Pre-collapse spaceborne deformation monitoring of the Kakhovka dam, Ukraine, from 2017 to 2023

The Kakhovka Dam, a crucial component of the Dnieper reservoir cascade in Kherson Oblast, Ukraine, was completed in 1956. Its primary functions included hydroelectric power generation, irrigation, and navigation. The dam's reservoir had a capacity of 18 cubic kilometers, supplying water for the Zaporizhzhia Nuclear Power Plant and irrigation in southern Ukraine and northern Crimea. The dam's structure consisted of a central section (overflow dam, hydroelectric power station, and navigation lock) connected to two earth dams. The Russian invasion of Ukraine led to observable damage to the dam, documented in satellite imagery from November 2022, showing new damage after Russia's withdrawal from Kherson. Further damage was observed on June 2nd, 2023, including road damage to the bridge. The catastrophic collapse of a substantial portion of the dam on June 6th, 2023, while under Russian control, was extensively documented through satellite optical and radar images, prompting investigations into the cause. While the cause remains debated (with theories ranging from intentional destruction to structural failure due to damage or increased water levels), evidence points toward an explosion as the primary cause of the collapse. This study employs advanced spaceborne monitoring techniques, specifically MT-InSAR, to investigate the dam's condition prior to the collapse, seeking to identify any pre-existing signs of distress.

Numerous studies have demonstrated the effectiveness of space-geodetic monitoring, particularly MT-InSAR, for assessing the stability of various infrastructure types, including bridges, levees, and dams. Previous research has highlighted the ability of MT-InSAR to detect subtle ground movements indicative of structural weakening and potential failures. Studies on other dams, both in Ukraine and internationally, have shown the correlation between InSAR-detected deformations and hydrological factors, thermal effects, and other potential stress mechanisms. The use of remote sensing methodologies in conflict zones to monitor critical infrastructure has also been established in several studies, showing its importance in obtaining vital information under challenging circumstances. The application of InSAR techniques in the case of the Mosul dam collapse, for example, showed the potential to detect pre-collapse movements and to provide critical information for risk assessments.

This study utilizes two Sentinel-1 InSAR datasets, comprising ascending and descending tracks, to monitor the Kakhovka Dam's deformation. The ascending track included 180 images acquired between July 2017 and June 2023, while the descending track comprised 226 images from November 2015 to June 2023. The MT-InSAR processing involved several steps: (1) initial data processing using SARPROZ software to generate interferograms; (2) selection of persistent scatterers (PSs) based on amplitude stability; (3) phase unwrapping to retrieve the displacement time series; (4) decomposing the line-of-sight (LOS) deformation into vertical and perpendicular-to-dam-wall (PDW) components; (5) removal of the thermal dilation component from the InSAR deformation signal to highlight anomalous trends using a linear model fitted to InSAR deformation and weather temperature data; and (6) analysis of correlation between water levels and the PDW component using Hydroweb River Water Level products and Visual Crossing weather data. The study focused on five defined polygons on the dam, analyzing the spatial and temporal patterns of deformation within each.

The analysis revealed distinct deformation patterns across the five polygons. Polygons I and III showed remarkable stability with minimal deformation throughout the study period. In contrast, polygons II, IV, and V exhibited significant changes, particularly from June 2021. Polygon V showed a linear deformation trend from 2015, accelerating significantly in June 2021. Polygons II and IV, located over the breached sections, showed the most pronounced deformation. Ascending and descending track data yielded slightly different velocity values, reflecting the combined effects of vertical and horizontal displacements. Decomposition of LOS deformation revealed subsidence (downward movement) starting in June 2021, with eastward lateral (upstream PDW) movements beginning in June 2022. Notably, a negative correlation was found between water level fluctuations and PDW deformation components, particularly in polygons II and IV. This correlation remained consistent even in the days leading up to the collapse. Similar negative correlations were observed in other dam sections, though the road segment between polygons II and III showed no significant correlation with water levels. The observed vertical and horizontal displacements are consistent with potential damage mechanisms such as overtopping, faulty gate operation, and sediment/debris build-up, all potentially affecting dam foundations.

The findings indicate that the Kakhovka Dam was experiencing significant subsidence and lateral movement in the years leading up to its collapse. The detected deformations, particularly the acceleration of subsidence in June 2021, coupled with the negative correlation between water levels and PDW movement, strongly suggest pre-existing structural weaknesses. While the InSAR data cannot confirm the explosion theory, the detected deformations provide evidence of potential contributing factors that could have compromised the dam's integrity. Neglect of dam maintenance and operation during the war might have exacerbated these pre-existing weaknesses. These findings are consistent with the hypotheses suggesting structural damage or lack of maintenance leading to the collapse, adding another layer of understanding to the event.

This study successfully utilized MT-InSAR to detect and quantify pre-collapse deformation at the Kakhovka Dam. The analysis revealed significant subsidence and eastward lateral movements starting in June 2021 and June 2022, respectively. These movements were negatively correlated with changes in water levels. The findings highlight the potential for InSAR as a powerful tool for proactive monitoring of critical infrastructure. Future research could explore the integration of InSAR data with other datasets (e.g., hydrological, meteorological, and structural health monitoring data) to develop more comprehensive models for dam stability assessment and risk management.

The study relies on the availability and quality of InSAR data. The accuracy of the deformation measurements depends on the density and quality of the PSs. The analysis primarily focuses on deformation patterns and does not directly assess the cause of the collapse. Other factors, beyond the scope of the InSAR analysis (e.g., material properties, construction details, specific maintenance activities), could have contributed to the dam's failure.