基于LR模型的管线地质灾害易发性评价

Susceptibility Evaluation of Pipeline Geological Hazard based on LR Model

山区长输油气管道通常穿越多种复杂地貌单元,沿线地质形貌复杂,地质灾害难以避免。故此对四川山区管线进行地质灾害易发性评价意义重大。文章以四川省东南部为研究区,通过现场调查获取斜坡坡度、斜坡坡形、日降雨量、斜坡结构等9个地质灾害易发性条件因子构建空间数据库,基于逻辑回归(LR)模型建立了研究区地质灾害易发性的预测模型。结果显示,LR预测模型具有良好的预测性能,其AUC高达0.949。研究区地质灾害易发性分区为4类:极低易发区(39.39%)、低易发区(17.42%)、中易发区(12.12%)、高易发区(31.06%),预测结果与研究区实际情况相近。因此采用LR模型能够较为客观准确地评价四川省东南地区的管线地质灾害易发性。

Long-distance oil and gas pipelines in mountain areas usually pass through a variety of complex geomorphic units,and the geological topography along the pipeline is complex, so geological disasters are diff icult to avoid. Therefore, it is very important to evaluate the susceptibility of pipeline geological disasters in mountainous areas of Sichuan province. In this paper,the southeast of Sichuan province was taken as the research area, and nine geological hazard susceptibility factors, including slope,slope shape, daily rainfall and slope structure, were obtained through f ield investigation to build a spatial database, and a prediction model of geological hazard susceptibility was established based on logistic regression(LR) model. The results show that the LR prediction model has good predictive performance with an AUC of 0.949. The susceptibility of the study area was divided into four categories: extremely low(39.39%), low(17.42%), medium(12.12%), and high(31.06%). The predicted results were similar to the actual situation of the study area. Therefore, the LR model can be used to objectively and accurately evaluate the susceptibility of pipeline geological disasters in southeast Sichuan Province.