Estimation of water quality parameter using soft computing approach:a case study of the Karun River, Iran

The quality of surface water is a critical factor in evaluating its suitability for irrigation. This study employeddata from the Mollasani gauge station to analyze the total dissolved solids (TDS) content of the water bycalculating various water quality parameters, including sodium adsorption rate (SAR), sodium percentage(Na%), Kelly index (KI), and permeability index (PI). SAR, KI, and Na% values ranged from ۱.۷۶ to ۹.۶۳meq/L, ۰.۴۱۲ to ۲.۰۸۶, and ۲۹.۸۸ to ۶۷.۶۴%, respectively. Approximately ۳۵% of the water samples wereclassified as suitable for irrigation based on their PI values. TDS values obtained from the study area variedbetween ۳۸۴ and ۲۱۲۵ mg/L. To effectively predict TDS values, an artificial neural network (ANN) modelwas developed using the aforementioned water quality parameters. Seven ANN models were trained, andtheir statistical performance was assessed using R۲ and RMSE. The best model achieved an outstandingtesting R۲ value of ۰.۹۴۸, demonstrating its superior predictive power. The strong association between theANN model's estimation results and experimental data (R۲ = ۰.۹۶) further confirms its effectiveness. Thisstudy establishes ANN as a reliable and efficient method for estimating TDS based on various water qualityparameters.

کلیدواژه ها:

artificial neural network ، Water quality prediction ، Machine Learning model ، Karun River.