Selection of smoothing parameter estimators for general regression neural networks - Applications to hydrological and water resources modelling

These are the guidelines for the program Generalised Regression Neural Networks (GRNNs). Engineering has been researching the use of artificial neural networks (ANNs) for water resources modeling applications, such as flow forecasting, water quality forecasting and water treatment process modeling since the early 1990s. While Multi‐Layer Perceptrons (MLPs) are the most widely used ANN architecture in water resources and hydrology, Generalised Regression Neural Networks (GRNNs) provide an alternative that is worth considering, especially as their structure is fixed and therefore does not have to be determined by trial‐and‐error. Consequently, this removes some of the uncertainty associated with the ANN model development process. In order to be able to implement GRNNs for research purposes, software code for developing GRNN models has been developed in PGI Visual Fortran 2008. The detailed theory and sample applications are given in 'Li