Rapid Building Damage Assessment Following the M7.7 Myanmar Earthquake: A Causal Bayesian Approach Using InSAR Imagery

On March 28, 2025, a devastating 7.7 magnitude earthquake struck Myanmar's Sagaing Region near Mandalay, killing approximately 5,350 people, injuring 7,860, and leaving hundreds missing. Our research presents comprehensive city-scale mapping of earthquake effects using a novel variational causal Bayesian network that integrates Interferometric Synthetic Aperture Radar (InSAR) derived change detection with empirical ground failure models and building footprints. This innovative approach enables four critical capabilities: (1) rapid large-scale estimation of building damage and ground failures from remote sensing, (2) causal attribution of building damage to specific failure mechanisms, (3) enhanced prediction of regional infrastructure-impacting landslides and liquefaction, and (4) fine-grained damage assessment across thousands of buildings simultaneously. By utilizing 30-meter resolution satellite imagery and building footprints, we achieved building-by-building damage classification within days of the earthquake. Our five-category damage classification system (no damage, slight, moderate, partial collapse, and collapse) was validated against high-resolution Maxar optical imagery, demonstrating exceptional accuracy in identifying damage patterns across the affected region. This work represents a significant advancement in rapid post-disaster damage assessment, providing critical information for emergency response and recovery planning following major seismic events.