Flood susceptibility prediction using multi criteria decision analysis and bivariate statistical models: a case study of Lower Kosi River Basin, Ganga River Basin, India

The present work focuses on the comparison between Analytical Hierarchy Process (AHP), the most commonly used Multi-Criteria-Decision-Analysis (MCDA) model, and three bivariate models: Evidential Belief Function (EBF), Weights of Evidence (WoE) and Frequency Ratio (FR) to predict the flood susceptible areas in the Lower Kosi River Basin of the Ganga River Basin. Twelve flood conditioning factors, topographic (altitude, slope, aspect, curvature, and geomorphology), hydrologic (rainfall, TWI, river density), anthropogenic (LULC, distance from road), and others (distance from river, soil) have been utilized for the spatial modelling. The results suggest that geomorphology, TWI, land-use, and river density are the most dominating factors. Area-under-Receiver-Operating-Characteristic (AUROC) method was used for validation of the used models. The value of the AUROC curve for AHP is found 0.837, outstands the bivariate models- FR (0.807), EBF (0.820) and WoE (0.787), suggests the MCDA model is more accurate in predicting the flood lands than bivariate models in the plains of Ganga River Basin.