Remote sensing assessment of the spatio-temporal transformation of the Kakhovka reservoir after dam destruction using Sentinel-2 data

Abstract

Purpose. To identify and characterise the spatio-temporal transformation of the Kakhovka Reservoir and adjacent landscapes after the destruction of the Kakhovka Hydroelectric Power Plant dam on 6 June 2023 using Earth remote sensing data. The study focuses on detecting changes in the open water surface, identifying the spatial structure of the exposed reservoir bed, and assessing early vegetation colonisation of drained bottom sediments under conditions where field surveys are limited or impossible due to active hostilities, mine contamination and safety risks.

Methods. Frequency distribution analysis of pixel values and visual interpretation of natural-colour imagery were used as additional tools to support the interpretation of spectral changes.

Results. The study is based on low-cloud multispectral Sentinel-2 MSI Level-2A imagery for 2023 obtained through the Copernicus Data Space Ecosystem. The Normalized Difference Water Index was used to identify open water surfaces and assess the spatial fragmentation of the reservoir water body. The Normalized Difference Vegetation Index was applied to detect vegetation cover and evaluate early plant colonisation on the exposed reservoir bed. Key temporal scenes before and after the dam destruction were compared to determine the main stages of landscape transformation. The NDWI analysis revealed a rapid reduction and fragmentation of the open water surface of the Kakhovka Reservoir during the first weeks after the dam destruction. The continuous water body that had characterised the reservoir before the disaster was transformed into a narrow Dnipro river channel, a system of isolated residual water bodies, and extensive areas of exposed bottom sediments. The NDVI analysis for the autumn period of 2023 showed the formation of vegetation cover across part of the drained reservoir bed. Positive NDVI values and a shift in the frequency distribution of pixel values toward vegetation-covered surfaces indicate rapid pioneer vegetation colonisation of areas that had been submerged only several months earlier.

Conclusions. The destruction of the Kakhovka dam caused a rapid reorganisation of the lower Dnipro hydroecosystem, with a transition from an artificial reservoir to a complex mosaic of riverine, wetland, meadow-like and exposed substrate areas. Sentinel-2 imagery combined with NDWI and NDVI indices provides an effective, objective and reproducible tool for documenting such transformations when direct field access is restricted. The detected early vegetation colonisation of the exposed reservoir bed has ambiguous ecological implications. On the one hand, it may indicate partial spontaneous restoration of floodplain habitats similar to those that existed before the reservoir was created. On the other hand, it may be accompanied by risks of secondary pollution, dust transport from contaminated sediments, fire hazards and uncontrolled landscape transformation. Further monitoring should combine optical and radar satellite data, multi-temporal NDWI and NDVI series, and, where possible, field validation.