Total Dissolved Solids modeling using machine learning algorithms in periods of low and high water (Case study: Khorammabad, Biranshahr and Alashtar watersheds, Lorestan province)

In this study, five soft computing techniques, GP-PUK, GP-RBF, M5P, REEP Tree and RF were used to predict the SL in Cham Anjir, Bahram Joo, Kaka Reza and Sarab Syed Ali hydrometry stations in Khorramabad, Biranshahr and Alashtar sub-watersheds, Lorestan province. Total data set consists of rain, discharge and solute load (SL) of three sub-watersheds out of which 70% data used to training and 30% data were used to testing phase. Finally, the models’ accuracy was assessed using three performance evaluation parameters, which were Correlation Coefficient (C.C.), Root Mean Square Error (RMSE) and Maximum Absolute Error (MAE). Results suggest that GP-PUK and GP-RBF models works well than other modeling approaches in estimating the SL in low and high water-periods. The result showed that, In the high-water period, in Cham Anjir, Sarab Said Ali and Kaka Reza stations the GP-RBF model and in the Bahram Joo station the GP-PUK model with the highest C.C and the lowest error were selected the optimal models in estimating the SL. Also, in the low water period, result shown that in Cham Anjir, Sarab Said Ali and Bahram Joo stations the GP-RBF model and in the Kaka Reza station the GP-PUK model were the best models in estimating the SL. Therefore, these models can be used to estimate the solute load of nearby rivers by/without hydrometry station for the management of the quantity and quality of surface water.