Prediction of the Pollution Index in Al Khamissiya Canal, Thi Qar Province, Iraq

Abstract The principal aim of this study is to ascertain the suitability of Main Outfall Drain water for Al Hammar Marsh by calculating the Comprehensive Pollution Index (CPI). Twenty physicochemical parameters, including Turbidity, PH, DO, EC, TDS, PO4, Cl, Ca, Mg, K, NH4, Na, SO4, NO2, Cr, Pb, Mn, Ni, Ba, and Zn were considered. This investigation was adopted to evaluate the efficacy of ANNs and MLR in predicting comprehensive pollution indexes. The results indicate the water of the Khamissiya Canal was categorized as severely polluted water by the CPI classification and that both MLR and ANN models are highly effective in forecasting CPI. The MLR model indicated that most variables were significant in the regression model’s construction. The MLR model generated the holistic pollution index’s predicted equation. The network structure with a configuration of 12-21-1 was found to be the most effective prediction ANNs model. Training, validation, and testing sets were yielded minimum Mean Squared Error (MSE) values of 9.537×10 −7 , 4.215×10 −8 , and 8.239×10 −8 , respectively, by the structure. Furthermore, it is obtained that the highest R-value of 1 for the testing set. The ANNs model’s precision level is high. As a result, it is the preferable option due to its iterative approach, which is suitable for the production of more precise results.