Evolution of the Dead Sea sinkholes
Published:January 01, 2006
- PDF LinkChapter PDF
Meir Abelson, Yoseph Yechieli, Onn Crouvi, Gidon Baer, Daniel Wachs, Amos Bein, Vladimir Shtivelman, 2006. "Evolution of the Dead Sea sinkholes", New Frontiers in Dead Sea Paleoenvironmental Research, Yehouda Enzel, Amotz Agnon, Mordechai Stein
Download citation file:
Over the past several years, the coastal area around the declining Dead Sea has undergone a catastrophic collapse. One of the major expressions of this process is the sudden appearance of hundreds of collapse sinkholes, causing a severe threat to the future of this region. Here we review results and inferences obtained from a multidisciplinary research conducted since 1999. Observations were obtained by geological mapping, aerial photographs, drilling, groundwater geochemistry, seismic refraction and reflection, and satellite radar interferometry. The suggested model for the formation of the Dead Sea sinkholes is based on the following observations: (1) presence of a thick salt layer (or layers) at depths between 20 and 50 m (depth of layer top), and sandwiched between aquiclude layers of clay and silt; (2) identification of cavities within the salt layer in sinkhole sites; (3) presence of water undersaturated with respect to halite in aquifers confined beneath the salt layer; (4) composition of the groundwater in the salt layer that indicates salt dissolution; (5) association between sinkhole sites and land subsidence; and (6) formation of sinkholes along and above buried faults. These observations combine to suggest that the primary cause of sinkhole formation is dissolution of the salt layer by undersaturated groundwater. The interface between the Dead Sea brine and this groundwater migrated eastward due to the Dead Sea decline. Undersaturated water accessed the salt layer via faults that cut through the soft aquiclude layers. The opening of these conduit-faults is likely due to differential compaction of the aquiclude layers, explaining the correlation between the land subsidence and sinkhole sites. It appears that the decline of the Dead Sea level affects the formation of sinkholes in three ways: (1) opening the way to eastward migration of the fresh-saline water interface and thus to undersaturated groundwater, (2) generating differential compaction of fine-grained sediments, and (3) destabilization of underground cavities, which catalyzes their collapse.
Collapse sinkholes started to appear along the Dead Sea coast in Israel and Jordan in the early 1980s. Sinkhole development has significantly accelerated since 2000 with an abrupt occurrence of hundreds of sinkholes. This regional-scale collapse is attributed to the rapid decline of the level of the Dead Sea (∼1 m/yr) (Arkin, 1993; Arkin and Gilat, 2000; Taqieddin et al., 2000; Wachs et al., 2000; Yechieli et al., 2006). The decline of the Dead Sea, which exceeds 25 m since the early 1930s, reflects human activities such as the interception of a fresh water supply from the Jordan River and the maintenance of large evaporation ponds by Dead Sea mineral industries in Jordan and Israel.
Thirty-six sinkhole sites (Fig. 1) are observed in two main sedimentary environments along the west coast of the Dead Sea: mud flats and alluvial fans. The alluvial fans are made of coarse gravel alternating with fine-grained sediments (silt and clay), whereas the mud flats are mainly fine-grained sediments. A single sinkhole can reach a diameter of 25 m and extend to a depth of 20 m. Deeper sinkholes are found in the alluvial fans and shallower ones in the wet mud flats. The diameter/depth ratio tends to be small in the alluvial fans and much higher in the wet mud flats.
Since 1999, the Geological Survey of Israel and the Geophysical Institute of Israel have conducted multidisciplinary research in order to understand how these sinkholes develop. This research employed a variety of tools, such as seismic reflection and refraction, electrical methods, microgravity, aerial photograph interpretation, drilling, groundwater sampling, satellite remote sensing, and general geological reconnaissance (e.g., Shtivelman et al., 1999; Wachs et al., 2000; Yechieli et al., 2002). Here we review geological and hydrological evidence implying that sinkholes are formed due to dissolution of a salt layer (Yechieli et al., 2006). We further present observations showing that sinkhole formation depends on the fault structure of the Dead Sea basin as well as land subsidence caused by compaction of fine-grained sediments (clay and silt) in the upper part of the sedimentary fill. Finally, we present a mechanism for the formation of sinkholes along the Dead Sea coast that attempts to reconcile all observations.
DISSOLUTION OF SALT LAYER AND SINKHOLE FORMATION—KEY OBSERVATIONS
Previous studies have shown that the primary cause for collapse sinkholes is the formation of cavities by dissolution within layers of soluble rocks (e.g., Martinez et al., 1998; Galloway et al., 1999; Neal and Johnson, 2002). At some stage, overlying layers fail to bridge the growing cavities and collapse structures may reach the surface, forming a sinkhole. In order to assess whether the Dead Sea sinkholes formed by a similar process, we have searched for layers of soluble rock in the upper section of the sedimentary fill of the Dead Sea rift and investigated groundwater chemistry for potential dissolution. The subsurface setting was explored by seismic refraction, boreholes, and sampling of groundwater from the boreholes. The following subsections review the key observations obtained by this exploration.
Shallow Salt Layer Buried within The Dead Sea Fill: Findings from Boreholes and Seismic Refraction
Seismic refraction profiles were conducted along most of the Dead Sea coast by the Geophysical Institute of Israel. These data, and 20 boreholes in the vicinity of seven sinkhole sites, indicate that a salt layer, several meters thick, is embedded within the upper part of the sedimentary section along the Dead Sea coast. A typical profile of seismic refraction in the Dead Sea region displays three layers with P-wave velocities of 600–800 m/s, 2000–2300 m/s, and 2900–3600 m/s for the upper to lower layers, respectively (Fig. 2). The upper two layers consist of uncemented or unconsolidated alluvial and fluvial sediments. The lower layer, where observed, is the salt layer. This stratigraphy was verified by boreholes in several sites (e.g., Hever-south, Ze'elim, En-Gedi, Shalem; Fig. 1), where a solid salt layer was penetrated at depths predicted by the seismic refraction. For example, the borehole Hever-2 at the Hever-south site penetrated an 11-m-thick salt layer at a depth of 24 m, as predicted by the refraction profile (Yechieli et al., 2002) (Fig. 2). The age of the salt layer was found to be ca. 10 ka, similar to the salt layer in the Ze'elim area (see Fig. 1 for location; Yechieli et al., 1993). The salt layer shows a broad range of P-wave velocities, between 2900 and 3600 m/s, perhaps due to the occurrence of both solid versus “crumbly” salt. Accordingly, we have used seismic refraction profiles cautiously to identify the extension of the salt layer.
The association between sinkhole occurrence and the subsur-face salt layer was corroborated by boreholes in seven sinkhole sites: Darga, Shalem, En-Gedi, Mazor, Hever-south, En-Bokkek, and Neve-Zohar (Figs. 1–3). At all of these sites, a salt layer was penetrated, supporting a dissolution-collapse origin for the Dead Sea sinkholes. The depth of the top of the salt layer ranges between 20–50 m, and in some locations the thickness of the salt layer exceeds 20 m. We do not yet know whether there is a single salt layer or multiple layers from several stratigraphic units. For this purpose, more dating of salt layer in several sites is required.
Cavities within the Salt Layer
In two of the seven sinkhole sites examined by boreholes, cavities were encountered within the salt layer. At the Hever-south site, one borehole (Hever-1) encountered an 11-m-thick salt layer. A second borehole (Hever-3), drilled 40 m south of Hever-1, encountered a cavity at 23–29 m depth, at the same stratigraphic level as the salt layer found in Hever-1 (Fig. 4). At the Shalem site (Mineral Beach; Fig. 1) the salt layer was penetrated at a depth of 19 m and a cavity was found at its base, over a depth range of between 28 and 31 m. A waterproof camera inserted into the cavity through the borehole indicated that the cavity wall is made of coarse-crystal salt (Fig. 3B). The diameter of the cavity is larger than the 1.5 m maximum spread of the caliper arms. The cavities found in the salt layers support the inference that salt dissolution causes the formation of the Dead Sea sinkholes.
Hydrology and Groundwater Chemistry
Alternating fine-grained (clay and silt) and gravel layers occur in the upper sedimentary section along the Dead Sea coast, forming several subaquifers (Fig. 4). In some locations (e.g., the En Gedi area), the groundwater head in the lower subaquifer is higher than in the upper subaquifer (Yechieli et al., 2004), indicating upward flow potential. The groundwater in the lower sub-aquifer beneath the salt layer is much less saline (Cl = 15 g/l; 78 g/l in various boreholes in En Gedi area) than the Dead Sea brine (Cl = 210 g/l) (Yechieli et al., 2004; 2006). Calculations indicate that whereas the Dead Sea brine is saturated with respect to halite, the dilute groundwater is far below saturation and therefore has the potential to dissolve salt. Furthermore, geochemical evidence proves that dissolution of salt does occur. This is best exhibited by the high Na/Cl ratio of groundwater (0.6) from within the cavity in the salt layer in the Mineral-2 borehole (Fig. 3B) compared to the Dead Sea brine (0.25) (Yechieli et al., 2004, 2006). The increased Na/Cl ratio reflects dissolution of the salt layer by groundwater consisting of mixed Dead Sea–type brine and more diluted groundwater. In Mineral Beach, the source of the fresher groundwater is thermal brine seepage from deep strata. In En Gedi and most other sites, the dilute groundwater is derived from the regional freshwater aquifer recharged in the mountains to the west (Yechieli et al., 2001). The active groundwater flow, which drains to the declining Dead Sea, maintains a continuous flux of undersaturated water through the salt layer, thereby enhancing ongoing dissolution.
APPEARANCE AND DISTRIBUTION OF SINKHOLES ALONG THE DEAD SEA COAST
Variations in Distribution of Sinkholes and Sinkhole Sites
Earlier studies (Raz, 2000; Itamar and Reizmann, 2000) show that sinkholes are not uniformly scattered along the Dead Sea shores, but rather occur as clusters. Presently, more than 30 sinkholes sites are known (Fig. 1), and the number of sinkholes at each site ranges between one (e.g., Hever-fan) to ∼100 (e.g., Lisan site, Fig. 1). In order to monitor sinkhole development, we used aerial and orthorectified aerial photographs from a number of years (Abelson et al., 2002). The data sets from the rectified photographs are incorporated in a geographic information system (GIS), which enables quantitative analysis of data acquired during monitoring of the sinkhole development (Fig. 5).
As observed in aerial photographs, sinkholes tend to develop along lineaments at most sites (Raz, 2000; Abelson et al., 2002, 2003). The linear development occurs in two modes: (1) new sinkholes form between older sinkholes without extending the lineament, and (2) sinkholes form at the ends of the lineament, causing its extension (Fig. 5). A representative measure of sinkhole evolution is the growth of the total area occupied by sinkholes at a given site (Yechieli et al., 2003). The Shalem sites (Fig. 5) develop along lineaments and show an abrupt increase in growth rate between 1999 and 2000. Such accelerations in the growth rate during 1999 and 2000 are found in most of the sinkhole sites of the Dead Sea northern basin.
On a regional scale, it appears that formation of new sinkhole sites propagates northward (Fig. 6), while older sites continue to grow in the south. Besides this northward propagation, the most intensive sinkhole activity has also migrated to the northern sites. This evolution may be explained by the shape and location of the western margin of the 10 ka salt layer: in the north, it is close to the present shoreline, whereas in the south, it is much further west. This notion is demonstrated by observations from boreholes. Two boreholes from the northern part of the Dead Sea coast show an absence of a massive salt layer; one is at Zukim nature reservation 150 m from the coastline, and the second is at the northwestern corner of the Dead Sea, ∼30 m from the coastline (Neve Midbar Beach). On the other hand, in the southern part of the northern basin, several boreholes drilled some 500 m away from the shoreline have encountered a thick salt layer.
Sinkholes above Concealed, Likely Active, Faults
The trends of sinkhole lineaments, exposed faults, and zigzagging segments of the rift escarpments show a striking similarity (Abelson et al., 2003) (Fig. 7). All features have a predominantly bimodal distribution with NNE and NW principal directions (Fig. 7). No similarity is observed between sinkhole lineaments and other surface features such as ancient or current Dead Sea shorelines or alluvial fans, implying a deeper origin for the sinkhole lineaments, such as faults concealed within the rift fill.
To confirm a linkage between buried faults and sinkhole lines, we conducted profiles of seismic reflection across and along sinkhole lines in six different sites. At these sites, the sinkhole lineaments were found to overlie prominent discontinuities. We present here an example (Fig. 8) showing clear interrupted reflectors beneath the sinkhole line in Shalem site and intact reflectors away from the sinkhole line. Another example, previously shown by Abelson et al. (2003), is the Hever-south site, where layers shallower than 20 m are offset beneath the sinkhole line. Carbon-14 dating of the salt layer at a depth of 27 m (sampled from a borehole at this site) indicates an age of about 10 ka, suggesting that the offset layers are younger than 7.5 ka. This implies that the faults are young and likely to be active, though they have no surface manifestation other than the sinkhole lineaments. The linkage between faults and sinkholes suggests that faults may play a major role in sinkhole formation. They may serve as conduits for undersaturated groundwater, enabling access across the aquiclude layers.
Sinkholes Association with Land Subsidence, Alluvial Fans, and Mud Flats
Subtle land subsidence along the Dead Sea coast was recently detected by interferometric synthetic aperture radar (InSAR) (Baer et al., 2002). The SAR scenes were obtained by the European Remote Sensing satellites ERS-1 and ERS-2 during the years 1992–1999. The interferograms span periods of 2–71 months. The InSAR analysis reveals gradual subsidence features that are a few hundred meters to a few kilometers long (Fig. 9), with subsidence rates of 0–20 mm/yr (Baer et al., 2002). This subsidence is attributed to consolidation of clay and silt layers (i.e., the aquiclude layers) due to the declining Dead Sea level and the associated drop in groundwater level (Baer et al., 2002). The drop in the groundwater level decreases the hydrostatic pressure (P h), which increases the effective stresses (P e), P e = P l − P h, where P l is the lithostatic overburden (Terzaghi, 1925). Layers with higher compression factors (i.e., the fine-grained clay and silt) respond to the increase of P e by compaction (Terzaghi, 1925; Galloway et al., 1998, 1999; Baer et al., 2002). Therefore, land subsidence is observed following the rapid Dead Sea level drop.
The subsidence areas form a narrow strip below and east of the topographic contour 400 m below sea level (Fig. 9). Most sinkhole sites (polygons in Fig. 9) are found within this strip in the vicinity of subsidence maxima (bright spots in Fig. 9), suggesting a link between sinkhole formation and aquiclude compaction. Furthermore, the relative timing of the Dead Sea decline as well as sinkhole appearance and acceleration (Fig. 10) show that the inception of sinkhole formation occurred as the Dead Sea level declined below the −400 m contour. Exceptions are found at seven sites (out of 34) located on the alluvial fans of major streams above the −400 contour (Fig. 10). It is noteworthy that the elevation range of the sinkhole sites, between the −400 contour and the shoreline, is systematically increasing with the advancement of the Dead Sea decline (Fig. 10). This observation can be explained by the ongoing exposure of landmass above the salt layer where the sinkholes originate.
MECHANISM OF FORMATION OF THE DEAD SEA SINKHOLES
As previously mentioned, the main trigger for the formation of sinkholes appears to be the declining level of the Dead Sea (Arkin and Gilat, 2000; Wachs et al., 2000; Abelson et al., 2003; Yechieli et al., 2003). This affects the formation of sinkholes in three ways, and is thus termed as the triple effect (Fig. 11): (1) By allowing invasion of undersaturated water from the west, (2) by generating differential compaction that fractures the aquiclude layers, and (3) by decreasing hydrostatic pressure and thereby increasing cavity instability in the salt layer. Whereas the first two effects promote the formation of cavities within the salt layer, the third one destabilizes an existing cavity rather than creating it.
The First Effect—Hydrological Factors (Fig. 11A)
The Dead Sea is hydraulically connected to the adjacent groundwater system, serving as a terminal base level for the flow system of the groundwater. Therefore, the drop of the Dead Sea level is accompanied by a drop of the groundwater level, though at a lower rate (Yechieli et al., 1995). For example, groundwater levels responded within a few days of the sharp rise in Dead Sea level in the winter of 1992 (Yechieli et al., 1995). The changes in groundwater level are greater near the shoreline (<1 km distance) than in more distant areas (Yechieli, 1993).
Groundwater chemistry varies significantly from freshwater springs (e.g., En Gedi and Arugot springs) to extremely saline, as the Dead Sea brine. Since the density of the Dead Sea brine is 1.23 g/cc, the fresh-saline groundwater interface is very shallow, ∼10 times shallower than that near the ocean (Yechieli, 1993, 2000), according to the Ghyben-Herzberg approximation for an equilibrium position. Time-domain electromagnetic (TDEM) sounding traverses across the Dead Sea coastal plain (Kafri et al., 1997, Yechieli et al., 2001) detected the fresh-saline water interface generally near the expected equilibrium position. The lowering of the Dead Sea and groundwater level is expected to be accompanied by the eastward migration of the fresh-saline interface (Yechieli, 2000, Yechieli et al., 2004), tending to bring undersaturated groundwater to the vicinity of the salt layer and dissolving it.
The Second Effect—The Role of Land Subsidence and Faults (Fig. 11B)
Eastward migration of undersaturated water and its emplacement below the salt layer is caused by retreat of the Dead Sea shoreline. However, boreholes show that the salt layer is usually underlain or sandwiched by aquiclude layers (i.e., silt and clay), suggesting limited access of undersaturated water to the salt layer. We suggest that the second effect of the declining Dead Sea level is the formation and/or reactivation of faults that cut through the fine-grained layers and enable the final access of undersaturated groundwater through these faults to the salt layer (Abelson et al., 2003). The decline of the Dead Sea level increases the effective stress and generates differential compaction of the aquiclude layers. The differential compaction results in shear deformation of the aquiclude layers, confining the salt layer along preexisting faults. Reactivation of these faults by differential compaction opens conduits for upward migration of undersaturated water. This upward migration is driven by the groundwater overpressure beneath the salt (e.g., the En Gedi site). The undersaturated water migrates upward and forms cavities that are the roots of the collapse sinkholes.
These coupled effects are inferred from several key observations: (1) The location of sinkhole sites clearly correlates with areas of subtle land subsidence recorded by InSAR (Fig. 9), attributed to differential compaction of aquiclude layers; (2) sinkhole clusters display linear shape and form above faults, implying water flow through faults; and (3) groundwater undersaturated with respect to halite is overpressurized beneath the salt layer and the fine-grained layers, with borehole data documenting larger hydraulic heads than in the upper phreatic subaquifer (Yechieli et al., 2004). The last observation suggests that this water can ascend through the faults toward the salt layer and dissolve it.
The Third Effect—Increase of Instability of Existing Cavities (Fig. 11C)
The increase in effective stress due to the Dead Sea level drop also directly destabilizes cavities in the salt layer, promoting the collapse of overlying sediments. This effect serves as a catalyst for sinkhole collapse rather than causing new underground cavities. Lowering of groundwater level, usually by pumping, induces sinkholes in many places in the world (e.g., Galloway et al., 1999). However, this mechanism is not yet clear and should be investigated further.
SPECIAL AREAS OF SINKHOLE OCCURRENCE
Elevated Sinkhole Sites in Alluvial Fans
Sinkhole sites located on the alluvial fans are exceptional in two aspects: (1) their elevation is above −400 m, and (2) they are located outside the zone of land subsidence recorded by the InSAR (Figs. 9). Considering the development of sinkholes above a salt layer, the first observation is possibly due to a combination of spatial distribution of the salt layer below surface and the rate of sediment accumulation above the salt since its deposition. As indicated from boreholes, the top of the salt layer in the alluvial fans is at a deeper topographic level than the salt top in the mud flats, e.g., −440 m in the Arugot fan near En Gedi versus −428 m at the mud flats 3 km southward. This suggests that (1) the higher accumulation rate of sediments in the alluvial fans since the salt deposition caused the elevated sinkhole sites in the alluvial fans, and (2) during salt deposition, no prominent alluvial fan is found under the present one, while the major paleoalluvial fans were probably deposited further to the west (Fig. 12). The apparent absence of land subsidence recorded by InSAR on the alluvial fans is explained by the domination of coarse gravel in the uppermost section typically found in boreholes from alluvial fans (see example in Fig. 2). This gravel is much less compressible than the fine-grained sediments. Nevertheless, similar to the mud flats area, a typical section through the alluvial fan contains fine-grained aquiclude layers above and below the salt layer (Fig. 2). Furthermore, most sinkhole sites in the alluvial fans develop above faults (see example in Fig. 13), as is found in the alluvial fans at Arugot, Hever, and Neve-Zohar. These two observations suggest that the mechanism combining differential compaction and breaching aquiclude layers is also valid in the alluvial fans. However, this differential compaction at depth is not expressed on the surface by the InSAR, probably due to the thick layers of coarse alluvial sediments.
The Hever-fan demonstrates a typical development of sinkholes at alluvial fans. The sites of Asa'el, Hever-south, and Hever-fan form a 1-km-long lineament trending 340°, suggesting sinkhole development above a major fault (Fig. 13), as also corroborated from seismic reflection (Abelson et al., 2003). In addition, the date of first appearance of these sites indicates propagation of sinkhole activity along this large-scale lineament from the subsiding mud flats below the −400 m contour into the rise of the alluvial fan (Fig. 13). The delay of sinkhole collapse on the alluvial fan may be caused by the more competent gravel relative to the weak, muddy sediments built of silt and clay, as well as an increase of thickness of gravel layers overlying the salt layer toward the center of the alluvial fan. A similar propagation of sinkhole activity from a lower area of land subsidence into elevated alluvial fan is found in the Arugot alluvial fan and possibly in Neve Zohar.
Evaporation Ponds in the Southern Subbasin
The Dead Sea basin is divided into two main subbasins by a sill extending westward from the Lisan Peninsula (Fig. 1). The northern subbasin is larger and deeper and contains the water of the present Dead Sea. The shallow southern subbasin is occupied by evaporation ponds of the local potash industry in both Israel and Jordan. The southern basin is naturally detached from the main Dead Sea by the sill of the Lintch Strait and artificially by the dams of the Dead Sea Works. Water supply from the northern to the southern basin is via canal. The water level in the southern basin declined continuously until the mid-1970s; since then, the water level decline has stopped, and the water level has recovered somewhat over the past 20 years. Accordingly, the development of sinkholes in the coast of the southern basin is dramatically lower than in the northern basin. For instance, in the northern basin, the growth rate in the last four years is 150–200 sinkholes per year, whereas in the southern basin, it is around one sinkhole per year. These observations strongly support the control of the decline in the Dead Sea level on the development of sinkholes.
We have shown that the primary cause for the formation of sinkholes along the Dead Sea coast is the dissolution of a subsurface salt layer. Groundwater in the vicinity of the Dead Sea coast contains two end-members of salinity, the brine of the hypersaline Dead Sea in the east and fresh groundwater originating on the western flanks of the Dead Sea rift. The declining Dead Sea level causes eastward migration of the fresh-saline interface and promotes the invasion of the salt layer by undersaturated groundwater. The contact between the undersaturated water and the salt layer is made possible by the differential compaction of aquiclude layers above and below the salt layer, which opens fractures that are conduits to the salt layer. This differential compaction is also induced by the decline of the groundwater level in response to the declining Dead Sea. This means that a level drop in the Dead Sea has the double effect of prompting the formation of cavities in the salt layer by (1) eastward migration of the undersaturated groundwater, and (2) opening of faults by differential compaction of aquiclude layers above and below the salt layer.
Assistance and discussions with Ittai Kurzon, Josh Steinberg, Eli Raz, Michael Beyth, Duba Primerman, and Mark Talesnik during the course of this research are greatly appreciated. Steve Ingebritsen, Peter Styles, and Yehuda Enzel are thanked for helpful reviews. ERS SAR data was provided by the European Space Agency under Category-1 project no. 1058.
Figures & Tables
New Frontiers in Dead Sea Paleoenvironmental Research
- aerial photography
- blind faults
- buried features
- collapse structures
- Dead Sea
- geologic hazards
- geophysical methods
- geophysical surveys
- ground water
- land subsidence
- landform evolution
- Middle East
- remote sensing
- satellite methods
- seismic methods
- solution features