Efficiency of artificial neural network in intelligent estimation of daily suspendedsediment in some of the selected gauging stations in Golestan province

The sediment estimation in rivers in order to mitigate flood damages is one of the important problems for the experts.Thus they have tried to develop hydrological and hydraulic models in order to predict the sediment load. In this study, artificial neural network model based on the multi-layer model with back propagation algorithm and non-linear TangAxon function was used for stimulating the suspended sediment in three gauging stations; Ramian in Ghare-Chai, Taghi Abad in Jaffar Abad and Sarmo in the Mohammad Abad River basins. 80 percent of the existing data records used to train the model and also 20 percent of rest for testing its. RMSE (Root Mean square Error), MAE and R2 indices were used to estimate the accuracy of the neural network model. The results obtained from calculating statistical indices indicated high accuracy of the artificial neural network models to predict suspended sediment with R2 greater than 0.99, RMSE lower than 0.0166 and MAE lower than 16.45. Also Nash-Sutcliff coefficient calculated for the stations showed high accuracy of the model to the prediction of suspended sediment.