Using Parametric and Nonparametric Methods to Model Total Organic Carbon, Alkalinity, and pH after Conventional Surface Water Treatment

摘要

Predicting the behavior of natural organic matter (NOM), alkalinity, and pH during drinking water coagulation is difficult because of the heterogeneous chemical nature of NOM and the complexity of carbonate chemistry. Parametric and nonparametric statistical regression methods were implemented to model the removal of NOM, as measured by total organic carbon (TOC), from raw water by conventional surface water treatment and to track the behavior of pH and alkalinity. The United States Environmental Protection Agency (U.S. EPA) Information Collection Rule (ICR) database was sampled for raw water and postsedimentation data from conventional surface water plants. All models were evaluated in terms of their fit and predictive capability, and for all variables explored, the nonparametric local polynomial models outperformed their parametric linear least-squares counterparts. This was most pronounced with the pH model, and was attributed to the nonlinear relationship found between pH and one of the predictors. Between the sedimentation basin and the plant effluent, alkalinity was found to remain relatively constant, TOC decreased by 12% by filtration, and pH increased, consistent with chemical additions required to minimize corrosion in the distribution system. Modeling efforts in this article are meant to be complementary to previous chemical and process models of water treatment.