SOUTHWESTERN ELEC. POWER CO. v. U.S. E.P.A.

STUART KYLE DUNCAN, Circuit Judge:

Steam-electric power plants generate most of the electricity used in our nation and, sadly, an unhealthy share of the pollution discharged into our nation's waters. To control this pollution, the Clean Water Act, 33 U.S.C. § 1251 et seq., empowers the Environmental Protection Agency to promulgate and enforce rules known as "effluent limitation guidelines" or "ELGs." Id. §§ 1311, 1314, 1362(11). For quite some time, ELGs for steam-electric power plants have been, in EPA's words, "out of date." 80 Fed. Reg. 67,838. That is a charitable understatement. The last time these guidelines were updated was during the second year of President Reagan's first term, the same year that saw the release of the first CD player, the Sony Watchman pocket television, and the Commodore 64 home computer. In other words, 1982. See id. (noting ELGs were "promulgated and revised in 1974, 1977, and 1982"). The guidelines from that bygone era were based on "surface impoundments," which are essentially pits where wastewater sits, solids (sometimes) settle out, and toxins leach into groundwater. Id. at 67,840, 67,851. Impoundments, EPA tells us, have been "largely ineffective at controlling discharges of toxic pollutants and nutrients." Id. at 67,840. Consequently, in 2005 the agency began a multi-year study to bring the steam-electric ELGs into the 21st century. Id. at 67,841.

In November 2015, EPA unveiled the final rule: the "Effluent Limitations Guidelines and Standards for the Steam Electric Power Generating Point Source Category," 80 Fed. Reg. 67,838 (Nov. 3, 2015). The rule updates guidelines for six of the wastestreams that issue from plants and foul our waters. Importantly, the Clean Water Act requires setting new ELGs based on the "Best Available Technology Economically Available" or "BAT." 33 U.S.C. § 1314(b)(2)(B). BAT is the gold standard for controlling water pollution from existing sources. By requiring BAT, the Act forces implementation of increasingly stringent pollution control methods. See NRDC v. EPA, 822 F.2d 104, 123 (D.C. Cir. 1987) (describing the Act as "technology-forcing").

We consider a challenge to the final rule brought by various environmental petitioners. They target two discrete parts of the rule: the new ELGs for "legacy wastewater" (wastewater from five of the six streams generated before a specific date) and for "combustion residual leachate" (liquid that percolates through landfills and impoundments). These two categories account for massive amounts of water pollution. For instance, leachate alone would qualify as the 18th-largest source of water pollution in the nation, producing more toxic-weighted pound equivalents than the entire coal mining industry. The environmental petitioners' basic complaint is that EPA set an unlawful BAT for these two categories. Whereas the BAT for the other streams adopts modern technologies, they claim the agency arbitrarily set BAT for legacy wastewater and leachate using the same archaic technology in place since 1982—namely, impoundments. It was as if Apple unveiled the new iMac, and it was a Commodore 64.

The environmental petitioners challenge those portions of the rule under the Administrative Procedure Act and the well-worn Chevron test governing review of agency action. See Chevron, USA, Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). For the reasons discussed below, we agree that the portions of the rule regulating legacy wastewater and combustion residual leachate are unlawful. Accordingly, we VACATE those portions of the rule and REMAND to the agency for reconsideration.

I. BACKGROUND

A. The Clean Water Act

The Clean Water Act ("CWA" or "Act"), 86 Stat. 833, as amended, 33 U.S.C. § 1251 et seq., was enacted over President Nixon's veto in 1972. See Train v. City of New York, 420 U.S. 35, 40, 95 S.Ct. 839, 43 L.Ed.2d 1 (1975). Few laws have shouldered a weightier burden—namely, "to restore and maintain the chemical, physical, and biological integrity of the Nation's waters." 33 U.S.C. § 1251(a); see also, e.g., City of Milwaukee v. States of Illinois and Michigan, 451 U.S. 304, 318, 101 S.Ct. 1784, 68 L.Ed.2d 114 (1981) ("Congress' intent in enacting [the CWA] was clearly to establish an all-encompassing program of water pollution regulation."); Am. Petroleum Inst. v. EPA, 661 F.2d 340, 343-44 (5th Cir. Unit A Nov. 13, 1981) ("API I") (noting CWA's "ambitious purpose"). To that end, the Act makes "unlawful" the "discharge of any pollutant by any person" into the nation's "navigable waters," unless otherwise permitted. 33 U.S.C. §§ 1311(a), 1362(7), (12).1

We have previously detailed the Act's "distinct, though interlocking, regulatory schemes." Chem. Mfrs. Ass'n v. EPA, 870 F.2d 177, 195 (5th Cir. 1989) ("CMA"), clarified on reh'g, 885 F.2d 253.2 Here we focus on one of the Act's key regulatory tools: "effluent limitation guidelines" ("ELGs" or "guidelines"), which are nationwide standards set by the EPA Administrator to govern pollutant discharges from point sources. See 33 U.S.C. § 1314(b) (authorizing Administrator to set "effluent limitation guidelines" for "classes and categories of point sources"); Tex. Oil & Gas, 161 F.3d at 927 ("ELGs are the rulemaking device prescribed by the CWA to set national effluent limitations for categories and subcategories of point sources").3

The Act requires ELGs to be based on technological feasibility rather than on water quality. Id. at 927 (citing E.I. du Pont de Nemours & Co. v. Train, 430 U.S. 112, 130-31, 97 S.Ct. 965, 51 L.Ed.2d 204 (1977); API I, 661 F.2d at 343-44). That is, the Administrator must "require industry, regardless of a discharge's effect on water quality, to employ defined levels of technology to meet effluent limitations." API I, 661 F.2d at 344; see also Tex. Oil & Gas, 161 F.3d at 927 (ELGs are "technology-based rather than harm-based" insofar as they "reflect the capabilities of available pollution control technologies to prevent or limit different discharges rather than the impact that those discharges have on the waters"). The Act therefore mandates a system in which, as available pollution-control technology advances, pollution-discharge limits will tighten. See, e.g., Nat'l Crushed Stone, 449 U.S. at 69, 101 S.Ct. 295 (the Act "provides for increasingly stringent effluent limitations") (citing 33 U.S.C. § 1311(b)); CMA, 870 F.2d at 196 (the Act requires compliance with "technology-based pollutant-effluent limitations that, in time, will become more stringent") (citing 33 U.S.C. §§ 1311(b), 1314(b)). The D.C. Circuit accurately described this aspect of the Act's scheme as "technology-forcing," meaning it seeks to "press development of new, more efficient and effective [pollution-control] technologies." NRDC v. EPA, 822 F.2d 104, 123 (D.C. Cir. 1987) ("NRDC I"); see also, e.g., NRDC v. EPA, 808 F.3d 556, 563-64 (2nd Cir. 2015) ("NRDC II") (describing ELG scheme as "technology-forcing, meaning it should force agencies and permit applicants to adopt technologies that achieve the greatest reductions in pollution") (citing NRDC I).4

The Act prescribes various technological standards to be used in setting effluent limitations. Two are relevant here: "best practicable control technology currently available" ("BPT") and "best available technology economically achievable" ("BAT"). Compare 33 U.S.C. §§ 1311(b)(1)(A); 1314(b)(1)(B) (BPT), with id. §§ 1311(b)(2)(A); 1314(b)(2)(B) (BAT). The less stringent of these two standards is BPT, which the Supreme Court has described as only "a first step toward [the Act's] goal." Nat'l Crushed Stone, 449 U.S. at 75 n.14, 101 S.Ct. 295; see also, e.g., BP Explor. & Oil, Inc. v. EPA, 66 F.3d 784, 789 (6th Cir. 1995) (describing BPT as "the first stage of pollutant reduction"). BPT applied to limitations on direct discharges of pollutants during an interim period (originally slated to end in 1984 but later extended to 1989). 33 U.S.C. § 1311(b)(1)(A); Tex. Oil & Gas, 161 F.3d at 927-28; CMA, 870 F.2d at 196. Instead of defining BPT, the Act lists various factors the Administrator must consider in determining it—including an explicit cost/benefit analysis: "the total cost of application of technology in relation to the effluent reduction benefits to be achieved from such application." 33 U.S.C. § 1314(b)(1)(B).5 We have explained that "BPT limitations are intended to represent the average of the best levels of performance by existing plants of various sizes, ages, and unit processes within the category or subcategory for control of conventional pollutants." CMA, 870 F.2d at 203 (citing 52 Fed. Reg. 42,525); see also, e.g., Nat'l Crushed Stone, 449 U.S. at 75-76, 101 S.Ct. 295 (discussing BPT).

The stricter of the two standards is BAT, which has applied to existing, direct discharges of toxic and non-conventional pollutants since March 31, 1989. See 33 U.S.C. §§ 1311(b)(2)(A); 1314(b)(2)(A); Tex. Oil & Gas, 161 F.3d at 927-28; see also BP Explor., 66 F.3d at 790 (describing BAT as "the second stage" of pollutant reduction). When pollutants are regulated under this standard, the EPA "must set discharge limits that reflect the amount of pollutant that would be discharged by a point source employing the best available technology that the EPA determines to be economically feasible across the category or subcategory as a whole." Tex. Oil & Gas, 161 F.3d at 928. We have held that BAT limitations must "be based on the performance of the single best-performing plant in an industrial field." CMA, 870 F.2d at 226. In describing the relationship between BAT and BPT, the Supreme Court has explained that a BAT must achieve "reasonable further progress" towards the Act's goal of eliminating pollution, and BPT serves as the "prior standard" for measuring that progress. See Nat'l Crushed Stone, 449 U.S. at 75, 101 S.Ct. 295 (explaining that "BPT serves as the prior standard with respect to BAT[`s]" reasonable further progress requirement). As with BPT, the Act lists factors the Administrator must consider in determining BAT. 33 U.S.C. § 1314(b)(2)(B).6 The Administrator has "considerable discretion" in weighing those factors. Tex. Oil & Gas, 161 F.3d at 928 (citation omitted). Unlike BPT, however, the BAT factors omit a cost/benefit analysis and replace it with a requirement to consider only "the cost of achieving such effluent reduction." Id; see also, e.g., Nat'l Crushed Stone, 449 U.S. at 71, 101 S.Ct. 295 (BPT and BAT factors are "similar ... except that in assessing BAT total cost is no longer to be considered in comparison to effluent reduction benefits"). Indeed, the Supreme Court has explained that, unlike BAT, "BPT limitations do not require an industrial category to commit the maximum economic resources to pollution control, even if affordable." Nat'l Crushed Stone, 449 U.S. at 75, 101 S.Ct. 295.7

B. The Final Rule

The rule at issue in this case regulates effluent discharges from steam-electric power plants. Those plants burn nuclear or fossil fuels to heat water in boilers, generating steam that drives turbines connected to electric generators. 80 Fed. Reg. 67,839 n.1. This process produces something nearly everyone regards as good: electricity. Indeed, the plants regulated by the rule provide most of the electricity annually produced in the United States. But the process also produces something everyone regards as bad: pollution. According to EPA, discharges from these plants account for "about 30 percent of all toxic pollutants discharged into surface waters by all industrial categories regulated under the CWA." Id. at 67,839-40; see also, e.g., Michigan v. EPA, ___ U.S. ___, 135 S.Ct. 2699, 2705, 192 L.Ed.2d 674 (2015) (addressing regulation of air pollution from power plants under the Clean Air Act); ConocoPhillips Co. v. EPA, 612 F.3d 822, 826 (5th Cir. 2010) (addressing regulation of cooling water systems at power plants). For instance, power plant discharges contain toxic metals such as mercury, arsenic, lead, and selenium, which bioaccumulate in fish, accumulate in lake and reservoir sediment, and pollute drinking water supplies. People who eat the tainted fish or drink the tainted water can suffer negative health consequences such as cancer, cardiovascular disease, neurological disorders, kidney and liver damage, and lowered IQs (in children). Id. at 67,840.

EPA first promulgated and then revised ELGs for steam-electric power plants in 1974, 1977, and 1982. See id.; see also 39 Fed. Reg. 36,186 (Oct. 8, 1974); 42 Fed. Reg. 15,690 (Mar. 23, 1977); 47 Fed. Reg. 52,290 (Nov. 19, 1982). Those guidelines are now, in the agency's words, "out of date," because "[t]hey do not adequately control the pollutants (toxic metals and other[s]) discharged by this industry, nor do they reflect relevant process and technology advances that have occurred in the last 30-plus years." 80 Fed. Reg. 67,840. The old rules and the processes they regulated are relics of the past:

Id. ("Surface impoundments" are ponds designed to allow particulates to settle out of wastewater by force of gravity. See infra.) Happily, though, EPA reports that, "[i]n the several decades since the steam electric ELGs were last revised," technologies that are more effective, "affordable," and "widely available" have "increasingly been used at plants." Id. Thus, EPA began a new rulemaking to update power plant ELGs. The agency conducted a detailed industry study8 from 2006-2009 and on June 7, 2013 issued a proposed rule, 78 Fed. Reg. 34,432, 34,439, that generated over 200,000 comments. 80 Fed. Reg. 67,844. On November 3, 2015 the agency issued a final rule entitled "Effluent Limitations Guidelines and Standards for the Steam Electric Power Generating Point Source Category" ("final rule" or "rule"). 80 Fed. Reg. 67,838.

The rule addresses these six streams produced by power plants:

80 Fed. Reg. 67,846-47.9 The rule treats another category ("legacy" wastewater), which is a subset of five other streams. Infra I.B.1.10 This diagram illustrates how such streams are produced:

When EPA originally regulated steam-electric effluents in the 1970s and 1980s, it did so under the less-stringent BPT standard, see supra I.A, and set BPT for bottom ash transport water and leachate as surface impoundments. See 80 Fed. Reg. 67,848-49. Surface impoundments, or "ash ponds," are essentially watery pits that "rely on gravity to remove particulates from wastewater" and were "the technology basis for the previously promulgated BPT effluent limitations for low volume waste sources." Id. at 67,840, 67,851. As the new rule describes, however, the ensuing three decades have rendered that BPT standard "out of date," because it "do[es] not adequately control the pollutants (toxic metals and other[s]) discharged by this industry, nor do[es] [it] reflect relevant process and technology advances that have occurred in the last 30-plus years." Id. at 67,840. Moreover, the Act required that the new guidelines for existing direct11 discharges conform to the stricter BAT standard. See 33 U.S.C. § 1311(b)(2)(A), 80 Fed. Reg. 67,848-49; see supra I.A.

EPA thus considered more advanced control methods, which it notes are "affordable technologies that are widely available and already in place at some plants." 80 Fed. Reg. 67,840. The agency describes those methods as follows:

From those options EPA selected the following technologies as BAT for the various wastestreams:

                                   Technology basis for the main
    Wastestreams               BAT/NSPS/PSES/PSNS regulatory options

FGD Wastewater                 Chemical Precipitation + Biological Treatment

Fly Ash Transport Water        Dry handling

Bottom Ash Transport Water     Dry handling/Closed loop

FGMC Wastewater                Dry handling

Gasification Wastewater        Evaporation

Leachate                       Impoundment (Equal to BPT)

80 Fed. Reg. 67848-49 (adapted from Table VIII-1-FINAL RULE: STEAM ELECTRIC MAIN REGULATORY OPTIONS). As shown, the rule set more advanced technologies as BAT for five of the six wastestreams. See also id. at 67,850, 67,852, 67,853 (explaining selection for each stream). For leachate and "legacy" wastewater, however, the rule selected "impoundment" as BAT, the same technology set as BPT in 1982. Id. at 67,854. Our focus is on the rule's treatment of those streams, and so we provide additional detail below.

1. Legacy Wastewater

Legacy wastewater is not a distinct type of wastestream. Instead, as the final rule explains, the term describes wastewater from five of the streams (FGD, fly ash, bottom ash, FGMC, and gasification wastewater) that is "generated prior to" a future date. 80 Fed. Reg. at 67,854. That date, which is determined by the permitting authority, is required to be "as soon as possible beginning November 1, 2020 but no later than December 31, 2023." 82 Fed. Reg. 43,496. Wastewater from streams generated before that date is denominated "legacy" wastewater and is not subject to the stricter BAT applicable to those streams. Id. Instead, the BAT for legacy wastewater is "equal to the previously promulgated BPT regulations" in effect since 1982—namely, impoundments. Id. This means that legacy wastewater is allowed by the final rule to contain the same quantity of toxic pollutants allowed since 1982. See id. (setting BAT for legacy wastewater "equal to the previously promulgated BPT limitations on [total suspended solids] in the discharge of fly ash transport water, bottom ash transport water, and low volume waste sources"); see also EPA Study Report, EPA 821-R-09-008, at 5-20 (table listing pollutant concentrations at several individual impoundments studied during the rulemaking process). The "legacy" category will thus encompass a massive amount of wastewater from the five composite streams. For instance, according to the EPA's Study Report, in 2008 alone the average plant produced over 2.7 billion gallons of fly ash transport water per year, as well as over 1.1 billion gallons of bottom ash transport water. Id. at 5-6, 5-7.

The rule imposes much more stringent limits on wastewater from these same streams generated after the date to be set by the permitting authority (again, between November 1, 2020 and December 31, 2023). For instance, EPA found that a combination of chemical precipitation and biological treatment was the BAT for treating pollution from non-legacy FGD wastewater, and that "dry handling" (a technique for disposing of fly ash and bottom ash without adding water) was the BAT for non-legacy ash wastestreams. Id. at 67,850-53. These technologies are significantly newer than surface impoundments, and EPA concluded they were the superior option for treating pollution in non-legacy wastewater. Id.

The rule accounts for the discrepancy between legacy and non-legacy wastewater regulations in various ways. For instance, it explains that legacy wastewater "already exists in wet form" and would thus not be amenable to dry handling, and also that EPA lacked data on whether legacy wastewater could be "reliably incorporated" into a closed-loop process "given the variation in operating practices among surface impoundments containing legacy wastewater." Id. at 67,854-55. The rule also asserts that EPA lacked sufficient data to determine whether chemical or biological treatment would be effective on legacy wastewater. Legacy wastewater, the agency explained, is often "commingled" —meaning different streams are mixed together in an impoundment—making testing and data collection difficult. Id. at 67,855. For instance, commingling may result in varying the concentration and "flow rate" of pollutants in an impoundment. Id. The rule acknowledges that multiple plants are in fact using chemical precipitation to treat commingled wastewater, but it nonetheless asserts that EPA lacks the requisite data from those plants. Id. at 67,855 n.29. Finally, the rule also acknowledges that a few plants discharge from impoundments containing non-commingled FGD legacy wastewater, but it nonetheless declines to establish a stricter BAT for that stream as well. Id. at 67,855. The rule explains that, in the agency's view, imposing the stricter technologies even on non-commingled legacy wastewater would create bad "incentives"—for instance, encouraging plants to begin commingling FGD with other wastewaters or to release FGD wastewater from impoundments on an "accelerated schedule" prior to the compliance date. Id.

2. Leachate

The final rule describes leachate as follows:

80 Fed. Reg. 67,847. Where leachate occurs in a lined landfill or impoundment, it is typically collected and transported to an impoundment, where it is either "discharge[d]... directly to receiving waters" or recycled to another impoundment prior to discharge. Id. Unlined landfills or impoundments simply "allow the leachate to potentially migrate to nearby ground waters, drinking water wells, or surface waters." Id. The rule explains that "surface impoundments are the most widely used systems to treat ... leachate." Id. Elsewhere, the rule acknowledges that "[g]round water contamination from surface impoundments" containing power plant wastewater "threatens drinking water, as evidenced by more than 30 documented cases." Id. at 67,840; see also EPA Study Report, EPA 821-R-09-008, at 3-24 (landfill leachate diagram). The EPA study detailed the size of leachate pollution: Given plants using current technologies (mostly surface impoundments), leachate pollution amounts to 70,300 toxic-weighted pound equivalents per year. See Technical Development Document ("TDD"), EPA-821-R-15-007, at 10-39. Leachate thus accounts for more equivalent pollution than the entire coal mining industry. Id.; Annual Effluent Guidelines Review Report, EPA-821-R-16-002, at 2-26 (listing pollution from other industries).

The final rule sets BAT for leachate equal to the previous BPT standard established in 1982. Id. at 67,854. The agency offers two primary justifications for its decision not to regulate leachate with any of the more advanced control technologies now available. First, the rule explains that EPA called for comments on leachate regulation during notice-and-comment rulemaking, but that "[c]ommenters did not provide information that the EPA could use to establish BAT limitations" for leachate. Id. at 67,854. Second, the rule asserts that leachate forms "a very small portion of the pollutants discharged collectively by all steam power plants." Id. The agency reasons that, because the new BAT limits established for wastewater from other streams will substantially curtail total power plant pollution, the new rule "represents reasonable further progress toward the CWA's goals" even without establishing any stricter controls on leachate. Id.

II. PROCEDURAL HISTORY

Four separate lawsuits challenging the final rule were originally brought in the Second, Fifth, Eighth, and Ninth Circuits.12 Different groups of petitioners challenged different parts of the rule. Various power companies ("Industry Petitioners") challenged the regulation of non-legacy FGD and gasification wastewater.13 Two water company associations ("Water Company Petitioners"), challenged the non-legacy FGD wastewater regulation.14 Finally, various environmental groups ("Environmental Petitioners" or "petitioners") challenged the regulation of legacy wastewater and leachate.15 The four cases were consolidated by the United States Judicial Panel on Multidistrict Litigation and randomly assigned to our court.16 The Utility Water Act Group ("UWAG") has since intervened to defend those portions of the rule challenged by the Environmental Petitioners.

In August 2017, we granted EPA's motion to sever and hold in abeyance the Industry Petitioners' and Water Company Petitioners' challenges to the final rule. In September 2017, EPA announced it would reconsider the rule's regulations concerning non-legacy FGD and bottom ash transport water. See 82 Fed. Reg. 43,494.17 As a result of these procedural developments, the challenges to the final rule raised by the Industry Petitioners and the Water Company Petitioners are not before us. We address only the challenges brought by the Environmental Petitioners.

III. STANDARD OF REVIEW

The Environmental Petitioners challenge the legacy wastewater regulation under the Administrative Procedure Act ("APA"). As relevant here, a court "shall... hold unlawful and set aside" agency action under the APA if it finds such action was "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." 5 U.S.C. § 706(2)(A). Under this "highly deferential" standard, Avoyelles Sportsmen's League, Inc. v. Marsh, 715 F.2d 897, 904 (5th Cir. 1983), we are "not empowered to substitute [our] judgment for that of the agency." Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). This is particularly so where the agency's decision turns on "its evaluation of complex scientific data within its technical expertise." BCCA Appeal Grp. v. EPA, 355 F.3d 817, 824 (5th Cir. 2003) (citing Baltimore Gas & Elec. Co. v. NRDC, 462 U.S. 87, 103, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983)). Indeed, "[i]f the agency's reasons and policy choices conform to minimal standards of rationality, then its actions are reasonable and must be upheld." Tex. Oil & Gas Ass'n, 161 F.3d at 934. Furthermore, the "EPA's choice of analytical methodology [in setting and enforcing standards] is entitled to a presumption of regularity," leaving challengers with a "considerable burden" to carry. Am. Petroleum Inst. v. EPA, 787 F.2d 965, 983 (5th Cir. 1986).

Our review under the APA is not toothless, however. We must set aside agency action if the agency "entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise." Motor Vehicle Mfrs. Ass'n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983) ("State Farm"); see generally, e.g., Atchafalaya Basinkeeper v. U.S. Army Corps of Eng'rs, 894 F.3d 692, 697 (5th Cir. 2018) (reciting State Farm standard). "[W]e must also ensure that the agency `examine[d] the relevant data and articulate[d] a satisfactory explanation for its action,'" and assess "`whether the [agency's] decision was based on a consideration of the relevant factors[.]'" 10 Ring Precision, Inc. v. Jones, 722 F.3d 711, 723 (5th Cir. 2013) (quoting State Farm, 463 U.S. at 43, 103 S.Ct. 2856); see also, e.g., Michigan v. EPA, ___ U.S. ___, 135 S.Ct. 2699, 2706, 192 L.Ed.2d 674 (2015) (explaining that "agency action is lawful only if it rests `on a consideration of the relevant factors'") (quoting State Farm, supra); U.S. Chamber of Commerce v. U.S. Dep't of Labor, 885 F.3d 360, 382 (5th Cir. 2018) ("Illogic and internal inconsistency are characteristic of arbitrary and unreasonable agency action."); Illinois Pub. Telecom. Ass'n v. FCC, 117 F.3d 555, 566 (D.C. Cir. 1997), decision clarified on reh'g, 123 F.3d 693 (unexplained and "seemingly illogical" decisions are arbitrary and capricious). Furthermore, we "may uphold agency action only on the grounds that the agency invoked when it took the action." Michigan v. EPA, 135 S.Ct. at 2712 (citing SEC v. Chenery Corp., 318 U.S. 80, 87, 63 S.Ct. 454, 87 S.Ct. 626 (1943)).

The Environmental Petitioners challenge the leachate regulation under the two-step framework articulated in Chevron USA, Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), governing judicial review of agency interpretations of statutes. See generally, e.g., BCCA Appeal Grp., 355 F.3d at 824 (discussing Chevron in context of challenge to Clean Air Act regulations). At step one, the court considers "whether Congress has directly spoken to the precise question at issue." Chevron, 467 U.S. at 842, 104 S.Ct. 2778. If Congress has directly spoken on an issue, that settles the matter: "[T]he Court, as well as the agency, must give effect to the unambiguously expressed intent of Congress." Id. at 842-43, 104 S.Ct. 2778. Only if the statutory text is ambiguous can the court proceed to step two, asking whether the agency's construction of the statute is "permissible." Id. at 843, 104 S.Ct. 2778. If the construction is permissible, it should be upheld. "[A] court may not substitute its own construction of a statutory provision for a reasonable interpretation made by the administrator of an agency." Id. "Chevron review and arbitrary and capricious review overlap at the margins," specifically at Chevron step two. Indep. Petroleum Ass'n of Am. v. Babbitt, 92 F.3d 1248, 1258 (D.C. Cir. 1996); see also Nutraceutical Corp. v. Von Eschenbach, 459 F.3d 1033, 1038 (10th Cir. 2006).

IV. ANALYSIS

A. Challenge to Legacy Wastewater Regulation

We first consider petitioners' challenge18 to the final rule's regulation of legacy wastewater. As already explained, legacy wastewater is not a separate wastewater stream but instead a subset of five of the other streams. Supra I.B.1. Specifically, legacy wastewater is defined by when wastewater is generated: Wastewater "generated prior to" the compliance date set for the new rule by a permitting authority is denominated "legacy" wastewater. See 80 Fed. Reg. 67,854. Instead of subjecting legacy wastewater to the more advanced and effective technologies that kick in after the rule's compliance date (i.e., chemical precipitation, biological treatment, dry handling, or evaporation), the rule sets BAT for legacy wastewater as equal to the BPT previously set in 1982 (i.e., surface impoundments). See id. at 67,854-55.

1.

Petitioners challenge that decision as arbitrary and capricious under the APA on two grounds. First, they claim the Act does not grant the Administrator authority to base BAT limits on when waste is generated, but instead requires setting BAT limits for "categories and classes of point sources" regardless of when waste is generated. See, e.g., State Farm, 463 U.S. at 43, 103 S.Ct. 2856 (explaining that, "[n]ormally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider"). Second, they claim that the rulemaking record refutes the conclusion that surface impoundments are BAT for legacy wastewater because, among other things, the final rule itself demonstrates that impoundments are ineffective at removing toxic pollutants. See, e.g., id. (an agency rule would also be arbitrary and capricious if it "offered an explanation for its decision that runs counter to the evidence before the agency"). We need not reach petitioners' first argument, because we conclude for multiple reasons that EPA acted arbitrarily and capriciously by setting a BAT limit for legacy wastewater equal to the outdated BPT standard of surface impoundments.

First, the final rule repeatedly recognizes that impoundments are "largely ineffective" at removing toxins from wastewater. 80 Fed. Reg. 67,840. Impoundments "rely on gravity to remove particulates from wastewater," but the rule explains that "gravity in surface impoundments" fails to "effectively or reliably" remove "[p]ollutants ... present mostly in soluble (dissolved) form, such as selenium, boron, and magnesium," and also fails to "effectively" remove the dissolved portion of "metals present in both soluble and particulate forms (such as mercury)." Id. at 67,851. When ingested by humans either through drinking water or through seafood, these metals can lead to serious harms including "cancer, cardiovascular disease, neurological disorders, kidney and liver damage, and lowered IQs in children." Id. at 67,840. Additionally, the rule informs us that various factors can alter chemical conditions in impoundments and thus compromise their effectiveness: For instance, low pH in the impoundment environment can convert particulate metals to soluble form, reducing the "settling efficiency in the impoundments" and "leading to increased levels of dissolved metals and high concentrations of metals in discharges from surface impoundments." Id. at 67,851. Even the changing seasons—in an effect called "seasonal turnover"—impair impoundments by cooling the upper layer of water and causing it to sink, resulting in "resuspension of solids ... and a consequent increase in the concentrations of pollutants discharged from the impoundment." Id.

These conceded defects in impoundments are in critical tension with EPA's choosing them as BAT for legacy wastewater. After all, BAT is supposed to be "the CWA's most stringent standard" for setting discharge limits for existing sources. Tex. Oil & Gas, 161 F.3d at 928; see also 33 U.S.C. §§ 1311(b)(2), 1314(b)(2). We are rightfully skeptical when EPA specifies impoundments as BAT while, in the same breath, detailing how bad those impoundments are in stemming the discharge of toxic pollution. See, e.g., CMA, 885 F.2d at 265 (remanding because EPA "failed ... [to] demonstrate a reasonable basis for its conclusion" that its chosen BAT was as effective as a proposed alternative).

Second, as the rule also recounts, the flaws of impoundments are precisely why EPA refused to set them as BAT for five of the six wastewater streams at issue here. See 80 Fed. Reg. 67,851-53. For instance, the rule states that "EPA did not select surface impoundments as the BAT technology for FGD wastewater because it would not result in reasonable further progress toward eliminating the discharge of all pollutants, particularly toxic pollutants." Id. at 67,851 (emphasis added); see also id. (explaining that EPA declined to set impoundments as BAT "[b]ecause many of the pollutants of concern in FGD wastewater are present in dissolved form and would not be removed by surface impoundments") (emphasis added). EPA likewise declined to set impoundments as BAT for fly ash transport water, bottom ash transport water, FGMC wastewater, and gasification water, and in each case explained that it did so "for the same reasons ... that EPA did not identify surface impoundments as BAT for FGD wastewater." Id. at 67,852 (fly ash), 67,853 (bottom ash, FGMC, and gasification wastewater).

In other words, for five of the six wastewater streams regulated by the final rule (the one exception is leachate, discussed in V.B. infra), EPA affirmatively rejected surface impoundments as BAT "because [they] would not result in reasonable further progress toward eliminating the discharge of all pollutants, particularly toxic pollutants." Id. at 67,851.19 And yet, having rejected impoundments as BAT because they would not achieve "reasonable further progress" toward eliminating pollution from those streams, EPA turned around and chose impoundments as BAT for each of those same streams generated before the compliance date. That paradoxical action signals arbitrary and capricious agency action. See, e.g., Chamber of Commerce, 885 F.3d at 382 ("Illogic and internal inconsistency are characteristic of arbitrary and unreasonable agency action."); see also, e.g., GameFly, Inc. v. Postal Regulatory Comm'n, 704 F.3d 145, 148 (D.C. Cir. 2013) (explaining that agency action "`illogical on its face'" may be arbitrary and capricious) (quoting Am. Fed'n of Gov't Emps., Local 2924 v. Fed. Labor Relations Auth., 470 F.3d 375, 380 (D.C. Cir. 2006)). It also strongly suggests that EPA has contravened the plain language of the CWA, which defines BAT as the technology that "will result in reasonable further progress" toward pollutant discharge elimination. 33 U.S.C. § 1311(b)(2)(A) (emphasis added); see also, e.g., Nat'l Crushed Stone, 449 U.S. at 74-75, 101 S.Ct. 295 (discussing "reasonable further progress" component of BAT).

Third, the final rule explains that the shortcomings of surface impoundments were a key factor in motivating EPA to conduct the 2006-2009 study and revise water pollution regulations for power plants in the first place. The rule describes the previous ELGs from 1974, 1977, and 1982 as "out of date," because they failed to "adequately control the pollutants ... discharged by this industry" and failed to "reflect relevant process and technology advances that have occurred in the last 30-plus years." 80 Fed. Reg. 67,840; see also id. (stating that, "[i]n the several decades since the steam electric ELGs were last revised, [more effective] technologies have increasingly been used at plants"). And the rule minces no words in laying the shortcomings of the prior ELGs at the feet of surface impoundments:

Id.

Thus, the final rule describes impoundments as an outdated and ineffective pollution control technology, and yet the same rule chooses to freeze impoundments in place as BAT for legacy wastewater. That is inconsistent with the "technology-forcing" mandate of the CWA. NRDC II, 808 F.3d at 563-64 (citing NRDC I, 822 F.2d at 123). To that point, the Supreme Court has explained that BAT has an inbuilt "reasonable further progress" standard and that "BPT serves as the prior standard with respect to BAT." Nat'l Crushed Stone, 449 U.S. at 75, 101 S.Ct. 295. Yet here EPA appears to have conflated the prior standard with the advanced one: It has selected as BAT the same three-decades-old technology previously set as BPT—a technology the current rule condemns as anachronistic and ineffective at eliminating pollution discharge. In other words, EPA asks us to believe that impoundments are both archaic and cutting-edge at the same time. That we cannot do. See GameFly, 704 F.3d at 148; Chamber of Commerce, 885 F.3d at 382.

Fourth, the final rule strongly indicates that other available technologies are far better than impoundments at removing pollutants from the various streams that comprise legacy wastewater. For instance, after explaining why impoundments are ineffective at removing toxic metals from FGD wastewater, the rule states that a combination of chemical precipitation and biological treatment is better at removing those pollutants. Id. at 67,850-51. Importantly, the rule explicitly concludes that "[c]hemical precipitation and biological treatment are more effective than surface impoundments at removing both soluble and particulate forms of metals." Id. at 67,851 (emphasis added). The rule also relies on that reasoning to justify rejecting impoundments as BAT for fly ash transport water, bottom ash transport water, FGMC wastewater, and gasification wastewater. Id. at 67,852-53. Moreover, the rule categorically states that more advanced control methods, such as chemical and biological methods, "are affordable technologies that are widely available, and already in place at some plants." Id. at 67,840.

These affirmative findings are difficult, if not impossible, to square with EPA's decision nonetheless to set 1980s-era impoundments as the BAT for legacy wastewater. To be sure, the agency's statements in the final rule do not prove that chemical precipitation or biological treatment (or some combination of the two) are BAT for legacy wastewater. That is for the agency to decide. But those statements do cast grave doubt on the agency's selection of impoundments as BAT for legacy wastewater. See, e.g., Nat'l Crushed Stone Ass'n, 449 U.S. at 74, 101 S.Ct. 295 (BAT calls for the "maximum use of technology within the economic capability of the [plant] owner or operator"). Once again, the EPA's own rule strongly indicates that it was arbitrary and capricious in doing so.20

Fifth and finally, our court has long recognized that "`Congress intended [BAT] limitations to be based on the performance of the single best-performing plant in an industrial field.'" Tex. Oil & Gas, 161 F.3d at 928 (quoting CMA, 870 F.2d at 226)); see also, e.g., Kennecott v. EPA, 780 F.2d 445 (4th Cir. 1985) ("In setting BAT, EPA uses not the average plant, but the optimally operating plant, the pilot plant which acts as a beacon to show what is possible.") (citing A Legislative History of the Water Pollution Control Act Amendments of 1972, 93d Cong., 1st Sess. (Comm. Print 1973), at 798). Yet here the rule says nothing to indicate that the choice of impoundments as BAT for legacy wastewater was based on anything like "the performance of the single best-performing plant" in the field. To the contrary, everything the rule says about the record of impoundments over the past three decades indicates that their performance in controlling discharges has been distinctly poor. See, e.g., 80 Fed. Reg. 67,840 (stating that "impoundments ... are largely ineffective at controlling discharges of toxic pollutants and nutrients"); id. (stating that "[g]round water contamination from surface impoundments ... threatens drinking water, as evidenced by more than 30 documented cases"); id. at 67,851 (declining to set impoundments at BAT for FGD wastewater because various dissolved toxic metals "are not effectively and reliably removed by gravity in surface impoundments"). Moreover, the rule also states that multiple plants are in fact treating legacy wastewater using chemical precipitation, id. at 67,855 n.29, a method the rule concedes is "more effective than surface impoundments at removing both soluble and particulate forms of metals[.]" Id. at 67,851. Yet the rule merely states— without explanation—that it lacks "data to characterize the effluent from these systems." Id. at 67,855 n.29. That unexplained assertion casts grave doubt on the agency's BAT decision. See also infra V.A.2 (discussing additional problems created by agency's lack of data excuse).

These shortcomings in the agency's explanations strongly indicate that its BAT decision simply defaults to the outdated BPT standard that has been demonstrated to be a poor performer by the agency's own analysis. That is antithetical to the statutorily-mandated BAT standard. See, e.g., Nat'l Lime Ass'n v. EPA, 233 F.3d 625, 634 (D.C. Cir.), as amended on denial of reh'g (2001) (remanding EPA rule for failure in agency's "clear statutory obligation to set emission standards" for various air pollutants); NRDC v. EPA, 863 F.2d 1420, 1433 (9th Cir. 1988) (despite EPA's asserted lack of "complete information" on availability of technology, declaring BAT limitation invalid because "Congress has demonstrated its intent to require industry to do as much as possible to control toxic discharges") (citing 33 U.S.C. § 1311(b)(2)(A)(i)).

In sum, we conclude that the EPA's decision to set surface impoundments as BAT for legacy wastewater was arbitrary and capricious. Far from demonstrating that impoundments are the "best available technology economically achievable" for treating legacy wastewater, the evidence recounted in the final rule shows that impoundments are demonstrably ineffective at doing so and demonstrably inferior to other available technologies. In light of this record, we cannot accept that an outdated, ineffective and inferior technology is BAT when applied to legacy wastewater. No record evidence affirmatively makes that case and, as we have explained, the evidence recounted in the final rule runs in the opposite direction.

2.

EPA defends its choice of impoundments as BAT for legacy wastewater by asserting that "it does not have the data" to justify choosing more advanced pollution control technologies. 80 Fed. Reg. 67,855. The agency explains that most plants "combine some of their legacy wastewater with each other and with other wastestreams," and that this "commingling" can meaningfully alter the characteristics (specifically, the "flow rate and pollutant concentration") of the impoundment water. Id. Because EPA lacked adequate examples of plants treating commingled wastewater "using anything beyond the surface impoundment itself," the agency concluded it lacked data to evaluate the performance of other technologies and therefore defaulted to "the previously promulgated BPT limitations" (i.e., impoundments). Id. We are unpersuaded.

First, EPA's arguments about the characteristics of commingled wastewater glide past the key issue before us, which is whether the agency arbitrarily chose impoundments as BAT. The agency may lack data on how other technologies interact with commingled wastewater, but it assuredly does not lack data on impoundments. To the contrary, we know that impoundments are ineffective at removing toxic pollutants from the various wastewater streams because the agency's own rule tells us so, repeatedly, based on over three decades of observation and analysis. See supra V.B.1. Nor does EPA's criticism of impoundments distinguish "legacy" from "non-legacy" wastewater: instead, the agency categorically states that "surface impoundments ... are largely ineffective at controlling discharges of toxic pollutants and nutrients" from "wastewater." 80 Fed. Reg. 67,840. To be sure, we do not pretend to second-guess EPA's assertions about the pertinent "flow rate" and "pollutant concentration" in commingled wastewater, id. at 67,855, matters beyond our expertise and authority. But those assertions side-step the legal issue we must decide, which is whether the agency was arbitrary in selecting impoundments as BAT.21 Once again, the agency's own pointed criticism of impoundments indicates that it was.

Second, the agency's "lack of data" excuse is untenable on its own terms. In a footnote, the rule concedes that multiple power plants have in fact been using chemical precipitation to treat commingled legacy wastewater. See id. at 67,855 n.29 (stating "EPA identified fewer than ten plants that use chemical precipitation to treat waster that contains, among other things, ash transport water"). Yet, the agency baldly asserts that it "does not have any data to characterize the effluent from these systems" and it raises this dearth of information to justify not regulating legacy wastewater under the same BAT standards as non-legacy wastewater (and to justify a demonstrably outdated technology as BAT). Id. We have previously rejected EPA's argument that an asserted lack of "sufficient data" justified the agency's failure to regulate. See API I, 661 F.2d at 357 (rejecting EPA's argument that its failure to regulate was justified by lack of "sufficient data" where EPA had failed to investigate "in light of ... new information" (internal quotation marks omitted)); see also, e.g., NRDC II, 808 F.3d at 573 (concluding that EPA's failure to gather data can be arbitrary and capricious when the lack of data is "a problem of EPA's own making"); see also infra V.B.2 (further discussing API I and NRDC II). We reject the argument again here. The final rule recounts (1) the long-recognized deficiencies of impoundments in controlling toxic discharges, 80 Fed. Reg. 67,840; (2) the demonstrated superiority of more advanced technologies in doing so, id. at 67,851-53; (3) the availability of those technologies in the industry, id. at 67,840, 67,844; and (4) multiple plants' actual use of one of those advanced technologies (chemical precipitation) to treat commingled legacy wastewater, id. at 67,855 n.29. Given those undisputed statements drawn from EPA's own rule, the agency cannot simply plead a lack of data to justify its decision to set impoundments as BAT. Again, we do not purport to tell the agency what technology it should choose as BAT for legacy wastewater. We decide only that, given the agency's own statements and evidence, it acted arbitrarily in selecting as BAT a pollution control method that decades of data have shown to be ineffective at controlling pollution.

Third, given EPA's heavy reliance on the characteristics of commingled legacy wastewater as a reason for declining stricter regulation, one would expect a different policy for non-commingled legacy wastewater. See 80 Fed. Reg. 67,855 & n.28 (asserting as key reason for defaulting to impoundments for legacy wastewater the fact that wastewater at the "vast majority" of plants is "commingled" with other streams). Yet we find the opposite: When EPA identified plants that discharge non-commingled legacy wastewater from impoundments, it still declined to impose more stringent controls and still defaulted to impoundments as BAT. See id. at 67,855 (declining to impose controls "other than surface impoundments" on plants that "discharge from an impoundment containing only legacy FGD wastewater"); id. at 67,855 n.30 (discussing three plants that use impoundments "where the FGD wastewater is not commingled with other process wastewaters in the impoundment") (emphases added). That striking inconsistency undercuts the agency's "commingling" rationale for not imposing the more stringent non-legacy BAT standards. See, e.g., Am. Fed'n of Gov't Emps., Local 2924 v. Fed. Labor Relations Auth., 470 F.3d 375, 380 (D.C. Cir. 2006) (an agency's decision is "arbitrary and capricious" if "illogical on its own terms"); see also Chamber of Commerce, 885 F.3d at 382 ("Illogic and internal inconsistency are characteristic of arbitrary and unreasonable agency action.").22 What is more, by selecting impoundments as BAT for any kind of FGD wastewater ("legacy" or not), the rule flatly contradicts itself: The rule states without exception that for FGD wastewater "[c]hemical precipitation and biological treatment are more effective than surface impoundments at removing" toxic pollutants. 80 Fed. Reg. 67,851.

Fourth, even assuming a lack of data prevented EPA from determining BAT for legacy wastewater, nothing required the agency simply to set impoundments as BAT. Instead, EPA could have declined to set nationwide effluent guidelines for legacy wastewater and allowed BAT determinations to be made by each facility's permitting authority through the NPDES permitting process on a site-specific basis. See 40 C.F.R. § 125.3(a), (c)(2) ("Technology based treatment requirements may be imposed ... [o]n a case-by-case basis."); Riverkeeper, Inc. v. EPA, 358 F.3d 174, 203 (2nd Cir. 2004) ("We see no textual bar in sections 306 or 316(b) [of the Clean Water Act] to regulating [certain] structures on a case-by-case basis."); Nat'l Wildlife Fed'n v. EPA, 286 F.3d 554, 566-67 (D.C. Cir. 2002) ("We believe EPA acted both reasonably and within its authority in adopting a case-by-case approach" to regulating certain pollutants from paper mills). The agency took that approach in the current rule by deferring setting BAT and other limits for metal cleaning wastes, after determining it lacked the necessary data. See 80 Fed. Reg. 67,863 (directing permitting authorities to "establish such requirements based on [best professional judgment] for any steam electric power plant discharg[ing]" such waste). Instead of deferring a nationwide effluent guideline and allowing case-by-case determination of BAT by permitting authorities, EPA unaccountably defaulted to impoundments—again, which its own rule recognizes as an out-of-date and ineffective pollution control technology. This is further indication that the rule respecting legacy wastewater is arbitrary and capricious.23

* * *

In sum, having examined the various justifications set forth for EPA's final rule on legacy wastewater, and finding each of those explanations wanting in light of the agency record, we conclude that EPA's rulemaking was arbitrary and capricious. We therefore set aside that part of the final rule and remand to the agency for reconsideration. See, e.g., Perdue v. FAA, 172 F.3d 866 (5th Cir. 1999) ("This court shall set aside agency action that is `arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.'" (quoting 5 U.S.C. § 706(2)(A))); Checkosky v. SEC, 23 F.3d 452, 491 (D.C. Cir. 1994) ("Section 706(2)(A) [of the APA] provides that a `reviewing court' faced with an arbitrary and capricious agency decision `shall'—not may—`hold unlawful and set aside' the agency action.").

We recognize that the agency is entitled to considerable deference in setting BAT limitations. See, e.g., BCCA Appeal Grp., 355 F.3d at 824 (EPA entitled to special deference where its decision turns on "its evaluation of complex scientific data within its technical expertise"); Tex. Oil. & Gas, 161 F.3d at 928 (recognizing EPA's "considerable discretion" in weighing BAT factors). Precisely for that reason, challenges to the agency's BAT determinations often fail because challengers ask the court to elevate itself as an expert over the agency.24 This case is different. We do not question the scientific or statistical methodologies relied upon by EPA, nor second-guess its weighing of the statutory factors. Instead, we rely on EPA's own scientific conclusions in the rule itself to conclude that its choice of an outdated and ineffective technology as BAT was unlawful under the Act. That is a legal—not a technical or scientific—conclusion that the APA requires us to make.

B. Challenge to Leachate Regulation

We turn to the challenge to the rule's regulation of combustion residual leachate. As explained, supra I.B.2, leachate consists of liquid that percolates through a landfill or impoundment and is eventually discharged into water. See 80 Fed. Reg. 67,847. The final rule sets BAT for leachate as impoundments, which is the same as the previous BPT for leachate established in 1982. See id. at 67,854 ("This rule identifies surface impoundments as the BAT technology basis for control of pollutants in combustion residual leachate ... establish[ing] a BAT limitation on [total suspended solids] in ... leachate equal to the previously promulgated BPT limitation on [total suspended solids] in low volume waste sources."). EPA offers two justifications for this decision: First, that "[c]ommenters did not provide information that EPA could use to establish [stricter] BAT limitations" for leachate, and, second, that because leachate forms a "very small portion" of overall discharges, the rule's stricter regulation of other wastestreams "already represents reasonable further progress towards the CWA's goals." See id.; supra I.B.2.

The Environmental Petitioners challenge the rule's leachate regulation under the Chevron test for reviewing agency interpretations of statutes. See Chevron, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694; see also generally, e.g., Acosta v. Hensel Phelps Constr. Co., 909 F.3d 723, 730 (5th Cir. 2018) (summarizing "the two-step framework established in Chevron"); RICHARD J. PIERCE, JR., ADMINISTRATIVE LAW TREATISE § 3.2 ("PIERCE") (discussing "[t]he Chevron Two-Step"). Petitioners assert that EPA's decision to set impoundments as BAT for leachate fails Chevron step one by contravening the plain text and structure of the Clean Water Act. Alternatively, petitioners argue that the agency's decision fails Chevron step two by adopting an impermissible construction of the Act. We address each argument in turn.

1.

At Chevron step one, we ask whether the pertinent statute "unambiguously foreclose[s]" the agency's challenged statutory interpretation; if it does, "`that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.'" U.S. Chamber of Commerce v. U.S. Dep't of Labor, 885 F.3d 360, 369 (5th Cir. 2018) (quoting Chevron, 467 U.S. at 842-43, 104 S.Ct. 2778); see also, e.g., Acosta, 909 F.3d at 730 (at Chevron step one, the court "must determine `whether Congress has directly spoken to the precise question at issue'" (quoting City of Arlington v. FCC, 569 U.S. 290, 296, 133 S.Ct. 1863, 185 L.Ed.2d 941 (2013))). To answer this question, we rely on "the conventional standards of statutory interpretation"— i.e., "text, structure, and the overall statutory scheme"—as well as "authoritative Supreme Court decisions." U.S. Chamber of Commerce, 885 F.3d at 369 (citing City of Arlington, 569 U.S. at 296, 133 S.Ct. 1863; Chevron, 467 U.S. at 843 n.9, 104 S.Ct. 2778). We are not to focus myopically on "a particular statutory provision in isolation" because "[t]he meaning—or ambiguity —of certain words or phrases may only become evident when placed in context." FDA v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 132, 120 S.Ct. 1291, 146 L.Ed.2d 121 (2000). Rather, we must read words "`in their context,'" interpreting the statute "`as a symmetrical and coherent regulatory scheme,'" and "`fit[ting], if possible, all [the statute's] parts into an harmonious whole.'" Id. at 133, 120 S.Ct. 1291 (quoting Davis v. Mich. Dep't of Treasury, 489 U.S. 803, 809, 109 S.Ct. 1500, 103 L.Ed.2d 891 (1989); Gustafson v. Alloyd Co., 513 U.S. 561, 569, 115 S.Ct. 1061, 131 L.Ed.2d 1 (1995); FTC v. Mandel Bros., Inc., 359 U.S. 385, 389, 79 S.Ct. 818, 3 L.Ed.2d 893 (1959)) (brackets added). Additionally, "we must be guided to a degree by common sense as to the manner in which Congress is likely to delegate a policy decision of such economic and political magnitude to an administrative agency." Texas v. United States, 787 U.S. 733, 760 (5th Cir. 2015) (quoting Brown & Williamson, 529 U.S. at 133, 120 S.Ct. 1291). The goal of our Chevron step one inquiry is ultimately to "ascertain whether the statute is silent or ambiguous in addressing the precise question at issue." Tex. Savings & Cmty. Bankers Ass'n v. Fed. Hous. Fin. Bd., 201 F.3d 551, 554 (5th Cir. 2000); see also PIERCE § 3.6 ("The question for the court [at step one] is whether the agency's construction of the language is within the range of meanings that could be plausibly attributed to the relevant statutory language.").

Petitioners' step one attack targets the agency's justifications for pegging leachate BAT to the same technology set as BPT in 1982—specifically, EPA's explanation that leachate forms a "very small portion" of collective industry discharges and that the rule's stricter BAT for other wastestreams represents reasonable overall progress in the industry. See 80 Fed. Reg. 67,854. Petitioners' various arguments may be grouped into three general categories for convenience of analysis. First, petitioners argue the agency's rationales contravene the CWA's text—requiring a BAT to eliminate discharges of "all pollutants" if "technologically and economically achievable," 33 U.S.C. § 1311(b)(2)(A) (emphasis added) —because, here, the agency has chosen not to regulate a significant pollution source (leachate) with a technology EPA admits is achievable (chemical precipitation). Second, petitioners argue the agency's decision on the proper control standard for leachate conflates BAT with BPT in a way that contravenes the structure of the Act. Finally, petitioners claim that the agency's proffered justifications for its leachate regulation appear nowhere in the factors mandated for determining BAT and, indeed, contradict those factors. See 33 U.S.C. § 1314(b)(2)(B) (listing BAT factors).

Petitioners' initial step one argument places too much weight on the phrase "all pollutants" in § 1311(b)(2)(A). We cannot agree that this statutory phrase, standing alone, squarely forecloses the agency's decision to maintain leachate BAT at the 1982 BPT standard. To the contrary, we agree with EPA (and with binding precedent) that we must interpret that phrase in context and with reference to the larger statutory scheme. See, e.g., Brown & Williamson, 529 U.S. at 132, 120 S.Ct. 1291 (at step one courts should not examine statutory provisions "in isolation" but must "interpret the statute as a symmetrical and coherent regulatory scheme" (internal quotes and citations omitted)). Section 1311(b)(2)(A) does not flatly require that a BAT standard "eliminate[ ] the discharge of all pollutants" solely if the Administrator finds such elimination "technologically and economically achievable," as petitioners claim.25 Rather, the phrase petitioners rely on is nested in a complex provision providing that certain effluent limitations

Id. (emphasis added). The italicized portion of § 1311(b)(2) cross-references the factors set forth in § 1314(b)(2) that are "to be taken into account" by the Administrator in determining BAT for a given point source. See id. § 1314(b)(2)(B) (requiring certain "factors relating to the assessment of [BAT]" shall "be taken into account in determining the best measures and practices available to comply with subsection (b)(2) of section 1311 of this title"). The statute requires us to read these two sections in harmony with each other. See, e.g., Doe v. KPMG, 398 F.3d 686, 688 (5th Cir. 2005) (requiring that we "read all parts of the statute together to produce a harmonious whole" (citation omitted)). And we have held that the Administrator has "considerable discretion" in evaluating the cross-referenced § 1314(b)(2) factors when making a BAT determination. See, e.g., Tex. Oil & Gas, 161 F.3d at 928 (citing NRDC v. EPA, 863 F.2d 1420, 1426 (9th Cir. 1988)). Consequently, it is not the case that § 1311(b)(2)(A)—standing apart from the factors in § 1314(b)(2)—unambiguously required EPA to set a stricter BAT for leachate. We therefore reject petitioners' step one argument based on the text of § 1311(b)(2)(A).

We agree with petitioners, however, that the leachate rule conflates the BAT and BPT standards in a way not permitted by the statutory scheme. The rule pegs BAT for leachate to the decades-old BPT standard, without offering any explanation for why that prior standard is now BAT. That is flatly inconsistent with the Act's careful distinction between the two standards. As explained, the difference between BAT and BPT is critical to the Act's "technology-forcing" scheme. Supra I.A (quoting NRDC I, 822 F.2d 104, 123). BPT is merely the "first step" towards the Act's pollution reduction goals and provides the "prior standard" against which the stricter BAT is to be measured. Nat'l Crushed Stone, 449 U.S. at 69, 101 S.Ct. 295; id. at 75 & n.14, 101 S.Ct. 295; CMA, 870 F.2d at 196. To that point, Congress designed BPT to reflect merely an average of the best performance levels of existing plants, CMA, 870 F.2d at 203, whereas it designed BAT to reflect "`the performance of the single best-performing plant in an industrial field.'" Tex. Oil & Gas, 161 F.3d at 928 (quoting CMA, 870 F.2d at 226). And this critical distinction is reflected in the Act's structure, which treats BAT and BPT in different subsections, implements them on different timelines, and evaluates them under different factors. Compare 33 U.S.C. §§ 1311(b)(1)(A), 1314(b)(1)(A), 1314(b)(1)(B) (specifying BPT applicability, timeline, and factors), with id. §§ 1311(b)(2)(A), (C), (D) & (F), 1314(b)(2)(B) (specifying BAT applicability, timelines, and factors); see also 80 Fed. Reg. 67,843 (distinguishing BPT and BAT).

Yet, in the face of this statutory structure, the rule sets BAT for leachate "equal to the [prior] BPT limitation"—i.e., impoundments. 80 Fed. Reg. 67,854. The rule reaches that decision without explaining why a technology selected in 1982 under the laxer BPT standard somehow meets the stricter BAT standard today. That is particularly inexplicable given the rule's recognition that impoundments have proven "largely ineffective" at pollution control over the past decades, id. at 67,840, 67,851-53; see supra I.B. And, as we have seen, it was the recognized shortcomings of impoundments —shortcomings with respect to leachate discharges as well as other wastestreams—that led the agency to revise the steam-electric effluent guidelines in the first place. Id. at 67,840; supra I.B.

To be sure, we do not say that EPA is precluded by the Act from ever setting BAT equivalent to a prior BPT standard. But given the plain distinction between the two standards marked out in the Act, the agency would at least have to offer some explanation for its decision that speaks to the statutory differences between BAT and BPT. Here we are given nothing along those lines. Consequently, the only conclusion we can draw from this record is that, in setting BAT for leachate, the agency simply defaulted to the prior BPT. As explained, however, the statutory scheme does not confer authority on the agency to collapse the carefully-wrought distinction between BAT and BPT in this manner. See, e.g., Texas v. United States, 497 F.3d 491, 501 (5th Cir. 2007) ("Chevron deference comes into play, of course, only as a consequence of statutory ambiguity, and then only if the reviewing court finds an implicit delegation of authority to the agency.").

We also agree with petitioners that the agency's proffered justifications for the leachate rule are not supported— indeed, are likely incompatible with—the factors set forth under the Act for determining BAT. "The first [Chevron step] determines whether Congress intended to give the agency any discretion," La Union Del Pueblo Entero v. FEMA, 608 F.3d 217, 220 (5th Cir. 2010), and we are unpersuaded that Congress gave the EPA discretion to rely on justifications like these. As explained, the agency excuses its lax leachate BAT by appealing to (1) the relatively "small" amount of pollutants discharged in leachate, and (2) the stricter BATs set for larger industry wastestreams. See 80 Fed. Reg. 67,854. Yet neither of these considerations implicates any of the factors the Act requires the Administrator to consider in determining BAT for a given point source. See 33 U.S.C. § 1314(b)(2)(B) (requiring that regulations "shall take into account" specified "[f]actors relating to the assessment of best available technology" with respect to "any point source ... within such categories or classes").

The Act specifies the following BAT factors:

Id. (brackets added). These factors cannot be stretched to accommodate the agency's rationales for its leachate BAT. That is, no factor allows the agency to consider the amount of pollutants generated by a one wastestream relative to other streams. Nor does any factor allow the agency to consider whether less stringent regulation of one wastestream may be set off against the benefits of regulating other streams more strictly. See, e.g., Tex. Oil & Gas Ass'n, 161 F.3d at 928 (explaining that "CWA specifies several factors that must be considered by the EPA in determining BAT limits" (emphasis added)). Thus, on their face, the justifications for the leachate BAT put forward in the rule fall outside of the factors mandated by the Act for determining BAT. See, e.g., API II, 858 F.2d at 264 ("Before EPA selects BAT-level limitations, it is required to address both (1) operational considerations, including `the process employed, the engineering aspects of the application of various types of control techniques [and] process changes,' and (2) cost" (emphasis added)).

What is more, the agency's benefit-weighing appears incompatible with the BAT factors and with the broader statutory scheme. Whether BAT factors always preclude weighing regulatory benefits against other considerations is an open question. See Entergy Corp. v. Riverkeeper, Inc., 556 U.S. 208, 221, 129 S.Ct. 1498, 173 L.Ed.2d 369 (2009) (declining to resolve whether "cost-benefit analysis is precluded under the [BAT] ... test[ ]").26 We need not address that question here, because we conclude that the agency's benefit-weighing in this case contravenes the statute. As explained, the rule explicitly justifies a less stringent BAT for leachate by touting the benefits of stricter BATs for other wastestreams. But the Act does not permit the agency to set a BAT by playing one pollution source off against another. As petitioners point out, the Act instead requires a BAT determination to be made with respect to a discrete "point source." 33 U.S.C. § 1314(b)(2)(B). A "point source" is defined as "any discernible, confined and discrete conveyance, including but not limited to any pipe, ditch, channel, tunnel, conduit, well, discrete fissure, container, rolling stock, concentrated animal feeding operation, or vessel or other floating craft, from which pollutants are or may be discharged." Id. § 1362(14). This broad definition easily includes leachate, and the rule leaves no doubt that it treats leachate as a distinct point source. See 80 Fed. Reg. 67,847 (treating "combustion residual leachate" from landfills or impoundments as a separate wastestream). The Act thus specifically requires the BAT factors be applied with respect to a specific point source—here, leachate. See 33 U.S.C. § 1314(b)(2)(B) (providing that regulations "shall ... specify factors to be taken into account in determining the best measures and practices available to comply with [the BAT requirements in § 1311(b)(2)] applicable to any point source ... within such category or classes" (emphasis added)). But in the final rule the agency has explicitly factored into its BAT determination the regulation of wastestreams other than leachate, which contravenes the plain text and structure of the Act.

EPA's principal argument in response is that the Act allows consideration of "other factors as the Administrator deems appropriate," 33 U.S.C. § 1314(b)(2)(B), which permits it to consider the "very small" size of leachate pollution relative to the overall industry. Because leachate represents only about 3 percent of overall industry pollution, EPA claims, the rule still represents "reasonable progress" towards eliminating pollution because of the rule's regulation of other sources. See 33 U.S.C. § 1311(b)(2)(A). Even accepting EPA's characterization of leachate pollution as "very small" (something we address at Chevron step two, infra), we reject the agency's reliance on the "other factors" clause. Like any statute, we must interpret the Clean Water Act by "looking at the full text of the statute, rather than one isolated clause, along with the statute's structure and its public safety purpose[.]" United States v. Transocean Deepwater Drilling, Inc., 767 F.3d 485, 496 (5th Cir. 2014). As petitioners correctly point out, the agency's "other factors" argument would undermine the concept of BAT altogether. The BAT factors are designed to support achievement of an effluent limitation that "shall require the elimination of discharges of all pollutants," if "technologically and economically achievable." 33 U.S.C. § 1311(b)(2)(A) (citing id. § 1314(b)(2)). But accepting the agency's expansive view of the "other factors" clause would allow it, in every case, to justify a less stringent BAT for one pollution source by claiming it was regulating other sources more strictly and thus making reasonable progress in the industry "as a whole." On this understanding, the BAT standard would cease to have any meaning. We therefore reject EPA's reliance on the "other factors" clause to support its justification for the leachate regulation.27

In sum, we conclude that the BAT determination for leachate fails step one of Chevron.

2.

Alternatively, we conclude that the leachate regulation fails step two of Chevron. For purposes of this analysis, we assume that the CWA is "silent or ambiguous" with respect to the question addressed by the rule, and we ask only "whether the agency's answer is based on a permissible construction of the statute." Acosta, 909 F.3d at 730 (quoting City of Arlington, 569 U.S. at 296, 133 S.Ct. 1863; Chevron, 467 U.S. at 843, 104 S.Ct. 2778). "We do not simply impose our own construction on the statute." BNSF Ry. Co. v. United States, 775 F.3d 743, 751 (5th Cir. 2015) (cleaned up) (quoting Chevron, 467 U.S. at 843, 104 S.Ct. 2778). "The agency's view `governs if it is a reasonable interpretation of the statute—not necessarily the only possible interpretation, nor even the interpretation deemed most reasonable by the courts.'" Acosta, 909 F.3d at 735 (quoting Coastal Conserv'n Ass'n v. U.S. Dep't of Commerce, 846 F.3d 99, 106 (5th Cir. 2017); Riverkeeper, 556 U.S. at 218, 129 S.Ct. 1498). While this is a highly deferential standard, an agency interpretation can fail Chevron step two if "it is contrary to clear congressional intent or frustrates the policy Congress sought to implement." Garcia-Carias v. Holder, 697 F.3d 257, 271 (5th Cir. 2012). Agency action that is "arbitrary, capricious, or manifestly contrary to the statute" also fails step two. Tex. Coal. of Cities for Util. Issues v. FCC, 324 F.3d 802, 807 (5th Cir. 2003) (quoting Chevron, 467 U.S. at 844, 104 S.Ct. 2778). Because Chevron step two and the APA share the "arbitrary and capricious" standard, "[t]he APA reflects the principles of Chevron," and analysis under the two standards proceeds similarly. Nutraceutical Corp. v. Von Eschenbach, 459 F.3d 1033, 1038 (10th Cir. 2006); see also, e.g., PIERCE § 3.6 (suggesting that Chevron step two has "complete overlap" with APA test of whether a rule adopts an "unreasonable" statutory interpretation) (quoting Animal Legal Def. Fund v. Glickman, 204 F.3d 229, 234 (D.C. Cir. 2000)).

Petitioners argue that the agency's decision to set surface impoundments as BAT for leachate is based on an impermissible interpretation of the Act. They raise arguments similar to the ones raised under step one—i.e., that the agency acted unreasonably by setting a leachate BAT based on its relative size and on the rule's stricter regulation of other streams. Additionally, petitioners argue that the agency rejected more effective, achievable control technologies (like chemical precipitation) in favor of a less effective technology like impoundments, which "is unreasonable because it cannot be squared with Congress's intent for BAT to be more stringent than BPT limits." We agree with petitioners that the leachate regulation is based on an impermissible interpretation of the Act. We therefore hold that the regulation fails Chevron step two and must be vacated on that alternative basis as well.

First, the rule unreasonably sets as BAT a technology the rule itself deems ineffective at controlling toxic discharges from leachate. As already explained, supra V.B.1, the final rule categorically recognizes that impoundments are ineffective at removing toxic pollutants from wastewater, 80 Fed. Reg. 67,840, 67,851, which is why the rule declined to set impoundments as BAT for five of the six wastestreams at issue, id. at 67,852-53. Nothing in the rule even hints, much less explains, that impoundments are somehow better at controlling harmful discharges from leachate. To the contrary, the rule recounts that groundwater contamination from impoundments (which are "the most widely used systems to treat ... leachate") has resulted in numerous documented cases of drinking water pollution. Id. at 67,840, 67,847; see also id. at 67,847 (defining "leachate" as "liquid ... that has percolated through or drained from waste or other materials placed in a landfill, or that passes through the containment structure... of a surface impoundment"); id. (explaining that "[u]nlined impoundments and landfills ... allow the leachate to potentially migrate to nearby ground waters, drinking water wells, or surface water"). The rule also refers to an environmental assessment document, id. at 67,840, which reports that "[c]ombustion residual leachate can migrate from the site in the ground water at concentrations that could contaminate public or private drinking water wells and surface waters, even years following disposal of combustion residuals." See Environmental Assessment Document No. EPA-821-R-15-006, § 3.3.2. Given these admitted deficiencies in impoundments, it was unreasonable to adopt them as BAT for leachate. See State Farm, 463 U.S. at 43, 103 S.Ct. 2856 ("Normally, an agency rule would be arbitrary and capricious if the agency has ... offered an explanation for its decision that runs counter to the evidence before the agency.").

Second, the rule unreasonably declines to set as BAT available technologies that are admittedly more effective at controlling leachate discharges. We have already detailed the rule's affirmation that available modern technologies like chemical precipitation "are more effective than surface impoundments at removing both soluble and particulate forms of metals." 80 Fed. Reg. 67,851; supra V.B.1. The rule never explains why this blanket statement does not apply to the use of impoundments to treat leachate. Indeed, EPA acknowledged during the rulemaking process that "chemical precipitation is an available and demonstrated technology for the treatment of combustion residual leachate." See EPA's Response to Public Comments, SE05958, at 7-20. And EPA's counsel conceded at oral argument that chemical precipitation is "available" as a treatment for leachate. Oral Argument Audio at 22:25-22:45 (Oct. 3, 2018). As with its treatment of legacy wastewater, the rule appears to select a BAT for leachate simply by defaulting to the decades-old and demonstrably ineffective BPT standard. We have already explained that this kind of regulation-by-inertia is inconsistent with the "technology-forcing" mandate of the CWA. See NRDC II, 808 F.3d at 563-64; NRDC I, 822 F.2d at 123; supra V.B.1. Moreover, a decision to leave BPT limitations in place for leachate, when those limitations were based on admittedly ineffective technology, does not reflect "a commitment of the maximum resources economically possible to the ultimate goal of eliminating all polluting discharges," which was the intent of Congress in enacting BAT standards in the first place. Nat'l Crushed Stone Ass'n, 449 U.S. at 74, 101 S.Ct. 295. The EPA's decision to rest on its laurels (questionable as they are) respecting leachate thus "frustrates the policy Congress sought to implement" in the CWA, see Garcia-Carias, 697 F.3d at 271, and cannot stand.

Our decision in Chemical Manufacturers Association v. EPA, 870 F.2d 177, reinforces our conclusion on this point. That case involved a challenge to BAT and NSPS regulations set for water pollution by organic chemicals and synthetics plants. Id. at 261. In setting NSPS limits for these plants, EPA did not consider recycling as a possible NSPS technology, even though the evidence before it showed that plants could achieve zero discharge by using recycling. Id. at 262-63. We concluded that the agency's failure even to consider recycling, an "available demonstrated technology," was arbitrary and capricious. Id. at 264. Our case is analogous: here we have the known shortcomings of the status quo technology (impoundments), and the demonstrated superiority of available alternatives (chemical precipitation). To be sure, EPA was entitled to deference on NSPS standards just as it is on BAT standards, id. at 263, but deference runs out when the agency fails to consider an option with all the indicators of being a superior choice. Just so here. An "[u]nexplained inconsistency in agency policy is a reason for holding an interpretation to be an arbitrary and capricious," and "[a]n arbitrary and capricious regulation of this sort is itself unlawful and receives no Chevron deference." Encino Motorcars, LLC v. Navarro, ___ U.S. ___, 136 S.Ct. 2117, 2126, 195 L.Ed.2d 382 (2016) (citing Nat'l Cable & Telecomms. Ass'n v. Brand X Internet Servs., 545 U.S. 967, 981, 125 S.Ct. 2688, 162 L.Ed.2d 820 (2005); United States v. Mead Corp., 533 U.S. 218, 227, 121 S.Ct. 2164, 150 L.Ed.2d 292 (2001)).

Finally, we note that the EPA has described leachate as being chemically similar to FGD wastewater, a wastestream admittedly susceptible to effective treatment by chemical and biological methods. See 80 Fed. Reg. 67,851; see also EPA's Response to Public Comments, SE05958, at 7-20. It puzzles us that the EPA has described two wastestreams as chemically similar, and susceptible to treatment by the same methods, and yet has set strikingly different BAT standards for each. As with legacy wastewater, the agency's rationales contradict themselves. The BAT determination for leachate is "illogical on its own terms" and therefore cannot stand. See, e.g., GameFly, 704 F.3d at 148 (D.C. Cir. 2013) (facially illogical determinations are arbitrary and capricious).

EPA offers two justifications for selecting impoundments as BAT for leachate— the first based on the agency's lack of data about alternatives, and the second based on the relative size of the leachate wastestream. We find neither persuasive.

First, we are unpersuaded by the agency's argument that "[c]ommentators did not provide information that EPA could use to establish BAT limitations" concerning leachate. 80 Fed. Reg. at 67,854. During its lengthy study period beginning in 2005, not only did EPA have adequate opportunity to collect data on various treatment options for leachate, but the agency did collect data on the size of leachate pollution and on the benefits of chemical precipitation. See 80 Fed. Reg. 67,854 ("EPA considered whether technologies in place for treatment of other wastestreams" could be used to treat leachate); TDD, EPA-821-R-15-007, at 10-39 (table showing potential reduction in leachate pollution of 33,800 toxic-weighted pound equivalents by adopting chemical precipitation). These data were certainly enough to demand further inquiry and, of course, they say nothing to justify setting impoundments as BAT.

We have before declined to accept lack of data as a valid excuse for an agency's failure to regulate activity that concededly creates pollution, and we decline again here. In API I, for instance, we rejected EPA's attempt to justify failing to regulate "stripper gas wells" based on its claim that "there was not sufficient data" to justify regulation. 661 F.2d at 357. We reasoned that, in the three years since EPA had declined to regulate, ample public data on the wells "belie[d] EPA's contention that there exists nothing to regulate." Id. We therefore remanded for further agency consideration "in light of this new information." Id. Here, impoundments have been in operation for over three decades, and, as we have discussed at length, the agency's own rule amply demonstrates their ineffectiveness in controlling discharges from wastestreams including leachate. The agency cannot rely on a lack of data to justify its setting a BAT standard based on demonstrably outdated and ineffective technology.28

EPA counters this point by asserting that it "identified no existing plants using chemical precipitation to treat their leachate." We take EPA at its word, as we must, but the agency misses the point. Under our precedent, a technological process can be deemed "available" for BAT purposes "even if it is not in use at all," or if it is used in unrelated industries. API II, 858 F.2d at 265. "Such an outcome is consistent with Congress'[s] intent to push pollution control technology." Id. In this case, technologies alternative to surface impoundments are in use in the steam-electric industry, just in separate wastestreams. See, e.g., 80 Fed. Reg. 67,855 n.29. If technologies from other industries can be considered, then, a fortiori, technologies within the same industry should be considered when the status quo technology in place for a wastestream is admittedly ineffective. The final rule itself recognizes this point. See 80 Fed. Reg. 67,843 ("BAT is intended to reflect the highest performance in the industry, and it may reflect a higher level of performance than is currently being achieved based on technology transferred from a different category or subcategory, bench scale or pilot studies, or foreign plants." (citing Am. Paper Inst. v. Train, 543 F.2d 328, 353 (D.C. Cir. 1976); Am. Frozen Food Inst. v. Train, 539 F.2d 107, 132 (D.C. Cir. 1976))). In any event, EPA's argument ignores the fact that during notice-and-comment it admitted that "chemical precipitation is an available and demonstrated technology for the treatment of combustion residual leachate," based on the technology's use in treating a similar wastestream (FGD wastewater). See EPA's Response to Public Comments, SE05958, at 7-20.

Second, we reject the EPA's argument that its regulation is justified by the fact that leachate pollution constitutes "a very small portion of the pollutants discharged collectively by all steam power plants." 80 Fed Reg. 67,854. As already explained at step one, supra V.B.1, this consideration finds scant support in the statutory scheme given that the relative size of a stream is absent from the statutory BAT factors. But even assuming the statute allows the agency to consider the relative size of a pollution source in setting BAT, we find the EPA's use of that consideration here to be unreasonable.

The agency's argument involves some sleight of hand. Leachate discharges may constitute "a very small portion" of pollutants, 80 Fed Reg. 67,854, but only by comparison to all pollution from the entire steam-electric power plant industry, which is largest collective source of water pollution in the country. Id. at 67,839-40.29 In an absolute sense, however, leachate pollution is not a "very small portion" of anything. If leachate were a separate industry, it would rank as the 18th-largest source of water pollution in the United States. See TDD, EPA-821-R-15-007, at 10-39 (table showing toxic-weighted pollution from leachate); Annual Effluent Guidelines Review Report, EPA-821-R-16-002, at 2-26 (listing pollution from other industries). Leachate alone produces more toxic-weighted pound equivalents than the entire coal mining industry, according to the EPA's own records. Id. And EPA's brief grudgingly admits that leachate pollution "might be considered comparable" to the total amount of pollution coming from other industries such as coal mining and sugar processing. Thus, with the distortions stripped away, EPA's "argument" for its leachate regulation turns out to be a mere statement of fact, and an empty one at that. Yes, leachate pollution may form a "very small portion" of a gargantuan source of water pollution. But leachate constitutes a gargantuan source of water pollution on its own terms. Both statements are true. Neither begins to justify the agency's choice of impoundments as BAT.30

At oral argument, EPA conceded that it lacked discretion to ignore leachate based on its allegedly small size, instead arguing it had regulated leachate—i.e., by setting BAT based on surface impoundments. Oral Argument Audio at 22:10-22:20. Thus we end where we began. The final rule itself tells us that surface impoundments have a poor record at controlling toxic water pollution. 80 Fed. Reg. 67,840. The agency's own comments tell us that "chemical precipitation is an available and demonstrated technology for the treatment of combustion residual leachate." Faced with these concessions, we must conclude that the leachate regulation rests on an impermissible interpretation of the Clean Water Act and therefore fails Chevron step two.31

V. CONCLUSION

In the Clean Water Act, Congress afforded EPA considerable policy-making discretion to formulate rules to mitigate pollution in our nation's waters. As a price for that delegation of authority, however, the agency must engage in "reasoned decisionmaking," Michigan v. EPA, 135 S.Ct. at 2706, and "stay[ ] within the bounds of its statutory authority." Utility Air Reg. Grp. v. EPA, 573 U.S. 302, 315, 134 S.Ct. 2427, 189 L.Ed.2d 372 (2014) (internal quotation marks and citations omitted). The regulations challenged in this case fall short of those judicially-enforceable limits on the exercise of agency discretion. We therefore conclude that the portions of the final rule regulating legacy wastewater and residual combustion leachate are unlawful. Accordingly, we vacate those portions of the rule and remand to the agency for reconsideration consistent with this opinion.

VACATED IN PART AND REMANDED.