A self-tuning ANN model for simulation and forecasting of surface flows

Artificial neural networks (ANN) are applicable for and forecasting without the need to calculate complex nonlinear functions. This paper evaluates the effectiveness of temperature, evapotranspiration, precipitation and inflow factors, and the lag time of those factors, as variables for simulating and forecasting of runoff. The genetic algorithm (GA) is coupled with ANN to determine the optimal set of variables for streamflow forecasting. The minimization of the total mean square error (MSE) is considered as the objective function of the ANN-GA method in this paper. Our results show the effectiveness of the ANN-GA for simulating and forecasting runoff with consistent accuracy compared with using pure ANN for runoff simulation and forecasting.