Estimation of the Qualitative Parameters of the Tajan River in Mazandaran Province Using Supporting Vector Machines and Expression Gene Programming Models

Due to the importance of predicting and monitoring river water quality parameters, in the present study, the capability of two models of support vector machines (SVM) and gene expression planning methods (GEP) used to estimate two water quality parameters namely the TDS and SAR in the Tajan River. To estimate the TDS and SAR, parameters which the measurement of these parameters are easy and has low cost, were used as input data to the model. These parameters included the electrical conductivity of water, acidity, sodium, calcium, potassium, the adsorption ratio of sodium, magnesium, chlorine, sulfate, bicarbonate, and river flow. SPSS was used to determine the correlation between independent and dependent variables. Based on data analysis using step-by-step method, different scenarios of combining input data for TDS and SAR prediction were considered. Comparison of the results using statistical criteria showed that the use of three variables, EC, SO4 and SAR in estimating TDS in Tajan river, had the highest correlation coefficient and the lowest error rate. The Na and EC variables also give the best results in estimating the river SAR parameter. In general, the obtained results show the high performance of gene expression planning method in estimating the quality parameters of the Tajan river and therefore it can be used to predict the qualitative parameters of rivers.