Voglio sapere a riguardo di…

Federal Register / Vol. 86, No. 156 / Tuesday, August 17, 2021 / Rules and Regulations
45883

contains an error in the source apportionment model that was discovered in December 2020.
Specifically, the commenter asserts that EPA does not know whether the specific beta version of the model used in their analysis contained the bug and associated source apportionment error.
The commenter further states that no known quantification of the error has been calculated and therefore it is unknown just how significant a change might be seen in the upwind state contribution to downwind receptors under the 2008 Ozone NAAQS or in potential application for future consideration of significant contribution under the 2015 Ozone NAAQS, and requests that EPA correct the source apportionment results and significant contribution calculations with the corrected version of the air quality model before making a final decision on removal of the I/M program from the Tennessee SIP.
Response B9: The commenter is correct that EPA relies on modeling developed for the Revised CSAPR
Update RCU to determine that removal of the I/M program will not interfere with other states ability to attain and maintain the 2008 ozone NAAQS under the good neighbor provision of the CAA.
The modeling was made available to the public in the proposed RCU on October 30, 2020. See 85 FR 68964. The comment period for that rulemaking was open from October 30, 2020, through December 14, 2020. Id.
Petitions for review of the RCU were due by June 29, 2021 in the D.C. Circuit.
See 86 FR 23054, 23164 April 30, 2021; see also CAA section 307b.
Additionally, EPA had previously determined that the CSAPR Update Federal implementation plan for Tennessee eliminated the States significant contribution to downwind ozone nonattainment or maintenance for the 2008 8-hour ozone NAAQS. See 81
FR 74504, 74508 October 26, 2016.
EPA disagrees that there was an error in the modeling that is material to this action. EPA used CAMX version 7, beta6
for the air quality modeling to support the RCU. This version of CAMX was the most up-to-date version of the model available at the time EPA performed air quality modeling for the RCU. The final CAMX version 7.0 was released by the model developer, Ramboll, in May of 2020. This version was a different version than the beta6 version EPA used in its modeling.
Following release of version 7.0, the commenter is correct that an error was identified in the model code that affected model predicted concentrations and, therefore, contributions when the
Continued
khammond on DSKJM1Z7X2PROD with RULES

these areas.30 Therefore, the net change in total anthropogenic emissions across the entire State of Tennessee would be much less than the projected 2 percent increase in NOX emissions.
On October 30, 2020, in the NPRM for the Revised CSAPR Update, EPA
released and accepted public comment on updated 2023 modeling that used a 2016 emissions platform developed under the EPA/Multi-Jurisdictional Organization MJO/state collaborative project.31 On March 15, 2021, EPA
signed the final Revised CSAPR Update using the same modeling released at proposal.32 In this modeling, EPA found that the highest contribution in 2023
from the entire State of Tennessee to any downwind receptor identified as having a nonattainment or maintenance problem for the 2008 ozone standard is projected to be 0.32 ppb. This amount of contribution is well below the 1
percent of the NAAQS threshold used in EPAs good neighbor framework for determining whether an upwind state contributes to a nonattainment or maintenance receptor under the 2008
ozone NAAQS i.e., 0.75 ppb. The small amount of projected increase in NOX emissions in Tennessee as a result of this action, combined with the fact that the highest modeled contributions from this state are well below the 1
percent threshold, support the conclusion that the projected increase in mobile source emissions does not affect EPAs prior decision that Tennessee has no remaining interstate transport obligations under the 2008 ozone NAAQS.
This final action does not make any finding regarding Tennessees interstate transport obligations for the 2015 8-hour ozone NAAQS. EPA has not yet taken final action on Tennessees good neighbor SIP submission for the 2015 8hour ozone NAAQS.
Comment B9: In response to EPAs April 2021 SNPRM, a commenter asserts that EPAs proposed conclusion that removal of the I/M program will not interfere with other states ability to attain and maintain the 2008 ozone NAAQS is based on an air quality modeling-based technique performed for the Revised CSAPR Update that 30 In 2022, emissions of VOC are projected to increase by 740 tons, or a 1.7 percent increase in total anthropogenic VOC emissions. In the context of interstate ozone transport, EPA focuses on NOX
as the key ozone precursor pollutant.
31 See 85 FR 68964, 68981 October 30, 2020.
The results of this modeling are included in a spreadsheet in the dockets for this action. The underlying modeling files are available for public review in the docket for the Revised CSAPR Update EPAHQOAR20200272.
32 See 86 FR 23054 at 23075, 23164 April 30, 2021.

VerDate Sep<11>2014

16:18 Aug 16, 2021

Jkt 253001

PO 00000

Frm 00029

Fmt 4700

Sfmt 4700

model was run using ozone or PM2.5
source apportionment tools SAT for calculating source contributions to pollutant concentrations. Specifically, when CAMX version 7.0 was run with SAT, the pollutant species emissions in the input point source emissions file did not match the species in the core model.
For example, it is possible that the model might have assigned point source emissions of nitric oxide NO in the input emissions file to SO2 in the model run. Thus, the effects on model predictions due to this type of mismatch of pollutant emissions and concentrations would likely be significant. Once this error was identified, it was quickly corrected.33
Further, the code error in version 7.0
did not occur in CAMX model runs when SAT were not invoked i.e., model runs without SA.
EPA contacted Ramboll to determine whether this coding error in the final release of version 7.0 was also present in any of the pre-release beta versions, particularly beta6, which EPA had used for the RCU. Ramboll initially stated that they had no information to determine whether or not the code error was in beta6 or any of the other version 7.0 beta codes. However, in consultation with Ramboll, EPA found that this could be determined by comparing the model predictions from a run without SAT to a companion model run with SAT invoked. If the predictions are the same, then the code EPA used for the RCU did not contain this coding error.
For the RCU, EPA had performed two CAMX model runs for 2023, one without SAT and one with SAT. Thus, to respond to this comment, EPA
compared the ozone predictions from these two runs to identify whether or not there are any notable differences between the two. As an example, Table 5 provides the maximum daily 8-hour ozone predictions for the 2023
emissions case for the month of July at the RCU nonattainment receptor site in Stratford, Connecticut. In addition, Figure 1, provided in the dockets for this final rulemaking, shows the maximum daily 8-hour ozone concentrations in each model grid cell on one of the days in June based on 2023 modeling without and with SAT.34
33 From CAM version 7.10 release notes, January X
5, 2021: Fixed bug that improperly mapped point source species to model species when running SAT.
Implications: Core model point source species were improperly injected affecting core model concentrations and by extension SAT
concentrations. See https camx-wp.azure websites.net/Files/Release_notes.v7.10.txt 34 Figure 1: Model-predicted 2023 maximum daily 8-hour ozone concentrations ppb for June 20
from CAMX v7beta6 model runs without SAT top
E:FRFM17AUR1.SGM

17AUR1