The presence of human-derived fecal coliform bacteria (sewage) in streams and rivers is recognized as a human health hazard. The source of these human-derived bacteria, however, is often difficult to identify and eliminate, because sewage can be delivered to streams through a variety of mechanisms, such as leaking sanitary sewers or private lateral lines, cross-connected pipes, straight pipes, sewer-line overflows, illicit dumping of septic waste, and vagrancy. A multiple-tracer study was conducted to identify site-specific sources of sewage in Accotink Creek, an urban stream in Fairfax County, Virginia, that is listed on the Commonwealth's priority list of impaired streams for violations of the fecal coliform bacteria standard. Beyond developing this multiple-tracer approach for locating sources of sewage inputs to Accotink Creek, the second objective of the study was to demonstrate how the multiple-tracer approach can be applied to other streams affected by sewage sources. The tracers used in this study were separated into indicator tracers, which are relatively simple and inexpensive to apply, and confirmatory tracers, which are relatively difficult and expensive to analyze. Indicator tracers include fecal coliform bacteria, surfactants, boron, chloride, chloride/bromide ratio, specific conductance, dissolved oxygen, turbidity, and water temperature. Confirmatory tracers include 13 organic compounds that are associated with human waste, including caffeine, cotinine, triclosan, a number of detergent metabolites, several fragrances, and several plasticizers.

To identify sources of sewage to Accotink Creek, a detailed investigation of the Accotink Creek main channel, tributaries, and flowing storm drains was undertaken from 2001 to 2004. Sampling was conducted in a series of eight synoptic sampling events, each of which began at the most downstream site and extended upstream through the watershed and into the headwaters of each tributary. Using the synoptic sampling approach, 149 sites were sampled at least one time for indicator tracers; 52 of these sites also were sampled for confirmatory tracers at least one time. Through the analysis of multiple-tracer levels in the synoptic samples, three major sewage sources to the Accotink Creek stream network were identified, and several other minor sewage sources to the Accotink Creek system likely deserve additional investigation.

Near the end of the synoptic sampling activities, three additional sampling methods were used to gain better understanding of the potential for sewage sources to the watershed. These additional sampling methods included optical brightener monitoring, intensive stream sampling using automated samplers, and additional sampling of several storm-drain networks. The samples obtained by these methods provided further understanding of possible sewage sources to the streams and a better understanding of the variability in the tracer concentrations at a given sampling site. Collectively, these additional sampling methods were a valuable complement to the synoptic sampling approach that was used for the bulk of this study.

The study results provide an approach for local authorities to use in applying a relatively simple and inexpensive collection of tracers to locate sewage sources to streams. Although this multiple-tracer approach is effective in detecting sewage sources to streams, additional research is needed to better detect extremely low-volume sewage sources and better enable local authorities to identify the specific sources of the sewage once it is detected in a stream reach.