Physically interpretable flood susceptibility mapping using spatial cross-validation and SHAP-derived environmental thresholds: A statistically validated framework
This study presents a statistically validated, physically interpretable framework for flood susceptibility mapping in Nepal's Triyuga River Watershed, utilizing spatial cross-validation and SHAP-derived environmental thresholds to demonstrate that a cumulative indicator of five critical hydro-geomorphic factors significantly enhances predictive accuracy and provides a transparent decision-support tool for risk management.