Skip to main content
Open access
Research
8 March 2023

Inequitable Exposures to U.S. Coal Power Plant–Related PM2.5: 22 Years and Counting

Publication: Environmental Health Perspectives
Volume 131, Issue 3
CID: 037005

Abstract

Background:

Emissions from coal power plants have decreased over recent decades due to regulations and economics affecting costs of providing electricity generated by coal vis-à-vis its alternatives. These changes have improved regional air quality, but questions remain about whether benefits have accrued equitably across population groups.

Objectives:

We aimed to quantify nationwide long-term changes in exposure to particulate matter (PM) with an aerodynamic diameter 2.5μm (PM2.5) associated with coal power plant SO2 emissions. We linked exposure reductions with three specific actions taken at individual power plants: scrubber installations, reduced operations, and retirements. We assessed how emissions changes in different locations have influenced exposure inequities, extending previous source-specific environmental justice analyses by accounting for location-specific differences in racial/ethnic population distributions.

Methods:

We developed a data set of annual PM2.5 source impacts (“coal PM2.5”) associated with SO2 emissions at each of 1,237 U.S. coal-fired power plants across 1999–2020. We linked population-weighted exposure with information about each coal unit’s operational and emissions-control status. We calculate changes in both relative and absolute exposure differences across demographic groups.

Results:

Nationwide population-weighted coal PM2.5 declined from 1.96μg/m3 in 1999 to 0.06μg/m3 in 2020. Between 2007 and 2010, most of the exposure reduction is attributable to SO2 scrubber installations, and after 2010 most of the decrease is attributable to retirements. Black populations in the South and North Central United States and Native American populations in the western United States were inequitably exposed early in the study period. Although inequities decreased with falling emissions, facilities in states across the North Central United States continue to inequitably expose Black populations, and Native populations are inequitably exposed to emissions from facilities in the West.

Discussion:

We show that air quality controls, operational adjustments, and retirements since 1999 led to reduced exposure to coal power plant related PM2.5. Reduced exposure improved equity overall, but some populations continue to be inequitably exposed to PM2.5 associated with facilities in the North Central and western United States. https://doi.org/10.1289/EHP11605

Introduction

Electricity generation from U.S. coal power plants was historically a major contributor to poor air quality—notably in the form of particulate matter (PM) and ozone from sulfur and nitrogen emissions.13 To curb pollution from coal power plants, the U.S. Congress enacted coal power plant emissions regulations under the 1990 Clean Air Act Amendments, and the U.S. Environmental Protection Agency (U.S. EPA) has further tightened emissions standards under the 1990 Amendments with the Acid Rain Program, NOx Budget Trading Program and State Implementation Plan (SIP) Call, Clean Air Interstate Rule, and Cross-State Air Pollution Rule.47 National Ambient Air Quality Standards (NAAQS), tightened periodically by the U.S. EPA and attained through state actions, have led to further emissions reductions.8 Since the early 2010s, much of the new electricity generation capacity in the U.S. electricity generation sector has come from natural gas; no new coal power plants have been built in the contiguous United States since 2015.9
These programs resulted in regional improvements in air quality and health.1015 In addition, interventions on individual facilities have been shown to improve air quality and population health in local populations.1618 It is less clear, however, whether the overall benefits of these regulations have accrued equally across different populations; researchers have identified exposure inequities related to U.S. power plant air pollution impacts,19,20 but evidence is lacking on how major emissions reductions over the last two decades have reduced disparities. Although the Clean Air Act Amendments did not address inequity directly, multiple executive orders21,22 have directed the U.S. EPA to address environmental justice (EJ). Notably, the Biden administration’s Justice40 initiative establishes the goal that 40% of certain federal investments will go toward “disadvantaged communities that are marginalized, underserved, and overburdened by pollution.” Harper et al. describe how quantitative comparisons between and within population health status and exposure are important for designing regulatory actions that address EJ.23 For regulations targeting specific air pollution source categories (e.g., electricity generation units), the U.S. EPA’s definition of EJ suggests the need to quantify health and/or exposure inequities associated with the regulated source and its change in response to regulations.
Because the coal power plant sector is a combination of individual point sources at which operational decisions and control installations attributable to regulatory action happen semi-independently, thorough understanding of evolving exposure associated with dramatic sector-wide emissions reductions is best achieved by assessing each power plant’s historical contribution to exposure and inequities.19,20,24 For example, in crafting its recent climate-focused Clean Power Plan (later replaced by the now-vacated Affordable Clean Energy Act25) the U.S. EPA identified many facilities located in neighborhoods with higher environmental burdens and vulnerable populations.26 The U.S. EPA continues to provide information on EJ issues for populations surrounding power plants through its Power Plants and Neighboring Communities online tool and communicates that these issues will continue to be accounted for in future power plant regulations.27
The U.S. EPA’s tool, however, only accounts for populations living within 3 mi of the facilities (thereby missing populations impacted by pollution transport), and this tool and previous work in this area have focused on a single year’s exposure. The desire to expand this previous work to investigate trends led to the development of the HYSPLIT with Average Dispersion (HyADS) model, which is computationally nimble enough to simulate individual unit spatial impacts over long periods. Although other reduced complexity models such as the Intervention Model for Air Pollution (InMAP), the Estimating Air pollution Social Impact Using Regression model (EASIUR), and the Air Pollution Emission Experiments and Policy analysis model (APEEP) have been applied to estimate population damages and power generation source exposure inequities,20,2834 the HyADS model is appropriate for long-term simulations because it accounts for both meteorological variability and emissions changes over time.
We sought to quantify nationwide long-term changes in exposure and inequities to PM2.5 associated with coal power plant SO2 emissions because particulate sulfate, an atmospheric product of SO2, has historically been associated with most of the PM2.5-related mortalities from coal (one estimate attributed 75% of annual mortalities from electricity generation to SO2 emissions in 2005, in comparison with 7% from NOx and 14% from primary PM2.5).35 Using HyADS, we estimated annual spatial-temporal PM2.5 source impacts related to SO2 emissions and other correlated precursor emitted species (termed “coal PM2.5” here) from each of over 1,200 coal electricity-generating units (EGUs) in the United States across 1999–2020, the most extensive data set of its kind to our knowledge. We report here the extent that SO2 emissions changes from each EGU influenced national PM2.5 concentrations, and we used facility-level information to quantify the influence of three interventions—retirements, reduced operations, and emissions controls installations—on population exposure. Finally, we performed a detailed disparities assessment that accounts for location-specific differences in racial/ethnic population distributions relative to locations of power plants. The accompanying publicly available data set has further potential for application in long-term nationwide exposure, epidemiological, and regulatory accountability studies.

Methods

This work applied the HyADS model to estimate PM2.5 associated with SO2 emissions from all U.S. coal power plants (coal PM2.5). Attributing changes in coal PM2.5 exposure and exposure disparities to actions taken at each of the over 1,200 coal electricity-generation units necessitated the choice of a reduced complexity model such as HyADS over traditional full-complexity photochemical grid models such as CMAQ, CAMx, and GEOS-Chem.

Emissions and Facility Attributes Data

We employed monthly data from the U.S. EPA Clean Air Markets Division’s Air Markets Program Data (AMPD) database, which houses facility attributes (e.g., location and fuel type) and emissions (SO2 in U.S. tons) information for 480 coal electricity-generation facilities operating in the United States since at least 1999.10 Most facilities comprise multiple units—the base denomination for emissions and emissions controls—and the entire database contains 1,237 units. We supplemented AMPD data with stack height information obtained from the U.S. EPA’s 2014 National Emissions Inventory. Of the 1,237 coal units, 69% have stack height information; for the others, we assigned the mean stack height of all available units (182m).

The HyADS Model

HyADS has been applied previously to address multiple topics relevant for exposure and health impacts studies, including ranking the influence of individual coal plants on specific locations16,36,37 quantifying long-term changes in coal PM2.5 source impacts18; and estimating the relative influence of meteorological and emissions variability on coal PM2.5 source impacts.38 In addition, HyADS has been used to address questions relevant for population health, including investigating the effect of long-term nationwide coal emissions changes on hospitalizations of adults 65 y of age18; the influence of short-term interventions at power plants with high local impacts on urban asthma outcomes16; and differential toxicity of coal PM2.5 source impacts relative to bulk PM2.5.18 A key feature of HyADS is its ability to resolve individual source contributions to population exposures to coal PM2.5. Although the overall model framework of HyADS has been described in previous applications, we have made updates reflected in the model description below and in the Supplemental Information (“Updates to the HyADS Model,” Figures S1–S3).
HyADS acts as a wrapper to initiate and process many runs of the HYSPLIT air transport and dispersion model,39,40 and the output is power plant–specific PM2.5 concentrations associated with SO2 emissions. The modeling approach was based on emissions events, for each of which we tracked 100 air parcels. HYSPLIT tracks the transport and dispersion of these air parcels for 7 d using wind fields from the NCEP/NCAR Reanalysis meteorological product, an assimilation of meteorological observations from multiple platforms.41 We repeated emissions events at 6-h intervals each day [2400 hours (12:00 A.M.), 0600 hours (6:00 A.M.), 1200 hours (12:00 P.M.), and 1800 hours (6:00 P.M.)], resulting in [365 days]×[4 emissions events]=1,460 HYSPLIT runs per year for each facility. Air parcel locations were summed by month over a 36-km grid; this resolution was selected because of the regional nature of annual PM2.5 related to SO2 emissions, which is discussed in detail below. At this stage in the modeling process, the intermediate output consisted of monthly gridded air parcel counts for each emissions source.
Next, we multiplied the gridded air parcel counts by each unit’s monthly SO2 emissions. We averaged the monthly spatial emissions-weighted HyADS impacts across the year, accounting for the number of days in each month.
The resulting unitless concentrations at this stage were not directly relatable to measured air pollution species because they corresponded to only a portion of total ambient pollution. To post-process the unitless concentrations to coal PM2.5, we regressed the unitless concentrations against coal PM2.5 source impacts estimated with the Hybrid CMAQ-DDM model.2,42 Hybrid CMAQ-DDM employs the Community Multiscale Air Quality (CMAQ) model with the Direct-Decoupled Method (DDM) to simulate PM2.5 source impact sensitivities from all U.S. coal sources and corrects them using an optimization-based adjustment technique to more closely match observed PM2.5 observations (the approach for estimating coal source impacts is described in detail by Ivey et al.2). We used PM2.5 source impacts for all U.S. coal sources from the Hybrid CMAQ-DDM model in 2005 to post-process the unitless HyADS exposures to PM2.5 source impacts. As described in detail in Henneman et al.37 and updated as detailed in the Supplemental Information (“Updates to the HyADS Model,” Figures S1–S3), the postprocessing was based on the following regression (the model was trained independently for a given period; therefore, we dropped temporal terms from the notation above):
log(PM2.5,iCMAQDDM)=β0+βHyADSu=1UHyADSi,u+βHyADS2(u=1UHyADSi,u)2+(u=1UHyADSi,u)XiTβX,HyADS+ϵi,
(1)
where PM2.5,iCMAQDDM is the Hybrid CMAQ-DDM PM2.5 coal source impacts for location i, XiT is the vector of meteorological variables at location i, HyADSi,u for u=1, 2,,U are the contributions of unitless HyADS concentrations from each of U units to that location, and ϵi is assumed iid normal. β0 is the intercept, which permits some constant amount of PM2.5 predicted by Hybrid CMAQ-DDM coal PM2.5 source impacts that are not explained by HyADS (e.g., PM2.5 related to NOx or primary PM2.5 emissions not otherwise correlated with SO2 emissions). βHyADS, βHyADS2, and βX,HyADS are parameters governing the relationships between each of the corresponding variables with Hybrid CMAQ-DDM PM2.5 coal source impacts, (βX,HyADS corresponds to interactions between HyADS and each meteorological variable). We employed meteorological data from the NCEP/NCAR North American Regional Reanalysis project, including average temperature, accumulated precipitation, relative humidity, and x and y wind vectors. The meteorological data (originally on a 32-km grid) and CMAQ-DDM Hybrid PM2.5 source impacts (36km) were spatially projected to the HyADS 36-km grid. A log link was used because Hybrid CMAQ-DDM PM2.5 coal source impacts has an approximately log normal distribution.
CMAQ-DDM Hybrid coal source impacts were available in 2005 and 2006, with year 2006 results used to evaluate the model trained in 2005 (more details and model evaluations are discussed in detail in the Supplemental Information).37 HyADS employs the model trained on 2005 Hybrid CMAQ-DDM results to predict total coal PM2.5 and unit-specific coal PM2.5. We predicted total coal PM2.5 using each year’s u=1UHyADSi,u and meteorology in Model 1, and we subtracted the intercept from the predictions to eliminate the portion of CMAQ-DDM Hybrid coal source impacts not accounted for by raw HyADS. To predict each unit’s annual spatial coal PM2.5, we distributed the coal PM2.5 from all sources to each unit based on its fractional contribution to the total u=1UHyADSi,u at each location.
The model relies on the assumption that SO2 emissions are an important contributor to PM2.5 source impacts through their atmospheric processing to particulate sulfate. To the extent that emissions of other atmospheric constituents such as NOx and trace metals are correlated with SO2, their influence may be reflected in coal PM2.5. A detailed evaluation of unit-specific HyADS coal PM2.5 against GEOS-Chem adjoint sensitivities from Dedoussi et al.43 shows that population-weighted PM2.5 exposure from the two models by state and across the United States compare at levels consistent with other reduced complexity model evaluations44 (details in the Supplemental Information section “HyADS model evaluations and applications,” Tables S1–S3 and Figures S4–S6). In the comparison with GEOS-Chem, HyADS showed similar agreement with both PM2.5 from only SO2 emissions and PM2.5 from SO2+NOx emissions, suggesting some potential that the HyADS coal PM2.5 metric captures at least some PM2.5 impacts from NOx as well as those attributable to SO2. The finding that coal PM2.5 agrees with directly comparable quantities from a more complex model provides evidence that the model was sufficient for the applications described below and highlights potential influence on the findings by uncertainties related to the simplified approach to atmospheric chemistry taken by HyADS. We mitigated these uncertainties by aggregating to regional and nationwide exposure averages and exploring long-term trends instead of year-specific results.

Population-Weighted Exposure (PWE)

We employed PWE to reduce multidimensional spatial exposure fields to a single dimension by emphasizing portions of the exposure field that impact areas with high population. For gridded coal PM2.5 source impacts, we calculated population-weighted exposure PWEu,y,d from each unit or group of units u in each year y for each demographic group d as:
PWEu,y,d=i=1I[CoalPM2.5]u,y,i×(Py,d,iPy,d,total),
(2)
where i=1, 2,,I denotes grid cell locations, Py,d,i is the population of a given demographic in each grid cell, and Py,d,total is the total population in the domain of the demographic group.
Annual grid cell population was calculated by spatially apportioning U.S. Census county populations to the 36-km HyADS grid. We used annual intercensal population estimates from 2000 to 2020.45 We assigned 1 April 2000 population estimates for year 1999 population. We calculated PWu,y,d for the following racial/ethnic groups (census names in parentheses): White (White alone), Black (Black or African-American population alone), Native (American Indian and Alaska Native population alone), Asian (Asian population alone), Pacific (Native Hawaiian and Other Pacific Islander population alone), and Hispanic (population of Hispanic origin and any race). Although county data and the HyADS 36-km grid are relatively coarse measures of air pollution and demographic spatial distribution, they are justified by the regional nature of coal power plant pollution.20,43,46,47 Recent findings, for example, have shown that cross-state EGU source impacts accounted for around half of total fatalities attributable to EGU emissions,20,43 likely because local contributions to sulfate PM2.5 from SO2 emissions are less likely due to elevated stack heights and the delay introduced by atmospheric processing of SO2 to sulfate.48 HyADS uses a 36-km spatial grid; the average land area of counties in the contiguous United States is 1,555km2,49 which is of similar order as the 1,296-km2 area of a 36-km grid cell. Daouda et al. previously used a county-level analysis and a portion of the data set described here to quantify racial disparities in preterm birth outcomes attributable to EGU SO2 emissions.50

Exposure Contributed or Avoided by Regulatory or Operational Activity

Population exposure to coal PM2.5 was reduced through various actions taken on individual coal EGUs across the study period, including reduced operations, emissions controls (“scrubbers”; control technologies identified by the following labels in the AMPD data set: Dry Lime FGD, Dry Sorbent Injection, Dual Alkali, Fluidized Bed Limestone Injection, Magnesium Oxide, Sodium Based, Wet Lime FGD, Wet Limestone, and Other), and retirements. Using PWE from HyADS and data from EPA AMPD, we calculated PWE contributed by operational facilities and PWE avoided through each of these three interventions.
We used dates of unit retirements and scrubber installations listed in the AMPD data set to designate each unit’s operational or emissions control status. Additionally, we employ each unit’s annual heat input—also available in the AMPD data set—to characterize units as operating at high capacity (annual heat input above each unit’s median annual heat input reported in operational years from 1999 to 2020) or low capacity (annual heat input below median heat input). This characterization of high vs. low operational capacity allows for the quantification of exposure avoided by reduced operations. Using this information, we characterize each unit into one of six categories: a) operating at high capacity without a scrubber, b) operating at low capacity without a scrubber, c) operating at high capacity with a scrubber, d) operating at low capacity with a scrubber, e) retired without previously installing a scrubber, and f) retired after operating with a scrubber. These six operational/control categories led to seven contributed and avoided exposure designations that could be calculated using modeled PWE across subsets of years for each unit (Table 1).
Table 1 Contributed and avoided PWE designations and calculation approaches.
Naming conventionPWE calculationApplied to each unit in years that meet these criteria
Contributed: uncontrolled
Average PWE in years that meet these criteria:
Operating
Heat input above median
Before scrubber installation
Operating
Heat input above median
Before scrubber installation
Contributed: scrubber
Average PWE in years that meet these criteria:
Operating
Heat input above median
After scrubber installation
Operating
Heat input above median
After scrubber installation
Avoided: reduced operation
Difference between average PWE in two sets of years:
High-operation years
Operating
Heat input above median
Before scrubber installation
Low-operation years
Operating
Heat input below median
Before scrubber installation
Operating
Heat input below median
Before scrubber installation
Avoided: reduced operation after scrubber
Difference between average PWE in two sets of years:
High-operation years
Operating
Heat input above median
After scrubber installation
Low-operation years
Operating
Heat input below median
After scrubber installation
Operating
Heat input below median
After scrubber installation
Avoided: scrubber
Difference between average PWE in two sets of years:
No scrubber years
Operating
Heat input above median
Before scrubber installation
Scrubber years
Operating
Heat input above median
After scrubber installation
Operating
After scrubber installation
Avoided: retirement
Average PWE in years that meet these criteria:
Operating
Before scrubber installation
Retired
Avoided: retirement after scrubber
Average PWE in years that meet these criteria:
Operating
After scrubber installation
Retired
After scrubber installation
Note: PWE, population-weighted exposure.
We calculated each quantity listed in Table 1 for each unit in years that met the corresponding criteria and presented the sum of each exposure class across units. We did not include the years of scrubber installation or retirement in the PWE averaging to avoid transition years. Each unit’s potential PWE designation among these five categories remained constant across any given range of years for which its scrubber and operational status did not change. We presented the annual results as a percentage of total potential exposure in each year.
The approach was designed to explore trends across years, and the calculated values were somewhat sensitive to the criteria listed in Table 1. Therefore, the results were not precise enough to diagnose a given year’s exposure distribution across the seven categories, and we focused on overarching trends in the discussion. Sensitivity of the results to the selection of the heat input value cutoff used to define high/low operating capacity is presented in Figure S7.

PWE Inequities Accounting for Spatial Distributions of Sources and Populations

We explored power plant exposure inequities through two lenses: a) comparisons of demographic-specific PWE relative to the total population PWE and b) comparisons of demographic-specific PWE relative to an “expected” exposure disparity based on regional demographic distributions. The second comparison acknowledges that, before high voltage transmission lines made long-distance electricity transport possible in recent decades, power plants were cited nearby the population that would use the electricity and were operated mostly independently across regions.51 Even as regional electricity transport has become more viable, historical citing policies continue to influence exposure to power plants. For example, in 2010 50% of operating coal plants had been in service for 38 y,52 and power plant operations are still managed somewhat independently across regions.20 The differences between the two types of exposure inequality measures highlights the extent that regional demographic differences influence interpretation of exposure inequality from coal power plants. We adapted multiple literature-based exposure metrics to calculate source-specific PWE inequities for groups of EGUs, and we developed novel region-specific “relative expected PWE” to account for location-specific demographic makeups.

PWE Inequities

To quantify exposure inequities in region- and state-specific coal PM2.5, we presented relative and absolute comparisons of each demographic’s PWEu,y,d to the total population PWEu,y,d=all as described previously.23 Absolute PWE differences contributed by coal facilities were calculated by subtracting the PW exposure for the total population from the PW exposure from a given group:
PWEu,y,dabsolute=PWEu,y,dPWEu,y,d=all,
(3)
where PWEu,y,d=all is the population-weighted exposure across the entire population. PWEu,y,dabsolute provides the difference in exposure contributed by a unit or group of units u in year y on population group d relative to the total population. The units are in concentration units (micrograms per cubic meter here); a value of zero signifies that the coal units in question do not inequitably expose the group relative to the population average. Relatedly, relative PW exposure was calculated as the ratio between the two terms in Equation 3:
PWEu,y,drelative=PWEu,y,dPWEu,y,d=all.
(4)
PWEu,y,drelative is unitless; values represent the fractional difference of a group’s exposure relative to the total population.
The absolute comparison (PWEu,y,dabsolute) quantifies inequity in concentration (micrograms per cubic meter) units and is useful for showing how health-relevant exposures differ in space and have evolved over time. The relative comparison (PWEu,y,drelative) is in fractional units, which are useful for comparing across years and regions with large absolute exposure differences (e.g., inequities may exist in the western United States, even though absolute exposure is low because of a paucity of coal power plants). A PWEu,y,drelative value of 1 signifies that group’s exposure is equal to that of the total population. We defined all disparity measures relative to the average member of the total population, which revealed different interpretation than what may be found, for example, by comparing to the worst- or best-off member of the population.23 Such a decision was justified by coal pollution’s regional nature.
Regional differences in average population demographics mean that a given facility located in an arbitrary location is expected to contribute to exposure disparities that simply reflect that area’s demographics. For example, the southeastern United States’ population is characterized by a higher percentage of Black or African-American people than any other region; therefore, a facility located anywhere in the Southeast would seem to inequitably impact the Black or African-American population relative to other regions. Assuming facilities must be located in a specific region (as is often the case to provide nearby power sources53) the most equitably sited facilities would impact demographic groups at a level equal to their expected exposure in that region. We calculated the PWE enhancement for a group of units in region (or state) r as the ratio between the relative PW exposure and the relative expected PW exposure:
PWEu,r,y,denhancement=PWEu,y,drelativePWer,y,drelative,expected.
(5)
PWer,y,drelative,expected is expected relative population-weighted exposure for facilities located in an average location in region (or state) r in year y on demographic group d. The lowercase “e” signifies that the expected exposure calculated is not in micrograms per cubic meter, as clarified below.
PWer,y,drelative,expected=PWer,y,dPWer,y,d=total.
(6)
To calculate regional control population-weighted exposure PWer,y,d, we started by assuming that a source could feasibly be located at any overland location, here implemented as the centroids of the 36-km HyADS grid cells. From each location, we calculated exposure as the inverse of distance from that location in all 36-km grid cells across the domain (distance for the containing grid cell is set at 18km), and we used the same population-weighting approach as above:
PWec,y,d=i=1I[1distc,i]u,y,i×(Py,d,iPy,d,total),
(7)
where c denotes grid cell centroids, and distc,i references the distance between centroid c and grid cell i. Expected PWE for region/state r in year y on demographic group d (PWer,y,d) is calculated as the sum of PWec,y,d for each centroid location in region r.
This approach controlled for population group spatial distributions in locations impacted by coal units and population demographic changes over time (PWer,y,drelative,expected is plotted in Figure S8). Changes in wind patterns and changes in emissions at coal facilities are not captured in this expected exposure, meaning the PWEu,r,y,denhancement measures the extent that these two variables increase or decrease exposure disparities. Step changes surrounding census years in PWer,y,drelative,expected reflect discontinuities in intercensal years; such step corrections signify some potential for bias surrounding 2010, but note that the steps are smaller in all cases than the overall trend across 1999–2020.

Results

Coal Power Plant SO2 Emissions, 1999–2020

Annual coal power plant SO2 emissions decreased from 11.8 million tons in 1999 to 0.8 million tons in 2020 (93%; Figure 1). Individual units’ emissions are mostly below 10,000 tons SO2 per year, but some units emitted more than an order of magnitude above that in some years. Monthly emissions have two peaks in most years: one each in the summer and winter. All units emitted <20,000 tons SO2 per year since 2018.
Figure 1. Line plot of monthly total coal electricity generating unit SO2 emissions from 1,237 units in the U.S. EPA Clean Air Markets database.10 Data for this plot are provided in Supplemental Excel Table S1.
SO2 emissions control installations (noted generically as scrubbers here) and coal unit retirements drove the large sectorwide emissions reduction seen in Figure 1. The total number of operational units in the AMPD database peaked in 2004 at 1,256 (Figure S9). The number of units reporting SO2 emissions each year peaked at 1,090 in 2009 (units listed as operational do not always report emissions greater than zero). After 2004, both the number of units in operation with a scrubber and the number of retired units increased. The number of units operating with a scrubber increased from 316 (25% of operating units) in 2006 to 503 (42%) in 2010, and peaked in 2016 at 545 (55%), and declined thereafter as units with and without scrubbers retired. After 2010, new scrubber installations slowed, and emissions changes were driven by coal unit retirements—an average of 70 units retired per year after 2010 in the contiguous United States. In 2020, 438 (80%) of operational units had scrubbers installed.

National Coal PM2.5 Source Impacts and PWE, 1999–2020

Annual average overland coal PM2.5 over the contiguous United States decreased by more than an order of magnitude from 1.17μg/m3 (interquartile range across spatial grid cells=0.101.75) in 1999 to 0.05μg/m3 (0.01–0.07) in 2020 (Figure 2). Average population-weighted coal PM2.5 decreased from 1.96 to 0.06μg/m3 across the same period (Figure 3). Higher population-weighted coal PM2.5 exposure relative to average exposure is attributable to the power plants’ proximity to large population centers in the eastern United States.
Figure 2. Gridded coal PM2.5 from all coal power plants in the United States between 1999 and 2020. The top panel provides boxplots displaying annual median with first and third quartile and outlier coal PM2.5 from coal power plants in the U.S. EPA Clean Air Markets database.10 Select years’ spatial distribution maps are provided as examples in the bottom panel. United States maps are from the USAboundaries R package.42 Data for this plot are provided in Supplemental Excel Tables S2 and S3.
Figure 3. Line plot of annual average population-weighted coal PM2.5 by demographic group averaged across all locations in the contiguous United States. Points denote population-weighted exposure for the total population. U.S. Census racial/ethnic designations are denoted by colors and line type and grouped as follows: White (White population alone), Black (Black or African-American population alone), Native (American Indian and Alaska Native population alone), Asian (Asian population alone), Pacific (Native Hawaiian and Other Pacific Islander population alone and any race), and Hispanic (population of Hispanic origin). Data for this plot are provided in Supplemental Excel Table S4.
In the early 2000s, the eastern United States saw elevated regional exposure, and the western United States saw hot spots near coal facilities. The highest exposures in earlier years are seen along the Ohio River Valley, which houses the highest density of large coal facilities in the country. In recent years, elevated concentrations across the eastern United States have been reduced, but hot spots remain, especially in a band stretching from eastern Texas to the northeast and along the Ohio River Valley.
Modeled coal PM2.5 is correlated (R2>0.5) with measured sulfate PM2.5 at rural IMPROVE monitors in the Northeast, North Central, and South in most years before 2014 (correlation falls off in the South after these years; Figure S4 and Figure S5). The comparison is imperfect, because ambient sulfate can originate from many sources, but the U.S. EPA estimates that electric utilities contributed 70% of total U.S. SO2 emissions from 1999 to 2014, thereafter falling to under 50% in 2020,54 suggesting that much of the ambient sulfate in the United States should be related to electricity generation. We restrict the evaluation to rural IMPROVE sites55 to limit interference from urban sources. Coal PM2.5 follows a similar annual regional trend as observed sulfate PM2.5 at monitoring sites, although the average predicted coal PM2.5 is about 1μg/m3 lower than observed sulfate PM2.5 concentrations at monitor locations (Figure S4). Coal PM2.5 changed at a similar rate in the Northeast, North Central, and South until around 2010, after which coal PM2.5 decreased faster than observed sulfate. Our data suggest little relationship between coal PM2.5 and observed sulfate in the West, where coal PM2.5 exposure is low. Although this evaluation suggests a potential negative bias of coal PM2.5, some negative bias is expected due to the focus on coal emissions specifically, and comparisons with more complex models do not show a similarly systematic negative bias (Figure S6).

National Exposure Contributed or Avoided by Regulatory or Operational Activity

Between 1999 and 2007, at least 85% of the contributed exposure—and 70% of the total (contributed + avoided) exposure—came from units without SO2 scrubbers (such units represented between 60% and 70% of annual input heat capacity in these years). Units with installed scrubbers contributed about 15% of total exposure between 1999 and 2020 (Figure 4). Between 2006 and 2013, both scrubber installations and reduced operations contributed to a large reduction in PWE. Avoided exposure from scrubbers increased from 5% of the total (avoided + emitted) PWE in 2006 to 50% in 2011, where it held steady until 2020. Avoided exposure from reduced operations in units without scrubbers was <10% of total PWE before 2008 and between 10% and 20% from 2009 to 2014, coinciding with decreased operations simultaneous with the economic recession. After 2010, reductions in PWE were driven further by retirements of facilities with and without scrubbers. Although the percentage magnitudes are somewhat sensitive to the high/low operational cutoff (larger avoided exposure is found for higher cutoffs; Figure S7), the overall trends are consistent across applied cutoffs.
Figure 4. Stacked area chart of PWE contributed and avoided relative to each unit’s (n=1,237) baseline PWE (methodological approach details are shown in in Table 1). Baseline is taken as the average exposure contributed before an intervention (either scrubber installation or retirement) in years the unit operated at least at its median capacity. “Avoided: retirement after scrubber” considers only PWE avoided by shutting down; exposure avoided by scrubber installment continues to contribute to “Avoided: scrubber” even after a unit is retired. The top-to-bottom order of categories in the legend is consistent with the order of the plotted data. Data for this plot are provided in Supplemental Excel Table S5. Note: PWE, population-weighted exposure.
Together, the facility retirement/control equipment counts and exposure related to both interventions tell a consistent story: Nationwide, reductions in exposure were primarily driven by scrubber installations from 2006 to 2010, reduced operations from 2008 to 2012, and facility retirements after 2010. Since 2015, avoided exposure has made up over 95% of total (combined + avoided) exposure. Of the small remaining contributed nationwide exposure since 2015, nonscrubbed facilities contributed between 25% and 50%, suggesting similar importance of scrubbed and nonscrubbed facilities for the relatively small remaining PWE.

PWE Inequities Accounting for Spatial Distributions of Sources and Populations

Nationwide population-weighted coal PM2.5 exposure has decreased for all demographic groups since 1999 (Figure 3). Overall, White population exposure tracked the population average because White population represents the largest population group across most of the country (Figure S10). Population average exposure changed little before 2005 and decreased approximately linearly thereafter until 2012. Annual exposure decreased at a slower pace after 2012. Exposure in the Black population was higher than the population average in 1999 and similarly remained relatively constant until 2005. Starting in 2005, Black population exposure decreased fastest among all population groups and by 2016 was at a level nearly identical to the population average. Other demographic groups were exposed at much lower levels than the average population, but these national results obscure regional differences.
The Black population is inequitably exposed in comparison with the population average by facilities located in the Northeast, South, and North Central regions (Figure 5; regions defined in Figure S11). In the South, the Black population was exposed to >0.2μg/m3 more coal PM2.5 most years between 1999 and 2010, and the absolute difference fell thereafter to close to zero by 2017. The relative exposure for the Black population in the South, however, is above 1.3 (30% above the population average) and declines only slightly across the period even as the absolute difference approaches zero. Other demographic groups are exposed at levels similar to (as in the White population) and less than the population average in the Northeast, South, and North Central.
Figure 5. Line plots of absolute and relative annual average population-weighted exposure differences (PWu,y,dabsolute and PWu,y,drelative) attributable to coal facilities located in each of four regions on demographic groups in all locations. U.S. Census demographic variables are denoted by colors and line type and grouped as follows: White (White population alone), Black (Black or African-American population alone), Native (American Indian and Alaska Native population alone), Asian (Asian population alone), Pacific (Native Hawaiian and Other Pacific Islander population alone), and Hispanic (population of Hispanic origin and any race). Data for this plot are provided in Supplemental Excel Table S6.
Multiple populations—notably the Native American, Pacific, and Hispanic populations—were exposed to emissions from facilities located in the West at levels higher than the population average. The relative exposure difference is pronounced in the Native American population, with values over 1.6 (160% of the population average; recall, however, that absolute differences in the West were very low after 2010). Exposure inequities are more pronounced for these populations in the relative exposure difference than the absolute difference because of the lower overall exposure in the western United States (there were many fewer coal plants in the West than in other regions during the study period).
Accounting for regional population demographic distributions and changes over time, many of the observed exposure inequities shown in Figure 5 in the Black population in the Northeast and South are reduced to below the population average (Figure 6). Although the Black population experienced slightly enhanced exposure before 2015, thereafter the population has been exposed at levels less than the population average. An opposite pattern is observed in the Black population exposure to emissions from facilities in the North Central region; accounting for the regional population distribution enhances the Black population’s average exposure inequity across the period from 7% over the average to 13% over the average (absolute differences in the North Central were very low after 2010).
Figure 6. Line plots of PWE enhancement from facilities in each region (top) and state (bottom) that account for regional demographics and change over time (PWEr,y,denhancement; denoted as Relative PWE enhancement and Rel. PWE enhmnt in the figure). U.S. Census demographic variables are denoted by colors and line type in the top plots and grouped as follows: White (White population alone), Black (Black or African-American population alone), Native (American Indian and Alaska Native population alone), Asian (Asian population alone), Pacific (Native Hawaiian and Other Pacific Islander population alone), and Hispanic (population of Hispanic origin and any race). Only states with coal power plants are shown in the lower plots (California and Idaho are omitted); blue shading and hashing denote PWEr,y,denhancement below one, and red shading and dots denote PWEr,y,denhancement above one. United States maps are from the USAboundaries R package.42 Data for this plot are provided in Supplemental Excel Tables S7 and S8. Note: PWE, population-weighted exposure.
In the West, exposure inequity in the Native American population persists (but is slightly mitigated) by controlling for regional population distribution. Other populations that see higher exposures relative to the population average without controlling for the regional population distribution (Pacific and Hispanic populations) are exposed to smaller levels of coal PM2.5 than would be expected for a facility in an arbitrary location throughout the region.
Regional averages of exposure enhancements obscure underlying heterogeneity: Facilities in some states contributed more exposure inequities than others (Figure 6, bottom). Facilities in small groups of states stand out as contributing inequitably to exposure in each of the demographic groups: Washington State (Asian population); Nevada and Louisiana (Hispanic population); Oregon, Nevada, Arizona, and New Mexico (Native population); and Washington State, Nevada, and Utah (Pacific population). Facilities in a small number of states stand out (30 in 1999, 25 in 2020) contributed enhanced exposure to the Black population.

Uncertainty and Limitations

The HyADS model is a reduced complexity model in that it uses the assumptions detailed above to approximate atmospheric transport and deposition. Further, the chemical conversion of SO2 to PM2.5 is approximated in each year using a statistical model trained on CMAQ-DDM output from a single year (2005). This approach, necessitated by limited availability of Hybrid CMAQ-DDM output, makes it difficult to quantify the uncertainty associated with relationships that changed over the time scale of this study. For example, although model (1) allows for varied relationships between meteorological influence on coal PM2.5, it does not account for long-term changes in the nitrate and ammonium emissions, which exert influence over PM2.5 mass formed by sulfur emissions.56,57 Our evaluations have shown, however, that although uncertainties in individual-unit coal PM2.5 are nonnegligible, high correlations in source impacts from individual units with complementary estimates from more sophisticated models suggest that the model is able to provide useful information for the applications above, particularly in the regional and national relative exposure long-term trends. Uncertainty in the exposure modeling with HyADS is not propagated into reported air quality or disparities metrics presented here, which motivated our approach to focus on temporal trends and relative exposure disparities that would be robust to modeling uncertainties that are not systematically related to time or the spatial distribution of different populations. There is the potential for future studies to explore reduced complexity approaches that more fully capture the atmospheric dynamics that are not accounted for in our approach.
The study is limited by its geographic resolution at the county level, which follows from the HyADS grid resolution (36km). As in any model based on a grid, there is potential for within-grid variation to introduce exposure misclassification and the possibility that this coarse resolution averages out intragrid exposure and demographic variability as addressed by Spiller et al.,58 but such potential is mitigated by annual averages used here and the regional nature of SO2-related coal PM2.5 impacts. Using PM2.5-emitting facility locations and a 2.5-mi (4-km) buffer, for example, Mikati et al. found higher relative disparities in year 2011 emissions burden than the values we report, albeit for a different group of facilities.59
Calculations of PWE enhancement begin with the assumption that coal power plants could be located at any overland location in a region, which is not actually the case. Locations are restricted by water availability, access to fuel sources, and feasibility to deliver electricity demand.53 Although the method uses inverse-distance weighting and therefore does not control for meteorological variability, such a control could be added in future work. One approach to undertaking this would be to run the HyADS model from each grid centroid location and using the output to calculate PWec,y,d. Given the high agreement between inverse-distance weighted exposure and HyADS output found in Henneman et al., however, we concluded the inverse distance approach provided sufficient exposure control.37
The analysis focuses on impacts from SO2 emissions reported at coal power plants. Coal combustion leads to emissions of other pollution species, including NOx, primary PM2.5, and mercury.35 Although emissions of some of these other pollution species may be captured in coal PM2.5 owing to their correlation with SO2 emissions and the reduced-complexity nature of the model, we have not quantified the extent that they have contributed to exposure inequities insofar as they are not coincidentally captured by HyADS and its conversion to coal PM2.5. Additionally, it is possible that locations of now-retired coal power plants were repurposed to house natural gas facilities; inequities contributed by such situations are not captured in this analysis.

Implications

Regulatory policies, technological improvements, and economics had wide influence on decreasing exposure to PM2.5 from coal power plant SO2 emissions since 1999. Three unit-level interventions taken at many individual coal power plants throughout the country resulted in decreasing exposure and exposure inequities— SO2 scrubber installations, unit retirements, and decreased operations. In recent years, more nationwide PWE is attributable to units with scrubbers installed than to uncontrolled emissions, and exposure inequities persist throughout the country. This finding suggests that reducing emissions from the 505 units still in operation in 2020, which contribute similar magnitudes of exposure from scrubbed and nonscrubbed units, offers potential benefits for overall exposure and exposure inequities reduction.
Although air pollution from coal power plants is generally regional in nature, this work reveals exposure inequities attributable to regional and state facility siting (Figure 6). These inequities persist even as emissions have been reduced from the entire source sector, for example in Black populations impacted by facilities in the North Central region and in individual states across the center of the country and in Native American populations impacted by facilities in the West. Exposure in both groups remained inequitably high in 2020, although average exposure after 2015 has been very low in the West.
We find that accounting for population demographics near sources in exposure-inequity calculations reduces estimated inequities relative to calculations that do not account for underlying population demographics in many cases. Results from the analyses may be of interest to policymakers at varying levels of government; for example, the U.S. EPA may be interested in the national inequities, whereas local decision makers may use results that account for regional population makeup.
In the United States, coal is being rapidly phased out as an energy source. This analysis, therefore, serves primarily as a detailed post hoc assessment of coal’s impacts across 1999–2020 and a period of dramatic improvements due to air quality regulations and the economics of coal price relative to other electricity sources. We find that coal electricity-generation SO2 emissions contributions to overall PM2.5 are diminishing to the point of having minimal effect on most locations’ ability to meet national ambient air quality standards.

Acknowledgments

This work was supported by research funding from the National Institutes of Health (NIH) NIHR01ES026217 and the U.S. EPA 835872. S.C.A. acknowledges support from NASA grant no. 80NSSC21K0511. Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the U.S. EPA. Further, the U.S. EPA does not endorse the purchase of any commercial products or services mentioned in the publication.
Code and data needed to reproduce the figures in this manuscript are available on Github (https://github.com/lhenneman/coal_exposure_longterm). Coal PM2.5 source impacts and facility information are available for download from Open Science Framework (https://osf.io/8gdau), and code for importing gridded coal PM2.5 into R is provided on Github (https://github.com/lhenneman/coal_unit_PM25).
The disperseR package enables parallelization of many HYSPLIT runs and provides tools for manipulating and summarizing the output data relevant for HyADS. All results presented here were run using the development version of the disperseR package maintained on Github (https://github.com/lhenneman/disperseR).

Article Notes

The authors declare they have nothing to disclose.

Supplementary Material

File (ehp11605.original.acco.pdf)
File (ehp11605.smcontents.508.pdf)
File (ehp11605.s001.acco.pdf)
File (ehp11605.s002.codeanddata.acco.zip)

References

1.
Squizzato S, Masiol M, Rich DQ, Hopke PK. 2018. PM2.5 and gaseous pollutants in New York state during 2005–2016: spatial variability, temporal trends, and economic influences. Atmos Environ 183:209–224, https://doi.org/10.1016/j.atmosenv.2018.03.045.
2.
Ivey CE, Holmes HA, Hu YT, Mulholland JA, Russell AG. 2015. Development of PM2.5 source impact spatial fields using a hybrid source apportionment air quality model. Geosci Model Dev 8(7):2153–2165, https://doi.org/10.5194/gmd-8-2153-2015.
3.
Blanchard CL, Tanenbaum S, Hidy GM. 2012. Source contributions to atmospheric gases and particulate matter in the southeastern United States. Environ Sci Technol 46(10):5479–5488. https://pubmed.ncbi.nlm.nih.gov/22475316/, https://doi.org/10.1021/es203568t.
4.
National Research Council. 2004. Air Quality Management in the United States. Washington, DC: National Academies Press. https://doi.org/10.17226/10728.
5.
U.S. EPA (U.S. Environmental Protection Agency). 2005. Regulatory Impact Analysis for the Final Clean Air Interstate Rule. EPA-452/R-05-002. Washington, DC: Office of Air and Radiation.
6.
United States Environmental Protection Agency Office of Air and Radiation Office of Air Quality Planning and Standards Office of Atmospheric Programs (6204J) 1200 Pennsylvania Ave, NW Washington, DC 20460. https://www.epa.gov/sites/default/files/2015-08/documents/ozonenbp-2004.pdf.
7.
U.S. EPA. 2015. Cross-State Air Pollution Rule Update for the 2008 Ozone NAAQS, Vol. 80, 74504–74650. https://www.epa.gov/sites/default/files/2017-09/documents/2014_full_report.pdf [accessed 27 February 2023].
9.
U.S. Energy Information Administration. Form EIA-860 Detailed Data with Previous Form Data (EIA-860A/860B). https://www.eia.gov/electricity/data/eia860/ [accessed 17 June 2021].
10.
U.S. EPA. Clean Air Markets Program Data. https://campd.epa.gov/ [accessed 22 July 2021].
11.
de Gouw JA, Parrish DD, Frost GJ, Trainer M. 2014. Reduced emissions of CO2, NOx and SO2 from U.S. power plants due to the switch from coal to natural gas with combined cycle technology. Earths Future 2:75–82, https://doi.org/10.1002/2014EF000196.
12.
Harrington W, Morgenstern R, Shih J-S, Bell ML. 2012. Did the clean air act amendments of 1990 really improve air quality? Air Qual Atmosphere Health 5(4):353–367, https://doi.org/10.1007/s11869-012-0176-5.
13.
Lin S, Jones R, Pantea C, Özkaynak H, Rao STT, Hwang S-A, et al. 2013. Impact of NOx emissions reduction policy on hospitalizations for respiratory disease in New York state. J Expo Sci Environ Epidemiol 23(1):73–80. https://pubmed.ncbi.nlm.nih.gov/22829050/, https://doi.org/10.1038/jes.2012.69.
14.
Godowitch JM, Gilliland AB, Draxler RR, Rao ST. 2008. Modeling assessment of point source NOx emission reductions on ozone air quality in the Eastern United States. Atmos Environ 42(1):87–100, https://doi.org/10.1016/j.atmosenv.2007.09.032.
15.
Russell MC, Belle JH, Liu Y. 2015. The impact of three recent coal-fired power plant closings on Pittsburgh air quality: a natural experiment. J Air Waste Manag Assoc 67(1):3–16, https://doi.org/10.1080/10962247.2016.1170738.
16.
Casey JA, Su JG, Henneman LRF, Zigler C, Neophytou AM, Catalano R, et al. 2020. Improved asthma outcomes observed in the vicinity of coal power plant retirement, retrofit and conversion to natural gas. Nat Energy 5:398–408, https://doi.org/10.1038/s41560-020-0600-2.
17.
Abrams JY, Klein M, Henneman LR, Sarnat SE, Chang HH, Strickland MJ, et al. 2019. Impact of air pollution control policies on cardiorespiratory emergency department visits, Atlanta, GA, 1999–2013. Environ Int 126:627–634. https://pubmed.ncbi.nlm.nih.gov/30856450/, https://doi.org/10.1016/j.envint.2019.01.052.
18.
Henneman LR, Choirat C, Zigler CM. 2019. Accountability assessment of health improvements in the United States associated with reduced coal emissions between 2005 and 2012. Epidemiology 30(4):477–485. https://pubmed.ncbi.nlm.nih.gov/31162280/, https://doi.org/10.1097/EDE.0000000000001024.
19.
Levy JI, Wilson AM, Zwack LM. 2007. Quantifying the efficiency and equity implications of power plant air pollution control strategies in the United States. Environ Health Perspect 115(5):743–750. https://pubmed.ncbi.nlm.nih.gov/17520062/, https://doi.org/10.1289/ehp.9712.
20.
Thind MPS, Tessum CW, Azevedo IL, Marshall JD. 2019. Fine particulate air pollution from electricity generation in the US: health impacts by race, income, and geography. Environ Sci Technol 53(23):14010–14019. https://pubmed.ncbi.nlm.nih.gov/31746196/, https://doi.org/10.1021/acs.est.9b02527.
21.
The President of the United States. 1994. Executive Order 12898. Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations. Fed Reg 59(32):7629–7633 (11 February 1994). https://www.govinfo.gov/content/pkg/FR-1994-02-16/html/94-3685.htm [accessed 4 January 2022].
22.
The President of the United States. 2021. Executive Order 14008. Tackling the Climate Crisis at Home and Abroad. Fed Reg 26:7619–7633 (27 January 2021). https://www.federalregister.gov/documents/2021/02/01/2021-02177/tackling-the-climate-crisis-at-home-and-abroad [accessed 4 January 2022].
23.
Harper S, Ruder E, Roman HA, Geggel A, Nweke O, Payne-Sturges D, et al. 2013. Using inequality measures to incorporate environmental justice into regulatory analyses. Int J Environ Res Public Health 10(9):4039–4059. https://pubmed.ncbi.nlm.nih.gov/23999551/, https://doi.org/10.3390/ijerph10094039.
24.
Buonocore JJ, Dong X, Spengler JD, Fu JS, Levy JI. 2014. Using the Community Multiscale Air Quality (CMAQ) model to estimate public health impacts of PM2.5 from individual power plants. Environ Int 68:200–208. https://pubmed.ncbi.nlm.nih.gov/24769126/, https://doi.org/10.1016/j.envint.2014.03.031.
25.
American Lung Association and American Public Health Association, Petitioners v. Environmental Protection Agency and Andrew Wheeler, Administrator, Respondents. 2021. U. S. Court Appeals Dist Columbia Circuit No. 19-1140 (D.C. Cir. 19 January 2021). https://statepowerproject.files.wordpress.com/2021/01/american-lung-assn-v.-epa-dc-cir.-no.-19-1140-per-curiam-decision.pdf [accessed 25 March 2022].
26.
U.S. Environmental Protection Agency Environmental Justice Screening Report Carbon Pollution Emission Guidelines for Existing Stationary Sources: Electric Utility Generating Units Docket: EPA–HQ–OAR–2013–0602 Office of Air and Radiation, Office of Air Quality Planning and Standards Research Triangle Park, NC 27711. https://19january2017snapshot.epa.gov/sites/production/files/2016-04/documents/ejscreencpp.pdf.
27.
U.S. EPA. Power Plants and Neighboring Communities. 2021. https://www.epa.gov/airmarkets/power-plants-and-neighboring-communities [accessed 29 March 2022].
28.
Tessum CW, Hill JD, Marshall JD. 2017. InMAP: a model for air pollution interventions. PLoS One 12(4):1–26. https://pubmed.ncbi.nlm.nih.gov/28423049/, https://doi.org/10.1371/journal.pone.0176131.
29.
Tessum CW, Apte JS, Goodkind AL, Muller NZ, Mullins KA, Paolella DA, et al. 2019. Inequity in consumption of goods and services adds to racial–ethnic disparities in air pollution exposure. Proc Natl Acad Sci 116(13):201818859. https://pubmed.ncbi.nlm.nih.gov/30858319/, https://doi.org/10.1073/pnas.1818859116.
30.
Goodkind AL, Tessum CW, Coggins JS, Hill JD, Marshall JD. Fine-scale damage estimates of particulate matter air pollution reveal opportunities for location-specific mitigation of emissions. Proc Natl Acad Sci USA 116(18):8775–8780. https://doi.org/10.1073/pnas.1816102116.
31.
Muller NZ, Mendelsohn R. 2007. Measuring the damages of air pollution in the United States. J Environ Econ Manag 54(1):1–14, https://doi.org/10.1016/j.jeem.2006.12.002.
32.
Heo J, Adams PJ, Gao HO. 2016. Reduced-form modeling of public health impacts of inorganic PM2.5 and precursor emissions. Atmos Environ 137:80–89, https://doi.org/10.1016/j.atmosenv.2016.04.026.
33.
Heo J, Adams PJ, Gao HO. 2017. Public health costs accounting of inorganic PM2.5 pollution in metropolitan areas of the United States using a risk-based source-receptor model. Environ Int 106:119–126. https://pubmed.ncbi.nlm.nih.gov/28633084/, https://doi.org/10.1016/j.envint.2017.06.006.
34.
Baker KR, Amend M, Penn S, Bankert J, Simon H, Chan E, et al. 2020. A database for evaluating the InMAP, APEEP, and EASIUR reduced complexity air-quality modeling tools. Data Brief 28:104886. https://pubmed.ncbi.nlm.nih.gov/31872009/, https://doi.org/10.1016/j.dib.2019.104886.
35.
Dedoussi IC, Barrett SRH. 2014. Air pollution and early deaths in the United States. Part II: attribution of PM2.5 exposure to emissions species, time, location and sector. Atmos Environ 99:610–617, https://doi.org/10.1016/j.atmosenv.2014.10.033.
36.
Henneman LR, Choirat C, Ivey CE, Cummiskey K, Zigler CM. 2019. Characterizing population exposure to coal emissions sources in the United States using the HyADS model. Atmos Environ 203(203):271–280, https://doi.org/10.1016/j.atmosenv.2019.01.043.
37.
Henneman LR, Dedoussi IC, Casey JA, Choirat C, Barrett SRH, Zigler CM. 2020. Comparisons of simple and complex methods for quantifying exposure to individual point source air pollution emissions. J Expo Sci Environ Epidemiol 31(4):654–663. https://pubmed.ncbi.nlm.nih.gov/32203059/, https://doi.org/10.1038/s41370-020-0219-1.
38.
Henneman LR, Mickley LJ, Zigler CM. 2019. Air pollution accountability of energy transitions: the relative importance of point source emissions and wind fields in exposure changes. Environ Res Lett 14(11):115003. https://pubmed.ncbi.nlm.nih.gov/33408754/, https://doi.org/10.1088/1748-9326/ab4861.
39.
Stein AF, Draxler RR, Rolph GD, Stunder BJB, Cohen MD, Ngan F. 2015. NOAA’s HYSPLIT atmospheric transport and dispersion modeling system. Bull Am Meteorol Soc 96(12):2059–2077, https://doi.org/10.1175/BAMS-D-14-00110.1.
40.
Draxler RR, Hess GD. 1998. An overview of the HYSPLIT_4 modelling system for trajectories, dispersion, and deposition. Aust Meteorol Mag 47:295–308.
41.
Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, et al. 1996. The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77(3):437–471, https://doi.org/10.1175/1520-0477(1996)077%3C0437:TNYRP%3E2.0.CO;2.
42.
Ivey C, Holmes H, Shi G, Balachandran S, Hu Y, Russell AG. 2017. Development of PM2.5 source profiles using a hybrid chemical transport-receptor modeling approach. Environ Sci Technol 51(23):13788–13796. https://pubmed.ncbi.nlm.nih.gov/29110467/, https://doi.org/10.1021/acs.est.7b03781.
43.
Dedoussi IC, Eastham SD, Monier E, Barrett SRH. 2020. Premature mortality related to United States cross-state air pollution. Nature 578(7794):261–265. https://pubmed.ncbi.nlm.nih.gov/32051602/, https://doi.org/10.1038/s41586-020-1983-8.
44.
Gilmore EA, Heo J, Muller NZ, Tessum CW, Hill JD, Marshall JD, et al. 2019. An inter-comparison of the social costs of air quality from reduced-complexity models. Environ Res Lett 14(7):074016, https://doi.org/10.1088/1748-9326/ab1ab5.
45.
U.S. Census Bereau. County Population by Characteristics: 2010–2019. https://www.census.gov/data/datasets/time-series/demo/popest/2010s-counties-detail.html [accessed 10 June 2021].
46.
Emami F, Masiol M, Hopke PK. 2018. Air pollution at Rochester, NY: long-term trends and multivariate analysis of upwind SO2 source impacts. Sci Total Environ 612:1506–1515. https://pubmed.ncbi.nlm.nih.gov/28915545/, https://doi.org/10.1016/j.scitotenv.2017.09.026.
47.
Hand JL, Schichtel BA, Malm WC, Pitchford ML. 2012. Particulate sulfate ion concentration and SO2 emission trends in the United States from the early 1990s through 2010. Atmospheric Chem Phys 12(21):10353–10365, https://doi.org/10.5194/acp-12-10353-2012.
48.
Levy JI, Spengler JD, Hlinka D, Sullivan D, Moon D. 2002. Using CALPUFF to evaluate the impacts of power plant emissions in Illinois: model sensitivity and implications. Atmos Environ 36(6):1063–1075, https://doi.org/10.1016/S1352-2310(01)00493-9.
49.
Mullen LA, Bratt J. 2018. USAboundaries: historical and contemporary boundaries of the United States of America. J Open Source Softw 3(23):314, https://doi.org/10.21105/joss.00314.
50.
Daouda M, Henneman LR, Kioumourtzoglou M-A, Gemmill A, Zigler C, Casey JA. 2021. Association between county-level coal-fired power plant pollution and racial disparities in preterm births from 2000 to 2018. Environ Res Lett 16(3):034055. https://pubmed.ncbi.nlm.nih.gov/34531925/, https://doi.org/10.1088/1748-9326/abe4f7.
51.
Vann A. 2010. The Federal Government’s Role in Electric Transmission Facility Siting. Washington, DC: Library of Congress, Congressional Research Service, 17.
52.
U.S. EPA. 2015. Regulatory Impact Analysis for the Clean Power Plan Final Rule. Research Triangle Park, NC: U.S. Environmental Protection Agency Office of Air and Radiation, Office of Air Quality Planning and Standards, 343.
53.
Wang S, Fisher EB, Feng L, Zhong X, Ellis JH, Hobbs BF. 2021. Linking energy sector and air quality models through downscaling: long-run siting of electricity generators to account for spatial variability and technological innovation. Sci Total Environ 772:145504. https://pubmed.ncbi.nlm.nih.gov/33581514/, https://doi.org/10.1016/j.scitotenv.2021.145504.
54.
U.S. EPA. Air Pollutant Emissions Trends Data. https://www.epa.gov/air-emissions-inventories/air-pollutant-emissions-trends-data [accessed 23 August 2022].
55.
Solomon PA, Crumpler D, Flanagan JB, Jayanty RKM, Rickman EE, McDade CE. 2014. U.S. National PM2.5 chemical speciation monitoring networks—CSN and IMPROVE: description of networks. J Air Waste Manag Assoc 64(12):1410–1438. https://pubmed.ncbi.nlm.nih.gov/25562937/, https://doi.org/10.1080/10962247.2014.956904.
56.
Vasilakos P, Russell A, Weber R, Nenes A. 2018. Understanding nitrate formation in a world with less sulfate. Atmos Chem Phys Discuss 2018:1–27, https://doi.org/10.5194/acp-2018-406.
57.
Holt J, Selin NE, Solomon S. 2015. Changes in inorganic fine particulate matter sensitivities to precursors due to large-scale US emissions reductions. Environ Sci Technol 49(8):4834–4841. https://pubmed.ncbi.nlm.nih.gov/25816113/, https://doi.org/10.1021/acs.est.5b00008.
58.
Spiller E, Proville J, Roy A, Muller NZ. 2021. Mortality risk from PM2.5: a comparison of modeling approaches to identify disparities across racial/ethnic groups in policy outcomes. Environ Health Perspect 129(12):127004. https://pubmed.ncbi.nlm.nih.gov/34878311/, https://doi.org/10.1289/EHP9001.
59.
Mikati I, Benson AF, Luben TJ, Sacks JD, Richmond-Bryant J. 2018. Disparities in distribution of particulate matter emission sources by race and poverty status. Am J Public Health 108(4):480–485. https://pubmed.ncbi.nlm.nih.gov/29470121/, https://doi.org/10.2105/AJPH.2017.304297.

Information & Authors

Information

Published In

Environmental Health Perspectives
Volume 131Issue 3March 2023
PubMed: 36884005

History

Received: 23 May 2022
Revision received: 23 January 2023
Accepted: 27 January 2023
Published online: 8 March 2023
Corrected: 15 May 2023

Authors

Affiliations

Department of Civil, Environmental, and Infrastructure Engineering; George Mason University, Fairfax, Virginia, USA
Munshi Md Rasel
Department of Civil, Environmental, and Infrastructure Engineering; George Mason University, Fairfax, Virginia, USA
Christine Choirat
Swiss Data Science Center, ETH Zürich and EPFL, Lausanne, Switzerland
Susan C. Anenberg
Department of Environmental and Occupational Health, George Washington University, Washington, District of Columbia, USA
Corwin Zigler
Department of Statistics and Data Sciences, University of Texas, Austin, USA

Notes

Address correspondence to Lucas R.F. Henneman, 4400 University Dr., MS-6C1, Fairfax, VA 22030 USA. Email: [email protected]

Metrics & Citations

Metrics

About Article Metrics


Citations

Download citation

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click DOWNLOAD.

Cited by

  • Impacts of Fuel Prices and Regulations on Electricity Generation Emissions and Urban Air Quality, ACS ES&T Air, 10.1021/acsestair.3c00034, (2024).
  • Invited Perspective: Leveraging Air Quality Research Advancements to Assess and Address Exposure Inequities, Environmental Health Perspectives, 10.1289/EHP12503, 131, 3, (2023).
  • Air quality policy should quantify effects on disparities, Science, 10.1126/science.adg9931, 381, 6655, (272-274), (2023).
  • Mortality risk from United States coal electricity generation, Science, 10.1126/science.adf4915, 382, 6673, (941-946), (2023).
  • Greenhouse Suppliers 100: A ranking of corporate producers of greenhouse gas precursors in the USA, The Japanese Political Economy, 10.1080/2329194X.2023.2252483, 49, 2-3, (212-230), (2023).
  • COVID-19 and Beyond: COVID-19 Interventions and Power Plant Emissions in the United States, ACS ES&T Engineering, 10.1021/acsestengg.2c00353, 3, 7, (923-931), (2023).
  • Cost-benefits analysis of ultra-low emissions standard on air quality and health impact in thermal power plants in China, Journal of Environmental Management, 10.1016/j.jenvman.2023.118731, 345, (118731), (2023).

View Options

View options

PDF

View PDF

Get Access

Restore your content access

Enter your email address to restore your content access:

Note: This functionality works only for purchases done as a guest. If you already have an account, log in to access the content to which you are entitled.

Media

Figures

Other

Tables

Share

Share

Copy the content Link

Share on social media