Identification of Destroyed and Damaged Residential Buildings in the City of Mariupol Using Sentinel-2 Satellite Imagery

The full-scale war in Ukraine has caused extensive destruction of residential areas, creating an urgent need for rapid damage assessment using remote sensing techniques. The aim of this study is to assess the extent of destruction and damage to residential buildings in the city of Mariupol using Sentinel-2 satellite imagery. The study area covers the territory of Mariupol, which experienced severe destruction during the military operations in 2022. Sentinel-2 satellite images acquired before the full-scale invasion and after the end of the active phase of hostilities were used for the analysis. Data preprocessing included clipping the imagery to the study area boundary and generating multispectral composites. Damaged areas were identified through supervised image classification using a training dataset developed from the spectral characteristics of the main land cover classes. Subsequent spatial analysis made it possible to identify residential buildings located within the detected damage zones and to estimate their number by the city's administrative districts. The results indicate that, as of May 8, 2022, a total of 21,243 destroyed or damaged residential buildings were identified, accounting for 45.89% of the total residential building stock. The Left Bank and Prymorskyi districts experienced the most severe destruction. Comparison of the obtained results with official data confirmed their overall agreement regarding the scale of the damage. The proposed methodology can be applied for rapid monitoring of the consequences of military operations, damage assessment, and decision support during the recovery of affected areas using open-access satellite imagery.