备注
为了减轻地质灾害对人类生产、生活的影响,通过易发性评价对地质灾害易发区进行划分,对于地质灾害防治具有重要意义。以雅砻江中上游新龙段为研究区,初选坡度、坡向、起伏度、坡形、工程岩组、降雨、与水系距离、与道路距离和与断层距离为评价因子。以236处地质灾害隐患点为依据,运用确定系数法计算各评价因子分级的确定性系数(CF),将CF值带入逻辑回归模型(Logistic),剔除未通过显著性检验的评价因子,建立了CF-Logistic模型。通过ArcGIS空间分析平台,对研究区进行了地质灾害易发性评价,并将单一的CF模型作对比,运用ROC曲线检验模型精度。结果表明:极高易发区最主要分布在贯通南北的G227线两侧,该处同时也是雅砻江主干两岸,面积为245.53 km2,占比2.69%; 高易发区主要分布在中部断层周围,面积573.48 km2,占比6.28%; 中、低、极低易发区面积分别为1 518.54,4 245.11,2 553.05 km2。模型精度结果显示:CF-Logistic模型(0.889)优于单一的CF模型(0.878),可以更加精确地划分研究区易发性分区。
In order to reduce the impact of geological disasters on human production and life, the areas prone to geological disasters were divided through the assessment of susceptibility, which is of great significance to the prevention and control of geological disasters. Taking the Xinlong section of the middle and upper reaches of the Yalong River as the research area, we selected the gradient, aspect, undulation, slope shape, engineering rock group, rainfall, distance from water system, distance from road and distance from fault as evaluation factors. Based on 236 hidden danger points of geological disasters, the determination coefficient method was used to calculate the certainty factor(CF)of each evaluation factor classification, the CF value was brought into the logistic regression model(Logistic), and the evaluation factors that had not passed the significance test were eliminated to establish CF-Logistic model. Through the ArcGIS spatial analysis platform, the susceptibilities of geological disasters of the study area were evaluated, and the single CF model was compared, and the accuracy of the model was verified by the ROC curve. The results show that the extremely high-prone areas mainly distribute on both sides of the G227 Highway that runs through the north and south, and these areas are also the main bank of the Yalong River, cover the area of 245.53 km2, and account for 2.69% of the total area; the high-prone areas mainly distribute around the central fault, cover the area of 573.48 km2, and account for the proportion 6.28% of the total area; the medium, low and very low prone areas are 1 518.54 km2, 4 245.11 km2, and 2 553.05 km2, respectively. The model accuracy results show that the CF-Logistic model(0.889)is better than the single CF model(0.878),and can more accurately divide the susceptibility zone of the study area.
