基于I-SVM耦合模型的区域滑坡易发性评价

REGIONAL LANDSLIDE SUSCEPTIBILITY EVALUATION BASED ON I-SVM COUPLED MODEL

区域滑坡易发性对于灾害防治与土地规划有着重要意义,本文选取云南省大关县作为实例,采用信息量-支持向量机(I-SVM)耦合模型作为评价方法进行区域滑坡易发性评价,采用高程、坡向、土地利用等13项指标评价因子构建模型评价体系,完成滑坡易发性区划。结果表明,I-SVM耦合模型预测结果精度较高,AUC值为0.966,优于传统I模型的0.769,是一种可靠的滑坡易发性评价模型。评价结果可为大关县的灾害防治提供一定的参考及建议。

Regional landslide susceptibility is of great significance for disaster prevention and land use planning.In this paper,Daguan County of Yunnan Province is selected as an example.The coupled Information-Support Vector Machine(I-SVM)model is used as an evaluation method for regional landslide susceptibility evaluation.Thirteen index evaluation factors such as elevation,slope direction and land use,are used to construct the model evaluation system and complete the landslide susceptibility zoning.The results show that the I-SVM coupled model has a high accuracy of 0.966 AUC,which is better than the traditional I model of 0.769.It is a reliable landslide susceptibility evaluation model.The evaluation results can provide some reference and suggestions for disaster prevention and control in Daguan County.