Oral argument — Heal Utah v. EPA

First case up today is Heal Utah v. EPA. It is docket 21-9509. And we are ready to hear the argument when counsel is ready to make it. Thank you. Good morning, Your Honors. Jenny Harbine on behalf of petitioners, I'm going to attempt to reserve three minutes of my time this morning for rebuttal. May it please the Court. Pacific corpse coal plants in Utah are some of the largest polluters in the West. Hunter Units 1 and 2 and Huntington Units 1 and 2 contribute significantly to haze pollution across the Colorado Plateau, and in particular in Utah's magnificent national parks, including Arches, Canyonlands, and Capitol Reef. This case is about EPA's decision to allow these plants to continue operating into the future without cost-effective industry standard controls to reduce their nitrogen dioxide or NOx pollution. And petitioners challenge EPA's flawed implementation of the Clean Air Act's requirement that such large old polluting sources install the best available retrofit technology to reduce such impairment. Before I turn to this BART requirement, I'll address the Court's question of whether EPA's recent Good Neighbor Rule moves this case, and the short answer is no. The Good Neighbor Rule is no replacement for the BART requirements that petitioners seek to reinstate in this case, because it establishes NOx emission limits as a backstop that are less stringent than EPA's BART requirements and would be implemented on a slower time frame. Additionally, the implementation of the new Good Neighbor Rule is uncertain, its timeline is uncertain, because Utah and Pacific Corp. and its co-owners already have sued in this circuit to challenge EPA's preceding action, which was the disapproval of state implementation plans under the Clean Air Act's Good Neighbor provisions, and those same parties have vowed to challenge EPA's rule that they issued last week. So while we believe the rule ultimately will be upheld, the time frame for its implementation could be called into question by a stay or injunctions or the like in this case has not moved. So it has no justiciability implications, does it form any part of the decisional law for the issues that you raise? It doesn't affect the judiciability at all or impact the administrative record in this case on which this case is based. What it does do, however, is highlight the need for these Hunter and Huntington units to clean up their NOx pollution, because not only do they impair visibility, which is what's at issue in this case, they impact the health of downwind air quality. And these reductions are feasible, available, and they've been a long time coming, and that brings me to BART. The Good Neighbor rule does require, would require them to install the catalytic devices on Huntington and the other one, Hunter. Potentially. That is what Pacificorp represented in its comments to EPA on the Good Neighbor rule. The emission limit is much less stringent than what this SCR technology is capable of achieving, and how that rule would ultimately be implemented in the state of Utah is a question. I just want to understand, your position is because the emission levels are less stringent under the Good Neighbor rule, that's why that does not affect the judiciability of this case? That's one reason, and of course the mootness standard is whether any effectual relief whatsoever is available. Certainly the relief we're asking for is the reinstatement of the BART requirements, which would establish, just to get in the weeds for a moment, an emission limit of .07 pounds per MMBTU NOx. The backstop emission limit that would apply not until 2030, it's phased in over time, under the Good Neighbor rule is twice that. That's assuming that you get complete relief in this case, right? Right. That it doesn't go back for findings, that it's the reinstatement of the original. Right, and provided any effectual relief is available, this case is justiciable. Starting with BART, in 2016, EPA went through the five-factor analysis required under the Clean Air Act and determined that BART for Hunter and Huntington units at issue was reflected by this SCR technology. And SCR, EPA determined, was a cost-effective and highly effective technology that would reduce NOx emissions from these plants by around 75%. The very next year, a new administration decided to reconsider that rule. And in 2020, EPA finalized a rule that replaced BART requirements with an alternative that, I'm sorry, my time appears to be flawed here. Can I pause and ask that to be reset? Please. Thank you. Okay. Thank you. In 2020, EPA replaced BART requirements with an alternative that eschewed SCR as BART and, indeed, any future emission reductions from the Hunter and Huntington BART units. And it replaced these requirements with an alternative that relied primarily on emissions reductions achieved through the closure of the carbon coal plant, also in Utah. And the closure was for reasons unrelated to regional haze. Before I get to Petitioner's chief arguments about the flawed approach here, I want to set out some important undisputed points. Let me ask you this. What is your answer to the EPA's view that the BART comparable emissions should not be reduced by 50% because of the hypothetical application of mass? What's your answer to that? They say that you can't do that because you have to compare each of them as to the baseline 2002. Yeah. As I understand EPA's argument, and your Honor, correct me if I'm misunderstanding your question, EPA argues that there should be an all or nothing approach to the emissions from this carbon coal plant. So under the BART alternative, they're eliminating all emissions, not just NOx, but also the SO2 emissions from the carbon coal plant. And then under the BART alternative, they're assuming that the plant would continue operating into the future at pre-closure emission levels. The problem with this is it assumes an impossibility. It's not an option for Pacific Corp to continue operating that plant. But their position is, I understand your position, but their position is in order to be comparing apples to apples, you have to use the 2002 baseline. And under the regulations, the BART alternative gets the benefit of zero emissions from carbon because it was separate regulation. And you have to keep the BART at the full emissions from carbon because of the baseline year 2002. That's their position, isn't it? That's their position. And let me explain what's wrong with this position and why it's not foreclosed by the rule. Petitioners don't argue that the alternative can't take credit for the elimination of emissions from the carbon coal plant due to that 2015 closure. But what the regional Hays rule provides is that for an alternative program, EPA has to determine that the emissions reductions under the alternative are surplus to those reductions resulting from measures adopted to meet the requirements. It's stated in the affirmative. They don't say you can't assume for the BART benchmark the reasonable future scenario that the plant would continue operating. The problem arises not under that test, but under the test that's set out in Section 308E2i, which requires in Step 5, or E, I believe it's labeled under the regulation, a reasonable comparison. A comparison between the emission reductions achievable through the BART benchmark and through the BART alternative. And the problem is in the BART benchmark scenario, you've set up this impossible future emissions scenario for the carbon coal plant that doesn't allow for a meaningful comparison. It's not an apples to apples comparison. It's apples to impossible. And that not only isn't required by the rule, it undermines the language and purpose of the Clean Air Act, which after all establishes as a floor for visibility improvement the level of improvement that's achievable from BART. Why isn't that a challenge to the regulation? It's not a challenge to the regulation because the regulation nowhere includes language that says you can assume or you must assume an impossibility for the BART alternative scenario. The only regulation that EPA has pointed to is 308E2i, which requires only this affirmative demonstration that the alternative program reductions are surplus to the baseline. It doesn't say, it doesn't condone this all or nothing approach in the BART benchmark scenario. So the regulation is just silent on this question? The regulation is silent on this question and for that matter so is the case law. Are there any cases involving plant closures like this one? In other words, are we the first one out of the chute on this issue? There's no case involving the situation such as this where you're relying on a plant closure where in the BART benchmark you would otherwise anticipate that the plant continues operating unlawfully. But there are cases about including non-BART emission reductions in the BART benchmark, right? There are both. Are those cases applicable here, the UARG 1 and 2? UARG 2 is the case that the respondents primarily rely on and it should not be extended to the facts here. UARG involved, and as did Yazzie from the Ninth Circuit, although not involving non-BART sources, but both of those cases involved the reduction of emissions under a BART alternative on the same sources that are subject to BART. And in the so-called UARG example, EPA determined that in fact compliance with that BART alternative, which was an earlier version of the good neighbor rule, would result in very similar controls on BART sources to BART itself. It simply wouldn't make sense to layer those programs on top of each other. It wouldn't make sense to assume that sources would comply both with the alternative and with BART in that circumstance. It shouldn't be extended to this case because compliance with both BART and this mercury and air toxics standard is eminently reasonable on separate units. It's quantifiable and in fact it's legally required. You answered to Judge Phillips when he said, you know, the regulation, he asked, the regulation answered his question and you said no. Isn't a reasonable application or interpretation of the regulation or at least an inference from it, is that it does allow this And that's certainly the way the agency has interpreted it. And as a consequence, isn't it entitled to our deference? EPA's position here is not entitled to deference to the extent that the regulation is silent. In the case of E2l4 on this point and unambiguous in terms of the test that set out in 308E2i, that test requires greater visibility improvement from the BART alternative as compared with BART. And here that simply can't be demonstrated if you set up an unreasonable comparison between BART and the BART benchmark. That's what's required here. Reasonable assumptions underlying an agency decision is just, you know, a bedrock requirement under administrative law. And EPA violated that with its unreasonable reliance on Pacific Corps modeling assumptions in this case. The second reason that EPA's approval of the BART alternative was arbitrary was because it wasn't shown to provide for greater visibility improvement as the regional HAYS rule requires. In the interest of time, I'll rely primarily on my briefs for this argument, but I do want to raise a couple of points here. So this is the issue where modeling expert Howard Gebhardt noted that the visibility benefits of the alternative, as shown by the computer model, were so exceedingly small, you know, one one-thousandth of the level of perceptibility, to be meaningless as a mathematical matter. Why isn't that a challenge to the regulation? As I noted, Your Honor, the regulation is unambiguous. All parties interpret it as requiring a greater visibility improvement. Petitioner's argument here is not that, you know, .0006 is not enough. Petitioner's argument is that that's effectively zero, that it's no visibility improvement. And no party is arguing that that is allowed under the test. When you say no visibility improvement, it sounds to me like you're saying no perceptible visibility improvement rather than no improvement. The issue here is not whether that level... Am I wrong? Yes, Your Honor. Okay, well tell me I'm wrong. The issue is not whether that level is perceptible. Of course, the perceptibility threshold is 1.0 deciviews. We're talking about one one-thousandth of that amount. And the issue is raised in our comments and in our briefs is that not only is that amount small, it's beyond the ability of the model to detect. Because of acknowledged uncertainties in the model. What is good enough for greater reasonable progress then? You say that's not, what is? And if you can't give an answer, why isn't the agency in charge of that? The agency would get deference to that point if they had provided a line drawing. If they had explained their reason for rejecting Mr. Gephardt's professional opinion that the uncertainties in the model render that very minimal visibility improvement effectively zero. But EPA was silent on that. One point I want to make that's not in the briefs is this court's 2017 Wild Earth Guardians case discusses agency deference in this context. And states that an agency has to provide evidence of sufficient volume and quality to support a choice among options. It has to actually provide data to support its decision. Particularly when it made the difference like it did here on a key issue. And here it meant the difference between requiring highly effective emission controls to reduce NOx pollution into the future or doing nothing. And the Clean Air Act requires more. Thank you. Good morning, Your Honors, and may it please the court. My name is Miranda Jensen, and I'm from the United States Department of Justice. Could you pull that microphone closer to you? There we go. I haven't got the best ears. I plan to share two minutes of respondents' time with the industry interveners. And it is my understanding that the state's counsel is present to answer questions but does not intend to present arguments today. Your Honors, EPA squarely complied with the regional Hays regulations in approving Utah's BART alternative. I'd like to highlight two points to help the court understand why EPA's action here should be upheld. First, EPA properly compared the average visibility differences between BART and the alternative when the air modeling concluded that the carbon plant's emissions reductions only were applicable in the alternative scenario. Second, EPA employed its technical expertise and reasonably relied on modeling results that showed the BART alternative demonstrated a greater overall visibility improvement as compared to BART. Before going into my first point, I just want to take a minute to clarify the regulatory framework here. This case involves a comparison between two policy scenarios to address regional Hays and Utah's SIP. The first option is BART, which here is based on upgraded combustion controls at BART units at Hunter and Huntington power plants. The second option is Utah's proposed BART alternative, which includes, or BART also includes the SCR controls as well. And then Utah's proposed BART alternative includes the upgraded controls at Hunter and Huntington BART units, as well as an additional Hunter unit, unit three. I'm sorry to interrupt you. I'm struggling with your position on, you know, there are briefs you described that the BART alternative achieved greater reasonable progress because the visibility improvement was small, but sufficed. Was it really anything? Could you just speak to the petitioner's point that it was just the modeling wasn't even capable of detecting a difference, that it was so statistically insignificant here? You describe it as small, but was it really anything at all? So the model improvement on the 20% worse visibility days, which is part of the dispersion modeling test, showed a 5,800 thousands of a deci view improvement. And while small, that is an improvement over BART. And to get to your point about the model, you know, EPA considered the petitioner's comments and explained why in EPA's view it determined the modeling results to be reliable here. For instance, EPA explained... It explained in the rule or did it explain in the brief? In the regulation and in the final rule, as well as in the proposed rule and the response to comments. It went over me if there was an explanation. Well, for instance, EPA explained that the CAMIX modeling, or the CAMIX modeling, which is a photochemical grid model that Pacific Core used, is a highly sophisticated model. And it has been approved for use in modeling long-range transport. The model has uncertainties, though. Correct, Your Honor. Yes, every model has uncertainties, and EPA acknowledged that. But it stated that, you know, to minimize those uncertainties, EPA has developed guidance on the model's use and evaluation. And in this case, it closely reviewed Pacific Core's modeling to ensure the results were reliable for the comparative analysis that's required here. Is there been a margin of error? So the petitioners raised a question about the margin of error, but the margin of error isn't relevant to this analysis. It's just, it may be a familiar concept when talking about data analysis, but that's only one way to assure results are reliable. And it's not relevant to how EPA uses models. Well, what if EPA said, yes, it is within a margin of error, and so we can't be sure whether or not .00058 really is a greater improvement? And yeah, it rules as it does. Why wouldn't that be arbitrary and capricious? So I don't think, those aren't the facts here, because EPA looked at the .00058 of a deciview improvement, and in its technical judgment, determined that their modeling was reliable, and that that did constitute an improvement here. And so this is a very technical analysis. Are you contending that we should defer to the EPA's decision about modeling assumptions based on technical deference? Yes, Your Honor. So is that your primary argument? I think it's important to think about the improvement here as a whole, because the petitioners seem to take issue with whether this amount qualifies as an improvement under the regional HAYS regulation. The regulation says overall improvement, right? Correct, Your Honor. So what does overall mean? It seems that you're interpreting overall to mean anything, any. So if we look at the regulation at E3, it states, there's two prongs. The first one is visibility does not decline in any Class I area. And a Class I area is the parks and wilderness areas that are protected and included in this analysis. And then the second prong is there's an overall improvement in visibility over all affected Class I areas at the end there. And so that overall improvement in visibility is looking at the overall visibility improvements across those Class I areas, and not a site-specific. What does overall mean? I guess you're talking about it in terms of geographic scope, not amount. We're not talking about it. See, I'm reading overall to mean any. Is that wrong? I think any is correct, yes. Any amount of improvement is enough. Okay, that's your position. Yes. Now, correct me if I'm wrong, but I understood the EPA's position is that the concept of margin of error is irrelevant because you are using, in a comparison type model like this, it's irrelevant. Isn't that their position? So EPA doesn't use a margin of error in its modeling at all, but in this case, the uncertainties that you had addressed, EPA determined that the modeling was reasonable for a particular purpose. Wait a minute. I understood them to say, is it margin of error in this comparison of BART alternative to BART benchmark, it's irrelevant because all the input is the same on each side of the balance. Isn't that their position? That's the correct kind of reasoning, but the margin of error concept is not used in EPA's modeling guidance or in its use of modeling. Did you address margin of error at the agency level? That is why it was irrelevant? So the petitioner's comments don't include a margin of error discussion, and that wasn't something that as a concept was addressed specifically. So you're saying there's an appellate waiver? I'm not sure, Your Honor. You're saying that they didn't use the expression margin of error at the agency. I understood, so you're saying they can't use it now. So, correct, to the extent that they didn't raise a question about the margin of error, that wouldn't be appropriate for this court to review in the first instance. Let me ask you this. As I understand it, that the BART benchmark is heavy on suppression of emissions, NOx emissions, and the alternative is heavy on sulfur dioxide. Correct? That was the modeling headfirst. It is explained in the proposed rule at 3567 to 68. Here's my question. If there is an input error or an uncertainty that would benefit the alternative because of its focus on sulfur dioxide, then you don't have, you're not comparing two things and you need to have a margin of error on that. Why is that wrong? So the margin of error concept, again, isn't part of this analysis, but EPA did, Pacific Corps submitted its modeling protocol to EPA for review, and EPA, in response to this, this bias for certain suggested revisions and made recommendations to improve the modeling. And Pacific Corps adopted all of those recommendations and made those adjustments to correct for those biases that you highlighted. And so EPA then concluded that the modeling was appropriate and reliable here for the purpose of this comparative analysis because any additional uncertainties like wind speed or atmospheric turbulence apply equally to both scenarios, so they're not likely to favor one scenario over another. And so then I also would like to discuss briefly why it was appropriate for excluding the carbon plant emissions reductions from the BART benchmark scenario because that analysis, EPA properly compared those average visibility differences between BART and the alternative to achieve that apples-to-apples comparison that Judge Murphy alluded to earlier. In the first instance, the court doesn't need to go further than the UARC cases because those cases address the same issue that the petitioners raised here, which is whether reductions from a separate Clean Air Act program included in a BART alternative should be reflected in the BART benchmark modeling. Could you respond to the petitioners' contention that if we relied on those cases, we would be extending them to the situation here? I assume your position is it would not be. Why not? Because the decision there reflects the regulation itself at 3083, which requires this comparative analysis. And so the court, in the UARC cases, the D.C. Circuit, properly concluded that petitioners' challenge, which was to the way that the scenarios were modeled, was barred by this Clean Air Act 60-day statute of limitations because it was attacking the validity of 308E3. And here, petitioners likewise claim that the CARBON shut down to comply with a separate Clean Air Act program. And as a result, the BART benchmark should take into account at least some reductions to account for CARBON's compliance with the MATS rule. But it's clear from the regulations that the BART benchmark should not include reductions from CARBON because CARBON's reductions are appropriately included in the alternative. And what the alternative is doing, or what the modeling is doing, is taking both scenarios and attributing the BART reductions to the BART benchmark and the alternative reductions to the alternative to create a comparative tool and assess which one achieves greater visibility improvement. So your position is that the case applies because it's the same reg, and that's it. There's no factual distinction here. I mean, the program is a little bit different. But that doesn't matter. Correct, that doesn't matter because what's important is the relevant timeframe under E2, Roman NETS 3 and 4, which allows the state to take credit in a BART alternative for any reductions between the baseline date of 2002 and the end of the first implementation period in 2018. And the closure that occurred here was the result of the MATS rule in 2012. And then the closure occurred in 2015, so that fits squarely within that timeframe. And for these reasons, the court should deny the petition for review. Thank you. What about the good neighbor rule? Do you have a different position? Yes. EPA's position is that it does not affect the justiciability of this case. Does not? Does not, correct. For the same reasons as your opposing counsel said? No. In our view, these two rules address separate Clean Act requirements. The good neighbor provision implements 42 U.S.C. Section 7410, A2, D, I, Roman NET 1 for the 2015 Ozone NETS. And the BART alternative is implementing the visibility requirements under Clean Act Section 7419, A and B. And there's also a lot of differences between how these programs work. So just one clarifying point. The good neighbor plan addresses interstate transport of ozone and its precursors through an emissions trading program that includes many states and allows facilities to trade allowances. And although the allowances budget in future years contemplate units at power plants will implement SCR technology, the rule itself does not require SCR installation. So it's not a guarantee that the BART units at Hunter and Huntington will install SCR. There's also other distinctions between the programs as well. Well, are the emission levels more relaxed for these two plans under the good neighbor rule than they are under the Hays rule that we're talking about here? So I'm not intimately familiar with the good neighbor provision. I did consult with EPA to respond to the justiciability questions. Is anybody here on your side going to address that? Well then, if somebody is designated to address that, let's have the person who knows about it. I think the key point is that the two programs address different obligations under the Clean Air Act. And so a level of reduction requirement really isn't relevant to whether this case is still justiciable. Do you have a further question that I can try to clarify? Well, Ms. Harbine said one of the reasons is because there are more relaxed emission levels, I think, under the good neighbor rule than there are under this rule. And that's why this remains justiciable. And I'm not sure I know your answer to that. Do you agree? I don't know that I agree with that position necessarily. You don't know whether you agree or disagree, right? But I agree that it's not justiciable for different reasons, I think. And that key reason is just because the programs address different requirements and that the good neighbor plan does not, while it contemplates the installation of SCR, it does not specifically require it because it's a trading program that allows for flexibility. Does the good neighbor rule have any effect on the second implementation phase? The good neighbor rule? I guess maybe let me ask you a different question. Is that okay if we're going over? Thank you. I'm trying to understand if we need to be thinking about or how we should be thinking about the implementation periods as they affect the petition for review. So the second implementation period, that does not negate the requirement to establish BART or a BART alternative in the first implementation period, because BART is not part of the analysis in the second implementation period. That's only done during the first one. So there's no justiciability question there. It's my understanding that the state has submitted its second implementation period SIP to EPA and that that proposal is pending. But there's no incongruity when deciding this case and that, because this case only involves the first implementation period. Correct, this case is the first implementation period. Thank you for your argument. Thank you. Good morning, and may it please the Court. My name is Carroll McGuffey. I represent Pacificorps. I recognize the respondent's time has gone a little bit over. I would ask the Court's discretion to allow me to answer a few of the questions that the Court has raised. I'll make it very brief. Just to start off, petitioners have said that the rule is silent about how to calculate the BART benchmark, that benchmark that the alternative must be better than. The rule is not silent. The rule was revised in 2005 by EPA to make clear how to make that calculation. That calculation does not allow for accounting reductions at non-BART sources, and that's in 51-308-2IC. It tells EPA exactly how to calculate the BART benchmark. Any reductions at the carbon plant cannot be included there because it's not a BART source. And what's important here is that test is a comparative test, and so it's the relative difference that matters. This helps answer the questions I've heard about margin of error. The comparative difference between the BART benchmark and the alternative is significant. You're talking about an improvement under the alternative. That is a 57 percent improvement over baseline. The 2002 baseline, Judge Murphy, that you mentioned, is required. The alternative is 57 percent better, and that is better than the BART benchmark, which is only 47 percent improvement. So both of these plans are significant improvements. But the improvement on the 20 worst days is still, remains, 58-10,000, correct? Yes, Your Honor, and I recognize that the petitioners have focused on the worst days, but their own comments say that the worst days are not a good metric. In JA856, their comments say that the worst days are dominated by wildfire, and they actually obscure the improvements of anthropogenic sources like my client's power plants. Well, even the 20 best days are not that significant. Well, the relative difference is significant, Your Honor. The difference is 19 percent better under the alternative. That's found in JA34. JA34 is part of the final rule. It contains a very helpful chart that shows all of the improvements, and you can see that the alternative chosen by Utah is significantly better from a relative perspective. Yes, all the numbers are small, but there's a reason for that. The impacts of these sources, the total impact is in the thousands. So when you're comparing two options for reducing those impacts by about half, yes, of course, those differences are going to be small, and this is not a case where there is a concern over margin of error. If you look at prior precedent in PCA, for instance, out of the Ninth Circuit in 2014, there the petitioners made a specific comment, margin of error is a specific number, and the result was below that. That is not a comment that was made in this case. Petitioners have not cited a specific margin of error. In fact, that phrase does not appear anywhere in their comments. They only talk about uncertainty, and EPA answered those questions. Thank you. Thank you. This feels luxurious, and I'll try to use my time wisely here. First, I want to go back to the question of what overall visibility improvement means and whether that's satisfied in this case, and EPA represented that any amount of improvement satisfies that test. Any amount is more than zero, and that's kind of beside the point. The argument here is whether we're talking about six ten-thousandths or five one-thousandths in the case of the 20% best days, whether that amount is more than zero. This question is one that certainly petitioners exhausted. I'll point to page 888 of the appendix where a petitioner's expert noted that the modeling uncertainty likely exceeds the magnitude of the modeled concentration, and on the preceding page where Mr. Kephart said this value was effectively zero. This is exhausted as demonstrated by the fact that EPA made attempts to respond to it in its response to comments. So, counsel, I think I'm struggling to understand your argument a little bit. Is it that you're not quibbling with the fact that the regulation doesn't provide for a quantum threshold of any kind, right? You're saying that it was arbitrary and capricious for the agency to have construed this as anything but zero. In light of the record evidence, yes. What was required of the agency, if it were to actually justify its decision in light of the expert comments in the record, what was required of the agency is what this court said in that 2017 Wild Earth Guardians case at page 1235, which is that the agency had to provide evidence of sufficient volume and quality that it supports a choice among options. And here EPA responded to the point that petitioners didn't make about being able to rely on levels of visibility improvement that are below the level of perceptibility. They made this point that Judge Murphy was getting at about how certain errors cancel each other out between the alternative and the benchmark. But what they didn't respond to is the numerous concerns that Mr. Gephardt raised on page 887 of the appendix that noted how these uncertainties may play out differently between the BART benchmark and the BART alternative. In response to these concerns, all EPA said, and this is on page 955 of the appendix, is that it had confidence in the results. The same thing that EPA's counsel referred to here, that it reviewed the modeling protocol and it remained confident in the results. That's not enough to satisfy the Wild Earth Guardians standard. And EPA's repeated assertion here today that the margin of error consideration was not part of its analysis simply affirms petitioner's point, which is that at a minimum, EPA was required to grapple with this issue, to explain its decision based on the evidence in the record, and it failed to do so. Was that issue raised at the agency level? It was absolutely raised at the agency level. Was it in the Gephardt? In the Gephardt report. Okay, we can find it. Okay, and again, it's 887 and 888 of the appendix. On the point that Pacific Corp's counsel was making, arguing that the rule speaks to what must be included in the BART benchmark, counsel pointed to 308E2c. I'm sorry for the recitation of these belabored citations. But that is the step in the alternative evaluation that requires EPA to determine BART for BART sources. EPA did that in 2016. There's no question what BART is. But that regulation doesn't define the entirety of the BART benchmark. Respondents simply read that constraint into the regional HAYS rule, and it's not there. Instead, it's subsection E where this concern arises. And subsection E is the one that requires a comparison between the BART alternative and the BART benchmark. Now, it doesn't say it requires a rational comparison. That requirement comes from the Administrative Procedure Act. But at a minimum, if you're going to satisfy the Clean Air Act's requirement to achieve a minimum level of visibility improvement reflective of BART on these sources, you have to set up a reasonable comparison. Otherwise, not only are you violating the Administrative Procedure Act, violating the regional HAYS rule, you're violating Congress's mandate in the Clean Air Act. Here, remand and vacater of this flawed rule are essential. As EPA noted, BART is a one-time requirement. It applies under the regional HAYS program only in the first implementation period. Not the good neighbor rule nor any other provision of the Clean Air Act has this technology-forcing visibility improvement mandate that's contained in the BART requirement. And vacater of this rule and reinstatement of the BART requirements are essential to vindicate the Clean Air Act's important purpose of cleaning up coal plant pollution to clear the air over our national parks and wilderness areas. Thank you. Thank you for your arguments and helpful briefing, counsel. The case is submitted and counsel are excused.