基于CF与Logistic回归模型耦合的地质灾害易发性评价——以北京市大清河流域生态涵养区为例

Geologic hazards susceptibility assessment based on coupling of CF model and Logistic regression model in the ecological conservation area of the Daqinghe river watershed of Beijing

在北京市大清河流域生态涵养区1450 km^(2)的区域内,以遥感影像解译为基础,结合1∶50 000地质灾害详细调查,获取全区888个地质灾害隐患点作为样本数据库,选取基岩类型、地貌类型、地形坡度、河流、公路、断裂6个评价因子,采用确定性系数(CF)与Logistic回归耦合模型评价地质灾害易发性,依照自然间断点分级法(Jenks)将研究区划分为极高易发区、高易发区、中易发区、低易发区和极低易发区。将未参与模型训练的20%地质灾害隐患点作为检验点与易发性分区结果进行叠加分析,通过频率比和ROC曲线进行精度检验。结果显示:基岩类型对地质灾害的发育具有控制作用;公路、断裂对地质灾害的空间分布影响明显;CF与Logistic回归耦合模型在实际应用中具有较高的准确性,是一种地质灾害易发性评价可靠性高的模型。

In the ecological conservation area of the Daqinghe river watershed of Beijing,888 geological hazard points are selected as analysis samples in an area of 1450 km^(2) through interpreting the remote sensing map and conducting a detailed field in-vestigation of geological hazards.Based on the relationship analysis between geological hazards distribution and 6 factors i.e.bed-rock type,geomorphic type,slope gradient,drainage,highway and fracture,this paper adopts coupling of CF model and logistic regression model to determine the weight coefficient of each factor.Using the Jenks method,the study area is divided into areas with extremely high,high,moderate,low and extremely low hazards susceptibility.The susceptibility maps are validated and compared using the frequency ratio and ROC to represent respective success rates.The results indicate that the development of geo-logical disasters is controlled by bedrock types and the geological distribution of the disasters is obviously influenced by highway and fracture.The coupling model of CF and logistic regression is highly rational,accurate and reliable in the evaluation of ecological hazard susceptibility.