研究降雨与区域地质灾害风险的关系，构建湟水流域地质灾害气象预警模型，为该流域地质灾害预警工作提供参考。以青海省湟水流域为例，利用信息量法进行地质灾害易发性评价，采用皮尔逊相关系数对各因子进行独立性检验并利用受试者特征曲线（receiver operating characteristic curve, ROC）与曲线下的面积（area under curve, AUC）对评价结果进行精度评定；结合降雨资料，利用逻辑回归方法建立地质-降雨耦合模型开展地质灾害气象风险预警，并进行验证。结果表明：湟水流域地质灾害的极高易发区与高易发区主要分布在河谷平原地带及其各级支流中下游两侧，该区域是地质灾害多发区；对易发性评价结果进行精度评定，AUC为0.785；地质-降雨耦合预警模型中的易发性值、当日降雨量、前1 d降雨量、前2 d降雨量及前3 d降雨量对灾害的影响均为正值，模型对地质灾害发生的预测准确率为84.2%，AUC为0.864；基于2020年8月29日湟水流域发生的5起典型地质灾害，对建立的地质-降雨耦合模型进行验证，发现以地质-降雨耦合模型预警，4起地质灾害达到一级预警等级，另外1起达到二级预警等级；结合预警雨量阈值，利用该模型实现对2024年9月5日红崖沟村山体滑坡的预警。可见所构建的湟水流域地质灾害气象预警模型预警效果良好，预测准确率较高，能为相关部门提供防灾减灾参考。

In order to study the relationship between rainfall and regional geological disaster risk, construct the meteorological warning model of geological disasters in Huangshui Basin, and provide reference for the geological disaster warning work in the basin. Huangshui Basin in Qinghai Province was taken as an example, where the susceptibility to geological hazards was evaluated using the information value method. The independence of each factor was tested using the Pearson correlation coefficient, and the accuracy of the evaluation results was assessed through the receiver operating characteristic curve (ROC) and the area under the curve (AUC). Subsequently, a geological-rainfall coupling model was also established using the logistic regression method, incorporating rainfall data, to carry out geological disaster meteorological risk warning and verification. The results show that Huangshui Basin geological disasters of very high susceptibility area and high susceptibility area is mainly distributed in the river valley plain zone and its tributaries at all levels in the middle and lower reaches of the two sides, the region is a geological disaster area. The results of the susceptibility evaluation are assessed for precision, and an AUC value of 0.785 is found. Geological-rainfall coupled warning model in the susceptibility value, the day rainfall, the first 1 day rainfall, the first 2 days day rainfall, the first 2 days of rainfall and the first 3 days of rainfall on the disaster are all positive, the model's prediction accuracy for the occurrence of geological disasters is 84.2%, and the AUC value is 0.864. Based on the five typical geological disasters that occurred in Huangshui Basin on 29th August 2020, the established geological-rainfall coupling model was verified, and it is found that with the geological-rainfall coupling model warning, four geological disasters reach the first warning level, and the other one reaches the second warning level. Combined with the warning rainfall threshold, the model was used to achieve the early warning of the landslide in Hongyagou Village on 5 September 2024. It can be seen that the constructed meteorological early warning model for geological disasters in the Huangshui Basin has a good early warning effect and a relatively high prediction accuracy rate, which can provide references for disaster prevention and mitigation for relevant departments.