摘要
潜在滑坡识别对地质灾害高发区的滑坡研究有重要意义。文章以中巴经济走廊灾害多发区某一试验区为例,结合由2014-2019的Sentinel1-A雷达数据计算出的地表变形数据,选取试验区地形、水文、地质、聚类共8种特征变量基于随机森林模型进行滑坡识别,滑坡识别结果总体精度为88.62%,Kappa系数为0.78。结果表明,结合地表变形数据的随机森林模型可以较好的识别潜在滑坡,聚类特征对于识别结果有重要贡献。
The identification of potential landslides is of great significance for landslide research in high-risk areas of geological hazards.Taking a test area in a disaster-prone area of the China-Pakistan Economic Corridor as an example,combined with the surface deformation data set calculated from the Sentinel1-A radar data of 2014-2019,eight characteristic variables of terrain,hydrology,geology,and clustering in the test area were selected.Landslide identification was performed using a random forest model.The overall accuracy of the landslide identification result was 88.62%,and the Kappa coefficient was 0.78.The results show that the random forest model combined with the surface deformation data set can better identify the distribution of potential landslides,and the clustering features make an important contribution to the recognition results.
作者
李萌
彭思佳
白艳萍
黄兆欢
LI Meng;PENG Sijia;BAI Yanping;HUANG Zhaohuan
出处
《科技创新与应用》
2020年第6期9-13,共5页
Technology Innovation and Application
基金
国家自然科学基金国际合作项目(编号:41661144046)
关键词
随机森林
中巴经济走廊
地质灾害
特征变量
random forest
China-Pakistan Economic Corridor
geological disaster
characteristic variable