The Effects of Daily Sequences of Meteorological and Hydrological Variables in Estimating of River Discharge A Case study of Kasilian Watershed

Rainfall-runoff relationship is one the most complicated issues in hydrological cycle and its accurate estimation is one of the most important concerns in water resources engineering and management. Nowadays, researchers have focused on the application of intelligent systems for forecasting of river discharge due to their high performance and addition no need for many data and complex relationship. Therefore, the multi-layer perceptron and radial basis function approaches were used in this research for river flow prediction of Kasilian watershed. Some variables such as precipitation, temperature, evapotranspiration, relative humidity and discharge in daily time scale over 42 year period were considered while 40 different scenario structures were assessed for river flow prediction. The results showed that among 5000 available models for estimation of river flow, the multi-layer perceptron including nine input variables and arrangement of 9-5-1-1 has better performance than radial basis function. The MSE, RMSE and MAE of the model in training stage were 0.19, 0.43 while in validation and testing stages was 0.0 2, 0.15 and 0.02 and 0.03, 0.16 , 0.03 respectively.