Unsupervised Detection of Building Destruction during War from Publicly Available Radar Satellite Imagery

Automated detection of building destruction in conflict zones is crucial for human rights monitoring, humanitarian response, and academic research. However, existing approaches rely on proprietary satellite imagery, require labeled training data that are often unavailable in war-affected regions, or depend on optical imagery obstructed by cloud cover. This study addresses these challenges by introducing an unsupervised method to detect building destruction using freely available Sentinel-1 synthetic aperture radar (SAR) images from the European Space Agency (ESA). By statistically assessing interferometric coherence changes over time, unlike existing approaches, our method enables the detection of destruction from a single satellite image, enabling near real-time assessments. We validate our approach across three case studies, Beirut, Mariupol, and Gaza, demonstrating its ability to capture both the spatial patterns and exact timing of destruction. Using open-access data, it offers a scalable, global, and cost-effective solution for detecting building destruction in conflict zones.