Flood risk and community resilience mapping at the Mouza level: Insights from the 2024 Malda Flood, India

Flooding remains one of the most recurrent and damaging natural hazards in the Lower Ganga Plain, where complex hydrology, low-lying topography, and intense monsoonal rainfall amplify exposure and vulnerability. Despite frequent flood events, a comprehensive understanding of spatial flood vulnerability, risk, and community-level resilience remains limited for highly affected districts such as Malda, West Bengal, India. This study aims to evaluate flood vulnerability, risk, and resilience across six administrative blocks of the Malda district by integrating field survey, satellite-based flood mapping w ith explainable machine learning techniques. An integrated framework was developed using Sentinel-1 SAR imagery to delineate flood inundation during the extreme flood events of August and September 2024, combined with a Random Forest (RF) model to assess flood vulnerability and SHapley Additive exPlanations (SHAP) to interpret predictor importance. Flood risk and resilience were subsequently quantified using hazard, vulnerability, and adaptive capacity indicators. Sentinel-1 data were selected due to their all-weather capability and effectiveness in capturing flood dynamics during peak monsoon conditions. Results indicate that approximately 13.74% (155.12 km²) and 15.04% (169.83 km²) of the study area were inundated during the August and September 2024 flood events, respectively. SHAP analysis reveals that rainfall intensity, maximum rainfall, September rainfall, and flood depth collectively contribute over 65% to flood vulnerability. High flood vulnerability (~19.47%) and risk (~14.74%) are predominantly concentrated in the Manikchak, Ratua-I, and Kaliachak-II blocks due to low elevation, river proximity, and repeated inundation. Resilience assessment shows that nearly 32.72% (369.3 km²) of the area exhibits very low to low flood resilience, resulting in an overall low average resilience index (0.19), primarily driven by weak adaptive capacity. The proposed framework offers a scalable and transferable approach for flood risk and resilience assessment, providing critical insights to support targeted mitigation, adaptive planning, and sustainable flood governance in flood-prone riverine landscapes.