GIS/EM4 - Modeling in an urban park environment: a water quality modeling case study in Central Park

Modeling in an urban park environment:

a water quality modeling case study in Central Park

GIS/EM4 No. 119

Mark R. Lenz
Cindy F.H. How

Abstract

In an effort to improve water quality in Central Park's eight man-made waterbodies, GIS was used with SWMM and WASP to develop and evaluate water quality management plans. This paper focuses on the land-based SWMM modeling and how spatial scale and temporal scale issues were quantified to accurately represent the unusual urban park environment. Large scale data and small time steps were required to account for the spatial and temporal scales. The existence of existing drainage piping, irrigation and fertilization management practices, and intense foot traffic affected watershed delineations, event mean concentrations (EMCs), and runoff curve numbers. The quantification of these affects was accomplished through the overlay and spatial analysis capabilities of GIS.

Keywords

Watershed model, water quality model, urban parks, microscale GIS, event mean concentration.

Introduction

With 20 million visitors each year, Central Park is the most frequently visited urban park in the United States. The eight man-made waterbodies in Central Park represent more than 15% of the total area of the Park and are an integral part of the visitor experience. While the lakes and ponds serve as Park attractions, they often become eyesores on the landscape, particularly during warm summer periods when they are prone to unsightly algal blooms. To address these problems, the New York City Department of Environmental Protection (NYCDEP) and Malcolm Pirnie, Inc. (Malcolm Pirnie) are conducting a Water Quality Management and Planning study. Part of this study involves using GIS in conjunction with a land-based watershed model (SWMM) and a lake water quality model (WASP5) to develop and evaluate alternative management plans.

Problem statement

In an effort to improve water quality in the Central Park waterbodies three water quality management plans were developed. Each plan includes combinations of structural and non-structural best management practices (BMPs) aimed at improving water quality. A system consisting of a GIS, the United States Environmental Protection Agency's (EPA) Storm Water Management Model (SWMM) runoff water quality model, and the Water Quality Analysis Simulation Program, version 5, (WASP5) lake water quality model was used to evaluate the anticipated benefits of each plan. Throughout the modeling process, the urban park environment presented several interesting issues and challenges. This paper will focus on the development, calibration, and application of the land-based watershed model (SWMM), how the GIS was used to facilitate the modeling, and three primary issues encountered during modeling: urban park environmental issues, spatial scale issues, and temporal scale issues.

Urban park environmental factors

The urban park environment within Central Park is exemplified by the waterbodies; they are man-made features that are designed to look like natural features of the landscape. Because the Park is "land-locked", artificial sources and outlets must be provided to prevent the lakes and ponds from becoming stagnant. The current source of flow for the waterbodies is potable water. Although this source ensures turnover, orthophosphate that is used to control corrosion within the drinking water distribution system artificially elevates the available nutrient concentrations within the waterbodies. Currently all of the ponds and lakes have a continuous discharge to the City combined sewer system.

Central Park is an unusual mix of land uses. It has large, grassy open spaces and various patches of dense tree cover, as well as heavily trafficked transverse roads that serve as the main thoroughfare between east and west Manhattan for a series of 51 blocks. In addition to automobile traffic, the Park sees large amounts of foot traffic that takes its toll on the land condition. In an effort to offset the effects of intense foot traffic, parts of the Park experience heavy maintenance regimes. These heavily maintained areas are subject to irrigation, fertilization, and pesticide application.

Built on a once-swampy area, a gravity-based drainage system was originally installed in the Park. This system has been altered throughout the years, though an extensive drainage network still exists. The drainage network was originally designed to intercept runoff that would saturate the soil. The intercepted water is subsequently discharged to Park waterbodies or City combined sewers, altering the topography based drainage patterns.

Modeling in a micro-scale environment

With a total area of 847 acres (1.3 mi2/3.4 km2), Central Park is a microcosmic complex environment. To effectively model the assorted land uses, the varied topography, and the complex drainage pattern of the Park, large-scale data was needed to accurately delineate watersheds and determine watershed parameters. To represent the complexity of the environment, small sub-watersheds were needed, at times approximately 1 acre in size. These small watersheds created the need to consider an appropriate time scale.

Temporal effects on a micro-scale environment

An effect of small watersheds is a short time of concentration (Tc). The time of concentration is defined as the time from the beginning of rainfall till flow from the entire watershed is contributing to the hydrograph. This parameter determines the shape of the runoff hydrograph. To ensure that the hydrograph represents the contribution of all the Park watersheds, a relatively small (15 minute) time step is required.

Background

As previously stated, this paper focuses on the land-based stormwater modeling and the GIS tools used to facilitate the modeling. Multiple methods are available to generate stormwater hydrographs and to determine water quality. GIS tools are available that help generate model input and aide in the creation of model input files. As a full discussion of these topics is beyond the scope of this paper, a brief introduction to some of these methods and tools in relation to their applicability in the urban park environment is provided.

Land-based stormwater modeling

Runoff quantity is modeled based on one of two determining factors: a percent impervious or a runoff curve number. Percent impervious is often used in urban areas to predict runoff, as urban nonpoint-source loads are generally directly related to impervious surface area (Loucks et al. 1981). The Soil Conservation Service (SCS) method for determining initial abstractions is based on runoff curve numbers. The runoff curve number defines the runoff potential of an area, with higher curve numbers indicating higher runoff potential. The curve number is determined based on the hydrologic soil group (HSG), land use, land condition, antecedent runoff condition, and treatment of the area (USDA 1986). Several routing methods, used to determine the shape of the runoff hydrograph, are also available. These include non-linear reservoir methods, the SCS hydrology method, and unit hydrographs. The non-linear reservoir method is based on Manning's equation and continuity (James et al. 1999). A unit hydrograph is a unit pulse response function that considers each watershed as a lumped linear system (Chow et al. 1988). The SCS hydrology method uses a time of concentration (Tc) and a shape factor to determine the shape of the hydrograph.

The water quality expected in runoff is modeled by two basic theories: buildup/washoff and event mean concentration (EMC). The buildup/washoff method is based on the theory that solids and other pollutants are accumulated during dry periods and washed off during storm events (UWRRC 1992). Pollutants accumulate based on time, independently or in conjunction with land use, curb length, or area. Washoff can be a function of time or a function of runoff. An EMC is a constant concentration assigned to runoff at all times. Loads differ between storm events based on variable flows. An event mean concentration is independent of time and is strictly a function of land use.

Available GIS Tools

Due to the spatial nature of watersheds, and the need to conduct overlay analysis to develop appropriate parameters, GIS can facilitate land-based modeling. Several tools are available to help develop model input parameters and to help generate model input files. Applications such as the Watershed Delineator, developed by the Environmental Systems Research Institute (ESRI), and CRWR-PrePro, developed at the Center for Research in Water Resources at the University of Texas at Austin, can delineate watersheds based on gridded topographic data in the form of digital elevation models (DEMs). CRWR-PrePro can also calculate a curve number grid based on soil and land use data, as well as a watershed's time of concentration (Olivera and Maidment 1999). Both the Watershed Delineator and CRWR-PrePro run in ESRI's desktop application, ArcView. Programs, such as PCSWMM GIS (Computation Hydraulics International), and Visual Hydro (CAiCE Software) have GIS tools for creating SWMM input files based or running the SWMM model on databases created through GIS interfaces.

Approach

The EPA Storm Water Management Model (SWMM) was used for land-based stormwater modeling. SWMM can simulate runoff quantity and quality from storm events through the previously discussed methods, creating hydrographs and pollutographs (graphs showing flow and pollutant load as a function of time). The selection of the appropriate modeling method was driven by the unusual environment in Central Park. In the urban park environment, which has characteristics of both urban and rural land uses, the percent impervious area is not the dominant driving force in determining runoff quantity; open spaces and tree-covered areas cover more than 75% of the Park. For this reason, the SCS curve number method was selected to model runoff capacity. SWMM uses the SCS routing method in conjunction with curve numbers to develop hydrographs.

The GIS application ArcView was used to aid in developing model input for both the SCS hydrology method and for applying EMCs. ArcView, with the CRWR-PrePro application, was used to aid in delineating watersheds in the Park and in determining the average runoff curve number for each watershed. Aerial photographs, 2-foot contours, street line files, building coverages, and hydrography data were acquired from the New York City Department of Environmental Protection (NYCDEP), which was conducting a project to acquire large-scale GIS data (1:800 scale) and aerial photography. Land use information was derived from leaf-off aerial photography and soils data was available as a soil survey previously conducted on the Park. The contours were used to develop a (DEM) that was used with the hydrography data to delineate watersheds. The land use and soil data was used with a lookup curve number table to develop runoff curve numbers using CRWR-PrePro.

The selection of EMCs for determining water quality was based on several factors: EMCs are applicable in both urban and agricultural environments, EMCs are consistent with the available data, and EMCs represent the level of detail required in determining load. As the purpose of the study was to predict long-term water quality within the waterbodies, predicting intra-storm variations in load was beyond the relevant scope of the study. Literature is also available with data on values found for EMCs for various land uses (USEPA 1983; Benaman 1996; Schueler and Claytor 1996).

ArcView was also used to help develop model input to apply EMCs; SWMM requires a percent land use for each subcatchment it models and an EMC is associated with each land use. The combination of land use percentages and land use-based EMCs determines the concentration associated with each subcatchment. The ArcView Spatial Analyst (ESRI) extension was used to tabulate the area associated with each land use within each watershed.

Methods

While literature values are available for both SCS hydrology and EMCs, calibration is the key step in applying SCS hydrology and EMCs to the urban environment. This step is used to adjust the literature values to reflect the conditions found within the Park. This section discusses the development and calibration of the model. Emphasis is placed on the adjustments made during calibration to account for the environmental factors and considerations necessary to deal with the small spatial and time scales.

Model development began with the delineation of watersheds. Originally, watersheds were delineated for each of the Park waterbodies based on topography. This did not, however, account for the gravity-based drainage system. The original watersheds were revised with information on the drainage system, then updated through field verification. After field verification of the piping, each watershed was divided into subwatersheds.. Many of these subwatersheds were small, but if they included impervious areas or heavily managed lawns they were found to contribute large volumes of flow and nutrient loads.

The drainage system also affected the time of concentration calculations and values. As discussed, calculating a time of concentration can be done in a GIS environment. However, these calculations do not account for urban alterations to topography-based drainage patterns. For this project, Tc values were done manually, calculated based on topography, land use, and the stormwater drainage system. The resulting values were all less than 1 hour and averaged less than 30 minutes. This was the dominant factor in selecting a15-minute time step to resolve the issues of temporal scale.

Both the drainage system and the intense use of the park affected the calibration of the runoff curve numbers input to the model. Recommended literature values for curve numbers based on land use and soil type (USDA, 1986) were used as initial values in determining curve numbers for each land use. However, heavy use of the park causes deteriorated soil conditions leading to increased curve numbers (Singleton, 1998). As part of the calibration process, curve numbers were increased throughout the park to account for the degraded conditions caused by human activity. Areas with underlying drainage systems were also found to have with higher than expected curve numbers, as water quickly permeated the soil then entered the drainage system, acting as runoff. Adjustments were made to the curve number to match the model-predicted hydrographs to the data measured in the field. Figure 1 shows modeled-predicted versus field measured volumes.

Figure 1. Model predicted versus field measured storm runoff volumes.

Similar to the process used in runoff flow calibration, data collected during the field sampling events was used to modify the EMC values for various land uses. Literature values were used as a basis for the modeling and were found to be reasonable; few changes were made to the literature based EMC values for most land uses. Additional nutrients were attributed to fertilization practices on Managed Lawns, with available records of fertilizer application showing the assumed values to be reasonable. The calibrated EMC values are listed in Table 1.

Table 1. Event mean concentration values (mg/L) for selected pollutants in Central Park.

Findings

While this study was conducted in an unusual setting, being both a micro-scale environment and a mixture between open space and highly trafficked urban environment, it was found that traditional modeling methods were effective in simulating the conditions of the Park. The SCS method, using curve numbers and a time of concentration, was able to match both the volume and peaks of the measured hydrographs with reasonable accuracy. EMC values were also found to reasonably reflect the data measured in the field.

Use of these methods is conditional, however, on having a high degree of knowledge of the Park environment and of Park management practices. Both had significant implications during modeling. The drainage in the Park played a significant role in determining the watersheds in the Park and field verification was key in confirming the drainage system. Figure 2 shows the evolution of the watersheds and Table 2 shows the drainage area of the waterbodies. This area does not include the surface area of the waterbody, nor does it include areas that drain into an upstream waterbody. While the Pool-Loch-Meer system is a flow through system, the watershed area listed for the Loch does not include the Pool or the drainage area of the Pool.

Figure 1. Changes in Central Park watersheds from drainage mapping and field verification.

Table 2. Original and updated watershed areas.

Discussion and Recommendations

Watershed modeling is, by nature, a spatial venture. This makes GIS an ideal tool to aid in the modeling process. Prior to the advent of GIS, determining watershed boundaries and curve numbers was a tedious manual task. GIS overlay analysis capabilities can also aid in developing watershed parameters such as curve numbers and EMCs. The GIS tools developed specifically for watershed modeling aid in delineating watersheds, developing model parameters, and creating model input files.

While there are many available tools that develop model input, the functions are constrained by their inability to account for existing infrastructure. Pipe systems can significantly alter drainage patterns and existing watershed tools do not account for the changes the drainage pattern or in times of concentration. Separate tools exist for creating SWMM input files, but they require accurately delineated watersheds and may require previously developed input.

A comprehensive watershed modeling tool would delineate watersheds, account for infrastructure, develop model input, and create model input files. And though the tools exist for most of these steps, accounting for infrastructure is presently the gap between the available tools. Manual adjustment needs to be made and can be extremely time consuming. In urban environments, drainage is often dominated by the infrastructure and may be delineated with topography as a secondary factor. However, in some cases such as the urban park environment, topography cannot be considered secondary.

While the urban park environment may not typically be of concern when considering watershed model applicability, the need to account for infrastructure in watershed delineation is not only a microscale problem. The New York City water supply watershed is an example where the need is apparent. The City's raw water supply consists of 19 reservoirs covering nearly 2000 mi2 (NYCDEP 1999). The reservoirs are interconnected through a series of aqueducts that convey the water from the reservoirs, across the Hudson, and to New York City for distribution. This system, more than three orders of magnitude larger than Central Park, also can not be accurately represented with the available tools due to their inability to incorporate drainage systems and is further reinforcement of the idea that infrastructure cannot be ignored in watershed modeling.

References used

Akan AO. 1993. Urban Stormwater Hydrology - A Guide to Engineering Calculations. Lancaster, PA.: Technomic Publishing Company, Inc. 268 p.

Benaman J. 1996. Modeling of Dissolved Oxygen in the Houston Ship Channel using WASP5 and Geographic Information Systems [thesis]. Austin, TX: University of Texas at Austin. 285 p.

Chow VT, Maidment DR, Mays LW. 1988. Applied hydrology. New York: McGraw-Hill, Inc. 572 p.

James W, Huber WC, Dickinson RE, James WRC. 1999. Water systems models hydrology: users guide to SWMM4 runoff and supporting modules. Ontario, CA: Computational Hydraulics International, Inc. 313 p.

Loucks DP, Stedinger JR, Haith DA. 1981. Water resource systems planning and analysis. New Jersey: Prentice-Hall, Inc. 559 p.

[NYCDEP] New York City Department of Environmental Protection. 1999 December 16. NYCDEP Celebrating New York City's Clean Drinking Water. . Accessed 2000 June 27.

Olivera F and Maidment DR. 1999 July 30. GIS Tools for HMS Modeling Support. . Accessed 2000 June 27.

Rogers EB. 1987. Rebuilding Central Park: A Management and Restoration Plan. Cambridge, MA: MIT Press. 160 p.

Schueler TR and Claytor RA. 1996. Design of Stormwater Filtering Systems. Silver Springs, MD: Center for Watershed Protection.

Singleton AD. 1998. Impacts of Site Use on Soil Quality and Runoff Potential in a Degraded Watershed in Central Park, New York City [thesis]. Ithaca, NY: Cornell University. 89 p.

[UWRRC] Urban Water Resources Research Council, [ASCE] American Society of Civil Engineers and [WEF] Water Environment Foundation. 1992. Design and construction of urban stormwater management systems. New York: ASCE No. 77 and WEF FD-20. 724 p.

[USDA] US Department of Agriculture, Soil Conservation Service. 1986. Urban Hydrology for Small Watersheds. Washington: US Department of Agriculture. Technical Release 55. Available from: NTIS, Sprigfield, VA; PB87-101580.

[USEPA] US Environmental Protection Agency, Water Planning Division. 1983. Results of the National Urban Runoff Program. Washington: US Environmental Protection Agency. Volume I, Final Report. Available from: NTIS, Springfield, VA; PB84-185552.

Authors

Mark R. Lenz, P.E., Senior Project Engineer
Malcolm Pirnie, Inc., 75-20 Astoria Boulevard Suite 350, Jackson Heights, New York, USA 11370.
Email: mlenz@pirnie.com, Tel: +1-718-446-0116, Fax: +1-718-446-4020.

Cindy F.H. How, Engineer
Malcolm Pirnie, Inc., 104 Corporate Park Drive, Box 751, White Plains, New York, USA 10602.
Email: chow@pirnie.com, Tel: +1-914-641-2887, Fax: +1-914-641-2730.