Assessment of the Distribution of Toxic Release Inventory Facilities in Metropolitan Charleston: An Environmental Justice Case Study

Abstract

Objectives. We assessed spatial disparities in the distribution of Toxic Release Inventory (TRI) facilities in Charleston, SC.

Methods. We used spatial methods and regression to assess burden disparities in the study area at the block and census-tract levels by race/ethnicity and socioeconomic status (SES).

Results. Results revealed an inverse relationship between distance to TRI facilities and race/ethnicity and SES at the block and census-tract levels. Results of regression analyses showed a positive association between presence of TRI facilities and high percentage non-White and a negative association between number of TRI facilities and high SES.

Conclusions. There are burden disparities in the distribution of TRI facilities in Charleston at the block and census-tract level by race/ethnicity and SES. Additional research is needed to understand cumulative risk in the region.

Toxic Waste and Race in America, published in 1987, was the first comprehensive national report to demonstrate that many people-of-color communities and disadvantaged populations are differentially burdened by environmental hazards and unhealthy land uses¹ These burden disparities lead to exposure disparities, increased health risks, and environmental health disparities.^1–3 Community activists, health advocates, researchers, and public health practitioners are working to address environmental injustice, which is driven by differential power and privilege embedded in how we zone, plan, develop, and regulate.^2,3 Environmental injustice is linked to the historic pattern of exploitation, commodification, and devaluation of place, space, and people, which leads to the production of unhealthy geographies and environmental disparities.^2,3 The environmental justice movement is a social movement that includes activists and advocates struggling for the health of communities affected by disparities in burden, exposure, and environmental health. Since the 1987 report, researchers in environmental justice science (the academic arm of the environmental justice movement) have shown that these disparities continue to exist.¹

Low-income populations and populations of color continue to live in communities that suffer from the exposure and burden of environmental hazards.^1–7 These hazards may include noxious land uses such as incinerators and landfills,⁸ Superfund sites,⁹ Toxic Release Inventory (TRI) facilities,^4–10 sewer and water treatment plants,^5,6 and other locally unwanted land uses.¹¹ This disproportionate burden results in increased exposure to harmful environmental conditions for affected communities. Constant exposure to these harmful conditions results in negative health outcomes, stressed communities, and reduction in quality of life and neighborhood sustainability.^2,12

As noted in many studies, people of color and poor populations exposed to environmental hazards show increased health risks that are heavily influenced by many social factors, including racism and classism, segregation, socioeconomic status (SES), and inequities in zoning and planning.^{1–7,11–16} Studies have also shown that socioeconomic vulnerability contributes to increased health disparities and variation in community health,^17,18 which further enhance the long-term effects of environmental injustice. Along with increased environmental exposure, communities overburdened by environmental hazards are also affected by associated psychosocial stressors.^2,12,14 Such stressors, coupled with inadequate health-promoting infrastructure (e.g., supermarkets, parks, open spaces, medical facilities), reduce the community’s ability to defend against the adverse health consequences of their differential burden and exposure.^2,12,14

Environmental injustice and related disparities continue to plague much of the southern United States, especially in port communities such as New Orleans, Louisiana; Savannah, Georgia; and Charleston, South Carolina. In South Carolina, limited work has been performed to assess and address burden disparities associated with the distribution of environmental hazards and unhealthy land uses. The Port of Charleston, with its 3 port terminals on the Charleston peninsula and another port terminal on the Wando River, is currently the fourth largest port in the country and the busiest in the southeastern United States.¹⁹ In 2000, it brought in more than 5% of total US imports.²⁰ The port has a tremendous impact on the local and state economy, as it generates $3.5 billion a year for the Charleston area and $23 billion statewide. The South Carolina State Ports Authority plans to build a new marine container terminal on the Cooper River (on the site of the former Charleston Navy Base in North Charleston) that will open in 2017.^21,22

Charleston has also become a major port of call for cruise liners, the volume of which tripled in only 3 years.²³ This development may contribute to increased air and water pollution in nearby communities. Furthermore, the Charleston region is undergoing rapid urbanization.²³ Metropolitan Charleston is among the top 100 fastest-growing metro areas in the United States, and population expansion could lead to more use of chemicals. The region has several chemical plants, a coal-fired plant, a paper mill, Superfund sites, an incinerator site, brownfields, and 2 major wastewater treatment plants that discharge into Charleston Harbor, with additional upstream discharges into the Ashley, Wando, and Cooper rivers. The region also has a large amount of car traffic on Interstate 26. Local pollution emissions are also caused by diesel truck traffic, rail traffic, and cargo ship traffic related to the movement of goods in the region, leading to air quality problems in heavily trafficked areas.

The Low Country Alliance for Model Communities (LAMC) is a community-based organization concerned about the differential burden of environmental hazards and unhealthy land use in the Charleston Metropolitan Statistical Area (MSA), particularly North Charleston.²⁴ LAMC organized in 2005 to address environmental justice and health issues in LAMC communities and develop partnerships in South Carolina to revitalize LAMC neighborhoods and other disadvantaged neighborhoods in the region. LAMC created a community–university partnership with the University of South Carolina and other stakeholders to study and address environmental justice and health issues in the region.²⁴ This study is part of a larger effort of the partnership to assess burden disparities associated with TRI facilities and other locally unwanted land uses in this region with significant port traffic.

We assessed spatial disparities in the distribution of TRI facilities in Charleston as part of a community-driven research program to assess the cumulative burden and impact of industrial facilities and unhealthy land use in the Charleston region. We aimed to ascertain whether the racial and SES composition of census tracts with a TRI facility differs from the composition of those that do not have a TRI facility. Examination of the effects of differences in distance from a TRI facility can help determine whether these populations may have potentially higher risks of exposure and negative health outcomes. With this knowledge, LAMC will be able to engage local policymakers in the mitigation of environmental injustice and revitalization of poor and disadvantaged neighborhoods in the Charleston MSA. Furthermore, because there is limited literature on the distribution of TRI facilities in traditional southern US port cities, this research may help environmental justice groups in these communities to develop their own assessment of the spatial distribution and burden of toxic facilities.

METHODS

In 2008, the population of the Charleston MSA, which includes the cities of Charleston and North Charleston, was about 644 000, an increase of 17% from the year 2000.²⁵ The population is 67% White, 28% Black, and 3% Hispanic.²⁵ The median household income in 2008 was $52 302,²⁵ up from $39 491 in 2000.²⁵ In North Charleston, where LAMC is located, the population is 70% Black and 26% White, which is a marked contrast with the racial composition of the Charleston MSA. In 2000, the area that covers the LAMC neighborhoods was nearly 83% Black, with a median household income less than $20 000.

Toxic Release Inventory Data

The US Environmental Protection Agency (EPA), under the authority of the Emergency Planning and Community Right-to-Know Act (42 USC §11004-11049 [1986]), requires that all facilities that use more than 10 000 pounds or process more than 25 000 pounds of any of the 650 TRI chemicals report their releases and waste management strategies.²⁶ We extracted year 2008 TRI data from an EPA database by using the EPA’s TRI Explorer²⁶ and mapped TRI facility locations in the Charleston MSA in ArcGIS 9.3 (Environmental Systems Research Institute, Redlands, CA), a geographic information system (GIS) software, using latitude–longitude coordinates. TRI data indicated that in 2008, South Carolina’s 510 TRI sites released more than 65 million pounds of contaminants into the air, water, and soil,²⁶ whereas the 63 TRI facilities in the Charleston MSA released nearly 17 million pounds of contaminants.²⁶ Therefore, although only 12.4% of the state’s TRI facilities in 2008 were in the Charleston MSA, they were responsible for more than 26% of the TRI chemicals released.

Demographic Data

This project used year 2000 demographic data from US Census Bureau Summary Files 1 and 3.^27,28 Census demographic information is available at 4 different geographic scales: zip code tabulation areas, tracts, block groups, and blocks. We used data at the census-tract and block level to enumerate the characteristics of populations burdened by TRI facilities, focusing on race/ethnicity and SES. These included percentage measures of the following: race/ethnicity (non-Hispanic White, non-Hispanic Black, and non-White), poverty (residents living below the national poverty level), educational attainment (population older than 24 years with < high school education vs those with a college degree), employment status (unemployed), homeownership status (homeowners vs renters), and age (younger than 5 years vs older than 60 years; not reported in this study). We also obtained median household income for each census tract in the Charleston MSA. Many of these variables were used previously to assess demographic disparities in the distribution of noxious facilities.¹¹

Geographic Information Systems Methods

We employed 3 GIS methods to ascertain the demographic profile of populations burdened by TRI facilities in the Charleston MSA: (1) mean distance analysis, (2) spatial coincidence, and (3) proximity analysis. In environmental justice assessments, distance is used to assess how close a population is to an environmental hazard or land use facility and can be compared with distance of another population.^11,29–31 Mean distance analysis is a simple but powerful technique that is used to assess spatial disparities in the distribution of hazards. For the mean distance analysis, we used ArcGIS 9.3 to calculate the mean distance of each TRI facility from the centroid of each census block (smallest census division) in the Charleston MSA.

In the spatial coincidence method, also known as the “unit-hazard coincidence” method,^11,29 we used ArcGIS to evaluate demographic information for census tracts and blocks that contained a TRI facility. Populations within census divisions that contained a TRI facility were considered exposed and those outside were considered unexposed.

Proximity analysis examines the characteristics of populations within specified distances of an environmental hazard using buffer zones.^11,29–31 The buffer zones act as proxies for the areas of impact and are represented as circles around each environmental hazard or locally unwanted land use.³⁰ We regarded the populations within these buffers as exposed and the populations outside the buffers as unexposed.^29,30 We used the ArcGIS 9.3 buffer tool to create 0.5-, 1-, 2-, and 5-km buffers and 0.5-, 1-, and 5-mile buffers around each TRI facility.^11,31 We used block centroids as aggregate population center proxies; for the population of a block to be counted “in” a particular buffer zone, the centroid of that block must have fallen within the buffer zone. If the centroid fell within the buffer zone, ArcGIS 9.3 estimated the demographic and socioeconomic characteristics of burdened populations within the buffers.

Statistical Methods

We used SAS version 9.2 (SAS Institute, Cary, NC) to perform statistical analyses for this study. We applied linear regression models to examine the association between distance to the nearest TRI facility (dependent variable) and SES variables at the census-tract level (independent variables). Additionally, to examine the association of a census tract containing a TRI facility with race/ethnicity and SES, we fitted a logistic regression model. We also used a linear regression model to evaluate the association between the number of facilities and the covariates (race/ethnicity and SES) at the census-tract level.

RESULTS

Figure 1 is a choropleth map (i.e., a map shaded with different colors, patterns, or intensity, used in illustrating geographic variation of any data of interest).³² The choropleth map, created in ArcGIS 9.3, illustrates the distribution of TRI facilities in relation to percentage of non-Whites in the Charleston MSA. A large number of the TRI facilities are located in North Charleston, which has a greater percentage of non-Whites, particularly Blacks, than the rest of the Charleston MSA. To illustrate the geographic variation of TRI facilities by race/ethnicity, we used a specific type of choropleth map known as a quartile map. Quartile maps represent a form of equal interval classification in which the range of possible values are divided into 4 intervals so that each class contains an equal number of values.^32,33 There were no TRI facilities in the lowest quartile, 30 in the second quartile, 22 in the third quartile, and 7 in the fourth quartile. As percentage of non-Whites increased, the number of TRI facilities increased, and then decreased again in the fourth quartile. We created a map for percentage living below the federal poverty level in relation to TRI facilities (not shown) and observed similar results.

FIGURE 1—

Map of Toxic Release Inventory (TRI) facilities (2008) in Charleston Metropolitan Statistical Area (MSA), by percentage of non-White population (2000).

Mean Distance and Spatial Coincidence Analyses

For census blocks in the Charleston MSA with a high proportion of White residents (more than 50%), the mean distance to the nearest TRI facility was 3.3 miles; in primarily non-White census tracts, the mean distance was 2.9 miles. Furthermore, linear regression models resulted in a coefficient of 1.15 (P < .001; Table 1). In the linear regression analysis, in which we adjusted for SES variables (Table 1), census tracts with TRI facilities were predominantly Black or non-White (all P values < .05 after adjustment for multiple testing). Additionally, in census tracts without a TRI facility, there were a larger number of high-SES than low-SES individuals.

TABLE 1—

Association Between Distance to Toxic Release Inventory Facilities, 2008, and Race/Ethnicity and Socioeconomic Status at the Census-Tract Level: Charleston Metropolitan Statistical Area, 2000

Variable	Regression Coefficient	P
% Poverty	−0.032	.004
% Unemployment	−0.031	.42
% With no high school diploma	0.021	.005
% Low income	−0.010	.56
% College graduates	−0.043	< .001
Median household income	−0.00012	< .001
% Renters	−0.038	< .001
% Black	3.99	.17
% White	1.15	< .001

Note. Population data are from 2000 US census; data for Toxic Release Inventory facilities are from 2008.

Proximity (Buffer) Analysis

The 2000 population of the Charleston MSA was 644 000. In this study’s buffer analysis, 87% of this total population lived within 5 miles of a TRI facility. The total Black population in the Charleston MSA was 169 888, or 26% of the total population, and nearly 88% of the total Black population in the Charleston MSA (148 638) lived within 5 miles of a TRI facility. We observed that fewer than 12% of the 192 435 non-White persons residing in the Charleston MSA (22 738) were found in a TRI buffer. Figure 2 shows that in all buffer analyses, P values were less than .001, which indicates that in all buffers, the population of non-White residents was significantly larger than that of White residents.

FIGURE 2—

Results of buffer analysis by percentage of Black and non-White population for Charleston Metropolitan Statistical Area at census tract level, 2000.

Figure 2 also shows an incremental decrease in the percentage of Black and non-White residents in the 0.5-mile, 1-mile, and 5-mile TRI buffers. In the 0.5-mile buffer, the population was 48% Black and decreased by 6% for the 1-mile buffer and by 17% for the 5-mile buffer, indicating that there are higher percentages of Black residents in areas closer to TRI facilities than at distances further from them. The population of non-Whites, which was about 52% in the 0.5-mile buffer, decreased to 47% for the 1-mile buffer and to 36% for the 5-mile buffer. The same trend was observed for the 0.5-km, 1-km, 2-km, and 5-km buffers. In the 0.5-km buffer, the population was 48% Black and 53% non-White. As the buffer distance increased, the proportions of Blacks and non-Whites were reduced by 3% (1-km buffer) and 7% (2-km and 5-km buffers), indicating a general trend of significantly higher numbers of Blacks and non-Whites at distances closer to TRI facilities compared with distances farther from the TRI facilities.

Statistical Analysis

We fitted a logistic regression to evaluate the association between presence in a census tract of a TRI facility and SES and racial/ethnic composition. After we adjusted for multiple testing (adjusted significance level = .0056), percentage of college graduates and percentage of Whites were the only 2 variables found to be significant (Table 2). The log-odds that a census tract contained a TRI facility decreased by 0.47 (odds ratio = 0.66, P < .001) for each 1% increase of college graduates and by 0.29 (odds ratio = 0.75, P = .001) for each 1% increase of White population.

TABLE 2—

Association Between Presence of Toxic Release Inventory Facility in Census Tract and Socioeconomic Status and Race/Ethnicity: Charleston Metropolitan Statistical Area

Variable	Odds Ratio	P
% Poverty	1.11	.2
% Unemployment	2.61	.07
% With no high school diploma	1.05	.37
% College graduate	0.66	< .001
% Low income	0.65	.061
Median household income	1.00	.9
% Renters	1.38	.026
% White	0.75	.001

Note. Population data are from 2000 US census; data for Toxic Release Inventory facilities are from 2008.

To evaluate the association between the number of TRI facilities in each census tract and corresponding SES and race/ethnicity, we used a linear regression model.

For percentage of people living in poverty, the percentage with no high school diploma, the percentage unemployed, and the percentage renting homes, we observed a positive and significant association with the number of TRI facilities. There was a positive and significant association between median household income and the number of TRI facilities, but the estimate was miniscule. Thus, in general, the worse the SES status was for individuals in a census tract, the larger the number of TRI facilities. As for the effect of race, for every 1% increase of White population, the number of facilities decreased by about 6 (Table 3).

TABLE 3—

Association Between Number of Toxic Release Inventory Facilities at the Census-Tract Level and Socioeconomic Status and Race/Ethnicity: Charleston Metropolitan Statistical Area

Variable	B	P
% Poverty	0.10	< .001
% Unemployment	1.00	< .001
% With no high school diploma	0.18	< .001
% College graduate	−0.20	< .001
% Low income	0.22	< .001
Median household income	0.00063	< .001
% Renters	0.40	< .001
% White	−5.84	.009

Note. Population data are from 2000 US census; data for Toxic Release Inventory facilities are from 2008.

DISCUSSION

Using 2000 census data, we found evidence of racial and sociodemographic disparities in the burden of TRI facilities in Charleston, SC. The results of mean distance analyses revealed that non-White populations were located closer to TRI facilities, which may be evidence of spatial disparities in the distribution of facilities that release toxic emissions in the Charleston MSA. In the 0.5-, 1-, and 5-mile buffer zones, we found that 52%, 47%, and 36% of the population was non-White, respectively. We observed a similar trend for the 0.5-, 1-, 2-, and 5-km buffer zones, suggesting that more non-Whites were burdened by and potentially exposed to emissions from TRI facilities than were Whites because of their closer proximity to TRI facilities.

We used logistic regression to evaluate the association between a census tract with a TRI facility and SES and race/ethnicity at the census-tract level. Percentage of college graduates and percentage of Whites were the only statistically significant variables that showed a negative relationship—as percentage of college graduates or of Whites increased in each census tract, the odds of that census tract having a TRI facility decreased. In addition, we used linear regression to evaluate the association between the number of TRI facilities in each census tract and corresponding SES and race/ethnicity. For several variables, including percentage living in poverty, percentage with no high school diploma, percentage unemployed, and percentage renting, we observed a positive and significant association with the number of TRI facilities. As the number of individuals with low SES increased, the number of TRI facilities at the census-tract level also increased. We observed the opposite relationship between percentage of Whites and the number of TRI facilities. These results indicate the role that race/ethnicity and socioeconomic composition play in whether a block or census tract will have a TRI facility and are an indication of burden disparities for low-SES populations as well as non-Whites in the Charleston MSA. These results are similar to results presented in the environmental justice literature.^1,4,10–14 Furthermore, they present a case for exploring the cumulative burden and impact of all noxious facilities in the region and potential linkages to environmental health disparities.

We believe that the approach outlined in this article is generalizable to other contexts where disadvantaged communities are affected by environmental injustice. This work adds to general environmental justice science as a case study for community-based environmental justice organizations to follow in the assessment of racial and SES disparities in facility location and potential use of results for equitable planning and development efforts. Maps showing burden disparities can be used by local residents in their efforts to show disproportionate impact and possible negative health outcomes caused by this burden. These maps may be even more useful when historic GIS is used to show changes in the density of pollution-emitting facilities like TRI facilities, chemical plants, and other environmental hazards and related emissions and their burden on disadvantaged and underserved communities affected by environmental injustice.

Limitations

This study has some limitations. We used 2008 TRI data and 2000 census data, which could have introduced some burden misclassification, and results provide only a snapshot of burden disparities of TRI facilities in the Charleston MSA. It is important to look retrospectively at both changes in the TRI distribution over time and changes in population demographic indices. This study also fell short in the inclusion of nonpopulated land features in the analysis. The exclusion of water features, roads, unpopulated blocks, or higher-level census divisions that included military installations might have improved the burden disparities assessment illustrated by this study.

An additional limitation was the focus on facility location but not emissions. Previous research has shown that to understand burden and exposure disparities, it is important to examine emissions from the facilities, including pollutant concentrations. In addition, other studies have used Risk Screening Environmental Indicators (RSEI) and National-Scale Air Toxics Assessment (NATA) data instead of TRI facility location and emissions data.^34,35 RSEI and NATA data have been shown to better represent the toxic potential of local emissions and may be a better measure of burden for populations who live near TRI facilities. We concur, and believe that examining facility distribution is just the first step in understanding burden disparities and cumulative impact.

Traditionally, researchers use census divisions such as census block groups and census tracts as proxies for neighborhoods.^36,37 This is because of the wealth of demographic data that is available at these divisions and that can be aggregated or disaggregated, allowing for GIS analyses at different levels. However, in the case of communities and neighborhoods affected by environmental injustice, the use of census divisions may not be appropriate in all cases. A neighborhood can overlap many blocks or block groups, or several tracts. In the case of affected communities in the Charleston MSA, such as the LAMC neighborhoods, the use of census tracts may not accurately capture the population burdened by TRI facilities. In future work, spatial techniques will be used to perform neighborhood-level analyses for the Charleston MSA.

Conclusions

This study has shown that there are burden disparities in the distribution of TRI facilities in the Charleston MSA at the block and census-tract levels, across varying levels of racial/ethnic composition and SES. Even with a few methodological limitations, this study found statistically significant associations between (1) block-level distribution and number of TRI facilities and race/ethnicity and (2) presence or absence of TRI facilities and non-White, renter, poverty, educational, and employment status at the census-tract level. We are confident that LAMC will be able to use the results of this study, particularly the burden maps, to influence local environmental decision-making, particularly around revitalization efforts in the region. In future work, we will explore the use of RSEI and NATA data for the assessment of burden and exposure disparities in the region. We will also employ other spatial methods, including cumulative distance function and geographically weighted regression, at different census levels to assess burden disparities.

In addition, we believe that this study contributes to environmental justice science and provides information that may be useful for other port cities in their local environmental decision-making and community planning and development. These findings are unique and timely given that the expansion of the Panama Canal in 2014 will open more southeastern US ports to increased port traffic and the additional industrial development that often follows. This study provides those communities an opportunity to learn from the lessons of Charleston so that the environmental disparities present there will not be replicated in other southeastern port communities.