基于多种组合模型的低山丘陵区地质灾害危险性评价及精度分析

1.1.华南理工大学土木与交通学院,广州 510640;2.2.广东省地质环境监测总站,广州 510510,;3.3.广东省地质工程公司,广州 510000

地质灾害;危险性评价;逻辑回归;多尺度地理加权回归

Risk Assessment of Geological Disasters in Low Mountain and Hilly Regions Based on Multiple Combined Models and Its Accuracy Analysis
YAO Xiaoyue1,SU Wenji1,LI Xiujuan2,ZHENG Zhiwen2,MEI Weibiao3

1.Department of Civil Engineering , South China University of Technology , Guangzhou 510640, China;2.Guangdong Geological Environment Monitoring Station , Guangzhou 510510, China;3.Geological engineering Company of Guangdong Province , Guangzhou 510500, China

Geological disaster; Risk assessment; Logistic regression; Multi-scale geographically weighted regression

DOI: 10.13512/j.hndz.2023.03.12

备注

为探索低山丘陵区地质灾害危险性评价模型的适用性和评价结果的合理性,以广东省平远县作为研究对象,依托研究区内历史地质灾害资料,开展基于多种组合模型的区域地质灾害危险性评价及精度分析研究。论文选取坡度、距断层距离、地层岩性、流域沟壑密度、植被归一化指数、年均降雨量等14项与地质灾害发生相关的孕灾环境因子和诱发因素因子,依据378处地质灾害点样本对各连续型评价因子开展快速聚类分析,进行状态分级,分别采用信息量—逻辑回归、证据权—逻辑回归、信息量—多尺度地理加权回归、证据权—多尺度地理加权回归4种组合模型,构建了相应的评价指标体系,并完成地质灾害危险性评价,最后对评价结果进行分析、比较和讨论。综合所有灾点样本在危险区的分布情况,以及3类精度分析方法的评价结果得出:在低山丘陵区地质灾害危险性评价方面逻辑回归组合模型均表现较为理想,其中证据权—逻辑回归(WL)组合模型更为合理和适用,而多尺度地理加权回归模型尽管在地质灾害危险性评价方面具有一定的创新性,但是针对本文的研究区广东省平远县,并不具有很强的适用性。
In order to explore the applicability of the geological disaster risk assessment model and the rationality of the evaluation results in low mountain and hilly regions, taking Pingyuan County, Guangdong Province as the research object, and relying on the historical geological disaster data in the study area, the regional geological disaster risk assessment and accuracy analysis based on multiple combined models were carried out. In this paper, 14 environmental factors and inducing factors related to geological disasters are selected, including slope, distance from fault, stratigraphic lithology, watershed gully density, vegetation normalization index, and average annual rainfall, etc. Based on the samples of 378 geological disaster points, rapid cluster analysis is carried out on each continuous evaluation factor to classify the status. Four combined models of information quantity-logical regression, evidence weight-logical regression, information quantity-multi-scale geographically weighted regression and evidence weight-multi-scale geographically weighted regression are used to construct the corresponding evaluation index system. then the paper completes the risk assessment of geological disasters, and finally analyzes, compares and discusses the evaluation results. Based on the distribution of all disaster point samples in dangerous area and the evaluation results of the three types of precision analysis methods, it is concluded that the logistic regression combination model performed better in the risk assessment of geological disasters in low mountain and hilly areas,and the weight of evidence-logistic regression model is considered to be more reasonable and applicable. Although the multi-scale geographically weighted regression model has certain innovation in the risk assessment of geological disasters,it is not highly applicable for the study area of Pingyuan County,Guangdong Province.
·