基于WCF-RF的地质灾害敏感性评价——以左贡县为例  

Sensitivity Evaluation Based on Weighted Random Forest Model:A Case Study of Zuogong County

作　　者：张根[1,2,3] 欧家婷 央金卓玛 胡嘉乐[1] ZHANG Gen;OU Jiating;Yangchen-Drolma;HU Jiale(Engineering College,University of Tibet,Lhasa 850000,China;Plateau Major Infrastructure SmartConstruction and Resilience Safety TechnologyInnovation Center,Lhasa 850000,China;School of Ecology and Environment University of Tibet,Lhasa 850000,China)

机构地区：[1]西藏大学工学院,西藏拉萨850000 [2]高原重大基础设施智慧建造与韧性安全技术创新中心,西藏拉萨850000 [3]西藏大学生态环境学院,西藏拉萨850000

出　　处：《高原科学研究》2025年第1期67-77,共11页Plateau Science Research

基　　金：西藏自然科学基金项目(XZ202401ZR0013);西藏中央引导地方项目(XZ202401YD0003);国家自然科学基金项目(U23A2047)。

摘　　要：为提高地质灾害敏感性评价的准确性和可靠性,文章以昌都市左贡县为例,提出一种结合随机森林模型(RF)与加权确定性系数模型(WCF)的地质灾害敏感性评价方法。通过层次分析法(AHP)对13个关键敏感因子进行权重计算,得到加权确定性系数值,采用信息量模型(Ⅰ)和确定性系数随机森林模型(CFRF)与WCF-RF模型进行对比分析,利用ArcGIS软件对模型数据进行处理,生成了极低、低、中、高、极高5类敏感区,统计得到不同敏感区对应的地灾点密度,通过ROC曲线下的面积(AUC值)进行模型精确度验证。WCF-RF模型极高敏感区的地灾点密度为15.65处/km^(2),高于Ⅰ模型的6.92处/km^(2)和CF-RF模型的10.87处/km^(2),说明WCF-RF模型更能揭示地质灾害点的分布特征;WCF-RF模型性能最优,Ⅰ模型、CF-RF模型、WCF-RF模型的AUC值分别为80.2%、89.1%、90.7%。结果表明,WCF-RF模型在左贡县的地质灾害敏感性评价中具有较高的准确性和可靠性,能够为地质灾害预防提供参考。To improve the accuracy and reliability of geological disaster susceptibility assessments,this study proposes a new geological disaster susceptibility evaluation method that combines the Random Forest(RF) model with the Weighted Certainty Factor(WCF) model,using Zuogong County as a case study.The study calculates the weights of 13 key sensitivity factors through the Analytic Hierarchy Process(AHP) to obtain weighted certainty factor values.Comparative analyses are conducted using the Information(I)model,the Certainty Factor Random Forest(CF-RF)model,and the WCF-RF model.ArcGIS software is utilized to process the model data,generating five categories of susceptibility zones:extremely low,low,medium,high,and extremely high.The corresponding densities of geological disaster points in different susceptibility zones are then calculated,and the accuracy of the models is validated using the Area Under the Receiver Operating Characteristic Curve(AUC value).The WCF-RF model demonstrates a geological disaster point density of 15.65 points/km~2 in the extremely high susceptibility zone,which is higher than the 6.92 points/km~2 of the I model and the 10.87 points/km~2 of the CF-RF model.This indicates that the WCF-RF model is more effective in illustrating the distribution characteristics of geological disaster points.Additionally,the WCF-RF model performs optimally,with AUC values of 80.2%,89.1%,and 90.7% for the I model,CF-RF model,and WCF-RF model,respectively.The results show that the WCF-RF model exhibits high accuracy and reliability in assessing geological disaster susceptibility in Zuogong County,providing valuable references for geological disaster prevention.

关 键 词：地质灾害敏感性 加权确定性系数 随机森林模型 层次分析法 

分 类 号：P642.22[天文地球—工程地质学]