首页|基于多源模型的粤北山区县域地质灾害危险性评价与驱动力分析

基于多源模型的粤北山区县域地质灾害危险性评价与驱动力分析

扫码查看
地质灾害危险性评价对降低地质灾害发生的不确定性及区域防灾减灾至关重要.本研究以粤北山区翁源县为例,选取与地质灾害密切相关的自然环境、区位距离与人类活动方面共13个评价指标,并基于皮尔逊相关系数进行评价指标相关性验证.采用信息量-随机森林耦合模型(Information-Random forest model,I-RF)对研究区地质灾害危险性进行评价,并结合地理探测器(Geodetector,GD)对地质灾害危险性驱动因子进行分析.研究结果显示:ROC曲线验证AUC值达到0.890,表明I-RF耦合模型可以较好地分析刻画研究区地质灾害危险性格局;研究区较高、高易发区主要分布河谷平原区和丘陵区,占比49.85%,中低易发区主要分布低山丘陵区,占比50.15%.县域中心龙仙镇遭受地质灾害的危险性较大,铁龙镇地质灾害危险性较低;因子探测结果显示高程、人口足迹和地质岩组是研究区崩塌、滑坡和泥石流三种主要地质灾害共有的诱发因子,交互探测结果显示高程是研究区地质灾害发生首要驱动因子,崩塌和滑坡主要受高程与距河流距离、地形起伏度和暴雨雨量交互驱动影响,泥石流则主要受高程与地形地貌、地质构造、水文气象和人类活动等方面评价指标交互驱动影响.总体来看,研究区地质灾害危险程度较高,应针对不同的地质灾害危险区和地灾类型,实施差异化的地质灾害防治措施.
Evaluation and Driving Force Analysis of Geological Disaster Risk in Mountainous Counties of Northern Guangdong Based on Multi-source Model
The uncertainties of time,space and scale of geological disasters bring great difficulties to disasters prevention and mitiga-tion.The risk assessment of geological disasters is very important to reduce the uncertainty of geological disasters and regional disaster prevention and reduction.Therefore,it is very necessary to evaluate the risk of regional geological disasters.This study takes Wengyuan County in the mountain area of northern Guangdong as an example,13 evaluation indicators of natural environment,location distance and human activities which are closely related to geological disasters are selected,and are verified for the correlation of evalua-tion indicators based on Pearson correlation coefficient.Firstly,the information value of each evaluation index classification interval is obtained based on the information quantity model,and then the random forest model is used to determine the weight of each evaluation index to weight the information quantity model to get the geological disasters prone result of I-RF model.Finally,the geographical de-tector is used to detect the factors of the results of the Information-Random forest(I-RF)model,and the influence degree of each eval-uation factor on the risk of geological disasters is quantitatively analyzed.The results show that:(1)The ROC curve verifies that the AUC value reaches 0.890,indicating that the I-RF model can better analyze and depict the risk pattern of geological disasters in the study area.(2)The area of high and high prone areas in the study area accounts for 49.85%,which is mainly distributed in valley plains and hilly areas where human activities are strong and ecologically sensitive and fragile,while middle and low prone areas account for 50.15%,mainly distributed in the low mountain hilly areas with little human disturbance,high vegetation cover,relatively low rainfall and rainfall erosion force.(3)From the county center,Longxian Town,which has good economic development,frequent con-struction activities and high development intensity,it is more likely to suffer from geological disasters,while for Tielong Town,which pays attention to ecological protection and relatively less intensity of human activities,it is less prone to geological disasters.(4)From the point of view of the driving factors of geological disasters,the factor detection results show that elevation,population footprint and geological rock groups are the common inducing factors of collapse,landslide and debris flow in the study area.The interactive detec-tion results show that elevation is the primary driving factor for the occurrence of geological disasters in the study area.Collapse and landslide are mainly driven by the interaction between elevation and the distance from the river,topographic relief and rainstorm rain-fall,while debris flow is mainly driven by the interaction between elevation and the evaluation indicators of topography,geological structure,hydrometeorology and human activities.On the whole,the risk degree of geological disasters in the whole county is on the high side.In the process of prevention and control of geological disasters in the future,different policies need to be implemented for different geological disaster risk areas and types of geological disasters.

geological disaster risk assessmentrandom forest modelinformation modelgeodetector

李亚、邓南荣、陈朝、陈进栋、王琦、李冠虹

展开 >

广东省科学院生态环境与土壤研究所华南土壤污染控制与修复国家地方联合工程研究中心广东省农业环境综合治理重点实验室,广州 510650

福建师范大学地理科学学院,福州 350007

地质灾害危险性评价 随机森林模型 信息量模型 地理探测器

广东省科学院发展专项资金项目国家自然科学基金项目国家自然科学基金项目

2019GDASYL-01050424187751442277479

2024

地球与环境
中国科学院地球化学研究所

地球与环境

CSTPCD北大核心
影响因子:0.875
ISSN:1672-9250
年,卷(期):2024.52(3)