POLLUTION LOCATOR|Caveats on Scorecard's Hazardous Air Pollution Data

cautions against "using the results of the National-Scale Air Toxics Assessment modeling exercise alone to draw real-world conclusions about current local conditions" because of the limitations involved in modeling exposures using 1996 emissions data. There have clearly been changes in emissions since 1996 in pollution sources and quantities that may affect the reliability of EPA's exposure estimates. These same changes may also impact the accuracy of Scorecard's source apportionment. Some pollution sources have come under substantial regulatory controls since 1996, and others have grown in importance.

The best approach to evaluating the accuracy of NATA estimates is comparison with both historical and current monitoring data:
Comparing NATA estimates with 1996 monitoring data, EPA found that "modeled estimates for most of the pollutants examined are typically lower than the measured ambient annual average concentrations." Comparing NATA estimates with more current monitoring data, Scorcard shows that modeled estimates are generally consistent with current concentrations. NATA's systematic underprediction of monitored 1996 concentrations is presumably due to gaps in its emissions inventory. Reductions in pollution releases since 1996 have effectively reduced the significance of these gaps, increasing the consistency between NATA estimates and current monitoring data.

Scorecard compares NATA estimates with monitoring data from several states, and provides online access to these data whenever they are available.

NATA's estimated air concentrations are based on the application of dispersion models to emissions data and depend on a number of assumptions. See EPA's discussions of the
limitations of its modeling. NATA has been evaluated positively by extensive peer review, but important uncertainties remain about resulting predicted ambient concentrations and responsible source categories. Errors in emissions inventories (e.g., missing or incorrectly located sources), variations in pollution species released by different source categories, regional and seasonal climatic variability, and uncertainties in environmental fate modeling can all contribute to misspecification of local concentrations.

Uncertainties in the accuracy of exposure data and source apportionment increase as the scale of geographic analysis decreases to the census tract or source-specific level. In its national
NATA report, EPA aggregates and presents information at the county-level or higher and strongly cautions that census tract-level estimates are not reliable. EPA recommends the county level of resolution because emissions inventory data for some pollutants and sources are only available at the county level and there are large uncertainties regarding exposure modeling parameters at the tract level.

Scorecard utilizes
conventional health risk assessment methods to characterize potential cancer and noncancer risk. Note that risk estimates are calculations based on models - they are useful for ranking purposes but are not necessarily predictive of any actual individual's risk of getting cancer or other diseases.

Risk assessment methods rely on the use of assumptions to address gaps in scientific understanding and data, which can lead to mis-estimation of health risks. Some assumptions err on the side of health protection and may result in overestimation (e.g., presuming animal carcinogens are potential human carcinogens). Other assumptions may result in underestimated health risk (e.g., presuming all people have equal susceptibility to toxicants). Given that NATA appears to underestimate exposures, it is prudent to use health-protective assumptions to generate cancer and noncancer dose-response factors.

NATA includes only chronic health effects from inhalation exposure to outdoor sources of air toxics. Effects from less-than-lifetime exposures (e.g., accidental chemical releases) and total exposure to air toxics (e.g., including indoor air pollution sources) require further evaluation.

Exposures to HAPs which are persistent or bioaccumulative toxicants may be significantly underestimated by EPA's National-Scale Air Toxics Assessment, because EPA's methodology only considers inhalation exposures. Ingestion (of contaminated food, water or soil) is likely to result in substantially greater human exposures than inhalation for the following compounds: lead, mercury, cadmium, polychlorinated biphenyls, dioxin, hexachlorobenzene, polycyclic aromatic hydrocarbons, and polycyclic organic matter.

Exposures to HAPs present largely as a result of secondary reactions in the atmosphere or affected by larger-scale, regional transport may be significantly underestimated by the ASPEN model, because it is not formulated to handle these processes. Pollutants potentially affected include: formaldehyde, acetaldehyde, acrolein, and lead.

Estimated health risks from chromium exposures are uncertain because source-specific information about the speciation of emissions is lacking, and Scorecard's general assumption that one-third of the total ambient chromium is present in the carcinogenic hexavalent form will not capture source-specific variations.

Seven HAPs modeled in NATA have not undergone the same degree of data quality control as the 33 priority pollutants that are the main focus of NATA. These seven pollutants are: styrene, xylene, toluene, ethyl benzene, propionaldehyde, hexane, and methyl tert butyl ether. Neither the emissions nor modeled concentrations from these pollutants have undergone state agency review or quality assurance checks.

Limitations in available emissions inventories for mobile sources may impact the accuracy of NATA results. In its assessment of onroad emissions, EPA uses population data as a surrogate to project vehicle miles traveled. This method results in underestimation of onroad emissions in more suburban counties, while largely overestimating onroad emissions in urban counties. In its assessment of nonroad emissions, EPA uses economic measures of construction activity as a surrogate to project emissions from nonroad engines. This method distorts the relative contribution of nonroad diesel sources in urban counties where housing and commercial building prices are extremely high. This distortion particularly impacts exposure and risk estimates for New York City metropolitan counties, where the relative contribution of nonroad diesel contributions are unrealistically high.