Machine Learning based Flood Susceptibility assessment in the West Rapti River Basin using Multi Sourced Geospatial Data
TL;DR
Variable importance analysis showed that distance from the river is the most influential factor, and the resulting flood susceptibility maps indicate that approximately 30% of the basin is highly susceptible to flooding, primarily in low lying areas near river channels.
Abstract
Floods are one of the most destructive natural hazards, causing significant human loss. Identification of flood-prone areas is a substantial need for effective planning and risk management. This study aims to generate flood susceptibility maps for the West Rapti River Basin incorporating three machine learning models: Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN). Among the models, XGBoost achieved the highest predictive performance with an accuracy of 0.9706, followed by RF (0.9510) and ANN (0.9412). Variable importance analysis showed that distance from the river is the most influential factor. The resulting flood susceptibility maps indicate that approximately 30% of the basin is highly susceptible to flooding, primarily in low lying areas near river channels. The findings demonstrate the effectiveness of machine learning approaches in flood susceptibility assessment. This study provides valuable insights for disaster risk reduction and early warning systems in the West Rapti River Basin.