CAO rules require that a quantitative or qualitative uncertainty evaluation be included in a Level 3 and Level 4 risk assessment. Specifying the assumptions and uncertainties inherent in the risk assessment helps place the risk estimates in proper perspective. Another use of uncertainty characterization can be to identify areas where a moderate amount of additional effort (such as better emission estimate methods, or characterization of chemical species) might significantly improve the evaluation.
Often it is difficult to quantitatively evaluate uncertainty. Generally, though, you can make a determination that an uncertainty will likely result in an underestimate or overestimate of risk. In some cases it will be unknown whether risks will be under- or over-estimated. There are various types of uncertainties associated with a risk assessment, including the following four major categories:
- Selection of Toxic Air Contaminants (TACs) for evaluation
- Emission rate calculations
- Exposure assessment assumptions
- Derivation of toxicity values
Selection of TACs for Evaluation
If there are no Risk Based Concentrations (RBCs) for some of the reportable emitted chemicals at a facility, state that risks from these chemicals cannot be quantitatively evaluated, which will likely result in an underestimation of risk. Also, some processes may emit chemicals that are not anticipated from currently available emissions data.
Often emission rates are calculated using published emission factors developed by EPA or other entities. In general, these factors are designed to be protective, and should not underestimate emissions and risk. If these factors are applied to similar processes for which the factors were not developed, emission estimates may be under or overestimated. You can partially quantify uncertainty in emission rates estimates by evaluating the range of emission factors that could be used. The use of source test data will reduce uncertainty related to emissions and associated risk.
Modeled and Measured Exposure Concentrations
Uncertainty associated with calculating actual exposures and therefore estimating risk can include potential inaccuracies in the emission inventory, variability in estimates of emission rates, uncertainties in air dispersion models, and protectiveness inherent in the exposure assumptions incorporated into the derivation of RBCs. TACs that are missed in the emission inventory or have underestimated emission rates will result in an underestimation of risk. Protective assumptions used in models will likely overestimate risk.
There is often high uncertainty associated with ambient monitoring air concentration results. To reduce uncertainty, a sufficient number of monitors need to be deployed, and they need to be appropriately placed to obtain representative data. Detection limits need to be adequate to detect TACs above RBCs. Uncertainty associated with evaluating non-detected concentrations needs to be discussed.
Sources of uncertainty for toxicity values can be discussed. For carcinogens, EPA has weight of evidence categories that can be presented and evaluated. For noncarcinogens, TRVs have associated uncertainty factors that can also be presented and discussed. Most risk assessments will include a summation of risk from multiple TACs. The assumption of dose additivity (inherent in the rule requirements for summing risk) does not consider possible synergistic or antagonistic effects. The potential for under- or over-estimating summed risks can be discussed. Evaluating noncancer risk by target organ (Appendix F) is one way to reduce uncertainty in the noncancer risk evaluation.