How can statistical and artificial intelligence approaches predict piping erosion susceptibility

摘要: It is of fundamental importance to model the relationship between geo-environmental factors and piping erosion because environmental degradation attributed soil loss. Methods that identify areas prone at regional scale are limited. The main objective this research develop a novel modeling approach by using three machine learning algorithms-mixture discriminant analysis (MDA), flexible (FDA), support vector (SVM) in addition an unmanned aerial vehicle (UAV) images map susceptibility loess-covered hilly region Golestan Province, Northeast Iran. In research, we have used 22 indices/factors 345 identified pipes as predictors dependent variables. maps were assessed area under ROC curve (AUC). Validation results showed AUC for mentioned algorithms varied from 90.32% 92.45%. We concluded proposed could efficiently produce map.