Utilizing artificial neural network for forecasting groundwater table depths fluctuations

Abstract

This study presented the model of predicting the water table fluctuation in flood plain of Sepidroud watershed (North of Iran-Gilan). The model for prediction of water table depth was developed leaning on artificial neural network. The neural network with different numbers of hidden layer neurons was developed by using 4 years (2004-2007) monthly rainfall, potential evapotranspiration and influencing wells as input and water table depth as output. The best model was selected based on mean square error. The results showed that artificial neural network could be used to predict water table depth in aquifer with good convergence and maximum error was 5% approximately.