Flood Susceptibility Assessment in the Jhelum Basin Using Machine Learning and GIS Techniques

Flood susceptibility assessment is critical for disaster risk reduction, particularly in flood-prone areas like the Jhelum Basin. This study integrates machine learning techniques, including random forest (RF), artificial neural networks (ANN), support vector machines (SVM), and an ensemble model with geographic information systems (GIS) to map flood susceptibility. Key environmental factors such as topography, hydrology, land use, and climate variables, including slope, elevation, stream power index, topographic wetness index, sediment transport index, terrain roughness index, soil texture, drainage density, normalized difference vegetation index, land use land cover, lithology, slope length factor, and rainfall were analyzed. The information gain ratio method optimized flood conditioning parameters, while historical flood events and environmental factors were used to train and validate the models. Model performance was evaluated using the area under the curve receiver operating characteristic (AUC–ROC). Results show that all models effectively predict flood susceptibility, with the ensemble model outperforming RF (0.828), ANN (0.902), and SVM (0.909) in terms of AUC (0.929), demonstrating superior predictive capability. These findings offer valuable insights into selecting machine learning algorithms for flood mapping. They can inform policymakers and disaster management authorities in developing effective mitigation strategies and emergency response plans for the Jhelum Basin.