基于空间逻辑回归模型的滑坡易发性评价
Evaluation of Landslide Susceptibility Based on Spatial Logistic Regression Model
郑雪 1唐章英 1宋超2
作者信息
- 1. 西南石油大学 地球科学与技术学院, 四川 成都 610500
- 2. 四川大学 华西公共卫生学院, 四川 成都 610041
- 折叠
摘要
滑坡是山地丘陵地区发生最频繁、危害性最强的地质灾害之一.进行滑坡易发性评价对于人们进行土地灾害评估和减轻滑坡相关损失是十分必要的.在过去的几十年里,已经开发了许多种模型用于滑坡评估和易发性分级,但这些模型大多数并未考虑数据的空间结构信息,预测精度还有待提高.研究使用随机森林模型筛选风险因子后对四川雅安芦山地区震后滑坡使用贝叶斯空间逻辑回归进行建模,并与普通未考虑空间结构信息的逻辑回归建模结果进行比较.贝叶斯空间逻辑回归的AUC值为 0.931,在传统逻辑回归的基础上提升了近 14%,为滑坡易发性评价带来了新的思路.
Abstract
Landslide is one of the most frequent and harmful geological disasters in mountainous and hilly areas.It is necessary to conduct landslide susceptibility assessment for people to assess land disasters and reduce landslide related losses.In the past few decades,many models have been developed for landslide assessment and susceptibility classification,but most of these models do not consider the spatial structure information of the data,and the prediction accuracy still needs to be improved.This study uses the random forest model to screen risk factor,and then uses Bayesian spatial logistic regression to model the post earthquake landslides in Lushan area,Ya′an,Sichuan,and compares the results with those of ordinary logistic regression modeling without considering spatial structure information.The AUC value of Bayesian spatial logistic regression is 0.931,which has increased by nearly 14%on the basis of traditional logistic regression,bringing new ideas for landslide susceptibility evaluation.
关键词
滑坡/贝叶斯/空间逻辑回归/风险制图/随机森林Key words
landslide/Bayesian/spatial logistic regression/risk mapping/random forest引用本文复制引用
基金项目
国家自然科学基金面上项目(42071379)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据