Flood susceptibility mapping in river basins: a risk analysis using AHP-TOPISIS-2 N support and vector machine

Due to the damage caused by floods, mapping areas susceptible to this natural phenomenon plays a fundamental role in environmental planning. Therefore, it becomes essential to understand and map the conditions and factors involved in areas affected by geo-hydro-meteorological events. In this context, we mapped areas susceptible to flooding using the AHP-TOPSIS-2 N, Support Vector Machine (SVM), and a hybrid model, AHP-SVM, the Sorocaba-Medio Tiete basin, that is a subtropical, densely populated river basin located in Brazilian territory. We considered 11 conditioning factors related to hydrogeomorphological and anthropological characteristics, and 382 historical flood and non-flood points. We assessed the accuracy of the modeling using the Area Under the Curve – AUC. The AHP-SVM model presented the best efficiency among the models analyzed (AUC = 0.962). The principal conditioning factors related to flooding were land cover and land use. We argue that models can be successfully applied as a scientific tool in the mapping of areas susceptible to flooding by public managers and risk managers since the resultant maps can help mitigate the negative impacts related to the flood event.