Prediction of River Water Quality Parameters Using Soft Computing Techniques

Water is the most important substance for life on earth and every living being need freshwater to survive. Besides various sources of water, river water is all-important source of freshwater. Due to rapid urbanization, industrialization, religious and social practices on the banks of rivers, the river water gets polluted and it is one of the major issues in India. So, the need of hour is to keep a continuous check on the quality of river water parameters. Various researchers have developed accurate prediction models to estimate the future quality of river water with least forecasting errors. Autoregressive time series models have been developed to generate linear forecast only and most of them are unable to handle nonlinear problems. To handle such nonlinear problems, artificial neural network (ANN) and adaptive neuro-fuzzy interface system are found to be most efficient tool for accurate prediction. Besides these methods, wavelet decomposition tool for analyzing nonlinear situations has been used to generate forecast values close enough to observed values. The biochemical oxygen (BOD) of river Yamuna at sample site of Nizamuddin (Delhi) is predicted using the past monthly averaged data. Statistical analysis provides basis to understand the nature of wavelet domain constitutive series. The prediction results obtained using neuro-fuzzy-wavelet coupled model generates more accurate outcomes as compared to neuro-fuzzy, ANN and regression models.