【摘要】 为有效降低地质灾害风险，本文采用地理探测器、多种机器学习模型，构建“因子诊断-主控识别-多模型对比”县域地质灾害易发性评价技术路径，并以湘潭县为例开展实证研究。结果表明：构造、坡度、工程岩组及剖面曲率是湘潭县地质灾害的核心主控因子，因子间交互作用呈显著放大效应，其中以构造与平面曲率的耦合作用最为突出；SVM(线性核函数)为最优预测模型，其受试者工作特征曲线下面积(AUC)达0.823,综合预测精度最优；湘潭县地质灾害易发性呈现“西南-东北高、中心-西北低”的空间异质性特征，极高易发性区域集中分布于南部、东南部及东北部的丘陵山地带。本文构建的评价技术路径兼具科学性与实用性，可为县域地质灾害风险管理及国土空间规划提供技术支撑与决策参考。更多还原

【Abstract】 In order to effectively reduce the risk of geological disasters, in this paper, sGeodetector and multiple machine learning models were employed to construct a technical pathway for the susceptibility assessment of geological disasters at the county-level, which involves “factor diagnosis, dominant controlling identification, and multi-model comparison”. An empirical study is conducted using Xiangtan County as a case study. The results show that tectonics, slope gradient, engineering rock group, and profile curvature are the core controlling factors for geological disasters in Xiangtan County. The interaction among these factors exhibits a significant amplifying effect, with the coupling effect between tectonics and plan curvature being the most prominent. The SVM(linear kernel function) emerged as the optimal prediction model, with an Area Under the Receiver Operating Characteristic Curve(AUC) reaching 0.823, indicating demonstrating the best overall predictive accuracy. The susceptibility of geological disasters in Xiangtan County shows spatial heterogeneity characteristics of “high in the southwest and northeast, low in the central and northwest regions”, with extremely high susceptibility areas concentrated in the hilly and mountainous zones of the south, southeast, and northeast. The extremely high susceptibility areas are concentrated in the hilly and mountainous regions in the south, southeast, and northeast. The evaluation technical pathway established in this paper is both scientific and practical, providing technical support and decision-making references for the geological disaster risk management at the county-level and territorial spatial planning.更多还原