Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 66


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 65
due to greater crossing distances. Diverting traffic may cause impacts at other intersections that may not be equipped to handle greater traffic volumes. Regional air quality mitigation measures. Regional measures may be considered to address nonattainment of pollutants. Regional air quality mitigation measures may include roadway or transit projects to reduce congestion or inspection and maintenance programs to reduce vehicle emissions. SELECTING AN APPROPRIATE METHOD OF ANALYSIS Most of the air quality and environmental justice assessment methods presented in this chapter examine each of the major categories of vehicle-generated air pollution individually. Although these techniques are the ones most readily combined with standard air quality models and assessment processes, they do not allow practitioners to consider additive or synergistic effects of pollutants or cumulative effects from all sources. One method, however, does allow for evaluation of cumulative pollutant levels from all sources. This method can be implemented as an extension of the most commonly used air quality assessment models. Examples are provided showing how results for estimated pollutant levels can be evaluated independently of current criteria pollutant standards. The methods presented here do not directly address questions about the causal connection between vehicle-generated pollutants and observed health effects such as asthma or increased cancer rates. Understanding this connection between environmental air quality (outdoor and indoor) and health effects is an emerging field of research, although most studies have not focused solely on transportation sources. This type of research is very time consuming and requires considerable expertise in health and epidemiology. Because the cost of health effects- based methods would be beyond the scope of all but a very few, extreme transportation situations, they are not included in this guidebook. See Carlin and Xia (1996), Hockman and Morris (1998), and Waller, Louis, and Carlin (1999) for examples of exposure-based environmental justice research on the health effects of air quality. METHODS Table 3-3 provides a summary of the four methods presented in this chapter. The methods used to analyze air quality impacts from transportation projects range from simple to complex, and from well understood to experimental and not yet prescribed by regulatory authority. Micro-scale analyses for transportation-related projects nearly always focus on CO and possibly PM, whereas regional analyses generally address CO, NO2, PM, and VOC. Method 1. General air quality review When projects do not warrant an air quality analysis, the assessment may extend only to a general discussion about air quality rules and the existing air quality environment. For purposes of environmental justice assessment, concentrations should be evaluated in terms of potential NAAQS violations to identify situations where alteration of the planned policy, program, or 66

OCR for page 65
project must be considered. In addition, the net change in concentrations, independent of NAAQS thresholds, can be determined to identify situations where there would be either beneficial or adverse effects to protected populations. Table 3-3. Summary of methods for analyzing air quality Assessment Appropriate Use Data Expertise Method level uses when needs required 1. General air Screening Project/corridor/system Initial analysis and Low Spreadsheet quality when air quality review effects are expected to be minimal 2. Detailed Detailed Project/corridor Project is Medium Modeling microscale controversial or there (CAL3QHC) analysis are potential environmental justice concerns and more detail of microscale effects is required by local guidance 3. Detailed Detailed Large projects/systems Transportation Medium Modeling regional conformity analysis is (MOBILE6) analysis required and there is potential for environmental justice concern 4. Analysis Detailed Project/corridor/system Air quality is a major High Modeling, using issue with protected database, pollution population groups and geographic surfaces previous methods information have not addressed all systems (GIS), issues statistical analysis When to use. This approach is advised in situations where detailed analysis is not warranted. It is intended to document both local and regional air quality in an area, as well as the presence of protected populations. The level of concern that there could potentially be distributive effects to protected populations is based on review of demographic patterns. Information gathered for this assessment can be used to perform detailed microscale analysis and detailed regional air quality analysis if findings indicate the potential for adverse effects to protected populations. Detailed microscale or regional air quality analyses may not be necessary if the project does not trigger the requirements for a conformity determination or if the project is within an attainment area or a maintenance area with specific guidance that does not require a detailed microscale analysis. A detailed air quality analysis may not be necessary within a nonattainment or maintenance area if a project is not regionally significant and if the project does not affect any transportation control measures. 67

OCR for page 65
Even within maintenance or nonattainment areas, programs may be in place that provide thresholds indicating whether a microscale air quality analysis is needed. An example guidance would require evaluation of the 10 highest volume intersections in a region. The local guidance may state that a microscale air quality analysis is required if the highest volume intersection within the project area is less than the 10th highest intersection volume in the region. Similar guidance may specify that intersections operating at a particular level of service (LOS) or better may not require micro-scale air quality analysis. Analysis. This analysis is a process of documenting regional and local air quality issues, and then documenting the presence of protected populations in the study area to determine (on a qualitative basis) whether the proposed transportation system change could potentially cause distributive effects. Based on local guidance, this analysis may include evaluating the highest traffic volume intersections within the project area, and/or evaluating the worst intersection LOS. Step 1 Document regional air quality. The first step is to document regulations and regional air quality in the study area. Much of the information needed for this step can be obtained from state and local agencies. At a national level, the EPA provides detailed information on regional air quality throughout the United States. Available information includes lists of regions that are in violation of the NAAQS, air pollution data and maps for ozone, PM, NO2, SO2, and air quality index (AQI) reports. The AQI is an index for reporting air quality to the public. The AQI is calculated daily for each criteria pollutant using standard formulas approved by the EPA and is tied to relative levels of acute health concern. For example, an AQI of 101 to 150 is unhealthy for sensitive groups or individuals, such as those with asthma. AQI values greater than 150 are considered unhealthy (151-200), very unhealthy (201-300), or hazardous (301-500) to all persons. The AQI is calculated on the basis of measurements taken at more than a thousand locations across the country each day. In large metropolitan areas, state and local agencies must report the AQI to the public daily. When the AQI is above 100, they must also report which groups, such as children, people with asthma, or people with heart disease, may be sensitive to the specific pollutant. Although it is not required, many smaller communities also report the AQI as a public health service (U.S. EPA 2003a). To complete Step 1, describe the regional pollutants of highest concern in the study area, common sources of those pollutants, and provide an indication of the general level of regional air quality based on state and local information and on information available from the U.S. EPA. The data developed for this step could include the region's attainment/nonattainment status and number of days above the NAAQS for the various pollutants. Step 2 Document local air quality. The second step is to document local air quality rules and potential locations with relatively higher pollutant levels in the study area. Discuss the regulations and guides that are used to evaluate local transportation air quality and transportation conformity. Describe how the attainment status in the area determines the level of air quality assessment that is necessary. Next, identify locations in the study area where air quality from mobile and point sources may be of greater concern. At a minimum, you should document areas where air pollution levels could be higher because of proximity to transportation facilities. These 68

OCR for page 65
locations could be areas near large concentrations of mobile sources, such as freeways, arterials, and congested intersections. If time and resources allow, it would be advisable to also identify other significant sources of air pollution that may affect the project area, such as electrical generation facilities and other point sources. Step 3 Document protected populations in the project area. Use any of a number of qualitative or quantitative methods for identifying protected populations in the project area. Many suitable techniques are discussed in Chapter 2. The best technique to use depends on the size of the project being evaluated. For policies, programs, and large projects with a sizeable area of potential effects, it is appropriate to use census data and any of the evaluation techniques that use census data, such as the Environmental Justice Index (EJI) or threshold analysis. For projects with smaller study areas, it is more appropriate to use data gathering techniques to characterize protected populations. Suggested techniques include local knowledge, field surveys, and public participation-based techniques such as focus groups. Step 4 Evaluate potential for unequal distributive effects. A map is probably the best way to evaluate the distribution of air pollution sources in the vicinity of protected populations. Either desktop methods or GIS can be used to map the location of significant point and mobile sources depending on available resources, the size of the study area, and the number of sources and demographic analysis units. Then, overlay the source information on maps showing the location of protected population groups. Because this is a screening test, the objective is not to identify statistically significant unequal location patterns. Rather, the intent is to identify areas where there is spatial clustering of air pollution sources and where protected populations are in very close proximity to air pollution sources. Document these sites as locations where local air quality and its effects on protected populations could be of greater concern than in other portions of the study area. Commonly used regional air quality assessment methods assume that pollutants at this scale are distributed relatively uniformly across large areas. Therefore, it will not usually be worthwhile to evaluate unequal geographic distributions unless the study area includes multiple regions. Rather, you will need to discuss regional air quality as mentioned above and describe the adverse effects that may be experienced by various population groups. Data needs, assumptions, and limitations. This type of analysis requires the following basic information on the project and the geographic area or region of the project: Project traffic study results, including traffic volumes and LOS; Maps showing transportation system intersections and links; Information on the location of point sources that emit air pollution; Regional attainment status, and local project analysis guidance; and Information on the location of protected populations. Information on point sources that emit air pollution can be found in numerous sources. Many states publish an annual database of permitted facilities. Some of these databases include information on the level of emissions, making it possible to map not only the locations of facilities but also to categorize facilities based on the volume and types of pollutants they emit. 69

OCR for page 65
At the national level, the EPA annually releases this type of information in the Toxics Release Inventory (TRI) (U.S. EPA 2003b). This technique assumes that proximity to air pollution sources is an indication of possibly higher local air pollution levels. This may not be a valid assumption in all situations because of factors such as wind speed and direction, topography, and pollutant dispersion characteristics, among others. Without further analysis, this technique merely documents local and regional air quality issues in the study area and identifies locations where air quality may be of greater concern to protected populations. Using this method, one approach to evaluating the distributive air quality effects of a project would be to assess whether the project would affect local air quality in the areas of greater concern. If so, a more detailed analysis in those areas should be considered. However, if the project would not have an effect in these areas and air quality is not a major issue to protected populations in the study area, it should generally be safe to conclude that the project would have no distributive air quality effects. Obviously, this method has major limitations. It relies on very basic data to assess the potential for localized air quality effects, and it assumes that the distribution of regional air quality is uniform. For areas where the potential for unequal effects is identified, the actual level of effect cannot be quantified. Because of these limitations, it is safe to assume that any results from this method of assessment would be challenged if a project were controversial. This technique is therefore best used as a screening technique only, and further assessment using more robust techniques should be performed unless no potential for unequal effects is found and the project is not controversial. Results and their presentation. The results of this analysis could include a table showing intersection volumes or LOS, where higher volumes indicate a greater potential for localized air quality effects. A second table showing total stack and fugitive emissions from point sources could be used to display sites with high emissions levels. A general discussion of the regional attainment status and guidance for air quality should be included. Probably the most useful presentation aid would be a map showing locations of protected populations and nearby mobile and point sources of air pollution. Assessment. A general air quality review is intended to disclose local and regional air quality concerns and regulations. For environmental justice assessment, comparison of mobile and stationary air quality sources and their proximity to protected populations is included. Due to the many limitations of this approach, further detailed analysis should be performed in any situation where protected populations could be affected by the policy, program, or project or where the transportation system change is controversial. Because no quantitative air quality analysis is conducted, this technique provides qualitative results and any conclusions are subjective. Method 2. Detailed microscale analysis The detailed microscale analysis is an extension of the hot spot analysis used to determine NAAQS conformance. The additional information and modeling needed to perform 70

OCR for page 65
environmental justice assessment are reasonable and should be within the resources of most agencies that must currently perform local air quality analysis under Clean Air Act and transportation conformity guidelines. When to use. A detailed analysis should be conducted for any regionally significant transportation project within a nonattainment or maintenance area. Because the additional data requirements necessary to perform environmental justice assessment are relatively minor, this method can be used to evaluate both projects and plans, such as TIPs or major investment studies. This analysis should also be conducted if a state agency requires it on the basis of its SIP, guidance, or rules. The analysis also is needed if the project is controversial and will be subjected to substantial public scrutiny due to perceived potential air quality impacts. Such an analysis is especially useful in situations where a general air quality review (Method 1) indicates the potential for air quality effects to protected populations. This type of analysis should be conducted (a) to determine that no established criteria pollutant standards are violated and (b) to evaluate effects to protected populations. Although this form of detailed assessment is only required in specific circumstances according to the Clean Air Act and transportation conformity requirements, the method is broadly applicable to all transportation plans and projects that would benefit from microscale air quality assessment. The microscale analysis is typically performed at selected worst case, or "hot spot" intersections. Environmental justice assessment involves evaluating the pattern of where these intersections are located in relation to the activity spaces of protected population groups and then, at an even finer level, the effects of hot spot intersections on specific sensitive receptors. Analysis. The following steps are specific to the environmental justice component of the local air quality assessment. For an overview of a standard procedure for performing the hot-spot analysis, see "Microscale air quality assessment," above. FHWA (2001) also provides a detailed discussion of the requirements for performing hot-spot analysis. Figure 3-3 provides an example of how a microscale analysis process can be used to evaluate environmental justice. Step 1 Describe the transportation system change and gather necessary data. The detailed microscale assessment can be performed for policies, plans, and projects down to the level of a specific intersection. You must define the nature of the transportation system change and the air quality objectives by which the change should be evaluated. For general NAAQS conformance with an environmental justice component, the objective could be stated as "document NAAQS compliance ensuring that the proposed projects cause no violations at worst case intersections and in areas where there is possible environmental justice concern." In situations where air quality is highly controversial and there is a history of concern over impacts to protected populations, the objective could be to obtain emissions reductions. In this case, the goal would be to "evaluate project performance and ensure that the transportation facility yields a net emissions reductions once it is fully implemented." Once the transportation system change and objectives are described, you must assemble all the data necessary to describe intersection-related traffic conditions and demographic characteristics of the intersections. The specific data needed are described later under "Data needs, assumptions, and limitations." 71

OCR for page 65
Step 2 Review demographics for all intersections. Once each intersection has been assigned a score characterizing its level of environmental justice concern, rank the intersections either categorically or numerically. If an environmental justice concern was quantified using a measure such as the EJI, select a threshold value denoting areas of relatively higher concern based on expert opinion and knowledge of both system user demographics and activity space demographics near the intersections. Tabulate the number of intersections with a high level of environmental justice (EJ) concern. Step 1 1a Policy/plan/project/intersection description 1b Determine air quality/regulatory objectives 1c Assemble all data pertaining to intersection-related traffic conditions 1d Assemble all data pertaining to intersection-related demographics Step 2 Step 3 Step 4 Multiple intersection Multiple intersection 4a Individual demographic review screening/ranking intersection modeling 4b Assemble data on traffic, meteorology, site characteristics, 3a 3c background Rank top 20 Model top 3 based by traffic on traffic volumes 4c volumes Locate receptors 3b 3d 4d Compute 1-hour Calculate Model top 3 based peak traffic LOS on LOS concentration using CAL3QHC 4e Apply persistence factor and background 4f 2a 2b Rank by Tabulate number of Model top 3 based concentration intersections of on EJ concern EJ concern 4g Identify relationship between EJ score and concentration Disproportionate effects determination Disproportionate effects determination Figure 3-3. Example of a microscale environmental justice assessment process Source: FHWA 2001. The results of this step would lead to a finding of distributive effects to protected populations in two cases: 72

OCR for page 65
Fifty percent or more of the affected intersections in the study area are in areas with high environmental justice concern. The proportion of intersections in the study area with high environmental justice concern is larger than would be expected for the comparison region. The expected proportion of intersections with high environmental justice concern can be estimated from reviewing the demographic characteristics of a large sample of similar intersections in the comparison area. The comparison area should be a jurisdiction large enough to encompass the entire study area, such as a county, metropolitan area, or state. Step 3 Screen and rank multiple intersections. As suggested in the FHWA guide, rank the intersections by traffic volumes and select the top 20 intersections. Then calculate the level of service for those intersections. Select for individual intersection modeling, at a minimum, the top three intersections based on traffic volume ranking, the top three intersections based on LOS ranking, and the top three intersections based on EJ concern ranking. The purpose of adding at least three intersections of environmental justice concern is to allow comparison of results for worst case intersections obtained in Step 4. Step 4 Model individual intersection. The purpose of this step is to compute 1-hour peak traffic-related concentrations at sensitive receptors through application of a hot-spot model such as CAL3QHC with the addition of any persistence or background air quality factors. The screening and ranking step (Step 3) is recommended because the level of modeling effort increases linearly with the number of intersections being evaluated. By applying the screening process, you can be reasonably certain that the worst case intersections are being evaluated. Once you have calculated emissions scores for each modeled intersection, rank the intersections in order of concentration. Interpretation of the results can be relatively simple, although various statistical techniques also can be used to compute rank-order correlations and determine significance of the findings. Initial findings, however, can be made simply by identifying where the "high-EJ concern" intersections fall in relation to the worst case intersections. In general, a finding of adverse distributive effects to protected populations would be warranted based on the following results: Fifty percent or more of the worst case intersections are in areas of high environmental justice concern. The proportion of worst case intersections with high environmental justice concern is larger than would be expected for the comparison region. A violation or exceedance is indicated at any intersection of high environmental justice concern. Projected pollutant concentration increases will be measurable between baseline and future-year scenarios, and the areas with the greatest increases are characterized as having high environmental justice concern. In contrast, results indicating that concentrations are lower at intersections characterized as having high environmental justice concern when compared to the worst case intersections may support a finding that protected populations are not adversely or disproportionately affected. A 73

OCR for page 65
finding of beneficial distributive effects would be supported by a result indicating that concentrations at the modeled intersections with high environmental justice concern will be measurably lower for future-year scenarios than for the baseline condition. Data needs, assumptions, and limitations. This analysis requires the following types of information: Background concentration levels (although this can often be estimated); Traffic impact study, including traffic volume, intersection geometry, signal timing, traffic speeds, location of sensitive receivers; Regional vehicle parameters and fuel information, including: Vehicle classification and mileage data, Vehicle age distribution, and Vehicle inspection/maintenance program details; Specific fuel characteristics (oxygenated fuel, or ethanol); and Geometric layout of intersections and locations of sidewalks and nearby sensitive receptors. In addition, information is needed to characterize the level of environmental justice concern of each intersection being analyzed. Any number of methods presented in Chapter 2 can be used to identify the relative level of environmental justice concern. As an initial screening step, or for large projects or system-wide analyses to review policies or programs, it would be useful and possibly necessary to use a census-data-based approach such as the Environmental Justice Index. However, due to the localized nature of effects being addressed by the microscale analysis, it is more appropriate to use demographic data collected for specific receptors using field survey, local knowledge, public input, and other detailed data-gathering techniques. A microscale intersection analysis is limited by the fact that only discrete receptors are analyzed. These locations are on such a small scale that the approach does not allow any means of generalizing or characterizing the overall air quality in an area. Thus, the approach can only be used to identify if any microscale distributive effects would result from the proposed transportation system change. It cannot be extended to address the general air quality of the local study area. It should be noted, however, that this approach does evaluate the maximum impact at intersections affected by a proposed project. So if the maximum impact is below the NAAQS, you can be reasonably confident that the project's impacts elsewhere also will be lower than the NAAQS. Pollutant concentrations typically decrease very rapidly at greater distances from a roadway, but it would be difficult to precisely extrapolate results beyond the specific receptors being evaluated. It could also be possible that a roadway is to be moved closer to a receptor. Under these conditions, a slight shift in roadway alignment closer to a receptor could result in an increased concentration, even if an overall air quality benefit is achieved. For these reasons, occasions may arise where an increase from no-build occurs, but a build analysis verifies that concentrations do not violate the NAAQS. 74

OCR for page 65
Results and their presentation. Microscale intersection analyses predict the pollutant concentrations at discrete receptors near worst case intersections and intersections of high environmental justice concern. The number of receptors can vary significantly (e.g., 1 to 20). The location of these receptors is usually close (within 200 ft) to the roadway. Most analyses of this type include a figure showing the receptor locations, along with a table detailing the concentrations at each receptor. The results will always include the project alternative results and sometimes existing and no-build scenarios. Assessment. Policy, program, and project microscale results are most often compared to the NAAQS. However, results are occasionally compared to a no-build scenario. This microscale environmental justice assessment method is based on FHWA microscale or "hot spot" NAAQS conformity analysis, and shows expected concentrations at discreet receptors in parts per million (ppm). Environmental justice is evaluated by comparing and ranking expected concentrations and level of environmental justice concern at the worst case intersections as defined by traffic volumes, LOS, and level of environmental justice concern. If an exceedance of the NAAQS is predicted or if it is found that areas with relatively greater environmental justice concern experience relatively higher emissions, mitigation measures should be included to prevent the exceedances and unequal emissions levels. Mitigation could be required depending on attainment status and local rules. Mitigation may include adding intersection capacity with additional traffic lanes, optimizing signal timing for air quality purposes, or diverting traffic to other locations. Potential impacts associated with these measures could include right-of-way acquisition, increased pedestrian conflict areas, or increased traffic volumes at other locations. Mitigation measures often must be completed before a project is finished and sometimes may be a result of a nearby site development. Each of these mitigation measures could have environmental justice-related issues other than air quality that would need to be evaluated using appropriate methods. Method 3. Detailed regional analysis When to use. A regional air quality analysis is conducted for regionally significant projects and when required by local guidance. This type of analysis might be conducted for a major transit project, for a new freeway connection, or even for adding capacity to a major regional connection in an urban area. Because this technique provides only regional estimates and does not provide geographic distinction below the regional level, environmental justice assessment of regional air quality merely involves documenting the protected populations in the region and the air quality concerns that have been raised by those populations. Many of the analysis steps and issues discussed as part of the general air quality review (Method 1) apply to this technique as well. Analysis. A detailed regional analysis is carried out in two steps: Step 1 Perform regional air quality analysis. Regional air quality analyses use regional travel demand models and MOBILE6 emission modeling. Regional travel demand models provide traffic volumes and speed on a link level. The MOBILE6 model provides emission rates that vary by speed and incorporates vehicle age, vehicle classifications, and other operating 75

OCR for page 65
characteristics. By applying these emission rates to each link of the regional travel demand model, total regional emissions can be determined. Generally, if capacity is added to a transportation system, delay decreases. While this may result in more trips or longer trips, the reduced delay results in an overall decrease in emissions because vehicles operate least efficiently when idling or in congested environments. Step 2 Document air quality concerns of protected populations. Once the regional air quality analysis is completed, present the findings to protected population groups and address any questions or concerns. Except in cases where the geographic scope of the policy, program, or project covers more than one air quality region, it is very difficult to assess unequal distributive effects using this method. That does not mean, however, that regional air quality has no environmental justice implications. Because unequal effects cannot be determined in most cases and regional air quality is similar across broad areas, the environmental justice assessment effort should focus on sharing information with protected population groups and understanding any air quality concerns that these groups may have. To understand how protected populations perceive regional air quality and its health effects, consider either conducting a survey or focus groups. You must be sure to obtain the demographic characteristics of each participant that are relevant to characterizing protected populations. Recommended survey questions are presented in Chapter 2. Data needs, assumptions, and limitations. This type of analysis requires the following information: Results of applying a regional travel demand model, MOBILE6 modeling variables and regional characteristics, and Information on protected populations. The models that must be used (according to regulation) assume a uniform distribution of the vehicle fleet across a region. The greatest limitation of this type of analysis for environmental justice assessment is that it provides no geographic distinction for results. As a result, extension of the technique to evaluate environmental justice is merely a matter of disclosure. You can do little more than share the results with protected population groups and understand their most important air quality concerns and how those concerns may be different (e.g., lesser, greater, or focused on specific health effects such as asthma) from the general population. Suggestions for understanding locations in a study area where there may be relatively higher pollution concentrations are explained under Method 1, but can be applied here as well. Results and their presentation. A regional air quality analysis generally includes a description of the air quality status of the region, a description of the guidance, and a tabular presentation of total pollutant emissions. The pollutants include CO, NO2, PM, and VOC. The results of the analysis would include a no-build and build comparison of annual tons of pollutant. Again, transportation projects usually represent an increase in capacity, which usually results in reduced pollutant emissions because free flowing traffic pollutes less than stop-and-go traffic. Assessment. This method can provide regional estimates of air quality impacts that would result from a proposed major transportation project, but it cannot give an indication of how protected 76

OCR for page 65
for each zip code in the study area. Technology groups are combinations of vehicle characteristics and operating characteristics that have been identified through empirical research as being highly predictive. For engine start and engine off modes, vehicle fleet characteristics are estimated for zonal analysis units such as TAZs and/or zip codes. Estimates for operating vehicles (i.e., vehicles in motion) associated with road segments are based on analysis of travel demand model links. In MEASURE, the mix of vehicles on various roadways is identified by developing estimates of local and regional fleet characteristics based on zip-code-level vehicle registration data. Step 1c. Estimate vehicle activity. Use the travel demand model to predict regional travel, including the number and location of peak hour or daily trip origins, road segment volumes, and average speeds based on volume-to-capacity ratios. Use this information to estimate distributions of speed and acceleration, and vehicle mode operations. Step 1d. Predict facility-level emissions. Transportation "facilities" include major roads, minor roads, and trip origin zones. Engine start emissions are linked to the zone (i.e., TAZ, census block group, or census block) of origin. Minor roads not usually included in travel demand models are assigned running exhaust emissions as zonal averages based on travel time, local road speeds, fleet composition estimates, and vehicle miles traveled. Major roads included in the travel demand model are assigned running exhaust emissions as linear (i.e., road segment) averages based on speed and acceleration characteristics and fleet composition estimates. Step 1e. Generate the mobile emissions inventory. This step involves converting the facility- based emissions estimates generated in Step 1d to grid-cell-based emissions estimates. Convert estimates for each polygon facility (origin zone or minor road zone) to a rate expressed as grams/square kilometer, and convert major road facilities to a rate expressed as grams per kilometer. The polygon and line facilities are overlaid on the output grid cells and the rates are allocated to the grid cells based on the proportion of a facility that falls into each cell. Compute emissions per grid cell by multiplying the rates by the areas and lengths of the facility/grid cell portions. Step 1f. Apply photochemical model (Optional). Use an appropriate photochemical model to generate estimates of ground-level pollutant concentrations across the study area. The ground- level pollutant concentrations are based on meteorological conditions and the emissions estimates obtained for each grid cell. The physical properties of the pollutant and distance decay functions are also accounted for in the model. Perform this step if you wish to evaluate the distribution of ground-level pollutant concentrations rather than transportation-related emissions (the output of Step 1e). By incorporating emissions information for point sources into the photochemical model, you can evaluate cumulative ground-level concentrations in addition to ground-level concentrations from transportation-related sources. However, application of photochemical modeling requires collection and processing of far more input data (including regional speciated emission inventories and three-dimensional fields and boundary conditions of all meteorological and air quality parameters) than any of the other methods described here. Therefore, this step generally would be impractical for evaluating individual transportation projects. Step 1g (SAVIAH example). Data collection and preparation. Unlike a model-based method that estimates emissions based on roadway geometry, traffic volumes, and vehicle fleet 81

OCR for page 65
emissions characteristics, statistical methods predict population surfaces by fitting regression models to observations at monitoring sites based on known values for predictor variables. Because existing monitoring sites (such as stations used to evaluate regional air quality and to develop AQI scores) are relatively sparse in most areas, you may want to consider additional monitoring over a prolonged period of time (i.e., many weeks or months). A larger monitoring network and a larger number of samples will yield a more accurate regression model. In addition, you will need to collect the necessary information to compute predictor variables. Commonly used information is listed in "data needs, assumptions, and limitations." Step 1h. Conduct exploratory study. The result of a multivariate analysis is a best-fit curve allowing estimates of the response variable to be derived from known values of predictor variables. The purpose of the exploratory study is to reduce the set of candidate predictor variables described in Step 1g to the set to be used in the regression model. Highly intercorrelated predictor variables should be eliminated, as should other variables found to have low correlation with the response variable. Use a variable reduction procedure to develop correlation matrices and identify candidate predictors that should be retained in the final model. Comrie and Diem (2001), for example, used principal component analysis to evaluate an initial set of candidate predictor variables. Although it is beyond the scope of this guidebook to provide an in-depth discussion of variable reduction, numerous books on applied regression analysis are available, such as Neter et al. (1996). Although this approach is not always applied, it may be worthwhile to consider developing multiple regression models, one for each commonly observed set of meteorological conditions. Comrie and Diem (2001), for example, developed independent regression models for four distinct clusters of CO monitoring data, where observed concentrations, temperature ranges, wind speeds, and atmospheric pressure were relatively constant within each cluster. Step 1i. Refine and select model. The result of Step 1h is a small subset of candidate regression models with a limited number of explanatory variables that provide good predictive ability. Step 1i results in selection of the final model, based on review of residual plots and analyses to identify lack of fit, outliers, and influential observations. The SAVIAH study (Briggs et al. 1997), for example, developed regression models to predict NO2 concentrations in Huddersfield, the United Kingdom; Amsterdam, the Netherlands; and Prague, the Czech Republic. In each city, the regression models were based on local predictors that provided results with the most predictive power. However, two constraints were placed on the regression models developed for each city. First, the models had to include terms for traffic volume, land cover, and topography because each of these is known to affect pollutant dispersion. Second, a common GIS-based buffering approach was used to compute predictor variables that had a spatial component, such as total traffic volume within 300 meters of a monitoring site. The resulting regression equation for average annual NO2 concentration in Huddersfield included traffic volume within 300 meters, high-density housing, and industrial land use proportions within 300 meters, altitude, and sampler height. Step 1j. Apply the model. Based on computed predictor variable values for each grid cell, use the selected regression equation to compute predicted pollutant concentrations. This yields a pollution surface for the entire study area. 82

OCR for page 65
The output of either the model-based or statistical approach is a pollution surface that provides estimates of either emissions or pollutant concentrations for each cell in the study area grid. This output is then combined with a population surface to perform an environmental justice assessment. Step 2 Develop population surface. A population surface is a raster, or grid-cell-based, representation of a population. Population surfaces are produced in GIS using any of various algorithms to convert census polygon-based, or zonal, population data to a grid cell-based format. A population surface is the best form of demographic data to use with model-based and statistical air quality prediction techniques because their results are also grid-based. By using the same set of grid cells to produce pollution surfaces and population surfaces, you create an information-rich dataset that can be evaluated to assess unequal distributive effects to protected populations. More detail on the process for developing population surfaces is provided in Chapter 2. The output of this step is a set of GIS grids depicting the distribution of protected populations, nonprotected or other populations, and total population. Protected populations can include race, income, age, sex, or any other protected population group. It is also possible to produce a "protected population surface" that is merely the sum of all population groups of interest. From these datasets, standard map algebra routines can be used to compute estimates of affected populations and population percentages for the entire study area or for smaller areas within a larger region. Step 3 Overlay pollution surface and population surface. The next step is to overlay the pollution surface created in Step 1 with the various population surfaces created in Step 2. You can do this using GIS and relational database software. Both tabular and map-based results are needed. Figure 3-6 depicts the process of combining population surfaces and pollution surfaces and the resulting outputs. Step 4 Visualize results. The power of this technique is the rich dataset it produces for evaluating geographically distributed effects. With additional processing of time-series data, such as prebuild, build, and future-year scenarios, temporal environmental justice aspects can also be assessed. Because of the richness of the dataset, it is both unwise and unnecessary to base conclusions about distributive effects on a single statistical test or data visualization approach. Rather, it is important to analyze and visualize the data in many different ways. This will lead to a detailed understanding of distributive effects patterns and effective control measures for altering present or future distributive patterns. The following assessment and data visualization routines are described below: Relative emissions burden calculation, Relative pollution burden graphs, Regional effects mapping and analysis, and Local effects analysis. Relative emissions burden calculation. Relative emissions burden is defined as the ratio between (a) the average ground-level pollutant concentration (or emissions level) available to 83

OCR for page 65
members of protected population groups and (b) the concentration available to members of non- protected population groups. Note that this is a measure of pollutant "availability," not exposure. "Burden" therefore means that pollutants are present and available for persons to be exposed to. The relative emissions burden evaluation was first used in 1998 by the U.S. EPA to determine if pollutant distributions in the Louisiana Industrial Corridor disproportionately affected minority populations (U.S. EPA 1998b). From Step 1 Inputs Zonal census Zonal census blocks block groups Pollution surface Grid cells Zonal/Surface conversion Population surfaces Protected Nonprotected Total population (PP) population (NPP) population (TP) Overlay Tabulated population characteristics and concentrations for each grid cell Outputs Average emissions burden for Map PP and pollutant Comparison charts PP, NPP, and TP concentration Determine distributive effects and evaluate significance Figure 3-6. GIS process for combining pollution surfaces and population surfaces The EPA approach based estimates of emissions burden on proximity to pollution sources and on estimates of pollutant emissions in pounds. The approach did not account for pollutant dispersion based on meteorological and chemical properties, nor did it develop estimates of ground-level pollutant concentrations. Unlike the approach presented here, the U.S. EPA study used zonal census data, buffers of pollutant-emitting facilities, and information on emissions volumes to evaluate distributive emissions patterns. One benefit of working with population surfaces is that 84

OCR for page 65
calculation of relative emissions burden is relatively simple and straightforward when compared to working with vector GIS datasets. Relative emissions burden is computed using the following equation: N PCi NPPi / ( NPP ) i= 1 R= N PCi NNP / ( NNP ) i i= 1 where R = relative emissions burden N = number of cells in grid PC = pollutant concentration NPP = number of persons in protected population group NNP = number of persons in non-protected population group Evaluation of results is also straightforward. If R is greater than 1.0, the average level of emissions experienced by members of protected population groups is greater than the average level experienced by members of other population groups. If, for example, R is computed to be 1.25, this means that the average emissions burden to members of protected populations is 25 percent greater than the average emissions burden to members of other population groups. Although it is worthwhile to evaluate relative emissions burden, comparison of average burden levels across a study area is only of limited value in determining to what extent there may be unequal distributive patterns and where those patterns are evident. This more detailed evaluation is better performed by evaluating the distributions in charts and maps. Examples are provided below. Relative pollution burden graphs. When evaluating the codistribution of air quality effects and protected populations, visualizing the information patterns in graphs and maps is often the most insightful form of analysis. To produce graphs, the pollution surface must be combined with a population surface, which yields an estimate of pollutant concentration and protected population characteristics for each grid cell. This dataset can be used to identify any areas where standards may be violated and to identify locations where strategies to reduce pollutant concentrations are required. For purposes of environmental justice assessment, it is also important to understand that many communities may have concern over pollutant concentrations that meet regulatory air quality standards. This could be because a community feels that the NAAQS are not protective enough of sensitive individuals or of individuals that receive greater exposure due to lifestyle. Or, it could be that a community is concerned about the additive and synergistic effects of exposure to multiple pollutants. 85

OCR for page 65
Evaluating distributive effects at concentration levels below the NAAQS is also a useful approach for dealing with concerns about measurement error. It is often practical to evaluate effects above the NAAQS, as well as to identify a "threshold of concern" for values that approach the NAAQS. The threshold of concern can be based on policy, expert information, or even community input. Figure 3-7 depicts results of an analysis to estimate annual average PM10 concentrations (see page 61) in a five-county metropolitan region. The annual average NAAQS for PM10 is 50 micrograms per cubic meter (g/m3). A decision was made, based on public input, that areas with concentrations above a 40 g/m3 threshold of concern should also be evaluated for distributive effects. The graphs show the regional pattern of concentrations available to protected population groups compared to other individuals. The top chart shows that a greater proportion of protected population groups reside in areas with concentrations between 31 and 60 g/m3meter. The bottom chart shows that a greater number of persons in protected populations also reside within areas with concentrations in this range. It is important when using this technique to compare to the rest of the population both (a) the percent of the protected population and (b) the total number of individuals in the protected population group that would experience adverse effects. This can be done by preparing one set of graphs with percent of population as the vertical axis and another set of graphs with the number of persons as the vertical axis (as in Figure 3-7). Both evaluations are necessary to determine distributive effects because in certain study areas a majority of the affected population may belong to protected groups. In the case of this particular dataset, evaluations by number of persons and population proportions both show patterns of unequal distribution. For other study areas, however, this may not be the case. This topic is addressed again in Chapter 10. Regional effects mapping and analysis. The graphic visualization of pollution and population surface results presented in Figure 3-7 is a very useful assessment technique for characterizing the distribution of pollutant levels among population groups. The graphs are even more useful, however, when combined with maps showing the pollution surface overlaid on the population surface. Figure 3-8 provides an example for the same five-county area discussed above. The top map depicts areas where there is a high proportion of protected population. The bottom map depicts areas where a large total number of members of protected population groups reside. As with the graphs, it is necessary to evaluate the population distribution patterns in both ways. The benefit of using both maps and graphs is that one form of visualization overcomes the limitations of the other: maps are very good at depicting geographic patterns, but all detail is lost when you try to quantify the geographic patterns. On the other hand, by viewing the graphs it is easy to speak in quantitative terms about the disproportionate pattern that seems evident in the map. If you rely on just the graphs, however, it is easy enough to understand that a disproportionate pattern exists but impossible, without the map, to determine where the patterns are located. Regional analysis is thus a combination of determining relative emissions burden and visualizing any potential disparities using graphs and maps. In the examples provided above, it could be concluded that areas where annual average PM10 concentrations exceed 40 g/m3 are of concern and that strategies must be implemented to reduce these concentration levels. It could also be concluded that there is an environmental justice issue because those concentration levels burden 86

OCR for page 65
a greater proportion of the protected population and a greater number of individuals in protected population groups. This form of analysis is useful for policies, programs, long-range planning, and regionally significant projects. Proportion of Specific Population Group 45 Protected Population 40 Other Population 35 Percent of Population 30 Threshold Above of Concern PM10 NAAQS 25 Unequal exposure at levels between 20 31 & 60 micrograms/cubic meter 15 10 5 0 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 PM10 Concentration (micrograms/cubic meter) Total Persons 200,000 180,000 Protected Population Other Population 160,000 140,000 Threshold Above Population Count 120,000 of Concern PM10 NAAQS 100,000 Unequal exposure at levels between 31 & 60 micrograms/cubic meter 80,000 60,000 40,000 20,000 0 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 PM10 Concentration (micrograms/cubic meter) Figure 3-7. Distribution of annual average PM10 concentrations in a regional study area 87

OCR for page 65
Percent of Total Population Number of Persons Figure 3-8. PM10 concentration and protected population patterns in a regional study area 88

OCR for page 65
By performing prebuild and postbuild modeling runs, this method also allows you to evaluate policy, program, and project impacts. With some change in current regulation, the geographic variability of pollutant concentrations that results from this method would also give transportation planners a much broader set of control measures to consider for achieving transportation conformity. Local effects analysis. The pollution surface technique can be used to evaluate regionally significant projects and transportation corridors. However, for smaller projects with the potential only for localized effects, it is more appropriate to combine regional analysis of the pollution surface with microscale analysis (Method 2). Regional pollution surface analysis is useful to help understand whether the area of effects is in a location where pollutant concentrations are predicted to be relatively high. Microscale analysis can then be used to evaluate the localized effects of small projects. To evaluate localized effects of regionally significant projects and transportation corridors, the project area of effects must be identified. Using the pollution surface technique, a reasonable approach is to apply a dispersion model to identify the area within which the project would have a measurable (or meaningful) effect on concentrations. This level would depend on the pollutant being evaluated as well as on the sensitivity of the model. For CO, the area potentially affected by the project would be highly localized, whereas for PM the area of effects could be far- reaching. Figure 3-9 shows the distribution of predicted postimplementation annual average PM10 concentrations. The regional pollution surface described above was used to tabulate results within the project area of effects defined for two alternative alignments. Alternative 1 is depicted in the top two graphs and Alternative 2 in the bottom two graphs. For each alternative, percent of population is depicted in the left-hand chart, and population count in the right-hand chart. A review of these graphs indicates that Alternative 1 may be preferable from an environmental justice perspective because there would be no unequal impact to protected population groups above 40 g/m3. In contrast, Alternative 2 would result in a greater proportion of protected population groups being exposed at these levels. Both the total number of persons and the total number of persons in protected population groups in areas with concentrations above 40 g/m3 would be similar under both alternatives. The biggest drawback of the pollution surface approach is that it requires a great deal of time and effort to produce the necessary data. That is why, although limited in their predictive power, the previous three methods are likely to be used in all but the most controversial situations. Both statistical and model-based methods for estimating the regional and local variability in air quality are highly experimental. Unlike the micro-scale analysis and regional air quality assessment methods, this method has not yet been mandated, or even widely accepted, as part of the regulatory process. The assessment of relative emissions burden is not the same as determining if people in the study area are exposed or if the level of exposure varies by person. To measure exposure, you must consider factors such as the amount of time persons spend outdoors and the ventilation properties 89

OCR for page 65
of buildings. Tests to evaluate the statistical significance of the relative emissions burden statistic (R) are available, but their application is problematic. This is because in most situations the sample set, in this case grid cells, will be large and even slight differences in average emissions burden would be deemed significant. It is also uncertain whether you can assume that the sample population is normally distributed about the mean, which is a requirement of most parametric tests. Because of these concerns, it is better to merely evaluate the value of R to determine the direction of the difference in average burden levels. Proportion of Specific Population Group Total Persons Alternative 1 Alternative 1 45 14,000 Protected Population Protected Population 40 Other Population 12,000 Other Population 35 10,000 Percent of Population 30 Threshold Above Threshold Above Population Count of Concern PM10 NAAQS of Concern PM10 NAAQS 8,000 25 No unequal exposure No unequal exposure at levels above at levels above 20 6,000 40 micrograms/cubic meter 30 micrograms/cubic meter 15 4,000 10 2,000 5 0 0 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 PM10 Concentration (micrograms/cubic meter) PM10 Concentration (micrograms/cubic meter) Alternative 2 Alternative 2 40 3,000 Protected Population Protected Population 35 Other Population Other Population 2,500 30 Percent of Population Threshold Above 2,000 Threshold Above 25 Population Count of Concern PM10 NAAQS of Concern PM10 NAAQS Unequal exposure Unequal exposure 20 1,500 at levels between at levels between 30 & 50 micrograms/cubic meter 6 & 55 micrograms/cubic meter 15 1,000 10 500 5 0 0 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 0-5 6-10 11-30 31-40 41-50 51-55 56-60 61-65 PM10 Concentration (micrograms/cubic meter) PM10 Concentration (micrograms/cubic meter) Figure 3-9. Comparison of alternative project alignments Data needs, assumptions, and limitations of population surfaces are discussed in Chapter 2. Results and their presentation. The graphics included in the discussion of Step 4, above, provide examples of results and how they may be presented. Depending on the audience, it may be necessary to simplify the presentation of the graphs and maps. Data needs, assumptions, and limitations. Application of a dispersion model such as MEASURE would require the following data to be collected (Bachman et al. 2000): 90

OCR for page 65
Spatial character information Vehicle mode activity information -- Land use boundaries -- Idle -- Census blocks -- Cruise -- TAZs -- Acceleration -- Roads -- Deceleration -- Travel demand model network -- Starts -- Output grid cells -- Engine off Temporal character information Trip generation information -- Hour of the day -- Land use Vehicle technology information -- Housing units -- Model year -- Socioeconomic characteristics -- Engine displacement -- Home-based work trips -- Transmission type -- Home-based shopping trips -- Fuel delivery technology -- Home-based school trips -- Supplemental air injection system -- Home-based other trips -- Catalyst configuration -- Nonhome-based trips -- Exhaust gas recirculation Application of a regression model such as the one implemented in the SAVIAH project would generate the following data needs (Briggs et al. 1997): Road network Topography Road type Concentrations from monitoring sites Distance-to-road Sample height Traffic volume Site exposure Land use/land cover Topographical exposure Assessment. Analysis of local and regional air quality using pollution surfaces and population surfaces is a very useful technique for evaluating distributive air quality effects to protected populations. Pollution surfaces can be developed using models that extend commonly used regulatory models to account for geographic variability in pollutant concentrations. Data visualization techniques using graphs and maps and findings based on expert opinion and public input generally are recommended over statistical tests that reduce the information-rich dataset to a single test for statistical significance. Although these methods are not widely used in current practice, considerable research has been performed. The drawback of this method is that it is extremely data intensive and has not yet received the level of regulatory approval that has been given to traditional microscale and regional air quality assessment methods. Given that techniques for understanding the geographic variability of air quality are important both for 91