基于日降雨量的木里县滑坡预测模型与预警系统的建立  

Establishment of Landslide Prediction Model and Early Warning System in Muli County Based on Daily Rainfall

作　　者：罗咪 张世卫 黄冉 Luo Mi;Zhang Shiwei;Huang Ran(Chengdu University of Information Technology,Chengdu 610200,China)

机构地区：[1]成都信息工程大学,成都610200

出　　处：《黑龙江科学》2024年第22期24-27,共4页Heilongjiang Science

基　　金：四川省大学生创新创业训练计划项目(202210621049)。

摘　　要：以四川木里藏族自治县为研究区,利用2014—2019年的日降雨量和历史滑坡数据分析了滑坡与降雨量的时空特征,采用有效降雨量概念建立了有效降雨强度-降雨历时的散点图,得到阈值方程,用历史数据建立Logistic预测模型,用ROC曲线和新滑坡事件对模型进行验证。结果表明,滑坡多发生在夏季和沿河地区,滑坡发生与前5天降雨量相关性最大,Logistic回归模型在样本数据中的预测精度达90.4%,新滑坡事件预测概率在70%~80%,说明基于ArcEngine和C#开发的系统具有数据处理、空间分析和可视化预警等功能。In this study,the daily rainfall and historical landslide data from 2014 to 2019 were used to analyze the spatial and temporal characteristics of landslide and rainfall in Muli Tibetan Autonomous County,Sichuan Province.The study conducted the concept of effective rainfall,established a scatter plot of effective rainfall intensity-rainfall duration,and obtained the threshold equation.Logistic prediction model was established with historical data,and the model was validated by ROC curve and new landslide events.The results show that landslides mostly occur in summer and along the river.The occurrence of landslides was mostly correlated with the rainfall in the previous five days.The prediction accuracy of the logistic regression model in the sample data was 90.4%,and the prediction probability of new landslide events was between 70%and 80%.The system developed based on ArcEngine and C#has functions,such as data processing,spatial analysis,and visual early warning.

关 键 词：滑坡概率 日降水量 LOGISTIC回归模型 滑坡预警系统 

分 类 号：P642.22[天文地球—工程地质学]