River flow simulation using a multilayer perceptron-firefly algorithm model

River flow estimation using records of past time series is importance in water resources engineering and management and is required in hydrologic studies. In the past two decades, the approaches based on the artificial neural networks (ANN) were developed. River flow modeling is a non-linear process and highly affected by the inputs to the modeling. In this study, the best input combination of the models was identified using the Gamma test then MLP–ANN and hybrid multilayer perceptron (MLP–FFA) is used to forecast monthly river flow for a set of time intervals using observed data. The measurements from three gauge at Ajichay watershed, East Azerbaijani, were used to train and test the models approach for the period from January 2004 to July 2016. Calibration and validation were performed within the same period for MLP–ANN and MLP–FFA models after the preparation of the required data. Statistics, the root mean square error and determination coefficient, are used to verify outputs from MLP–ANN to MLP–FFA models. The results show that MLP–FFA model is satisfactory for monthly river flow simulation in study area.