程斌, 杨根兰, 覃乙根,等.基于CF-LR模型的典型红层地区地质灾害易发性评价——以赤水市为例[J].河南理工大学学报（自然科学版）,2023,42(6):51-60.

CHENG B, YANG G L, QIN Y G,et al.Disaster susceptibility evaluation in typical red-bed areas based on  CF-LR coupling method：A case study of Chishui City[J].Journal of Henan Polytechnic University(Natural Science) ,2023,42(6):51-60.

基于CF-LR模型的典型红层地区地质灾害易发性评价——以赤水市为例

程斌1, 杨根兰1, 覃乙根2, 黄凯杰1, 史文兵1, 易庆波3, 吴运武4

1.贵州大学 资源与环境工程学院，贵州 贵阳  550025;2.中山大学 地球科学与工程学院 广东省地球动力作用与地质灾害重点实验室，广东 广州  510000;3.中国电建集团贵阳勘测设计研究院有限公司，贵州 贵阳  550081;4.贵州开阳经开产业投资发展有限公司，贵州 贵阳  550300

摘要:为探究地质灾害与红层地区风化调控作用的关系，以贵州省典型红层地区赤水市为研究区，选取太阳辐射、高程、坡度、地层分布、降雨量、NDVI、道路缓冲区和地形起伏度8个影响因子，引入皮尔森相关系数对影响因子进行共线性分析，采用CF模型、LR模型及CF与LR耦合模型在GIS平台上进行地质灾害易发性分析，并利用ROC曲线法对3种模型的精度进行检验。结果表明：3种模型的AUC值分别为0.854，0.875和0.905，其中，耦合模型AUC值最高，说明耦合模型评价精度更高，灾害易发性分区更准确。研究结果可为赤水市地质灾害预防工作与红层地区地质灾害易发性评价研究提供参考。

关键词:确定性系数模型;逻辑回归模型;易发性评价;红层;太阳辐射;皮尔森相关系数

doi:10.16186/j.cnki.1673-9787.2021060082

基金项目:国家自然科学基金资助项目（42067046）；贵州省科技厅研究项目（黔科合基础-ZK［2021］一般200）

收稿日期:2021/06/22

修回日期:2021/08/04

出版日期:2023/11/25

Disaster susceptibility evaluation in typical red-bed areas based on  CF-LR coupling method：A case study of Chishui City

CHENG Bin1, YANG Genlan1, QIN Yigen2, HUANG Kaijie1, SHI Wenbing1, YI Qingbo3, WU Yunwu4

1.College of Resource and Environment Engineering，Guizhou University，Guiyang  550025，Guizhou，China;2.Guangdong Provincial Key Lab of Geodynamics and Geohazards，School of Earth Science and Engineering，Sun Yat-Sen University，Guangzhou  510000，Guangdong，China;3.Power China Guiyang Engineering Corporation Co.，Ltd.，Guiyang  550081，Guizhou，China;4.Guizhou Kaiyang Economic Development Industry Investment and Development Co.，Ltd.，Guiyang  550300，Guizhou，China

Abstract:To explore the relationship between geological hazards and weathering regulation in red-bed areas，Chishui City，a typical red-bed area in Guizhou Province，was selected as the research area.Eight influencing factors were selected，including solar radiation，elevation，slope，stratigraphic distribution，rainfall，NDVI，road buffer and topographic relief，and Pearson correlation coefficient was introduced to analyze the covariance of the selected influencing factors.The CF model，LR model and the coupling model of CF and LR were used to analyze the geological hazard susceptibility on GIS platform，and the accuracy of the three models was tested by ROC curve method.The results show that the AUC values（offline area） of the three models are 0.854，0.875，and 0.905，respectively.Among them，the coupled model has the highest AUC value，indicating higher evaluation accuracy of the coupled model and more accurate zoning of disaster susceptibility.The research results can provide reference for the prevention of geological disasters in Chishui City and the evaluation of geological disaster susceptibility in the red-bed areas.

Key words:deterministic coefficient model;logistic regression model;vulnerability evaluation;red-bed;solar radiation;Pearson correlation coefficient

007_2021060082_程斌_L.pdf