Use of Multiple Regression Analysis for Predicting Trihalomethane Formation in Water Supply

Abstract

Trihalomethanes trihalomethanes represent between 5 and 20% of the total chlorinated products formed during the chlorination process in water treatment. Trihalomethanes are carcinogenic in nature and consequently their presence in drinking water has raised much concern both at national and international levels. The water quality data was collected for Mumbai water supply for the concentrations of individual species of trihalomethanes, chloroform, bromodichloromethane, dibromochloromethane and bromoform in finished water at the outlet of treatment process and master balancing reservoir. A multiple regression model for chloroform formation with respect to pH, turbidity and total organic carbon was generated for predicting trihalomethanes in the finished water at the outlet before leaving the treatment plant for the consumers, using the data of Panjrapur, Bhandup, Tulsi and Vehar treatment plants. In general this model was found to give acceptable fits. Overall 80% of the predicted values are within the 95% confidence interval.