At times, a facility can be in compliance with permits, approvals, and regulations, and yet still be subject to intense scrutiny over its reported environmental impact. A case in point is the attention some facilities have received since U.S. EPA's 2018 release of the latest National Air Toxics Assessment (NATA), which in some areas of the U.S. set off a flurry of activity among sources, regulatory agencies, and the public.

NATA Figure 1 and 2

What is the National Air Toxics Assessment?

The NATA is EPA's ongoing, periodic review of outdoor air quality reflecting the impact of emissions of air toxics across the U.S and down to the census tract level. It is intended to provide information regarding the long‐term risks to human health (both cancer and non‐cancer health effects) based on exposure to air toxics over many years. This information is used to educate the public and to provide data to guide state and federal rulemaking processes. NATA results, including very large database files, are all publicly accessible once the assessment is finalized. EPA has conducted six assessments: the most recent NATA was released in August 2018 and is based on data from 2014; the previous assessments were based on data from 1996, 1999, 2002, 2005, and 2011.

The Role of the Human Health Risk Assessment Process within NATA

The NATA begins with a national emissions inventory of all outdoor air toxics sources, which includes source characterization (including air emissions information characterization data like emission or stack parameters, emissions rates, and source information), and then uses air dispersion modeling to arrive at predicted concentrations of air toxics. Exposure is then quantified by identifying health risk hazards established though EPA's Integrated Risk Information System (IRIS) and other tools. The end result of this process, as shown in Figure 1, is a quantitative assessment of health risks, both cumulatively and for individual toxics. Given the complexity of the process, it is understandable why the NATA is updated only every three years.

1. Toxicity Information

The IRIS Program, managed by EPA's National Center for Environmental Assessment (NCEA) in the Office of Research and Development (ORD), is EPA's main source of toxicity data. The program involves evaluation of risk determinations for various air toxics to ensure they remain protective of human health. As depicted in Figure 2, this is a lengthy process based on review of the scientific literature and health risk study data, and assessments undergo multiple rounds of internal and public review before being published.

Because IRIS assessments for an air toxic are subject to revision, facilities emitting air toxics should remain aware of IRIS updates pertinent to their operations. One can easily determine those pollutants that have been reassessed and those for which risk assessments are in development by accessing the NCEA website for the list of Assessments and Assessments in Review.

In late 2016, for example, the IRIS assessment process for ethylene oxide was concluded and resulted in an update of the inhalation unit risk estimate that was significantly lowered (making it more stringent). This revised risk value was then utilized in the NATA released in 2018.

2. Emissions Inventory Data

The NATA uses data from the National Emissions Inventory (NEI), which is a comprehensive and detailed estimate of air emissions of criteria pollutants, criteria precursors, and hazardous air pollutants from air emissions sources (see Figure 5). The NEI, publicly released every three years, is based primarily upon data provided by state, local, and tribal air agencies for all sources of air emissions in their jurisdictions. These agencies develop the data from their emission inventory reporting programs established under the Clean Air Act, including those which require permitted sources of air emissions to annually report to these agencies (this annual emissions reporting process is frequently linked to emissions fees and programs set by agencies). Stack parameters, detailing how emissions are emitted to the atmosphere, are also inputs to the NEI through the same agencies' reporting process.

NATA Figure 3

Notice the importance of the NEI data in the NATA process. If this emissions and stack parameter information (generated ultimately at the source level, by the source itself) is not accurate, the NATA results will have systemic errors. Thus, sources are encouraged to annually review reported parameters-stack heights, pollutant emission rates, stack location coordinates, stack exit velocity, stack diameters, and stack exhaust temperature-because (as shown in Figure 5) these data are the basis of input data for the initial air dispersion modeling analysis, which the NATA uses to predict the concentrations of toxics. Unfortunately, as the NEI system feeds the NATA, in the worst case, the old adage may apply: "garbage in, garbage out." Therefore, sources should ensure that their reported data are accurate, as that data will be used to model the source's contribution to predicted concentrations.

NATA Figure 4 and 5

With respect to source data accuracy, one further point of caution is worth mentioning: the importance of consistency across toxics emissions databases. As noted, the NATA is initially built upon NEI data, which is developed from source inventory information collected from annual reports of emissions and associated exhaust parameters submitted to air agencies. However, another toxics database is readily available to the public through the Toxics Release Inventory (TRI) program, authorized under Section 313 of the Emergency Planning and Community Right-to-Know Act (EPCRA). This database stores historical emissions information for chemicals required by TRI, many of which are the same chemicals in the NEI. Although the NATA process includes several quality assurance/quality control steps to ensure data accuracy, and reported air emissions in TRI should be the same as the reported air emissions in the state annual reporting process, it is unfortunately not uncommon for NEI data for a particular toxics source to differ from TRI data for the same source. In this era of readily available data, differences between TRI and NEI data can become opportunities for discussion, confusion, and consternation.

New NATA Implications

Nationwide, total emissions of air toxics are declining, and air quality monitoring data show that concentrations of many toxics in the air are trending downward. The new NATA estimates that the nationwide average cancer risk from air toxics exposure is 30 in 1 million. Per the fact sheet1 released by EPA with the new NATA, "About half of that risk comes from the formation of formaldehyde - produced when other pollutants chemically react in the air. This is known as secondary formation, and comes from emissions from industries, mobile sources, and natural sources. The other half of the nationwide cancer risk comes from pollution that is directly emitted to the air."

Although air quality and health risks from toxics exposure may be improving nationwide, local conditions can remain a challenge. Of particular concern for both sources and communities are localized elevated cancer risk due to toxics exposure, particularly at the census tract level (e.g., smaller aggregations of neighborhoods within a county). The new NATA estimates that industrial emissions of three pollutants - ethylene oxide, chloroprene, and coke oven emissions - contribute to most of the risk in localized census tracts with an elevated cancer risk (which constitute less than 1% of all U.S. census tracts).

EPA defines an elevated cancer risk as a potential cancer risk greater than 100 in 1 million and uses this risk level to determine if sources need to reduce emissions. At this level, sources will fall under additional scrutiny that will require more detailed analysis and actions to address the questions below.

What are the Actual Emissions of the Toxic?

If there is an elevated risk predicted by NATA, oftentimes the first inquiry will be concerning the accuracy of the emissions information relied upon in the NEI, which may be at the source level and points to the challenge of how to quantify the actual emissions of a toxic. EPA has long-standing guidance for quantifying emissions, which recognizes that the process may be very simple or extremely complex, depending on the facility size, the nature and number of processes, and the emission control equipment in place. EPA has also recognized that the emissions to be reported, the availability of data, and the cost considerations should be considered. EPA developed a chart to consider the reliability of risk sensitivity estimates based on various emissions estimations procedures.2

When an elevated cancer risk, as predicted through NATA or any other modeled or monitored data, is being investigated, and a local source or sources may emit the toxic contributing to that risk, EPA or the local air agency will generally request for the "best" possible emissions data - typically data derived from very specific, protocol-driven testing and continuous emissions monitoring programs. As shown in Figure 6, with improved accuracy comes increased costs. Even with highly sophisticated testing protocols, the variety of scenarios under which a source or specific equipment at a source may operate, the limitations of the testing method, and the duration of the test, may yield less than conclusive results regarding emissions from a particular operation. Batch manufacturing processes, unlike continuous processes, present a particularly challenging testing environment.

Speaking of emissions quantification, another challenge is quantifying fugitive emissions - how good are the source data with respect to emissions that may not be directed to a stack or vent, but are emitted through unintended leaks or process upsets? If a source is under scrutiny for elevated risk, it should expect to defend both stack and fugitive emission estimation methodologies.

Are the predicted modeling concentrations of NATA accurate?

As described above, the NATA utilizes air dispersion modeling to predict ambient air concentrations. And, while modeling is a reasonably effective tool, particularly for predicting the general, longer-term concentrations, it has inherent limitations under many scenarios (e.g., releases involving fugitive emissions, nearby buildings/structures, some meteorological conditions, etc.) that affect its accuracy at specific locations.

Oftentimes, the solution to the "reasonable effectiveness" of modeling is to gather more actual concentration data through an ambient monitoring system. Unfortunately, because there are many different types of toxics, and because many of these pollutants are location-specific, EPA does not operate a large national network of air toxics monitors. EPA currently has a total of 27 sites across the U.S. to monitor over 100 air toxic pollutants, with the main purpose of this network being to assess trends over time and the effectiveness of emission reduction programs. Twenty-seven sites across the entire U.S. is anything but a dense network. For example, the relatively populous state of Illinois has only a half-dozen monitoring locations in the entire state, which monitor only a very limited number of metal and organic air toxics.

If a source is under scrutiny for an elevated risk, it should expect a discussion about establishing a fence line monitoring data collection system to validate predicted concentrations from modeling. This type of ambient monitoring system is already required by some EPA rules, such as the 2015 Risk and Technology Review for the Petroleum Refinery Sector (also known as the Refinery Sector Rule), which requires continuous fence line monitoring for benzene.

NATA Figure 6

Will the Risk Estimated by NATA Lead to More Regulation?

In some instances, the toxic pollutants evaluated under NATA are also classified as Hazardous Air Pollutants (HAPs). There are 187 toxic air pollutants regulated as HAPs by EPA, and these HAPs are regulated to meet technology-based standards known as National Emissions Standards for Hazardous Air Pollutants (NESHAPs). On its website3 for ethylene oxide (a HAP), which is also an air toxic evaluated under NATA, EPA notes that it has begun to review its air toxics emissions standards for miscellaneous organic chemical manufacturing (MON). EPA last updated this rule in 2006 and is required to complete its risk and technology review (RTR) of the rule by March 2020. In this RTR, EPA will consider risks to human health and the environment, along with advances in work practices, processes, or emission controls that can further reduce emissions.  As part of this process, EPA typically gathers additional information on industrial HAP emissions, including where emissions occur, how those emissions can be controlled, and how current emission controls can be improved. The relevance here is that EPA has also indicated it may seek information from emissions testing at sources that emit ethylene oxide, as part of this MON RTR process, focusing first on areas where NATA estimates that there is an elevated cancer risk.

NATA Can Have Far-reaching Consequences

Finally, in any elevated cancer risk analysis prompted by the NATA, much of the information generated by EPA or state, local, and tribal environmental agencies will be made public, per EPA's transparency provisions under Next Gen Monitoring initiatives (see related Trinity EQ article, Fall 2017, "Next Gen - A Case Study in Advanced Monitoring and Data Transparency").  Once data becomes public, it can be used in various ways, including calls from interested public parties for more study, more testing, more real-time data, and more transparency. And, as we have seen in the past, accusations may be made against alleged perpetrators for negligence, trespass, nuisance, or even adverse health impacts, leading to individual and class action lawsuits.

And sometimes, this all starts with NATA.


2Introduction to Stationary Point Source Emission Inventory Development, May 2001, prepared by Eastern Research Group, Inc. for the STAPPA ALAPCO EPA Emission Inventory Improvement Program, Point Source Committee, Figure 1.4-1