See the latest EHS federal and state regulatory updates due to COVID-19

PM2.5 FactoryOn March 4, 2013, the U.S. EPA published draft guidance for PM2.5 permit modeling with the intent of advising PSD permit applicants on how to demonstrate that they will not cause or contribute to a violation of the NAAQS and PSD increments for PM2.5. The guidance incorporates previous related modeling guidance and addresses how the applicant should account for secondary PM2.5 (formed through the chemical reaction of certain precursor compounds, primarily, NOx and SO2) for the first time. This guidance is based on and consistent with EPA’s Guideline on Air Quality Models, (40 CFR 51, Appendix W), although it extends the regulatory modeling techniques to include the effects of precursor emissions. Because each permitting action will be considered on a case-by-case basis, according to EPA, this document does not preclude any particular approach that either an applicant or permitting authority may take to conduct the required compliance demonstrations. Nonetheless, EPA guidance typically carries considerable weight when evaluating viable options.

Status of EPA’s Screening Tools

As stated on page 10 of EPA’s draft guidance, “the EPA has historically allowed the use of screening tools to help facilitate the implementation of the PSD program and streamline the permitting process in circumstances where proposed construction is projected to have an insignificant (or de minimus) impact on air quality.” This allows sources to determine what further data collection and modeling analyses may be required as part of the permit application. These tools have included Significant Emission Rates (SERs), Significant Impact Levels (SILs), and Significant Monitoring Concentrations (SMCs). Because of a January 22, 2013 ruling by the U.S. Court of Appeals for the District of Columbia Circuit, all of these tools are not necessarily available for use at this time for PM2.5.

The use of dispersion modeling to demonstrate that ambient concentrations are less than SMCs is no longer allowed. In 2010, EPA promulgated an SMC for PM2.5 to provide a screening tool that would allow a permit applicant to be exempt from the statutory requirement to collect and compile preconstruction ambient air monitoring data. The basis for the exemption would be dispersion modeling. Even though the District court has vacated the PM2.5 SMC due to perceived inconsistency with Section 165(e)(2) for the CAA because the court declared that collecting preconstruction monitoring was a “rigid mandate,” PSD permit applicants may continue to meet preconstruction monitoring requirements. These monitors must be found to be representative of background concentrations by the permitting authority consistent with information offered by EPA in a March 4, 2013 Questions and Answers document on the topic.

Dispersion modeling may still be used to preclude full NAAQS and PSD increment modeling by demonstrating ambient concentrations that are less than SILs.

Regulatory Milestones in PM2.5 NAAQS Development
July 18, 1997 EPA published an annual and 24-hour NAAQS for PM2.5 
Oct 23, 1997 EPA established the PM10 Surrogacy Policy to alleviate the difficulties associated with PM2.5 monitoring, emissions estimation, and modeling
Sept 21, 2006 The 24-hour PM2.5 standard was lowered from 65 μg/m3 to 35 μg/m3 
May 16, 2008 EPA published final implementation rule regarding New Source Review for PM2.5 and restrictions on the use of the PM10 Surrogacy Policy
July 1, 2009 EPA issued an administrative stay (that was subsequently extended) of the grandfathering provision that has allowed PSD permit applicants to continue use of the PM10 Surrogacy Policy in certain situations
Feb 11, 2010 EPA proposed repealing the grandfathering provision and ending the PM10 Surrogacy Policy for approved SIPs
Mar 23, 2010 EPA issued a guidance memo recommending interim procedures related to the statistical form of the standard and addressing impacts of secondary PM2.5 
Jan 7, 2011 NACAA Workgroup provided modeling recommendations related to emissions inventories, secondary PM2.5 emissions, and representative background concentrations
May 18, 2011 EPA published the Implementation of the New Source Review (NSR) Program for PM2.5, repealing the grandfathering provision
Jan 4, 2012 EPA granted a petition requesting specification of air quality models for ozone and PM2.5 for use in PSD permitting
Dec 14, 2012 The annual standard was lowered from 15 μg/m3 to 12 μg/m3 
Jan 22, 2013 The US Court of Appeals for the District of Columbia Circuit Court granted a request to EPA to vacate and remand the SILs for PM2.5 and to vacate the SMCs for PM2.5 
Mar 4, 2013 EPA released Questions and Answers document on the SIL/SMC court decision containing considerations for implementation of permitting for PM2.5 until EPA revises regulations in accordance with court decision, and draft guidance for PM2.5 permit modeling addressing secondary PM2.5 
Mar 18, 2013 New annual PM2.5 NAAQS is fully promulgated at 12 μg/m3 


However, this technique must first pass an additional hurdle to check if the difference between the PM2.5 NAAQS and the monitored PM2.5 background concentration is less than the SIL. If the difference between the NAAQS and background is less than the SIL, EPA recommends that a cumulative impacts analysis (NAAQS and PSD increment) be completed without regard to the project’s modeled impacts against the SIL. Also, even the value of the SIL may be in question as the PM2.5 SILs were also vacated. In this case, EPA was left with the option of proposing alternate SILs. Until corrected SIL regulations are proposed, EPA recommends how to use the SILs in a manner that is consistent with CAA intentions, namely modeling project emissions increases against the SIL to determine whether a cumulative impact analysis is required, but only if the full SIL is available. Specifically, if the difference between the PM2.5 NAAQS and the measured PM2.5 background concentrations are greater than the SILs, one may conclude that a source with an impact below the SIL will not cause or contribute to a violation of the NAAQS and thus, forego a cumulative impact analysis. This, of course, should be with the agreement of the state permitting authority on a case-by-case basis.

The 2008 Significant Emissions Rates (SERs) of 10 tpy for direct emissions of PM10 and 40 tpy for both NOx and/or SO2 remain in place to indicate whether proposed projects will emit sufficient amounts of the pollutants to warrant further review under the PSD program. Currently, the SER for each precursor species is the same as if these were primary pollutants. Recent work reported by the National Association of Clean Air Agencies (NACAA) in a 2011 study, PM2.5 Modeling Implementation for Projects Subject to National Ambient Air Quality Demonstration Requirements Pursuant to New Source Review, suggested that because only a portion of these emissions will convert to PM2.5, perhaps the SERs for NOx and SO2 should be much greater. Final consideration by EPA will determine whether the SERs for precursors will remain as is or be increased. New or modified sources of direct and precursor emissions that exceed these levels must apply one of four levels of analysis that are proposed in Table III-1 of the Draft Guidance for PM2.5 Permit Modeling. The application of these analyses, assessment cases, or approaches (all names used in the EPA draft guidance) is described in the following section.

Table 1. EPA Recommended Approaches for Assessing
Primary and Secondary PM2.5 Impacts - Primary
Assessment CaseDescription of Assessment CasePrimary Impacts ApproachSecondary Impacts Approach
Case 1: No Air Quality AnalysisDirect PM2.5 emissions < 10 tpy SER
NOx and SO2 emissions < 40 tpy SER
Case 2: Primary Air Quality Impacts OnlyDirect PM2.5 emissions ≥ 10 tpy SER
NOx and SO2 emissions < 40 tpy SER
Appendix W preferred or approved alternative dispersion modelN/A
Case 3: Primary and Secondary Air Quality ImpactsDirect PM2.5 emissions ≥ 10 tpy SER
NOx and/or SO2 emissions ≥  40 tpy SER
Appendix W preferred or approved alternative dispersion model
  • Qualitative
  • Hybrid qualitative/quantitative
  • Full quantitative photochemical grid modeling
Case 4: Secondary Air Quality Impacts OnlyDirect PM2.5 emissions < 10 tpy SER
NOx and/or SO2 emissions ≥ 40 tpy SER
  • Qualitative
  • Hybrid qualitative/quantitative
  • Full quantitative photochemical grid modeling

Source: Page 21 Table III-1 of EPA's Draft Guidance for PM2.5 Permit Modeling, March 2013


Assessment Cases

To assist permit applicants and regulatory authorities in conducting a significant impact analysis and a cumulative impact analysis if required, EPA has provided four Assessment Cases, as shown above in Table 1. These cases can be used to determine the appropriate method of air quality analysis needed to demonstrate whether a project is less than or greater than the SIL (if the full SIL is available) and compliance with the PM2.5 NAAQS. Each case is contingent on the projected increase in tons per year of direct PM2.5 and NOx and SO2 emissions and their comparison to the SER—10, 40, and 40 tpy, respectively.

In Case 1, where both direct and precursor emissions are below the SERs, no further analysis is required. In Case 2, where only direct PM2.5 emissions exceed the SER, then only direct emissions must be addressed with dispersion modeling using the preferred regulatory model, AERMOD. In Case 3, where direct PM2.5 emissions exceed the SER and either NOx or SO2 exceed the SER, then direct emissions must be addressed with AERMOD and precursor emissions must be characterized through either qualitative analysis, a hybrid qualitative and quantitative analysis, or a full quantitative photochemical grid modeling analysis. Finally, in Case 4, where direct emissions are below the SER, but either NOx or SO2 exceed the SER, only those precursors must be assessed through the same three options as in Case 3.

Significant Impacts Analysis

In Cases 2 and 3 above, the impact of direct PM2.5 emissions can generally be estimated by using AERMOD, the Appendix W preferred model for near-field PM2.5 modeling (less than 50km), or an acceptable alternative (which generally requires a demanding exercise following the steps provided in Section 3.2 of the Guideline on Air Quality Models). For Cases 3 and 4, where the SERs for NOx and/or SO2 are exceeded, the applicant must choose between the qualitative, hybrid qualitative/quantitative (hybrid), or full quantitative approaches. Although EPA expects the full quantitative approach using a photochemical grid model to be rarely needed, “rarely needed” has not yet been defined.

Qualitative Assessment of Secondary PM2.5 Emissions
In certain cases, the permit applicant may be able to rely on qualitative analysis of the emissions source and nearby environmental conditions to demonstrate that its PM2.5 precursor emissions will not cause or contribute to a PM2.5 NAAQS or increment violation. Among the factors that may be considered are:

  • Background PM2.5 concentrations with speciation (i.e., sulfate, nitrate, organic, etc.), seasonality, and long-terms trends identified
  • Background concentrations of pollutants that can form secondary PM2.5, such as NH3, VOC, and ozone, as well as any mitigating factors, e.g., particularly low levels of NH3
  • Characterization of nearby air emissions sources
  • Representative meteorological conditions and their seasonal affects on PM2.5 formation
  • Existing photochemical grid modeling performed for other regulatory purposes that may inform the estimation of secondary PM 2.5 formation in the area

EPA provides an example of a qualitative analysis in the draft guidance for a permit issued to Shell for its Discoverer drill ship in the Chukchi Sea off Alaska. This example, while covering all of the main areas described by the draft guidance, is not useful for more typical applications concerning sources in urban areas or even rural areas, most of which are not located in such an isolated location as the Discoverer drill ship. Typical background concentrations are higher than those described; more nearby PM2.5 sources are generally present; and levels of other participant species (e.g., ammonia and VOCs) are generally higher.

Hybrid Assessment of Secondary PM2.5 Emissions
When a qualitative approach cannot justify the conclusion that a source’s secondary PM2.5 emissions combined with its direct emissions will not cause or contribute to a violation, air quality modeling results regarding the NOx and SO2 emissions may also be needed. Existing model results may be acceptable if they are sufficiently representative of the proposed source and nearby conditions. “Pollutant offset ratios”1 may be acceptable to convert precursor emissions to equivalent direct emissions which can then be combined with true direct emissions and modeled with AERMOD. However, the offset ratios used must be demonstrated to be appropriate for both the source type as well as the region of the country, a task that should involve EPA collaboration. Concerns in this methodology include whether the location of maximum direct PM2.5 will be commensurate with that location where the formation downwind of the secondary PM2.5 takes place.

Two other issues related to conversion and past tools used for defining sources and emissions used in modeling relate to 1) the method of dividing emissions by distance (Q/D) to determine sources to exclude in the modeling and 2) using AERMOD with 100 percent conversion assumption as a conservative screening tool. Both methods were rejected by EPA and not included in the draft guidance.

Quantitative Assessment of Secondary PM2.5 Emissions
In the rare cases that photochemical grid modeling is found to be necessary, the model must meet Appendix W criteria for an alternative model and modeling procedures. Using gridded meteorological data, these models provide a comprehensive assessment considering emissions, chemical transformation, pollutant transport, and deposition. The Comprehensive Air Quality Model with Extensions (CAMx) and the Community Multiscale Air Quality (CMAQ) models are two potentially appropriate models for this purpose, although the suitability of these and similar models for estimating single source impacts is still under evaluation. In lieu of extensive guidance on how to apply photochemical grid models for NAAQS compliance demonstrations, EPA recommends extensive consultation between the permit applicant and reviewing authority to develop a comprehensive modeling protocol.

Comparison to the SIL

EPA recommends that the approach for comparison to the SIL will also depend on how the direct and precursor PM2.5 emissions compare to the SER, as detailed in Cases 2, 3, and 4. For Case 2, where only direct emissions are a concern, the applicant can compare AERMOD results to the SIL with the following refinements (due to the form of the PM2.5 NAAQS):

  • For National Weather Service (NWS) meteorological data - Use the highest of the 5-year averages of the maximum modeled 24-hour or annual PM2.5 concentrations for each year at each receptor using 5 years of NWS data
  • For onsite meteorological data - Use the highest modeled 24-hour or annual PM2.5 concentrations predicted across all receptors based on one year of site-specific meteorological data, or the highest of the multi-year averages of the maximum modeled 24-hour or annual PM2.5 concentrations predicted each year at each receptor, based on two or more years, up to 5 complete years of onsite meteorological data

PM2.5 RefineryFor Case 3 (where both direct and precursor PM2.5 emissions exceed the SERs), the applicant must account for both precursor emissions and direct emissions to compare to the SIL. The estimate of direct emissions can be derived using the same approach as Case 2 above. However, the precursor emissions present more of a challenge. For example, it would be difficult to utilize the qualitative approach unless a compelling argument could be made that very little of the precursors would convert to PM2.5 (due to a lack of NH3 in the local atmosphere, for example). For a full quantitative analysis with photochemical grid modeling, the analyst must determine an appropriate method to combine estimates of direct emissions from AERMOD with estimates of indirect emissions from the photochemical model. Combining peak concentrations from each comprises the most conservative approach, however, the applicant may elect a different method should the most conservative approach produce problematic results. Full temporal and spatial pairing of the primary and secondary PM2.5 impacts is difficult due to the difference in receptor treatment between AERMOD and photochemical grid models. The applicant should confer with the regulatory authority to negotiate an agreed upon approach.

For Case 4, where precursor PM2.5 emissions exceed the respective SER but direct emissions do not, the SIL comparison need only address the precursor emissions and can be based on the qualitative, hybrid, or quantitative approach. In practice, however, due to the uncertainties around the qualitative and hybrid approaches, the applicant will likely either have to proceed with a full photochemical grid modeling analysis or forego the SIL comparison and focus on NAAQS compliance using a full cumulative analysis.

Cumulative Impact Analysis

When an applicant is unable to apply the screening approach using the SIL to demonstrate that it will not cause or contribute to a NAAQS or increment violation, a cumulative impact analysis must be performed to account for the emissions of the new or modified source, nearby sources, and background concentrations in the area. Specifically, the analysis should include the following:

  • Primary impacts from direct PM2.5 emissions and secondary impacts from precursor emissions from the new or modified source
  • Primary impacts from direct PM2.5 emissions from relevant nearby sources
  • Monitored background concentrations (for NAAQS analyses only)

Once again, the primary impacts from direct emissions should be estimated using the AERMOD dispersion model; the estimates of secondary PM2.5 impacts will depend on which of the three allowed approaches is used; and the background concentrations will be derived from monitoring data. One change from previous guidance now allows adding the 98th percentile modeled value (rather than the peak modeled value) to the 98th percentile background value for comparison to the 24-hour PM2.5 NAAQS. This method of determining the cumulative impact is more consistent with the form of the NAAQS and typical practice for other pollutants.

EPA acknowledges that many states lack adequate PM2.5 inventories and has indicated that recommendations for developing PM2.5 emissions inventories for use in PSD applications will be forthcoming. Thus, the challenge of compiling representative inventory data may remain a problematic issue for modeling of a subject source and these recommendations need to be provided by EPA sooner rather than later to reduce uncertainties in permit applications.

In terms of the background concentrations needed for the analyses, cumulative impact modeling analyses should include either nearby sources or use nearby monitoring to account for nearby sources but not both. Previous guidance had mentioned a Tier 3 approach whereby modeled concentrations and background could be paired in time, but this is not mentioned in the current draft guidance.

Practical Considerations (Living with the Draft Guidance)

If your new or modified facility must address ambient PM2.5 impacts through dispersion modeling in a permit application, the following offers a series of questions to consider in deciding on an approach.

  • How much PM2.5 do you emit? You will need to know your facility’s direct PM2.5 emissions as well as its precursor PM2.5 emissions to determine which category of Case Assessment applies. This will assist in decisions on whether to change the project or plan for additional controls while in the formative stage. Design changes could allow the project to fall below critical SER levels for direct PM2.5, precursors, or both and thus, avoid some or all PSD requirements for air quality impacts.

  • How sophisticated is the state agency with respect to modeling requirements? The availability of existing photochemical model results to include in a hybrid analysis will vary from state to state with states that have already participated in regional transport programs like CAIR or CSAPR being more likely to have information available from large scale PM2.5 modeling. Additionally, the level of expertise to participate in protocol development and review full photochemical modeling analyses will vary depending on whether agency personnel have prior experience with SIP development for ozone or PM2.5 or, if needed, whether a regional planning organization completed modeling on their behalf.

  • Is the facility close to a “representative” PM2.5 monitor? The presence of a nearby PM2.5 monitor may help in justification that it is representative of air quality surrounding your facility, but the applicant should take care to ensure that sources of direct PM2.5 and precursors are not double counted if they are included in monitored concentrations and also included in the modeling analysis.

  • How close to the NAAQS is the representative PM2.5 monitoring data? If there is little “headroom” available, applicants may need to develop rationale for using a more distant PM2.5 monitor to represent “true” background. Alternatively applicants may need to be prepared to move directly to cumulative impact modeling and possibly a cause or contribute analysis that demonstrates that the proposed source or project does not have impacts above the SIL at the time and location of modeled NAAQS exceedances.

  • Can photochemical models be used for single sources? Photochemical models are designed to consider the complex interaction of multiple chemical reactions occurring with respect to atmospheric pollutants. These reactions can be a function of meteorology (i.e., temperature or humidity), background pollutant concentrations, and pollutant concentrations contributed by nearby stationary and mobile sources. Because of the complex nature of photochemical modeling, an individual source or facility cannot be accurately modeled without consideration of all of these other contributors to atmospheric chemical reactions. This means that, in general, photochemical modeling is much more time and resource intensive, requiring modeling of a full base case and an alternative case to determine impacts of operation of new or modified emission sources. EPA has made some effort to develop new methods using existing models to reduce the time and effort necessary to consider secondary PM2.5 formulation from individual new or modified sources, however there is not yet clear consensus or guidance regarding how single source photochemical modeling will be or even can be used for regulatory purposes.

  • Will the timing on the permit application be potentially extended by a secondary PM2.5 analysis? If additional consultation with regional EPA offices is necessary due to the expected case-by-case nature of secondary PM2.5 analysis, especially in the near term, the time to issuance of a permit may be increased. If so, capital and construction plans may need to be adjusted to account for additional delays in obtaining an air permit, which has frequently been a critical path item in construction projects even before the additional complication of secondary PM2.5 became a factor.



EPA’s Draft Guidance for PM2.5 Permit Modeling provides a general framework for how to address secondary formation of PM2.5 in regulatory dispersion modeling analyses. However, by no means does it provide a complete set of instructions regarding how these analyses must be completed. It remains vitally important that applicants for PSD permits work closely with regulators to develop modeling techniques that are scientifically sound and consistent with regulations and guidance to address impact of both direct and secondary PM2.5, when necessary. In general, this means that AERMOD should be used to estimate impacts of direct PM2.5 emissions and that the impacts of secondarily formed PM2.5 should also be considered using qualitative, hybrid, or quantitative techniques. Alternative models may be appropriate for selected situations, subject to approval by the EPA Regional Office. Additionally, monitoring alone may also be appropriate in certain situations where there is no appropriate model.

PM2.5 Flowchart

Due to the complex chemistry involved in secondary PM2.5 formation as well as the nature of the source and its emissions, EPA offers this draft PM2.5 guidance as a tool but ultimately the methodology must be specified by the local permitting agency.In order to keep the permit application on track, a modeling protocol is essential to ensure that a permit applicant provides the necessary analysis to meet agency requirements.

It is also important that those who will be impacted by guidance issued by EPA on PM2.5 permit modeling, including industrial facilities and utilities who may be subject to PSD permitting requirements, provide input on the process so that methodologies recommended by EPA through guidance are useable. To allow stakeholders to provide input, EPA has requested public comment on the draft guidance document. The public comment period for the draft guidance, which began March 4, 2013 and was initially scheduled to close on April 17, 2013, has been extended until May 31, 2013 with expected final guidance expected by late summer or early fall, 2013.


1 PM2.5 Modeling Implementation for Projects Subject to National Ambient Air Quality Demonstration Requirements Pursuant to New Source Review, Report from NACAA PM2.5 Modeling Implementation Workgroup, Washington, DC, January 7, 2011; as included in EPA’s 2008 Implementation of the New Source Review Program for PM2.5 (73FR28321) rule concerning interpollutant trading provisions for PM2.5 under state nonattainment area NSR programs for PM2.5.