GENERATION OF FLOOD SUSCEPTIBILITY MAP USING ARTIFICIAL INTELLIGENCE: A CASE STUDY, TABUK, KSA

Abstract

This study focused on flood susceptibility mapping (FSM) for Tabuk, Saudi Arabia, using artificial intelligence and advanced techniques in an effort to find areas that are more prone to flooding. The study thus aims at generating a reliable tool for urban planning and flood risk management in a flash-flood-prone arid region. The acquisition of data involved various sources, among others, Digital Elevation Models (DEM), land use and land cover (LULC), hydrological data (Topographic Wetness Index, Stream Power Index), and noted flood records. Four ML models- Random Forest (RF), Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Decision Trees (DT)-were used to assess the environmental conditions and produce an FSM. They were validated using metrics like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R²) to assess their predictive performance. The results seem rather to show that the low-lying areas, proximity to streamlines, and several topographic features contribute significantly to the flood susceptibility of the town. Shortages of historical flood data are one of the limitations that can provide obstacles to the prediction ability of the models used for flood risk assessments, with no consideration given to socio- economic factors. Recommendations for improvement in the relational modeling for better forecast of flood vulnerability include more accurate data, collecting long-term historical records of flood occurrence, and considering socio-economic factors into integrated flood risk models for providing proper flood management plans.