Landslide susceptibility evaluation models can directly affect the accuracy of research results.To investigate the influence of different coupled models on landslide susceptibility analysis,a total of eight evaluation factors,including elevation,slope,engineering rock group and so on,were selected in Longling County,Yunnan Province,as the study area to construct a landslide susceptibility evaluation system.The multi-layer perceptron method(MLP)was coupled with the information volume model(IV)to carry out landslide susceptibility evaluation and was compared with the traditional information volume-logistic regression model(IV-LR),and the accuracy was verified by the AUC value.The results show that the AUC values of IV-MLP,IV-LR,and IV models were 0.862,0.813,and 0.799,respectively.The AUC value of the IV-MLP coupling model is higher than that of the IV-LR coupling model,and the MLP model performs better in the landslide susceptibility coupling anal-ysis than the IV model.The AUC values of IV-MLP and IV-LR coupling models were higher than those of the single IV model by 0.063 and 0.014,respectively,and the coupled model is more effective than the single model in the landslide susceptibility zoning.The results can provide a scientific theoretical basis for assessing landslide susceptibility in Longling County and a ref-erence for disaster prevention.
关键词
滑坡易发性/信息量法/逻辑回归/多层感知器/龙陵县
Key words
landslide susceptibility/information volume method/logistic regression/multi-layer perceptron/Longling county
维普
万方数据