Goodall, 1995 - Apend 2
Goodall, 1995 - Apend 2
Goodall, 1995 - Apend 2
Appendix 2
by
(University of Aberdeen)
O Ni,
Late Quaternary Arabian desert models of
Permian Rotliegend reservoirs
by K.W. Glennie, J.M.Pugh & T. M.Goodall (Aberdeen University)
Introduction
It is common practice to study patterns of modern sedimentation so that the internal
geometry of hydrocarbon revervoirs in ancient rocks can be better understood. The writers
studied the desert areas of the United Arab Emirates and Oman with the aim, through a better
understanding of their origins, of improving the modelling of hydrocarbon reservoirs in the
Permian Rotliegend of NW Europe.
The senior author”s first involvement with present-day deserts (Glennie, 1970) was a
direct outcome of the realisation in 1963 of the size of the giant Groningen gas field, where
the reservoir was thought to represent an ancient desert sequence. Because of the difficulty
of studying reservoir geometry in unconsolidated modern sands, Weber (1987) studied
outcrops of Permian dune sand in Canyon De Chelly (Ariz., USA) as a guide for developing
the Leman gas field in UK waters.
Three or four decades ago, deserts were considered to be entirely 'modern” in terms of
their activity. Itis now realised that James Hutton's dictum “the present is the key to the past”
does not quite apply to the Emirates desert. For a proper understanding of this desert area,
it has to be realised that many of its sedimentary features (e.g. dunes, wadi sediments) are
mostly products of events long past in terms of humsn history, and are only mildly active
today. Indeed, some of these features are currently more subject to the effects of deflation
than to deposition.
Both units described in this article, the Quaternary sequences of SE Arabia and the
Permian Rotliegend, consist essentially of desert sediments (Glennie, 1972). They also have
one other important feature in common. To a large extent, sedimentation was controlled by
their location relative to the Earth's poles and Equator, and the presence or absence of high-
latitude glaciations centred in North America and Scandinavia during the Quaternary and in
Gondwana during the Permian. These glaciations had a profound effect, not only on the
climate of their times but also on sea-level changes.
during earlier glaciations. The last glaciation took a long time (about 70 Ka) to reach its
maximum extent around 18 Ka BP, but then collapsed rapidly in about 10 Ka. At the height
of the last glaciation, the Arabian Gulf was dry and the joint Tigris-Euphrates River flowed
into the Indian Ocean south of the Strait of Hormuz (e.g. Sarnthein, 1972).
The large areas of very high atmospheric pressure associated with each of the ice caps had
the effect of squeezing all other air-pressure zones towards the eguator, resulting in an
increase in global wind-velocity (Figure 3). In desert areas, the wind probably blew at sand-
Oxygen isotope global sea-level history (after Shackleton 1987)
using “0 data from Planktonic and Bethonic fauna
-200-h
INTERGLACIAL
Í
LOW VALUES
INTERGLACIALS |,
AJ | [DA Lota [|
o.
8' 0 ABOUT ZERO MEAN
HIGH VALUES
GLACIALS
GLACIAL
| | | Dr. No
0 50 100 150 200 250 300 350 400 450 500 550
Modified trom Boutton, 1993, Fig 21.3 AGE Ka BP
Figure 1. Variation in oxygen-isotope composition of oceanic foruminifera over the past 575 Ka. Their
composition reflects the amount of land ice on Earth. The oceans are isotopically heavy during glaciations
and light during interglacials. The record indicutes six major glacial eycles with intervening temperate
interglacials. The glacial-interglacial boundary roughly matches that used in Figure 2. Modified from
Boulton, 1993.
Oxygen isotope global sea-level history (after Shackleton 1987)
using "0 data from Planktonic and Bethonic fauna
0 sm Aos
Sealevel in metres below present sea level
Figure 2. Oxygen-isotope sea-level history for the past 140 Ka using HO data from planktonic and
benthonic fauna (simplified from Shackleton, 1987). There is some evidence to suggest heavier rainfall in
SE Arabia prior to about 70 Ka, which possibly indicates interglacial rather than glacial conditions. A
tentative oversimplified boundary benween these two climatic states is therefore taken ai -50m.
Exploration Bulletin 274 (1994/3) Page 3
SHAMAL
(WINTER)
Figure 4. Map showing the present distribution of sand dunes (aeolianite) over Arabia, the occurrence of
carbonate-rich dune sand (miliolite) derived from earlier exposed sea floor, and contours showing
tentative time spans in thousands of years during which the Arabian Gulf was subaerially exposed during
past 120 Ka (based on the sea-level curves of Figure 2 and bathymetric map of Kassler, 1973).
a
ii
>
NYNO
SNIVINNOIN
Figure 5. Landsai false-colour Themaric Mapper mosaic of the Emirates. From the east coast of Qatar to
the west coast of the Musandam Peninsula is a little over 500 km. Sabkha Masti is at the west margin of the
dune area. Al Liwa is centre bottom. Note the interference pattern of dune uxes east of Abu Dhabi. A Abu
Ditubi, D Dubai, H Jebel Hajir, M Sabkhat Matri, MU Musandam Peninsula, O Qatar.
Exploration Bulletin 274 (1994/3) Page 5
wind. Landsat imagery indicates a probable linear origin for the large dunes of the
northern and eastern Emirates under the influence of strong winds that were forced by
the Oman Mountains to splay out in a range of orientations from WSW to WNW (Figure
5). The upwind (western) end of the linear dunes east of Abu Dhabi island seem to have
originated from the adjacent coastal sabkhas or even the Gulf itself. This observation
emphasises the probability that their sands were derived from across the southern Gulf
when its floor was exposed during the last glaciation. Since then, the prevailing wind
direction has changed. Over the northeastern Emirates, the large linear dunes have been
partly eroded and their sands redeposited in interdune areas by winds that blew from the
west. To the south, these former linear dunes have developed south-facing slip faces
under the new wind regime. This change to a transverse form implies that the modern
winds are weaker than their glacial-age predecessors. Further west, however, Landsat
imagery shows that about 50 km north of the Al Liwa dunes (Figure 5) the surfaces of
NW-SE trending large linear dunes are being eroded into much smaller linear forms
trending NNW-SSE (parallel to the main road; see also Figure 8), indicating a shift in
wind direction of about 45º. To their south, this complex of linear dunes grades into
small transverse forms, which encroach on the northern edge of Al Liwa.
When the Gulf was dry at the height of the last glaciation, the water table beneath the
Emirates was probably well below the land surface. Sand that originated in Syria, Jordan, Iraq
or northemn Saudi Arabia migrated down the Gulf, across the Emirates and into the Rub al
Khali. Progressive late glacial flooding of the Gulf stopped the supply of wind-blown sand
to the Emirates. Althoughits direction changed, the wind did not stop blowing, and especially
in coastal areas, deflation removed all sand down to the level of the water table. Those low-
relief, coastal areas that were flooded by the post-glacial sea developed into the classic
coastal sabkhas of Abu Dhabi while further onshore, inland sabkhas occupy low-lying areas
such as the extensive Sabkhat Matti, which storm-driven waves could not reach. Along the
steeper eastern coasts of the Emirates, the development of long-shore bars resulted in lagoons
and sandbanks capped by coastal sabkhas.
MEDITERREAN
SEA THE ARABIAN PENINSULA
O 100 200 300 400 500
KILOMETRES
Sabkhat
Matt U.8.E.P
Exploration Bulletin 274 (1994/3) Page 7
sand dunes migrating to the southeast from the Saudi-Qatar border area. Since the late glacial
flooding of the Gulf, however, Sabkhat Matti has been the site of deflation of the older fluvial
sands and gravels down to the level of the water table. These sediments are now gypsum
cemented at a height slightly above the level of the present water table. Sands and silts are
now being deposited in the wetter (former interdune?) areas of sabkha. With the limit of
deflation controlled by the water table, Sabkhat Matti forms a very extensive 'Stokes surface”
(Stokes, 1968; Fryberger et al., 1988).
Further downwind, dune sands are developed over increasingly widespread local highs
inthe southern part of the sabkha or along its southeastern margin. In the latter area, the dunes
generally follow a pattern of development that starts as a featureless sandsheet, only a few
metres thick, which build up by the southward migration of aeolian ripples. The sandsheets
are fairly well vegetated, which results in the creation of nebkha (sand accumulating within
and behind small bushes). Downwind, the sandsheets assume the form of a rolling topogra-
phy studded with zibars (slipfaceless dome-shaped dunes), possibly the result of earlier
deflation of small barchans. Further on, this changes to barchanoid and transverse ridges, to
oblique (asymmetric) barchans and finally to linear dunes.
Al Liwa.
In conjunction with the rising post-glacial sea level in the Gulf, the water table also rose
tointersect interdune areas in Al Liwa resulting in sabkhas whose surfaces, within the borders
of the Emirates, are some 80-100 m above sea level.
The Al Liwa comprises an area of large (up to170 m high), virtually stationary, reddened
transverse dunes (draa) with SSE-facing slipfaces and intervening sabkhas (Figure 8).
Because of deflation of the coastal areas, the dunes at the northern edge of the Liwa are clearly
being encroached from the north by smaller (10-20 m high), active, finer-grained transverse
dunes; their precursors are now climbing over the windward slopes of the draa to create
higher and much more complex dune forms.
The surfaces of the interdune sabkhas of the Liwa are covered with a thin crust of halite,
and a thin gypsum layer is commonly present about 50 cm below the surface. The sands
beneath the gypsum layer consist of SSE-facing aeolian slipfaces identical to those on the
southem slopes of the adjacent draas. It seems likely, therefore, that there has been no major
change in direction between the winds immediately before and subsequent to the creation of
these interdune sabkhas. Today, the local presence of patches of sand dunes (1-5 m high),
many partly stabilised by halophytic vegetation, indicates a rough balance in the interdune
areas between sedimentation and deflation.
Reservoir Geometry.
If the desert sediments of the Emirates are to be of any use as an analogue for the
Rotliegend reservoirs of NW Europe, it is essential that their bedding geometry is known in
greater detail than can be deduced from surface morphology.
K is notoriously difficult to study the sub-surface bedding attitudes of unconsolidated
dune sand; as soon as a pit is dug, the dry sand collapses. The sands are given reasonable
coherence if they are dampened, but in desert areas the supply of water is generally limited.
Inthe Autumn of 1992, an experimental survey using Ground-Penetrating Radar was
Page 8 Exploration Bulletin 274 (1994/3)
TRES
o € AREA
ga Oui a ed REM MES TA
a SAR
PRN e EX N
N É ye
A
t) 20) a( + “ fa E E pesar
[a E ju E ur»
EA Sm amp”. FO”,
Figure 8. Landsat false-colour Thematic Mapper image ofthe N & W Liwa. The transverse
nature of these
dunes is clear. Moisture associated with the interdune sabkhas shows up in blue. Note
light-coloured dunes
encroaching from the NW. Across the line of the main road from the north, the surface
of the older, large
NW-SE longitudinal dunes is toduy being eroded and replaced by a cover of much smaller
NNW-SSE
trending dunes. A change in wind direction of some 45 has taken place between the
mo dune-forming
events.
undertaken at the NE margin of Al Liwa. Up to that time, no such survey of modern dune
sands had ever been published. The survey was carried out with the aid of Dr Charlie Bristow
of Birkbeck College, London, who had previously used the technique on the Cretaceous
Greensand of England.
The depth of penetration and degree of resolution obtained from the equipment depends
upon the operating frequency of the radar. A high-frequency transducer (e.g. 1000 MHz) will
give a low depth of penetration and high resolution, while with a low frequency (e.g. 70 MHz)
the depth of penetration will be much greater but the resolution poorer.
To carry out the survey, a SIR 10 (Subsurface Interface Radar) system was fitted with a
300 MHz general purpose transducer to give medium resolution and depth of penetration.
Survey Procedure
For the purposes of the survey, its site had to be chosen with care. Not only must it be
possible to obtain records of the subsurface (experimental lines were taken over three sites)
but, as an independent check, it was planned to dampen the dune sand and cut a trench through
it; the site had to be accessible, therefore, to a water bowser.
Once selected, a detailed topographic survey was carried out to record the surface
geometry of the dune. A 90 by 90 metre grid was marked out with lines parallel (340º-160º)
and normal (070º-250º) to the present-day prevailing wind (Figures 9a, b). The lines were
spaced at 10-m intervals with a 5-m spacing in areas of greatest interest. A total of 24 GPR
lines were surveyed to give 2160 metres of radar coverage. Surveying and GPR recording
took two days.
Once the GPR survey had been completed, a line parallel to the prevailing wind was
chosen for excavation (Figure 9c). The dune here was dampened with approximately 25,000
gallons (over 110,000 litres) of water and left to soak overnight. The next day a trench was
dug through the now coherent sand using a 'D8” bu lldozer. The trench faces were cleaned
with trowels and the exposed internal structure of the dune was compared with the results
obtained by radar.
"CATCH ROAD
Figure 9a. Low-relief dune in the NE Al Liwa over which GPR survey was undertaken. Approximate
location of lines 7.5 and G are shown. Prevailing wing from left to right.
JPIUOZIOY D[X INOQGD 9JDIS [DIJU9A “falas 4OÍ Polo24409UF)
Exploration Bulletin 274 (1994/3)
“DMIT IN UI 2UNP D JO SIS DU NO (1) SU) PDUMOU-PUIM PUD (ǺZ QUI) JoJD4DA-PuiM “Q6 AANBLA
am
coa tem E TO ag SS RSS O adia ER ——
T Fa nem
bes” RE aj e age VR “e ai uai “Ez; A seio eres . Acomto RO tda end FR]
O AA — sê: . nt ' . Talita
) E qe ale doll
al pm 3
BETA O A il RU O reartes
A Pr
EO 0d atear LeoaDE cdrr
Mi PPS a ig
e E
PTD e
PONCICIA A a ous
K
EA E oa o»
PÉ O ia,
+ p
aeee RR
pasa EE
EE a na
AE
ms mi dE
ga
RARE ES
gs
U , a io a mae tra a
As no." , no =
Page 10
Exploration Bulletin 274 (1994/3) Page 11
Figure 9c. Dampened sand exposed in a trench beneath wind-parallel line of Figure 9a. confirms that the
GPR record is that of dune bedding prograding down-slope.
Survey Results
Numerous sub-surface reflectors were located by the radar along each of the survey lines.
Most noticible is the difference between the wind-parallel and normal survey lines.
The parallel lines comprise strong, large-scale, gently dipping reflectors that truncate
weaker, small-scale, steeply dipping surfaces. Some of the large-scale features project
beyond the base of the survey, while small-scale reflectors generally become more diffuse
in character with depth (Figure 9b).
The normal lines comprise east-west trending troughs with an offset stacking pattern that
results in truncation. The troughs become progressively wider and deeper with depth and can
extend beyond the base of the survey. Again, the reflectors generally become more diffuse
with depth. It should be noted that none of the profiles have been depth corrected, and thus
have an over-steepened appearance.
Interpretation
In the wind-parallel example, the steeply dipping reflectors have been interpreted as third-
order growth surfaces, and the gently dipping reflectors as second-order stacking surfaces
(Figure 9b). These correlate well with the stacked sequence of downward-climbing dunes
exposed in the trench (Figure 9c ). The diffuse nature of the reflectors at depth is caused by
the absorption of high-frequencies by saline ground water. The water greatly increases the
specific conductivity of the dune sand, resulting in signal attenuation.
Growth surfaces form as the the dune grows, responds to wind changes, and migrates.
Stacking surfaces result from the partial erosion of a dune, followed by stacking a second
dune on top of the first (Fryberger, 1993). Bounding surfaces act as the loci for the
accumulation of fine sediment, and it is this contrast between the well sorted foreset laminae
and the more poorly sorted sediment of the bounding surface that produces the radar
reflection. The variation in reflection strength between the two types of bounding surface
results from differing concentrations of fine material. The stacking surfaces are often the sites
of muddy or evaporitic interdune deposition (Fryberger, 1993) and give a stronger reflection.
In the wind-normal example the troughs have also been interpreted as second- and third-
order bounding surfaces (Figure 9b). The dunes in the survey area have an aklé dune pattern,
which consists of sinuous ridges transverse to the wind. At the base of each active slip-face
is a hollow bounded on either side by a small longitudinal ridge. As the dune migrates,
deposition within these hollows produces trough-stratified units normal to the wind
(Brookfield, 1977).
Exploration Bulletin 274 (1994/3)
"uonpAouad
“DpD4 Aodoop poruoadad 4opm auips
Page 12
Exploration Bulletin 274 (1994/3) Page 13
The downward-climbing nature of the stacking surfaces parallel to the wind results
from the superimposition of the aklé dunes on the back of a larger bedform. This large
dune, or draa, has no slipface, thus allowing the aklé dunes on its back to climb down its
leeward slope.
South of the survey site are the high dunes (draa) of the Liwa proper. These draas are
also transverse in nature, and have aklé-style dunes climbing up their windward and part-
way down their lceward slopes. The aklé dunes terminate, however, at a 40-50 m high
slipface that extends down to the interdune floor. The slipfaceless draa where the survey
was conducted and the draas of the Liwa proper, 2-3 km to the south, have similar width/
length ratios, which suggests similar origins. The slipfaceless draas in the north are
probably the product of dune degradation following the climatic optimum, when post-
glacial flooding etfectively cut off the northern supply of sand. Similarly, the draas south
of the survey area have also undergone degradation, but to a lesser degree, and hence
retain part of their original slipface.
This interpretation is supported by the contrasting colours of the sands on windward
slopes. The residual coarse sands (1-2 mm diam.) have a strong red coating and are more
plentiful on the lower slopes, in contrast to the finer-grained and more mobile yellowish sand
grains, which dominate the crests of the large draas (see also Besler, 1982).
Palaeowind directions are normally deduced from the attitudes of steeply dipping
slipfaces. The GPR survey has shown that slipfaceless dunes (zibar) can form by a
combination of deflation and down-wind migration. Another GPR survey just east of
Sabkhat Matti (not shown), reveals the internal structure of a zibar to consist of sigmoidal,
tabular, cross-bedded sets. The wind-ripple laminae mimic the leeward slope as the zibar
migrates downwind. Numerous truncation or bounding surfaces cross-cut bundles of wind-
ripple laminae, probably as a result of shifting winds.
A much more limited GPR survey was carried out in Sabkhat Matti, but no bulldozer-
assisted trenches could be dug. One line (Figures 10a, b), taken over the alluvial gravel plain
that flanks the western margin of Sabkhat Matti, clearly shows a number of fluvial channels
that cannot be discerned solely from their surface expression.
The experimental survey using Ground-Penetrating Radar proved that the technique gives
a geometric representation of bedding in both dune and fluvial sands down to the level of the
water table. By using a higher-frequency transducer and digital recording, it would be
possible to generate a much more sophisticated modem analogue of ancient reservoir sands.
Figure 11. The Southem and Northern Rotliegend basins of NW Europe. The desert lake
along the axis of
the Southern Basin is essennally land-locked. The limit of Zechstein flooding indicates
that the surface of
the late Rotliegend desert lake was well below global sea level. Fluvial transport was
mainly northward
uway from the Hercynian highlands.
Exploration Bulletin 274 (1994/3) Page 15
The salts and clays of the Rotliegend Silverpit Lake form a seal for gas trapped in underlying
Carboniferous reservoirs. The long-term role of the Arabian Gulf as both source and
transport area for aeolian sand has so far prevented the deposition of a sealing facies apart
from within the coastal sabkhas.
Glacially induced sea-level changes have probably had a much more profound effect
on the history of sedimentation in the Emirates than with the Rotliegend. It has recently
been claimed (Dudley et al., 1994) that the effects on aeolian sand supply of a
climatically shifting desert-lake shoreline can be recognised in both erg-margin and erg-
interior environments in the Rotliegend of the UK offshore. A similar, but perhaps more
striking control of sand supply has been suggested for the Permian aeolianites of the
United States, where acolian sandstones were periodically and rapidly truncated and
overlain by marine strata (Blakey and Middleton, 1983). The Jurassic aeolian sequences
of North America (see e.g. Blakey et al., 1988) have also been separated by repeated
marine incursions, although in this case, the changes in sea level are probably related to
sea-floor spreading rather than to glaciations.
Like the dune facies of the Rotliegend, its modern counterpart would also make an
excellent reservoir for hydrocarbons; indeed, parallel to the northern edge of the high
Liwa dunes the sands form the aquifer in what is reputed to be the largest oasis in the
o +00km world. The Rotliegend sands, however,
ARABIAN GULF
o were deposited on the flank of a basin that
was subsiding rapidly, and therefore had a
very high preservation potential. Mio-
Pliocene alluvial sediments in the Rub al
GULF OF OMAN
Khali Basin are reported to be up to 600
metres thick (McClure, 1978). The very
limited, if any, later subsidence in south-
ern Arabia seems to follow the axis of the
Rub al Khali desert, which passes through
the Emirates and can be discerned on
Figures 4 & 7. At the western end of the
) Rub al Khali, however, dune sands lie at
an elevation of some 1200 metres above
sea level.
Rotliegend wadi sands form fairly good
reservoirs where deposited as alluvial fans
Zamnsunses that were well above the water table (e.g.
DEFLATION PLAIN À A . Groningen and Roughfields; see Glennie,
1990). Quaternary wadi sands and gravels
Figure 12. Alluvial fans (red) flanking the Oman
Mountains. The dune sands (yellow) were emplaced, occur on both flanks of the Oman Moun-
seasonally, by either the Late Pleistocene winter tains (Figure 12). Those on the west flank
Shamal or, in the Wahiba, the summer Southwest have (rare) dates indicating the latest ma-
Monsoon. Sequences of fluvial sands and gravels jor deposition in the range of 40 Ka and 70
were exhumed in the areas marked “E”, probably Ka BP. Si
. Since then, and probably mostl
becuuse they were subjected to deflation rather than a p y y
sand deposition by both wind systems. M Masirah during the last glacial maximum, the bet-
Is. MP Musandam Peninsula, R Rams, U Umm al ter cemented fluvial channels have be-
Qawain, US Umm as Samim. come exhumed by deflation and now stand
Page 16 Exploration Bulletin 274 (1994/3)
A Abu Dhabi
AL Al Ain N
D Dubai
F it Faiyah +
H lebel Hafit
É so
J Jebel Dhanna
Arabian Gulf
M Mender
s Sila
SM Sabkha Motti
Figure 3a. Wind patterns deduced from the bedding attitudes of carbonate-cemented Late Pleistocene
dune sand; numbers refer to sample localities. The surface can be subdivided into several discrete aareas.
A Coastal Sabkha and Sabkhat Matti, B Deflated linear megadunes, C Dunes of type B parily swamped by
small transverse dunes, D a close-knit sand sea of small transverse dunes, E Mega barchanoid dunes and
interdune sabkhas of Al Liwa, F N-S trending small linear dunes east of Sabkhat Marti.
E |
Leal eae R SS
Es ig É
Conclusions
Quaternary desert sediments of SE Arabia provide a good analogue for the Permian Upper
Rotliegend sequences of NW Europe. In both cases, sedimentation took place in a Northern
Hemisphere Trade Wind desert that was strongly influenced by high-latitude glaciations.
The glacially-induced sea level fall that has been so important in controlling the supply of
aeolian sand in the Emirates did not affect the Rotliegend; instead, the Rotliegend desert lake
probably received occasional incursions of sea water during interglacial high-stands of sea
level. In both cases, sedimentation has been influenced by more than one glaciation.
Rising post-glacial sea level in the Gulf cut off the supply of dune sand to the Emirates.
The lack of a fresh supply of dune sand resulted in deflation down to the new associated
higher level of water table, and the creation of a major “Stokes” surface in Sabkhat Matti.
Comparable lake-margin stoppages in the supply of dune sand to the Rotliegend have only
recently been recognised away from the shore line. Changes in sea level also played a part
in alternating continental and marine sequences in the United States of both Permian and
Jurassic ages. In Al Liwa, the end-glacial high water table intersected the interdune surface
and led to the deposition of widespread but very thin evaporites in interdune sabkhas. Similar
sabkhas have been recognised in some Rotliegend lake-margin dune fields as thin field-wide
sequences of adhesion ripples.
An experimental Ground-Penetrating Radar survey in the NE Al Liwa proved that the
technique gives a geometric representation of bedding in dune sands down to the level of the
water table. By using a higher-frequency transducer and digital recording, a much more
sophisticated analogue of reservoir sands could be generated.
Acknowledgements
The management of Abu Dhabi Onshore Operating Co. (ADCO) gave the writers logistic support
for several field seasons; they also provided the surveyor, water bowser and bulldozer for the GPR
survey, and considerable LANDSAT coverage including Figure 5. Without that support, much less
would have been achieved and the GPR survey would not have been nearly so valuable. Goodall and
Pugh are in receipt of full funding from NAM and Shell Expro respectively and have studied
appropriate Rotliegend cores made available by them. Shell Expro funded the GPR survey at
relatively short notice, and many of the figures were drafted by Shell Expro. All are thanked for their
help and generosity.