The overall effect of urban infrastructure is that the natural water cycle is disrupted, with large areas of impermeable surfaces that short circuit paths of runoff, infiltration, and evapotranspiration (fig. 4). During heavy rainfall, excess stormwater runoff can cause localized flooding and changes in streamflow. Urban flooding is prevalent in many Great Lakes cities, and although not unique to this region, communities near the Lakes may be at a greater risk of flooding, than in the past, because of climate change. The communities in the region are typically older with inadequate infrastructure and are not designed to minimize the effects on the receiving waters of the Lakes. A study of Cook County, Illinois, identified 177,000 property damage insurance claims during a 5-year period from 2007 to 2011 (Center For Neighborhood Technology, 2014). This is the equivalent of one in six properties in the county making a claim with total claims amounting to $660 million during the 5 years examined. Figure 5 shows where properties are at risk from flooding in the Great Lakes region and the predicted percent increase in the number of properties at risk in 30 years. Many areas adjacent to the lakes, like the greater Chicago, Ill., and Gary, Indiana areas, are particularly vulnerable, in part because of large expanses of impervious surfaces in these densely populated areas combined with predicted intensification of storms as a result of climate change (First Street Foundation, 2020; figs. 5 and 6).

Despite their large area, the Great Lakes are sensitive to the effects of a wide range of pollutants including debris, oils, greases, and other substances that wash off from streets. Because outflows from the Lakes are relatively small (less than 1 percent per year) in comparison with the volume of water contained in them, contaminant discharges are retained in the system and become more concentrated with time (U.S. Environmental Protection Agency, 1995). Residence time varies widely from lake to lake, with the longest being Lake Superior at almost 200 years and the shortest, which varies depending on water levels, being in Lakes Erie and Ontario which is generally less than 20 years (Quinn, 1992).

Stormwater can transport a wide variety of contaminants such as metals, road salt, nutrients, bacteria, and organic compounds. In urban drainage basins, excess stormwater can also increase sedimentation, increase water temperature, reduce dissolved oxygen, degrade aquatic habitat structure, contribute to loss of fish and other aquatic populations, and decrease water quality in receiving waters (U.S. Environmental Protection Agency, 2016a). Most of the of 31 Great Lakes International Joint Commission designated Areas of Concern (AOCs) flow through densely populated areas (26 in the United States and 5 shared with Canada; fig. 7). In most of these AOCs, urban runoff contributed to the degradation of the ecosystems that led to their designation as AOC (U.S. Environmental Protection Agency, 2020).

Runoff from impervious areas contributes accumulated plastics (like bottles, bags, synthetic clothing, and diapers) that are washed off streets into stormwater drains. In a study where water samples were collected from 29 tributaries to the Great Lakes, litter-related plastics (fragments, foams, and films) were found at higher concentrations in stream samples from drainage basins having a higher percentage of urban land use (Baldwin and others, 2016). Plastic litter is not only more prevalent in urban drainage basins than in areas with other land covers, but it also is more mobile because impervious surfaces and storm sewers facilitate conveyance of plastics to receiving water bodies during runoff-event conditions (figs. 8 and 9).

Chloride concentrations in streams in the northern United States have increased substantially, with average concentrations approximately doubling from 1990 to 2011, outpacing the rate of urbanization (Corsi and others, 2015). Deicing of roads and walkways in the seasonally cold Great Lakes “rust belt” region contributes to higher chloride (salts) in water and adds challenges specific to GI design. The effect of chloride on aquatic life increases with time; more than a quarter of stream sites studied by Corsi and others (2015) exceeded the concentration for the EPA chronic water-quality criteria of 230 milligrams per liter (mg/L) by an average of more than 100 individual days per year during 2006–11. Salts tend to mobilize certain metals in water and may transform metals to more bioavailable species (Schuler and Relyea, 2018). Lakes are particularly susceptible to long-term salinization because of their longer water residence time and their large drainage basins, which combine many salt sources (Dugan and others, 2020). For most midwestern lakes, especially those in urban areas or along highways, chloride concentrations were predicted to be greater than 20 mg/L with concentrations exceeding 230 mg/L for many of the lakes studied. Figure 10 shows the relative risk of chloride contamination to lakes in the region.

The Great Lakes receive nutrients from many tributaries draining areas ranging from pristine forests to intensively farmed areas and large urban centers (Robertson and Saad, 2011). Nutrients in runoff can cause excessive plant and algae growth in nearshore areas leading to degradation of water quality. Urban areas contribute nutrients to all the Great Lakes in numerous ways, including runoff from urban land and inputs from wastewater treatment facilities. Much of the nutrient loading from wastewater treatment happens in communities with combined sewer systems which are designed to collect and transmit wastewater and stormwater to publicly owned treatment works through a single network of pipes. Nearly 57 percent of phosphorus loading to Lake Ontario is from urban sources with wastewater treatment contributing about 41 percent and urban runoff about 16 percent (Robertson and others, 2019). Wastewater treatment also contributes a large proportion of nutrients to Lakes Michigan and Erie (fig. 11). Large precipitation events can cause combined sewer overflows (CSO) when the total volume of stormwater and wastewater entering the combined system exceeds the capacity of the wastewater treatment system, and as a result raw untreated sewage is released directly to waterbodies (U.S. Environmental Protection Agency, 2016a). CSO events can be detrimental to human health and the environment because they introduce pathogens, bacteria and other pollutants to receiving waters, causing beach closures, contaminating drinking water supplies, and impairing water quality (U.S. Environmental Protection Agency, 2016a). Fish and other aquatic populations also can be affected by the depleted oxygen concentrations that CSOs can cause (U.S. Environmental Protection Agency, 2016a).

There are 184 combined sewer systems in the Great Lakes Basin; in 2014 there were 1,482 events discharging an estimated 22 billion gallons of untreated wastewater into the Lakes (figs. 12 and 13; U.S. Environmental Protection Agency, 2016a). The greatest number of communities that discharge directly into the lake or into tributaries are in the Lake Erie (93) and Lake Michigan (72) Basins (fig. 12).

Stormwater management is an increasing challenge for communities throughout the Great Lakes Basin. In areas where CSO systems exist, GI can slow down the rapid transfer of runoff from impervious surfaces that can quickly overwhelm the combined sewer network and lead to CSOs. The slowing down of runoff transfer is done primarily through retention and storage of water in soils; the water gradually infiltrates into shallow aquifers or transfer to the combined system through controlled release via subsurface drainage or underdrains. Therefore, GI can reduce the threat of excessive flows to combined sewers, and it can also improve water quality by trapping sediment and nutrients (Great Lakes Commission, 2018).

The use of GI across the region often remains limited to local efforts and with little coordination among various levels of government despite promising GI solutions to stormwater management challenges (Great Lakes Commission, 2018). This is at least in part because of a lack of clear standards for stormwater, low funding, and a general uncertainty in the type and expected performance of GI infrastructure best suited for a given situation. Coordination among all levels of government along with clear management guidelines can reduce the barriers for municipalities to adopt GI and better manage stormwater regionally resulting in improved water quality in the Great Lakes.

Because stormwater challenges including funding and innovation are often local in nature, they are heavily dependent on Federal, State, and provincial policy (Great Lakes Commission, 2018). City planners, engineers, and political leaders often see investment in GI as riskier than other alternatives despite studies by both EPA and the American Society of Landscape Architects that found that, in most cases, GI either reduced or did not affect costs (Sinha and others, 2017). Policies that encourage or require municipalities to transition away from traditional “pipes and ponds” are lacking. The result is that for many communities the uncertainty of using new techniques often overshadows the benefits of GI (Chaffin and others, 2016).

Maintenance and long-term performance are often perceived to be the biggest barrier to GI implementation among stormwater management professionals (Basu and others, 2020). Installation and maintenance costs are also a major concern to municipalities. The next major barrier is political, including a lack of acceptance among local leaders, municipal staff, and practitioners, followed by conflicting codes and ordinances. Resistance from the regulatory community is considered the fourth largest barrier, and although not as large of a concern as the first three barriers to many municipalities, it remains a concern in some areas.

Rain gardens, bioretention ponds or cells, and planter boxes (collectively referred to as vegetated basins) are shallow depressions built to receive runoff generated from nearby impervious surfaces by providing a temporary holding basin while runoff infiltrates into the ground (U.S. Environmental Protection Agency, 2021c; fig. 15). Vegetated basins can range in size from small front yard gardens to large, vegetated ponds. Planter boxes are essentially rain gardens with vertical walls with either open or closed bottoms and are ideal for space-limited sites in dense urban areas and function also as a streetscaping element.

Vegetated basins mimic natural hydrology by collecting and absorbing runoff from rooftops, sidewalks, and streets and allowing excess water to either infiltrate, evaporate, or evapotranspire (U.S. Environmental Protection Agency, 2021c). Vegetated basins help manage storm water more sustainably by creating a hydrologically functional landscape that reduces flooding and improves water quality. Key components of vegetated basins include a substrate that allows infiltration and healthy growth of deep-rooted plants. Rain gardens are generally smaller than bioretention ponds, with both often requiring soil amendments to allow adequate percolation to the subsurface (U.S. Environmental Protection Agency, 2021c). Rain gardens typically manage water that falls on residential urban lots whereas bioretention ponds treat larger expanses of impervious surfaces like parking lots or streets. Bioretention ponds may have a gravel base that extends below the frost line with an embedded tile drain so that water can be released slowly along the tile as it moves down gradient to natural soils approximating natural hydrologic processes (U.S. Environmental Protection Agency, 2021c). In cases where the tile outlet drains to a storm sewer, the bioretention pond may behave hydrologically like a bioswale.

Vegetated basins are a very popular stormwater management practice because of relatively low cost and ease of installation. Many communities support the installation of rain gardens by hosting workshops and seminars or by having grant programs, making them more attainable to the general public. There are numerous rain garden and bioretention installations in the Great Lakes region, and several studies that have done extensive monitoring of soil moisture, and frequency of inflow, outflow, and overflow volume (fig. 16). Sidebar 1 highlights a USGS investigation that evaluates the hydrologic response to the installation of a rain garden in a parking lot that was previously continuous impervious cover.

Construction methods and native site characteristics play important roles in vegetated basin performance. Construction methods vary from a simple depression to a highly engineered basin with filtration substrates. Plants in vegetated basins may range from a monoculture of low vegetation like grass to a diverse variety of trees, shrubs, grasses, and ground covers. Generally, the more deep-rooted plants in the installation, the greater the benefit (U.S. Department of Agriculture, undated). Soil conditions vary widely across the region and often vary substantially within a single basin; therefore, soil analysis alone may not be a reliable way to predict the performance of the installation.

Vegetated channels including bioswales and grassed swales are designed to slow water flow so that some of the water has time to either infiltrate, evaporate, or transpirate as it is routed from one location to another (fig. 17). They are generally linear channels or swales that are vegetated or mulched. In contrast to vegetated basins that are designed to retain all runoff at a site, vegetated channels are often used to route runoff water from sidewalks, streets, or parking lots to storm sewers or natural channels.

Vegetated channels can be effective at infiltrating stormwater from a large impervious surface; channel performance is dependent on the permeability of underlying soils. Grassed swales are simple vegetated channels constructed over minimally engineered substrate, whereas bioswales are engineered with specific soil media mixtures and underdrains. Both types of swales are often used in conjunction with other structures such as detention ponds (Illinois Sustainable Technology Center, 2012).

Vegetated channels are sloped to move water through the system in an efficient manner, can range from a few feet to hundreds of feet long, can be wet or dry, and the surface can be covered with grass or other types of vegetation. Bioswales are sometimes covered with rocks instead of vegetation and are particularly useful along streets or in parking lots where runoff can be redirected from the curbside (U.S. Environmental Protection Agency, 2021c). Plants and soils help treat stormwater runoff to reduce flow and contaminants to storm sewers and may improve water quality in downstream creeks and streams. They can absorb low flows or carry runoff from heavy rains to storm sewer inlets or directly to surface waters. Vegetated channels can improve water quality by infiltrating the first flush of storm water runoff and filtering the large storm flows they convey (U.S. Environmental Protection Agency, 2010). In the Great Lakes region, vegetated channels are a common sight along roadways and parking lots (fig. 18).

Permeable pavements allow water to infiltrate between spaces in the surface material, through layers of crushed rock, and then into soil and groundwater below (Winston and others, 2018). The purpose of the pavement is to infiltrate most of the stormwater onsite to reduce the amount of water and contaminants flowing into storm sewers or directly to rivers and streams (U.S. Environmental Protection Agency, 2021c; fig. 19). Permeable pavements may maintain more stable base flows in streams, reduce flood peaks, reduce stream bank erosion, and may help protect water quality (U.S. Department of Agriculture, 2009).

Permeable pavements can be installed in a variety of applications like parking lots, sidewalks or walking paths, or as an integral component in green streets and alleys. Because there is a high maintenance component associated with permeable pavement, one of the best applications may be at the periphery of less costly traditional paving so that the permeable part allows for infiltration of small, frequent rains and the first flush of larger storms which wash off most pollutants. Stormwater may leave the permeable pavement system through one or more of the following four hydrologic pathways: evaporation, exfiltration (infiltration into the underlying soil), underdrain discharge (if an underdrain is present), or surface runoff (fig. 19).

Permeable pavement consists of either porous paving, like concrete or asphalt, or interlocking modular paver blocks, typically of concrete or brick, that achieve porosity by virtue of the spacing between blocks or pores in pavement. Permeable pavers have been installed in several locations throughout the Great Lakes (fig. 20) and include experimental and permanent installations in a variety of environmental settings.

Trees reduce and slow stormwater by intercepting precipitation in their leaves and branches (fig. 21). Many cities have programs to plant trees in areas where they were lost to development and thus restore some of the benefits of trees (fig. 22). According to the authors of a report on urban planning, “trees interact with the urban hydrologic cycle by intercepting incoming precipitation, removing water from the soil via transpiration, enhancing infiltration, and bolstering the performance of other GI technologies” (Berland and others, 2017, p. 167); however, the interactions of trees with the urban hydrologic cycle is not well understood, particularly at the spatial and temporal scales that are relevant to stormwater management. Therefore, the contribution trees can make on overall stormwater control depends on further research of how and to what extent trees interact with stormwater, as well as information on the location and best vegetation for a given geographical location.

Nongovernmental entities including homeowners, businesses, and community groups can also participate in planting and maintaining trees in urban areas. In some cities, engineered solutions have been developed that divert stormwater through systems designed to allow infiltration and uptake by trees and thus reduce runoff.

Urban tree canopies may be most effective when paired with engineering structures that capture and route stormwater runoff through the root systems of, for example, sidewalk trees; however, more study is needed to adequately understand these interactions especially at the spatial and temporal scales relevant to stormwater management (Berland and others, 2017).

According to the EPA, green roofs fit into a subcategory of on-site stormwater control measures, termed low-impact development (LID) technology (Berghage and others, 2009). Building roofs make up a sizable proportion of the impervious surfaces associated with urban development along with roads and parking lots (fig. 23). Green roofs thus have the potential to reduce the stormwater effects of development because they can detain and retain stormwater.

Like other GI, green roofs can be incorporated with other stormwater controls and included in a municipal stormwater management plan. Although green roofs retain and remove stormwater through evapotranspiration, excess runoff may add to stormwater control problems that many urban and suburban communities face. Implementation of other stormwater control measures in urban areas may not be practical in some circumstances because of small available surface area and other concerns. Location of example green roof studies is shown in figure 24.

One measure of determining the effectiveness of vegetated basins is the capacity of the installation to infiltrate a 100-year storm event from a contributing area six times larger than the rain garden or retention basin (Illinois Sustainable Technology Center, 2012). Often methods of evaluating the performance of vegetated basins include comprehensive monitoring of all components of the water cycle; however, it is prohibitive in cost and the effort to include an adequate number and range of intensity of storm events. For this reason, there are few studies reporting performance of vegetated basins (especially for rain gardens). Alternative methods of evaluating performance based on visual inspection, infiltration testing, and synthetic drawdown testing have been developed and tested by Gulliver and Anderson (2008) and Asleson and others (2009) and may be useful for rapidly assessing multiple installations. A whole water-cycle monitoring approach may be needed to fully assess the role of GI interventions on performance of inflows captured, duration of outflow drainage, hydrologic losses (because of evaporation and transpiration), and groundwater table dynamics that indicate deeper infiltration losses; however, low-cost monitoring strategies may be suitable for addressing selected monitoring objectives and design questions (Shuster and Darner, 2018).

Most studies indicate that rain gardens are an efficient, cost-effective way to reduce local runoff. A visual assessment of 70 vegetated basins in the Chicago region determined that all basins functioned to some degree and even the most poorly maintained looked as if they could be restored with routine weeding (Illinois Sustainable Technology Center, 2012). Shuster and others (2017) found that despite sediment loads clogging the rooting zone and overall lower-than-design infiltration rates, rain gardens maintained relatively high overall retention effectiveness. They found that 50 percent of total inflow was detained in the rain gardens, and overflow to the sewer system was delayed by an average of 5.5 hours during the 4-year study duration (Dumouchelle and Darner, 2014; St. Francis Apartments, fig. 16). Twelve rain gardens installed in the Slavic Village neighborhood, Cleveland, Ohio (fig. 16), provided considerable additional detention capacity to the sewershed and prevented or delayed some proportion of event stormwater from directly entering into the local combine sewer system (Shuster and Darner, 2018). Eight of 12 rain gardens in Minnesota that were evaluated by visual inspection performed well after several years. The other four rain gardens were determined to be nonfunctional based on (1) presence of ponded water, (2) presence of hydric soils, (3) presence of emergent wetland vegetation, and (4) failing vegetation, suggesting that lack of maintenance was the primary reason for failure, respectively (Asleson and others, 2009).

Soil drainage and provisions for subsurface drainage like an internal water storage zone (areas designed to store water at depth before slowly releasing to surrounding native soils or outlets), especially in areas with poorly drained soils or shallow water tables, are key design features for properly functioning vegetated basins (Shuster and others, 2007; Winston and others, 2016). Soil infiltration capacity was measured in 15 of 70 Chicago rain gardens with all but 1 estimated to be able to successfully infiltrate the water from a 100-year storm event (Illinois Sustainable Technology Center, 2012). Rain gardens constructed with engineered soils had higher infiltration capacity than native soils; however, almost all the native soil rain gardens performed better than would be predicted using U.S. Department of Agriculture Natural Resource Conservation Service soil survey data (Illinois Sustainable Technology Center, 2012). Generally, systems in sandy soils have higher infiltration rates than high clay soils. Selbig and Balster (2010) observed median infiltration rates that were an order of magnitude greater for rain gardens with sandy soils (Owen Conservation Park, fig. 16) than for those with clay (Madison Water Supply pump house, fig. 16) regardless of vegetation type; however, spatial variability was noted within rain gardens having similar soil conditions. Several studies have reported a high variability in infiltration rates within individual rain gardens, even those constructed of engineered soils where soil conditions are expected to be similar throughout the garden. Lower infiltration rates were observed near inflow and in the deepest locations of the rain gardens, suggesting that surface infiltration may decrease with time in these areas because of compaction of soils caused by concentrated flows or surface soil clogging from particle settling (Illinois Sustainable Technology Center, 2012).

Plant diversity and density are critical for creating successful installations, and failing to achieve adequate vegetative communities will ultimately cause the practice to fail because vegetation diversity and density affects infiltration capacity in vegetated basins (Rainer and West, 2015). Prairie vegetation can have greater infiltration rates than turf grass in rain gardens (Selbig and Balster, 2010). Vegetated basins with high infiltration rates may serve to restore hydraulic connectivity in areas where surface and groundwater are otherwise decoupled, providing a pathway for recharging of groundwater (Stewart and others, 2017).

Implementation of an internal water storage zone allows for continued exfiltration (infiltration into the underlying soil) during interevent periods. Exfiltration rate and internal water storage zone thickness were the primary determinants in successfully reducing runoff volume, with lateral exfiltration the primary pathway (Winston and others, 2016). Runoff reduction for three bioretention ponds with internal water storage zones in Ohio (Ursuline College, Holden South, and Holden North ponds at Holden Arboretum [fig. 16]) was 60, 42, and 36 percent, respectively, and was primarily related to the exfiltration rate of the underlying soil (Winston and others, 2015). The maximum rainfall depth that could be retained in the basins without outflow is a result of storage capacity in the internal water storage zone (a function of zone depth and exfiltration rate) with the greatest volume reduction taking place in basins with the highest infiltration rate and deepest internal water storage zone.

Like vegetated basins, vegetated channel success is dependent upon infiltration capacity, the presence of an internal storage zone, and the density and diversity of vegetation in the installation. Even where soils have poor hydraulic conductivity, a 12-foot (ft) trench can reduce stormwater runoff from a typical road to about 25 percent of total precipitation (Capitol Region District, 2021). Most of the studies reviewed reported reduction in stormwater volume and peak flow. Water-quality effects can include reduction of contaminants from road runoff being retained in the bioswale and a resultant reduced contaminant load to surface waters (Kumar and others, 2017). Studies indicate that vegetated channels designed to slow but not necessarily capture 100 percent of runoff are successfully used to effectively capture and reroute enough stormwater runoff to prevent overflow onto nearby roads during small storm events (Kumar and others, 2017).

At two sites in the Chagrin River watershed, Northeastern Ohio (fig. 18), frequent residential street flooding led to the installation of a bioswale and a rain garden (Darner and Dumouchelle, 2011; Darner and others, 2015). The combination of the rain garden and bioswale performed better than expected when several precipitation events that exceeded 0.75 inches (in.) did not result in overflows (Darner and Dumouchelle, 2011), and total runoff was reduced by 45 percent (Darner and others, 2015). During a 5-year monitoring period that was adjusted for rainfall variation, the runoff reduction decreased with time; however, the lag time from precipitation to peak flow remained constant (Darner and others, 2015). The combination of increased runoff volume with no change in timing may possibly be because of preferential flow paths developing in the structure, increasing capacity for internal drainage. If clogging of surface or subsurface sediments in the bioswale is causing increased overflow, the system may be inadequate to reduce street flooding in future years (Darner and others, 2015). Sediment deposition in bioswales is likely not to be uniform, resulting in decreased performance in parts of the channels where sediments accumulate. Infiltration rates at bioswales at Cermak Road and Blue Island Avenue corridor in Chicago, Ill. (fig. 18), decreased with time because of soil compaction and deposition of fine-grained sediments from stormwater runoff (Kumar and others, 2017). Greater amounts of sediment were deposited close to the curb cut compared to the center of the bioswale.

Vegetated channels in conjunction with other practices can be an effective contributor to reducing runoff. Few studies were found that focus solely on quantifying the benefits of vegetated channels. A combination of bioswales, permeable pavers, and infiltration zones (tree boxes and planter boxes) installed in Chicago were designed to infiltrate 80 percent of rainfall from the 2-year 30-minute design storm, which is equivalent to approximately 1.12 in. of rainfall; however, during the study period, the area received rainfall at much higher intensities and still the practices were able to infiltrate almost 100 percent of runoff (Kumar and others, 2017). Biowales were included as a part of low-impact development in a new residential area in Wisconsin (Selbig and others, 2004) that was studied to determine the effect of residential construction on a nearby Class I trout stream. Collectively, the stormwater management and erosion control practices reduced stormwater runoff and solids transport during construction and provided sufficient protection against degradation of the stream. A combination of vegetated swales and detention basins were able to retain predevelopment runoff characteristics and avoid negative consequences of the residential neighborhood development, indicating that implementing GI practices before residential construction can help prevent stream degradation because of land disturbance. Grassed swales were installed as part of a low-impact development catchment which included forested hillslopes, lawns, and a detention pond in Cross Plains, Wisconsin (fig. 18). The swales captured 96 percent of the potential runoff volume (based on rainfall in the basin) before reaching the infiltration basin and provided the greatest benefit for high-intensity storm events that exceeded infiltration rates of underlying soils (Selbig and Bannerman, 2008).

Every study reviewed for this GI practice reported that, at least initially, all permeable paver sites captured and reduced stormwater runoff (Kumar and others, 2017; Lebens and Troyer, 2012; Selbig and Buer, 2018; Traver, 2002; Winston and others, 2016, 2018, 2020). As many as 80 percent of storm events were completely captured by permeable pavements. The reduction in stormwater volume resulting from the installation of permeable pavements varied widely (13 to nearly 99 percent), indicating that factors beyond the pavers themselves contribute to a large percentage of the success (Winston and others, 2015). Factors contributing to the effectiveness (or lack of effectiveness) in permeable pavement installations included characteristics of the site and installation design, level of long-term maintenance, and characteristics of individual storm events.

Site characteristics and installation design were important for achieving stormwater reductions. Greater volume reductions were achieved by having a larger permeable zone relative to the watershed size, incorporating an internal water storage zone beneath the crushed stone base layer, and installing permeable pavement systems in areas with higher permeability underlying soils (Winston and others, 2015). Stormwater was completely captured during 4–80 percent of observed storm events (Winston and others, 2018), with volume reductions from 16 to 32 percent at four separate installations (Old Woman Creek National Estuarine Research Reserve, Orange Village Park, Perkins Township Police Administration, and Willoughby Hills, parking lots northern Ohio, [fig. 20]). Those studies found that the loading ratio for exfiltration (a measure of the capacity of the permeable pavement system in relation to the runoff area) and drawdown rate (the amount of water that can flow through the system into the soil) affect the performance of the system (Winston and others, 2018). High loading ratios and low drawdown rates resulted in the worst overall performance in completely capturing events. The study determined that it is important to locate permeable pavements over the most permeable soils at the development site and increase the permeable pavement surface area to reduce the loading ratio and improve runoff mitigation. Even in areas with poorly drained soils, permeable pavements with well-engineered substrates can be implemented successfully through the addition of an underlying impermeable geomembrane (Eisenberg and others, 2015). Harsh winter climate with multiple freeze/thaw cycles and the addition of deicing compounds can also affect the performance of permeable pavement; however, recent studies indicate that using specially designed concrete mixtures and the correct deicing compounds can enhance performance of these systems in cold regions such as the Great Lakes, as illustrated by a study in Minnesota (Schaefer and others, 2012)

For four sites in northern Ohio, the minimum rainfall depth to produce outflow ranged from 0.11 to 0.99 in., indicating there is a wide variability in the hydrologic performance of these systems (Winston and others, 2015). Construction and materials also affect performance; permeable asphalt was found to absorb higher runoff volume than permeable interlocking pavers and permeable concrete (Selbig and Buer, 2018), but asphalt lacked the same degree of water-quality treatment. Rainfall intensity and volume were important in the overall effectiveness of permeable pavement installations (Winston and others, 2018). Stormwater was captured completely for as much as 80 percent of observed storms, and substantial peak flow mitigation was observed for all rainfall events not producing surface runoff.

A reduction in stormwater runoff is frequently cited as a benefit of increasing urban tree canopy (U.S. Environmental Protection Agency, 2014b), yet few studies exist that quantify the effect that urban tree canopy has on stormwater runoff quantity and quality (fig. 22). The USGS is collaborating with the U.S. Forest Service, EPA, and the University of Wisconsin-Madison to measure the hydrologic effects of urban tree canopy to fill some of these data gaps (see sidebar 5).

A study of green roofs in Pennsylvania used paired field and laboratory tests to determine design specifications for optimal performance in limiting runoff and assess water quality effects (Berghage and others, 2009). Water quantity and quality properties measured included flow, turbidity, and nutrients. The results indicate that a 3.5–4 in. deep green roof can retain 50 percent or more of annual precipitation. Green roof functioning can be further improved by understanding the interactions among soils and vegetation and other roof ecosystem elements (Oberndorfer and others, 2007).

Three major factors affect the water quality effect of vegetated basins: (1) the type and makeup of soils (native or engineered), (2) the presence of an internal water storage zone, and (3) the type and number of plants (Winston and others, 2015). Particulate matter is captured near the surface of soils through filtration, resulting in capture of particulate phosphorus, total suspended solids, and heavy metals (Davis, 2007). Dissolved components of water quality were reduced because of interactions with clay particles and organic matter during infiltration. The initial content of phosphorus in the soil can be critical in determining if the basin acts as a net phosphorus sink or source (Hunt and others, 2006). Saturated conditions cause phosphorus to be ineffectively stored in the submerged zone. Plants provide a mechanism for volume reduction through transpiration and nutrient and metal uptake from soil water (Lucas and Greenway, 2008). Nitrogen fate within basin soils is regulated mainly by aerobic or anaerobic conditions of the soils and plant uptake (Davis and others, 2009). Planted basins have higher soil-saturated hydraulic conductivity and higher nitrogen and phosphorus assimilation than unplanted basins (Bratieres and others, 2008; Lucas and Greenway, 2008). Vegetated basins in sandy soils allow more infiltration than heavier soils like clays, but this may also allow for greater pollutant mitigation because of greater throughflow and less accumulation of contaminants within the basin; however, sediment and heavy metal concentration reductions were observed in nearly every field- and laboratory-based study on bioretention regardless of soil or vegetation in the basin (Davis, 2007; Davis and others, 2009; Feng and others, 2012; Hunt and others, 2008). A study was done in the Shepherd Creek catchment in Cincinnati, Ohio (fig. 16), where 81 rain gardens were installed on 30 percent of the properties in 4 experimental subcachments. Results indicated that the rain gardens were successful at mitigating stormwater at the property level but that only minor effects on streamflow volume and water quality, with no changes in biotic health, were observed in Shepherd Creek (Roy and others, 2014; Thurston and others, 2008). Their study concluded that improvement of overall stream health is unlikely without additional treatment of major impervious surfaces (like roads and parking lots) and that further research is needed to define the minimum effect threshold for retrofitting catchments if the goal is to improve stream ecosystem health.

Water quality results indicated that large amounts of pollutants, like metals, nutrients, sediment, and chloride, entered the bioswale from road runoff and were retained in the structure, preventing contamination from entering sewers directly (Kumar and others, 2017). Although grassed swales are effective at reducing runoff velocity and peak discharge, they do not provide adequate pollutant removal benefits to act as a stand-alone best management practice (Minnesota Pollution Control Agency, 2021).

Reduction of stormwater volume is the primary design function of permeable pavement; however, there may be other benefits to well-designed permeable pavement systems including improved quality of runoff. All the studies referenced in the previous section on permeable pavements monitored inflow and outflow from permeable pavement to determine volume and peak reduction but only a subset monitored water-quality properties (Selbig and Buer, 2018; Winston and others, 2016 ,2020). Of these, results varied and indicated a decreased effectiveness with time. In one study, a combination of primary treatment and final capture in a cistern contributed to a 96 percent decrease in total suspended solids and turbidity concentrations and a 99.5 percent decrease in total suspended solids and loads that resulted in a greater than 40 percent reduction of sediment-bound nutrient forms and total nitrogen (Winston and others, 2020). Winston and others (2020) also analyzed for several metals and found that copper, lead, and zinc were initially sequestered in the subsurface during the first 5 months of monitoring; however, during the second half of the monitoring period, metals were observed to leach back into the runoff. The cause of leaching may have been because of decreased pH within the cistern or a result of desorption of metals from the sediment caused by residual chloride from salt used for deicing.

One of the few studies used for this review that included water quality was a study of green roofs in Pennsylvania (Berghage and others, 2009). Water-quality properties measured in addition to flow included turbidity and nutrients. The results indicate that a 3.5–4 in. deep green roof limited nitrate loading, and the authors suggest that performance can be improved further by routing the roof discharge to other GI practices such as vegetated basins or channels. An added benefit to both urban tree canopy and green roofs is they can help mitigate the urban heat island effect (U.S. Environmental Protection Agency, 2021c).

An evaluation of catchment characteristics is the first step in ensuring good performance of GI installation. The success (or failure) of GI performance can be determined by whether the installation meets certain design criteria. Even though there are a multitude of design criteria, there is no national standard for sizing GI to address stormwater challenges. Communities must evaluate the objectives of local programs and establish their own constraints for determining required water volume and depth capture based on local rainfall analysis, catchment characteristics, and targeted removal of stormwater pollutants.

Site design techniques that reduce impervious cover, conserve natural areas, and use pervious areas more effectively increase the likelihood of success for restoring natural hydrologic conditions (Minnesota Pollution Control Agency, 2021). Individual practices that are relatively small but densely distributed, like rain gardens and bioswales, are relatively inexpensive, can be considered an aesthetic part of landscaping, and may provide cost-effective alternatives to more traditional larger centralized practices like detention ponds. An additional benefit of increasing catchment capacity through multiple distributed systems is to enhance chemical transformations and storage of pollutants via longer soil residence times to attenuate entrained constituent and particulate matter transport to a stream, river, or other water body (Golden and Hoghooghi, 2018). The maximum contributing area should not exceed five acres (Christopher Burke Engineering, LLC, undated). Other key site factors affecting performance include slope (less than 20 percent), depth to bedrock, water table depth, underlying soils, and ratio of contributing area to GI surface area (Christopher Burke Engineering, LLC, undated). The optimal ratio of connected impervious contributing area to GI surface area differs for each practice ranging from 3:1 for permeable pavement to 5:1 for vegetated basins and swales. If the ratio is too large, the installation is overwhelmed and cannot capture the required stormwater volumes (Christopher Burke Engineering, LLC, undated). Although undersized practices still offer some treatment benefits, such design limitations may affect longevity, maintenance requirements, and the resilience to accommodate extreme rainfall events (Taguchi and others, 2020).

The infiltration capacity of a GI is driven by the substrate on which it is installed. The substrate may be of unaltered or amended soil and may include underdrains to improve infiltration in areas with low permeability. Unaltered soils vary widely throughout the Great Lakes region, with highly permeable well-drained sandy soils in much of Wisconsin and Michigan to very poorly drained high clay soils in Ohio and Indiana (fig. 2B). Often soils in urban areas are low in organic matter and highly compacted, reducing infiltrability. The urban soil condition with the greatest infiltration potential is noncompacted sand (Pitt and others, 1999; fig. 25). Compaction has the greatest effect on infiltration rates in sandy soils, with little detrimental effects associated with soil moisture. Compaction and moisture have about the same effect on infiltration on clayey soils, with saturated-compacted clayey soils having the least effective infiltration (Pitt and others, 1999; fig. 25). Because it is difficult to predict the level of infiltration for urban soils, doing infiltration tests during saturated conditions is crucial for a site-specific design of GI installations. Soils that have previously been disturbed by development should be considered compacted. Compaction typically affects the first 18 in. below the surface; therefore, infiltration testing should extend below the compaction depth (Pennsylvania Department of Environmental Protection, 2006). For clay or compacted soil, it is typical to excavate to the depth of the planting soil and replace with an engineered soil to support healthy plant growth (U.S. Environmental Protection Agency, 2014a).

A study done in Madison, Wisc., from 2004 to 2008 compared the capability of rain gardens with different soil and vegetation types to infiltrate stormwater runoff (Selbig and Balster, 2010). Two side-by-side rain gardens were installed on sandy soil, and two additional side-by-side rain gardens were installed on clay soil (fig. 26). For each soil type, one of the rain gardens was planted with turf grass and the other with a native prairie species. Results indicated that regardless of soil type or vegetation, the rain gardens stored and infiltrated at least 99 percent of the runoff during the 4-year study period. Median infiltration rates in sand were greater than clay. Although infiltration rates were reduced during winter months, the hydraulic function of the rain gardens was not appreciably altered (Selbig and Balster, 2010). In Toledo, Ohio, a study of bioswales built on clay soils indicated the system was able to retain about 64 percent of annual runoff volume and peak flows were reduced by 60–70 percent despite the relative imperviousness of the underlying soils (U.S. Environmental Protection Agency, 2014a). Initially an underdrain was installed because of the existence of clay soils, but it was later closed to promote infiltration into the native soil.

Soil amendments are recommended by the U.S. Department of Agriculture if infiltration rates are below 0.5 in. per hour and organic matter is below 2 percent (U.S. Department of Agriculture, undated). Although amendments are intended primarily for supporting vigorous plant growth, all growing media should have a field-tested infiltration rate between 1 and 8 in. per hour according to the Minnesota Stormwater Manual (Minnesota Pollution Control Agency, 2021). If phosphorus is a water quality concern, recommended amendments include substituting organic matter less prone to phosphorus leaching, like coir or biochar. Soil amendments that are effective in increasing chemical sorption of dissolved phosphorus include iron filings, steel wool, native iron-rich soils, oxide-based sorbtive media, and drinking water treatment byproducts containing aluminum and (or) iron hydroxides. Other water-quality concerns have different media requirements.

During site evaluation, the depths to the seasonal high groundwater table and to bedrock should also be measured to ensure adequate infiltration. At least 2–3 ft of separation from the GI bed bottom to bedrock or seasonal high groundwater table is recommended (Minnesota Pollution Control Agency, 2021; Pennsylvania Department of Environmental Protection, 2006). Underdrains convey subsurface water in a GI practice to facilitate infiltration when rates are too slow and are highly recommended where parent soils have low or very low infiltration rates when wetted, have permanent high-water table, or contain a claypan or clay layer at or near the surface (Minnesota Pollution Control Agency, 2021; U.S. Environmental Protection Agency, 2014a). A maximum 72-hour dewatering time is recommended to prevent extended surface ponding (Minnesota Pollution Control Agency, 2021).

Vegetation selection and maintenance is important for GI success. Under-vegetated structures encourage erosion and heavy inflows of salt and other urban pollutants, which kill plants. Often plantings are treated as horticultural objects rather than a diverse system of roots that play a major role in filtration, water uptake, transpiration, and erosion control (Rainer and West, 2015; fig. 27). Bioretention basin plantings require many different plants woven together in tight mosaics to effectively address the wide array of hydrologic conditions in a small area; for example, there are dry rims in the upper part of the installation, moist centers, and perpetually wet areas near outflows and drains. Plant diversity also provides a range of different benefits: cool-season growers hold the soil in winter, warm-season growers tolerate drought, short-lived plants reseed and fill gaps, long-term plants create stability, deep rooted plants act as a sponge, and shallow rooted plants resist erosion (Rainer and West, 2015).

All GI practices require some periodic maintenance to sustain performance through time. Neglecting to include the long-term cost and support of maintenance will eventually lead to reduced performance or failure. The level of maintenance ranges from low for revegetation (for example tree canopy) to high for permeable pavement (Northeast Ohio Regional Sewer District, 2012). Maintenance considerations for bioretention ponds, rain gardens, bioswales, and green roofs are primarily associated with maintaining vegetation (table 2). Maintenance can be greatly reduced in vegetated systems with diverse, densely packed plantings. Vegetated systems that have few species and widely spaced plantings require more weeding and attention to erosion damage. Trash and debris removal are also a concern when contributing areas are frequently littered.

Erosion and sedimentation are also maintenance concerns for bioretention ponds, rain gardens, bioswales, and permeable pavements especially when these practices receive runoff from roads and parking lots (table 2). Many practices fail because of sod or rock inlets that have filled with sediment and cause stormwater to bypass the structure. In areas with high sediment loading, pretreatment devices, such as sediment traps and forebays (fig. 28), have the capacity to remove at least 50 percent of the annual sediment load (U.S. Environmental Protection Agency, 2016b). If run-on water to the practice is contaminated with nutrients, sediment, metals, or other contaminants, they will accumulate in engineered soils and may require replacement periodically. Li and Davis (2008) found that sediment and heavy metals concentrate in the top 5–10 cm of bioretention media, so removal and replacement of surface layers may revitalize water quality performance. Depending on the contaminants present, these soils may be considered hazardous and require suitable disposal. If sediment buildup reaches 25 percent of the ponding depth, it should be removed (Christopher Burke Engineering, LLC, undated). Bioretention and bioswales that have underdrains or overflow structures may require periodic maintenance to prevent clogging and potential overflow of the installation (table 2).

Maintenance of permeable pavements is critical to providing long-term reductions in stormwater runoff (Kumar and others, 2017). All the reviewed permeable pavement installations became clogged with sediment through time, reducing performance (Kumar and others, 2017, Lebens and Troyer, 2012; Selbig and Buer, 2018; Selbig and others, 2019; Traver, 2002; Winston and others, 2015, 2016, 2018, 2020). Run-on from nearby pervious and impervious surfaces carry sediment and debris that clog surface pores of the pavement (fig. 29). Locations beneath trees, draining pervious areas, and with higher traffic flows are more susceptible to clogging. Maintenance includes periodic vacuuming, sweeping, or power washing to alleviate clogs in the pavement. When the pavement becomes clogged, often the only remedy is to remove or replenish the top layer of stone or sand between joints (Danz and others, 2020).

A study of two sites in Chicago reported that a decline in paver infiltration rates was on the road with higher traffic intensity and specifically the side of the road where daily commuters parked their cars (Kumar and others, 2017). Infiltration rates of pavers recovered after cleaning; manual cleaning with complete replacement of aggregate fill material was more effective than mechanical cleaning.

Surface cleaning to maintain the permeable properties of porous pavement is recommended at least twice per year using industry methods such as regenerative air or vacuum sweeping (Eisenberg and others, 2015; Wisconsin Department of Natural Resources, 2016). Selbig and Buer (2018) tested several methods of surface cleaning and all produced appreciable gains in infiltration (fig. 30). A review of numerous studies including some done in cold climates found that the common key to success was proper and regular maintenance (Weiss and others, 2015).

Quantitative recommendations made by the company Sustainable Technologies after an evaluation of several permeable pavement installations include twice-yearly vacuuming and joint material replacement when the measured surface infiltration rate is less than 150 millimeters per hour (Sustainable Technologies Evaluation Program, 2018). If surface drainage performance remains unacceptable, all the bedding and joint fill in the affected area should be replaced. Other recommendations include twice-yearly cleaning of inlet structures within or draining to infiltration beds to prolong their lifespans. Finally, maintaining planted areas adjacent to the permeable pavement prevents soil washout onto the pavement, and if washout happens, it should be cleaned immediately to prevent further clogging. Although permeable pavement systems usually require less deicing treatment than standard pavement, sand or other abrasive deicing treatment can cause clogging if applied near or on the permeable pavement.

Because each type of practice has associated unintended consequences, the greatest negative consequence of many GI practices may be the infiltration of stormwater contaminants to groundwater (Andres and others, 2018). Frequently, the only design criterion for stormwater infiltration is the rate of infiltration; however, infiltrated stormwater can carry pollutants (nitrogen; pesticides and fertilizers; metals, oil, and grease from road surfaces and gas stations; hazardous waste spills; and salts used in road deicing) and cause hydraulic problems like mounding, slope instability, and subsurface flooding of infrastructure (Andres and others, 2018). Contaminants that are water soluble (for example, deicing salts, nitrogen, and some pesticides) are more likely to infiltrate into shallow groundwater, whereas contaminants that bind to soils (metals, some pesticides, and orthophosphate) tend to accumulate in soils underlying the GI installation. Accumulated contaminants in soils may eventually release to groundwater with time; for example, deicing agents containing chlorides, applied during winter months, can mobilize metals, and carry them to groundwater. Deicing compounds containing chlorides are highly soluble and may accumulate in GI sediments to be released later during subsequent runoff events. In addition, salt-laden runoff from deicing compounds can mobilize previously sorbed toxic metals from filtration media (Paus and others, 2014). Likewise, hydrocarbons are trapped in the first few centimeters of soil and are potentially taken up by microorganisms; it is not uncommon for groundwater from infiltrated stormwater to contain organic compounds such as polycyclic aromatic hydrocarbons (Taguchi and others, 2020). As pointed out in Andres and others (2018, p. 1), “…using shallow groundwater for the disposal of todays' stormwater problems may only be delaying a problem rather than solving it.”

The pathways for contaminated stormwater to be transported from GI to groundwater vary depending on the practice and the construction. Vegetated basins and vegetated channels generally do not have lined bottoms so that stormwater can freely infiltrate deep into aquifers (Andres and others, 2018). Sediments below permeable pavements may accumulate nutrients, metals, deicing compounds, and hydrocarbons and with time may percolate to deeper substrates and eventually the groundwater table (Taguchi and others, 2020). Practices, such as vegetated basins and channels, disconnect impervious rooftops and sidewalks from local waterways, and storm drains generally do not have lined bottoms so that stormwater freely infiltrates into the ground, posing potential risks to groundwater (Andres and others, 2018). Potential contamination to groundwater from accumulation of nutrients, metals, and hydrocarbons in sediments below permeable pavement also is a concern (Taguchi and others, 2020). Fertilizers in contributing areas run-on and have the potential to wash into a permeable pavement system and infiltrate to groundwater. Wash off from streets onto permeable pavement may include metals or hydrocarbons.

During high inflow events, underdrains in some permeable pavement and bioswale systems may release stormflow almost directly to storm sewers. This reduced residence time in the GI basin may limit the filtering effect and increase the potential for contaminants to move to storm drains and eventually to streams (Stewart and others, 2017). The internal water storage system may also release increased phosphorus concentrations and loads in outflows during large events.

Performance of GI depends on infiltration to happen. Sediment transported to the site from construction and other activities can greatly reduce infiltration rates through clogging and compaction. Winter sand application clogged some sites after just 1 year when spring runoff carried high sediment loads (Selbig and others, 2019). The area most prone to clogging at permeable pavement sites may be the permeable/impermeable interface where the largest sediment load is deposited (Winston and others, 2015).

Vegetation and soil media can cause unintended consequences for vegetated basins and channels, tree canopy, and green roofs when materials selected are inadequate for the practice. Vegetation within these practices provide biological treatment that can permanently remove certain organic contaminants through biodegradation depending on the health of the vegetation and soil microorganisms (Taguchi and others, 2020); however, overuse of compost, fertilizer, irrigation, or other common landscaping practices can result in accelerated nutrient accumulation and lead to increased contaminant concentrations in stormwater throughflow (Taguchi and others, 2020). Vegetated basins and channels are often successful in reducing metal concentrations (common in street runoff), and export of phosphorus from accumulated sediments has been observed (Taguchi and others, 2020). Salt used for de-icing can also be harmful to vegetation in GI installations.

Trees that border streets contribute organic matter and nutrients, in the form of bulk solids as particles including leaves, pollen, nuts, and fruits, to streets (Taguchi and others, 2020). Janke and others (2017) demonstrated a strong relation among nitrogen and phosphorus concentrations and tree canopy over impervious surfaces. Coarse organic solids from trees are likely a major source of solids to storm drains (and to GI practices), increasing maintenance costs (Taguchi and others, 2020). Winter leaching of leaves may also be a major source of nutrients to winter stormwater (Taguchi and others, 2020). Tree canopy along roadways may funnel water to impervious surfaces. Locations beneath trees, draining pervious areas, and with higher traffic load were also more susceptible to clogging (Winston and others, 2015). Street sweeping can greatly reduce nutrient loading to stormwater (Selbig, 2016).

Atmospheric deposition and the addition of nutrients are the primary sources of contaminants from green roofs. Nutrient rich stormwater draining from green roofs can reach storm drains or infiltrate to groundwater if not routed to a secondary GI practice (Taguchi and others, 2020). Green roofs do not have contributing areas beyond the roof, therefore, run-on from other surfaces do not cause accumulation of contaminants. During dry conditions, green roofs may need irrigation. When vegetation dies and is not replaced, vegetated basins, vegetated channels, tree canopy, and green roofs lose the pollutant uptake and evapotranspiration benefits provided by the plants (U.S. Environmental Protection Agency, 2016b). Lack of vegetation can lead to lower porosity of soil media, preventing infiltration. Mulch requires less maintenance and provides some direct pollutant and sedimentation removal; however, it is organic and decomposes with time, potentially releasing nutrients to the environment. In addition, improper balance of soil amendments could also lead to nutrient export. Because vegetated basins and channels have the capacity to accumulate contaminants with time, contaminated soils may need to be periodically removed and replaced.

Vegetated basins, vegetated channels, urban tree canopy, and green roofs require planting, watering, fertilizing and other maintenance. Many of the unintended consequences associated with GI are because of lack of maintenance, and when not addressed properly this can lead to failure and high costs associated with restoration (U.S. Environmental Protection Agency, 2016b). Practices that serve highly impervious roadways and parking lots (such as vegetated basins, vegetated channels, and permeable pavements) require a high level of maintenance; without maintenance, these practices can be a source of high nutrient and sediment loads, heavy metals, and organic compounds to streams and groundwater. Increased sedimentation, because of lack of maintenance, may also cause localized flooding around the GI installation, and in some cases, cause worse flooding than preinstallation conditions; for example, the pavers at the Juarez High School (part of the Cermak-Blue Island Streetscapes Corridor, fig. 20) were never cleaned and had substantial deterioration in terms of infiltration performance and often pond water at the depression areas in the plaza and sidewalks (Kumar and others, 2017).