基于GIS的Logistic回归模型在地质灾害危险性区划中的应用  

The application of GIS-based logistic regression for geological hazards zonation

作　　者：王卫东[1,2] 钟晟[2] 

机构地区：[1]中南大学商学院,长沙410075 [2]中南大学土木建筑学院,长沙410075

出　　处：《工程勘察》2009年第11期5-10,共6页Geotechnical Investigation & Surveying

基　　金：交通部西部项目(200331880201)

摘　　要：地质灾害危险性区划是将区域内产生崩塌、滑坡等地质灾害的条件、诱发地质灾害发生的外界因子,以及地质灾害现状综合统计分析,定量或半定量地评价区域内地质灾害的危险性程度等级。以贵州省为研究区域,首先将单个灾害因子分级,并在GIS环境下结合已发生的地质灾害进行二元空间统计分析,得到单因子各级别下灾害面积和灾害指标。然后,利用Logistic回归模型、GIS空间分析工具和统计软件寻求最合适的模型描述灾害是否发生(因变量)和5个灾害因子(高程、地层岩性、河流切割程度、坡度、年平均降雨量等)之间的关系。最后,应用Logistic回归模型计算危险性概率,绘制贵州省地质灾害危险性区划图。该区划图将贵州省分为低、中等、高、很高等4个危险性区域,分别占贵州省面积的16.37%、26.34%、43.30%和13.99%。Geological hazards zonation is making a comprehensive statistical analysis of conditions or external factors that induced hazards of collapse and landslide, and historic geological hazards in an area, then it defines levels of danger of the geological hazards by using quantitative or semi-quantitative methods. Firstly, the study divided every factor which induces hazards to several subclasses, made bivariate statistical analyses combined with historical geological hazards in Guizhou Province, and gotten the hazards area and index of every subclass under GIS environment. Secondly, the use of logistic regression, GIS spatial tools and statistical software is to find the best fitting function to describe the relationship between the presence or absence of hazards （dependent variable） and 5 independent factors （elevation, lithology, annual rainfall, etc. ）. Lastly, the study calculated the hazards probability by using logistic regression model, created the geological hazards zonation. Using the predicted map of probability, the study area was classified into four categories of landslide susceptibility： low, medium and high and very high which take up 16.37%, 26.34%, 43.30% and 13.99% of the total study area, respectively.

关 键 词：地质灾害 危险性区划 LOGISTIC回归 GIS技术 

分 类 号：P642.22[天文地球—工程地质学]