Easier access to most parts of the Abu Dhabi coastline, combined with satellite imagery, has enabled a more detailed examination of the Holocene strata than was possible twenty or thirty years ago. Whilst the basic principles of coastal progradations documented in those early days are still valid, they portrayed an oversimplified picture of the Holocene coastal geomorphology and sabkha sedimentology. New data re-emphasises the importance of antecedant topography in imposing depositional complexity on the Holocene System. Former Holocene shorelines can now be mapped with greater clarity and reveal its highly embayed nature. The Holocene transgressive limits are locally redefined. Some palaeo-highs, that formed earlier Holocene peninsulas, have been completely removed by deflation. Two phases of sabkha anhydrite are recognised and their distributions are largely predictable by remote sensing.


The Abu Dhabi coastal sabkhas have provided a basis for sedimentologically interpreting many ancient carbonate-evaporite formations, such as the Arab Formation (Wood and Wolfe, 1969), since the highly illuminating research of the 1960’s (Curtis et al, 1963; Shearman, 1966; Kendall and Skipwith, 1969; Evans et al., 1969; Kinsman, 1969; Butler, 1970). Most early research concentrated on the mainland coastal sabkhas near Abu Dhabi Island (Figure 1) largely because of their accessibility during the early days of the oil boom when the Emirate was relatively undeveloped.

The progradational sabkha model, encapsulated in the schematic illustration of the famous ‘Evans line’ (Evans et al., 1969), has perhaps been the most influential in assisting geologists to interpret ancient coastal sabkha sequences. That the model which the early researchers outlined has withstood the test of time is a monument to their comprehensive descriptions of Abu Dhabi’s Holocene sabkha stratigraphy. Yet, in hindsight, their work was essentially only reconnaissance in nature (G. Evans, personal communication).

It is unfortunate that the coastal sabkhas to the northeast of Abu Dhabi Island where Butler (1970), Kinsman and Park (1976), Park (1977), McKenzie et al. (1980), Butler et al. (1982) and Patterson and Kinsman (1981, 1982) did much of their famous research are no longer available for study because the natural environment has been irrevocably disturbed by road and channel constructions, etc.

Undisturbed conditions prevail locally to the east and west of the Abu Dhabi urbanisation but civil engineering projects and oil field developments are progressing so rapidly that even the more remote coastal sabkhas are in danger of severe damage to their crucially important hydrogeological regimes. Fortunately, the ‘Evans line’ (which McKenzie et al. also used) is still locatable although industrial projects, pipeline constructions, etc. are beginning to impinge on it. For instance, the Tertiary high along the ‘Evans line’ has been completely excavated for use as industrial ballast.

Nevertheless, a study of Landsat TM and SPOT Panchromatic satellite images combined with generally easier access, deep trenching for pipe laying and channel constructions have enabled a fuller understanding of the Holocene sabkha sedimentology, stratigraphic controls and evolution of the area. Recourse to aerial photographs of the pre-urbanized Abu Dhabi has also enabled interpretational continuity along the mainland coast.


Transgressive and regressive stages of the Holocene sediments which were deposited above a Pleistocene aeolian substrate were identified along the ‘Evans line’ (Evans et al., 1969). It established lateral and vertical variations in sedimentary and diagenetic facies which have formed the basis of idealised sabkha sequences illustrated by various authors (Wood and Wolfe, 1969; McKenzie et al., 1980; Shinn, 1983). Most workers have therefore understandably accepted that the ‘Evans line’, depicting a simple transgressive-regressive cycle, was representative of most of the mainland coast although it is now apparent that there is considerable heterogeneity in the Holocene system.

Although Holocene storm beaches and spits have formed along most of the Abu Dhabi mainland coast (Figures 2 and 3), the storm beach of the ‘Evans line’ is part of a localised, crescentric collection of small beach ridges (Figure 4) that formed adjacent to an inlier of Miocene strata (G. Evans and P. Bush, personal communication) which protruded up through the Pleistocene and Holocene sediments of the coastal sabkha. Similar inliers with crescentric beach ridges or ‘winged spits’ exist elsewhere along the coast as discussed below.


Mesas of Miocene rocks presently form isolated headlands at intervals along much of western Abu Dhabi, between Jebel Barakah and Tarif (Figure 1). They are essentially northwest-southeast trending yardangs shaped largely through aeolian deflation by northwesterly winds (Shamals) that have operated with varying strengths at least since the Late Miocene. ‘Winged spits’ developed on the east and west flanks of these headlands and extended into the intervening embayments that were slowly being back-filled by sediment and evaporites. Calcite-cemented, Pleistocene, carbonate seif dunes similarly form headlands with ‘winged spits’ along the mainland coast between Abu Dhabi Island and Ras Ghanadha (Kirkham, in press).

Headlands do not currently exists along the mainland coast between Tarif and Abu Dhabi Island. Instead, satellite imagery reveals a coastline dominated by numerous parallel storm beaches that have stacked seawards like a series of cheniers (Figure 2). A closer examination of this storm beach trend, however, reveals ‘winged spit’ systems existed at intervals along their landward side (Figures 5 to 7). They are indicative of former highs (now completely deflated) which were presumably Miocene mesas or Pleistocene outcrops. Many isolated Miocene mesas still exist as inliers immediately behind the parallel storm beaches. Even here the coastline was once dominated by peninsulas and embayments that have been completely back-filled with sediment to form the relatively straight, present day coastline that coincides approximately with the northern edge of the parallel storm beaches.


With this basic understanding of the palaeo-geomorphological controls on mainland coastal processes it is possible to map out the various strandline positions along the mainland coast during the Holocene regression (Figure 4). The strandline continuities have become heavily masked by urbanization in the vicinity of Abu Dhabi Island, but it is obvious that at least two headlands once existed on the mainland coastal sabkha immediately east of the Island. Butler (1970) mapped the Holocene thicknesses in the embayment between these two headlands, where the Flandrian transgression locally appears to have reached the escarpment of Miocene-Pleistocene strata at the back of the coastal sabkha.

It is interesting to note that the storm beach system encountered on the ‘Evans line’ appears to extend largely hidden beneath the sabkha and is exposed in the banks at the eastern end of the Mussafah channel (Kirkham, in review). The channel was evidently constructed along the southern flank of a former embayment. Whilst part of the “Evans line” beach system may also curve southwestwards beneath the sabkha surface and merge with the parallel storm beaches, part of it post-dates them.

West of Abu Dhabi Island, the limit of the Flandrian transgression is more difficult to locate, although there are local indications from satellite imagery. For instance, one or two additional, very low ridges occur landward of the chenier-like storm beaches, and represent even older beach deposits or zones of supratidal sabkha which were elevated by the vertical expansion of subsurface anhydrite. Where two ridges occur, the most landward is interpreted as approximating to the Flandrian transgressive limit. The anhydrite zone, generally about 0.5 kilometer (km) wide, is the more continuous. The less continuous older beach deposits tend to occur landward of the anhydrite zone (Figure 6). These two ridges combine to form the seaward limit of meteoric and/or marine water-ponding on the sabkha surface.

The offshore barrier islands have enlarged landwards mainly by spit construction under the process of Shamal-dominated leeward accretion that is connecting islands (e.g. Al Dabb’iya in Figures 1 and 7) to the mainland to form tombolos (Purser and Evans, 1973; Figure 7).


Aeolian deflation is a dominant feature of the coastal sabkhas which are ‘equilibrium aeolian deflation surfaces’. The level of the sabkha surface is essentially a ‘Stokes surface’ (Stokes, 1968) adjusting to the top of a groundwater capillary zone. Large areas of the coastal sabkhas that are unprotected by more resistant Holocene sulphates or beach sands, beyond the Flandrian transgressive limit, consist of Pleistocene, uncemented, quartzose aeolian sand that is deflated to the slightly lower elevations where water ponding occurs. Towards the inner margins of the sabkha it is locally overlain by Pleistocene carbonate aeolianites which in turn display other ‘Stokes surfaces’. The aeolianites form part of the indented scarp along the inland margins of the coastal sabkhas, where they are actively being deflated and sometimes eroded into natural sculptures (“hoodoos”) along their windward extremities. These carbonate aeolianites are often informally called ‘miliolite’ because analogous sediments in Iran are occasionally rich in miliolid foraminifera.

Seawards, the sabkha sediments comprise Holocene marine sediment with evaporites (mainly gypsum and anhydrite) overlying the Pleistocene aeolian sand, and coated usually by a veneer of Recent aeolian sand with a surficial halite crust. Any topographic highs are prone to deflation. For instance, the above-mentioned beach ridges are now so deflated that they are barely discernible above the general level of the sabkha. Even more dramatic is the complete deflation of several pre-Holocene outcrops immediately behind these storm beaches between Bu Labyad and Abu Dhabi Island. The main evidence of their former existence is provided by the ‘winged spits’ and slight seaward protrusions of the parallel beach ridges on their windward sides. The only topographic expressions of these former highs may be provided by deflated beach sands which had accreted above or on their seaward flanks.


Many geologists’ perception of the Abu Dhabi coastal sabkhas is governed by the ‘Evans line’ that depicts a thin transgressive sequence which culminated 4 to 7 thousand years (ka) before present (BP) and gave way to a thicker regressive sequence. The entire cycle is about 1.5 meters thick. Exposures of the Holocene sabkha sediments at the eastern end of the Mussafah channel, which intersects the ‘Evans line’ (Figure 4), reveal the true complexity of the stratigraphy (Kirkham, in review) of which perhaps its most notable feature is the cap of regressive supratidal anhydrite.

Many geologists have visited the Abu Dhabi sabkhas expecting to find anhydrite buried just beneath the surface, but experienced difficulties in locating it. Butler (1970) stated that the bulk of the anhydrite occurs in the vicinity of Abu Dhabi. The anhydrite distribution initially appears to be rather sporadic, but systematic sampling of the sabkha with the aid of satellite images reveals that anhydrite is indeed quite predictable and yet its origins are more complex than described by earlier workers. There have been two phases of anhydrite development, referred to as early and late Holocene for convenience.

Early Holocene Anhydrite

The early Holocene anhydrite is best developed on the mainland coastal sabkha in a zone about 0.5 km wide and extending eastwards from Bu Labyad. Its continuity along strike was broken by probable Miocene inliers protruding through the sabkhas. It is evident on satellite images as a zone that sub-parallels the continuous storm beaches on their landward side (Figures 5 and 6). It has not been recognized beyond the eastern limit of these storm beaches and may therefore be absent from both the Mussafah region and Butler’s (1970) ‘Abu Dhabi sabkha’. It is, however, locally developed in most smaller sabkhas separated by headlands of Miocene strata to the west of Bu Labyad.

East of Bu Labyad, trenches dug at least every 0.5 km on transects from the storm beaches to beyond the anhydritic zones revealed a consistent depositional model that is summarised in Figure 8. The anhydrite has a sharp, irregular base and developed as nodules and occasionally as contorted, often truncated bands within brown, quartzose, Pleistocene aeolian sand. Its displacive growth within the sediment elevated the sabkha as a ridge that partly forms the seaward limit of surface water ponding after severe winter rainfalls and/or marine floods by storm-assisted high tides. Any significant evaporite found within the sabkha beneath the ponding areas is essentially limited to intergranular gypsum cementation of the aeolian sand. There is no large-scale hydration to gypsum as the anhydrite decreases in abundance towards the ponding areas along its landward flank.

The top of the anhydrite may be 30 cms or more below the sabkha surface. It gets deeper seawards in the Holocene sequence and is overlain successively by overlapping layers of gypsum mush and lagoonal muds with thin microbial laminations. The anhydrite gradually thins seawards and disappears whilst large disc-shaped gypsum crystals, typically forming near the water table in modern intertidal-lagoonal sediments, become common within the muds.

This composite vertical succession is the reverse of what is typically described in idealised sabkha successions and so is interpreted as transgressive, although a regressive gypsum mush is commonly found above the subtidal muds and so part of the anhydrite may also be early regressive (Figure 9). It is postulated that the deflated and/or marine reworked aeolian dune fields along the palaeo-coastline formed the substrate for supratidal anhydrite growth in the latest stages of the Flandrian transgression. Sea-level continued to rise slightly, causing the gypsum mush and microbial muds to be deposited sequentially above the anhydritic sand. By analogy with the modern peritidal environments, the sequence reflects the progressive change from continental through supratidal to intertidal/lagoonal environments within the transgressive tract. The sequence is correlated with the transgressive Holocene strata described by Kinsman and Park (1976) and Kenig et al. (1989) from the mainland coastal sabkhas near Abu Dhabi Island. A thin, more recent veneer of aeolian sand generally completes this part of the Holocene sequence.

From existing information, the maximum extent of the Flandrian transgression is represented by discontinuously preserved and very highly deflated beach ridges just landward of the anhydritic zone (Figure 6). It is these relict beach ridges that form an alternative seaward boundary to the surface water ponds wherever the anhydritic ridges do not form the same function.

The original northerly extent of the early Holocene lagoon is unknown. The mush layers and marine muds extend seawards at least as far as the continuous deflated storm beaches. The beaches indicate a significant increase in energy of the depositional system that led to possible reworking of the lagoonal sediments. It is equally possible that the storm beaches represent the barrier behind which the lagoons and evaporitic sabkha originally formed.

The coastal sabkhas are narrower and more dissected between Jebel Barakah and Bu Labyad, mainly because the plateau escarpment of Miocene strata approaches the present strandline and isolated Miocene mesa headlands are more abundant. The sabkhas are, therefore, more restricted and largely confined to the embayments between the Miocene peninsulas although some neighbouring sabkhas are linked around the landward sides of the mesas. Facies distributions within these individual, smaller sabkhas are less predictable, even with the aid of satellite imagery. Continuous storm beaches did not develop due to the overriding tendency for ‘winged spits’ to form on the flanks of headlands. The stratigraphic evolution of these sabkhas is, therefore, less well understood and transgressive anhydrite has not been recognized although evaporitic facies typical of the easterly sabkhas are encountered.

Everywhere, this anhydrite and its associated sediments have been deflated to some degree. Vast areas of ‘mature’ mush are exposed to provide a glistening gypsum carpet on the sabkha surface. Outliers of a gypsum mush, over 30 cms in height, sometimes provide relatively conspicuous relief on sabkha surfaces and form the vestiges of formerly more extensive accumulations. They locally provide minimum estimates of deflationary sediment loss. Vertically orientated gypsum discs, often several centimeters in diameter and typically developed near a water table within buried intertidal-lagoonal sediments, locally protrude from the surface in abundance due to deflation of their overlying strata.

Late Holocene Anhydrite

This is the anhydrite encountered seawards of the “winged spit” system along the ‘Evans line’ (Evans et al., 1969). Its areal extent within the former Holocene embayment immediately west of the ‘Evans line’ is not yet resolved, but it is well exposed in the banks of the Mussafah channel. It can be traced with confidence along the supratidal, seaward side of the continuous storm beaches on the mainland coast of Khor Al Qirqishan and Khor Al Bazm at least as far as Bu Labyad. It is essentially non-existent along the less protected coastal areas to the west of Bu Labyad and is also apparently absent from the barrier islands. Unlike the early Holocene anhydrite, it does not have a clear remote sensing signature but at least the continuous storm beaches between the Mussafah channel and Bu Labyad provide a useful guide to its approximate occurrence.

This late Holocene anhydrite appears to be best developed beneath the Mussafah sabkha and in what Butler (1970) designed as the ‘Abu Dhabi sabkha’, which is the wide, mainland coastal plain located immediately east of Abu Dhabi Island. It is progressively underlain in vertical section by gypsum mush, microbial mat and lower intertidal/subtidal sediments. Such a sequence typifies a classic, progradational, peritidal carbonate-evaporite sequence and so the anhydrite is clearly regressive in origin.

The anhydrite is typically nodular but several, well developed and variably contorted anhydrite bands often 2 to 3 cm thick may also occur at any one locality, especially near the very top of the regressive sequence. These contorted bands are frequently exposed and truncated by deflation at the sabkha surface, but the degree of deflation is by no means as severe as that experienced by the earlier Holocene sediments primarily because the relative sea-level fall has not been as great.


The interpretation of ancient evaporites has long been a subject of debate. When Curtis et al (1963) and Shearman (1966) recognised the applicability of the regressive sabkha evaporite model to many ancient evaporites, two schools of thought developed with protagonists supporting either supratidal or subaqueous modes of sulphate accumulations. The debate is frustrated by loss of primary features due to mineral transformation (hydration, dehydration or recrystallisation) tending to blur the original facies characteristics.

Geologists have tried to differentiate between sabkha and salina evaporites (Warren and Kendall, 1985; Alsharhan and Kendall, 1994; Al Silwadi et al., 1996). Warren and Kendall stated that unequivocal interpretations of supratidal anhydrites require evidence of a trinity of subtidal, intertidal and erosion-capped supratidal sediments. Sabkha sequences are generally considered as representing the later marine stages of regressive system tracts and supratidal anhydrites are usually interpreted, by default, as marking the terminal phases of parasequences. The early Holocene sabkha anhydrites between Bu Labyad and Mussafah instead appear to be partly transgressive and demonstrate the potential for very different interpretations that could significantly impact geological modelling based upon high resolution sequence stratigraphic approaches to the rock record. The late Holocene anhydrite is very clearly of regressive origin.

The specific dating of the early Holocene anhydrite is not yet achieved although there are strong indications of its relative age. Given that the early Holocene evaporitic carbonates have been significantly deflated from sabkhas along the entire coastline, they clearly formed at higher sea levels and are being removed as the land surface is reduced to the lower level of the capillary zone. Evans et al. (1969) concluded that the Flandrian transgression climaxed at approximately 1 to 2 meters above present day sea-level along the Abu Dhabi coastline and revealed how the late Holocene anhydrites have formed since about 4 ka BP. This provides a minimum age for the earlier Holocene anhydrite. Kenig (1991) dated Flandrian, transgressive microbial mats at the eastern end of the Mussafah channel as 5,110 ± 167 ka BP, although Kinsman and Park (1976) dated a transgressive mat as 7 to 8 ka BP on the ‘Abu Dhabi sabkha’. These differences may simply reflect diachroneity but it is reasonable to assume that this microbial mat formed penecontemporaneously with part of the early Holocene anhydrite. Dalongeville et al. (1993) dated a higher sea-level stand at 6 ka BP in the northern Emirates and elsewhere in the Gulf.

Although the Abu Dhabi progradational sabkha model has provided the basic interpretational principles for many carbonate-evaporite formations in the last thirty years it is surprising that even in this classic environment the regressive anhydrites are possibly subordinate to the earlier transgressive ones. Butler’s (1970) view that the bulk of the (regressive?) anhydrites occur near Abu Dhabi Island is debatable now that our ability to predict and locate anhydrite is much improved with the aid of remote sensing and easier to access nearly all parts of the coastal sabkhas. His views are even more doubtful because he may have mistakenly interpreted some Miocene anhydrite as Holocene (Kirkham, in review; P. Bush, personal communication).

It has long been widely perceived that the (regressive) Abu Dhabi sabkha model provides the best modern analogue for ancient sabkha environments. The analogue is now complicated if some of the Abu Dhabi anhydrite is transgressive instead of regressive. Ancient sabkha deposits need to be critically examined to correctly assign them to their true stratigraphic context, although dehydration of gypsum mush to anhydrite with deep burial masks the sequential ordering of facies and makes interpretation very difficult.

As previously stated, sedimentological interpretations of the anhydritic sequences within the Arab Formation have long relied upon relatively simple perceptions of coastal sabkha processes arising largely from the Abu Dhabi Holocene model. The regionally correlative, thick anhydrite members within the Arab Formation have historically been assumed to represent major sabkha accumulations, although it is becoming increasingly apparent that subaqueous precipitation has played a significant role in creating the thick, regionally correlatable anhydrites of the Upper Arab Formation and the overlying Hith Formation (Leeder and Zeidan, 1977; Alsharhan and Kendall, 1994; Saner and Abdulghani, 1995; Al Silwadi et al., 1996). The accredited role of sabkhas in creating the Arab Formation anhydrites has, therefore, perceptibly diminished. Indeed, the areally restricted Abu Dhabi sabkha anhydrites question the possibility that regionally extensive progradations of anhydritic sabkhas are possible. Holocene anhydrites of Abu Dhabi have been confined to the supratidal areas of localized lagoons and embayments, which infers that similar geographic settings may be crucial to the formation of ancient sabkha evaporites although the process is not necessarily scale-dependent. Such settings are, however, likely to have been as depositionally complex as the Abu Dhabi Holocene setting and it is not surprising that extreme heterogeneity characterises the more anhydritic upper Arab Formation of Abu Dhabi (Kirkham and Twombley, 1995).


Remote sensing has enabled a better understanding of Holocene coastal evolution in Abu Dhabi. The early Holocene coastline was characterized by peninsulas of Miocene strata and Pleistocene aeolianites separated by embayments that have since been largely filled by carbonate sedimentation and evaporite precipitation. Such peninsulas persist today along much of the Abu Dhabi coastline but some of these palaeo-highs have been completely deflated in the area immediately west of Abu Dhabi Island. Evidence of their former existence is provided by remote sensing of deflated beach ridges forming “winged spits” around their seaward extremities. Remote sensing has also assisted in locally defining the limits of the Flandrian transgression in some areas.

Two generations of Holocene anhydrite are recognized as discontinuous zones that sub-parallel the present day coastline. The earlier anhydrite and its associated gypsiferous sediments accumulated when sea-level was over a meter higher but are now undergoing deflation. Its still slightly elevated character contributes to the northern limit of regular water-ponding on the sabkha and its trend is traceable over a 50 km distance along the strike of the extensive mainland sabkha west of Abu Dhabi Island. It is partly transgressive and is locally associated with traces of Holocene beach deposits demarcating the known maximum limit of the Flandrian transgression on its landward side. Further west, this early Holocene anhydrite developed along the inner reaches of the smaller sabkhas between peninsulas of Miocene rocks. The entirely regressive late Holocene anhydrite has formed along the mainland supratidal zone of Khor Al Qirqishan and the eastern part of the Khor Al Bazm. It is separated areally from the early Holocene anhydrite by a storm beach system and lagoonal deposits.


The author very much appreciates the encouragement and support for this work given by BP Exploration and Abu Dhabi Company for Onshore Oil Operations. Thanks also go to the U.K. National Remote Sensing Centre; to Gulf PetroLink; and to John Stillwell (Abu Dhabi National Oil Company) for assistance with the illustrations. Valuable discussions on the subject were held with Graham Evans and Peter Bush.


Anthony Kirkham was employed by BP Exploration for 20 years as a Sedimentologist and Senior Development Geologist. He mostly worked within reservoir engineering teams on international development projects and had long-term assignments in Norway, Egypt and Turkey. From 1994 to early 1997, Anthony worked as a Geological Specialist with Abu Dhabi Company for Onshore Oil Operations but is currently a Senior Technical Consultant with Landmark Graphics in Abu Dhabi. He received his BSc from Aberystwyth University in 1970, MSc from Imperial College, London in 1972 and PhD from Bristol University in 1977. Sedimentology, reservoir characterization and 3-D reservoir modeling are his particular areas of interest. Anthony is Secretary of the Society of Explorationists in the Emirates and a member of the Editorial Advisory Board of GeoArabia.