Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 101


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 100
Use of these models provides the basis for developing a more standardized "quantitative" approach to environmental justice assessment of hazardous materials concerns in the transportation field. Practically speaking, these models (or adaptations of them) would generally be used for larger and more complex transportation projects. Figure 4-2. Census data from LandView III Source: EPA 1998. METHODS Environmental justice assessment of transportation-related hazardous materials effects should use methods that match the overall complexity of the project or program being evaluated. Using a "tiered" process, the assessment should be initiated using practical desktop review methods and elevated to more complex analysis and computer modeling only as dictated by project requirements. By using a tiered assessment process, you can develop an efficient approach to environmental justice assessments within your agency's objectives and resource limitations. The following methods provide examples of how hazardous material data can effectively be used to perform environmental justice assessment. The techniques presented here may be adapted or modified to meet specific project or program needs. Table 4-2 provides a summary of four methods for evaluating environmental justice with respect to hazardous materials. 102

OCR for page 100
Method 1. Phase 1 desktop assessment When to use. This approach can be used as the initial environmental justice review to evaluate distributive effects of potential hazardous materials exposure in most project or corridor studies. The examples provided below are for performing the evaluation as part of a Phase I ESA. Desktop assessment is also well suited to assessing the distribution of hazardous materials sites with respect to demographic patterns during all phases of transportation planning. Additionally, it is appropriate for evaluating environmental justice concerns related to construction staging areas, transportation maintenance facilities, transportation projects where physical property will be acquired or altered, and patterns of known hazardous materials spills. Table 4-2. Summary of methods for analysis of hazardous materials effects Assessment Appropriate Use Data Expertise Method level uses when needs required 1. Phase 1 Screening Initial During evaluation of Low Simple data desktop assessment of proposed construction analysis assessment the presence of corridors hazardous waste sites 2. Phase 1 Screening Second-tier When desktop assessment Medium Geographic computer- assessment of indicates possible information based the presence of problem areas systems assessment hazardous (GIS), waste sites Statistical analysis 3. Hazardous Screening Initial During evaluation of Low Simple data materials assessment of proposed construction analysis transport transport routes corridors screening for hazardous study materials 4. Hazardous Detailed Risk modeling Screening methods Medium/ Fault-tree and materials of hazardous indicate a significant high other risk transport-- materials potential for exposure to analysis probability exposure or hazardous materials methods, GIS modeling release Cost of mitigation or remediation is high Analysis. The approach combines Phase I ESA database and map review with desktop demographic review. It involves evaluating the presence of both hazardous materials sites and protected populations in the study area. When the two are present in the same area, there is the potential for environmental justice concern and the need to perform further review. Used in this manner, the approach serves as a useful screening technique so that the resources to perform more in-depth analysis can be targeted to the areas where they are needed. 103

OCR for page 100
Step 1 - Conduct environmental assessment. Review national, state, and local databases to identify locations where hazardous materials and waste are likely to be produced, stored, or used. The results of a Phase I ESA presented in map form are ideal for this purpose. The Phase I ESA process is discussed above, and a list of useful databases is provided in the resources section of this chapter. Step 2 Perform demographic review. Collect information on the presence of protected populations using any number of the techniques described in Chapter 2. Especially useful techniques include use of local knowledge, threshold analysis using block and block-group-level census data, field survey, and the Environmental Justice Index (EJI). Whatever technique or combination of techniques is applied, the intent is to identify locations in the activity space of protected population groups. Step 3 Tabulate results. Results of the environmental and demographic reviews can be compiled in numerous ways. Probably the simplest approach is to mark up a Phase I ESA map to show minority or low-income neighborhoods and work places and activity centers that are predominantly used by members of protected population groups. Then it is relatively straightforward to list the sites where further environmental justice review, such as a thorough field survey, should be performed. Data needs, assumptions, and limitations. The environmental review should include the following information sources: EPA National Priorities List (Superfund) sites, Sites on the state Priority List, Leaking underground storage tanks (LUST), Solid waste landfills, incinerators, and transfer stations, Registered underground storage tanks (UST), Sites with previous hazardous materials spills, and Sites that generate hazardous waste. The demographic review should be based on readily available information according to the method used to identify the protected population. More information on data sources is provided in Chapter 2. The desktop assessment technique is limited to identifying hazardous materials sites near areas used by protected populations. A more thorough review should be performed in situations where such locations are identified. This semi-quantitative approach does not use statistical analysis. Application of this technique alone is not recommended for controversial projects where more thorough analysis would be required. The technique is not useful in situations where hazardous materials transport and release should be evaluated. Results and their presentation. The best form of presentation is maps showing hazardous materials sites, small-area demographic data, neighborhoods, and sites of interest to protected populations. Figure 4-3 provides an example. 104

OCR for page 100
Assessment. For most project and corridor environmental justice assessments, this is the logical hazardous materials screening evaluation to perform. In many cases, further study will not be necessary. In cases where there is a need for further analysis, consider using GIS to perform the environmental and demographic reviews as this would make it easier to use the results in further studies. This method can also be used as a way to evaluate benefits to protected populations by identifying areas where environmental cleanup activities are planned. Figure 4-3. Results of a Phase I desktop assessment Method 2. Phase 1 computer-based assessment When to use. This approach is a modification of the Phase I desktop assessment that uses GIS and includes a statistical test. Consider using this technique when projects are controversial or large or if the desktop assessment indicates the potential for environmental justice concern. This approach, although somewhat data intensive, is also suitable for reviewing regional or even statewide hazardous materials programs. Analysis. Steps 1 and 2 are the same as those in Method 1, Phase 1 desktop assessment, although the information should be stored electronically in GIS. Step 3 Statistical test. Various statistical tests are available to determine if hazardous material sites are located predominantly in protected population areas. The specific test that should be applied depends in large part on the amount of data being evaluated and on the experience and qualifications of the person performing the analysis. Often it is appropriate to use simple techniques that are easily performed manually or with the aid of spreadsheet software. The following example illustrates the application of a chi-square test for independence. This test can be used to determine if hazardous material sites in the study area occur more frequently in areas with protected populations than would be expected if they were randomly distributed. To perform the chi-square test, first divide the study area into sub-areas. In general, it is best to use a high level of resolution, because the chi-square test is more robust with larger numbers of 105

OCR for page 100
observations. For a typical transportation project, an analysis based on census block or block- group subareas provides adequate resolution. Once you have determined the subareas, characterize each in terms of their relative density of protected population using one of the techniques described in Chapter 2. Using the EJI, for example, you could define subareas of higher environmental justice concern as census block groups with an EJI greater than 40 (or another value that is appropriate for the study area in question). Next, characterize each subarea in terms of presence or density of hazardous materials, based on the Phase 1 ESA. The characterization could be based on, for example, the number of hazardous materials sites within 1 mile of the subarea. If subareas vary greatly in size, it may be necessary to convert the score to an area-weighted measure, such as the number of hazardous facilities per square mile. Convert the quantitative risk estimates to a two- or three-point scale (e.g., high, medium, and low availability). Table 4-3 is an example of results of a hypothetical analysis. Table 4-3. Example analysis results EJI Risk of exposure Sub-area > 40 40 Low Medium High 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Table 4-4 shows the same data cast into a 2 by 3 contingency table. The values in italics are the expected frequencies for each cell if the EJI did not vary with the presence of hazardous materials. The chi-square test compares the actual distribution of values within the table with the expected frequencies and determines the probability that the discrepancies between the two could have occurred from sampling error alone (in other words, that there is not a statistically significant difference between the two distributions). 106

OCR for page 100
Simply reading the table, it appears from visual inspection of this table that the environmental impacts of this hypothetical project are not equally distributed. Only 20 percent of the sub-areas with low hazardous materials availability have EJI ratings greater than 40, whereas 100 percent of the sub-areas with high hazardous materials availability have EJI ratings greater than 40. This is only an impression, however, which can be confirmed (or not) using the chi-square test. This test can be conducted using virtually any standard statistical software package. Table 4-4. Contingency table for example data Hazardous materials presence Low Medium High Total 1 2 5 8 > 40 2.67 2.67 2.67 EJI 4 3 0 7 40 2.33 2.33 2.33 Total 5 5 5 15 Note: Expected values for the chi-square computation are in italic. In this example, the computed value of chi-square (X 2) is 6.97. The statistical significance of the value of X 2 is determined by a table look-up (e.g., Siegal and Castellan 1988, Table C), with 2 degrees of freedom (df). (Statistical software programs provide this information.) The degrees of freedom are based on the number of rows and columns in the contingency table: df = (# of rows 1) X (# of columns 1) = 1 X 2 = 2 In this example, the value of X 2 (6.97, df = 2) is significant at the 5 percent error level. That is, there is a 5 percent or less probability that the observed discrepancy between observed and expected frequencies would occur by chance alone. Thus, the subjective impression that the availability of hazardous materials is not equitably distributed between protected and non- protected populations is confirmed statistically. Data needs, assumptions, and limitations. The following data are required for the calculation of X 2: Phase 1 Environmental Assessment results by census blocks or block groups and Demographic data detailing the density of protected and nonprotected populations in each census block or block group. When the expected frequencies are very small, the X 2 test should not be used. Siegal and Castellan (1988) list several criteria that should be met, including the following: When the degrees of freedom is 1 (i.e., rows = 2 and columns = 2) and the total number of observations (census blocks, in the example given) is less than or equal to 20, X 2 should not be used. In these cases, the Fisher exact test may be used. 107

OCR for page 100
When the degrees of freedom are greater than 1, the X 2 test should not be used if more than 20 percent of the cells have an expected frequency of less than 5 or if any cell has an expected frequency of less than 1. (Note that the example given does not meet this criterion. It has been simplified for purposes of description.) The chi-square test is widely used and has the benefit of simplicity. However, it is limited in that it does not take into account order effects. In the example presented, the value of the chi-square statistic would be the same regardless of the order in which the columns (or rows) of the contingency table were placed. This means information is available that is not used in the analysis. Siegal and Castellan (1988) describe several nonparametric statistical tests for two independent samples that make use of order information, although they are computationally more complicated than the chi-square test. These tests (e.g., Wilcoxon-Mann-Whitney test, Rank Order test, and Kolmogorov-Smirnov Two Sample test) are more powerful than the chi-square test for ordinal-scale data. That is, they are more likely to indicate whether a statistically significant effect is present than is the chi-square test when applied to the same data. Results and their presentation. Use maps or a GIS to plot hazardous materials and socio- demographic overlay information. The value of chi-square, degrees of freedom, and the level of significance should be presented along with the contingency table. Text should be provided documenting the methodology used, data collected, assumptions made, and interpretations derived. Assessment. This method makes use of quantitative data similar to that gathered in Method 1 for the desktop assessment of hazardous materials distribution. It has the benefit of using an inferential statistic to validate or reject any subjective impressions that may have arisen from the desktop assessment. The proposed statistic, chi-square, is easy to use and to interpret, although it may not be as powerful as alternative, more complicated statistical tests that make use of order information in the data. Method 3. Hazardous materials transport screening study When to use. Assessing environmental justice aspects of accidental hazardous materials releases in transportation corridors is based on the risk to protected populations compared to that of the rest of the population. In this context, the term risk implies a combination of two factors--the probability of an accidental release and the impact of the release on the populations. This method and Method 4 are suitable for assessing risk of exposure to accidental releases of hazardous materials in transportation corridors. This screening method is used to obtain a rough estimate of the possible risk of a hazardous materials release in different segments along a route or set of routes and to determine whether the risk is disproportionate in areas with protected populations. Use this method as a screening step to determine whether a more detailed risk assessment needs to be done (Method 4, below). Transport screening studies rely on existing sources of information to determine the following: Routes in the study area that are available for hazardous materials transport, 108

OCR for page 100
Sources and destinations of hazardous materials transport in the study area, and Distribution of protected and nonprotected populations in close proximity to the routes under study. This method does not include a detailed analysis of the probability that an accidental release might occur. However, if results indicate that protected populations would be more likely to be impacted if a release were to occur, a more detailed analysis of exposure risks should be conducted. By the same token, if a preliminary study indicates that there is not a distributive effect, the more detailed and costly risk analysis method is probably unnecessary. Analysis. To apply this method, you must first make a rough determination of the likely routes for hazardous materials transport within the study area. The second step is to determine the number of people protected and nonprotected populations living near enough to the roadway to suffer the consequences of a spill. The third and final step is to apply a statistical test to determine whether the protected populations would be disproportionately impacted if a spill were to occur. Step 1 Identify hazardous materials routes. The objective of this step is to identify roadways on which there is reason to expect that hazardous materials will be transported. Review national, state, and local databases to identify locations where hazardous materials and waste are likely to be produced, stored, or used. The results of a Phase I ESA, presented in map form or a similar presentation, are ideal for this purpose. The Phase I assessment process is discussed above, and a list of useful databases is provided in the resources section of this chapter. Use these data to identify possible sources and destinations of hazardous materials. Next, contact the national, state, and local agencies that issue permits for hazardous material transport. Use the information previously gathered regarding sites where hazardous materials are produced, stored, or used to identify transport permit holders. In some states (e.g., Georgia), holders of hazardous materials transport permits are required to file annual reports detailing the type and amount of hazardous materials transported and the origins and destinations of transport. If available, this information can help to determine the routes of hazardous material transport within the study area. In some areas (states, counties, or cities), the transport of hazardous materials is restricted by statute to certain routes. For example, routing information for the state of Texas may be found online (TXDOT 2003). If this information is available, it should be analyzed to determine whether the designation of preferred, prohibited, or alternate routes impacts the transport of hazardous materials in the project area. For the purposes of this method, a hazardous materials transport route is any section of roadway that is designated by state or local statute to be a hazardous materials route or that is a preferred route to or from a location identified as the destination or source of hazardous materials. GIS- based routing analysis can be used to identify preferred routes if that information is not directly available from other sources. Step 2 Perform demographic review. Mills and Neuhauser (2000) describe a method for determining the density of protected and nonprotected populations living in proximity to a 109

OCR for page 100
roadway--in this case, a roadway used for transport of hazardous materials. Their method consists of analyzing census data for the blocks in proximity to the roadway. This is straightforward, except for the issue of how to define "proximity." The Argonne National Laboratory conducted a study (Brown et al. 2000) to determine the Initial Isolation Zone (IIZ) and Protective Action Distance (PAD) for accidental releases of various classes of chemicals that are toxic by inhalation (TIH) or that produce TIH gases when they react with water (TIHWR). They define the IIZ as the radius of a zone around a release from which all people not directly involved in emergency response are to be kept away. The PAD is the downwind distance from a release that defines a zone in which persons should be either evacuated or sheltered-in-place. The authors computed the IIZ and PAD for small and large spills of various materials. IIZs and PADs are given for both daytime and nighttime spills. Nighttime IIZs and PADs are greater than daytime values. It may be reasoned that the IIZ represents the zone of immediate and significant impact. The largest IIZ for any of the materials in this study was 3,000 feet (e.g., the nighttime IIZ for a spill of over 55 gallons of liquefied toxic gas). Thus, for the purpose of this analysis, the figure of 3,000 feet can be used as a conservative definition of proximity to a release. If the nature of the hazardous materials being transported in the study area is known or if it is known that only small quantities (55 gallons or less) of the most dangerous materials will be transported within the area, it may be appropriate to use a smaller value than 3,000 feet, based on the worst case IIZ for the known conditions. Indeed, the definition of proximity may differ from one section of the roadway to another, depending on local conditions. For example, if you determine in Step 1 that a plant producing anhydrous ammonia is located on a road segment that is not designated as a hazardous material route, it may be assumed that anhydrous ammonia is the only hazardous material likely to be shipped on that road segment. In this case, it would be appropriate to use the worst case IIZ for anhydrous ammonia (200 feet) as the definition of proximity for the purpose of this analysis. Once proximities have been defined, you can use buffer techniques in GIS and perform small-area interpolation as described in Chapter 2 to characterize the demographics of the population in the proximate zone or zones. Step 3 Analyze the findings. In Step 1, roadways on which hazardous materials may be transported are identified. In Step 2, a buffer zone around the roadways is defined and the census data for the blocks or block groups falling within the buffer zone or zones is compiled. To analyze these findings, compare the protected population proportions and the nonprotected population proportions for the proximate zone or zones. For example, if the study area has a total of 10,000 low-income persons (or members of any protected population group) and through small-area interpolation it is estimated that 2,000 live in a zone near hazardous materials transport routes, then an estimated 20 percent of the protected population lives in the proximate zone. This calculation is then repeated for the nonprotected population. If the proportion of the total study area protected population living in proximate zones is greater than the total study area non-protected population living in those zones, the protected population 110

OCR for page 100
group may be differentially affected by hazardous materials spills if they are expected to occur randomly along hazardous material route segments. Step 4 Optional statistical test. A discrepancy is defined as a difference in the proportion of the protected population and the proportion of nonprotected population in proximity. A statistically significant difference exists if the observed difference could not be explained by chance alone. If a discrepancy is observed that is very unlikely to occur under random and independent assignment, then the discrepancy is statistically significant. Note that a significant discrepancy is a necessary but not sufficient condition to show a disparate impact. It is insufficient because the statistically significant result could be due to the fact that the premise of random and independent assignment of individuals to locations is not appropriate. Nonetheless, this kind of evaluation serves as a useful starting point for evaluating a potential disparate impact. To compute the test statistic, you must calculate the proportion of the total protected population that is in proximity (this is defined as p1) and the proportion of the total nonprotected population that is in proximity (this is defined as p2). If p1 and p2 are different from one another, this is evidence of a discrepancy. The test statistic is figured as the difference p1 - p2 divided by the standard error of the difference, where the standard error is computed under the assumption that the two true proportions, p1 and p2, are equal. Under this assumption, the expected value of p1 - p2 is zero. The standard error is interpreted as the amount by which the observed difference p1 - p2 might differ from zero just due to chance variability. Thus, taking the observed difference relative to the standard error indicates whether the observed difference is "far" away from zero. A general rule of thumb is that if the ratio p1-p2 divided by the standard error is greater than 2 or 3, then one can conclude that p1 is statistically significantly greater than zero. The formula for the test statistic (P) is thus: ( p1 - p2) P= 1 1 p ^ ) + ^ (1 - p n1 n2 where p1 = the proportion of the total protected population that is in proximity p2 = the proportion of the total nonprotected population that is in proximity ^ = the overall proportion of the total population that is in proximity p n1 = the total number of individuals in the protected population group in the population n2 = the total number of individuals in the nonprotected population group in the population (see for example Bain and Engelhardt 1989) The confidence interval for the ratio p1/p2 is computed as: 111

OCR for page 100
p explog p 1.96 1 n + 11 ( ) + (n 1 -1 1 + ) + (n 1 -1 21 + ) + (n 1 -1 2 + ) 1 -1 2 2 2 2 2 where n1 = the total number of protected persons in the population1 n2 = the total number of nonprotected persons in the population n11 = the number of protected persons in the population in proximity n21 = the number of nonprotected persons in proximity (see, for example, Agresti 1990) Data needs, assumptions, and limitations. This method relies on the availability of information about which roadways are designated by state or local statute as hazardous materials transport routes. If no such statutes are in place, you must assume that all roadways in the study area are available for hazardous material transport. Additional information from the Phase 1 ESA may also be used to designate roadways as hazardous materials transport routes by virtue of being preferred access or egress routes for facilities that produce or use hazardous materials. Because no hard information is used regarding the type or volume of hazardous materials actually transported, the method relies on very conservative assumptions about what materials are being transported on what roadways. Results and their presentation. Considering the preliminary nature of this method, elaborate statistical tests are not required, although one has been included to aid in interpreting results. The objective of the method is merely to give an impression of whether the proximate buffer zone has a higher proportion of members of protected populations compared to the proportion of members of nonprotected populations. To quantify the possible distributive effect, simply compute the ratio (p1/p2) of the protected population proportion living in the proximate zone (p1) to the nonprotected population proportion living in the proximate zone (p2). A ratio greater than 1.0 indicates that there is a possible disproportionate pattern. Assessment. This is a semi-quantitative screening method to assess the impact of hazardous materials transport within the environmental justice framework. It only crudely quantifies the probability of a transport-related release by attempting to determine what routes are used for hazardous material transport. It relies on worst-case assumptions about the volume, time, and composition of possible spills. Within those limitations, however, it can provide a high-level determination as to whether there is an environmental justice issue that needs further, more careful analysis. Method 4. Hazardous materials transport--probability modeling When to use. This method is used to analyze the risk to protected and nonprotected populations associated with accidental release of hazardous materials in transit. Unlike Method 3, this 1 Protected persons are defined as individuals who belong to a protected population group. 112

OCR for page 100
method makes use of a hazardous material flow survey to estimate the types and volumes of materials transported over various segments of a transportation corridor. Thus, it allows a more detailed analysis of the distribution of hazardous materials exposure risk between protected and nonprotected populations. Analysis. This method depends heavily on the performance of a material flow survey, as described in greater detail in Guidance for Conducting Hazardous Materials Flow Surveys (U.S. DOT 1995). Step 1 Conduct hazardous materials flow surveys. The objective of this step is to derive the hazardous material flow data for each segment of the project area and for any areas that will be used for comparison with the project area as a whole. A flow survey is an empirical technique that involves monitoring the hazardous material transport on a given route segment. It is accomplished by stopping all trucks that display U.S. DOT hazardous materials placards and examining their shipping papers. For comprehensive guidance on conducting hazardous material flow surveys, see U.S. DOT (1995). The following is a brief summary of the major steps described in that document: a) Identify the specific purpose of the study. In the present context, the reason for performing a flow study is to develop an accurate and defensible estimate of the probability that an accidental release of hazardous material will occur in the study area. An estimate of probability in turn relies on an accurate determination of the following information: Number of trips involving hazardous material transport in any week, Volume of hazardous material transported in each trip, Type of material transported in each trip, and Type of container. In some cases, the scope of the analysis may be limited to certain types of material. Any decision to limit the scope of the study should be based on an initial survey of the types of materials transported in the study area. For example, if it were known that the project corridor is or will be used for transport of spent fuel from a nuclear power station, the motivation for a risk analysis might be limited to accidental release of radioactive material. b) Gather baseline information. Before conducting the actual flow survey, review existing information to determine the routes within the study area over which hazardous materials will be transported, as described in Method 3, above. In addition, gather information about the condition and other attributes of the route, such as lane widths, road capacity, and shoulder conditions. The U.S. DOT Highway Performance Monitoring System (HPMS) is a good source for this information. Gather route-specific information such as total traffic volume, volume of truck traffic, and accident history. Finally, use the techniques described under Method 3 to estimate the types of hazardous materials that might be transported in the study area. c) Design the study. Using the baseline information, determine what route segments are to be studied. Establish optimal locations for survey stations where trucks can be stopped for inspection with minimal disruption to the carrier and the flow of traffic. Decide over what 113

OCR for page 100
time periods the survey will be conducted. At a minimum, continuous 24-hour surveys of truck traffic over several days during at least two distinct seasons of the year is desirable. Based on the number of survey stations and the duration of sampling, determine the personnel needs for the survey. Two surveyors and several state police are the minimum staff that will be needed for each survey station. d) Perform the surveys. Inspect all trucks displaying hazardous materials placards that indicate the truck is carrying the type of material being studied. In general, it should not be necessary to physically inspect the contents. Trucks carrying hazardous materials are required to have shipping papers containing all the necessary information. A standard checklist should be developed and used to ensure that all essential information is obtained for each truck inspected. e) Analyze the data. Depending on the particular objectives of the study, the survey findings should be collated according to the type of truck, the type and volume of material carried, and the type and size of any containers used. It may also be desirable to analyze the density of hazardous material transport as a function of the time of day. Step 2 Estimate the probability of accidental release. For each route segment, estimate the probability of an accidental release using event-tree analysis. The basic information required for this analysis--volume of traffic for each material type, volume of shipment, and container type--comes from the material flow survey performed in Step 1. The following factors may be taken into consideration: Type and volume of hazardous material, Roadway condition, Traffic density, and Type of container. The data gathered in Step 1 supports the event-tree analysis. The method for conducting the event-tree analysis, including normative probabilities for various types of accidental release scenarios, is given in Battelle (2001). An example event-tree is shown in Figure 4-4. An event-tree analysis involves assigning probabilities to each branch of the event-tree. The combined probability for each "leaf" of the tree (termini on the right of the event-tree) is computed by combining the probabilities for each branch leading into the leaf. Additional data sources for computing probabilities used in the event-tree analysis include the following: U.S. Bureau of the Census Commodity Flow Survey (1997). U.S. DOT Hazardous Materials Information System. U.S. DOT Hazardous Materials Incident Data and Summary Statistics for Incident Years 19932002. U.S. DOT, Federal Motor Carrier Safety Administration Motor Carrier Management Information System (MCMIS). 114

OCR for page 100
Urban area Large release Rural area Release Accident occurs Small release No release Figure 4-4. Example event-tree for release of hazardous material The outcome of this step is a probability estimate for both small and large spills for each type of material in each segment of the study area. Step 3 Estimate the impact of accidental release. Whereas Step 2 consists of determining the probability that an accidental release will occur, this step involves estimating the level of impact a given type of release will have on people in the surrounding area. Define an impact function for each type and volume of material transported per the materials flow survey. Use published nighttime (i.e., worst-case) PADs to estimate the maximum size of the impact area for each material type and volume of spill. A simple dispersion model can be used to weight the impact based on the distance from the roadway and the PAD; for example, an impact score of 100.0 can be assigned at the site of the release and 0.0 at the PAD distance from the roadway, with linearly decreasing scores at intermediate distances from the roadway. This relationship can be expressed as follows: Px - d I x (d ) = Px where Ix(d) = the impact at distance d of a spill of type x for which the PAD is Px If desired, a more accurate estimate of the impact can be obtained using air dispersion modeling, taking into account such factors as the volatility of the material, the influence of terrain, prevailing wind direction, and other meteorological conditions common in the study area. In that case, the impact function would not be assumed to be uniform in all directions. 115

OCR for page 100
Note that the impact is in arbitrary units. The scaling of the impact is unimportant, as long as it is proportional to the relative consequence of each type of spill at a given distance from the roadway. Step 4 Develop a risk surface. Using the impact function and probability of each type of spill (type and volume of material), compute a risk function for each using the following equation: R( d ) = ( I x ( d ) px ) x where R(d) = the total risk score at distance d from the roadway for all types x of spills Ix(d) = the impact at distance d of each type of spill px = the probability of each type of spill Note that the value of d should be no greater than the PAD for each type of spill; that is, do not use negative values for any Ix(d). Use the risk function to develop a risk surface similar to the pollution surface described in Method 4 of Chapter 3. The risk surface amounts to a GIS layer that indicates for each grid cell the maximum risk of exposure to an accidental release of hazardous material. Step 5 Perform demographic analysis. Using GIS, develop a population surface as described in Method 4 of Chapter 3. If road use analysis indicates that a significantly disproportionate number of members of protected populations use the roadway, this should also be taken into consideration. To determine this, develop the estimated numbers of protected and nonprotected individuals traveling on the road segment over a given time interval and compare these demographics to the maximum risk scores computed for each segment. The time interval chosen should be equal to the driving time at average speed to travel a distance equal to the average PAD for the materials studied. When analyzing the road use data, you should only count individuals who are traveling through the study area, not those living in the study area, so as to avoid double counting. Step 6 Evaluate distributive effects. Using the techniques described in Steps 3 and 4 of Method 4 in Chapter 3, overlay the risk surface with the population surface to analyze potential distributive effects on risk of exposure to hazardous materials for protected versus nonprotected populations. Data needs, assumptions, and limitations. This method is data intensive, relying on existing databases and reports for historical accident data, roadway conditions, and demographic information. In addition, it requires data derived from hazardous material flow surveys, which are costly and labor intensive. In both cases, the method necessarily relies on extrapolation from relatively sparse data. Due to the high cost involved in collecting or developing truly complete information, it is necessary to assume worst case conditions. For example, because the volume of material flow data is unlikely to be great enough to allow modeling of diurnal patterns, the worst case nighttime PADS should be used to describe the area of impact of a spill. Finally, two aspects of 116