in consequence (Fig. 10.3). In
systems
heavily industrial regions the trapping of smoke and other pollution under the
same stable layer can give rise to quite gloomy and noxious conditions at the
surface.
The other factor affecting cumulus shape is the presence of wind shear (i.e.
vertical shear of horizontal wind) in the layer containing the cumulus. Such shear is
never as marked as in the surface boundary layer, but it can nevertheless be strong
enough to make the cumulus lean quite noticeably. Though the shear at cloud level
need not parallel the wind there, which in turn need not parallel the surface wind
and shear, in fact these shears are often nearly parallel, with the result that cumulus
are seen to lean nearly enough down the direction of the surface wind and shear. In
very windy conditions the shear may be so strong as to inhibit the orderly develop-
ment of cumulus, producing a ragged, heavily sheared cloud known as scud or
fractocumu/us when it is accompanied by rainy cloud at higher levels, as is often
the case in strong winds (Fig. 10.4 and section 9.11).
In the trade-wind zones there is a large population of cumulus, much of it taller
than small cumulus because of the rather different stratifications of temperature
and humidity which prevail there, which tend to lean backwards against the surface
winds (Fig. 8.13) in what is apparently the wrong way, at least by comparison with
observation at higher latitudes. The reason for this is that the trade-wind flow is
quite shallow, extending no more than a few kilometres above the oceans, where
their reliability was once so widely exploited by sailing ships. For example, in the
northern hemisphere the characteristic north-easterly flow fades rapidly with
height and is replaced by a westerly flow in the middle of the low troposphere,
which tends to persist and increase with height (Fig. 10.5). The trade-wind cumulus
are therefore rising through a layer in which wind shear and near-surface winds are
nearly opposed in direction, and their upper parts are therefore sheared backwards
in relation to their bases, unlike the case of such clouds at higher latitudes, where
shear and flow usually have much the same direction.

Fig. 10.4 Fractocumulus

(scud) silhouetted against an
advancing clearance of
overhanging frontal
nimbostratus.

334
 10.3 Cumulus development
12

10

2
ITrade
~w i nds
Fig. 10.5 Vertical profile of
zonal wind component in and
E' ly
above winter trade winds. In
20 10 0 10 20 summer the upper westerlies
may be weak or replaced by
Zonal wind speed (m s" 11 easterlies. (After (27])

10.3 Cumulus development

Unless there is a convectively stable layer aloft, some small cumulus seem to avoid
the dissolution described in the previous section, and carry on growing until they
become elevated hills and mountains of cloud (Fig. 10.6), likely to produce
precipitation and therefore by definition become cumulonimbus. Let us consider
the process of development of such cumulus congestus (swelling cumulus) from
small cumulus, and the factors which encourage it on some occasions.
The updraughts in small cumulus, and in the cloud-free thermals which feed
them from beneath, are rather weak, seldom exceeding 2m s- 1 • As mentioned in
sections 7.12 and 10.2, this is consistent with a condition of near balance between
buoyancy and the combined effects of form and entrainment drag. The buoyancy
is weak because the temperature excess in a thermal is usually less than 1 °C, and is
partly offset by the weight of cloud water borne by the updraught above cloud

Fig. 10.6 Cumulus

congestus, showing typical
mountainous shape.

335
10 Smaller-scale weather base, and the temperature excess is small because of the weakness of solar warming
systems
of the terrestrial surface, and because of the readiness of atmospheric turbulence to
dissipate temporary inhomogeneities. In these conditions the loss of buoyancy
associated with the evaporative cooling which occurs all over the upper surfaces of
a cumulus, and continues throughout its life, is potentially very important
(appendix 10.1). Indeed, its tendency to make the flanks of a cloud sink rather than
rise is often quite clearly visible on larger clouds. Cloudy updraughts which can
avoid or delay the onset of such destruction of their buoyancy have a considerable
advantage over their fellows.
Suppose that conditions are especially favourable for the formation and persis-
tence of cumulus: because the temperature stratification aloft is neutral or even
slightly unstable (see section 5.12), the ambient air above cloud-base level is suffi-
ciently moist to minimize evaporative cooling, and there is an adequate supply of
buoyant moist air rising from the surface layers. The instability can be induced by
large-scale factors such as the warming of surface layers by the sun or by advection
over a warmer surface, or by the cooling of upper layers by the differential
advection which is such a feature of flow in extratropical cyclones (section 11.4).
Indeed in the latter there are effective mechanisms involving large-scale vertical
motion too. Gentle but persistent wholesale uplift can reduce stability by raising a
layer whose lower parts are sufficiently more humid to saturate before the upper
parts, as exemplified in section 10.6. Another effective mechanism is the differ-
ential uplift associated with vertical stretching and synoptic-scale horizontal con-
vergence (section 7.13) whose effect can be seen by raising and stretching a layer
with a sub-adiabatic lapse rate on a tephigram.
In such conditions many small cumulus form, and some are reinvigorated by
new thermals rising into their bases before their initial impetus is lost. These new
thermals are able to rise through the pre-existing cumulus without evaporative loss
of buoyancy until they emerge from the protective body of the cloud, and
obviously tend to reach this stage with greater buoyancy than the pioneering
thermals which first produced the cloud in the teeth of evaporative destruction of
buoyancy. Cumulus receiving such additional buoyancy must therefore be con-
siderably advantaged over their contemporaries, who have to make do with a single
initial input, and tend to grow larger and live longer as a result. Indeed they are
likely to receive repeated additions of buoyancy from beneath cloud base by the
same mechanism, each new influx being protected from destructive evaporation by
rising through successively larger depths of accumulated cloud before temporarily

i1.
taking the brunt of pushing the cumulus congest us further up into the unsaturated
~ environment. The end result is that a fraction of the total number of cumulus-type
clouds grows much larger than their fellows in these conditions, often extending up
into the middle troposphere (- 5 km above sea level).
The size of the fraction becoming cumulus congestus in any particular situation
' 1 base
varies considerably with the temperature and humidity stratification of the lower
Fig. 10.7 Outlines from a and middle troposphere, as does the average and maximum heights reached by the
sequence of photographs of a clouds, and these in turn are affected by the mechanisms already cited. Note that
cumulus congestus taken at
one-minute intervals. Over
the convergent mechanism can be enhanced by frictional convergence in the
the whole sequence the planetary boundary layer of a low-pressure system, as is believed to be important in
chosen cloudy thermal grows tropical cyclones (section 12.3). When sufficiently encouraged, cumulus congestus
from being a small lump on may outweigh small cumulus in terms of cloud volume, but are usually out-
the edge of the congestus to
become its main turret.
numbered by the small cumulus which continue to coexist.
(Redrawn from [63]) Careful observation of individual cloudy thermals by time-lapse photography
shows that as they rise they tend to expand as if bounded by an imaginary cone,
whose vertex has a semi-angle of about 15° (Fig. 10.7). From this and from
336 laboratory work with similar convection in water, as well as from a variety of
theories, it is known that conical expansion is a characteristic feature of such pene- 10.4 Cumulonimbus
trative convection, and one which results from net entrainment. The familiar shape
of the congestus, on the contrary, shows a marked narrowing with height
(Fig. 10.6). The reason for this inversion in the aggregate of the behaviour of the
individual cloudy thermals lies again in the role of the established body of cloud as
protector of new cloudy thermals: those on the flanks of the cloud suffer much
more evaporative dissolution than those in its core, so that the congestus as a whole
converges upward despite the tendency of its individual constituents to diverge.
The sacrifice of the flanks, however, allows the central core to rise relatively effi-
ciently into the middle and upper troposphere, a virtual prerequisite for the
production of cumulonimbus.

10.4 Cumulonimbus

Though occasionally a few spots of rain may reach the surface from a small
cumulus congestus, and low stratocumulus may produce slight drizzle, it is a
familiar fact that most showers fall from clouds which have already grown to a very
substantial height, and hence warrant the title cumulonimbus. Indeed it is the dark
bases of such tall clouds, shadowed by efficient backscatter of sunlight by the thick
cloud above, which give the clouds their nimbus title and are correctly regarded as
threatening rain.
The reasons for the association of cloud thickness with capacity for precipitation
follow directly from the discussion of section 6.11 and problem 6.8. It takes a
considerable length of time, probably at least 20 min, for significant numbers of
cloud droplets (or crystals) to grow selectively to the millimetric size at which their
fall speeds are big enough to outweigh updraught speeds, and they are able to
survive evaporation in the sub-cloud layer. Only fairly substantial clouds last as
long as this, and even if a shallow cloud managed to avoid dissolution, fall speed
and updraught would have to be very nicely balanced for some minutes to allow
precipitation to develop in such small clouds.
Most showers fall from cumulonimbus which have very recently developed from
cumulus congestus. Where cloud base is reasonably high and the shower is viewed
against a pale background sky, the virga or long trails of precipitation reaching
down from cloud base to surface can be visually very striking (Fig. 6.17), par-
ticularly if they are curved by wind shear, as is often the case in the blustery winds
which accompany showers in cool equatorward airflows in middle latitudes. The
visible thinning of the precipitation arises as the smaller precipitation droplets are
evaporated to extinction before the larger, the great sensitivity of extinction
distance to droplets size (Appendix 6.4) ensuring that selective evaporation is
always significant. If the air is so cold that even the low troposphere is below 0 o C,
the precipitation will fall as snow, which has a characteristically pale and diffuse
appearance beloved of painters such as Turner (Fig. 10.8).
As a cumulonimbus drifts along with the wind over land, it leaves a trail of
precipitation (rain or snow, often accompanied by hail) which is as wide as the
active core of the cloud (usually no more than a few kilometres) and as long as the
distance the cloud is able to travel during its active lifetime. Since this life often
lasts about an hour, the length of the affected strip may vary from only a few kilo-
metres in very light winds to more than 50 km in a gale. Comparing clouds with 337
10 Smaller-scale weather
systems

Fig. 10.8 Winter

cumulonimbus precipitating
snow in north-west England.

similar precipitation production and lifetime but different drift speeds, we see that
the conventionally measured rainfall on any part of the strip is inversely propor-
tional to the total length of the strip and is greatest when the shower is stationary.
Showers occurring in very weak airflow are therefore particularly sporadic and
heavy in the affected areas, but even in a vigorous airflow the proportion of surface
area subject to showers may be quite small. The treatment of such uneven dis-
tributions in observation and forecasting has been mentioned in section 2.8.
Typical rainfalls (or the equivalent rainfalls of snow and hail showers) from
showers of normal intensity are observed by standard (stationary) rain gauges to
range from about I to 10 mm. (Unusually intense storms are discussed in section
10.6) Though such totals are quite modest, and compare with rainfalls from weak
to moderate fronts, they usually fall in a small fraction of an hour and therefore
correspond to short bursts of moderate to heavy rain, with rates of rainfall -
30 mm h-I, contrasting with the sustained periods of mainly light rain which are
typical of large areas of frontal rainfall. In weak airflows the prolongation of such
falls in limited areat~ can easily overload the natural drainage there, causing
localized flooding.
It often seems that a cumulus congestus, having taken 20 min or so to reach the
point of becoming a cumulonimbus by building slowly up to a vertical extent of
several kilometres, begins to grow much more conspicuously as precipitation
begins. As mentioned in section 6.11, in middle latitudes this transformation seems
to be associated with widespread glaciation of the upper parts of the cumulo-
nimbus. The rapid release of latent heat of fusion as large amounts of supercooled
water cloud freeze must tend to enhance buoyancy and further growth (appendix
10.1 ), as must the loss of some of the dead weight of cloud by precipitation. It is
certainly often apparent that as precipitation begins to fall from the cloud's base,
its top begins to grow more rapidly into the upper troposphere, often by sprouting
a cloud tower which at least initially is considerably narrower than the pyramidal
338 heap of congestus beneath. Rates of rise of 5 to 10m s - I are observed to be quite
normal, and are consistent with typical updraught speeds in the cloudy interior 10.4 C umulonimbus
measured and estimated by a variety of direct and indirect methods. Soon veiled by
glaciation, this tower is often quickly stretched very asymmetrically by strong wind
shear in the high troposphere, producing the anvil-shaped top which is so charac-
teristic of the mature cumulonimbus (Fig. 10.9).

Fig. 10.9 Distant summer

cangestus and cumulonimbus
with nearly symmetrical
anvils forming over sunlit
western slopes of ,the Pen nine
chain.

July 9 1959
The top o f the anvil is usually fairly flat, indicating that the updraughts are being
halted at a fairly definite altitude which is often observed to coincide with the
tropopause. The situation is sketched in Fig. 10.10, which shows how the
updraughts are inhibited by the convectively very stable low stratosphere. A rela-
tively simple treatment (Appendix 10.2) confirms observation by aircraft that
updraughts can overshoot their ultimate equilibrium level by several hundreds of
metres on account of their vertical momentum before sinking back ahd spreading
out. Sometimes there is a stable layer some way below the tropopause which serves
as the upper limit of cumulonimbus development. Whichever is the case, the
collision between the towering updraughts and the stable layer feeds large volumes
of ice cloud laterally into the anvil - a flux which is maintained throughout the
active life of the cloud. In the course of an hour's vigorous activity, huge amounts Fig. 10.10 Tephigram of
of ice cloud may be disgorged in the high troposphere. Since anvil cloud is relatively conditions in the troposphere
slow to evaporate, because vapour-density differences between the cloud and the on the occasion of an intense
cumulonimbus (a severe local
cloudless ambient air are very small (even when the ambient air is very dry, as can storm) in southern England.
be seen by extrapolating Fig. 6.2 to the very low temperatures typical of the high Radar and aircraft
troposphere), the resulting anvil may stretch many tens of kilometres down the observation showed tower
direction of the wind shear. The lower edge of the anvil rises progressively upwards penetration of the
stratosphere and anvil tops at
with increasing distance from the main cumulonimbus tower as smaller and smaller
the levels sketched
ice crystals sink more and more slowly into the underlying clear air (Fig. 10. 10). (horizontal structure is
Sometimes the undersurface of the anvil near the tower may hang down in smooth unrelated to the tephigram) .
breast-like shapes known as mammatus (Fig. 10. 11), which indicate the presence The dashed line is a saturated
of pockets of negative buoyancy probably driven by the evaporation of anvil adiabat. (After [42])

material into the dryer air of the middle troposphere.

Like the much smaller cumulus, cumulonimbus have a fairly clearly defined life
cycle. Initial growth as cumulus congestus is followed by the transition to cumulo- 339
10 Smaller-scale weather
systems

Fig. 10.11 Mammatus on the

underside of the anvil of a
summer cumulonimbus in
southern England.

nimbus, frequently accompanied by rapid extension into the high troposphere.

Precipitation falls in intensive shafts through the interior of the cloud body and
into the clear air beneath cloud base. As these develop, air in their vicinity begins to
fall, partly through frictional drag, and partly by chilling of unsaturated air by
evaporating precipitation (section 5.11). When they reach the underlying surface
these downdraughts are felt as the chilly gusts which usually accompany all but the
slightest showers. Eventually the updraughts are overwhelmed by the down-
draughts and the cloud begins to die, essentially as a result of its earlier success in
producing cloud and precipitation. The main trunk of cloud dissolves into a
temporary veil of precipitation, which looks much less solid than cloud because its
consists of a relatively very small population of very large droplets and particles.
After another 10 min nothing may be left of the recent glistening mountain of
cloud apart from the tuft of anvil drifting and slowly dissolving in the high tropo-
sphere and the long track of surface which has received precipitation- symbols of
the vertical exchange which underlies all convection.
Many cumulonimbus are associated with hail and thunder. As mentioned in
section 6.10, hail is produced by the relatively rapid accretion of ice and super-
cooled water (mainly the latter) onto a falling frozen embryo. The growing hail
stone soon has a relatively high fall speed through the ambient cloudy air, and in
fact is able to grow to a substantial size only if the updraught of cloudy air is strong
enough to hold it inside the body of the cloud for some minutes. However, cumulo-
nimbus are typified by their strong updraughts, and very large cumulonimbus by
their very strong updraughts - often exceeding 10 m s _,_ Hail is therefore typical
of larger cumulonimbus, though in the British Isles comparatively weak winter
cumulonimbus often produce little conical pellets of spongy hail (groupe/). The
association of maximum particle size with updraught strength explains why the
largest hail fall from summer cumulonimbus in middle latitudes, the potential for
the buoyancy which powers all such convection being greatest then. Hailstones
more than a few millimetres across remain unmelted even after falling through a
340 kilometre or more of warm air and often appear as a narrow shaft of hail in or
beside the core of heaviest rain (some of which is probably melted smaller hail). In 10.4 Cumulonimbus
lower latitudes the low troposphere is so warm that even quite large hail is melted
before it reaches the surface.
The same strong updraughts which encourage the growth of hail usually separate
electrical charge efficiently enough (section 6.12) to cause lightning within the
cumulonimbus and between its lower parts and the underlying surface. Most
substantial cumulonimbus produce thunder in this way, and even those
small ones which do not thunder audibly produce crackles of radio noise which
indicate that weak sparks are jumping between regions of separated charge. The
frequency of lightning strokes, like the maximum size of hail, is a measure
of updraught strength, which in turn is a measure of the vigour of the parent
cumulonimbus.
Civil aircraft are normally routed to avoid cumulonimbus because of the
discomfort and even danger of flying quickly through alternating up- and
downdraughts. Sudden downward accelerations in particular (widely but quite
misleadingly attributed to the presence of air pockets) can throw coffee cups, and
occasionally even unstrapped passengers, upward relative to the aircraft's interior
- a potentially dangerous example of how an unexpected acceleration seems like a
force in the opposite direction. In a very few cases such loads have caused struc-
tural failure of the aircraft and subsequent catastrophe. Although a few accidents
have been attributed to lightning strikes on aircraft, this risk is relatively small,
since the aircraft tends to keep close to the electrical potential of the ambient air.
Aircraft icing is potentially a more important hazard (section 6.13). Larger aircraft
carry a small radar which can detect the shafts of heavy precipitation which are
associated with the active cores.
Consider a typical sequence of events experienced at a point on the ground lying
in the path of a well-developed summer cumulonimbus. The air is warm and
humid, both factors contributing to the convective potential of the situation. The
sky becomes gradually overcast as the anvil of the advancing cloud reaches over-
head and beyond, though the details are often visually very unclear if the low
troposphere is hazy, as it often is in such conditions. The wind, which may have
been blowing gently in much the same direction as the storm movement, dies as a
persistent draw of air toward the base of the updraught sets in and locally offsets
the overall airflow near the surface. (Sometimes there is no obvious relation
between surface flow and storm motion, which is essentially with the airflow
several kilometres above the surface, and occasionally the two are opposed.) As yet
there is no obvious evidence of the storm: there is no precipitation and usually no
evidence of thunder or lightning unless it is already after dusk, but there is wide-
spread awareness of the 'lull before the storm', which is confirmed as the main
cloud mass begins to loom overhead. The depth of cloud is such that it may become
dark enough to trigger automatic street lights.
Quite suddenly there is a sound of wind, and nearby trees begin to move a few
seconds before the squall of downdraught reaches the observer. If there is time to
notice it before the onset of precipitation, the squall is obviously cooler than the
sultry air it has replaced. The rain begins very suddenly, almost as if switched on,
and may be heavy and contain bursts of small hail. As the same time the sky
overhead begins to lighten considerably, giving the appearance of a pale grey vault.
This is the tail column of downdraught where the cloud has been largely replaced
by the much less optically dense veil of precipitation. This too is the region where
the electrical activity is most obvious, with lightning strikes to ground, and less
distinct flickering and rumbling from lightning in the middle troposphere. The sky
lightens even further as the rain begins to ease off, and the squaJI winds subside. As
the storm passes away downwind it may become clearly visible for the first time as a 341
10 Smaller-scale weather glistening mountain of cloud. The deposited carpet of cool, fresh downdraught
systems remains, to be warmed again by the sun if there is enough of the day left.

10.5 Clouds and their environment

So far we have distinguished between cloudy and cloud-free air as if they were as
separate as they appear to be visually. But in sections 5.12 and 5.13 we considered
how convection, like any mixing, tends to produce an environment which is
unaltered by further convection. To the extent that this tendency is realized, there-
fore, the ambient air around thermals below cloud base, and around the cloudy
thermals above, is not independent of the buoyant masses which are currently
active - it has probably already been processed by many such thermals through
successive entrainment and detrainment. If the individual thermals are scarce and
weakly interactive with their surroundings, the ambient air may wait so long
between successive interactions that it is seriously out of equilibrium with the
thermals, and the distinction between thermal and environment may be quite
marked in consequence. But in fact most cumulus types of convection are so
vigorous that thermals and environment are only slightly out of equilibrium. This
is why the temperature excesses inside clouds are usually so small, less than 1 K in
cumulus humilis and usually less than 5 Kin even quite substantial cumulonimbus.
It is also why the various types of cumulus convection are associated with charac-
teristic vertical profiles of temperature and humidity which we will now briefly
summarize.
Figure 10.12 depicts typical vertical profiles of potential temperature in the
presence of various types of cumulus convection. The vertical scale has been made
roughly proportional to the logarithm of height to enlarge the details of structure in
the lower troposphere which would otherwise be inconveniently small. Remember
that conversion from potential to actual temperature would simply tilt the whole
towards the top left of the diagram by about 10 K km -t (section 5.6). The profiles
represent conditions outside the cloud masses which, remember, occupy only a
small fraction of the total horizontal area at any level above cloud base. They there-
fore represent what is almost always recorded by radiosondes in these conditions,
since only a small minority of sondes actually ascend through any substantial depth
of such cloud. Conditions inside a cloud of the cumulus family are always very

Fig. 10.12 (a) Vertical <a> (b)

profile of ambient potential
temperature in the presence
of cumulus and flattened
cumulus, with a saturated
adiabat shown dashed.
Dotted portions correspond to
flattened cumulus. (b) As s:::
0>
(a) but with vertical scale ·;;;
I
considerably reduced to
accommodate the full height
of a cumulonimbus. (Based on
[40))

342 Potential temperature Potential temperature

 10.5 C louds and their
close to saturation, and are therefore moister than the air outside the cloud, but are
environment
usually only a very little warmer, as discussed above.
Profile (a) in Fig. 10.12 represents the conditions accompanying small cumulus,
either in the early stages of a day over land which will later give rise to cumulus
congestus, or (dotted) confined beneath a subsidence inversion, or similar con-
vectively stable layer, a little above cloud base. The sharply superadiabatic layer
(sharp lapse of 0) typical of the surface boundary layer over land (section 9.9) is
apparent, as is the nearly adiabatic layer reaching up to cloud base. Careful
measurements suggest that this layer is not exactly dry adiabatic (uniform 0), in
that 0 lapses by - I Kin the first few hundred metres above the surface boundary
layer, and recovers by the same amount up to cloud base. If this is a general result,
it must arise from details of the way in which thermals vary with height, which are
still neither clearly observed nor understood. Near cloud base the profiles differ
somewhat, depending on whether we are beneath a cloud or a gap between clouds.
Rising into cloud base the temperature profile converts from a dry adiabat to some-
thing close to the appropriate saturated adiabat, as expected (section 5.10).
However, rising toward the level of cloud base whilst beneath a clear gap between
clouds, 0 tends to begin to rise before cloud base is reached, for reasons which are
no doubt complex in detail, but which in principle depend on a very important
aspect of cloudy convection which is worth brief attention.
The cloudy updraughts are only one part of an unsteady convective cycle which
must include downdraughts as well. Some of the downdraughts occur within the
clouds, as mentioned in descriptions of cumulonimbus, but most of the sinkage
occurs in the clear air between clouds. These up-and-down motions must maintain
a close balance of mass flux (only very slightly biased by the presence of any large-
scale convergence or divergence- section 7 .13). But the rising air tends to follow a
saturated adiabat, whereas most of the sinking air must tend to follow a dry
adiabat. Air sinking dry adiabatically after cloudy ascent will as a result tend to
become warmer than the rising cloudy air. In the main this tendency is held in check
by the frequency and vigour of the interactions between clouds and environment,

Fig. 10.13 Cumulus

congestus , probably kept
unusually narrow by strong
convective instability. The
sharp contrast between
conditions in the cloudy rising
air and the clear ambient air
is visually obvious, as is the
Iorge area of the intervening
mixing zone.

343
10 Smaller-scale weather mentioned already. However, it seems that the interactions at around cloud-base
systems level are sufficiently weak to allow slightly warmer air to extend a little way down
into the sub-cloud layer between clouds.
Above cloud base the potential temperature profile deviates significantly from
the expected saturated adiabat, tending consistently to lower temperatures (to the
left in Fig. 10.12), but still maintaining a rise of potential temperature with height.
This too is a consequence of the different adiabatic processes being followed by the
rising and sinking air. If both branches followed the dry adiabatic, then according
to the general law of mixing (section 5 .12) the equilibrium profile approached by
continued mixing of this type would be a dry adiabat, as is very nearly the case in
most of the sub-cloud layer. If both branches followed the wet adiabatic process,
then that would be the equilibrium profile. Since the branches differ, the
equilibrium must fairly obviously lie somewhere between the dry and wet adiabatic
profiles, at a position determined by details of the mixing effected by any particular
population of clouds, including for example the vertical profile of rates of entrain-
ment and detrainment. This is borne out by observation, though incompleteness of
current theory and observation prevents detailed matching. Observed environ-
mental lapse rates in layers populated by small cumulus often lie about one third of
the way towards the dry adiabatic from the relevant saturated adiabatic, the
fraction decreasing with increasing cloud cover. As already mentioned, tempera-
tures inside the clouds differ very little from those outside. In fact some small
cumulus seem to be capable of operating on zero temperature excess, relying on
their excess water vapour to maintain buoyancy.
Above the layer of small cumulus the potential temperature profile typical of
cloudy convection may extend through a deep layer populated by cumulus
congestus, or potential temperature may rise sharply, indicating the presence of a
convectively stable layer (dotted line on Fig. 10.12). In the latter case the cloudy
updraughts lose buoyancy very quickly as they rise into the warmer ambient air.
Terminal detrainment quickly follows the cessation of upward motion, but a little
exchange of air between the convecting and stable layers occurs in the process.
When the stable layer is maintained by subsidence of the lower middle tropo-
sphere in an anticyclone, it may persist for days in a type of dynamic equilibrium
with the underlying convection. If convection has the advantage, as may become
the case if the air moves over warm land or warm water, the convecting layer eats
upwards into the stable layer, but at a rate which diminishes as the erosion
proceeds, since this tends to sharpen the 'elbow' in the temperature profile and
hence increase the stability of the immediately overlying air. If the subsidence has
the advantage, either through weakness of convection or enhanced anticyclonic
vigour, then the convecting layer is squeezed downwards into a progressively
shallower layer, producing stratocumulus as mentioned in section 10.1, and even
leading to the extinction of all cloud if the base of the stable layer sinks below the
lifting condensation level. The dynamic interaction at the base of the stable layer
exchanges sensible heat and water vapour between the subsiding and convecting
layers, the subsiding layer being cooled and moistened and the convecting layer
being dried and heated as a result. Estimates have shown that an appreciable
fraction of the heat input into the bottom layers of an anticyclone over land is
brought down from the subsiding air in this way, most of the remainder coming
through direct absorption of sunlight by the ground.
Profile (b) in Fig. 10.12 represents conditions accompanying cumulus congestus
and cumulonimbus convection. The details up to levels reached by small cumulus
are essentially the same as in profile (a), which is consistent with the observation
that large cumulus are normally accompanied by small cumulus, though not the
344 stunted type associated with anticyclones. Above these levels the potential
temperature profile remains intermediate between the dry and saturated pro- 10.5 Clouds and their
files, for reasons already discussed, but tends in the upper troposphere to return environment
toward the saturated adiabat through conditions at cloud base. In fact the equili-
brium buoyancy level indicated by drawing a saturated adiabat upwards from
the lifting condensation level associated with screen level conditions often gives
a rough but observably reasonable value for the level of cumulonimbus tops
(Appendix 10.3).
The tendency for ambient temperature profiles to recover toward the saturated
adiabat from the lowest limit of the convection no doubt reflects the concentration
of detrainment and dissolution by cumulonimbus, especially in the upper tropo-
sphere. It is as if the middle troposphere were partly bypassed by the deep convec-
tion currents linking low and high troposphere, so that the environment in these
intermediate levels is influenced only at second hand by the tumult of small
cumulus rapidly processing the low troposphere and the scattering of cloudy
towers reaching toward the top of the troposphere to disgorge their fans of icy
anvil. The intertropical convergence zone is densely populated by many of the
largest cumulonimbus on Earth, and it is significant that these are often technically
termed hot towers, on account of their role in piping latent and sensible heat from
the warm surface (most of it ocean) to the vicinity of the equatorial tropopause
about 15 km above. In fact such hot towers collectively represent the ascending
branch of the Hadley circulation outlined in section 4. 7.
The existence of a layer of minimum interacting between updraughts and their
middle-troposphere environment is associated with a basic property of entrain-
ment and detrainment: that it is least effective in proportion to the mass of rising
air when updraughts are largest. Basically this is yet another example of the impor-
tance of the decline of surface-to-volume ratio with increasing scale which applies
to phenomena as diverse as the heat economies of mice as compared with
elephants, and the slowing rate of growth of cloud droplets growing by conden-
sation (section 6.10): the mass of a thermal is proportional to its volume whereas its
rates of entrainment and detrainment are proportional to surface area. The
insulating effects of large scale are confirmed by more technical arguments which
follow from an extension of the treatment of entrainment begun in section 7 .12.
These indicate that small cumulus should mix themselves out of existence in the
short time taken for their tops to rise a few hundreds of metres, as indeed is usually
observed, whereas updraughts more than a kilometre or so in breadth should be
able to rise through the full depth of the troposphere without losing more than a
fraction of their mass by mixing with the environment. The advantage of scale is
apparent too in the mechanism outlined for the development of cumulus congestus
(section 10.2).
An inevitable consequence of such minimal interaction is that it encourages
maximum differences between clouds and environment: temperature excesses in
cloudy interiors are largest in the mid-troposphere and often amount to several Kin
cumulonimbus. Buoyancy is thereby enhanced and this together with the reduction
in entrainment drag per unit mass (section 7.12) encourages strong updraughts.
Compared with the fretful inefficiency of small cumulus, cumulonimbus, are rela-
tively very efficient in all that they do, and are capable of producing quite vigorous
weather locally, as described in the previous section. If conditions are such that
cumulonimbus produce weather which is so severe as to be locally damaging at the
surface, and this applies to a small but significant fraction of the total, then the
cumulonimbus are called severe local storms.

345
10 Smaller-scale weather
systems 10.6 Severe local storms

Certain land areas are seasonally subject to unusually intense cumulonimbus when
intensity is judged by rates of precipitation, maximum size of hail, squally wind
speeds, and electrical activity. Such storms also quite often produce one or more
tornadoes (Fig. 10.14): intense localized vortices extending from cloud base to
surface, where they produce damaging winds (but not always the devastation
reported by the media- see later in this section) .

Fig. 10.14 A powerful

tornado in the United States
(10 April1979) showing the
parent cloud base tapering
downwards into the vortex
core. The great breadth of the
cloudy core shows that the
pressure in its centre is very
low. Near the ground. violent
winds maintain a plume of
centrifuging debris.

Severe local storms have been studied quite extensively in recent years because of
the very considerable damage they can do. Much of this study has been carried out
in North America, partly because of the considerable incidence and severity of
storms there, but also because that particularly affluent and sophisticated popula-
tion demands detailed forecasting of such events, and supports a greater per capita
number of forecasting and research meteorologists than anywhere else in the
world. However, it is important to realize that severe local storms occur in many
regions scattered throughout the world, including the British Isles, Europe and
European Russia, north India, north Indochina and China, with many of the
features mentioned below in the North American context.
Despite gross differences of geography and climate between different affected
regions, it seems that there are two features which are usually present when such
storms occur. The first is a mechanism for accumulating and then fairly suddenly
releasing substantial convective instability. Without such a temporary bottling up
of energy, the atmosphere releases its instability as soon as it appears, in weak,
unspectacular cumulonimbus, which do little more than contribute a little to local
precipitation. The second feature is the presence of significant wind shear dis-
tributed through a considerable depth of the troposphere. Though much less
obviously relevant than the accumulation and release of convective potential, wind
shear appears to play a crucial role in making storms especially vigorous and long-
lasting, and therefore severe, and it does so in two quite distinct ways which
correspond to two different types of storm - the multicell and the superce/1
storms. Let us consider these two features, and the ways in which they are believed
346 to contribute to the appearance of severe local storms.
 The accumulation and sudden release of convective instability occurs in 10.6 Severe local storms
different ways in different geographical regions, and usually depends on quite
specific local distribution of sea, land and topography, though the release of condi-
tional instability (section 5.12) is often a common feature . Consider, for example,
the severe local storms which occur in the central United States in the spring and
summer. These are amongst the world's most spectacular storms, and often occur
ahead of weak cold fronts which have crossed the Rocky Mountains and are
moving slowly across the Great Plains. A typical situation is depicted in Fig. 10.15,
which shows a weak cold front orientated from roughly south-west to north-east.
The associated jet stream is similarly orientated (according to the thermal-wind
relationship- section 7 .l 0) and provides the necessary wind shear throughout the
troposphere in a fairly broad zone on either side of the jet core. As will be outlined
in the next chapter, the flow of air in the vicinity of fronts is much more complex
than is apparent on either horizontal or isobaric surfaces. There is always a three-
dimensional interlacing of airflows from source areas which can be hundreds or Fig. 10. 15 Airflows
even thousands of kilometres apart, and which look quite unconnected with the associated with outbreaks of
system on purely horizontal or isobaric charts. severe local storms in the
In the case depicted, the air in the low troposphere well ahead of the advancing southern central United
States. Warm moist air from
cold front is flowing northwards from the Gulf of Mexico, and is consequently very the Gulf of Mexico is labelled
warm and humid. Such air is potentially very good convective fuel, on account of G , and hot dry air from the
its high wet-bulb potential temperature, but the potential would be frittered away Mexican Highlands is labelled
in rather ordinary cumulonimbus convection if its release were not temporarily M. (After [42))
inhibited by the presence of an overlying 'lid' of warmer air which confines any
convection in the warm, Gulf air below a convectively stable interface between the
two layers (Fig. 10.16). The warmer air is dry and seems to originate over the arid
highlands of New Mexico and Mexico to the south-west, where the strong sun
shining on the dry, elevated surfaces produces air with relatively high potential
temperature, but low wet-bulb potential temperature. This air is then drawn into
the confluent sandwich of air streams which is so typical of the vicinity of fronts ,
where it overlies the Gulf air.
The release of the convection potential of the troposphere as a whole occurs
when the underlying moist Gulf air becomes warm enough to convect through the
last vestiges of the weakening stable layer of the lid. This warming seems to come

Fig. 10.16 Tephigram of

Gulf and Mexican air streams
at position X in Fig. 10.15.
The chained line shows how
the saturated Gulf air would
rise in general uplift. (After
[27] and [64))

347
10 Smaller-scale weather about partly as a result of solar heating of the Gulf air as it over-runs the warmed
systems
Plains in the middle or latter part of the day, and partly as a result of the general
uplift of the whole sandwich of airflows in the lower half of the troposphere which
is associated with persistent large-scale convergence in the vicinity of the frontal
system (sections 7.13 and 12.1). The general uplift begins to warm the moist Gulf
air faster than the dry Mexican air as soon as cloud begins to form in the Gulf air,
since the latter will then tend to rise wet-adiabatically while the unsaturated
Mexican air continues along a dry adiabat (Fig. 10.16). In fact as the lid fades
away, the steep temperature lapse rate of the Mexican air directly surmounts the
cloudy Gulf air beneath, comprising a composite layer which is convectively
extremely unstable. Clouds begin to tower up into the Mexican air, accumulating
buoyancy at least to the level of the upper middle troposphere, as they rise and
become progressively warmer than their immediate environment, until the most
favoured reach the tropopause. These then offer efficient paths for the rapid ascent
of more warm, moist Gulf air into the high troposphere, with consequent copious
production of cloud, rain, hail and thunder.
It is now that the presence of wind shear through a deep layer begins to have its
effect, achieving rather similar results by interestingly different means in the cases
of the multicell as compared with the supercell storms.

Fig. 10.17 (a) Horizontal lbl

wind vectors in the low,
middle and high troposphere C motion
over ground
(L, M and Has observed from 4
the surface) on an occasion I
I
such as shown in Fig. 10. 15. / l relative
The vector difference

~~"'
between M and H agrees in
direction with frontal
alignment through the
thermal wind relation. (b) The
same winds relative to a / ~0 r~la
storm cell C moving with the
middle troposphere flow.
(c) Consistent development of
new cells on the right flank
pulls apparent storm motion
to the right of individual cell
motion. In the multice/1 storm, the wind shear between lower and middle levels
(somewhat above the disintegrating lid) is outlined in Fig. 10.17 for the situation
described above. The individual cumulonimbus tends to move with the flow at
middle levels, so that we can find the airflow at low levels (and any other level of
interest) relative to the travelling cumulonimbus by vectorially adding an overall
movement which exactly opposes the middle-level flow. When this is done in the
present case we find that, from the viewpoint of an observer riding on an individual
cumulonimbus, the low-level flow of warm, moist air flows in from a direction
which is distinctly to the right of the direction of motion of the storm. Here the
humid air rises in the powerful updraught, producing the great column of cloud
and consequent precipitation and electrical activity which constitutes the currently
active convective cell. However, as the inevitable downdraught begins to choke the
updraught, the inflowing surface air begins to develop a new updraught on the
right flank of the previous one, climbing over the spreading flood of downdraught.
This new cell flourishes at the expense of its parent until it too is replaced by
another on its right flank. The sequence is depicted on Fig. 10.17, and has no
obvious end for as long as there is potentially warm and cool air to be vertically
348 exchanged by convection. Although close inspection shows the sequential develop-
ment of new cells, the overall impression is that there is a single, powerful storm 10.6 Seve re local storms
which is moving distinctly to the right of the flow at middle levels. The significance
of the title multicell is obvious. The sequence may run on until evening cooling of
the surface air reduces its convective potential, or it may break down earlier
because some local topographical feature inhibits the development of a new cell.
Many quite ordinary cumulonimbus appear to have the multi cell structure too, but
when the convective potential is high enough to produce severe storms, the multi-
cell structure ensures that the severity is persistent and distributed over substantial
tracts of surface.
In the superce/1 storm the prevailing deep layer of wind shear allows a single
powerful convective cell to become organized into an essentially steady, though
highly dynamic, state. The detailed dynamics of the development and maturity of
these storms are not well understood, and are far beyond the level of an intro-
ductory text, but it is possible to see one basic principle at work as follows. The
downdraughts associated with the falling and evaporating precipitation originate
in the upper middle troposphere, where the air is moving much more quickly than
the air near the surface, on account of the presence of pronounced shear. The
downdraught air tends to maintain its initial momentum as it sinks down to lower
levels, flooding north-eastwards in the particular case of the situation outlined
above and in Figs. 10.15 and 10.16, and thereby helping to 'shovel' the warm,
moist air into the base of the updraught, which for the same reason adopts a corres-
ponding slope backward relative to the storm's direction of motion (Fig. 10.19).

Aughton 1200Z 19 May 1989

Fig. 10.18 The troposphere
on the occasion of unusually
heavy rainstorms near
Halifax (north-central
Britain). The area of the
shaded zone on the
tephigrom is a measure of the
potential energy available for
conversion in air rising wet-
adiabatically from cloud bose
(remaining strong buoyant up
to the 300 mbor level) . A
conglomeration of cells of
heavy precipitation was
observed by radar moving
slowly over several hours.
(After [65])

349
10 Smaller-scale weather Wi nd profile
systems I
Key I
I
Relat ive to
surface
Fig. 10.19 Vertical section Relative to
through a supercell storm storm
showing updrought and .::::;:>
Storm motion
downdrought, inflows and relative to
outflows, as observed from a surface
frame moving with the storm
(i.e. with the flow in the m s-1
lower middle troposphere).
The dashed line indicates the
path of a Iorge layered
hailstone from effective The result is that in at least the lower half of the troposphere the rising and falling
origin 0 to impact I on the airflows are adjacent and parallel, rather than entangled as they tend to be in the
surface. The squall or gust
front S is the leading edge of
more vertical, ordinary and multicell cumulonimbus, and are tilted so that the
the cool downdrought, and updraught overhangs the downdraught. The burden of precipitation no longer
tornadoes ore most likely in falls back through the parent updraught; instead it falls into the downdraught,
the region T. (After [27] and enhancing it by drag and evaporative cooling. The self-choking tendency of more
(42]) vertical cumulonimbus is therefore avoided, and the single supercell could, on this
simple view, continue to operate for as long as the convective potential of the
overall situation was maintained.
We might expect the supercell storm to travel with the air in the middle tropo-
sphere, as is the case with ordinary cumulonimbus and the individual cells of the
multicell storms, but this is apparently not generally the case. The movement is
often distinctly to the right of the mid-tropospheric flow, but leftward motion has
also been observed, and neither of these tendencies is properly explained. What-
ever is the storm motion, it is the flow relative to the storm which is nearly steady
and which is depicted in a simple two-dimensional vertical section in Fig. 10.19.
The three-dimensional pattern of relative flow in the case of a right-moving storm
is depicted in plan view in Fig. 10.20, where it is apparent that the single supercell is
maintaining a configuration of flow and storm motion which is similar to that of
the composite multicell.
In the space of the several hours which is the normal lifetime of supercell storms,
their relatively rapid movement over the ground means that they can cover many
tens of kilometres, leaving a carpet of precipitation, chilled air and more or less
serious damage.
The separation of updraughts and downdraughts is obviously consistent with the
Fig. 10.20 Plan view of a tremendous power of supercell storms. Updraughts become horizontally very large
right-moving supercell storm, (up to 10 km), and the minimization of drag and maximization of buoyancy which
showing low-level inflow and
climbing flow (dashed), and
follow from the advantages of scale ensure that they also become very strong.
high-level flow and outflow Vertical speeds of as much as 50 m s- 1 (100 milh) are believed to occur occa-
(solid). Concentric envelopes
contain areas of cloud, rain
(R) and hail (H). The vault Vis
largely free of radar echo Hi gh level
because the updrought there outflow m anvil
has few precipitation
particles though it is full of
dense cloud. A hook-shaped
radar echo is often observed
beside V just before
tornadoes appear in the some
Low
area. (Redrawn from (8]) level
inflow
350
sionally, on the basis of some exceedingly bumpy traverses by military aircraft and 10.6 Severe local storms
dramatic inadvertent ascents by gliders, and from indirect evidence such as the
height of updraught overshoot into the stratosphere (Appendix 10.2). Such
powerful updraughts maintain high rates of production of cloud and precipitation,
and permit the growth of giant hail (section 6.10), whose size provides further
indirect evidence of updraught strength, through the inferred fall speeds
(Fig. 6.10) which obviously must be exceeded by updraught speeds throughout
much of the life of the growing stone. In fact the onion structure observed in some
giant hail suggests a vertical oscillation of the growing stone as depicted in
Fig. 10.19- the stone being first carried aloft in the great updraught, then falling
out of the forward overhang of the anvil back into the updraught again, and
repeating this cycle until it is big enough to have a fall speed substantially greater
than the core of the updraught, through which it finally falls to the surface. The
large variations in temperature and cloud conditions between the extremes of the
looping path are consistent with the alternating zones of clear and opaque ice, as
mentioned in section 6.11. Giant hailstones five or more centimetres across can
clearly be very damaging, but they are produced in such small numbers by the
ultimate example of selective growth that reports of injury are rare.
The very rapid convergence of air at the base of such powerful updraughts can
generate tornadoes by concentrating residual synoptic or mesoscale vorticity (by
the spinning-skater effect- section 7.14). According to section 7.15 the fall of
pressure in a horizontal traverse to the core of a tornado can be as much as
100 mbar. Since this is comparable with the vertical lapse of pressure in the first
kilometre of the atmosphere, the base of the cumulonimbus cloud is often drawn
right down to the surface, as air spirals violently into the axis of the vortex before
ascending, to form the sinuous trunk of cloud which is typical of the mature
tornado (Fig. 10.14). Air whirls around the tornado vortex with tangential speeds
probably exceeding 100m s- 1, though direct measurement is virtually impossible
in such extreme and localized conditions - the violent core of the vortex being
seldom more than a couple of hundreds of metres across. Most buildings suffer
severe damage in such conditions, with roofs taken off and unframed walls
demolished. Those stout enough to remain intact at the first onslaught may
succumb to a tendency to explode if they do not leak internal pressure fast enough
to match the sudden fall of external pressure as the vortex core passes over. The dis-
located components are then whirled away by the fierce winds just outside the
vortex core, contributing to the cloud of fast-moving debris which continually
sprays tangentially out of the foot of the tornado and is potentially lethal to any
exposed animal or man (Appendix 7.7). Fortunately, the conditions for the main-
tenance of a tornado are only marginally attained in even the most vigorous storm,
and the majority of individual vortices seldom remain at full damaging strength for
more than 20 min, but that is cold comfort to anyone whose farm or house has
been traversed in that brief, violent life.
The downdraught too can attain great strength through separation from the
updraught, and through the effects of the very heavy precipitation, and it is not
unusual to record gusts of hurricane strength ( > 33 m s - 1) where fallen air spreads
out at the surface. Though much less intense than tornadoes, these winds can cause
damage to crops (which can become irreversibly lodged on the ground by being
battered down when wet) and buildings over a much wider path than that of a
tornado.
The various types of very severe weather mentioned above are not confined
exclusively to supercell storms, since many very severe storms are of the multicell
type. However, the efficiency of the supercell structure means that virtually all
supercell storms are especially severe. 351
10 Smaller-scale weather The combination of hurricane-strength winds, torrential rains, giant hail,
systems
intense thunder and lightning, and the possibility of tornadoes, makes the severe
local storm one of the most damaging types of weather system, and this together
with the intrinsic interest of such powerful systems attracts the attention of
research scientists and forecasters. The North American examples are amongst the
most spectacular, and include a very considerable incidence of tornadoes, as news
photography confirms each year when towns or villages are unlucky enough to be
traversed by them. Because of this, there are well-developed ways of forecasting
the particular conditions known from long experience there to be likely to produce
very severe storms, and of monitoring weather radars for the tell-tale signs of the
imminence of tornado formation (in particular the hook echo in Fig. 10.20). When
these are seen, local broadcasting is interrupted to give very specific local warnings
to areas at risk, which are repeated and amplified if tornadoes are actually sighted.
Little can be done to protect against damage, but taking cover in storm cellars or
sheltered rooms in solidly constructed houses very considerably reduces the risk of
injury by flying debris.
Although less spectacular than the tornado, one of the most expensive types of
associated damage is that done to crops by the deluges of moderate-sized hail which
often form in the core of each precipitation shaft (Fig. 10.21). Such stones are only
a tiny fraction of the weight of giant hail, but they are nevertheless heavy enough to
strip the head from a mature cereal plant, and numerous enough to reduce the
yields from affected areas almost as if they had been visited by locusts. In fact the
economic danger to farmers in parts of the United States which are prone to such
damaging hail storms rs such that it is prudent to insure against hail damage, and
one enterprising meteorologist has investigated the distribution of hail from a
series of storms by plotting the locations of fields subject to insurance claims for
hail damage.
In some countries affected by similar storms (Russia and northern Italy for
example) there have been persistent but largely inconclusive attempts to reduce hail
damage by firing explosive rockets or shells into threatening clouds in the hope of

Fig. 10.21 A maize crop

destroyed by a shower of
centimetric-scale hail.
Smaller stones lack the
damaging fall speeds (Fig.
6.10), and larger ones cannot
fall densely.

352
inhibiting the growth of hailstones to the damaging size. Although such efforts 10.7 Cumulus systems
probably help psychologically more than they do physically, there is some evidence
that they can occasionally be effective. The explosive dispersal of freezing nuclei
has also been tried and at least has the merit of being useful in principle, being an
attempt to use cloud seeding to increase the numbers and therefore decrease the
sizes of the hail by deliberate exaggeration of techniques used elsewhere for
precipitation enhancement (section 6.13).
In different regions, different aspects of severe surface weather are accentuated
at the expense of others. In most, the electrical activity causes damage to power
lines etc. In many areas it is the torrential rainfall as well as the hail. In North
America it is the hail and tornadoes. There is strong evidence that devastating
tornadoes are extreme examples of a spectrum of vortices. Patient work by
volunteers in Britain [95], sifting eye-witness accounts and inspecting damage,
shows that tornadoes capable of causing substantial damage, rather than
devastation, are quite widespread in southern Britain, and probably in Western
Europe (and probably, by inference, in most areas visited by severe local storms).
In some fairly dry regions like central northern India, which are nevertheless
subject to cumulonimbus vigorous enough in at least some respects to be classed as
severe local storms, the downdraught is the most significant feature for surface
conditions. This arises because of strong evaporative chilling of rain falling
through the very deep sub-cloud layer. This can even entirely remove the rain, like
torrential virga tapering to nothing (section 6.1 0), leaving the surface dry, but sub-
jecting it to a blast of cold air which raises vast dust clouds from the arid land.

10.7 Convective systems

We have already had two examples of the way in which individual cumuliform
elements can combine and cooperate to produce more extensive and persistent
systems. Individual cloudy thermals tend to combine to form the aggregate which
we call cumulus congestus when there is sufficient convective instability through a
substantial fraction of the troposphere (section 10.3). And in the presence of wind
shear through a similar fraction, individual cumulonimbus can cooperate to
produce the multicell sequence (section 10.6). There are other examples of con-
vective cooperation which produce distinctive systems of cumuliform elements on
scales ranging from - 100 m (small scale) to more than 1000 km (synoptic scale),
and these we will now briefly and selectively review.
Photographic reconnaissance by aircraft over warm seas shows that small
cumulus often appear in remarkably straight, parallel lines called streets
(Fig. 10.22), which can individually run downwind for as much as 100 km. Lateral
separation of adjacent lines is about 1 km. Over land, such lines are much more
irregular and almost always run downwind from a sun-facing slope or some other
especially effective source of thermals, but the ocean surface is normally too homo-
geneous to suppose that there could be a corresponding origin for maritime streets.
Although the mechanism is still not properly understood, it is clear that maritime
cumulus streets are a symptom of a larger-scale dynamic structure in the convecting
layer which has the effect of encouraging convection along long parallel lines, and
discouraging it along adjacent intervening lines. Given the marginal buoyancy of
small cumulus, the encouragement and discouragement need be only very slight to 353
10 Smaller-scale weather
systems

Fig. 10.22 Cumulus streets

aligned with trade winds.

produce a very marked degree of alignment in what would otherwise presumably

be a random cloud distribution. There is some evidence that opposing wind shears
above and below a low-level wind maximum plays a significant role, with repulsion
between the opposing horizontal vorticities of rising and sinking air currents
encouraging their lateral separation [66]. Observation of soaring sea birds suggests
that similar streets of thermals are quite widespread in the maritime boundary layer
even when there are no cumulus aloft, and this has been supported by recent
observation by high-power radar which is able to detect cloudless thermals by
backscatter from their small-scale turbulence.
Satellite pictures of showery air streams over oceans in middle latitudes almost
always show a characteristic distribution of ring-like patches or cells (Fig. 10.23)
which was quite unsuspected before there were any weather satellites, since the cells
are too small to be resolved by the very incomplete synoptic network there, and too
large to be coherently viewed by normal aircraft. Each cell seems to consist of an
extensive cloudless area surrounded by a roughly hexagonal ring of well-developed
cumulus, including congestus and cumulonimbus, and the whole showery zone is
made up of a close-packed mesh of such cells. Cells range in diameter from about
20 to 200 km, depending on larger-scale meteorological conditions, especially the
depth and instability of the convecting layer. Similar cellular structure is observed
on the western margins of the great maritime subtropical anticyclones, where
gradually destabilizing air flows polewards and eastwards. All such cells are termed
open on account of their open centres, and to distinguish them from the related but
contrasting closed variety.
Oceanic areas covered by extensive stratocumulus are observed to be patterned
by closed cells of a somewhat smaller horizontal scale and a much smaller vertical
scale than open cells. These consist of extensive pancakes of thicker stratocumulus
bounded by roughly hexagonal margins in which the stratocumulus is fragmentary
or absent. Again the whole zone is covered by a fairly close-packed mosaic of these
354 mesoscale elements, each of which in turn consists of an assembly of small-scale
 10.7 Cumulus systems

Fig. 10.23 Close-up of open cells of showery convection in the centre of Fig . 2 . 11 (55 " N .
20 " W }. Cells ore 20 to 30 km ocross i n this zone .

stratocumulus elements. Closed cells are most often observed on the eastern flanks
of maritime anticyclones, where the subsidence inversion is relatively low and
pronounced, and where the ocean surface is often kept cool by upwelling of cold
water from the ocean depths.
Open and closed cellular convection is also observed on a scale of centimetres in
laboratory experiments pioneered by Benard at the end of the last century. Small
tanks of water heated gently from below are observed to form closed cells
(Fig. 10.24), while air in the same situation forms open cells. In pioneering theoret-
ical studies of convection, Rayleigh was able to show that the crucial difference
between the two types is the vertical profile of viscosity: in the air tank the viscosity
is largest near the warm base, while in the water tank the viscosity is smallest there,
because of the opposite dependence of viscosity on temperature in gases and
liquids. On the relatively enormous scales of atmospheric cellular convection, the
effects of molecular viscosity are overwhelmed by eddy viscosity (section 7 .2), but
the fact that the Rayleigh number regimes (eqn (10.12)) are similar, if molecular
viscosity is replaced by realistic estimates of eddy viscosity in the atmospheric case,
is highly suggestive of a dynamic similarity between the two very different scales.
Rayleigh numbers then lie in the range known to be associated with cellular con-
vection (laminar and regularly patterned), rather than the range of larger values >
- 50 000 known to be associated with penetrative convection (turbulent and 355
10 Smaller-scale weather
systems

Fig. 10.24 Closed convection

cells in o small tank of water
heated gently from beneath.
Motion is revealed by
reflection streaks from
aluminium powder in a time-
exposure photograph. (From
[57))

chaotic). Now atmospheric cellular convection is far from laminar on the small
scale, since the convective elements are the intrinsically turbulent cumulus, but on
the mesoscales of the cells themselves the turbulence is so small in scale for the indi-
vidual eddies to be indistinguishable. Moreover the vertical profiles of eddy
viscosity resemble their molecular counterparts: in open cells there is an upward
lapse of eddy viscosity from a maximum close to the relatively warm sea surface
(the normal state of affairs in and above the turbulent boundary layer); while in the
case of closed cells, radiative cooling of the upper parts of the stratocumulus layer
produces negative buoyancy which encourages an elevated layer of enhanced
turbulence. It seems likely that the small laboratory models with their associated
simplified (but far from simple) dynamics isolate the meso-scale dynamics which
gently but persistently constrain the cumulus elements to form these large patterns
of cells in the atmosphere [67].
Relatively recent detailed studies of precipitation distribution in fronts show that
there are quite pronounced patterns on the meso-scale (Fig. 10.25). Bands of much
more intense precipitation are embedded in the synoptic-scale areas of frontal
precipitation, often aligned fairly closely with the nearest front. Although these
bands are too small in scale to be resolved by the synoptic surface network, their
presence is consistent with the large variations in rainfall rate which are such a
familiar feature of most fronts (Fig. 2.5). Weather radars show that these bands
coincide with bands of cumulonimbus (whose shafts of precipitation show up as
vertical stripes of radar echo on a range-height display) embedded in the frontal
nimbostratus. The instability producing these arises from the differential
advection which is typical of flow in the vicinity of fronts (Section 11.4), in which
air differing streams are brought together in a complex vertical sandwich, as
mentioned in the previous section. Direct instability (vertical lapse of potential
temperature) does not occur, but if a dry stream should flow over a saturated
stream with much the same potential temperature but higher wet-bulb potential
temperature, then the composite layer is unstable for cloudy convection, and
356 clouds will mushroom up from the saturated layer. This is very similar to the
 10.7 Cumulus systems

Fig. 10.25 Meso-scale

patterns of enhanced
precipitation in and near the
fronts of a mid-latitude
depression. The broad arrow
is the warm conveyor belt
(section 11 .4) of ascending air
in which bands of moderate
mechanism cited in the previous section, but here the instability may be quite rain are embedded (light
gentle, adding a cumuliform pattern to an existing substantial stratiform system stipple), which in turn contain
patches of heavy convective
(Fig. 10.26). Though the cumulonimbus are often quite feeble, they are neverthe-
rain (dark stipple). (From
less capable of enhancing the precipitation rate severalfold at the surface, either [68])
directly or by interactions between their precipitating ice crystals and underlying
layers of supercooled water cloud associated with the ambient nimbostratus. In
cold fronts the embedded cumulonimbus tend to be much more vigorous,
producing bursts of torrential rain and hail, vigorous gusts and lightning. The
alignment of meso-scale rainbands parallel to the ambient or nearby fronts (for
some bands form outside the frontal cloud masses) presumably arises because the
airflow at middle levels, in which the convection is embedded, is similarly aligned
by the thermal wind effect, but the precise mechanism is not well understood.
One type of linear alignment of cumulonimbus has been known about for a
considerable time, because it is big enough to show clearly on synoptic weather
maps, and because the cumulonimbus are often extremely vigorous, often being
severe local storms. This is the squall line of the central and eastern United States.
Typical conditions for the formation of a squall line are identical with those
already described in section 10.6 and Fig. 10.15 in relation to the formation of
severe local storms in the same region. On this scenario the cumulonimbus begin to
erupt along or just ahead of the cold front which is moving from the west or north-
west. By some sort of dynamic cooperation the cumulonimbus become very clearly
and closely aligned, to the extent that the very blustery leading edges of their squalls
merge into a nearly continuous line, giving the phenomenon its title. The line may
be several hundreds of kilometres long, acting like a very sharp forward extension
of the cold front. However, the ability of the advancing wedge of cool, dense air to
help 'shovel' the warm air ahead up into the updraught seems to encourage the
squall line to advance faster than the parent cold front, so that it becomes a
detached line of powerful cumulonimbus. Over the day or two in which the line
squall is identifiable it may advance several hundreds of kilometres into the warm
sector as a narrow band of intense rain and hail (Fig. 10.27). Forecasters are par-
ticularly interested in the development and progress of line squalls because of their
tendency to be or become sites of very severe weather, including tornadoes. In fact
it seems that the updraughts and downdraughts can become separated in much the
same way as in the supercell storm (section 10.6). Lines of cumulonimbus appear in

Fig. 10.26 A vertical section

across a warm front , showing
embedded weak vertical
convection whose
precipitating ice and snow
crystals seed the underlying
cloud sheet. (After [ 69))

357
10 Smaller-scale weather
systems

40

30

30
Fig. 10.27 A classic case of
a North American squall line,
extending over 1500 km from
Texas to Michigan and lying ; ::; Light rain
//
between 200 and 500 km
ahead of a doublet of cold p.-Y Squall tine ,.,{o5 mbar i sobar
fronts. (After [27] and [70])

other parts of the world (west Africa and northern India for example), and
although their mode of formation varies from place to place, and few are as
energetic as the American line squalls, they all seem to derive some dynamic advan-
tage from their linear cooperation.
There is considerable observational evidence to show that tropical cyclones can
be regarded as spiral assemblages of very vigorous cumulonimbus clustered round
the relatively quiet eye. The characteristic spiralling arms and the massive wall
which surrounds the eye are composed of very large and vigorous cumulonimbus.
The violent winds spiralling inwards toward the eye wall feed the powerful
updraughts of the cumulonimbus, which in turn sustain the very high rainfall rates
and the copious production of ice cloud which fans out aloft to produce the dense
white shield observed by satellite (Fig. 12.10). Other types of tropical weather
system too seem to be cooperating assemblages of cumulonimbus, although the
modes of cooperation are largely unknown. Tropical weather systems in general
and hurricanes in particular are outlined in sections 12.3 and 12.4. Even the vast
intertropical convergence zone is in essence a planetary-scale girdle of very tall
cumulonimbus linearly constrained (and constraining, since as usual there is no
clear chicken or egg) by the Hadley circulation (Fig. 8.14).

10.8 Atmospheric waves

There are many examples of atmospheric waves on the small and meso-scales
which are at least superficially similar to waves on the ocean surface.
358 Acoustic echo sounders have been used in recent years to probe the atmospheric
 10.8 At mospheric waves
Acoustic Sounder rat LilnCiister "'-'ghtoo~oO•lo
AcolhhC" ~. '• , l L:<~11r .t•'r AI ~Jioi.-61 ft.._laO II '.1
ll")-'h1 bo-,Jo M '~. · t 2300 rJ>'VII 93m above MSL .al 2300 Gfo.tT
on 18·11 · 197& l _.,t

liM& .. ,

(a) (b)

Fig. 10.28 Vertical sections through (a) gravity waves and (b) shear waves in weak flow in the
nocturnal boundary Ioyer, as observed by acoustic echo sounder in north-west England.

boundary layer by detecting backscatter from variations in air density on the scale
of half the acoustic wavelength used (- 10 em). Individual variations are asso-
ciated with micro-scale turbulence, but they are patterned by events on a somewhat
larger scale. At night they are seen to be almost incessantly patterned by events
which are clearly wavelike (Fig. 10.28a). Activity like this was largely unsuspected
before acoustic sounders enabled effectively continuous monitoring of the
nocturnal boundary layer, but it has now been observed to be typical of overland
convectively stable boundary layers from the tropics to the poles.
Many of the patterns are obviously gravity waves, in which vertical displacement
from equilibrium is restored by gravity, with subsequent under- and overshooting
of the equilibrium level giving rise to a series or train of waves. As such they
resemble surface water waves, but actually they are best compared with internal
water waves which are observed under water on horizontal boundaries between
water of slightly different densities. In the atmospheric case the boundaries are
always more or less spread out in the vertical by mixing, but the dynamics are essen-
tially the same. A vertically diffused layer in which density lapses with increasing
height (the opposite would of course be convectively unstable) is simply a con-
vectively stable layer, of the type discussed in section 5.12. It can be shown
(appendix 10.4) that if a parcel of air is displaced adiabatically and vertically in this
layer and then released, it will oscillate about the level of neutral buoyancy with
what is called the Brunt-Vaisala frequency Nl21r, where

N= ( L ao )v. (10.1)
(} dZ
and iJ(J!iJz is the vertical gradient of ambient potential temperature, which
expresses the degree of convective stability of the layer. This expression confirms
what is intuitively reasonable: that the frequency of oscillation increases with
stability, just as the rate of vibration of a spring-mounted mass increases with the
stiffness of the spring.
We can easily imagine how the waves in Fig. 10.28a come about. Nocturnal
cooling of the land surface established the stable layer, and airflow over humps and
hollows in the surface give the air its initial vertical push. The waves then propagate
through the air, just as waves propagate over the water surface, and produce the
patterns shown by raising and lowering a slightly turbulent layer above the acoustic 359
10 Smaller-scale weather sounder. The speed of wave propagation over the sounder is the algebraic sum of
systems the celerity of the waves through the air, and the speed of airflow. If the two speeds
were exactly equal and opposite, as might be in the case of waves rippling upwind
from a fixed obstacle, then we would have standing waves, which could be seen
only if we had several acoustic sounders operating instantaneously along the flow.
A fundamentally different type of wave is apparent in Fig. 10.28b, again from
acoustic sounding of a nocturnal boundary layer. Although it may slightly
resemble a breaking sea wave, this is a misleading coincidence of appearance, since
these interlocking wave shapes are believed to result from the Kelvin-Helmholtz
instability of a strongly sheared layer. Such shearing instability is responsible for
the flapping of a flag, which is encouraged by the difference in wind speeds
either side of the cloth, and restrained by the stiffness and strength of the cloth. In
the atmospheric case the sheared layers are horizontal and the restraining force is
the convective stability of the sheared layer. A dynamic analysis of the balance
between rippling by shear and smoothing by convective stability is beyond the
scope of this book, but yields as a critical parameter the Richardson number Ri
quoted in the discussion of turbulence in section 9.8:

Ri =
az I (~
o~
__¥__
az )2

When Ri falls below 0.25 (according to the detailed theory) then the convective
stability is unable to restrain the tendency to ripple, and the shear waves grow in
amplitude to the point where orderly flow breaks down in turbulence. The whole
process has the effect of converting some of the kinetic energy of the original
orderly flow into turbulent kinetic energy (and ultimately heat), and considerably
deepening the originally shallow sheared layer. The prevalence of the interlocked
wave patterns in the nocturnal boundary layer strongly suggests that strongly
sheared layers spend a considerable proportion of their time in the process of
growing these shear waves, which are quite distinct from the gravity waves
considered earlier, and are much shorter in wavelength. There is some inter-
connection between the two types when conditions encourage large-amplitude
gravity waves, which concentrate shear at their crests and troughs which then
generates shear waves and turbulence.
Gravity and shear waves play a very considerable role in the activity of stable
boundary layers, the shear waves in particular being an important link in the
production of turbulence which transports momentum, heat and material (water
vapour, etc.) to or from the cool surface. However, each type is also important at
much larger scales and heights in the atmosphere.
As mentioned already (section 9.11), airflow over hills and mountains often
maintains standing waves in the lee of the obstruction. These are basically the same
as the much smaller standing waves mentioned above. The theory of such waves is
quite well developed but is subtle in principle and complicated because of differ-
ences arising from the vast range of conditions of atmosphere and terrain. A
number of points can be made in relation to the outline in Fig. 10.29.
1. The dominant wavelength A induced by an obstacle is given approximately by
assuming that air flows through the steady waves at the undisturbed wind speed
V, swinging up and down at the Brunt-Vaisala frequency Nl21r. Since

V = frequency x wavelength
we have

A = 2 1r
v
360 N
 10.8 Atmospheric waves

Fig. 10.29 Vertical section

\

'\ of airflow containing hill and

\
\ lee waves and a turbulent
\
\ rotor R turning in a zone of
.c \
nearly separated flow. Nate
r
"'
·c;;
the typical increase of
wavelength with height,
upstream shift of leading
peaks and troughs, and less
typical decrease of vertical
displacement (see text). Flow
is usually conspicuously mare
turbulent in the lowest
clouds, and is violently so in
the rotor.
If we substitute reasonable values for the stability of the air flow (appendix
10.4), corresponding to a lapse rate of about 75o/o of the dry adiabatic value,
then we have the useful rule of thumb that the dominant wavelength in kilo-
metres is related to the average wind speed in metres per second by
'A =: 0.6V
the fraction (and wavelength) decreasing with increasing stability. Typical
wavelengths of - 3-20 km follow from typical wind speeds.
2. As implied by the absence of terrain profile in the above, the wavelength of the
lee waves is independent of terrain shape (i.e of any 'wavelength' there), but
this is not the case for the amplitude, the vertical oscillations in the waves. The
largest-amplitude waves are produced by terrain whose wavelength along the
flow is considerably smaller than the dominant lee wavelength, but whose
wavelength across the flow is very long. A hill or range which is very long in the
direction of flow is ineffective because the air merely climbs slowly up one side
and sinks as slowly down the other, without further vertical motion. And a hill
which is short in the cross-flow direction is easily bypassed by lateral airflow.
Amplitude obviously tends to increase with terrain height, but quite small hills
can be surprisingly effective if other conditions are optimum.
3. For a given terrain, the largest-amplitude waves arise when there is a significant
increase with height in wind speed toward the barrier, and a significant decrease
in stability. Largest amplitudes often appear in the vicinity of a stable layer and
may considerably exceed the height of the mountain or hill at heights far above
hill or mountain top.
4. Motion is everywhere smooth and wavelike except in the immediate lee of a
sharply scarped hill, and under the first wave crest downwind where there may
be a violently turbulent rotor in which trapped air is continually rotating.
5. For subtle dynamic reasons the axes of the wave crests and troughs shift upwind
with increasing height throughout the region in which the wave train is
exporting energy upward. This is believed to be significant on the large scale in
communicating surface drag to the middle and upper troposphere, just as the
wake of tied waves increases the drag on a boat and spreads its equal and
opposite effect through a large water mass.
Hill and lee waves become conspicuous when a layer of air is lifted above its
lifting condensation level in the wave crests, producing lenticular clouds whose
smoothness and stability amid the flowing air nicely depict the wavy nature of the
flow (Fig. 10.30). The dependence on atmospheric stability often enhances the
number and size of lenticular clouds at dusk, producing clouds kilometres above 361
10 Smaller-scale weather
systems

Fig. 10.30 Lenticular clouds

over the Yosemite National
Park, USA. The larger clouds
(middle right) are at a lower
level and they and the edges
of the clouds above them
show that the humidity
structure of the middle
troposphere is finely
structured in the vertical.

hills which are only a few hundred metres high. Updraughts and downdraughts in
large-amplitude wave clouds are sufficient to make flying conditions uncom-
fortable and even dangerous. The danger can be compounded by severe turbulence
in the vicinity of a steep hill or rotor, and many aircraft accidents have been
attributed to such encounters, including the fatal crash of an airliner which broke
up while allowing its passengers a view of Mt Fujiyama in the lee of that steep,
conical mountain (height 3.8 km) .
Lenticular clouds have been noted for at least as long as they have been speci-
fically named, but only with the coming of the satellite has it been clear that vast
trains of lee waves are the norm in the vigorous cloudy airstreams associated with
extratropical cyclones (Fig. 10.31). These would of course be invisible from the
ground since they are embedded in thick nimbostratus, but they most probably give
rise to at least some of the strong patterns of rain and snowfall long observed in
hilly terrain. Detailed mechanisms are complex and difficult to confirm, but it is
clear that cloud production is enhanced on the upwind side of a wave crest. If the
wavelengths are long enough, like the ones apparent on satellite pictures such as
Fig. 10.31, then precipitation too may be enhanced before the cloud is thinned in
the lee of the wave crest. A related possibility is that precipitation falling through
the denser cloud on the upwind side is able to grow much more rapidly by collision
and coalescence, or that ice crystals falling through air suddenly forced toward
water saturation by uplift grow rapidly by the Bergeron-Findeisen mechanism
(section 6.11 ). Mechanisms such as these are believed to be involved in the enhance-
ment of rainfall on the upwind flanks of hills and mountains, and its depletion on
the near downwind side, but pictures such as Fig. 10.31 suggest that the influence
could be much more pervasive, though hidden amid the inherent structure and
variability of weather systems. It seems clear from some detailed studies of rainfall
patterns that they may at least trigger the formation of some of the meso-scale rain
bands mentioned in the previous section.
Shear waves too are important at much greater heights than the boundary layer
in which they were noted above. In the strongly sheared zones below the core of a
jet stream in particular, the Richardson number may be held at about 1 throughout
substantial volumes of the high troposphere by the large-scale structure. Local
362 enhancements of shear or reductions in stability may then produce sheets of
turbulence far above and completely detached from the turbulent boundary layer. 10.8 A t mospheric waves
When first encountered by high-flying aircraft, these were judged most surprising
in cloudless air, where there was no obvious convective source of turbulence, and
became known as clear-air turbulence (CAT), but they are also found in the cloudy
parts of similarly sheared zones. Turbulence of small scale and amplitude is wide-

Fig. 10,31 Satellite view of north-westerly flow over the northern British Isles (i.e. top left to
bottom right), showing extensive wove trains associated with hills and mountains there. There
appear to be some Iorge closed cells in the south-west of the picture. Infra-red image from sun-
synchronous NOAA-6, 0836 z.
21 August 1980. 363
10 Smaller-scale weather
systems

Fig. 10.32 Billow clouds in strong wind shear in the high troposphere downwind of Mt Teide,
Canary Islands. Continued observation shows that each wove has only a limited life before it
breaks up in turbulence. (Reproduced by permission from Clouds and Storms: the behaviour and
effect of water in the atmosphere, by F.H. ludlam, Pennysylvanio State University Press,
University Pork , USA, 1980)

spread in the high troposphere, as may be judged on commercial flights by looking

out for slight but persistent vibration, which feels as if the aircraft were taxiing over
soft cobblestones. CAT can be much more violent on occasion, presumably as the
aircraft encounters large shear waves or the large eddies which are their immediate
aftermath, and the problem is sufficiently serious to have encouraged considerable
study of ways avoiding severe CAT by forecasting risky zones, or detecting current
CAT by special radar. The problem remains, however.
Shear waves occasionally produce clouds which pick out the crests of the
turbulent wavelets arranged across the shear (Fig. 10.32). These are sometimes
called billow clouds, which is unfortunate since breaking sea waves have no equiva-
lent of the shear which is their crucial ingredient. Shear waves and CAT are impor-
tant on a global scale because they ensure a distribution of frictional destruction of
the energy of large-scale airflows through substantial volumes of the troposphere,
rather than have it concentrated entirely in the turbulent boundary layer, or
communicated from there by gravity waves.

Appendix 10.1 Evaporative loss of buoyancy

Consider a parcel of unit mass containing air, vapour and cloud, and suppose that
some of the cloud water in the parcel evaporates, reducing the parcel's cloud
content and increasing its vapour content. If the process is adiabatic, the heat
needed to evaporate the cloud water must come from the parcel, chilling the
364 mixture of air, vapour and cloud.
 The heat needed to evaporate a mass m of water is simply the product of the mass Appendix 10.1
and the specific latent heat L of evaporation. The fall ~Tin temperature is given
by
~T = (L m)/(parcel heat capacity)
In full detail the parcel's heat capacity is the sum of the heat capacities of the air,
the vapour and the cloud water, and is complicated by the exchange of water
substance from water to vapour during the cooling. However, vapour accounts for
such a small proportion of the total parcel mass (exceeding 1o/o only in the warmest
and most humid parts of the troposphere), and cloud accounts for even less, that
we can closely approximate the parcel heat capacity by assuming that the whole
parcel is dry air, using the specific heat capacity CP at constant pressure. This
underestimates the actual heat capacity a little, since the specific heat capacity of
cloud water is over four times, and that for vapour is nearly twice the value for dry
air, but the small specific masses mean that the net underestimation is only a few
percent at most. Note the use of CP rather than c., since the former is relevant to all
meteorologically relevant processes (sections 5.3 and 5.11).
Because we are assuming unit parcel mass

~T= Lm (10.2)
cp
Inserting values for Land CP we see that the parcel temperature falls by about 2.5 K
for each gram of cloud water evaporated in a kilogram parcel. Realistically we
might expect to have a cloud water content of several g per kg, but on the edge of a
cumulus this would evaporate only if the cloudy parcel were mixed with an
unsaturated, cloudless parcel of much the same mass. The original calculation
therefore still gives a reliable order of magnitude for the cooling effect. Since
temperature excesses in cumulus are only - 1 K, evaporative chilling is clearly
potentially very significant.
The negative effect of such chilling on parcel buoyancy is offset by two effects:
the reduction of the burden of cloud water reduces the parcel density, and the
increase in the relatively low-density vapour component (molecular weight only
18/29 that of air) decreases it further. A detailed examination of these effects on
parcel density shows that they are equivalent to a warming of about 0.43 K for each
g per kg cloud water evaporated at typical low troposphere temperatures, and
slightly less at lower temperatures. The net effect on parcel density of all of these
effects (chilling by evaporation, loss of cloud burden, increase of low-density
vapour content), is therefore equivalent to a cooling of about 2.1 K for each g per
kg evaporated, which is still very considerable, and readily accounts for the losses
of buoyancy visible on the sides of cumuliform clouds.
The process described above is entirely reversible, showing that the condensation
of 1 g per kg of water vapour to form cloud warms the parcel by 2.5 K and
decreases its density as if by a simple warming of 2.1 K, with obvious relevance to
the production of buoyancy by cloud formation.
The chilling or warming which respectively follow melting of ice cloud or
freezing of water cloud can be found from eqn (10.2) by treating Las the latent heat
of melting of ice (or freezing of water). The temperature changes are only about
one seventh of those associated with evaporation or condensation since this is the
ratio of the latent heats. There is no offsetting correction to apply because the
vapour content is unchanged and the slight difference in density between water and
ice has quite negligible effect on the parcel density as a whole.
Direct sublimation from ice to vapour involves an L which is the sum of the Ls 365
10 Smaller-scale weather for melting and evaporation at the temperature involved. Chilling by sublimation
systems would therefore be especially destructive of buoyancy if the very small differences
in vapour density between cloud and environment at the low temperatures involved
(below - 25 °C) did not make the process very slow. Corresponding warming by
direct condensation from vapour to ice, such as is envisaged in the Bergeron-
Findeisen process, is correspondingly effective in producing buoyancy, and is quite
fast for reasons discussed in section 6.11.

Appendix 10.2 Convective overshoot

Figure 10.33 represents an idealized version of the vertical temperature profile in

the vicinity of the tropopause, in which an isothermal stratosphere surmounts a dry
adiabatic high troposphere.

Fig. 10.33 Idealized

temperature- height profile
around the tropopause. Solid
lines depict ambient
temperatures and the dashed
line traces on overshooting .c
c::n
air parcel. Continued ·a;
overshooting often maintains J:
a temperature inversion
below the isothermal
stratosphere, enhancing the
sharpness of the tropopause.
and gradually raising its
altitude. Temperature

Consider an air parcel in the updraught of a cumulonimbus which is neutrally

buoyant as it rises through the dry adiabatic layer (remember that saturated and
dry adiabats are nearly parallel in the very cold high troposphere). As it rises above
the tropopause the parcel continues to cool dry adiabatically and so becomes colder
than the surrounding stratosphere. At height z above the tropopause the tempera-·
ture deficit of the parcel will be z rd. According to Appendix 7.11 the resulting net
downward buoyant force per unit parcel mass will therefore be g(z r d)/ T, which as
expected increases with increasing penetration of the stratosphere. This force
opposes the continued rise of the overshooting parcel and eventually brings it to a
halt before sending it back down again.
Ignoring frictional and entrainment drag, the vertical component of the
equation of motion is
dw zrd
- = -g-- (10.3)
dt T
To solve this equation for the variation of w with z, note that
dw dw dz dw
=-- w--
366 dt dz dt dz
 Appendix 10.3
Rearranging eqn (10.3) to separate wand z terms and integrating, we find

grd r
J wdw =-rJ zdz +A

Ignoring relatively slight variations in T, we can evaluate the integrals from the
tropopause (z = 0, updraught w) to the level at which the speed of updraught has
been reduced to zero (height z, w = 0), to find the height of penetration of the
stratosphere in terms of the updraught speed w at the tropopause.

z= ( g ~ d ( w =( Tg~P ) ~ w
Substituting for CP and g, and assuming a realistically low value of 220 K for T, we
find (in SI units)
z == 48w
which would suggest that an updraught of 10m s- 1 should penetrate nearly 500 m
into the stratosphere. This is bound to be an overestimate, since drag must tend to
reduce the penetration, though the reduction cannot be reliably estimated.
Observations of the tops of severe local storms indicate penetrations of more than
1 km on occasion, which implies updraught speeds greater than 20m s- 1 • After
coming to a halt, the air sinks back to the top of the troposphere, and after some
oscillation (damped by drag) settles there.

Appendix 10.3 Cumulonimbus tops

Figure 10.34 represents a sounding on a showery autumn day in north-west

England. The showers were forming over a wide area of the northeastern Atlantic,
as a cool north-westerly airstream over-ran the relatively warm sea (which is
warmer there than even the western margins of the land throughout the winter).
The sounding therefore portrays an air mass which is well adjusted to the showers it
contains.
The cloud base at Aughton is very low, and the situation at low levels is
complicated by a weak convectively stable layer which represents the influence of
local land-based cooling (sunset was over an hour earlier). Over the slightly warmer
sea, the lifting condensation level was probably as depicted. On the simplest
adiabatic model of the showery convection, the rising air follows the saturated
adiabat from the lifting condensation level, and is therefore warmer than the
ambient air mass up to the level marked CT, where the saturated adiabat and
environmental temperature profiles re-cross. On this model the air accelerates to
produce a maximum updraught speed at CT, then overshoots and oscillates about
that level before detraining there, as discussed in appendix 10.2.
In fact the inevitable entrainment must dilute rising air so that with increasing
height it drifts to the cold side of the saturated adiabat, and widespread glaciation
and consequent release of latent heat (Appendix 10.1 - a little earlier the evening
sky was full of fresh and decaying anvils) must tend to reverse this trend in the 367
10 Smaller-scale weather Augnton t200Z 14 Nov 1972
systems
Pressure
lmbarl

Fig. 10.34 Tephigram for a

showery day. The solid line
traces the radiosonde
temperature profile, and the
dash-dotted line shows the
construction used to estimate
the altitude of cumulonimbus
tops.

upper part of the convecting layer, with the result that the level predicted by the
too-simple model agrees reasonably with observed heights of the tops of the highest
cumulonimbus. The presence of both entrainment and frictional drag keeps the
updraught speeds well below the frictionless values (given by the hatched area on
the sounding and having value 20 m s- 1 on this occasion), and ensures that they are
strongest in the upper middle of the convecting layer, because of the entrainment
minimum there, rather than at the top.

Appendix 10.4 Oscillations in a stable

atmosphere

Figure 10.35 depicts a convectively stable layer of air in which a parcel is displaced
vertically and dry adiabatically from its original position of neutral buoyancy at 0.
According to Fig. 10.35 a raised parcel is cooler than its surroundings and conse-
quently negatively buoyant, whereas a depressed parcel is warmer and positively
buoyant. If a parcel is released from any initial displacement it will therefore
accelerate towards 0, overshoot, slow to a halt, accelerate back toward 0 again,
and continue oscillating about 0 until the motion is damped out by drag, leaving
the parcel at 0. Let us examine the dynamics of such oscillation, slightly changing
and generalizing the method used in appendix 10.2.
368 If the sub-adiabatic lapse rate of the stable layer is r, and the dry adiabatic lapse
rate of the profile traced by the parcel is r d• then the temperature excess of the Appendix 10.4
parcel at any height z above 0 is given by z (f - r d). Since r < r d• this expression
shows that upward displacement (positive z) produces a temperature deficit, while
downward displacement (negative z) produces a temperature excess, consistent
with Fig. 10.35. Note that this is a more general form of the expression used in
appendix 10.2, where r was zero since the stable layer was isothermal. Algebraic
manipulation shows that the expression for parcel temperature excess can be
simplified to - z dO/dz, where dO/dz is the potential temperature gradient of the
stable layer, and 0 is the constant potential temperature of the parcel. There is only
one temperature gradient in this form of the expression for the parcel temperature
excess because the potential temperature gradient corresponding to dry-adiabatif:
parcel motion is zero.
We can use either of these expression for the parcel temperature excess to derive Temperature

the buoyant force per unit parcel mass, as in appendix 10.2, and hence write down Fig. 10.35 Temperature-
the vertical component of the frictionless equation of motion: height diagram of vertical
oscillation of an air parcel in
dw g dO a convectively stable
-- ---z environment.
dt 0 dz
dw 2
Since -- - d- z- , t h e equatiOn
· can b e wntten
· · t he farm
m
dt dt2
d 2z
-- - N2 z (10.4)
dt2
g dO
where N2
0 dz
If the amplitude of the oscillation is small enough to allow the oscillating parcel
to remain within a layer with uniform dO/dz, then N is a constant, and eqn (10.4)
becomes a perfect example of the equation for simple harmonic motion - the
idealized oscillatory motion which governs small-amplitude swinging of
pendulums, and serves as the model for many natural and man-made oscillating
systems. Standard procedure, or simple substitution, shows that solutions are of
the form
z =A sin Nt (10.5)
where A is the amplitude of the oscillation (the maximum distance the parcel moves
above or below 0), which, however, remains undetermined unless we define the
parcel's starting speed, direction and height z, as was done in appendix 10.2. By
contrast N is fully determined by the conditions already described (crucially by the
convective stability of the ambient layer). According to the behaviour of sine
functions, eqn (10.5) describes an oscillation which completes one cycle in a time
period 21r IN, which means in other words that the frequency of oscillation about 0
is Nl21r. This is the Brunt- Vtiistilti frequency of adiabatic oscillation in a con-
vectively stable layer, and it is apparent that the frequency increases with stability.
Substitution of values shows that the isothermal layer considered in appendix 10.2
has a Brunt-Vaisala period of almost exactly 5 min. {Indeed it should now be clear
that the analysis in appendix 10.2 was of one half of a Brunt-Vaisala cycle, with
special interest in the amplitude rather than the frequency of the oscillation.) Less
stable layers have longer periods (lower frequencies), tending to infinity as the layer
approaches neutral convective stability (a dry adiabatic lapse rate).

369
10 Smaller-scale weather
systems Problems

LEVEL 1

1. Soaring birds and gliders are usually observed to be continually turning. Why
do they need to do this?
2. Given that the top of a small cumulus rises at half the speed of the fastest
updraught within it, estimate the typical speeds of such updraughts from the
simple observation that many small cumulus grow to a depth of 100 m in
5 min.
3. Fractocumulus often appears at lower levels than any cloud observed before
the arrival of the overlying nimbostratus. Suggest a reason for such behaviour.
4. In a certain north-westerly flow in the northern hemisphere, showers are
observed to arrive in small groups separated by clear periods which range from
5 min to an hour. Interpret these observations in terms of the motion of a
regular pattern of open cells, and deduce their breadth given that the pattern is
moving at 15 m s -I.
5. As a cumulonimbus slides overhead it becomes very dark. Where has the light
gone?
6. An aircraft flies through a train of billows with a fairly uniform wavelength of
1 km at an air speed of 250 m s -I. Given that the billows are moving with the
ambient air, find the time period between successive jolts of the aircraft.
7. What is the minimum downward acceleration of an aircraft which can begin to
throw passengers up out of their seats?

LEVEL 2

8. A certain cumulonimbus is moving at 20 km h- 1, and is raining at an even rate

over a zone which is 2 km long in the direction of motion. Given that 5 mm of
rain is measured by all gauges passed over by the full length of the shower, find
the rainfall rate in mm per hour during the cumulonimbus' life.
9. Find the increase in temperature of air in which I g kg- 1 of water substance
behaves in each of the following ways: (a) water cloud condenses from vapour;
(b) supercooled cloud water is frozen; (c) ice cloud condenses from vapour.
10. Using the tephigram in Figs 5. 5 and 5. 6 and referring to Fig. 10.15, trace the
thermodynamic path of air moving adiabatically from a surface source in the
Mexican Highlands (temperature 30 oc, pressure 900 mbar) to the 800 mbar
level over the position X in the United States. Compare this with air which rises
in dry and saturated convection from a surface source at X with surface
pressure 1000 mbar and temperature and dew-point 26 and 23 oc respec-
tively. What do you conclude about convection extending to greater heights?
1I. Find the time period of vertical oscillation (the Brunt-Vaisala period) of an air
parcel embedded in air with a temperature inversion of I oc per 50 m.
I2. The core updraughts of some cumulonimbus are observed to overshoot the
tropopause by 500 m, penetrating a low stratosphere which has a temperature
inversion of 5 oc km -I. Find the implied updraught speed at the tropopause
in the absence of drag.
13. A large area of wave clouds is seen in the high troposphere by a satellite on a
370 certain occasion, with a coherent wavelength of about 50 km apparent for
 many wavelengths downwind from a hilly source region. Using the rule of Problems
thumb for lee wavelengths, estimate the wind speed in the high troposphere.
How would this compare with a more stable atmosphere?

LEVEL 3

14. Consider the likelihood of a report of 8 octals of cumulus humilis being

accurate.
15. Discuss in detail the role of condensation and evaporation in creating and
destroying buoyancy in small cumulus.
16. Before the outbreak of a summer thunderstorm, the oppressive air has
temperature 30 oc and dew point 24 °C. In the cool squall accompanying the
main precipitation, these fall to 22 and 19 oc respectively. By following each
up to the 700 mbar level on a tephigram (Fig. 5.5), at first dry adiabatically
and then wet-adiabatically, find the implied temperature difference there.
This is clearly a measure of the buoyancy driving the cumulonimbus. What
surface observations give us this temperature difference most directly?
17. A certain cloudy air parcel is 1 oc warmer than ambient cloud-free air. If the
cloudy and cloudless parcels mix in equal masses, and I g kg -I of cloud water
is evaporated in the process, find the resultant temperature of the mixed air
compared with the ambient air. Find the effective temperature of the mixed air
as far as buoyancy is concerned.
18. Discuss the microphysical processes at work in the upper parts of a cumulo-
nimbus as it glaciates and produces an anvil, mentioning their relevance to
buoyancy and precipitation.
19. Use the method of problem 6.20 to consider the growth of a giant hailstone of
radius 5 em in a cloud with 10 g kg -I of water substance in collectable form,
explaining how such stones can grow in a cloud which is seldom more than
I2 km tall.
20. Describe and discuss the tendency of cumulus of all types to appear in non-
random patterns.
21. A passenger aircraft flies from vertically static air into a downdraught of
20m s- 1 • Given that the aircraft adopts the new ambient motion in a hori-
zontal distance of 500 m, and the plane is flying at 300 m s -I, find its implied
downward acceleration and describe the sensations and behaviour of the
passengers. Given that the pilot finally zeros the aircraft's sink rate (by
climbing through the sinking air) just as the aircraft flies out of the
downdraught into a 20 m s- 1 updraught, repeat your calculations and
descriptions.
22. Surface air contaminated by the Chernobyl reactor fire reached Northern
England on 3 May I986 just as thunder showers were breaking out near the
centre of a slow-moving weak depression. Assuming convective updraughts of
3 m s -I I 000 m above the surface, use the method of problems 7.23 and 24 to
find the associated convergence in this sub-cloud layer and estimate the
reduction in horizontal area in 30 min. Consider the possible fate of
contaminated aerosol particles in the converging layer.

37I
CHAPTER
Large-scale weather
11 systems in middle latitudes

11.1 Historical

Although people have known for centuries that the great storms of middle latitudes
must be very large, much larger than can be appreciated from any single viewpoint
on the Earth's surface, that knowledge has remained fragmentary and elusive until
comparatively recently. While the European climate deteriorated from its
thirteenth-century optimum, there were several storms which caused very wide-
spread damage and loss of life by coastal flooding. Given the primitive communi-
cations, awareness that each of these disasters was caused by a single very large
storm developed only in retrospect. The conspicuous fall of atmospheric pressure
before the arrival of a gale was an added straw in the wind available from the mid-
seventeenth century, but it took the journalistic effort and skill of Daniel Defoe to
compile the first comprehensive account of the effects of a very severe gale, the
great storm which struck southern Britain on 26 November 1703, and that account
was completed a year after the event [71]. On 21 October 1743 Benjamin Franklin
was prevented from seeing an expected lunar eclipse in Philadelphia by the sudden
arrival of a north-easterly gale and associated cloud. He was surprised to learn later
that the eclipse had been visible at Boston, 300 km to the north-east, and that the
gale did not arrive there until the next day. His curiosity aroused, he wrote to
people living along the connecting path and was able to show that the storm
apparently moved from Philadelphia to Boston, against its constituent winds [72].
This is reckoned to be the first recognition that storms did not simply blow along
with their own winds. During the next hundred years patient reconstruction of scat-
tered observations of a few storms gradually revealed the great wheel-like
(cyclonic) distribution of winds around a low-pressure centre which is now such a
commonplace feature of weather maps, and confirmed that these wheels often
moved as quickly as Franklin's storm. (Note that the term cyclonic is now used
technically to mean wheel-like rotation anticlockwise in the northern hemisphere
and clockwise in the southern, which is the observed sense of rotation around low-
pressure centres. The opposite rotation, observed round high-pressure centres, is
called anticyclonic.) Piddington and others concentrated such behaviour into an
empirical law of storms which applied in low latitudes as well as high, and which
worked well enough to allow sailors to plot a safe course in the vicinity of a
372 potentially dangerous storm.
 11.1 Historical

Fig. 11.1 Fitz Roy's picture of

contrasting warm (dashed)
and cold (solid) flows in
cyclonic disturbances around
the British Isles. (From (73])

Until the development of the electric telegraph in the middle of the nineteenth
century, there was no point in trying to organize networks of people to observe and
possibly forecast the movement of storms over extensive areas, because many
storms moved and changed at least as quickly as the fastest available means of
communication - the galloping horse. However, the arrival of the telegraph
encouraged the formation of early synoptic networks (section 2.8), and quickly
established and extended the picture of cyclonic storms, especially those found in
middle latitudes. In Britain the young Meteorological Office flourished briefly
under Admiral FitzRoy, who years earlier had captained the Beagle as it carried
Charles Darwin around the world. FitzRoy was a good organizer and a perceptive
observer (despite his fanatical opposition to Darwin's later theories), and Fig. 11.1
shows that he realized that the contrast between warm and cold air masses was an
essential feature of the mid-latitude (i.e. extratropical) cyclone. There was a con-
siderable delay in recognizing that these contrasts were concentrated in narrow
bands of especially active weather, so that the modern picture of the extratropical
cyclone and its associated fronts did not really emerge until the rise of a small group
of gifted observers, analysts and theorists in the Norwegian meteorological service
between 1910 and 1930. Cut off from many European observations during the First
World War, they looked closely and perceptively at what observations they had
and derived what has become known as the Norwegian cyclone model (Fig. 11.2).
Although the Norwegians made good use of visual observations of middle and
high clouds to define behaviour throughout the troposphere, their model depended
almost entirely on the surface synoptic network (section 2.9). But as aircraft
became more numerous and adventurous, the need for upper-air observations
grew. The Second World War triggered the establishment of radiosonde stations
(section 2.10) on a wide scale, and these quickly confirmed the presence of a jet
stream in the upper troposphere in the vicinity of each well-marked front, and
indeed of the huge circumpolar vortices of fast westerly flow in the middle latitudes
of each hemisphere- features which had been known about previously only rather
haphazardly. The whole composite picture of structure in the low, middle and high
troposphere was magnificently confirmed as soon as meteorological satellites
began to operate in the early 1960s, but their capacity for high resolution over wide
fields of view also revealed the presence of much meso-scale structure which had 373
11 Large-scale weather
systems in middle latitudes

Fig. 11.2 The classical

Norwegian cyclone model
after Bjerknes and Solberg
(73]. The top part is a plan
view of surface conditions,
with airflow, isobars, fronts,
and areas of cloud and rain.
The bottom shows a vertical
section from A to 8, showing
flow in the plane of the
section, cloud and
precipitation. Subsequent
amendments have added the
familiar triangles and humps
on the cold and warm fronts
respectively, raised the tops
of the deepest cloud masses, Ca
added jet streams aloft 9.Km.
(Fig. 11.1 O) and altered cloud
type abbreviations (see
appendix 2.3). Ca. 500KI:2 -

slipped through the rather coarse synoptic-scale observation network. Despite

much dedicated effort by meteorologists, the meso-scale structure revealed by
satellites and weather radars is still rather poorly integrated with synoptic-scale
structure encapsulated in the Norwegian cyclone model, but the synoptic-scale
picture of events in both low and middle latitudes is now established beyond
dispute.

11.2 Extratropical cyclones

ORIGINS

The weather map in Fig. 11.3 shows a typical situation in the northern Atlantic: a
front is strung out from west to east across several thousand miles of ocean, and
along it are several large weather systems in various stages of development, each
marked by a wave-like distortion of the front. This is a family of extratropical
cyclones, each of which has a characteristic common structure and life cycle. In
fact since each moves eastwards along the extended front during its life cycle, the
sequence from west to east on Fig. 11.3 corresponds to stages in that common
374 cycle.
 11 .2 Extratropical cyclones

Fig. 11.3 A family of

depressions in the north
Atlantic at 1200 Z on 6 July
I 972. like all actual cases it
departs from the ideal, there
being no really mature
member between the old
occluding system and the
young waves to the west.
Fronts in the far north-east
are debris of the earliest
members of the family.
Notice the ridge of high pres-
sure over the British Isles
Longitude r'Wl
ahead of the warm front.

The front which links these depressions (using their familiar name) is called the
polar front in the Norwegian terminology which has become universal, and is a
region of relatively sharp temperature contrast separating the warm air of low
latitudes from the cold air of high latitudes. Various factors can sharpen the back-
ground meridional temperature gradient imposed by the unequal distribution of
solar input (section 8.7), distribution of land and sea being one of the most
effective in the northern hemisphere. For example, in the north Atlantic in winter
there is a particularly sharp contrast between the cold polar continental air in the
interior of North America, and the warm tropical maritime air over the west
Atlantic warmed by the powerful poleward flow of very warm water in the Gulf
Stream. Although such air masses are clearly defined only in the lower troposphere
(since the upper-troposphere westerlies flow freely by aloft) they are a useful
descriptive concept, because they define surface atmospheric conditions, and
because the strong thermal contrasts between them are observed to be favoured
sites for cyclogenesis - the birth of depressions. The north-western Atlantic
breeds depressions which drive across the Atlantic to arrive in a mature state in the
western approaches to Europe, and the north-western Pacific and the northern
Mediterranean (in winter) act similarly to spawn depressions is the mid-latitude
westerlies (Fig. 11.4).
Fig. 11.4 Zones of
cyclogenesis in the northern
hemisphere in (a) summer
and (b) winter. Note the
northward migration in
summer and the associated
collapse of the Mediterranean
zone, on which the local
tourist industry depends.
(After [72])

375
11 Large-scale weather The Norwegian school believed it was the polar front itself which generated
systems in middle latitudes
depressions through a wave-like instability, like a huge version of the waves on a
rippling flag. Though current theory does not support this view in detail (section
11.6), it is such a persuasive interpretation that it is widely used as a descriptive
device. Whatever the details, it is clear that cyclogenesis depends on the juxta-
position of contrasting air masses.

LIFE CYCLE AND STRUCTURE

The stages in the life cycle of a depression are shown in Fig. 11.5- another classic
diagram from the Norwegian school. In the first stage (a) the pre-existing front
begins to develop a large elongated mass of deep stratiform cloud in response to
persistent uplift throughout the troposphere. As low-level air converges, to replace
the rising air, the planetary vorticity fis concentrated (section 7.14) and the air
begins to rotate cyclonically, twisting the warm air polewards to the east of the
centre and the cold air equatorwards to the west and sharpening the horizontal
temperature gradients there, especially in the low troposphere.
These developments continue in the second stage (b), by which time large areas
of continuous precipitation have appeared close to the deforming front. Satellite
pictures (Fig. 11.6) clearly show the large area of middle- and high-level cloud,
often with a spume of cirrostratus fanning north-eastward over the crest of the
frontal wave. The rightward curvature of this high cloud shows that the flow there
is twisting anticyclonically in strong divergence above the main uplift (Fig. 7.29).
By now sea-level pressure is falling sharply near the wave crest as divergence in the
upper troposphere removes air from the depression centre faster than convergence
in the low troposphere replaces it.
Conventionally the fronts in Fig. 11.5 are marked in their surface positions. In
fact the regions of strongest horizontal temperature gradient (the frontal zones)
slope upwards at such very small angles (Figs 7.16 and 11.2) that in the upper
troposphere they lie hundreds of kilometres poleward of their surface positions. As
the depression develops, the frontal zones become more and more sharply defined,
especially in the low troposphere to the west of the crest of the frontal wave.
Temperature gradients - 5 oc per 100 km are observed where the cold polar air
undercuts the warm tropical air at the cold front, forming a very sharp thermal
boundary between the warm sector and the cold air to the north-west. The warm
front acts as a weaker eastern boundary.
Development of the depression continues with further growth in amplitude of

Fig. 11.5 The life cycle of an

........_______
-------= ~
SltoTo!!.pn er e
extratropical cyclone in the
northern hemisphere as lr oo•c ol /J. or
depicted by the Norwegian
school. (Modified from [75]}
t:3:",, ,,,,, 1 :: ~J.tt .
~ ) ~, ,.,...,...,.... ,,,, , ,~
j~ I'- --~F
::..---- ~
L ~ -

~
a(- · _ _ _

--- --

~-
~(C ~ - ~
/.

-
~ ~
~

376
~ , . ,,~
b
~~
c
0 d f:
 11 .2 Extratropical cyclones

Fig. 11.6 Infra-red satellite picture of young and old depressions in the eastern Atlantic.
A
young system still in the open-wave stage lies over Spain and France, moving north-east
along
the front trailing from the old occluded system whose comma-shap ed cloud pattern
is centred
over Scotland. Notice the large meso-scale swirl of cloud to the west of Ireland. {1917
Z 14
December 1980)
0

the frontal wave and falling of the pressure minimum at its crest. Figures 11.2 and
11.5(c) are convention ally regarded as showing the beginning of the mature stage
of the life cycle. The minimum surface pressure may now be tens of millibars below
the undisturbe d value, and surface winds may reach gale force over much of the
area of cyclonic flow. Large masses of nimbostra tus blanket the warm and cold
frontal zones, giving substantia l falls of rain or snow to the surface, depending on
season and latitude. An extensive flange of altostratus and cirrostratu s reaches 377
11 Large-scale weather
systems in middle latitudes
polewards and eastwards ahead of the warm front, enlarging the 'spume' first
noted in Fig. 11.6. Associated with the well-marked warm and cold frontal zones,
there are powerful cores of the westerly polar front jet stream in the high
troposphere. The whole situation is described more fully in section 11.4.
In the early stages the meridional extent of the frontal wave (its north-south
'height') was smaller than its zonal wavelength. Now however the two are
comparable and the warm sector is narrowing quickly, like an ocean swell
steepening in shoaling water (but remember the cyclone 'wave' is not a vertical
gravity wave like a swell). This narrowing leads to the occlusion of the warm sector,
as the cold front overtakes the warm front and forms a composite or occluded front
(Figs 11.5(d) and 11. 7). During occlusion the still-deepening low-pressure centre
moves polewards and westwards from its previously central position in the
depression, drawing out the occluded front which connects it to the rump of the
warm sector. The jet-stream core is still aligned with the trailing cold front to the
west, but not with the occluded front, which it cuts somewhat poleward of the
warm surface air (Fig. 11.18). This lack of alignment shows (through the thermal-
wind relation) that the thermal contrast across the occluded front is much less than
it was in the cold front. There is still some contrast however, since the fresh polar
airstream west of the occluded front is often colder than the 'older' cool air ahead
of it.
After occlusion the depresssion progressively loses the marked temperature
contrasts which were such important features of its growth to maturity. In fact the
centre of the circulation is now predominantly cold (giving the term cold core), and
spinning well to poleward of the nearest warm air mass. The occluded front may
continue to maintain widespread cloud and precipitation for several days, during
which its poleward end may spiral round the low-pressure centre, producing a
striking 'comma' shape on satellite pictures (Fig. 11. 7)- about the only time that
the low-pressure centre, such an obvious feature in analysed weather maps,
appears clearly in satellite pictures. The cyclonic rotation in the low troposphere
may persist similarly, maintaining strong winds for several more days, but it is like
a vast eddy shed after vigorous formation, inexorably slowing through friction and
filling as convergence below exceeds divergence aloft (Fig. 7.29).

FAMILIES OF DEPRESSIONS

The eastward and poleward motion of the depression from birth to occlusion is not
well represented in Fig. 11.5. The open waves (a) and (b) in particular may move
and develop rapidly in the prevailing westerly flow, testing the skill of forecasters,
especially in areas of sparse data like the central Atlantic. Eastward motion
continues through maturity but dies away after occlusion, leaving the frontless
swirl of Fig. 11.5(e) turning about a nearly fixed centre. However, the residue of
the cold front, which was such a prominent feature of earlier stages, lies like an
uneven string unwound from the equatorward rim of the depression as it rolled
eastwards. This residual front still contains considerable contrast between polar air
flowing equatorwards down the western side of the depression and the tropical
air at lower latitudes (often the poleward flank of a subtropical anticyclone).
This thermal contrast may encourage further cyclogenesis, beginning to wave the
front again as in Fig. 11.5(a) and passing through the same life cycle along the
track of the trailing front to merge with its decaying parent after occlusion. The
second wave in turn leaves a trailing front which can encourage the growth of a
378 third, and so on through a family of six or more depressions, all decaying into
 11 .2 Extrat ropical cyclones

Fig. 11.7 Infra-red satellite picture of an occluded depression . The warm sector remnant lies
over France, with an extensive occluded frontal cloud mass lying meridionally over the North
Sea, southern Norway and the Norwegian Sea. The great westward hook of cloud surrounds the
low-pressure centre at sea level, and lies to the north of the mottled pattern of shower clouds in
the cool westerlies . (0910 Z 4 January 1980)

379
11 Large-scale weather and re-invigorating the same final swirl begun by the founding parent. The
systems in middle latitudes
sequence comes to an end when no further cyclogenesis occurs on the trailing front,
often as a ridge of high pressure begins to build there, extending polewards from
the local subtropical anticyclone. The old vortex, cut off from further
invigoration, finally dwindles and dies, leaving little trace of the enormous weather
systems which have troubled a considerable fraction of the middle latitudes for two
weeks or more.

11.3 The mature depression

Let us examine in more detail the structure of a mature extratropical cyclone. As

shown in Fig. 11.5 (c), the mature system has well-developed cold and warm fronts
separating the warm sector from cold air to the east and even colder air flowing
equatorwards to the west. The structure is summarized in the classical plan and
vertical cross-section of the Norwegian Cyclone Model published in the 1920s
(Fig. 11.2a and b).

PRESSURE AND WIND

The low-pressure centre lies close to the apex of the surface fronts, and the isobars
form a large pattern with many closed isobars when they are drawn at intervals of
one or two millibars, as is normal practice. The isobars curve round fairly smoothly
in the cold air but run almost straight across the warm sector. The pattern of fronts
and their associated weather usually moves parallel to the warm sector isobars- a
useful rule of thumb in forecasting. The isobaric curvature absent from the warm
sector appears to be concentrated at the two surface fronts, especially at the cold
front where a deep kink in the isobars is often observed. These sharp kinks are
hydrostatically consistent with a boundary between cold dense air and warm less
dense air (Fig. 11.8) which slopes up from the position of the surface front.
Moving horizontally from the surface front into the cold air, the increasingly deep
layer of denser air produces a relatively sharp increase of surface pressure. In the
presence of the overall fall of pressure towards the low-pressure centre, this sharp
pressure gradient appears as a sharp clockwise turn of the isobars as we go from the
warm sector into the cold surface air across the surface cold front. Of course there
Fig. 11.8 Horizontal are no actual discontinuities of air density in the real atmosphere, but the
pressure patterns near a boundaries can be quite sharp on the synoptic scale, particularly at the cold front.
front. The solid lines in (a)
show the isobar pattern
associated with uniform Ia) Plan (b) Vertical section x- y
westerly geostrophic flow in
the northern hemisphere.
These change to the dash-dot
......
......
....... ::.~
tN
pattern when we add a cold P,
......
......
front as shown (b), to .........
I
. . . . i> "
Oo'~"m
accommodate the associated p2 y
wedge of denser air. ..........
......... I
.......
p3
380 p3 p2
 Associated geostrophically with these kinks in the isobars, the winds too must 11.3 The mature depression
turn sharply clockwise as we go across the surface cold front from its warm to its
cold side. Clockwise rotation of wind direction is known as veering, whereas anti-
clockwise rotation is termed backing. In the particular layout in Fig. 11.2(a) the
geostrophic wind veers from azimuth 260° to about 310° in going across the
surface cold front, which is roughly equivalent to a veer from a south-south-
westerly to a north-westerly in compass directions. Of course surface winds are not
accurately geostrophic because of friction (section 7 .11), but although reduced and
backed by the effects of friction, they too must veer sharply across the cold front.
In fact the surface position of a vigorous cold front is often the site of vicious
squalls from cumulonimbus embedded on the edge of the nimbostratus, in which
the sharp veer can be concentrated into a few hundred metres on the edge of the
squalls. This behaviour was well known to sailors in square-rigged sailing vessels,
who often had to go aloft to reset the angle of the spars, and to reef in the larger
sails, as the squally wind veer came upon them. Struggling with heavy wet canvas
cracking like gunshot under the strain of winds gusting well above gale force while
lashed by bursts of associated rain and hail (and lightning - see section 6.13)
would be highly uncomfortable in any situation, but 30 m above a deck heaving in
the confused sea produced by the rapid veer of wind, it was an unforgettable and
highly dangerous experience which exacted a heavy annual toll of death and serious
injury for the centuries in which such ships were widely used. By contrast the wind
veer associated with warm fronts (as we cross from the cold to the warm side of the
surface front) is usually smaller and more gentle. Such veers of wind and changes in
pressure tendency are commonplace at fixed observation positions as fronts drive
across from the west; and the sharp rise of pressure after the surface passage of the
cold front is a useful telltale which often enables it to be positioned accurately on a
synoptic chart.

CLOUD AND PRECIPITATION

Figure 11.2(b) represents a vertical section through the warm sector and its
bounding fronts, along the lineA-Bin the plan view (a). Working westwards from
its eastern boundary we have the sequence of observations and weather as it would
be experienced at a fixed geographical location as the structure moves across from
the west. Firstly there is fair weather in the little ridge of high pressure which
usually precedes the warm front, with stunted cumulus if there is any convection at
all. Then the outlying edge of the frontal cloud sheet is heralded by the appearance
of hooked cirrus (mare's tails), produced by patchy formation of ice cloud in the
high troposphere (Fig. 2. 7). The motion and appearance of these clouds is quite
informative. Their progression across the sky indicates the speed of approach of
the depression while their individual movement along the overall pattern reveals
the relative flow of the air in the high troposphere. According to the thermal wind
relation (section 7 .10) this air should move so that cooler air is on the left looking
downwind in the northern hemisphere. The flow should therefore be out of the
paper in Fig. 11.2(b), which is from right to left when viewed by an observer still
ahead of the approaching front, watching the hooked cirrus as they invade the sky.
This is readily confirmed by observation, since the mare's tails are usually
unobscured by lower cloud. The curved hooks are formed as streamers of the larger
ice crystals in the cloud fall into the slightly lower and slower air.
The hooked cirrus usually gives way to a film of cirrostratus which fills the high
troposphere and slightly diffuses the sunlight (Fig. 11.9). The shallow layer of 381
11 Large-scale weather
systems in middle latitudes

Fig. 11.9 The 22° halo

around the sun indicating the
presence of a thin veil of
cirrostratus which is often a
forerunner of an approaching
warm front.

highly regular, hexagonal ice crystals can produce by refraction a faint but striking
halo around the sun, whose angular radius is 22° [76]. The cirrostratus thickens
steadily as its bottom edge lowers, and by the time this has sunk down to middle
levels (below the 7 km level technically) the sun's position is barely visible. The
cloud layer now consists of cirrostratus and altrostratus, though quite often only
the altrostratus is visible from the surface. Aircraft observations suggest that such
thick layers are usually built up of several thin layers rather than a single thick one.
When seen over hilly land, there are often darker patches where layers of moist air
in middle levels form lenticular clouds in standing waves over the hills (section
10.8). Shortly after this the cloud thickens and darkens to nimbostratus and the
precipitation begins. Nearer the surface position of the front most of the lower
troposphere too is filled with cloud, so that there is a multi-layered sandwich of
cloud filling most of the troposphere and producing gloom and moderate or heavy
precipitation at the surface.
The precipitation consists of large ice crystals sifting down through the very
gentle uplift of air in the front. If the air in the low troposphere is cold enough, no
more than a degree or so above freezing, then the precipitation will fall as snow.
This tends to happen in winter if the depression is at a fairly high latitude or in a
continental interior in middle latitude, or of course if the surface is high enough
above MSL. In a maritime climate such as Britain's, even in winter the
precipitating snow normally melts at least several hundred metres above sea level
and falls as rain on low ground. The large wet flakes in the melting zone produce a
strong horizontal echo on weather radars (Fig. 6.27) known as the melting band.
There are usually large quantities of supercooled water cloud in the lower tropo-
sphere, some of which is seeded by ice crystals precipitating from above to form
meso-scale patches of heavier rain at the surface (section 10.7). Apart from these,
precipitation in warm fronts tends to be more steady than in any other type of
382 weather system.
 As precipitation falls into the sub-cloud layer it tends to moisten the layer by 11 .3 The mature depression
continual evaporation. Although there is no sun in these conditions to drive con-
vection, the strong winds and their associated shear maintain mechanical con-
vection (section 9.8) in low layers. The tumbling masses of moistened air reach
their lifting condensation levels here and there to produce the torn cloud fragments
called fractocumulus (Fig. 10.4) which are characteristic of such conditions.
Sometimes rain falling into cold air overlying the surface at the narrow end of the
wedge of cold air near the surface front will saturate the cold air to produce frontal
fog.
The vertical cross-section (Fig. 11.2b) is probably unrealistic in indicating that
the upper surface of the warm frontal cloud mass descends toward the warm air,
though such is the respect for the Norwegian model that the feature is reproduced
faithfully in many texts. Aircraft and satellite observations show that the upper
surface is fairly horizontal across the front, although there is some meso-scale
sculpting which can cast dramatic shadows in oblique illumination. The frontal
cloud mass is therefore wedge-shaped in section with maximum thickness close the
surface front.

THE JET STREAM

The Norwegians originally had little upper-air data, and had to rely on visual
observations from the surface. Though they knew that there were high winds in the
upper troposphere near fronts, they had almost no direct measurements of them,
and consequently did not emphasize this feature. However, the radiosonde
network has changed all that, and the polar-front jet stream is now an integral part
of the model, and one which is of great practical importance to the large amount of
air traffic which is routed at these levels. The core of the jet stream is presented in
Fig. 11.10 together with an outline of the patterns of Fig. 11.2. Notice that the jet
cores lie in the warm air (although at these levels there is nothing like the thermal
contrast typical of the low troposphere) roughly vertically above the positions of
the frontal zones in the middle troposphere, as is to be expected from the thermal
wind relation. Core speeds of as much as 75 m s -I are quite typical in vigorous cold
fronts. The hooked cirrus ahead of the warm front is moving on the eastern flank
of the jet core there: hence the visible equatorward motion noted earlier. Fig. 11.10 An idealized
picture of a mature
extratropical cyclone (similar
to that in Fig. 11.2) showing
plan positions of the centre of
low pressure (L), the warm
and cold fronts, areas of
stratiform cloud (hatched)
and precipitation (cross-
hatched), and the axis of the
associated polar-front jet
stream centred on the
300 mbar level , with barbs in
It is tempting to imagine from the appearance of Fig. 11.2(b) that the great regions of highest speed .

wedge of warm frontal cloud is produced by air creeping up the frontal slope in the
plane of the frontal cross-section. Though there certainly is some motion in that
plane, it is much weaker than the components of motion perpendicular to the plane
(i.e. parallel to the front)- the jet core being the extreme example. It is flow up the
slight but very extensive slopes along the fronts which begins and maintains the
production of cloud and precipitation. In the case of the cold front the air is 383
11 Large-scale weather climbing as it travels polewards (into the page in Fig. 11.2), whereas in the warm
systems in middle latitudes
front the air in the upper troposphere is climbing as it flows equatorwards (out of
the page). In fact the three-dimensional motion of the air is quite difficult to deduce
accurately from observation, and is very complex insofar as it is known, as will
appear in the next section.

WARM SECTOR AND COLD FRONT

Continuing with the westward section of Fig. 11.2, we have now reached the warm
sector. This is often largely free of substantial cloud sheets, at least clear of the
poleward extremity where stratus and even fog may form as the humid air is carried
polewards over colder and colder surfaces (a classical example of advection fog).
Being a tropical air mass, the warmth of the warm sector can be quite striking at
and near the surface. In winter this warmth is an important source of heat for
middle latitudes, and is an obvious example of the warm air advection considered
more generally in section 8.12. In summer the warmth may become positively
balmy when the sun adds to the intrinsic warmth of the air by heating the under-
lying land surface. However, on some occasions there may be much more than the
quite typical scattering of small cumulus, even to the extent of there being patches
of nimbostratus and cumulonimbus. Convective instability often increases on the
western side of the warm sector (the North American squall line being an extreme
example- see section 10. 7). When winds are strong, as they often are especially in
winter, the pleasant balmy feeling of the warm sector is spoilt for us, though the air
temperature may be just as high. The temperature contrasts between the air before,
in and behind the warm sector can obviously be masked or accentuated by local
cooling or heating, especially overland. The dew-point temperature is a much more
secure indicator of air-mass differences, since it is unaltered by simple isobaric
heating or cooling (section 5.5). The dew point is highest in the warm sector,
indicating its origin in the humid warmth of lower latitudes.
The cold front may arrive with great vigour, with heavy bursts of precipitation,
squally winds and thunder, and the wind veer and pressure kick mentioned already.
On other occasions it may approach much less dramatically, though the sky
generally fills with cloud much more rapidly than it does with a warm front.
Precipitation rates are characteristically higher and more variable than in warm
fronts, consistent with there being quite vigorous cumulonimbus embedded in the
nimbostratus, often in groups as mentioned in section 10.7, and with associated
hail and thunder.
The cold frontal cloud mass is usually considerably narrower than the warm
front because of the steeper frontal zone associated with the stronger temperature
contrast in the lower troposphere (section 7 .10). Its unpleasant weather therefore
does not last too long, though note the different behaviour in young depressions
mentioned below. The clearance at the back edge of the frontal cloud mass can be
correspondingly much more rapid than its counterpart, the advance of the leading
edge of the warm front, and occasionally offers a panoramic view of a nimbo-
stratus edge stretching from horizon to horizon as it retreats from the observer
(Fig. 11.11)- one of the few occasions when a surface-based observer gets a sense
of the vast horizontal scale of large weather systems. Facing the retreating edge,
small-scale structure such as aircraft condensation trails, etc. embedded in the
upper part of the edge moves quickly from right to left with the jet stream there.
The observer is now in the cold polar air behind the front, which is kept in a con-
384 vectively unstable state by equatorward flow over progressively warmer surfaces.
 11 .3 The mature depression

Fig. 11.11 The clearance

behind a cold front moving
south (left) over the Irish Sea
in late autumn. The
overhanging flange of high
and middle cloud reveals the
backward slope of the frontal
zone (Fig. 11.2, bottom).
Small cumulus are growing in
the cold air as it flows south
over the relatively warm sea.

These north-westerlies (as they usually are) are therefore showery, with cumulo-
nimbus hurrying along singly or in groups in the brisk flow of cool clear air.
Sometimes there are larger areas of merged cumulonimbus or nimbostratus-like
segments of cut-off occluded front which give short bouts of unpleasant weather,
and of course it may become very cold in winter as the winds veer gradually to
become northerlies.

DIVERSITY OF WEATHER

If this depression is not the last member of a family, the showers will gradually die
away as they are suppressed by the ridge of high pressure preceding the warm front
of the next wave depression advancing from the west. The observer is now about to
repeat the sequence of observation and experience recounted above as another
system moves across, but of course no two systems are the same, so that the next
one may differ very considerably in intensity, path, speed, and maturity at time of
passage. The orientation of the fronts relative to their direction of motion in
particular can have a dramatic effect on the conditions experienced at any par-
ticular geographical location. In the mature state the fronts make quite large angles
with their direction of propagation, with the result that they pass breadthwise
across any location, and so do not last very long if they are moving quite quickly, as
they usually are. However in their young state, when the frontal wave is still very
open (see Fig. 11.5(b)), they tend to move nearly lengthwise, with the result that
most of the adjacent surface is unaffected by cloud and precipitation while an
unlucky stretch passes under almost the entire length of both the warm and cold
fronts, experiencing therefore one or more days of nearly incessant rain or snow
which may fade and re-intensify as the slightly buckled fronts meander laterally
overhead.
A location in the vicinity of the great terminal vortex of the depression family,
which may persist with little motion other than rotation for a week or two, may
suffer conditions which have very little to do with the sequence associated with the 385
11 Large-scale weather mature depression. If the location is to the east of the centre of the old low, the
systems in middle latitudes
winds are persistently southerly, and periods of rain and cloud may recur as the
debris of old fronts pass over, with little thermal contrast or organized cloudy
structure. If the location is to the west a persistent showery northerly flow is expe-
rienced, with longer periods of rain or snow as frontal debris swirls round from the
east. Regular inspection of weather maps shows that a huge variety of conditions
and sequences of conditions can arise depending on the orientation of the
depression track. This has been assumed in Fig. 11.5 to be nearly westerly, but in
fact it may have any orientation, though easterly motion is rare and usually
sluggish. The classic pattern and motions depicted in Fig. 11.2 should therefore
not be allowed to limit the actual range of behaviour, and indeed the Norwegians
themselves were most careful and flexible in applying their model to the analysis of
actual situations, as are all professional forecasters.
Note that the descriptions in this and the previous section have been written for
the northern hemisphere. It is a useful exercise to re-examine them to see how they
should be changed to apply to the southern hemisphere, bearing in mind that the
direction of map rotation about the local vertical, and all related phenomena,
reverses there. Figure 11.12 outlines a depression and its fronts in southern hemi-
sphere mid-latitudes.

MSL analysis for 0000 Z 27 19 181

Fig. 11.12 Sea-level

pressure at 0000 Z, 27
September 1981, showing a
quadrant of the southern
hemisphere with high-
pressure zones in the
subtropics and high latitudes
and depressions in mid-
latitudes. Airflow is
anticlockwise round the
anticyclones and clockwise
round the depressions.

SEA WAVES AND STORM SURGES

The great wheel of strong winds of a mature depression can have obvious and
familiar effects on underlying surfaces. Trees, buildings and crops are liable to
damage in the more vigorous storms, the damage increasing rapidly as the average
winds rise above gale force (about 19m s- 1) and the strongest gusts rise much
higher. The sea surface too becomes increasingly disturbed, with waves increasing
in height, wavelength and celerity as the winds strengthen, and more and more
386 crests break into seething turbulence. This response of the sea, and the difficulties
it posed for mariners, provided early motivation for the development of 11 .3 The mature depression
meteorology, but though modern ships are much better able to withstand an angry
sea than the relatively frail and complex sailing ships of earlier times, they are by no
means immune. At the end of January 1953 a vigorous depression was passing to
the north of the British Isles. The strong winds raised mountainous seas in the
narrow North Channel between Scotland and Northern Ireland just as the car ferry
Princess Victoria was making her way westwards. An unus:~ally large wave struck
the vessel's stern and buckled the car loading door there. The engine room flooded
and the ship drifted helplessly to the south-west, calling for assistance. Sea condi-
tions were so bad that despite the best efforts of lifeboats and several searching
ships, the ferry was still alone several hours later when she foundered in broad
daylight within a few miles of the nearest coast with heavy loss of life. The cox-
swain of one searching lifeboat claimed that his boat was thrown bodily out of the
water by the steep high waves typical of storms in confined seas.
Coastlines are very significantly shaped by short periods of unusually violent
waves, and the scouring currents they indirectly drive, and this role of the great
storms has often exposed unwisely sited coastal settlements to discomfort, damage
and even disaster. Some of these difficulties also arise from an interaction between
wind and water which is much less obvious, but even more potentially damaging,
than the great wind-driven waves of a storm. This is the so-called storm surge.
As a vigorous depression moves across a body of water, the sea over a
considerable area may respond in ways which produce substantial alterations to the
water level expected on the basis of the closely predictable astronomical tide. There
are two types of response: a static one in which the sea surface behaves like an
inverted water barometer, and a dynamic one in which shallow water in confined
zones is set in vast oscillation by the moving and changing tangential wind stress.
The static effect is relatively simple in principle. If an area of sea surface is subject
to lower atmospheric pressure than its surroundings, then the surface there will rise
up and the surrounding surface sink down until there is uniform total pressure on
the highest horizontal surface which is below the water surface everywhere. If the
area of low pressure is much smaller than the surrounding area of normal atmo-
spheric pressure, then we can simply concentrate on the elevation of water in the
region of low atmospheric pressure, and ignore the surrounding compensation. We
can use the hydrostatic approximation in the water to find the elevation of the sea
surface which accommodates the fall in atmospheric pressure. It is easy to show
that because the density of water is 1000 kg m- 3 a fall of 1 mbar in atmospheric
pressure should support an elevation of almost exactly 1 em in water level. Since
minimum pressures in vigorous systems can be depressed more than 50 mbar, we
see that water levels can be raised by more than half a metre- a significant value to
those in charge of coastal defences, and likely to cause increasing concern as
average sea levels rise because of global warming (section 4.10).
The dynamic effects can be even larger. If, for example, a northerly gale blows
into the constricted North Sea between North Europe and the British Isles, such as
may happen as a vigorous depression moves eastwards over Scandinavia, then the
large area of southward tangential stress sets up an insistent water movement which
may subsequently affect the whole region of coastal waters down to the coast of
France. The water motion is actually a ponderous wave of extremely long wave-
length, like the slopping movement of water in a shaken wash-basin, and like that
slopping it can become very large if there is resonance between the imposed shake
and the natural period of the water basin. Such time periods are many hours long in
substantial water bodies like the North Sea, and in that time the turning of the
Earth imposes a movement to the right in the northern hemisphere, just as it does
on large-scale atmospheric flow. The result is that in the case of the North Sea, the 387
11 Large-scale weather
systems in middle latitudes

Fig. 11.13 Heavy seas at

Morecambe (north-west
England) during the storm
surge of 26 February 1990,
which caused the damage
shown in Fig. 4. 16(b) and
extensively flooded Towyn
(North Wales). The huge burst
of water is caused by collision
between incident and
reflected waves, from which
high winds whipped water
inland over the sea wall and
road.

surge of highest water moves south along the east coast of Britain before swinging
anticlockwise around the Belgian, Dutch and Danish coasts. The height of the
surge varies considerably from place to place along the coast because of subtle
interactions with coastal shape and bottom topography.
Just such a storm surge occurred on I February 1953, as the vigorous depression
whose gales had sunk the Princess Victoria on the previous day moved eastwards
toward Norway. The surge concided with the peak of a high spring tide as it moved
down the east coast of Britain. The extra 2-3 m of water topped by an angry sea
breached the sea defences of East Anglia and the Thames estuary, before going on
to inundate an appreciable fraction of Holland, drowning several thousand
people. The devastation and loss, and the realization that it could easily happen
again, spurred great improvement to sea defences, including thirty years later an
adjustable barrage in the Thames to protect the heart of London, and a whole
range of investigation which has made the February 1953 storm surge the most
intensively studied event of its type, though much less destructive than the cyclone
surges of the Bay of Bengal (section 12.4).

11.4 Three-dimensional airflow

Let us consider the paths (trajectories) traced out by air parcels as they move in
large-scale weather systems. Horizontal winds are so much stronger than vertical
ones that you might suppose that motion is effectively horizontal, and that we can
find the trajectories of moving air parcels simply by examining conventional hori-
zontal or isobaric weather maps and using the isobars or contours as streamlines of
flow. For example, air moving in the low troposphere around the equatorial side of
a mature depression would be expected to describe a cyclonically curved path deter-
mined by the combination of the instantaneous pattern of isobars (as in Fig. 11 .2)
and the rate of eastward translation of that pattern. However, two points should
make us wary of such an approach: air approaching a front near the surface
388 obviously cannot pass through it as isobaric motion apparently requires, since no
available process is able to convert cold air into the warm moist air or vice versa; 11.4 Three-dimensional
airflow
and in the vicinity of fronts especially, wind speed and direction vary so strongly
with height that quite small height variations could transform the shape of air
trajectories (compare the north-westerlies near the surface just behind the cold
front in Fig. 11.2 with the south-westerly jet stream only about 10 km above the
same location).
Large-scale vertical motion would indeed be insignificant if the atmosphere had
comparable vertical and horizontal scales, but in fact the atmosphere is so grossly
flattened that the weak vertical movements are just about as important in relation
to the shallow depth of the troposphere as are the relatively strong horizontal winds
in relation to its considerable breadth. For example, a typical updraught in a front
is estimated below to be about 10 em s - 1• At this rate it would take an air parcel
about 28 h to rise 10 km (i.e. through most of the depth of the mid-latitude tropo-
sphere), which compares very closely with the time taken for an air parcel moving
at 20 ms - 1 to traverse a system with horizontal dimension 2000 km. It seems that
large-scale weather systems arrange themselves so as to equalize the time periods
involved in vertical and horizontal traverse by constituent air parcels, which is a
feature which distinguishes the genuinely large-scale system from a large conglo-
meration of small-scale systems, such as an extensive field of cumulus and cumulo-
nimbus for example. In the present context this means that relatively slight vertical
movements are just as important as the much more obvious horizontal ones if we
wish to examine the three-dimensional motion of air in large-scale weather
systems.
The large-scale vertical motions (remember there are embedded small-scale
movements as well - section 10. 7) in large weather systems are very small indeed,
and we must consider briefly how they can be determined. It is quite inconceivable
that a rate of rise of a few centimetres per second could be measured by a vertical
anemometer several kilometres above the surface in the disturbed conditions
typical of an active front. In fact such values have to be estimated by indirect
methods, one of the most physically obvious of which uses observed background
rates of rainfall. Warm fronts in particular produce fairly steady precipitation
rates for long periods over extensive areas, in which much more vapour is
condensed and precipitated than is instantaneously present in the full depth of the
troposphere. It follows that there must be a reasonable balance between the
upward flow of water vapour in the rising air and the downward flow of condensed
water in the precipitation. As shown in appendix 11.1, a typical background rate of
rainfall (3 mm per hour) is consistent with an uplift speed of just under 10 em s- 1,
assuming a realistic vapour content in the low troposphere. This is a useful order-
of-magnitude estimation, but it assumes too much to be used accurately in any
particular case.
Dynamic reasoning can be applied to observations of large-scale convergence
and divergence etc. to calculate the associated large-scale rates of rise or sink of air
(implicit in section 7.13 and Fig. 7 .29), with similar results. However, this requires
a subtle approach if it is not to demand impossibly high precision from the basic
observations. Such methods underlie the forecasting of rates and amounts of
precipitation, though it is only quite recently that numerical forecasts have become
sophisticated enough to provide usefully accurate predictions of rainfall by such
dynamic methods. However, a thermodynamic method has been used from time to
time in recent decades to gain glimpses of large-scale vertical air motion, and with it
a picture of the three-dimensional flow in large systems. Though used for research
rather than routine forecasting, it is relatively straightforward, at least in principle,
and depends on the assumption introduced in section 5.6, that vertical motion in
large-scale weather systems is sufficiently fast for the procession of air parcel states 389
11 Large-scale weather to be nearly adiabatic (often termed isentropic in this context since entropy is con-
systems in middle latitudes served in an adiabatic process). When the air is unsaturated the process is nearly
dry adiabatic, and when the air is cloud-filled the process approximates to a satu-
rated adiabatic. These assumptions have already been used in the analysis of
cumulus convection, where the air trajectories are effectively vertical, but now we
are applying them to air motions which are very slightly but extensively sloped.
In the procedure of isentropic analysis an array of radiosonde data from the
weather system under examination is analysed to pick out winds and other observa-
tions on each of several surfaces of constant potential temperature (isentropic
surfaces). The most generally conservative label to use is the wet-bulb potential
temperature (section 5.11 ), though in practice the ordinary potential temperature is
used in cloudless parts. Figure 11.14 outlines a simple vertical cross-section
through isentropic surfaces. Higher values of potential temperature correspond to
greater heights at any location, otherwise the air would have a lapse of potential
temperature and be explosively unstable (section 5.12), and isentropic surfaces
slope upwards towards the cold air in the presence of a horizontal temperature
gradient, since potential and actual temperatures on an intersecting isobaric
surface must fall towards the cold air.

Fig. 11.14 Idealized vertical

section showing how sloping
synoptic-scale flow con -10km
produce a layered vertical
profile at X, as observed by
radiosonde or equivalent. In
reality, such flows are always
three-dimensional, as shown
in Fig. 11.15.

The horizontal flow (the winds) on an isentropic surface can be assumed to

represent the flow in the tilted plane of the surface because the angles of tilt of such
surfaces are observed to be very small ( < 1°) in large-scale weather systems, even in
the vicinity of fronts. However, it would usually be misleading to use the instan-
taneous winds directly as streamlines of flow since the isentropic surfaces in most
weather systems move with much the same speeds as the systems themselves (-
10 m s -t in the case of many depressions). Just as in the analysis of flow in a super-
cell storm (section 10.6), the system velocity is therefore subtracted vectorially
from each observed wind velocity to obtain the wind velocities relative to the
moving weather system, and provided the system is not changing intensity or shape
too rapidly, these can be used to draw streamlines of airflow in the travelling
system. The provisos are obviously met best in mature systems, and as a result it is
mature systems which are usually investigated by such isentropic analysis.
Figure 11.15 depicts the typical airflow in a mature depression, as revealed by
isentropic analysis of many systems in North America and the north Atlantic [77].
The major feature is the extensive flow of air upwards and polewards in and ahead
of the advancing cold front, rising from the low troposphere in the warm air deep in
the warm sector, and turning gradually eastwards in the high troposphere in the
forward and poleward flange of the warm front. This has become known as the
warm conveyor belt because the air seems to be rising along an invisible ramp fixed
in the moving system, and because this is the warmest of a number of such flows.
Since the air entering the conveyor belt is quite moist, it does not travel far up the
slope before it reaches saturation and fills with stratiform cloud. The air continues
rising, but rather more steeply now that it is moving on a saturated adiabat, and it
390 soon becomes part of the nimbostratus deck of the cold front. The crowding of the
 11 .4 Three-dimensional
airflow

Key
Low, middl e Fig. 11. 15 Major sloping
........, cloud
airflows and cloud decks
~H i g h cloud associated with a mature
+ WCB edge depression. WCB and CCB are
respectively the warm and
- . . - Over laid
cold conveyor belts.
-£:.- CCB edge
Intersections of climbing
WCB
-);;r Overlai d flows with isobaric surfaces
- · - · Isobar in CB are labelled in millibars. The
----Overlai d coldest air in the system is
descending equatorwards off
the western edge of the
diagram. {After (76])

isobars indicates its relatively rapid ascent through the lower middle troposphere
(around the 750 mbar level), where the endless production of cloud maintains the
background precipitation of the cold front. Ascent continues into the upper parts
of the warm front, contributing to some of the precipitation from upper levels and
maintaining the anticyclonically curving plume of cirrostratus noted in Fig. 11.6.
Notice that the updraught speed of air at any position on the trajectory is given by
the product of the relative wind speed and the trajectory slope. Typical values in the
middle troposphere are 20m s- 1 and 1/too, giving an updraught of 20 em s- 1 ,
which is consistent with values found by other methods already mentioned, bearing

Fig. 11.16 Haze top at the

base of the subsidence
inversion of a summer
anticyclone, made especially
stark by urban pollution
(Montreal is virtually invisible
under the haze) and by being
viewed horizontally from a
descending aircraft.

391
11 Large-scale weather in mind that in the present method we are concentrating on just one slice of the
systems in middle latitudes
rising air, and probably the most rapidly rising one.
In the low troposphere ahead of the warm front, air ascends in the cold conveyor
belt, initially parallel to the warm front, but then swinging round in a grand anti-
cyclonic arc which sometimes produces a pronounced flange of middle-level cloud
poleward and westward of the apex of the warm sector. The north-westward flow
of the cold conveyor belt along the warm front might seem to disobey the thermal
wind relation, which requires that air flows south-eastwards there, but remember
that the flow in Fig. 11.15 is relative to the eastward-moving depression, and that
relative westward flow will occur whenever the air ahead of the warm front moves
eastwards more slowly than the front itself. Since the front and the whole system
tends to move more or less in step with the air in the middle troposphere, the
slower-moving air in parts of the low troposphere is being overtaken and appears as
relative easterly flow in the pattern of relative flow. The cold conveyor belt con-
tributes particularly to the production of cloud and precipitation in the lower-
middle troposphere of the warm front. Other features of the flow in Fig. 11.15 are
the largely horizontal flow of air in the high troposphere, catching up rapidly with
the system and sinking slightly as it flows from the north-west, and then swinging
cyclonically and rising a little to join the north-western flank of the south-westerly
jet stream associated with the cold front. Notice that the air in the middle tropo-
sphere well behind the cold front is sinking and turning cyclonically as it flows to
lower latitudes. This is part of the equatorward sink of cool air which, together
with the poleward rise of warm air, accomplishes the meridional and vertical
exchange of air which is required for hemispheric energy balance (section 8.12).
It is clear from this brief examination that the flow pattern in a depression is
quite different from what might have been deduced from a series of isobaric or
horizontal charts. In particular it is more complex and asymmetric, with the cold
front looking simpler and more fundamental to the basic scheme (the meridional
and vertical exchange of air) than the warm front. In fact the pattern is inescapably
three dimensional, which greatly complicates any attempt to portray it on a two-
dimensional page. The presence of gently sloping flows is obviously consistent with
the term slope convection used in Chapter 8 to describe this vast and structured
combination of convection and advection.

11.5 Anticyclones

Even casual barometric observation in mid-latitudes shows that fine weather is

associated with high (especially rising) pressure, just as wet and windy weather is
associated with low (especially falling) pressure. By the early nineteenth century it
had become clear that high surface pressure occurs in the middle of anticyclonically
rotating air masses (section 7 .17), and the name anticyclone began to be used.
Although much less intense than cyclonic weather systems, anticyclones are
nevertheless distinctive weather systems in their own right, with characteristic
structure and behaviour, and several variants of the basic type. The basic features
of most types of anticyclone, including the slowly subsiding air which fills its broad
core, and the subsidence inversion which separates this from the local convecting
boundary layer, are most clearly seen in the subtropical anticyclone, and are fully
392 described in section 12.1.
RIDGES 11 .5 Anticyclones

In middle latitudes the ridges of high pressure which separate members of a family
of depressions can be regarded as relatively narrow poleward extensions of the
subtropical high along whose flanks they apparently slide (Fig. 12.1). As seen from
a fixed location, the poleward limb of subsidence stunts and then suppresses the
showers in the cool air behind the previous cold front, often giving a few hours of
cloudless sky before the advance of hooked cirrus from the west heralds the
approach of the next warm front (Fig. 11.3). In the summer in particular, the
associated brief sunny spell, combined with the fresh, clean low level air, can
produce some of the most pleasant weather experienced in the industrial areas of
Britain and Western Europe, now that more persistent anticyclonic weather is so
often marred by thick industrial haze in the surface layer. But the effect is always
short-lived, as the ridge moves on in step with the preceding and following
members of the family of cyclonic systems.

BLOCKS

Sometimes an anticyclone may form in the mid-latitudes well to poleward of the

normal position of the subtropical highs. The development is associated with a
temporary halt to the normal eastward procession of depressions in middle lati-
tudes, with the result that it has become known as a blocking high or block.
Once established, a block may persist virtually motionless for days and even
weeks, causing large departures from the seasonally typical climate over a wide
area, as mentioned in section 4.9. In winter in particular, the presence of a
pronounced, low-subsidence inversion can trap smoke and other pollution from
large industrial regions and increase surface concentrations of noxious materials to
the point of irritation and even danger. Combined with a deck of stratocumulus
thick enough to reduce considerably the weak winter sun, the result has been
appropriately nicknamed anticyclonic gloom. The situation can be considerably
worsened if fog forms, as discussed in section 9.12. In summer the enhanced con-
vection tends to maintain deeper convecting layers, but encourages the photo-
chemical reactions underlying the Los Angeles type of smog (Fig. 11.16 and section
12.1).
Climatic anomalies can be very marked in the vicinity of a block. For example,
the whole of the British Isles lay under and slightly to the east of a block throughout
January, February and part of March 1963, and the combination of loss of warm
air advection, absence of thick cloud cover (which normally shields the surface
from cooling through the long winter nights), and the advection of very cold air
from the interior of Europe, reduced the monthly mean temperatures over much of
southern Britain by about 4 °C. Populations of small birds were decimated,
transport and industry were disrupted, and the eastern sides of evergreen shrubs
bore brown scorch marks for a decade afterwards. In summer the presence of a
block is likely to lead to unusually high temperatures, since solar warming through
the long, bright days far exceeds the net long-wave cooling. In July and August
1976 the British Isles lay under a block which came at the end of a period of more
than a year in which rainfall had been consistently well below the normal because
of an unusually high incidence of blocking. With reservoirs not fully restocked by
the rather dry winter, and the prolonged excess of evaporation over precipitation in
the hot, dry summer, Britain's normally very heavy domestic and industrial
consumption of water could not be sustained, and an unusually severe drought 393
11 Large-scale weather ensued. Variations in the frequency of blocking are believed to underlie some of the
systems in middle latitudes
climatic variations observed in recent centuries and decades (section 4.9).

CONTINENTAL HIGHS

Anticyclones develop regularly in winter over continental interiors, the most

pronounced being the Siberian High which extends over Siberia and northern
asiatic Russia, and influence a considerable portion of the Asian continent in
winter. These are known as continental highs, but though they exhibit the usual
anticyclonic circulation in the low troposphere (as they must to be geostrophically
consistent with the pressure pattern) they differ otherwise from the other types of
anticyclone. In particular they are relatively shallow systems, rarely extending their
anticyclonic circulation above the middle troposphere. Radiosonde profiles
through the centres of continental highs show that air temperatures are substan-
tially lower than those of the surrounding air masses throughout the low tropo-
sphere, rising only slowly from the very low temperatures of the snow-covered land
surface. It follows that, in contrast to the majority of warm-cored types of anti-
cyclone (page 147), the pressure excess in the centre of the cold-cored type falls
rapidly with increasing height.
In early winter, the cooling of the land is greatly accelerated by the first snow
falls, which practically eliminate the solar warming while maintaining the net loss
by terrestrial radiation. The surface and overlying air cools progressively day by
day, as in an extended night. The whole mass of overlying air sinks with the
shrinking layer of cooling air near the surface, and it seems that the air aloft is able
to converge faster than the air near the surface is able to diverge, with the result that
the surface pressure rises. After being set up, the surface pressure remains high
despite a dynamic equilibrium between convergence and divergence, and as in the
case of the other types of anticyclone, this is not easily explained. Despite persistent
subsidence, the absence of an underlying convective boundary layer prevents the
appearance of anything like a subsidence inversion. In fact there is simply a rela-
tively smooth inversion which can extend from the surface for kilometres into the
troposphere, once the high has become well established (Fig. 11.17). The sub-
sidence is sufficient to keep the low troposphere free of cloud, which has the effect
of enhancing the surface cooling and making the anticyclone self-maintaining.
Cloud and precipitation can occur in the unaffected middle and upper tropo-

Fig. 11.17 A vertical

temperature profile in the
Siberian winter anticyclone
(Yakutsk, 62 °N, 129 °E),
showing a very strong and
deep surface-based
temperature inversion. The
standard British tephigram
has had to be extended to
include the very low
temperatures observed in the
low troposphere. (After [78])

394
sphere, but the cyclonic weather systems which would be the primary source of 11.6 Waves in the
cloud are discouraged by the inert mass of cold air replacing the normal boundary westerlies
layer, so that continental interiors tend to be largely cut off from the westerlies for
much of the winter. The persistence of continental highs throughout much of the
middle and latter parts of the winter in Siberia gives rise to extremely low mean
surface temperatures, which are lower even than Antarctic values when accen-
tuated by local topography (section 9.11).

11.6 Waves in the westerlies

The great mid-latitude weather systems look like flat waves on weather maps,
waves which almost always travel eastwards as they develop. The Norwegian
school recognized this, and tried to look for their origins in instability of the Polar
Front, similar to the Kelvin-Helmholtz instability which makes a flag flap in
response to a difference in wind speed on either side of its fabric (section 10.8).
That particular theoretical approach made little headway against the many
subtleties which make large-scale atmospheric waves among the most complex
phenomena to be treated by fluid dynamics (probably second only to turbulence in
this respect). It is beyond the scope of an introductory text to dwell on such matters
but, for example, careful analyses of weather maps show that the wave-like dis-
turbances of the great midlatitude weather systems usually move at different
speeds at different altitudes, so that it is not strictly possible to describe any one
system as a single wave. However, the general idea of treating such weather systems
as waves (though not waves on a narrow frontal zone) has proved very fruitful since
the late 1930s, as perusal of modern texts on dynamical meteorology will confirm
[44]. In this section we will briefly consider some of the most obvious properties of
large-scale waves in the westerlies which have emerged from such studies.

CYCLONE WAVES

Previous treatment of the mid-latitude depression has given many examples of its
wavy structure, especially in its formative stages (Fig. 11.5). This is especially
obvious when isobaric contours are plotted in the middle and upper troposphere,
and indeed a sequence of such pictures of a developing depression creates an irresis-
tible impression of an amplifying travelling wave (Fig. 11.18). Theoretical treat- Fig. 11 . 18 A sequence of
ment of the dynamics of the baroclinic westerlies by J . Charney and E.T. Eady [80] simplified pictures of flow ot
the level of the polar-front jet
in the late 1940s showed that they are unstable in a way which leads to the stream during the
formation of waves which, at least in their early stages, strongly resemble those development of a mid-
seen in weather maps of the middle and upper troposphere. latitude depression_
Sequences at all levels down
to the lower troposphere
3 similarly show on amplifying
wavy disturbance of mainly
westerly airflow. (After [79])

395
11 Large-scale weather The theoretical methods used are subtle and complex in mathematical detail, but
systems in middle latitudes in essence they explore the stability of the initial unwaved flow by imposing very
small-amplitude meridional perturbations and looking for subsequent develop-
ment. They show that wavy perturbations grow exponentially, and at an initially
realistic rate, in a range of wavelengths (i.e. west-to-east separations of adjacent
troughs or crests) centred on the few thousand kilometres confirmed by inspection
of weather maps. One of the most important destabilizing factors is the vertical
shear of zonal wind shear in the initially undisturbed flow. When this exceeds
about 1 m s -·I per kilometre, a range of unstable wavelengths opens up and widens
with increasing shear. Now according to the thermal-wind relation (section 7.1 0)
such a shear is associated geostrophically with a layer-deep temperature gradient
across the flow, i.e. with the north-south baroclinity ofthe zonal flow, and for this
reason this type of instability is known as baroclinic instability. In fact you can
quickly show that a shear of the critical value is associated with a horizontal
temperature gradient of about 1 oc per 400 km in middle latitudes, or about 5 oc
across the middle 20° of latitude. This is considerably less than the temperature
gradient maintained by the meridional gradient of net radiative warming, so that it
appears that the troposphere in its normal state is baroclinically unstable in middle
Fig. 11.19 A vertical zonal
section through the zonally latitudes.
displaced axes of waves of
temperature and pressure
(and therefore flow) in Eady's
model of the troposphere of a
developing depression. The 1 Mid troposphere
arrow shows the direction of I
I
wave propagation. (After
[80]) West Honzontal East

As already mentioned, the details of the theoretically produced waves have

certain observably realistic features, one of the most important of which is the
offset or phase shift between the waves in the patterns of temperature and flow.
Figure 11.19 is taken from Eady's treatment, and clearly shows that the crest of the
thermal wave (i.e. the north-south axis of the warmest air) lags behind the crest of
the wave in the mid-tropospheric isobaric contours (which represents the flow
pattern there through the geostrophic relation), by one quarter of a wavelength.
This has the important consequence that warmer air is being carried polewards and
cooler air is being carried equatorwards, so that the net effect of the system is to
contribute to the poleward advection of heat required to balance the meridional
heat budget. This phase shift is observable in many ways in the real atmosphere. It
is directly observable in pictures such as Fig. 11.6, where the great plume of cloud
indicating the presence of the warm conveyor belt pours north-eastward on the
western side of the wave in the middle tropospheric flow. It is also indirectly
observable in the westward shift of wave axes with increasing height (Fig. 11.20)
which appears in upper air charts of isobaric contours. For example, the presence
of a deep layer of warm air to the west of the sea-level ridge of high pressure means
that the isobaric surfaces aloft are domed upward over the warm layer so that the
axis of the ridge in the middle and upper troposphere lies hundreds of kilometres
westward. And the deep layer of cold air to the west of the low-pressure centre at
sea level similarly means that the axis of low pressure is shifted westward aloft. This
has been a familiar feature of upper air charts since they were first drawn routinely,
and was noticed but not recognized by observers in the nineteenth century who
reported that barographs from stations on mountain tops (which were then quite
popular) lagged behind those nearby at sea level in indicating the passage of a
396 pressure trough. The westward tilt of the axes of both theoretical and observed
 11 .6 Waves in the
westerlies

Fig. 11.20 Westward shift of

low-pressure axis with
increasing height. The surface
low pressure is centred at the
apex of the worm centre
while the axis of the
500 mbor trough is located
about 1200 km further west.
(After (72])

systems is very pronounced, the axes making angles of only a degree or so to the
horizontal.

LONG WAVES

Another factor which greatly complicates the identification of weather systems

with individual waves is the simultaneous presence of several identifiable wave-
lengths in any particular observed situation. There are indications that there are
smaller wavelengths associated with the baroclinic waves mentioned so far, which
may relate to some meso-scale structures observed in frontal cloud and precipita-
tion, and may even represent a frontal instability of the type suspected by the
Norwegians. However, there has been no doubt for several decades that there are
waves which are considerably longer in zonal wavelength than those associated
most directly with individual depressions and ridges. These are known generally as
long waves or planetary waves, but are sometimes named after Rossby who first
discussed their significance and origin, though strictly speaking the term Rossby
wave is now reserved for a particular kind of long wave.
Long waves do not depend on the presence ofbaroclinity, although they may be
modified by it. But they do depend crucially on the tendency of absolute vorticity
to be conserved in very large-scale flow in the middle troposphere (where there is
little horizontal divergence for reasons discussed in section 7.13 and apparent in
Fig. 7.29 and 12.5). On such vast scales there is little relative vorticity about the
local vertical, so that the main component is the planetary vorticity, which is the
Coriolis parameter /(section 7 .14). If we now imagine the very large-scale flow to
be perturbed so that it is angled slightly poleward, the air will move to higher and
higher latitudes, where/is larger and larger. To conserve absolute vorticity the air
develops balancing negative relative vorticity, which at its simplest means anti-
cyclonic curvature (Fig. 11.21). The flow therefore curves back towards its original
latitude, overshoots it and then curves cyclonically back again as it develops
positive relative vorticity. The oscillation about the original latitude continues
indefinitely in the absence of any damping, and we can identify the zonal separa-
tion of adjacent crossing of the initial latitude as the wavelength L. It seems clear
that the meridional variation of the Coriolis parameter must play an important
part in determining the wavelength and other features of these long waves. The rate 397
11 Large-scale weather
systems in middle latitudes - --------- L -----------

Fig. 11.21 Idealized plan of ~ ~ l •neol latitude

a Rossby wave with 7 ~
wavelength L in westerly
flow.

of change of/with northward distance along a meridian is represented by {3, and is

shown in appendix 11.2 to have values of about 1.6 x I0- 11 m - I s -lin middle lati-
tudes. In the specially simple case analysed by Ross by, there are no other important
factors, and relatively straightforward analysis shows that such Rossby waves
propagate westwards relative to the prevailing air flow with wave speed (i.e. speed
of propagation of wave shape) {3 L 2!21r. This means that the wave speed relative to
the Earth's surface is given by

C= u- _p_£_
4 7r2

where U is the speed of the initially uniform westerly flow of air. The form of this
relationship shows that Rossby waves propagate eastwards over the Earth's
surface at speeds which decrease with increasing wavelength, and that there is a
certain wavelength at which the waves are stationary, given by
L = 27r { U/{3 }'1'
If the zonal flow speed is assumed to be 20 m s - I , then it follows that in middle
latitudes the wavelength for stationary waves is about 7000 km, four of which
would girdle the Earth there. Even longer wavelengths will propagate westward.
Observation is not straightforward because of the somewhat shorter baroclinic
waves with which middle latitudes abound, but when the flow in the middle or
upper troposphere is averaged over a few days, the fast-moving baroclinic waves
are largely smoothed away and the slow-moving long-wave pattern is exposed
(Fig. 11.22). These shift majestically with time, usually moving eastwards at a few

Fig. 11.22 Contours of the

500 mbar surface averaged
over seven days to reveal the
underlying pattern of long
waves. The surface fronts at
on instant in the middle of the
period comprise a series of
cyclone waves associated
with the long-wove trough.
(After [42])

398
degrees of longitude per day, but transforming from time to time by extensive 11.6 Waves in the
decay and regrowth in different meridians. westerlies

TOPOGRAPHICAL LOCKING

There is clear evidence that there are preferred positions for long waves in relation
to large-scale dispositions of land masses and topography, this being most obvious
in the northern hemisphere. One mechanism for such linkage can be seen by
considering the behaviour of large-scale zonal air flow as it impinges on a long
meridional barrier like the Rocky Mountains (Fig. 11.23). As the air reaches the
western (upwind) side of the mountains, it is progressively squeezed into a thinner
layer by the rising land surface beneath. The squeezing effect is reduced by bodily
ascent of the whole atmosphere, and by ponding on the upwind side of the barrier,
but these cannot eliminate the effect completely. Since the air is incompressible on
this large scale, the vertical squeezing is compensated by horizontal divergence
(section 7 .14), which in turn acts on the underlying planetary vorticity to produce
anticyclonic curvature (i.e. southward in the northern hemisphere) in the
previously westerly flow. When the air passes over the crest of the mountain ridges,
the falling topography enforces large-scale horizontal convergence which in turn
produces cyclonic curvature, tending to re-establish the initial absolute vorticity.
However, the air is now at a somewhat lower latitude, where the planetary vorticity
fis smaller, so that the air is left with residual cyclonic relative vorticity in the form
of cyclonic curvature. There is consequently a lee trough on the downwind side of
the mountain barrier (Fig. 11.23). Beyond this the air moves polewards, losing
cyclonic curvature as it moves to regions with larger f, and beginning to curve anti-
cyclonically as it passes the latitude of its original flow west of the barrier. The flow
is now in exactly the situation considered at the beginning of the treatment of
Rossby waves above, and will behave accordingly, with long-wavelength
undulations apparently anchored to the mountain barrier. In practice such
anchorage appears only statistically, there being a somewhat greater than average
concentration of long-wave troughs just east of the Rockies.
Very significantly, all long waves seem to exert a controlling influence over the
faster-moving baroclinic waves both by steering and amplitude control. For
example, there is a tendency for baroclinic waves to be suppressed to the west of a
long-wave trough, and enhanced to the east, the latter probably corresponding to a
zone of repeated cyclogenesis. And once formed, the depressions tend to move as if
steered by the contours of long-wave patterns in the middle troposphere.

Eleva1ion Fig. 11.23 Formation of a

large-scale trough in the lee
of o meridionally extensive
mountain range.
Plan

+-E
N

399
11 Large-scale weather soE SOE
systems in middle latitudes Low index
~~

" '-,
""
Fig. 11.24 Extremes of
index cycle os shown by two
18 0 t------'1.--+tt-''t-+-+--'<::::±::2"...__,:-'t-"..,__---l+-i---1 0 180 I
-
\\ ~0
examples of seven-day
averages of sea-level J
pressure in the northern
hemisphere. In the low-index
example, the high pressures
I
are centred over North
America and Central Russia. ~~

(After [81]) sow sow

INDEX CYCLE

On an even grander scale, the long-wave pattern round the whole hemisphere seems
to vacillate (i.e. oscillate irregularly) between two extremes: one of maximum
meridional amplitude of waviness, and the other of minimum amplitude
(Fig. 11.24). The maximum sinuosity is usually associated with one or more well-
developed blocks under the nearly cut-off long-wave ridges, and the minimum
corresponds to unusually straight zonal flow over extensive regions, with fast-
developing baroclinic waves moving quickly east in the vigorous westerly flow. The
long-wave set-up is quantified by the zonal index which is defined to be the zonally
averaged pressure difference across an agreed meridional segment of middle
latitudes at some particular level- for example between latitudes 35° and 55° at
the 500 mbar level. The index corresponds to an average geostrophic wind in the
zone and is sometimes expressed as such. Regardless of units used, it is small in
contorted flow, and large in zonal flow, and the vacillation from one extreme to the
next (say from minimum to minimum) is called an index cycle. Such cycles have
larger amplitudes in winter than in summer, and generally last between 3 and 8
weeks. They clearly correspond to significant events on the very largest possible
scale of tropospheric flow - variations in the huge circumpolar vortex. And they
can extend their influence far above the tropopause by complex vertical trans-
mission of wave energy, but are not yet susceptible to forecasting of even the most
rudimentary kind. Since the periodicities of the zonal index, and the adjustments
of very large-scale flow in middle latitudes which they represent, often occupy a
significant fraction of a season, they probably account for a considerable part of
the year-to-year variations which are such a feature of short-term climatic
variations.

11.7 Polar lows and heat lows

It seems likely that there are several different types of low-pressure system at work
400 in mid-latitudes in addition to the extratropical cyclone described by the
Norwegian cyclone model, though the latter is undoubtedly the most common and 11.7 Polar lows and heat
lows
significant system. As an example let us consider one quite dynamic type, the polar
low of the eastern Atlantic, and the relatively static and ephemeral heat low.

POLAR LOWS

In the eastern Atlantic and western European region there is a type of weak low-
pressure system which occurs occasionally in the winter and is well-known in
Britain for its ability to produce disruptive snow falls. A classic example appears in
Fig. 11.25, which shows the typical synoptic situation in which they occur. A fairly
gentle northerly flow, to the west of an old stagnant complex low over Scandinavia,
is bringing very cold air off the Greenland ice cap and over Iceland to reach the
British Isles from the north-north-west. Weak surface lows form in the vicinity
of Iceland and move south-eastwards with the air in the low troposphere, bringing
substantial amounts of nimbostratus, cumulonimbus and quite heavy falls of snow
to the British Isles, before moving on into northern France and weakening.
It was once thought that these systems were essentially conglomerations of fairly
shallow cumulonimbus, but fairly recent analysis [25] of this particular situation,
amongst others, suggests that they are shallow baroclinic waves forming in
strongly baroclinic northerly flow. Though there are snow showers from cumulo-
nimbus in the wake of the system, just as there are showers in the wake (i.e. to the
west) of a normal depression, these seem to be secondary to the mass of nimbo-
stratus associated with the shallow low. A sophisticated radar, capable of
measuring the motion of precipitation from the Doppler shift of reflected radar
waves, indicated the presence of meso-scale convergence (and therefore ascent -
see section 7 .13) in the low troposphere, but failed to detect any updraughts strong
enough to suggest the presence of cumulonimbus embedded in the nimbostratus.
Isentropic analysis of the flow revealed a flow structure rather like that found in
normal depressions, but confined to the lower half of the troposphere, limited to
little more than 500 km in horizontal extent, and of course moving southwards
instead of eastwards. If there were fronts they were too weak and on marginally too
small a scale to show on the radar or in the standard synoptic observations which
were also used.
Theoretical analysis [83] confirms that the shallow but strongly baroclinic zone
between the cold airflow off the Greenland ice cap and the much warmer air to the
west is capable of generating baroclinic waves of the observed dimensions and
speeds of propagation, and at the observed rates, provided that wind speeds are not
strong enough to encourage strong damping by turbulent friction(> 10m s- 1).
Analysis of more subtle effects, arising from heating by the relatively warm sea
surface, suggests that development is inhibited unless the airflow is almost perpen-
dicular to the sea-surface isotherms. Since local sea-surface isotherms are largely
zonal, this may well be the reason why polar lows develop only in northerly flows.
Between them these restrictions ensure that polar lows develop in only a small
fraction of northerly air streams in the eastern Atlantic in winter.
The two lows apparent in Fig. 11.25 produced substantial snow falls in the
British Isles, including some exceeding 250 mm in average depth (i.e. about 25 mm
equivalent rainfall) which almost totally disrupted traffic in parts of southern
England. Although some of the disruption arose because of that area's combina-
tion of very high population density and very low preparedness for snow, it is
worthwhile noting that there are two quite obvious features of snowfalls which
make them especially disruptive of road traffic and other activities. One is the low 401
11 Large-scale weather
systems in middle latitudes NS

55

Fig. 11.25 Two polar lows

near the British Isles at 1800 Z
on 8 December 1967. Sea-
level isobars show that the
shallow lows were embedded
in geostrophic northerlies.
Areas of falling snow are
hatched. (After [82])

density of fresh snow, which arises because the spiky arms of the snow crystals
prevent close packing. Fresh snow lies to an average depth which is at least ten
times the equivalent depth of rainfall. The other feature is the tendency for fallen
snow to drift in even quite moderate winds, moving from exposed surfaces and
gathering in great depth in eddies forming in the lee of walls and hedges. Fenced or
hedged roads abound in lowland Britain and are often blocked by several feet of
snow in windy conditions while surrounding meadows have only a skim of snow.
The disruption caused on the occasion of Fig. 11 .25 served one useful purpose,
however: it attracted the attention of British meteorologists and led to a
considerable increase in understanding of these weather systems. However, fore-
casting of polar lows is still very difficult because they usually form in the observa-
tional desert north of Iceland and quickly traverse the fairly short distance down to
the British Isles.

HEAT LOWS

In warm summer weather in middle latitudes it is common to see weak puddles of

low pressure developing daily over islands and peninsulas in the absence of any
strong synoptic-scale pressure gradients. Sometimes they may be associated with
showery activity in the afternoon and evening, and sometimes the skies may remain
largely clear. These lows are a meso-scale, even synoptic-scale, response to the
rapid development of strong temperature contrasts between land and adjacent sea.
Figure 11.26 represents a typical heat-low situation over England. During the
previous night the land cooled somewhat below the temperature of the surrounding
sea, but only a few hours of strong sunshine are enough to reverse the situation
completely, because the effective heat capacity of land is so much smaller than that
of the sea (section 9.4). The land surface may soon be 20 oc or more warmer than
the sea, and the temperature difference in the overlying layer of air may be half as
large. This layer expands ~s it warms, raising the overlying air so that isobaric
surfaces aloft dome upwards slightly overland. For example if a layer of air 300 m
deep warms isobarically by 10 oc in roughly 300 K, its density must fall by about
30Jo according to the equation of state for a perfect gas. Assuming that the layer is
402 prevented by its surroundings from expanding laterally, it must expand upwards,
 11.7 Polar lows and heat
lows

Fig. 11.26 Sea-level isobars

showing a heat low over
south-east England at 1500 Z
on 20 June 1960.
Temperatures were just over
25 °( throughout the area of
low pressure. Although not
plotted here, a succession of
earlier maps shows that the
low developed as the land
heated strongly in the
preceding few hours. (From
(84))

its upper surface rising by about 30Jo -i.e. about 10m in the example chosen.
The position is now as sketched in Fig. 11.27(a), and there is obviously a
tendency for the overlying air (virtually the whole troposphere since only a shallow
layer is warmed substantially) to flow outwards off the domed boundary layer. The
air movement needed to accomplish this on a reasonable time scale (say 3 h) is very
small indeed: appendix 11.3 shows that radial speeds - 1 em s-• will suffice for a
land surface of radius 100 km. If we assume that the movement is complete when
the isobars at the top of the heated layer become horizontal again (Fig. 11.27b),
then it is clear that air equivalent in mass to a 10 m layer of the low troposphere has
been removed, which corresponds to a loss of pressure of a little more than 1 mbar
at the surface and throughout the heated layer, as is typically observed. The sea-
level pressure over the warming land has now fallen a millibar or so below the sur-
rounding values, and the weak heat low has appeared on the synoptic charts.
Notice that the upward movement of the top of the warming layer is slow (-
1 mm s-• in the worked example) and spread over the whole land surface, and is
not to be confused with the updraughts of the thermals embedded in the layer. The
latter are strong (- 1 m s -•), localized and almost completely offset by intervening
sinks (section 10.5), and are the means whereby sensible heat is pumped into the
warming layer from the underlying surface. Even after the reduction of pressure by
shedding of overlying air, the warm layer is strongly baroclinic around the edges of
the warmed land, since the isopycnals dip down inland much more steeply than the
isobars (Fig. 11.27b). The air in and immediately above the shallow baroclinic
layer responds by developing the sea-breeze circulation described in section 9 .11.
The inflow of cool, dense sea air and outflow of warm land air tends to fill the heat
low, but the tendency is impeded by friction in the shallow layer, and by Coriolis
deflection of the relatively fast flows, so that the low is depleted only slightly before
evening cooling begins to destroy it by reversing the process by which it was created
in the first place.
If the warming is not fully offset by the subsequent nocturnal cooling, because of
a change in the sypnotic situation, for example, then the shallow residual low may
develop further with the next day's heating, and so on for several days. This applies
better to larger-scale land masses, such as France or Spain, where the scale is too 403
11 Large-scale weather !a I
systems in middle latitudes
----- - --
'
--'
' '',Warmed layev
' - --/

''
/

'
/

Fig. 11.27 Idealized vertical •• J ' Land 'l . . .... ~

sections showing the Sea Sea

formation of a heat low. In (a)
the isobars (solid lines) dome
slightly upwards in and above (b)
the warmed layer (isotherms
dashed}, shedding air aloft. In
(b) the shedding is complete,
-,,
', /
/
/ -
"
leaving isobars dished ' ' / /
/

' /
'
downwards in the warmed ........ j ' '
Land
/ /

l
layer. Sea Sea

large for the offsetting sea-breeze circulation to reach more than a small propor-
tion of the total area. Coriolis deflection is the crucial factor here; if the sea breeze
persists for more than a few hours, it suffers Coriolis deflection (to the right in the
northern hemisphere) and tends to reach quasi-geostrophic equilibrium with the
pressure field and friction. The flow in the low troposphere begins to move
cyclonically round the shallow low, forming a dynamic barrier to further filling,
and confirming the feature as a consistent low-pressure system on the weather
map.
Meanwhile the warmed layer inland may be deepening as convection reaches
higher above the warming surface, so that the pressure deficit in the heat low
increases. However, the vertical development of convection will depend heavily
on how convective stability is being influenced by other factors. If it is being
reduced, by synoptic-scale vertical stretching associated with the approach of a
weak front for example (section 10.3), then the convection may quickly become
quite deep, producing widespread outbreaks of heavy showers and thunder. The
associated cloud reduces the surface heating, but the deep, warm, cloudy layer may
persist for some days as baroclinic adjustment encourages it to rise further in rela-
tion to its cooler surroundings. Such thundery lows often develop over France in
the summer and give spells of humid, thundery weather in Britain as conglomera-
tions of showers swing north-west in the weak circulation. If however the con-
vective stability is being maintained or increased, by subsidence associated with an
overall anticyclone, for example, the convection from the warmed surface may
remain confined in a fairly shallow layer, and the depth of the heat low be limited
by the temperature excess of the heated layer over adjacent cooler layers. Some-
thing like this happens in summer over land masses influenced by subtropical anti-
cyclones; Figure 12.6(b) shows the effects of this on sea-level pressures over North
Africa, where a semi-permanent heat low over the western Sahara maintains a pres-
sure deficit of about 20 mbar relative to the Azores High to the west, in obvious
association with very high surface temperatures over land. Nothing comparable is
apparent on the winter maps (Fig. 12.6a). Comparison of summer and winter over
the Indian subcontinent shows that there are even more dramatic effects associated
with the monsoon there.

11.8 Middle-latitude climates

We have seen that the middle-latitude troposphere is a chronically active zone:
404 opposing radiative imbalances at lower and higher latitudes (section 8. 7) maintain
 11 .8 Middle-latitude
climates

Fig. 11.28 Mean surface

temperatures in (a) January
and (b) July. (After [85])
(b)

a strong meridionally baroclinic zone whose deep westerly flow is intrinsically

unstable (section 11.6), generating long waves and cyclone waves, and through
them the whole gamut of smaller-scale disturbances from fronts down to showery
and even smaller-scale convection; the seasonal march of the sun produces large
seasonal variations in solar and terrestrial radiative fluxes, and their net balance,
driving seasonal rhythms of mean temperature and diurnal temperature range
which are especially significant overland (Fig. 11.28), an account of the small
effective heat capacities there.
The climatic potential in the seasonal and meridional variations of solar eleva-
tion is very great: at latitude 65° the noon solar elevation (Fig. 8.8) ranges from
only 1.5° in midwinter to 48.5 ° in midsummer, and at latitude40° the range is from
26.5 to 73.5. Daylength [51] varies from 2.7 h to 20.2 hat latitude 65°, and from
8.7 h to 14.2 hat latitude 40°, severely limiting the daily insolation and enhancing
the net radiative output in winter, and reversing the effects in summer (section 9 .3). 405
11 Large-scale weather These constraints would impose a very sharp meridional gradient of climate, if
systems in middle latitudes
they were not partly offset by the smoothing actions of the tropospheric distur-
bances mentioned above, and of the heat capacities of the underlying surfaces
(section 9.4). The vast oceanic heat capacity in particular has a very large
smoothing effect which, together with the poleward heat flows in major ocean
currents like the Gulf Stream and Kuroshio (section 8.13), and ready supply of
water vapour, maintains much greater and steadier cyclonic activity over the
oceans and western margins than in continental interiors. This maintains a marked
west-east zonation of climate from maritime to continental (Fig. 11.28) which
dominates the following discussion.

WESTERN MARGINS

Prevailing westerlies and the weather systems they bear allow only very modest
seasonal variations of temperature and rainfall (Figs 11.28 and 12.15); they also
raise annual average temperatures significantly above the zonal average (in the
northern hemisphere, because it contains almost all middle-latitude lands), mainly
by raising winter temperatures. Families of depressions are normal throughout
the year, interspersed with irregular periods of blocking anticyclones, whose
distribution varies only slightly with location and season. Contrasts between
tropical and polar air masses are held in check by rapid accommodation to the local
sea-surface temperature: for example, showers very rapidly warm the cool polar
flows in which they flourish, by pumping latent heat aloft. Summer heat lows may
develop inland in slack flows, giving thundery weather; but their warmth is
tempered by cloudy loss of sunlight. However, what might seem almost boring
uniformity to inhabitants of more extreme climates in fact contains a wealth of
day-to-day variation which excites endless comment in native populations and
complex climatological commentary which draws attention to subtle singularities
[86], such as the over-dramatically named European Monsoon - a frequent
increase in cyclonic activity in late June, highlighted nowadays by its effects on
tennis at Wimbledon (London).
Temperature variations arising from the onset of blocking (warming in summer,
cooling in winter- section 11.5), or vigorous cyclonic activity (the reverse), can be
so nearly comparable with the small seasonal variations that the warmest days of
winter may be warmer than the coolest days of summer. The virtual absence of
rainfall in the blocks gives rise to much of the relatively small rainfall variation
between corresponding seasons and years at any location. However, the prevailing
westerlies maintain strong concentrations of rainfall on western coasts and hills,
where there is a fine structure of relief rainfall and rainshadows, with overall reduc-
tions inland on scales of 10-100 km (section 9.11). If the relief is modest, as in
much of Europe and European Russia, these eastward reductions continue on a
larger scale reflecting the inland reduction in cyclonic activity already mentioned;
but strong meridional barriers like the Rockies and Andes can compress the
transition from maritime to continental climates into much narrower corridors
(Fig. 11.28).
Prevailing westerly winds sweep inland, weakening with the increase of surface
roughness (section 9.9), and fading patchily in the lee of north-south ridges; and
they decrease over longer distances too as the cyclonic activity which drives them
decreases inland. On western-facing coasts daily runs of wind (section 2.4) of more
than 600 km at the 10 m standard level (an average of Force 4- appendix 2.2) are
406 commonplace, offering the opportunity of reliable generation of electrical power.
 Diurnal temperature variations at screen level increase quite quickly inland, as 11 .8 Middle-latitude
climates
the decreasing thermal capacity of the upwind fetch allows larger and larger
response to the net radiative cycle. This in turn encourages a diurnal cycle in
convection and associated cloud and precipitation which reverses seasonally: in
winter, showery air masses (like the polar maritime air which sweeps in behind cold
fronts) quickly lose their showers as the air flows over progressively colder land,
whereas in summer marginally stable flows are progressively destabilized as they
pass over warmer and warmer land. The former adds to the normal relief rainfall
shadow inland from the western margins. Inland cooling sharply increases the
incidence of frost and lying snow in winter.
Many of the climatic features of western margins appear in more limited form
on smaller scales, ranging from the North Mediterranean in winter (where
cyclogenesis is encouraged by the strong baroclinity between the warm sea and the
cold southern European land mass) to the eastern margins of the North American
Great Lakes, and the Black and Aral Seas.

CONTINENTAL INTERIORS

Just as western margins have maritime climates because they lie only a short fetch
from the sea down the prevailing westerly wind, so the rest of the continental areas
have continental climates on account of their much longer fetches from the sea.
Even eastern margins are largely continental in climate (Fig. 11.28), since
cyclogenesis in the adjacent seas (the the East) only briefly affects the margins
before the young depressions eastward move out of range.
For the reasons given above, continental climates are typically much more
extreme, meridionally, seasonally and diurnally, than maritime climates. For
example, Fig. 11.28 shows that, at latitude 50°, the winter-summer swing in
seasonal mean temperatures is from about 5 to 15 oc on the west coasts of North
America and Europe, while it is from about -10 to +20 °C in north-eastern
America (longitude 80 °W) and -15 to +25 °C in Asiatic Russia (longitude
80 °E). The seasonal swings in the continental interiors are broadly associated with
a transition from cold winter anticyclones to summer heat lows, but considerable
differences between North America and Eurasia arise from their particular
geographies and are best dealt with separately.
The smaller seasonal temperature range over the North American continent
arises from its smaller size, a tendency for weak cyclogenesis in the lee of the
Rockies (section 11.6), and the presence of a large warm water surface to the south
(the Gulf of Mexico). The cold anticyclone in winter is a weak, unstable feature in
comparison with its great Asiatic counterpart.
Much of the land north of latitude 60° has a near-Arctic climate, with wide-
spread permafrost (permanently frozen ground under a shallow top layer which
thaws in summer). In winter, northerlies in the lee of the Rockies bring bitterly cold
conditions from this Arctic continental source, which contrast strongly with
conditions associated with the southerly flows from the Gulf at other times. Strong
outbreaks of the very cold air (cold waves) are associated with temperature falls of
10 oc or more in 24 hours. Even sharper contrasts occur when unusually strong
westerlies force a very warm, dry Chinook down the eastern slopes of the Rockies
(section 9.11). Winter precipitation (snow) is sparse in the central interior, but
increases sharply to the east of the Great Lakes (and Hudson Bay, before it freezes
over in early winter). Unlike its great Asian counterpart, the cold anticyclone of the 407
11 Large-scale weather in North American winter is therefore a statistical feature arising from tendencies
middle latitudes systems
underlying considerable day-to-day variability.
In summer, southerly flows of very warm, humid air from the Gulf are asociated
with very vigorous outbreaks of severe local storms along sharply defined cold
fronts (section 10.6). Less vigorous cyclonic activity draws vapour from the same
source to water the east of the continent, but much summer precipitation falls in
heavy showers not associated with any air mass contrast. Though driven ultimately
by the strong daytime solar heating of the land, these often occur at night,
apparently in response to subtle changes in low-level airflow set off by initial
nocturnal cooling. The balance between precipitation and evaporation in some of
the great grain-growing regions is rather fine, apparently leaving them vulnerable
to quite small changes in global climate.
The much greater zonal extent of the Eurasian continent (over twice that of
North America however the limits are defined) allows the effects of oceanic
remoteness full play, despite the lack of any equivalent of the Rocky Mountain
barrier to the west. A maximum winter-summer swing of seasonal mean
temperature from -40 to + 15 oc is reached at around longitude 130°.
Winters are extremely harsh over a wide area. Permafrost reaches south of the
Arctic circle from about 100 °E to the eastern extremity of Siberia (about 180 °E).
Maritime influence from the Arctic ocean is lost when polar pack ice extends to the
north coast, and the barriers of the Tibetan and Himalayan massifs to the south
prevent any northward intrusion of warm air from the Indian ocean. Snow settles
permanently in late autumn over a wide area and remains until spring, raising the
ground surface albedo and reflecting what little potential solar input there is
(section 9.3). Below the developing winter inversion (Fig. 11.17), falling
temperatures reduce the maximum vapour content and with it the local greenhouse
effect, allowing the surface to cool even faster by long-wave emission to space
(section 8.5). Though the shallow layer of cold air produced in this way is readily
channelled by valleys and dammed by ridges, it spills out over a wide area
throughout the winter, producing intensely cold waves over North China and
Korea. Precipitation is minimal over a wide area.
After a rapid transition in spring, summer conditions arrive with little of the lag
apparent in maritime climates. Temperature gradients are modest over a wide area
(Fig. 11.28), and a reasonable precipitation total falls mainly as showers (though
the northern flanks of the Tibetan massif are semi-desert). China, however,
experiences substantial maritime influence, with shallow depressions advancing
north-east in late spring (part of the Asian Monsoon), and occasional typhoons in
late summer). In the autumn the positive net radiation balance quickly dies and
reverses, and rain and snow alternate unpredictably (especially in the west, as both
Napoleon and Hitler found to their cost), before winter proper sets in.

Appendix 11.1 Uplift and rainfall

Consider a steady-state weather system in which the upward flux of water

substance in the form of vapour is exactly balanced by the downward flux in the
form of precipitation. Let us examine this equality in a vertical column of unit hori-
408 zontal cross-section. The fact that air may be rising in a gentle slope rather than
vertically up the column is irrelevant provided there is no significant net horizontal Appendix ll . 2
divergence of water substance as a result. Note that we need only the vapour flux
into the base of the cloud, since the details of updraught and vapour content at
higher levels make no difference to the overall balance.
Suppose that the vapour rises into the base of the cloud in air of density p,
specific humidity q (g kg- 1) and updraught w. Then the upward mass flux of water
vapour in kg m- 2 s- 1 is
I0-3 p q w
Now if the equivalent rainfall rate of the precipitation is r mm h - 1 and the density
of water is taken to be 1000 kg m- 3, the downward mass flux of precipitation in
kg m- 2 s- 1 is
2.78 x I0- 4 r
Equating these fluxes, and rearranging to isolate the updraught w, we find

w = 0.278 _r_
pq
Air enters most cloudy updraughts in the low troposphere where the air density is
close to 1 kg m - 3 .To this approximation therefore, and removing the superfluous
accuracy of the numerical factor,

w = 0.3 _!__
q
where wis in m s- 1 when ris in mm h- 1 and q ising kg- 1•
In middle latitudes the specific humidity of air rising into the base of a cloudy
mass is - 10 g kg- 1 • It follows that a typical frontal rainfall rate of 3 mm h- 1
implies an updraught - 8 em s - 1• A rainfall rate more typical of a heavy shower
is 30 mm h - 1, and this implies w - 0.8 m .s - 1• However, the presence of evapora-
tive loss from the sides of cumulonimbus, and significant efflux of cloud in
the anvil, mean that the simple estimation of w is likely to be considerably too
small.

Appendix 11.2 Variation off with latitude

Consider the rate of change of the Coriolis parameter f with y, the poleward
distance along a meridian. According to Fig. 11.29 and the definition of radian Fig. 11.29
measure a very small increment dy corresponds to a similarly small increment in
latitude cp through

dy = R dcp

where R is the radius of the Earth. If follows that we can rewrite

f3 = aJ
ay 409
11 Large-scale weather df
systems in middle latitudes
R d¢

Substituting 20 sin ¢ for f and differentiating we find

20 cos¢
R
1.6 x I0- 11 m -I s -I at about latitude 45°.

Appendix 11.3 Radial winds and heat lows

Consider a horizontal circular area A over which the surface pressure falls by t:,.p in
time C:,.t. Assuming hydrostatic balance, the implied reduction in air mass resting
on the area is (Al:,.p)lg, so that the average rate of reduction is
-rr R2 t:,.p (12.4)
g t:,. t
where R is the radius of A.
In the simplest model of a heat low this is achieved by air moving radially
outwards at all levels above the warming and expanding boundary layer. Figure
v II.30 depicts a cylinder of air standing on A. If the radial flow has speed Vbetween
heights z and z + dz, it is apparent that the contribution to outward mass flux in
this range of heights is given by
Vp2-rrR dz
Fig. 11.30
Integrating through the full depth of the atmosphere (ignoring the relatively
shallow boundary layer) and replacing p dz by dp/g through the hydrostatic
approximation, we find that the rate of export of mass is

2-rrR f v dp
P,
g 0

where p, is the surface pressure. If Vis uniform, or is the appropriately weighted

average radial speed, this expression becomes
2-rrR V
- - - p, (11.2)
g
The balance between observed depletion rate and rate of export by radial flow is
then given by equating expressions (1I.I) and (I1.2). After rearrangement we have

v = _!!_ f:,.p
2p, C:,.t

If we substitute for the case of a rather large and rapid pressure fall (3 mbar in 3 h)
we will find an upper limit to the required radial flow speeds. For a value of R
corresponding to a small but substantial island (I 00 km) we find V to be a little over
I em s - 1 • Notice that this corresponds to a horizontal divergence of a little over
410 10-1 s-1.
 The whole analysis can be applied, with V and f:).p! /:).t reversed, to the Problems
formation of a cold continental high-pressure area.

Problems

Problems for Chapter 11 are included at the end of Chapter 12.

411
CHAPTER
Large-scale weather
12 systems in low latitudes

Middle- and high-latitude weather and climate are strongly affected by the seasonal
march of the sun, even over the oceans. We now consider large-scale weather and
climate in the rest of the troposphere - the lower latitudes where the seasons,
though often very pronounced, are not so directly related to the sun's progress
from solstice to solstice. Here seasonal temperature variations are often small, or
out of step with the solar calendar, or even show two or more maxima per year; and
they may in any case be subordinate to seasonal variations in rainfall or wind
direction, or diurnal or irregular day-to-day variations.
More than half of the Earth's surface is included in these lower latitudes (noting
that exactly half lies between latitudes 30 oN and S), but the historical growth of
meteorology in middle latitudes, the inherent complexity of lower-latitude weather
and climate, and the extensive oceans there, have combined to slow our progress in
understanding this major part of the Earth's atmosphere. Fortunately the satellite-
observation network (section 2.11) is redressing the balance, and the global
comprehensiveness required of modern numerical forecasting models is forcing
attention on outstanding problems.

12.1 Subtropical anticyclones

These archetypal anticyclones nearly girdle the Earth in the tropics and maintain
the subtropical high-pressure zones which are such a major feature of large-scale
atmospheric structure (section 4. 7). They are very large, persistent, zonally
elongated areas of high sea-level pressure, and are particularly well-developed over
the oceans, especially in winter. Fig. 12.1 depicts a typical situation in the North
Atlantic, where the local segment is known as the Azores High in Europe and the
Bermuda High in North America, because it often straddles one or both of those
islands. Observations show that the weather is mostly fine, with few reports of
precipitation, and those mostly very light apart from some showers which are
usually confined to the western flanks of the high. Where skies are not clear, clouds
are generally reported to be low, with stratocumulus being the most widespread
412 type. There is very little cloud in the middle troposphere, as is particularly obvious
 12.1 Subtropical
anticyclones

Fig. 12.1 A surface map

showing the Azores High in a
well-developed state.
Depressions and fronts ore
moving eastward on its
northern flank , and a ridge of
high pressure is extending
northwards at about 50 °W.
Notice the lowering of sea·
level pressures over north·
west Africa in association
with the summer solar
heating there. (From the
Daily Weather Report of the
o"' "· n·· (•
, 2 ... of- - t .. lJ)"'o)...., ·• •'''"'" iJ •• Meteorological Office, by
'+" ,...~b V~ T t•"1'
permission)

when observations are made from aircraft, when the sky is seen to be quite empty
above whatever low cloud there may be (Fig. 12.2).
Radiosondes rising through such an anticyclone show clearly why the middle
troposphere is so free of cloud. Above the layer occupied by surface-based
convection (Fig. 12.3), there is a relatively shallow layer which is either nearly
isothermal or contains a temperature inversion (temperature increasing with
height). In either case the potential temperature increases sharply with height, and
the layer therefore represents a convectively very stable interface separating the low
troposphere from potentially much warmer air aloft. The warmer air has very low
relative humidities, often lower than 200Jo, so that cloud formation or maintenance

Fig. 12.2 A deck of

stratocumulus under a
cloudless middle troposphere
in an anticyclone. Wisps of
high cloud suggest that air in
the high troposphere is
relatively moist.

413
11 Large-scale weather Aughton 1200Z July 16 1972
systems in low latitudes

Fig. 12.3 A radiosonde

ascent through the interior of
an anticyclone, showing a
well-mixed convecting
boundary layer surmounted
by a subsidence inversion and
a deep layer of very dry
subsiding air.

is out of the question. Indeed even liquid aerosol cannot escape evaporation, with
the result that the air in middle levels is remarkably clear of haze. This contrasts
sharply with the air in the low troposphere, below the temperature inversion, which
is often quite hazy, especially over or downwind of continents. Indeed observa-
tions from mountains or aircraft often show a very sharp haze top which coincides
with the base of the inversion (Fig. 11.16 shows the same feature in a blocking
anticyclone at higher latitudes) and marks the upper limit of the relatively humid
and hazy air of the convecting layer.

SUBSIDENCE

The potential warmth and the low humidity of air in the middle troposphere of an
anticyclone are both consequences of the gentle but persistent downward motion
which is known from a variety of evidence to be present over large areas of the
system. As air sinks down from high levels it warms, and because there is virtually
no cloud present (usually there is nothing more than a few wisps of cirrus or aircraft
condensation trail), the warming tends to be dry-adiabatic. And because there is no
cloud or precipitation to evaporate in the warming air, the specific humidity in the
air is conserved, with the result that the relative humidity falls as the air warms.
Pretending for a moment that actual descent in time corresponds approximately to
instantaneous descent on Fig. 12.3, you can check the magnitude of this effect by
sliding a dew-point temperature down an isopleth of saturation humidity mixing
ratio (Appendix 12.1) from assumed saturation in the high troposphere, noting the
very large dew-point depression which has developed by the time the air has sunk to
lower middle levels. If the descent of air were truly dry-adiabatic, the temperature
414 would similarly slide down a dry adiabat from the high troposphere in Fig. 12.3.
The actual temperature profile tends more or less steadily to lower potential 12.1 Subtropical
temperatures at lower altitudes, which suggests that the sinking air cools anticyclones
diabatically. This cooling is maintained by the net effects of radiation (the only
mechanisms capable of reaching air isolated from convection by its own convective
stability), in which weak direct solar warming is consistently exceeded by net terres-
trial cooling (section 8.6). As noted in the previous section, it is dangerous to
assume that three-dimensional air motion in large-scale weather systems can be
approximated by horizontal or vertical motion, but in fact more thorough
investigations largely confirm the above interpretation. This is largely because the
winds in the upper troposphere of an anticyclone are relatively weak, so that air
parcels remain in its vicinity for days or longer.
The stable layer in the low troposphere of an anticyclone (Fig. 12.3) is called the
subsidence inversion because it is maintained by the subsidence of the warm, dry
air which makes up the bulk of the air in the system. It forms at the interface
between the large-scale descent and the small-scale convection which is maintained
by the solar heating of the underlying surface, and is often sufficiently intense to
contain a temperature inversion. However, the title subsidence inversion is often
used even when the layer is not quite stable enough to produce an inversion. This is
unfortunate but not too misleading, since the physically important factor is the
presence of strong convective stability, rather than a temperature inversion as
such.

CLOUD AND SMOG

If the air in the convective boundary layer is normally humid, the lifting condensa-
tion level (section 5.11) will be a few hundred metres below the base of the sub-
sidence inversion, and stunted cumulus will form whenever there is convection
from the surface. When such convection is persistent, a layer of stratocumulus
forms, which may be thick enough to reduce the sunlight considerably, and even
produce a little drizzle. Seen from above, the backscatter of sunlight makes the
cloud layer look quite bright by contrast (Fig. 12.2). Once formed, such a cloud
layer tends to maintain itself by convection between its top surface, which is cooled
more by terrestrial radiation than it is warmed by sunlight, and its base, which is
kept warm by the exchange of terrestrial radiation with the underlying surface. The
consequent overturning also maintains the globular appearance of the cloud, even
when there is no convection below cloud base.
If the air in the convecting boundary layer is dry enough, then the lifting conden-
sation level will be so high that convecting air will still be unsaturated when its
upward motion is stopped by the base of the subsidence inversion, and there will be
no low cloud. The presence of the subsidence inversion may still be seen from the
obvious haze top which is often visible, even from the oblique viewpoint of a
surface observer (Fig. 11.16). In these conditions the solar warming of the surface
may become intense in the middle of the day, warming the convective boundary
layer as a whole and tending to eat into the base of the subsidence inversion.
However, the large-scale subsidence is often observed to be able to contain such
convection below the lifting condensation level for days on end, and is sometimes
strong enough to lower the subsidence inversion to within a few hundred metres of
the surface. In such circumstances, industrial pollution of the confined layer may
produce significant enhancement of concentrations there. Indeed the so-called Los
Angeles smog is produced there and in many other cities regularly or occasionally
under the influence of anticyclonic subsidence (Fig. 12.4). This type of smog is 415
11 Large-scale weather
systems in low latitudes

Fig. 12.4 Los Angeles Smog

in late summer. The stunted
cumulus above the distant
hills are trapped below the
anticyclonic subsidence
inversion, as are the pollutant
gases and particles and the
resulting smog over the city
basin. However a much lower
inversion (produced by
nocturnal cooling or an
invasion of cool sea air) is
trapping the thickest smog in
a much shallower layer and
obscuring the city details in
the centre of the picture.

especially marked in summer because the ultraviolet component of the strong

sunlight encourages photochemical production of ozone and other biologically
damaging substances in the confined boundary layer.
On other occasions the convective boundary layer may be several kilometres
deep, and yet still not reach the lifting condensation level, because of the dryness of
the air as a whole; this can happen over continental subtropical deserts such as the
Sahara in North Africa.
The systematic absence of substantial cloud means that rainfall in the sub-
tropical highs is very low. In continental areas the result is arid desert, and indeed
there is a belt of hot deserts in the subtropics of each hemisphere, including for
example the interior of Australia in the southern hemisphere. Rainfall is sparse,
intermittent and unreliable, coming in deep convection during periods of weakness
of the anticyclonic circulation. The rain when it comes may fall in torrential
showers, running off the baked landscape in flash floods which rush down the
gullies (wadhis to the arabic nomads who have long eked a living in such harsh
conditions in North Africa and Arabia) maintained by similar sporadic events in
the past. Though the brief, inefficient moistening of the ground surface is enough
to trigger the ecological miracle of the flowering desert, the water soon evaporates
and the desert returns. Over the oceans the persistent excess of evaporation over
precipitation in the subtropics keeps the surface waters significantly more saline
than elsewhere.

AIRFLOW

The anticyclonic circulation which is so pronounced in the lower troposphere, and

which gives the whole type of system its title, can be related to the simplified
vertical section in Fig. 12.5. Since air is persistently subsiding in the middle tropo-
sphere it must be persistently converging in the high troposphere and diverging in
the low troposphere. The convergence aloft concentrates the vorticity there,
making it more cyclonic (section 7 .14), but it seems that the initial vorticity is often
very feeble, with the result that the circulation in the upper parts of a deep anti-
cyclone is often only weakly cyclonic. As this air sinks into the low troposphere,
416 however, it diverges strongly and its absolute vorticity is reduced so sharply that it
 12.2 Monsoons

~:c
77777777
Fig. 12.5 A simplified model
of vertical and radial airflow
in the troposphere in an
anticyclone. Regions of
synoptic-scale horizontal
convergence and divergence
becomes anticyclonic relative to the weather map. Although pronounced, the mag- are labelled C and 0
nitude of this anticyclonic vorticity is limited as described in section 7.14, so that respectively. Compare with a
winds in anticyclones never reach the strengths which are quite common in cyclonic similar picture of a cyclonic
systems. system (Fig. 7.29).
Near the surface, turbulent friction encourages divergence by enabling air to
flow with a substantial component of motion outward across the isobars toward
lower pressure. On the western and polar flanks of the subtropical highs such
diverging flow feeds air into the warm sectors of the mid-latitude depressions as
they sweep eastward, maintaining their warmth and supplying air to the warm
conveyor belts there (section 11.4). On the equatorial flanks the diverging flow
feeds the trade winds which angle westwards down to the subtropical convergence
zone, fuelling the deep convection there (section 4. 7).
During the development of a particular subtropical high-pressure cell, con-
vergence aloft obviously exceeds divergence below, in a reversal of what happens
during the formation of a depression, where divergence above exceeds convergence
below. However, during the quite long periods in which the intensity of an anti-
cyclone changes little, divergence and convergence must balance closely in a
dynamic equilibrium. No simple argument explains the continuing presence of the
high surface pressure, despite the unwise claims of some descriptive texts. For
example, it is not a direct dynamic consequence of the downward motion of air in
the main mass of air (like the pressure of a hose squirting water down onto a floor),
since this would imply a gross failure of the hydrostatic approximation (section
4.4), which actually must be especially accurate in the quiet conditions prevailing
there. It seems that dynamic constraints such as the Coriolis effect maintain the
excess of atmospheric mass in the central parts of the anticyclone, once it has been
established by the initial excess of convergence. The relative warmth of the
subsiding air means that air pressure falls more slowly with increasing height than
is the case in the surrounding cooler air masses (section 4.3), with the result that the
upward doming of isobaric surfaces (or equivalently the concentration of high
pressure on horizontal surfaces) increases with height, often into the upper tropo-
sphere. The system is therefore warm and deep, unlike the shallow, cold highs
which develop over continental interiors in winter (section 11. 7). Over the Sahara
in summer, the subsiding warm core is superimposed on a shallow heat low at the
very hot surface (section 11.5), with the result that the depression of isobaric
surface near the land surface gives way to upward doming in the middle and upper
troposphere; a deep high therefore overlies some of the depressed sea-level
pressures apparent over North Africa in Fig. 12.6(b).

12.2 Monsoons

The great northward migration of cloud and rain over the Indian subcontinent
which takes place early each summer is known to virtually everyone through direct 417
12 Large-scale weather la l
systems in low latitudes

Fig. 12.6 Mean pressure and

airflow at the bose of the
troposphere in low latitudes
in January and July. Isobars
of mean sea-level pressure (b )
(from [86)) ore labelled in
tens and units of millibars.
Arrowheoded thick lines ore
streamlines of flow near the
surface (from [26]). Dashed
lines mark zones of
confluence (also from [26]),
some of which correspond to
the showery intertropical
convergence zone described
in section 12.2.

experience, school geography lessons or news reports. It is often simply called the
Indian monsoon, which correctly implies that there are monsoons in other regions,
but tends to underplay the scale and interconnection of the huge seasonal events
which affect Indo-China, Indonesia and China as well as India. Popular discussion
also tends to focus on the wet summer, whereas this is only one part (very impor-
tant because of the water it brings) of an annual cycle.
The word monsoon means season in Arabic, and dramatic seasonal change is a
common thread which links a wide variety of climatic events occurring annually in
low latitudes. Figure 12.6 shows how the airflow in the low troposphere shifts
between January and July. In most places there is some meridional migration of
the convergence zones which are the local components of the intertropical con-
vergence zone (section 4.7). Since these zones are dotted with cloud and rain
maintained by the convergence of moisture-laden winds (section 8.13), their
presence in any area indicates the rainy season there. One of the largest meridional
shifts occurs in the vicinity of the Indian subcontinent. In January the nearest con-
vergence zone lies about 10° south of the equator (i.e. in the summer hemisphere),
whereas in July, the nearest equivalent feature is the heat low and associated
monsoon trough extending along the axis of the Ganges and Indus rivers, just
south of the great Himalayan mountain barrier. There is another very large shift
from northern Australia to an ill-defined location in northern China, and a
substantial shift in eastern Africa. By contrast, the convergence zones over the
Atlantic and Pacific oceans migrate comparatively short distances, and the conver-
gence zone in the eastern Pacific remains in the northern hemisphere and close to
the equator throughout the year.

MECHANISMS

The much larger seasonal shifts of airflow over land as compared with ocean
418 clearly must be related to the relatively very small effective heat capacities of land
surfaces (section 9.4): as the zenith sun moves from the tropic of Capricorn in late 12.2 Monsoons
December to the tropic of Cancer in late June, so the zone of maximum thermal
response by surface and atmosphere tries to march with it, and that march must be
faster where the total heat capacity is smaller, as it is over land. But the response is
made complex and indirect by the role of water vapour and cloud: solar input to
moist land surfaces is used largely to evaporate water (section 9.4), which may
travel some way before giving up its latent heat in cloud formation. And the
atmosphere tends to organize convergent flows of moistened air which gather
latent heat from extensive ventilated areas before releasing it in narrow zones of
cloudy uplift. Such convergence happens on a range of scales from the Hadley
circulation (Fig. 8 .13) downwards, all of which are too subtly sensitive to dis-
tributions of topography, land and sea to be easily understood.
On the smallest scale, atmospheric response is localized in cumulonimbus, even
the largest of which would be individually much too small to appear on maps like
Fig. 12.6 (even if they were not smoothed by seasonal averaging). The ascent which
constitutes the rising branch of the Hadley circulation actually takes place in a large
number of hot towers (section 10.5)- narrow, vigorous updraughts embedded in
much larger volumes of relatively static or sinking air. And these are organized in a
variety of types of large tropical weather system, few of which are observed or
understood as clearly as the large-scale weather systems of middle latitudes.

RAINY SEASONS

In terms of local climate, the result of all this is that regions within the range of
seasonal excursions of the ITCZ tend to have rainy seasons when it is close by, and
dry seasons otherwise. In the simplest picture, places between the extremes of
seasonal migration of the convergence zone would have two rainy seasons per year,
but there are so many special factors at work in any particular location, especially
overland, that the northward and southward migrations are often very different,
and only one may be effective as a rainy season.
A rainy season is not simply a period of unceasing rain, or an unrelieved
succession of showers: the rains come in bursts as one or other of the low-latitude
weather systems develops or moves by, and a particular rainy season may be
unusually wet or dry depending on the number, type and intensity of systems
occurring. Indeed the combination of the rather short length of many rainy
seasons, and the fact that individual weather systems last at least as long, at any
location, as mid-latitude systems, mean that year-to-year variability of rainfall is
higher than in middle latitudes. This is especially obvious in regions with short
rainy seasons, but appears even in well-watered low latitude regions (compare
Tables 12.1 and4.1).
Table 12.1 Annual rainfall at Canton Island 2°48'S, 171° 43' W

Year Rainfall Year Rainfall

1957 1269 mm 1962 402 mm

1958 1597 1963 713
1959 759 1964 519
1960 492 1965 1433
1961 508 1966 1101

mean 879 range + 718

- 477
standard deviation 414 ie. 47%
419
 12 Large-scale weather THE INDIAN MONSOON
systems in low latitudes
Although the rains enter Burma in April and May, they do not reach India until late
May or early June. Then they often sweep across the southern half of the sub-
continent in a spectacular burst, whose mechanism has intrigued generations of
meteorologists, and which may owe more to re-arrangements of flow in the high
troposphere than to the prior establishment of the heat low in the north west
(Fig. 12.6) which used to be regarded as a prime cause.
The monsoon usually becomes established over the Indian subcontinent by late
June, by which time there is a more or less continuous southerly flow of warm,
moist surface air toward, but not effectively into, the monsoon trough lying across
the north of the peninsula. (Northerly flow to this zone is blocked by the highlands
of Tibet and Afghanistan.) The southerly flow is in fact a continuation of the
south-east trades of the southern hemisphere. As these cross the equator north-
wards, the Coriolis deflection switches from leftward to rightward, and the flow
veers to become south-westerly. This flow of already moist air picks up further
large quantities of water vapour from the Arabian Sea (warmed to 28 or 29 oc by
very strong sunshine before the onset of the monsoon) and reaches the west coast of
India as the south-west monsoon. Large quantities of rain fall as a result of ascent
forced or triggered along the western edges of the great plateau, and substantial
flows continue northwards and merge with the southerly flow from the Bay of
Bengal. In the middle and lower middle troposphere there are important westerly
flows of air probably originating in Africa and even the Mediterranean.
Slow-moving synoptic-scale low-pressure systems, such as subtropical cyclones
and monsoon depressions (the latter being one of the few types to bring rain to the
monsoon trough) form in association with these flows and can produce large areas
of rain and quite strong winds even though surface pressures are depressed by only
a few millibars.
With the surface thermal equator established locally just a little north of latitude
25 °N, and substantial volumes of air rising in weather systems on its equatorial
flank, the normal baroclinity is reversed through almost the full depth of the tropo-
sphere between this warmth and the relatively cool troposphere nearer the equator.
The thermal wind relationship (section 7 .10) then ensures that the tropical easterly
jet stream is a semi-permanent feature of that region in the summer months at
latitudes about 15 °N (Fig. 12.7). The weather systems associated with the
monsoon therefore form and act in the context of south-westerly flow at low levels
and easterly flow aloft, and it is not surprising that their movements are often
sluggish and easterly, and their mechanisms complex.
Fig. 12.7 Streamlines The south-westerly monsoon begins to retreat in September and usually
(arrowed solid) and isotachs
completes its withdrawal from the Indian subcontinent by the end of November.
(dashed, labelled in knots) on
the 200 mbar surface at
Well before the end of this period, the easterly jet aloft is replaced by a narrow
0300 Z on 25 July 1955, branch of the subtropical (westerly) jet skirting the southern flank of the
showing a strong easterly jet
stream over southern India.
The zero isotach marks the
90 120
boundary between westerlies
and easterlies, and a south-
westerly jet stream lies
south-east of the Black Sea, 30
in air flowing towards the
ON
north of the Tibetan massif.
(After [27] and [88]) 15 ~._,._,_<:..__..,_

420 Equator
Himalayas. In the north of the subcontinent the ensuing drier seasons are also 12.3 Tropical weather
systems
much cooler because of the moderately high latitudes and flows of cold air from
neighbouring high ground. Weak low-pressure systems move south-eastward
along what was the axis of the monsoon trough, and later in the winter these may be
followed by north-westerly surges of cold air which move right across the peninsula
from the great reservoir which is building to the north and east of the Tibetan
massif. But this barrier largely protects India from the fierce surges of the bitterly
cold north-eastern monsoon which affects China in winter (section 11.8). Before
the onset of winter proper, coastal areas of the Indian subcontinent may have been
influenced by an important type of low-latitude weather system which we will
consider shortly - the tropical cyclone.

12.3 Tropical weather systems

The baroclinity of the tropical troposphere is weak and unstructured by com-

parison with higher latitudes. Solar heating of surface and troposphere reaches a
broad maximum between the tropics, and terrestrial radiation partly offsets this by
an amount which leaves a radiative excess varying little with latitude (Fig. 8. 7). The
result is that the tropical troposphere (meaning the whole troposphere between the
tropics, including the equatorial zone) is nearly barotropic in comparison with the
essentially baroclinic troposphere of higher latitudes. Horizontal temperature
gradients at any level are usually very small, and differences of more than - 5 oc
are comparatively rare away from the immediate influence of land and sea
surfaces. By the same token, seasonal changes in temperature are so small that the
seasons in low latitudes are measured more by their contrasts in weather, especially
rainfall, than they are by their temperature contrasts.
Although the tropical troposphere is relatively barotropic, the small horizontal
temperature gradients which do arise are associated with very significant patterns
of cloud and associated weather. Tall cumulonimbus (hot towers) may be only a
couple of degrees warmer than the mid-troposphere they penetrate, which may in
turn be a couple of degrees warmer or cooler than its counterpart outside the
conglomeration of cumulonimbus which constitutes any particular weather
system. In addition, the weakness of the Corio lis effect at low latitudes means that
most of the disturbances of the tropical troposphere are not even quasi -geostrophic
in their mutual adjustment of wind and pressure fields. This means that the
pressure field cannot be used to bolster the observed wind field in the way which is
so useful at higher latitudes. (More than a few degrees of latitude away from the
equator however, the undisturbed environment is still geostrophic to the extent
that for example the thermal-wind relationship holds at least qualitatively.) As a
result, quite a vigorous synoptic-scale weather system producing tens of centi-
metres of rainfall per day and winds of nearly gale force may be associated with a
depression in sea-level pressure of only a few millibars, and with horizontal
temperature differences aloft of only a very few degrees. In operational
meteorology such systems are known as tropical depressions unless surface winds
exceed gale force (section 12.4). This sensitivity of linkage between behaviour and
ambient conditions poses severe problems for observational, operational and
theoretical meteorology - problems which are compounded by the sparseness of
observations enforced because so little of the tropical zone is land, and even less is 421
12 Large-scale weather populated. Fortunately, the development of the meteorological satellite in recent
systems in low latitudes decades, especially the geosynchronous type (section 2.11), has improved things
considerably. Analysis techniques have been developed to extract wind profiles
from the movements of large clouds between consecutive pictures, and cloud and
air temperatures from simultaneous measurements of terrestrial radiation at
several wavelengths.
In the zones poleward of the equatorial confluence zones, the trade winds
maintain an easterly component of flow in the low troposphere, and there is clear
evidence from series of consecutive satellite pictures of these zones that much of the
cloud and rain over the oceans occurs in large clusters of cloud several thousands of
kilometres apart, which remain observably coherent while moving westwards at
between 5 and 10 m s -t - a speed which is usually a little slower than the airflow in
the low troposphere. Over land, the picture is confused by topography and a strong
diurnal rhythm, so that many clusters develop and decay without obvious motion.
So although land-based weather systems are the more immediately relevant to
people, the oceanic systems are the more obviously coherent and better under-
stood. In addition, many weather systems which form over the tropical oceans
(which account for about 75ft/o of the Earth's surface between the tropics) signifi-
cantly affect adjacent land areas, especially eastern margins.
Many of the cloudy systems over the sea seem to move in association with weak
troughs in sea-level pressure (Fig. 12.8) whose wave-like appearance on a weather
map gives th6m their name - easterly waves. Note that the reversed pressure
gradient compared with the mid-latitude norm of northward pressure lapse gives a
superficial impression of a pressure ridge. In fact the trough represents a weak
poleward extension of the convergence zone trough whose zonal axis lies some I oo
of latitude equatorward of the heart of the wave.
Fig. 12.8 An easterly wave In some cases there is marked asymmetry of cloud distribution, cumulonimbus
in the central Pacific in plan
and vertical section. On the and some altostratus being concentrated with rain and fairly strong winds to the
map, sea-level pressures and
surface winds show a wave
with trough line approaching
200 12
Kwajalein (the Marshall
Islands), followed by another
wave about 3000 km to the 300 10
east. Synoptic surface winds
are plotted as arrows with a 8
400
full barb representing each
Vertical ~
10 kt . Column totals of water .0 500 6
section E
vapour as a proportion of
600

~~~~~mMUfi
atmospheric mass are plotted 4
as dashed lines and labelled 700
in parts per thousand, with
2
dry and moist zones marked 800
D and M respectively. The ~~3m
dotted line was traversed by
instrumented aircraft whose
observations of cloud height
20
are represented on the
vertical section - average
heights by hatched cloud
towers and extremes by clear
Plan 10
extensions. (Horizontal
extents of towers are purely
diagrammatic .) (After [27] N
and [89])
0
160
422 E
east of (i.e. behind) the trough axis. Such concentration seems to be associated with 12.4 Tropical cyclones
gentle convergence in the low troposphere there and corresponding divergence to
the west of the trough axis, as follows. The convergence implies quite substantial
vertical stretching of the lower troposphere which serves to destabilize the layer
(sections 7.13 and 10.3) and thereby encourage the growth of cumulonimbus. It
also helps concentrate the fuel for cloudy convection, the water vapour
evaporating into the boundary layer from the sea surface. The stretching and con-
vergence also deepen the layer with easterly wind components, so that the
westerlies are confined to the upper troposphere (Fig. 12.8), and increase the
absolute vorticity of the lower troposphere (section 7 .14). The latter shows up in
the poleward excursion (increasing f) and cyclonic curvature of the airflow as it
overtakes the wave crest. A nearly balancing slight divergence seems to be con-
centrated in the high troposphere, giving a much more asymmetrical vertical dis-
tribution of vorticity than is typical in middle latitude systems (Fig. 7 .29). To the
west of the wave axis each of these effects is reversed to maintain relatively clear
skies in the more northerly flow.
In other cases the cloud is banded zonally across the trough axis, and in others
again the cloud is distributed fairly homogeneously. There is conflicting evidence
about whether or not all these cloudy systems are warmer or cooler on aggregate
than their surroundings, i.e. whether the systems are warm- or cold-cored, but it is
clear that they are not conspicuously warm-cored like their much more energetic
relatives, the tropical cyclones. The westward movement of such systems across the
great oceans at intervals of a few days provides much of the rainfall in these zones,
and in the coastal and island areas they affect. They have been studied relatively
intensively in the north central Atlantic by American meteorologists interested in
their effects on the Carribean and surrounding regions and their capacity to
develop occasionally into tropical cyclones (next section). It is clear that a substan-
tial number of the systems reaching the Carribean originate in the lines of cumulo-
nimbus which move westwards across north-west Africa in the summer. Others
form in the Carribean convergence zone and on fragments of fronts trailing down
from mid-latitudes. The annual total observed in the Atlantic remains fairly
constant at about 100, of which about 70 reach the Carribean.
The subtropical cyclones mentioned in the previous section are not limited to
monsoon zones, and can give rise to several closed isobars (at the standard 2 mbar
interval) at sea level, even though they appear to originate in the upper tropo-
sphere. They are cold-cored systems, though the temperature depressions are quite
small and they can occasionally become slightly warm-cored through unusually
large release of latent heat in the embedded clouds. There is evidence of banded
cloud structure, and they are often individually very persistent.

12.4 Tropical cyclones [91]

Of the 70 tropical disturbances (mostly easterly waves) reaching the Caribbean

each year, about 25 deepen to the point where there are several closed isobars on
surface charts and strong cyclonic circulation, cloud and rain over a substantial
area. When winds exceed gale force (19m s- 1 at 10m) they are called tropical
storms, and are equivalent in intensity to rather weak middle-latitude depressions,
though they have no fronts. In summer and autumn, a variable number of these 423
12 Large-scale weather storms, averaging about eight in the Caribbean, intensify to the point where winds
systems in low latitudes
exceed hurricane force (33m s- 1) at the 10m level, whereupon they are known as
tropical cyclones (popularly hurricanes). As is well known by news reports, and
even fiction [90], hurricanes are most intense in a ring surrounding a much calmer,
very low-pressure eye. In the week or so in which they remain in the mature, violent
state, they usually travel westwards and then polewards. They are found in many
sea areas other than the central west Atlantic and Caribbean (Fig. 12.9), being
called cyclones in the Indian Ocean and around Australia, and typhoons in the rest
of the Pacific Ocean. Regardless of local nomenclature, these are all the same type
of weather system, which in its combination of intensity and extent is the most
powerful disturbance of the Earth's atmosphere.
Tropical cyclones form only over oceans where surface temperatures exceed
about 26.5 oc over a substantial area at latitudes of at least 5°, but they occur in
every such eligible zone except the south Atlantic. The annual total of about 80
tropical cyclones is small in comparison with totals of extratropical cyclones, and
yet hurricanes (as we will now call them all for simplicity) continue to attract the
attentions of meteorologists to an extent which is out of all proportion to their
number. There are probably several reasons for this, but two will suffice. First,
they can be almost unbelievably destructive if they pass over vulnerable areas:
buildings and crops suffer severely in winds exceeding hurricane force, and coastal
areas risk inundation by the huge waves and exaggerated tides which can be
generated (over 200 000 people were drowned in 1970 when a cyclone from the Bay
of Bengal made landfall in the mouths of the Ganges, raising the local sea and
estuary levels by several metres for a few hours). Second, they are almost unique
among tropical weather systems in having a very sharply defined structure which
positively invites investigation.
Although hurricanes have been studied intensively for many decades now, their
mode of formation from relatively very weak tropical disturbances is still far from
clearly understood. Since they form in effectively barotropic conditions there is no
obvious source of available potential energy (section 13.2) to compare with the
baroclinity of mid-latitudes. However, there is some evidence that residual baro-
clinity leaking down from higher latitudes in the form of a weak upper trough may
help trigger their dramatic development. In the space of a couple of days the central
sea-level pressure then falls by 20 mbar or more (normal enough in middle latitudes
but unusual in the tropics), and the stage is set for the ultimate intensification. This
may still be prevented by factors such as making premature landfall and thereby
losing the crucial supply of water vapour from the warm sea surface, or entraining
unusually cold air into the centre of the vortex in the high troposphere (which
douses the developing warm core of the storm), but if it is not, then the mature
hurricane develops rapidly by formation of a relatively narrow ring of extremely
strong winds and updraughts around a central, relatively inactive eye.

STRUCTURE

Figure 12.10 is a typical satellite view of a mature hurricane. The dense white ring
extending a couple of hundred kilometres out from the centre is the shield of cirrus
fanning out in the high troposphere above the ring of updraught surrounding the
central eye. Often a small dull spot in the centre and a thin dull ring mark the inner
and outer limits of this dense shield, whose appearance amid the more diffuse mass
of high cloud is a sure sign that the hurricane has tightened into full maturity. Pene-
424 tration by special reconnaissance aircraft then reveals the full majesty of the storm.
 12.4 Tropical cyclones
60

40
N Fig. 12.9 Representative
20 tracks of tropical cyclones
(arrowed thick lines)
0
compared with mean sea-
20
surface isotherms (25 °(
s 40 dotted, 27 °( solid) for
September in the northern
60 hemisphere and March in the
180 90 w 0 E southern hemisphere. Note
that the Mercator projection
seriously underestimates the
The spiral arms of cloud stretching hundreds of kilometres outward from the fraction of the Earth's surface
area affected by such storms.
central structure, often mainly equatorwards and eastwards, are seen to be long (After [27] and [70])
curving lines of cumulonimbus. The inner zone contains a dense mass of cumulo-
nimbus and sheets of cloud spreading from them at all levels, circling faster and
faster as the centre is approached (Fig. 12.11), until the fastest flow is reached in
the vicinity of a nearly solid ring of giant cumulonimbus which often completely
encircles the eye, and produces a ring of particularly strong echo on weather radars

Fig. 12.10 Typhoon Clara at 0600 Z on 19 September 1981, centred east of Luzon (The
Philippines), as seen in the visible by the Japanese Meteorological Agency's geostationary
satellite. The eye and surrounding dense cirrus shield are particularly marked, as is the
anticyclonic diverging swirl of thinner high cloud further out on the equatorial flank. 425
12 Large-scale weather
systems in low latitudes

Fig. 12.11 Vertical radial 10

section through hurricane
E - 30
Helene on 26 September
""' I
- 20
1958. On the right-hand side
E
0>
·a; \
- 10
the solid lines ore isotherms J: 5 I
\

labelled in °C, and the 0

dashed lines are isotochs of
tangential wind speed 10
labelled in metres per 20
second. The left-hand side is
a mirror image of the right 150 100 R;dar 0 50 100 150
showing distributions of r i ng ~
cloud, precipitation, the plan Eye Rad ial dis1ance (km )
position of the strongest
radar echo, and the
boundaries of the eye wall as (Fig. 12.12). The eye itself is relatively cloud-free: a quiet amphitheatre a few tens
defined by the zone of
maximum temperature
of kilometres across surrounded by the massive, rotating eye wall of encircling
gradient. (Simplified from cumulonimbus. The temperature rises sharply inward at all levels below the very
[27]) highest troposphere, reaching a warm plateau in the vicinity of the eye wall. The
sea-level pressure falls toward the centre, slowly at first and then more and more
rapidly to the minimum value which applies across the eye and is a useful measure
of the severity of the storm. Most severe hurricanes have minimum sea-level
pressures of 950 mbar or less. Such values are not very different from the minima
in vigorous extratropical cyclones, although a few hurricanes have much lower
minima, but the low pressures in hurricanes are concentrated in very much smaller
areas, and are associated with maximum pressure gradients (in the ring of
maximum winds) which can be - 1 mbar km -I - nearly an order of magnitude
larger than the largest synoptic-scale pressure gradients associated with extra-
tropical cyclones. In fact the vigorous central parts of a hurricane are structured on
the meso-scale rather than on the synoptic scale, which rendered their detailed
observation very problematic before there were meteorological aircraft or
satellites. On a conventional surface-weather map, with sea-level isobars drawn at
2 mbar intervals, the innermost 10 or so isobars are not resolvable by synoptic
observation, and are interpolated using a simple pattern of concentric circles
spaced to match estimates of maximum winds.

WINDS

The very high winds of the hurricane are quite literally its most striking feature, and
they exceed hurricane force at the surface in an annulus usually a few tens of kilo-
metres wide. Highest winds occur to the right of the centre, facing in the direction
of the storm's overall motion, where the speeds of rotation and translation are
added (Fig. 12.13). Sustained speeds as high as 75 m s- 1 at the 10m level have
been recorded occasionally, though the chances of an anemometer being in such
fairly limited zones (and surviving) are quite small.
The effect of wind speeds on surfaces and structures increases roughly in propor-
tion to the kinetic energyofthe flow (i.e. to the square of the sustained wind speed),
though there is selective sensitivity to embedded gusts depending on their
frequency. Wind speeds above 50 m s -I have enormous effects. Over the sea they
426 raise waves - 10 m high from trough to crest, and drive sheets of spray and foam
 12.4 Tropical cyclones

DRO
77

Fig. 12.12 Radar PPI display from Brownsville, Texas of hurricane Anita centred off the Fig. 12.13 Winds in a very
Mexican coast on 2 September 1977. Note the echo-free eye surrounded by the solid ring of echo severe hurricane in the
in and beyond the eye wall. Range markers ore at 45 km intervals. northern hemisphere. Solid
streamlines in (a) show air
spiralling inwards in the low
Storm
................... propagation troposphere . while the dotted
<al (b ) streamline shows air
spiralling outwards in the
high troposphere - initially

--.... ......,
cyclonically l ike the low-level
... flow, but then
anti cyclonically. (b) The
'\ \ associated pattern of isotachs
\
I
at the 10m level. labelled in
I
m s - l. The black zone has
I

I
I winds in excess of 80 m s - 1
I
I (nearly 160 kt). (Mostly from
I
I
[91])

~~ -
--
427
11 Large-scale weather over their broken surfaces to the extent that the interface between air and water is
systems in low latitudes
blurred. Even modern well-found ships do not brave such conditions by choice,
and an accidental encounter between a United States battle fleet and a typhoon in
the Philippine Sea in 1944left 3 ships sunk, 146 aircraft destroyed aboard carriers
and 790 men dead.
As in the case of mid-latitude depressions, the areas of strong winds produce
storm surges (section 11.3) in coastal waters, and sea levels can be raised or lowered
by as much as 4 m compared with the predicted astronomic tide for a few hours,
and by half as much again in bays and estuaries. The actual response of the water is
a very complex function of storm size, strength and movement, as well as coast and
sea-bed topography, but the net result in storms making perpendicular landfall on
a coast is a substantial mound of water to the right of the storm centre (in the
northern hemisphere). In some situations resonance by surge reflection can lead to
a dwindling series of resurgences, the first couple of which may be high enough to
cause trouble. If a high surge happens to coincide with a high astronomic tide there
is obviously a serious risk of sea defences being breached in vulnerable areas,
especially since walls, dunes, etc. are then unusually exposed to battering by large
waves.
Over land, hurricane winds flatten crops, uproot trees, severely damage or
destroy weakly constructed buildings, and smash most unshuttered windows.
Exposed livestock and people obviously risk death and injury by falling debris.
Unfortunately in a way, hurricanes are sufficiently rare events at any particular
location for it not to be worthwhile for societies to be fully adapted and protected
against visitation. So the emphasis is on reliable forecasting to allow time for
shuttering of windows and evacuation of areas likely to be flooded. Unfortunately
this touches on another aspect of hurricane behaviour which is not properly under-
stood. As the outline storm tracks in Fig. 12.9 suggest, most hurricanes travel west-
ward initially in the easterlies which give them birth, but later tend to recurve
polewards. Individual paths can be quite contorted in detail for reasons which are
not well understood, making it difficult to forecast future positions.
How are such winds produced and sustained for the considerable life of a
hurricane? The concentration of strong updraughts around the eye obviously
requires a substantial inflow of suitable air, which means considerable horizontal
convergence. This is concentrated particularly in the lowest kilometre where
friction encourages the flow to spiral in across the nearly circular isobars, but it is
known that convergence must persist throughout the layer with significant cyclonic
rotation, which is usually at least 10 km deep (Fig. 12.11). In this layer con-
vergence continually concentrates pre-existing vorticity, which in the absence of
any synoptic-scale residue is the planetary vorticity f. It is simpler to go back to
basics at this point and consider the conservation of angular momentum about a
vertical axis in the centre of the eye. Then according to appendix 12.2, the
tangential wind speed U reached after a ring of air has contracted from radius R 0 (at
which it was stationary on the weather map) to a much smaller radius R, is given to
a good approximation by
U = f R2o (12. 1)
2R
At latitude 20° a contraction from nearly 145 km would produce a wind speed of
50 m s -I at radius 30 km (a typical value for the radius of the ring of maximum
winds), whereas at latitude 5° the contraction would have to be from nearly 500 km
for the same result. In fact both of these contractions are considerable under-
estimates because some surface friction is communicated throughout the rotating
428 mass by deep convection, eroding original angular momentum as the air converges.
It seems that the convergence required to generate sufficient wind speeds at 12.4 Tropical cyclones
latitudes less than about 5° is larger than the tropical troposphere can manage, so
that hurricanes are not found in this equatorial zone.
The effect of friction is to reduce the radial dependence of tangential winds from
the R- 1 predicted by the conservation of angular momentum to about R- 0 .6.
However, the diverging flow at the very top of the troposphere (which spreads the
canopy of anvil seen by satellite) is effectively above the reach of this friction,
because it originates largely from the updraughts in the eye wall, and tends to
preserve the angular momentum of the ring of maximum winds more nearly as it
diverges. The result is that, in zones outside the ring of maximum winds, wind
speeds aloft are systematically much lower than at lower levels (Fig. 12.11). In fact
when air spreads out to radii - 200 km it begins to rotate more slowly than the
Earth's surface and hence turn anticyclonically on the weather map. The resulting
anticyclonic curvature of upper-troposphere flow beyond the dense, central
canopy makes the diverging high cloud in the outer parts conform to much the
same spiral formation as the converging cyclonically curved lines of cumulonimbus
in the lower parts of the troposphere, so that together they contribute to the charac-
teristic spiral 'galactic arm' appearance which is used as the symbol for a mature
hurricane on large weather maps (Fig. 12.13).

BALANCE

The reduction in tangential wind speeds in most of the upper parts of the central
annulus of the storm (Fig. 12.11), maintained by the above mechanism, plays a
vital role in enabling the storm to balance its wind and thermal structures. In this
very active zone the horizontal accelerations arising from rotation are so large that
they must match the horizontal pressure gradient forces as closely as is the norm in
middle latitudes. However, as a rough estimate will show, the centripetal accelera-
tions arising from rapid motion around the storm's centre on the weather map far
outweigh the hidden Coriolis component, with the result that the balance is cyclo-
strophic rather than geostrophic (section 7 .12). Assuming for simplicity that all the
air at a radial distance R from the storm's centre is actually moving around that
circle (actual trajectories must obviously be complicated in detail by the radial
motion and the translation of the storm), then the cyclostrophic balance between
pressure gradient and centrifugal force is given by
V2
(12.2)
R P aR
This balance reconciles the extremely strong horizontal pressure gradients and
wind speeds found in the inner parts of the storm, but it also implies a cyclostrophic
equivalent of the thermal wind equation (section 7.10 and appendix 12.3):
gR aT
---- (12.3)
T aR
relating the vertical gradient of the square of the wind speed to the radial tempe-
rature gradient. The sense of this relation is the same as for the normal thermal
wind: positive shear is associated with low temperatures to the left of the shear in
the northern hemisphere. In the hurricane this means that the distinctively warm
core on the left requires negative shear (upward decrease of cyclonic winds) for
balanced flow- exactly what is maintained by the combination of surface friction 429
12 Large-scale weather and conservation of angular momentum. Inserting realistic wind speeds from
systems in low latitudes Fig. 12.11 into eqn (12.3) shows that the radially inward rise in temperature
through the region of maximum winds should be 3-4 oc, in reasonable agreement
with observation. The warm core of the hurricane is therefore dynamically
confined by the deep vortex of high winds whose strength decreases upwards, in
much the same way as adjacent warm and cold air in middle latitudes is confined by
the dynamic barrier whose elevated core is the associated jet stream. In each case
the warm air would spill out over the cold air, destroying the baroclinity in a few
hours, if the dynamic barrier were removed. However, in the case of the mid-
latitude front, the dynamic balance involves the Earth's rotation quite directly,
through geostrophic balance, whereas in the hurricane the cyclostrophic balance
has been constructed from the weak Coriolis effect of low latitudes by powerful
convergence. Moreover, the sense of the balanced flow around the warm core of
the hurricane requires a kind of deep, inverted jet stream with a circular core in the
lower troposphere, with obvious consequences in terms of violent surface winds.

WARM CORE

The very warm core of the hurricane is a conspicuous feature, and one which distin-
guishes it from other types of tropical disturbance, most of which are cool-cored on
the synoptic scale (though shot through by many hot towers which are warm-cored
on their much smaller scale). Though less spectacular than the winds, or the tor-
rential rain and electrical activity which mark the inner parts of a hurricane, the
warm core is quite noticeable without specialized observations, and many factual
and fictional accounts mention the oppressive warmth of the eye, where the more
violent distractions are temporarily reduced. The evolution and maintenance of the
warm core is vital for the birth and prolonged maturity of the hurricane, but at first
glance it is not easy to see how it is achieved. Hurricanes develop only over exten-
sive warm sea surfaces, where deep cumulonimbus convection is endemic anyway,
and the troposphere is consequently maintained in a nearly neutrally stable state
for such convection (section 5.12). How is it possible for hurricane cumulonimbus
to achieve the observed substantial rise in core temperature in the absence of any
hotter surface? To try to answer this question we must consider the mechanisms of
Fig. 12.14 Temperature hurricane development.
soundings in and around Figure 12.14 contains several temperature soundings in and around hurricanes.
hurricanes, plotted on a
tephigram. Curve M is the
September Caribbean
average. Curve A represents
saturated adiabatic ascent
from mean surface
conditions, whereas B
represents ascent after
isothermal depressurization
to 950 mbar (XB in inset).
Curve E represents typical
conditions in a hurricane eye.
The difference between B and
E in the upper troposphere is
largely removed by allowing
for liberation of lotent heat of
fusion in glaciating cloud.
(Based on [27))

430
The average sounding for the hurricane season is conditionally unstable through- 12.4 Tropical cyclones
out most of the depth of the troposphere, so that when convection is triggered
locally, a deep cumulonimbus will grow. However, if many showers break out in a
smallish region, the effect is for the local troposphere to become less encouraging
to deep convection if other factors do not intervene. One such factor which has
been closely studied since it was proposed by Charney and Eliassen in 1964 involves
positive feedback arising from frictional convergence in the boundary layer
supplying convective fuel to the bases of the cumulonimbus. Synoptic-scale
convergence concentrates and deepens this particularly moist layer with the
result that more cumulonimbus convection is triggered and so on in an expo-
nential development which has realistic properties when it is investigated in
sophisticated detail. The process is called conditional instability of the second kind
(CISK).
Another much simpler effect becomes important when the central surface
pressure has already fallen significantly (say by at least 20 mbar) below values
outside the developing storm. The air spiralling in towards lower pressures in the
boundary layer is then made potentially warmer through isothermal warming by
contact with the uniformly warm sea surface in the presence of falling pressure, so
that its buoyancy relative to the distant external atmosphere is significantly
increased. The process is depicted in Fig. 12.14, where you can see the advantage
gained by having the warming take place at a pressure significantly lower than the
surrounding sea-level value. Air rising in deep convection then follows a profile
which is warmer at all levels, just as if it had come from a significantly warmer sea
surface at 1000 mbar. Cumulonimbus are therefore encouraged and more heat is
pumped into the active core of the storm. This too becomes a mechanism for
development through positive feedback if the increased heat supply causes a
further fall in surface pressure, as will be seen to be the case in a moment. Meteoro-
logists are therefore especially keen to identify the cause of the initial deepening
which allows this mechanism to become effective, but the puzzle is still not clearly
resolved. In mature hurricanes the very low surface pressures ensure that the effect
is very large, equivalent to several °C, and contributes to the relatively high
efficiency of the hurricane as a thermodynamic engine.

PRESSURE

Although vertical accelerations in some of the strong updraughts (and down-

draughts) are just about significant in comparison with g, the resultant deviations
from hydrostatic balance are localized and relatively small. Thus despite the
storm's fury, hydrostatic equilibrium still prevails quite accurately on the meso and
synoptic scales, and the pressure at any level is simply a measure of the weight and
therefore of the mass of air above that level. The warm core is significantly less
dense than the surrounding air, so that volume for volume it is significantly less
massive. Since no compensating cold air or extra mass (for example in the strato-
sphere) is observed aloft, the surface pressure below the warm core must therefore
be considerably depressed. If for example the core is warmer on average by 6 K in
300 K, the decadal and other scale heights are increased by 20Jo, and it is readily
shown (Appendix 12.4) that this produces a 4.6% fall in surface pressure- nearly
50 mbar. The observed relation between severity and minimum pressure now
becomes intrinsically reasonable, to the extent that the core warmth is obviously a
measure of the energy of the storm. 431
12 Large-scale weather THE CENTRAL ZONE
systems in low latitudes
To summarize the spectacular energy and relatively fine structure of the inner
zones of a mature hurricane, imagine that you are observing the westward
approach of the centre of a hurricane in the northern hemisphere, from the safe
vantage point of a well-found lighthouse for example. The north-north-westerly
winds (crossing the isobars at more than 20° because of surface friction) have been
well above hurricane force for more than an hour, and the very low cloud base is
only just visible through the driving, torrential rain and the blizzard of spray
whipping off the very heavy seas. Lightning flickers frequently but indistinctly
through the murk, but thunder is inaudible above the roar of wind and sea. A
tornado writhes briefly by, leaning toward the south in the strong wind shear, and
marking a region of small-scale intense convergence below a particularly strong
updraught. The barometer stands at 945 mbar and has fallen 15 mbar in the last
hour. Suddenly the sky begins to lighten to the east, and the winds begin to ease.
Five minutes later they have fallen to gale force, the rain has largely stopped and
through breaks in the low cloud you can see that the main mass of cloud is
retreating to the west, giving glimpses of an enormous wall of dense cloud (the eye
wall) down which some cloud fragments are falling like a waterfall in slow motion.
A few blinks of sun show the sea running high and confused, as wave trains from
different parts of the eye wall cross and interfere, but the water surface is now
largely free of spray and foam under the dying wind. The barometer has steadied at
944 mbar, but the still well in the lighthouse basement shows that the sea level is
half a metre above the expected tide and rising, as part of the storm surge in the
local coastal waters, including an inverse barometer effect of the very low atmo-
spheric pressure. Some exhausted birds are perching on the balcony of the light-
house. The wind dies further and begins to change direction uncertainly. After
nearly an hour of abatement the wind begins to freshen again, this time from the
south, and half an hour later the opposite eye wall is glimpsed approaching from
the east and promising a virtual repeat of the earlier experience with winds raging
from the south-south-east.

12.5 Low-latitude climates

The Hadley circulation (Fig. 8.4) supplies a framework in which to summarize

significant features of low-latitude weather and climate. This covers the largely
barotropic region across which the sun is within about 10° of the zenith at least
once per year. Air-mass contrasts are weak by comparison with higher latitudes,
and there are no fronts, although thermal contrasts between land and sea can be
significant, especially when sharpened by sea breezes or monsoon flows on the
equatorial sides of the subtropical anticyclones. There are very large meridional
gradients of water content, cloud and precipitation between the arid subtropics and
the wet equatorial zone, and comparable zonal gradients reflecting distributions of
land and sea, many of which show large seasonal variation (Fig. 12.15).
The subtropical high-pressure zones include all of the hot land deserts of the
Earth. Widespread subsidence inhibits cloud and precipitation throughout the
year, with the result that the land is permanently arid and what little atmospheric
432 water vapour there is has to be advected from adjacent moist areas. The virtual
 12.5 Low-latitude cl imotes

(c)

Fig. 12.15 Mean annual

precipitation in (c)
December-February and (b)
(b) June-August. (After [92])

absence of evaporative cooling during the day, and the strong radiative cooling
through the wide open window (little cloud or vapour) at night can maintain a
diurnal surface temperature range of 50° or more (section 9.4).
On the equatorial flanks there are brief, unreliable rainy seasons when the ITCZ
reaches its poleward extent. In the North African Sahel, gross inter-annual rainfall
variation and largely unexplained longer-term variations have contributed to
human misery on an awesome scale in recent decades (section 4.10). In India, the
disposition of sea, land and mountain barriers encourage a vast monsoonal
invasion of cloud and rain in summer, which despite producing some of the highest
rainfall totals in the world on exposed hills, is still sufficiently unreliable to cause
large-scale human distress from time to time, especially near its northern
extremity.
The trade winds are fed by air flowing equatorwards from the subtropical highs.
As the air is transformed by injection of water vapour from beneath (especially
over the oceans), it becomes populated by cloudy weather systems which drift west-
wards providing much of the rainfall in those zones, and imposing substantial
rainfall enhancement on eastern sides of islands and larger land masses. The north-
ward flow across the Equator in the Indian summer monsoon is turned westwards
however, giving rainfall enhancement on western coasts and slopes (Fig. 9.11). 433
12 Large-scale weather Over the warmest western sides of oceans at latitudes of at least 5° (i.e. with a large
systems in low latitudes enough Coriolis parameter - section 12.4), some of these disturbances develop
into tropical cyclones. In addition to their more notorious behaviour, these storms
can deliver very large (sometimes catastrophic) rainfalls to the islands and coasts
they cross. Indeed some recurve polewards right out of low latitudes still raining
intensely (producing notable floods in the north-eastern USA, for example).
The trade winds from each hemisphere's Hadley cell feed the intertropical con-
vergence zone, supplying air and water vapour to the updraughts of organized
populations of tall cumulonimbus. These billow upwards to the high equatorial
tropopause, emptying torrential showers onto the underlying surfaces, and
incidentally maintaining most of the several thousand thunderstorms usually
active in the Earth's atmosphere. The organization of such hot towers is subtly
sensitive to local geography and the diurnal cycle, giving each location its
distinctive rainy season, and producing an endless display of slowly blossoming
canopies of high ice cloud on satellite pictures like Fig. 1.3.

ENSO

The structuring along zonal axes obvious in Fig. 12.15 reflects slight but significant
zonal patterning of uplift and subsidence which was first proposed by Sir Gilbert
Walker in 1923, and confirmed and elaborated by more recent observation and
computation (Fig. 12.16(a)). This Walker circulation has been found to undergo
irregular reversals every few years which were called the Southern Oscillation
because they influence weather over a large part of the south central Pacific
(Fig. 12.16). More recently the Southern Oscillation has been linked to the
notorious El Nino (the Christ Child - a local name for a hesitation in the
Humboldt Current which occurs each December and only occasionally develops
into a major breakdown), the temporary failure of the cold, Humboldt Current off
Peru, which blights the fishing industry there every few years. In fact the whole
system is now known as El Nino-Southern Oscillation (ENSO).
In the normal Walker circulation, uplift and rainfall are encouraged over
Indonesia and suppressed over the Humboldt current by a slight but persistent
vertical circulation superimposed along the centre of the Hadley circulation.
Subsidence along the Equator can split the local ITCZ by a cloud-free lane centred
along the Equator, which is clearly visible in satellite pictures, and can reach as far
west as Canton Island (central Pacific - Table 12.1 ). In the ENSO, subsidence

Flg.12.16 Vertical Ia I
equatorial section of the
Walker circulation in (a)
normal and (b) ENSO modes,
according to computations by
Tourre (1984). (After [85])

434
 Appendix 12.1
weakens over the failed Humboldt current, and uplift and rainfall are encouraged
in a large zone centred on Canton Island. Major ENSOs occurred in 1957-8,
1972-3 and 1982-3, the first of these being associated with markedly increased
rainfall in Table 12.1. In fact Fig. 12.16 shows that ENSO affects the vertical
circulation along the full length of the Equator, and there is statistical evidence that
it is related to shifts in climatic patterns at higher latitudes as well.
Schemes of teleconnections are proposed to explain these distant harmonies in
tropospheric activity, many involving interactions between atmosphere and ocean.
Though largely speculative so far, they point to a degree of interconnectedness in
both space and time hardly guessed until20 years ago, and begin to encourage hope
that some forecasting of short-term climatic fluctuations may become possible
eventually. In particular the crucial role of air-sea interactions in imposing longer-
term order on shorter-term atmospheric chaos is attracting close scrutiny.

Appendix 12.1 Drying by descent

Figure 12.17 shows the upper part of a tephigram, as relevant to the middle and
upper troposphere. The point P represents the state of air which has risen to the
300 mbar level Gust over 9 km above sea level) by saturated adiabatic ascent with
wet-bulb potential temperature 20 oc (equivalent potential temperature 61 °C}. It
is saturated by 0.53 g kg- 1 water vapour and its temperature and dew point are
- 37 °C. Its potential temperature is 60 °C, and the closeness of this to the equiva-
lent potential temperature is a measure of how little vapour there is left to condense
at such low temperatures.
If this air sinks dry-adiabatically to the 500 mbar level (about 5.5 km above sea
level) it arrives at the point Q, found by moving down the dry adiabat through P. In
fact all the potential temperatures are unaltered in the process, though the simple
temperature has risen to 0 °C. Since the vapour content of the parcel too is
unaltered by the dry adiabatic process, the dew point at the 500 mbar level is found
by sliding down the saturated specific humidity isopleth through P to R. This
shows that the dew point rises by only a little over 5 K in the process, so that the
dew-point depression at 500 mbar is over 31 K. From the ratio of saturation

Fig. 12.17 Drying by dry

adiabatic descent in the high
troposphere (on a
tephigram).

435
11 Large-scale weather specific humidities at temperature and dew point it is apparent that the relative
systems in low latitudes humidity of the subsided air is only about 7o/o. In fact such subsidence is generally
accompanied (indeed made possible) by radiative cooling. If this were such as to
cool the sinking air along the 20 oc saturated adiabat (but notice this is dry rather
than saturated descent), the air would reach the 500 mbar level at S with a
temperature of - 9 °C, and a relative humidity which, as you should check, is
about 13%.

Appendix 12.2 Convergence and conservation of

angular momentum

Consider a narrow circular horizontal ring of air which begins at rest with radius R 0
on the Earth's surface, and then shrinks to radius R while conserving angular
momentum about the local vertical. Initially it has only the angular velocity of the
surface itself, which isf/2, wherefis the Coriolis parameter. Arbitrarily assuming
unit mass for the ring, the initial angular momentum about the centre of the circle is
R'ijf/2. After shrinking to radius R, suppose that the ring is turning with tangential
speed Urelative to the map. This contributes an additional angular velocity U/ R to
the planetary f/2, so that the total angular momentum is now R 2 (U/R + f/2).
Equating initial and final expressions for angular momentum and rearranging we
find

U = _j_(R~ - R 2)
2R u

When the first term in brackets is much larger than the second we have

fR~
U=. --
2R
We can rearrange the latter to find the initial radius needed to produce a required U
at a given R, 50 m s- 1 at 30 km being reasonable for an intense hurricane.
2RU
f
Notice that R 0 increases with decreasing!, i.e. decreasing latitude.

Appendix 12.3 Thermal wind in cyclostrophic

balance

The isobaric form of the equation for cyclostrophically balanced flow at speed V
436 round a circle of radius R is
 az Appendix 12.4
V 2 =Rg--P-
aR
where az/aR is the upward and outward slope of the isobaric surface which
supplies the inward force to maintain the centripetal acceleration of the whirling air
parcels. Proceeding as in appendix 7.9 we apply this basic equation to two isobaric
surfaces and subtract to find

= Rga
V22 - V I2 aR- (ZP2 - Z PI )

where the right-hand side contains the radial thickness gradient. As in appendix 7. 9
we replace the thickness gradient by the gradient of the column mean temperature
T.
Vz2- Vl2 Rg aTP
Zp2 - Zp 1 T aR
If we consider an extremely thin layer then the differential form appears
Rg arp
----
T aR
The strongly negative radial temperature gradient in a hurricane therefore requires
an upward reduction of wind speed.

Appendix 12.4 Low pressure and warm core

Suppose that the full height of the uniformly warm core of a hurricane is z. If the
scale height of the isothermal core isH, then according to eqn (4.8) the surface
pressure p, is related to the pressure p 1 at the top of the core by

p, = P 1 exp ( ~ )

For a fixed z and p 1 we can find the sensitivity of p, to H by differentiating

dp, - z
- p, H2
dH
or
( ~) ~
dp, -
-
p,
The latter expression shows that the fractional (or percentage) change in surface
pressure is proportional to the fractional change in scale height, the constant of
proportionality being z/ H. The minus sign ensures that a rise in scale height is asso-
ciated with a fall in surface pressure. Since His directly proportional to the core
temperature T (eqn (4. 7)), we have established the sensitivity of surface pressure to
core temperature. In the case of a hurricane z may be as large as 16 km, so that z/H
may approach 2.3, the ratio of decadal to exponential scale heights.
Note that the above is a computational procedure to find the sensitivity of p, to H
and therefore T, and incidentally to demonstrate another usage of scale height. The 437
11 Large-scale weather relationships above do not show how the surface pressure in a particular core may
systems in low latitudes develop during the life of a hurricane, except in so far as z and p 1 do actually remain
constant, which is a sweeping oversimplification.

Problems on Chapters 11 and 12

Note that the northern hemisphere is assumed unless otherwise stated.

LEVEL 1

1. In not more than one sentence in each case, cite observational evidence for the
following large-scale properties of developing extratropical cyclones: broad-
scale uplift; convergence in the low troposphere; divergence in the upper
troposphere.
2. Whereabouts is the observer of the following situation most likely to be in
relation to the classical model of a family of depressions? Pressure is rising and
falling by about 20 mbar every couple of days; winds are strongly westerly,
backing a little as cloud and rain move in during each fall of pressure, and
veering in the subsequent clearances.
3. A square-rigged ship is making northwards in strong westerly winds just to the
east of an advancing cold front. If no action is taken by the crew, what is likely
to be the situation of the ship after the passage of the surface front?
4. To the north of an old centre of low pressure a mass of cloud is observed
advancing from the east, with the edge of its high cloud moving quickly
northwards. Decide whether the front is a warm or cold front, and outline
your reasoning.
5. Given that air at two different levels 2 km apart on a vertical sounding have
risen in parallel conveyor belts with a gradient of about 1 in 100, estimate the
horizontal separation of their surface sources.
6. In the conveyor-belt picture of airflows in fronts, a large region of heavy
rainfall is likely to be associated with what properties of the flow?
7. Cold damage to evergreen foliage was concentrated on the eastern sides of
plants in the severe winter of 1963 in the British Isles. What can you say about
the position of the centre of the blocking high?
8. Outline the reason for saying that the rapid onset of continental winters in
middle and high latitudes is often associated with the radiative behaviour of
snow.

LEVEL 2

9. A low-pressure centre with sea-level pressure 980 mbar lies not far from a ridge
with pressure 1012 mbar. Find the difference in sea level associated hydro-
statically with this pressure difference.
438 10. If air has been risen to the 500 mbar level along a sloping warm conveyor belt
 with a gradient of about 1 in 75, estimate the geographical position of the Problems on Chapters 11 and
region in which it was in the atmospheric boundary layer. 12
11. Air enters the upper troposphere of an anticyclone (at the 400 mbar level) with
a temperature of - 30 oc and is saturated with respect to supercooled water.
If it sinks to the 700 mbar level in 3 days, cooling by net radiation at an
equivalent isobaric rate of 2 oc per day, find its temperature and relative
humidity at the 700 mbar level. Use the tephigram in Fig. 5.5.
12. Using the method of problems 7.23 and 24, find the average horizontal diver-
gence which is needed in the lower troposphere of an anticyclone to accom-
modate a rate of subsidence of 1 em s- 1 in the middle troposphere there.
13. Using the expression for the speed of eastward propagation of a Ross by wave,
estimate the speed of a cyclone wave of wavelength 2000 km in middle
latitudes. Comment briefly on the inconsistency of the implied assumption.
14. Estimate the depth of a seasonal heat low, such as that over north Africa,
associated with a layer of air 2 km deep which is on average 10 oc warmer
than comparable air over adjacent seas.
15. Using the method of appendix 12.2, follow air at latitude 20° as it flows
outwards from a position in the high troposphere just inside the eye wall of a
hurricane, where it was moving cyclonically at speed 30 m s - 1 round a circle of
radius 30 km. Assuming conservation of absolute angular momentum about
the vertical in the centre of the eye, find the radius at which the outflow will
stop turning cyclonically relative to the surface.

LEVEL 3

16. Compare and contrast the isobaric and isentropic views of air flow in a mature
depression, citing the observations which the latter fits, as well as those it does
not.
17. Compare and contrast the synoptic-scale structures of a depression and a
blocking anticyclone in middle latitudes.
18. Explain carefully how a subsidence inversion is maintained in the low
troposphere of an anticyclone.
19. In problem 12.12, the subsidence associated with radiative cooling has been
ignored. Assuming that the low troposphere as a whole is cooling at 2 °C per
day in the anticyclone, estimate the associated rate of subsidence in the mid-
troposphere. The original omission is therefore justified.
20. Draw up a list of all meteorological effects which involve the meridional
variation of the Coriolis parameter, including a brief description of the
involvement in each case.
21. What differences in regional climate would you expect if the Earth were
entirely covered by ocean? (Note that climatologists might then be dolphins!)
22. Discuss the validity of saying that the troposphere in low latitudes is largely but
not exactly barotropic.
23. How can tropical cyclones develop conspicuously warm cores over nearly
isothermal ocean surfaces, and how are they maintained?

439
CHAPTER
The atmospheric
13 eng1ne
•

13.1 Introduction

It was mentioned in Chapter 1 that when flows of energy are considered, the atmo-
sphere is seen to be acting like a very large, self-regulating heat engine, whose winds
and weather systems correspond to the moving parts (pistons, flywheels etc.) of
man-made engines. It is true that the atmospheric engine differs from the ones we
construct in many important respects: it is entirely fluid, and has the enviable
ability to maintain and regenerate its structures in the course of normal operation;
it is very complex, and its activity ranges in intensity from the barely measurable to
the overwhelmingiy powerful on the human scale; and of course it operates on an
enormous range of space and time scales, as detailed throughout this book. But
despite such apparent differences, the comparison between the atmosphere (and
other natural heat engines) and man-made heat engines is quite strict, and in
particular they all obey the same physical laws. Two which constrain them overall
and in every part are amongst the most fundamental physical laws known to
mankind - the first and second laws of thermodynamics - and we conclude this
introduction to meteorology by applying these two laws to highly simplified
models of the troposphere and stratosphere as connected heat engines. Figure 13.1
shows the two engines in schematic form: there is a powerful primary engine in the
troposphere and a weak secondary one in the stratosphere. We will trace the
production and subsequent life history of the energy which passes through the
atmospheric engines, keeping them working and maintaining the activity and
structure of the atmosphere in the process.
Energy comes in a variety of apparently very different forms, such as heat,
kinetic energy, potential energy and chemical energy, but it is the great strength of
the concept of energy that each of these is seen to be a different form of the same
fundamental thing, and that they can transform according to fixed rates of
exchange which have been established by theory and experiment. In fact a common
currency (the joule) is now in use in all but the most specialized areas. The ultimate
unification is expressed in the principle of conservation of energy which states that
if all forms of energy relevant to any particular system are accounted for, including
fluxes into and out of the system, then the energy total is conserved. When heat is
one of the forms of energy involved, the principle is known as the first law of
440 thermodynamics. The first law (as we will simply call it) is fundamental to all
 13.1 Introduction
Solar Terrestrial
input output

Upper
,----------.,
Ip I
Stratosphere
1-----;-:---i r3 r4 1---'-:--------.; Lower
L_____ ;_~ __ J
/ / / / / //1////////////// //i:J// / / / / , / / / / / / / / / / Tropopause

Fig. 13.1 Schematic outline

. .-a-
: E ---.-,i +'02
of the tropospheric heat
engine (E) and the
:
I
T2 M:
I
j
:
Troposphere
stratospheric heat pump (P).
-t·····'···-····1 Friction f-- Radiant-energy fluxes are
r,
1
1 1
represented by wavy lines,

-
I I
: ' o, : heat fluxes by straight solid
'---- ---"' lines, and fluxes of
mechanical energy by dotted
Surface
lines.

energy budgets, since it constrains them to balance in whole and in part. In fact it
can be used to determine a component which cannot be measured well enough in its
own right, provided all the other components of the energy budget for that
particular situation are known. This approach has already been used at various
points in the book, especially in Chapter 8, where the imbalance in radiative energy
fluxes was used to point to the presence and importance of the convective and
advective heat fluxes. We will now use the concept of energy more generally,
including the mechanical forms (potential and kinetic energy), to establish an
overall framework for the whole range of dynamic atmospheric behaviour, much
of it weather-related.
The first law is powerful but undiscriminating. It requires that energy budgets
balance, but usually does not determine how the total should be divided between
the relevant components in any particular situation. Towards the end of this
chapter we will be able to make a little use of the second law of thermodynamics,
which was developed out of experience with early steam engines but which is now
seen to be one of the cornerstones of the natural sciences, largely because of its
ability to describe the direction of energy flow in systems and to discriminate
between the very many apparently possible types of mechanism and the much
fewer types which actually occur.
The atmospheric engine has basic features in common with all heat engines,
however different they may be in detail. It has heat sources, where heat energy
effectively enters the system, locations of energy transformation (and our interest
is especially in the zones of production of mechanical energy), and heat sinks,
where heat leaves to go to some other system. For present purposes all the input Q 1
of solar heat to the tropospheric engine is supposed to occur at a single temperature
T 1 which is presumably close to the average temperature of the base of the
troposphere. This ignores the substantial minority· of solar input which is
distributed through the depth of the troposphere (Fig. 8.6). However, this fraction
is a secondary feature since it is usually more than offset by the net output of
terrestrial radiation from the same zone.
Using this surface heat input, the tropospheric engine generates mechanical
energy M by a wide variety of mechanisms, examples of which have appeared
throughout this book, but which obviously include the Hadley circulation in low
latitudes, the monsoons, the depressions and anticyclones of middle and higher
latitudes, and the sporadic cumulus convection found at all latitudes. However, 441
13 The atmospheric engine any particular package of mechanical energy is present only temporarily in the
troubled state of atmosphere and ocean, since it is eventually transformed to heat
by friction of various types. Some of it then joins the much larger flux of heat
which has passed through the tropospheric engine in the convective and advective
heat fluxes (labelled Q2 in Fig. 13.1), before being exhausted to space in the form of
terrestrial radiation. The rest joins the cascade of terrestrial radiation at lower
altitudes, including the surfaces boundary layer with its concentration of turbulent
friction. For simplicity the output too is assumed to occur at a single temperature
T2 , though it must in principle involve a wide range of tropospheric temperatures
because of the variation of tropospheric transparency with radiative wavelength.
However, it is reasonable to assume that the value of T2 should be close to the
temperature of the highest opaque layer of the atmosphere (section 8.4). Above
this level, which is in the upper troposphere on account of the vertical distribution
of water vapour, cloud and carbon dioxide, convection and advection are
insignificant and the outward energy flux is carried mainly in the form of terrestrial
radiation with no further dynamical significance. Notice that the heat flux through
the tropospheric engine in Fig. 13.1 does not contain the considerable but purely
radiative fluxes passing through the window and in the cascade of absorption and
emission (section 8.4), since they pass through the atmosphere without requiring or
encouraging any atmospheric motion (a point considered further in section 13.4).
The oceans interact thermally and dynamically with the troposphere, and so in
principle should be represented in Fig. 13.1. But, as mentioned in Chapter 9, the
oceans are dynamically passive in that they are driven by the troposphere to pump
heat polewards in surface currents like the Gulf Stream. In addition their huge
thermal capacity has a powerful smoothing effect on seasonal variations in surface
temperature. Hence the main effects of the oceans are to assist the atmosphere in
the advection of heat and to smooth the seasonal and latitudinal variations of the
temperature T1 of the heat input to the tropospheric engine. There is a small,
intrinsically oceanic engine which transports heat polewards, and which is driven
by the descent of cold, dense polar waters, but the heat flux it carries is trivial in
comparison with the wind-driven flux.
In comparison with the troposphere, the stratosphere is relatively inactive, as
mentioned in Chapters 3 and 8. In the radiative budget of the stratosphere
(Fig. 8.6) the absorption of solar ultraviolet is quite closely balanced by the output
of terrestrial radiation, according to the scheme outlined in section 3.2. Actually
the upper stratosphere in particular is far from inactive, but the amounts of
mechanical energy corresponding to the motion of such a tenuous gas are very
small in comparison with tropospheric values. In fact there must be slight
imbalances of the stratospheric budget because the low stratosphere in particular is
known to a:ct as a heat pump (a heat engine working in reverse), forcing heat from
the relatively cool and high equatorial zone (at temperature T3) to the warmer and
lower polar zone at T4 • The mechanism is believed to involve synoptic-scale eddies
driven by a small leakage of mechanical energy from tropospheric long waves (44].
For simplicity friction has been ignored in the stratosphere, and the temperatures
of source and sink assumed to have single values as usual. After transformation,
the heat pumped to the polar regions contributes a small component to the outward
flux of terrestrial radiation there.

13.2 Forms of energy

442 Energy exists in the atmosphere as mechanical energy and as heat. Mechanical
energy comes in two forms: kinetic energy, which is the energy which a body has an 13.2 Forms of energy
account of its motion, and potential energy, which is the energy it has on account of
its position in a field of force. Since gravity is the only force field which is
significant in the lower atmosphere (electrostatic and electromagnetic forces are
relatively very weak even in the immediate vicinity of thunderstorms), the only type
of potential energy we have to consider is gravitational potential energy. This will
be called simply potential energy from now on, the 'gravitational' being
understood. (In the upper atmosphere by contrast electromagnetic forces can be
very important [11].) An expression for the rate of change of mechanical energy of
an air parcel of unit mass can be derived from the equation of motion, as shown in
Appendix 13.1.
d
dt (\12 V2+ g z) = POW- FW (13.1)

The terms in brackets on the left-hand side are the familiar expressions for the
kinetic and potential energies of a body moving with speed Vat height z above an
arbitrary datum level, which is usually mean sea level in the atmospheric case. The
terms on the right-hand side are written as convenient initials to avoid needless
complication. PG W represents the rate of working of ambient pressure gradients
on the parcel, and FW represents the rate of working of external friction. Notice
that there are no Corio lis terms in the equation. As shown in appendix 13.1 these
disappear because their associated apparent forces do not work and hence have no
effect on any type of energy. If the right-hand side is zero, then the sum of the
kinetic and potential energies of the parcel is fixed. However, the apportionment
between the two energy forms varies with parcel height, kinetic increasing at the
expense of potential when the parcel falls and vice versa. This is the familiar
behaviour of a freely falling body, but the tendency remains even when the right-
hand side is not zero, and is an example of the conversion from one type of energy
to another which results from the action of a force, in this case the force of gravity.
In a neutrally stable atmosphere (section 5.1 0) the conversion is exactly opposed by
the working of the vertical pressure gradient, with the result that no kinetic energy
is evolved as the parcel moves up or down.
Heat energy exists in the atmosphere as both sensible and latent heat. The first is
measured by the internal energy of the air parcel (section 5 .2), which is the kinetic
energy of its constituent molecules in their chaotic thermal motion, and the second
is the energy which is latent in the presence of water vapour, and can be realized
(i.e. converted to sensible heat) by condensation. For simplicity we will retain only
the sensible heat terms in the following. This omission will not matter too much
when we consider large masses of air rather than small parcels because the role of
latent heat is then largely internal, for example tending to raise the altitude of
effective solar warming towards cloud level without altering the total rate of input.
We can derive an expression for the rate of change of internal energy very quickly
from the first law of thermodynamics, as mentioned in Appendix 13.1:

~ (C T) = dQ + PC (13.2)
dt v dt
The first term on the right-hand side is the diabatic heating term, which in the
present context is the parcel's net rate of gain (or loss) of heat by radiation, either
by direct atmospheric absorption and emission, or by convective contact with a
radiatively warmed or cooled surface, or both. The term PC represents the
compressing effect of ambient pressure in raising internal energy. When dQ/dt is
zero, eqn (13.2) describes the adiabatic warming or cooling of an air parcel as
discussed in Chapter 5. 443
13 The atmospheric engine We find the expression for the rate of change of the total energy (kinetic plus
potential plus internal) of a parcel simply by adding eqns (13.1) and (13.2).

~(Y2V2+gz+C T) = dQ + PW- FW (13.3)

dt v dt
The term PW is the sum of PG Wand PC, and represents the total rate of working
by the pressure field on the parcel.
In dealing with the atmosphere on a large scale we are not so much interested in
the instantaneous behaviour of individual air parcels as with the average behaviour
of large air masses. This might seem likely to complicate matters, since we will have
to sum each of the very variable terms in eqn (13.3) throughout extensive volumes
and over considerable time periods. But though it is true that a full treatment leads
to involved algebra, there are some very important ways in which the aggregated
form of eqn (13.3) is simpler than a full aggregation of parcel expressions.
First let us consider a column of air extending from the Earth's surface to the top
of the atmosphere. The column total of internal energy per unit horizontal area is
simply

IE= Cv J: Tdm (13.4)

where dm is the mass per unit area of a very thin horizontal slice of the column, and
m is the total mass of the atmosphere resting on unit horizontal area. A similar
expression gives the column total kinetic energy:

KE = Y2 J~ V2 dm
However, if we assume hydrostatic equilibrium, as we can to a high level of
accuracy in most atmospheric behaviour, we find (Appendix 13 .2) that the column
total of potential energy can be transformed from its obvious form

PE = g J: zdm

into the expression

PE = R [Tdm (13.5)

where R is the specific gas constant for air. Comparison of eqns (13.4) and (13.5)
shows that PE is directly proportional to IE, the constant of proportionality being
RICv. The physical reason for this is fairly obvious: in hydrostatic equilibrium the
vertical distribution of atmospheric mass, and hence of potential energy, is
determined by the air density, which is inversely proportional to absolute air
temperature according to the equation of state. If air temperature rises, air density
falls and the column expands vertically, raising its potential energy in step with its
internal energy.lt follows from eqns (13.4) and (13.5) and the definition of CP that
we can define a total potential energy (TPE) which is the sum of PE and IE and is
given by

TPE == CP [ Tdm (13.6)

From one viewpoint this describes the effective heat capacity of the column of air
444 (another example of the general relevance of CP as distinct from Cv - see section
5.3 and appendix 13.3). For example, in the simple case of a column extending to 13.2 Forms of energy
the top of the atmosphere with uniform temperature T and mass m per unit area,
the heat capacity per unit area is CPm, and the temperature will rise by QI(Cpm) if
there is a heat input of Q. From another viewpoint the relation

TPE = IE + PE = c. TPE + _B_ TPE

cp cp
describes the apportionment of heat input between IE and PE, and suggests that
both are reservoirs of energy potentially capable of conversion into kinetic energy
(KE). So whereas all fluids and mobile material can convert PE into KE, a readily
compressible fluid like air can in addition convert IE in ways which we will not
consider in detail but which obviously include the loss of IE in adiabatic
decompression.
Now let us now extend the aggregation of the column-totalled terms in eqn (13.3)
to cover a wide horizontal area bounded by vertical walls thousands of kilometres
apart. The pressure working term effectively vanishes since there is no pressure
working across the top or bottom surfaces, and the import and export of energy
across the relatively very shallow walls is very small. (The constituentPWterms act
overwhelmingly to redistribute energy within the very large bounded volume.) To
this extent the equivalent version of eqn (13.3) for the very large volume simplifies
to

~(KE+TPE) = dQ- FW (13.7)

dt dt
where the terms now represent horizontal aggregations of the column totals
mentioned above. If the atmospheric volume is extended to cover the entire globe,
or one hemisphere averaged over one year, then the observed nearly steady state
implies that the aggregated total energy KE + TPE is nearly constant and the left-
hand side of eqn (13.7) is as nearly zero. In an adiabatic and frictionless
atmosphere this is clearly consistent with the fact that each term on the right-hand
side is permanently zero. In the real atmosphere it implies that the diabatic and
frictional terms effectively cancel. At first sight such a balance seems unlikely,
since we have assumed at many previous points that the Earth's surface and
atmosphere are in thermal equilibrium in the long term, implying zero dQ/dt.
However, the dQ/ dt in eqn (13. 7) is the portion of the heat input to the atmospheric
system which is converted into mechanical energy. After this and subsequent
conversions associated with details of the dynamical behaviour of the atmosphere,
it is reconverted back to heat by frictional working FW, producing heat at a rate
which, at least on the large space and time scales assumed here, exactly balances the
diabatic heat input dQ/dt. The significance of the negative frictional term should
now be clear, as should its role in restraining the atmospheric motion maintained
by the diabatic heat input. The result is that although heat has been temporarily
extracted to drive the atmospheric system, overall thermal equilibrium is
maintained by its return through frictional degradation of the kinetic energy of
atmospheric motion.
Let us now consider the flow of energy through the global atmosphere assuming
a steady, dynamic state (Fig. 13.2). The flow follows the scheme outlined verbally
in the previous section: energy enters in the form of net radiative input Q which
warms the air in a variety of direct and indirect ways. As the air warms it converts
the input into TPE. The warmer, taller columns of air may then give rise to
atmospheric motion in ways to be described, converting some of the TPE further
into KE. Once in kinetic form the energy begins to be degraded by friction, ending 445
13 The atmospheric engine Solar input Terrestrial output

Fig. 13.2 Schematic outline

of the flow of energy through
dI
the atmosphere in a
thermally steady state. Terms dO =I P.K I = FW
are defined in the text. dt

up as heat which is exhausted to space as part of the terrestrial radiation. According

to the steady-state assumption, the rate of input dQ/dt, the rate of conversion from
TPEto KE (represented by {P.K} ), and the rate of frictional degradation of kinetic
energy are all equal.
The boxes in Fig. 13.2 represent the reservoirs of TPE and KE which are main-
tained by the dynamic equilibrium of the atmosphere, and we can easily estimate
their approximate values. Assuming a mean atmospheric temperature of 260 K
and mass per square metre of 10 000 kg, the TPE in a column resting on unit
horizontal area is 2.6 x 103 MJ m- 2 • However, typical wind speeds of about
15 m s -I, together with the same mass, are equivalent to a KE value of just over
1 MJ m- 2 • This is very much smaller than the pool of TPE, but a little thought
shows that any surprise we feel at this inequality arises from a quite unwarranted
assumption that TPE should be completely converted to KE by atmospheric
systems. Zero TPE would be reached only if the atmosphere were reduced to
absolute-zero temperature and zero volume at sea level. In reality of course the
temperature of any substantial part of the atmosphere is unlikely to fall far below
the appropriate average value, which means that a large part of its TPE is
unavailable for conversion toKE. It is therefore more meaningful to replace TPE
by the very small portion of it which is available potential energy (APE), i.e.
realistically available for conversion to kinetic energy. Though the large
unavailable portion corresponds to the warmth and other vital static aspects of the
atmosphere, it plays no role in the chain of energy transformations.
We can get an impression of just how little TPE is available by considering a very
simple incompressible model of mid-latitude baroclinity. Two columns of incom-
pressible 'air' stand side by side, and are initially identical. Then one is heated so
that it expands vertically, increasing its PE by raising its centre of gravity. The
columns are now as depicted in Fig. 13.3(a)- a set-up which is clearly unstable in
the sense that the columns can be rearranged to lower their common centre of
gravity, freeing some of their PE (IE is irrelevant in an incompressible fluid) for
conversion to KE. A very simple re-arrangement is depicted in Fig. 13.3(b), in
which the denser air spreads horizontally under the heated air, which spreads
evenly over the top without mixing. Calculation of PE before and after
rearrangement (appendix 13.3) shows that the fall in PEas a fraction of the total
PE is given by

Fig. 13.3 Vertical columns of Ia I lbl

more and less dense fluid (a)
before and (b) after
rearrangement to minimize r-fl6z 6zj2
their gravitational potential

tl:J'
energy.

446
 13.2 Forms of energy
6PE /).p
PE 4p
if the fractional difference 6p/ p between initial densities is small. Being in fact a
perfect gas, the thermal expansion of air is such that a small fractional change in
density is equal to the small fractional change in temperature causing it, so that the
fractional drop in PEcan be written in the following form to relate to atmospheric
conditions:
6PE 6T
(13.8)
PE 4T
If we imagine that the model represents adjacent columns of polar and tropical air,
then 6 TIT is about 20/300, and it follows that in this very simple case only about
one sixtieth of the potential energy is made available by rearrangement. Figure 13.4
also depicts the even more obvious instability which results when denser air overlies
less dense air. When the layers are vertically interchanged without mixing, the fall
in PE is twice that in eqn (13 .8), but the corresponding atmospheric value is smaller
because convection limits 6 Tto a very few degrees by preventing accumulations of
convective instability.

!al (b) Fig. 13.4 Horizontal layers

of more and less dense fluid

E8
(a) before and (b) after
~ rearrangement to minimize

Li_] 1
their gravitational potential
energy. Condition (a) is
convectively unstable.

Calculations of this sort have been made for a wide variety of model
atmospheres since they were first attempted by the Ukrainian pioneer dynamicist
Margules in 1903 when he invented the concept of APE. The values of APE/TPE
which we have just found are in fact rather large: considerably smaller values are
found when continuous gradients of density are allowed, and when the slight but
systematic vertical stability of most of the real atmosphere is included in the bam-
clinic case (convective instability being confined to a small fraction of the
atmosphere at any instant). Regardless of detail, all models show that an essential
requirement for the release of any potential energy is that the initial warming
should be inhomogeneous. If the two columns in Fig. 13.3(a) had been heated
equally there would have been no APE at all, as would have been the case in the
convective model if the upper layer had been heated as much as the lower. Even the
simplest examination of APE therefore points to the very important fact that not
all heat inputs will lead to the production of KE- spatial inequality of heat input is
crucial. And even when there is production of KE, the conversion from PE (TPE in
the compressible case) is very small.
In the actual compressible atmosphere the available potential energy of any
particular configuration can be defined as the maximum TPE which can be made
available by adiabatic rearrangement (being compressible there are temperature
changes as the air rearranges but no heat is gained or lost overall if the movements
are adiabatic). The necessarily detailed and complex calculations needed to find the
APE of typical dispositions of atmospheric densities (in fact potential density, by
analogy with temperature and potential temperature) indicate that in the real
atmosphere ratherless than 0. 5Olo of the TPE is available at any instant. This means
that the APE corresponding to the initial estimation of typical atmospheric TPE is
probably less than 13 MJ m - 2 , much nearer the estimated KE value, though still 447
13 The atmospheric engine somewhat larger. In Fig. 13.2 therefore the TPE box should be relabelled APE,
and its much smaller value understood, but the rate of transformation {P.K} toKE
is unchanged in both physical significance and numerical value. If we regard the
value of APE as a measure of the degree to which the atmosphere is kept off
balance by the continual throughput of solar energy, then it seems that this
imbalance is very small, presumably because the solar input is fairly gentle, its
gradients quite small, and the mobility and rapidity of response of the atmosphere
are sufficient to prevent any sizeable accumulation of available potential energy.

13.3 Energy paths

The chain of energy transformations leading to atmospheric motion begins in the

unequal heating of the atmosphere by imbalance or warming by solar radiation and
cooling by terrestrial radiation. Of the two main inequalities, the concentration of
net warming in low latitudes leads to the motion systems associated with large areas
of substantial horizontal winds such as the trades and the mid-latitude westerlies,
including the much smaller but still quite extensive areas of very strong horizontal
winds such as jet streams. By contrast, the concentration of net warming in the base
of the troposphere leads to widespread but highly localized and frictionally limited
vertical motion associated with the whole range of vertical convection, from the
shimmering dance of air over heated land surfaces to towering cumulonimbus.
Because observation and theory of vertical convection is still relatively incomplete,
we will concentrate in the following on the large-scale systems of horizontal winds,
using the energy types introduced in the previous section to outline cycles of energy
transformation and, in the process, summarizing many aspects of the general
circulation of the troposphere.
We must expect the variation of net radiative warming with latitude to maintain
APE in the form of warmer and taller air columns in lower latitudes which will
continually tend to over-run their shorter and cooler neighbours at higher lat~tudes
(like Fig. 13.3 in principle), producing KE in the form of horizontal winds in the
process. The thermal-wind relation associated with the meridional temperature
gradient through the quasi-geostrophic nature of flow in all but the lowest latitudes
would lead us to expect these winds to be westerlies. In conventional jargon this
would be called conversion from zonal APE to zonal KE, written as Az and Kz
respectively in Fig. 13.5, where the rate of this conversion is represented by
{Az.Kz}. The actual algebra of the terms represented by such symbols is quite
complex, and varies with the precise nature of the averaging used and the
simplifications made in the theoretical framework, but extensive work in recent
decades has produced a fair consensus of data and interpretation. Data from the
northern hemisphere still predominate, and behaviour is more clearly observed and
understood in middle latitudes than it is close to the equator, but the global annual
picture is agreed in broad outline, together with seasonal and gross meridional
deviations. It seems clear from this that the expected direct conversion from zonal
APE to zonal KE is relatively unimportant in the annual picture.
The explanation for this surprising result depends largely on the presence of the
long (planetary) waves which are such a basic feature of the middle and upper
troposphere in middle latitudes (Fig. 11.22). Though the zonal APE is certainly the
448 first stage in the energy cycle after radiative heat input, it is apparently followed by
a radical redistribution of APE in which huge fingers of warm air from low 13.3 Energy paths
latitudes are interleaved around the hemisphere with equally large fingers of cold
air from high latitudes. In technical terms there is a transformation from zonal
APE to eddy APE (Aeon Fig. 13.5) at a rate symbolized by {Az.Ae}, the symbols
again representing complex terms arising from the aggregated energy equations.
The term 'eddy' here refers to any deviation from a hemispheric average around a
zone of latitude, and therefore includes the effects of monsoons and other geogra-
phically tied variations, as well as the mobile cyclone waves and planetary waves of
middle latitudes. Descriptively, the initial contrast between hemispheric zonal
rings of warmed air in low latitudes and cooled air in high latitudes has been
localized and sharpened in the contrasts between adjacent warm and cold fingers.

0.5
Fig. 13.5 Energy reservoirs
and rates of conversion from
data covering 12 months in
the northern-hemisphere
troposphere. Fluxes and
conversion rates are
expressed in W m - 2 and
reservoir contents (within
box) are expressed in
1.8 MJ m-2. (Adapted from (44])

The stage is now set for the conversion of APE to KE. The weather systems
which feed on and maintain the baroclinity of the middle latitudes convert eddy
APE toKE by sporadic, repeated eruptions of baroclinic instability, producing
families of depressions and their associated fronts and jet streams. The conversion
is mainly to eddy KE in the first place almost by definition, since planetary and
cyclone waves are associated with localizations of KE, such as jet stream cores and
areas of surface gales. The rate of transformation is symbolized by {Ae.Ke}.
The crucial feature of all systems converting PE to KE is that less dense (i.e.
warmer) air rises while more dense air sinks, so that the overall centre of gravity
falls, just as in the simplest model (Fig. 13.3). This is known as a thermally direct
circulation, and all self-driving motion systems ranging from centimetric-scale
convection to extratropical cyclones and the Hadley circulation are of this type. It
is sometimes said that all systems which produce KE must therefore be warm-
cored, which is ultimately true, but may not be very helpful when the system has a
complex structure. For example, on a superficial view we might be tempted to
regard the core of a mature depression as the axis of the pressure minima at
successively higher levels which slopes upwards and westwards from the surface
low, especially since the latter is often called the centre of the low-pressure system.
Since this 'core' contains relatively cool air (in fact the pressure trough aloft is
enhanced by the consequent depression of thickness), and the presence of some
precipitation shows that there is some rising air, it would therefore seem that the
depression is a thermally indirect system, consuming KE by converting it into PE.
However, more perceptive observation suggests that the warm conveyor belt
(Fig. 11.15) is the true warm core, together with any cooler conveyor belts there
may be which are still warmer than the area average, and measurement shows that
their KE production far outweighs KE consumption in rising cold air. In fact any
complicated system is likely to have parts in which local energy conversions run
counter to the overall trend, i.e.localized thermally indirect parts which are driven
by the main thermally direct circulations nearby.
The sporadic and extensive production of eddy KE in the large weather systems 449
13 The atmospheric engine of middle latitudes maintains the endless westerly flow there by subsequent
conversion to zonal KE. In fact westerly zonal motion in a ring of air bounded by a
lower latitude is enhanced by large-scale eddies if poleward motion across that
latitude is systematically associated with stronger westerly flow, and equatorward
flow with weaker westerlies. In the northern hemisphere this means that southerly
and westerly winds should be positively correlated, which is generally observed to
be the case, though it does not apply in every instance. For example such correla-
tion is apparent in the fairly obvious tendency for the south-westerly jet associated
with the cold front (which includes the warm conveyor belt in its lower parts) to be
stronger than the north-westerly jet lying ahead of the warm front. It is also
apparent in the tendency for the axes of cyclone waves to be angled somewhat from
south-west to north-east (Fig. 11.10). More fundamentally, both of these
correlations result from the tendency to conserve angular momentum as air moves
polewards and reduces its perpendicular distance from the Earth's axis of rotation
(section 7 .14). The resulting rate of conversion from eddy KEto zonal KE is repre-
sented by the symbol {Ke.Kz}.
These conversions and others to be mentioned below are depicted in Fig. 13.5,
together with numerical values taken from one of many detailed budgets calculated
from annual data from the northern hemisphere [44]. The production of APE
from net radiative warming of low latitudes and cooling of high latitudes starts at
the top left, and proceeds at a rate of just over 3 W m- 2 , when expressed as an
energy flux per unit area of the hemispheric surface. The precise numerical value
varies considerably between analyses, but is never more than a few watts per square
metre, which is only a small fraction of the vertical heat flux borne by large-scale
motions of the atmospheric engine (- 50 W m- 2, according to section 9. 5). This
rate of production maintains a reservoir of APE which is only 4 MJ m- 2 ,
compared with the figure of nearly 3000 MJ m - 2 in the earlier estimation of TPE.
The ratio of reservoir content to rate of throughput defines a residence time of just
over two weeks for energy in the form of zonal APE, with a further five days in the
form of eddy APE. Most but not all of this APE is in turn converted into eddy KE
in large-scale weather systems. The importance of middle latitudes can be gauged
from energy budgets for individual vigorous depressions, which show that
conversion rates approach 20 W m - 2 over the affected areas. It seems that four or
five of these systems in action around a hemisphere at any one time can account for
a substantial proportion of the total conversion. Notice that the reservoir of
atmospheric KE is almost equally divided between the eddy and zonal components,
although in terms of rates of conversion, only one fifth of the eddy KE is passed on
to zonal KE. The implied residence times for energy are nearly four days in the
form of eddy KE, which is consistent with the significant role assigned to synoptic-
scale weather systems. The much longer residence time of over 20 days in the form
of zonal KE is consistent with the view that the reservoir of zonal KE acts as a big
zonal flywheel kept going by the rather small flux of energy reaching the end of the
sequence of transformations.
Energy is lost from three corners of the array in Fig. 13.5. The first of these in
order of energy flow corresponds to radiative wastage of eddy APE, in which warm
fingers of air lose some of their warmth and height by net radiative cooling before
their contribution to the total APE can be converted into KE. Some of the original
input has therefore been wasted, from the point of view of production of
atmospheric KE, by what is termed radiative dissipation. No doubt a more detailed
picture of the earliest stages in the energy cycle would reveal temporary production
of even more zonal APE than is depicted in Fig. 13.5, followed quickly by wastage
through radiative dissipation. It therefore appears that not only is most of the
450 atmospheric TPE unavailable for conversion into kinetic energy, but in addition a
 13.3 Energy paths
substantial fraction of what little is available is not in fact converted. The
atmospheric engine is in this sense rather inefficient- a fact which is confirmed by
a more fundamental criterion in the next section.
Warm anticyclones such as those in the subtropics provide clear examples of
radiative dissipation. Their domes of warm air represent sizeable contributions to
both zonal and eddy APE. However, these largely static warm cores are
particularly prone to radiative cooling because of the near absence of cloud above
the subsidence inversion (section 12.1), so that they are continually collapsing,
wasting APE which could otherwise have been converted into KE, and maintaining
their cloudless cores by the associated subsidence.
The frictional dissipation of KE appears on the right-hand side of the array in
Fig. 13.5. The relatively large value of the destruction of eddy KE as compared
with zonal KE arises because the highest wind speeds in both the low and high
troposphere appear in the eddy category almost by definition. In particular, the
strong wind shears in the boundary layers of vigorous depressions act as powerful
concentrations of frictional loss. On a simple view, the rate of destruction of KEin
the boundary layer is proportional to the cube of the geostrophic wind speed
(appendix 13.4), biasing the loss heavily toward regions of strong winds. The
presence of large-scale horizontal convergence in the lower parts of cyclonic
disturbances maintains strong winds and associated frictional dissipation by
conservation of angular momentum (section 7.14). There is some significant
frictional dissipation at higher altitudes, particularly in the regions of strong
vertical shear below and above the cores of jet streams, where clear-air turbulence
(section 10.8) is evidence of turbulent dissipation of larger-scale flow and
associated KE.
By contrast, the gentler shears associated with large-scale zonal motion, both in
and above the boundary layer, maintain a relatively small rate of frictional
dissipation per unit area and allow the reservoir of zonal KEto remain comparable
with the eddy KE despite the much smaller throughput of kinetic energy. The zonal
'flywheel' therefore runs relatively freely on a small conversion from eddy KE and
a very small leakage from zonal APE.
The values of sequences of Fig. 13.5 correspond to the annual average picture of
the general circulation of the hemisphere. If the seasonal extremes are examined
separately (and this complicates the analysis considerably) the direct conversion
from zonal APE to zonal KE, expected at the beginning of this section, reappears,
especially in winter. Examination shows that this arises mainly from the action of
the Hadley circulation of low latitudes (section 8.13) -the huge thermally direct
circulation powered by the liberation of latent heat in the intertropical convergence
zone. The conversion rate {Az.Kz} is affected by the conservation of angular
momentum as air in the upper troposphere diverges toward the subtropics (section
4. 7). Some of the energy flow therefore tends to cut across the top of the equivalent
of Fig. 13.5, partly short-circuiting the eddy forms of energy. Analysis by latitude
zones confirms this interpretation, and shows that in middle latitudes there is
conversion in the opposite sense (from zonal KEto zonal APE) in the thermally
indirect Ferrel cell (section 4. 7) which in simpler data such as average winds is all
but swamped by the unsteady eruption of planetary and cyclone waves. In the
northern-hemisphere summer the Asiatic monsoon blurs the effect of the Hadley
circulation on the energy sequence by introducing eddy components at lower
latitudes. The smallness of {Az.Kz} in the hemispheric annual picture (Fig. 13.5)
therefore arises because of opposing conversions in different latitude zones, as well
as the partial disruption of the Hadley circulation by the summer monsoons.
As mentioned at the beginning of this section, we have deliberately concentrated
on the energy pathways in large-scale weather systems rather than the pathways in 451
13 The atmospheric engine convective (cumulus-scale) systems, because the latter are so poorly understood. In
fact the energy fluxes through the two types of system are believed to be
comparable on a global scale, as suggested by the comparability of their associated
heat fluxes (Chapter 8). The two types are similar in that they represent the
temporary conversion of APE into KE, though the APE for convection arises from
the convectively unstable vertical stratification of density, whereas the APE for
large-scale weather systems arises from the baroclinically unstable horizontal
zoning of density. However, unlike the case in large-scale systems, the APE in
convection is converted into KE very quickly and locally, and is similarly reduced
to heat by the overwhelming combination of turbulent form drag and entrainment
drag (section 7 .12), with the result that motion is transient and quickly stifled in all
but the most vigorous storms. The effect of all this friction is to keep the reservoir
of associated kinetic energy very small indeed although the energy throughput is
comparable with that of large-scale motions. This is of course consistent with
observation -vertical wind speeds are an order of magnitude down on horizontal
wind speeds even in the active cores of cumulus, and drop by another order on
averaging through large volumes of the atmosphere. (In principle it is not correct to
assume that all KE associated with cumulus is associated with vertical motion, but
in terms of contribution to global budget, gales and tornadoes in cumulonimbus
are exceptions which prove the general rule.) The smallness of the ratio of reservoir
to throughput is simply a measure of the shortness of the residence time of energy
in the form of KE, which is consistent with the observed brevity of cumulus
lifetimes. The weakness, transience and turbulent complexity of cumulus systems
severely hinder our observation and understanding of them and their energy path-
ways, so that nothing comparable in measured detail with the scheme for large-
scale motion has been deduced so far. Some of the most important energetic
differences between large and convective scale types of motion are summarized in
the following graphic analogy drawn by the writers of a recent textbook [8]: large-
scale motion systems are like a person coasting down a certain hill on a racing
bicycle, whereas convective systems are like a person coming down the same hill on
rusty roller skates. Obviously the cyclist reaches much higher speeds than the
skater, although the same initial potential energy is available to each.
Consider finally a descriptive outline of the energy pathways and trans-
formations in the stratosphere (Fig. 13.6). The temperature profile is so stable con-
vectively that the stratosphere cannot support any self-driven activity - even
baroclinic instability is suppressed. In fact it seems likely from detailed studies that
such muted activity as it shows is driven by eddy kinetic energy propagating up
from the great planetary waves of the troposphere (bottom right in Fig. 13.6).
Strictly speaking this leakage from the troposphere should appear as a component
in the tropospheric cycle in Fig. 13.5, but is omitted because it is smaller than the
uncertainties in the numerical values there. Because there is so little friction, this
input to the stratosphere maintains a very considerable zonal circulation and eddy
and zonal APE by a thermally indirect circulation in which warmer air sinks and

Fig. 13.6 Energy reservoirs

and flows in the stratosphere.
(After (44])

From
troposphere
452
cooler air rises. This sinuous zonation of APE then leaks heat to space by weak net 13.4 Efficiency of the
atmospheric engine
radiative cooling which is concentrated in the relatively warm, low stratosphere of
higher latitudes. The reversed flow of energy in Fig. 13.6 as compared with
Fig. 13.5 in fact corresponds to the action of a heat pump rather than a heat engine,
the heat being taken from weak net radiative warming in the cold, high base of the
equatorial stratosphere, and pumped to high latitudes by the input of mechanical
energy from beneath. It is a kind of gentle refrigerator, keeping the equatorial
stratosphere cool by pumping the small radiative excess of heat to higher latitudes.
The life cycle of atmospheric mechanical energy is now complete in outline:
produced largely in the low troposphere from a portion of the heat input to the sun-
warmed surface, most is degraded to heat again in the troposphere by friction of
various types, while a very small part is used to drive the stratospheric heat pump.
The prominent role of friction in the tropospheric engine apparently differs
from that in man-made engines, which are of course designed to have minimum
friction. But the proper comparison is with man-made engines including the
machines that they drive. When this comparison is made, then in man-made
engines too, all mechanical energy is eventually degraded to heat, mostly by the
action of friction, though this may happen hundreds of kilometres from the engine
proper if the power has been transmitted electrically to the point of use. Together
with the large throughput of heat enforced by their necessarily limited mechanical
efficiency (see next section), this inevitable degradation of mechanical energy to
heat means that all energy passing through man-made engines is converted to heat
sooner or later, either at source or elsewhere. Man's engines are now so numerous
and powerful that unintentional heating is having significant effects in localized
areas such as cities, rivers and estuaries. The effects may soon become globally
significant too: if our economic, social and industrial systems permit a modestly
increased population of five thousand million to produce energy at the per capita
rate of citizens of the United States in 1950 (Appendix 13.5), then the total rate of
artificial heat production will exceed 1OJo of the flow of mechanical energy through
the atmosphere. If the necessary power were to be drawn from sunshine, wind,
waves or tides, then it would be little more than a temporary redirection of part of
the natural energy flow, but of course at the moment the overwhelming proportion
of the power comes from coal, oil, nuclear fission and the irreversible destruction
of trees, and is therefore in addition to the natural flow.

13.4 Efficiency of the atmospheric engine

The first law of thermodynamics is a budgeting law, requiring that energy accounts
always balance: the second law of thermodynamics shows that not all balancing
budgets are permissible. For example, in terms of Fig. 13.1, the first law requires
of the tropospheric engine that
(13.9)
The second law states that heat must flow from a warm source to a cool sink if a
positive quantity of mechanical energy M is to be produced by conversion, however
temporarily. In terms of Fig. 13.1, the second law requires that T 1 be higher than
T2 • Applied to the stratospheric heat pump, ·the second law requires that
mechanical energy must be supplied if heat is to be taken from a cooler source and 453
13 The atmospheric engine delivered to a warmer sink. The stratosphere could not for example pump heat
from T3 to a higher T4 and produce mechanical energy. More specifically, the first
and second laws together set an upper limit to the fraction of the heat input to an
engine which can be converted into mechanical energy, i.e. to the mechanical
efficiency of the engine. This limit is reached in an ideally efficient engine in which
there is no friction and in which processes are so smoothly ordered as to be
reversible by a minute alteration of conditions. In the terminology of Fig. 13.1,
this reversible limit is given by

(13.10)

as shown in appendix 13.6. Actual engines must be less efficient than this to the
extent that they include friction, which of course they all do, and involve processes
which are not in equilibrium, which is also always the case, but there are no simple
criteria for estimating this shortfall.
From eqn (13.10) it is clear that we can estimate the optimum mechanical
efficiency of the tropospheric engine provided that we can define the source and
sink temperatures T 1 and T2 • If the value of the heat input Q1 is known in addition,
then the optimum production of mechanical energy can be found at once, and this
places an inescapable upper limit on the mechanical activity of the atmosphere.
Note that the terminology of Fig. 13.1 does not specify whether the Qs and Mare
quantities of energy or rates of energy (powers), since the first and second laws of
thermodynamics apply in either case. For comparison with the analysis of the last
section it is natural to consider powers.
The simplifications of section 13.1 will be accepted to identify reasonable single
values for T 1 and T2 • The global annual average temperature of the Earth's surface
is 288 K, and this will be assumed to be the value for T 1 • The effective radiative
temperature of the planet Earth is close to 255 K (section 8.2), and this will be
accepted as the value for T 2 , though it must in principle be a little above the
temperature of the highest opaque layer (the most obvious temperature to choose
for the heat sink) on account of the small output coming through the atmospheric
window from the relatively warm surface. It follows that to the extent that these
values are realistic, the mechanical efficiency of the tropospheric engine cannot
possibly exceed 33/288, or approximately 11 OJo. Because of the relatively small
temperature difference between source and sink, therefore, at least 890Jo of the heat
taken into the tropospheric heat engine must pass through without conversion into
mechanical energy, and be wasted to the heat sink. (That is wasted from the point
of view of the mechanical energy of the engine- not of course from the viewpoint
of the comfort of the inhabitants of middle latitudes who bask in the warm
draught.) This is apparent in the convective and advective currents of warm and
cold air and water described in Chapter 8 and elsewhere and which are associated
with the whole range of weather systems large and small. It is likely that a fuller
allowance for the complex multiplicity of realistic values for T 1 and T2 would result
in an even smaller value for the maximum mechanical efficiency and an even
greater heat transfer without mechanical effect. And of course the irreversible
aspects of actual atmospheric behaviour will depress the mechanical efficiency still
further.
It is difficult to identify precisely the value of the heat input Q 1 to the tropo-
spheric engine. Considering the scheme of Fig. 8 .6, it seems reasonable to focus on
the convective heat flux from the underlying surface, since many of the radiative
fluxes are not necessarily associated with the existence of a heat engine at all. For
example, the sunlight scattered out to space plays no role, and the net flux of
454 terrestrial radiation between the surface and the base of the troposphere could
continue as well in a completely static atmosphere. (You cannot be too dogmatic 13.5 Epilogue
about this though, since the distributions of cloud, water vapour, carbon dioxide
and temperature which determine this flux are all themselves maintained by
atmospheric motion.) The vertical convective flux which balances the radiative
budget corresponds to a global annual average of 30 units or just over IOO W m - 2 •
If we choose this for Q 1 we may be neglecting advection between latitude zones
which is implicit in the solar input of I9 units (about 65 W m - 2) to the troposphere,
since some of this is likely to be offset by net long-wave cooling at higher latitudes
rather than locally. However, the nearly parallel gap between the meridional
profiles of radiative input and output in Fig. 8. 7(b) is very close to IOO W m - 2 at
all latitudes, suggesting that this is a value which fairly represents the combination
of convective and advective heat fluxes. So we will identify this value with Q 1 for
the purposes of estimating the maximum possible mechanical efficiency for the
atmospheric engine.
Actually a more detailed examination would almost certainly reduce the value
for Q1, since a sizeable fraction of it must correspond to heat fluxes passing across
temperature differences considerably less than the 33 oc assumed above. You can
practically see this effect in the vertical distribution of cloud tops in a field of
cumulus, tall clouds being so much rarer than shallow clouds. And it is confirmed
by the rapid decrease of convective heat flux with increasing height apparent in
Fig. 8.II. But since we have no good way of judging the size ofthe overestimation,
let us accept the IOO W m- 2 as a round figure, knowing that it is likely to be a
substantial overestimate.
We have now established that the maximum mechanical efficiency of the
atmospheric engine is II OJo and that the input of heat from the heat source is
IOO W m- 2 at most. It follows from eqn I3.10 that the maximum rate of
production of mechanical energy by the engine is 11 W m- 2 , which is quite
reasonably consistent with the value of 3 W m - 2 indicated by the more detailed
studies quoted in the previous section, bearing in mind the simplicity of the
argument, and the fact that it establishes an upper limit rather than a likely value.
Some of the difference between the two values no doubt arises from the difficulty
of determining T~o T2 and Q1, especially the last, but some must also arise because
the atmospheric engine runs substantially below the thermodynamic optimum
because of irreversible processes. The value of this simple analysis is that it shows
that, as well as any incidental inefficiencies arising from particular details of atmo-
spheric behaviour, there is an inescapable limitation to its efficiency overall, which
is actually quite severe.

13.5 Epilogue

So in the end we find that the incessant, often impressive, and occasionally very
destructive activity of the atmosphere is maintained by a conversion to kinetic
energy of only a small fraction of the total flux of energy driven through the atmo-
sphere by the Sun, which in turn is only an infinitesimal part of the Sun's prodigal
output. The enormous power of the hurricane, the trade winds which dominate
over one third of the Earth's surface, the gales which so frequently trouble the seas
and western continental margins in middle latitudes - all are manifestations of an
average rate of production of kinetic energy in an atmospheric column resting on 455
13 The atmospheric engine an area the size of a domestic table which could barely light a small light bulb if it
were entirely converted to electrical power. However, this is not as unlikely as it
might seem at first glance. As mentioned in section 13.3, the atmosphere is in some
respects like a large flywheel which is maintained in nearly steady rotation by the
balance between power input from a weak engine and power loss by friction. The
balance achieved is such that there is a substantial reservoir of kinetic energy in the
flywheel at any time - a quantity representing about one week's production of
kinetic energy according to Fig. 13.5. Moreover all the more spectacular motion
systems of the atmosphere represent considerable concentrations of kinetic energy:
depressions, jet streams, cumulonimbus, tornadoes and so on, all occupy only a
very small fraction of the available atmospheric volume, and the most intense
(tornadoes on this list) generally occupy the smallest fraction. In these small
volumes kinetic energies rise far above the global average. Note that for the most
part these motion systems represent concentrations of production, rather than
concentrations of pre-existing kinetic energy (though the tornado may be an
exception here). According to Fig. 13.5 the flywheel is kept spinning mainly by
continual nudging from little, transient, energetic wheels which represent
individual weather systems. Even the Hadley circulation is kept turning by the
continual eruption of hundreds of cumulonimbus in the intertropical convergence
zone.
Because the atmosphere is in a state of nearly steady activity, the rate of destruc-
tion of atmospheric kinetic energy by friction must equal its rate of production by
solar heating. Therefore to say that the reservoir of kinetic energy corresponds to
about a week's production is to say equally that it corresponds to about a week's
frictional destruction. If that production were suddenly to cease by some
unimaginable solar catastrophe, the flywheel which has been spinning for aeons
would grind to a halt in a period of a few days or weeks. The precise length of the
period depends on how we imagine the largely surface-based friction to reach to the
middle and upper troposphere, and on the effects of the large amounts of heat
stored in the surface layers of the ocean, but whatever the details it is bound to be
almost infinitely short compared with the timescale on which the Sun has been
keeping our atmosphere chronically active. The builders of pre-history who sited
their megaliths to mark the annual cycle of the Sun, and the religious devotees,
poets and artists of those and subsequent years who have striven to do it homage in
many and varied ways, can hardly in their wildest dreams have conceived how
absolutely and continually crucial is that 'sovereign eye' to the maintenance of the
Earth's surface and atmosphere in their familiar condition.

Appendix 13 .1 The energy equation

Multiply the vertical component of the equation of motion (7 .12(c)) by the vertical
wind speed w to obtain
dw w ap
w - = - - - · - g w - wFz (13.ll)
dt p az
Since w= dz!dt and g is effectively constant we can rewrite g was d(gz)/dt, and by
elementary differentiation w dwldt can be rewritten as d(Yz w 2 )/dt. Equation
456 (13 .11) can therefore be rewritten in the form
 d w ap Appendix 13.2
- ( Yz w2+ g z) = - - - - (13.12)
dt p az
Applying the same procedure in turn to the x andy components of the equation
of motion (eqn 7.12(a and (b)), we find
d
- ( Y2 u 2) =f
u
uv - -
ap
- - uF
df p ax X

and
d v ap
- ( Y2 v 2) = - f v u - - - - vF
dt P ay y

When these two equations are added the Coriolis terms cancel, showing that the
Coriolis effect leaves kinetic energy unchanged. This can be shown to apply very
generally to the full array of Coriolis terms mentioned in section 7. 7, for the
basically simple reason that Coriolis forces, being perpendicular to the relative
velocities from which they arise, can perform no work on an air parcel as it moves,
and cannot therefore change its kinetic energy. If we add the sum of these two
equations to eqn (13.12) we obtain the full expression for the rate of increase of
kinetic and potential energy
d
-(Yl VZ+gz) = PGW- FW (13.13)
dt
where V 2 = u 2 + v 2 + w 2 , and PGWandFWrepresent the total rates of working
by the pressure gradients and friction respectively, whose components are readily
identified in the above equations.
The rate of working represented by PG W arises from the work done on a parcel
as it moves down a pressure gradient. Another kind of working PC occurs when
ambient pressure squeezes a parcel and increases its internal energy, producing a
rise in temperature. The effect is described quantitatively by a rearrangement of
eqn (5.1):
PC= _ p dVol =C dT _ dQ ( 13 . 14)
dt v dt dt
where small changes have been converted into rates of change for compatibility
with eqn ( 13 .13). The sum of PG Wand PC represents the total rate of war king P W
by pressure on the air parcel, and eqns (13.13) and (13.14) can now be combined to
form an expression for the rate of increase of the total energy (kinetic, potential
and internal) of the air parcel
d dQ
-(YlVZ + gz + CvT) =-d + PW- FW (13.15)
dt t

Appendix 13.2 Potential and internal energies of

an air column

The total gravitational potential energy of a column of air resting on unit

horizontal area of surface is given by 457
13 The atmospheric engine
PE =g [ zdm (13.16)

where m is the total mass of the column. According to the hydrostatic

approximation, the mass dm of a thin horizontal slice of the column is related to
the pressure difference dp between its top and bottom by dm = - dp!g, where the
minus sign shows that we are arbitrarily assuming that dm is positive upwards (i.e.
that the total mass of the atmosphere is accounted from bottom to top), unlike
pressure increments dp. Hence

PE = -
o
J z dp =
ip, z dp
p, 0

where p, is the pressure at the base of the atmospheric column (zero height).
Integration by parts gives

PE = [p z]g· - [ p dz

The first term on the right-hand side is zero because the product pz is zero both at
the top and bottom of the atmosphere: at the top because pressure is zero at finite
height, and at the bottom because height is zero at finite pressure. Hence

PE = [pdz

We can use the equation of state to substitute for p. The resulting specific gas
constant can be moved outside the integral because it is a constant, leaving

PE = R [ Tpdz

Since the product p dz is simply dm we have

PE = R [Tdm (13.17)

showing that the gravitational potential energy of the column Is directly

proportional to its internal energy

IE= Cv [ Tdm

A slight extension of this analysis confirms the special usefulness of the specific
heat capacity at constant pressure CP. If we consider the potential energy of an air
column extending from the surface to a height z at which the pressure has non-zero
value p, it follows from the above that

PE = R [ T dm - pz

i:
and that the total potential energy TPE (i.e. PE +IE) is given by

TPE = CP T dm - pz (13.18)

If this layer is now warmed by the input of a quantity of heat !:::. Q while the pressure
458 pat the column top remains constant, the heat energy must be used both to increase
the total potential energy of the layer and to do work by expanding vertically and Appendix 13.3
raising the overlying air:
!::..Q = !::..TPE + p !::..z
However,

!::..TPE = CP J: !::..Tdm- p !::..z

c
and it follows that

!::..Q = CP !::..Tdm

showing that the apparent complications arising from the finite depth of the layer
and the presence of the overlying air have zero net effect, and that the whole
specific energy capacity of the layer is fully described by CP.

Appendix 13.3 Available potential energy

Figure 13.3(a) represents two adjacent vertical columns of unit horizontal cross-
section, one of which is slightly taller and less dense than the other. Assuming the
material of column 1 to have uniform density, its centre of gravity is at height z/2
above its base. Multiplying by the column weight gp 1Z, we find its gravitational
potential energy to be gp 1 z 212. In the same way the potential energy of column 2 is
found to begp 2 (z + !::..z} 212. The sum of these expressions is the aggregate potential
energy of the columns in their initial configuration.
Now suppose that the columns are rearranged so as to minimize their aggregate
potential energy. If we assume their material to be incompressible and require that
their horizontal area (two units in total) is unchanged, then the minimum is
achieved as in Fig. 13.3(b}, with the denser material underlying the less dense. The
aggregate potential energy in this final configuration is found by the above
procedure to be
gp 1z 214 + gp 23z 214 + gp 2 !::..z (z+ !::..z/4)
Subtraction from the initial aggregate potential energy and simplification shows
that the reduction !::..PE in potential energy effected by the rearrangement of the
columns is given by

(13.19)

Since the fractional differences in density and initial column height are assumed to
be small, the initial aggregate potential energy PE is closely approximated by gpz 2 ,

r]
where pis the average column density. Writing p1 - p2 as !::..p, it follows that

%E :: Y4 [~P + ( ~z
If we consider the atmospherically realistic case that the fractional differences in
density and initial column height are proportional and very small, the second term 459
13 The atmospheric engine in the brackets must be much smaller than the first. The fraction of the initial
potential energy made available by feasible rearrangement of the columns is
therefore given approximately by !:::.p/4p.
Note that any calculation of available potential energy requires precise
prescription of the re-arrangement process, since the fraction available is otherwise
undefined. In this simple case density and horizontal area are conserved. In more
realistic compressible models, the rearrangements are usually assumed to be
adiabatic, but no single model applies to all atmospheric situations.

Appendix 13.4 Frictional dissipation

Figure 13.7 represents a slab of air with unit horizontal area being dragged forward
(i.e. in the direction of the ambient wind) by the turbulent frictional stress exerted
by the faster-moving air above, and being dragged backward by friction with the
slower-moving air beneath. According to Fig. 9.21 and the discussions of sections
T• dr 9.9 and 9.10, the stress at the lower level is slightly larger throughout the depth of
Fig. 13.7 Frictional stresses
frictional influence, increasing from nearly zero at the gradient level to a maximum
on the top and bottom faces of r, in the surface boundary layer (SBL), in which r, is effectively constant. The net
of a slab of air in the frictional force on the slab in Fig. 13.7 therefore opposes the air motion and is
atmospheric boundary layer. given by the difference dr. If the speed of the slab is V, the rate of working by
friction against the wind is V dr, and the aggregated rate of dissipation of kinetic
energy in a column extending through the planetary boundary layer is given by

L Vdr (13.20)

This must account for nearly all of the FW term in eqn ( 13. 7) since most of the
frictional dissipation is concentrated within the planetary boundary layer.
Figure 13.8 contains a plot of V against r which is consistent with the above
features. The geostrophic wind speed V8 is reached at the gradient level, where r is
zero; wind speeds fall slowly through the Ekman layer (EL) as r builds up towards
the maximum value r, which Qccurs in the SBL, where V falls almost vertically to
zero. The aggregated rate of dissipation is the area under the curve, which can be
seen to be equal to V8 r, nearly enough to be used in an order-of-magnitude
estimation.

G
Fig. 13.8 Graph of Vg
horizontal wind speed V
;--1 Area v9 r 5
L.-~
against horizontal frictional
stress from the surface S v
j
r.~dT
TS

through the atmospheric 0

boundary layer to the
gradient level G.

We can use eqn (9.24) to express the frictional stress in the SBL in terms of the
square of the 10 m wind speed V 10 and the appropriate drag coefficient C 10 :
460 FW::: p C 10 Vfo Vs (13.21)
Substituting a reasonable value of 1 kg m- 3 for the air density p, and 3 x I0- 3 for Appendix 13.6
C 10 (which corresponds to a roughness length of 1 em and a 10m observation
height), and using the reasonable rule of thumb that V10/ Vg = 0.4, we have finally
FW ==
- sx I0- 4 Vg3
Although only very approximate, this relation emphasizes the great sensitivity of
frictional dissipation rate to wind speed. For example the rate implied is
0.06 W m- 2 for a geostrophic wind speed of 5 m s- 1 , but is 4 W m- 2 for a
geostrophic wind speed of 20 m s- 1 (equivalent to about 8 m s- 1 at 10m).

Appendix 13.5 Human energy production

According to detailed studies of fuel total industrial and domestic fuel

consumption in the United States [93], the per capita energy production in 1952
was 62.1 MW h, which is the energy corresponding to a power of 62.1 MW
maintained for one hour. Dividing by the number of hours in a year, we find that
this corresponds to a power of 7.1 kW continued throughout the year. The total
power consumed by 5000 million people at this per capita rate would be 3.6 x
1013 W. Assuming realistically that almost all of this is degraded to heat after use
(in principle a little must be stored in intermolecular binding energies of long-lived
manufactured materials), the implied rate of heat production must almost equal
this value. Dividing by the surface area of the Earth, we find the corresponding
heat flux in the same units (W m - 2 ) as are used in section 13.3. The value is 7 x
I0- 2 W m - 2 , which is a little over 20Jo of the value quoted for heat production by
frictional degradation at the end of the atmospheric mechanical energy cycle. Since
the uncertainty in the latter value is about 50%, it is safest to quote the percentage
as of order 1.

Appendix 13.6 Optimum mechanical efficiency

of a heat engine

In standard thermodynamics textbooks it is shown that there are several equivalent

statements of the second law of thermodynamics, one of which (known as
Clausius' inequality) can be expressed in terms of the tropospheric engine of
Fig. 13.1 as follows:
Q1 ~ Qz (13.22)
T1 ""' T2
The temperatures T1 and T2 must be expressed in the absolute scale. The equality
describes the ideal reversible limit, while the inequality describes the behaviour of
all realistic engines, natural as well as man-made. Equation (13.22) can be simply
rearranged: 461
13 The atmospheric engine Tz ~~
Tt"" Qt
From the first law of thermodynamics we can rewrite the expression for the
mechanical efficiency of the engine as follows:
M = Qt - Qz = 1- ~
Ql Ql Ql
Combining this with Clausius' inequality we find
M~T1 -T2
Qt"" Tt
where the equality again describes the relationship at the reversible limit, which is
now seen to be associated with maximum efficiency. Since T2 is always greater than
absolute zero (very much greater in the case of the atmospheric engine), it follows
that the optimum efficiency of even the most perfect engine is less than unity. In
addition, all real engines are irreversible to some extent and therefore fall short of
the optimum for their particular source and sink temperatures T1 and T2 •
Although irreversibility appears as a kind of imperfection in classical thermo-
dynamics (which concentrates on the reversible ideal), recent advances in the study
of irreversible systems of many types [94] suggest that many properties of real
systems are associated with their intrinsic irreversibility and non-linearity. The
irreversible nature of weather systems remains largely unexplored, but is being
opened up for scrutiny by chaos theory (section 1.7).

Problems

LEVEL 1

1. In a heat engine, what happens to the heat which is not converted into
mechanical energy?
2. In the style of Fig. 1.6 sketch (a) a heat engine and (b) a heat pump operating
between a temperature T1 and a lower temperature T2 , showing the direction of
the flows of heat and mechanical energy in each case.
3. Write down the first law of thermodynamics for case (b) in problem 2, where
heat Q 2 is absorbed from the cool heat source, Q 1 is delivered to the warm heat
sink, and mechanical energy Misused.
4. By considering the case of a domestic refrigerator running in a steady thermal
state, show that the resultant effect must be to warm the room in which it
operates. Where has the heat come from?
5. Present a non-technical argument to explain why so little of the atmosphere's
considerable total potential energy is available for conversion to kinetic
energy.
6. Consider the likely dynamic behaviour of an atmosphere which is being heated
uniformly throughout its bulk by net radiation.
7. Make a verbal summary of the sequence of inputs, transformations and
outputs summarized by Fig. 13.5.
462 8. Why is the tropospheric engine so inefficient?
LEVEL 2 Problems

9. Find the ratio of available potential energy to potential energy for the case of a
two-layer fluid (Fig. 13 .4) in which the upper layer is 1OJo denser than the
lower. Assume the relationship mentioned in section 13.2.
10. Find the maximum mechanical efficiency of the intrinsic oceanic heat engine
(i.e. ignoring the considerable activity driven by the atmosphere) given that its
heat source is in tropical surface waters (temperature about 25 oq and its heat
sink is in polar surface waters (about 0 °C}.
11. By assuming conservation of the sum of potential and kinetic energies (i.e.
ignoring the right-hand side of eqn (13.1)), find the speed of a rain drop after
falling 1 km from rest. Since actual fall speeds are - 5 m s - 1, estimate the
amount of energy per unit mass of raindrop which seems to have disappeared
at this point, and discuss where it has gone.
12. If 10 MJ m -z of heat energy is pumped into a vertical column of air extending
through the full height of the atmosphere, find the resulting rise in air
temperature, and find also the increases in internal and potential energies.
(Assume that the column maintains its horizontal area throughout.)
13. Using the approximate relationships in appendix 13.4, find the frictional rates
of dissipation of kinetic energy in the boundary layer corresponding to
Beaufort wind forces 2, 4, 8 and 12.
14. Find the column total of KE in the active part of a mid-latitude depression,
assuming wind speeds of 20 m s - 1 from the 1000 to the 500 mbar levels,
60 m s- 1 from 500 to 200 mbar, and 20m s- 1 in the rest of the atmosphere.
(Hint: convert the pressure slabs into equivalent masses per unit area assuming
hydrostatic balance.)

LEVEL 3

15. Sketch and discuss qualitative modifications of Fig. 13.5 needed to describe
(a) convective activity in the troposphere, and (b) large-scale activity in the
Hadley circulation.
16. Using the method of appendix 13.3, show that the ratio of available PE to
actual PE (relative to a datum level at the base of the lower layer) for the
overturning depicted in Fig. 13.4 is given approximately by b.p/2p.
17. Consider eqn (13.1) applied to a slightly buoyant air parcel, and discuss the
relative sizes of the terms in a discussion of the parcel's behaviour, firstly in the
absence of and then in the presence of drag.
18. Show from the identity of eqn (13.5) and its immediate predecessor that the
altitude of the effective centre of gravity of an isothermal atmosphere is one
exponential scale height above its base. (Note that this altitude Z is defined to
be such that mgZ = PE, where m is the atmospheric mass.)
19. Outline the sequence of energy types and conversions for large-scale
atmospheric activity in the northern hemisphere on the timescale of a year,
qualitatively defining all terms used.
20. Speculate on the sequence of events which would occur in the Earth's surface
and atmosphere over a period of a year if the unthinkable happened and the
Sun's output suddenly and completely ceased.

463
 Bibliography, references
and other sources

Textbooks

The following is a selection of books which I have found to be useful in ways mentioned. The
brief opinions are, of course, personal and each of the following deserves close scrutiny by
anyone considering their use. The order is alphabetical.
Atkinson, B. W. (Ed.) (1981) Dynamical meteorology: an introductory selection. Methuen,
London. Although the wide range of authors is responsible for a variety of levels of treat-
ment, the overall coverage is excellent, with good discussions and just enough formalism
to show the way to more thorough treatments. Graphics are adequate.
Hess, S.L. (1959) Introduction to Theoretical Meteorology. Holt, Rinehart and Winston,
New York. An old text, famous for its uncluttered clarity of treatment of
thermodynamics, radiation and dynamics (without vectors!). Minimal graphics and little
explicit comparison with observation detract but not critically.
Holton, J .R. (1972) An Introduction to Dynamic Meteorology. Academic Press, New York.
Coverage as implied by the title but pitched at people with considerable experience in
applied mathematics. Though the treatment is quite formal throughout, the associated
discussion always encourages physical understanding. The treatment follows the con-
ventional bias toward synoptic-scale dynamics.
Lamb, H.H. (1982) Climate, History and the Modern World. Methuen, London. A brief
outline of the ideas and methods of climatology followed by a fascinating survey of
climate, mainly since the last glaciation, with special emphasis on social effects.
Lockwood, J.G. (1979) Causes of Climate. Edward Arnold, London. A compact, semi-
quantitative survey of climate and climatic change mostly over the last 100 000
years.
Ludlam, F.H. (1980) Clouds and Storms: the behaviour and effect of water in the
atmosphere. Pennsylvania State University Press, University Park and London. A very
large and thorough treatment of many aspects of precipitating and non-precipitating
clouds (but not thunderstorm electricity). Unusual for combining cloud physics and
small, meso and synoptic-scale meteorology, this book is posthumous testimony to the
physical insight and elegant treatment typical of this remarkable scientist, whom I was
privileged to have as Ph.D. supervisor. Though not a teaching text, it touches usefully on
a wide range of observable meteorological phenomena.
Mcintosh, D.H. and Thorn, A.S. (1969) Essentials ofMeteorology. Wykeham Publications
Ltd (Taylor & Francis Ltd), London. A small, concise book covering a wide range of
meteorological essentials in surprising depth. The treatment of the tephigram is parti-
464 cularly useful. Dynamics are treated using vector notation. Minimal graphics.
Mason, B.J. (1962) Clouds, Rain and Rainmaking. Cambridge University Press. A classic Journals
introduction to the physics of clouds, precipitation and thunderstorm electricity.
Neiburger, M., Edinger, J.G. and Bonner, W.D. (1982) Understanding our Atmospheric
Environment. W.H. Freeman and Company, San Francisco. A substantial introductory
text, mostly about meteorology but with just enough about pollution and other human
interaction to justify the title. The treatment is mostly qualitative but is nevertheless
meteorological (rather than physical geographical) in style. Good graphics.
Oke, T.R. (1978) Boundary Layer Climates. Methuen, London. A good introduction to
boundary-layer meteorology and climatology, with a substantial section on anthro-
pogenic effects. The treatment is mainly semi-quantitative but there are many useful
tables of values.
Palmen, E. and Newton, C. W. (1969) Atmospheric Circulation Systems. Academic Press,
New York. A classic survey of understanding of synoptic- and larger-scale weather
systems, with some treatment of organized vigorous small-scale convection. Although the
treatment is not at the introductory level, the discussion is full and rewards patient study.
Riehl, H. (1978) Introduction to the Atmosphere. McGraw-Hill Kogakusha Ltd, Tokyo.
Probably the best purely qualititative introduction to meteorology, with good photo-
graphs in the later editions.
Rogers, R.R. (1976) A Short Course in Cloud Physics. Pergamon Press, Oxford. A
thorough treatment of the whole range of meteorological cloud physics with the curious
exception of thunderstorm electricity.
Scorer, R.S. (1978) Environmental Aerodynamics. Ellis Horwood Ltd (associated with
John Wiley & Sons), Chichester. An unusual and stimulating book, beginning with a
formal vector treatment of fluid dynamics which is well beyond introductory level, but
then proceeding to a discussion of many smaller-scale aspects of atmospheric behaviour
with useful emphasis on physical reality as well as theoretical nicety. It includes related
topics such as chimney-plume dispersion and bird soaring.
Simpson, R. H. and Riehl, H. (1981) The Hurricane and its Impact. Basil Blackwell, Oxford.
A largely qualitative description of these fearsome storms with considerable emphasis on
their destructive power.
Wallace, J.M. and Hobbs, P.V. (1977) Atmospheric Science: an introductory survey.
Academic Press, New York. A modern text on fairly traditional lines, with very solid
thermodynamics and radiation, but a much lighter treatment of dynamics. The coverage
of cloud, precipitation and electrical processes is particularly detailed. Diagrams and
photographs abound.
Wayne, R.P. (1985) Chemistry of Atmospheres. Clarendon Press, Oxford. A very full
account of the chemistry of the terrestrial and other planetary atmospheres.
Wickham, P. G. (1970) The Practice of Weather Forecasting. HMSO, London. An excellent
summary of synoptic-scale weather analysis and manual forecasting with lots of maps and
soundings from actual cases.

Journals

Most meteorological journals are pitched well beyond the introductory level, though they
can be skimmed with profit by looking at diagrams and conclusions before trying to under-
stand the physical meaning of the more formal sections. The popular journal of the Royal
Meteorological Society (Weather) is an important exception. Though the level of articles is
quite variable, there is a good range from the practical to the theoretical in each monthly
issue, with a useful policy of public education.
Royal Meteorological Society, 104 Oxford Road, Reading, Berkshire, RG I 7LJ.
The American Meteorological Society publishes a monthly Bulletin for its members and
others which contains technical reviews and updates which are often informally and
accessibly expressed. 465
Bibliography, references and 45 Beacon Street, Boston, Ma 02108, USA.
other sources The Scientific American publishes review articles on climatological and other
atmospheric matters (amongst others) in a clear and graphic style by world authorities in the
subjects.

Tephigrams

British Meteorological Office blank tephigrams (Metform 2810) from Her Majesty's
Stationery Office can be ordered through The Government Bookshop, PO Box 569, London
SE1 9NH. Other materials are listed and may be available through the Marketing Services
section of the Meteorological Office, London Rd, Bracknell, Berks RG12 2SZ, UK.

References (numbered in the text)

1 J. Gleick, Chaos (London, Sphere, 1987).

2 I. Stewart, Does God Play Dice (Harmondsworth, Penguin, 1989).
3 Lucretius, The Nature of the Universe, trans. R. Latham (Harmondsworth, Penguin,
1966).
4 Meteorological Office, Handbook of Meteorological Instruments, Part I (London,
HMSO, 1969).
5 Meteorological Office, The Observer's Handbook (London, HMSO, 1982).
6 J .S.A. Green, Holiday meteorology: reflections on weather and outdoor comfort,
Weather, 22:4 (1956), 128-131.
7 Meteorological Office, Hygrometric Tables, Part II screen values, Part III aspirated
values (London, HMSO, 1964).
8 J.M. Wallace and P.V. Hobbs, Atmospheric Science (New York, Academic Press,
1977).
9 R.C. Ward, Principles of Hydrology (London, McGraw-Hill Publishing Company,
London).
lO World Meteorological Organization, International Cloud Atlas, Vol. I (revised 1975).
Simple summary in: Meteorological Office, Cloud Types for Observers (London,
HMSO, 1972).
ll J .K. Hargreaves, The Upper Atmosphere and Solar- Terrestrial Relations (Wokingham,
Van Nostrand Reinhold (UK), 1979).
12 C. C. Delwiche, The nitrogen cycle, in The Biosphere, A Scientific American Book (San
Francisco, W .H. Freeman, 1970). Also: I.M. Campbell, Energy and the Atmosphere: a
physical-chemica/ approach (Chichester, Wiley, 1977).
l3 P. Cloud and A. Gibor, The oxygen cycle, in The Biosphere, A Scientific American
Book (San Francisco, W.H. Freeman, 1971). Also: I.M. Campbell (as reference [12]).
14 J.D. Shanklin and B.G. Gardner, British Antarctic Survey (Cambridge, UK, 1989).
15 B. Bolin, The carbon cycle, in The Biosphere, A Scientific American Book (San
Francisco, W.H. Freeman, 1971).
16 Scientific American, The Biosphere (San Francisco, W.H. Freeman, 1971).
17 Siegenthaler and Oeschger, Tel/us, 39b (1987).
18 Department of the Environment, Global Climate Change (London, HMSO, 1989).
19 G.E. Likens, R.F. Wright, J.N. Galloway and T.J. Butler, Acid rain, Scientific
American, 241:4 (1979), 39-47. See also I.M. Campbell, ref. [12].
20 D.J. Spedding, Air Pollution (Oxford, Clarendon Press, 1974). And ref. [42].
466 21 F. Press and R. Siever, Earth (San Francisco, W.H. Freeman, 1974).
22 E.J. Nisbet, The Young Earth (London, Allen and Unwin, 1987). References
23 Wayne (1985) (see Textbooks).
24 J. Lovelock, The Ages of Gaia (Oxford, Oxford University Press, 1988).
25 T.W. Harrold and K.A. Brown, The polar low as a baroclinic disturbance, Quart. J.
Roy. Met. Soc., 95:406 (1969), 710-723.
26 Mintz (1974) (Diagram source).
27 E. Palmen and C.W. Newton, Atmospheric Circulation Systems, their Structure and
Physical Interpretation (New York, Academic Press, 1969).
28 G. Manley, Central England temperatures: monthly means 1659to 1973, Quart. J. Roy.
Met. Soc., 100:425 (1974), 389-485.
29 H.H. Lamb, Climate, History and the Modern World (London, Methuen, 1982).
30 J .G .L. Lockwood, The Causes of Climate (London, Edward Arnold, 1979).
31 A.H. Perry and J.M. Walker, The Ocean-Atmosphere System (London, Longman,
1977).
32 Jones and Wigley, University of East Anglia, UK (Diagram source).
33 J.T. Houghton, G.J. Jenkins and J.J. Ephraums (Eds), Climate Change- the IPCC
Assessment (Cambridge, Cambridge University Press, 1990).
34 B.S. John (Ed.), The Winters of the World (North Pomfret (Vt), David & Charles,
1979).
35 F.W. Sears, M.W. Zemansky and H.D. Young, University Physics, 6th edn (Reading
(Mass.), Addison-Wesley, 1982).
36 Warner and Telford, Journal of Atmospheric Sciences (1967).
37 A. Letestu (Ed.), International Meteorological Tables (Geneva, Secretariat of the
World Meteorological Organization, 1966).
38 Hess (1959) (see Textbooks).
39 F.H. Ludlam and J.F.R. Mcilveen, The lag of the humidity sensor in the British
radiosonde, Meteorological Magazine, 98 (1969), 233-246.
40 F.H. Ludlam, Cumulus and cumulonimbus convection, Tel/us, XVIII:4 (1966),
688-698. And ref. [27].
41 B.J. Mason, Clouds, Rain and Rainmaking (Cambridge, Cambridge University Press,
1962).
42 F.H. Ludlam, Clouds and Storms (Pennsylvania State University Press, University
Park, Pa 16802, 1982).
43 A. Eliassen, Geostrophy, Quart. J. Roy. Met. Soc., 110:463 (1984), 1-12.
44 J.R. Holton, An Introduction to Dynamic Meteorology (New York, Academic Press,
1972).
45 R. Hide, Some laboratory experiments etc., in G.A. Corby (Ed.), The General
Circulation of the Atmosphere (London, Royal Meteorological Society, 1970).
46 R.S. Scorer, Environmental Aerodynamics (Chichester, Ellis Horwood, 1978).
47 S.L. Valley (Ed.), Handbook of Geophysics and Space Environments (New York,
McGraw-Hill Book Company, 1965).
48 W.S. Von Arx, An Introduction to Physical Oceanography (New York, Addison-
Wesley, 1962).
49 J. London and T. Sasamori, Space Research, XI (1971), 639-649.
50 M.I. Budyko, Climate and Life (New York, Academic Press, 1974).
51 J.L. Monteith, Principles of Environmental Physics (London, Edward Arnold, 1989).
52 G.L. Pickard and W.J. Emery, Descriptive Physical Oceanography: An Introduction
(Oxford, Pergamon Press, 1982).
53 F. Hoyle, The Black Cloud (Harmondsworth, Penguin Books, 1975).
54 T.R. Oke, Boundary Layer Climates (London, Methuen, 1978).
55 R.E. Munn, Descriptive Micrometeorology (New York/London, Academic Press,
1966).
56 H.H. Lettau and B. Davidson (Eds), Exploring the Atmosphere's First Mile, Vols I and
II (London, Pergamon Press, 1957).
57 D.J. Tritton, Physical Fluid Dynamics (Wokingham, Van Nostrand Reinhold (UK),
1977).
58 S. Pond and G.L. Pickard, Introductory Dynamic Oceanography (Oxford, Pergamon
Press, 1981). 467
Bibliography, references and 59 Jarvis eta/. (1976) (Diagram source).
other sources 60 R.G. Barry, Mountain Weather and Climate (London/New York, Methuen, 1981).
61 J .E. Simpson, D.A. Mansfield and J .R. Milford, Inland penetration of sea-breeze
fronts, Quart. J. Roy. Met. Soc., 103, (1977), 47-76.
62 C.J.M. Aaneuson (Ed.), Gales in Yorkshire in February 1962, Meteorological Office
Geophys. Mem (London, HMSO), 14:108.
63 P.M. Saunders, An observational study of cumulus, J. Meteorol., 18 (1961), 451-467.
64 Newton (1963) (Diagram source).
65 Collinge eta!. (1990) Radar observations of the Halifax Storm. Weather, 45, 10.
66 J.P. Kuettner, Cloud bands in the Earth's atmosphere: Observations and theory, Tell us,
23 (1971) 404-426.
67 E.M. Agee, T.S. Chen and K.E. Dowell, A review of mesoscale cellular convection,
Bull. Amer. Met. Soc., 54:10 (1973), 1004-1012.
68 K.A. Browning, Radar measurements of air motion near fronts (Part II), Weather, 27
(1971), 320-340.
69 T.J. Matejka, R.A. Houze and P.V. Hobbs, Microphysics and dynamics of clouds
associated with mesoscale rainbands in extratropical cyclones, Quart. J. Roy. Met.
Soc., 106 (1980), 29-56.
70 Bergeron (1954) (Diagram source).
71 D. Defoe, The Storm (1704). Quoted inS. Brown, World of the Wind (London, Alvin
Redman, 1962).
72 B. Franklin (1744). Quoted inS. Brown (1962) (see ref. [71]).
73 S. Petterssen, Weather Analysis and Forecasting, Vol. 1, Motion and Motion Systems
(New York, McGraw-Hill Book Company, 1956).
74 Bjerknes and Solberg (1921) (Diagram source).
75 C.L. Godske, T. Bergeron, J. Bjerknes and R.C. Bundgaard, Dynamic Meteorology
and Weather Forecasting (Boston, Mass., American Met. Soc. and Washington, D.C.,
Carnegie Inst., 1957).
76 M. Minnaert, The Nature of Light and Colour in the Open Air (New York, Dover
Publications, 1954).
77 T.N. Carlson, Air flow through midlatitude cyclones and the comma cloud pattern,
Mon. Wea. Rev., 108: Oct (1980), 1498-1509.
78 P.E. Lydolph, Climates of the Soviet Union, Vol. 7 of H.E. Landsberg (Ed.), World
Survey of Climatology (Amsterdam, Elsevier Scientific Publishing Co., 1977).
79 H. Flohn (1959) (Diagram source).
80 E.T. Eady, Long waves and cyclone waves, Tellus, 1:3 (1949), 33-52.
81 Willett (1944) (Diagram source).
82 C.M. Stevenson, The snowfalls of early December 1967, Weather, 23 (1968), 156-160.
83 D.A. Mansfield, Polar lows: The development of baroclinic disturbances in cold air
outbreaks, Quart. J. Roy. Met. Soc., 100:427 (1974), 541-554.
84 P.O. Wickham, The Practice of Weather Forecasting (London, HMSO, 1982).
85 R.G. Barry and R.J. Chorley, Atmosphere, Weather and Climate (Methuen, London,
1989 and earlier editions).
86 G. Manley, Climate and the British Scene (London, Collins, 1952).
87 C.S. Ramage, Monsoon Meteorology (New York/London, Academic Press, 1971).
88 Koteswaram (1958) (Diagram source).
89 Malkus and Riehl (1981) (Diagram source).
90 R. Hughes, In Hazard (various publishers, 1938).
91 R.H. Simpson and H. Riehl, The Hurricane and its Impact (Oxford, Basil Blackwell,
1974).
92 Moller (1951) (Diagram source).
93 C.M. Cipolla, The Economic History of World Population (Harmondsworth, Penguin
Books, 1970).
94 I. Prigogine, From Being to Becoming: time and complexity in the physical sciences (San
Francisco, W.H. Freeman, 1980).
95 Rowe, M.W. and Meaden, G.T. (1985) Britains Greatest Tornado Outbreak. Weather,
40, 8, 230-234.
468
Glossary

The Glossary lists terms which are sufficiently important, widespread or peculiar to warrant
separate summary definition. Many terms used in the definitions are themselves listed else-
where in the Glossary to encourage you to refer to several entries to clarify any important
definition. Except for very basic terms whose meaning is assumed, all technical terms can be
traced through the Index to their definition in the main text, whether or not they appear in
the Glossary.

absolute acceleration Acceleration measured relative to an unaccelerated (non-inertial)

reference frame.
absolute vorticity Vorticity measured relative to a non-rotating reference frame.
absorptivity The proportion of incident electromagnetic radiant power absorbed by a
surface, expressed as a fraction or percentage.
acceleration Rate of change of velocity with time.
accretion Growth of a body of water or ice by the collection of smaller particles or
droplets.
adiabatic In thermal isolation. A thermal process is said to be adiabatic if no heat enters or
leaves the system in which it is occurring.
advection Horizontal transport (of mass, heat etc.) effected by horizontal exchange of air.
advection fog Fog formed by advection of warm air over a colder surface.
aerodynamic roughness length The height above the impermeable base of a rough surface
at which wind speed falls to zero by extrapolating a log wind profile down through the
overlying surface boundary layer.
aerosol Suspension of very small (maximum diameters of order 1 1.1m) particles or droplets
in the atmosphere, especially concentrated in the low troposphere.
air The gaseous component of the turbosphere.
air mass A synoptic-scale segment of troposphere whose temperature and humidity are
usefully related to a geographical source region. Usefulness is often marginal in the
presence of typically strong wind shears in the vertical.
air parcel A small body of air which, at least in principle, is coherent and identifiable
throughout a useful period, for example a buoyant thermal in a cumulus cloud.
albedo The proportion of incident electromagnetic radiation (usually solar) reflected or
backscattered by a surface, expressed as a fraction or percentage.
alto Prefix to cloud type, meaning 'in the middle troposphere'; as in altocumulus.
altostratus Sheet cloud in the middle troposphere (typical of fronts).
anabatic flow or winds Airflow up sloping terrain.
anemometer Instrument for measuring wind speed or run of wind.
angular momentum or moment of momentum The product of the tangential momentum
of a body and its radial distance from the axis of rotation.
469
Glossary angular velocity The rate of turning of a body about an axis expressed in angle turned per
unit time (usually radians per second).
anticyclone Synoptic-scale weather system with winds blowing anticyclonically (clockwise
in the northern hemisphere) in the low troposphere.
anvil A fan of ice cloud at the top of a cumulonimbus.
aphelion Position or time of maximum separation of planet and sun.
apparent g Gravitational acceleration measured relative to a fixed point on the Earth's
surface.
Arctic sea smoke Steam fog produced when very cold air from polar land masses flows
onto the sea.
atmospheric boundary layer or planetary boundary layer The part of the low troposphere
(- 500 m deep) most directly influnced by the underlying sea or land surface, including
the laminar, surface and Ekman boundary layers.
atmospheric waves Horizontal or vertical undulations of airflow on scales from the metric
to the hemispheric, sometimes in short trains and sometimes solitary (e.g. see cyclone and
gravity waves).
available potential energy That part of the atmospheric potential energy which could be
converted to kinetic energy by realistic redistribution of air parcels.
azimuth Wind direction specified by the angle in degrees from which the wind is blowing,
counting clockwise from zero at true north.

backing Wind direction changing so that azimuth is decreasing.

baroclinic The condition of the atmosphere when large-scale isopycnic and isobaric
surfaces are misaligned, usually in association with horizontal temperature gradients.
baroclinic instability Dynamic instability associated with baroclinic zones and their
associated thermal winds, giving rise to cyclogenesis.
barotropic The condition of the atmosphere when it is not baroclinic.
barometer An instrument for measuring atmospheric pressure.
barograph A barometer with a graphical output.
Beaufort scale A scale of wind strength assessed by effects on surface features such as trees
and sea state.
Bergeron-Findeis en A mechanism for the development of precipitation by preferred
growth of ice crystals in clouds consisting largely of supercooled water.
biosphere The interacting network of organisms living on the Earth.
blackbody An ideal body which absorbs all incident electromagnetic radiation and emits it
with maximum thermodynamic efficiency.
boundary layer A layer of fluid whose dynamic behaviour is directly influenced by an
adjacent solid or liquid surface. Meteorological examples are the atmospheric boundary
layer and the sheath of air enveloping a falling raindrop.
Bowen ratio The vertical flux of sensible heat expressed as a fraction or percentage of the
latent heat flux at a particular site or across a geographical region.
Brunt-Viiisiilii frequency The frequency of vertical oscillation of an air parcel released
after displacement from its equilibrium position in a convectively stable fluid layer.
buoyancy The net upward force experienced by a body immersed in a fluid of different
density in the presence of a hydrostatic pressure gradient.
Buys-Ballot's law The law defining the relationship between the directions of horizontal
pressure gradient and associated geostrophic wind.
cellular convection Convection organized in quasi-regular cells, as in stratocumulus or the
cellular clustering of shower clouds.
Celsius scale The temperature scale in which the freezing and boiling points of pure water
at standard atmospheric pressure are 0 and 100 units ( C) respectively.
0

centripetal Toward the centre of a circle, as in centripetal acceleration.

centrifugal Away from the centre of a circle. Ignoring a centripetal acceleration incurs an
apparent centrifugal force.
chaos Apparently random behaviour of a deterministic system.
chinook A fohn wind near the Rocky Mountains.
cirrus A fibrous cloud of ice particles.
470 climate Weather conditions and their range typical of a region or site.
cloud A dense population of water droplets and/or ice crystals, each of which has Glossary
diameter of order 10 11m.
cloud seeding Encouragement of precipitation from clouds by freezing some supercooled
cloud droplets and speeding the Bergeron-Findeisen mechanism.
cold front See front.
collision and coalescence Precipitation development by accretion, in which larger particles
or droplets grow as they fall through populations of smaller ones, collecting some by
collision and subsequent coalescence.
condensation The growth of water or ice by net diffusion from contiguous vapour - the
reverse of evaporation. In the case of ice growth, either process is often called
sublimation.
condensation nuclei The components of an aerosol population which act as nuclei for the
formation of individual cloud crystals or droplets in a realistically supersaturated vapour.
conditional instability Convective instability which arises when a previously unsaturated
layer of air fills wholly or partly with cloud.
conduction of heat Heat transfer through spreading of enhanced thermal agitation by
molecular impact.
continuity equation Formal description of the conservation of the mass of a body of
moving air, relating the net mass flux into a fixed volume to the rate of accumulation of air
mass there.
contour A height contour of an isobaric surface.
convection Usually the small-scale vertical exchange of air parcels, driven either by
buoyancy (thermal convection - although density differences do not always depend
entirely on temperature differences) or wind shear (mechanical convection). Thermal
convection is usually assumed unless mechanical convection is specified. See slope
convection for application to the large scale.
convective instability Tendency toward buoyant convection arising from disposition of
atmospheric density.
convective stability Inhibition of buoyant convection arising from atmospheric density
disposition.
convergence Negative divergence.
Coriolis acceleration Component of absolute acceleration arising from the motion of a
body relative to the rotating terrestrial reference frame. In meteorology it usually means
the component in the local horizontal arising from horizontal flow.
Coriolis parameter The factor 2 1l sin t/>, where 1l is the Earth's angular velocity and t/> is the
angle of latitude.
covariance Average value of the product of deviations of matched pairs of values (of
vertical and horizontal wind speeds, for example) from their respective mean values.
cryosphere Ice caps, pack ice, glaciers, permafrost - the frozen water substance at and
below the Earth's surface.
cumulus The family of hill-shaped clouds ranging from small cumulus to large
cumulonimbus. The term on its own usually denotes small cumulus or only slightly larger.
cumulus congestus A moderate or large cumulus cloud, growing conspicuously.
cumulonimbus A precipitating cumulus, often but not necessarily having an anvil.
cyclone Synoptic-scale weather system with winds in the low troposphere rotating about
the local vertical in the same sense as the local terrestrial surface but faster (i.e. anti clock-
wise on a northern hemisphere map). Also a local name for a severe tropical cyclone in the
Indian Ocean.
cyclone wave The wave-like deformation of flow in the middle and upper troposphere
associated with an extratropical cyclone.
cyclonic shear Horizontal flow sheared in such a sense as to cause an embedded parcel to
rotate cyclonically.
cyclogenesis Cyclone formation, usually applied to extratropical cyclones.
cyclostrophic flow Horizontal circular air motion in which centripetal acceleration on the
weather map is effectively maintained by centripetal pressure gradient force, Corio lis and
friction terms being negligible.
density Mass concentration expressed as mass per unit volume.
depression See extratropical cyclone. 471
Glossary detrainment Loss of mass from a rising thermal to surrounding air. The reverse of
entrainment.
dew Water condensed onto a solid surface which has cooled to the dew point of the
contiguous air.
dew point or dew-point temperature The temperature at which a given parcel of air
becomes saturated during isobaric cooling with conservation of vapour content.
dimensions The dependence of a measurable quantity on the fundamental measurables -
mass, length, time and temperature. Thus the dimensions of velocity are length divided by
time.
dimensionless number Any fundamenal parameter (made up of a dimensionless assembly
of observable factors) which usefully describes a regime of dynamic or thermodynamic
behaviour, e.g. a Reynolds number.
diurnal With a daily cycle.
divergence (of velocity) The resultant rate of stretching of a fluid as given by the sum ofthe
longitudinal gradients of the three components of flow velocity. In meteorology it usually
refers in particular to the stretching components of air in the local horizontal.
drag coefficient The dimensionless ratio of the drag force on a body immersed in a fluid
flow to the incident or adjacent fluid momentum flux.
drizzle Precipitation of water droplets with radii - 100 /lm.
dry adiabat A line describing a dry adiabatic process on a thermodynamic diagram such as
a tephigram.
dry adiabatic Describing an adiabatic movement or process of an air parcel in which there
is neither evaporation nor condensation of the water substance - also called isentropic,
because entropy is conserved.
dry adiabatic lapse rate Temperature lapse rate equal to the value found in dry adiabatic
ascent or descent of an air parcel.
dry air Normal turbospheric air without water vapour.
dynamics Study of forces and associated motional deformation.
dynamical coefficient of viscosity Tangential shearing stress arising from viscosity,
divided by lateral shear (spatial velocity gradient). Also known as the Newtonian
coefficient of the same.
easterly Flowing or moving from the east.
easterly wave A synoptic-scale wave-like disturbance of the low tropospheric easterlies of
low latitudes.
eddy Individual transient element of turbulence - often pictured as a translating and
rotating air parcel.
eddy viscosity Apparent viscosity arising from turbulence.
electromagnetic radiation Travelling waves of electric and magnetic disturbance, capable
of passing through empty space. Of the full electromagnetic spectrum, solar and
terrestrial radiation are the most important components in meteorology.
Ekman layer The topmost layer of the atmospheric boundary layer, between the free
atmosphere and the surface boundary layer.
emissivity The actual emittance of a radiating surface expressed as a fraction of the
blackbody value for the surface temperature.
emittance The flux of electromagnetic radiant energy emitted by unit area of a radiating
surface.
energy The capacity to do work. See kinetic, gravitational potential, radiant and heat
energies for meteorologically important types.
entrainment Incorporation of surrounding air by a rising thermal.
entrainment drag The effective drag on a thermal arising from continual sharing of
momentum by entrainment of nearly static ambient air.
equation of motion Equation of relative acceleration with net force acting on an air parcel
of unit mass - a form of Newton's second law of motion.
equation of state Equation relating pressure, density and temperature of a parcel of an
ideal gas or mixture of ideal gases.
equivalent potential temperature The potential temperature of an air parcel after all water
vapour has been condensed by adiabatic decompression.
472 equinox A time when daylight and night are equally long.
evaporation Reverse of condensation. Glossary
exponential growth (or decay) Profile of growth (or decay) of a quantity whose gradient is
directly proportional to the quantity itself.
extratropical cyclone A cyclone outside the tropics, distinguished from a tropical cyclone
by greater scale, the presence of one or more fronts and the absence of great central
intensity.
fall speed See terminal velocity.
fetch Distance upwind from point of observation to a significant location, as in 'standard
synoptic anemometers should be sited to have an unobstructed fetch of at least 100 m'.
flux of energy, etc. The rate of flow of energy, etc. per unit time through a real or
imaginary surface (normal to the flow unless otherwise specified).
flux density of energy, etc. The flux per unit area of real or imaginary surface.
fog Dense cloud in contact with a land or water surface, with density specified by visibility.
fohn A strong, dry, katabatic wind, produced by prior enforced ascent of air over high
parts of the European Alps.
free atmosphere The atmosphere above the atmospheric boundary layer (i.e. above the
gradient level).
freezing fog A fog of supercooled water droplets, freezing on impact with any solid
surface.
freezing nuclei Those solid components of an aerosol population which act as nuclei for
the freezing of supercooled cloud droplets.
frequency The number of events per unit time.
friction velocity A basic wind speed parameter defined by .Jrl p, where r is the turbulent
horizontal wind stress in the surface boundary layer and p is the air density there.
front A swath of cloud and precipitation which is synoptic-scale in length and at least
large-meso-scale in breadth, and is associated with a significant horizontal temperature
gradient in the low troposphere of an extra tropical cyclone. A front is called warm or cold
depending on whether the warmer or colder air is advancing, and is called occluded when
it connects the warm sector to a separated surface pressure minimum.
frontal zone A region of continually or seasonally preferred cyclogenesis. Also the meso-
scale zone of maximum lateral horizontal temperature gradient associated with a front.
frost Deposition of ice on a land surface by diffusion and sublimation. When thick enough
to produce marked whitening of vegetation (especially grass), it is called hoar frost.

gale A wind whose ten-minute average speed at height 10 equals at least 37 knots.
gas constant The constant in the equation of state for an ideal gas. In the universal form, a
single universal gas constant applies to all gases. In the meteorological form, a different
specific gas constant applies to each gas or uniform mixture.
geostrophic wind A horizontal wind in which the Coriolis acceleration is exactly main-
tained by the horizontal pressure gradient force (or equivalently the Coriolis and pressure
gradient forces exactly balance), i.e. when friction and acceleration relative to the Earth's
surface are zero.
glaciation The process of conversion of supercooled water cloud into ice cloud.
geostationary orbit An equatorial satellite orbit such that the satellite is stationary relative
to the Earth.
gradient The spatial rate of change of an observable in the direction of maximum rate of
increase (as in temperature gradient).
gradient level The lowest level in the troposphere which is so free of surface drag that
actual and gradient winds are indistinguishable.
gradient wind A horizontal wind in which the combined Coriolis and centripetal
accelerations relative to the Earth's surface, arising from cyclonic or anticyclonic flow,
are exactly maintained by the horizontal pressure gradient force.
gravitational acceleration or apparent g The downward acceleration of a body falling
freely in vacuo, as measured from a frame fixed to the Earth's surface. Equivalently it is
the gravitational force per unit mass of the body, as measured from the same frame.
gravity waves Waves of fluid disturbance in which gravity is the predominant restoring
force.
greenhouse effect The elevation of surface and low-troposphere temperatures which is 473
Glossary maintained by the atmosphere's transparency to solar radiation and opaqueness to terre-
strial radiation.
gust A short positive departure from the ten-minute average wind speed.
Hadley circulation The background vertical and meridional circulation of air in low
latitudes, consisting of two opposing Hadley cells, each having air rising in the inter-
tropical convergence zone and sinking in a subtropical anticyclone.
hail Millimetric or larger precipitation particle of ice, formed by the accretion of ice
crystals and rapidly freezing supercooled water droplets.
heat The energy of a material which is stored in the form of sensible heat (kinetic energy of
rotational, vibrational and translational thermal motion of constituent atoms and
molecules) and/or latent heat.
heat capacity The amount of heat required to raise the temperature of a body by one
degree kelvin. In air or any other highly compressible fluid, its value differs considerably
depending on whether heating occurs at constant volume or constant pressure, the latter
being the larger.
high pressure zones (highs) Regions of raised atmospheric pressure at mean sea level. Also
known as anticyclones.
hoar frost See frost.
humidity The vapour content of air. See relative and specific humidity.
humidity mixing ratio The mass of vapour as a fraction of the mass of dry air with which it
is mixed in a moist air parcel. Numerically indistinguishable from specific humidity in all
but the most humid air.
hurricane See tropical cyclone.
hydrologic cycle The network of pathways of the water substance through the oceans, land
surfaces and atmosphere.
hydrosphere The shell of water which nearly envelopes the Earth in the form of oceans and
inland seas. Sometimes includes the cryosphere, groundwater and atmospheric water
substance.
hydrostatic equation The formal expression of pure hydrostatic equilibrium.
hydrostatic equilibrium Balance between gravitational downward pull on an air parcel and
upward pressure gradient force arising from vertical pressure lapse. Such equilibrium
applies very accurately to all but the most violently disturbed parts of the atmosphere.
hygrograph A recording hygrometer.
hygrometer An instrument measuring humidity, often by measuring relative humidity.
hygroscopic Tending to attract and condense ambient water vapour.
ideal gas A gas which behaves as if its molecules were infinitely small, interacting only by
perfectly elastic collision at the instant of collision, and therefore obeying the equation of
state for an ideal gas.
ideal gas constant See gas constant.
inertial reference frame A reference frame with zero absolute acceleration.
infra-red radiation Electromagnetic radiation lying beyond the red end of the visible
spectrum, but not far beyond.
insolation Solar irradiance of a surface, sometimes totalled over a finite time period such
as a day. ·
instability The condition of a body or system which responds to a specified disturbance by
increasing the disturbance until an irreversible change has taken place. Sometimes used in
context to mean convective instability.
internal energy The sensible heat capacity of an ideal gas as assessed from its specific heat
at constant volume.
intertropical convergence zone (ITCZ) The zone of persistent convergence of airflow in
the low troposphere in very low latitudes.
inversion An increase of atmospheric temperature with height (an inversion of the normal
tropospheric lapse).
ionosphere The region of the upper atmosphere (usually reckoned above 50 km height)
which is chronically in a significantly ionized state.
irradiance The energy flux density of electromagnetic radiation impinging on a real or
imaginary surface.
474 isentropic See dry adiabatic.
isobar An isopleth of atmospheric pressure. Glossary
isopycnal or isopycnic An isopleth of air density.
isopleth A line or surface with uniform value of a specified property.
isotach An isopleth of wind speed.
isotherm An isopleth of air temperature.
isotropic Independent of direction; thus microscale turbulence is nearly isotropic while
larger scales are vertically squashed.
jet stream A relatively narrow and shallow stream of fast flowing air, usually in the high
troposphere (see polar front and subtropical jet streams).
joule The unit of energy.
katabatic winds Airflow down sloping terrain.
kelvin The unit or scale of absolute temperature.
kinetic energy The energy which a body has in respect of its motion.
knot A wind speed of one nautical mile per hour.
laminar flow Smooth, viscosity dominated flow.
lapse rate See temperature lapse rate.
latent heat The heat which is given out when gases and liquids liquify or solidify, and
which is absorbed when solids or liquids melt or evaporate.
lateral acceleration Acceleration across the direction of flow, e.g. centripetal acceleration
in cyclonic or anticyclonic flow.
lee trough A synoptic-scale low-pressure system formed or maintained in the lee of a long
ridge of high ground.
lifting condensation level The level at which air would become saturated if lifted dry-
adiabatically conserving its vapour content (often referring to initial conditions at screen
level).
lightning The large electric sparks produced in and around thunderclouds.
linear acceleration Acceleration in the direction of flow.
log wind profile The nearly linear profile of wind speed in the surface boundary layer when
plotted against log height.
mass The quantity of material in a body as assessed by its reluctance to accelerate when
acted on by a force when free of all other constraints.
mechanical convection Vertical exchange of air parcels accomplished by turbulence driven
by vertical wind shear.
melting band The horizontal band of enhanced radar echo produced by snow melting at
the 0 oc level in extensive precipitating clouds.
meridional Along a line of meridian.
mesosphere The region of the upper atmosphere lying between the stratosphere and the
thermosphere.
mesoscale Spatial scales intermediate between small and synoptic scales of weather
systems.
microphysics of clouds Physical processes active on the scale of individual cloud and pre-
cipitation droplets and particles.
millibar (mbar) The meteorological unit of pressure; 100 pascals.
mole A mass of pure material whose mass in grams is numerically equal to the material's
atomic or molecular weight, and therefore contains Avogadro's number of atoms or
molecules.
momentum The product of the mass and velocity of a body- a vector quantity.
monsoon Pronounced seasonal variation of airflow in the lower troposphere in tropical
and subtropical regions - for example the Indian monsoon. Sometimes used for similar
but weaker behaviour in higher latitudes.
neutral stability The state of a system which accepts a specified disturbance without fur-
ther response. Often used to mean neutral convective stability: the state of an atmospheric
layer which accepts imposed vertical displacement of air parcels without positive or nega-
tive response (i.e. which is on the boundary between convective instability and stability).
nimbostratus An extensive layer of cloud precipitating rain and/ or snow.
non-inertial reference frame A reference frame with non-zero absolute acceleration. 475
Glossary normalize To divide a quantity by a more fundamental quantity of the same dimensions to
produce a non-dimensional ratio.
Normand's theorem A particular thermodynamic relation between temperature, ther-
modynamic wet-bulb temperature and dew-point temperature, expressed in terms of a
construction on a tephigram or other thermodynamic diagram.
northerly Flowing or moving from the north.

occluded front See front.

order of magnitude The typical magnitude of a quantity to the nearest integral power of
10.

pascal The SI unit of pressure.

period The interval between consecutive similar stages in a repeating cycle of events.
planetary boundary layer See atmospheric boundary layer.
photochemistry Chemical processes encouraged by incident photons of electromagnetic
radiation (solar radiation in meterology).
photodissociation Molecular dissociation caused by absorption of photons of
electromagnetic radiation (solar radiation in meteorology).
photoionization Ionization caused by absorption of photons of electromagnetic radiation
(solar radiation in meteorology).
photosynthesis Synthesis of water and carbon dioxide to form oxygen and sugars in the
presence of sunlight and photosynthetic organisms.
Poisson's equation Equation relating initial and final absolute temperatures and pressures
of an ideal gas undergoing dry adiabatic compression or decompression.
polar front The middle-latitude frontal zone separating air flowing from tropical and
polar source regions. A favoured site of cyclogenesis.
potential energy The energy of a body (or system of bodies) in respect of its position in a
field of force. Used in meteorology to mean gravitational potential energy.
potential temperature The temperature of an air parcel after dry adiabatic compression or
decompression from its actual pressure to 1000 mbar.
power Energy input or output per unit time.
precipitation Ice particles or water droplets large enough (- 100 ~m or larger) to fall at
least 100 m below cloud base before evaporating. See drizzle, hail, rain and snow.
pressure The apparently continuous and isotropic force exerted on unit area of any real or
imaginary surface because of bombardment by molecules of contiguous fluid.
pressure gradient force The net force on an air parcel arising from its location in a pressure
gradient.
pressure tendency The rate of change of pressure at a fixed location, usually at the Earth's
surface.
partial pressure The pressure which would be exerted by a particular component of a
gaseous mixture if all other components were removed without other change.
psychrometer An instrument measuring vapour content by means of wet- and dry-bulb
thermometry.

radian The angle subtended at its centre by an arc of a circle equal to its radius.
radiometer An instrument to measure fluxes of electromagnetic radiation.
radiosonde A balloon-borne package of thermometer, barometer, hygrometer and radar
reflector, for sensing the troposphere and low stratosphere during free ascent.
radiation See electromagnetic radiation.
radiant flux Flow of electromagnetic radiant energy per unit time.
radiant flux density Radiant flux through unit area of real or imaginary surface, usually
perpendicular to the rays.
radiation fog Fog produced by net radiative cooling of the underlying surface.
rain Precipitation in the form of millimetric-sized water droplets (as distinct from drizzle).
rain gauge An instrument to measure totals or rates of rainfall, including drizzle and
melted hail and snow.
reference frame A set of moving or stationary spatial axes used as a basis from which to
measure body motion and dynamics.
476 reference process See thermodynamic reference process.
reflectivity The proportion of incident radiation reflected by a surface, expressed as a Glossary
fraction or percentage.
relative acceleration Acceleration measured relative to a specified reference frame, which
may itself have absolute acceleration.
relative humidity The vapour content of air (measured as vapour density or pressure) as a
percentage of the vapour content needed to saturate air at the same temperature.
relative vorticity Vorticity about the local vertical relative to the local tangential plane of
the Earth.
residence time The average time spent by a particle in one particular component of a
system of components in dynamic equilibrium (for example, by a water molecule in the
oceans between precipitation and the next evaporation).
reversible Of a thermodynamic process which is always so close to equilibrium that it can
be reversed by a minute change in the set-up.
Reynolds number The dimensionless ratio of fluid acceleration and accelerations induced
by viscosity typical of a particular flow regime.
Richardson number The dimensionless ratio of velocity shears and buoyancy forces
typical of a particular flow regime.
ridge An elongated zone of high pressure.
Rossby number The dimensionless ratio of relative accelerations and Coriolis
accelerations typical of a particular regime of synoptic-scale flow.
run of wind The length of airflow registered by an anemometer in a certain time period
(often an hour or a day).

saturated adiabat A line on a tephigram, or other thermodynamic diagram, corresponding

to a saturated adiabatic process.
saturated adiabatic lapse rate temperature lapse rate associated with saturated adiabatic
ascent or descent of an air parcel.
saturated adiabatic process An adiabatic process in which air is kept in an exactly
saturated state by evaporation or condensation of water substance.
saturation The state of water vapour which is in dynamic equilibrium with a contiguous
plane surface of pure water or ice through balanced fluxes of evaporating and condensing
water molecules.
scale height The vertical height interval in which a certain atmospheric property (such as
pressure) decreases to a certain proportion (such as 1110) of its value at the base of the
interval.
scale analysis Analysis of the rough magnitudes of individual terms in dynamic and
thermodynamic equations, to determine their relative significance.
scattering The process of chaotic deflection of electromagnetic radiation by impact on
rough surfaces.
screen See Stevenson screen.
seeding of clouds See cloud seeding.
severe local storm A very severe cumulonimbus.
shear Cross-flow gradient or difference of the direction and/ or speed of fluid flow (speed
if direction is unspecified).
shear stress The tangential force per unit area of sheared fluid which is associated with the
shear through molecular or turbulent exchange across the flow.
slope convection Synoptic-scale ascent and descent associated with extratropical cyclones
and anticyclones.
smog Polluted fog.
snow Precipitation in the form of well-developed hexagonal ice crystals, either singly or in
conglomerate flakes.
solar constant The average value of normal solar irradiance just outside the Earth's
atmosphere. The term can be used for other planets.
solar radiation Electromagnetic radiation from the Sun, especially in the wavelength range
0.3 to 3 ~m which contains nearly all the total irradiance.
solarimeter An instrument to measure solar irradiance.
specific mass The mass of a certain material as a proportion of the total mass of mixture in
which it is mixed and of which it forms a part. 477
Glossary specific lleat capacity The quantity of heat needed to raise the temperature of unit mass of
a particular substance by one degree kelvin.
specific humidity The mass of water vapour as a proportion of the total mass of moist air
of which it forms a part.
spectrum The distribution of variance or power across wavelength or frequency.
squall A sudden onset of strong, gusty winds.
squall line or line squall A linear organization of vigorous cumulonimbus in the United
States.
sound Longitudinal pressure waves audible to the human ear.
stability The condition of a body or system which responds to a specified disturbance by
opposition and suppression. Often used in meteorology to refer to convective stability in
particular.
station pressure Atmospheric pressure measured at station level before correction to sea
level.
steam fog Fog formed by evaporation from a relatively warm water surface into cooler air.
Stevenson screen A white, wooden ventilated box used to allow thermometers to measure
air temperature without radiative influence.
storm A general term applied to any type of weather system associated with strong surface
winds.
stratosphere The relatively quiet atmospheric layer between the troposphere and
mesosphere.
streamline A line which is instantaneously parallel to fluid flow.
sublimation See condensation.
subsidence inversion The temperature inversion in the low troposphere of the subsiding
core of an anticyclone.
subtropical high pressure Zones of systematically raised surface pressure between lati-
tudes 20° and 40°.
subtropical jet stream A westerly jet stream centred near the 200 mbar level at the polar
extremity of each Hadley cell.
superadiabatic lapse rate A rate of fall of temperature with height which exceeds the dry
adiabatic lapse rate.
supergeostrophic wind Wind with speed larger than the local geostrophic value.
supersaturation Greater vapour pressure or density than is required for saturation at the
prevailing temperature.
synoptic observations Observations made simultaneously at the established synoptic grid
of positions.
synoptic scale The minimum horizontal spatial scale of weather system which is well
defined by the synoptic observation network.
temperature The concentration of sensible heat in a body, measured according to an.
arbitrary scale such as the Celsius and Kelvin scales.
temperature lapse rate The rate of fall of temperature with increasing height. See dry and
saturated adiabiatic lapse rates.
tephigram A thermodynamic diagram having temperature and potential temperature as
perpendicular axes.
terminal velocity or fall speed The speed at which a particular body's weight is balanced by
its drag as it falls through a particular fluid.
terrestrial radiation Electromagnetic radiation emitted by materials at terrestrial surface
and atmospheric temperatures.
thermal A buoyant eddy, especially those of size - 100m and larger.
thermal conductivity The ratio of heat-flux density to temperature gradient for any
particular heat-conducting material.
thermal convection Buoyant convection driven by temperature differences.
thermal diffusivity Thermal conductivity divided by the product of density and specific
heat capactiy of a heat-conducting material.
thermal wind The vertical shear of geostrophic wind.
thermocline A region below a water surface in which temperature increases upward.
thermodynamic diagram A diagram whose axes are thermodynamic parameters chosen sr
478 that area represents energy.
thermodynamic reference process An adiabatic (dry or saturated) series of thermo- Glossary
dynamic states of an air parcel used for comparison with complex atmospheric reality.
thermodynamic wet-bulb temperature The lowest temperature to which an air parcel can
be cooled by adiabatic evaporation of water into it.
thermograph A recording thermometer.
thickness The vertical depth of a slab of air bounded by chosen isobaric surfaces.
thunder The audible noise made by lightning.
tornado Intense, cloud-cored vortex extending from the base of a severe local storm to the
surface.
torque The turning moment of a force.
total potential energy The sum of the internal and gravitational potential energies of an
atmospheric column.
trade winds The conspicuously reliable winds blowing obliquely in the low troposphere
from a subtropical high to the intertropical convergence zone.
trajectory The path traced by a moving air parcel.
transmissivity The proportion of a flux of electromagnetic radiation which is transmitted
through the atmosphere (or part of it).
tropical cyclone A cyclonic disturbance in tropical regions. When surface winds exceed
33 m s -t it is called a hurricane in the Atlantic, a typhoon in the Pacific, and a cyclone in
the Indian Ocean.
tropopause The limit of the troposphere (usually well marked and lying between 10 and
15 km above sea level).
troposphere The layer of air continually stirred by weather systems (bounded by the
surface and the tropopause).
trough An elongated zone of low atmospheric pressure at a horizontal surface.
turbosphere The atmospheric layer in which diffusive equilibrium is swamped by convec-
tive equilibrium (from the surface to about 90 km above sea level).
turbulence Apparently chaotic fluid motion.
typhoon See tropical cyclone.
ultraviolet radiation Electromagnetic radiation between the violet end of the visible
spectrum and X-radiation.
universal gas constant The constant linking pressure, volume and absolute temperature of
a mole of ideal gas.
unstable The condition of a system which amplifies a particular type of imposed
disturbance, the type of instability depending on the type of disturbance, e.g. convective,
dynamic and baroclinic.
upper-air observations Synoptic observations made by radiosonde or equivalent
technique.
vacillation Irregular fluctuation between minimum and maximum sinuosity of middle
tropospheric flow in middle latitudes.
vapour The gaseous form of a substance. Often used as an abbreviation for water vapour.
vapour pressure The partial pressure of water vapour.
vector A vector quantity has both magnitude and direction.
veering Wind direction changing so that azimuth is increasing.
velocity The rate of change with time of the position of a body in a specified reference
frame.
viscosity The friction in gases and liquids arising from molecular exchange and impact.
Sometimes called molecular viscosity to distinguish it from eddy viscosity.
vorticity The measure of fluid rotation. In synoptic scale meteorology the term usually
means the relative vorticity about the local vertical.
warm front See front.
warm sector The wedge of relatively warm troposphere between a warm front and the
following cold front.
water substance Water in any of its physical states -liquid, ice or vapour.
watt One joule per second - the SI unit of power.
wave cyclone An extratropical cyclone (sometimes applied especially to the immature
phase). 479
Glossary wavelength The shortest distance between adjacent wave crests (or any other similar parts)
in a train of waves. In a solitary wave it is taken to be the length of the perceptible
distortion.
waves Regular or nearly regular distortions of physical properties which repeat regularly
(or nearly so) in space and time.
westerly Flowing or moving from the west.
wet-bulb depression The depression of wet-bulb temperature below dry-bulb temperature.
wet-bulb potential temperature The wet-bulb temperature after an air parcel is taken
adiabatically to 1000 mbar. Equivalently it is the temperature at which the saturated
adiabat of the parcel crosses 1000 mbar on a tephigram.
wet-bulb temperature The temperature registered by a wet-bulb thermometer.
wet-bulb thermometer A thermometer which registers the temperature reached by free
evaporation from a saturated wick.
wind Air motion relative to the Earth's surface, usually its horizontal component.
wind shear Often the vertical shear of horizontal wind. See shear.
wind stress The horizontal stress exerted on a real or imaginary surface because of
adjacent wind shear.
wind vane A device to measure wind direction.
zonal Along a line of latitude.
zonal index A pressure index of sinuosity of large-scale tropospheric flow in middle
latitudes.

480
Notes on selected
problems

Chapter 1

5 Pressure is scalar but pressure gradient is vector. 7 About 11100 of the horizontal. 8 4. 9
0.1 N. 10 1000 m 3 • 11 Add isotherms to make a subsidiary temperature maximum or
minimum or both between A and B. 12 Anticyclone over the British Isles is weakening and
belt of rain is moving from Atlantic into NW Ireland. 13 Work out length/timescales ratio
for fictional hurricane and compare with Fig. 1.5. 14 Consider atmospheric equivalents of
warmed bearing, tyres, ground and slipstream of car.

Chapter 2

9 42.8 units. 10 78.5 Pa. 11 12 mm and 4.8 m. 12 46.1 mbar. 13 1053 Z, 190.4 km and
53.5 °E of N. 14 0.11 °C. 15 Glass reduces solar radiation and breezes; warm interior
radiates infra-red. 16 For same height rise, pressure fall is directly proportional tog, hence
much less on Moon. Barometer fall is same as on Earth (but what about an aneroid baro-
meter?). 17 Drops chill window by evaporation until inside glass surface falls below interior
dew point. 18 Snowfall distorted by airflow, like rain but much more so (why?). Unlike rain,
after falling, snow is not redistributed by gravity, but is by moderate winds or stronger.
19 Heat flux to bulb directly proportional to bulb surface, but heat capactiy of bulb directly
proportional to bulb volume. Hence speed of response proportional to surface/volume, i.e.
11R in sphere radius R.

481
Notes on selected problems
Chapter 3

1 No, since each type will give uniformity. 8 1.11 grams, depleted by respiration. 9 10- 4 kg
1- 1, 1.25 x 102 1 molecules 1- 1• 10 21.8 yr, 403 units. 11 Increase from combustion, further
increase by reduction in photosynthesis and increase of decay, slow decrease as chilling
ocean absorbs more. 15 Atmospheric source trivial, ice-cap source considerable but very
slow. River flooding? Folk memory of inundation of Mediterranean? Coastal inundation
after explosion of Thera? 16 Sloping terrain at coasts. Sea level rises, but so does land freed
from ice burden. 17 Assume all vapour precipitates and use chain rule.

Chapter 4

9 0.465 kg m- 3 • 10 2497, 2355, 3576 m. 11 246.6 Kat 70 °N, 267.3 Kat 30 °N. 12 (a) 10
days, (b) 13 days. 13 51.3 and 95.2 minutes. 14 0.2 (core speed 60 m s- 1), 3.4 x 10- 3 • 15
Notep = p RdTwherep and pare air pressure and density. Tv = 290.8 K, cf. T = 288.2. 16
10.9 km for Mars, low atmosphere temperature 220 K. 17 Troposphere much shallower,
tropopause and low stratosphere much warmer. Westerlies aloft and easterlies below
removed. No rotation, daylight 6 months long. 18 High albedo ice reduces solar input,
produces more ice. Baroclinity drives weather systems, precipitates more snow but produces
more blanketing cloud.

Chapter 5

2 Note tornado core has low pressure. 8 4. 7 oc. 9 10.6 and 7. 7 g kg- 1, 730?o. 10 2550 m. 11
540?o. 12 - 26.6 °C. 13 55.9 °C, about 18 oc, about 57 oc. 14 1.3 °C. Cooling could be
furtherreduced.15020 °C,q9 gkg- 1,e14.5mbar, Td 12.2 oc, Tw =Ow= 15.2 °C,LCL
pressure 892 mbar, Oe 46 oc. 16 - 8 oc. Warming by contact with warm interior, heaters
etc. 17 Implies 1000 mbar T 55.9 oc, Tw 17 oc, surface T about 19 oc, looks like ocean
surface below 30 °C latitude in winter, 40 o in summer.

Chapter 6

4 Lower RH being lower in the sub-cloud layer, sunshine stronger. 8 Large drips not yet at
break-upspeedinshortavailableh eight.112.14 X w- 3 and2.25 X w- 2 kgm- 3 . Sutface
air grossly subsaturated. 12 7.3%, 5000 m. 13 0.36, 3.57 ~m. 14 0.56 sat 11 oc, 5.2 sat
-20 °C assuming same D value. 15 13.4 min to 500 ~m. another 116 s to 1 mm. 16 Thunder
482 begins 0.61 s after flash, loud bang centred about 3 s after flash, end of rumble 6.1 s after
flash. 17 9.3 x 1021 molecules m- 2 s- 1 deposited, 8.6 x IOUi impacts. 18 0.62, 6.181-'m; Notes on selected problems
0.124, 0.012o/o. 19 Bases- explosive growth, well-mixed layer; edges- rapid evaporation
in subsaturated air. Deactivated invisible, activated highly visible. 20 About 830 m.

Chapter 7

1 No vertical contribution. 3b Reading enhanced. 7 About6 m s- 1 compared with 100m s- 1

needed if air used. 11 Assuming craft contents are weightless, i.e. it is in ballistic orbit or
flight. 14 300 m s - 1 ! Ignore Corio lis. 15 0.24 m s - 1• 16 Note longitudinal accelerations
- 4 x w- 3 m s - 2 • 17 39.7, 68.8, 394 m s- 1 • The last is very unrealistic. Pressure gradients
like this occur in hurricanes, but balance is then cyclostrophic not geostrophic. 18 Each
speed 4 x greater. 5.1 x w- 4 in all cases (i.e. 51 m per 100 km). 19 20m s- 1 • 21 3.28 x
w- 2 m s- 2 , 8.1 m s- 2 • 22 I0- 2 m s- 2 , i.e. about 1/3. 23 4215, 1776, 749 km 2 • 24 I, 3 and
5 em s- 1• 251.98, 3.93, 7.78 absolute and0.98, 2.93, 6.78relativevorticities, all in units of
w- 4 s- 1• 26 Note deep tropospheric drag by mountain waves and deep convection. 28
Obviously more isotropic. Effective height of solar heat input might not then be at the
surface, giving marine-like inactivity in lower troposphere. 31 Sloping frontal zone gives
sharp T gradient in only shallow zone on any vertical. 32 Slope 10 : 1 (not far from vertical
- slowly diverging funnel where pressure = LCL value).

Chapter 8

3 Note this is the opposite of the conventional radiometer output, which would make the cat
look similar in theIR and visible wavelengths. 9 6.5 o. 16 2.696, 1.397, 0.605 kW m - 2 • 17
244 K. 18 448.3, 463.8, 339.6, 233.0, 167.7 W m - 2 • 19 Sc- uniform darkish field, darker
than visible; Cb - white tops in dark field; frontal decks - white stripes amid darker on
dark field. 20 Stratosphere/troposphere = 1.43. 21 699, 295 W m- 2 • 25 Solar constant 5.6
x present value, Earth much hotter, terrestrial radiation in near IR, solar radiation red,
wavelength overlap between solar and IR. 27 T~ = 2T!, T~ = S/[u (I - a/2) ]. With S =
350 W m - 2 , T8 is 318 K, Ta is 267 K. 28 E.g. snow-covered low ground not too cold, hence
warmer and darker than high cloud in IR though indistinguishable in visible. 29 439 W m - 2
above atmosphere, compared with 300 W m - 2 observed suggests effective albedo 0.32.
10 Polewards and eastwards. 14 ITCZ air cooler and saturated; STH air warm and very
unsaturated. 22 Water 19 oc, atmosphere 79 °C, ice depth 0.32 m. 23 About 14 kW per
head of 4 billion population. 24F = 31.3 W m- 2 , reasonable agreement. 33 See problem 22.
34 Find mass of moving water to be 5 x 105 kg m- 2 • The average poleward heat flux is 31 MJ
m - 1 • Fair agreement, though a little low.

483
Notes on selected problems
Chapter 9

3 0.125. 7 The transparency since it exposes a huge mass. 10 Values keeping well above 0.25
even in the coldest (pre-dawn) period. 17 906, 87,574 W m- 2 • 18 428,41,271 W m- 2 • 19
11.84, 5.59 MJ m- 2 • 20 0 °, i.e. on the horizon, 47 °. 21 About 0.3 m s- 1 depending on
choice of 0 (note can treat 0 as T). 22-2 m 2 s - 2 , .J2 m s -I, 1. 7 N m - 2 • 23 6.25 em, 7.32 m
s -I, 0.58 m s -I. 24 2 g kg -I (assuming not far from 1000 mbar at comparison level and
apparently assuming saturated rising air, but see next problem). 25 780 mbar, 2.3 km. Note
that the maximum upslope/downslope T difference is preserved throughout the upslope
sub-cloud layer. 26 Little cloud, hot summers, cold winters, snow, big range of day length
and noon solar elevation. 27 Consider turbulent fluxes of sensible heat and momentum in
the vertical. 28 Maximum solar elevation from problem 20. Solar day 24 hr long, solar input
by method of problem 19 34.1 MJ m- 2 • The LW loss depends heavily on humidity and cloud
cover of sky. Crudely, ground Tis 283 K, sky T 233 K, all day input gives 17 MJ m -- 2 • 29
1.303.

Chapter 10

2 0.66 m s -I. 4 Consider hexagonal cells crossing observer, some diametrically, some
tangentially. 54 km. 7 Aretheythrownoraretheyleft? 8 50 mmhr- 1 • 92.5, 0.33, 2.83 oc.
10 At 800 mbar T values are 20.2 and 16 °C. Convection inhibited. 11 ao;az is
approximately 1150 °C m- 1 and assuming 0 273 K, B-V time period is 3.9 min. 12 12.8 m
s- 1 usingappendix 11.2replacingrd by(rd + 1/200).1383 ms- 1 (i.e. a jet core). An under-
estimate. 14 No room for compensating sink. 15 Buoyancy increases on condensation,
decreases on evaporation. Locate active zones on simple cloud model. 16 6. 7 oc, surface
wet-bulb temperature. 17 Cooling by 1.25 °C. After appendix 11.1 this is partly offset to
give effective cooling of 1.05 °C. 19 Need about 25 km height to fall through assuming
average air density 0.8 kg m- 3 • Large stone held while falling through updraught at 10 m
s -I for 30 min followed by 12 km fall to ground, gives effective total 24 km. Suggests such
growth needs encounter with rich droplet source when large (near base of updraught?). 21
12 m s - 2 , i.e. 1.2 g (accelerating faster than free fall) followed by 2.4 g jolt upward. This is
why civil aircraft keep clear of big Cb! 22 Sub-cloud layer area reduces to 11220 in 30 min
(vertically stretched to same extent). Some radioactive aerosol in converged and uplifted air
was rained and washed to surface in rain so contaminated that public health warnings were
broadcast.

Chapters 11 and 12

2 Well to west of old low centre where open waves form and move quickly. 3 Taken aback
(i.e. blown backwards) by near headwinds against sails still set NW -SE. 9 32 em. 10 About
400 km on simplest straight-line geometry relative to the moving system. Path curvature
484 could reduce this somewhat, but eastward system motion could add considerably to
geographical distance of source. 11 6.8 oc and 8.9"7o RH. 12 2 x I0- 6 s -I. 13 1.6 m s -I Notes on selected problems
slower eastward than the airflow in mid troposphere (e.g. if latter is 20 m s -I E'wards then
cyclone wave moves eastwards at 18.4 m s- 1.) Assume broad airstream and no divergence or
convergence. 14 Use method of appendix 12.7 to find 7.6 mbar for average T 300 K and
ambientp5 1000 mbar. 15 192 km. 16 If use dry adiabatic flow, then cloudy uplifts under-
estimated. Model in Fig. 11.15 does not describe intense activity in low troposphere near
surface cold front. 19 About 0.45 mm s -I as mean temperature falls.

Chapter 13

4 In steady state, inward heat leakage balances outward heat pumped. Additional heat
generated by motor and inefficient heat pump is net gain. 5 Do not use technical jargon. 6
No dynamic behaviour. 8 Heat source and sink temperatures quite close, much friction and
irreversibility. 9 0.5%. 10 8.4%. 11140 m s- 1• About 9797 J per kg of raindrop has to be
converted into heat by friction and shared with ambient air. 12 0.98 °C. 2.87 MJ m- 2 into
PE, 7.13MJm- 2 intoiE.13UseFW= 7.5 X w- 3 v~o(appendix 13.4)tofind0.1,2.3, 51,
262 W m- 2 • 14 6.93 MJ m- 2 • 15 (a) No zonal and eddy categories, but horizontal and
vertical possible instead. Some A from A. - convection in fronts. Turbulent loss
dominates. 17 POW (upwards) slightly exceeds d(gz)/dt (buoyancy), giving relatively small
rate of increase of KE (w 2 /2). Friction nearly completely offsets buoyancy giving much
smaller KE increase.

485
Appendix

Section equivalents between Fundamentals of Weather and Climate (FWC) and Basic
Meteorology (BM). All details are omitted where numbering is identical (though titles may
differ), and new sections in FWC are listed without BM equivalence.

FWC BM
Section 1.4,1.5 1.4
1.6 1.5
1.7
Section 3.1 3.1, 3.2, 3.3
3.2, 3.3 3.4
3.4, 3.5 3.5
3.8
Appendix 3.1
Section 4.9 4.9
4.10
Appendix 4.3
4.4 4.3
4.5 4.4
Appendix 5.2
5.3 5.2
5.4 5.3
Section 7.2 7.2, 7.3, 7.4
7.3 7.5
7.4 7.6
7.5 7.7
7.6 7.9
7.7 7.10
7.8 7.11
7.9 7.12
7.10 7.13
7.11 7.14
7.12 7.15
7.13 7.16
7.14 7.17
486 Appendix 7.5 Section 7.8
 7.6 Appendix 7.5 Appendix
7.7 7.6
7.8 7.7
7.9 7.8
7.10 7.9
7.11 7.10
7.12 7.11
7.13 7.12
7.14 7.13
Section 8.9
8.10 9.1
8.11 9.2
8.12 9.3
8.13 9.4
8.14 9.5
Appendix 8.3 9.1
8.4 9.2
8.5
Chapter 9, Sections and Appendices Chapter 10, Sections and
Appendices
except
Section 9. 12 10.12, 10.13
Chapter 10, Sections and Appendices Chapter 11, Sections and
Appendices
Section 11.1 12.1
11.2 12.2
11.3 12.3
11.4 12.4
11.5 12.5
11.6 12.6
11.7 12.7
11.8
Appendix 11.1 12.1
11.2 12.3
11.3 12.4
Section 12.1 12.5
12.2 12.8
12.3 12.9
12.4 12.10
12.5
Appendix 12.1 12.2
12.2 12.5
12.3 12.6
12.4 12.7

487
Index

Acceleration 181, 186-8 in hurricanes 428-9, 436

Accretion 163 large scale eddy flux of 222
Acid rain 60 Anisotropy 3
Acoustic echo sounding 60 Anticyclone 61, 91, 93, 204, 215, 392-5,
Activation see Condensation nuclei 412-17
Adiabatic blocking 93, 393-4
expansion 75, 116-25 continental 394
reference process 116 gradient balance in 236
Advection 261, 264 subtropical 91, 412-17
Advective heat fluxes 264, 269, 280, 281 Anticyclonic gloom 393
meridional profiles 269, 279-80 Anvil 165
Aerobic photosynthesis 68 Aphelion 96
Aerodynamically rough 301 Archimedes' principle 174, 210, 237
Aerodynamic roughness length 309 Arctic sea smoke 316
Aerosol 47, 61, 150, 151, 153 Argon 51
fallspeed 149, 175 Atmospheric
Aggregation see Accretion boundary layer see Boundary layers
Air heat engine 9, 10, 440-56
density 74, 81-2 efficiency 453-5, 461, 462
dry 47-9 stratification 3
masses 83, 373 waves 358-64, 397-400
moist 47 window 253
parcel 8, 73, 75, 76, 110 Axes, meterological 181
temperature 16, 17 Azimuth 20
Aircraft Azores High 412
bumpy flight 341, 361-2
icing 168
Aitken, J. 150 Backing wind 313, 381
nuclei see Condensation nuclei Baroclinic 86, 91, 201
Albedo instability 396
planetary 247 waves 395-6
surface 259, 292-3 Barograph 23
Altocumulus 41-2 Barometers 21-3
Altostratus 42, 382 Barotropic 86
Anabatic wind 315 tropical atmosphere 421
Anemometer 20, 21, 40 Beaufort
lag 40 Sir Francis 40
Angular momentum 90, 220 letters 40
conservation of 220-2 scale 20, 41
488
Beer's law 278 hooked 381 Index
Bergeron, T. 162 CISK 431
Bergeron-Findeisen process 163, 167, 362 Clausius-Clapeyron equation 172
Bermuda High see Azores High Climate 68, 82-100, 253-75
Billow clouds 364 global models 11
Blackbody 245 middle latitude 404-8
radiation 245, 248, 275-7 continental 407-8
Blocking anticyclone see Anticyclone, western margins 406-7
blocking low-latitude 432-5
Blue-green algae 68 North American 407
Boundary layers 286-7 North Mediterranean 407
atmospheric 286 Climate change (replacing climatic varia-
laminar 286, 299-301 tion) 5, 67, 94-7, 253-5, 434
nocturnal 360 anthropogenic 97-100
planetary 3, 286 Climatic
surface 287, 306-13 singularities 406
turbulent 287 variation see Climate change
Bowen ratio 268, 313 Climatology 5
British Meterological Office 31 Cloud 140-56
Brunt-Vaisala frequency 359, 360, 369 accessory 42
Buoyancy 128-9,211,237,344-5,364-6 base 145-6, 332
force 210, 369 billow see Billow cloud
Buys-Ballot's law 196, 201 chamber 148, 163
Butterfly effect 12 classification 40-2
environment 342-5
formation 65, 109, 112, 141-50
Campbell-Stokes sunshine recorder 27 frontal 274
Cambrian 68,69 genera 40
Capillatus 42 initial development 154-5
Carbon dioxide 57, 249 lenticular see Lenticular cloud
growth 58-9, 67-8 mixed 162
residence time 57 mother see Mother cloud
seasonal oscillation 57 orographic see Orographic cloud
Carboniferous 68, 69 seeding 167
Cause and effect 8, 9 species 42
Celerity 360 supplementary features 42
Cellular convection varieties 42
closed 355 Cold core 378, 423
open 355 Cold water sphere 273
Centrifugal Cold waves 407
force 187 Collision and Coalescence 161, 163, 176,
pressure gradient 230-1 177
Centripetal acceleration 186-7 Component
CFCs 55, 56, 254 forces and accelerations 181-3
Chaos 10-13 Condensation 141, 172 see also Cloud
Charney, J. 395
formation
Chernobyl 371
nuclei 61, 148, 154-5
Chinook 320
Chlorine 56 activation 154
Chlorofluorocarbons see CFC Aitken 61, 154-5
Circular motion 106-7, 186-7 giant 62, 154-5
Circulation large 61, 154-5
general see General Circulation Conditional instability 129
thermally direct 449 of the second kind see CISK
Circumpolar vortex 87, 201, 207 Congestus 42
Cirrocumulus 42 Conservation of energy 110
Cirrostratus 42, 381 Constant flux layer 306
Cirrus 42, 162 Continental anticyclone see Anticyclone 489
Index Continuity Decay, biological 52-4
equation 213-6 Deformation 212-9, 239
of mass 238 Dendrities 163
Convection 211 see also Cumulus, Denitrification 52
Cumulonimbus Depression see Extratropical cyclone
cellular see Cellular convection Determinism 10-11
cumuliform 270, 274 Laplacian 10
development 130-3 Detrainment
mechanical 131 terminal 333, 344
penetrative 355 Dew point 20
slope see Slope convection depression 20
thermal 131 in anticyclones 414
vertical 27 4 lapse rate 171
Convective meter 20, 115, 146
equilibrium 48-51, 130-4 profile 118, 145-7, 172
heat flux 266-9, 270-5, 280-3, 3ll Diabatic heat input 443-5
instability 128, 129, 130, 131, 132, 339, Diffuse reflection 255
347 Diffusion 48, 160, 173
near-neutral stability 130-4 equation 173
neutral stablitiy 129, 133 Diffusive equilibrium 47-51
overshoot 365-7 Dimensionless numbers 194
stability 127-31, 368-9, 415 Dimension 226-9
Convergence 218 Dishpan 206
Conveyor belts 390-2 Diurnal tide 206
Corio lis Divergence 216, 239
acceleration 187 anticyclone 394
effect 188-91, 231-3 Downdraughts 350-2
force 199, 203 Drag 211
parameter 195 coefficient 309
variation of 409 Droplet
Corona discharge 165 fallspeed see Terminal velocity
Covariance 305 growth 153-5, 161-4, 172-4, 176-7
Critical droplet radius 153, 156 in equilibrium 154
Cumulonimbus 41, 42, 121, 126, 164, 209, Dry adiabat 117-24, 126, 129, 146
267,274,337-42,344 Dry adiabatic
anvil 164 lapse rate 121-2, 127, 130-2
precipitation 337-8, 340, 350-2 reference process 116-19
temperature excess in 121 temperature profile 131, 145
tops 339, 367 Dust devils 208-10
Cumulus 5, 42, 121,211, 267,274, 331-4 Dynamic similarity 195
congestus 335, 338, 343-5
development 335-7 Eady, E. T. 395-6
humilis 42, 331-2, 343 Easterlies 87-90, 421-3
streets 289, 353 Easterly waves 421-3
systems 353-9 Eddy 301
temperature excess in 121 available potential energy 448-52
Cyclogenesis 375 flux 304-7, 325-7
zones of 375 stress 186, 305-7
Cyclone 199, 423 viscosity 184
extratropical see Extratropical cyclone coefficient of 184, 311
tropical see Tropical cyclone Ekman, 287
Cyclostrophic balance 208-10 spiral 313
in hurricanes 429-31 Emissivity 293
in tornadoes 208-10 Emittance 245
'thermal wind' 436-8 Energy see also Kinetic, Mechanical and
Potential energies
Damping depth 296 budget
490 Dart leader 165 atmospheric 256-9, 450-5
 dissipation 445, 451-3, 457, 460-1 Index
global 246-8, 255
surface 256-8 drag 205
conservation of 110 forces 184
conversion terms 443, 446, 448-52 heating 461
equation 443-5,456-7 Front 90, 201, 270
forms 442-8 cold 376, 384
human production of 461 occluded 378
internal 443-4 polar 375
paths 448-53 warm 376, 380
total 444 Frontal
total potential 444-8 precipitation 355-8, 381-5, 389-91
ENSO 434 slope 202
Entrainment 211, 332 zones 376
drag 212 Frost hollows 315
Entropy 118, 136 Froude number 210
Environmental viewpoint 127
Equations Gaia 70
of motion 8, 181-2, 190-1, 231-2 Gas constant
of state 73, 101, 110-13, 117, 133, 171 specific 73, 102, 110, 112
Equilibrium vapour 141, 150 universal 73, 101
Equinoxes 259 General circulation 86-9
Evaporation 141 Geopotential height 79, 103-4
cooling 364-6 Geostrophic
of falling drops 17 balance 192
Exponential decay 104-6 departures 203-7
Extratropical cyclone 5, 90, 204, 215, 218, equilibrium 197, 201, 206
225,261,265,270,273,374-92,412 flow 195, 197, 215, 233-4
development of 375 wind 199
families 374-5 shear 200
mature 380-6 speed 196
weather in 385 Geothermal flux 247
Norwegian model 373, 380 Glaciation 95, 97
cloudy 162, 338
Fahrenheit 17 Global warming 97-100, 255
Fair weather electric field 166 Gradient
Ferrel cell 89 level 313
Fick's law 172 wind 204
Fitz Roy, Admiral 373 Graupel 159, 163, 340
Fog Gravitational
advection 322-4 acceleration 21, 183
frontal 282 apparent 183
hill 321 force 183
radiation 321 potential energy 443
steam 315 Gravity waves 358, 360
Fohn 319 Greenhouse
Forecasting 5 effect 253-6
limits 12 gases 56, 254
Greenwich mean time 30
Fossil fuels 53, 59
Greybody 277
Fractocumulus 333, 382 Gulf of Mexico 347
Fractus 42 Gulf Stream 272, 286
Franklin, B. 164, 372
Fraunhofer lines 244 Hadley
Free atmosphere 3, 195, 287 cell 88, 91
Freezing nuclei 162, 353 circulation 206, 220, 270, 273, 344, 358,
Friction 432,451
velocity 299 Hail 313
Frictional giant 350
convergence 218 Half life 105 491
Index Har 320 Ionosphere 51
Haze 62, 155 Irradiance 245
Heat surface 277, 289-90
conduction 265, 295-8, 300 Irreversibility 461-2
engine 9 Isentropic 136
latent Ill analysis 389
lows 402-4, 410, 417 Isobaric
pump 442, 453 contours 234-5
sensible Ill temperature change 111
sinks 9, 454-5 Isobars 3, 85, 120, 125, 196, 224, 234
sources 9, 454-5 lsopleths 3, 120, 146
Heterogenous nucleation 162 lsopycnals 85, 224
Hoar frost 289 Isotherms 3, 85, 120, 146
Homogeneous nucleation 151, 162
Hot towers 344, 419 Jet core 201
Humidity Jet stream 190, 383
measurement 18 easterly 419-21
mixing ratio 112 polar front 91, 197, 203, 383
profile 144 subtropical 91, 221
relative 18-20, 114, 145
specific 63, 112, 123, 125, 146, 171 Katabatic 315
saturation 146 Kelvin-Helmholtz instability 359
Humilis 42 Kinetic energy 441-53, 456-7
Hurricane 5, 38, 208, 423-32 destruction see Frictional dissipation
eye 424-5, 431 eddy 448-53
wall 425 zonal 448-53
formation 430-1 Kipling, R. 324
low pressures 431-2, 437 Kirchhoff's law 247, 249,263, 275
pressure gradient 425 Knot 20
recurvature 428 Kohler curve 153,179
warm core 430-1
winds 425, 428-9 Lag
Hydrologic cycle 63, 65 coefficient 40
Hydrosphere 63 instrumental 18-23, 39-40, 105
Hydrostatic Land breeze 317
equation 79-81, 102, 104, 235 Laplace 10
equilibrium 79, 102, 171, 192, 201, 264 Lapse rate (temperature) 75, 122
Hygrograph 18 dry adiabatic 121-2, 127, 130-2
Hygrometer 18 environmental 127-31
hair 18 parcel 131
skin 18 superadiabatic 109, 131
Hygroscopic 20, 150 Latent heat of vaporization 109, 112, 115,
nuclei 154, 161 122-4, 267, 306, 362
Lateral acceleration 203
Ice ages 68, 95-9 Lee trough 399
Ice nuclei 162, 167 Lenticular cloud 42, 360-4
Ideal gas 73, 101 Lenticularis 42
laws 73, 101-2 Lifting condensation level 126, 146, 332,
Index cycle 400 344
Inertial reference frame 186 Lightning 165
Insolation conductor 169
daily 324-6 fork 166
surface 324-5 humanrisk 171
Internal energy 110, 443-4 return stroke 165
column total 444-5 sheet 166
International cloud atlas 26, 40 Linear acceleration 203
Intertropical convergence zone 88, 206, Little ice age 94, 96, 98
492 220, 271 Log wind profile 307, 326
London, J. 255 solar observation 325 Index
Long waves 397-401 turbidity 294
Lorenz, E. 11 Optical depth 278
Lucretius 15, 164 Order of magnitude 1
Orographic cloud 319-20
Mammatus 339 Oxygen 52,67-9
Margules 447 depletion 52
Mass flux divergence 213-7 odd 55
Mechanical efficiency 454 residence time 52
Mechanical energy 9, 441-3, 455 Ozone 53, 249
Melting band 169, 382 hole 55, 56
Mercator projection 279 ultraviolet absorption by 53-6
Meridional 83
advection profiles 279 Paleoatmosphere 67
temperature profiles 84 Parcel viewpoint 127
thickness profiles 83 Partial derivative, differential 127, 182
winds 88 Partial pressure 102, 112
Mesopause 77 Past weather 26, 42, 44
Meterology 5 Pea souper 323
Methane 254 Perihelion 96
Microclimate 294-7, 320-4 Permafrost 407
Milankovitch mechanisms 96 Photodissociation 50, 53, 70
Mixing law 133 Photoionization 50
Mole 73, 101 Photons 263, 275-6
Momentum 181 Photosphere 244
angular see Angular momentum Photosynthesis 52, 57
vertical flux 299 Pileus cloud 42
Monsoon 89,404,417-21 Pitot tube 21
depressions 419 Planck 276
European 406 function 276
Indian summer 420-1,433 Planetary
burst of 420 albedo 247
trough 418 boundary layer 3, 8, 57, 184, 192-5,
Mother cloud 42 205,218,286
radiative equilibrium 246
waves 397-400
Nacreous cloud 66 Plasma 51
Newton Poisson's equation 117
viscosity law 185 Polar lows 400-2
cooling law 39 Potassium 51
second law of motion 8, 181, 186, 211, Potential energy 441-53
218 available 446-8, 459-60
Nimbostratrus 41-2 column total 444, 457-9
Nitrogen 51 see also Denitrification gravitational 443
fixation 51 Potential temperature 118, 123
residence time 52, 70 Precipitable water 63, 134
Noctilucent cloud 66 Precipitation 23-4,65, 140, 143, 157, 169,
Non-linearity 12 337-8
Normalization 228, 309 evaporation of 122
Normand's theorem 126-7, 146 measurement 23
Norwegian model see Extratropical cyclone mean seasonal 433
Number density 49, 54 mechanism 161-5
near hills 318-20
Observation 15-44 Present weather 26, 42-4
networks 26-39 Pressure
surface 30-3 atmospheric 79, 106
Oceanic correction to MSL 30
heat fluxes 272 gradient 197-8, 209, 234 493
Index Pressure cont. Reynolds
in hurricane 426 number 175, 194,228, 229, 303
force 182, 229 stresses 305, 313
horizontal distribution 83 Richardson, L.F. 301
mean surface 91 number 303, 359
measurement 21-3 Ridge
station 23 weather in 393, 417
tendency 22 Ring vortex 332
vertical pressure 77, 78, 105, 106, 191 Ross by
Princess Victoria, the 387 number 195, 199, 204, 206
Psychrometer 20 waves 397
aspirated 20 Rotor 361
Psychrometric equation 115 Run of wind 21, 40

Sahel 433
Quasi-geostrophic balance 193
Satellite 34
geosynchronous 36, 38
Radar 24, 31, 36, 169, 356 radiometry 36-8, 248
Radiation sun synchronous 36
cosmic 166 Saturated
diffuse 256 adiabat 120, 123, 126, 129, 344
direct 256 adiabatic reference process 116, 122,
electromagnetic 244-5 126, 129
solar 17, 26, 36, 62, 76, 131, 258 pseudo-adiabatic process 125
terrestrial 9, 17, 26, 36, 76, 130-2, Saturation 18, 114, 141-2, 150
247-52,258,292 vapour density 140
window 253 vapour pressure 140-1, 171
Radiative Scalar 8
dissipation 450-1 Scale
energy budget analysis 191
atmospheric 360 small 206-1
global 244-8 synoptic 191-4
surface 256-8 height 1
fluxes 259-61 binary 84, 102
diurnal variations 263-4 decadal 3, 77, 80, 105
meridional variations 259-61 exponential 80, 105, 228
seasonal variations 261-3 length
forcing 254-5, 281-2 horizontal 191
Radiatus 42 vertical 191
Radioactive decay 105 meso 330
Radiometer 36-7 see also Satellite micro 330
radiometry net 28 small 330
Radiosonde 33, 34, 36, 75, 117, 127, 197, space 7
367,390,414 time 7, 191
Rainfall 92 see also Frontal and Screen level 31
Cumulonimbus precipitation and Scud 318, 333
uplift 409 Sea breeze 316, 403
Rain gauges 23 front 316
exposure 24 Sea defences 100
tipping bucket 23 Sea level rise 98
Rain shadow 24, 318, 406-7 Sea storm waves 386, 428
Rainy season 419 Separation (of flow) 317
Raoult's law 151 Shaw, Sir N. 9
Rayleigh number 305, 356 Shear 184 see also Wind shear
scattering 256 stress 184, 230
Residence time 52, 70 waves 359-60
Respiration 52 shimmer 304
494 Reversible processes 118, 461 Simple harmonic motion 369
Simpson, G. 165 inversion 343, 415 Index
Slope Subtropical
convection 261, 270, 274, 389-92 cyclones 419
winds 314-6 deserts 88, 416
Smog 322 high pressures 88, 271, 392, 412, 432-4
London 322 Sudden warming 77
Los Angeles 416 Sulphur dioxide 59
Snow 157-60, 162 Sunspot minima 96
falls 401 Superadiabatic 128, 131
Solar Supercooling 157, 162, 167
constant 246, 248 Supergeostrophic 203
energy 8 Supersaturation 145, 148, 153, 160, 173
granulations 7 Surface
photosphere 244 climate 298
Solarimeter 27 thermal response 294-9
Solar radiation Synoptic
flux 245-7,255,261,268 network 28-35
annual average 268 surface 30-3
measurement 26-8, 36 upper air 33-5
spectrum 244-5 scale 28, 191, 199
Southern Oscillation 434 low 193-6
Specific heat capacity weather systems 372-432, 437
at constant pressure 110-2 stations 31-3
at constant volume 110
Specific mass 47-9, 73-4, 102 Teleconnections 435
Spectral Temperature 15-16
absorptivity 263, 275 absolute 150, 171
emissivity 263, 275 dew-point 20, 115, 121, 146, 384
radiative emittance 263, 275 equivalent potential 125-7, 136-7
Squall line 164, 357 global trend 98
Stagnation of flow 317 grass minimum 27
Static stability and instability see Convective horizontal distribution of 85
instability inversion 62, 76, 131, 311
Station pressure 23 lapse 252
Stefan-Boltzmann constant 248, 276 mean surface 405
Stefan's law 245, 276 measurement 16-19
Stepped leader 165 potential ll8, 123
Stevenson screen 17, 20, 31, 147 profile of 128, 130, 343
Stoke's regime (fallspeeds) 175 thermodynamic wet-bulb 115
Storm vertical profile 74-7, ll8, 121,125,128,
multicell 346, 348, 351 130, 253, 265, 312, 339, 342, 347,
severe local 164, 345-53 366-7,414,431
supercell 346, 349-52 virtual 91
surge 386, 428 wet-bulb ll5, 126
North Sea 386-88 potential 125-7, 137, 347
Stratocumulus 41, 333, 344, 354 Tephigram ll7-22, 146
Stratopause 77 skeleton 133-6
Stratosphere 3, 34, 53, 76-7, 85 Terminal velocity 150, 159, 161, 173-5
Antarctic 55, 56 Terrestrial radiation 9, 247,249, 251-2
Stratospheric heat engine 440, 442, 452-3 fluxes 255, 259, 261-2, 267
Stratus 41 measurements 26-8, 36, 249
Streamlines 204 Thermal
Subcloud layer 144-6 capacity 18
Subgeostrophic 204 conductivity 265
Sublimation 141 diffusivity 295
nuclei 162 wind 199-202,235
Subsidence 88, 271, 412-7 equation 200-2
drying 409-10, 435 relation 200 495
Index Thermals 314, 353 flux 172, 311-13
cloudy 336 measures 18-20, 112-16
Thermocline 295 pressure 20, 115, 142, 171
permanent 273 specific gas constant 112
Thermodynamics 109-37, 443, 453-5, Vector 8
461-2 Veering wind 380
first law 110, 123, 133, 440 Vegetation 320-2
second law 453-54 Venus greenhouse effect 68, 253-6
Thermograph 17 Verkhoyansk 316
Thermometers 15, 17 Vertical motion see also Downdraughts,
minimum 17 Subsidence, Updraughts
response time 39-40 small scale 209-13
wet-bulb 20, 114, 122, 126 synoptic scale 191
Thermosphere 50 Virga 42-3, 159, 337
Thomson effect 150 Viscosity 184
Thunder 166, 340 dynamic coefficient of 184
storms 164, 340-1, 351 eddy 184, 194, 356
Thundery lows 404 coefficient of 184
Topography and air flow 317 vertical profile of 356
Topside observation 36 kinematic coefficient of 184
Tornado 208,226,345,350-2 Viscous
forecasting 28, 351 drag 184
Torque 218 force 230
Torricelli 21 Von Karman constant 307
Trade winds 87,206, 271, 333, 432-4 Vorticity 223
Trajectory 204, 387 absolute 224
Transmissivity 278 anticyclonic 225
Tropical cyclonic 225
cloud clusters 421 geostrophic 225-7
cyclones see Hurricanes mesoscale 350
easterlies 87-90, 421-3 negative relative 224
easterly jet stream 419 planetary 224
storm 423 relative 224
troposphere 420 synoptic scale 350
Tropopause 3,66, 76,85,215
Troposphere 3, 34, 47, 57, 75-6, 85,
88-90,162,191,214,420 Wadhis 416
Tropospheric Walker circulation 434
circulation 86-90 Warm
heat engine 9, 440-56 core 423,437,449
radiative absorption 250, 255, 258 front 376, 380-2
Turbopause 50-1 sector 376, 380, 384-5
Turbosphere 50-1 , 77, 80 Warm water sphere 273
Turbulence 184, 193, 301-7 Water
clear air 362 precipitable 63, 134
Turbulent sunlight in 325
friction 194, 205, 207, 211 turbid 294
momentum transport 304, 325-7 substance 62-3
residence times 63, 70
Ultraviolet radiation 53-7, 258 vapour see Vapour
Updraughts 21, 332, 335, 339, 341, 350 Wave(s)
Uplift 389-91, 428 clouds 42, 360-4
hill and mountain 317-20
Vacillation 399 in stormy seas 386-8
Vane 20 in the westerlies 395-401
Vapour (water) trains 358-65
density 18, 112 115, 172,312 Weather and climate 90
496 equilibrium 150-3 Westerlies 86-90, 202, 378, 383, 395-400
Wet-bulb depression 19 vane 20 Index
temperature 114-15 response time 40
Wien's law 246, 249, 277 Wordsworth 148
Wilson, C. T. R. 165 World Meterological Organization (WMO)
Wind see also Easterlies, Westerlies 28, 38
direction 20-21, 31, 86-90 World Weather Watch (WWW) 28-38
measurement 20-1 , 31 , 40
meridional 88-90 Zenith transmissivity 279
power 406 Zero plane displacement 310
shears 184, 193,201,304, 334, 348 Zonal 83
sondes 34 available potential energy 448-53
speed 20, 31, 40 index 400
strength 40-1 winds 87, 89-90

497