Assessing flood susceptibility in a Triyuga watershed, Nepal using statistical models.

Floods are among the most damaging natural disasters, posing significant threats to socio-economic stability and environmental sustainability. This study addresses an important research gap by evaluating flood susceptibility in a small watershed (< 500 km), where no detailed susceptibility mapping has been conducted before. Flood susceptibility in the Triyuga Watershed, Nepal, was evaluated using three statistical models: Frequency Ratio (FR), Logistic Regression (LR), and Weight of Evidence (WoE), and the distinct hydrological behaviours of small watersheds were highlighted. A flood inventory map was developed from field surveys, identifying 190 flood and non-flood locations, with 70% allocated for training and 30% for validation. Eleven influential factors: LULC, distance from river, slope, flow direction, profile curvature, rainfall, DEM, TPI, TWI, NDVI, and aspect, were selected with no multicollinearity among them. The results revealed that: (1) the LR model exhibited the highest predictive accuracy with an AUC of 0.89, the lowest Brier Score (0.1186), and the highest Brier Skill Score (0.5254); (2) both the WoE and FR models also showed strong performance with AUC values of 0.85 and competitive Brier Scores and BSS values; and (3) the LR model's ability to handle multiple predictors simultaneously and capture complex relationships likely contributed to its superior performance, as reflected by its higher AUC and more favourable Brier validation metrics. These findings offer valuable insights for flood risk management and emphasize the necessity of precise flood susceptibility mapping to guide disaster preparedness and sustainable land-use planning.