五峰集镇地质环境背景条件复杂，特别在汛期降雨影响下滑坡地质灾害频发，因此开展五峰集镇滑坡空间预测评价和降雨阈值研究，不仅对研究区滑坡的防灾减灾具有实际指导意义，同时对基于降雨阈值的滑坡危险性评价方法研究也具备较好的参考价值。本文以降雨型滑坡多发的五峰集镇为例，利用地理探测器法准确选取研究区评价因子，综合层次分析模型与 BP 神经网络模型计算全区易发性指数，得到基于斜坡单元的滑坡易发性分区；同时，统计每个滑坡点的降雨历时及有效降雨强度，分析研究区滑坡的致灾雨型，绘制诱发鄂西山区滑坡灾害发生的临界降雨 I‐D 阈值曲线，得到设计工况下的时间概率；综合易发性分区和时间概率得到基于有效降雨阈值的鄂西山区五峰集镇滑坡危险性分区图。研究结果表明：五峰集镇滑坡高和极高易发区占研究区总面积的 27.12%，主要分布于研究区大型河流两岸；五峰集镇滑坡发生 10%、50%、75%概率的阈值曲线分别为 I = 31.42 × D-0.786 94 、I = 68.11 × D-0.786 94 、I = 84.74 × D-0.786 94 ；五峰集镇滑坡高和极高危险区占研究区总面积的 19.33%，主要分布于研究区中部及东南部河流两岸。本文所构建的五峰集镇滑坡危险性评价体系，以及适用于鄂西山区的基于降雨阈值滑坡危险性评价方法，能够为山区地质灾害防控以及危险性评价研究的不断完善提供有益参考。
Given the complex geological environment conditions， especially for frequent rainfall-induced landslide events during the flood season， the landslide spatial prediction assessment and rainfall threshold in Wufeng town was performed. This study provides for practical guide to the landslide disaster prevention and mitigation in the study area， and the landslide risk assessment method based on rainfall threshold. This paper focused on the Wufeng town that is prone to landslides induced by rainfall. The evaluation factors of the study area were selected by using the geographical detector method. The susceptibility index of the whole area was calculated by combining the AHP with BP neural network models， and the landslide susceptibility zoning based on slope units was obtained. Simultaneously， the rainfall duration and effective rainfall intensity of each landslide were counted， the rainfall regimes that trigger landslide hazards in the study area was thus analyzed， and the I-D threshold curve of critical rainfall triggering landslides in the mountainous area of western Hubei was mapped to obtain the temporal probability under design conditions. The landslide hazard zoning map of the town in western Hubei based on effective rainfall threshold was obtained by combining the susceptibility zoning and time probability. The results show that the high and extremely high landslide prone areas in the town account for 27.12% of the total area， which are mainly distributed on the banks of large rivers in the study area. The threshold curves of 10%，50% and 75% probability of landslide occurrence in the town are I=31.42×D-0.786 94，I=68.11×D-0.786 94，I=84.74×D-0.786 94. The high and extremely high risk areas of landslide in the town account for 19.33% of the total area， which are mainly distributed in the middle and southeast of the study area. The landslide risk assessment system and method based on rainfall threshold proposed in this paper， can provide useful reference for the prevention and control of geological disasters in mountainous areas.