Flood Susceptibility Mapping using GIS and Evidential Belief Function Model in Lam Chiang Krai Watershed, Nakhon Ratchasima Province, Thailand

Abstract

Flood events are recognized as highly impactful natural hazards, frequently resulting in considerable economic losses across ecosystems, agriculture, and infrastructure. Identifying areas most susceptible to flooding has become a crucial component of flood mitigation plans. The objective of this study is the application of Geographic Information Systems (GIS) integrated with the Evidential Belief Function (EBF) model to flood susceptibility mapping in the Lam Chiang Krai watershed, Nakhon Ratchasima province, Thailand. A flood inventory map was initially generated, and the data were subsequently partitioned into a training dataset (70%) and a validation dataset (30%) for model development and accuracy assessment. Ten flood conditioning factors, such as elevation, slope, curvature, stream power index (SPI), topographic wetness index (TWI), distance to stream, geology, soil texture, land use and land cover (LULC), and mean annual rainfall, were used as thematic layers in the analysis. The potential redundancy among these variables was examined using multicollinearity analysis. An analysis of the spatial associations between the flood inventory and the explanatory variables was carried out using the EBF model, through which the flood susceptibility index (FSI) was derived. The FSI was classified using the geometrical interval classification scheme into five flood susceptibility classes. The model was validated using the relative operating characteristic (ROC) and area under the curve (AUC). The analysis demonstrated that the AUC values for the success and prediction rates were 0.778 and 0.816, respectively. These results validate the efficacy of the EBF model in accurately generating flood susceptibility maps for the study area.