An Investigation of Antecedent Precipitation Index Role in River Flow Forecasting Using Artificial Neural Network (Case Study: Bakhtiari River)

In hydrology, there has been a virtual explosion in the use of artificial neural networks (ANNs) over the last 10 years. However, most of the recent ANN research have been devoted to comparing ANNs and more established rainfall runoff models or to assessing ANN training algorithms, while norms are still lacking that would help hydrologists to create and train efficient ANN models in a systematic way. In the present study was used multi layer perceptron for forecasting of Bakhtiari's of daily inflow. For this purpose was examined role of antecedent precipitation index in rainfall runoff process and this parameter added to model inputs and considerable improvement was resulted in forecasting result (with determination coefficient of 0.94 in verification step). The results of sensitivity analysis verified that one day ago inflow and 7 days ago precipitation in Tangpanj station are important parameters in river daily inflow forecasting.