Improving flood hazard susceptibility assessment by integrating hydrodynamic modeling with remote sensing and ensemble machine learning

Floods are natural disasters with significant economic and infrastructural impacts. Assessing flood susceptibility in mountainous urban regions is particularly challenging due to the complicated interaction in which urban structures and mountainous terrain affect flood behavior. This study employs two ensemble machine learning algorithms, Extreme Gradient Boosting (XGBoost) and Random Forest (RF), to develop flood susceptibility maps for the Hunza-Nagar region, which has been experiencing frequent flooding for the past three decades. An unsteady flow simulation is carried out in HEC-RAS utilizing a 100-year return period flood hydrograph as an input boundary condition, the output of which provided the spatial inundation extents necessary for developing the flood inventory. Ten explanatory factors, including climatic, geological, and geomorphological features namely elevation, slope, curvature, topographic wetness index (TWI), normalized difference vegetation index (NDVI), land use land cover (LULC), rainfall, lithology, distance to roads and distance to rivers are considered for the flood susceptibility mapping. For developing flood inventory, random sampling technique is adopted to create a spatial repository of flood and non-flood points, incorporating the ten geo-environmental flood conditioning factors. The models’ accuracy is assessed using the area under the curve (AUC) of receiver operating characteristics (ROC). The prediction rate AUC values are 0.912 for RF and 0.893 for XGBoost, with RF also demonstrating superior performance in accuracy, precision, recall, F1-score, and kappa evaluation metrics. Consequently, the RF model is selected to represent the flood susceptibility map for the study area. The resulting flood susceptibility maps will assist national disaster management and infrastructure development authorities in identifying high flood susceptible zones and carrying out early mitigation actions for future floods.