Viii lid technology: case studies and watershed restoration

J. T. Bushey¹, C.M. Fleischmann², E.D. Jackson³, C. Atkinson-Palombo⁴

Stormwater runoff, and its associated pollutants, is considered a problem in many urban watersheds. One way to control runoff is low impact development (LID) which aims to mimic natural hydrologic processes by breaking up the impervious surface coverage with various forms of vegetation and other pervious land cover such as porous pavement. Some forms of LID have the added benefit of calming traffic, increasing the amount of greenspace and enhancing the walkability of neighborhoods. Hydrologists have a variety of ways to measure how effective LID is from a hydrological perspective, using individual sites as the geographic unit of analysis. Yet, few studies have investigated impacts at the neighborhood scale, and to date, no studies have investigated the impact of LID on transportation. We evaluated LID implementation scenarios for a range of percentage of streets converted for a neighborhood in Hartford, CT. Hydrologic evaluation was performed through a SWMM model while transportation evaluation was performed via VISSIM and TransCAD. Hydrologically, swales, bioretention basins and porous pavement obtained runoff reductions of approximately 17% with 100% conversion. However, transportation evaluation revealed that the distance and time traveled increased dramatically with only 5.7 km of lane miles converted (23%). This suggests that transportation needs may limit the potential for LID implementation in such developed neighborhoods.

Stormwater management in urban areas represents an environmental challenge and primary focus of sustainable development (USEPA, 2007; USEPA, 2003). Historically, stormwater management aimed to rapidly collect and deliver surface runoff from developed lands to streams, lakes and rivers (Seybert, 2006). This approach has altered the hydrologic cycle in urban areas; minimizing infiltration to groundwater and enhancing runoff via impervious surfaces (Paul and Meyer, 2001; Arnold and Gibbons, 1996) leading to “urban stream syndrome” (Walsh et al., 2005). Additionally, runoff from impervious surfaces and roadways delivers contaminants such as heavy metals (Watts et al., 2007) to waterways without treatment.

Recent efforts to mitigate stormwater impacts have shifted from conventional engineered systems towards a concept of low-impact development (LID). LID is a site design strategy with the goal of maintaining or replicating the pre-development hydrologic regime. Various best management practices (BMPs; e.g., bioswales, bioretentioin basins, porous pavement, tree boxes, rain barrels) have been used in retrofitting existing development and in planning for new development to achieve hydrologic landscape objectives (Lai et al., 2005; USEPA, 2000). Hydrologically, the primary directive of LID is to mimic natural ecosystem processes and to foster the use of green spaces and natural infiltration processes when possible. However, LID is not limited to hydrologic benefits, but provides additional improves regarding community sustainability (Garrison and Hobbs, 2011). Benefits of LID include increased greenspace, decreased greenhouse gas emissions, healthier communities, and walkability (USEPA, 2012). Traffic calming has been demonstrated to be a benefit of site-specific LID design (e.g., Matel, 2010), with site-specific improvements designed to facilitate walkability and bicycle use (e.g., Portland, OR; Garrison and Hobbs, 2011).

The hydrologic and traffic calming benefits of LID site designs have been demonstrated at the site level (e.g., Hager, 2003; Lehner et al. 1999; Matel, 2010; UNH Stormwater Center, 2012). Prior research efforts regarding LID optimization largely have addressed site-specific factors (Montalto et al. 2007; Dreelin et al., 2006; UNH Stormwater Center, 2007). However, the implementation of LID and other community stormwater and traffic management strategies occurs over multiple scales, from site to neighborhood to watershed level (Damodaram et al., 2010; Williams and Wise, 2006). Design factors and challenges differ for large scale LID implementation at the watershed level relative to concerns at the site level, a fact recognized by the USEPA for urban watersheds (Lai et al., 2005). Characterizing the neighborhood-scale role of LID as a stormwater mitigation measure is of particular importance given the recent USEPA Memorandum of Understanding and Letter of Intent (USEPA, 2012) supporting the use of Green Infrastructure for stormwater and CSO management. However, studies at the neighborhood scale are limited with those which exist largely scaled up from those performed at the site level (Arnold and Gibbons, 1996). In a recent report, only Milwaukee, WI, and Philadelphia, PA, are listed as having performed system-wide modeling analysis, with Milwaukee’s assessment only covering six acres. Additionally, factors that may be beneficial at the site scale such as traffic calming may inhibit large-scale implementation at the watershed level. Decreased roadway widths can lead to negative transportation impacts on a neighborhood scale, and particularly is important for public transit corridors (Jackson et al., 2012). Yet, the analysis of LID impacts on transportation at this scale has not been examined.

To address this gap in scale between LID implementation, stormwater management, and traffic requirements, we assessed the potential benefits in stormwater runoff reduction at a neighborhood scale, evaluating the ability of various types and amounts of LID features to mitigate runoff in a dense residential urban neighborhood. Concurrently, we evaluated the effect of LID placement and coverage on traffic flow. Our review provides an assessment of the potential for LID features to alleviate stormwater runoff at larger scales under the constraints of space and traffic requirements in an existing urban neighborhood. Five common LID technologies (permeable pavement, swales, bioretention cells, rain barrels, and tree box filters) were assessed at a variety of implementation levels within the watershed using an EPA SWMM model to determine changes in runoff reduction (Fleischmann et al., 2012). LID features were selected with transportation considering transportation requirements and modeled for the effect on traffic flow (Jackson et al., 2012).

Study Area

A small urban watershed located northwest of downtown Hartford, CT was selected for analysis (Figure 1). The Granby sub-section of the North Branch Park River Watershed is characterized as an urban high-density residential neighborhood composed predominantly of privately-owned properties. The 167 ha study area is contained approximately by Granby Street to the west, Blue Hills Avenue to the east, Burnham Street to the north and Westbourne Parkway to the south. Total roadway distance in this area is just over 24 km. The average percent impervious cover (% IC) for the 119 subcatchments is 45% and predominantly consists of transportation infrastructure (parking lots, driveways, roadways) and roof tops. Roof tops comprise the greatest percentage of the impervious cover within the watershed at 19% of the total land area followed by roads, driveways and parking lots at 14%, 10.6%, and 1.4%, respectively. Soils throughout all 119 catchments have been classified as moderately well-drained sandy and silty loam soils (USDA, 2003). While the base soil is suitable for infiltration

and the installation of LID measures, surface soils have been heavily modified, representative of an urban environment.

The design of this neighborhood typifies an urban residential design layout: the transportation infrastructure is a gridded pattern with wide, curbed streets flanked by pedestrian walkways. Little commercial development exists in the community; less than seven percent by area of the watershed is town owned/commercial properties. This area was designed to facilitate travel via vehicle to shopping districts located at the north and south end of the neighborhood with a high percentage of residents also commuting to work outside of the neighborhood. The two main transportation arteries through this neighborhood are Granby Street and Blue Hills Avenue. These routes run north-south and serve at the primary arterials into Hartford. While there are bus stops along Blue Hills and Granby Streets as well as ample sidewalks, the primary mode of transportation in this

watershed in northwest Hartford, CT, USA.

For this analysis, we investigated five LID technologies: porous pavement (PP), swales, bioretention cells, rain barrels and tree boxes. Our approach highlights a top-down watershed-level implementation evaluation. As such, we did not evaluate specific individual site suitability as has been the focus of previous studies (Garrison and Hobbs, 2011). Instead, we addressed the overall potential of LID in watershed management decisions. The LID features considered were divided into two categories: (1) roadway – those LID options (PP, swales, bioretention) that would be implemented in the roadway and could potentially alter traffic patterns, and (2) non-roadway – LID technologies that would not interfere with traffic (rain barrel, tree boxes). For roadway design options, we maintained at least one travel lane and one parking lane following LID implementation. The implementation of the two non-roadway LID technologies was designed to correspond to the roadway patterns used for the roadway LID technology implementation (Fleischmann et al., 2012) based on design recommendations for tree boxes (Virginia Department of Conservation, 1999) and assuming one rain barrel per house with that rain barrel draining half of the roof. Green roofs were not considered due to concerns over the structural limitations of the existing roofs and the lack of publically-owned roof acreage. Increased vegetative coverage was also not considered as tree coverage is already relatively dense in the current green spaces.

We evaluated the effectiveness of LID features for minimizing watershed runoff using SWMM Version 5.0.022, a hydrologic model developed and updated by the USEPA (USEPA, 2011). SWMM was selected as: (1) an existing, verified SWMM model of the area of interest was available (The Metropolitan District Commission, Hartford, CT) and (2) Version 5.0.022 has the ability to model various LID features. SWMM is a dynamic rainfall-runoff watershed simulation model designed for modeling urban areas to predict the resultant runoff from each sub-catchment in response to precipitation. Each sub-catchment is parameterized by percent pervious/impervious, average slope, storage and infiltration. In addition to watershed runoff/infiltration, the SWMM model incorporated engineered stormwater infrastructure (e.g., stormwater pipes, catch basins) as well as potential groundwater contributions to streams and the

piping network.

Transportation methods

The impacts of LID features on the transportation network were modeled in transportation simulation models, TransCAD and VISSIM. The traditional 4-step planning model (Trip Generation, Trip Distribution, Mode choice and Assignment) was used to simulate traffic on the network. Census tract data were used to estimate the number of trips generated from and attracted to each zone within the network. This resulted in an origin-destination (O-D) matrix. This matrix was assumed to be static and not impacted by LID improvements. Also for simplicity, the mode choice was assumed to be negligible and not impacted by LID improvements. The resulting O-D Matrix was then assigned to routes throughout the network to get travelers from their origin to their destination. These assignments were made based on the current characteristic of each link of the network. As the proposed LID improvements were applied stepwise to the network, the O-D matrix was assigned to the network based on the new characteristics of each link. For example, if the 2.4 km of roadway were to be converted to a one-way street with a grassy swale, the links would be changed to one way travel links, thus restricting simulated traffic to use this link only for one way traffic. The resulting change in traffic flow and patterns were noted for proposed LID scenario and level of implementation. Overall changes in vehicle miles traveled (VMT), vehicle hours traveled (VHT) and number of vehicles traveling on each link (flow) were summarized.

We compared the hydrologic benefit of each LID feature and the transportation impact of LID implementation over a range of percent coverage. LID options were implemented from baseline conditions in approximately 1.9 km increments through 11.3 km with additional hydrologic analysis at 75% and 100% coverage. The streets selected for implementation were evenly distributed throughout the watershed while accounting for transportation needs (Fleischmann et al., 2012). Due to public transportation corridors, and the primarily north-south traffic flow, street selection focused on east-west secondary roadways (Jackson et al., 2012). As implementation coverage increased, secondary east-west streets were selected followed by secondary north-south streets. Once specific streets were determined, the roadway length was converted to a total implementable distance in each of the 119 subcatchments for entry into SWMM by assigning the selected streets to their respective subcatchments. Hydrologic simulations were performed using a 1-yr storm event, the minimum design storm for most LID technologies (ISUIT, 2009).

A comparison of percent runoff reduction to implementation distance was conducted using SWMM for each LID technology (Figure 2). Porous pavement, bioretention and vegetated swales were comparable in terms of runoff reduction per implementable distance. All of the LID technologies assessed ranged from 1% percent reduction at the 2 km implementation distance through 17% reduction for full implementation (100% or 24.4 km of roadway implementation, Figure 2). The trends were approximately linear with variation of increased coverage due to street-specific differences (Fleischmann et al., 2012). Certain streets, and therefore catchments, have a greater potential to reduce runoff with the implementation of a pervious surface (LID). Rain barrels and tree box filters were less effective methods, with maximum runoff reduction potential of 4% and 6%, respectively (Figure 2). Rain barrels account for a very small decrease in percent reduction as they do not hold a substantial amount of water. Tree box filters also were not as effective in capturing runoff due to their small size (3.34 m²) and the large amount of space suggested between the boxes (30.5 m) in order to maximize performance (Virginia Department of Conservation, 1999). Based on the comparison of percent runoff reduction with implementable distance, porous pavement, vegetated swales and bioretention cells would be appropriate options for maximizing runoff reduction in this type of urban watershed.

NOTE: Total linear road length in the watershed is 24 km. Runoff reduction evaluated using SWMM model validated for the watershed for porous pavement (PP), swales, bioretention basins (Bio), rain barrels, and tree boxes. Lines do not reflect trends.