Determining Potential Areas for Flood Inundation Using Remote Sensing and Artificial Neural Networks

This study aims to analyze the morphological indices of river channels in Shiraz, located in southwestern Iran, through the examination of morphological parameters and the application of remote-sensing (RS) indices to delineate flood zones. RS indices, including the automated water extraction index, superfine water index (SWI), water ratio index, normalized difference vegetation index, and normalized difference water index, were derived from Sentinel-1 and Sentinel-2 satellite imagery. These indices were used to delineate flood zones in the study area before and after the flood event on March 11, 2019, and April 9, 2019. Additionally, a correlation coefficient matrix was computed between various Sentinel bands and the optimum index factor to analyze the main river within the study area. Subsequently, to evaluate the status of flood-prone river channels, morphological parameters such as river sinuosity, the central angle of rotation, and the longitudinal profile of the main river were examined. Finally, a multilayer perceptron (MLP) model was employed to predict the most effective remote-sensing index based on the river’s morphometric characteristics. Our principal results demonstrate that the SWI was superior for flood inundation zones, achieving an R2 of 0.89. The morphometric analysis revealed that the river system is highly susceptible to flooding, with 66.57% of channels exhibiting meandering configurations and 66.6% of the area being at high risk. Furthermore, the MLP model (with eight neurons in the hidden layer and a single output neuron) successfully predicted SWI values with exceptional accuracy (R2=0.98). The main conclusion of this study is that Shiraz’s river morphology, marked by high sinuosity and a very gentle slope, is the key driver intensifying floods within the city limits. This integrated framework offers a strong, practical tool for proactive flood risk assessment and urban planning in comparable semiarid catchments.