融合地理探测器和随机森林的地质灾害易发性评价——以山西省乡宁县为例  

Geological hazards susceptibility evaluation based on geodetector and random forest:Taking Xiangning County of Shanxi Province as an example

作　　者：刘晓慧 张金雨 徐玲 LIU Xiaohui;ZHANG Jinyu;XU Ling(School of Surveying and Geo-Informatics,Shandong Jianzhu University,Jinan 250101,China)

机构地区：[1]山东建筑大学测绘地理信息学院,山东济南250101

出　　处：《山东建筑大学学报》2025年第1期62-70,共9页Journal of Shandong Jianzhu University

基　　金：山东省高校科技计划项目(J16LH05);山东省自然科学基金项目(ZR2016DQ06)。

摘　　要：为了全面评价山西省乡宁县地质灾害的易发性,文章利用研究区内144个以地面塌陷和滑坡为主的地质灾害隐患点及地层岩性、地质断层等多源数据,分别构建信息量模型、随机森林加权信息量(Information-Random Forest, I-RF)模型、地理探测器和随机森林加权信息量(GeoDetector-Information-Random Forest, GD-I-RF)模型,开展地质灾害易发性评价。结果表明:研究区内地质灾害沿道路、建筑用地和断层分布特征明显,高易发区主要分布于人类工程活跃的区域和断层附近;信息量模型、GD-I-RF模型和I-RF模型的曲线下面积值分别为0.818、0.842和0.863,加权后的信息量模型评价效果更好;相较于信息量模型和I-RF模型,GD-I-RF模型易发区更接近实际灾害隐患点的空间分布,且可以利用较少的评价因子降低评价过程的计算量,同时提高了评价精度。In order to comprehensively evaluate the susceptibility of geological hazards in Xiangning County,Shanxi Province,this study utilized 144 geological hazard sites mainly consisting of ground subsidence and landslides and multi-source data such as stratigraphic lithology and geologic faults.It constructed the Information quantity model,the Information-Random Forest(I-RF)model and the GeoDetector-Information-Random Forest(GD-I-RF)model respectively to evaluate the susceptibility of geological hazards in the study area.The results revealed that geological hazards in the study area are predominantly distributed along roads,construction zones,and fault lines,with high-susceptibility areas concentrated in regions of active human activity and near geological faults.The Area Under Curve(AUC)values for the Information quantity model,GD-I-RF model and I-RF model were 0.818,0.842 and 0.863 respectively,indicating superior performance of the weighted information model.Compared to the information quantity model and the I-RF model,the GD-I-RF model aligns more closely with the spatial distribution of actual geological hazard sites.Moreover,the GD-I-RF model uses fewer evaluation factors,which can reduce the calculation amount of the evaluation process and improve the evaluation accuracy.

关 键 词：地质灾害 易发性评价 地理探测器 随机森林 信息量 

分 类 号：X43[环境科学与工程—灾害防治]