Beyond single techniques: can a hybrid MCDM approach re-define flood risk mapping and decision-making in Sub-Himalayan North Bengal?

Flood susceptibility mapping in the Sub-Himalayan piedmont of North Bengal is difficult because flood footprints shift and multiple drivers interact. We propose a transparent GIS-based hybrid MCDM ensemble that integrates TOPSIS, VIKOR, EDAS and PROMETHEE-II after min-max normalization and direction-of-influence harmonization (higher values indicate higher susceptibility). To retain interpretability, the four model outputs are equally weighted, while PROMETHEE-GAIA provides post-hoc insight into factor contributions. Validation using an independent multi-event flood inventory shows improved discrimination and calibration over single methods (AUC = 0.80, 95% CI 0.76–0.82; Brier = 0.200; DeLong significant). Five susceptibility classes identify High-Very High hotspots along the Torsa-Raidak-Jaldhaka corridors and low-lying confluences, with >70% of High-Very High cells within ~2 km of major channels. Tufanganj I-II and Mathabhanga I emerge as priority blocks for screening and mitigation. Polar-plot diagnostics complement probability metrics, supporting corridor-focused drainage maintenance in data-limited Himalayan piedmonts.