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

Area of Science:

• Hydrology and Environmental Science
• Natural Hazard Assessment
• Geospatial Analysis

Background:

• Floods are major natural disasters impacting socio-economic and environmental stability.
• Small watersheds often lack detailed flood susceptibility mapping, creating a research gap.
• Understanding localized flood dynamics is crucial for effective disaster management.

Purpose of the Study:

• To evaluate flood susceptibility in the small Triyuga Watershed, Nepal.
• To compare the performance of three statistical models: Frequency Ratio (FR), Logistic Regression (LR), and Weight of Evidence (WoE).
• To highlight the distinct hydrological behaviors relevant to small watershed flood susceptibility.

Main Methods:

• Utilized a flood inventory map from field surveys (190 locations) for training (70%) and validation (30%).
• Selected eleven influential factors (e.g., LULC, slope, rainfall, DEM, NDVI) with no multicollinearity.
• Applied and compared FR, LR, and WoE statistical models for susceptibility mapping.

Main Results:

• The Logistic Regression (LR) model demonstrated superior predictive accuracy (AUC=0.89, Brier Skill Score=0.5254).
• Frequency Ratio (FR) and Weight of Evidence (WoE) models also showed strong performance (AUC=0.85).
• LR's effectiveness is attributed to its ability to handle multiple predictors and complex relationships.

Conclusions:

• Precise flood susceptibility mapping is essential for disaster preparedness and land-use planning.
• The study provides valuable insights for flood risk management in small watersheds.
• Statistical modeling, particularly LR, is effective for assessing flood susceptibility in data-scarce regions.