Reconstruction of river water quality missing data using artificial neural networks

Abstract

The monitoring of river water quality is important for human life and the health of the environment. However, water quality studies in many parts of the world, especially in developing countries, are restricted by the existence of missing data. In this study, the efficiency of the multilayer perceptron (MLP) and radial basis function (RBF) networks for recovering the missing values of 13 water quality parameters was examined based on data from five stations located along the Maroon River, Iran. The monthly values of other existing water quality parameters were used as input variables to the MLP and RBF models. According to the achieved results, the hardness missing values were estimated precisely by both the MLP and RBF networks, while the worst performance of the networks was found for the turbidity parameter. It was also found that the MLP models were superior to the RBF models to reconstruct water quality missing data.