2024-23341. Air Quality State Implementation Plans; Partial Approval, Partial Disapproval and Promulgation; Texas; Regional Haze

Table 1—Class I Areas Included in AOI Analyses of the 2021 Texas Regional Haze Plan

Site	Code	State	County	Latitude	Longitude
Big Bend National Park	BIBE1	TX	48043	29.3027	−103.178
Breton Island	BRIS1	LA	22075	30.10863	−89.76168
Caney Creek	CACR1	AR	05113	34.4544	−94.1429
Great Sand Dunes	GRSA1	CO	08003	37.7249	−105.5185
Guadalupe Mountains National Park	GUMO	TX	48109	31.833	−104.8094
Hercules-Glades	HEG1	MO	29213	36.6138	−92.9221
Mingo	MING1	MO	29207	36.9717	−90.1432
( print page 83349)
Rocky Mountain National Park	ROMO1	CO	08069	40.2783	−105.5457
Salt Creek	SACR1	NM	35005	33.4598	−104.4042
Upper Buffalo Wilderness	UPBO1	AR	05101	35.8258	−93.203
Wheeler Peak	WHPE1	NM	35055	36.5854	−105.42
White Mountain	WHIT1	NM	35027	33.4687	−105.5349
Wichita Mountains	WIMO1	OK	40031	34.7323	−98.713

As explained above, the EPA concluded in the 1999 RHR that “all [s]tates contain sources whose emissions are reasonably anticipated to contribute to regional haze in a Class I area,” and this determination was not changed in the 2017 RHR.^[50] Critically, the statute and regulation both require that the cause-or-contribute assessment consider all emissions of visibility impairing pollutants from a State, as opposed to emissions of a particular pollutant or emissions from a certain set of sources. Consistent with these requirements, the 2019 Guidance makes it clear that “all types of anthropogenic sources are to be included in the determination” of whether a state's emissions are reasonably anticipated to result in any visibility impairment.^[51]

While Texas identified Class I areas within and outside of the State that are potentially impacted by Texas sources, Texas did not conduct an AOI analysis for the Bosque del Apache Class I area.^[52] Texas justifies this decision based on “past SIP and FIP documentation” but provides no additional context or explanation of why that decision remains appropriate for this planning period.^[53] In contrast, Texas's CAMx PSAT ^[54] modeling identified Bosque del Apache as having impacts from Texas sources. According to Texas's PSAT modeling, Texas sources contribute over seven percent of the total visibility impairment at Bosque del Apache.^[55] Specifically, the 2021 Texas Regional Haze Plan identifies that the influence due to particulate sulfate from Texas sources is more than five times the influence of New Mexico sources, and the influence due to particulate nitrate from Texas sources is nearly twice the influence of New Mexico sources.^[56] Thus, Texas's PSAT modeling suggests that emissions from Texas sources are reasonably anticipated to contribute to visibility impairment at the Bosque del Apache Class I area given the low threshold for visibility impact on Class I areas discussed previously.^[57] Therefore, Texas did not complete its obligation under 40 CFR 51.308(f), which provides that each state's plan “must address regional haze in each mandatory Class I Federal area located within the State and in each mandatory Class I Federal area located outside the State that may be affected by emissions from within the State,” and (f)(2), which requires each state's plan to include a long-term strategy that addresses regional haze in such Class I areas.

D. Calculations of Baseline, Current, and Natural Visibility Conditions; Progress to Date; and the Uniform Rate of Progress

Section 51.308(f)(1) requires states to determine the following for “each mandatory Class I Federal area located within the State”: baseline visibility conditions for the most impaired and clearest days, natural visibility conditions for the most impaired and clearest days, progress to date for the most impaired and clearest days, the differences between current visibility conditions and natural visibility conditions, and the URP. This section also provides the option for states to propose adjustments to the URP line for a Class I area to account for visibility impacts from anthropogenic sources outside the United States and/or the impacts from wildland prescribed fires that were conducted for certain, specified objectives.^[58]

In Chapter 4 of the 2021 Texas Regional Haze Plan, Texas determines and presents the baseline, natural, and current visibility conditions for both the 20 percent most anthropogenically impaired days and the 20 percent clearest days for the State's two Class I Areas consistent with the EPA's RHR and guidance. In the 2021 Texas Regional Haze Plan, the TCEQ used visibility data from IMPROVE monitoring sites to calculate baseline visibility conditions. Consistent with the RHR, Texas calculated baseline visibility based on data from 2000-2004. For Big Bend specifically, baseline visibility conditions are based on valid data for 2001 through 2004 because 2000 did not meet completeness criteria.^[59] Baseline visibility indices for Big Bend and Guadalupe Mountains are presented in the 2021 Texas Regional Haze Plan in table 4-4. In our review, we identified that the information provided by Texas in Chapter 4 of its 2021 Regional Haze Plan as to the baseline and current conditions on the 20 percent clearest days is inconsistent with the IMPROVE monitoring data and information presented in Chapter 8. Based on the information in table 8-42 of the 2021 Regional Haze Plan, Texas identifies the correct data set for where this information is located but presents the incorrect data in Chapter 4. Based on the data source that Texas identified in Chapter 8, we present information in tables 2 and 4 consistent with information in Chapter 8 of its Plan and the IMPROVE monitoring data.^[60]

( print page 83350)

Table 2—Estimate of Baseline Visibility Conditions (2000-2004) for Class I Areas in Texas

Class I area	Most impaired haze index (dv)	Clearest haze index (dv)
Big Bend	15.57	5.78
Guadalupe Mountains	14.60	5.92

Using the revised IMPROVE algorithm ^[61] and the methodology described in the 2018 Visibility Tracking Guidance, the TCEQ determined natural visibility conditions for Big Bend and Guadalupe Mountains, presented in table 4-3 of the 2021 Texas Regional Haze Plan, and included in the following table 3.

Table 3—Estimate of Natural Visibility Conditions for Class I Areas in Texas

Class I area	Most impaired haze index (dv)	Clearest haze index (dv)
Big Bend	5.33	1.62
Guadalupe Mountains	4.83	0.99

The current visibility conditions, which are based on 2014-2018 monitoring data, are presented in the 2021 Texas Regional Haze Plan in table 4-5 with corrected values included in the following table 4.

Table 4—Estimate of Current Visibility Conditions (2014-2018) for Class I Areas in Texas

Class I area	Most impaired haze index (dv)	Clearest haze index (dv)
Big Bend	14.06	5.17
Guadalupe Mountains	12.64	4.73

While the 2021 Texas Regional Haze Plan does not specifically present the differences between current visibility conditions and natural visibility conditions as well as the progress to date, we include these calculations using the corrected information in tables 5 and 6.

Table 5—Progress to Date

(Differences Between Baseline and Current Conditions)

Class I area	Most impaired (dv)	Clearest haze (dv)
Big Bend	1.51	0.61
Guadalupe Mountains	1.96	1.19

Table 6—Differences Between Current and Natural Conditions

Class I area	Most impaired (dv)	Clearest haze (dv)
Big Bend	8.73	3.55
Guadalupe Mountains	7.81	3.74

The Regional Haze Rule allows states the option to adjust the 2064 glidepath endpoint to account for both international anthropogenic and certain prescribed fire impacts at Class I areas. In the EPA's September 2019 Availability of Modeling Data and Associated Technical Support Document for the EPA's Updated 2028 Visibility Air Quality Modeling memorandum ^[62] (EPA 2019 Modeling TSD), the EPA used 2028 modeling results to quantify the international and prescribed fire impacts at Class I areas on the 20% most anthropogenically impaired days. Texas used its own CAMx modeling results to adjust the URP to account for international anthropogenic emissions consistent with the approach used by the EPA in the TSD associated with the EPA's Updated 2028 Visibility Air Quality Modeling memorandum. Texas's adjusted URP for Big Bend and Guadalupe Mountains are presented in Figures 8-28 and 8-29 of its 2021 Texas ( print page 83351) Regional Haze Plan.^[63] Texas's adjusted URP in 2028 on the 20% most impaired visibility days is 14.38 deciviews for Big Bend and 12.81 for Guadalupe Mountains.^[64] These values for Big Bend and Guadalupe Mountains are within the range of 2028 adjusted glidepath values provided for in the EPA 2019 Modeling TSD.^[65]

The EPA finds that the visibility condition calculations for the two Texas Class I Areas meet the requirements of 40 CFR 51.308(f)(1). Therefore, the EPA proposes to approve the portions of the 2021 Texas Regional Haze Plan relating to 40 CFR 51.308(f)(1).

E. Long-Term Strategy for Regional Haze

Each State having a Class I area within its borders or emissions that may affect visibility in a Class I area must develop a long-term strategy for making reasonable progress towards the national visibility goal.^[66] As explained in the Background section of this notice, reasonable progress is achieved when all states contributing to visibility impairment in a Class I area are implementing the measures determined—through application of the four statutory factors to sources of visibility impairing pollutants—to be necessary to make reasonable progress.^[67] Each state's long-term strategy must include the enforceable emission limitations, compliance schedules, and other measures that are necessary to make reasonable progress.^[68] All new ( i.e., additional) measures that are the outcome of four-factor analyses are necessary to make reasonable progress and must be in the long-term strategy. If the outcome of a four-factor analysis and other measures necessary to make reasonable progress is that no new measures are reasonable for a source, that source's existing measures are necessary to make reasonable progress, unless the State can demonstrate that the source will continue to implement those measures and will not increase its emission rate. Existing measures that are necessary to make reasonable progress must also be in the long-term strategy. In developing its long-term strategies, a State must also consider the five additional factors in § 51.308(f)(2)(iv). As part of its reasonable progress determinations, the State must describe the criteria used to determine which sources or group of sources were evaluated ( i.e., subjected to four-factor analysis) for the second implementation period and how the four factors were taken into consideration in selecting the emission reduction measures for inclusion in the long-term strategy.^[69]

1. Source Selection

a. Overview of Texas's Source Selection

Under 40 CFR 51.308(f)(2)(i), states must evaluate and determine the emission reduction measures that are necessary to make reasonable progress by considering the costs of compliance, the time necessary for compliance, the energy and non-air quality environmental impacts of compliance, and the remaining useful life of any potentially affected anthropogenic source of visibility impairment.^[70] In doing so, states should consider evaluating major and minor stationary sources or groups of sources, mobile sources, and area sources as part of their long-term strategy for regional haze. Furthermore, the State must include in its implementation plan a description of the criteria it used to determine which sources or groups of sources it evaluated. States may rely on technical information developed by the RPOs of which they are members to select sources for four-factor analysis and to conduct that analysis, as well as to satisfy the documentation requirements under 40 CFR 51.308(f)(2). Texas, however, conducted its own analysis separate from CenSARA's analysis to select sources for further evaluation using the four statutory factors.

Texas focused on sources of NO_X and SO₂ emissions in its control strategy analysis for the second planning period. Texas explained these are the main anthropogenic pollutants that affect visibility at Class I areas in Texas and Class I areas in neighboring states. Texas further stated that, “on an individual basis, point sources are the largest contributors to SO₂ and NO_X ,” and thus Texas elected to focus on point sources in this planning period.^[71]

Texas's source selection methodology relied on a two-step approach. As the first step for source selection, Texas developed areas of influence (AOIs) for thirteen ^[72] Class I areas (in Texas and nearby states) to identify areas that may contain sources of NO_X and SO₂ that were expected to contribute to visibility impairment at these areas. The AOIs are graphical representations of the extinction weighted residence time (EWRT), which combines air flow patterns with ammonium sulfate and ammonium nitrate extinction measured at IMPROVE monitors at the Class I areas on the 20% most impaired days. The TCEQ used the AOI of a Class I area as a brightline cutoff to define the boundaries within which to further evaluate sources located within that area. As the second step, Texas then applied a Q/d threshold for NO_X and for SO₂ of greater than or equal to five to point sources located within the geographical area of the selected AOI threshold.^[73] As a result, any source within the AOI boundaries with a Q/d less than five or any source, regardless of its Q/d, that fell outside of the AOI boundaries were eliminated from further consideration.

Although Texas determined AOIs for 13 Class I areas in Texas and nearby states, Texas's 2021 Regional Haze Plan focused only on those Class I areas where sources with a Q/d greater than or equal to five fell within the AOI boundary.^[74] Following this methodology, Texas selected 18 sources for further analysis for only four Class I areas: Wichita Mountains, Caney Creek, Guadalupe Mountains, and Salt Creek.^[75]

( print page 83352)

Table 7—Texas' s Source Selection for Its 2021 Regional Haze Plan ⁷⁶

Company/site name	Unit(s)	Class I area(s)	Pollutant(s)
Coleto Creek Power/Coleto Creek Power Station	(1) coal boiler	Wichita Mountains	SO 2 .
Southwestern Electric Power/Welsh Power Plant	(2) coal boilers	Caney Creek & Wichita Mountains	SO 2 .
AEP/Pirkey Power Plant	(1) coal boiler	Caney Creek & Wichita Mountains	SO 2 .
NRG Energy/Limestone Electric Generating Station	(2) coal boilers	Wichita Mountains	SO 2 .
Vistra Energy/Martin Lake Electric Station	(3) coal boilers	Caney Creek & Wichita Mountains	SO 2 .
San Miguel Electric Cooperative/San Miguel Elec. Plant	(1) coal boiler	Guadalupe Mountains & Wichita Mountains	SO 2 .
Public Service Co. of Oklahoma/Oklaunion Power Station	(1) coal boiler	Wichita Mountains	SO 2 & NO X .
Vistra Energy/Oak Grove Steam Electric Station	(2) coal boilers	Wichita Mountains	SO 2 .
Holcim Texas LP/Midlothian Plant	(2) cement kilns	Wichita Mountains	SO 2 .
Vitro Flat Glass/Works No. 4 Wichita Falls Plant	(2) glass melting furnaces	Wichita Mountains	SO 2 & NO X .
Graphic Packaging International/Texarkana Mill	(4) boilers: (2) black liquor solids & NG; (1) NG & fuel oil; (1) NG, fuel oil, & other materials	Caney Creek	NO X .
El Paso Natural Gas Co./Keystone Compressor Station	(15) reciprocating engines	Guadalupe Mountains & Salt Creek	NO X .
El Paso Natural Gas Co./Cornudas Plant	(6) turbines	Guadalupe Mountains	NO X .
El Paso Natural Gas Co./Guadalupe Compressor Station	(1) turbine	Guadalupe Mountains	NO X .
GCC Permian/Odessa Cement Plant	(2) cement kilns	Guadalupe Mountains	NO X .
Orion Engineered Carbons/Orange Carbon Black Plant	(1) incinerator; (4) dryers; (2) tail gas and NG boilers; (1) flare	Caney Creek	SO 2 .
Oxbow Calcining/Oxbow Calcining-Port Arthur	(4) coke calcining kilns	Caney Creek	SO 2 .
Trinity Lightweight Aggregate (TRNLWS)/Streetman Plant	(1) lightweight aggregate kiln	Wichita Mountains	SO 2 .

b. EPA's Evaluation of Texas's Source Selection Methodology

In identifying the required emission limits, schedules of compliance, and other measures as may be necessary to make reasonable progress toward meeting the national goal, States first select sources for consideration of the four statutory factors.^[77] Under the RHR, States have flexibility in conducting their source selection; however, Texas's source selection methodology was neither well-reasoned nor adequately justified.^[78] Notably, Texas did not select any sources for further analysis of control measures that may be necessary for inclusion as part of the long-term strategy to make reasonable progress for Big Bend National Park and did not select any SO₂ sources for consideration for Salt Creek. Moreover, the EPA finds the source selection methodology used by Texas was not adequately or accurately described. As such, the threshold Texas applied to define its AOIs was not justified. Without the proper justification, it is unclear how, despite these deficiencies, Texas makes reasonable progress at these Class I areas.

i. The TCEQ Failed To Adequately Describe the Criteria It Used To Select Sources

Under 40 CFR 51.308(f)(2)(i), States are required to include a “description of the criteria it used to determine which sources or groups of sources it evaluated.” Based on our review of the 2021 Texas Regional Haze Plan, the methodology Texas described in its SIP to develop its AOIs is inconsistent with, and would not result in, the AOIs presented in Texas's SIP. Texas states in its SIP that the AOIs were determined by dividing the EWRT for each cell by the sum total of all the EWRTs ( i.e., EWRT for each cell) across the entire domain.^[79] However, based on the documentation the EPA obtained during early engagement in the Fall of 2020 and comparing it to what was in its 2021 Regional Haze Plan, Texas actually divided the EWRT for each cell by the maximum EWRT in the domain for each respective pollutant. There was thus an inconsistency between what Texas said its methodology was, and what was in its 2021 Regional Haze Plan submission. Specifically, in the 2020 early engagement document, Texas stated, “. . . prior to plotting the AOIs, the weighted probabilities were scaled to 1 by dividing the weighted probabilities in each cell by the maximum value in

( print page 83353)

a cell in the domain.” ^[80] The EPA compared the plotted AOIs Texas had submitted during the 2020 early engagement period with the plotted AOIs Texas submitted with its 2021 Regional Haze Plan. These AOIs are the same, confirming that, despite what Texas stated in its 2021 Regional Haze Plan, Texas was actually following its articulated methodology in the 2020 early engagement document.

This early engagement information was not included in the proposed SIP Texas published during its state-level notice-and-comment process. Thus, Texas's SIP failed to accurately or adequately describe the criteria actually used in its 2021 Regional Haze Plan submission to determine which sources, or groups of sources, it chose to evaluate for additional control measures as required by 40 CFR 51.308(f)(2)(i). Without an accurate and adequate description of Texas's source selection methodology, it is not clear from its 2021 Texas Regional Haze Plan how Texas evaluated and determined the emission reduction measures that are necessary to make reasonable progress for its second planning period long-term strategy. We discuss the AOI methodologies and these inconsistencies further in the Technical Support Document (TSD) included in the docket for this action.

ii. Texas Failed To Adequately Justify Its AOI Threshold

As noted in the previous section and more fully explained in the EPA's TSD, Texas selected sources using AOIs it developed for each Class I area then followed with a Q/d analysis. The AOIs established a brightline geographic boundary within which Texas selected sources with a Q/d of greater than or equal to five. In other words, Texas did not consider a source, regardless of the size of its emissions, if it was not first within the geographic area defined by the chosen AOI threshold.

To define the brightline geographic boundaries of the AOIs, Texas applied a threshold of 0.1 or 10% of the maximum EWRT value for that AOI.^[81] Texas did not provide any discussion or justification for its selection of this threshold, nor did Texas explain how this threshold resulted in evaluating a meaningful set of sources for possible controls measures to improve visibility impairment. Further, Texas did not evaluate whether the selected threshold provided for AOIs that included a sufficiently large area to capture the sources with the highest emissions, or Q/d values, that impact visibility at certain Class I areas. The need for a justification is crucial when a State is applying the threshold as a brightline when selecting sources to evaluate for additional control measures, such as what Texas did here. The AOIs generated from EWRTs represent the general location that air parcels are coming from when visibility extinction is high. However, unless an appropriate threshold value is applied, they do not necessarily capture the specific sources of emissions that are contributing to visibility impairment at the Class I area.^[82] Texas's approach did not consider the size or location of point sources, despite articulating a specific focus on point sources,^[83] or the total emissions captured to support that their approach and chosen threshold resulted in a reasonable identification of sources for analysis in development of the long term strategy. This problem is evident in Texas's 2021 Regional Haze Plan, where several AOIs contained no sources identified for further consideration and several large emission sources with Q/d values far exceeding Texas's Q/d threshold of five being excluded from further consideration because they were located outside of Texas's generated AOIs.

For example, W A Parish is located just outside of Texas's ammonium sulfate AOIs for both Caney Creek and Wichita Mountains, and outside of Texas's ammonium sulfate AOI for Big Bend.^[84] The SO₂ Q/d values for W A Parish are 32.2 for Caney Creek, 28.2 for Wichita Mountains, and 25.1 for Big Bend.^[85] Tolk Generating Station is also located outside of Texas's ammonium sulfate AOI for Salt Creek; however, it has a Q/d value of over 84.^[86] Ammonium sulfate is the largest contributor to observed light extinction at Salt Creek ^[87] but Texas did not identify any source of SO₂ emissions for further analysis due to the application of their AOI brightline test and selected EWRT threshold, despite the large SO₂ emissions from Tolk and the relative proximity of the facility to Salt Creek.^[88] Given the large emissions from these facilities, these sources likely are meaningfully contributing to visibility impairment, even if they happen to fall outside of the chosen Texas AOIs. Based on its analysis of other coal-fired EGUs with no controls or underperforming controls, had Texas selected these sources for further evaluation under the four factors, Texas may have found cost-effective controls available, resulting in emission reductions that may have been necessary for inclusion in its long-term strategy to make reasonable progress toward meeting the national goal. Moreover, Texas did not explain how not evaluating these high-emitting sources nonetheless results in a long-term strategy that makes reasonable progress toward the national goal.

We therefore find Texas's unjustified use of its selected threshold and resulting AOIs as a brightline cutoff in ( print page 83354) source selection to be unreasonable. Texas's methodology resulted in several of the highest emitting SO₂ stationary point sources in the State of Texas not being selected for further evaluation of controls to improve visibility impairment at the Class I areas they likely impact, and in the case of some Class I areas, no sources selected at all for further analysis using the four statutory factors for those areas.

iii. PSAT Modeling Results Further Demonstrate Unreasonableness of Texas's Source Selection Methodology

The 2019 Guidance identifies photochemical modeling and the use of source apportionment modeling as possible methods to assess PM species impacts from sources or groups of sources for source selection.^[89] Texas conducted photochemical source apportionment modeling (known as the Particulate Matter Source Apportionment Technology, or PSAT, function of CAMx modeling) as part of its 2021 Regional Haze Plan to evaluate the impact of emissions from source categories on visibility in Class I areas.^[90] While Texas did not conduct PSAT modeling for the explicit purpose of source selection, Texas nevertheless included the results of the PSAT modeling in its SIP.^[91] The EPA finds Texas's own PSAT modeling results illustrate the flaws in Texas's source selection methodology.

The TCEQ failed to address in its 2021 Regional Haze Plan how its source selection approach and resulting failure to select sources for further analysis to address visibility impairment at Big Bend are consistent with the CAA's statutory goal and Regional Haze Rule requirements.^[92] TCEQ's source selection methodology did not identify any sources for further analysis of control measures that may be necessary to include in its long-term strategy to make reasonable progress at Big Bend. The TCEQ's PSAT model results indicate that emissions from Texas anthropogenic sources account for over 10% of the total light extinction at Big Bend, and 67% of the light extinction due to U.S. anthropogenic emissions.^[93] The influence from Texas sources on light extinction at Big Bend is approximately double the influence from anthropogenic sources in the rest of the U.S. combined.^[94] While Texas states that visibility at Big Bend is heavily influenced by international emissions, the TCEQ has already accounted for this by adjusting the glidepath for its Class I areas to remove visibility impairment from international emissions, consistent with the EPA's guidance, and thus should not be used as a rationale for not evaluating sources for additional control measures. CAA 169A(a)(1), (b)(2) and the RHR require states to make reasonable progress towards addressing anthropogenic impairment from U.S. sources in the second planning period in furtherance of Congress's national goal.

The influence of Texas sources on sulfate and nitrate concentrations at Big Bend shows that emissions from Texas sources are projected to account for approximately 65.4% of the particulate sulfate concentration and 59.3% of the nitrate concentration due to U.S. anthropogenic emissions.^[95] The vast majority (93.9%) ^[96] of the Texas influence on particulate sulfate concentrations at Big Bend can be attributed to Texas anthropogenic emissions from electricity generating unit (EGU) point and non-EGU point sources.^[97] Therefore, these data demonstrate that Texas's AOI analysis and threshold selection for Big Bend did not adequately identify the relevant sources that impact visibility impairment for further analysis necessary to develop a long-term strategy to make reasonable progress at Big Bend.

Similarly, Texas's PSAT modeling also underscores inadequacies with its source selection for Class I areas in New Mexico, for example, Salt Creek. As noted above, Texas's AOI analysis for Salt Creek identified no sources of SO₂ in Texas for consideration for further analysis. However, the results of Texas's PSAT modeling show that Texas sources account for almost 12% of the light extinction at Salt Creek.^[98] The largest contributor to light extinction at Salt Creek is sulfate.^[99] Focusing on modeled U.S. anthropogenic impacts alone, Texas anthropogenic sources account for approximately 51.3% of the particulate sulfate concentrations at Salt Creek.^[100] Texas's chosen approach for source selection failed to identify any SO₂ point sources, despite accounting for over half of all the U.S. anthropogenic particulate sulfate concentrations at Salt Creek.

Class I areas like Salt Creek that are not projected to be on or under the glidepath are subject to additional requirements in the RHR. Under 40 CFR 51.308(f)(3)(ii)(B), Texas must provide a robust demonstration that there are no additional emission reduction measures for anthropogenic sources or groups of sources in the State that may reasonably be anticipated to contribute to visibility impairment in the Class I area that would be reasonable to include in its own long-term strategy.^[101] The influence from Texas's point sources on particulate sulfate concentrations at Salt Creek is more than double the amount of New Mexico's total (point source, non-point source, and mobile source) influence on particulate sulfate concentrations at Salt Creek.^[102] Meaning, SO₂ emissions from Texas sources contribute more to visibility impairment at Salt Creek than SO₂ emissions from New Mexico sources. Given the meaningful contribution to visibility impairment demonstrated by its PSAT modeling, Texas's decision not to select any SO₂ sources for further analysis and consideration of the four statutory factors (or to adequately justify the decision not to select these sources) fails to satisfy the requirement to provide for a robust demonstration for those Class I areas projected to be above the glidepath, as required by 40 CFR 51.308(f)(3)(ii)(B).

iv. EPA's Conclusions and Proposed Action on Source Selection

The EPA finds the source selection methodology used by Texas was not adequately described as required by the RHR.^[103] Nevertheless, the EPA was able to discern the state's approach to ( print page 83355) developing its AOIs which relied upon drawing a boundary based on a threshold of ten percent of the maximum EWRT values. Texas, however, did not provide any rationale or justification for this ten percent threshold. The boundaries of the AOIs were used as a brightline cutoff, with sources outside the AOIs not given any further consideration. As demonstrated in previous sections, Texas's methodology was unreasonable because it resulted in the selection of no sources for further evaluation at Big Bend and no SO₂ sources for further analysis at Salt Creek. Texas's own PSAT modeling results confirm that its methodology was unreasonable because the results show significant contribution from Texas anthropogenic sources to visibility impairment at Big Bend and Salt Creek. Texas made no attempt to explain the disconnect between its PSAT results and its source selection approach.

The selection of a reasonable set of sources is a necessary first step in identifying the required emission limits, schedules of compliance, and other measures as may be necessary for inclusion in its long-term strategy to make reasonable progress toward meeting Congress's goal of preventing any future, and remedying any existing, impairment at Class I areas after consideration of the four statutory factors.^[104] It is evident that developing a long term strategy to make reasonable progress cannot be met, if no sources of pollutants shown to be meaningful contributors to impairment are selected for further evaluation. It is further evident that, at least for Big Bend for both NO_X and SO₂ and for Salt Creek for SO₂, Texas's method of establishing an AOI is not adequate to identify sources of visibility impairment in Texas.

Therefore, the EPA is proposing to disapprove the portion of Texas's 2021 Regional Haze Plan addressing the regulatory requirements of the long-term strategy under 40 CFR 51.308(f)(2).

2. Four Factor Analysis

This section discusses the technical bases and information Texas relied on in the evaluation of emission reduction measures necessary to make reasonable progress in each Class I area affected by emissions from Texas when developing its long-term strategy for the second planning period. As discussed in the preceding section, Texas selected 18 sources for evaluation of emissions reductions necessary to make reasonable progress.^[105] If a source triggered analysis for both NO_X and SO₂ , control strategies for both pollutants were analyzed separately and concurrently.^[106] Of the 18 sources selected for evaluation, eight are EGU sources and 10 are non-EGU sources.

Based on the statutory and regulatory requirements, Texas evaluated emission reduction measures that are necessary to make reasonable progress by considering the four statutory factors listed in CAA § 169A(g)(1) and 40 CFR 51.308(f)(2)(i) for these selected sources. The four statutory factors are (1) the cost of compliance; (2) the time necessary for compliance; (3) the energy and non-air quality environmental impacts of compliance; and (4) the remaining useful life of any potentially affected sources. This is commonly referred to as “the four-factor analysis.” The four statutory factors must be considered when evaluating and determining the emissions reductions measures that are necessary to make reasonable progress.^[107] Although visibility impact is not one of the factors required for consideration under the CAA and the RHR, Texas opted to evaluate and consider the visibility benefits from selected control measures evaluated in the four-factor analysis by conducting photochemical sensitivity modeling.^[108] In the subsections that follow, we discuss Texas's analysis of the four statutory factors.

a. Identification of Potential Controls

In accordance with EPA's 2019 Guidance, “the first step in characterizing control measures for a source is the identification of technically feasible control measures for those pollutants that contribute to visibility impairment.” ^[109] The EPA's 2019 Guidance does not define the term “technically feasible;” however, EPA's Regional Haze Regulations and Guidelines for Best Available Retrofit Technology (BART) Determinations (the BART Guidelines) states:

Control technologies are technically feasible if either (1) they have been installed and operated successfully for the type of source under review under similar conditions, or (2) the technology could be applied to the source under review. Two key concepts are important in determining whether a technology could be applied: “availability” and “applicability.” . . . a technology is considered “available” if the source owner may obtain it through commercial channels, or it is otherwise available within the common sense meaning of the term. An available technology is “applicable” if it can reasonably be installed and operated on the source type under consideration. A technology that is available and applicable is technically feasible.^[110]

A reasonable four-factor analysis will consider the full range of potentially reasonable options for reducing emissions.^[111] In order to provide guidance on what control measures should be included in their four-factor analysis, the RHR Guidance lists examples of different types of control measures that states may consider.^[112]

For EGUs without existing controls, Texas considered and evaluated dry sorbent injection (DSI), spray dryer absorber (SDA), and wet limestone scrubbing systems (wet FGD) as potential SO₂ control options, and selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR) as potential NOx controls.^[113] For EGUs with existing SO₂ controls, Texas considered and evaluated upgrading the control efficiency of the controls to 95%.^[114] For non-EGUs, Texas considered various NO_X and SO₂ control options depending on the type of source and whether it had existing controls.^[115]

For selected sources where Texas could not identify any feasible control options for a particular source-type, that particular source and pollutant was not further evaluated in the four-factor analysis. Texas stated that it only considered control technologies that have been demonstrated as technically feasible for units at each source type and evaluated those control technologies using available unit-specific data. Texas deemed a given control technology to be “demonstrated to be technically feasible” if it was identified in the EPA's Reasonably Available Control Technology/Best Available Control Technology/Lowest Achievable Emission Rate (RACT/BACT/LAER) Clearinghouse or operated in industrial applications for units within an industry type not in a performance “trial” phase.^[116] Texas further explained that a control measure or technique that has been established as technically demonstrated or feasible ( print page 83356) in one industry type was not considered to extend automatically to other industry types. Based on Texas's approach, Texas determined that there were no technically feasible controls for three of the 18 sources selected for further evaluation using the four factors: the Orion Carbon Black facility in Orange County, the Oxbow Calcining facility in Jefferson County, and the Streetman facility in Navarro County. These three determinations are discussed in more detail in the following paragraphs.

Initially we note that Texas's search for available controls relied primarily on the RACT/BACT/LAER Clearinghouse. BACT and LAER are terms associated with EPA's “New Source Review” (NSR) permitting program and is triggered when a company is planning to build a new plant or modify an existing plant such that air pollution emissions will increase by a large amount. EPA established the RACT/BACT/LAER Clearinghouse to provide a central data base of air pollution technology information (including past RACT, BACT, and LAER decisions contained in NSR permits) to promote the sharing of information among permitting agencies and to aid in future case-by-case determinations.^[117] We note that many of the petroleum coke calcining plants and carbon black plants were constructed prior to the start of EPA's NSR permitting program and have generally not been modified in ways that would trigger the permitting programs.^[118] As a result, Texas's reliance on that RACT/BACT/LAER Clearinghouse is not a sufficient search for these types of facilities.

In fact, several groups commented during Texas's state-level comment period that there were technically feasible controls available for petroleum coke calcining facilities similar to the Oxbow facility. For example, commenters referenced a report which includes a discussion of a petroleum coke calcining plant that currently operates a DSI system to control emissions.^[119] Additionally, the report identifies a Tesoro facility that operates a semi-dry scrubber combined with a wet electrostatic precipitator that reduces SO₂ emissions in excess of 95%.^[120] In response to these comments, Texas stated that:

The control technology the commenter provided may be technically feasible for petroleum coke calcining manufacturing sites but would not necessarily be considered technically demonstrated directly on the kilns such that this technology could be implemented at Oxbow's Port Arthur facility as suggested by the commenter. The possible control options suggested by the commenter would require modification to a site's operational process such that a potential SO₂ post-combustion control strategy could technically be implemented to control SO₂ emissions from petroleum coke calcining kilns. The TCEQ notes these potential strategies would be implemented downstream of the kiln, or kilns, and not directly on the kiln. The operational process modification would require additional process units to the site to make the potential post-combustion SO₂ control measure technically feasible, thereby increasing capital expenditures not directly associated with the new, additional control measure but necessary for the control measure to effectively function and control SO₂ emissions from the petroleum coke calcining kiln. The TCEQ contends these higher-level control analysis approaches require much broader and resource intensive engineering and economic analyses, and they may not result in the potential control strategy being deemed cost-effective or reasonable and necessary for making reasonable progress for long-term strategies for a planning period.^[121]

While Texas's response indicates that such control technologies may not be cost effective based on the modifications that may need to occur at the site, such a determination would necessarily come out of a four-factor analysis; it does not explain why Texas's SIP continued to find that such control measures were not technically feasible.^[122] In fact, it acknowledges that such control technologies may be technically feasible. To the extent Texas is relying on the fact that the costs of this control technology would be prohibitive, Texas needed to provide a cost analysis to document and support such an assumption.^[123]

Furthermore, information provided by Oxbow during Texas's comment period acknowledge that while there is limited publicly available information there are “a few commercially operating post-combustion SO₂ controls systems installed on petroleum coke kilns.” ^[124] Oxbow also provided a four-factor analysis conducted by Sargent & Lundy.^[125] Specifically, Sargent & Lundy concluded that, based on engineering judgment and information from control system vendors, several control technologies were technically feasible and commercially available including: a DSI system with a fabric filter; ^[126] a spray dryer flue gas scrubber system; ^[127] a wet limestone scrubbing system; ^[128] and a circulating dry scrubber system.^[129] Despite information provided to Texas to the contrary, the State continued to find that control technologies were not technically feasible. Therefore, Texas's determination that such control technologies were not technically feasible for petroleum coke calcining facilities was not reasonable. As a result, because Texas selected this source for further evaluation of control measures, it was unreasonable for Texas to not take into consideration the four statutory factors to determine whether there were cost-effective measures that were thus necessary for reasonable progress in fulfillment of their long-term strategy requirements for the second planning period.^[130]

Texas received similar comments regarding Texas's determination that there were no feasible controls for the Orion carbon black plant. Notably, the commenter states that the EPA had entered into consent decrees with several carbon black manufacturing companies that required control of SO₂ emissions to 95%.^[131] In response to these comments, Texas stated that while these consent decrees required certain control efficiencies, installing controls on carbon black facilities had yet to be demonstrated in practice. However, the EPA entered into a consent decree with ( print page 83357) the carbon black manufacturing company Cabot, which required the installation of wet gas scrubbers to control SO₂ emissions from their carbon black units. While the compliance dates were delayed,^[132] Cabot completed construction of the wet gas scrubber at its Canal Plant in 2020.^[133] Thus, the available information identifies technically feasible and available control technologies for carbon black facilities. Therefore, Texas's determination that no control technologies were technically feasible was unreasonable. Texas should have conducted a four-factor analysis for the Orion carbon black plant considering these available controls to determine whether cost-effective control measures were necessary for reasonable progress in fulfillment of its long-term strategy requirements.

Texas also received comments during Texas's state-level public comment period that there were technically feasible controls identified for lightweight aggregate plants like Streetman's plant.^[134] Specifically, commenters referenced EPA's AP-42 emission factor documentation ^[135] for lightweight aggregate manufacturing. Among other information, the document identifies that emissions from kilns at these lightweight aggregate facilities are controlled with wet scrubbers as well as fabric filters and electrostatic precipitators (ESPs). In response to this information, Texas stated that review of the data and information in the EPA's AP-42 emission factor dataset led the TCEQ to conclude that “while wet scrubbers designed for PM control may result in some emissions reductions of SO₂, the TCEQ does not view this as a control strategy for the direct control of SO₂ that could result in meaningful SO₂ emissions reductions.” ^[136]

While the EPA's AP-42 emission factor documentation discusses the use of scrubbers to control PM emissions, it also provides information and emission factors related to the control of SO₂ emissions from the installation of wet scrubbers.^[137] Several of the studies referenced in the documentation were done to measure SO₂ emissions.^[138] This information together shows reductions in emissions of SO₂ from the installation of wet scrubbers at lightweight aggregate plants.^[139] Regardless of whether the main pollutant of concern from these types of facilities is PM or SO₂, Texas does not adequately or reasonably explain how a proven control technology, installed within the same industry type and for which reduces the pollutant of concern (SO₂ ), becomes technically infeasible based on the fact that it also reduces PM. Texas's determination that there were no technically available controls for lightweight aggregate plants such as the Streetman facility was unreasonable and unsupported by information provided to Texas during its public comment period.^[140] Therefore, it was unreasonable for Texas not to have evaluated potential control measures for the Streetman facility using the four statutory factors to determine whether control measures were necessary for reasonable progress in fulfillment of their long-term strategy requirements.

b. Cost of Compliance

Texas evaluated the cost of compliance for each control option determined to be technically feasible for each selected EGU and non-EGU to arrive at an annualized cost and cost per ton of emissions reduced ($/ton), also referred to as a cost-effectiveness calculation, for each control option.^[141] Texas stated that as part of the cost analysis, individual units at a source selected for evaluation with NO_X or SO₂ emissions of less than five percent of the facility's total emissions of the same pollutant were eliminated from further analysis.^[142] Texas explained that excluding such units with smaller emissions is reasonable with respect to application of the cost of compliance criterion because controlling these smaller units would not be justified at this time considering both the cost to control and the anticipated improvement in visibility. Using this approach, Texas focused on the units with relatively greater NO_X and SO₂ emissions at a given source.

In the cost analysis for EGUs without existing controls, Texas stated it estimated the capital cost and annual operating and maintenance costs of technically feasible air pollution control options using the most recent data available from Sargent & Lundy.^[143] In the cost analysis for upgrading scrubbers at EGUs, Texas provided an example cost, but did not explain how that example was used.^[144] In the cost analysis for non-EGUs, Texas stated it estimated the capital cost and annual operating and maintenance costs of technically feasible air pollution control options using cost data and information from the EPA and available industry literature.^[145] For one non-EGU source, the Works No. 4 Glass Plant, Texas relied on vendor cost information for capital cost and annual operating and maintenance costs of control equipment.^[146] For all sources, Texas estimated annualized capital costs by multiplying the capital costs by the capital recovery factor.^[147] The capital recovery factor accounts for source financing of air pollution control equipment and is based on the assumed equipment life and interest rate. Texas stated that “capital recovery factors were estimated using the techniques listed in the EPA's Control Cost Manual” where it found appropriate.^[148] Texas estimated the capital recovery factor assuming an interest rate of 10 percent and an equipment life of five, 15, and 30 years. Ultimately, Texas chose to base its cost analysis on a ( print page 83358) capital life of 15 years for all selected sources.^[149]

Texas stated that annual operating and maintenance costs associated with each control option evaluated “were estimated from the same data and information used for estimating capital costs for each source.” ^[150] Texas added the annualized capital cost and the annual operating and maintenance cost to arrive at the total annualized cost for each control option for each source.^[151] After estimating the potential emission reductions of each control option using baseline emissions from the EPA's 2018 Clean Air Markets Program Data (AMPD) emission data for EGUs and 2016 TCEQ point source emission inventory data for non-EGUs, the total annualized cost was divided by the tons of pollutant emissions reduced to estimate the cost per ton of emissions reduced ($/ton), or cost-effectiveness.^[152] Texas then applied a cost-effectiveness ($/ton) threshold of $5,000/ton for NO_X and SO₂ emissions reduced to eliminate controls from further consideration by explaining that this allowed for the identification of sources to which potential control measures could be applied cost-effectively.^[153] The results of Texas's cost analysis are presented in the following tables.^[154]

Table 8—Texas' s Cost Estimates of SO ₂ Controls for EGUs Without Existing Controls

Source	SO 2 baseline emissions (tons/yr)	Control	Control efficiency (%)	SO 2 reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Coleto Creek Unit 1	13,213	DSI SDA Wet FGD	90 95 98	11,892 12,552 12,949	$3,261 6,720 7,406	$3,022 3,884 4,215	$2,976 3,340 3,603
Welsh Unit 1	7,528	DSI SDA Wet FGD	90 95 98	6,775 7,152 7,377	4,406 11,380 12,032	4,029 6,481 6,812	3,957 5,540 5,811
Welsh Unit 3	6,694	DSI SDA Wet FGD	90 95 98	6,025 6,359 6,560	4,814 12,622 13,357	4,394 7,179 7,558	4,314 6,135 6,445

Table 9—Texas' s Cost Estimates of SO ₂ Wet Scrubber Upgrades for EGUs

Source	Unit size (MW)	SO 2 baseline emissions (tons/yr)	Capital cost ($)	Annual operating and maintenance costs ($)	SO 2 reduction due to scrubber upgrade at 95% control efficiency (tons/yr)	Cost-effectiveness ($/ton)
						5-Year life	15-Year life	30-Year life
AEP Pirkey Unit 1	721	5,085	99,921,030	2,740,188	3,874	$7,511	$4,098	$3,443
Limestone Unit 1	893	4,156	123,757,947	3,393,881	3,212	11,222	6,123	5,145
Limestone Unit 2	957	4,164	132,627,498	3,637,115	3,259	11,853	6,467	5,434
Martin Lake Unit 1	793	19,282	109,899,275	3,013,827	16,172	1,979	1,080	907
Martin Lake Unit 2	793	17,167	109,899,275	3,013,827	14,101	2,270	1,238	1,040
Martin Lake Unit 3	793	19,749	109,899,275	3,013,827	16,458	1,945	1,061	891
San Miguel Unit 1	410	12,006	56,820,558	1,558,221	2,001	8,270	4,512	3,791
Oklaunion Unit 1	720	2,191	99,782,444	2,736,387	1,826	15,913	8,682	7,295

Table 10-Texas' s Cost Estimates of NO _X Controls Oklaunion Unit 1

Source	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Oklaunion Unit 1	6,804	SNCR SCR	50 98	3,402 6,668	$4,705 11,222	$4,152 6,455	$4,046 5,541

Table 11—Texas' s Cost Estimate of SO ₂ Wet Scrubber Upgrades for Midlothian Plant

Unit	SO 2 baseline emissions (tons/yr)	Baseline SO 2 control efficiency of wet scrubber (%)	Capital cost ($)	Annual operating and maintenance costs ($)	SO 2 reduction due to scrubber upgrade at 95% control efficiency (tons/yr)	Cost-effectiveness ($/ton)
						5-Year life	15-Year life	30-Year life
Cement Kiln No 1	522	90	8,196,683	224,782	261	$9,138	$4,986	$4,189
Cement Kiln No 2	856	90	8,300,438	227,627	428	5,647	3,081	2,589

( print page 83359)

Table 12—Texas' s Cost Estimate of Tri-Mer Cat Controls for Vitro Flat Glass Works No 4 Plant

Unit	Pollutant evaluated	Baseline emissions (tons/yr)	Control efficiency evaluated (%)	Emissions reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Glass Melting Furnace Line No.1	SO 2 NO X	136 674	80 80	109 539	$15,100 15,100	$10,300 10,300	$9,400 9,400
Glass Melting Furnace Line No. 2	SO 2 NO X	301 2,533	80 80	241 2,026	4,600 4,600	3,200 3,200	2,900 2,900

Table 13—Texas's Cost Estimates of NO _X Controls for Graphic Packaging Texarkana Mill

Unit	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Power Boiler No 1	109	LNB SCR	40 80	44 87	$21,788 36,200	$10,859 26,350	$8,762 24,469
Power Boiler No 2	692	LNB SCR	40 80	277 554	3,525 7,100	1,757 5,254	1,417 4,956
Recovery Boiler/Furnace No 1	275	LNB SCR	40 80	110 220	7,438 11,800	3,707 9,248	2,991 8,755
Recovery Boiler/Furnace No 2	674	LNB SCR	40 80	270 539	3,619 7,000	1,804 5,395	1,455 5,089

Table 14—Texas' s Cost Estimates of NO _X Controls for El Paso Natural Gas Company Keystone Compressor Station

Unit	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Reciprocating Internal Combustion Engine, A01	131	LEC SCR	40 80	53 105	$1,091 7,956	$544 6,754	$439 6,523
Reciprocating Internal Combustion Engine, A02	7	LEC SCR	40 80	3 6	19,209 129,200	9,573 108,036	7,724 103,974
Reciprocating Internal Combustion Engine, A03	133	LEC SCR	40 80	53 106	1,078 7,900	537 6,677	433 6,449
Reciprocating Internal Combustion Engine, A04	14	LEC SCR	40 80	6 11	9,989 67,500	4,978 56,494	4,017 54,381
Reciprocating Internal Combustion Engine, A05	24	LEC SCR	40 80	10 19	5,964 40,600	2,972 33,990	2,398 32,729
Reciprocating Internal Combustion Engine, A06	17	LEC SCR	40 80	7 13	8,664 58,600	4,318 49,085	3,484 47,253
Reciprocating Internal Combustion Engine, A07	14	LEC SCR	40 80	6 11	10,278 69,400	5,122 58,102	4,133 55,928
Reciprocating Internal Combustion Engine, A08	18	LEC SCR	40 80	12 24	4,851 33,100	2,418 27,769	1,915 26,743
Reciprocating Internal Combustion Engine, A09	16	LEC SCR	40 80	6 13	9,154 61,900	4,562 51,821	3,681 49,885
Reciprocating Internal Combustion Engine, A10	60	LEC SCR	40 80	24 48	2,377 16,600	1,185 13,940	956 13,437
Reciprocating Internal Combustion Engine, A11	34	LEC SCR	40 80	14 27	4,178 28,600	2,083 24,011	1,680 23,127
Reciprocating Internal Combustion Engine, A12	8	LEC SCR	40 80	3 6	18,554 124,800	9,247 104,367	7,461 100,443
Reciprocating Internal Combustion Engine, B01	29	LEC SCR	40 80	12 23	6,727 39,100	3,353 32,227	2,705 30,914
Reciprocating Internal Combustion Engine, B02	83	LEC SCR	40 80	33 66	2,365 14,200	1,179 11,755	951 11,293
Reciprocating Internal Combustion Engine, B03	66	LEC SCR	40 80	26 53	2,958 17,600	1,474 14,543	1,189 13,965

Table 15—Texas' s Cost Estimates of NO _X Controls for El Paso Natural Gas Company Cornudas Plant

Unit	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Gas Turbine, A1	69	LNB SCR	40 80	28 55	$1,913 27,700	$954 21,972	$769 20,879
Gas Turbine, A2	50	LNB SCR	40 80	20 40	5,823 37,742	2,902 29,958	2,341 28,464
Gas Turbine, A3	63	LNB SCR	40 80	25 51	4,623 30,292	2,304 24,112	1,859 22,926
Gas Turbine, B1	104	LNB SCR	40 80	42 83	3,748 22,878	1,868 17,982	1,507 17,042
Gas Turbine, C1	18	SCR	80	14	129,955	101,694	96,270
( print page 83360)
Gas Turbine, C2	18	SCR	80	14	129,955	101,694	96,270

Table 16—Texas' s Cost Estimates of NO _X Controls for El Paso Natural Gas Company Guadalupe Compressor Station

Unit	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Gas Turbine, C-1	56	LNB SCR	40 80	23 45	$13,897 69,485	$6,926 54,975	$5,588 52,190

Table 17—Texas' s Cost Estimates of NO _X Controls for GCC Permian Odessa Cement Plant

Unit	NO X baseline emissions (tons/yr)	Control	Control efficiency (%)	NO X reduction (tons/yr)	5-Year life cost- effectiveness ($/ton)	15-Year life cost- effectiveness ($/ton)	30-Year life cost- effectiveness ($/ton)
Cement Kiln No 2	427	LNB	40	171	$3,163	$1,576	$1,272

i. Texas Did Not Adequately Document the Technical Basis and Cost Information on Which It Based Its Cost of Compliance Analyses as Required by the Regional Haze Rule

Texas did not adequately document the technical basis and cost information on which it based its evaluation of the cost of compliance for all control measures considered as required by the Regional Haze Rule.^[155] The SIP submittal discusses Texas's general approach for estimating the cost of the various control options considered, but only provides sum total estimates of the capital costs and annual operating and maintenance costs without providing individual line items or calculations for review. Texas received comments during the State's public comment period on the proposed Texas RH SIP for the second planning period stating that the proposed SIP did not include proper documentation of the cost estimates of the various control measures, including the actual spreadsheets showing the calculations that inform the results of the cost analyses as part of the TCEQ's four-factor analysis.^[156] Despite these comments, Texas did not directly address why calculation spreadsheets and other necessary documentation of the cost analysis were omitted from the proposed SIP, nor did Texas make changes to the final SIP submittal or include adequate documentation of the cost analysis in the final SIP submittal in response to these comments. With respect to the capital and annual costs of scrubber upgrades, Texas provided one additional piece of information in its response stating that it relied on prior studies and work conducted on potential scrubbing system upgrades to estimate the capital and annual costs to inform total annualized costs.^[157] However, the response does not explain what “prior studies and work conducted on potential scrubbing system upgrades” it relied on or how it relied on those studies to estimate the capital and annual cost of scrubber upgrades. This documentation is critical to ensuring that Texas's consideration of cost of potential control measures, as required by the RHR and the CAA,^[158] was reasonable and based on sufficiently reliable information.^[159]

The EPA has recommended that costs of compliance and the remaining useful life should be calculated consistent with the methods set forth in the EPA's Control Cost Manual in order to allow for comparisons between different sources within a State, and cost analyses in other states.^[160] To that end, states relying on EPA's Control Cost Manual need only reference the manual as the documentation necessary to meet the requirements of the RHR to document the technical basis, including cost information, on which the State is relying.^[161] When a State uses cost methods other than the EPA's Control Cost Manual, it is necessary for those differences to be reasonable and sufficiently documented to meet the requirements of the RHR to document the technical basis, including cost information, on which the State is relying.^[162] In response to comments, Texas acknowledged that it deviated from EPA's Control Cost Manual in certain instances, but failed to provide adequate documentation and justification of its costs in light of its deviations.^[163]

One important element of a cost analysis is the remaining useful life of ( print page 83361) the equipment. This is important because equipment life, while related to the “remaining useful life” factor of the four-factor analysis, also factors into the consideration of cost of compliance due to the annualization of cost in estimating the cost-effectiveness ($/ton reduced). The EPA's 2019 Guidance explains that, generally, states can consider the remaining useful life factor by considering the useful life of the control system.^[164] Typically, the remaining useful life of the source itself will be longer than the useful life of the emission control system under consideration. Thus, annualized costs of compliance are typically based on the useful life of the control equipment rather than the life of the source, unless the source is under an enforceable requirement to cease operation or otherwise reduce its emissions ( i.e., switching from coal to natural gas).^[165]

The Control Cost Manual generally assumes a remaining useful life of equipment of 30 years for scrubbers and SCR.^[166] Texas, however, assumed a remaining useful equipment life of 15 years for all sources.^[167] Texas explained that some of the sources it evaluated in the four-factor analysis could not reasonably be expected to operate an additional 30 years,^[168] but that most could reasonably be expected to continue to operate longer than five years. Therefore, Texas determined that a remaining useful life of 15 years was a reasonable “mid-point” to use in the four-factor analysis. However, Texas did not provide any specific documentation to support its determination that all of the sources it selected could not reasonably expected to operate an additional 30 years nor did it point to any enforceable commitments to retire or otherwise reduce its emissions contained in the SIP. The selection of a 15-year useful life inflates the cost of controls because those costs are amortized over a shorter period of time, thereby increasing the calculated annualized cost and the cost-effectiveness ($/ton reduced). This impacted Texas's identification of cost-effective controls and ultimately, their assessment of aggregate annualized costs. For example, Texas considered SCR as a potential NO_X control for the Texarkana Mill.^[169] Using a 15-year equipment life resulted in an annualized capital cost for SCR on Boiler No. 2 of $853,383 and a cost effectiveness of $5,254 ($/ton).^[170] Using a 30-year equipment life resulted in an annualized capital cost of $688,550 and a cost effectiveness of $4,956 ($/ton).^[171] Because Texas used a cost-effectiveness threshold of $5,000, Texas did not further consider SCR for Boiler No. 2 in determining what measures may be necessary to include in the long-term strategy in order to make reasonable progress.^[172]

Another important element of the cost analysis is the interest rate used. According to the EPA's Control Cost Manual, if a company-specific interest rate is not available for use in evaluating the cost of controls in the four-factor analysis, the use of the current bank prime rate is the appropriate default.^[173] The bank prime rate is reflective of the typical rate for borrowing among large firms. The bank prime rate was 3.25 percent for at least six months leading up to Texas's public comment period,^[174] and remained so when Texas submitted the final SIP to the EPA in July 2021.^[175] Texas instead used a 10 percent interest rate, assuming that industrial sources could not obtain the bank prime rate. However, Texas did not provide any documentation to support this general assertion. In addition, the use of the higher 10 percent interest rate serves to increase the total annualized cost.

Finally, based on Texas's response to comments, Texas included certain costs inconsistent with the “overnight” cost methodology used in the EPA's Control Cost Manual, which resulted in increased costs for the control options considered. However, it is unclear from the information included in Texas's SIP submission, how and for which sources Texas included these costs. In the absence of adequate documentation and justification to support the basis for its cost analysis, we find that Texas's cost analyses are not sufficiently reliable to support its control determinations.

Thus, we find that Texas did not adequately document the technical basis and cost information on which it based its evaluation of the cost of compliance of controls, which is a RHR requirement under 40 CFR 51.308(f)(2)(iii). Without this information, it is unclear how Texas's methodology results in a long-term strategy that includes all measures necessary for reasonable progress in the Second Planning Period.

ii. Texas's Cost Analysis for Scrubber Upgrades Was Unsupported and Unreasonable

Texas's cost analysis of SO₂ scrubber upgrades for EGUs was unreasonable because many assumptions made by Texas in estimating the cost of scrubber upgrades were inadequately justified and based on outlier information that led to unreliable and inflated cost estimates. As explained in the previous section, the 2021 Texas Regional Haze Plan did not document or adequately explain Texas's methodology for estimating the capital costs and operation and maintenance costs of scrubber upgrades, which is a requirement under 40 CFR 51.308(f)(2)(iii).^[176] Rather, the 2021 ( print page 83362) Texas Regional Haze Plan merely provided an “example” which indicates that the average capital cost of wet scrubber upgrades is $37.84/kW and the average operating and maintenance cost is $3.09/kW-year for a 537 MW EGU.^[177] The significance of the example provided in the 2021 Texas Regional Haze Plan is unclear. An examination of the total capital costs included in the 2021 Texas Regional Haze Plan reveals that Texas did not use an assumption of $37.84/kW to estimate capital costs of scrubber upgrades, and in fact used a cost assumption that was over three times higher than the referenced “average” value. To illustrate, Texas estimated the capital cost of scrubber upgrades at AEP Pirkey Unit 1 to be $99,921,030, as shown in table 9. This capital cost estimate is not the equivalent of $37.84/kW, but rather $138.59/kW. In examining the other scrubber upgrades, Texas applied the $138.59/kW to all scrubber upgrade estimates.^[178] Thus, the example provided by Texas indicating that the average capital cost of wet scrubber upgrades is $37.84/kW is misleading and an inaccurate representation of Texas's methodology for estimating the capital cost of wet scrubber upgrades.^[179]

Because Texas did not include adequate documentation of its cost analysis, the EPA requested additional supporting information and data from Texas regarding its technical analyses to aid in our review. In response to this request, the Texas provided additional files to the EPA, including Excel spreadsheets, that were not made available to the public during Texas's public comment period and were not included in the final SIP submitted to the EPA.^[180] One of these files is an Excel spreadsheet that documents and provides additional information on Texas's methodology for estimating the capital costs and annual operating and maintenance costs of scrubber upgrades.^[181] Reviewing the spreadsheet demonstrates Texas's cost methodology relied on certain cost assumptions based on outlier information.^[182]

The files documenting the scrubber upgrades analysis confirm that Texas used an assumption of $139/kW to calculate the capital costs of scrubber upgrades. This $139/kW assumption is the highest capital cost $/kW value out of several scrubber upgrades cost estimates for EGUs compiled from a National Park Service (NPS) spreadsheet from 2010 found on the Western Regional Air Partnership (WRAP) legacy website and relied upon by Texas.^[183] Furthermore, this $139/kW assumption is an outlier value, which corresponds to upgrades at the Coal Creek Power Plant in North Dakota.^[184] The costs for upgrades at this facility included additional project elements other than upgrades to the existing scrubber, such as coal drying.^[185] Texas did not explain why using cost assumptions from a project, which included additional coal pre-treatment project costs like coal drying, is appropriate or reasonable in estimating the capital costs of the scrubber upgrades it was considering. The next highest capital cost $/kW value included in the spreadsheet is an upgrade project that was estimated to cost $52.39/kW.^[186] The average $/kW capital costs provided in the spreadsheet, even including the $139/kW outlier is approximately $38/kW, with costs as low as $4/kW for some units.^[187] Scrubber upgrade costs are site-specific, depending on existing scrubber design and available upgrades.^[188] Therefore, it is inappropriate to rely on cost assumptions that are based on outliers, especially absent any discussion of why the higher cost is more reflective of upgrades necessary at a particular source, because they are not representative of the anticipated cost of scrubber upgrades at these units. Had Texas instead relied on the average capital cost found in the spreadsheet, and presented as the example calculation in its SIP, the capital costs contained in the SIP would have been significantly lower.

To illustrate this point, the EPA recalculated the scrubber upgrade costs for Martin Lake, San Miguel, and Pirkey using the average capital cost ^[189] as well as the average operation and maintenance costs contained in Texas's Excel spreadsheet and identified in their example calculation in appendix B of the 2021 Texas Regional Haze Plan.^[190] EPA focused on these three sources as these were the scrubber upgrades that Texas identified as meeting its cost-effectiveness threshold of $5,000/ton. These recalculated values are found in table 18.

( print page 83363)

Table 18—TCEQ vs. EPA Recalculated Scrubber Upgrade Costs

Company/site name	Unit	2018 EIA electric capacity/2016 EI capacity or engine rating	Capital costs		15 Year life total annual cost		Emissions removed (tpy)	15 year life cost effectiveness ($/ton)
			TCEQ	Avg.	TCEQ	Avg.		TCEQ	Avg.
American Electric Power/Pirkey Power Plant	Unit 1	721 MW	$99,921,030	$27,279,969	$15,877,183	$5,817,383	3,874	$4,098	$1,502
NRG Energy/Limestone Elec. Gen. Station	Unit 1 Unit 2	893 MW 957 MW	123,757,947 132,627,498	33,787,812 36,209,335	19,664,805 21,074,153	7,205,163 7,721,546	3,212 3,259	6,123 6,467	2,244 2,370
Vistra Energy/Martin Lake Electrical Station	Unit 1 Unit 2 Unit 3	793 MW 793MW 793MW	109,899,275 109,899,275 109,899,275	30,004,182 30,004,182 30,004,182	17,462,700 17,462,700 17,462,700	6,398,313 6,398,313 6,398,313	16,172 14,101 16,458	1,080 1,238 1,061	396 454 389
San Miguel Electric Cooperative/San Miguel Electric Plant	Unit 1	410 MW	56,820,558	15,512,881	9,028,634	3,308,081	2,001	4,512	1,653

Based on this information, and utilizing Texas's selected 15-year remaining useful life assumption, the 15-year total annual costs for scrubber upgrades at these three facilities decrease from $77,293,916 to $28,320,403, a reduction in total annual costs of $48,973,513. If the outlier value was excluded in determining the average capital cost, the total annualized costs would be even lower. Thus, the reliance on this outlier value in estimating the capital costs significantly inflates the total annualized costs provided in the 2021 Texas Regional Haze Plan. Without an explanation as to why this was reasonable, this reliance is unjustified. Furthermore, had Texas used the average capital cost, the costs of upgrading the scrubbers at both units at Limestone would have been below its $5,000/ton cost-threshold. Based on Texas's analysis, upgrading the controls on both units at Limestone would result in a reduction in over 6,400 ^[191] tpy of SO_2. The inflation of total annualized costs is also important, as discussed later in the notice, because Texas relies on the combined total annualized costs of control measures in part to determine that no additional measures are necessary to include in its long-term strategy to make reasonable progress.^[192]

We are proposing to find that Texas's cost analysis of SO₂ scrubber upgrades for EGUs does not meet the requirements under 40 CFR 51.308(f)(2)(iii) to document the technical basis, including costs, that the State is relying on to determine the emission reduction measures that are necessary to make reasonable progress. Furthermore, in estimating the cost of scrubber upgrades as part of its four-factor analysis, many assumptions made by Texas were not adequately justified, and thus unreasonable, as it resulted in inflated and unreliable cost estimates. Because of these flaws, we find that Texas did not reasonably consider the cost of compliance as required by the RHR and CAA.^[193]

c. Time Necessary for Compliance

In its 2021 Regional Haze Plan, despite the time necessary for compliance being one of the four statutory factors a State must consider when determining what control measures are necessary for reasonable progress,^[194] Texas stated in its submission that the time necessary for compliance was not a critical factor in the determination of applicable additional controls for Texas sources.^[195] That being said, Texas determined that the time necessary for a source to design, build, and install SO₂ scrubbing technology would be approximately three years and that the time necessary to build and commence operation of DSI technology could be less given that scrubbing vessels would not need to be constructed.^[196] Texas also assumed that the time to design, build, and install NO_X control technologies would be approximately three years. While we are proposing to disapprove Texas's long-term strategy for the reasons provided elsewhere in Section IV.E of this notice, we note that Texas's assumptions of the time necessary for compliance for the controls evaluated are reasonable.

d. Energy and Non-Air Quality Environmental Impacts of Compliance

Where quantifiable for a particular control option, energy impacts of compliance are reflected in the cost estimate and were considered by Texas under the cost of compliance factor.^[197] For instance, electricity costs necessary to operate fans, pumps, and other ancillary equipment as well as waste disposal costs were factored into the cost of compliance calculations for dry and wet scrubbers, DSI systems, SCR systems, and SNCR systems.^[198] Texas stated that control systems that require only modifications to alter fuel-air mixing and combustion temperatures are not expected to have additional electricity or steam demands or to generate wastewater or solid waste.^[199] For reasons explained throughout section IV.E we are proposing to disapprove Texas's long-term strategy.

e. Remaining Useful Life

As we have discussed in detail in section IV.E.2.b. of this proposed rule, we disagree with Texas's generalized assumption of a 15-year equipment life. Without additional discussion explaining why the EGUs and non-EGUs evaluated in the four-factor analysis could not be expected to operate more than 15 years or a federally enforceable commitment to cease operations or otherwise reduce emissions at these units within 15 years, Texas's generalized assumption of a 15-year equipment life is not reasonable and results in the overestimation of the annualized capital costs and the cost-effectiveness of controls.

f. Texas's Control Determinations

After characterizing the four statutory factors, States must consider and weigh the factors to determine what control measures are necessary to include in its long-term strategy to make reasonable progress.^[200] In determining what control measures were necessary to make reasonable progress, Texas weighed the costs of compliance factor and projected visibility benefits of potential controls. Specifically, Texas relied on both the total annualized costs of controls in ( print page 83364) considering the costs of compliance, which it calculated was over $200 million, and the “less than perceptible visibility benefit” it projected in determining that no additional control measures were necessary to include in its long-term strategy to make reasonable progress.^[201]

Texas derived the total annualized cost by adding together the annualized costs of controls at each source that met its $5,000/ton cost effectiveness threshold. Table 19 presents a summary of the estimated total annualized cost of the controls that met Texas's $5,000/ton threshold.^[202]

Table 19 ²⁰³ —Texas' s Potential Control Strategy Summary

Pollutant	Total emissions reductions (tons/yr)	Estimated total annualized cost
NO X	3,171	$9,335,087
SO 2	79,285	195,539,404
Total Costs		204,874,491

In conjunction with total annualized costs, Texas also considered the potential visibility benefits of controls by conducting three different photochemical modeling sensitivity runs representing different control scenarios. Similar to how it calculated the total annualized costs, Texas only included those control measures at sources for which the cost of the control measures met the $5,000/ton threshold for NO_X or SO₂ .^[204] While Texas's 2021 Regional Haze Plan did not specifically identify (in Chapter 7 or Chapter 8 of its SIP) which sources or control measures were actually included in the sensitivity modeling, the information the TCEQ included in PowerPoint presentations used for consultation indicates control measures for 11 out of the 18 sources selected for evaluation under the four factor analysis were included in the sensitivity modeling.^[205] Each sensitivity scenario reduced NO_X and/or SO₂ emissions at specific EGU and non-EGU sources for the modeled 2028 scenario.^[206] Scenario 1 ^[207] involved the removal of emissions from the Oklaunion Power Station as its owners had announced its retirement in 2020. Scenario 2 ^[208] included SO₂ reductions at all units with identified cost-effective SO₂ controls in addition to Scenario 1. Scenario 3 ^[209] included NO_X reductions at all units with identified cost-effective NO_X controls in addition to Scenario 2. We note that the additional visibility improvements provided by the inclusion of NO_X controls in Scenario 3 provided little additional visibility benefit on the average across the 20 percent most impaired days, yet the associated costs of these controls resulted in several millions of dollars being included in the total annual costs Texas calculated in its 2021 Texas Regional Haze Plan.^[210] The results of this modeling analysis were used to estimate the overall visibility benefit these controls would have on the 20 percent most impaired days at the Class I areas impacted by Texas's emissions. The projected visibility improvements at Class I areas impacted by Texas sources under Scenario 3 are presented in table 20.^[211]

Table 20—Texas' s Estimated Haze Index Improvements for Affected Class I Areas ²¹²

Class I area	Haze index improvement (dv)
Big Bend	0.07
Caney Creek	0.56
Guadalupe Mountains	0.03
Salt Creek	0.07
Upper Buffalo	0.21
White Mountain	0.02
Wichita Mountains	0.23

Texas ultimately determined that any visibility benefit for each Class I area would be “imperceptible.” Thus, combining the “imperceptible” projected visibility benefit for each Class I area with the corresponding total annual costs associated with the controls included in the modeled control strategy, Texas concluded that no additional control measures were necessary to make reasonable progress. As discussed below, the EPA finds Texas's conclusion to be unjustified, unreasonable, and inconsistent with the CAA and the RHR.

i. Texas's Consideration of Costs To Support Its Determination That No Additional Measures Are Necessary To Make Reasonable Progress Was Unjustified and Unreasonable

Texas determined that the total annualized cost of controls of ( print page 83365) approximately $205 million was too high, but provided no context or support as to why total annualized cost was an appropriate decision metric in consideration of the cost of compliance, what range of total annualized cost would be reasonable, and why $205 million was not reasonable. While the RHR does allow for the evaluation of sources on either a source-by-source basis or based on the evaluation of groups of sources, in almost any case, a State could, as Texas has here, aggregate the annualized control costs for a large number of sources such that the State could find the total cost to be “too expensive;” and therefore, determine that no additional controls are necessary to make reasonable progress. This is especially true in States like Texas given the vast number of sources in the State and the number of Class I areas impacted by the emissions from these sources. Thus, a reasonable source selection for a State like Texas would necessarily identify several sources for evaluation of potential control measures for which total annualized costs would end up being “large.” Therefore, it is unsurprising that Texas found that total annualized costs of controls were over $200 million; however, without a relative scale to compare against, this $200 million figure is meaningless and does not necessarily support Texas's determination that no control measures are necessary for inclusion in its long-term strategy to make reasonable progress. This concern is supported by EPA's 2019 Guidance which states that, “EPA does not believe it is reasonable to solely use a threshold for the capital cost or annualized cost to determine that a measure is not necessary to make reasonable progress. Large capital costs considered in isolation may not provide complete information about the potential reasonableness of a measure . . .” ^[213] Furthermore, if this approach were replicated in each successive planning period, no controls would ever be found to be cost-effective and necessary to make reasonable progress, which would result in no long-term strategy. Rather, all that can be determined from Texas's use of the total annualized cost is that it represents the sum total of the costs of controls for a group of sources that impact one or more Class I areas in Texas or nearby States. Therefore, Texas's use of total annualized cost was unjustified and unreasonable.

In addition to failing to justify how consideration of total annualized cost was reasonable, Texas also failed to explain and justify the apparent contradiction between considering controls to be cost effective on a source specific basis using a threshold of $5,000/ton, but then dismissing those same controls as too costly when presented as total annualized costs. The need to support and justify this apparent contradiction is critical considering that Texas selected its $5,000/ton cost effectiveness threshold to “identif[y] the potential control measures for each source that could be applied in a cost-effective manner,” ^[214] and thus eliminate those control measures which they deemed too costly. Texas's reliance on the total annual costs of all controls considered cannot outweigh or otherwise negate the fact these controls were all below Texas's selected cost-effectiveness threshold of $5,000/ton. Furthermore, we note that the controls that make up this total annualized cost have an average $/ton cost-effectiveness of less than $2,500/ton SO₂ reduced and less than $3,000/ton for NO_X reduced.

Additionally, while the EPA finds that Texas's use of total annualized costs was unjustified and unreasonable, even if such a metric were appropriate, Texas's total annualized cost of approximately $205 million included unreasonable costs associated with the scrubber upgrades it evaluated. As previously explained in section IV.E.2.b, Texas's calculation of the costs associated with upgrading the scrubbers at Martin Lake, Pirkey, and San Miguel used an unsupported outlier value in determining the costs, which resulted in an inflated cost estimation. Had Texas used the average costs rather than the outlier value, the total annualized cost of the scrubber upgrades would have decreased by approximately $49 million, and the total annualized cost of controls would have decreased from approximately $205 million to $156 million.^[215] Thus, this one decision significantly and unreasonably inflates the total annualized cost. Even assuming the total annualized cost metric is a reasonable way of considering costs, because Texas failed to describe or justify why $205 million was too high, and what range of costs would be reasonable, we cannot determine whether Texas would have found this lower total annualized cost reasonable such that the measures are necessary for inclusion in its long-term strategy to make reasonable progress. Thus, the EPA finds that Texas failed to justify how its use of total annualized costs to dismiss controls was reasonable and consistent with the CAA and RHR requirement to include those measures necessary to make reasonable progress in its long-term strategy.

ii. Texas's Reliance on the Lack of Perceptible Visibility Benefits To Support Its Determination That No Additional Measures Are Necessary Was Unreasonable and Inconsistent With the CAA and the RHR

Texas's determination that visibility benefits are only meaningful if it results in a perceptible change in visibility was unjustified and unreasonable. As previously explained, after identifying which of the 18 sources selected for further analysis using the four statutory factors had potential control measures meeting the $5,000/ton cost-effectiveness threshold for NO_X or SO₂, those emission reductions associated with those control measures were then included in the photochemical modeling sensitivity runs conducted by the TCEQ. The projected visibility benefits are presented in table 20. Because the results of the modeling analysis showed that the visibility benefit of the modeled control strategy for each Class I area fell within a range that was “imperceptible,” (which Texas defines as less than 1.0 deciview), Texas found that this amount of visibility improvement was too small to support requiring any additional control measures for purposes of making reasonable progress during this planning period.

The CAA and RHR are clear that the four statutory factors must be considered when determining the enforceable emissions limitations, schedules of compliance, or other measures that are necessary for reasonable progress toward meeting the national goal.^[216] As the EPA has previously explained, while visibility may be considered along with the four statutory factors, it must be done in a reasonable way.^[217] For example, visibility modeling can be used to compare the visibility benefits of cost-effective controls selected through four-factor analysis to determine which controls produce the greatest visibility benefits compared to their costs, or prioritizing which among several sources should install controls during a planning period.^[218] Nowhere in the statute or regulations is there a requirement that control measures produce perceptible visibility ( print page 83366) improvements to be considered necessary to make reasonable progress at a particular Class I area; therefore, consideration of visibility benefits cannot outweigh the results of the analysis based on the four factors explicitly prescribed in statute.^[219] Furthermore, if a State uses a visibility benefit threshold to evaluate control measures, it must explain how its approach is consistent with the requirement to consider the statutory factors in making reasonable progress determinations. Texas did not explain how the use of perceptibility as a threshold to assess visibility benefits is consistent with the requirement to make reasonable progress.

Section 169A(a) of the CAA establishes as a national goal the “prevention of any future, and the remedying of any existing, impairment of visibility in mandatory Class I Federal areas which impairment results from manmade air pollution.” Nowhere in the CAA or the RHR is there a requirement to make a minimum amount of visibility improvement in determining that potential control measures are necessary to make reasonable progress. Rather, States are to make “reasonable progress” towards natural visibility conditions every planning period. What is necessary for reasonable progress, as described throughout this section and this notice, is determined by a consideration of the four statutory factors. To that end, the EPA has reiterated that visibility thresholds used for BART and other analyses in the first planning period ( e.g., 0.5 deciviews or 1 dv) are, in most cases, not appropriate thresholds for evaluating the impact of controls for reasonable progress in the second planning period and beyond.^[220] This is because regional haze is visibility impairment that is caused by the emission of air pollutants from numerous anthropogenic sources located over a wide geographic area.^[221] At any given Class I area, hundreds or even thousands of individual sources may contribute to regional haze.^[222] This necessarily means that to meet Congress's goal of preventing any future, and addressing any existing impairment, States must address these numerous sources of manmade air pollution which cause visibility impairment at Class I areas. Given the iterative nature of the regional haze program, evaluation of control measures for relatively smaller sources (with commensurate smaller visibility benefits) will be needed to continue making reasonable progress towards the national goal. As such, states should consider the magnitude of modeled visibility impacts or benefits in the context of its own contribution to visibility impairment. That is, whether a particular visibility impact or change is “meaningful” should be assessed in the context of the individual state's contribution to visibility impairment. At several Class I areas that Texas evaluated in its 2021 Regional Haze Plan, sulfate was the largest cause of anthropogenic visibility impairment, with the largest contribution coming from Texas anthropogenic sources.^[223] Texas's own modeling also showed that, for multiple Class I areas, relative to the home State in which the Class I area is located, Texas's contribution to sulfate concentrations at the Class I area was more than the home State itself. For example, Texas's sulfate contribution at Caney Creek is nine times the amount of Arkansas's contribution (Texas anthropogenic contribution to particulate sulfate is 40.81 percent compared to Arkansas's anthropogenic contribution of 4.4 percent).^[224 225] At Wichita Mountains, Texas's sulfate contribution is over four times Oklahoma's contribution.^[226] Yet, by using a threshold of perceptibility, Texas found that despite these impacts, the visibility benefits were too small to warrant requiring any additional control measures to make progress towards reducing this contribution. Such an approach is unreasonable as the approach results in maintaining significant visibility impairment in contradiction to Congress's expressly stated goal of remedying manmade impairment.^[227] This concern in part is why the EPA has explained that “the existence of an impact above a perceptibility threshold is not a statutory or regulatory factor to be used when determining whether a source or sources contribute to visibility impairment or when determining measures needed for reasonable progress.” ^[228] Thus, Texas's determination that the lack of perceptible visibility benefits weighed in favor of its determination that no additional measures were necessary was unreasonable and failed to result in a long-term strategy that encompassed all of the measures necessary to make reasonable progress in the second planning period.

Contrary to Texas's own conclusions, the EPA finds that the modeled TCEQ control scenarios are projected to achieve meaningful reductions in impairment. In table 21, based on Texas's own modeling and considering visibility impairment using light extinction units, the control scenarios provide for meaningful progress in reducing visibility impairment, particularly at Caney Creek. Considering just US anthropogenic impairment in 2028, Texas is responsible for 43 percent of the total U.S. anthropogenic impairment on the 20 percent most impaired days at Caney Creek. The modeled 3.18 Mm-1 reduction in impairment under Texas's Scenario 2 represents a 10.6 percent reduction of the total US anthropogenic impairment in 2028 and 25 percent reduction of the Texas contribution to anthropogenic impairment. In consideration of the statutory goal to remedy “ any existing impairment of visibility in mandatory Class I Federal area which impairment results from manmade air pollution,” ^[229] it is not reasonable to dismiss a potential 10.6 percent reduction in the nationwide total anthropogenic impairment and a 25 percent reduction in the Texas contribution to impairment as insignificant, especially since Texas found all of the modeled controls to be below their chosen cost-effectiveness threshold of $5,000/ton.

( print page 83367)

Table 21 ²³⁰ —Visibility Benefit of Texas's Control Scenario 2

Class I area	IMPROVE monitor	2028 extinction (Mm-1)	Texas anthro (%)	Non- Texas U.S. anthro (%)	Total US anthro (Mm-1)	Total anthro from Texas (Mm-1)	Texas % of total US anthropogenic impairment	Scenario 2 extinction reduction (Mm-1)	Scenario 2 reduction of total US anthropogenic (%)	Scenario 2 reduction of Texas anthropogenic contribution (%)	Scenario 2 extinction reduction (dv)	Scenario 2 extinction reduction compared to Natural conditions (dv)
Caney Creek	CACR1	55.4	23	31	29.92	12.74	43	−3.18	−10.60	−25.00	−0.56	−1.32
Big Bend	BIBE1	41.2	10	5	6.18	4.12	67	−0.31	−5.00	−7.50	−0.07	−0.18
Upper Buffalo	UPBU1	53.4	13	38	27.23	6.94	25	−1.2	−4.40	−17.30	−0.21	−0.48
Wichita Mountains	WIMO1	53.2	18	33	27.13	9.58	35	−1.19	−4.40	−12.40	−0.22	−0.61
Hercules-Glades	HEGL1	57.2	9	48	32.60	5.15	16	−0.78	−2.40	−15.20	−0.13	−0.31
Salt Creek	SACR1	40.3	12	34	18.54	4.84	26	−0.27	−1.50	−5.60	−0.06	−0.16
Guadalupe Mountains	GUMO1	34	11	11	7.48	3.74	50	−0.1	−1.30	−2.70	−0.03	−0.06

Texas's consideration of visibility benefit is also unreasonable because Texas only considered the potential visibility benefits relative to “dirty background” conditions. Because estimates of the visibility benefits of emission control measures help guide regulatory decisions, relying solely on a quantification of visibility benefits relative to “dirty background” ^[231] as Texas did in its 2021 Regional Haze Plan ( i.e., conditions with greater impairment than natural background visibility conditions) obscures the full potential benefits of control measures and makes it less likely that a measure, or measures, would appear reasonable from a visibility benefit perspective.^[232] Thus, this approach to considering visibility benefit serves to maintain the current impairment at Class I areas, which is inconsistent with the statutory goal of the CAA § 169A(a)(1) to eliminate future, and remedy existing manmade visibility impairment. Texas's own modeling results show that had Texas considered the visibility improvement associated with the control scenarios it modeled relative to natural background, the visibility improvement would have been considerably larger. For example, under control Scenario 2, the visibility improvement at Caney Creek would be considerably larger (1.32 deciviews) than the values documented by Texas (0.56 deciview).^[233] The right most column in table 21 shows Texas's modeled visibility benefits calculated relative to natural visibility conditions. Because Texas's consideration of projected visibility benefits was limited to a dirty background basis and did not consider the full potential benefits associated with each control scenario it evaluated, Texas's determination that the visibility benefits did not support requiring any additional control measures was unreasonable.

Recent annual emissions data from EPA's Clean Air Markets Program Data also contradict Texas's conclusion that no controls are needed due to the lack of perceptible visibility improvement. Across all states, Texas EGU SO₂ emissions ranked 1st and has ranked 1st over the past several years.^[234] Within the group of sources analyzed by Texas, Martin Lake and Coleto Creek ranked 6th and 31st, respectively, in facility-wide SO₂ emissions across the United States.^[235] The magnitude of SO₂ emissions from the sources Texas included in its sensitivity run, as well as all of Texas's EGUs statewide, is demonstrated in the model results showing Texas's large contribution to the total U.S anthropogenic visibility impairment. This, combined with the outcome of the four-factor analyses, emphasize that emission reductions from additional SO₂ controls on the sources Texas considered are cost-effective would result in meaningful progress towards remedying visibility impairment from manmade pollution at impacted Class I areas.

Therefore, the EPA finds that Texas's use of perceptibility as a visibility threshold to support its decision to dismiss controls was unreasonable, resulted in an unjustified long-term strategy for the second planning period, and is inconsistent with the CAA and the RHR.

g. EPA's Conclusions and Proposed Action on Texas's Four Factor Analysis

As explained in the preceding sections, due to numerous flaws in its evaluation of the four-factors and the resulting control determinations, Texas failed to submit to the EPA a long-term strategy that includes “the enforceable emissions limitations, compliance schedules, and other measures that are necessary to make reasonable progress” as required by 40 CFR 51.308(f)(2).^[236] Consequently, the EPA is proposing to find that the 2021 Texas Regional Haze Plan does not satisfy the requirements of 40 CFR 51.308(f)(2).

3. Additional Long-Term Strategy Requirements

Aside from the long-term strategy requirements already discussed, States must also meet the requirements specified in 40 CFR 51.308(f)(2)(ii)-(iv) when developing their long-term strategies for the second planning period. Under 40 CFR 51.308(f)(2)(ii) states are required to consult with other states that have emissions that are ( print page 83368) reasonably anticipated to contribute to visibility impairment in Class I areas to develop coordinated emission management strategies. Texas included documentation of its consultation with other states and the FLMs in appendix A of its 2021 Regional Haze Plan.

In addition to our analysis on Section 51.308(f)(2)(iii) above, this section also requires that the emissions information considered to determine the measures that are necessary to make reasonable progress include information on emissions for the most recent year for which the State has submitted triennial emissions data to the EPA (or a more recent year), with a 12-month exemption period for newly submitted data. Texas's 2021 Texas Regional Haze Plan included 2011, 2014, and 2017 statewide NEI emission data for NO_X, SO₂, PM, VOCs, and NH₃. For the base case CAMx modeling, Texas also relied on 2018 emissions from EPA's AMPD, and 2016 emissions data reported to the State of Texas Air Reporting System (STARS) database for non-EGU sources.

Finally, in developing their long-term strategies, States must consider five additional factors specified under 40 CFR 51.308(f)(2)(iv). The five additional factors are: emission reductions due to ongoing air pollution control programs, including measures to address reasonably attributable visibility impairment; measures to mitigate the impacts of construction activities; source retirement and replacement schedules; basic smoke management practices for prescribed fire used for agricultural and wildland vegetation management purposes and smoke management programs; and the anticipated net effect on visibility due to projected changes in point, area, and mobile source emissions over the period addressed by the long-term strategy. Chapter 7 of Texas's 2021 Regional Haze Plan includes a description of these additional factors.

Regardless, as explained in preceding sections, due to flaws and omissions in its source selection, four-factor analyses, and the resulting control determinations, we find that Texas failed to reasonably “evaluate and determine the emission reduction measures that are necessary to make reasonable progress” by considering the four statutory factors as required by CAA section 169A(b)(2)(A), CAA section 169A(g)(1), and 40 CFR 51.308(f)(2)(i). We also find that Texas failed to adequately document the technical basis that it relied upon in evaluating potential emissions reduction measures, as required by 40 CFR 51.308(f)(2)(iii). In so doing, the EPA finds that Texas failed to submit to the EPA a long-term strategy that includes “the enforceable emissions limitations, compliance schedules, and other measures that are necessary to make reasonable progress” as required by 40 CFR 51.308(f)(2). Consequently, the EPA finds that the Texas's 2021 Regional Haze Plan does not satisfy the long-term strategy requirements of 40 CFR 51.308(f)(2). Therefore, we are proposing to disapprove these corresponding portions of Texas's SIP submission.

F. Reasonable Progress Goals

Section 51.308(f)(3) contains the requirements pertaining to RPGs for each Class I area. Texas is host to two Class I areas and is therefore subject to § 51.308(f)(3)(i) and, if appliable, to (ii). Section 51.308(f)(3)(i) requires a State in which a Class I area is located to establish RPGs—one each for the most impaired and clearest days—reflecting the visibility conditions that will be achieved at the end of the implementation period as a result of the emission limitations, compliance schedules and other measures required under paragraph (f)(2) to be in states' long-term strategies, as well as implementation of other CAA requirements. The long-term strategies as reflected by the RPGs must provide for an improvement in visibility on the most impaired days relative to the baseline period and ensure no degradation on the clearest days relative to the baseline period. Section 51.308(f)(3)(ii) applies in circumstances in which a Class I area's RPG for the most impaired days represents a slower rate of visibility improvement than the uniform rate of progress calculated under 40 CFR 51.308(f)(1)(vi). Under § 51.308(f)(3)(ii)(A), if the State in which a mandatory Class I area is located establishes an RPG for the most impaired days that provides for a slower rate of visibility improvement than the URP, the State must demonstrate that there are no additional emission reduction measures for anthropogenic sources or groups of sources in the State that would be reasonable to include in its long-term strategy. Section 51.308(f)(3)(ii)(B) requires that if a State contains sources that are reasonably anticipated to contribute to visibility impairment in a Class I area in another State, and the RPG for the most impaired days in that Class I area is above the URP, the upwind State must provide the same demonstration.

Texas established RPGs based on projected visibility improvements from emission reductions associated with the Federal CAA, the Texas Clean Air Act, Texas' ozone SIP revisions and rules, and agreements between the EPA and petrochemical refineries and carbon black manufacturing plants for nitrogen oxides (NO_X) and sulfur dioxide (SO₂) emissions reductions. As part of establishing the RPGs, the TCEQ evaluated the impact of emissions reductions from these adopted measures on visibility in Class I areas using photochemical modeling. Further, the TCEQ evaluated the impacts of additional controls beyond those already adopted using photochemical modeling in a sensitivity analysis. Based on the results of Texas's four-factor analysis and the sensitivity analysis, the TCEQ concluded that additional measures for visibility improvement at Texas Class I areas and Class I areas affected by Texas emissions are not reasonable for this planning period.

The TCEQ elected to perform CAMx modeling to develop its future year visibility projections to establish its reasonable progress goals and evaluate the impact of identified emissions reductions on visibility in Class I areas. The CAMx modeling was based on the EPA's Modeling Guidance and consistent with the modeling protocol included in appendix G of its SIP (Modeling Protocol). The photochemical modeling used to support the 2021 Texas Regional Haze Plan consisted of base case model runs, future year model runs, including source apportionment runs, and three sensitivity runs.^[237] The TCEQ describes the development of its emission inventories for use in each modeling scenario in appendix E of its SIP (Emissions Modeling).

The TCEQ elected to use the adjusted Uniform Rate of Progress (URP) in its 2021 Regional Haze Plan SIP to evaluate its reasonable progress goals. The TCEQ presents the visibility for Class I areas on the 20% clearest days and 20% most impaired days for the 2014-2017 period, 2028 projected future year, and the 2028 adjusted glidepath and are shown in table 8-43 of its SIP and presented here in table 22. ( print page 83369)

Table 22—Visibility for Class I Areas on 20% Most Impaired Days and 20% Clearest Days

Class I area (IMPROVE ID, state)	2014-2017 20% clearest days (dv)	Future year (2028) 20% clearest days (dv)	2028 adjusted glidepath (dv)	Future year (2028) 20% most impaired days (dv)
Big Bend National Park (N.P.) (BIBE, TX)	5.2	4.9	14.4	14.2
Bosque del Apache Wilderness Area (W.A.) (BOAP, NM)	4.6	4.2	9.9	9.6
Breton Island W.A. (BRIS, LA)	11.8	11.3	19.8	18.3
Caney Creek W.A. (CACR, AR)	8.2	7.8	18.8	17.1
Great Sand Dunes W.A. (GRSA, CO)	2.9	2.6	8.2	7.3
Guadalupe Mountains N.P. (GUMO, TX)	4.5	4.1	12.8	12.2
Hercules-Glades W.A. (HEGL, MO)	9.8	9.1	19.6	17.4
Mingo W.A. (MING, MO)	11.2	10.6	20.2	18.6
Rocky Mountain N.P. (ROMO, CO)	1.3	1.1	9.2	7.3
Salt Creek W.A. (SACR, NM)	6.7	6.2	13.5	13.9
Upper Buffalo W.A. (UPBU, AR)	8.4	7.9	19.2	16.7
White Mountain W.A. (WHIT, NM)	2.6	2.2	10	9.5
Wheeler Peak W.A. (WHPE, NM)	0.3	0.1	6.5	5.3
Wichita Mountains W.A. (WIMO, OK)	8.4	7.7	17.4	16.7
Source: Texas 2021 Regional Haze SIP, Table 8-43.

Texas included baseline haze indices for Big Bend and Guadalupe Mountains in Chapter 8 of its 2021 Regional Haze Plan. Baseline visibility for the Big Bend Class I area is 5.78 dv for the 20% clearest days and 15.57 dv for the 20% most impaired days. Baseline visibility for the Guadalupe Mountains Class I area is 5.92 dv for the 20% clearest days and 14.60 dv for the 20% most impaired days. As Texas notes in its 2021 Texas Regional Haze Plan, and as shown in the data presented in table 22, the RPGs Texas established for both Big Bend and Guadalupe Mountains are below the adjusted glidepath.

Texas emissions impact visibility at one Class I area, Salt Creek Wilderness Area, in New Mexico, that is projected to be above the glidepath. Section 51.308(f)(3)(ii) requires states to demonstrate for Class I areas with a 2028 reasonable progress goal for the 20% most impaired days above the 2028 URP that there are no additional emission reduction measures for sources in the State that would be reasonable to include in the long-term strategy. The TCEQ states in its 2021 Texas Regional Haze Plan that the New Mexico Environment Department had not yet established a reasonable progress goal for Salt Creek Wilderness Area or developed its long-term strategy at the time Texas prepared its SIP; however Texas states that its analysis in the long-term strategy is robust, in accordance with 40 CFR 51.308(f)(3)(ii), and that it has provided a thorough evaluation of the Texas sources that impact Class I areas in and around Texas and consideration of whether any additional emission reduction measures are reasonable for the second planning period.^[238]

As previously discussed in Section IV.E.1, using its source selection methodology, Texas did not select any sources of SO₂ for further evaluation at Salt Creek, despite Texas's PSAT modeling showing that Texas sources are responsible for almost three times the amount of influence due to particulate sulfate and more than one and half times the influence due to particulate nitrate as the home State of New Mexico.^[239] Focusing on modeled U.S. anthropogenic impacts alone, Texas anthropogenic sources account for approximately 51.3% of the particulate sulfate concentrations at Salt Creek, with more than half of the Texas anthropogenic contribution coming from Texas EGUs.^[240] Furthermore, the sensitivity modeling Texas conducted showed that potential SO₂ and NO_X reductions from the aggregate group of control measures considered would provide for an estimated 0.07 dv improvement in visibility at Salt Creek. This is despite the fact that only a few of the NO_X sources included in the sensitivity analyses were included based on their impact to Salt Creek and no SO₂ sources were selected based on their impact at Salt Creek. The 0.07 dv improvement is calculated from a reduction in extinction of 0.27 Mm⁻¹ and represents a 1.5 percent reduction of total U.S. anthropogenic contribution and a 5.6 percent reduction of Texas's total anthropogenic contribution to visibility impairment at Salt Creek. While New Mexico had not established an RPG for Salt Creek at the time Texas submitted its SIP, contrary to its assertion, Texas's analysis did not meet the requirements of 51.308(f)(3)(ii)(B) to conduct a robust analysis with respect to Salt Creek as evidenced by the fact Texas did not evaluate sources of SO₂ despite PSAT modeling showing the substantial impact on the area from Texas.

Under 40 CFR 51.308(f)(3)(iv), the EPA must evaluate the demonstrations the State developed pursuant to 40 CFR 51.308(f)(2) to determine whether the State's reasonable progress goals for visibility improvement provide for reasonable progress towards natural visibility conditions. As previously explained in section IV.E., we are proposing to disapprove Texas's long-term strategy for failing to meet the requirements of 40 CFR 51.308(f)(2). Therefore, we also propose to disapprove Texas's reasonable progress goals under 40 CFR 51.308(f)(3) because compliance with that requirement is dependent on compliance with 40 CFR 51.308(f)(2).

G. Reasonably Attributable Visibility Impairment (RAVI)

The RHR contains a requirement at §  51.308(f)(4) related to any additional monitoring that may be needed to address visibility impairment in Class I areas from a single source or a small group of sources. This is called “reasonably attributable visibility impairment,” ^[241] also known as RAVI. ( print page 83370) Under this provision, if the EPA or the FLM of an affected Class I area has advised a State that additional monitoring is needed to assess RAVI, the State must include in its SIP revision for the second implementation period an appropriate strategy for evaluating such impairment. The EPA has not advised Texas to that effect, and the FLMs for the Class I areas that Texas contributes to have not identified any RAVI from Texas sources.^[242] For this reason, the EPA proposes to approve the portions of Texas's 2021 Regional Haze Plan relating to 40 CFR 51.308(f)(4).

H. Monitoring Strategy and Other Implementation Plan Requirements

Section 51.308(f)(6) specifies that each comprehensive revision of a state's regional haze SIP must contain or provide for certain elements, including monitoring strategies, emissions inventories, and any reporting, recordkeeping and other measures needed to assess and report on visibility. A main requirement of this subsection is for states with Class I areas to submit monitoring strategies for measuring, characterizing, and reporting on visibility impairment. Compliance with this requirement may be met through participation in the Interagency Monitoring of Protected Visual Environments (IMPROVE) network.

Texas discusses its monitoring strategy in Chapter 5 of its 2021 Regional Haze Plan. Haze species in Texas are measured and analyzed via the Interagency Monitoring of Protected Visual Environments (IMPROVE) network.^[243] Table 23 of this preamble lists the IMPROVE stations representing visibility at Texas Class I areas. Due to the close proximity of the Class I areas, Carlsbad Caverns (New Mexico) and Guadalupe Mountains (Texas) share the same IMPROVE monitor.

Table 23—IMPROVE Stations at Federal Class I Areas in Texas

Monitor ID	Class I area	Sponsor	Years operated
BIBE1	Big Bend National Park	National Parks Service	1988-Present.
GUMO1	Guadalupe Mountains National Park	National Parks Service	1988-Present.

Section 51.308(f)(6)(i) requires SIPs to provide for the establishment of any additional monitoring sites or equipment needed to assess whether reasonable progress goals to address regional haze for all mandatory Class I Federal areas within the State are being achieved.

The IMPROVE program reviewed its aerosol monitoring sites in 2006 to set priorities for maintaining the sites in the event of federal budget cuts affecting the IMPROVE program.^[244] This review determined that the IMPROVE aerosol samplers at Texas's two Class I areas represent conditions different from the conditions at the nearest Class I area IMPROVE monitors. Texas's two Class I IMPROVE monitors are not candidates for discontinuation since other IMPROVE monitors cannot represent conditions at Big Bend or Guadalupe Mountains.

Section 51.308(f)(6)(ii) requires SIPs to provide for procedures by which monitoring data and other information are used in determining the contribution of emissions from within the State to regional haze visibility impairment at mandatory Class I Federal areas both within and outside the State. In its 2021 Texas Regional Haze Plan, Texas stated that future assessments of visibility impairment and progress in reducing visibility impairment at Texas's two Class I areas, and at Class I areas in other states that Texas's emissions may potentially affect, will use the revised IMPROVE algorithm and will use data as prescribed in the EPA's Regional Haze Rule (40 CFR part 51, subpart P—Visibility Protection). The assessment will follow, as appropriate, the EPA's guidance including the 2019 Guidance and the 2018 Visibility Tracking Guidance.

Section 51.308(f)(6)(iii) does not apply to Texas, as it has a Class I area.

Section 51.308(f)(6)(iv) requires the SIP to provide for the reporting of all visibility monitoring data to the Administrator at least annually for each Class I area in the State. As noted above, the monitoring strategy for Texas relies upon the continued availability of the IMPROVE network. The TCEQ does not directly collect or handle IMPROVE data. The TCEQ will continue to participate in the IMPROVE Visibility Information Exchange Web System (VIEWS). The TCEQ considers VIEWS to be a core part of the overall IMPROVE program. The TCEQ will report IMPROVE data from the two Class I areas in Texas to the EPA using the VIEWS web system.

If Texas collects any visibility monitoring data through the state's air quality monitoring networks, the TCEQ will report those data to the EPA as specified under the Performance Partnership Grant agreement negotiated with the EPA Region 6. All validated data and data analysis results from any TCEQ visibility-related special studies are public information. The TCEQ will continue its practice of sharing the data and information with the EPA. Texas supports the continued operation of the IMPROVE network through both State and Federal funding mechanisms.

Section 51.308(f)(6)(v) requires SIPs to provide for a statewide inventory of emissions of pollutants that are reasonably anticipated to cause or contribute to visibility impairment, including emissions for the most recent year for which data are available and estimates of future projected emissions. It also requires a commitment to update the inventory periodically. Texas provides for emissions inventories and estimates for future projected emissions by participating in the CenSARA RPO and complying with the EPA's Air Emissions Reporting Rule (AERR). In 40 CFR part 51, subpart A, the AERR requires states to submit updated emissions inventories for criteria pollutants to the EPA's Emissions Inventory System (EIS) every three years. The emission inventory data is used to develop the NEI, which provides for, among other things, a triennial state-wide inventory of pollutants that are reasonably anticipated to cause or contribute to visibility impairment.

Chapter 6 of the 2021 Texas Regional Haze Plan includes a discussion of the NEI data, and Section 8.3 details specific emission inventories and emissions inputs developed for the regional haze photochemical modeling ( print page 83371) conducted by the TCEQ. The source categories of the emissions inventories included are: (1) point sources, (2) area sources, (3) non-road mobile sources, (4) drilling rigs, (5) commercial marine vessels and locomotives, (6) airports and (7) on-road mobile sources. Statewide pollutant summaries by source category for the years 2011, 2014, and 2017 are provided in tables 6-1, 6-2, and 6-3 of Texas 2021 Regional Haze Plan. Summaries are for the following pollutants: SO₂, NO_X, PM₁₀, PM_2.5, VOCs, CO, and NH₃ . Texas also provided a summary of anthropogenic SO₂ and NO_X emissions for each source type for 2011, 2014, and 2017 and are presented in tables 6-4 and 6-5 of the 2021 Texas Regional Haze Plan.

Section 51.308(f)(6)(v) also requires states to include estimates of future projected emissions and include a commitment to update the inventory periodically. Texas estimated 2028 future year emissions by applying growth projections and accounting for known existing federal, State, and local controls. The development of Texas's 2028 modeling emissions for the 2021 Texas Regional Haze Plan includes some methods used in previous SIP modeling for ozone, such as the Federal Tier 3 Vehicle Emission and Fuel Standards program, the Mass Emissions Cap-and-Trade (MECT) Program in the Houston-Galveston-Brazoria area, the Highly Reactive VOC Emission Cap-and-Trade (HECT) Program in Harris County, the Midlothian Cement Kiln caps and related agreed orders in the Dallas-Fort Worth area, and the EPA's final Cross-State Air Pollution Rule (CSAPR) update. Summaries of the primary data sources for the development of the future case modeling emissions are provided in the 2021 Texas Regional Haze Plan, appendix E, table 1-4: Summary of Future Case Point Source Emission Data Sources, table 1-5: Summary of Future Case On-Road Mobile Source Emission Data Sources, and table 1-6: Summary of Future Case Non-Road Mobile, Off-Road, Area, Oil-and-Gas, and Biogenic Source Emission Data Sources. The gridded photochemical modeling input files for the 2016 and 2028 emissions were provided along with the full emission processing message log files during Texas's public comment period. For point sources, Texas evaluated large stationary sources of emissions, such as electric generating units (EGUs), smelters, industrial boilers, petroleum refineries, and manufacturing facilities. Point source emissions were developed for the January 1 through December 31, 2016, annual episode with a 2028 future year projection. The data sources for development of the point source modeling emissions are summarized in the 2021 Texas Regional Haze Plan, appendix E, table 2-1: Sources of Point Source Emissions Data.

The EPA proposes to find that Texas has met the requirements of 40 CFR 51.308(f)(6) as described above, including its continued participation in the IMPROVE network and the CenSARA RPO and its on-going compliance with the AERR, and that no further elements are necessary at this time for Texas to assess and report on visibility pursuant to 40 CFR 51.308(f)(6)(vi).

In sum, for all the reasons discussed in this section, the EPA is proposing to approve Texas's 2021 Regional Haze Plan as meeting the requirements of 40 CFR 51.308(f)(6).

I. Requirements for Periodic Reports Describing Progress Towards the Reasonable Progress Goals

Section 51.308(f)(5) requires that periodic comprehensive revisions of states' regional haze plans also address the progress report requirements of 40 CFR 51.308(g)(1) through (5). The purpose of these requirements is to evaluate progress towards the applicable RPGs for each Class I area within the State and each Class I area outside the State that may be affected by emissions from within that State. Sections 51.308(g)(1) and (2) apply to all states and require a description of the status of implementation of all measures included in a state's first implementation period regional haze plan and a summary of the emission reductions achieved through implementation of those measures. Section 51.308(g)(3) applies only to states with Class I areas within their borders and requires such states to assess current visibility conditions, changes in visibility relative to baseline (2000-2004) visibility conditions, and changes in visibility conditions relative to the period addressed in the first implementation period progress report. Section 51.308(g)(4) applies to all states and requires an analysis tracking changes in emissions of pollutants contributing to visibility impairment from all sources and sectors since the period addressed by the first implementation period progress report. This provision further specifies the year or years through which the analysis must extend depending on the type of source and the platform through which its emission information is reported. Finally, § 51.308(g)(5), which also applies to all states, requires an assessment of any significant changes in anthropogenic emissions within or outside the State have occurred since the period addressed by the first implementation period progress report, including whether such changes were anticipated and whether they have limited or impeded expected progress towards reducing emissions and improving visibility.

The 2021 Texas Regional Haze Plan describes the status of measures of the long-term strategy from the first implementation period to address the requirements found in 40 CFR 51.308(g)(1) and (2). Control measures to reduce emission within and outside the State are found in the 2021 Texas Regional Haze Plan, Chapter 7: Long-Term Strategy to Establish Reasonable Progress Goals, Section 7.4: Federal Programs that Reduce Stationary Source Emissions, Section 7.5: Federal Programs that Reduce Mobile Source Emissions, and Section 7.6: State Air Pollution Control Programs. Control measures in the State are included in Section 7.6: State Air Pollution Control Programs, which discusses both State stationary and mobile source emissions control measures; Section 7.6.2: Best Available Control Technology (BACT) Requirements, which discusses air permitting requirements for new and modified sources of air pollution; and finally Section 7.6.3: Additional Measures, which discusses other measures addressing air pollution from mobile sources, construction activities, and fires, and measures addressing energy efficiency. Emissions reductions are found in the 2021 Texas Regional Haze Plan, Chapter 6: Emissions Inventory, Section 6.8: NO_X and SO₂ Emissions Trends, table 6-4: Anthropogenic NO_X Emissions by Source Type, and table 6-5: Anthropogenic SO₂ Emissions by Source Type.

The EPA proposes to find that Texas has addressed the requirements of 40 CFR 51.308(g)(1) and (2) because the 2021 Texas Regional Haze Plan describes the measures included in the long-term strategy from the first implementation period, as well as the status of their implementation and the emission reductions achieved through such implementation.

Section 51.308(g)(3) requires that for each Class I area within the State, the State must assess the following visibility conditions and changes, with values for most impaired, least impaired and/or clearest days as applicable expressed in terms of five-year averages of these annual values. The 2021 Texas Regional Haze Plan includes summaries of visibility conditions in Chapter 4: Assessment of Baseline and Current ( print page 83372) Conditions and Estimate of Natural Conditions in Class I Areas, Section 4.2: Baseline Visibility Conditions, Section 4.3: Natural Visibility Conditions. Changes in visibility conditions are displayed in Chapter 8: Reasonable Progress Goals, Section 8.4: Reasonable Progress Goal Status. The EPA therefore proposes to find that Texas has addressed the requirements of 40 CFR 51.308(g)(3).

Pursuant to § 51.308(g)(4), Texas evaluated emission trends for reasonable progress for the 2021 Texas Regional Haze Plan and presented those data in Chapter 6: Emissions Inventory, Section 6.7: Emissions Summaries, table 6-1: 2011 Statewide Pollutant Summary by Source Category, table 6-2: 2014 Statewide Pollutant Summary by Source Category, table 6-3: 2017 Statewide Pollutant Summary by Source Category, table 6-4: Anthropogenic NO_X Emissions by Source Type, table 6-5: Anthropogenic SO₂ Emissions by Source Type. The EPA is proposing to find that Texas has addressed the requirements of § 51.308(g)(4) by providing emissions information for NO_X, SO₂, PM₁₀, PM_2.5, VOCs, and NH₃ broken down by type of source.

Since the 2009 and 2014 Texas regional haze SIP revisions, reductions in anthropogenic emissions within and outside the State have occurred from the following: (1) ongoing rules and regulations for nonattainment areas in Texas (see the 2021 Texas Regional Haze Plan Chapter 7: Long-Term Strategy to Establish Reasonable Progress Goals, Section 7.6: State Air Pollution Control Programs); (2) closing several major coal-fired plants in Texas, which have permanently reduced emissions (see Chapter 7, Section 7.6.3.8: Potential Effects of Economically Driven Coal Burning Power Plant Closures); (3) continuing reductions in mobile emissions through the incentives like Texas Emissions Reduction Plan (TERP) (see Chapter 7, Section 7.6.3.1: Texas Emissions Reduction Plan); (4) ongoing energy efficiency state-wide, which has continued to increase (see Chapter 7, Section 7.6.3.3: Energy-Efficiency (EE) Programs and Renewable Energy (RE) Measures); and other items discussed in Chapter 7 of the 2021 Texas Regional Haze Plan. Texas uses the emissions trend data in the 2021 Texas Regional Haze Plan ^[245] to support the assessment that anthropogenic haze-causing pollutant emissions in Texas have decreased during the reporting period and that changes in emissions have not limited or impeded progress in reducing pollutant emissions and improving visibility. Texas's 2017 emission inventories for NO_X, SO₂, PM₁₀, PM_2.5, VOCs, and NH₃ were lower than their 2014 emission inventories for those same pollutants emissions.^[246] The EPA is proposing to find that Texas has addressed the requirements of § 51.308(g)(5).

In sum, because Texas's 2021 Regional Haze Plan addresses the requirements of 40 CFR 51.308(g)(1) through (5) as required by 40 CFR 51.308(f)(5), the EPA is proposing to approve Texas's 2021 Texas Regional Haze Plan as meeting the requirements of 40 CFR 51.308(f)(5) for periodic progress reports.

J. Requirements for State and Federal Land Manager Coordination

Section 169A(d) of the Clean Air Act requires states to consult with FLMs before holding the public hearing on a proposed regional haze SIP, and to include a summary of the FLMs' conclusions and recommendations in the notice to the public. In addition, section 51.308(i)(2)'s FLM consultation provision requires a State to provide FLMs with an opportunity for consultation that is early enough in the state's policy analyses of its emission reduction obligation so that information and recommendations provided by the FLMs' can meaningfully inform the state's decisions on its long-term strategy. If the consultation has taken place at least 120 days before a public hearing or public comment period, the opportunity for consultation will be deemed early enough, Regardless, the opportunity for consultation must be provided at least sixty days before a public hearing or public comment period at the State level. Section 51.308(i)(2) also provides two substantive topics on which FLMs must be provided an opportunity to discuss with states: assessment of visibility impairment in any Class I area and recommendations on the development and implementation of strategies to address visibility impairment. Section 51.308(i)(3) requires states, in developing their implementation plans, to include a description of how they addressed FLMs' comments.

The TCEQ consulted with the FLMs about the impact of Texas's emissions on regional haze at the regional Class I areas through conference calls. The TCEQ gave a presentation in March 2020 and discussed impacts to Class I areas in the region. An additional meeting was held October 8, 2020, where NPS presented its evaluation of the Texas SIP. NPS requested Texas look at 15 additional sources that were not included in the TCEQ's four-factor analysis. NPS also requested the TCEQ consider impacts to three New Mexico Class I areas: Bandelier, Salt Creek, and Carlsbad Caverns. NPS also identified inconsistencies between the TCEQ's SIP and the CAA. Both the NPS and FS submitted comment letters during the TCEQ's public comment period.

Texas responded to the FLM comments and included the responses in appendix A of their 2021 Regional Haze Plan. Notices of the proposed SIP, availability and the public hearing were published on TCEQ's website and in the Texas Register, the Fort Worth Star Telegram, the Houston Chronicle, the Austin American-Statesman, and the El Paso Times. A virtual public hearing on the proposed SIP revision was held on December 8, 2020, and was available for participation via internet or phone. Written comments relevant to the proposal were accepted until the close of business January 8, 2021.

Additionally, Texas's 2021 Regional Haze Plan includes a commitment to revise and submit a regional haze SIP by July 31, 2028, and every ten years thereafter. The state's commitment includes submitting periodic progress reports in accordance with 40 CFR 51.308(f) and a commitment to evaluate progress towards the reasonable progress goal for each mandatory Class I Federal area located within the State and in each mandatory Class I Federal area located outside the State that may be affected by emissions from within the State in accordance with § 51.308(g).

Regardless of the consultation described above, compliance with 40 CFR 51.308(i) is dependent on compliance with 40 CFR 51.308(f)(2)'s long-term strategy provisions and (f)(3)'s reasonable progress goals provisions. Therefore, because the EPA is proposing to disapprove Texas's long-term strategy under 51.308(f)(2) and the reasonable progress goals under 51.308(f)(3), the EPA is also proposing to disapprove the State's FLM consultation under 51.308(i). While Texas did take administrative steps to provide the FLMs the opportunity to review and provide feedback on the State's draft regional haze plan, the EPA cannot approve that consultation because it was based on a plan that does not meet the statutory and regulatory requirements of the CAA and the RHR, as described in this notice of proposed rulemaking. In addition, if the EPA finalizes our proposed partial approval and partial ( print page 83373) disapproval of the 2021 Texas Regional Haze Plan, the State (or the EPA in the case of a FIP) will be required to again complete the FLM consultation requirements under 40 CFR 51.308(i). Therefore, the EPA proposes to disapprove the FLM consultation component of the 2021 Texas Regional Haze Plan.

V. Proposed Action

For the reasons discussed in this notice, under CAA section 110(k)(3), the EPA is proposing approval of the portions of Texas's 2021 Regional Haze Plan relating to 40 CFR 51.308(f)(1): calculations of baseline, current, and natural visibility conditions, progress to date, and the uniform rate of progress; 40 CFR 51.308(f)(4): reasonably attributable visibility impairment; 40 CFR 51.308(f)(5): ^[247] progress report requirements; and 40 CFR 51.308(f)(6): monitoring strategy and other implementation plan requirements. The EPA is proposing disapproval of the remainder of Texas's 2021 Regional Haze Plan, which addresses 40 CFR 51.308(f)(2): long-term strategy; 40 CFR 51.308 (f)(3): reasonable progress goals; and 40 CFR 51.308(i): FLM consultation.

VI. Environmental Justice Considerations

Information on Executive Order 12898 (Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations, 59 FR 7629, February 16, 1994) and how EPA defines environmental justice (EJ) can be found in the section, below, titled “VII. Statutory and Executive Order Reviews.” For informational and transparency purposes only, the EPA is including additional analysis of environmental justice associated with this proposed action.

EPA conducted screening analyses using EJSCREEN, an environmental justice mapping and screening tool that provides EPA with a nationally consistent dataset and approach for combining various environmental and demographic indicators.^[248] The EJSCREEN tool presents these indicators at a Census block group (CBG) level or a larger user-specified “buffer” area that covers multiple CBGs.^[249] An individual CBG is a cluster of contiguous blocks within the same census tract and generally contains between 600 and 3,000 people. EJSCREEN is not a tool for performing in-depth risk analysis, but is instead a screening tool that provides an initial representation of indicators related to environmental justice and is subject to uncertainty in some underlying data ( e.g., some environmental indicators are based on monitoring data which are not uniformly available; others are based on self-reported data).^[250] To help mitigate this uncertainty, we have summarized EJSCREEN data within larger “buffer” areas covering multiple block groups and representing the average resident within the buffer areas surrounding the sources. We present EJSCREEN environmental indicators to help screen for locations where residents may experience a higher overall pollution burden than would be expected for a block group with the same total population. These indicators of overall pollution burden include estimates of ambient particulate matter (PM_2.5 ), ozone, nitrogen dioxide, and diesel particulate matter concentration, a score for traffic proximity and volume, percentage of pre-1960 housing units (lead paint indicator), and scores for proximity to Superfund sites, risk management plan (RMP) sites, and hazardous waste facilities.^[251] EJSCREEN also provides information on demographic indicators, including percent low-income, unemployment, communities of color, linguistic isolation, and education.

The EPA prepared EJSCREEN reports covering a buffer area of approximately 6-mile radius around each source identified in this proposed rulemaking. Table 24 presents a summary of results from the EPA's screening-level analysis for a few of the areas in Texas compared to the U.S. as a whole. The full, detailed EJSCREEN report for all areas is provided in the docket for this rulemaking.

Table 24—EJSCREEN Analysis Summary for Sources

[see detailed EJSCREEN report for all sources]

Variables	Values for buffer areas (radius) for each source and the U.S. (percentile within U.S. where indicated)
	Limestone	Martin Lake	Oklaunion	San Miguel	U.S.
Pollution Burden Indicators:
Particulate matter (PM 2.5 ), annual average	8.13 μg/m3 (49%ile)	8.8 μg/m3 (69%ile)	6.94 μg/m3 (17%ile)	8.38 μg/m3 (58%ile)	8.45 μg/m3 (—)
Ozone, annual average of the top ten 8-hour daily maximums	61 ppb (53%ile)	56.9 ppb (32%ile)	57.2 ppb (33%ile)	61.7 ppb (56%ile)	61.8 ppb (—)
Nitrogen dioxide, annual average	3.7 ppb (11%ile)	3.2 ppb (8%ile)	3.6 ppb (11%ile)	2.9 ppb (6%ile)	7.8 ppb (—)
Diesel particulate matter	0.0574 μg/m3 (11%ile)	0.0572 μg/m3 (11%ile)	0.0496 μg/m3 (8%ile)	0.0384 μg/m3 (4%ile)	0.191 μg/m3 (—)
Toxic releases to air score *	320 (39%ile)	9400 (92%ile)	32 (14%ile)	92 (23%ile)	4,600 (—)
Traffic proximity and volume score *	12,000 (5%ile)	9,900 (4%ile)	59,000 (13%ile)	28,000 (8%ile)	1,700,000 (—)
Lead paint (percentage pre-1960 housing)	0.061% (29%ile)	0.12% (38%ile)	0.51% (74%ile)	0.08% (32%ile)	0.3% (—)
Superfund proximity score *	0 (0%ile)	0.014 (56%ile)	0 (0%ile)	0 (0%ile)	0.39 (—)
RMP proximity score *	0.14 (39%ile)	0.18 (42%ile)	0.32 (53%ile)	0.084 (30%ile)	0.57 (—)
( print page 83374)
Hazardous waste proximity score *	0.058 (15%ile)	0.055 (15%ile)	0 (0%ile)	0 (0%ile)	3.5 (—)
Underground storage tank proximity score *	0.022 (29%ile)	0.18 (36%ile)	0.11 (34%ile)	0.000039 (26%ile)	3.6 (—)
Wastewater discharge score *	52 (50%ile)	50 (49%ile)	0.35 (18%ile)	14 (38%ile)	700,000 (—)
Drinking water noncompliance, points	2.7 (87%ile)	9.9 (92%ile)	2.2 (87%ile)	0.86 (77%ile)	2.2 (—)
Demographic Indicators:
People of color population	21% (37%ile)	33% (51%ile)	43% (60%ile)	44% (61%ile)	40% (—)
Low-income population	33% (60%ile)	28% (52%ile)	41% (72%ile)	15% (29%ile)	30% (—)
Unemployment rate	3% (45%ile)	4% (55%ile)	5% (62%ile)	9% (79%ile)	6% (—)
Linguistically isolated population	1% (59%ile)	0% (56%ile)	4% (71%ile)	0% (57%ile)	5% (—)
Population with less than high school education	11% (60%ile)	8% (50%ile)	30% (91%ile)	29% (91%ile)	11% (—)
Population under 5 years of age	4% (47%ile)	9% (80%ile)	5% (54%ile)	0% (13%ile)	5% (—)
Population over 64 years of age	27% (83%ile)	17% (53%ile)	17% (55%ile)	35% (92%ile)	18% (—)
* The traffic proximity and volume indicator is a score calculated by daily traffic count divided by distance in meters to the road. The Superfund proximity, RMP proximity, and hazardous waste proximity indicators are all scores calculated by site or facility counts divided by distance in kilometers. The underground storage tank proximity indicator is the weighted count within a 1,500-foot block group. The toxic releases to air indicator is the modeled toxicity-weighted concentration. The wastewater discharge indicator is the modeled toxicity-weighted concentrations divided by distance in meters.