基于信息熵与KPSO聚类法滑坡敏感性分析  

Landslide Susceptibility Analysis Based on Entropy and KPSO Clustering

作　　者：孙树林[1,2] 余文平[1] 刘小芳[1] 王天宇[1] 魏永耀[3] 

机构地区：[1]河海大学地球科学与工程学院,江苏南京210098 [2]水文水资源与水利工程科学国家重点实验室,江苏南京210098 [3]江苏省地质调查研究院,江苏南京210098

出　　处：《环境保护科学》2014年第6期88-96,共9页Environmental Protection Science

基　　金：河海大学水文水资源与水利工程科学国家重点实验室开放研究基金(2005408911);留学回国人员科研基金(20071108);江苏省环境保护厅项目(201029)

摘　　要：选取南京地区滑坡为研究对象。基于DEM及遥感影像数据,选取样本,提取滑坡影响因素,计算影响因素熵值。利用K折粒子群优化(K PSO)方法,在GIS环境下针对地质资料缺乏情况,生成一个可靠的南京地区滑坡敏感图。自组织特征映射网络(SOM)法用作一个并行研究,结果作为K PSO法结果对比。结果表明,K PSO法聚类准确率为85%,自组织映射(S OM)法聚类结果准确率为80%,由此说明K PSO聚类法在地质资料不足前提下形成滑坡敏感区是一种行之有效的方法。In this paper, the landslide in Nanjing is selected as the study target. On the basis of DEM and remote sensing image data, landslide samples are selected for extraction of influential factors of landslide and calculation of the entropy of the factors. By use of K fold particle swarm optimization （KPSO） method, a reliable landslide susceptibility map of Nanjing in GIS environment is generated under the condition of insufficient geological data. Furthermore, self-organization feature map （SOFM） network is used as a parallel study and the study results are used for comparison with those from KPSO. The results indicate that the accuracy of KPSO clustering is 85% and that of SOFM is 80%. It can be concluded that KPSO is an effective method for landslide susceptibility analysis under the precondition of insufficient geological information.

关 键 词：滑坡敏感性 信息熵 聚类分析 粒子群优化 GIS 

分 类 号：P66[天文地球—地质学]