基于Logistic回归的降雨型滑坡概率预测研究

Study on Probability Prediction of Rainfall-type Landslide Based on Logistic Regression

为明确降雨对滑坡的影响规律,本文选取赫章县2000年以来已经发生的滑坡作为研究对象,基于Logistic回归,采用降雨因子进行统计分析,并建立降雨型滑坡的预测模型。研究表明,滑坡发生前3~4 d,降雨量对滑坡的影响最大;发生滑坡的判断正确率为83.8%,未发生滑坡的判断正确率为93.5%,预测模型的总判断正确率为88.7%。

In order to clarify the impact of rainfall on landslides,this paper selects the landslides that have occurred in Hezhang County since 2000 as the research object,uses rainfall factors for statistical analysis based on Logistic regression,and establishes a prediction model of rainfall-type landslides.The study shows that rainfall has the greatest impact on the landslide 3~4 d before the landslide;the correct judgment rate of landslide is 83.8%,the correct judgment rate of no landslide is 93.5%,the total correct judgment rate of the prediction model is 88.7%.