AirToxScreen Assessment Methods
The AirToxScreen analysis includes four main steps:
- Compile a National Emissions Inventory (NEI)
- Estimate ambient concentrations of air toxics across the United States
- Estimate population exposures
- Characterize potential health risks from inhalation
Compile a National Emissions Inventory (NEI)
To get the emissionsThe release of pollutants into the ambient air. data for this assessment, EPA starts with the latest National Emissions Inventory (NEI). If the AirToxScreen data-year is more recent than the latest NEI, we update the inventory to reflect that year (see our Frequent Questions for more details). This inventory reflects reviews and quality assurance checks made by state and local agencies and other reviewers. We also conducted certain processing steps (for example, placing individual chemicals into groups and converting the emissions into the formats needed for the air quality models).
AirToxScreen includes these emission source types:
- major stationary sources, also referred to as point sources (including large waste incinerators and factories);
- area and other smaller stationary sources, also referred to as nonpoint sources (including dry cleaners and small manufacturers);
- onroad and nonroad mobile sources (including cars, trucks and boats);
- fires (including wildfires, prescribed fires and agricultural burning); and
- biogenic sources (emissions from trees, plants and soil microbes).
We also include estimates for:
- background concentrations – air toxics emissions from distant sources, emissions from prior years that persist in the environment, and natural source emissions other than those modeled as biogenic sources; and
- secondarily formed pollutants – chemicals that form in the air through chemical reactions. Please note that biogenic sources also directly emit some pollutants that also form secondarily, including formaldehyde, acetaldehyde and acrolein.
For more information on emission inventories, see the Air Emissions Inventory web page. You can find more detailed information on development of AirToxScreen’s emissions data in Section 2 of the AirToxScreen Technical Support Document (TSD), Emissions. Section 2 also includes more detailed information on how emissions and background concentrations were processed for modeling.
Estimate ambient concentrations of air toxics across the United States
Using two models, EPA estimated ambient concentrationsA way to describe how much of a pollutant is in the outdoor air. Concentration is usually shown as an amount, or mass, of pollutant per certain volume of air. In AirToxScreen, most concentrations are in micrograms (µg) of air pollutant per cubic meter (m3) of air (a “box” of air one meter on each side). of air toxics across the United States. We used the chemical transport model CMAQ and the dispersion model AERMOD, taking advantage of the strengths of each model.
CMAQ is used to conduct urban- to regional-scale simulations of multiple air-quality issues. CMAQ can model how air pollution disperses and chemically transforms as it travels through the air. CMAQ includes several strengths that make it well suited for AirToxScreen. It conserves mass (that is, if some pollution leaves an area, it is accounted for in the new area); allows long-range pollutant transport; and estimates concentrations of secondarily-formed pollutants such as formaldehyde. We modeled over 50 air toxics plus diesel particulate matter (diesel PM) in CMAQ using a 12-km by 12-km grid resolution. Fires and biogenic emissions were only modeled in CMAQ. Secondary formation of pollutants was also only estimated in CMAQ. CMAQ provides an ambient air concentration for each grid.
AERMOD, EPA’s preferred dispersion model for regulatory purposes, can estimate ambient concentrations at many closely spaced points (called receptors). We based AERMOD receptor locations on the centroids of populated census blocksThe smallest geographic areas that the U.S. Census Bureau uses. Blocks are bounded by visible or virtual features such as streets, streams, and city or town boundaries. Census blocks are usually small in area and population (typically around 50 residents)., monitoring site locations, and five evenly distributed points within each 12-km CMAQ grid cell. We then calculate census tractLand areas defined by the U.S. Census Bureau. Tracts usually contain from 1,200 to 8,000 people, with most having close to 4,000 people. Census tracts are usually smaller than 2 square miles in cities, but are much larger in rural areas.-level concentrations by averaging all the calculated block-level values. We modeled all AirToxScreen pollutants using AERMOD. We included all emission source types except fires, biogenic and secondary sources, and background estimates. AERMOD provides an ambient air concentration for each receptor.
For the air toxics modeled in both CMAQ and AERMOD, we used a special hybrid modeling method. This method combines the fine receptor spacing allowed by AERMOD with the full treatment of chemistry and transport that CMAQ provides. Using this hybrid approach, we can “anchor” the ambient AERMOD receptor concentrations to the ambient CMAQ grid concentrations. This allows mass to be conserved and avoids double counting.
In AirToxScreen, we also add background concentrations, which represent the contributions to ambient concentrations of air toxics resulting from:
- natural sources not already accounted for under biogenic emissions;
- emissions of persistent chemicals that occurred in previous years; and
- long-range transport from distant sources.
These background concentrations represent “lingering” pollutant levels found in a place even with no nearby emissions of those pollutants during the year of the AirToxScreen assessment. We included background directly in all CMAQ modeling. For AERMOD, we calculated a background concentration and added it to the modeled concentration. More information about background concentrations can be found in Section 3 and Appendix D of the TSD.
Using the hybrid approach (with background concentrations added where appropriate), we estimated ambient concentrations at every populated and nonpopulated census block receptor. For each census tract, we then used the area of each block in the tract to calculate an area-weighted average concentration across the census tract. (This is the ambient concentration reported for each tract in AirToxScreen’s results.) Then for each tract, we excluded nonpopulated census blocks and weighted the remaining blocks’ concentrations by their population. This gives us a population-weighted average concentration, which we use to calculate tract-level population exposures and risks (see next AirToxScreen steps).
More information about these models and the process for generating ambient concentration data may be found in Section 3 of the TSD.
Estimate population exposures
In AirToxScreen, we don’t directly measure exposure. Neither AERMOD nor CMAQ directly addresses certain important exposure variables (including human activity patterns, etc.). Instead, we use modeled concentrations as surrogates, or stand-ins, for exposure. This step is important, because the average concentration of a pollutant that people actually breathe may be much higher or lower than the concentration at some fixed modeled location.
To estimate exposure, EPA uses the census tract concentrations calculated in the previous step as inputs to the screening-level inhalation exposure model HAPEM7. HAPEM7 uses census data, human-activity-pattern data, ambient air quality levels, climate data, and indoor/outdoor concentration relationships to calculate an expected range of apparent inhalation exposure concentrations for groups of people. It then provides a best estimate of exposure for a hypothetical typical person for a given census tract.
EPA conducted HAPEM simulation modeling for the 1999 National Air Toxics Assessment (NATA) analysis, the predecessor of AirToxScreen, for each combination of pollutant, census tract and source group. Because running HAPEM requires a lot of time and resources, for the 2002 and 2005 NATAs, we relied on the exposure factors derived for the 1999 analysis and applied them to the modeled ambient concentrations. For the 2011 and 2014 NATAs, and for the 2017, 2018, and 2019 AirToxScreen assessments, EPA ran HAPEM for several surrogate pollutants (coke oven emissions, diesel PM, benzene, 1,3-butadiene, generic PAHs, chromium VI, and nickel). We then applied these exposure factors to other similar pollutants (for example, benzene was the surrogate for all other gas-phase pollutants).
Characterize potential health risks from inhalation
In the last step of AirToxScreen, EPA calculates cancer risksThe probability of contracting cancer over the course of a lifetime, assuming continuous exposure (assumed in AirToxScreen to be 70 years). and noncancer hazard indexesA way to express the combined impact of several toxics on human health. A hazard index (HI) of 1 or lower means air toxics are unlikely to cause adverse noncancer health effects over a lifetime of exposure. by pollutant. To do this, we combine the census tract-level exposure concentration estimates (generated in the previous step) with available unit risk estimatesUpper-bound estimate of a person’s chance of contracting cancer over a lifetime of exposure to a particular concentration: one microgram of the pollutant per cubic meter of air. Risks from exposures to concentrations other than one microgram per cubic meter are usually calculated by multiplying the actual concentration to which someone is exposed by the URE. and inhalation reference concentrationsAn estimate of a continuous inhalation exposure unlikely to cause adverse health effects during a person’s lifetime. This estimate includes sensitive groups such as children, asthmatics and the elderly. for each pollutant.
For this AirToxScreen release, we calculated cancer risks or hazard indexes for about 140 pollutants. You can find more details about unit risk estimates, inhalation reference concentrations, and risk characterization in EPA’s Dose-Response Assessment web page; in Section 5 of the TSD, Characterizing the Effects of Air Toxics; and in Section 6 of the TSD, Characterizing Risks and Hazards in AirToxScreen.
The 2019 AirToxScreen results section presents ambient and exposure concentrations, cancer risks, and noncancer hazard indexes at the tract, county and state levels. For noncancer effects, we estimated the effect of air toxics on the body’s various organ systems. Past assessments have shown that the effect on the lungs and other parts of the respiratory system (known as the respiratory endpoint) tends to drive noncancer risks. However, in the 2019 AirToxScreen results, we also included results for 13 other noncancer endpoints.
When using these results, it is important to consider that AirToxScreen is a screening-level assessment and only suitable to answer certain questions. Avoid using AirToxScreen results to compare or rank states or regions of the country. Air toxics emissions data reported to EPA vary in level of detail from state to state, making comparisons difficult. Certainly, air toxics concentrations, exposures and risks do differ across the United States. It is important to have an idea of how these factors vary throughout the country. But comparisons within a state are more appropriate than comparing one state or region to another.
EPA seeks to protect health with reasonable confidence. But it's important to keep in mind AirToxScreen's variability and uncertainty, which are part of any risk assessment. Scientific estimates of air concentrations, exposures and risks always involve assumptions. These assumptions simplify things to make an assessment possible, but they also introduce uncertainties. For example, the smaller the area, the more uncertain the results. A more complete discussion of both variability and uncertainty is found in Section 7 of the TSD, Variability and Uncertainty Associated with AirToxScreen.