Figure 3. Daily rainfall (top) and ENT concentrations (bottom) during flood (perigean tide) and baseline conditions throughout June and July 2022. The red line represents the EPA’s single sample maximum threshold concentration for safe public use (104 MPN 100 mL^-1), and the grayed area represents the minimum detection limit (below 10 MPN 100 mL^-1 for samples diluted 10:1). Gray and purple points correspond to measurements taken in Taylor’s Creek during basline and flood conditions at the three sampling locations, respectively, while teal points correspond to measurements taken from roadway floodwaters above a single stormwater catchment grate in the Queen Street network. Samples were processed in duplicate, and duplicate measurements are connected by vertical lines to visualize uncertainty.

We found there were significant differences in ENT concentrations by sampling location (alpha < 0.05). During baseline conditions, the Orange Street Outfall had the highest median concentration of ENT in the waterway (31 MPN 100 mL^-1) while the sampling locations near the Museum and Overland Outfalls showed lower median ENT concentrations (both 10 MPN 100 mL^-1, which corresponds to the minimum detection limit). Inter-site variability between outfall sampling locations was also noted by Price et al. (2021), who posited that site-specific infrastructure can cause localized water quality dynamics.

For June, the median ENT concentration was 20 MPN 100 mL^-1 and had an interquartile range of <10 to 23 MPN 100 mL^-1 while the July median concentration was 41 MPN 100 mL^-1 and had an interquartile range of 10 to 86 MPN 100 mL^-1. Rainfall occurred more frequently in July than in June. Periods of increased rainfall are known to correspond with increased FIB concentrations in waterways due to runoff-driven contamination and elevated surficial groundwater elevations, which can also lead to greater sewage exfiltration (Secru et al., 2011). The modeling results of baseline conditions, which show significant relationships between log₁₀ ENT and prior 3, 6, 12, 24, 48, and 72 hour rainfall totals (alpha < 0.05), reflect the influence of rainfall on the variance of bacterial concentrations (Table 1).

Table 1. Linear regression models of baseline conditions (top) and perigean tide conditions (bottom) between log₁₀-transformed ENT concentrations (log₁₀ MPN) and tidal or rainfall factors.

3.3 Roadway floodwaters had consistently high ENT concentrations

We visually observed nine minor floods localized to the Queen Street stormwater network, and no roadway flooding via the Orange Street and Museum stormwater networks during perigean tides. The presence of flooding was corroborated by the storm drain water level measurements and the pipe network model. The floodwater samples exceeded the EPA’s single sample maximum threshold concentration for safe public use (104 MPN 100 mL^-1) for 8 of the 9 high tide flood events, in many cases by an order of magnitude (Fig 3). These ENT concentrations are comparable to the concentrations reported by Macías-Tapia et al. (2021, 2023) in Norfolk, VA in 2017, when 95% of the ENT samples exceeded the single sample maximum threshold concentration and ENT concentrations ranged from 30 to >24,000 MPN 100 mL^-1.

In June, the median floodwater concentration was 710 MPN 100 mL^-1 and concentrations ranged from 41 to 9,804 MPN 100 mL^-1 (Fig 3). By comparison, in July, the median floodwater concentration was 9,563 MPN 100 mL^-1 and concentrations ranged from 1,882 to >24,196 MPN 100 mL^-1. The perigean high tide elevations were similar between June and July (0.35-0.51 and 0.34-0.45 m MHHW, respectively), as was the modeled inundation percentages at higher high tide (75-87 and 73-81%), so we hypothesize that differences in concentrations stem from antecedent rainfall factors. These findings suggest that antecedent rainfall can enhance fecal contamination in roadway floodwaters during high tide flood events. But, importantly, given that the June floodwaters were still more than 6 times higher than the EPA standard in the absence of rainfall, complete inundation of stormwater networks by tides in drier conditions can create concerning water quality conditions in roadway floodwaters.

The sources of fecal bacteria in the roadway floodwater samples are unknown. We believe the primary fecal bacteria sources during the high tide floods sampled were within the stormwater network. We observed fecal contamination in roadway floodwaters in the absence of rain, due solely to stormwater network inundation by tides, indicating a source from either Taylor’s Creek or the stormwater network. However, water quality samples collected near stormwater outfalls in Taylor’s Creek did not have similarly high concentrations (Fig 3), so Taylor’s Creek is likely not a primary source of fecal bacteria to the floodwaters. Lastly, the majority of the roadway floods were small and confined to puddles around the storm drains (Fig 4), indicating that the floodwaters likely did not flush fecal matter from the land surface.