权重线性组合与逻辑回归模型在滑坡易发性区划中的应用与比较

以贵州省为研究区对滑坡易发性进行研究,对预测方法进行比较.其步骤为:首先,选取年均降雨量、岩性、高程、坡向、地形地貌、坡度、至构造线距离、至水系距离、至铁路距离和至公路距离10项评价指标作为影响滑坡发生的致灾因子,分别采用主客观权重线性组合(WLC)模型及结合确定性系数的逻辑回归模型(LR)分析研究区内各致灾因子与滑坡发生的相关性并建立预测模型:其次,基于ArcGIS平台完成贵州省滑坡易发性区划图,结合历史滑坡分布,应用熵值法对2个模型的区划结果进行验证和比较.最后,从模型原理、运行环境和适用条件等方面对2个模型进行比较和讨论.研究结果表明:WLC模型和LR模型的熵权分别为0.255和0.745,表明逻辑回归模型(LR)的区划结果与实际的滑坡情况吻合度更高.

Application and comparison of weighted linear combination model and logistic regression model in landslide susceptibility mapping

Taking Guizhou Province as an example,the landslide susceptibility mapping was assessed,and a comprehensive comparison of the two approaches was conducted.The procedures are as follows.Firstly,a total of ten landslide triggering factors including average annual precipitation,lithology,elevation,aspect,topography,slope,proximity to structural lines,proximity to drainage network,proximity to railways and proximity to roads were considered.The weighted linear combination(WLC) model and logistic regression(LR) model were applied to investigate the correlation between the presence or absence of landslides and the set of triggering factors,and then the prediction models were produced respectively.Secondly,two landslide susceptibility maps were produced in ArcGIS environment.The zonation results derived from the two models were verified using the past failures and compared using the entropy method.At last,the two models were compared and discussed in terms of their principles,operating environments and applicable conditions.The results show that the entropy for WLC and LR model are 0.255 and 0.745,respectively,indicating that the LR model has better performance than the WLC model.