冯显杰, 李益敏, 邓选伦,等.高山峡谷地区地质灾害易发性评价——以怒江州为例[J].河南理工大学学报（自然科学版）,doi:10.16186/j.cnki.1673-9787.2022040062

FENG X J, LI Y M, DENG X L,et al.Evaluation of the susceptibility of geological disasters in the alpine and canyonareas:Taking Nujiang Prefecture as an example[J].Journal of Henan Polytechnic University(Natural Science) ,doi:10.16186/j.cnki.1673-9787.2022040062

高山峡谷地区地质灾害易发性评价——以怒江州为例(网络首发)

冯显杰1, 李益敏2,3, 邓选伦2, 赵娟珍1, 杨一铭2

1.云南大学国际河流与生态安全研究院，云南 昆明 650500；2.云南大学地球科学学院，云南 昆明 650500；3.云南省高校国产高分卫星遥感地质工程研究中心，云南 昆明 650500

摘要:为探究高山峡谷地区地质灾害易发性空间分布状况，以怒江州为研究区，选择高程、坡度、 坡向、曲率、起伏度等12个评价因子，构建区域易发性评价指标体系，采用信息量Gnformation value， IV）模型、信息量-BP 神经网络（information value-back propagation neural networks，IV-BPNN）耦 合模型与信息量-支持向量机（information value-support vector machine， IV-SVM）耦合模型进行地 质灾害易发性评价。结果表明：用实际地质灾害点验证易发性结果具有较好的空间分布一致性；高 与极高易发区呈线状分布，主要集中在怒江、澜沧江、独龙江等水系沿岸地区、道路附近和地质构 造活跃的区域。此外，IV模型、IV-BPNN模型与IV-SVM模型的受试者曲线（receiver operating characteristic curve，ROC）的曲线下面积（areaunder the curve，AUC）分别为 0.884， 0.889，0.901， 即IV-SVM模型精度更高。研究结果可为当地政府制定地质灾害防治措施提供参考。

关键词:地质灾害;信息量模型;BP神经网络;支持向量机;易发性评价;怒江州

中图分类号:P694

doi:10.16186/j.cnki.1673-9787.2022040062

基金项目:国家自然科学基金资助项目(41161070)； 云南省科技厅-云南大学联合基金重点资助项目(2019FY003017)； 云南大学大湄公河次区域气候变化研究省创新团队项目(2019HC027)； 云南大学第二届专业学位研究生实践创新项目(ZC-22222175)

收稿日期:2022-04-22

修回日期:2022-06-18

网络首发日期: 2022-12-22

Evaluation of the susceptibility of geological disasters in the alpine and canyonareas：Taking Nujiang Prefecture as an example（Online）

FENG Xianjie1, LI Yimin2,3, DENG Xuanlun2, ZHAO Juanzhen1, YANG Yiming2

1.Institute of International Rivers and Eco-Security， Yunnan University， Kunming  650500， Yunnan， China；2.School of Earth Sciences， Yunnan University， Kunming  650500， Yunnan， China；3.Research Center of Domestic High-Resolution Satellite Remote Sensing Geological Engineering， Kunming  650500， Yunnan， China

Abstract:To investigate the spatial distribution of geological hazard susceptibility in the alpine and canyon areas， we select Nujiang Prefecture as the research area. Meanwhile， 12 evaluation factors such as elevation， slope， aspect， curvature， undulation， etc.， are used to build a regional susceptible evaluation index system. Then， a total of three models， the information value （IV） model， information value-back propagation neural networks （IV-BPNN） coupled model， and information value-support vector machine （IV-SVM） coupled model， are applied to evaluate the susceptibility of geological hazards. The results of the study show that the susceptibility results have better space distribution consistency with the actual geological disaster points； the high and extremely high susceptibility areas are linearly distributed and mainly concentrate in the areas on both sides of rivers such as the Nujiang， Lancang and Dulong rivers, as well as near roads and areas with active geological formations. In addition, the areas under thecurve (AUC) of the receiver operating characteristic curve (ROC) of the IV model, IV-BPNN model and IV-SVM model are 0.884, 0.889, and 0.901, respectively, which indicates that the accuracy of IV-SVM model is higher. Theresults of the study can provide a reference for the local government to develop geological disaster prevention andcontrol measures.

Key words:geological hazard;information model;BP neural network;support vector machine;susceptibility evaluation;Nujiang Prefecture

CLC:P694