Earth fill dams, because of their low construction cost and relatively simple design, are the most common hydraulic structures built around the world. The safety of a dam throughout its operating life is of paramount importance and requires constant surveillance and monitoring of deformation to assure its safety and structural integrity.
Dams are susceptible to two kinds of deformations [1
]. First, horizontal deformation caused by water impounded in the reservoir. Second, vertical deformation resulting from the weight of the dam. Several factors can contribute to dam deformation, such as hydrostatic loading from water storage, construction parameters, and site geology. The latter is associated with planes and zones of weakness in the foundation and abutment rocks.
Evaporite rocks, such as gypsum and halite, are chemically more soluble than other rock types and pose much more difficult problems in dam engineering [2
]. It causes two of the most common problems at the dam site—piping and karstification, which could lead to failure. These two phenomena require significant pre- and post-construction treatments under and around the dam foundation to: (a) prevent excessive water seepage through subsurface solutioned openings, and (b) collapse of large solution-induced voids that may cause subsidence and possible dam failure leading to catastrophic flooding downstream [3
Differential Synthetic Aperture Radar (SAR) Interferometry (DInSAR) calculates the interference pattern created by phase difference between two radar signals across a common point over the earth surface. Multi-temporal analysis of SAR images provides very accurate monitoring of individual deformation down to mm-level accuracy [4
]. One of the robust multi-temporal DInSAR analysis techniques is the Persistent Scatterer Interferometry (PSI), which allows precise estimation of the deformation displacement-time series [7
] and has been applied to monitor dams successfully [8
]. The PSI is a suitable technique to monitor dam deformation that allows to gather data on its subsidence [8
]. A detailed description of the DInSAR and PSI and their applications can be found in [7
Although some studies have been published on deformation monitoring of Mosul Dam (MD) in NW Iraq [21
] using DInSAR techniques such as small baselines (SBAS) [1
] and PSI [21
], they lack validation of their results with in-situ data. It is important that remote sensing-derived analyses and interpretation are validated by comparing the results with proper differential GNSS and/or GNSS stations results.
In this paper, we briefly describe the history and geological setting of MD. Next, we explain the methodology used in the study; and finally, we present and discuss results of the geodetic and DInSAR surveys by including a thorough consideration of the dam site geology and validating the DInSAR results by in-situ GNSS measurements. We have also been able to establish the cause and effect relationship of MD deformation with site geology.
PSI technique has become an extremely important tool for monitoring deformation in engineering structures, as it provides a robust estimation of displacement parameters [61
] that can be utilized as an early warning system. The PSI method requires a large number (at least 30) of SAR images [64
], which nowadays are readily available through the Sentinel-1 mission. The availability of frequent repeat cycles of Sentinel-1A SAR data for use in the PSI method for identification of potential instability of strategic structures such as the MD is a basic requirement.
shows the distribution of 96 PSI points in the study area. It was found that the maximum subsidence occurred between PSI-1 and PSI-6, which are located in the center of the Dam, and above the old course of the Tigris River channel (Area A). Subsidence velocity decreases as one moves away from PSI-1. The X-coordinate profile in Figure 18
confirms progressive decrease in subsidence velocity away from PSI-1. The X-coordinate profile shows a good polynomial fit between the velocity of displacements in the PSI points and their X-coordinates. The best fit curve depicted in red represents a deformation pattern of the MD. The graph in Figure 19
, which shows the relationship between the distance of each PSI point measured from PSI-1 (X-axis) and the computed velocity of PSI (Y-axis), clearly indicates a strong and direct correlation between them.
The PSIs selected have a global minimum of the optimized criteria, where the sigma of velocity is minimum (< ±0.7 mm·yr−1
) and the coherence is maximum (>0.8) [47
]. The sigma of the velocity in our case for small area analysis is reasonable [66
]. Although sigma of the velocity for PSIs and the velocity of the displacements have a linear trend, it seems that sigma of the velocity slightly increased towards the LOS direction (Figure 12
shows the difference in velocity (error) and the distance between PSI and the 10 GNSS stations for validation of PSI measurement points. Generally, the error increases as the distance between the GNSS station and the PSI increases. Therefore, the absence of GNSS stations in the same location of the PSIs is one of the major causes of the difference in the velocity values between the PSIs and GNSS stations.
Despite the lack of any supporting geophysical and geological measurements of the area under the old course of the Tigris River, we believe that a major karst feature or subsurface channel exists below area (A) that has caused subsidence under this specific area (Figure 14
). Accordingly, the displacement time curve shows a concave shape (Figure 17
), with the old course of the Tigris River located at the bottom of the curve. It can also be observed that both sides of the dam are stable, and the velocity of the PSI points are either neutral (zero) or positive. This indicates that solutioning is more pronounced under the old course of the Tigris River than elsewhere. The presence of the Sinjar-Dohuk-Amadiya-Fault beneath the MD [40
] contributed to the development of cracks and channels in rock formations that accelerated the karstification process. In addition, Butma East anticline plunges very close to the MD, (Figure 4
) providing a pathway for movement of groundwater toward the MD foundation [43
Subsurface karstification is a significant geological problem in civil engineering projects that leads to differential settlement of dam and its eventual failure. The measurement of deformation patterns might be used as a warning of potential safety problems in dams and other infrastructures. It also helps decision makers in adopting appropriate measures to prevent losses [67
]. The safety problems assume critical importance for earth-fill dams constructed on soluble evaporitic rocks. In the western U.S. for example, 60 % of earth dams higher than 15 m have failed due to the process of piping and karstification in soluble evaporitic sediments [3
]. The presence of the thickest gypsum layer under the old course of the Tigris River (Figure 7
) resulted in the formation of large-size voids that seem to be interconnected due to solutioning activities, providing an easy outlet for water discharge. In addition, the maximum hydrostatic pressure is recorded in the vicinity of the old course of the Tigris River because this area has the lowest elevation in the Mosul Reservoir. Based on these observations, we conclude that area A (Figure 11
), which has the fastest subsidence velocity at the MD and includes PSI-1, is located on the thickest layer of gypsum.
B and Figure 17
show that the displacement along the old course of the Tigris River, which was less in 2014, has been increasing with time, with significant increases occurring in 2019. Maximum precipitation in the study area occurs during the winter months, specifically between December and April (Figure 3
]. The displacement curves in Figure 14
clearly show higher subsidence between March and April (peak to the downward) than the general trend for the years 2015, 2016, 2017, and 2018. Therefore, we recommend that grouting prior to or during the precipitation season should be increased to maintain safety of the MD.
Accelerated subsidence may also be the result of termination of grouting in the MD that occurred during the capture and control of the MD by ISIS. We believe that the hiatus in grouting for several months has allowed the subsidence beneath the MD to increase rapidly. Hence, a long and continuous grouting program will be needed to restore the situation to what it was before 2014. While grouting is definitely important for MD safety, lowering the water level in Mosul Reservoir is also important for controlling dissolution rate.
Variation in subsidence in the same year is due to higher hydrostatic pressures caused by greater storage of water in the reservoir (also evident from increase of water level observed by the intensity scenes), which, in turn, results in increased dissolution of gypsum layers. As seen in Figure 14
B and Figure 17
, there was a slight decrease in MD subsidence from April to October of 2017, which is attributed either to the resumption of grouting or decrease of water level in the Mosul Reservoir.
Milillo et al. [21
] monitored the MD until March 2016. This research that studied subsidence at the MD confirms our results. However, the velocity of subsidence in our study is slower than those reported by Milillo et al. [21
]. We also emphasize that subsidence increases with time specially for the area A. However, this conclusion is accurate for the periods between 03 October 2014 and 02 April 2017, and between 23 October 2017 and 29 December 2018. subsidence at MD after 02 April 2017 and until 23 October 2017 showed a slight decrease due to the resumption of grouting (Figure 14
In order to maintain safety, we strongly recommend the continuation of a grouting program and maintaining the water level at 319 m (a.s.l.) in the Mosul Reservoir. Increasing the water level beyond 319 m will speed up failure of the MD. Figure 17
shows the influence of higher water levels on increase in subsidence, which was clearly seen during the winter of 2014–2015 and 2016–2017, fall 2015 and 2016, and spring 2016 and 2017. Besides evaluating other frequencies such as X-band, further studies should be planned to determine the relationship between water level fluctuations and displacement of MD. At the same time, other ways to store Tigris River water should be explored to ensure safety of the MD.
This study demonstrates the successful application of the Persistent Scatterer Interferometry (PSI) approach for mapping unstable areas and determining geo-hazard risks at Mosul Dam. Seventy-eight Single Look Complex (SLC) scenes from Sentinel-1A, in ascending geometry, acquired from 03 October 2014 to 27 June 2019 were processed using PSI that offered an accurate estimate of dam deformation. Ninety-six PSI measurement points have been used. The maximum subsidence velocity in MD was at PSI-1 (−7.4 mm·yr−1). The area facing the powerhouse, designated subsidence area A, including PSI-1 to PSI-6, had a mean subsidence velocity of about −6.27 mm·yr−1. Area A includes thick gypsum layers, where solutioning activities have created larger voids due to the presence of the Sinjar-Dohuk-Amadiya-Fault under the MD foundation. This area also accounts for the maximum hydrostatic pressure due to its lowest topographic elevation in the Mosul Reservoir, causing greater subsidence than elsewhere in the dam.
Subsidence at the old course of the Tigris River had significantly increased after 2014 due to termination of grouting activities by ISIS, leading to increased dissolution beneath the MD.
In general, maintaining stability and safe operation of MD calls for continuous grouting and a controlled rise of water level in the Mosul Reservoir. Relevant institutions should explore other ways to store water of the Tigris River to meet various needs. Use of other SAR frequencies such as X-band are also encouraged for further studies.