摘要
重庆乌江河谷地区地形陡峻,滑坡地质灾害沿河谷地带多发,开展乌江高陡岸坡滑坡地质灾害易发性评价对减少地质灾害风险与线路工程规划具有重要意义。选取乌江龙溪-石朝门段高陡岸坡为研究区,提取高程、坡度、斜坡结构、斜坡形态、冲沟、岩组分类、地质构造和道路评价指标因子信息,采用随机森林模型对研究区滑坡地质灾害易发性进行了评价与精度分析,并研究了各评价指标因子的贡献程度。结果表明:斜坡物质和斜坡结构是影响研究区滑坡地质灾害发育的主要因素;通过混淆矩阵和ROC曲线对随机森林模型的预测精度进行评估,结果显示随机森林模型在高陡岸坡滑坡地质灾害易发性评价中的预测精度较高,是一种滑坡地质灾害易发性评价的可靠方法。
The Wujiang River valley in Chongqing is characterized by steep terrain and frequent landslide hazards.For landslide risk reduction and road engineering planning,it is necessary to carry out landslide susceptibility assessment of the high and steep bank slopes along Wujiang River.This paper selected the Longxi-Shichaomen section of Wujiang River as the research area and extracts altitude,terrain slope,slope structure,slope shape,gully,engineering geological rock group,geological structure and roads as the evaluation indexes.The paper applies the random forest model to evaluating and analyzing the landslide susceptibility in the study area.The results show that the slope material and slope structure are the main factors affecting the landslides in the study area.According to the analysis of confusion matrix and ROC curve,the results show that the prediction accuracy of random forest model is high in the assessment of landslide susceptibility in high and steep bank slopes,the random forest model is a reliable methed in the assessment of landslide susceptibility.
作者
杨硕
李德营
严亮轩
黄元
王明哲
YANG Shuo;LI Deying;YAN Liangxuan;HUANG Yuan;WANG Mingzhe(Faculty of Engineering,China University of Geosciences(Wuhan),Wuhan 430074,China)
出处
《安全与环境工程》
CAS
CSCD
北大核心
2021年第4期131-138,共8页
Safety and Environmental Engineering
基金
国家自然科学基金项目(41772310、41842062)。
关键词
高陡岸坡
滑坡地质灾害
易发性评价
随机森林模型
模型精度
high and steep bank slope
landslide hazards susceptibility assessment
random forest model
model accuracy
