A novel multi-scenario mitigation model for rainstorm flood disasters

Global warming induces frequent heavy rain disasters, resulting in significant casualties and economic losses. Seeking for an effective flood disaster mitigation method is crucial for mitigating the impact of disasters. Therefore, this study focuses on Nanhai district, starting with an analysis of the key disaster-causing factors for floods. Then, the Gradient Boosting Decision Tree (GBDT) is employed to quantify the contribution rates of disaster-causing factors, and determine the risk levels of the disaster. On this basis, a GBDT-based flood disaster mitigation model (GB-FDMM) is constructed by optimizing key disaster-causing thresholds. It then explores the changes in risk before and after the implementation of mitigation measures, thereby explaining the effectiveness of the FDMM under different heavy rainfall scenarios. The results indicate that: (1) the constructed GB-FDMM with threshold optimization method demonstrates an effective mitigation effect under extreme weather conditions. (2) the very-high-risk areas are mainly concentrated in the eastern region, and the GBDT algorithm can accurately evaluates the risk of flood disasters; (3) the main key disaster-causing factors are flood depth, submerge duration, population density, and GDP density. This study aims to provide valuable theoretical reference for enhancing the overall disaster reduction capacity in similar regions.