Machine Learning-Based Geological Hazards Susceptibility Mapping in Anxi County, Fujian Province

This study focused on Anxi County, Fujian Province, to evaluate and compare the performance of five widely used machine learning models in geological disaster susceptibility mapping. The objective was to establish a scientific foundation for geological hazard mitigation and regional land-use planning. A diverse set of geospatial and environmental datasets was incorporated. Training and validation datasets were systematically constructed, and the predictive capability of each model was rigorously evaluated using multiple performance indicators. The results revealed that the extra trees model outperformed the others across all metrics, with a particularly strong performance in the F1 score. In light of these results, this study employed the extra trees model to generate a geological hazard susceptibility zoning map for Anxi County, Segmenting the study area across five distinct hazard levels: extremely high, high, medium, low, and extremely low. These findings demonstrate the robustness established by ensemble learning approaches when applied to predict geological hazard susceptibility in complex geological conditions, making rich methodological contribution to landscape planning and geological risk assessment.