基于GIS的分组数据Logistic模型在斜坡稳定性评价中的应用

分组数据Logistic回归是针对因变量为定性变量、自变量为分类变量的一种解决方案,加权最小二乘法可用来求解该方程.将巫山县新城西区作为试验区,选取岩性、坡度、高程、地下水位埋深、距最近有影响构造线距离5种因素为斜坡稳定性影响因素,以试验区历史滑坡发生为因变量,建立了区域斜坡稳定性评价的分组数据Logistic回归方程,进行了回归方程显著性检验和回归系数显著性检验,最后利用回归方程对全区斜坡稳定性进行预测.模型拟合精度为:以滑坡发生概率0.157 9为判据,滑坡发生样本的判对率为72.55%,滑坡不发生样本的判对率为79.69%.

Application of GIS-Based Grouped Data Logistic Model in Evaluation of Slope Stability

Grouped data Logistic model is a scheme for the solving the problem of whose dependent variable is a qualitative variable and independent variable is a categorical variable. Weighted least square is used to solve grouped data Logistic equation. New west area in Wushan county was chosen as experimental area. A grouped data Logistic model was set up based on that lithological characters, gradient, height, groundwater level depth and nearest distant of effected structure line were taken as independent variables. Historical landslide was taken as dependent variable. The second step was prominent testing of equation and regression coefficient. At last, slope stability was predicted by regression equation in the experimental area. The model predicts actual landslide occurrence at 72.55% accuracy and landslide non-occurrence at 79.69% when criterion of probability is equal to 0.157 9.