Quantifying and Characterizing
Near-Port Populations in the
Conterminous United States

National Results Using High-Resolution
Population Data from EPA's EnviroAtlas

SEPA

United States
Environmental Protection
Agency


-------
Quantifying and Characterizing
Near-Port Populations in the
Conterminous United States

National Results Using High-Resolution
Population Data from EPA's EnviroAtlas

This technical report does not necessarily represent final EPA decisions
or positions. It is intended to present technical analysis of issues using
data that are currently available. The purpose in the release of such
reports is to facilitate the exchange of technical information and to
inform the public of technical developments.

Transportation and Climate Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency

NOTICE

United States
Environmental Protection
Agency

EPA-420-R-24-021
December 2024


-------
Executive Summary

Marine ports are critical for commerce and economic growth, and they play a significant role in
the goods movement supply chain. However, port-related activity often contributes to air pollution
emissions from a variety of diesel-powered mobile sources that operate in port areas such as trucks,
locomotives, cargo handling equipment, harbor craft, and ocean-going vessels. These sources can have
important impacts on local and distant air pollution, including fine particulate matter, nitrogen oxides,
air toxics, which are associated with increased risk of adverse health outcomes among those who are
exposed, as well as carbon dioxide emissions. Communities near ports may be exposed to harmful local
emissions related to port activity and may benefit from efforts to make ports cleaner. However, the
number of people living near ports and the demographics of these populations remains poorly defined,
in part due to the complex and numerous ways to describe a port and port operations. In this study, we
identified populations near ports in a more sophisticated manner than traditional proximity analyses by
leveraging a 2010 high-resolution population dataset of the conterminous United States (CONUS, the
lower 48 states and the District of Columbia) and port geometries from two different Federal agencies.
We also characterized the sociodemographic attributes of these near-port populations to identify
potential disproportionalities in community demographics that may be indicative of potential
environmental justice (EJ) concerns for near-port populations.

Depending on the port boundaries used, at least 16.1M or 31.1M people live within
5000m of major ports in the conterminous U.S.

A total of 123 major ports in the conterminous United States were investigated in this study
based on their inclusion in the U.S. Army Corps of Engineers (ACE) 2017 Principal Ports list of the top 150
ports by tonnage throughput and their representation in two different Federal datasets: the EPA's
National Emissions Inventory (NEI) and ACE's Master Docks Plus. By overlapping 2010 population data
with either the NEI or ACE port geometries, we estimated that 16.1 or 31.1 million individuals live within
5000m (~3.1 mi.) of ports included in this study. We also estimated that 2.6 or 4.7 million individuals live
within 1000m (~0.6 mi.). We caution that these values likely underestimate the total number of people
impacted by port operations, as the precise numbers presented in this study are highly dependent on
the port geometry used, the set of ports included, and the distance used to define 'near port'.

Near-port populations have higher shares of sociodemographically vulnerable groups
than comparison populations.

We also assessed the sociodemographic characteristics of these near-port populations based on
variables that are widely used across published EJ tools and represent a wide range of vulnerabilities.
Using two different comparison groups, we identified several vulnerable sociodemographic groups that
are overrepresented in the near-port population. For the racial and ethnic groups quantified, there were
higher percentages of Hispanic, Non-Hispanic Black, Non-Hispanic Asian, and people of color in the near-
port populations as compared to neighboring populations and the general population of CONUS. We
also detected disproportionalities among near-port populations for several socioeconomic factors,


-------
including the percentages of individuals living below twice the poverty level, renters, individuals living in
areas of persistent poverty, adults with less than a high school education, households in linguistic
isolation, and households living below a Quality of Life income threshold. The differences in
sociodemographic characteristics between the near-port populations and their comparison groups were
surprisingly consistent using either the NEI or ACE port geometries. This consistency increases
confidence in the overall takeaway that certain sociodemographically vulnerable groups are
overrepresented in proximity to major ports in CONUS.

These national near-port demographic disproportionalities are not driven byjust a few
ports, but instead point to broader trends around ports.

We also conducted a supplemental analysis in which we subset the 123 ports included in this
study into the top 10 ports by tonnage and the remaining 113 ports. The purpose of this supplemental
analysis was to understand whether the sociodemographic patterns observed in the primary
analysis were driven solely by the busiest ports by tonnage, which also aligned with major metro areas.
Further, we sought to understand if there were meaningfully different disproportionalities or
sociodemographic characteristics surrounding the top 10 busiest ports, which may also contribute the
most emissions related to port activity. In general, the national results were supported by the
supplemental analysis, with some nuances depending on the port geometry or comparison group that
was used. Further, we found that the populations living within 5000m of the top 10 ports accounted for
30-40% of the total near-port population around all 123 major ports that were included in this study.
Therefore, actions taken to lower emissions at these top 10 ports with the largest tonnage throughput
may have an outsized impact on the nation's near-port population.

A key challenge of this work is the complexity of mapping and defining port operations
geospatially.

Across the spectrum of port-related federal activities conducted by EPA, U.S. Department of
Transportation (DOT), ACE, and others (including grants and programmatic work), there are varying
definitions for a 'port' that come from statutory language or administrative requirements. In addition to
there being multiple definitions of 'port' in use, there is also not a single authoritative source for the
geospatial extent of U.S. ports or the extent over which vehicles and equipment serving a port may
operate. This study navigates these complexities by using the best available data from two different
agencies (ACE and EPA) for the same set of 123 ports and by making simplifying assumptions to capture
nearby impacts of port activities. The total near-port population estimates from this study varied
substantially depending on the port geometry used, which underscores the critical role of source
geometry in proximity analyses.

iii


-------
Table of Contents

Executive Summary	ii

List of Figures	v

List of Tables	vi

List of Acronyms & Abbreviations	vii

1.	Introduction	1

2.	Methods and Approach	3

2.1	Port Geospatial Data	3

U.S. Army Corps of Engineers Master Docks Plus Shapefiles	4

EPA Port Polygons from the National Emissions Inventory	5

Port Inclusion Criteria	6

2.2	Geodesic Buffers Around Port Geometry to Describe Near-Port Populations	8

2.3	Higher Resolution Population Estimates	9

2.4	Sociodemographic Variables	10

2.5	Comparison Groups	13

Conterminous United States	14

Intra-County Comparison Group	14

2.6	Developing a National Analysis	15

2.7	Top 10 Ports Supplemental Analysis	16

3.	Results	18

3.1	National Estimates of Near-Port Populations and Comparison Groups	18

3.2	National Disproportionalities of Near-Port Populations	19

Overrepresentation of People of Color in Near-Port Populations compared to Comparison Groups. 22

Multiple Metrics of Income Point to Economic Disproportionalities between Near-Port Populations &
Comparison Groups	24

No detectable disproportionalities among vulnerable age groups in Near-Port Populations	27

3.3	Top 10 Ports Supplemental Analysis Supports Conclusions of National Analysis	27

Estimates of Near-Port Populations & Comparison Groups	27

Disproportionalities of Near-Port Populations	32

4.	Discussion	37

4.1	Summary of Results	37

4.2	Differences between the EPA and ACE Shapefiles and Impact on Population Totals	38

iv


-------
4.3 Other Study Limitations	40

5. Conclusion	41

Works Cited	42

Appendix	46

A.	List of Ports Included in Study	46

B.	Summary of Geospatial Port Data Sources	47

C.	Summary of Port Definitions from Various Federal Agencies and Programs	48

Environmental Protection Agency	48

U.S. Department of Transportation	48

U.S. Army Corps of Engineers	51

D.	Additional Notes about the Dasymetric Model	51

E.	Data Processing	52

F.	Supplemental Figures	53

G.	Alternative Comparison Groups Considered	56

Neighboring Block Groups	56

Rural-Urban Continuum Codes	57

Balance of State Population	57

H.	Authors and Acknowledgements	57

List of Figures

Figure 1. Map of Boston, MA with port geometries from ACE (red points) and EPA (yellow polygons)	4

Figure 2. Map of 123 ports featured in study. Each unique port geometry has been presented as a single point for

SIMPLIFICATION OF VIEWING PORT LOCATIONS USED IN THIS STUDY ACROSS THE CONTERMINOUS U.S	7

Figure 3. Schematic of how the subset of ports was selected for this study. A full list of the 123 ports used in this study
is shown in Table A-l. List of ports included in study (n=123)	8

Figure 4. Illustration of the dasymetric model using a census block near Sacramento, CA with a cemetery, and

RESIDENTIAL HOUSING ALONG THE EASTERN BORDER. THE ENVIROATLAS DASYMETRIC METHOD ALLOCATES ZERO POPULATION TO
THE CEMETERY AND DENSER POPULATION ALONG THE EASTERN BORDER. ADAPTED FROM FIGURE 6 WITHIN BAYNES ET AL., 2022.10

Figure 5. Schematic to illustrate Intra-County Comparison Group. For simplicity to illustrate the approach that was

USED TO DEFINE THE INTRA-COUNTY COMPARISON GROUPS, BLOCK GROUPS ARE SHOWN AS SQUARE POLYGONS, AND PORT BUFFERS
ARE SHOWN AS CIRCLES. IN REALITY, THE BLOCK GROUPS, COUNTY BOUNDARIES, AND PORT BUFFERS THESE ARE IRREGULARLY SHAPED

POLYGONS	14

Figure 6. Map of the top 10 ports featured in supplemental analysis (as dark triangles) and remaining 113 ports (light

BLUE CIRCLES) BY TONNAGE IN THIS STUDY. EACH UNIQUE PORT GEOMETRY HAS BEEN PRESENTED AS A SINGLE ICON FOR
SIMPLIFICATION OF VIEWING PORT LOCATIONS USED IN THIS STUDY ACROSS THE CONTERMINOUS U.S	17

Figure 7. The relative share of tonnage by ports contributing >1% relative share, among the 123 ports featured in this
STUDY, BASED ON ACE PRINCIPAL PORT DATA FROM 2010-2019	18

V


-------
Figure 8. Comparison of the percentage of people of color between near-port populations and comparison groups for

123 PORTS INCLUDED IN THE PRIMARY ANALYSIS	22

Figure 9. Pyramid plot of the percentage of the population belonging to selected racial and ethnic groups by near-port

POPULATIONS AND COMPARISON GROUPS	23

Figure 10. Comparison of percentage of the population living below twice the poverty threshold between near-port

POPULATIONS AND COMPARISON GROUPS FOR 123 PORTS INCLUDED IN THE PRIMARY ANALYSIS	24

Figure 11. Comparison of median household income between near-port populations and comparison groups for 123

PORTS INCLUDED IN THE PRIMARY ANALYSIS	25

Figure 12. Pyramid plot of the percentage of the population belonging to selected socioeconomic groups by near-port

POPULATIONS AND COMPARISON GROUPS	26

Figure 13. Comparison of Quality of Life index, equal to the percentage of the population living below the Quality of
Life income threshold, between near-port populations and comparison groups for 123 ports included in the

PRIMARY ANALYSIS	27

Figure 14. Comparison of Very Near Port, Near Port, and Intra-County Comparison Group populations for the

SUPPLEMENTAL ANALYSIS	28

Figure 15. Comparisons by race/ethnicity between near-port populations and comparison groups for the 123, top 10,

AND REMAINING 113 PORTS. THE DIFFERENCE IN PERCENTAGE BY RACE/ETHNICITY BETWEEN THE NEAR-PORT POPULATIONS AND
COMPARISON GROUP IS PRINTED NEXT TO EACH BAR; BARS WITH POSITIVE VALUES INDICATE A HIGHER PERCENTAGE OF THAT
DEMOGRAPHIC GROUP IN THE NEAR-PORT POPULATIONS THAN THE COMPARISON GROUP, WHILE NEGATIVE VALUES INDICATE A
HIGHER PERCENTAGE IN THE COMPARISON GROUP. ONLY DEMOGRAPHIC CHARACTERISTICS WITH DIFFERENCES IN PERCENTAGE
GREATER THAN 1% PT. IN THE PRIMARY ANALYSIS ARE SHOWN	33

Figure 16. Comparisons by poverty-related factors between near-port populations and comparison groups for the 123,

TOP 10, AND REMAINING 113 PORTS. THE DIFFERENCE IN PERCENTAGE BETWEEN THE NEAR-PORT POPULATION AND COMPARISON
GROUP IS PRINTED NEXT TO EACH BAR; BARS WITH POSITIVE VALUES INDICATE A HIGHER PERCENTAGE OF THAT GROUP IN THE NEAR-
PORT POPULATION THAN THE COMPARISON GROUP, WHILE NEGATIVE VALUES INDICATE A HIGHER PERCENTAGE IN THE COMPARISON
GROUP. ONLY SOCIOECONOMIC CHARACTERISTICS WITH DIFFERENCES IN PERCENTAGE GREATER THAN 1% PT. IN THE PRIMARY

ANALYSIS ARE SHOWN	35

Figure 17. Comparisons by other socioeconomic between near-port populations and comparison groups for the 123, top

10, AND REMAINING 113 PORTS. THE DIFFERENCE IN PERCENTAGE BETWEEN THE NEAR-PORT POPULATION AND COMPARISON
GROUP IS PRINTED NEXT TO EACH BAR; BARS WITH POSITIVE VALUES INDICATE A HIGHER PERCENTAGE OF THAT GROUP IN THE NEAR-
PORT POPULATION THAN THE COMPARISON GROUP, WHILE NEGATIVE VALUES INDICATE A HIGHER PERCENTAGE IN THE COMPARISON
GROUP. ONLY DEMOGRAPHIC CHARACTERISTICS WITH DIFFERENCES IN PERCENTAGE GREATER THAN 1% PT. IN THE PRIMARY
ANALYSIS ARE SHOWN	36

Figure 18. Comparison of median household income between near port populations and comparison groups for the top

10 PORTS	37

Figure 19. Comparison of Quality of Life index, equal to the percentage of the population living below the Quality of

Life income threshold, between near port populations and comparison groups for the top 10 ports	37

Figure 20. Map comparing the differences between the EPA (left) and ACE (right) port geometries and the resulting

IMPACT ON DIFFERENCES IN THE EXTENT OF BLOCK GROUPS IN NEAR PORT POPULATIONS (SHOWN IN LIGHT BLUE) AND THE EXTENT
OF THE INTRA-COUNTY COMPARISON GROUPS (SHOWN IN PEACH) USING CHICAGO AS AN EXAMPLE. NOTE BLOCK GROUPS WITH
ZERO POPULATION ARE INCLUDED IN THE FIGURE ABOVE	39

Figure 21. Visualization of the number of near-port block groups captured by the ACE port geometries, the EPA port

GEOMETRIES, OR BOTH	39

List of Tables

Table 1. Table of the variables that were of interest in this study by demographic category and data source,

12


-------
Table 2. Summary of 2010 Populations within Very Near and Near Port Population and Comparison Groups by Port

Geometry (ACE or EPA)	19

Table 3. National Estimates of Near-Port Populations by Buffer Distance from Port and Port Geometry (ACE or EPA).

	19

Table 4. Summary of Differences in Percentage of Sociodemographic Groups between Very Near Port Populations and

Comparison Groups (within 1000m)	20

Table 5. Summary of Differences in Percentage of Sociodemographic Groups between Near Port Populations and

Comparison Groups (within 5000m)	21

Table 6. Sociodemographic Characteristics of Very Near Port Populations (within 1000m)	30

Table 7. Sociodemographic Characteristics of Near Port Populations (within 5000m)	31

List of Acronyms & Abbreviations

Analytical Tools Interface for Landscape Assessments ATtlLA

vii


-------
1. Introduction

Marine, coastal, river, and Great Lake ports are vital to the Nation's economy, serving as pivotal links
in local, regional, and global supply chains. Ports rely on a variety of vessels, vehicles, and other mobile
equipment to move passengers and cargo to and from shore and onto the next link in the supply chain.
The mobile equipment serving ports, including trucks, locomotives, cargo handling equipment, and
vessels, are typically diesel-powered. Consequently, port-related activity often contributes to air
pollution emissions and can have important impacts on local air pollution, including fine particulate
matter, nitrogen oxides, air toxics, and carbon dioxide, in addition to impacts farther downwind.1 These
and other air pollutants are associated with increased risk of adverse health outcomes. Research
indicates that people living in close proximity to mobile sources and those who are exposed to higher
concentrations of mobile source- and traffic-related air pollution have higher rates of adverse health
outcomes, including asthma onset and acute respiratory infections in children, adverse birth outcomes
(e.g., small for gestational age, childhood leukemia, asthma onset and lung cancer in adults, and
premature death.2'3'4'5'6,7 Moreover, many studies have found that air pollution, including from mobile
sources, is higher in areas where people of color and low-income populations represent a higher fraction

1	For more information about ports emissions, see U.S. EPA Ports Initiative (2022) Port and Goods Movement
Emission Inventories

2	Laden, F., Hart, J., Smith, T., Davis, M., & Garshick, E. (2007). Cause-specific mortality in the unionized U.S.
trucking industry. Environmental Health Perspectives, 115(8), 1192-1196. doi: 10.1289/ehp. 10027

3	Peters, A., von Klot, S., Heier, M., Trentinaglia, I., Hormann, A., Wichmann, H., & Lowel, H. (2004). Exposure to
traffic and the onset of myocardial infarction. New England Journal of Medicine, 351(17), 1721-1730. doi:
10.1056/NEJMoa040203

4	Zanobetti, A., Stone, P., Spelzer, F., Schwartz, J., Coull, B., Suh, H., Nearling, B.D., Mittleman, M.A., Verrier, R.L., &
Gold, D. (2009). T-wave alternans, air pollution and traffic in high-risk subjects. American Journal of Cardiology,
104(5), 665-670. doi: 10.1016/i.amicard.2009.04.046

5	Adar, S., Adamkiewicz, G., Gold, D., Schwartz, J., Coull, B., & Suh, H. (2007). Ambient and microenvironmental
particles and exhaled nitric oxide before and after a group bus trip. Environmental Health Perspectives, 115(4),
507-512. doi: 10.1289/ehp.9386

6	Boogaard, H., Patton, A., Atkinson, R., Brook, J., Chang, H., Crouse, D., Fussell, J.C., Hoek, G., Hoffmann, B.,
Kappeler, R., Kutlar Joss, M., Ondras, M., Sagiv, S.K., Samoli, E., Shaikh, R., Smargiassi, A., Szpiro, A.A., Van Vliet,
E.D.S., Vienneau, D., Weuve, J., Lurmann, F.W., Forastiere, F. (2022). Long-term exposure to traffic-related air
pollution and selected health outcomes: A systematic review and meta-analysis. Environment International, 164,
107262. doi: 10.1016/i.envint.2Q22.107262

7	Boothe, VL.; Boehmer, T.K.; Wendel, A.M.; Yip, F.Y. (2014) Residential traffic exposure and childhood leukemia: a
systematic review and meta-analysis. Am J Prev Med 46: 413-422. doi: 10.1016/i.amepre.2013.11.004

1


-------
of the population compared against the general population.8,910 11121314 15 Given the volume of mobile
source activity at ports, there is a need to quantify and characterize populations that may be affected by
emissions at port facilities.

The primary objective of this study is to meet this need by answering three questions:

1.	How many people live in close proximity to major U.S. port operations and their associated
mobile sources of air pollution?

2.	What are the racial, ethnic, age, and socioeconomic characteristics of people living in close
proximity to major U.S. port operations?

3.	Are there disproportionalities between people who live near U.S. ports compared to those who
do not with respect to their sociodemographic characteristics?

To answer these questions, we have conducted a near-port proximity analysis. Proximity analyses
are a common approach used by the EPA to estimate the size of a potentially affected community and
screen for potential environmental justice (EJ) concerns.1617 Proximity analyses use distance from a
source as a proxy for risk or exposure when actual observations or models of risk or exposure are not
readily available. Therefore, we do not quantify risk from or exposure to port-related diesel activity or
air pollution in this study. To use distance from a source as a proxy for risk or exposure, proximity
analyses have two underlying assumptions: 1) those within a specified distance of the source experience
different conditions than those beyond that distance and 2) the underlying data appropriately represent
the source. In this study, we addressed these assumptions by exploring how different distances and port
geometries used to define 'near-port' impacted our results.

8	Rowangould, G. M. (2013). A census of the near-roadway population: public health and environmental justice
considerations. Transportation Research Part D: Transport and Environment, 25, 59-67. doi:

10.1016/i.trd.2013.08.003

9	Marshall, J. D. (2008). Environmental inequality: Air Pollution exposures in California's South Coast Air Basin.
Atmospheric Environment, 42(21), 5499-5503. doi: 10.1016/i.atmosenv.2008.02.005.

10	Mohai, P., Pellow, D., & Roberts, J. T. (2009). Environmental Justice. Annual Review of Environment and
Resources, 34, 405-430. doi: 10.1146/annurev-environ-082508-094348.

11	Jbaily, A., Zhou, X., Liu, J., Lee, T.-H., Kamareddine, L, Verguet, S., & Dominici, F. (2022). Air pollution exposure
disparities across US population and income groups. Nature, 601(7892), 228-233.

doi: 10.1038/s41586-021-04190-v.

12	Collins, T. W., & Grineski, S. (2022). Racial/Ethnic Disparities in Short-Term PM2.5 Air Pollution Exposures in the
United States. Environmental Health Perspectives, 130(8). doi: 10.1289/EHP11479

13	Weaver, G.M., & Gauderman, W.J. (2018). Traffic-Related Pollutants: Exposure and Health Effects Among
Hispanic Children. American Journal of Epidemiology, 187(1), 45-52. doi: 10.1093/aie/kwx223.

14	Tessum, C. W., Paolella, D. A., Chambliss, S. E., Apte, J. S., Hill, J. D., & Marshall, J. D. (2021). PM2.5 polluters
disproportionately and systemically affect people of color in the United States. Science Advances, 7(18). doi:
10.1126/sciadv.abf4491

15	Valencia, A., Serre, M., & Arunachalam, S. (2023). A hyperlocal hybrid data fusion near-road PM2.5 and N02
annual risk and environmental justice assessment across the United States. PLOS ONE. doi:

10.1371/iournal. pone.0286406

16	U.S. EPA (2016), "Technical Guidance for Assessing Environmental Justice in Regulatory Analysis"

17	U.S. EPA (2023), "DRAFT Technical Guidance for Assessing Environmental Justice in Regulatory Analysis"

2


-------
There is no single, united definition of what a port is18, nor does the port-related activity end at the
port's gate. Vessel anchorage areas may be miles downriver or offshore from a port; dray trucks may
use local roads to reach nearby warehouses or queue to enter the gate; and locomotives may haul cargo
to and from nearby railyards. Typically, studies examining near-port populations have either relied on a
patchwork of representative point locations to describe ports or have focused on a small sample of
ports. For example, Greenburg (2021) used a combination of port names, coordinates, and aerial
photography to develop point locations for 50 ports across the U.S., and analyzed the population within
2, 5, and 10 miles (~3.2, ~8.0, and ~16.1km) of the centroids of these points.19 Rosenbaum et al. (2011)
analyzed health risk disparities of near-port communities but did not expound on how the 43 harbor
areas in the study were defined geospatially.20 Several Federal agencies have developed geospatial
representations of ports in pursuit of mission-specific purposes; however, the use of these geospatial
data in near-port studies is limited. One notable exception is Gillingham and Huang (2022), who used
coordinates of 27 coastal ports sourced from the U.S. Army Corps of Engineers (ACE) as part of their
study on racial disparities in the health effects from air pollution at ports.21

We emphasize that the plurality of ways to define ports as described above is an important
challenge of conducting a near-port proximity analysis, and there is a need for more unified port
geometries or on-the-ground efforts to corroborate our findings in the future. Nevertheless, this study is
a first effort to quantify and characterize the near-port population in the U.S., and we have conducted a
robust analysis that leverages a uniquely high-resolution population data set and port geometries from
two different Federal agencies. Using this approach, we generated a range of near-port population
estimates that are likely to underestimate the total number of people living near port operations.
Furthermore, we were able to characterize the near-port population and, using two comparison groups,
identify several vulnerable sociodemographic groups that are overrepresented in the near-port
population.

2. Methods and Approach
2.1 Port Geospatial Data

This study required high quality geospatial representations of ports due to the nature of
proximity analyses, which use distance to a source as a proxy for potential risk or exposure from primary

18	There is no single, government-wide definition for what constitutes a port, and various federal agencies have
defined ports differently depending on their mission. For purposes of this analysis, we define a port as "places
alongside navigable water with facilities for the loading and unloading of passengers and/or cargo from ships,
ferries, and other vessels". This is consistent with the definition in the EPA Ports Initiative Primer for Ports and
recent Diesel Emission Reduction Act Grant program requests for applications (see EPA Ports Primer: Glossary and
2021 DERA RFA I. B.7.b.2). For additional definitions of 'port' used by Federal programs, see Appendix C.

19	Greenburg, M. R. (2021). Ports and Environmental Justice in the United States: An Exploratory Statistical
Analysis. Risk Analysis, 41(11). doi: 10.1111/risa. 13697

20	Rosenbaum, A., Hartley, S., & Holder, C. (2011). Analysis of Diesel Particulate Matter Health Risk Disparities in
Selected US Harbor Areas. American Journal of Public Health, 101(S1), S217-S223. doi:10.2105/AJPH.2011.300190

21	Gillingham, K., & Huang, P. (2021). Racial Disparities in the Health Effects from Air Pollution: Evidence from Ports.
National Bureau of Economic Research. doi:10.3386/w29108

3


-------
air pollutants. However, as there is not a single, unified geospatial dataset for all U.S. ports, we used
shapefiles published by two different federal agencies: U.S. EPA (EPA) and U.S. Army Corps of Engineers
(ACE). These distinct shapefiles describe ports geospatially as a series of polygons (in the case of EPA) or
as a series of point locations (in the case of ACE). As Figure 1 illustrates, these two datasets represent
port locations very differently from one another in pursuit of each agency's specific mission and are not
official designations of any port authority's jurisdiction or a terminal operator's area of control.
Proximity analyses depend on the geospatial representation of the source of interest. Our choice to use
these two port geometries offers an opportunity to demonstrate the implications of source geometries
on the results of proximity analyses. For the purposes of this study, we assume that these shapefiles
may approximate areas where port operations and associated mobile source emissions may occur,
including port-related emissions from vessels, cargo-handling equipment, locomotive, and some dray
truck operations. However, neither port geometry is expected to perfectly capture the extent of port
operations or all ports in the U.S. Therefore, we emphasize that the resulting near-port populations are
likely underestimations of the total number of people potentially impacted by port operations and that
our results should be corroborated with on-the-ground experience.

Winthrop

Harbor Islands
\

86 ft

Boston Harbor
Islands St Park

Long
Island

Pemberton
P«int^

Figure 1. Map of Boston, MA with port geometries from ACE (red points) and EPA (yellow polygons).

Bayside

U.S. Army Corps of Engineers Master Docks Plus Shapefiles

U.S. Army Corps of Engineers (ACE) Institute of Water Resources maintains the database Master
Docks Plus22, which provides geospatial data for over 40,000 port and waterway facilities along coastal,

22 U.S. Army Corps of Engineers Navigation and Civil Works Decision Support Center, Waterborne Commerce
Statistics Center. "Master Docks Plus", Accessed May 2019. https://www.iwr.usace.armv.mil/About/Technical-

Centers/WCSC-Waterborne-Commerce-Statistics-Center-2/WCSC-Navigation-Facilities/

4


-------
Great Lakes, and inland ports across the U.S. This dataset covers a significant geographic extent using
latitude and longitude coordinates and is widely used across various federal agencies.

For this study, we used a May 2019 query of Master Docks Plus, featuring 41,697 distinct
waterway facilities. The extract was filtered to isolate only dock facilities, which we assumed represent
points where vessels come to land, and thus represent areas where goods or people move from water
to shore. All other facility types, such as locks, dams, mile point markers, and virtual docks, were
removed from the dataset, as they were considered to not be immediately part of the goods movement
activity occurring at ports. To focus on 'active' docks, we only included those associated with ports for
which 'Service Terminated' was not equal to 'Yes'. The final dataset included 15,320 active dock
locations associated with 834 distinct port IDs. These docks were represented geospatially as points,
with latitude and longitude coordinate data and port IDs, five-digit alphanumeric strings unique to each
port in the database The port IDs associated with dock coordinates were used to help match these
coordinates to the same ports in the EPA shapefiles.

EPA Port Polygons from the National Emissions Inventory

For this study, we used a composite of the 2011 and 2014 port shapefiles used in the 2011 and
2014 National Emissions Inventories (NEI), totaling 534 polygons representing 404 ports in CONUS.
These port shapefiles were first developed by Eastern Research Group under contract by EPA's Office of
Air Quality Planning and Standards for the 2008 NEI; they were then updated for the 2011 and 2014 NEI
publications and have not been updated since due to methodology changes.23 The purpose of these
shapefiles was to aid in the allocation of emissions from marine vessels to specific counties. The original
shapefiles were developed using a variety of resources including GIS shapefiles provided directly from
ports, maps or port descriptions from local port authorities, satellite imagery from tools such as Google
Earth and street layers from StreetMap USA, and feedback from the U.S. Army Corps of Engineers.24

The 2011 iteration of these shapefiles attempted to approximate landside boundaries of ports
that reported Marine Engine Category 3 vessel activity. The demographics analysis team determined
that these were the approximate areas where significant cargo handling equipment, rail, and drayage
truck activity occurs, as well as the docking locations for vessels visiting the port. A total of 381 polygons
representing 337 unique ports were included in the 2011 NEI. A subset of these (n=211 ports) were
represented as circles with a quarter mile radius (n=217 circles).

The 2014 NEI took a different approach, replacing all landside and circular boundaries with
simplified waterside boundaries. The goals of this effort were to generate a file that represented
primary areas where vessel hoteling25 and maneuvering activities were conducted and to simplify and

23	The 2011 National Emissions Inventory represents modeled emissions from 2011 and the port shapefile was
published in August 2014. The 2014 National Emissions Inventory represents modeled emissions from 2014 and
the port shapefile was published in May 2017. The 2017 National Emissions Inventory, published in April 2020,
uses the same port shapefile as the 2014 National Emissions Inventory.

24	Eastern Research Group, 2010. Project report: Documentation for the Commercial Marine Vessel Component of
the National Emissions Inventory Methodology. Eastern Research Group No. 0245.02.302.001, March 30, 2010.
Available via 2008 NEI Reference List as "cmv_report4.pdf".

25	Accurately capturing landside port boundaries was not the intent of the EPA NEI port shapefiles, but the 2011
dataset nevertheless contains some polygons stretching landside. The purpose of the NEI port shapefiles is to help
allocate marine vessel emissions to specific counties for totalling emissions.

5


-------
process the shapes so that they would be more suitable for modeling. The resulting shapefile used 914
polygons to represent 489 unique ports.

The composite shapefile used for this study was created by compiling ports represented by
landside boundaries from NEI 2011 polygons and remaining ports represented by waterside boundaries
from NEI 2014 polygons. Landside boundaries (NEI 2011) were considered preferable to waterside (NEI
2014) boundaries because nearby populations residing on land can be exposed to emissions from ports'
landside operations - such as dray trucks and cargo handling equipment - in addition to marine vessel
emissions. For the purposes of this study, the circular polygons included in the 2011 NEI were
considered a lower-quality depiction than the specific shapes that were used for every port in 2014 and
were removed in the composite file. Combining the two versions of NEI allowed for a larger number of
port boundaries to be represented geospatially. The shapefile combining two versions of NEI contains
164 NEI 2011 landside polygons representing 126 ports and 480 NEI 2014 waterside polygons
representing 363 ports.

Port Inclusion Criteria

This study focused on a subset of ports that are represented in both the EPA and ACE datasets
and, to align with the population data used in this study (see Section 2.3: Higher Resolution Population
Estimates section for more), fall within the conterminous U.S. (CONUS, the adjoining 48 states and the
District of Columbia).26 To further narrow the scope to ports that likely have the most diesel-related
activity and therefore represent the greatest mobile source emissions and potential impact on air
quality, this study only included ports with the highest tonnage levels, using the ACE 2017 Principal Ports
list27 to determine tonnage throughput. This approach, using port tonnage as a proxy for air quality
impacts, has been previously described by Gillingham and Huang (2021).28 Using these criteria, 123 ports
were selected for inclusion in this study. A map of these ports is shown in Figure 2, a complete list of the
123 ports can be found in Table A-l, and a high-level schematic of how ports were selected is shown in
Figure 3.

26	Findings related to Alaska, Hawaii, Puerto Rico, and the U.S. Virgin Islands are currently outside of the scope of
this research.

27	Each year, ACE reports a list of the top 150 Principal Ports by tonnage throughput within port limits determined
by ACE. The list of 2017 USACE Principal Ports can be found at:

https://usace.contentdm.oclc.Org/digital/collection/pl6021coll2/id/3114/rec/5. Because this analysis is limited to
the contiguous United States, the 14 following 2017 ACE Principal Ports are not considered: Valdez, AK; Honolulu,
HI; San Juan, PR; Barbers Point, Oahu, HI; Nikishka, AK; Kahului, Maui, HI; Anchorage, AK; Kivilina, AK; Hilo, HI;
Kawaihae Harbor, HI; Nawiliwili, Kauai, HI; Unalaska Island, AK; Ponce, PR; Ketchikan, AK.

28	Gillingham, K., & Huang, P. (2021). Racial Disparities in the Health Effects from Air Pollution: Evidence from Ports.
29108. Retrieved February 2023 from https://ideas.repec.Org/p/nbr/nberwo/29108.html.

6


-------
Figure 2. Map of 123 ports featured in study. Each unique port geometry has been presented as a single point for simplification of viewing port locations used in this study across
the conterminous U.S.

7


-------
Docks with Service ^
Terminated
(inactive docks)

ACE Master Docks Plus
n= 41,697 Points
n=19,098 Docks
n=l,412 Ports

ACE Master Docks Plus
Port Shapefiles
n= 834

NEI 2011
Port Shapefiles after
consolidation & review
n= 126

NEI 2014
Port Shapefiles after
consolidation & review
n= 363

Combined NEI 2011 + 2014
Port Shapefiles
n= 484

Ports without corresponding
ACE Shapefiles

n= 5	

Ports Outside of the
Conterminous U.S.
n= 14

Ports Outside of the
Conterminous U.S.
n= 14

Ports without corresponding
NEI Shapefiles

.	n= 8	

1.	Central Louisiana Regional Port LA

2.	Henderson County Riverport, KY

3.	Port ofKaskaskia, IL

4.	Natchez, MS

5.	Port of New Madrid County; MO

6.	Owensboro, KY

7.	Paducah-McCracken County
Riverport, KY

8.	Rosed ale, MS

Detroit, Ml

Hickman-Fulton County
Riverport, KY
Port of Virginia, VA
Grand Haven, Ml
Green Bay, Wl

Figure 3. Schematic of how the subset of ports was selected for this study. A full list of the 123 ports used in this study is shown
in Table A-l. List of ports included in study (n=123).

2.2 Geodesic Buffers Around Port Geometry to Describe Near-Port Populations

Port operations can occur over a much larger geographic area than just the immediate location
where cargo and passengers are loaded and unloaded from vessel to the shore. On-water activities,
including positioning of tug and pilot vessels, can happen well outside of a port's waterside boundary,
while port-related cargo handling equipment, dray trucks, locomotives, and other port-related mobile
sources can operate well beyond a port's official boundaries inland. As a result, emissions from ports
often have a wide geographic range of impact on nearby air quality. For that reason, we developed
buffers around the various ports' geometries; we estimated dasymetric population counts within
geodesic29 buffers drawn 1000m and 5000m from the port boundary, consistent with the smaller

29 Geodesic buffers account for the shape of the earth when drawn on a map and are more accurate over large
distances than Euclidean buffers. For more, see: https://www.esri.com/news/arcuser/0111/geodesic.html

8


-------
vicinities used in recent near-port studies.30,31 The geodesic buffer of 5000m captures areas farther from
the port's immediate vicinity that may still experience negative impacts from nearby port-related
operations, including, but not limited to, worsened air quality. We considered expanding this near-port
buffer to capture a larger area that may be impacted by air pollution from port operations. For example,
Agrawal et al. estimated that emissions from ocean-going vessels operating at the San Pedro ports
contributed 8.8% of the total PM2.5 emissions at a site nearly 10,000m away from the port.32 However,
we ultimately decided to limit our buffer distances to 5000m to ensure that the results of this proximity
analysis would provide an underestimation of the total population that is potentially impacted by
primary port-related emissions, including hazardous air pollutants like CO, ultrafine particles, metals,
elemental carbon (EC), NO, NOX, and VOCs that can have steeper near-source gradients than PM2.5.33.
Furthermore, these areas may also experience other port-related impacts, such as noise and traffic. We
also considered smaller buffers of 100m, 200m, and 500m, but primarily report the sociodemographic
characteristics of the populations living within the 1000m and 5000m buffers, referred to as Very Near
Port and Near Port populations, respectively.

2.3 Higher Resolution Population Estimates

Recognizing that people are not uniformly distributed geographically, we used the
dasymetric34 population dataset, developed and refined by EPA's Office of Research and Development
(ORD), to improve population estimates. ORD's dasymetric population model intelligently allocates 2010
U.S. Census counts from 2010 Census boundaries in the lower 48 states and the District of Columbia of
varying sizes and shapes to a standard 30x30-meter grid.35 This method identifies uninhabitable areas
(e.g., open water, emergent wetlands, railroad lines, cemeteries, areas with a slope of more than 25%)
and excludes populations from those areas, as illustrated in Figure 4. Within habitable landscape
categories, more people are allocated to areas where populations are more likely to live (e.g., developed
areas) and fewer people where they are less likely to live (e.g., forests) while maintaining Census block
level population counts. These habitable areas vary in population density by the intensity of
development. The intelligent dasymetric allocation of populations across a gridded area improves
residential population estimates compared to assuming equal residential population distribution across
irregularly shaped census geographies. The dasymetric model offers significant improvements compared
to equal area near-port population estimates. Demographic variables (e.g., people of color, low income)

30	Greenburg, M. R. (2021). Ports and Environmental Justice in the United States: An Exploratory Statistical
Analysis. Risk Analysis, 41(11). doi: 10.1111/risa. 13697

31	Svendsen, E. R., Reynolds, S., Ogunsakin, O. A., Williams, E. M., Fraser-Rahim, H., Zhang, H., & Wilson, S. M.
(2014). Assessment of Particulate Matter Levels in Vulnerable Communities in North Charleston, South Carolina
prior to Port Expansion. Environmental Health Insights, 8, 5-14. doi: 10.4137/EHI.S12814.

32Agrawal, H., Eden, R., Zhang, X., Fine, P. M., Katzenstein, A., Miller, J. W.,... Cocker, D. R. (2009). Primary
Particulate Matter from Ocean-Going Engines in the Southern California Air Basin. Environmental Science &
Technology, 43(14), 5398-5402. doi: 10.1021/es8035016

33	Karner, A.A.; Eisinger, D.S.; Niemeier, D.A. (2010). Near-roadway air quality: synthesizing the findings from real-
world data. Environ Sci Technol 44: 5334-5344. Doi: 10.102l/esl00008x

34	"Dasymetric" means "density measurement" and is derived from the Greek dasys (dense) and metro (measure).

35	Baynes, J., Neale, A., & Hultgren, T. (2022). Improving intelligent dasymetric mapping population density
estimates at 30 m resolution for the conterminous United States by excluding uninhabited areas. Earth Syst Sci
Data, 14(6), 2833-2849. doi: 10.5194/essd-14-2833-2022

9


-------
can be proportionally allocated from lower resolution Census geometries (e.g., tracts or block groups) to
the dasymetric population counts.

¦—I Census

'—1 Btoc*

Population
Density

SO
S t
S 2
¦ s3
M ss
B s 10
Hi >

Figure 4. Illustration of the dasymetric model using a census block near Sacramento, CA with a cemetery, and residential
housing along the eastern border. The EnviroAtlas dasymetric method allocates zero population to the cemetery and denser
population along the eastern border. Adapted from Figure 6 within Baynes et al., 2022.

The underlying population density raster36 used for this analysis is available to the public via EPA's
EnviroAtlas37, a public resource of geospatial data and tools. The data may be accessed as a feature layer
through the EnviroAtlas Interactive Map, and as downloadable raster file through the EnviroAtlas Data
Downloads webpage. Since this analysis was conducted, ORD's dasymetric model has been updated to
use 2020 U.S. Census decennial census counts and geometries and further expanded to include Alaska,
Hawaii, and other U.S. territories. Future analyses may incorporate the updated dasymetric model

2.4 Sociodemographic Variables

This study explored demographic and socioeconomic variables that are used widely across
published EJ tools and represent a wide range of vulnerabilities. The list of sociodemographic variables
included in this report is shown in Table 1. The list of demographic variables was generated after
reviewing available data and key variables in leading EJ screening tools (e.g., EPA EJScreen,
CalEnviroScreen, Climate and Economic Justice Screening Tool.) and speaking with EPA Office of
Environmental Justice and External Civil Rights subject matter experts. Data are available for each of

36	U.S. EPA. EnviroAtlas. Dasymetric Allocation of Population for the Conterminous United States,
2010. Retrieved: May 26, 2021, from https://www.epa.gov/enviroatlas/data-download

37	Pickard, B. R., Daniel, J., Mehaffey, M., Jackson, L. E., & Neale, A. (2015). EnviroAtlas: A new geospatial tool to
foster ecosystem services science and resource management. Ecosystem Services, 14, 45-55. doi:

10.1016/i.ecoser.2015.04.005.

10


-------
these variables to identify vulnerable populations at the block group level nationwide.38 This suite of
demographic metrics includes aspects of race, ethnicity, sex, age, language, and multiple metrics of
economic status. While this analysis was not related to any regulatory effort, these demographic
variables follow the first recommendation within the 2023 draft of the EPA's Technical Guidance for
Assessing Environmental Justice in Regulatory Analysis: "When achievable, analysts should present
information on estimated health and environmental risks, exposures, outcomes, benefits, or other
relevant effects disaggregated by race, ethnicity, income, and other demographic categories."39 Finally,
the vintages of demographic data used reflect a broad period of time, from 2010-2019 and were the
best available data at the time of analysis; many of these data sources have since been updated to
reflect the 2020 Census.

38	For more on Census geographic hierarchies, terms, and geographic classifications, see U.S. Census's
'Understanding Geographic Identifiers (GEOIDs)'

39	U.S. EPA (2023), "DRAFT Technical Guidance for Assessing Environmental Justice in Regulatory Analysis", page 18

11


-------
Table 1. Table of the variables that were of interest in this study by demographic category and data source.

Category | Data Source

Demographic Characteristics

Race/Ethnicity

2010 Decennial
Census

People of Color (not Non-Hispanic White)

Non-Hispanic White

Non-Hispanic Black/African American

Hispanic/Latino

Non-Hispanic American Indian/Alaska Native

Non-Hispanic Asian

Non-Hispanic Native Hawaiian/Pacific Islander

Non-Hispanic Two or More Races

Non-Hispanic Other Race

Income

2014-2018 American
Community Survey

U.S. EPA EnviroAtlas
(2017)

U.S. DOT RAISE
Persistent Poverty
Tracts (FY2021)

Population in Households Below 2x Poverty Level40

Population in Households Below lx Poverty Level

Population in Households Below 0.5x Poverty Level

Median Household Income

Households Below Quality of Life Income Threshold41

Population in Persistent Poverty Area42

Age

2010 Decennial
Census

Population Less than 5 Years Old

Population Less than 18 Years Old

Population 18-64 Years Old

Population Greater than 64 Years Old

Language

2014-2018 American
Community Survey

Households in Linguistic Isolation

Educational
Attainment

2014-2018 American
Community Survey

Population 25 Years Old or Older with Less than a High School Education

Housing

2010 Decennial

Occupied Housing Units

Census

Renter-Occupied Housing Units

For all variables, data were the latest available at the time that the geospatial analysis was conducted.

40	The percentage of the population in households below twice, once, or half their poverty threshold comes from
the ACS income-to-poverty ratio. This metric is calculated by comparing a family's income to a poverty threshold
that is based on household size and number of children. For more, see U.S. Census Poverty Glossary and Poverty
Thresholds.

41	Quality of Life Threshold Income refers to the regionally adjusted income where the basic needs of life are met,
including food, shelter, health care, and leisure time. For more, see the Threshold Income for Quality of Life,
EnviroAtlas Data Fact Sheet (July 2019)

42	Areas of Persistent Poverty, as defined by the Bipartisan Infrastructure Law (Sec 21202, §6702) are areas
meeting at least one of the following criteria: 1) county with greater than or equal to 20% of the population living
in poverty in the 1990 decennial census, the 2000 decennial census, and the most recent annual Small Area
Income and Poverty Estimates as estimated by the Bureau of the Census; 2) any Census Tract with a poverty rate
of at least 20 percent as measured by the 2014-2018 5-year data series available from the American Community
Survey of the Bureau of the Census; 3) Any U.S. Territory.

12


-------
The available resolution of Census data limits meaningful reporting of demographics at smaller
distance intervals from a port. While the dasymetric model can report population counts at a 30x30-
meter resolution, the Census demographic data used in this study are at the block group level. The
average block group found within 5000m of ports included in this study covered an area of 1.1 or 1.3
square miles (2.9 or 3.4 million square meters) using the ACE or EPA port geometries, respectively. The
dasymetric population data are so highly spatially resolved, that there are over 3,300 dasymetric model
grid cells within the average near-port block group; each of these grid cells would be allocated
sociodemographic characteristics from the same block group for which Census data are available. At
smaller buffer distances from a port, there are often few or no new block groups (and their associated
demographics) captured by the next largest buffer. Comparing demographic characteristics for
populations living within 100m or 200m from a port was not prioritized in this study since the underlying
Census data used in both groups would be similar. Instead, we compared the demographics of
populations within 1000m and 5000m of a port (Very Near Port and Near Port, respectively) against
comparison groups that live further from a port, which are described in more detail in Section 2.5 below:
Comparison Groups.

2.5 Comparison Groups

We followed EPA's Technical Guidance for Assessing Environmental Justice in Regulatory
Analysis43 to develop our comparison groups for this demographic analysis. This guidance document
emphasizes the importance of defining the general population and appropriate comparison groups to
meaningfully characterize the population of interest. These categorizations are used to estimate
differentials between populations with potential EJ concerns and populations without those concerns.
Specifically, we aimed to create comparison groups that are as similar as possible to near-port
populations found within the geodesic buffers drawn around the 123 ports of interest but that are
assumed to be far enough away from the port as to be unaffected by port-related mobile source
emissions. Populations farther from a port may have demographic characteristics that differ from
populations living closer to a major U.S. port; however, moving too far out in the comparison group may
pull in different population dynamics unrelated to the major port. For example, many ports are co-
located with major metropolitan areas, and we did not want to inadvertently compare urban areas to
non-urban areas by drawing the near-port buffer too large. The EPA's technical guidance also suggests
considering a variety of comparison groups to provide a more complete view of potential differences
between population groupings. For this study, we used two comparison groups: the Intra-County
Comparison Group and CONUS. Additional comparison groups considered for this study are noted in
Appendix G of this report.

43 U.S. EPA. "Technical Guidance for Assessing Environmental Justice in Regulatory Analysis" June 2016.
https://www.epa.gov/sites/default/files/2016-06/documents/eitg 5 6 16 v5.1.pdf

13


-------
Conterminous United States

The first comparison group used in this study was the entire Conterminous United States (i.e.,
the lower 48 states and the District of Columbia), including near-port areas. This comparison group is
considerably larger than the populations of interest and captures areas very far from the ports included
in this study. This comparison group was used to broadly contextualize near-port and nationwide
demographic patterns.

Intra-County Comparison Group

The Intra-County Comparison Group was intended to identify sociodemographic differences
between near-port populations and those less affected by port activity more locally than CONUS. The
Intra-County Comparison Group was developed by first identifying counties that contain any population
within the 5000m buffer of any of the 123 ports included in this analysis These are classified as port
counties. Within these port counties, the block groups that intersect the 5000m buffer are classified as
near-port block groups; the remaining block groups within the port counties comprise the Intra-County
Comparison Group, as shown in Figure 5. Because ACE and EPA define port geometry differently and
thus have differently sized and shaped 5000m buffers, unique Intra-County Comparison Groups were
developed for the ACE and EPA datasets separately.

Near Port Block Groups:
Block groups within
5000m of a port

Remove: Population that is within
Near Port Block Group, but outside
5000m buffer

br

Remove: Counties that do not contain
any block groups with population
within 5000m of a port have been
removed from this analysis.

Intra-county comparable block groups:
All block groups in the same county as
block groups within 5000m of any port

Exception: Block groups that contain
land, but no population, within
5000m of a port, are considered
Intra-county comparable block
groups.

CountyY

County Z

Figure 5. Schematic to illustrate Intra-County Comparison Group. For simplicity to illustrate the approach that was used to
define the Intra-County Comparison Groups, block groups are shown as square polygons, and port buffers are shown as circles.
In reality, the block groups, county boundaries, and port buffers these are irregularly shaped polygons.

14


-------
This comparison group allows for a subnational, regional, or subregional look into demographic
dynamics relevant to areas containing busy ports. Across the U.S., there is a wide range of heterogeneity
with respect to racial, ethnic, and economic demographics. Focusing on these smaller geographic
subsets allows for more precise comparisons between near-port and non-near port populations and
avoids confounding by factors that can influence demographics regionally, such as being coastal vs. non-
coastal or urban vs. non-urban. Not accounting for regional and sub-regional differences can mask
demographic or socioeconomic patterns among people who live near or far from a port. The Intra-
County Comparison Group is similar to the population group of concern, but we assume that it is outside
of the area more affected by port operations. Therefore, this comparison allows us to target the
differences between near-port populations and those who are less impacted by port activity more
precisely than a comparison against the general public.

2.6 Developing a National Analysis

In this study, we established a national estimate for the number of people living near ports,
described the demographic characteristics of these near-port populations, and compared how their
characteristics differ from those of the comparison groups described previously. Due to the significant
differences between the EPA and ACE port geographies, this national analysis was conducted using both
the EPA and ACE port geographies separately, resulting in two distinct analytical tracks and sets of
national results.

For each buffer distance tested (100, 200, 500, 1000, and 5000m; see Section 2.2: Geodesic
Buffers Around Port Geometry to Describe Near-Port Populations), we estimated the total near-port
population by first dissolving the boundaries of all 123 ports' geodesic buffers into a single layer. This
approach prevented double-counting of individuals who lived in proximity to more than one port. The
dissolved buffer layer was then overlaid on the dasymetric population model to estimate the total
population living within the specified distance of any of the 123 ports included in this study.

We leveraged block group-level sociodemographic data from Census and DOT (see Table 1) to
characterize the near-port population. For each near-port block group, we allocated its
sociodemographic composition to the population of the block group that fell within the near-port
buffer, which could be less than its total population if the block group was bisected by the buffer. Using
this approach, the near-port population was characterized under the assumption that sociodemographic
percentages are uniform across a block group. After the Census and DOT data were allocated to the
near-port population of each block group, the resulting subpopulations by sociodemographic group
were summed. These totals represented the national near-port population broken down by
sociodemographic characteristic.

For each of the near-port populations, median household income and the Quality of Life index
were calculated using a dasymetric population-weighted average. This approach accounted for the
population of each block group that fell within a near-port buffer, so that low-population block groups
or those that were only partially overlapping with the near-port buffer would not skew the median
household income or Quality of Life index for the entire near-port population. Similarly, the Quality of
Life index and median household income of the Intra-County Comparison Group was calculated as the

15


-------
population-weighted average of the relevant block groups. Direct comparisons were made across the
near-port and comparison groups for these two variables.

The CONUS comparison group reflects the total population and combined sociodemographic
features of all block groups within the lower 48 states and District of Columbia, including the near port
populations. The Intra-County Comparison Group represents the total population and combined
demographic features of all block groups that were within a port county but did not overlap a 5000m
port buffer. For instances in which a block group was bisected by the 5000m buffer, the population that
fell outside the geodesic buffer was not included in either the Intra-County Comparison Group or the
near-port population.

After estimating the breakdown of the near-port population by sociodemographic characteristic
using the approach described above, we compared it against those of the comparison groups. We
identified a difference in proportion between the near-port and comparison group populations of
greater than or equal to 1 percentage point as an indicator of disproportionate populations. This
threshold of 1 percentage point was selected as an initial screening threshold for the variables included
this study. A post hoc analysis found differences of less than 1% between variables that we would expect
to be proportional between comparison groups, the share of males and females (Figure A-4).44
Identifying disproportionalities between the near-port populations of both port geometries (ACE and
EPA) and their comparison groups (CONUS and Intra-County) can be considered stronger evidence of a
national EJ concern than identifying differences using just one comparison group or port geometry.

2.7 Top 10 Ports Supplemental Analysis

While the national analysis supports a general understanding of sociodemographic dynamics
among near-port populations and their comparison groups, we also pursued a supplemental analysis to
understand if the national results were being driven by a subset of the busiest ports as determined by
tonnage. The results of this supplemental analysis also provide insight into how representative national
disproportionalities are of typical near-port populations across the U.S. For this supplemental analysis,
we replicated the approach described above for two subsets of the 123 ports included in the study.
These subsets were the 10 ports that had the largest quantity of tonnage during 2010-2019 (a period
that aligns with the demographic data featured in this study) and the remaining 113 ports. The 10 ports
with the largest quantity of tonnage are listed below and shown in Figure 6.

1.

Port of South Louisiana, LA

6.

Port of Corpus Christi, TX

2.

Houston Port Authority, TX

7.

Port of Long Beach, CA

3.

Port of New York and New Jersey, NY and NJ

8.

Port of Baton Rouge, LA

4.

Port of Beaumont, TX

9.

Port of Los Angeles, CA

5.

Port of New Orleans, LA

10.

Port of Mobile, AL

44 Demographic proportions of male and female populations are not presented in the primary results of this work,
as these were not among the sociodemographic variables that we considered for exploring vulnerable
sociodemographic groups listed in Table 1.

16


-------
Figure 6. Map of the top 10 ports featured in supplemental analysis (as dark triangles) and remaining 113 ports (light blue
circles) by tonnage in this study. Each unique port geometry has been presented as a single icon for simplification of viewing port
locations used in this study across the conterminous U.S.

From 2010 to 2019, these 10 ports accounted for 45.6% of all tonnage handled across the top
150 ports in the U.S. and 49.5% of all tonnage handled across the 123 ports in this study (Figure 7).

17


-------
Relative Share of Tonnage by Port, 2010-2019

Houston Port Authority, TX



New York, NY&NJ

[5.7%]

Beaumont. TX

[3.7% j

New Orleans. LA

¦ [ 3.6% |

Corpus Christi, TX

[3.6% 1

Port of Long Beach, CA

13.5% |

Baton Rouge, LA

[2.9%]

Port of Los Angeles, CA

(2.7%)

Mobile, AL

[2.4% |

Lake Charles Harbor District, LA

12.4% |

Plaquemines Port District, LA

[2.4%]

Texas City, TX

[2.0%]

Huntington-Tristate, KY, OH, WV

11.9% J

Baltimore, MD

[1.8%]

Port of Savannah, GA

[1.6%]

Duluth-Superior, MN and Wl

[1.5% i

Port Arthur, TX

[1.5%]

St. Louis, MO and IL

11.5% |

Tampa Port Authority, FL

11.4% |

Cincinnati-Northern KY, Ports of

11.3%)

Pittsburgh, PA Port of

11.2% |

Port of Pascagouia, MS

[1.2%]

Richmond, CA

11.1%]

Philadelphia Regional Port, PA

11.1% |

Seattle, WA

[1.0%]

Tacoma, WA

[1.0%]

Port Freeport, TX

[1.0% I

o%

2%

4%	6%

Relative Share by Port

10%

12%

Shares reported as share of tonnage by port relative to all 123 ports
included in this study according to the top 150 Principal Ports lists from 2010-2019.

Ports with <1 % of relative share excluded from figure-
Data Source: USACE, 2010-2019

Figure 7. The relative share of tonnage by ports contributing >1% relative share, among the 123 ports featured in this study,
based on ACE Principal Port Data from 2010-2019.

3. Results

3.1 National Estimates of Near-Port Populations and Comparison Groups

The national analysis indicates that 16.1M or 31.1M people live within 5000m of major ports in
CONUS using the EPA or ACE port geometries, respectively (Table 2). These values correspond to 5% and
10% of the total CONUS population. Additionally, Table 2 summarizes the total size of the two
comparison groups used in the study. The population of the Intra-County Comparison Group was over
four times larger or over two times larger than the Near Port population for the EPA or ACE port
geometries, respectively.

18


-------
Table 2. Summary of 2010 Populations within Very Near and Near Port Population and Comparison Groups by Port Geometry
(ACE or EPA).





Very Near Port

Near Port

Intra-County

Conterminous





Population

Population

Comparison

United States





(within 1000m of
any port in this
study)

(within 5000m of
any port in this
study)

Group



EPA Port
Geometry

Total Population

2,618,779

16,124,914

70,241,778

306,563,500

Number of Block Groups

3,473

15,029

47,890

216,017

Number of Counties

157

186

182

3,107

ACE Port
Geometry

Total Population

4,676,199

31,058,906

70,125,603

306,563,500

Number of Block Groups

6,328

27,621

46,915

216,017

Number of Counties

203

243

241

3,107

Table 3. National Estimates of Near-Port Populations by Buffer Distance from Port and Port Geometry (ACE or EPA).

Population living within indicated 	Port Geometry	

distance of any port	ACE	EPA

100m	24,473	448,940

200m

123,354

629,820

500m

1,134,075

1,294,281

1000m (Very Near Port population)

4,676,199

2,618,779

5000m [Near Port population)

31,058,906

16,124,914

3.2 National Disproportionalities of Near-Port Populations

The full list of results for differences in percentages of sociodemographic groups between the
two port geometries and their comparison groups are presented in Table 4 and Table 5.

19


-------
Table 4. Summary of Differences in Percentage of Sociodemographic Groups between Very Near Port Populations and Comparison Groups (within 1000m).

Category

Demographic Feature

Differences in Percentage of Sociodemographic Groups for
Very Near Port Populations (<1000m) vs. Comparison Groups

(% pt.)

EPA Port Geometry

ACE Port Geometry

vs. ICC

vs. CONUS

i/s. ICC

vs. CONUS

Race/Ethnicity

All People of Color

6.4

16.0

10.1

16.1

Non-Hispanic (NH) White

-6.4

-16.0

-10.1

-16.1

NH Black

8.5

10.0

3.1

3.3

Hispanic

-0.5

5.0

6.2

9.8

NH Asian

-1.8

1.0

0.8

3.2

NH American Indian/Alaska Native

0.1

0.0

0.0

-0.4

NH Native Hawaiian/Pacific Islander

0.0

0.0

-0.1

0.0

NH Other Race

0.0

0.0

0.1

0.2

NH Two or More Races

0.0

0.0

-0.1

0.0

Income

Below 0.5x Poverty Threshold

4.6

4.0

4.1

2.9

Below lx Poverty Threshold

9.3

8.0

8.3

5.8

Below 2x Poverty Threshold

13.3

10.0

10.5

5.7

Living in Area of Persistent Poverty

21.0

26.0

11.5

13.4

Age

Less than 5 years old

0.2

0.0

-0.4

-0.6

Less than 18 years old

-2.1

-2.0

-5.2

-5.2

Greater than 64 years old

-1.5

-2.0

-1.2

-1.2

Language

Households in linguistic isolation

0.2

3.0

5.1

6.7

Educational Attainment

Less than HS education

4.9

5.0

4.6

3.9

Housing

Occupied Housing Units

-4.4

-3.0

-3.5

-1.3

Renter-Occupied Housing Units

16.9

19.0

31.2

30.3

NB: Instances where the very near port population has at least a 1% pt. higher share of population with the listed demographic characteristic than the
noted comparison group are shown in bold and outlined. ICC: Intra-county comparison group; CONUS: Conterminous U.S. comparison group

20


-------
Table 5. Summary of Differences in Percentage of Sociodemographic Groups between Near Port Populations and Comparison Groups (within 5000m)

Category

Race/Ethnicity

Income

Demographic Feature

Differences in Percentage of Sociodemographic Groups for
Near Port Populations (<5000m) vs. Comparison Groups (% pt.)

EPA Port Geometry

ACE Port Geometry

Age

Language
Educational Attainment

Housing



vs. ICC

vs. CONUS

vs. ICC

vs. CONUS

All People of Color

8.8

18.0

12.3

18.3

Non-Hispanic (NH) White

-8.8

-18.0

-12.3

-18.3

NH Black

9.6

12.0

8.9

9.1

Hispanic

-1.2

4.0

2.6

6.2

NH Asian

0.1

2.0

0.6

3.0

NH American Indian/Alaska Native

0.1

0.0

0.0

-0.4

NH Native Hawaiian/Pacific Islander

0.0

0.0

0.0

0.0

NH Other Race

0.1

0.0

0.1

0.2

NH Two or More Races

0.2

0.0

0.0

0.1

Below 0.5x Poverty Threshold

3.8

3.0

3.4

2.2

Below lx Poverty Threshold

7.5

6.0

7.1

4.5

Below 2x Poverty Threshold

11.0

8.0

10.2

5.5

Living in Area of Persistent Poverty

15.4

20.0

10.1

11.9

Less than 5 years old

0.3

0.0

0.2

0.0

Less than 18 years old

-1.7

-2.0

-2.0

-2.0

Greater than 64 years old

-1.3

-2.0

-1.0

-1.1

Households in linguistic isolation

0.8

3.0

3.4

5.1

Less than HS education

3.9

4.0

4.2

3.5

Occupied Housing Units

-3.0

-1.0

-2.2

0.0

Renter-Occupied Housing Units

17.0

20.0

20.7

19.8

NB: Instances where the near port population has at least a 1% pt. higher share of population with the listed demographic characteristic than the
noted comparison group are shown in bold and outlined. ICC: Intra-county comparison group; CONUS: Conterminous U.S. comparison group

21


-------
Overrepresentation of People of Color in Near-Port Populations compared to Comparison
Groups

The results of this analysis indicate an overrepresentation of people of color, defined as anyone
who does not identify as Non-Hispanic White, living near the 123 major ports in CONUS that were
included in this study. This overrepresentation was consistent using either the ACE or the EPA dataset
when compared against both the Intra-County Comparison Group and CONUS (Figure 8). Using the ACE
port geometries, 52% of the Very Near Port (within 1000m of any port in this study) population and 54%
of the Near Port (within 5000m of any port in this study) population are people of color. In comparison,
only 36% of the population of CONUS are people of color. This overrepresentation of people of color in
proximity to a port also holds true in comparison to the Intra-County Comparison Groups (42% for ACE;
45% for EPA).

60% -

o
ra

"5 40%"

CL

o

CL

o

20% -

<1)

CL

o%-

Figure 8. Comparison of the percentage of people of color between near-port populations and comparison groups for 123 ports
included in the primary analysis.

For other racial and ethnic data evaluated, both the EPA and ACE near-port buffers contain a
higher share of Non-Hispanic Black population and a lower share of Non-Hispanic White population than
the CONUS and Intra-County Comparison Groups (Figure 9). The near-port population around ACE port
geometries also captures a higher share of Hispanic/Latino population and Non-Hispanic Asian
population than the comparison groups. Using the EPA port geometries, however, the difference in
percentage of Hispanic populations living near ports versus the Intra-County Comparison Group was not
greater than 1% pt., indicating that the share of Hispanic populations near these port geometries is not
substantially different from nearby populations, and the disproportionality detected against CONUS may
reflect regional differences. For both the EPA and ACE port geometries, there were no difference greater
than 1% pt. in the share of Non-Hispanic Other Race, Non-Hispanic Two or More Races, Non-Hispanic
Hawaiian and Pacific Islanders, or Non-Hispanic Native Americans between the near-port population and
the CONUS or Intra-County Comparison Groups; these racial groups also comprise a smaller portion of
the population (0-2% of CONUS).

22

Share of People of Color Population Differences between Near Port and Comparison Populations

EPA Port Geometry

Very Near Port Near Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group

Very Near Port Near Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group


-------
5%
7%
7%
5%

¦

%

100% 90% 80% 70%

60%

50%
EPA

I

<1%

<1% I
<1%
<1% ]

2%

2%

2%

2%

<1%

<1% I
<1%

1% I

<1%
<1%
<1%
<1%

NH Asian

NH Other Race

NH Two or More
Races

NH American
Indian/Alaska Native

NH Native Hawaiian/
Pacific Islander

8%
8%
7%
5%

<1%

I <1%
<1%
<1%

I

2%
2%
2%
2%

<1%
<1%
<1%
| 1%

<1%
<1%
<1%
<1%

40% 30% 20% 10%

0%

0%

Very Near Port
(0-lkm)

I Near Port
(0-5km)

I Intra-County
Comparison Group

I Conterminous United
States (CONUS)

10% 20% 30% 40%

50%
ACE

60% 70% 80% 90% 100%

Figure 9. Pyramid plot of the percentage of the population belonging to selected racial and ethnic groups by near-port populations and comparison groups.

23


-------
Multiple Metrics of Income Point to Economic Disproportionalities between Near-Port
Populations & Comparison Groups

Both analytical tracks revealed the presence of disproportionate socioeconomic conditions
between near-port populations and comparison groups. 43% of the Very Near Port population around
the EPA port geometries, and 38% of the Very Near Port population around the ACE port geometries
were found to be living below twice the poverty threshold (Figure 10). These levels are nearly 10% pt.
higher than the share of population in the Intra-County Comparison Group for both port geometries
(29% for Intra-County Comparison Group populations around EPA port geometries; 28% for Intra-County
Comparison Group populations around ACE port geometries). The percentage of individuals living below
twice the poverty threshold was also at least 10% pt. higher among near-port populations than across
the Conterminous U.S. (32%).

Share of Population living up to Twice the Poverty Threshold in Near Port and Comparison Populations

EPA Port Geometry

ACE Port Geometry

o

'to 30%

38%



Very Near Port Near Port Population, Intra-county Conterminous US	Very Near Port Near Port Population, Intra-county Conterminous US

Population, 0-1000m 0-5000m Comparable Group	Population, 0-1000m 0-5000m Comparable Group

Figure 10. Comparison of percentage of the population living below twice the poverty threshold between near-port populations
and comparison groups for 123 ports included in the primary analysis.

We also observed a difference for near-port populations using median household income as an
indicator of socioeconomic status (Figure 11). The median household income for the Very Near Port
populations is $58k around EPA geometries and $70k around ACE geometries; the median household
income for Near Port populations is $61k around EPA geometries and $65k around ACE geometries.
Meanwhile, the ACE Intra-County Comparison Group has a median household income that is nearly 15%
higher than it is among the Very Near Port populations ($79.6k vs. $69.5k). The EPA Intra-County
Comparison Group has a median household income that is almost 33% higher than it is among the Very
Near Port population ($77.6k vs. $58.4k). This difference is also reflected when compared to the median
household income of CONUS, which is 10% higher than the Near Port population for EPA port
geometries and 5% higher than the Near Port population for ACE port geometries. However, the median
household income of the ACE Very Near Port population is 2% higher than the median household
income for CONUS.

24


-------
Summary of Household Median Income between Near Port and Comparison Populations

EPA Port Geometry

Very Near PortNear Port Population, Intra-county Conterminous US Very Near PortNear Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group	Population, 0-1000m 0-5000m Comparable Group

Figure 11. Comparison of median household income between near-port populations and comparison groups for 123 ports
included in the primary analysis.

In addition to these income and poverty metrics, we also identified a considerably higher
percentage of renters, households with less than a high school education, and households living in
linguistic isolation in Very Near Port and Near Port populations as compared to the Intra-County
Comparison Group and CONUS (Figure 12 ). While these characteristics are not necessarily reflective of
household wealth or poverty, the differences further underscore sociodemographic differences
between populations living near ports and those living father away from ports.

Finally, we examined the index of the threshold income for quality of life, a metric found within
EPA's EnviroAtlas that estimates the baseline income for a "positive quality of life and accompanying
emotional well-being", with which the basic needs of life are met, including the cost of housing, food,
and health care.45 The index is based on a national value of $75k in 2009 dollars (equal to $110k in 2024
dollars), and further adjusted to reflect county-level cost of living. The higher the index value, the
greater share of the population found to be living below the county-level threshold income. As shown in
Figure 13, the Quality of Life index for all near-port populations explored in this study ranged from 73 to
76, while the comparison groups had values between 61 and 64, further supporting the quantifiable
differences between near-port populations and those living farther away discussed above.

45 U.S. EPA. (2019) EnviroAtlas Fact Sheet EPA, EnviroAtlas Fact Sheet, 2019, Accessed September 6, 2024

25


-------
88%
89%f

92% r

90%

60%

Occupied Housing
Units

Renter-Occupied
Housing Units

Less than High School
Education

Households in
Linguistic Isolation

In an Area of
Persistent Poverty

Below 2x Federal
Poverty Level

Below lx Federal
Poverty Level

Below 0.5x Federal
Poverty Level

48%
46%

38%
38%

89%
L?0%
92%
90%

64%

Very Near Port
(0-lkm)

I Near Port
(0-5km)

I Intra-County
Comparison Group

I Conterminous United
States (CONUS)

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

EPA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

ACE

Figure 1212. Pyramid plot of the percentage of the population belonging to selected socioeconomic groups by near-port populations and comparison groups.

26


-------
Summary of Quality of Life Index between Near Port and Comparison Populations

80-

*60-

(1)

"a
_c

0)

¦540-

"ro
O

20-

o-

Figure 13 13. Comparison of Quality of Life index, equal to the percentage of the population living below the Quality of Life
income threshold, between near-port populations and comparison groups for 123 ports included in the primary analysis.

No detectable disproportionalities among vulnerable age groups in Near-Port Populations
There were no detectable disproportionalities for vulnerable age groups in the near-port
populations compared to the comparison groups. In fact, there were smaller shares of people aged 64 or
older in the Very Near (11.3% and 11.8% for EPA and ACE, respectively) and Near Port populations
(11.4% and 12% for EPA and ACE) compared to both the Intra-County Comparison Group (12.8% and
13% for EPA and ACE) and Conterminous U.S. (13%; see Figure A- 3 in the Appendix F). We observed a
similar trend among the two younger populations examined as well. For children less than 5 years old,
there was less than 1 percentage point difference between the near-port populations and the
comparison groups. Children less than 5 years old made up 6.6% and 5.9% of the Very Near Port
populations (for EPA and ACE, respectively); they made up 6.4% (EPA) or 6.3% (ACE) of the Intra-County
Comparison Group and 6.5% of the CONUS population. Children less than 18 years old made up 21.6%
and 18.8% of the Very Near Port populations for EPA and ACE geometries, respectively, and they made
up 22% of the Near Port population using both port geometries. These proportions were lower than
both the Intra-County Comparison Groups (23.6% for EPA and 24% for ACE) and the Conterminous U.S.
(24%). Taken together, these results indicate that there is not an overrepresentation of vulnerable age
groups in near-port communities.

3.3 Top 10 Ports Supplemental Analysis Supports Conclusions of National Analysis

Estimates of Near-Port Populations & Comparison Groups

As discussed earlier, a supplemental analysis of the top 10 ports by tonnage was developed to
further evaluate the validity of the national-level results of the primary analysis. The purpose of this

27

EPA Port Geometry

Very Near Port Near Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group

Very Near Port Near Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group


-------
supplemental analysis was to understand whether the sociodemographic patterns observed in the
primary analysis were driven solely by the busiest ports by tonnage, which also aligned with major
metro areas. Further, we sought to understand if there were meaningfully different disproportionalities
or sociodemographic characteristics surrounding the top 10 busiest ports, which may also contribute the
most emissions related to port activity.

While the top 10 ports were selected based on tonnage thresholds, the populations surrounding
these ports also contain a large subsample of the total population featured in the primary analysis. As
shown in Figure 14, the Near Port population of the top 10 ACE port geometries is 13.1M, which is
approximately 42% of the total Near Port population in the primary analysis of 123 ports (31.1M); the
Near Port population of the top 10 EPA port geometries is 5.1M, which is approximately 31% of the total
Near Port population in the primary analysis of 123 ports (16.1M). Similarly, 55% of the Very Near Port
population of ACE port geometries are in proximity to one of the top 10 ports (2.6M vs. 4.7M), and 41%
of the Very Near Port population of EPA port geometries are in proximity to one of the top 10 ports
(1.1M vs. 2.6M).

Population by Port Geometry and Distance for Supplemental Analysis

60,000,000

c

40,000,000

o_
o
CL

20,000,000-

0-

EPA Port Geometry

70.2M

53.2M

18.3M

16-1M

11.1M

5.1 M

2.6M

ACE Port Geometry

70.1 M

53.8M

31.1M

19.9M

13.1M

2.6M

18_1M

Top 10 Ports Remaining 113 Ports 123 Ports

Top 10 Ports Remaining 113 Ports 123 Ports

Population Group ¦	"""" ¦ XST™ I	»„

Figure 14. Comparison of Very Near Port, Near Port, and Intra-County Comparison Group populations for the supplemental
analysis.

There are important sociodemographic differences between the populations in proximity to the
top 10 busiest ports by tonnage and those in proximity to the remaining 113 ports included in this study.
Generally, there is a higher proportion of people of color near the top 10 ports than there is near the
remaining 113 (Table 6 and Table 7). For both port geometries (EPA and ACE) and buffer sizes (1000m
and 5000m), this pattern is largely driven by differences in the percentage of Hispanic and Non-Hispanic
Asian populations; among the Very Near Port population of the EPA port geometries, there is also a
notably higher percentage of Non-Hispanic Black for the top 10 ports as compared to the remaining 113.
Near-port populations surrounding the top 10 ports also have a higher proportion of households in
linguistic isolation, renter-occupied housing units, and individuals with less than a high school education

28


-------
than the populations surrounding the remaining 113 ports. Although the top 10 ports are located in
areas with higher median household incomes and a lower percentage of the population living in areas of
persistent poverty, their surrounding populations also have a greater proportion of households below
the Quality of Life income threshold as compared to those of the remaining 113 ports.

29


-------
Table 6. Sociodemographic Characteristics of Very Near Port Populations (within 1000m)

Port subset

Total 123	Top 10	Remaining 113

ACE Port Geometries

Race/Ethnicity (% of population)

NH White	47.9	39.8	57.7

People of Color	52.1	60.2	42.3

NH Black	15.6	15.9	15.2

Hispanic	26.2	33.0	17.9

NH Asian	7.7	9.1	6.1

NH Other Race	0.4	0.4	0.3

NH Two or More Races	1.9	1.6	2.2

NH American Indian/Alaska Native	0.3	0.2	0.5

NH Native Hawaiian/Pacific Islander	0.1	0.0	0.1
Socioeconomic status (% of population)

Occupied Housing Units	89.1	91.0	86.7

Renter-Occupied Housing Units	53.9	70.8	56.4

Less than High School Education	16.8	18.5	14.7

Households in Linguistic Isolation	11.6	14.5	8.0

In an Area of Persistent Poverty	47.6	47.1	48.2

Below 2x the Poverty Threshold	38.1	36.8	39.6

Below lx the Poverty Threshold	20.1	19.2	21.3

Below 0.5x the Poverty Threshold	9.4	8.7	10.2
Median Household Income $ 69,519.87 $ 74,448.98 $ 62,934.26

Quality of Life Index	7AA	753	72.1

EPA Port Geometries

Race/Ethnicity (% of population)

NH White	48.2	36.6	56.1

People of Color	51.8	63.4	43.9

NH Black	22.8	30.9	17.2

Hispanic	21.1	25.7	17.9

NH Asian	5.2	4.8	5.4

NH Other Race	0.2	0.2	0.3

NH Two or More Races	2.0	1.4	2.4

NH American Indian/Alaska Native	0.4	0.3	0.5

NH Native Hawaiian/Pacific Islander	0.1	0.1	0.2
Socioeconomic status (% of population)

Occupied Housing Units	87.8	88.7	87.1

Renter-Occupied Housing Units	53.5	52.7	54.2

Less than High School Education	17.8	20.5	15.9

Households in Linguistic Isolation	7.5	9.2	6.3

In an Area of Persistent Poverty	60.3	59.4	61.1

Below 2x the Poverty Threshold	42.6	43.1	42.2

Below lx the Poverty Threshold	22.2	22.2	22.3

Below 0.5x the Poverty Threshold	10.4	10.4	10.4
Median Household Income $ 58,358.87 $ 58,098.73 $ 58,485.34

Quality of Life Index	75.6	75.5	75.6

30


-------
Table 7. Sociodemographic Characteristics of Near Port Populations (within 5000m)

Port subset

Total 123	Top 10	Remaining 113

ACE Port Geometries

Race/Ethnicity (% of population)

NH White	45.6	36.3	52.5

People of Color	54.4	63.7	47.5

NH Black	21.4	21.5	21.3

Hispanic	22.6	29.7	17.5

NH Asian	7.6	10.1	5.7

NH Other Race	0.4	0.5	0.3

NH Two or More Races	2.0	1.6	2.2

NH American Indian/Alaska Native	0.2	0.2	0.4

NH Native Hawaiian/Pacific Islander	0.1	0.1	0.1
Socioeconomic status (% of population)

Occupied Housing Units	90.3	91.7	89.4

Renter-Occupied Housing Units	53.9	61.7	48.0

Less than High School Education	16.4	18.9	14.6

Households in Linguistic Isolation	9.9	13.9	7.0

In an Area of Persistent Poverty	46.1	43.8	47.7

Below 2x the Poverty Threshold	37.8	37.1	38.2

Below lx the Poverty Threshold	18.9	18.4	19.2

Below 0.5x the Poverty Threshold	8.7	8.1	9.0
Median Household Income $ 65,147.06 $ 68,567.21 $ 63,572.33

Quality of Life Index	7^4	75X)	72.6

EPA Port Geometries

Race/Ethnicity (% of population)

NH White	45.7	36.9	49.8

People of Color	54.3	63.1	50.2

NH Black	23.9	25.0	23.4

Hispanic	20.4	27.0	17.4

NH Asian	7.0	8.8	6.1

NH Other Race	0.3	0.2	0.3

NH Two or More Races	2.1	1.6	2.4

NH American Indian/Alaska Native	0.4	0.2	0.4

NH Native Hawaiian/Pacific Islander	0.1	0.1	0.2
Socioeconomic status (% of population)

Occupied Housing Units	89.2	89.8	88.9

Renter-Occupied Housing Units	53.6	58.3	51.4

Less than High School Education	16.8	20.3	15.2

Households in Linguistic Isolation	8.1	11.5	6.6

In an Area of Persistent Poverty	54.7	52.4	55.7

Below 2x the Poverty Threshold	40.2	40.7	40.0

Below lx the Poverty Threshold	20.4	20.3	20.5

Below 0.5x the Poverty Threshold	9.5	9.0	9.8
Median Household Income $ 61,302.89 $ 63,215.48 $ 60,406.98

Quality of Life Index	75.0	75.9	74.6

31


-------
Disproportionalities of Near-Port Populations
Race and Ethnicity Demographic Features

In the primary analysis of this study, we detected higher percentages of people of color and
Non-Hispanic Black populations living within 5000m of a port (using either EPA or ACE geometries)
compared to the Intra-County Comparison Group and CONUS. When subset to understand the
demographic differences around the top 10 ports and the remaining 113 ports, there is still a higher
percentage of Non-Hispanic Black population living in near-port populations compared to the CONUS
and Intra-County Comparison Groups for both port geometries studied (Figure 15).

Related to people of color, the supplemental analysis generally supported the findings of the
primary analysis; however, it did reveal some surprising nuances. Among populations near the top 10
ports (using either EPA or ACE geometries), there is still a disproportionately higher share of people of
color as compared to CONUS. However, the difference in percentage of people of color between near-
port populations and their Intra-County Comparison Group is much smaller for the top 10 ports as
compared to the primary analysis of 123 ports. In fact, the relationship reverses around EPA port
geometries, with the percentage of people of color within the Near Port population being 2 points lower
than it is in the Intra-County Comparison Group. These findings indicate that the overrepresentation of
people of color in the Near Port population versus the Intra-County Comparison Group is not being
driven by the population around the top 10 ports, but rather by the 113 remaining ports.

We also observe that the attenuated overrepresentation of people of color near the top 10
ports versus the Intra-County Comparison Group is partially driven by Hispanic and Non-Hispanic Asian,
among whom there is an even more pronounced reversal of population dynamics. For example, the
Hispanic population comprises 27% of the Near Port population around the top 10 EPA port geometries,
which is 11 percentage points higher than CONUS (16%), but 11 percentage points lower than the Intra-
County Comparison Group (38%). This finding suggests that the near-port population around the top 10
most active ports is comprised of relatively fewer people who are Hispanic or Non-Hispanic Asian than
the neighboring populations. As a result, we do not observe an overrepresentation of total people of
color around the top 10 ports as compared to the Intra-County Comparison Group, even though this
overrepresentation for Non-Hispanic Black populations is similar or greater than it is in the primary
analysis of 123 major U.S. ports.

32


-------
Near Port (within 5000m)

Very Near Port (within 1000m)

People of Color-

NH Black

Hispanic-

NH Asian

People of Color

NH Black

Hispanic

NH Asian -

0	10 20 30 -10 0	10 20

Difference in Percentage, 5000m and Comparison Groups (% pt.)

10 0	10 20 30 -10 0	10 20

Difference in Percentage, 1000m and Comparison Groups (% pt.)

Remaining 113 Ports H Top 10 Ports H 123 Ports

Remaining 113 Ports H Top 10 Ports | 123 Ports

Figure 15. Comparisons by race/ethnicity between near-port populations and comparison groups for the 123, top 10, and remaining 113 ports. The difference in percentage by
race/ethnicity between the near-port populations and comparison group is printed next to each bar; bars with positive values indicate a higher percentage of that demographic
group in the near-port populations than the comparison group, while negative values indicate a higher percentage in the comparison group. Only demographic characteristics
with differences in percentage greater than 1% pt. in the primary analysis are shovt/n.

33


-------
Socioeconomic Features

For almost all socioeconomic variables, the results of the supplemental analysis for the top 10
and remaining 113 ports support the national findings of the primary analysis. Mainly, there was
consistent overrepresentation in the near-port populations of individuals living below twice, once, and
half the poverty threshold as compared to CONUS and the Intra-County Comparison Groups. Using the
ACE port geometries, we also observed patterns that were generally consistent in the supplemental
analysis and in the primary analysis for other socioeconomic variables, like the percentages of occupied
housing units, renter-occupied housing units, adults with less than a high school education, and
households in linguistic isolation. Using the EPA port geometries, evidence of disproportionalities in
proximity to the top 10 ports based on renter occupancy, educational attainment, and linguistic isolation
tended to be stronger in comparison to CONUS than the Intra-County Comparison Group. For example,
the percentage of adults with less than a high school education in the population living within 5000m of
one of the top 10 ports was 7 points higher than in CONUS (20% vs. 13%); it was only 1 percentage point
higher than the Intra-County Comparison Group (20% vs. 19%).

Based on the comparison group that was used, some contrasting results did emerge from the
supplemental analysis for measures of income and poverty. For example, the proportion of the
population living in areas of persistent poverty was higher in the top 10 near-port populations than it
was in CONUS (by 18 percentage points using EPA port geometries) but lower than it was in the Intra-
County Comparison Groups (by 13 percentage points using EPA port geometries). Interestingly, the
percentage of households living below the Quality of Life income threshold in the top 10 near-port
populations was still higher than it was in the Intra-County Comparison Groups. We hypothesize that the
largest ports are important contributors to economic activity in their metropolitan areas or regions, but
that disproportionalities in who is most likely to live near these ports still remain. This hypothesis can
also help explain why median household income and the percentage of households living below the
Quality of Life threshold are both higher in the top 10 supplemental analysis than they are in the
primary analysis (Table 6 and Table 7). Although there may be wealthier regions that surround the top
10 ports, we still identified consistent socioeconomic disproportionalities in their near-port populations
as compared to CONUS and the Intra-County Comparison Groups. The supplemental analysis illustrates
the importance of using comparison groups in the same parts of the country as the near-source groups.

34


-------
Near Port (within 5000m)

Very Near Port (within 1000m)

EPA Port Geometry

ACE Port Geometry

-10	0	10 20	-10	0	10 20

Difference in Percentage, 5000m and Comparison Groups (% pt.)

0 10 20 30 -10 0 10 20
Difference in Percentage, 1000m and Comparison Groups (% pt.)

Remaining 113 Ports H Top 10 Ports H 123 Ports

Remaining 113 Ports H Top 10 Ports | 123 Ports

Figure 16. Comparisons by poverty-related factors between near-port populations and comparison groups for the 123, top 10, and remaining 113 Ports. The difference in
percentage between the near-port population and comparison group is printed next to each bar; bars with positive values indicate a higher percentage of that group in the near-
port population than the comparison group, while negative values indicate a higher percentage in the comparison group. Only socioeconomic characteristics with differences in
percentage greater than 1% pt. in the primary analysis are shovi/n.

35


-------
Near Port (within 5000m)

Very Near Port (within 1000m)

Occupied.
Housing Units

Renter-Occupied
Housing Units

Education Less
than High School

Households in
Linguistic Isolation

Occupied
Housing Units*

Renter-Occupied.
Housing Units

Education Less
than High School

Households in
Linguistic Isolation

0	10	20	30

Difference in Percentage, 5000m

0	10	20

and Comparison Groups (% pt.)

0 10 20 30 40 0 10 20 30
Difference in Percentage, 1000m and Comparison Groups (% pt.)

Remaining 113 Ports

Top 10 Ports H 123 Ports

Remaining 113 Ports H Top 10 Ports | 123 Ports

Figure 17. Comparisons by other socioeconomic between near-port populations and comparison groups for the 123, top 10, and remaining 113 Ports. The difference in
percentage between the near-port population and comparison group is printed next to each bar; bars with positive values indicate a higher percentage of that group in the near-
port population than the comparison group, while negative values indicate a higher percentage in the comparison group. Only demographic characteristics with differences in
percentage greater than 1% pt. in the primary analysis are shown.

36


-------
Sioo.ooo

I 590,000
580,000
I 570,000
560,000
I 550,000
540,000
I 530,000
I 520,000
I 510,000

5-

$72,995 ¦ $68396

I Very Near Port
(0-lkm)

I Near Port
(0-5km)

I Intra-County
Comparison Group

¦ Conterminous United
States (CONUS)

Figure 18. Comparison of median household income between near port populations and comparison groups for the top 10 ports.

100
90
80
70
60
50
40
30
20
10
0

¦ Very Near Port
(0-lkm)

i Near Port
(0-5km)

I Intra-County
Comparison Group

I Conterminous United
States (CONUS)

Figure 19. Comparison of Quality of Life index, equal to the percentage of the population living below the Quality of Life income
threshold, between near port populations and comparison groups for the top 10 ports.

4. Discussion
4.1 Summary of Results

There is not a single authoritative source for the geospatial extent of U.S. ports; therefore, we
estimated the total population living in proximity to a subset of major U.S. ports (n=123) using data from
two different federal agencies: U.S. EPA (EPA) and U.S. Army Corps of Engineers (ACE). Using the EPA
geometries, we estimate that 16.1M people live within 5000m of a major U.S. port, equal to ~5% of the
total population of CONUS (2010). Using the ACE geometries, we estimate that 31.1M people, or ~10%
of the total population of CONUS (2010), live within 5000m of a major U.S. port.

37


-------
Furthermore, we have identified sociodemographic patterns across the 123 ports included in
this study that show consistent disproportionalities, which may suggest potential disparities that exist
among near-port populations at a national scale. Specifically, we have found that there is a
disproportionate overrepresentation of people of color, non-Hispanic Black, and Hispanic populations
living within 1000m and 5000m of U.S. ports when compared to neighboring populations (i.e., Intra-
County Comparison Group) and to the rest of the conterminous United States. The results of a
supplemental analysis further supported these national-level findings, with clear evidence that these
trends are not solely attributable to the demographic characteristics of populations in proximity to the
largest ports in the U.S. The results of this study also reveal important socioeconomic
disproportionalities among near-port populations as represented by factors associated with income,
poverty, and housing. Specifically, we observed overrepresentation among near-port populations as
compared to their neighboring communities and CONUS of individuals living below twice the poverty
threshold, renters, individuals living in areas of persistent poverty, adults with less than a high school
education, households in linguistic isolation, and households living below the Quality of Life income
threshold.

While we did not estimate air quality or exposure in this study, our results do characterize the
populations that live in close proximity to ports and are most likely to be exposed to harmful air
emissions from port activities, which are also the populations that are most likely to benefit from efforts
to lower port-related mobile source emissions. These communities are overrepresented by people of
color, non-Hispanic Black, Hispanic, and low-income populations. Additionally, actions taken to make the
top 10 ports with the largest tonnage throughput cleaner may have an outsized impact on the national
near-port population, as their near-port populations account for approximately 30-40% of the total
population living in proximity to a major port in CONUS.

4.2 Differences between the EPA and ACE Shapefiles and Impact on Population Totals

We identified significant qualitative differences between the two geometries at the individual
port-level, such as the differences shown in Figure 2. There are only 4 ports (out of the 123 studied) for
which all the points corresponding to active docks from the ACE dataset are wholly contained within the
EPA polygons.46 Furthermore, the majority of ACE active dock points fall outside of the corresponding
EPA polygon at almost 75% of the ports (92 out of 123). While we used these datasets to approximate
where both waterside and landside mobile source activity may occur, all of the ACE points and 40 of the
EPA polygons do not geospatially cover potential landside operations, and as a result may
underestimate the spatial extent of where port activity is.

To further illustrate the distinct geometries between the two port geometry datasets and the
resulting impact on population found withing the geodesic buffer, consider the way that each dataset
describes the port of Chicago in Figure 20, and the resulting impact on the defined near port population
for each analytical track. Nationally, the total area enclosed by the 5000m buffer around ACE port

46 The ports with active docks from ACE that are wholly contained within the EPA shapefiles are: Port Dolomite Ml;
Marcus Hook PA; Silver Bay, MN; and Two Harbors MN.

38


-------
geometries was over 10,000 square miles, which is 54% larger than the area enclosed by the 5000m
buffer around EPA port geometries.

1=1

Figure 20. Map comparing the differences between the EPA (left) and ACE (right) port geometries and the resulting impact on
differences in the extent of block groups in Near Port populations (shown in light blue) and the extent of the Intra-County
Comparison Groups (shown in peach) using Chicago as an example. Note block groups with zero population are included in the
figure above.

Number of near-port block groups captured by
one or both of the port geometries studied

1000m	5000m

Figure 21. Visualization of the number of near-port block groups captured by the ACE port geometries, the EPA port geometries,
or both.

39


-------
To depict the overlap of Census data that were allocated to the EPA and ACE near-port
populations, we have visualized the number of Very Near Port (within 1000m) and Near Port (within
5000m) block groups that were similar or different between the two (Figure 21). We see that the
majority of EPA near-port block groups (61% for Very Near Port and 93% for Near Port) were also ACE
near-port block groups. Yet, each port geometry retained distinct populations not found in the vicinity of
the other. Because the ACE port geometries captured more near-port block groups in total, a lower
percentage were also shared by the EPA near-port populations (33% for Very Near Port and 51% for
Near Port).

Given these differences in overall area and the distinct populations captured as 'near-port'
between the two geometries, we chose to analyze and report the results of the two port geometry
datasets separately, rather than as ranges or averages. As Figure 21 shows, there is a large degree of
overlap in near-port block groups of the EPA and ACE port geometries, but we observe that the two
near-port populations are distinct.

This distinction is most clearly manifested in the stark differences in the population totals (Table
3) and further underscores the pivotal role that underlying source geometries play in developing
proximity analyses. Neither ACE nor EPA port geometries represent official port boundaries, and port-
related operations can occur outside these estimated limits. Questions remain about the NEI polygons
that comprise the EPA dataset in this analysis, including how the specific set of ports were determined
given that ACE's Master Docks Plus included many more ports. Due to uncertainties involved in defining
areas of port activity and port boundaries as well as recent port reorganizations, we cannot accurately
report port-level results that represent current conditions on the ground using either the EPA or the ACE
dataset. Improvements in the geospatial depiction of port operations, including geometries that are
corroborated with on-the-ground experience, would greatly enhance future quantifications of near-port
populations.

4.3 Other Study Limitations

While this study improves upon past work and is reflective of the most recent agency guidance,
it does have limitations. One limitation of this analysis stems from aggregating results to the national
level. The 123 ports included in this study are diverse, representing a variety of geographic regions,
urban/rural classifications, and port activities. We reasonably assume that their near-port populations
are also sociodemographically unique, as some areas of the country have greater shares of certain racial,
ethnic, and socioeconomic groups than others. Aggregating all these near-port areas in a national-level
analysis may mask demographic patterns and disproportionalities that exist on a port-by-port or
regional basis.

Another limitation of this analysis stems from the differing resolutions of the Census data and
the dasymetric population model. To allocate Census sociodemographic characteristics to the more
highly resolved dasymetric population counts, we assumed that sociodemographic characteristics are
distributed evenly throughout a block group. Because equally high-resolution Census data are not
available, we cannot perfectly characterize the differences between the populations on opposite sides
of the near-port buffer boundary when census block groups are bisected by the buffer.

40


-------
We were unable to include Alaska, Hawaii, Puerto Rico, the U.S. Virgin Islands, and other
territories and insular areas outside of CONUS because of the coverage of the dasymetric population
model used in this study. Ports in these areas serve as vital economic links to the rest of the country. By
excluding these areas, we underestimate the total number of people living near ports and may have
especially missed certain demographic groups in near-port populations (e.g., Native Hawaiian and Other
Pacific Islander or American Indian and Alaska Native). The 2020 dasymetric population model has been
released since the bulk of this study was completed, and in addition to updated data to reflect 2020
populations, this dataset includes Alaska, Hawaii, Puerto Rico, and the U.S. Virgin Islands. As a result,
future work for this effort may apply these methods to the updated population dataset to encompass
key ports beyond CONUS.

Finally, the total populations presented here are likely to be underestimations of the total
population impacted by port activity. The population data used reflects those residing near ports and
does not capture the share of people working at or around ports who may be occupationally exposed to
emissions from ports. Additionally, this study did not include every port across the country, as it was
limited to ports represented by both port geometries, the geographic extent of the dasymetric data, and
the ACE Principal Ports list. As a result of these inclusion criteria, the populations presented here
exclude those living in proximity to ports outside of the conterminous U.S., ports that primarily serve
cruise passengers or offer ferry services, and ports used by Tribes for non-commercial activities.
Additionally, because complementary shapefiles were not available in both datasets, two major U.S.
ports were excluded: the Port of Virginia and the Port of Detroit. Finally, the choice of buffer distances
used in the study (1000m and 5000m) cover a proximity that may exclude areas that are impacted by
port operations and air quality pollutants, such as particulate matter, which can be transported over
much larger distances than 5000m.

5. Conclusion

We have conducted a national analysis of 123 major ports in CONUS to understand the
sociodemographic characteristics of the 2010 near-port population in the U.S. We leveraged two
different federal datasets to represent unique definitions of ports and port-related activity. Further
characterizing the sociodemographic breakdown of two comparison groups also allowed us to identify
disproportionalities that may be indicative of disparities between near-port populations and neighboring
communities or CONUS. Ultimately, we observe that the near-port population is overrepresented by
non-Hispanic Black, Non-Hispanic Asian, and Hispanic, people of color, and individuals belonging to
socioeconomically vulnerable groups as compared to neighboring populations and the general
population of CONUS. Efforts to reduce port-related emissions could benefit as many as 16-31 million
people living near major CONUS ports and address environmental injustices for vulnerable groups who
are overrepresented in proximity to port operations.

41


-------
Works Cited

Adar, S., Adamkiewicz, G., Gold, D., Schwartz, J., Coull, B., & Suh, H. (2007). Ambient and

microenvironmental particles and exhaled nitric oxide before and after a group bus trip.
Environmental Health Perspectives, 115(4), 507-512. doi:10.1289/ehp.9386

Agrawal, H., Eden, R., Zhang, X., Fine, P. M., Katzenstein, A., Miller, J. W., . . . Cocker, D. R. (2009).

Primary Particulate Matter from Ocean-Going Engines in the Southern California Air Basin.
Environmental Science & Technology, 43(14), 5398-5402. doi:10.1021/es8035016

Baynes, J., Neale, A., & Hultgren, T. (2022). Improving intelligent dasymetric mapping population density
estimates at 30 m resolution for the conterminous United States by excluding uninhabited
areas. Earth Syst Sci Data, 14(6), 2833-2849. doi:10.5194/essd-14-2833-2022

Boogaard, H., Patton, A., Atkinson, R., Brook, J., Chang, H., Crouse, D., . .. Van Vliet, E. (2022). Long-term
exposure to traffic-related air pollution and selected health outcomes: A systematic review and
meta-analysis. Environment International, 164, 107262. doi:10.1016/j.envint.2022.107262

Boothe, V. L., Boehmer, T. K., Wendel, A. M., & Yip, F. Y. (2014, April). Residential traffic exposure and
childhood leukemia: a systematic review and meta-analysis. American Journal of Preventative
Medicine, 46(A), 413-22. doi:10.1016/j.amepre.2013.11.004

Collins, T. W., & Grineski, S. (2022). Racial/Ethnic Disparities in Short-Term PM2.5 Air Pollution
Exposures in the United States. Environmental Health Perspectives, 130(8).
doi:10.1289/EHP11479

Eastern Research Group. (2010). Documentation for the Commercial Marine Vessel Component of the
National Emissions Inventory Methodology. Filed as 'cmv_report4.pdf' within the 2008 NEI
Reference List Download. Retrieved from https://www.epa.gov/air-emissions-inventories/2008-
national-emissions-inventory-nei-data

Flater, D., & ESRI Geoprocessing Development Team. (2011). Understanding Geodesic Buffering.

Retrieved from ArcUser Online: https://www.esri.com/news/arcuser/0111/geodesic.html

Gillingham, K., & Huang, P. (2021). Racial Disparities in the Health Effects from Air Pollution: Evidence
from Ports. National Bureau of Economic Research. doi:10.3386/w29108

Greenburg, M. R. (2021). Ports and Environmental Justice in the United States: An Exploratory Statistical
Analysis. Risk Analysis, 41(11). doi:10.1111/risa.13697

Jbaily, A., Zhou, X., Liu, J., Lee, T.-H., Kamareddine, L., Verguet, S., & Dominici, F. (2022). Air pollution
exposure disparities across US population and income groups. Nature, 601(7892), 228-233.
doi:10.1038/s41586-021-04190-y.

Karner, A., Eisinger, D., & Niemeier, D. (2010). Near-roadway air quality: synthesizing the findings from
real-world data. Environmental Science & Technology, 44(14), 5334-5344.
doi:10.1021/esl00008x

42


-------
Laden, F., Hart, J., Smith, T., Davis, M., & Garshick, E. (2007). Cause-specific mortality in the unionized
U.S. trucking industry. Environmental Health Perspectives, 115(8), 1192-1196.
doi:10.1289/ehp. 10027

Marshall, J. D. (2008). Environmental inequality: Air Pollution exposures in California's South Coast Air
Basin. Atmospheric Environment, 42(21), 5499-5503. doi:10.1016/j.atmosenv.2008.02.005

Mohai, P., Pellow, D., & Roberts, J. T. (2009). Environmental Justice. Annual Review of Environment and
Resources, 34, 405-430. doi:10.1146/annurev-environ-082508-094348

Peters, A., von Klot, S., Heier, M., Trentinaglia, I., Hormann, A., Wichmann, H., & Lowel, H. (2004).

Exposure to traffic and the onset of myocardial infarction. New England Journal of Medicine,
351(17), 1721-1730. doi:10.1056/NEJMoa040203

Pickard, B. R., Daniel, J., Mehaffey, M., Jackson, L. E., & Neale, A. (2015). EnviroAtlas: A new geospatial

tool to foster ecosystem services science and resource management. Ecosystem Services, 14, 45-
55. doi:10.1016/j.ecoser.2015.04.005

Rosenbaum, A., Hartley, S., & Holder, C. (2011). Analysis of Diesel Particulate Matter Health Risk

Disparities in Selected US Harbor Areas. American Journal of Public Health, 101(S1), S217-S223.
doi:10.2105/AJPH.2011.300190

Rowangould, G. M. (2013). A census of the near-roadway population: public health and environmental
justice considerations. Transportation Research Part D: Transport and Environment, 25, 59-67.
doi:10.1016/j.trd. 2013.08.003.

Svendsen, E. R., Reynolds, S., Ogunsakin, O. A., Williams, E. M., Fraser-Rahim, H., Zhang, H., & Wilson, S.
M. (2014). Assessment of Particulate Matter Levels in Vulnerable Communities in North
Charleston, South Carolina prior to Port Expansion. Environmental Health Insights, 8, 5-14.
doi:10.4137/EHI.S12814

Tessum, C. W., Paolella, D. A., Chambliss, S. E., Apte, J. S., Hill, J. D., & Marshall, J. D. (2021). PM2.5

polluters disproportionately and systemically affect people of color in the United States. Science
Advances, 7(18). doi:10.1126/sciadv.abf4491

U.S. Army Corps of Engineers Institute of Water Resources. (2018). Principal ports of the United States,
2017. Retrieved from USACE Digital Library:

https://usace.contentdm.oclc.Org/digital/collection/pl6021coll2/id/3114/rec/5

U.S. Army Corps of Engineers Navigation and Civil Works Decision Support Center, Waterborne

Commerce Statistics Center. (2019). Master Docks Plus Public Extract. Retrieved May 2019, from
https://ndclibrary.sec.usace.army.mil/resource/ed0949e6-19al-4767-9fbd-17d0de5f727e

U.S. Census Bureau, (n.d.). Understanding Geographic Identifiers (GEOIDs). Retrieved from Guidance for
Geography Users: https://www.census.gov/programs-surveys/geography/guidance/geo-
identifiers.html

U.S. Department of Transportation. (2021). Areas of Persistent Poverty & Historically Disadvantaged
Communities. Retrieved from RAISE: https://www.transportation.gov/RAISEgrants/raise-app-
hdc

43


-------
U.S. Environmental Protection Agency. (2014). 2011 Port Shape Files. Retrieved from 2011 National

Emissions Inventory (NEI) Data: https://www.epa.gov/air-emissions-inventories/2011-national-
emissions-inventory-nei-data

U.S. Environmental Protection Agency. (2015). Dasymetric Allocation of Population, Conterminous U.S.,
2010 v.3, Raster. Retrieved May 2021, from https://www.epa.gov/enviroatlas/data-download

U.S. Environmental Protection Agency. (2016). Technical Guidance for Assessing Environmental Justice in
Regulatory Analysis. Retrieved July 2024, from https://www.epa.gov/sites/default/files/2016-
06/documents/ejtg_5_6_16_v5.1.pdf

U.S. Environmental Protection Agency. (2017). 2014 Port Shape Files. Retrieved from 2014 National

Emissions Inventory (NEI) Data: https://www.epa.gov/air-emissions-inventories/2014-national-
emissions-inventory-nei-data

U.S. Environmental Protection Agency. (2018). EnviroAtlas Data Fact Sheet: Threshold Income for Quality
of Life. Retrieved September 2024, from EnviroAtlas Fact Sheets:

https://enviroatlas.epa.gov/enviroatlas/DataFactSheets/pdf/Supplemental/Thresholdincomefor
qualityoflife.pdf

U.S. Environmental Protection Agency. (2020). Retrieved from 2017 National Emissions Inventory (NEI)
Data: https://www.epa.gov/air-emissions-inventories/2017-national-emissions-inventory-nei-
data

U.S. Environmental Protection Agency. (2020). Ports Primer for Communities Glossary: "Port". Retrieved
February 2023, from Ports Initiative:

https://sor.epa.gov/sor_internet/registry/termreg/searchandretrieve/glossariesandkeywordlists
/search.do?details=&vocabName=Ports%20Primer%20Glossary

U.S. Environmental Protection Agency. (2021). Request for Applications for 2021 Diesel Emissions
Reduction Act (DERA) National Grants. Section I.B.7.b.2; page 19. Retrieved from
https://www.epa.gov/sites/default/files/2021-03/documents/2021-3-2-dera-rfa-final.pdf

U.S. Environmental Protection Agency. (2022). Port and Goods Movement Emission Inventories.

Retrieved 2024, from U.S. EPA Ports Initiative: https://www.epa.gov/ports-initiative/port-and-
goods-movement-emission-inventories

U.S. Environmental Protection Agency. (2023). Analytical Tools Interface for Landscape Assessments
(ATtlLA) Toolbox. Retrieved July 2024, from https://www.epa.gov/enviroatlas/attila-toolbox

U.S. Environmental Protection Agency. (2023). DRAFT 2023 Technical Guidance for Assessing
Environmental Justice in Regulatory Analysis. Retrieved from
https://www.epa.gov/system/files/documents/2023-
ll/ejtg_revision_110823_508compliant_0.pdf

U.S. Environmental Protection Agency. (2024). Clean Ports Program. Retrieved September 2024, from
U.S. EPA Ports Initiative: https://www.epa.gov/ports-initiative/cleanports

U.S. Environmental Protection Agency. (2024). EPA Collaboration on International Air Pollution

Standards for Ships. Retrieved September 2024, from Regulations for Emissions from Vehicles

44


-------
and Engines: https://www.epa.gov/regulations-emissions-vehicles-and-engines/epa-
collaboration-international-air-pollution-0

U.S. Environmental Protection Agency, (n.d.). Diesel Emissions Reduction Act(DERA) Funding. Retrieved
September 2024, from https://www.epa.gov/dera

U.S. Environmental Protection Agency, (n.d.). Learn About How Mobile Source Pollution Affects Your

Health. Retrieved 2024, from https://www.epa.gov/mobile-source-pollution/learn-about-how-
mobile-source-pollution-affects-your-health

U.S. Environmental Protection Agency, (n.d.). Ports Initiative. Retrieved September 2024, from
https://www.epa.gov/ports-initiative

Valencia, A., Serre, M., & Arunachalam, S. (2023). A hyperlocal hybrid data fusion near-road PM2.5 and
N02 annual risk and environmental justice assessment across the United States. PLOS ONE.
doi:10.1371/journal. pone.0286406

Weaver, G. M., & Gauderman, W. J. (2018). Traffic-Related Pollutants: Exposure and Health Effects
Among Hispanic Children. American Journal of Epidemiology, 187(1), 45-52.
doi:10.1093/aje/kwx223

Zanobetti, A., Stone, P., Spelzer, F., Schwartz, J., Coull, B., Suh, H., . . . gold, D. (2009). T-wave alternans,
air pollution and traffic in high-risk subjects. American Journal of Cardiology, 104(5), 665-670.
doi:10.1016/j.amjcard.2009.04.046

45


-------
Appendices

A. List of Ports Included in Study

Table A-l. List of ports included in study (n=123).

Port Name

Port ID

Port Name

Port ID

Albany, NY

C0505

Morehead City, NC

C0764

Alpena, Ml

L3617

Mount Vernon, IN

12332

Anacortes, WA

C4730

Nashville, TN

12370

Ashtabula, OH

L3219

New Castle, DE

C0299

Baltimore, MD

C0700

New Haven, CT

C1507

Baton Rouge, LA

C2252

New Orleans, LA

C2251

Beaumont, TX

C2395

New York, NY and NJ

C0398

Boston, MA

CO 149

Oakland, CA

C4345

Bridgeport, CT

C0311

Olympia, WA

C4718

Brownsville, TX

C2420

Palm Beach, FL

C2162

Brunswick, GA

C0780

Panama City, FL

C2016

Buffington, IN

L3737

Pascagoula, MS

C2004

Burns Waterway Harbor, IN

L3739

Paulsboro, NJ

C5252

Calcite, Ml

L3620

Penn Manor, PA

C0298

Camden-Gloucester, NJ

C0551

Pensacola, FL

C2007

Charleston, SC

C0773

Philadelphia, PA

C0552

Chattanooga, TN

12372

Pittsburgh, PA

12358

Chester, PA

C0297

Plaquemines, LA, Port of

C2255

Chicago, IL

L3749

Port Angeles, WA

C4708

Cincinnati-Northern KY, Ports of

12338

Port Arthur, TX

C2416

Cleveland, OH

L3217

Port Canaveral, FL

C2160

Conneaut, OH

L3220

Port Dolomite, Ml

L3627

Coos Bay, OR

C4660

Port Everglades, FL

C2163

Corpus Christi, TX

C2423

Port Fourchon, LA

C1910

Drummond Island, Ml

L3813

Port Hueneme, CA

C4150

Duluth-Superior, MN and Wl

L3924

Port Inland, Ml

L3803

Escanaba, Ml

L3795

Port Jefferson, NY

C0522

Everett, WA

C4725

Port Manatee, FL

C2023

Fairport Harbor, OH

L3218

Portland, ME

C0128

Freeport, TX

C2408

Portland, OR

C4644

Galveston, TX

C2417

Portsmouth, NH

C0135

Gary, IN

L3736

Presque Isle, Ml

L3845

Grays Harbor, WA

C4702

Providence, Rl

C0191

Greenville, MS

12271

Redwood City, CA

C4340

Gulfport, MS

C2083

Richmond, CA

C4350

Guntersville, AL

12371

San Diego, CA

C4100

Helena, AR

12293

San Francisco, CA

C4335

46


-------
Hopewell, VA

C0738

Savannah, GA

C0776

Houston, TX

C2012

Searsport, ME

C0112

Huntington - Tristate

12348

Seattle, WA

C4722

Iberia, LA

C2030

Silver Bay, MN

L3928

Indiana Harbor, IN

L3738

South Louisiana, LA, Port of

C2253

Jacksonville, FL

C2017

Southeast Missouri Port, MO

12301

Kalama, WA

C4626

St. Clair, Ml

L3509

Kansas City, MO

12385

St. Louis, MO and IL

12310

Lake Charles, LA

C2254

St. Paul, MN

12320

Lake Providence, LA

12269

Stockton, CA

C4270

Long Beach, CA

C4110

Stoneport, Ml

L3619

Longview, WA

C4622

Tacoma, WA

C4720

Lorain, OH

L3216

Tampa, FL

C2021

Los Angeles, CA

C4120

Terrebonne, LA, Port of

C2224

Louisville, KY

12333

Texas City, TX

C2404

Marblehead, OH

L3212

Toledo, OH

L3204

Marcus Hook, PA

C5251

Tulsa, Port of Catoosa, OK

16109

Marquette, Ml

L3844

Two Harbors, MN

L3926

Matagorda Port Lv Pt Com, TX

C2410

Vancouver, WA

C4636

Memphis, TN

12294

Vicksburg, MS

12276

Miami, FL

C2164

Victoria, TX

C2411

Milwaukee, Wl

L3756

Wilmington, DE

C0554

Mobile, AL

C2005

Wilmington, NC

C0766

Monroe, Ml	L3202

B. Summary of Geospatial Port Data Sources

Table A-2. Summary of geospatial port data sources used in this study.

Geospatial Representation
featured in this study

Data Source(s)

EPA Port Polygons
(EPA)

Polygons: 135 polygons
representing 100 ports from the
2011 NEI; and 40 polygons
representing 23 ports from the
2014 NEI

U.S. Army Corps of
Engineers Active
Docks (ACE)

Points: 6,728 points representing
123 ports in the Conterminous U.S.

2011 and 2014 National Emissions Inventory.
Accessed May 2019 from:
https://www.epa.gov/air-emissions-
inventories/2011-national-emissions-
inventory-nei-data and
https://www.epa.gov/air-emissions-
inventories/2014-national-emissions-
inventory-nei-data

Master Docks Plus Public Extract. Accessed May
2019 from:

https ://ndcl i bra ry.sec.usace.army.mil/resource/
ed0949e6-19al-4767-9fbd-17d0de5f727e)

47


-------
C. Summary of Port Definitions from Various Federal Agencies and Programs

There is no single definition of a "port" and below is an illustrative list of port definitions from various
Federal agencies and programs. This list is meant to illustrate the variety of definitions in use and does
not necessarily reflect all definitions currently in use.

Environmental Protection Agency

From EPA Clean Ports Program (source: Clean Ports Program: Zero Emission Technology Deployment
Competition Notice of Funding Opportunity)

•	Water port: Places on land alongside navigable water (e.g., oceans, rivers, or lakes) with one or
more facilities in close proximity for the loading and unloading of passengers or cargo from
ships, ferries, and other commercial vessels. This includes facilities that support non-commercial
Tribal fishing operations.

From EPA Diesel Emission Reduction Act (source: Diesel Emissions Reduction Act Request for
Applications, FY2021)

•	Ports: Places alongside navigable water with facilities for the loading and unloading of
passengers and/or cargo from ships, ferries, and other vessels.

U.S. Department of Transportation
U.S. Maritime Administration

From MARAD Port Infrastructure Development Program (source: Port Infrastructure Development
Program under the Infrastructure Investment and Jobs Act and Consolidated Appropriations Act, 2024
Notice of Funding Opportunity (A.3. Definitions)

•	Coastal seaport: A port on navigable waters of the United States or territories that is subject to
the U.S. Army Corps of Engineers regulatory jurisdiction for oceanic and coastal waters under 33
C.F.R. 329.12 or that is otherwise capable of receiving oceangoing vessels with a draft of at least
20 feet (other than a Great Lakes port).

•	Great Lakes port: A port on the Great Lakes and their connecting and tributary waters as
defined under 33 C.F.R. 83.03(o).

•	Inland river port: A harbor, marine terminal, or other shore side facility used principally for the
movement of goods that is not at a coastal seaport or Great Lakes port.

•	Small Port: A coastal seaport, Great Lakes, or inland river port to and from which the average
annual tonnage of cargo for the immediately preceding three calendar years from the time an
application is submitted is less than 8,000,000 short tons, as determined by using U.S. Army
Corps of Engineers data or data by an independent audit if the Secretary determines that it is
acceptable to use such data instead of using U.S. Army Corps of Engineers data.

48


-------
From MARAD Glossary of Shipping Terms (2008):

•	Port: Harbor with piers or docks.

Bureau of Transportation Statistics

From BTS Port Performance Freight Statistics Program Glossary:

•	Port:

(1)	The land, facilities, and adjacent body of water located on a coast, river, or Great Lake where
cargo is transferred between ships and other ships, trucks, trains, pipelines, or storage facilities.
A port is typically located within a harbor;

(2)	A place in which vessels load and discharge cargoes and passengers. Facilities normally
include berths, cargo handling equipment and personnel, cargo storage facilities, and land
transportation connections. Often with a city, town, or industrial complex.

Excerpt from BTS 2016 Port Performance and Freight Statistics Annual Report:

Section 2.1: Definition of Ports

Ports are commonly recognized as places where cargo is transferred between ships and trucks,
trains, pipelines, storage facilities, or refineries. Ports are more difficult to define for statistical
purposes when such places are close to one another or when activity related to a port blends in
with surrounding neighborhoods. Many ports are located adjacent to closely related land uses
(e.g., railyards and truck depots) or to other ports. Continuous waterfront may be divided into
separate ports by administrative boundaries, such as the Ports of Los Angeles and Long Beach, or
the series of Mississippi River terminals in Louisiana divided between the Ports of New Orleans
and Baton Rouge. In contrast, the Port of New York and New Jersey and the Ports of Cincinnati
and Northern Kentucky are treated as single entities, even though the former has a river and a
state line dividing its facilities and the latter has terminals that stretch along 226 miles of two
rivers through two states. Further, for more detailed performance assessments, the appropriate
entity may be an individual terminal, not a port comprised of multiple terminals with diverse
ownership, cargo, and operating methods.

The Federal government defines ports in many different ways. For example, U.S. Customs and
Border Protection (CBP) defines some "ports" as a single port and others as units comprising
multiple ports. The U.S. Census Bureau relies on the CBP definitions for reporting on trade. The
USDOT Maritime Administration (MARAD) defines a port as "a harbor with piers or docks" in its
Glossary of Shipping Terms.

The U.S. Army Corps of Engineers (USACE) identifies ports in different ways for planning and
managing port and waterway improvement projects and for the collection and tabulation of
waterborne commerce statistics. The USACE Waterborne Commerce Statistics Center (WCSC)
aligns ports with their enacting legislation. In contrast, a USACE project area may encompass
multiple ports along a shared stretch of water (like the Ports of Los Angeles and Long Beach
which are both assigned to the same harbor), or multiple projects might be encompassed by a
single port (as is the case with the Port of New York and New Jersey).

49


-------
Ports are organized and governed in a variety of ways, with implications for port definition and
data availability. Most ports are governed by port authorities or harbor districts, usually part of
local government Some governing bodies are state entities (e.g., the Maryland or Georgia Port
Authorities) or interstate authorities (e.g., The Port Authority of New York and New Jersey). A
port's jurisdiction typically extends over land, where it may include concession and construction
approval and policy decision-making authorization, and over water, where it is primarily focused
on navigation.

A port authority is a government entity that either owns or administers the land, facilities, and
adjacent water body where cargo is transferred between modes. A port authority promotes
overall port operating efficiency and development, maintains port facilities, and interacts with
other government bodies. Additional activities include business development and infrastructure
finance. While the structure, powers, and role of port authorities vary, the American Association
of Port Authorities (AAPA) states that they "share the common purpose of serving the public
interest of a state, region or locality." Port authorities may act as:

o Landlords, building and maintaining terminal infrastructure and providing major capital
equipment, but are not engaged in operations. The Ports of Los Angeles, New York and
New Jersey, and Oakland are examples of landlord ports. Ports may also offer
concessions to tenants that make infrastructure improvements. For example, the
Maryland Port Administration granted a 50-year concession for the Baltimore Seagirt
Marine Terminal that included a commitment by the concessionaire to deepen the
channel.

o Operating ports, directly operating some or all of the terminals in the jurisdiction. For
example, the Port of Houston Authority is an operating port.

o Jurisdictional bodies, under which private terminals are responsible for providing and
operating their infrastructure. For example, the Ports of Cincinnati and Northern
Kentucky is a jurisdictional body.

A port may own and operate an extensive range of facilities over a large area, many of which
may not be water related. Several port authorities (e.g., Port of Oakland, Massachusetts Port
Authority) also operate airports. The Port Authority of New York and New Jersey operates
airports, tunnels, bridges, and transit systems as well as the seaport.

Some states, such as South Carolina and Georgia, have statewide port authorities to administer
some or all of the ports within their jurisdiction. These entities are typically led by boards of
appointed members. They may also directly operate port facilities within the state. A state port
authority may be a separate state department, or be located within that state's DOT.

Some port authority jurisdictions cross state boundaries. The Port Authority of New York and
New Jersey and the Ports of Cincinnati and Northern Kentucky are examples.

Port authorities typically have jurisdiction over public terminals. Private (usually bulk) terminals
are normally outside the public port authorities' jurisdiction although they are still subject to U.S.
Coast Guard and Federal regulation.

50


-------
U.S. Army Corps of Engineers

From USACE Engineering Pamphlet 1130-2-520:

• Port Area:

(1)	Port limits defined by legislative enactments of state, county, or city governments.

(2)	The corporate limits of a municipality.

Geospatial data of defined port areas represented as polygons is available from USACE
Geospatial Platform.

From USACE Master Docks:

Geospatial data for docks designated by port ID are represented as points and are available from
USACE Waterborne Commerce Statistics Center Master Docks Plus.

D. Additional Notes about the Dasymetric Model

This study used a highly resolved population distribution model to quantify the population living
near ports (see 2.2 Higher Resolution Population Estimates for more) to align with a larger cross-EPA
effort to quantify populations living near transportation infrastructure. Table A-3 shows the comparison
of between how the near port populations by port geometry and buffer distances differed between
these two population models. Note, the population totals for the EPA Port Geometries listed below
differ from the totals included and discussed in the body of this report due to using a pre-processed raw
data file to accurately compare equal area and dasymetric estimates.

Table A-3. Comparison of Equal Area and Dasymetric Population Distribution Totals

Near-port population estimates using equal area approach versus dasymetric model approach

Buffer size



O-lOOm

0-200m

0-500m

O-lOOOm

0-5000m

ACE (n=123)

Dasymetric model estimate

24,473

123,354

1,134,075

4,676,199

31,058,906

Equal area estimate

84,631

279,276

1,407,661

4,840,956

30,988,627

Difference a

60,158

155,922

273,586

164,757

-70,279

% difference

245.8%

126.4%

24.1%

3.5%

-0.2%

NEI (n=404)b

Dasymetric model estimate

481,571

697,418

1,531,146

3,299,055

22,748,983

Equal area estimate

601,780

818,132

1,590,573

3,268,649

22,600,496

Difference a

120,209

120,714

59,427

-30,406

-148,487

% difference

25.0%

17.3%

3.9%

-0.9%

-0.7%

a Negative difference and percent difference values indicate that the population estimates were lower using the equal area
method than they were using the dasymetric method.

b The near-port population estimates using NEI port geometries included in this table do not align with the near-port
population estimates from the primary analysis (see Table 3). This inconsistency is because equal area estimates were only
available for an earlier version of NEI port geometries that represented more than the 123 major CONUS ports that were
included in this study. The comparison of these equal area and dasymetric population model approach estimates still
illustrate the potential differences in results that could be expected between the two proximity analysis methodologies.

51


-------
We found that near-port population estimates derived using an equal area approach tended to
overestimate the population living in very close proximity (O-lOOm, 0-200m, and 0-500m) as compared
to the dasymetric population model approach. This finding implies that there are uninhabitable areas in
very close proximity to ports that are captured by the high-resolution dasymetric population model. The
dasymetric near-port populations living in proximity to the ACE port geometries were more disparate
from the equal area approach estimates than those in proximity to the NEI port geometries. This finding
may be illustrative of the differences in the two port geometries' land versus waterfront coverage. The
overestimation using the equal area approach as compared to the dasymetric population model
approach attenuated with larger buffer sizes.

Additionally, for some Census block groups (n=46), there is a small discrepancy between the
Census total population and the dasymetric modeled population (difference of less than 1). This
discrepancy occurs when a pixel of the dasymetric grid excludes the population of a very small block
group (e.g., a block group that is 10 meters wide). In these cases, a portion of the population of the
geographically small block group is reallocated to a neighboring block group. This reapportionment
accounts for the reallocation of approximately 1,000 people in the entire Conterminous United States.
(<0.001% of the population modeled).

E. Data Processing

EPA's Office of Transportation and Air Quality (OTAQ) reviewed published NEI Shapefiles and
Master Docks Plus coordinates, selected a subset of ports, and provided port shapefiles to ORD. ORD
used ArcGIS Pro to create geodesic buffers at 100m, 200m, 500m, 1000m, and 5000m around each port
geometry. The Analytical Tools Interface for Landscape Assessments (ATtlLA) ESRI ArcGIS toolbox47 was
used to extract the dasymetric raster population from these buffers and assign them to the relevant
overlapping Census block group. ATtlLA was run a total of four times: for both the EPA and ACE port
geometries, once with all ports together as a single, multi-polygon object, and once with all ports
separately, categorized by their assigned name. ORD then provided the within-buffer dasymetric
population estimates by block group in tabular form to OTAQ.

OTAQ merged the tabularized output from ORD with block group-level demographic data
obtained from the U.S. Census (2010 Decennial Census and 2014-2018 American Community Survey),
EnviroAtlas (2017), and tract-level data for the U.S. DOT RAISE Areas of Persistent Poverty (FY2021).
Data was processed using R version 4.4.1.

47 U.S. EPA. ATtlLA Toolbox: Analytical Tools Interface for Landscape Assessments (ATtlLA). Accessed July 2024:
ATtlLA Toolbox I US EPA

52


-------
F. Supplemental Figures

Summary of Racial and Ethnic Demographic Differences between Near Port and Comparison Populations

48%	[46%

55% |

64%

164% J

52%	[54%j

145% |

54%)

24% |

116% |















12% '



112% I

0.7M



6.6M



8.8M



37.6M

126% |

21% I	120%

22% I



23% I

120% J













116% |













1.2M



7M



14M



50.3M

fo%1

0%





	[o%]	







<0.1M







0%

<0.1 M





0.2M



0.6M

®	 d%]

(2%| >%	L|%

^ l°*T

	To%

Very Near Port Near Port Population, Intra-county Conterminous US	Very Near Port Near Port Population, Intra-county Conterminous US

Population, 0-1000m 0-5000m Comparable Group	Population, 0-1000m 0-5000m Comparable Group

Figure A-1. Summary of Racial and Ethnic Demographic Differences between Near-Port and Comparison Populations

53


-------
Summary of Economic Demographic Differences between Near Port and Comparison Populations

EPA Port Geometry	ACE Port Geometry

100%-	88%

60%- 54%

o o%-[
^ 50% n

o%-

25% -

10%

Very Near Port Near Port Population, Intra-county Conterminous US Very Near Port Near Port Population, Intra-county Conterminous US
Population, 0-1000m 0-5000m Comparable Group	Population, 0-1000m 0-5000m Comparable Group

Figure A-2. Summary of Economic Demographic Differences between Near-Port and Comparison Populations

54


-------
Summary of Other Demographic Differences between Near Port and Comparison Populations

Very Near Port Near Port Population, Intra-county Conterminous US	Very Near Port Near Port Population, Intra-county Conterminous US

Population, 0-1000m 0-5000m Comparable Group	Population, 0-1000m 0-5000m Comparable Group

Figure A-3. Summary of Other Demographic Differences between Near-Port and Comparison Populations

55


-------
G. Alternative Comparison Groups Considered
Neighboring Block Groups

The 'Neighboring Block Groups' border near-port block groups (i.e., those that intersect the 0-
5000m buffer (Figure A-l). This comparison group was created by EPA ORD at OTAQ's request as a first
attempt to create a smaller and more regionally representative comparison group than the entire
Conterminous United States. This population is assumed to not be adversely affected by port
operations' mobile source emissions.

K. - ' \

5*..

. \

} \ %
t - ¦ \ A

: .A/

¦

Figure A-4. Rendering of Neighboring Block Groups, shown in red, around the Port of Wilmington, NC (using the EPA port
polygon).

The benefits of the Neighboring Block Groups comparison group are that it is hyper-localized
and may offer more insight into regional demographic dynamics than a nationwide comparison.
Additionally, this method utilizes entire block group populations as it is a 'contiguity-based neighbor'
classification. By doing so, it avoids further sub-setting block group populations and applying any
assumption about population demographic distributions within these neighboring block groups.
Furthermore, we can easily compare the demographics between the near-port and comparison groups,
because the Neighboring Block Groups do not have any overlap with the near-port populations.

Upon generating these comparison groups, it became clear that there were significant
challenges associated with them. First, the method used to create the Neighboring Block Groups is
complex and computationally intensive.48 Additionally, block groups are not uniformly sized or shaped,
and the decision to rely on a contiguity-based neighbor classification may have made it seem like there
was an epidemiologic-based rationale for such a close neighbor, while the exact distance from a port at
which a population is no longer burdened is still under consideration. Because of these complexities, the

48 Finding the shared boundary of polygons can be addressed using the Polygon Neighbors analysis tools in ArcGIS
Pro; other methods are available through other geospatial analytical tools, such as R's spdep package.

r\

J*. _

AAAAA
A? <--y

'-••-J

56


-------
Neighboring Block Groups were only created for the EPA port polygon dataset, after which it was
determined that the Intra-County Comparison Group would be faster and simpler to create, while
offering similar benefits as the Neighboring Block Groups.

Rural-Urban Continuum Codes

We considered using the U.S. Department of Agriculture's Economic Research Service county-
binning Rural-Urban Continuum Codes— (RUCC) to compare populations around ports with populations
in different landscapes. However, port buffers often span multiple RUCC codes; additionally, there were
insufficient populations near rural ports in comparison to urban and small city near-port areas. Because
the RUCC divisions did not produce robust sub-national groupings, we decided to not use them in this
analysis.

Balance of State Population

The balance of a state's near-port population was also deemed to be an unsuitable comparison
group because port districts often cross state lines. Regional categorization of ports by East Coast, West
Coast, Gulf Coast, Inland, and Great Lakes was also considered, but we determined this option to be
inappropriate because of how ports can fall into more than one category (e.g., Florida ports, river ports
branching to/from Great Lakes).

H. Authors and Acknowledgements

This report is the result of a cross agency team including: Chad Bailey, Deirdre Clarke, Sarah
Froman, Sarah Harrison, Marion Hoyer, Ali Kamal, and Grace Kuiperfrom EPA's Office of Transportation
and Air Quality; Jeremy Baynes, Annie Neale, and Jeremy Schroeder from EPA's Office of Research and
Development; and Laurina Bird, Stepp Mayes, and Asa Watten from the Oak Ridge Institute for Science
and Education (ORISE) Research Participation Program hosted by U.S. EPA.

The authors thank Margaret Zawacki for her review of an early draft of this document, and peer
reviewers Michael Aldridge, Elizabeth Chan, and Harold Rickenbacker. The authors also want to thank
Eloise Anagnost of OTAQfor her assistance in optimizing the layout and formatting of the report.

49 United States Department of Agriculture. Rural-Urban Continuum Codes. Accessed July 2024 Rural-Urban
Continuum Codes

57


-------