基于斜坡单元划分和逻辑回归模型的滑坡易发性评价——以山阳县高坝店镇为例

Landslide Susceptibility Assessment Based on Slope Unit Division and Logistic Regression Model-Taking Gaobadian Town of Shanyang County as an Example

斜坡单元作为基本评价单元,常被用于地质灾害易发性评价中,在地质灾害易发性评价中具有极其重要的作用。使用GIS平台中的水文分析工具提取高坝店镇斜坡单元,选取坡度、地势起伏度、海拔高度、灾害点密度、距断层距离、距水系距离、距道路距离、NDVI指数8个评价因子,进行多重共线性计算,据逻辑回归模型求得各因子逻辑回归系数及常量,制出高坝店镇滑坡易发性概率图。采用自然间断法将其划分为极高、高、中、低、非易发区,面积分别为19.41 km^(2)、27.20 km^(2)、46.42 km^(2)、60.52 km^(2)、94.99 km^(2);占研究区总面积的7.81%、10.94%、18.68%、24.35%、38.22%;各区灾害比率随易发性等级升高而增加。采用受试者工作特征(ROC)曲线方法检验评价模型结果的可靠性,得到AUC为0.8117。说明对高坝店镇进行基于斜坡单元划分和逻辑回归模型的滑坡地质灾害易发性评价具较好的适用性。

As a basic evaluation unit,slope unit is often used in the evaluation of geological disaster susceptibility,and plays an important role in the evaluation of geological disaster susceptibility.In this paper,the slope unit of Gaobadian Town is extracted by using the hydrological analysis tool in GIS platform.Eight evaluation factors are selected,including slope,relief amplitude,altitude,disaster point density,distance from fault,distance from water system,distance from road and NDVI index.After multicollinearity calculation,the logistic regression model is used to obtain the logistic regression coefficients and constants of each factor,and the probability map of landslide susceptibility in Gaobadian Town is drawn.The natural discontinuity method is used to divide it into extremely high,high,medium,low and non-prone areas.The area is 19.41 km^(2),27.20 km^(2),46.42 km^(2),60.52 km^(2),94.99 km^(2);it accounts for 7.81%,10.94%,18.68%,24.35%and 38.22%of the total area of the study area.The disaster ratio of each district increases with the increase of susceptibility level.The reliability of the evaluation model results was tested by the receiver operating characteristic(ROC)curve method,and the AUC was 0.117.It shows that the susceptibility evaluation of landslide geological disasters based on slope unit division and logistic regression model in Gaobadian Town has good applicability.