عنوان مقاله [English]
نویسندگان [English]چکیده [English]
Runoff resulting from rainfall is a complex and non-linear process and, therefore, its modeling is not so easy. The aim of this study was the application of the Gamma test to select the optimal combination of input variables for runoff modeling in Amameh watershed. M-test was used to identify the optimal number of required data for modeling. Data of rainfall (P(t)) and runoff (R(t)) in daily time scale were used in the period of 2000 to 2009. Totally, eight input variables namely four variables of daily streamflow: lag-1 (R(t-1)), lag-2 (R(t-2)), lag-3 (R(t-3)) and lag-4 (R(t-4)) as well as four daily rainfall variables: without lag time (P(t)), lag-1 (P(t-1)), lag-2 (P(t-2)) and lag-3 (P(t-3)) were used. Streamflow modeling was performed based on the optimum number of the selected variables using the artificial neural network (ANN) and local linear regression (LLR) methods. The results showed that the six variables of P(t), P(t-1), P(t-2), P(t-3), R(t-1) and R(t-2) belonged to the optimum combination of variables in streamflow modeling of the mentioned watershed. Moreover, based on the M-test output, only 1405 points were found to be adequate for modeling in the training section. Results indicated the fact that the LLR method had greater accuracy in training process compared to ANN. However, the ANN had large amount of accuracy in the model testing process. In training section the R2 and RMSE values were found to be equal to 0.96 and 1.7, respectively.