Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 315


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 314
APPENDIX C USING GEOGRAPHIC INFORMATION SYSTEMS TO EVALUATE ENVIRONMENTAL JUSTICE INTRODUCTION This appendix provides an overview of how geographic information systems (GIS) can be used to evaluate environmental justice as it relates to proposed transportation system changes. The appendix discusses in more detail a number of GIS-related topics introduced in the guidebook chapters. In addition, this appendix briefly describes certain GIS-based methods for evaluating environmental justice that have not been discussed elsewhere in the guidebook. Two recent NCHRP research reports have discussed GIS and its use in environmental justice and social economic impact assessment (Forkenbrock and Weisbrod 2001; Cambridge Systematics 2002). Much of the content in this appendix was borrowed from the above referenced research reports. They are valuable information sources if you desire more information on these topics. BACKGROUND GIS allows you to analyze and present the spatial nature of predicted social and economic effects to protected populations. Using various types and scales of maps, it is possible to compare effects between one location and another to determine how various population groups may be differentially affected by a proposed transportation system change. Thus, GIS can be used to display the patterns of distributive effects at varying scales. In this appendix, we assume that you have a least a basic working knowledge of GIS. Our intent is not to teach GIS basics, but rather to discuss the particular data, techniques, and software issues that may arise during GIS-based analysis of the social, economic, and environmental effects of a transportation project. GIS products are designed to combine and analyze layers of information about a place or location. Most, if not all, of the variables considered in an environmental justice assessment have a spatial component. Because of this, GIS is extremely well suited for analysis of the distribution of benefits and burdens. GIS technologies are now being used throughout the world by a diverse group of technicians from all different disciplines. Historically, demographers and planners have used GIS as a tool to store, manipulate, and display data. However, GIS can be implemented as a modeling, decision making, and general spatial statistical analysis tool as well. GIS is unique among computer-based analysis tools for several reasons: 1) GIS allows geo-referenced variables and data from diverse sources and scales to be overlaid and viewed so a more complete picture can be developed of a geographic area. 2) GIS allows aggregation and disaggregation of data to different scales so analysis can be done at an appropriate scale or at multiple scales allowing more robust results. 3) GIS facilities mapping and visualization of information. 323

OCR for page 314
Most commonly, GIS is used for querying spatial databases to find locations that fit criteria for mapping demographics, displaying trends or historical data, displaying assets like transportation infrastructure, and visualizing areas and points of capital investment; as well as processing points, polygons, and lines to find numeric descriptors of data that can then be used in spatial- statistical analysis. Spatial statistical analysis is the description of patterns in space with mathematics and statistics. This type of analysis allows geographic patterns and trends to be numerically described. Geographic and demographic data The U.S. Census Bureau has a Web site with numerous demographic data available for downloading. Other basic data needed for GIS analysis can be accessed or developed from existing public records, such as tax and real estate databases. Table C-1 lists the data and possible data sources commonly used when mapping transportation-related effects. Not every analysis requires all the types of data listed, but the table provides a sense of where to search for location-specific data. Table C-1. Data and data sources for GIS analysis Data type Source Demographic U.S. Census Bureau, local planning departments (for updates and detailed forecasts) Topographic U.S. Geological Survey (USGS), metropolitan planning organizations, state departments of natural resources, local planning departments Street network TIGER/Line Census files (available from the U.S. Census Bureau) local planning/engineering departments, commercial GIS data vendors Land use Local planning departments, city public works departments Accessibility points of interest (local Local planning departments, neighborhood landmarks/activity centers) organizations, geocoded addresses Activity centers (major employers, schools, Geocoded addresses, local or regional economic houses or worship, shopping, and public development or planning departments services) Source: Forkenbrock and Weisbrod 2001. Topographic data usually must be collected on a local-to-regional level. The United States Environmental Protection Agency (U.S. EPA) mandates reporting on some environmental factors; thus, the U.S. EPA may have relevant data available. Otherwise, good sources for obtaining environmental features include the United States Geological Survey (USGS), state departments of natural resources, and metropolitan planning organizations. Data on road networks are usually available from the U.S. Census Bureau in the form of Topologically Integrated Encoding and Referencing (TIGER/) Line files. If greater detail or accuracy is 324

OCR for page 314
desired, local planning and engineering departments and state departments of transportation may have files depicting road networks, or such files usually can be purchased from commercial GIS data vendors. Data issues Conversion. Once data for measuring the effects of a transportation project have been obtained, several issues may arise. Data are available in different formats for use with GIS software packages, including several different types of coordinate systems and projects. Projections relate to how spherical data from the surface of the earth are transformed to the flat surface of a map. Some distortion is inevitable. Fortunately, most GIS software packages now include data and file translators that automate the conversion of data to the projection currently being used for the impact analysis. It still is important to remain aware of the projection used for each data file. Although conversions from one projection to another are not generally likely to be a problem, travel demand or other impact models may only be able to use one specific projection. Another data incompatibility issue arises when data have been created using different software packages than the one being used for the impact analysis. Because most GIS software is capable of converting incompatible file types to data that are recognizable by the software, this form of incompatibility is rarely a serious problem. Privacy and data suppression. Obtaining detailed household data raises privacy issues. The U.S. Census Bureau publishes income and other sensitive data (such as welfare status) only at the block-group levelnot at the census-block level that is desirable for many project impact analyses. Typically, there are about 30 blocks in a block group. Often, however, local city or regional city planning departments may have done their own estimates or personal income and poverty at the block level; those data can be useful for transportation-related analyses. If no other estimates exist, it is possible to estimate block levels from data at higher levels of aggregation. Example. Being careful to use explanatory variables available at both the block and the block group, you can fit a regression model that predicts the percentage of persons living in poverty at the block level (see Forkenbrock and Schweitzer 1999). Once the coefficients of the regression have been estimated at the block-group level, it is possible to apply these coefficients to explanatory variables at the block level to predict the number of persons living in poverty at the block level. Forkenbrock and Schweitzer (1999) built such a model for a metropolitan area using three variables (median home value, percent of homes that are owner-occupied, and percent of population over 65 years of age) at the block-group level to predict the percentage of persons at the block level. 325

OCR for page 314
Software requirements Several GIS software packages are suitable for analyzing the distributive effects of transportation projects. Choosing which specific software package to use can be difficult, however, because of the wide variety of functions and tools available in the different packages. It is necessary to assess which tools and techniques will be needed to complete the analysis as well as what data type and format is being used. Almost any analysis of social and economic effects will require basic GIS tools such as the following: Database query capability, Geocoding and address matching, and Data projection and translation. Very few GIS packages consist of only these tools; most contain a comprehensive array of capabilities for spatial analysis. For more complicated functionssuch as barrier or buffer analysisa GIS software package needs additional tools. Most GIS packages include scaled-down tools for statistical analysis, but often it is necessary to use a more powerful, separate statistical analysis package. When this is the case, data must be exported out of the GIS software and into the statistical package. The tabular data in the GIS software must be compatible with the data format that the statistical package uses and vice versa. Usually, this presents only minor problems. The most complex analyses of the environmental justice-related effects of transportation projects require GIS tools that are often not included with basic GIS software packages. These tools normally are available in groupings with other complex analysis tools as extensions or add-ons to the software. Three examples of complex tools that may only be available in this way are grid processing, irregular polygon information aggregation, and triangulated irregular network (TIN) creation and analysis. TIN creation and analysis is important for impact analysis because of its ability to depict areas of equal effects at various distances from a roadway (e.g., to generate noise level contours from sample sound receptor locations). Irregular polygon information aggregation allows identification and counting of households and their associated demographic characteristics within selected contours (see Chapter 2, Method 3). Grid processing and analysis allows you to perform surface analysis and certain forms of spatial disaggregation (see Chapter 2, Method 6, and Chapter 3, Method 3). Common types of GIS-based analyses Many different types of GIS-based analysis can be used to evaluate environmental justice. Most require the use of network analysis for transportation impacts. Network analysis can include travel demand analysis and traffic simulation studies. Such analyses can be extremely complex and may require the use of large data sets and powerful computers to predict effects on a road network. 326

OCR for page 314
Estimating the social and economic effects of transportation projects may require the use of any or all of the following analysis techniques, which are briefly discussed in turn: Mapping and visualization, Buffer analysis, Barrier analysis, and Overlay analysis. Mapping and visualization. Maps almost always are an important first step in an analysis. Sometimes they reveal patterns that would not be obvious in numeric or statistical charts. They allow hypotheses to be made and then provide visual backup of any results and statistical testing that is done on the hypotheses. Much can be learned about a problem or situation simply by viewing the variable(s) over space. Simple choropleth, or graduated color, maps of Census 2000 demographic characteristics by block group compared to Census 1990 demographic characteristics can be an important initial analysis tool, for example, for understanding how population characteristics of a particular area have changed (see Chapter 2, Method 7). It is easy for the human eye to pick up differences in spatial patterns when they are displayed on a map. This method is a good way of demonstrating how environmental justice status can change over time. Transportation networks and infrastructure may remain relatively unchanged through time, while settlement patterns of various demographic groups may change considerably. Slightly more complex maps can be developed that combine different types or dimensions of information into a single visual display. The environmental justice index (EJI) described in Chapter 2 is an example. For the Atlanta Benefits and Burdens project (Cambridge Systematics 2002), information from the Census Transportation Planning Package (CTPP) journey-to-work data were extracted, formatted for use in GIS, and processed to find the most common journey- to-work flows in the study area. The largest flows were displayed on maps as lines with graduated line widths to represent the magnitude of commuter flows. In addition to the line maps, a pie chart was displayed on the origin district of each major flow depicting the distribution of mode split for the journey to work. These maps thus communicate three important dimensions of travel patterns--origin/destination, volume, and means of travel--in a single visual display. Buffer analysis. A buffer is an area of specified width that surrounds one or more map features. Buffer analysis is used when examining areas affected by activities or events that take place at or near these map features (Caliper Corp. 1996). It is most often used in environmental justice assessment as a screening tool to determine if effects actually would exist in the predicted impact area before proceeding with a more in-depth analysis. To perform buffer analysis, it is necessary to know the specific width (i.e., distance) from a map feature within which an effect may occur. Most GIS software packages include an analysis tool dedicated to creating buffers. Performing the analysis simply involves selecting the map feature to be buffered and then selecting the buffer tool. GIS packages include dialogue boxes designed to guide users through the buffering process. The program will ask for a specified distance at which to buffer the map feature or features. Many software packages offer different options for buffering, such as creating buffers 327

OCR for page 314
of different specified sizes, creating evenly spaced buffers, or creating buffers of variable sizes using a database field as a reference. Carrying out a buffer analysis typically involves four steps: 1. Select the map feature(s) to be buffered. Map features may include specific locations (i.e., points), road network segments (i.e., lines), or established areas and districts (i.e., polygons). 2. Determine the distance(s) necessary to buffer the selected map feature. This distance should reflect the expected spatial extent of the social or economic effect under consideration (e.g., the area within a quarter-mile of bus stops, as an indication of transit service availability). 3. Using a GIS buffer tool, create the buffer and overlay it on appropriate demographic or economic data, generally displayed at the census-block level. 4. Observe the resultant map and determine whether potential social or economic issues exist. If potential issues are observed, then proceed further with a more in-depth analysis. When evaluating a transportation project, it is often important to be able to summarize what will be impacted within a certain distance of the project. For example, as part of the Boston area's Silverline project, the Metropolitan Boston Transit Authority (MBTA) project team analyzed the economic level of the residential population and the number of jobs within half-mile radius buffers around the proposed Silverline boarding points. First, the locations of the boarding points were mapped as a layer in the GIS. Then, half-mile radius buffers were constructed around each boarding point. Next, the demographic data were collected at the smallest scale possible and joined to geographic boundary layers in the GIS. Specifically, an extract of the Census of Population and Housing 1990 Summary Tape File 3A (see Appendix D) was used that included variables about poverty and household income at the block-group level and data on number of workers at the Transportation Analysis Zone (TAZ) level that was created by Boston's Central Transportation Planning Staff (CTPS). Since neither TAZ boundaries nor block group boundaries nest neatly in the circular buffer rings, a script was applied to each boundary that returned what proportion of the boundary fell within the buffers. Then, the proportions were applied to the demographic data and summed for each buffer. The results could be reported to the MBTA and FTA as approximate numbers of low-income people living or working within a half-mile of the boarding points. Barrier analysis. Barrier analysis involves the creation of a barrier such as a road construction zone or a road that prohibits nonmotorized travel across it. The analysis estimates the change in level of access that has occurred due to the creation of the barrier. A GIS-based analysis can provide useful insights into changes in the accessibility of important destinations. To assess the relative change in access to common and important destinations on the part of protected versus other populations, four general steps can be followed: 1. Determine the general locations of households using a GIS-based geocoding function. Census-block data are the most geographically specific data available, so the coordinates of the block's centroid may be used as a proxy for the locations of households in the block. It is from these centroids that distances and travel times are computed. 2. Geocode the locations of important destinations. Locational data are readily available for businesses, agencies, and most households in the United States. Analyses related to 328

OCR for page 314
schools, houses of worship, shopping sites, human service agencies, and major employment centers are likely to be among the most common. 3. Compute the shortest paths between origins and destinations. The shortest path is that which minimizes the distance between two locations over a street network. The network capabilities of GIS and related software enables shortest paths to be computed from all block centroids. 4. Estimate the changes in access. The analysis can be carried out using the existing transportation system, both before and after the barrier is in place on the network. Results can be expressed in terms of units distance, time en route, or number of persons crossing a street corridor. Overlay analysis. Overlay analysis involves the integration of several discreet data layers. Analytic operations in estimating most types of effects require two or more data layers to be joined. Overlays, or spatial joins, can integrate spatial data on concentrations of different population groups with the incidence of one or more types of effect. To perform an overlay analysis, it is necessary to have data layers already created. For example, to estimate the number of persons who would be affected by noise pollution resulting from a proposed transportation project, layers containing data on (1) population characteristics, and (2) the estimated air or noise pollution extent (represented by contours) must already exist. In most GIS packages, it is possible to choose the overlay tool and then follow the instructions in the dialogue boxes for inputting the desired layers. Generally, four steps are required: 1. Using transportation noise modeling software (such as the Minnesota Department of Transportation's MNNOISE model; see Minnesota Department of Transportation [Mn/DOT]1991), generate noise levels and point distances from the transportation project. Distances can be specified by the user with an x,y coordinate plane and standard units of measure (e.g., feet or meters), or distances can be calculated using geocoded locations. 2. Create TIN structures by triangulating the values between points using extrapolation. Equal value noise contours will be created with this process. This can be done using the TIN creation and analysis tool within the GIS software. 3. Overlay the noise contours on the street network or transportation project area and demographic data layer. 4. Use a GIS spatial analysis feature to count the number or calculate the percentage of persons within the noise contour considered likely to experience an effect. Spatial linear models. Simple spatial econometric and regression models can be used to test for environmental justice. For example, a regression model could be developed to relate the amount of transportation benefit from a project (expressed as a change in accessibility or travel time savings) to the percentage of minority population in a census tract, TAZ, or other spatial unit of analysis. A model could be constructed to predict transportation access as a function of different demographic characteristics. The error in the model also can be examined geographically. Every linear model is expected to have error, but in a good spatial regression model, that error would be evenly distributed through space. If the error from the model results in a spatial pattern, there is likely some additional spatial variable that explains the variation in the dependent variable. 329

OCR for page 314
GIS is not required to construct a spatial linear model, but it can facilitate the analysis. GIS can be used to aggregate data to the scale at which the model is operating, to map the results of the model, to map the error or residuals of the model, and to group or cluster the geographic entities that behave similarly in the model. For example, a spatial linear model was built in Atlanta to examine the geographic distribution of vehicle ownership. It was found that several variables--percent black, percent low-income, and average household size--have a statistically significant explanatory relationship with the dependent variable, average vehicles per household. Results showed that with the other variables (income and average household size) held constant, as the percent of the population that is black increases, vehicles per household decreases. Similarly, as percent low- income population increases, vehicles per household decreases, holding percent black constant. Conversely, as average household size increases so too does vehicles per household. This analysis was done at the census-block-group level but could be done at any scale. Spatial autocorrelation tests for phenomena. Spatial phenomena can be tested for spatial autocorrelation. Spatial autocorrelation means that like values are clustered geographically. Typically, if spatial autocorrelation exists, inequity also exists. For example, if transit stops are spatially autocorrelated, that means that they are clustered and that the areas where there is not a cluster of transit stops are not receiving the same level of transit service. Tests to see if positive outcomes of the transportation plan or project are spatially autocorrelated can be done in a GIS setting to help determine the proportionality in the distribution of benefits and burdens. Spatial indices - index of dissimilarity. For comparability purposes, it is important to be able to describe patterns or distributions with numbers. The field of spatial statistics seeks to do just this. There are many simple statistics that can be calculated to describe the location, centrality, and dispersion of a spatially distributed variable. The most simple of these is probably a population weighted centroid. Population weighted centroids were computed to describe how the distribution of the black population in Atlanta has changed since 1980. As expected, the population weighted centroid of the black population has been moving away from the central city since 1980. Another set of statistics is used to describe the dispersion of a variable. The nearest neighbor statistic can be used to calculate how clustered or spread out a population is. The numeric nearest neighbor statistic allows description of a pattern that is clustered, random, or regularly spaced. This statistic is useful when calculated for many sets of data across time so the results can show historical trends. There also are many statistical indices that measure or compare the spatial patterns between variables. One that is very useful in environmental justice is the index of dissimilarity. This index is also sometimes referred to as the segregation index. This index measures the degree to which two spatial variables are distributed differently within a specified area. The index of dissimilarity can be used to calculate whether the spatial distribution of nonminorities and the spatial distribution of minorities are similar or dissimilar, and thus help to assess the degree to which transportation needs are being met by existing and proposed services. Data on the number of nonminorities and the number of minorities would need to be gathered at a geographic scale smaller than a county (preferably a scale that allows a significant number of entities within the county so statistical significance can be determined). 330

OCR for page 314
For example, if there are at least 20 TAZs in a county, the TAZ level would be an appropriate scale at which to collect data on number of nonminority populations and number of minority residential populations. This index was calculated for each county in Georgia. It was known that the percentage of population that is black in one particular county increased substantially between 1990 and 2000. However, that does not indicate that at a smaller scale, the population within the county necessarily is becoming less segregated. The index of dissimilarity compared the percentage of the black population in each block group to the percentage of the white population in each block group. The results of the analysis showed that, although a larger percentage of the county's population is black, the county is not more integrated. In fact, segregation at the block-group scale has increased since 1990. This kind of analysis should be calculated at different scales. For example, data could be collected at the county level and the index could be calculated for the state, or data could be collected at the census-block level and calculated for the TAZ. GIS could be used to automate the calculation of indices like these at multiple scales. The index of dissimilarity formula is applied as follows: I = 100(0.5 (x-y)) where I = index of dissimilarity x = the percentage of the county's non-minority population in a TAZ y = the percentage of the county's minority population in a TAZ Sum for all TAZs, divide by two, and multiply by 100. The index can vary from 0 to 100. An index of zero reflects a perfect similarity between the distributions of minorities and the distribution of nonminorities. Conversely, an index approaching 100 reflects a large dissimilarity between the two populations and means that within the county, the minority population is clustered (segregated) and not spread evenly across space. Spatial disaggregation and population surfaces. GIS-based spatial disaggregation techniques can be used to disaggregate population and impact data in order to better estimate the distribution of impacts by population group. For example, a GIS raster module can be used to disaggregate zone-based population data and network-based impact data to grid cells. This allows impacts calculated for different types of spatial units to be more precisely overlaid on population data. For example, emissions from a transportation network can be assigned to the grid cells corresponding to the network, and then overlaid with population data that is assigned from census tracts to the corresponding grid cells. This approach was demonstrated in the System for Planning and Research in Towns and Cities for Urban Sustainability (SPARTACUS) project undertaken in three European cities: Helsinki, Naples, and Bilbao (Commission of the European Communities 1998). Raster-based data disaggregation also has been applied in the Salt Lake City, Utah, metropolitan area, although impacts were not analyzed by socioeconomic group. SPARTACUS is based on an underlying "engine" that combines the integrated transportation-land use model, MEPLAN, with a GIS 331

OCR for page 314
raster module to calculate and display 100-meter grid cell micro-scale indicators. For example, emissions and noise dispersion models were used within SPARTACUS to calculate the impacts resulting from the transportation network. In the Helsinki analysis, it was found that about 29 percent of the metropolitan population would feel disturbed by traffic noise in the peak hour under the baseline transportation scenario. This percentage does not vary significantly by population group. While the SPARTACUS project admittedly was a large-scale modeling project with substantial data and resource requirements, it does indicate that large amounts of data can be aggregated down to a small number of indicator values. The project illustrates approaches for quantifying the equity and social justice implications of alternative scenarios. While MEPLAN, in particular, may require some data that may not be readily available in all U.S. urban areas, the GIS-based analysis of emission and noise exposure can be applied independently of the land use model (see Chapter 3, Method 3). RESOURCES 1) Environmental System Research Institute (ESRI). 1999. GIS for Everyone. Redlands, CA: ESRI Press. This book is a basic beginner's guide to GIS. It includes detailed GIS data, a full working version of GIS software, and tutorial exercises. No previous experience with GIS is necessary, but experience with computers is very helpful in understanding the tutorials. 2) Mitchell, Andy. 1999. The ESRI Guide to GIS Analysis, Volume 1: Geographic Patterns and Relationships. Redlands, CA: Environmental System Research Institute (ESRI) Press. This book offers a review of basic GIS concepts and provides an easy-to-understand guide to GIS analyses. Many real-world examples are used to illustrate the GIS analyses presented. This is not an introductory text; it assumes some prior knowledge of GIS concepts. 3) DeMers, Michael. 1996. Fundamentals of Geographic Information Systems. New York, NY: John Wiley and Sons. This book is a comprehensive text that presents information on GIS without excessive detail. It covers all basic GIS concepts and most advanced concepts. This text may be too advanced for persons with no GIS background. Internet sites 4) http://www.gis.com/jumpstation/ GIS.com is ESRI's Web site providing beginner-level discussion of GIS. The jump-station is a searchable index that provides links to other GIS Web sites including federal, state, and local government agencies; commercial; noncommercial; and universities. REFERENCES Caliper Corp. 1996. TransCAD 3.0 User's Guide. Newton, MA: Caliper Corporation. 332

OCR for page 314
Cambridge Systematics, Inc. 2002. Technical Methods to Support Analysis of Environmental Justice Issues. Final report of project NCHRP Project 8-36(11). Transportation Research Board, National Research Council. Washington, DC: National Academy Press. Chakraborty, Jayajit, Lisa A. Schweitzer, and David J. Forkenbrock. 1999. "Using GIS to Assess the Environmental Justice Consequences of Transportation System Changes." Transactions in GIS, Vol. 3, No. 3 (June), pp. 239-258. Commission of the European Communities. 1998. SPARTACUS Final Report. Available at http://www.Itcon.fi/spartacus/default.htm. Forkenbrock, David J., and Lisa A. Schweitzer. 1999. "Environmental Justice in Transportation Planning." Journal of the American Planning Association, Vol. 65, No. 1 (Winter), pp. 96- 111. Forkenbrock, David J., and Glen E. Weisbrod. 2001. Guidebook for Assessing the Social and Economic Effects of Transportation Projects. NCHRP Report 456. Transportation Research Board, National Research Council. Washington, DC: National Academy Press. Also available at http://trb.org/trb/publications/nchrp/nchrp_rpt_456-a.pdf. Minnesota Department of Transportation (Mn/DOT). 1991. Noise Analysis: Stop and Go Trajfic Procedures. St. Paul, MN: Mn/DOT Noise Group of Environmental Engineering, Engineering Services Section. 333