基于证据权和支持向量机模型的威信县滑坡易发性评价

滑坡易发性评价研究对滑坡灾害防治具有重要意义。多模型耦合在滑坡易发性评价中运用广泛,但将(Weight of evidence,WOE)和支持向量机模型(Support Vector Machine,SVM)模型耦合进行滑坡易发性评价研究较少。以滇东北山区威信县为研究区,选取坡度等12个滑坡评价因子建立滑坡易发性评价指标体系,根据证据权计算得到证据权对比度、滑坡栅格占比和分级栅格占比,对指标因子进行分级,构建WOE-SVM模型计算得到滑坡易发性指数(Landslide susceptibility index,LSI),利用GIS平台得到研究区易发性分级图。结果表明：滑坡极高和高易发区主要分布河流流域和人类工程活动频繁区域,SVM和WOE-SVM模型评价结果与滑坡空间位置分布基本一致,但耦合模型精度高于单一SVM模型,其评价结果也更加合理有效,可为当地滑坡灾害的治理与预防提供一定参考价值。

Landslide susceptibility assessment in Weixin County based on evidence weight and support vector machine model

The study of landslide susceptibility evaluation is of great significance for landslide disaster prevention and control. Multi-model coupling is widely used in landslide vulnerability assessment, but there are few studies on combining Weight of Evidence (WOE) and Support Vector Machine (SVM) model in landslide vulnerability assessment. Taking Weixin County in the mountainous area of northeast Yunnan as the research area, 12 landslide evaluation factors such as slope were selected to establish the landslide susceptibility evaluation index system. According to the weight of evidence, the evidence weight contrast, landslide grid proportion and hierarchical grid proportion were calculated, and the index factors were graded. The WOE-SVM model was constructed to calculate Landslide susceptibility index (LSI), and GIS platform was used to obtain the susceptibility classification map of the study area. Results show that the landslide is extremely high, and high rock mainly river basins and human engineering activities frequently area, SVM and WOE - SVM model evaluation results was basically in accord with the landslide spatial distribution, but the coupling model accuracy is higher than single SVM model, the evaluation result is more reasonable and effective, but for local governance and landslide disaster prevention provide a certain reference value.