Limitations in the 1996 National- Scale Air Toxics Assessment

Given the broad scope of this assessment, it is subject to a number of limitations with respect to the information and analysis tools on which it depends. Largely, the limitations relate to the massive amounts of data required to perform the assessment, EPA’s limited ability to gather all such data, and the assumptions which are made when the data are less than perfect. In some cases, these assumptions are built into a computer simulation model, and in others, they are incorporated into the data processing steps performed before and after computer modeling. The general limitations are listed below, followed by those limitations specific to each step in the assessment. Despite these limitations, it represents an important step in characterizing air toxics risks nationwide. Continued research will enable future assessment activities, both at the national level and at more local refined levels, to yield improved assessments of cumulative air toxics risks.

General limitations

• Limited to outdoor sources: Potentially important sources and pathways of risks to public health are beyond the scope of this assessment. Specifically, while we recognize that indoor sources of air toxics likely contribute substantially to the total exposures that people experience for a number of toxic air pollutants, these emissions were not included in this initial assessment.
• Limited to inhalation exposure: For some of the toxic air pollutants, other routes of exposure such as ingestion are generally known to be the main route of exposure.
  
• Limited to 34 Air Pollutants: EPA is focusing on the 33 air toxics identified as priority pollutants in EPA's Integrated Urban Air Toxics Strategy, plus diesel particulate matter. The 33 air toxics are a subset of EPA's list of 188 air toxics. EPA does not have all of the scientific information necessary to perform the full exposure assessment for the remaining 155 air toxics. As this information is developed, EPA will determine when the assessment can be completed for the remaining pollutants (2001 or later).
• Results represent 1996: Since the results represent estimates for 1996, they do not reflect potentially significant air toxics reductions resulting from the implementation of Federal and State programs or from industry initiatives since 1996.

Limitations in the Emissions

The National Toxics Inventory (NTI) is a compilation of quantitative information concerning the mass of air toxics emitted into the atmosphere (through smokestacks, tailpipes, vents, etc.). The 1996 NTI is the underlying basis for all emissions estimates used in this assessment (with the exception of diesel particulate matter). Consequently, limitations in the 1996 NTI contribute to limitations in the subsequent steps in the assessment. The massive amounts of data in the NTI give rise to its limitations:

Consistency: The 1996 NTI is a composite of emissions estimates generated by state and local regulatory agencies, industry, and EPA. Because the estimates originated from a variety of sources and estimation methods, as well as for differing purposes, they will in turn vary in quality, included pollutants, level of detail and geographic coverage. However this compilation of emissions estimates represents the best available information to date.

Variability in quality & accuracy of emission estimation methods: The accuracy of emission estimation techniques vary with pollutants and source categories. In some cases, an estimate may be based on a few or only one emission measurement at a similar source. The techniques used and quality of the estimates will vary between source categories (e.g., some have been better studied than others) and between major, area, and mobile source sectors.

Default parameters: Air dispersion models require a great deal of detail in order to predict concentrations. Where these details were not available in the original NTI data, certain default assumptions were made. These include:

• Emissions parameters (stack height, flow rate, temperature, etc.) - these are added when point sources were missing these details using averages based on similar source types.
• Geographic locations - where specific latitude and longitude coordinates did exist or were in error, a facility was randomly located in the county where it resides (only a small percentage of the total 1996 NTI emissions was subject to this default).
• Chemical species characteristics (e.g., vapor versus particle ratios, individual chemicals included in groups, etc.) - where such details were not available, assumptions are made based on engineering judgment or previous modeling assumptions. For example:

  EPA grouped individually reported toxic air pollutant species (e.g., lead oxide) into toxic air pollutant groups (e.g., lead compounds) based on what will be most useful for subsequent risk assessments.

  EPA also adjusted specific metal compound species emissions (e.g., cadmium chloride) to account for only the weight of the metal part of the compound, which will subsequently be used in risk assessments and monitoring data comparisons.

• Actual annual emissions estimates for some major sources - where an annual emission estimate was not available for a source for 1996, EPA used an adjustment to convert available emissions (e.g., actual hourly or maximum allowable annual emissions) to annual.
• Spatial allocation of emissions - since the emissions data for several area and other source emission categories and all mobile source emission categories were provided on an aggregated county-level basis, modeling assumptions were made to distribute these emissions among the census tracts in the county. Depending on the type of source, emissions are distributed based on surrogates such as land-use (e.g., industrial land) and census data (e.g., population density) available for each tract in each county.

The ASPEN (Assessment System for Population Exposure Nationwide) Model is the computer simulation model used to estimate ambient air toxics concentrations in this assessment. To enable the simulation of a large number of pollutants nationwide, a number of simplified approaches must be incorporated into the model, giving rise to the following limitations:

Simplified approaches used to simulate atmospheric deposition and transformation: The complex processes associated with the wet and dry deposition of particles and gases from the atmosphere to the ground are simulated using simplified algorithms in order to permit timely computation for the full modeling domain. Similarly, atmospheric transformations are included in the simulation, but are handled using a computationally simplified approach.

Terrain impacts on dispersion not included: For simplicity, no terrain features have been included in the modeling approach. While this omission will have a substantial impact on a limited number of populated locations, it has been shown to have limited impact on a national scale assessment of this type.

Daily emission patterns held constant: For simplicity, annual average concentrations are based on emission patterns which repeat daily. For some source types, this assumption may have a significant effect, but it’s impact on annual average concentrations is expected to be minimal.

Meteorological data extrapolated from nearest measurement site: Meteorological data used in calculating pollutant transport are derived from measurements taken at the nearest National Weather Service (NWS) monitoring station, which is usually the nearest airport. In some cases, this may be far from the source and result in errors in locating maximum impacts from individual sources, but it has been shown to have a limited impact on a national-scale assessment of this type.

Long-range transport not included in the simulation: The dispersion modeling approach simulates only pollutant transport within 50 kilometers of any individual source. To correct this shortcoming, a uniform background concentration is added for some pollutants for which information is available. This simplified approach will impact the ability to predict concentrations for some pollutants more than others. Additionally, much of the data used to estimate background concentrations are representative of historical conditions and may not accurately represent current conditions.

Simplified source modeling: For timely simulation, all "area and other" source emissions are modeled as if they occur in a number of discreet locations rather than using more complex area, volume or line source modeling algorithms which may better reflect the actual characteristics of the release. Additionally, a Gaussian dispersion modeling approach is used which contains simplified assumptions.

Source contributions to predicted ambient levels limited: The current national-scale application is limited to providing source contributions from only four large emission sectors (major, area and other, onroad mobile, and nonroad mobile). More detailed source contribution will require additional, more refined simulations, perhaps on a localized scale.