摘要
为探究湖南省湘乡市滑坡灾害的临界降雨量阈值,该文收集了湘乡市近30余年降雨与滑坡灾害相关数据,并以湘乡市地区91个降雨型滑坡作为研究对象,依据2种常用滑坡事件降雨场的划分标准,分别对湘乡市的滑坡事件降雨场予以划分。通过研究2种降雨场划分标准与传统3种阈值模型(I-D、E-D、E-I)相互组合时对于湘乡市滑坡灾害降雨量阈值的适用性,将滑坡事件按照80%:20%的比例选取训练集与测试集,通过80%的训练集数据获取不同组合方式下的降雨量阈值曲线。其次,利用剩余20%的测试集数据对不同组合方式所得的降雨量阈值曲线进行验证,以验证结果的准确率,定量评估得出最适用于湘乡市滑坡灾害降雨量阈值的组合方法。最后,应用最优组合方法分别得到研究区域内滑坡灾害在不同岩性下的降雨量阈值,并采用普通最小二乘回归法(ordinary least squares regression,OLSR)划分临界阈值曲线,进一步深化降雨量阈值的应用。研究表明,以连续3 d的降雨量大于0.5 mm的时刻作为前期有效降雨量开始计时的时间,将滑坡灾害发生当日雨量作为当日激发雨量的滑坡事件降雨场划分标准与I-D阈值模型相互组合下所得降雨量阈值曲线准确率最高、最适用。
The rainfall data and landslide-related data in the past 30 years are collected in this study in order to investigate the critical threshold rainfall for landslide disasters in Xiangxiang City,Hunan Province.The research focuses on 91 rainfall-induced landslides in Xiangxiang City,and the rainfall fields of landslide events in Xiangxiang City are classified according to the two commonly used classification criteria.By coupling the two classification criteria with three traditional threshold models(I-D,E-D,E-I),the study assesses their applicability for determining the rainfall threshold values for landslide disasters in Xiangxiang City.Landslide events are selected as training and test sets in an 80%:20%ratio.80%of the landslide events are used as a training set to obtain various combinations of rainfall threshold curves.The remaining 20%is then used as a test set to validate these curves.The accuracy of the verification results are used to quantitatively evaluate the most suitable combination method for determining the rainfall volume threshold value for landslide disasters in Xiangxiang City.Finally,an optimal combination method are applied to determine rainfall thresholds of landslide disasters in different lithologies within the study area,and the ordinary least squares regression(OLSR)is then used to create critical threshold curves,further deepening the application of rainfall thresholds.The results indicate that combining the landslide event rainfall field classification criteria with I-D threshold model,using rainfall exceeding 0.5 mm for three consecutive days as the starting point for the effective pre-precipitation period and the rainfall on the day of the landslide as the triggering rainfall,yield highly accurate and applicable results.
作者
李昕鑫
刘洋
张联志
丁海洋
涂远卓
陶家珞
伍冬子
LI Xinxin;LIU Yang;ZHANG Lianzhi;DING Haiyang;TU Yuanzhuo;TAO Jialuo;WU Dongzi(School of Earth Sciences and Spatial Information Engineering,Hunan University of Science and Technology,Xiangtan 411201,China;Geospatial Survey and Monitoring Institute of Hunan Province,Changsha 410129,China)
出处
《自然灾害学报》
北大核心
2025年第5期57-69,共13页
Journal of Natural Disasters
基金
湖南省自然科学基金面上项目(2022JJ30244)
湖南省湘乡市1:10000地质灾害调查和风险评价项目(湘地调[2022]22号)。
关键词
降雨型滑坡
降雨场划分
阈值模型
I-D阈值曲线
最小二乘回归
rainfall-type landslide
division of rainfall field
threshold model
intensity-duration threshold curve
least squares regression
