Pedro Avellaneda

In urban to peri-urban watersheds such as those surrounding San Francisco Bay, stormwater runoff is a major pathway by which contaminants enter aquatic ecosystems. We evaluated the occurrence of 154 organic contaminants via liquid chromatography coupled to tandem mass spectrometry, including organophosphate esters (OPEs), bisphenols, per- and polyfluoroalkyl substances (PFASs), and a suite of novel urban stormwater tracers (SWCECs; i.e., vehicle-derived chemicals, pesticides, pharmaceuticals/personal care products, benzothiazoles/benzotriazoles). Time-averaged composite sampling focused on storms in highly developed watersheds over four wet seasons, with complementary sampling in less-urban reference watersheds, near-shore estuarine sites, and the open Bay. Of the targeted contaminants, 68 (21 SWCECs, 29 OPEs, 3 bisphenols, 15 PFASs) were detected in ≥10 of 26 urban stormwater samples. Median concentrations exceeded 500 ng L⁻¹ for 1,3-diphenylguanidine, hexa(methoxymethyl)melamine, and caffeine, and exceeded 300 ng L⁻¹ for 2-hydroxy-benzothiazole, 5-methyl-1H-benzotriazole, pentachlorophenol, and tris(2-butoxyethyl) phosphate. Median individual PFAS concentrations were −1, with highest concentrations for PFHxA (180 ng L⁻¹), PFOA (110 ng L⁻¹), and PFOS (81 ng L⁻¹). In six of eight urban stormwater samples analyzed for 6PPD-quinone (a tire rubber-derived transformation product), concentrations exceeded coho salmon acute toxicity thresholds, suggesting (sub)lethal impacts for sensitive species. Observed concentrations were generally significantly higher in highly developed watersheds relative to reference watersheds, but not statistically different in near-shore estuarine sites, suggesting substantial transient exposure potential at stormwater outfalls or creek outflows. Results emphasized the role of stormwater in contaminant transport, the importance of vehicles/roadways as contaminant sources, and the value of monitoring broad multi-analyte contaminant suites to enable comprehensive source and toxicity evaluations.

As stormwater control measures (SCMs) capture surface runoff from impervious areas, a shift in the water balance and flow regime components may emerge in urban watersheds, but the amount of SCM treatment needed to detectably shift these components may vary. We used the Soil and Water Assessment Tool (SWAT) hydrologic model to assess the sensitivity of 16 hydrologic metrics as an increasingly dense rain garden SCM network was applied across the West Creek watershed, near Cleveland, Ohio (USA). As the area treated by SCMs increased, annual baseflow increases matched decreases in surface runoff, while water yield and evapotranspiration changes remained small. The stream's peak response to rainfall decreased with SCM implementation across storm sizes, ranging from the threshold rainfall depth (4.8 mm) to values higher than the design storm of a single rain garden (19 mm). SCM networks draining >20% of directly connected impervious area (DCIA) significantly decreased the magnitude of discharges with a return period of less than 1 year, the percentage of time above mean flow, and flashiness. Recession slopes and annual 1‐ and 7‐day low flows exhibited a slight response that fell within uncertainty limits of the model. Water balance and rainfall response metrics exhibited the greatest sensitivity to different intensities of stormwater management, while infrequent high and low flows were resistant to detectable change even at high levels of SCM treatment when model uncertainty was included.

Small streams often lack reliable hydrological data. Environmental agencies play a key role in providing such data; however, these agencies are often challenged by the growing monitoring needs and lack of funding. Given the spatial mismatch between observed data and small watersheds/headwaters, local volunteers can act as potentially valuable research partners. We examine how CrowdHydrology, a citizen science program that collects stream stage and stream temperature observations, improves a hydrologic model of the Boyne River, Michigan, USA. Volunteers provided observations at four calibration sites with different interarrival times of the observations. We tested whether stream stage and stream temperature observations (measured by volunteers) improved the performance of a Soil and Water Assessment Tool (SWAT) model of the Boyne River. Observations were integrated into the model using the ensemble Kalman filter. This framework allowed us to integrate observation error, track the variability of model parameters, and simulate daily streamflow and stream temperature across the watershed. Measures of daily model performance included the Nash‐Sutcliffe efficiency, modified Nash‐Sutcliffe efficiency (E_f‐mod), refined index of agreement (d_r), and relative bias (Bias). For all calibration sites, estimates of streamflow improved after data assimilation compared to simulations based on initial/default SWAT parameters. Different measures of model performance emerged based on the interarrival times of the observations. Results demonstrate that observations collected by local volunteers, with a certain temporal resolution, can improve SWAT hydrological models and capture central tendency.

Improving hydrological models with the assimilation of data collected by citizens

Boyne River computational model

Deleterious effects of urban stormwater are widely recognized. In several countries, regulations have been put into place to improve the conditions of receiving water bodies, but planning and engineering of stormwater control is typically carried out at smaller scales. Quantifying cumulative effectiveness of many stormwater control measures on a watershed scale is critical to understanding how small‐scale practices translate to urban river health. We review 100 empirical and modelling studies of stormwater management effectiveness at the watershed scale in diverse physiographic settings. Effects of networks with stormwater control measures (SCMs) that promote infiltration and harvest have been more intensively studied than have detention‐based SCM networks. Studies of peak flows and flow volumes are common, whereas baseflow, groundwater recharge, and evapotranspiration have received comparatively little attention. Export of nutrients and suspended sediments have been the primary water quality focus in the United States, whereas metals, particularly those associated with sediments, have received greater attention in Europe and Australia. Often, quantifying cumulative effects of stormwater management is complicated by needing to separate its signal from the signal of urbanization itself, innate watershed characteristics that lead to a range of hydrologic and water quality responses, and the varying functions of multiple types of SCMs. Biases in geographic distribution of study areas, and size and impervious surface cover of watersheds studied also limit our understanding of responses. We propose hysteretic trajectories for how watershed function responds to increasing imperviousness and stormwater management. Even where impervious area is treated with SCMs, watershed function may not be restored to its predevelopment condition because of the lack of treatment of all stormwater generated from impervious surfaces; non‐additive effects of individual SCMs; and persistence of urban effects beyond impervious surfaces. In most cases, pollutant load decreases largely result from run‐off reductions rather than lowered solute or particulate concentrations. Understanding interactions between natural and built landscapes, including stormwater management strategies, is critical for successfully managing detrimental impacts of stormwater at the watershed scale.

In this study, we evaluated the cumulative hydrologic performance of green infrastructure in a residential area of the city of Parma, Ohio, draining to a tributary of the Cuyahoga River. Green infrastructure included the following spatially distributed devices: 16 street‐side bioretention cells, 7 rain gardens, and 37 rain barrels. Data consisted of rainfall and outfall flow records for a wide range of storm events, including pretreatment and treatment periods. The Stormwater Management Model was calibrated and validated to predict the hydrologic response of green infrastructure. The calibrated model was used to quantify annual water budget alterations and discharge frequency over a 6 year simulation period. For the study catchment, we observed a treatment effect with increases of 1.4% in evaporation, 7.6% in infiltration, and a 9.0% reduction in surface runoff. The hydrologic performance of green infrastructure was evaluated by comparing the flow duration curve for pretreatment and treatment outfall flow scenarios. The flow duration curve shifted downward for the green infrastructure scenario. Discharges with a 0.5, 1, 2, and 5 year return period were reduced by an average of 29%. Parameter and predictive uncertainties were inspected by implementing a Bayesian statistical approach.