华南地质2026，Vol.42Issue(3)：542-552,11.DOI:10.3969/j.issn.2097-0013.2026.03.014
基于信息量和支持向量机耦合模型的滑坡地质灾害易发性评价
Landslide Geological Hazard Susceptibility Assessment Based on the Coupling Model of IV and SVM:A Case Study of Renhuai City
摘要
Abstract
[Objective]Conducting regional landslide susceptibility assessment is of great significance for sup-porting disaster prevention and mitigation as well as territorial space planning in Renhuai City.[Methods]Ta-king Renhuai City as the research object,the Pearson correlation coefficient was first used for variable screening,and the tolerance(TOL)and variance inflation factor(VIF)were combined to conduct multico-llinearity diagnosis.Finally,nine factors including topographic slope,aspect,elevation,engineering rock group,distance to geological structure,distance to water system,distance to road,annual average rainfall,and normalized vegetation index(NDVI)were determined as the core indicators for landslide susceptibility evaluation.On this basis,a coupled model of information value(IV)and support vector machine(SVM)(IV-SVM)was constructed.Relying on the physical interpretability advantage of the IV model in the quantitative assignment of environmental factors,combined with the strong learning ability of SVM in handling nonlinear relationships and high-dimensional data,a comparative analysis of landslide susceptibility evaluation was conducted using three models(IV,SVM,and IV-SVM).[Results](1)The susceptibility distribution patterns generated by the three models(IV,SVM,and IV-SVM)have a high spatial consistency with the spatial distribution of historical landslide points;(2)The AUC values of the IV model and the SVM model are 0.766 and 0.851 respectively,The AUC value of the IV-SVM coupled model reached 0.881,which was approximately 15%higher than that of the single IV model,and the discrimination ability was significantly improved.[Conclusion]The IV-SVM coupled model has higher discrimination accuracy and is more suitable for regions with complex geological conditions and strong human interference,such as Renhuai City.It can accurately identify high-susceptibility areas for landslides and provide scientific basis and technical support for regional territorial spatial planning,disaster risk prevention and control,and emergency management.关键词
滑坡/信息量模型/支持向量机模型/易发性评价/黔北Key words
Landslide/Information Volume Model/Support Vector Machine Model/Susceptibility Evaluation/Northern Guizhou分类
天文与地球科学引用本文复制引用
杨毅,江攀和..基于信息量和支持向量机耦合模型的滑坡地质灾害易发性评价[J].华南地质,2026,42(3):542-552,11.基金项目
贵州省地质局地质科研项目(黔地质科合[2025]14号) （黔地质科合[2025]14号）
