Proxies for Carbon and Disinfection By-Product Precursor Concentrations


With growing concerns about the effects of DBPs on human health and associated changes in the regulatory standards, there is much interest in identifying the sources of DBP precursors and developing tools to better monitor and predict carbon and DBP concentrations. To identify source‑water attributes that are good predictors or “surrogates” for DBP formation, relations between concentrations of THM and HAA (in finished water and DBPFP samples) and organic carbon concentrations and optical properties were examined using Spearman rank correlations (table 11). While many of the variables were significantly positively correlated with concentrations of THM4 and HAA5 in finished water as well as with THMFP and HAAFP, the best DBP predictors were DOC concentration, FDOM and several fluorescence peaks, and total component loadings from components C2, C3, and C4. In general, the correlations between concentrations of TOC and DOC and DBPs in drinking-water samples were somewhat higher for HAA5 compared with THM4, although all were significant (p <0.001).


Laboratory DBPFPs were highly correlated with concentrations of DOC and TOC, turbidity, UVA, fluorescence peak intensities, PARAFAC component loadings, and various other indicators (r >0.9, p <0.001; table 11); r values for THMFP and HAAFP were similar. Correlations between STHMFP, filtered and unfiltered, and other constituents were not that high (r <0.4), although some correlations were significant (table 11). In contrast, SHAAFP was highly correlated with a host of other indicators; for example, SHAAFP in filtered-water was significantly correlated (r = 0.78, p <0.001) with SUVA (table 11). 


Given that DBP precursors are strongly correlated with DOC concentrations, there is much interest in identifying DOC proxies, especially for water systems where DBPs are approaching drinking-water standards. Laboratory bench-top measurements of FDOM were highly correlated (r = 0.98; p <0.001) with DOC concentration (fig. 23) as well as laboratory THMFP and HAAFP (table 11). In addition, continuous in-situ FDOM measurements were highly correlated with laboratory FDOM (r = 0.99; p <0.001; appendix F2) and provided similarly high correlations with DOC concentration (r = 0.96; p <0.001) as did the laboratory based optical measurements (fig. 23; appendix F2). These strong correlations provide convincing evidence that in-situ FDOM measurements can be effective at tracking concentrations of DOC and DBP precursors in source water. In addition to providing high-frequency, real-time data, the in-situ measurements do not require the sample collection, processing, and analyses needed for UVA measurement.


The ability to monitor DWTP inflow water quality to predict finished-water DBPFP continuously, in real-time, would be of great benefit for DWTP operations. In addition to source-water quality, however, the amount of DBPs that form during treatment is also influenced by DWTP operations, including coagulation, disinfection type and dose, contact times, pH, temperature, and other factors. Much can be learned by using this information along with monitoring feedback to adaptively manage treatment plants and optimize for DBP precursor removal.


During this study, finished-water DBPs were determined approximately monthly, resulting in 18 data points to compare in-situ FDOM measurements to finished-water THM4 and HAA5 concentrations. The correlation between in-situ FDOM and finished-water HAA5 concentration was significant (r = 0.83; p <0.001; appendix F3); applying the equation derived from this relation, the high-frequency in-situ FDOM data were used to estimate finished-water HAA5 concentrations (fig. 24). If accurate, these estimates would suggest HAA5 concentrations did not exceed the 0.06 mg/L MCL during this study. While the correlation between in-situ FDOM and chloroform was significant (r = 0.64; p <0.01), the relation between FDOM and finished-water THM4 was not, so estimates of continuous finished-water THM concentrations were not generated for this study. Further research on this topic is warranted given that fluorescence was a good predictor of laboratory THM formation potential, here and elsewhere (Hua and others, 2007; Marhaba and others, 2009; Kraus and others, 2010). It should also be emphasized that the relation between source-water quality and DBP formation potentials conducted on untreated water in the laboratory under uniform conditions is expected to be stronger than the correlation with treated (finished) water that has undergone coagulation (table 11). 


The significant, positive correlation between SHAAFP and SUVA (table 11) suggests the HAA precursor pool is made up of, if not linked to, chromophoric DOM, which may explain why there is a stronger correlation between FDOM and HAA5 concentrations compared to THM4. Prior studies indicate HAA precursors are more aromatic compared to THM precursors (Croué and others, 2000; Liang and Singer, 2003; Hong and others, 2008), and HAA precursors have been associated with terrestrially derived fulvic and humic acids (Kraus and others, 2008, 2010). However, other studies have also observed a link between HAA precursors and algal‑derived DOM (Chen and others, 2008; Hong and others, 2008; Kraus and others, 2011). 


First posted February 11, 2013