Low-impact-development (LID) facilities such as green roofs, bioretention swales, rainwater cisterns, rain gardens, impervious runoff dispersion, and permeable pavement can be used to reduce stormwater runoff.
The effectiveness of different LIDs to reduce stormwater runoff depends on the individual LID facility, the native soil, and the local climate. In the United States, there is a great range of regional climates and climatic conditions that affect the effectiveness of different LIDs.
Clear Creek Solutions used continuous simulation hydrologic computer modeling of LIDs to evaluate the effectiveness of different LIDs for these different regions of the country. The LID modeling was based on Hydrological Simulation Program–Fortran (HSPF) using the Western Washington Hydrology Model (WWHM) interface. Results are presented in terms of the percent reduction in mean annual stormwater runoff for each LID with each major soil type in each of the major regions represented by Philadelphia, PA; Atlanta, GA; Seattle, WA; and Los Angeles, CA.
Overview
LIDs use two mechanisms to reduce stormwater runoff: infiltration and evapotranspiration. LIDs slow the movement of runoff as it travels toward a stormwater conveyance system for collection and removal. This gives the native soil more opportunity to infiltrate some or all of the runoff. At the same time the opportunity for additional evaporation and transpiration (evapotranspiration) is increased.
Infiltration is limited by the infiltration characteristics of the native soil. Alluvial soils, such as those found in Philadelphia, have higher infiltration rates than the glacier-compacted till soils found in Seattle or the clay soils of Los Angeles and Atlanta. This difference plays a large role in the relative effectiveness of different LIDs to reduce stormwater runoff volume in these two different soil environments.
Evapotranspiration is limited by the atmospheric conditions. Potential evapotranspiration (PET) is typically computed using pan evaporation data and a pan evaporation multiplication factor to convert pan evaporation data to PET data. Actual evapotranspiration (AET) occurs when the atmosphere has capacity to hold additional water vapor and there is sufficient atmospheric energy to convert water from liquid to gas. This energy transfer is dependent on the relative humidity, solar radiation, air temperature, and wind speed. AET is always less than or equal to PET. Seattle, with its long, gray, wet, overcast winter days has an average of only 3.5 inches of PET between the months of November and March, while its average rainfall for this period is 25 inches. This limited amount of winter PET reduces the effectiveness of LIDs to decrease stormwater runoff by evapotranspiration. By comparison, Los Angeles averages only 14 inches of precipitation but has 8 inches of PET during this same period. As discussed below, LIDs generally provide greater runoff reduction in Los Angeles than Seattle, but are useful in both climates.
Modeled LID Facilities
Six LIDs were modeled with WWHM4:
- Impervious area disconnection
- Rain garden
- Grass bioswale
- Bioretention
- Permeable pavement
- Green roof
The Philadelphia LIDs were modeled using 57 years of Philadelphia International Airport National Weather Service hourly precipitation data (1949 to 2005). The Seattle LIDs were modeled using 49 years of Seattle-Tacoma International Airport National Weather Service hourly precipitation data (1949 to 1997). Los Angeles LIDs were modeled using 57 years of Burbank Valley hourly precipitation (1949 to 2006). Atlanta LIDs were modeled using 86 years of Albany hourly precipitation (1921 to 2006).
Except for the green roof, all of the LIDs reduce stormwater runoff using both infiltration and evapotranspiration. Green roofs reduce stormwater runoff using only evapotranspiration (further discussion below).
Impervious Area Disconnection Modeling. Impervious area disconnection, also known as lateral flow dispersion, occurs when runoff from an impervious surface is dispersed on an adjacent pervious area. This movement of the impervious surface runoff across the adjacent pervious surface gives the surface runoff an opportunity to infiltrate into the soil before it is collected into a stormwater conveyance system. The amount of infiltration is a function of the soil’s properties and antecedent soil moisture
conditions.
For comparison purposes, the conventional site was modeled with the 1.00-acre impervious area draining directly to a stormwater collection system and the adjacent 0.25-acre pervious lawn doing the same. The impervious area disconnect LID collects the total runoff only after it flows off of the 0.25-acre pervious lawn area.
The specified soil characteristics are input in terms of Natural Resources Conservation Service (NRCS) soil categories (A, B, C, and D). For this example, I selected a B soil for Philadelphia. For Atlanta, Seattle, and Los Angeles, C soil is appropriate.
Rain Garden Modeling. The terms rain garden and bioretention are often used interchangeably. For the purposes of this modeling work we are defining a rain garden as a bioretention facility that does not have an underdrain and are modeling it accordingly. Bioretention (described below) has an underdrain.
In a rain garden the native soils have been excavated and replaced with amended soil. At the downstream end of the rain garden a berm or weir controls the surface discharge from the rain garden and detains runoff, encouraging it to infiltrate into the amended soil. Infiltration from the amended soil to the native soil is also possible, depending on the properties of the native soil. In WWHM4, rain gardens are represented by the “bio swale element.”
WWHM4 also uses HSPF Special Action algorithms to route runoff first into the surface storage portion of the rain garden and then checks to determine the amount of runoff that can infiltrate into the top amended soil layer. If the inflow exceeds the amended soil infiltration rate or capacity, then water ponds on the surface and is available for surface discharge to the next downstream conveyance feature. Infiltration into the native soil is allowed, if desired. The native soil conditions often play a critical role in determining the effectiveness of a rain garden to reduce stormwater runoff. WWHM4 simplifies the native soil input to a constant infiltration rate (inches per hour). The actual amount of infiltration during a single time step is calculated in WWHM4 using the user-input infiltration rate and the size of the infiltration area. For the purposes of this modeling work, the native soil infiltration rate was assumed to be equal to 0.39 inch per hour for the Philadelphia site and 0.04 inch per hour for the Seattle, Atlanta, and Los Angeles sites.
For comparison purposes, the conventional site was modeled with the 0.25-acre impervious roof area draining directly to a stormwater collection system and the adjacent 0.025-acre pervious lawn doing the same. For the rain garden LID, the lawn area is replaced by the rain garden, and all of the roof runoff is directed into the rain garden area.
I selected a light loamy medium coarse sand with a Ksat = 10 centimeters per hour for Philadelphia, Atlanta, and Los Angeles for the amended soil layer. For Seattle, an amended soil with a Ksat = 6.4 centimeters per hour (2.5 inches per hour) was selected for the amended soil layer based on local experience.
Grass Bioswale Modeling. The grass bioswale LID is potentially a combination of a rain garden and an open channel. There is opportunity for infiltration into the soil underlying the open channel.
For comparison purposes, the conventional site was modeled with a 1.00-acre impervious parking area draining directly to a stormwater collection system and the adjacent 0.25-acre pervious lawn draining to an open channel. For the grass bioswale LID, both the runoff from the impervious parking area and the lawn area are routed through the grass bioswale.
WWHM4 open channel element represents the grass bioswale. The user inputs soil characteristics in terms of NRCS soil categories (A, B, C, or D). For this example, I selected a B soil for Philadelphia. For Seattle, Atlanta, and Los Angeles, a C soil is appropriate.
Bioretention System Modeling. As described above, the difference between a bioretention LID and a rain garden LID is the use of an underdrain. Bioretention has an underdrain; a rain garden does not.
Other than the addition of an underdrain, the methodology is the same for both types of LID. The native soils have been excavated and replaced with amended soil. At the downstream end a berm or weir controls the surface discharge from the bioretention and detains runoff, encouraging it to infiltrate into the amended soil. Infiltration from the amended soil to the native soil is also possible, depending on the properties of the native soil. In WWHM4, bioretention is represented by the bio swale element.
The underdrain is assumed to be located at the bottom of the bottom amended soil layer, just above the native soil. The user inputs the underdrain diameter and the underdrain orifice diameter. The orifice diameter controls the discharge from the underdrain.
For comparison purposes, the conventional site was modeled with a 1.00-acre impervious parking area draining directly to a stormwater collection system and the adjacent 0.10-acre pervious lawn doing the same. For the bioretention LID, the lawn area is replaced by the bioretention facility, and all of the parking area runoff is directed into the bioretention area.
I selected a light loamy medium coarse sand with a Ksat = 4 inches per hour for Philadelphia, Atlanta, and Los Angeles for the top amended soil layer and gravel for the second layer. For Seattle, an amended soil with a Ksat = 2.5 inches per hour was selected for the top amended soil layer and gravel for the second layer based on local
experience.
Permeable Pavement Modeling. Permeable pavement has been in use for the past 30 years, although applications have been limited due to concerns about the water infiltrating and weakening the soil under the pavement. Cost considerations also have been an issue.
WWHM4 represents permeable pavement with multiple material layers underlain by native soil. Each material layer has a depth and porosity specified by the user. Typically permeable pavement has two layers: the upper layer represents the pavement depth, and the lower layer is the gravel subgrade placed between the permeable pavement and the native soil. If appropriate, the user can then infiltrate the water that travels through the permeable pavement and the gravel layer into the native soil.
To evaluate the effectiveness of permeable pavement in reducing runoff volume, a permeable pavement area (roadway) was modeled with the permeable pavement layer 0.33 foot thick with a porosity of 0.25. Below the permeable pavement is a 0.66-foot gravel subgrade with a porosity of 0.40. An underdrain with a diameter of 3 inches has been placed 0.33 foot above the gravel-native soil interface.
The native soil conditions play a critical role in determining the effectiveness of a LID to reduce stormwater runoff. For the purposes of this modeling work, the infiltration rate was assumed to be equal to 0.39 inches per hour for the Philadelphia site and 0.04 inches per hour for the Seattle, Atlanta, and Los Angeles sites.
Green Roof Modeling. Green roofs (also known as ecoroofs or vegetated roofs) have the potential to reduce peak flows and total runoff volume from roofs that are otherwise impervious. Green roofs can be retrofitted on commercial buildings in the urban core where other stormwater mitigation measures are not feasible. Green roof retrofitting can be used to reduce urban stormwater inflows to a city’s combined sewer system and eliminate combined sewer overflows (CSOs) to comply with state and federal regulations.
Unlike the other LIDs, green roofs cannot infiltrate water into the native soil. Infiltration is limited to the depth of growing media on the roof. With limited available soil moisture storage, the major stormwater reduction benefit comes from evapotranspiration. The size and type of vegetation determines the green roof’s interception storage (water storage on the plant surfaces) and actual evapotranspiration.
To define the green roof, the user provides the depth of material (growing media), roof slope, vegetative cover type (ground cover, shrubs, or trees), and the rooftop length (maximum flow travel distance to a drain).
For the purpose of this modeling exercise, the depth of material was set equal to the growing media depth (4 inches) plus drainage layer depth (2 inches) for a total depth of 6 inches. The roof slope was input as 0.02 foot per foot. The vegetative cover is ground cover, due to the small growing media depth. The rooftop length is 65 feet.
Unlike the other LIDs, the only difference in modeling the green roof for the four locations is the input meteorological data (precipitation and potential evapotranspiration).
Model Results
The WWHM4 model results are highly influenced by the assumed native soil characteristics. These characteristics (described above) control how much runoff is infiltrated into the native soil and removed from the stormwater conveyance system. The only LID where native soil characteristics are not important is the green roof, where infiltration into the native soil is not possible.
Table 1 shows that, of the six LIDs modeled, permeable pavement is the most effective LID in reducing runoff volume. This is true for all four
locations.
Permeable pavement offers the greatest annual runoff volume reductions of the six LIDs modeled because there is sufficient runoff storage in the gravel subgrade layer to allow for long-term slow infiltration into the underlying native soil. If infiltration into the native soil is not allowed, then the effectiveness of the permeable pavement is greatly reduced.
Somewhat surprisingly, permeable pavement was less effective in reducing annual runoff volume in Los Angeles than in the other three modeled locations. This is partially because of the low native soil infiltration rate (0.04 inch per hour), but Seattle and Atlanta used the same low rate. More important was the distribution of the rainfall. Los Angeles has the lowest annual precipitation of the four sites modeled (16 inches), but when it does rain the rainfall intensity is high and at times overwhelms the storage capacity of the permeable pavement’s gravel sublayer. This results in water ponding on the surface of the pavement and surface runoff in addition to flow through the underdrain. By comparison, Seattle has more than double the Los Angeles annual precipitation (38 inches versus 16), but Seattle’s rainfall intensities are much lower and only produce underdrain runoff. It should be noted that underdrain runoff is routed to a conveyance system and is not included in the runoff volume reduction.
Rain gardens, bioretention, and grass bioswales all perform similarly. They all provide some additional evapotranspiration and infiltration to the native soil (if allowed). Their effectiveness in reducing annual runoff volume is highly dependent on the native soil infiltration rate. As a result, the Philadelphia site with its B soil shows a greater reduction in annual runoff volume for these LIDs than the C soil sites (Los Angeles, Seattle, and Atlanta). Precipitation and evapotranspiration play a role in the relative effectiveness of the rain gardens, bioretention, and grass bioswales, but to a lesser extent than the native soil infiltration rate.
Impervious area disconnection appears to be the least effective LID of the six modeled LIDs in reducing annual runoff volume. As with the rain garden, bioretention, and grass bioswale LIDs, the effectiveness of impervious area disconnection is directly dependent on the native soil.
Unlike the other five LIDs, the effectiveness of green roofs in reducing annual runoff volume is independent of the native soil infiltration rate. This is because the green roof does not (and cannot) infiltrate runoff to the native soil; the effectiveness of the green roof LID is dependent on the site’s evapotranspiration. More specifically, the effectiveness of the green roof LID is dependent on the difference between the precipitation and the potential evapotranspiration during the wet season. This is when the green roof receives most of its precipitation and is most likely to produce stormwater runoff. Atlanta, Philadelphia, and Los Angeles all show similar annual runoff volume reductions in the range of 38% to 44%. Seattle lags behind with 22% reduction. The reason for Seattle’s lower value is the relatively meager potential evapotranspiration of only 3.5 inches total for the winter period of November through March; the corresponding total precipitation for this period is 25 inches. The other three regional sites do not have a comparably large difference between the precipitation and evapotranspiration.
Summary
The WWHM4 model results are highly influenced by the assumed native soil characteristics. These characteristics control how much runoff is infiltrated into the native soil and removed from the stormwater conveyance system. The only LID where native soil characteristics are not important is the green roof, where infiltration into the native soil is not possible.
Permeable pavement offers the greatest annual runoff volume reductions of the six LIDs modeled because there is sufficient runoff storage in the gravel subgrade layer to allow for long-term slow infiltration into the underlying native soil. The most important regional factor is the distribution of the rainfall. Los Angeles has the lowest annual precipitation of the four sites modeled (16 inches), but when it does rain the rainfall intensity is high and at times overwhelms the storage capacity of the permeable pavement’s gravel sublayer.
Rain gardens, bioretention, and grass bioswales all perform similarly. They all provide some additional evapotranspiration and infiltration to the native soil (if allowed). Their effectiveness in reducing annual runoff volume is highly dependent on the native soil infiltration rate.
Impervious area disconnection appears to be the least effective LID of the six modeled LIDs in reducing annual runoff volume. The effectiveness of impervious area disconnection is directly dependent on the native soil.
The effectiveness of green roofs in reducing annual runoff volume is independent of the native soil infiltration rate. The effectiveness of the green roof LID is dependent on the difference between the precipitation and the potential evapotranspiration during the wet season.
Summary conventional and LID annual stormwater runoff volumes are shown in Table 2.
Conclusions
Different LIDs perform differently in different climates. Green roofs, which rely on evapotranspiration, provide greater stormwater volume reduction benefits in climates where the potential evapotranspiration is relatively high compared to the precipitation.
The different LID source area to infiltration area ratios modeled for the different LIDs makes it difficult to directly compare the effectiveness of each LID. LIDs like permeable pavement with 1:1 source area to infiltration area ratio are the most effective in reducing stormwater runoff volume because of the low source area to infiltration footprint ratio.
The regional model results presented above also are highly influenced by the assumed native soil characteristics. These characteristics control how much runoff is infiltrated into the native soil and removed from the stormwater conveyance system. Based on the assumptions made for this LID modeling, the infiltration rates used in the models were the dominant factor in reducing stormwater runoff volume.
The infiltration rates combined with the source area to infiltration area ratio largely determined the effectiveness of each LID modeled. This should be true regardless of whatstormwater modeling software package is used.
