2024-15826. Air Plan Disapproval; Missouri; Interstate Transport of Air Pollution for the 2015 8-Hour Ozone National Ambient Air Quality Standards

Table 1—The EPA's March 2018 Memorandum—Downwind Receptors With Missouri Contributions Above 0.70  ppb

Site (monitor, county, State)	2023 Projected average DV (ppb)	2023 Projected maximum DV (ppb)	Missouri projected contribution (ppb)	Comments
260050003, Allegan, MI	69.0	71.7	2.61	Maintenance receptor.
261630019, Wayne, MI	69.0	71.0	0.92	Maintenance receptor.
484392003, Brazoria, TX	74.0	74.9	0.88	Nonattainment receptor.
482011039, Harris, TX	71.8	73.5	0.88	Nonattainment receptor.
550790085 Milwaukee, WI	71.2	73.0	0.93	Nonattainment receptor.
551170006, Sheboygan, WI	72.8	75.1	1.37	Nonattainment receptor.

Table 2—Missouri Contributions Above 0.70  ppb to Receptors Based on the EPA's 2016 v 2 Modeling

Receptor ID	Location	Nonattainment/maintenance	2023 Projected average design value (ppb)	2023 Projected maximum design value (ppb)	MO projected contribution (ppb)
550590025	Kenosha, Wisconsin	Maintenance	69.2	72.3	1.66
550590019	Kenosha, Wisconsin	Nonattainment	72.8	73.7	1.08
170317002	Cook, Illinois	Maintenance	70.1	73.0	0.94
551010020	Racine, Wisconsin	Nonattainment	71.3	73.2	0.92

Using what it describes as a “weight of evidence” approach, the State analyzed each of the four receptors in table 2 using 2016v2 contribution modeling results and the EPA's August 2018 memorandum described in section I.C.

For the Racine, Wisconsin receptor (Site ID: 551010020), the MoDNR noted that Missouri's projected contribution to this receptor is 0.92 ppb, which is less than 1 ppb. The MoDNR observed that the 1 ppb threshold would capture 67 percent of the total contribution from all upwind States and that the contribution captured by the 1 ppb threshold is 92.23 percent of the amount captured by the 0.70 ppb threshold at this receptor. The MoDNR asserted that the 1 ppb threshold would capture a substantial amount of total upwind States' contribution to ozone concentrations at this receptor, which will lead to meaningful emissions reductions to ensure attainment of the NAAQS at this monitor in 2023. Therefore, the MoDNR relied on a 1 ppb threshold to conclude that its existing SIP sufficiently addresses the good neighbor obligation for the 2015 ozone NAAQS with respect to this receptor.

For the Kenosha-Chiwaukee, Wisconsin receptor (Site ID: 550590025), the MoDNR noted that its projected contribution to this receptor is 1.08 ppb, which is more than 1 ppb. The MoDNR observed that the 1 ppb threshold would capture 86.9 percent of the total upwind contributions and that a 2 ppb threshold would capture 71 percent of the total upwind State contributions. The MoDNR also observed that an alternative 2 ppb threshold would capture 81.7 percent of the upwind State contributions captured under a 1 ppb threshold. Using these data, the MoDNR asserted that a 2 ppb threshold is appropriate because it would capture at least 70 percent of the total upwind State contributions and thus the burden should fall on States other than Missouri ( i.e., only on those States contributing above 2 ppb) to provide emissions reductions that will help ensure attainment of the NAAQS at the site. The MoDNR also asserted that the primary contributors to the projected ozone concentrations at the monitor in Kenosha-Chiwaukee include emissions from Illinois, Indiana, and Wisconsin. The MoDNR cited the EPA's 2016v2 modeling projecting that emissions from these States would contribute a combined 30.79 ppb in 2023 to the Kenosha-Chiwaukee receptor. The MoDNR pointed to the Lake Michigan Air Directors Consortium's (LADCO's) interstate transport modeling results for the 2015 ozone NAAQS to support these claims. The MoDNR asserted that LADCO's analysis indicates that the ozone levels at the Wisconsin shoreline of Lake Michigan are heavily affected by the emissions from Illinois, Indiana, and Wisconsin.^[39]

The MoDNR further pointed out that two other monitoring sites in Wisconsin (Site IDs: 551270006 and 551330027) further inland from Lake Michigan have no projected problems with attaining and maintaining compliance with the 2015 ozone NAAQS. Based on its assessment of this information, the MoDNR concluded that its existing SIP sufficiently addresses its good neighbor obligations for the 2015 ozone NAAQS with respect to the Kenosha-Chiwaukee receptor based only on this Step 2 weight of evidence analysis.

For the Kenosha-Water Tower, Wisconsin Site receptor (Site ID: 550590019) the MoDNR noted that its projected contribution to this receptor is 1.66 ppb, which is more than 1 ppb. The MoDNR observed that the 1 ppb threshold would capture 88.5 percent of the total upwind contributions and that a 2 ppb threshold would capture 71.8 percent of the total upwind State contributions. The MoDNR also observed that an alternative 2 ppb threshold would capture 81.1 percent of the upwind State contributions captured under a 1 ppb threshold. Using these data, the MoDNR asserted that a 2 ppb threshold is appropriate because it would capture at least 70 percent of the total upwind State contributions and thus the burden should fall on States other than Missouri ( i.e., only on those States contributing above 2 ppb) to provide emissions reductions that will help ensure attainment of the NAAQS at the site. The MoDNR then noted that its projected contribution of 1.66 ppb is less than 2 ppb. The MoDNR also asserted that the primary contributors to the projected ozone concentrations at the monitor in Kenosha-Water Tower include emissions from Illinois, Indiana, and Wisconsin. The MoDNR cited the EPA modeling projecting that emissions from these States would contribute a combined 28.47 ppb in 2023 to the Kenosha-Water Tower receptor.

The MoDNR again asserted its interpretation that the LADCO analysis indicates that the ozone levels at the Wisconsin shoreline of Lake Michigan are heavily affected by the emissions from Illinois, Indiana, and Wisconsin.

The MoDNR again pointed out that two other monitoring sites in Wisconsin (Site IDs: 551270006 and 551330027) further inland from Lake Michigan have no projected problems with attaining and maintaining compliance with the 2015 ozone NAAQS. The MoDNR concluded that the nonattainment receptor at Kenosha-Water Tower is heavily influenced by local transport emissions and lake breeze effects over Lake Michigan. Based on its assessment of this information, the MoDNR concluded that its existing SIP sufficiently addresses the good neighbor obligation for the 2015 ozone NAAQS with respect to the Kenosha-Water Tower receptor based only on this Step 2 weight of evidence analysis.

For the Cook County Chicago-Evanston, Illinois receptor (Site ID: 170317002), the MoDNR noted that its projected contribution to this receptor is 0.94 ppb, which is less than 1 ppb. The MoDNR observed that the 1 ppb threshold would capture 69.7 percent of the total contribution from all upwind States and that the contribution captured by the 1 ppb threshold is 92.9 percent of the amount captured by the 0.70 ppb threshold at this receptor. The MoDNR asserted that the 1 ppb threshold would capture a substantial amount of total upwind States' contribution to ozone concentrations at this receptor, thus the burden should fall on States other than Missouri ( i.e., only on those States contributing above 1 ppb) to provide meaningful emissions reductions to ensure attainment of the NAAQS at this site. Therefore, the MoDNR relied on a 1 ppb threshold to conclude that its existing SIP sufficiently addresses the good neighbor obligation for the 2015 ozone NAAQS with respect to this receptor.

The MoDNR further pointed out that the other nine receptors in Cook County, Illinois show 40 percent less impact from Missouri in the 2016v2 modeling than the County Chicago-Evanston, Illinois receptor. The MoDNR concluded that this difference indicates uncertainty on any conclusion that emissions from Missouri are contributing significantly to this single monitor in Cook County, while at the same time not contributing significantly to any other monitor in Cook County.^[40]

The MoDNR cited a 2019 Lake Michigan Ozone Study (LMOS or the study) relating to high ozone monitor concentrations near Lake Michigan. According to the MoDNR, the study recommends a finer grid resolution to better characterize ozone concentrations near large bodies of water. The MoDNR interprets the study as showing that upwind States' NO_X emissions may have little to no impact on ground level ozone concentrations that are linked to downwind receptors because on high ozone level days the ozone concentrations in these areas are sensitive to emissions of VOCs and not NO_X . The MoDNR also contends that, based on information included in the EPA's “Air Quality Modeling for the 2016v2 Emissions Platform Modeling TSD” and the receptor specific data included in the EPA's document titled “CAMx 2016v2 MDA8 O3 Model Performance Stats by Site,” the EPA's 2016v2 modeling is “severely underperforming” in this region of the country where all of Missouri's linked receptors in the EPA's 2016v2 modeling are located.^[41]

For these reasons the MoDNR concludes that the weight of evidence analyses provided for these receptors in the November 2022 Submission shows that Missouri's current SIP is adequately addressing its good neighbor obligations with respect to each of these four receptors.^[42] In its conclusion of Steps 1 and 2, the MoDNR has stated it believes Missouri's good neighbor obligations are met at Steps 1 and 2 for the 2015 ozone NAAQS. However, in its submission, the MoDNR goes on to acknowledge there are uncertainties due to model performance and recent year NO_X emissions from Missouri EGUs that exceed the assurance levels of the CSAPR NO_X Ozone Season Trading Program. Consequently, the MoDNR developed a Step 3 analysis to address these uncertainties ( i.e., proceeding from an assumption that Missouri is linked at Step 2) and conducted an analysis of potential emissions control opportunities to ensure that the “good neighbor obligations are indeed satisfied.” ^[43]

2. Information Provided at Step 3

In Step 3, the MoDNR begins by identifying controls that could be implemented at units for the 2023 ozone season. The MoDNR's evaluation at Step 3 gives some consideration to NO_X emitting sources at certain EGUs, as well as certain sources in certain non-EGU sectors, including cement and cement products manufacturing, glass and glass products manufacturing, and pipeline transportation of natural gas industries.

For 2023, the MoDNR made a list of 10 coal-fired EGUs currently equipped with Selective Catalytic Reduction (SCR) in the State. The MoDNR observed that four of these units have Prevention of Significant Deterioration permits requiring continuous operation of their NO_X control equipment (Hawthorn unit 5A, Iatan units 1 and 2, and John Twitty Energy Center unit 2). Based on the existence of these permits, the MoDNR stated that no additional NO_X control requirements would be cost-effective for these units.^[44] The MoDNR observed that the six other units do not currently have enforceable requirements to ensure the continuous operation of their control equipment ( i.e., SCRs) during the ozone season. The MoDNR claims that, based on its assessment, substantial and timely emissions reductions are both available and cost-effective at five units and that the sixth unit could choose to operate their controls less efficiently or not at all.^[45]

The MoDNR also observes that there are two EGUs in Missouri that have Selective Non-Catalytic Reduction (SNCR) but do not currently have enforceable requirements to ensure the continuous operation of their control equipment during ozone season. The MoDNR determined that substantial and timely emissions reductions are both available and cost effective at these two units.

The MoDNR observed that there are nine remaining coal-fired EGUs that have no SCR or SNCR. At the time of submission, four of these units at two facilities (Rush Island and Meramec) are expected to retire by 2026, which is the next applicable attainment date under the 2015 ozone NAAQS following the 2023 deadline for Moderate areas.^[46] The MoDNR also stated that the other five units (Labadie and Sikeston) are expected to continue operating at least through 2026. The MoDNR claims that no timely reductions ( i.e., by 2023) are available for any of these nine units.^[47] The MoDNR determined, however, that new enforceable requirements at the Labadie and Sikeston facilities to ensure the continued operation of the existing NO_X controls ( i.e., low NO_X burners, etc.) would help guard against potential backsliding and lock in the emissions reductions these facilities have already achieved.

To determine the new enforceable requirements at the coal-fired EGUs currently equipped with SCR (New Madrid, Thomas Hill facilities and John Twitty Unit 1), the MoDNR first analyzed historical NO_X emissions rates for each unit with SCR to determine what the starting point for a new limit should be. The MoDNR selected what it identified as the third best ozone season emissions rate for five of the units at these facilities, as the MoDNR determined those years to be reflective of continuous SCR operation. Next, the MoDNR took the average of what it identified as the third best emissions rates for these units, which was 0.102 lbs/mmBtu. The MoDNR then decided that a compliance margin of approximately 20 percent was appropriate for this limit, putting the limit at 0.12 lbs/mmBtu. The MoDNR claims that this numeric limit combined with a stipulation that the sources continuously run their controls during the ozone season were enough to ensure meaningful reductions were achieved.^[48]

The MoDNR attempted to use the same analysis for the units with SNCR, however the units only had one year of full operation using the control equipment. The MoDNR selected the larger emissions rate of the two units for that year and again added the 20 percent compliance margin to arrive at a numeric NO_X ozone season emissions rate of 0.18 lbs/mmBtu.^[49] The MoDNR also used this historical emissions rate analysis with the 20 percent compliance margin to establish the limits for the units without SCR and SNCR to prevent backsliding, arriving at a 0.12 lbs/mmBtu rate for one facility and 0.13 lbs/mmBtu rate for a second facility.^[50] The MoDNR acknowledged that for these units without existing post-combustion controls, this rate would not achieve new emissions reductions, but rather avoid potential increases in emissions.

The MoDNR estimated that these limits, combined (0.12 lbs/mmBtu on coal-fired units with existing SCR, and 0.18 lbs/mmBtu on coal-fired units with existing SNCR, and anti-backsliding limits on coal-fired units without SCR or SNCR), would achieve ozone season NO_X reductions of 6,713 tons annually compared to emissions levels at these units in 2021, assuming these units operate at the limited rate.^[51]

The MoDNR then used these new emissions reductions it claims would be achieved through these new NO_X emissions limits to evaluate the impact to linked downwind receptors. The MoDNR used the EPA's Ozone AQAT for this analysis.^[52] The MoDNR analyzed the cost of different control strategies for the EGUs discussed above to determine what controls it viewed as cost-effective.^[53] The MoDNR evaluated annual costs for the operation of SCR at the six units with existing SCR described above that do not currently have limits requiring the operation of SCR. This did not include capital cost because the units already have SCR. The MoDNR performed the same analysis for the Sioux facility that is equipped with SNCR. Similar to the SCR units, these annual costs do not include capital costs because the units are already equipped with these controls. The MoDNR then summed these costs and divided the costs by the modeled improvement in ppb at each linked receptor. The MoDNR identified the cost per 1 ppb improvement at the linked receptors as the “cost effectiveness results.” ^[54]

The MoDNR states that it considered Missouri's modeled contribution to the four linked receptors relative to in-state and other upwind States contributions, the conclusions it drew at Step 2 of their November 2022 Submission, and the costs and the corresponding reduction of ozone concentrations at the four linked receptors resulting from the new NO_X emissions limits at certain facilities.^[55] With consideration of those factors, the MoDNR asserts that the projected emissions reductions in 2023 from the new NO_X emissions limits would fully satisfy Missouri's good neighbor obligation for the 2015 ozone NAAQS.^[56]

The MoDNR also reviewed the cost of new post-combustion controls included in the proposed Good Neighbor Plan (87 FR 20036, April 6, 2022) for 2026. In the proposed Good Neighbor Plan, the additional emissions reductions required for EGUs in 2026 are based primarily on the potential retrofit of additional post-combustion controls for NO_X on most coal-fired EGUs and a portion of oil/gas-fired EGUs that are currently lacking such controls. In the proposed Good Neighbor Plan, the EPA identified SCR retrofits for coal-fired EGUs as part of the strategy to eliminate significant contribution from States linked in the 2026 analytic year. The proposed Good Neighbor Plan also included new emissions limitations for non-EGU point sources including pipeline natural gas transportation, cement and concrete manufacturing, glass and glass product manufacturing, basic chemical manufacturing, petroleum and coal products manufacturing, and pulp, paper, and paperboard manufacturing. Because Missouri was among the States found linked through 2026 in the 2016v2 modeling, the EPA proposed in the Good Neighbor Plan to apply these requirements for non-EGU sources in the State.

In the MoDNR's submission, it acknowledges that the EPA's 2016v2 modeling results for 2026 show that Missouri continues to contribute above 1 percent of the NAAQS in 2026 to one nonattainment receptor, but less than 1 ppb, and also continues to contribute above 1 percent of the NAAQS in 2026 to three maintenance receptors: to two maintenance receptors above 1 percent of the NAAQS but less than 1 ppb and to one maintenance receptor above 1 ppb. However, the MoDNR asserts that the Kenosha, Wisconsin, receptor (Site ID: 550590025) will no longer be a maintenance receptor in 2026 due to emissions reductions under the Good Neighbor Plan and the new NO_X limits that would be implemented by Missouri if the November 2022 Submission is approved by the EPA.^[57] The MoDNR concluded that Missouri has no remaining good neighbor obligations for the 2015 ozone NAAQS in 2026 because Missouri is projected to contribute less than 1 ppb to the other three receptors before considering the impact of the EPA's proposed Good Neighbor Plan in 26 States.^[58]

The MoDNR further claims that the cost-effectiveness of the retrofit of SCR at the Labadie facility and Sikeston facility using the expected remaining life of those units is more than ten times the average cost-effectiveness of the State's planned controls associated with better operation of controls at EGUs with existing SCR and SNCR. The MoDNR then divided the cost by the modeled improvement in ppb at each of the three receptors to which Missouri contributes above 1 ppb in 2023. The MoDNR used those values and cost-effectiveness to calculate the cost effectiveness of the SCR retrofits for each of the three receptors in maintenance or nonattainment in 2026.^[59] The MoDNR asserted that SCR retrofits at existing coal-fired EGUs are not cost-effective, nor required for the purpose of satisfying Missouri's good neighbor obligations under the 2015 ozone NAAQS.^[60]

The MoDNR also purported to analyze the cost of controls included in the proposed Good Neighbor Plan for non-EGUs. According to the MoDNR, the EPA's proposed Good Neighbor Plan included proposed requirements for preheater/precalciner kilns at a limit of 2.8 lbs. NO_X /ton of clinker produced. The MoDNR indicated that the State's rule, 10 CSR 10-6.380 Control of NO_X Emissions from Portland Cement Kilns, covers the cement kilns identified by the EPA in the “Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026” included in Docket ID No. EPA-HQ-OAR-2021-0668-0191.^[61] The MoDNR further asserts that the State's rule includes a more stringent requirement of 2.7 lbs NO_X /ton of clinker produced during the regulatory ozone season (May—September). The MoDNR concluded that no cost-effective emissions reductions in this source category in Missouri are available.^[62]

The MoDNR also observed that the proposed Good Neighbor Plan included new control requirements in the glass and glass product manufacturing industry. Two facilities in Missouri were listed in the “Screening Assessment of Potential Emissions Reductions, Air Quality Impacts, and Costs from Non-EGU Emissions Units for 2026” as potentially subject to these proposed requirements: Pittsburg Corning Corporation in Sedalia, Missouri, and Piramal Glass USA Inc. in Park Hills, MO. The MoDNR addressed these two facilities by first identifying that the Piramal plant in Park Hills, Missouri, was expected to close in March of 2022.^[63] The MoDNR then observed that the EPA estimated a total cost of $5.8 million for the NO_X controls at the Pittsburg Corning Corporation in the proposed Good Neighbor Plan. The MoDNR further notes that the 2020 and 2021 NO_X emissions at this facility were 17 and 44 tons, respectively. The MoDNR concluded from this that emissions reductions for the glass manufacturing sector in Missouri are not cost effective and that no further requirements are needed under this source category to address Missouri's good neighbor obligations under the 2015 ozone NAAQS.^[64]

The MoDNR observed that the EPA identified four pipeline natural gas transportation facilities in Missouri as potentially subject to new controls in the proposed Good Neighbor Plan. The EPA projected the average annual cost per ton NO_X reduced for Missouri as $5,452 for this industry category. The MoDNR calculated cost effectiveness values in terms of annual dollars spent in Missouri per 1 ppb improvement at the remaining three downwind linked monitors using the EPA's estimated cost per ton reduced figure from the proposed Good Neighbor Plan, similar to the analysis the MoDNR performed for the SCR retrofit for EGUs. The MoDNR concludes that emissions reductions for the pipeline natural gas transportation sector in Missouri are not cost-effective and that no further requirements are needed under this source category to address Missouri's good neighbor obligations under the 2015 ozone NAAQS.^[65]

3. Information Provided at Step 4

In Step 4, the MoDNR lists several EGU sources with which the State has developed “Consent Agreements” (Agreements or the Agreements) with NO_X emissions limits based on the MoDNR's Step 3 analysis.^[66] First, the MoDNR explains the requirements for EGUs with SCR. The Agreements require each facility to operate the existing SCR system control devices at least 95 percent of the time during the ozone season when burning coal. The MoDNR asserts that the five percent allowance for non-operation of the SCR is necessary to account for operational issues that SCRs might experience such as catalyst maintenance, plugging issues, and potential supply issues of the SCR reagent.

Next, the MoDNR explains the numeric limits for the controlled units and how the State arrived at the limit of 0.120 pounds per million British Thermal Units (lbs./mmBtu) for the ozone season (see Step 3 discussion detailed earlier in this section). The MoDNR also states that the Agreements contain necessary monitoring, recordkeeping, and reporting to verify compliance with these limits.

Next the MoDNR explains several other additional terms in the Agreements. The terms include provisions for Startup, Shutdown, and Malfunction (SSM). The MoDNR explains what constitutes SSM for each facility and when the hours of operation may be excluded from the requirement to operate the SCR when burning coal at least 95 percent of the time in ozone season. The Agreement for the John Twitty plant defines startup as ending when the unit reaches minimum gross load offered to the Southwest Power Pool (the Regional Transmission Operator) and exempts those hours on the front end. For the New Madrid and Thomas Hill plants, the Agreements provide that they may exclude startup hours following the process in the State SSM rule, 10 CSR 10-6.050. All three facilities are subject to the process in the State SSM rule to exempt hours for shutdown and malfunction when determining compliance with the percent operating time requirement. The MoDNR clarified that the SSM exemptions do not apply to the numeric emissions rate limit of 0.120 lbs./mmBtu.

Next the MoDNR explained that the Agreements also include a “regulatory safety valve” that suspends the numeric emissions rate limits under certain circumstances. The MoDNR explains that the purpose of this provision is to provide regulatory relief in the event that the SCR system could not be operated, but the unit was needed to ensure electric grid reliability/stability. The Agreements have several notification and justification requirements to use this mechanism. The MoDNR asserts that the regulatory safety valve was designed for use only during rare, unexpected grid emergency situations.

In the next section, the MoDNR explains the requirements for EGUs with SNCR. The Agreement for the Sioux Energy Center is designed very similarly to the Agreements the MoDNR made for EGUs with SCR. The MoDNR asserted that the SNCR agreement stipulates a 90 percent operating time requirement as opposed to a 95 percent operating time requirement for the SCR control units, because it was necessary to allow for the weekly tuning procedures for the complementary over-fire air NO_X control system at the Sioux facility. The MoDNR determined that a numeric emissions rate limit for the SNCR controlled units of 0.18 lbs/mmBtu is appropriate based on analysis of historic emissions rates. The MoDNR added similar SSM provisions to the Agreement with the John Twitty facility described above. The MoDNR also included the same regulatory safety valve language for this Agreement as described above.

Next the MoDNR explained the Agreements with Labadie Energy Center and Sikeston Power station. These Agreements include numeric limits of 0.12 lbs/mmBtu and 0.13 lbs/mmBtu respectively. Both facilities are required to continuously operate their currently installed control technologies, generally consisting of combustion-control measures. The MoDNR explains that both Agreements have the same SSM provisions, but they do not include a regulatory safety valve like the Agreements for the SCR and SNCR controlled units because the State determined that they were not necessary.

III. The EPA's Evaluation of Missouri's November 2022 Submission

A. Evaluation of Information Provided by Missouri Regarding Steps 1 and 2

In the November 2022 Submission, the MoDNR provided its interpretation of the 2016v2 modeling results for 2023 and 2026 to eliminate receptors and its linkages identified in the prior modeling released with the March 2018 memorandum, and to identify projected nonattainment and maintenance receptors in 2023 and 2026 as well as Missouri's contributions to them.

The MoDNR utilized a 1 ppb or 2 ppb threshold at Step 2 as it found it needed to reach a conclusion that Missouri was not “linked” to particular downwind nonattainment or maintenance receptors. The EPA had suggested in its August 2018 memorandum that with appropriate additional analysis it may be acceptable for States to use a 1 ppb contribution threshold, instead of the 1 percent of the NAAQS threshold that the EPA has traditionally used, for the purposes of identifying linkages to appropriate downwind receptors, so long as appropriate circumstance-specific information and justification was included.

The MoDNR argued for application of an alternative 1 ppb or 2 ppb threshold, depending on whether Missouri's contributions were below 1 ppb or 2 ppb, by presenting the different numerical percentages of collective contribution that the respective thresholds would capture, and then asserting that the percentages of upwind contribution captured from the 1 ppb or 2 ppb threshold would be sufficiently meaningful. Stated differently, the MoDNR's logic is that so long as the States contributing above these thresholds will shoulder the burden of implementing their own emissions reductions to eliminate significant contribution to the shared receptors, it is appropriate for Missouri to use these thresholds for the purpose of excluding Missouri's emissions sources from having any such obligations. (We note that no other States linked to these receptors included any emissions reductions in their interstate transport SIP submissions for the 2015 ozone NAAQS.)

The EPA proposes to find that the MoDNR did not justify the use of either an alternative contribution threshold of either 1 ppb or 2 ppb. As an initial matter, the MoDNR's theories for use of a 1 ppb or 2 ppb threshold repeat the same arguments that the EPA considered and rejected in acting on Missouri's first submission (June 2019 Submission). See87 FR 9541-43 (February 22, 2022); 88 FR 9358 (February 13, 2023). The EPA is not reopening those determinations; they apply with equal force to the MoDNR's attempts to again justify a higher threshold here; and so are incorporated by reference.

Second, the EPA has, in the SIP Disapproval and Good Neighbor Plan actions carefully evaluated a variety of issues associated with the August 2018 memorandum and potentially recognizing alternative Step 2 contribution thresholds and found in these notice-and-comment rulemakings that, absent an adequate circumstance-specific justification, a 1 percent threshold is the most appropriate threshold for identifying States that “contribute” to downwind ozone receptors for the 2015 ozone NAAQS. 88 FR 9342; 88 FR 36678 (June 5, 2023).^[67] Consistency with past interstate transport actions such as CSAPR, and the CSAPR Update and Revised CSAPR Update rulemakings (which used a Step 2 threshold of 1 percent of the NAAQS for two less stringent ozone NAAQS), is important. Continuing to use a 1 percent of the NAAQS approach ensures that as the NAAQS are revised and made more stringent, an appropriate increase in stringency at Step 2 occurs, to ensure an appropriately larger amount of total upwind-State contribution is captured for purposes of fully addressing interstate transport. See88 FR 9370-72. Accord76 FR 48237-38 (August 8, 2011). In addition, the Agency has explained through both its Disapproval Action and Good Neighbor Plan rulemakings that consistency and equity among States is an important consideration in addressing interstate ozone pollution, which weighs in favor of consistency absent a strong justification otherwise. See 88 FR 9371-75. Larger thresholds such as 1 ppb or 2 ppb would reduce the amount of cumulative upwind State emissions that would be captured, whereas the purpose of the threshold at Step 2 is simply to serve as a de minimis screening threshold, to screen in States for further evaluation of emissions control opportunities, or, stated differently, to screen out States with de minimis contributions from further analysis even if they do have cost-effective emissions reduction potential. See88 FR 9371.

The EPA has not rescinded the August 2018 memorandum, but at the same time, the Agency does not view that memorandum as completely endorsing the use of a threshold higher than 1 percent of the NAAQS. Rather, the memorandum invited State agencies to provide technically sound analytical justifications for use of a 1 ppb or any other threshold based on state-specific circumstances. The MoDNR in this Submission has not advanced new arguments that have not already been considered. The need for consistency in application of a threshold is important. The EPA did not approve the use of a 1 ppb threshold for any State for the 2015 ozone NAAQS. 88 FR 9370-75. Further, it is now clear (to a degree that it may not have been in 2018-2019 when Missouri and other States were developing their SIP submissions) that no States linked even at the higher thresholds the MoDNR asserts are appropriate actually proposed to implement any emissions reductions to benefit these or any other receptors. States' use of the higher thresholds to avoid implementing emissions reductions is contrary to the August 2018 Memo, which provides that “the use of a 1 ppb threshold to identify linked upwind States still provides the potential, at Step 3, for meaningful emission reductions in linked upwind States in order to aid downwind States with attainment and maintenance of the 2015 ozone NAAQS.” ^[68] As the EPA identified in the Disapproval Action, States' reliance on incidental, hypothetical air quality benefits from other contributing States as a basis to justify use of a higher threshold to dismiss their own contribution is improper under the EPA's longstanding approach to evaluating States' obligations on a consistent and equitable basis. This would introduce an inter-dependency into the solution of the “collective contribution” problem that ozone pollution poses and is inconsistent with requiring each State to eliminate its own significant contribution. See Response to Comments at 295-297 in the final Disapproval Action docket (EPA-HQ-OAR-2021-0663). In the proposed and final Good Neighbor Plan, the EPA evaluated all other States linked to these same receptors to which Missouri is linked using the 1 percent of NAAQS threshold. See88 FR 36678. For these reasons, the EPA proposes to find that the 1 percent threshold is appropriate to use to establish whether Missouri's emissions “contribute” to other States' ozone receptors, and the EPA proposes that Missouri's November 2022 Submission did not adequately justify use of any higher thresholds.

In addressing Missouri's linkages in 2026, the EPA notes that the 2016v2 modeling used by the MoDNR indicated four linkages above the 1 percent threshold to nonattainment or maintenance receptors in that year. The EPA finds the MoDNR's dismissal of three of those linkages for being below a 1 ppb threshold is unsatisfactory for the same reasons described above regarding Missouri's linkages in 2023. With respect to Missouri's contribution of 1.53 ppb to the Kenosha-Chiwaukee receptor in 2026, the MoDNR argued that this site will not actually be a receptor, once the emissions reductions in the proposed Good Neighbor Plan are implemented, as well as the emissions controls the MoDNR purports to require in the current November 2022 Submission. This argument is flawed for several reasons. First, the EPA's analysis in the proposed Good Neighbor Plan indicated that the Kenosha-Chiwaukee receptor would remain a maintenance receptor after the implementation of retrofits of post-combustion emissions controls at EGUs in the linked upwind States, including Missouri.^[69] Thus, the available evidence at the time of the MoDNR's submission does not support the stated conclusion. Second, the MoDNR's argument provides no justification why the other States linked to this receptor should be subject to the full stringency of the proposed Good Neighbor Plan (which would include post-combustion control retrofits at EGUs and the non-EGU control measures) while Missouri's sources should enjoy the benefit of only implementing the near-term EGU emissions control strategies relying on existing installed control technologies that the MoDNR purports to require in its November 2022 Submission. (Note that even if the MoDNR's reasoning were applied to the EPA's 2016v3 (rather than 2016v2) modeling and the final (rather than proposed) Good Neighbor Plan (discussed further below), the EPA would reach the same conclusion: Missouri is linked to the Sheboygan receptor through 2026 with a 1.68 ppb contribution, and Sheboygan is projected to remain a receptor through 2026, even with the full implementation of the final Good Neighbor Plan emissions control strategies in all of the upwind States, including Missouri, linked to that receptor through 2026.^[70] )

The MoDNR presents further arguments identical to arguments provided by the MoDNR in public comment on the proposed disapproval of the MoDNR's first submission:^[71] specifically, that due to claimed model underperformance in 2016v2 and other concerns the MoDNR has about modeling ozone near Lake Michigan, the weight of evidence analyses provided for these receptors in the November 2022 Submission shows that Missouri's current SIP is adequately addressing its good neighbor obligations with respect to each of these newly identified four receptors in the 2016v2 modeling.

The EPA addressed the MoDNR's assertions about the LMOS in responding to its comments on the proposed disapproval of Missouri's June 2019 Submission, which were nearly verbatim to the November 2022 Submission on this topic.^[72] The EPA's response to this comment can be found on pages 139 and 152 to 153 of the Response to Comment document supporting the previous final SIP Disapproval Action.^[73] The EPA is not reopening its determinations in the SIP Disapproval Action on this topic, and incorporates that analysis by reference.

As the EPA concluded in the final SIP Disapproval Action, the EPA's 2016v3 modeling is reliable for evaluating good neighbor obligations for the 2015 ozone NAAQS. As stated above, the EPA invited and received comments on the 2016v2 emissions inventories and modeling. In response to these comments, the EPA made a number of updates to the 2016v2 emissions inventories and model design to construct a 2016v3 emissions platform, which was used to update the air quality modeling. Model performance issues noted by the MoDNR in the November 2022 Submission have been addressed in the EPA's 2016v3 modeling. See88 FR 9344-45.

The EPA found that model performance for 2016v3 modeling is improved over the performance for 2016v2 modeling and that the 2016v3 modeling performed well within the range of bias and error performance criteria recommended in the scientific literature, which alleviates the performance concerns in the Midwest asserted by the MoDNR in its November 2022 Submission.^[74 75] The EPA is not reopening these determinations in this action and incorporates its prior analysis by reference.

To be clear, the EPA is not disapproving the November 2022 Submission for using the 2016v2 modeling. Under either the 2016v2 modeling or the 2016v3 modeling, Missouri would be linked to at least one receptor in another State. Therefore, the EPA proposes to find that Missouri is obligated to further evaluate its emissions to determine what portion of its contribution, if any, constitutes “significant” contribution. In fact, the MoDNR conducted such an analysis, as discussed and evaluated further in section III.C.

B. Results of the EPA's Step 1 and Step 2 Modeling and Findings for Missouri

As explained in section I., the EPA is relying on the EPA's 2016v3 modeling and violating-monitor methodology for this action, which is the same set of data the EPA used for all other States in the final SIP Disapproval Action and the final Good Neighbor Plan, including for the State of Missouri.

To summarize what was found in these prior actions for Missouri: based on the EPA's updated 2016v3 air quality modeling and considering contributions to violating-monitor receptors, Missouri is projected to contribute more than 1 percent of the NAAQS ( i.e., 0.70 ppb), 1 ppb, and even 2 ppb, to multiple downwind nonattainment and maintenance receptors in 2023. Specifically, as shown in table 3, Missouri is projected to contribute 1.87 ppb to a nonattainment receptor in Sheboygan County, Wisconsin, (Site ID: 551170006) and 1.87, 1.39, and 1.01 ppb to three maintenance-only receptors, respectively, in Wisconsin and Illinois in the 2023 analytic year. As shown in table 5, Missouri is also projected to contribute above 1 percent of the NAAQS to four violating-monitor receptors at locations in Michigan, and Wisconsin, in the 2023 analytic year. Furthermore, data for 2026 in table 4 indicate that emissions from Missouri will continue to contribute greater than 1 percent of the NAAQS to one maintenance-only receptor in Wisconsin. In addition, Missouri's contribution exceeds 1 ppb at four receptors in 2023 and one receptor in 2026.

Table 3: Projected Missouri Linkage Results Based on the EPA Updated 2023 Modeling

Receptor ID	Location	Nonattainment/maintenance	2023 Average design value (ppb)	2023 Maximum design value (ppb)	Missouri contribution (ppb)
551170006	Sheboygan, Wisconsin	Nonattainment	72.7	73.6	1.87
170317002	Cook, Illinois	Maintenance-Only	68.5	71.3	1.39
551010020	Racine, Wisconsin	Maintenance-Only	69.7	71.5	1.19
550590019	Kenosha, Wisconsin	Maintenance-Only	70.8	71.7	1.01
Source: Final Good Neighbor Plan AQM TSD

Table 4: Projected Missouri Linkage Results Based on the EPA Updated 2026 Modeling

Receptor ID	Location	Nonattainment/maintenance	2026 Average design value (ppb)	2026 Maximum design value (ppb)	Missouri contribution (ppb)
551170006	Sheboygan, Wisconsin	Maintenance-Only	70.8	71.7	1.68
Source: Final Good Neighbor Plan AQM TSD.

Table 5: Missouri 2023 Linkage Results Based on Violating-Monitor Maintenance-Only Receptors

Receptor ID	Location	2021 Design value (ppb)	2022 Design value (ppb)	2021 4th high (ppb)	2022 4th high (ppb)	Missouri modeled contribution (ppb)
261210039	Muskegon, Michigan	74	79	75	82	2.95
260050003	Allegan, Michigan	75	75	78	73	2.18
550890008	Ozaukee, Wisconsin	71	72	72	72	1.64
550590025	Kenosha, Wisconsin	72	73	72	71	1.54
481211032	Denton, Texas	76	77	85	77	0.70
Source: Final Good Neighbor Plan AQM TSD.

C. Evaluation of Information Provided by Missouri Regarding Step 3

To determine what, if any, emissions significantly contribute to nonattainment or interfere with maintenance and, thus, must be eliminated under CAA section 110(a)(2)(D)(i)(I), at Step 3 of the 4-step interstate transport framework, a State's emissions are further evaluated, in light of multiple factors, including air quality, levels of emissions controls, and cost considerations.

To evaluate effectively which emissions in the State should be deemed “significant” and therefore prohibited, States generally should prepare an accounting of sources and other emissions activity for relevant pollutants and assess potential additional emissions reduction opportunities and resulting downwind air quality improvements. The EPA has consistently applied this general approach ( i.e., Step 3 of the 4-step interstate transport framework) when identifying emissions contributions that the Agency has determined to be “significant” (or interfere with maintenance) in each of its prior Federal, regional ozone transport rulemakings, and this interpretation of the statute has been upheld by the Supreme Court. See EME Homer City, 572 U.S. 489, 519 (2014). While the EPA has not directed States that they must conduct a Step 3 analysis in precisely the manner the EPA has done in its prior regional transport rulemakings, State implementation plans addressing the obligations in CAA section 110(a)(2)(D)(i)(I) must prohibit “any source or other type of emissions activity within the State” from emitting air pollutants that will contribute significantly to downwind air quality problems. While the Good Neighbor Plan has defined for purposes of a FIP the obligations for Missouri sources based on extensive modeling, analysis, and technical and policy determinations applied by the EPA, States including Missouri may submit, and the EPA will approve, alternative approaches to defining “significant contribution” that meet the Act's requirement to determine whether and to what degree emissions from a State should be “prohibited” to eliminate emissions that will “contribute significantly to nonattainment” or “interfere with maintenance” of the NAAQS in any other State.

In this section, the EPA evaluates the information provided by the MoDNR in its submission, summarized in section II.B., in support of the conclusions the MoDNR draws at Step 3. Although the MoDNR has stated it believes its good neighbor obligations are met at Steps 1 and 2 for the 2015 ozone NAAQS (statements with which the EPA disagrees as explained in the preceding section), the MoDNR then goes on to state that there are uncertainties due to model performance. The MoDNR also acknowledges that there have been recent exceedances of the assurance levels of the CSAPR NO_X Ozone Season Trading Program by Missouri EGUs. See88 FR 36797-98. Consequently, the MoDNR developed a Step 3 analysis to address these so-called “uncertainties” ( i.e., proceeding from an assumption that Missouri is linked at Step 2) and conducted an analysis of potential emissions control opportunities to ensure that its “good neighbor obligations are indeed satisfied.” ^[76]

After reviewing the MoDNR's Step 3 analysis, the EPA finds several shortfalls in the MoDNR's analysis, including in its evaluation of improved emissions performance opportunities at EGUs with existing post-combustion controls, its evaluation of additional emissions control opportunities at EGUs, and its assessment of non-EGU emissions control opportunities, including its use of a “weighted” approach to identifying significant contribution applying a dollar-per-ppb (or “$/ppb”) metric.

In general, the EPA observes that in the Good Neighbor Plan, the EPA as statutorily required pursuant to CAA section 110(c), and based on extensive record evidence, defined the amount of emissions that constitutes significant contribution from Missouri for purposes of the 2015 ozone NAAQS. In the Good Neighbor Plan, the EPA defined a level of emissions control under Step 3, based on strategies of optimizing existing post-combustion controls and installing or upgrading combustion and post-combustion control equipment on EGU sources as well as installing certain controls on impactful non-EGU sources, and then, at Step 4, established through regulations particular implementation methods to achieve that level of emissions control. The EPA has acknowledged and continues to acknowledge that States are free to develop a SIP to replace a Good Neighbor Plan FIP that adopts a different suite of control measures if they meet good neighbor obligations for the 2015 ozone NAAQS. See88 FR 36838-43 (discussing options for States to replace the FIP with a SIP).

The MoDNR's submission pre-dates the final Good Neighbor Plan; however, in some respects, as discussed further below, the MoDNR modeled aspects of its analysis on the proposed Good Neighbor Plan. As noted further in this section, it is clear that the level of emissions control the MoDNR offers in the November 2022 Submission is less than the level of emissions control the EPA proposed and ultimately found necessary to meet good neighbor obligations in the final Good Neighbor Plan.^[77] As such, the Submission is clearly not “equivalent” in achieving the elimination of an amount of emissions that the EPA determined constituted “significant contribution” in the Good Neighbor Plan. See88 FR 36838-43 (discussing options for States to replace the FIP with a SIP). This is most clearly evident in the Submission's failure to include additional emissions control strategies for EGUs and non-EGUs that the EPA found warranted for those States that remain linked to one or more out-of-state receptors through 2026, which, as noted in the previous section, Missouri is.

The EPA finds that this aspect of the November 2022 Submission runs counter to the guidance the EPA has provided States in the proposed and final Good Neighbor Plan (and prior interstate transport rulemakings dating back to at least 2005) that the FIP establishes an “important benchmark” and that the EPA generally anticipates that SIP submissions that do not meet that benchmark are not likely to be approvable (see discussion in the Executive Summary of this document, section I.A.). However, recognizing that the MoDNR presented alternative arguments as to why its emissions reduction obligations to eliminate “significant contribution” should be less than what was finalized in the Good Neighbor Plan, and these arguments were developed before the EPA's final action issuing the Good Neighbor Plan, the EPA will evaluate additional aspects of the technical, policy, and legal merits of the alternative approaches the MoDNR put forward. In doing so, the EPA will highlight where relevant methodological choices the MoDNR made are not sufficiently technically justified, create inconsistencies or are not reconciled with the good neighbor obligations that the EPA has set for other States linked to shared receptors, and/or otherwise result in a plan submission that does not meet good neighbor obligations for the 2015 ozone NAAQS. In the following subsections, the EPA will evaluate important factors considered by the MoDNR and/or the EPA within the Step 3 multifactor test. These include evaluation of levels of emissions controls on EGUs and non-EGUs in Missouri, potential air quality resulting from these levels of controls, and the MoDNR's use of a “dollar-per-ppb” metric to assess cost effectiveness of controls.

1. Evaluation of Potential Level of Emissions Controls on Missouri EGUs

In Missouri's November 2022 SIP Submission, the controls the State identified to eliminate significant contribution (from all sources in the State) is an “optimized” emissions rate of 0.12 lb/mmBtu for application to certain specifically-named coal-fired EGUs with SCR post-combustion controls and a 0.18 lb/mmBtu rate for coal-fired EGUs with SNCR post-combustion controls already installed. Optimization of existing post-combustion controls on coal-fired EGUs is a well-established strategy that the EPA has recently applied in multiple good neighbor rulemakings, including the CSAPR Update, the Revised CSAPR Update, and the Good Neighbor Plan. However, in the Good Neighbor Plan, the EPA identified that on an ozone-season average, fleetwide basis, sources with existing SCR controls are generally capable of achieving an emissions rate around 0.08 lb/mmBtu. See88 FR 36721.

To understand this discrepancy, the EPA evaluated the level of EGU emissions controls identified by the MoDNR as well as several alternatives based on emissions control opportunities evaluated in the final Good Neighbor Plan. The emissions control opportunities evaluated in the Good Neighbor Plan included optimizing existing SCR and SNCR post-combustion NO_X controls at units that currently have this technology and installing state of the art combustion controls (SOA CC) at units that currently lack them and retrofitting of SCR post-combustion controls at units that currently do not have those controls installed.

The EPA first evaluated the 0.12 lb/mmBtu rate identified by the MoDNR for coal-fired EGUs with SCR. Similar to the methodology described in the EPA's NO_X Mitigation Strategies Final Rule TSD, the EPA focused on the third-lowest ozone season emissions rate for the coal-fired EGUs with SCR systems in Missouri, and calculated a third best average rate, which it determined to be 0.086 lb/mmBtu specific to these units. This value is well below the 0.12 lb/mmBtu rate the MoDNR ultimately determined was appropriate for these units. The EPA then examined the third best average rate; however, the EPA did so for all units across the United States with cyclone boilers with SCR (similar to those in Missouri where the 0.12 lb/mmBtu rate would apply) and found a value of 0.073 lb/mmBtu. This value is also well below the 0.12 lb/mmBtu rate identified by the MoDNR.

Next, going beyond fleetwide average emissions rates, the EPA examined the historical operation of each individual unit for which the MoDNR proposed to apply the 0.12 lb/mmBtu rate to identify whether there was a justifiable reason for selecting an “optimized” rate well above the identified third best average rates (either using the sources included in the EPA's evaluation, or the subset of sources in Missouri that the MoDNR used). The EPA found that essentially all units have achieved emissions rates well below 0.12 lb/mmBtu, and in almost all cases below at least 0.08 lb/mmBtu on a monthly or seasonal basis.^[78] Thus, in general, the emissions rate the MoDNR identified for these sources is roughly 33 percent less stringent than appears to be achievable based on the relevant data.

To investigate the source of this discrepancy further, the EPA revisited its assessment of all coal-fired EGUs with SCR for potential optimization completed for the Good Neighbor Plan. The EPA examined costs for full operation of SCR controls for units that already have this technology installed.^[79] This includes the cost of catalyst replacement and disposal, the costs of reagent, and the cost for returning a partially operating SCR to full operation. The EPA evaluated nationwide coal-fired EGU NO_X ozone season emissions data from 2009 through 2021 and calculated an average NO_X ozone season emissions rate across the fleet of coal-fired EGUs with SCR for each of these thirteen years and considered the third best average emissions rate. The EPA did not consider the lowest or second-lowest NO_X emissions rates since these may reflect SCR systems that have all new components and therefore not representative of ongoing achievable NO_X emissions rates considering broken-in components and routine maintenance schedules. The units identified for control under Missouri's SIP submission are included in the subset used in the EPA analysis. The analysis, which includes the costs required to increase reagent and routine maintenance, resulted in an optimized rate of 0.08 lb/mmBtu. Based on the results of this assessment, we still do not find adequate justification for a 0.12 lb/mmBtu rate for these sources.

As discussed previously, the MoDNR initiated its analysis of Step 3 by focusing on improvements of existing SCR operation at certain coal-fired EGUs. The MoDNR follows an assessment similar to those developed by the EPA in evaluating optimization of SCRs at coal-fired EGUs in prior good neighbor rulemakings. In the development of its analysis to identify an emissions rate, however, the MoDNR excludes the best operating coal-fired units with SCRs in Missouri before considering the third best average emissions rate. In addition, in establishing the emissions rates in the Consent Agreements, the MoDNR added a 20 percent compliance margin, but the MoDNR's November 2022 Submission does not contain any analysis or data supporting this approach to allowing increased emissions far beyond the selected rate. Additionally, the Consent Agreements already include provisions addressing instances of variability from startup, shutdown, and malfunction of equipment. Thus, the identified emissions rate is inconsistent with a review of historical data for the subset of units the MoDNR considered for controls in the SIP Submission. The EPA's analysis of the data at the sources at which Missouri's November 2022 Submission proposes to apply the 0.12 lb/mmBtu limit shows that historical NO_X emissions rates ranging from 0.06-0.10 lb/mmBtu are attainable for these units.^[80] Allowing increases above historical emissions rates by an additional 20-100 percent does not reflect an optimized emissions performance level for these coal-fired EGUs with SCR and in fact allows for degradation in emissions performance from observed achievable historical rates. The MoDNR provided no cost or feasibility information regarding why these rates that were previously achieved at these EGUs are not attainable at these EGUs going forward.

The MoDNR's analysis of near-term emissions control opportunities at EGUs was incomplete in certain other respects. The EPA agrees that Missouri EGUs do not have SOA CC potential upgrades available at this time.^[81] However, the MoDNR did not evaluate additional SCR optimization at oil and/or gas fired EGUs. The EPA's analysis suggests that there may be additional emissions reductions through cost-effective optimization of combined cycle units in Missouri available in the short-term, which the MoDNR did not consider.^[82] Finally, the Consent Agreements for Labadie and Sikeston include ozone season emissions rates of 0.12 lb/mmBtu and 0.13 lb/mmBtu, respectively. Historical ozone season emissions rates for Sikeston for the previous five years ranged from 0.10-0.12 lb/mmBtu and for Labadie was 0.09 lb/mmBtu for the same period.^[83] While the MoDNR acknowledges in its submission that these rates would not necessarily achieve any reductions in emissions, instead serving as an assurance of the operation of controls at the units in these facilities, the emissions rates in the Consent Agreements actually represent an increase of their historical and consistently lower emissions rates.

Without adequate technical justification for applying the rates selected and included in the Consent Agreements, and with no additional emissions controls identified to eliminate a comparable amount of significant contribution in compensation for that reduced level of emissions control, the November 2022 Submission fails to eliminate the amount of emissions that the EPA has found achievable through near-term EGU control strategies and which comprises a portion of the amount of emissions that should be prohibited to eliminate significant contribution for the 2015 ozone NAAQS (as well as for the 2008 ozone NAAQS previously).

However, the EPA conducted further evaluation of the consequences of the application of this rate with respect to estimated total emissions reductions in the State and the air quality effects of those reductions at downwind receptors.

The EPA applied the emissions rates identified by the MoDNR as well as the emissions rates the Agency identified in the final Good Neighbor Plan to the suite of units each agency identified and calculated the resulting ozone season State-level EGU emissions for the State (including the emissions from units whose emissions rates remained unchanged) at various levels of stringency based on 2021 emissions.^[84] The resulting state-wide ozone season emissions levels for various levels of stringency based on the EGU fleet in 2023 and 2026 is shown in table 6. For 2023, both the emissions levels under the MoDNR`s emissions controls and the final Good Neighbor Plan represent a reduction in emissions from the base case. However, under the final Good Neighbor Plan an additional 784 tons are reduced beyond the emissions reductions identified by the MoDNR.

The difference is even more pronounced in 2026, with additional reductions accruing from emissions reductions commensurate with the installation of SCR controls under the Good Neighbor Plan. Such controls could achieve 4,518 additional tons of ozone-season emissions reductions, as under the final Good Neighbor Plan, beyond the level of reduction specified in the November 2022 Submission.

Table 6—2023 and 2026 EGU Ozone Season NO _X Emissions (Tons) at Various Levels of Stringency

Year	Engineering analysis (EA) base case	Missouri consent Agreement NO X limits (0.12-0.18 lb/mmBtu on existing units)	SNCR and SCR optimized (existing SCRs optimized at 0.08 lb/mmBtu rate)	New SCRs added and SCR optimized (existing SCRs and optimized to 0.08 lb/mmBtu rate)
2023	20,094	13,382	12,598	(not applicable)
2026	18,612	11,899	11,116	7,381

The EPA notes that these figures do not account for roughly an additional 2,065 tons of projected ozone-season NO_X emissions reductions that were identified in the Good Neighbor Plan associated with non-EGU emissions control strategies to eliminate significant contribution from the State of Missouri. The EPA addresses the non-EGU analysis provided in the Missouri SIP submission in section III.C.4.

Based on this evaluation, the EPA determines the information provided in the November 2022 Submission regarding the potential cost-effective emissions control strategies at EGU sources in Missouri is inadequate.

2. Evaluation of Projected NO_X Reductions on Downwind Linked Receptors

The effects of emissions control strategies on downwind receptors to which upwind States are linked is one of the important factors that States and the EPA typically assess at Step 3 for purposes of defining the amount of “significant contribution.” Further, while the EPA does not view achieving precisely the same degree of projected air quality improvement at receptors as its FIP achieves as necessarily required for a SIP to be approvable, the EPA considers the Good Neighbor Plan's evaluation of air quality improvement to supply an important benchmark that allows for a reasonable assessment of the sufficiency of alternative programs States may put forward. The MoDNR included data in their submission showing the projected effect of its chosen emissions control strategies on ozone levels at the downwind receptors. While this data indicates that the MoDNR's proposed approach achieves some improvement in ozone levels at the identified receptors to which Missouri is linked, this amount of air quality improvement is less than the degree of improvement that occurs with the application of the level of emissions control that the EPA had determined in the Good Neighbor Plan is necessary to eliminate Missouri's significant contribution. This finding supports the EPA's proposed conclusion that the November 2022 Submission is not adequate to address Missouri's good neighbor obligations for the 2015 ozone NAAQS.

The EPA evaluated the air quality analysis conducted by the MoDNR that used the Air Quality Assessment Tool, or AQAT, from the proposed Good Neighbor Plan. The MoDNR utilized the AQAT to estimate the air quality effects of Missouri's proposed emissions reductions on the receptors that the MoDNR identified as potential receptors in 2023. The EPA independently checked the emissions reductions projected to be achieved at the level of emissions control identified by the MoDNR and then using the same AQAT (from the proposed Good Neighbor Plan) confirmed that the MoDNR's results are accurate.

Next, since the final Good Neighbor Plan had updates to the air quality modeling that form the basis of the AQAT, we repeated the air quality analysis using the air quality modeling and AQAT from that rule. Starting with the 2021 emissions from the engineering analysis used in the final Good Neighbor Plan and applying the emissions rates identified by the MoDNR, we found slightly different emissions levels for 2023. We also found emissions levels for 2026, which included emissions changes due to retirements and/or other changes (see the Ozone Transport Policy Analysis Final Rule TSD from the final Good Neighbor Plan for details; also available in this docket). In addition to the emissions control level identified by the MoDNR, to ensure a thorough review, we evaluated several other emissions levels using the engineering analysis from the final Good Neighbor Plan ( i.e., optimization levels for existing SCR controls of 0.08 lb/mmBtu (or better) as well as optimization of existing SNCR controls).^[85] Additionally, we evaluated the potential effects of installing SCR controls with rates of 0.05 lb/mmBtu on all coal-fired units that are greater than 100 MW. The resulting analysis allows for a comparison of reduction in ozone levels that the MoDNR's emissions reductions strategies for EGUs would achieve at downwind receptors as compared to what the final Good Neighbor Plan's emissions reduction strategies would achieve. The comparative estimated air quality contributions to each of the potential receptors resulting from these emissions stringency levels can be found in table 7.

We observe that the emissions reductions proposed by the MoDNR result in some air quality improvements for each receptor. However, we observe that for the stringency levels identified and selected to eliminate significant contribution in the final Good Neighbor Plan, there are more emissions reductions and that these emissions reductions result in more air quality improvements (through reductions in the amount of ozone to which Missouri's emissions contribute) at the downwind receptors relative to the stringency level that has been adopted by the MoDNR in its November 2022 Submission. Based on an extensive record, the EPA had found the emissions reductions called for in the Good Neighbor Plan broadly cost-effective on industry-wide bases through national-scale analysis in light of the large geographic scale and persistent nature of the interstate ozone transport problem for the 2015 ozone NAAQS. As discussed here and throughout section III.C., the MoDNR has not put forward an adequate technical justification explaining why Missouri's EGU and non-EGU sources should be subject to a substantially less stringent emissions control program delivering proportionately less air quality benefits to receptors in other States.

Table 7—2026 Air Quality Contributions (ppb) Estimated Using AQAT to Each of the Receptors to Which Missouri Contributes Greater Than or Equal to 0.70 ppb at Various Levels of Emissions Reductions ⁸⁶

Receptor ID #	State	County	AQ Model Base Case	Engineering Analysis (EA) Base Case	SNCR and SCR optimized to proposed Missouri NO X limits a	SNCR and SCR optimized b	New SCRs added and SCR optimized c	New SCRs added and SCR optimized with Non-EGU reductions d e
170317002	Illinois	Cook	1.258	1.316	1.242	1.233	1.192	1.169
550590019	Wisconsin	Kenosha	0.912	0.955	0.900	0.894	0.863	0.847
551010020	Wisconsin	Racine	1.061	1.125	1.043	1.033	0.988	0.963
551170006	Wisconsin	Sheboygan	1.689	1.778	1.664	1.650	1.587	1.552
260050003	Michigan	Allegan	1.969	2.068	1.942	1.927	1.857	1.818
261210039	Michigan	Muskegon	2.649	2.685	2.638	2.633	2.607	2.593
481211032	Texas	Denton	0.634	0.661	0.661	0.661	0.661	0.661
550590025	Wisconsin	Kenosha	1.381	1.449	1.362	1.352	1.303	1.276
550890008	Wisconsin	Ozaukee	1.479	1.551	1.459	1.448	1.397	1.369
a  Proposed limit is 0.12 lb/mmBtu on existing units.
b  Existing SCRs optimized at 0.08 lb/mmBtu rate.
c  Existing SCRs optimized to 0.08 lb/mmBtu rate.
d  Existing SCRs optimized to 0.08 lb/mmBtu rate.
e  This is the stringency set by the Good Neighbor Plan.