Using Simulation to Estimate Vehicle Emissions in Response to Urban Sprawl within Geauga County, Ohio

Urban sprawl oft en leads to rapid expansion and haphazard developments of low density residential land uses that are spatially disjoined. Populations occupying these new developments are expected to contribute to increased traffi c volumes and vehicle emissions through increased home-work journeys. Computer simulation is one of few feasible approaches to model projected trends of local communities to understand how they evolve and better plan their future courses. Th e VERTUS model was developed as a planning tool to estimate vehicle emissions in response to urban sprawl. Th e model is specifi c towards estimating vehicle emissions at the local and highway levels during the home-work journey. Th e model was applied to Geauga County, Ohio to estimate how an increase in housing over a 20-year period from 2000-2020 will infl uence vehicle emissions generated. Results indicate that emissions are currently highest in the western part of the county where the greatest number of households is located. Th is geographic distribution remains when emissions are estimated for growth in housing. While additional housing translates to more vehicle emissions, this research found that diff erences exist among the county’s individual municipalities in terms of emissions generated. In several instances, municipalities with a smaller growth in housing generate a greater amount of emissions than a municipality with a larger growth in housing. Th ese diff erences result from variations in the commuting characteristics of each municipality’s residents and provide insight into how household travel patterns relate to vehicle emissions. OHIO J SCI 109 (3): 52-66, 2009 1Address correspondence to Timothy J. Dolney, Assistant Professor of Earth Sciences, 215 Hawthorn Building, 3000 Ivyside Park, Penn State University -Altoona College, Altoona, PA 16601. Email: [email protected]. INTRODUCTION Computer simulation has become an eff ective tool for urban planners to manage urban growth at the local community level. Th is is signifi cant as the landscape in the United States has come under considerable alteration through the rapid expansion of housing developments into rural and suburban areas along city edges (Ewing, 1994; Sutton, 2003; Wolman et al., 2005). Th roughout the years, this movement has grown into the phenomenon termed urban sprawl. Urban sprawl is typically characterized as an undesirable form of development. Open green space, critical nature areas and, in many cases, prime farm lands are being overrun by housing and pavement for additional roads. Many are located in areas that are the least accessible from the built-up urbanized areas. Th is increases geographic separation over space between the location of housing developments and work locations. Th e population therefore travels greater distances to work, shopping and recreation, among others, to overcome this spatial division. Longer travel distances translate to increasing vehicle emissions that negatively impact the environment. Th e spatial dimension of “dirty-air” primarily being an urban problem may now extend to suburban and rural areas. Growth-induced air pollution is an important and critical issue. However, the extent of the problem is relatively unknown as household travel patterns have been discussed and related to land-use patterns in urban growth literature, but have not been extended to how they aff ect vehicle emissions (Frank et al., 2000). As a result, the Vehicle Emissions Related to Urban Sprawl (VERTUS) model was developed for estimating vehicle emissions in response to urban growth. Th e model addresses the shortcomings of urban growth models that solely focus on land use eff ects, Urban Growth Simulator (UGS) (Lee, 2003), WhatIf ? (Klosterman, 2001), and UrbanSim (Waddell, 2002); and models that estimate emissions for a given area but not resulting from urban growth, MOBILE (US Environmental Protection Agency [EPA], 2003), Mobile Emission Assessment System for Urban and Regional Evaluation (MOBILE) (Bachman et al., 2000), and ONROAD (Yu, 1998). Given a level of urban growth, VERTUS estimates the amount of vehicle emissions generated as the population travels from new housing developments to their place of work. Emphasis is placed solely on home-work journeys as most of the population participates in this trip purpose on a daily basis. VERTUS’ design was implemented within the UGS to off er users a simulation tool that quantifi es the environmental impact on land use and air quality resulting from urban sprawl. Th is paper fi rst provides a brief overview of the design and functionality of VERTUS and its incorporation into UGS as a planning tool for estimating vehicle emissions. Second, it demonstrates the applicability of the model and its methodological design for this purpose through a case study of residential growth in Geauga County, Ohio at the municipality (township and village) level. Results from this analysis provide insight into the relationship between urban growth, household travel patterns, and vehicle emissions. MATERIALS AND METHODS VERTUS’ design is based on the premise that given locations of new housing, the amount of vehicle emissions generated during the population’s home-work journey can be estimated. Emphasis is solely placed on home-work journeys as most of the population participates in this trip purpose on a daily basis. VERTUS is standalone simulation engine that can run independent of other urban growth models. For the purpose of this research, it was coupled with UGS to link together the components of urban sprawl, trip generation, and vehicle emissions to estimate emissions at the municipality level. Urban Sprawl Model UGS was developed as an impact assessment tool for urban sprawl within a 15-county region in northeast Ohio but can be applied nationally at several geographic levels: township, village, city, county, or community level (Lee et al., 2002). With the simulator, users enter a projected amount of residential, commercial, or industrial developments with the average lot size in acres. Users can further defi ne locations of new developments as either along OHIO JOURNAL OF SCIENCE 53 T.J. DOLNEY a road frontage or away from the road in clusters. Users also have the option to incorporate any or all growth management strategies: open space management, avoiding development on environmentally critical areas, avoiding development of farmlands, and limiting developments in a pre-set growth boundary. With user inputs defi ned, the simulator develops cells until the projected growth for each community is accommodated. When simulation is complete, UGS displays a map showing locations of newly developed areas (Fig. 1). Users can also view statistics indicating how much agricultural and critical nature area is lost and the amount of nutrients loaded into the soil resulting from additional development. It should be noted that the algorithms within UGS are such that the simulated housing locations will vary with each simulation even if the housing inputs are the same for each. However, results are consistent as the simulator places new developments in areas with current or proposed waterand sewer-service. Users can also export locations of simulated housing development as a polygon shapefi le, the Geographic Information System (GIS) data format readable by most GIS soft ware packages. Th is capability is essentially the fi rst step of VERTUS where users export the locations of simulated housing developments to serve as the initial starting points for home-work journeys. Trip Generation Model Th e trip generation model establishes the variables required to model home-to-work journey travel, including: number of workers number of non-commuters number of home-work journeys locations of employment centers percentage of workers that travel to each employment center Th e US Census 2000 Transportation Planning Package (CTPP) data-set were utilized to model these components. CTPP are special tabulations from the decennial census designed for transportation planners that contains information by place of residence, place of work, and fl ows between home and work. It is the only Census product that summarizes data by place of work and provides information on the travel fl ow between home and work. Because the data are based on the decennial census, the data are reliable and accurately refl ect the characteristics of the surveyed population. Additionally, CTPP data have been used in other studies (Wang, 2000; Boyce and Bar-Gera, 2003; Cho et al., 2001; Gottlieb and Lentnek, 2001). With these data, algorithms were developed that calculate home-work journey variables depending on the municipality being simulated. Estimating the number of workers for newly developed areas is a function of the number of housing units input into UGS. VERTUS multiplies the number of new housing units and the number of workers per household for that particular municipality to estimate the number workers. Th e model next calculates how many of these workers participate in the Figure 1. UGS interface representing Parkman Township, Geauga County, Ohio -Locations of new housing are circled. 54 VOL. 109 ESTIMATING VEHICLE EMISSIONS home-work journey as some are non-commuters; work at home, walk or ride their bike, and carpool to work (CTPP categories). CTPP data also provide the commuting patterns of home-work journeys; where home-work journeys travel for work and what percentage travels to each work location. Last, to model homework journey travel, locations of employment were established for each county of work at the Traffi c Analysis Zone (TAZ) level. Using a methodology developed by Giuliano and Small (1991 and 1999), employment density and total employment threshold values were determined for the region of study. TAZs within each county that met the threshold values are considered employment centers. In using this methodology, a single TAZ or multiple TAZs may be identifi ed as prospective employment centers. However, to streamline the modeling process and reduce user input, VERTUS establishes one employment center per county by calculating the geographic center where multiple exist. Vehicle Emissions Model In estimating vehicle emissions emitted during the home-work journey, VERTUS partitions a number of variables to better represent reality. First, estimates are provided at two geographic scales; local and highway. Th ese represent the assumption that commuters start their journey from their home, travel through local streets, and eventually gain access to a highway for faster travel to work. Th us, local emissions are those generated during the commute from home to highway access points (HAPs) with highway emissions representing those from the start of the highway to places of work. Th is diff erentiation allows users to see emissions generated at the local and regional level across multiple counties. VERTUS has a local (Fig. 2) and highway (Fig. 3) interface to model these emissions. Second, emission rates are calculated for two broad categories of vehicles fueled by gasoline: passenger cars (PCs) and light-duty vehicles (LDVs). LDVs consist of trucks, vans, and sport utility vehicles (SUVs). Categorizing the vehicle fl eet is signifi cant as diff erent classifi cations of vehicles emit diff erent amounts of emissions. Typically, LDVs emit more than PCs as they have less restrictive emissions, lower fuel effi ciency, and bigger engines (Davis and Truett, 2000). Algorithms within the model calculate the number of PCs and LDVs according to vehicle registration, income, and number of persons per household. Research has shown that larger and higher income households are more likely to purchase LDVs than PCs (Zhao and Kockelman, 2000; Niemeier et al., 2001). Based on these fi ndings, the model allocates a greater percentage of LDVs to municipalities with larger and higher income households. Model output is provided for the following fi ve vehicle emissions: hydrocarbons (HC), nitrogen oxides (NOx), carbon monoxide (CO), particulate matter 10 (PM10), and carbon dioxide (CO2). Figure 2. VERTUS local emissions interface representing Parkman Township, Geauga County, Ohio. Lines represent the street network with points representing housing locations. Th e circled point is a highway access point. OHIO JOURNAL OF SCIENCE 55 T.J. DOLNEY Th e fi nal output of the model expresses the amount of emissions generated as tons per year. Estimates are based on the distance traveled from new housing developments to the HAP (local emissions) and from the HAP to employment centers (highway emissions). Travel distance is calculated as the EPA and other organizations do not have emissions as a rate per time basis. Travel simulation and all computational components were integrated using Microsoft Visual Basic (VB) programming language and ESRI’s (Environmental Systems Research Institute, Redlands, California) MapObjects and NetEngine. Output can be saved in a text fi le format for further analysis (Fig. 4). Case Study VERTUS, in conjunction with UGS, were applied to Geauga County, Ohio, (Fig. 5) as a case study to estimate how projected growth in housing would impact the spatial distribution of vehicle emissions. Th e county serves as a suitable study area because of past and current growth in housing and the anticipated growth expected to occur in the future. Th e county is subdivided into 22 municipalities (townships and villages) and is one of several counties surrounding Cuyahoga County that has absorbed a large portion of suburbanites moving away from the city of Cleveland. Over the 30-year period from 1970 to 2000, Geauga County’s population increased from 60,977 to 90,895, a 49 percent increase. From 2000 to 2030, the county estimates a 23 percent increase in population (Geauga County Planning Commission, 2003). Similar to the county population trends, the number of housing units per square mile is expected to increase over the 30-year period by 43 percent. Also, about 88 percent of the county (230,388 acres) is zoned for residential purposes with the average lot size ranging from 1.5 to fi ve acres (Geauga County Planning Commission, 2003). Using data from the US Census Bureau and the county’s Comprehensive Housing Improvement Strategy (CHIS) plan, the Geauga County Planning Commission estimate an additional 7,226 (22 percent) households from 2000 to 2020. Growth is primarily due to in-migration of residents from surrounding counties. Th e plan directs growth in housing where adequate infrastructure exists to ensure aff ordable single-family housing for residents. Based on current infrastructure, therefore, most development will occur in the western municipalities of the county. Year 2000 housing characteristics (Table 1) serve as the baseline for establishing vehicle emissions. Locations of housing were represented using the UGS residential land use layer. Future emission estimates were then calculated based on housing growth from 2000 to 2020 (Table 2). Projected housing estimates were fi rst simulated in UGS to derive their locations. Th ese locations were then simulated in VERTUS to estimate the additional amount of emissions generated at the local and highway levels. Home-work journey emissions from Geauga County residents can only emanate from the private automobile as no public transportation exists within the county. A HAP was established in each municipality to model local emissions. CTPP data indicates that a total of 11 counties serve as locations of employment for Geauga County residents. One employment center was established in each of these counties. In performing simulation for each municipality, the projected number of additional households were entered into UGS with half cluster and half frontage, each one acre in size; for example Auburn Township, 1,320 additional households from 2000-2020, Figure 3. VERTUS highway emissions interface representing Parkman Township, Geauga County, Ohio. 56 VOL. 109 ESTIMATING VEHICLE EMISSIONS