Estimating of Soil Erosion Covered with Rolled Erosion Control Systems Using Rainfall Simulator (Neuro-fuzzy and Artificial Neural Network Approaches)

Abstract

5 Abstract: Due t o di fficulties in di rect measuring of soil erosion by simulated r ainfall; at present research a lternative met hods consisting of neuro-fuzzy and artificial neural network have been us ed for p redicting of ru n off w ater a nd splash of so il particles fo r e ach run in soils covered with RECs and bare s oils. This study investigates reduction of splash det achment and overland flow of a clayey soil after covering by different synthetic geotextiles using rainfall simulator and small flume f acilities. The soil sample w as taken from an unstable sloping m arl hill in I ran. Then, the data s et was divided into t wo s ubsets for calibration ( 80%) and testing ( 20%) of the model s. For predicting of mentioned parameters, neuro-fuzzy, artificial neural network and multivariate r egression were applied. It is found that the neuro-fuzzy model gives better estimates than the other techniques. A fter n euro-fuzzy model, artificial n eural n etwork had better accuracy than multivariate regression for p rediction of ru n off a nd splash.