Flood Risk Spatial Analysis via Integrating SCS-CN and Logistic Regression Methods (Case Study: Kalateh-ye Qanbar Drainage Basin in Nishabur)

Flood-prone area analysis within drainage basins plays a crucial role in flood risk management and planning. This study evaluated the performance of the Logistic Regression (LR) model and the Soil Conservation Service–Curve Number (SCS-CN) method for flood forecasting and zoning. A total of ۱۷ criteria were incorporated, including elevation, slope, aspect, plan and profile curvature, precipitation, drainage density, distance to rivers, lithology, land use/cover, soil type, Normalized Difference Vegetation Index (NDVI), Slope Length Factor (SLF), Stream Power Index (SPI), Topographic Position Index (TPI), Terrain Roughness Index (TRI), and Topographic Wetness Index (TWI). In the SCS-CN model, groundwater infiltration (S) and surface runoff (Q) were estimated, while the Analytical Hierarchy Process (AHP) was applied to assign weights to the factors. Flood hazard maps for ۵, ۱۵, ۲۵, and ۵۰-year return periods were then generated based on weighted layers. The Receiver Operating Characteristic (ROC) curve was used to validate zoning results against the LR model. Findings indicated that the CN model outperformed LR, achieving RMSE = ۰.۱۴۳, MSE = ۰.۰۲۱, AUC = ۰.۹۲۱, and STD = ۰.۳۱۱. Within the LR model, lithology, proximity to rivers, and NDVI were the most influential predictors, while in the CN model, aspect, drainage density, precipitation, and land use were also dominant factors. Results further showed that ۴۵–۵۰% of the basin could be classified as having medium to high flood potential over the ۵ to ۵۰-year periods, particularly in downstream reaches and riverbeds. Given the concentration of settlements and agricultural activities, these areas represent critical zones for flood crisis management and mitigation.

Keywords:

Flood , zoning , CN model , ROC Curve , Logistic regression , risk