基于IV-IOE-LR耦合模型的滑坡易发性评价——以云南省泸水市为例  

Evaluation of landslide susceptibility based on IV-IOE-LR coupling model-a case study of Lushui city,Yunnan

作　　者：陈羽 杨海林[1] 钟延江 Chen Yu;Yang Hailin;Zhong Yanjiang(School of Land Resources Engineering,Kunming University of Science and Technology,Kunming Yunnan 650093,China;Kunming Engineering Corporation Limited,Kunming Yunnan 650000,China)

机构地区：[1]昆明理工大学国土资源与工程学院,云南昆明650093 [2]中国电建集团昆明勘测设计研究院有限公司,云南昆明650000

出　　处：《化工矿物与加工》2023年第10期62-71,79,共11页Industrial Minerals & Processing

摘　　要：开展滑坡易发性评价有利于提高当地滑坡风险管控水平。以云南省泸水市241个滑坡点为研究对象,选取高程、坡度、曲率、地形起伏度、坡向、岩组类型、断裂缓冲区、河流缓冲区、年均降雨量、道路缓冲区等10个影响因子构建滑坡易发性评价体系,将建模数据划分为70%的训练数据和30%的测试数据,分别利用信息量模型(IV)、信息量-熵指数耦合模型(IV-IOE)、信息量-逻辑回归耦合模型(IV-LR)和信息量-熵指数-逻辑回归耦合模型(IV-IOE-LR)进行了滑坡预测,并通过ROC曲线对模型准确性进行了检验,结果表明:4种模型的预测结果均较理想,A UC值均大于0.7,3种耦合模型的效果明显优于IV模型,其中IV-IOE-LR耦合模型的精度最高,预测效果最好;对未参与训练的30%滑坡栅格进行预测精度分析,得到IV、IV-IOE、IV-LR模型和IV-IOE-LR模型的A UC值分别为0.725、0.736、0.741、0.794,IV-IOE-LR模型较其他3种模型的预测成功率分别提高了9.5%、7.9%、7.2%。Conducting landslide susceptibility assessment facilitates improvement of the local management and control level of landslide risk.An evaluation system was established with 10 influencing factors including elevation,slope,curvature,terrain undulation,slope orientation,rock formation type,fault buffer zone,river buffer zone,annual rainfall,and road buffer zone,while taking 241 landslide points as the study object.The modeling data was divided into 70%of training data and 30%of testing data.The information volume(IV)model,information volume-index of entropy(IV-IOE)coupling model and the information volume-logistic regression coupling model(IV-LR)and the information volume-index of entropy-logistic regression coupling model(IV-IOE-LR)were employed for landslide prediction,of which the accuracy was tested via ROC curves.The results show that the predicted data from the four models were acceptable as expected,presenting with A UC values of over 0.7.The performance of the coupled models was significantly better than that of the IV model,the application of IV-IOE-LR coupled model exhibited the highest accuracy of prediction and the most satisfactory results.Analysis of the prediction accuracy of the 30%landslide grids which were excluded from training indicated that the A UC values of the IV model,IV-IOE model,IV-LR model,and IV-IOE-LR model were 0.725,0.736,0.741,and 0.794 respectively.The successful prediction likelihood achieved with the last IV-IOE-LR model was 9.5%,7.9%and 7.2%higher than that with the former three models respectively.

关 键 词：滑坡 易发性评价 耦合模型 信息量 熵指数 逻辑回归 

分 类 号：P642.22[天文地球—工程地质学]