Artificial neuron models for hydrological modeling

Artificial neural networks (ANNs) have been successfully employed for hydrological modeling in the last two decades or so. Most ANN hydrologic models use the McCulloch and Pitts´ artificial neuron (MPAN) as the building block of the ANN models. This paper presents the results of a study employing an artificial neuron called Generalized Neuron (GN). Specifically, two neural network models are presented in this study. The first is a traditional feed-forward neural network model trained using back-propagation algorithm (BPNN) and the second is a GN model. The rainfall and flow data from the Kentucky River Basin, USA are used to develop and test the two models. The results obtained in this study indicate that a GN model is able to capture the complex relationships hidden in rainfall and flow data. The GN model was found to perform better than the BPNN in terms of certain error statistics considered in this study.