Ensemble models of GLM, FDA, MARS, and RF for flood and erosion susceptibility mapping: a priority assessment of sub-basins

The mountainous watersheds are increasingly challenged with extreme erosions and devastating floods due to climate change and human interventions. Hazard mapping is essential for local policymaking for prevention, planning the mitigation actions, and also adaptation to extremes. This study proposes novel predictive models for susceptibility mapping for flood and erosion. Furthermore, this study elaborates on prioritizing the existing sub-basins in terms of erosion and flood susceptibility. A comparative analysis of generalized linear model (GLM), flexible discriminate analyses (FDA), multivariate adaptive regression spline (MARS), random forest (RF), and their ensemble is performed to ensure highest predictive performance. Furthermore, the priority of the sub-basins in terms of sensitivity to erosion and flood was determined based on the best model. The results showed that the GLM, FDA, MARS, RF, and ensemble models had an area under curve (AUC) 0.91, 0.92, 0.89, 0.93 and 0.94, respectively, in modeling the flood susceptibility. Also, the GLM, FDA, MARS, RF, and ensemble models had an AUC of 0.93, 0.92, 0.89, 0.96, and 0.97, respectively, in determining erosion susceptibility. Priority assessment based on the best model, the ensemble approach, indicated that the sub-basins SW3 and SW5 were found to have high sensitivity to the flood and soil erosion, respectively.