Estimation of Mordagh Chay River water quality using gaussian process regression method

Abstract

Accurate modeling of river quality parameters is essential for environmental planning, optimal operation of reservoirs, designing of hydraulic structures and irrigation planning. Considering that direct measurement of quality parameters is time consuming and costly, it is possible to predict these parameters with less time and cost and with high accuracy using artificial intelligence methods. In this regard and in the present study, the electrical conductivity parameter of Mordagh Chay River has been estimated using the gaussian process regression method. Based on this, the amounts of calcium, magnesium, sodium, chlorine and sulfate of this river on a monthly scale over a period of 47 years (1971-2018) were used as input parameters of the models. Statistical parameters of correlation coefficient, scattering index and Wilmott’s index were also used to compare the obtained results with the observed values. Finally, the obtained results showed that the GPR5 model with SI of 0.093 and WI of 0.995 had the best performance. It was also generally concluded that using the models used in this study, the EC value in the Mordagh Chay River can be estimated with appropriate accuracy.