Spatio-temporal prediction of water quality parameters of a reservoir using genetic programming and least square support vector machine

Abstract

In water quality, spatio-temporal modelling is used for classification or pattern recognition by using statistical techniques. Sometimes due to extreme environmental conditions or insufficient water quality testing locations, it is difficult to find water quality parameters of previous time series. In this study, an attempt has been made to develop spatio-temporal prediction model using genetic programming and least square support vector machines. Four water quality parameters from Gangapur, Kadwa and Nandur Madmeshwar reservoirs are used for the prediction of the same parameters for the next time step of Nandur Madmeshwar reservoir. As input data is small and only from three locations, it is a great challenge for the prediction. Performance of models is assessed by coefficient of determination, root mean square error and correlation coefficient. Models are also evaluated by using band error, to know the upper limits of the water quality parameters for which the water quality standards have been violated