Water Quality Sleuthing: Applying Multiple Methods to Identify Pollutant Sources in Urban Streams

Introduction
Urban streams are impacted by a wide range of hydrologic stresses and pollutant sources. Localities that seek to maintain water quality standards in local streams, must identify pollution sources or causes of degradation. Although this can be readily accomplished for many streams, other water bodies, especially in urban settings, present unusual water quality patterns or have pollutants of unknown origin. This paper describes how a combination of traditional and modern methods was applied to understand pollutant sources and improvement opportunities in two urban streams The City of Durham (NC) identified water quality issues on two streams through water quality monitoring. Sandy Creek Tributary A receives stormwater runoff from highly impervious commercial area (Figure 1), and had unusually low dissolved oxygen concentrations, detected during routine ambient monitoring (City of Durham, 2017). Warren Creek drains mixed land uses and had elevated concentrations of metals and polycyclic aromatic hydrocarbons (PAHs) in sediments (AECOM, 2018). The City's existing pollution investigation methods, available data and mapping information did not point to obvious sources/causes of these effects. The City initiated a more detailed investigation in 2020, with goal of identifying both pollutant sources and potential improvement projects. ###Methods
The project team applied multiple methods to identify causes of low dissolved oxygen in Sandy Creek Tributary A. The study began in fall 2020 with evaluation of the stream channel and flow conditions, which documented the extent and nature of urban hydrologic impacts. Additional water quality sampling was performed to detect indicators of potential sources (e.g., wastewater) and characterize spatial variability in water quality within the stream. In fall 2021, the team visually inspected and sampled portions of the stormwater sewer system to identify illicit discharges. This effort included microbial source tracking with the HF183 human marker. A thermal imaging survey was conducted in the stream to identify potential seeps of low-oxygen groundwater to the stream. For Warren Creek, the project team performed additional sampling of metals and PAHs in sediments and water. Sampling locations were designated to provide additional insights into the spatial distribution of these constituents, including whether they demonstrated a hotspot versus distributed pattern. Risk assessments were performed to determine if the observed concentrations were of concern to human or ecological health.
Results
The field investigations revealed several pollutant sources in the upper drainage area of Sandy Creek Tributary A, including a sewer leak, a broken water pipe, and a leaking dumpster (Figure 2). The sewer leak was noteworthy because traditional wastewater indicators (e.g., ammonia, CBOD5) were either low or only slightly elevated in the stream, but the leak was confirmed by both visual inspection and extremely high levels of the HF183 human marker. The effects of these pollutant sources were exacerbated by urban hydrologic impacts on the stream. These included rip rap dams and debris blockages that created stagnant conditions during dry weather periods (Figure 3). Water quality sampling and the thermal imaging survey confirmed that the stream receives inputs of low-oxygen groundwater that help maintain low-dissolved conditions for several thousand feet of stream length (Figure 4). The Warren Creek investigation revealed that metals and PAH concentrations were relatively low at most locations. Although some concentrations exceeded screening levels, the more detailed risk assessments concluded that they were not of concern to human or ecological health. The array of PAHs by molecular weight was similar at all sites, which pointed toward a regional or distributed source rather than hot spots or remediable sources.
Recommended Improvement Projects
Based on the results of the investigation, recommended improvement projects to increase dissolved oxygen levels in Sandy Creek Tributary A included: (1) fixing the identified pollutant sources in the stormwater system, such as the sewer and water line break; (2) in-stream hydrologic improvements to increase stream velocity and aeration; and (3) continued application of MS4 control measures. It was specifically recommended to prioritize the watershed for illicit discharge detection and elimination. Because the Warren Creek contaminants were at relatively low levels and not associated with specific hot spots, the team recommended monitored natural attenuation in combination with MS4 control measures such as good housekeeping, public education, and management of construction site runoff.
Summary and Conclusions
Ambient water quality monitoring networks are useful for characterizing status and trends in local water quality. But more in-depth investigations are needed to accurately characterize pollutant sources and identify effective improvement projects. Traditional ambient water quality sampling methods can be limited because the typical indicators of pollutant sources might be diluted or rapidly attenuated. For example, in this study, ambient concentrations of CBOD5 and ammonia were not high enough to conclusively identify sewage as a pollutant source in Sandy Creek Tributary A. Visual inspection and microbial source tracking were required to identify this important and fixable pollutant source. The study also emphasized the need to understand urban stream hydrology when diagnosing water quality problems. Although pollutants exert oxygen demand, the ability of urban streams to reaerate themselves is an important factor in the stream oxygen budget (Blaszczak and others, 2019). Shallow groundwater tends to be low in dissolved oxygen, and so surface water – groundwater interactions can influence in-stream oxygen concentrations (Harvey and others, 2013). Thermal imaging represents an innovative method for identifying groundwater seeps to streams. This method is most effective in streams that receive groundwater inputs as discrete seeps or springs rather than as highly diffuse inputs.