Modeling the water quality index and climate variables using an artificial neural network and non-linear regression

This study aims to investigate the relationship between the water quality index (WQI) for irrigation purposes and four independent climate variables. Our case study was conducted on the Euphrates River within Karbala city, Iraq over the period between 2008 to 2016. The Bhar-gava WQI was calculated using nine physicochemical parameters, the electrical conductivity (EC), total dissolved solids, turbidity, pH, and calcium, magnesium, sodium, chloride and sulfate levels. The Bhragava WQI classified the Euphrates river as generally "good". Artificial neural network (ANN) and non-linear regression models were developed and used to forecast the relationship between the WQI and four independent climate variables (temperature, relative humidity, and rainfall depth and sunshine duration). The non-linear regression model was adopted to predicate the WQI because the coefficient of determination and minimum error value were better than those obtained with the ANN model. The non-linear model matched the calculated Bhargava WQI values and recorded meteorological data with a coefficient of determination (R2) = 78.2 and standard error = 2.1.