The Dead Sea Sinkholes are expediently developing over the past 30 years due to the desiccation of the Dead Sea. We followed their evolution for almost a decade using terrestrial and airborne laser scans. The acquired point clouds allowed us to evaluate their change over the years and better understand their dynamics.

The Dead Sea, the lowest place on Earth, drains an extensive region in its surrounding countries. However, when a dam was built in the mid-1960s to divert fresh water from the its main drainage basin, an expedite artificial desiccation process begun, and the lake today is rapidly disappearing, at a rate of about 1 m a-year. This meteoric drop in lake level is undermining the stability of the surface around the lake, triggering a chain of reactions with disastrous effects to both natural and man-made environments.

Salty crust formed along the coast of the Dead Sea. In the background, power lines and infrastructure of the north-most pumping station.

The most prominent response is the accelerated development of collapse sinkholes, which started to appear in early 1990s on both Israeli and Jordanian sides. Scientists commonly contributed their appearance to the dissolution of salt (halite) layers by fresh underground water (aka “groundwater”) some 20-40 m underneath the surface. This creates underground cavities that later collapse leading to what we see on the surface as sinkholes . However, the reason why some sinkholes appear as solitary individuals and some are turning into hundred of metres dense fields was not clear.

To answer that question, we used both airborne and terrestrial laser scans for almost 10 years and followed sinkholes dynamics in one of the largest alluvial fan along the Dead Sea coast. The Ze’elim alluvial fan covers an area of approximately 17 km2, and was, at the time, a focal point for planned construction. Therefore, it was not important to understand these sinkholes only from a scientific point of view, but it was essential for future planning. Figuring out their behaviour would serve us in building sustainable constructions that will not fail by the constant collapsing of new sinkholes.

We start from point zero – the airborne scan of December 2005.

In order to better view the surface, we preset the point clouds as hillshade maps. More information about such maps can be read here.

Base line: December 2005

In December 2005 we carried the first airborne scan of the region. At 4 points per square meter, we got a general picture of the current state of the fan. This include gullies (channels) that flow to the lake and several sinkholes, mostly scattered arbitrarily along the fan. Using a detection method (detailed here) we automatically mapped all sinkholes in the fan – 35 of them in total. Based on these mappings, we marked four regions with multiple sinkholes that seem to have the potential of growing.

Automatic detection of sinkholes

First epoch: May 2011

A second scan was carried on May 2011, almost six years after the first scan. During that time, the fan surface has changed dramatically. A 150 m of land has been exposed because of the 6 m drop of sea level (1 m/yr approximately). The gullies, adapting to the new sea level, cut the land further east. Quantitative analysis can trace the “unseen” changes (here you can find more information on how to detect changes). Subtracting of one dataset from the other shows the a 40 cm compaction of the fan surface, probably due to the desiccation of the mudflat.

Difference map 2005-2011 – The colours represent height differences between the epochs.

The differentiation map also depicts the changes in the sinkholes’ regions. Groups #II and #IV have been developed extensively, while groups #I and #III hardly changed. Group #IV has five times more sinkholes than before, turning into a dense field of 50 sinkholes. The new sinkholes are wider and deeper than those in 2005: they are 5–10 m wide and up to 6 m deep. The new sinkholes are arranged vertically to the coast, as if advancing toward the lake. Group #II also grew into a dense and clustered field. It now holds 24 sinkholes, also oriented towards the lake. Slide the sliders below to see the difference…

Group #II in 2005 (left) and 2011 (right).
Group #IV in 2005 (left) and 2011 (right).

As we could not scan the whole fan terrestrially, we had to decide on which region we should focus. Since group #II exhibited the most changes, and had the first interface with gullies in the fan, we chose to focus on it. This way we could also monitor small changes.

Second epoch: December 2011

We visited group #II in December, just after the first rain of that year. While most sinkholes seemed to maintain their shape, the differentiation between the acquired point clouds showed that they have deepened by up to 3 m. Their edges are not stable, and they have subsided by 1.5 m. A small new sinkhole has emerged, seamlessly connected to a gully. A closer look shows that it serves as a water catchment to the water streaming from the gully, as a small alluvial fan has developed within it – proving a loss of energy as the water is drained in the sinkhole. All and all, about 4500 m3 of soil was removed from the existing sinkholes.e could not have spotted with the naked eye.

Group #II in May 2011 (left) and December 2011 (right). The borders of the terrestrial scan are marked in white line.

Third epoch: March 2012

After four months, we could hardly recognize the site: there were sinkholes everywhere, and the region has changed completely. The group, which has now turned to a dense field, covers an area of 310×180 m2 – almost eight times its size four months earlier. Fifteen new sinkholes have emerged, while the existing ones either united with each other or deepened considerably. About 54.150 m3 of soil was removed from the region in these four months, a rate of 13,525 m3 /month, almost 13 times faster than the rate between May and December of last year.

Field II in March 2012

Fourth epoch: June 2013

In June 2013 we carried another airborne scan. Similar analysis showed that there was hardly any compaction since 2011. However, more land was exposed due to the lake retreat, and the gullies are still at work, adapting themselves to the new lake level.

As for the sinkholes, field #II keeps developing eastwards, and is now almost 380 x 180 m2. Its expansion rate is slowed down, but it is still deepening. Our computations show that approximately 35.200 m3 of soil was removed since the last epoch. Group #III, which hardly changed between 2005 and 2011, is now rapidly expanding. In fact, we can trace the initiation of this development to the beginning of winter that year. Now it is a clustered field, also oriented to the Dead Sea. As for Field #IV – the differentiation between the point clouds shows that it expanded in the same direction as before, accentuating the orientation to the lake, but the middle of the field is stabilized. It is worth noting that group #I hardly changed since we started monitoring.

Difference map 2011-2013 – The colours represent height differences.

The point clouds role in studying the sinkholes

Besides the extensive picture of the fan evolution that is drawn over the years, meaning – the new shorelines, the compaction of the fan surface, and the incision of gullies, the point clouds allow us a closer look on the development of sinkholes. In fact, an extension to the prevailing theory of sinkholes evolution is derived by analysing the above processes. To summarize, the entire evolution follows four main phases:

(1) Initial formation: The drop in lake-level modifies the groundwater regime, so that fresh water dissolves the subsurface salt layer and triggers the formation of sinkholes. This stage is controlled solely by subsurface processes, and therefore seems arbitrary on the surface. The 2005 scan demonstrates this phase, where scattered sinkholes are scattered along the fan.

(2) Accelerated development: As the sinkholes and the drainage system interact, runoff water from floods penetrates the surface and transported underground. There it triggers the development of new sinkholes on its way to the lake, with a limited lateral extent. This stage is characterised by a gradual increase in the connectivity between surface and underground processes. It can be seen by the development of sinkholes with clear orientation towards the lake.

(3) Deepening and spatial localisation: After expanding, the sinkholes begin a deepening phase: most of the new collapses will occur within the cluster of existing sinkholes, and almost no new sinkholes will be formed outside of those clusters. Though the fields deepen, they are spatially stabilised. This we can see at the last epoch for field #II. A similar phenomena was seen also in field #III a few years later.

(4) Stabilisation: The final stage. The field stops developing, and only small collapses occur within its boundaries. Examples of this stage are seen in two different sinkhole fields along the fan: first, field #IV came to a developmental halt when field #II diverted all runoff into a subsurface tunnel, cutting off its runoff supply; second, group #I ceased its development during the entire monitoring period due to its disconnection from the active runoff system.

Understanding that runoff and the development of sinkhole fields are inter-related has vast impact on future planning. Diverging runoff into designated locations will decrease sinkhole development and allow better and sustainable design in such environments. This includes not only the Dead Sea but also similar hazardous environments worldwide, particularly in areas characterised by highly soluble layers in the subsurface, e.g., NE Spain, south and SW China, western Iran, or northern Germany.

Technical Details

Acquisition was carried by both terrestrial and airbone laser scanners.

Terrestrial laser scanner (Leica c10)

  • 20-35 points per cm2
  • Modelled surface precision: 2 mm
  • Angle measurement precision: 60 millirad
  • Registration error: 3 mm

Airborne laser scanner (Optech 2050)

  • 4-8 points per m2
  • Measurement accuracy of 2 cm horizontal and 5 cm vertical
  • Registration error: 10 cm (based on GPS survey)

Participants in Research

  • Laser Scanning and Photogrammetry Lab, Technion – Israel Institute of Technology
    • Reuma Arav
    • Sagi Filin
  • Geological Survey of Israel
    • Yoav Avni

Some of the images in this post were reprinted from Geomorphology, Volume 351, Reuma Arav, Sagi Filin, and Yoav Avni, Sinkhole swarms from initiation to stabilisation based on in situ high-resolution 3-D observations, 106916. Copyright (2019), with permission from Elsevier