山东建筑大学学报2025，Vol.40Issue(1)：62-70,9.DOI:10.12077/sdjz.2025.01.008
融合地理探测器和随机森林的地质灾害易发性评价
Geological hazards susceptibility evaluation based on geodetector and random forest:Taking Xiangning County of Shanxi Province as an example
摘要
Abstract
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Key words
geological hazards/susceptibility evaluation/geodetector/random forest/information quantity分类
资源环境引用本文复制引用
刘晓慧,张金雨,徐玲..融合地理探测器和随机森林的地质灾害易发性评价[J].山东建筑大学学报,2025,40(1):62-70,9.基金项目
山东省高校科技计划项目(J16LH05) （J16LH05）
山东省自然科学基金项目(ZR2016DQ06) （ZR2016DQ06）
