延长县位于陕北黄土高原东部，区内遍布第四系黄土，地质灾害频繁发生，当地人民生命财产安全受到巨大威胁。本文结合地质灾害的危险性以及灾害发生可能造成的损失，对延长县地质灾害风险性大小进行分区评价，为延长县以后的发展规划提供一定的理论依据。通过收集相关资料得到延长县地质灾害发育特征，根据黄土地区地质灾害的独特性，选取坡度、坡向、曲率等9个评价因子，基于栅格单元和斜坡单元创建RF模型、SVM模型，以及使用CRITIC法获得的各评价因子分级信息量，将带有信息量值的因子属性数据作为属性值创建CRITIC-RF模型，得到研究区地质灾害易发性评价，再依据降雨和人类工程活动对地质灾害的影响完成研究区地质灾害危险性评价；选取人口密度、道路密度、建筑物密度和耕地密度作为评价因子，使用CRITIC-AHP组合赋权法完成研究区地质灾害易损性评价，根据危险性评价和易损性评价结果完成延长县地质灾害风险性评价。主要研究成果如下：

（1）延长县地处陕北黄土高原，区内地质构造简单，人类工程活动强烈，共发育73处地质灾害，其中滑坡38处（均为黄土滑坡），崩塌35处（岩质崩塌22处，黄土崩塌13处）。滑坡多分布于黄土沟谷区、黄土沟间区及河谷阶地区，岩质崩塌主要发育在公路沿线，黄土崩塌主要发育于人工斩坡平基建宅地段。地质灾害发育时间多在集中降雨时期。

（2）选取了高程、坡度、坡向、曲率、地形起伏度、地层岩性、黄土厚度、植被归一化指数（NDVI）、距河流距离、距道路距离和土地利用类型11个评价因子，经过相关性和共线性分析，剔除高程和地形起伏度这两个高相关评价因子，使用剩下的9个评价因子建立研究区地质灾害易发性评价因子指标体系。采用CRITIC法得到各评价因子分级信息量值，分析可知，在第四系黄土层、距道路距离＜1000m、黄土厚度125~150m和距河流距离＜1000m条件下更易发生地质灾害。

（3）基于9个评价因子分别制作了栅格单元和斜坡单元因子图层，采用RF、SVM和CRITIC-RF模型，对研究区进行地质灾害易发性评价，并通过分区统计以及ROC曲线对评价结果进行了合理性检验和预测精度分析，结果表明，基于栅格单元的RF模型预测得到的易发性评价结果更为合理和准确（AUC=0.898）。结合降雨量和人类工程活动对地质灾害的影响与易发性评价结果综合完成研究区地质灾害危险性评价。

（4）选取人口密度、道路密度、建筑物密度和耕地密度作为评价因子，使用AHP-CRITIC法对研究区进行地质灾害易损性评价，得到研究区地质灾害易损性评价分区图。将危险性评价与易损性评价结果采用风险概念模型（R=H×V）进行叠加，完成地质灾害风险性评价。

YanChang County is located in the eastern part of the Loess Plateau in northern Shaanxi. The area is covered with Quaternary loess. Geological disasters occur frequently, and the safety of local people's lives and property is greatly threatened. Based on the risk of geological disasters and the losses caused by disasters, this paper evaluates the risk of geological disasters in YanChang County, and improves the theoretical basis for the future development planning of YanChang County. By collecting relevant data, the development characteristics of geological disasters in YanChang County are obtained. According to the uniqueness of geological disasters in loess area, 9 evaluation factors such as slope, aspect and curvature are selected. Based on grid unit and slope unit, RF model and SVM model are created, and the classification information of each evaluation factor obtained by CRITIC method is used. The factor attribute data with information value are used as the CRITIC-RF model created by attribute value to obtain the susceptibility evaluation of geological disasters in the study area, and then the risk evaluation of geological disasters in the study area is completed according to the influence of rainfall intensity on geological disasters. The population density, road density, building density and cultivated land density were selected to complete the vulnerability assessment of geological disasters in the study area by CRITIC-AHP combination weighting method. According to the results of risk assessment and vulnerability assessment, the risk assessment of geological disasters in Yanchang County was completed. The main research results are as follows :

(1) Yanchang County is located in the Loess Plateau of northern Shaanxi. The geological structure in the area is simple and human engineering activities are strong.There are a total of 73 geological disasters in YanChang County, including 38 landslides (loess landslides) and 35 collapses (22 rock collapses and 13 loess collapses). Landslides are mostly distributed in the loess gully area, the loess gully area and the valley terrace area. Rock collapses are mainly developed along the highway, and loess collapses are mainly developed in the artificial slope flat foundation building area. The geological disasters in the study area are mostly developed in the years and months of concentrated rainfall.

(2) 11 evaluation factors, such as elevation, slope, aspect, curvature, topographic relief, stratigraphic lithology, loess thickness, normalized difference vegetation index (NDVI), distance to river, distance to road and land use type, were preliminarily selected. After correlation and collinearity analysis, two high correlation evaluation factors, elevation and topographic relief, were eliminated, and the remaining nine evaluation factors were used to establish the index system of geological disaster susceptibility evaluation factors in the study area. The CRITIC method is used to obtain the classification information value of each evaluation factor. The analysis shows that geological disasters are more likely to occur under the conditions of Quaternary loess layer, distance to road <1000m, loess thickness of 125~150m and distance to river <1000m.

(3) Based on the nine evaluation factors, the grid unit and slope unit factor layers are made respectively. The RF, SVM and CRITIC-RF are used to evaluate the susceptibility of geological disasters in the study area. The rationality test and prediction accuracy analysis of the evaluation results are carried out by zoning statistics and ROC curve. The results show that the susceptibility evaluation results obtained by the RF model based on the grid unit are the most reasonable and accurate (AUC = 0.898). The rainfall and susceptibility evaluation results are superimposed to complete the risk assessment of geological disasters in the study area.

(4) The population density, road density, building density and cultivated land density were selected as evaluation factors, and the AHP-CRITIC method was used to evaluate the vulnerability of geological disasters in the study area, and the zoning map of vulnerability evaluation of geological disasters in the study area was obtained. The risk assessment results and vulnerability assessment results are superimposed using the risk conceptual model (R=H×V) to complete the risk assessment of geological disasters.