Uncontrolled stormwater runoff volume is a legacy stressor on sewer-system capacity that is further compromised by the effects of aging infrastructure. Green stormwater infrastructure (GSI) has been used in a variety of designs and configurations (for example, bioretention) with the goal of increasing evapotranspiration and infiltration in the local water cycle. In practice, GSIs have variable effectiveness in reducing runoff volume.

An urban residential site near Detroit, Michigan, called RecoveryPark was monitored for 8 years before and after GSI construction to evaluate how effectively the GSI reduced volumes of stormwater flowing to Detroit’s Water Resource Recovery Facility through combined sewer systems. In addition to the GSI, the study site included an urban farm where salad crops were grown in hoop houses. The monitoring approach was to characterize the urban water cycle through high-frequency measurements of inflows and outflows. Datasets included meteorological data, soils and sediment characteristics, groundwater levels, flows within the combined sewer system, and soils and water chemistry with specific focus on the disposition of road salt.

Although land cover within the RecoveryPark sewershed was high-density residential in the 1950s, the sewershed included only one residence within the 8.74-acre sewershed during this study. Measurements of annual precipitation at the site exceeded long-term annual averages by more than 10 inches during 3 of the 8 years of study. Potential evapotranspiration was often greater than the measured precipitation that averaged 28–34 inches per year. As compared to underlying clay-rich sediments, soils data indicated relatively permeable sediments near land surface with estimated hydraulic conductivity of 0.75 inches per hour; however, these values decreased with increasing depth. Groundwater-level data revealed increases in groundwater storage as indicated by increases in seasonal groundwater levels and development of a groundwater mound adjacent to the GSI. These increases in groundwater levels were directly adjacent to swales designed to infiltrate stormwater and only became evident after installing the GSI.

Flows within the combined sewer system included rainwater, septic effluent, groundwater infiltration, leakage from water-supply lines, and release of water stored in abandoned foundations. Dry-weather flows (no rain fell within the prior 3 days) averaged 7–10 gallons per minute, which were much greater than flows estimated by septic outflow alone. A set of estimated water budgets were compiled, and results showed large discrepancies in unaccounted flows. To further examine these discrepancies, dye-tracing within the combined sewer system helped examine the sources of water by relating flow volumes to drainage area. For one of the monitoring sites within the combined sewer system along the southeast side of the study area, flows estimated by dye concentrations were more than 10 percent greater than those measured by standard methods. Through peak-flow-regression analysis, a minimum of 2.4 million gallons of water per year were infiltrated or lost to evapotranspiration because of GSI construction. After site modifications were made by excavating gravel drains to improve drainage characteristics, estimated stormwater volumes within the combined sewer system returned to near preconstruction levels. The GSI was effectively bypassed to address slow infiltration rates and standing water; the bypass all but eliminated the potential benefits of volume reduction.

Late in the project, a water-quality study was added to examine the transport of road salt and associated chloride within the GSI and the combined sewer system. Continuous specific conductance was used as a surrogate for chloride concentrations to estimate that 2,790 pounds of dissolved chloride passed through the sewershed during the winter months of late 2020 through early 2021. These data were collected after GSI modification, therefore most, if not all, of the chloride was transported directly to Detroit’s Water Resource Recovery Facility via the combined sewer system. Mixing diagrams using chloride and bromide concentrations of road salt, potable water, rainwater, groundwater, and water from the combined sewer system confirmed that water within the combined sewer system is a mix of these sources. The poor condition of the combined sewer system pipes and resulting unaccounted inflows added to the challenge of accurately monitoring and identifying sources and sinks of water within the RecoveryPark sewershed.

Our research results suggest that—along with clear and quantifiable objectives—the catchment and site conditions should be well-characterized before determining the GSI design. In addition, the work presented in this report provides implications and lessons learned for effectiveness and future studies of GSI in urban settings. These efforts can be improved through increased communication between stakeholders, use of high-quality soils in GSI that have suitable hydraulic characteristics, redundant data-collection networks for critical data streams, and focusing meteorological-data collection within the GSI to obtain relevant evapotranspiration data.