Table 5 summarizes the models for predicting HAAs in PP
and HWT. The R2 values for the linear regression models range from 0.83 to 0.96. The RMSE for the nonlinear models were
observed to be 22.3 and 14.6 for the HAAs in the PP and HWT,
respectively. The analysis of variance (ANOVA) showed that
the models were statistically significant (P < 0.0001), and
residuals were normally distributed and presented no visible
trend. Consistent with the THMs, HAAs and free chlorine
residuals in the WDS were found be statistically significant in
all linear regressions, while temperature in the PP and TOC at
the WDS were not found to be significant in any model. This
may be explained by the fact that higher temperatures in the
PP accelerate microbiological activity, thereby reducing HAAs
in the PP. Conversely, higher temperature accelerates reactions
between residual NOM and free chlorine residuals in the
PP, which may increase HAAs formation in the PP. Due to
these combined effects, implications of temperature on HAAs
formation in the PP may not be noticeable. A better understanding
of the role of temperature on microbiological activity
in the PP may be essential to obtain a clearer picture of the
scenarios. Figs. 5 and 6 show the plots for the model predictions
and measured HAAs in the PP and HWT, respectively. In
the case of HAAs in the PP, the models successfully predicted
the peak values as well as the low values. For the HAAs in the
HWT, models predicted most of the peak values consistently
(Fig. 6).