A Flood Risk Analysis Framework Based on Causal Bayesian Networks

In recent years, the intensification of global climate change has exacerbated the severity of flood disasters, particularly urban flooding and rainstorm-induced floods, which pose significant threats to economic development and public welfare. Existing flood analysis methods often suffer from high computational complexity and limited model interpretability. This study proposes a novel flood risk analysis method based on causal Bayesian learning. By analyzing observational data, a Bayesian network is constructed to uncover the latent causal relationships among flood risk factors and quantify their associated uncertainties. Compared to classical flood analysis methods, the proposed approach demonstrates notable improvements in both accuracy and F1 score. Additionally, by identifying the underlying causal relationships between risk factors, Bayesian networks enhance interpretability. The findings highlight the potential of causal Bayesian learning to enhance disaster preparedness and mitigation efforts. This research offers a scalable, data-driven approach to tackling the complex challenges of flood risk management.