Precipitation-Based Modeling of Water Quality Within an Urban Recreational Watershed

In coastal cities, recreational water quality can be substantially impacted by stormwater runoff leading to beach advisories and potential health hazards. The Poenisch Park beach, a segment of Corpus Christi Bay, was placed on the 2014 Texas Water Quality Inventory and the 303(d) list of impaired waters for not meeting the contact recreation single sample criteria of no more than 25 percent of the measured values exceeding 104 Colony Forming Units (CFU) per 100mL for the indicator bacteria (Enterococci). A Total Maximum Daily Load (TMDL), which is a process with goals to restore impaired waters, was initiated. Components of a TMDL include modeling and monitoring of the supporting watershed. For this study, our analysis was initially based on historical Texas Beach Watch (TBW) water quality measurements. These historical measurements were complemented by flow rate and water quality measurements at the Poenisch Park stormwater outfall. Results showed that there are two distinct components to the water quality impairment in this segment: runoff from the adjacent watershed and a dry weather load of unknown origin which was estimated as 68 CFU for all model calibrations. All precipitation information for the study was based on Multi-sensor Precipitation Estimates (MPE) as there are no rain gauges located within the watershed. MPE is a radar-derived time series that covers areas of about 4km by 4km, therefore only one MPE cell was needed to assess the precipitation in this small, 0.26 km², urban watershed. Based on the TBW data, a statistical analysis of the correlation between measured Enterococci concentrations in nearshore water and precipitation within the watershed showed that elevated bacteria levels tend to occur within 72 hours following the onset of precipitation. The analysis identified 409 wet weather sampling events. Enterococci concentrations for these wet cases had a mean of 401 CFU/Most Probable Number (MPN) and a median of 35. Thirty-three percent (136 cases) of the concentrations exceeded the threshold of 104 CFU/MPN. There were 275 dry events with mean Enterococci concentrations of 68 CFU/MPN, a median of 10 CFU/MPN, and 12 percent (34 cases) exceeding the 104 CFU/MPN threshold. During the study, measured Enterococci concentrations at the stormwater outlet ranged between 17,329 MPN to 155,310 MPN for the three events sampled for water quality, while measured Enterococci concentrations in the nearshore bay waters were between 228 MPN to 55,000 MPN. Three modeling approaches were applied for the watershed all based on the same MPE time series. A numerical model took into account the characteristics of the stormwater network, a simplified step model assumed that 1/2 of the outflow was concurrent with precipitation followed by 1/3 and 1/6 of the runoff 1 hour and 2 hours later, while a purely statistical model linked directly the precipitation and bay water quality measurements. The first two modeling approaches predicted runoff volume and load at the stormwater outlet based on precipitation. Both models were based on the same informational land use map that included runoff and event concentration loads based on the respective land uses. While the numerical model started to experience saturation of its predictions and had a minimum one-hour delay between precipitation and water flowing through the system, the simplified step model did not saturate and reacted without delay to the onset of new precipitation. Furthermore, the simplified step modeling approach was more transparent, did not require special computing resources, and reflected measurements at the stormwater outlet adequately. The models linked either load predictions or directly the MPE to indicator bacteria concentrations in the recreational waters. The relationship was modeled by an exponential function with two parameters, a multiplicative mixing factor and a decay rate. The models were finally applied to several years of MPE data prior to analysis. A suggestion that was drawn from the linkage between Enterococci concentrations and precipitation was that excluding wet weather events (measurable precipitation within the past 72 hours) lowered the frequency of Enterococci measurements above the 104 CFU/MPN per 100 mL threshold by 12.4 percent. In addition, the main recommendation of the study was to avoid recreational contact at the start of precipitation within the watershed and up to 72 hours after the end of the precipitation event.