Full opinion text
Opinion for the Court filed PER CURIAM. PER CURIAM: In these consolidated petitions, several states, counties, and industrial entities challenge the Environmental Protection Agency’s promulgation of area designations for the annual national ambient air quality standard applicable to fine particulate matter, a category of air pollutants consisting of miniscule airborne particles known to present adverse health risks. Insisting that EPA’s methodology for designating areas as “nonattainment” for the fine particulate matter standard violates section 107(d) of the Clean Air Act, which governs such designations, and that this methodology and the individual designations it produced are otherwise arbitrary and capricious,, petitioners ask us to vacate the nonattainment designations and to send EPA back to the drawing board. With one minor exception, we deny the petitions for review. Faced with the complex task of identifying those geographic areas that contribute to fine particulate matter pollution, EPA both complied with the statute and, for all but one of the 225 counties or partial counties it designated as nonattainment, satisfied — indeed, quite often surpassed — its basic obligation of reasoned decisionmaking. I. Title I of the Clean Air Act charges EPA with formulating national ambient air quality standards (NAAQS) for air pollutants that may reasonably be anticipated to endanger public health and welfare. 42 U.S.C. §§ 7408-09. NAAQS set maximum ambient air concentrations for those pollutants. Id. While each state has “primary responsibility for assuring air quality” within its borders and, in particular, for developing a state implementation plan (SIP) for achieving and maintaining the NAAQS for each air pollutant, 42 U.S.C. § 7407(a), the Act triggers more or less stringent requirements depending on the quality of an area’s ambient air. Thus, before a state can design an appropriate SIP, it must know which areas within its boundaries comply with the NAAQS and which do not. This is where CAA section 107(d) comes in. It requires EPA to designate areas as “attainment,” “nonattainment,” or “unclassifiable” depending on their compliance with the relevant NAAQS. “Attainment” areas are those that meet the relevant standard; “nonattainment” areas are those that exceed the standard or that “contribute[] to ambient air quality in a nearby area” that exceeds the standard; “unclassifiable” areas are those that permit no determination given existing data. § 7407(d)(l)(A)(i)-(iii). In nonattainment areas, the Act requires stricter pollution controls. For instance, states must implement controls that will achieve attainment “as expeditiously as practicable” in nonattainment areas, id. § 7502(a), (c)(1), whereas states need only implement measures that will prevent “significant deterioration of air quality” for attainment and unclassifiable areas, id. § 7471. In addition to setting the criteria for attainment and nonattainment, section 107(d)(1) prescribes the designation process. Upon promulgation of new or revised NAAQS, states must submit to EPA their own “initial designations” of all areas within their borders. § 7407(d)(1)(A). EPA must then promulgate the submitted designations or modify them as it deems necessary. § 7407(d)(1)(B). Specifically, CAA section 107(d)(l)(B)(ii) provides that: In making the promulgations required ..., the Administrator may make such modifications as the Administrator deems necessary to the designations of the areas (or portions thereof) submitted [by the states] under subparagraph (A) (including to the boundaries of such areas or portions thereof). Whenever the Administrator intends to make a modification, the Administrator shall notify the State and provide such State with an opportunity to demonstrate why any proposed modification is inappropriate. The Administrator shall give such notification no later than 120 days before the date the Administrator promulgates the designation, including any modification thereto. § 7407(d)(l)(B)(ii). This case involves the NAAQS for fine particulate matter. Known as PM2.5, fine particulate matter consists of airborne particles that are 2.5 micrometers in diameter or smaller — i.e., less than one-thirtieth the thickness of a human hair. Air Quality Designations and Classifications for the Fine Particles (PM2.5) National Ambient Air Quality Standards (“PM25 Designations Rule ”), 70 Fed.Reg. 944, 945 (Jan. 5, 2005) (codified at 40 C.F.R. pt. 81). A “significant association” links elevated levels of PM2.b with adverse human health consequences such as premature death, lung and cardiovascular disease, and asthma. Id. And significantly for the primary issue before us — EPA’s method for identifying the geographic origins of elevated ambient PM25 concentrations — PM2t5 can travel hundreds or thousands of miles. In 1997, EPA abandoned its practice of regulating all particulate matter, both coarse and fíne, under a unified standard. Instead, it established specific PM2.5 NAAQS for the first time. National Ambient Air Quality Standards for Particulate Matter, 62 Fed.Reg. 38,652 (July 18, 1997). EPA promulgated annual and 24-hour PM2.5 NAAQS, setting the annual standard — the one at issue here — at 15 micrograms per cubic meter. Id. at 38,-677; see also 40 C.F.R. § 50.7. Although section 107(d) required EPA to promulgate area designations for the new standard “as expeditiously as possible,” § 7407(d)(l)(B)(i), litigation here and in the Supreme Court waylaid the designation process until we finally upheld the standard five years later. See Am. Trucking Ass’ns v. EPA 283 F.3d 355 (D.C.Cir. 2002), on remand from Whitman v. Am. Trucking Ass’ns, 531 U.S. 457, 121 S.Ct. 903,149 L.Ed.2d 1 (2001), affg in part and rev’g in part Am. Trucking Ass’ns v. EPA 175 F.3d 1027 (D.C.Cir.1999). In the meantime, Congress had passed legislation requiring the deployment of a nationwide PM2.5 monitoring network and extending the deadline for the designations until three years of monitoring data had been collected. See Transportation Equity Act for the 21st Century (“TEA-21”), Pub.L. No. 105-178, §§ 6101-02, 112 Stat. 107, 463-65 (1998). Then, following our decision upholding the PM25 NAAQS and EPA’s initiation of the designation process, Congress amended the Clean Air Act to add section 107(d)(6), which set a firm deadline for the PM2.5 area designations. See Pub.L. No. 108-199, § 425(a), 118 Stat. 3, 417 (2004) (codified at § 7407(d)(6)). Thus, amended section 107(d)(6)(A) now provides: Notwithstanding any other provision of law, not later than February 15, 2004, the Governor of each State shall submit designations referred in paragraph (1) for the July 1997 PM2.5 national ambient air quality standards for each area within the State, based on air quality monitoring data collected in accordance with any applicable Federal reference methods for the relevant area. § 7407(d)(6)(A). And section 107(d)(6)(B) provides: Notwithstanding any other provision of law, not later than December 31, 2005, the Administrator shall, consistent with paragraph (1), promulgate the designations referred to in subparagraph (A) for each area of each State for the July 1997 PM2.5 national ambient air quality standards. § 7407(d)(6)(B). In April 2003, EPA issued a guidance document initiating the PM25 designation process. Known as the Holmstead Memo, the document explains the timeline for state submissions and, more importantly, the criteria EPA would employ in reviewing those submissions. Memorandum from Jeffrey R. Holmstead, Assistant Administrator, to Regional Administrators, Regions I-X (Apr. 1, 2003) (“Holmstead Memo”). Noting “recent evidence that violations of the PM2 5 air quality standards generally include a significant urban-scale contribution,” the Holmstead Memo explains EPA’s “inten[t] to apply a presumption that the boundaries for urban nonattainment areas should be based on Metropolitan Area boundaries.” Id. at 2. In other words, the Memo announces EPA’s view that if any area within a metropolitan area exceeds the annual PM2.5 NAAQS, then all areas within the metropolitan area presumptively “contribute” to that violation within the meaning of section 107(d)(1)(A) and therefore warrant “nonattainment” designations. The Holmstead Memo further explains that EPA would define metropolitan boundaries by reference to Office of Management and Budget definitions of metropolitan statistical areas (MSAs) and consolidated metropolitan statistical areas (CMSAs) (interchangeably, “C/MSAs”)- Id. Thus, under this approach, a violation of the PM25 NAAQS in the District of Columbia, for instance, would trigger the presumption that seventeen counties in northern Virginia and five counties in Maryland — -all inside the applicable MSA, though only some are contiguous with the District— contribute to elevated PM2.5 levels in the city and warrant “nonattainment” designations. An attachment to the Holmstead Memo elaborates on EPA’s basis for adopting the presumption, as well as the circumstances that would warrant a departure. As to the rationale for the presumption, the Holmstead Memo explains that after “exam-in[ing] the geographic distribution of total PM25 concentrations in and near many metropolitan areas,” the agency had “found an association of higher PM2.B concentrations with greater levels of urban activity” such as “motor vehicle use and home heating[,] as well as industrial activities.” Holmstead Memo, Attach. 2, Guidance on Nonattainment Area Designations for PM2.5, at 4-5 (“Holmstead Memo Guidance”). Thus, “[the] presumption reflects EPA’s view that, in the absence of evidence to the contrary, violations of the PM2.5 NAAQS in urban areas may be presumed attributable at least in part to contributions from sources distributed throughout the Metropolitan Area.” Id. at 5. That said, the Holmstead Memo also recognizes that appropriate boundaries of urban nonattainment areas may well be smaller or larger than the applicable C/MSA. EPA would therefore “consider requests for urban nonattainment area definitions that deviate from OMB’s metropolitan area definitions on a case-by-case basis.” Id. at 6. The Holmstead Memo lists nine factors to guide that case-by-case analysis: (1) emissions in the potentially contributing areas; (2) air quality in those areas; (3) population density and degree of urbanization in those areas; (4) traffic and commuting patterns; (5) expected growth; (6) meteorology; (7) geography and topography; (8) jurisdictional boundaries; and (9) level of control of emissions sources. Id. at 7. The Memo encourages states submitting designations that depart from the metropolitan presumption to justify such designations by reference to all nine factors. Id. With that, the designation process was underway. The states submitted their initial designations; EPA responded with its proposed modifications, as well as explanations of its analysis under the nine-factor test; the states then submitted any responses; and EPA promulgated the final area designations in December 2004. See PMZ£ Designations Rule, 70 Fed.Reg. at 946. In a technical support document accompanying the Rule, EPA explained the basis for the designations and the analytical tools that it had developed and applied to assess the nine factors. See EPA, Office of Air Quality Planning and Standards, Technical Support Document for State and Tribal Air Quality Fine Particle (PM2.5) Designations (2004) (“Technical Support Document”). Recognizing that 2004 monitoring data would soon become available, EPA also invited states to submit any new data that might support an amended designation. PMZ5 Designations Rule, 70 Fed.Reg. at 948. Based on the new data, EPA then revised designations for eight areas from nonattainment to attainment and four areas from unclassifiable to attainment. See Air Quality Designations for the Fine Particles (PM2.5) National Ambient Air Quality Standards — Supplemental Amendments, 70 Fed.Reg. 19,844, 19,844 (Apr. 14, 2005) (codified at 40 C.F.R. pt. 81). In the end, EPA applied the C/MSA presumption so that nonattainment boundaries were coextensive with metropolitan boundaries (and unchanged by the nine-factor analysis) in only seven of thirty-nine metropolitan areas. See EPA Br. 54 (citing relevant portions of the Technical Support Document). In every other metropolitan area, applying the nine-factor analysis and finding that the presumption misjudged the nature of the PM2.5 problem, EPA designated the urban nonattainment area as either an area larger than the OMB-defined metropolitan area (for instance, in Chattanooga, Tennessee) or smaller (for instance, in Washington, DC). Several states, local governments, and industrial entities petitioned EPA for reconsideration of the designations; others filed petitions for review directly in this court. We stayed proceedings in this court while EPA considered the petitions for reconsideration. Once EPA resolved the petitions for reconsideration, we consolidated all petitions for review. Petitioners’ challenges fall into four broad categories. First, they raise procedural challenges to the Holmstead Memo and to the Designations Rule, arguing that EPA illegally bypassed notice and comment for each. Second, they raise various objections to EPA’s statutory authority to adopt and implement the C/MSA presumption and the nine-factor test. Third, they argue that even if section 107(d) permits EPA to adopt the C/MSA presumption and the nine-factor test, EPA’s analysis nonetheless suffers from such serious methodological deficiencies and inconsistencies as to render the entire Designations Rule arbitrary and capricious. Finally, as a last resort, petitioners request that we vacate certain area designations that affect them, claiming that EPA at least acted arbitrarily and capriciously in making these particular nonattainment and unclassifiable designations. We review petitioners’ challenges under section 307(d)(9) of the Clean Air Act, which requires the court to set aside EPA’s final actions when they are in excess of the agency’s statutory authority or otherwise arbitrary and capricious. 42 U.S.C. § 7607(d)(9)(a). II. Before addressing petitioners’ arguments, however, we think it helpful to review some technical background. Fine particulate matter includes both “primary” particles (e.g., carbonaceous particles and so-called “crustal” particles like dust) that pollution sources emit directly into the atmosphere, as well as “secondary” particles (e.g., sulfate and nitrate particles) that form in the atmosphere as a result of chemical reactions between PM25 precursors that sources emit. PM25 Designations Rule, 70 Fed.Reg. at 945. Power plants, diesel and gasoline powered engines in mobile sources like cars and trucks, and other industrial sources produce most carbonaceous particles; agriculture, mining, and other activities that cause soil or metals to be suspended in the atmosphere account for the crustal component. See Technical Support Document § 3.1; EPA, Office of Air Quality and Planning Standards, The Particle Pollur tion Report 6 (2004), available at http:// www.epa.gov/air/airtrends/aqtrnd04/pm. html. The chemical precursors to secondary PM2.5 include sulfur dioxide (S02), emitted in substantial part by power plants; nitrogen oxides (NOx), emitted in substantial part by mobile sources, power plants, and other industrial sources; and ammonia, emitted from agricultural sources, mobile sources, and power plants. See PM2£ Designations Rule, 70 Fed.Reg. at 945; Technical Support Document § 3.1. Atmospheric chemical reactions between these gases yield secondary PM2,5 in the form of sulfate and nitrate particles. PM2.5 Desig nations Rule, 70 Fed.Reg. at 945. The PM2.5 NAAQS set a 15g/m3 annual limit for all fine particulate matter without distinguishing among the various kinds (or “species”) of PM2.6. Even so, “speciation data” that breaks the total PM2.5 concentration into its constituent components is quite useful for the area designation process. Because such data reveals the kinds of particles (carbon, sulfate, nitrate, crustal particles, etc.) that most account for an area’s PM2.5 problem, it suggests, by extrapolation, the kinds of sources most responsible for the problem as well. While the Holmstead Memo announced EPA’s methodology for determining whether an area with passable PM25 concentrations nonetheless deserves a “nonattainment” designation, EPA regulations set forth the technical procedures for measuring ambient PM2 5 concentrations in the first place. Exhaustive technical specifications regulate the states’ operation of a network of air monitors that collect air quality data for any given area. See 40 C.F.R. pt. 50, Apps. L, N. These monitors measure ambient PM2.5 concentration, what’s called the “design value,” on any given day. The annual design value— which determines whether an area complies with the PM2.5 standard- — -is then computed by averaging every quarter’s worth of daily design value samples (typically collected every third or sixth day), averaging those quarterly numbers to obtain an annual average, and then averaging three years of annual numbers to yield the final annual design value. See 40 C.F.R. pt. 50, App. N. Samples from these monitors can be further analyzed to yield the speciation data described above. The majority of the PM2.5 designations at issue here drew on monitoring data collected from 2001 to 2003. Based on a judgment that no petitioner challenges, EPA decided that “the county boundary ... [would] determin[e] the extent of the area reflected by [a violating] PM2.5 monitor.” PM2.5 Designations Rule, 70 Fed. Reg. at 946; see also id. at 946-47 (“[I]f a PM2.5 monitor was violating the standard based on the 2001-2003 data, at a minimum we designated the county where that monitor is located as nonattainment. We made exceptions ... in a few very large western counties where a significant geographic feature such as a mountain range divided a county....”). Thus, for instance, if a monitor in an industrial area of downtown Detroit registered a violation, all of Wayne County would be designated as nonattainment — including even its more idyllic corners like the town of Grosse Pointe. Cf. id. at 980 (designating Wayne County nonattainment). To be clear, this reflects quite a separate judgment from that underlying the C/MSA presumption. EPA’s selection of the county as the unit of analysis resolved a problem inherent in the monitoring process, namely, that a monitor only measures air quality in its immediate vicinity. Because of this, EPA had to determine how much compass to give any monitored measurement, which it did by choosing the county as the unit of analysis. The C/MSA presumption, by contrast, addresses a different problem, namely, how to identify those areas that, although deemed to be meeting the standard themselves, are contributing to nearby violations. We have already described the nine factors that EPA evaluated to determine whether to depart from the C/MSA presumption, see supra at 8, and in the hundreds of pages that comprise the Technical Support Document, EPA explained its findings for each metropolitan area on each of the nine factors. EPA assessed these factors with the help of several analytical tools and models it had developed. See Technical Support Document §§ 3.0-5.9 (explaining analytical tools). We describe the most important ones here in general fashion, reserving additional elaboration for those portions of the opinion that require it. To start, given the hundreds of miles that PM2.5 can travel in the atmosphere, EPA thought it important to isolate the portion of urban PM2.5 that originates from a metropolitan area’s local sources as opposed to regional sources much farther away. Thus, under its “urban excess” analysis, EPA paired an urban monitoring site with an upwind rural monitoring site— i.e., a rural site where prevailing winds move in the direction of the metropolitan area — and “subtracted] the rural concentration from the measured urban concentration.” Id. § 3.1. True to name, this simple arithmetic allowed EPA to estimate the portion of urban PM25 levels that arises from urban activities by cancelling out the “rural background” that would exist regardless of those activities. Id. EPA calculated the urban excess for each PM2.5 species and then summed those numbers to yield a total urban excess measure. Id. § 3.2. EPA next used the urban excess calculations to develop “weighted emissions scores” for each county in a C/MSA. Id. § 4.0. The notion underlying these scores is intuitive: if the urban excess numbers for the District of Columbia, for instance, reveal that all local PM2 5 pollution is in the form of carbon, it would make little sense to think that a C/MSA county plays a significant part in the District’s monitored violation if the county has zero carbon emissions. Thus, for purposes of evaluating the first of the nine factors — the “emissions in areas potentially included versus excluded,” Holmstead Memo Guidance at 7- — -EPA determined that raw emissions data is usually less suggestive of contribution than data adjusted to account for the PM25 species that actually comprise the urban excess. Calculating weighted emissions scores required a number of steps. See Technical Support Document §§ 4.1-4.3. First, EPA determined the total metropolitan emissions of carbon, S02, NOXj and crustal particles by summing the counties’ individual emissions of each pollutant. For those pollutants, EPA then divided each county’s emissions by total C/MSA emissions, calculating each county’s percentage share of total metropolitan emissions. For example, if County A emits 50 tons of carbon and if total C/MSA carbon emissions equal 100 tons, then the ratio would be fifty percent. Next EPA “weighted” these percentages by multiplying them by the proportion of urban excess attributable to the relevant pollutant. To continue with our hypothetical, then, if carbon accounted for forty percent of urban excess, EPA would multiply fifty percent by four-tenths, and County A’s weighted carbon score would be twenty. Having calculated a county’s weighted scores for each PM2.5 species, EPA then added these scores together to derive a county’s total weighted emissions score. Importantly, because these scores scale a county’s raw emissions based on attributes specific to an individual C/MSA — i.e., the urban excess number and the total level of metropolitan emissions— they only provide a measure for comparing counties within the same C/MSA. Finally, EPA developed so-called “pollution roses” that depict 2001-2003 monitoring and meteorological data for each PM2,5 air monitor. See, e.g., Technical Support Document 6-11 to -12. Each pollution rose consists of concentric circles, with the circles’ center representing the location of the air monitor. EPA then plotted dots around the circles, with each dot representing one monitored reading, the dot’s size representing the magnitude of the reading, the dot’s spatial location representing the prevailing wind direction on the day of the reading, and the dot’s distance from the center of the circle representing the average wind speed on that day. With this technical background in mind, we turn to petitioners’ four primary arguments. III. Petitioners first lodge procedural challenges against EPA’s promulgation of the final designations rule and the Holmstead Memo. Petitioners argue that EPA violated the Administrative Procedure Act, 5 U.S.C. § 553, by failing to publish both the Rule and the Holmstead Memo for notice and comment. They are in error as to both. A. Final Designations Rule Designation of nonattainment areas is governed by section 107(d) of the Clean Air Act. § 7407(d). The parties disagree as to which subsection of section 107(d) provides EPA’s authority to promulgate designations. Industry petitioners argue that designations are promulgated under section 107(d)(6), which states: “Notwithstanding any other provision of law, not later than December 31, 2004, the Administrator shall, consistent with paragraph (1), promulgate the designations referred to in subparagraph (A) for each area of each State for the July 1997 PM2.5 national ambient air quality standards.” § 7407(d)(6)(B). By contrast, EPA argues that the source of its authority is a provision of section 107(d)(1), entitled “Promulgation by EPA of designations,” which states in relevant part: Upon promulgation or revision of a national ambient air quality standard, the Administrator shall promulgate the designations of all areas (or portions thereof) submitted under subparagraph (A) as expeditiously as practicable, but in no case later than 2 years from the date of promulgation of the new or revised national ambient air quality standard. § 7407(d)(l)(B)(i). The distinction between these provisions is important because the statute exempts designations under section 107(d)(1), among others, from the APA’s section 553 notice-and-comment requirements; it does not, however, exempt designations under section 107(d)(6). See § 7407(d)(2)(B) (“Promulgation or announcement of a designation under paragraph (1), (4) or (5) shall not be subject to the provisions of sections 553 through 557 of title 5 of the United States Code (relating to notice and comment), except nothing herein shall be construed as precluding such public notice and comment whenever possible.”). EPA is entrusted with administering the Clean Air Act, of which section 107 is a part, and thus we review the agency’s construction of the statutory provisions under the familiar two-step framework set out in Chevron U.S.A. Inc. v. Natural Resources Defense Council, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Under Chevron step one, we ask “whether Congress has directly spoken to the precise question at issue.” Id. at 842, 104 S.Ct. 2778. If at that point we determine that “the intent of Congress is clear, 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.” Id. at 842-43, 104 S.Ct. 2778. We proceed to Chevron’s second step only “if the statute is silent or ambiguous with respect to the specific issue.” Id. at 843, 104 S.Ct. 2778. At the second step, we determine “whether the agency’s answer is based on a permissible construction of the statute.” Id. Here we need go no further than the first step because the intent of Congress is clear. Petitioners’ reliance on section 107(d)(6) is misplaced. Subparagraph (B) of section 107(d)(6) requires that, “not later than December 31, 2004, the Administrator shall ... promulgate the designations referred to in subparagraph (A).” § 7407(d)(6)(B). Subparagraph (A), in turn, states: “Notwithstanding any other provision of law, not later than February 15, 2004, the Governor of each State shall submit designations referred to in paragraph (1) [of § 7407(d)] for the July 1997 PM2.5 national ambient air quality standards .... ” § 7407(d)(6)(A) (emphasis added). Thus, contrary to petitioners’ claim, section 107(d)(6)(B), when read in conjunction with section 107(d)(6)(A), shows that section 107(d)(6) does not itself authorize the promulgation of designations. Rather, section 107(d)(6) merely governs the timing of PM2.5 designations, which are made under the authority contained in section 107(d)(1) — a provision that the statute expressly exempts from notice-and-comment requirements. See § 7407(d)(2)(B). B. Holmstead Memo Petitioners’ argument that the Holmstead Memo had to undergo notice and comment stems, in part, from their erroneous belief that the final designations were subject to notice and comment. Our determination above, that the statute exempts the nonattainment designations from notice-and-comment procedures, suggests that the Holmstead Memo — which was simply the first step in the promulgation of designations — is also exempt. Petitioners’ further argument, that the Holmstead Memo is a legislative rule that must undergo notice and comment independent of the final rule to which it relates, is also unavailing. For support, petitioners point to General Electric Co. v. EPA, 290 F.3d 377 (D.C.Cir.2002), in which we explained that whether an agency action is the type of action that must undergo notice and comment depends on “whether the agency action binds private parties or the agency itself with the ‘force of law,’ ” id. at 382 — that is, whether “a document expresses a change in substantive law or policy (that is not an interpretation) which the agency intends to make binding, or administers with binding effect,” id. at 382-83 (quoting Robert A. Anthony, Interpretive Rules, Policy Statements, Guidances, Manuals, and the Like — Should Federal Agencies Use Them to Bind the Public?, 41 Duke L.J. 1311, 1355 (1992)). General Electric further explained that “an agency pronouncement will be considered binding as a practical matter if it either appears on its face to be binding ... or is applied by the agency in a way that indicates it is binding.” Id. at 383 (internal citation omitted). In response, EPA argues that the Holmstead Memo is merely a policy statement, not a legislative rule, because it does not create or modify legally binding rights or obligations. As EPA correctly notes, the APA expressly exempts policy statements from notice-and-comment requirements. See 5 U.S.C. § 553(b)(A) (specifying that, except when required by statute, the section 553 requirements for notice and comment do not apply “to interpretative rules, general statements of policy, or rules of agency organization, procedure, or practice”). EPA has the better of this dispute. First, the Holmstead Memo is not binding on its face. It specifies that it merely “provides guidance to State and local air pollution control agencies ... on the process for designating areas for the purpose of implementing the fine particle national ambient air quality standards.” Holmstead Memo at 1. It then explicitly states that it is “not binding” on the states or EPA and notes that it provides only EPA’s “current views” on the designation process, suggesting that those views are open to revision. Id. at 2. Unlike the agency documents at issue in General Electric and CropLife America v. EPA, 329 F.3d 876 (D.C.Cir.2003), which petitioners also cite, the Holmstead Memo does not impose binding duties on states or the agency. It merely clarifies the states’ existing duties under the Clean Air Act and explains the process EPA suggests for states to follow in providing their initial designations. As we explained above, see supra at 27-28, the Memo establishes a rebuttable C/MSA presumption and outlines nine factors for EPA to consider in its final designations, see Holmstead Memo Guidance at 5-7. Petitioners point to language in the Holmstead Memo that they view as evidence of its binding character vis-a-vis the states: “A demonstration supporting the designation of boundaries that are less than the full metropolitan area must show both that violations are not occurring in the excluded portions of the metropolitan area and that the excluded portions are not source areas that contribute to the observed violations.” Id. at 7. But this language does not create a new burden on the states; it merely reiterates the statutory requirements. See § 7407(d)(1)(A)® (requiring Governors to designate as non-attainment “any area that does not meet (or that contributes to ambient air quality in a nearby area that does not meet)” the NAAQS). Nor does the Holmstead Memo bind EPA. The Memo announces the C/MSA presumption as a rebuttable presumption, which preserves the agency’s discretion to deviate from the boundaries of a C/MSA in the final designations. See Panhandle Producers & Royalty Owners Ass’n v. Econ. Regulatory Admin., 822 F.2d 1105, 1110 (D.C.Cir.1987) (“ ‘An agency pronouncement is not deemed a binding regulation merely because it may have some substantive impact, as long as it leave[s] the administrator free to exercise his informed discretion.’ ... Presumptions, so long as rebuttable, leave such freedom.” (quoting Brock v. Cathedral Bluffs Shale Oil Co., 796 F.2d 533, 537 (D.C.Cir.1986) (internal quotation marks omitted))). Further, EPA has not applied the Holmstead Memo in a binding manner. Petitioners again cite General Electric for the proposition that an agency document will be considered binding if “the affected private parties are reasonably led to believe that failure to conform will bring adverse consequences.” Gen. Elec., 290 F.3d at 383 (quoting Anthony, supra, at 1328). The Memo “encouraged” states to address all nine factors EPA identified, but did not require them to do so. Holmstead Memo Guidance at 7. Some states did not address all or even any of the factors. See, e.g., Letter from Robert G. Burnley, Dep’t of Envtl. Quality, Commonwealth of Virginia, to Donald S. Welsh, U.S. EPA Region III (Feb. 13, 2004) (recommending that all of Virginia be designated attainment without addressing any of the nine factors); Letter from Stephanie R. Timmermeyer, West Virginia Dep’t of Envtl. Prot., to Donald S. Welsh, U.S. EPA Region III (Feb. 13, 2004) (recommending PM2.5 nonattainment areas to match the existing ozone nonattainment areas without addressing any of EPA’s other eight factors). EPA considered such submissions and did not impose “adverse consequences,” notwithstanding the states’ failure to address the factors listed in the Holmstead Memo. Compare State of West Virginia PM2.5 Designations — Preliminary Recommendations, with PM2.5 Designations Rule, 70 Fed. Reg. at 1014-15 (showing that EPA designated as attainment an area that West Virginia had proposed be designated non-attainment, despite the fact that West Virginia did not address eight of EPA’s nine factors). In sum, we deny petitioners’ procedural claims because EPA was not required to submit either the final designations rule or the Holmstead Memo for notice and comment. IY. Petitioners next claim that EPA violated section 107(d) of the Clean Air Act by applying the C/MSA presumption and nine-factor test to identify areas that contribute to nearby PM25 violations. As we explained above, we review EPA’s interpretation of the Clean Air Act under Chevron, asking whether Congress has “directly spoken to the precise question at issue,” 467 U.S. at 842, 104 S.Ct. 2778, and if so, whether it has unambiguously foreclosed the agency’s statutory interpretation, e.g., Sierra Club v. EPA, 536 F.3d 673, 677 (D.C.Cir.2008). But if the statute is either silent or ambiguous on the specific question at issue, we defer to EPA’s statutory interpretation so long as it is reasonable. Id. At the outset we observe that section 107(d) is replete with the kinds of words that suggest a congressional intent to leave unanswered questions to an agency’s discretion and expertise, see Nat’l Cable & Telecomms. Ass’n v. Brand X Internet Servs., 545 U.S. 967, 980, 125 S.Ct. 2688, 162 L.Ed.2d 820 (2005) (statutory ambiguity is delegation to the agency “to fill the statutory gap in reasonable fashion”). Section 107(d) requires EPA to designate an area as nonattainment if it “contributes to ambient air quality in a nearby area” that exceeds the relevant standard, § 7407(d)(l)(A)(i), yet the statute defines neither “contributes” nor “nearby” — words that we have expressly found ambiguous as used in other sections of the Act. See Envtl. Def. Fund v. EPA, 82 F.3d 451, 459 (D.C.Cir.1996) (“contributes to” ambiguous in another section of the Clean Air Act); Sierra Club v. EPA, 719 F.2d 436, 443-44 (D.C.Cir.1983) (same, as to “nearby”). It authorizes EPA to revise state-submitted designations whenever it “deems” such modifications “necessary,” yet it says nothing of what precisely will render a modification “necessary.” § 7407(d)(l)(B)(ii). And section 107(d) requires states to submit PM2.5 designations “based on air quality monitoring data collected in accordance with any applicable Federal reference methods,” § 7407(d)(6)(A), yet it fails to define “based on” and “[tjhere is no question that the phrase ‘based on’ is ambiguous,” Sierra Club v. EPA, 356 F.3d 296, 305-06 (D.C.Cir.2004); accord. Nuclear Energy Inst, Inc. v. EPA, 373 F.3d 1251, 1269 (D.C.Cir.2004). “[A]mbiguities in statutes within an agency’s jurisdiction to administer are delegations of authority to the agency to fill the statutory gap in reasonable fashion.” Brand X, 545 U.S. at 980, 125 S.Ct. 2688. Because it conveys no clear-cut approach for determining whether an area contributes to a nearby PM25 violation, section 107(d)’s text is consistent with such a delegation. To be sure, a statute may foreclose an agency’s preferred interpretation despite such textual ambiguities if its structure, legislative history, or purpose makes clear what its text leaves opaque. Cf. Ariz. Pub. Serv. Co. v. EPA, 211 F.3d 1280, 1287 (D.C.Cir.2000) (court must “exhaust[ ] traditional tools of statutory construction” at Chevron step one). Notwithstanding petitioners’ torrent of arguments to the contrary, this is not such a case — indeed, it isn’t even close. We start with the argument that petitioners judge to be their best. See Oral Arg. at 1:16-3:00. Pointing to section 107(d)(4), petitioners insist that the statute’s express mandate that EPA apply the C/MSA presumption in other contexts conclusively proves that Congress intended to preclude its use here. Enacted as part of the 1990 Amendments to the Clean Air Act, section 107(d)(4) “revise[s] ... by operation of law” the boundaries of certain urban ozone or carbon monoxide nonattainment areas “to include the entire metropolitan statistical or consolidated metropolitan statistical area,” unless EPA determined that some portions “do not contribute significantly to the violation of the national ambient air quality standard.” § 7407(d)(4)(A)(iv)-(v). In contrast, section 107(d)(6) says nothing about the C/MSA presumption. Instead it provides that the PM2,5 area designations must be “based on air quality monitoring data” and promulgated in accordance with section 107(d)(l)’s general provisions for area designations. § 7407(d)(6)(A). Citing the familiar canon of statutory interpretation that “[w]here Congress includes particular language in one section of a statute but omits it from another section of the same Act, it is generally presumed that Congress acts intentionally and purposely in the disparate inclusion or exclusion,” Russello v. United States, 464 U.S. 16, 23, 104 S.Ct. 296, 78 L.Ed.2d 17 (1983) (internal quotation marks and brackets omitted), petitioners insist that the statute’s exclusive inclusion of the C/MSA presumption for the ozone and carbon monoxide designations demonstrates Congress’ unambiguous intent to preclude EPA from adopting the presumption here. Although petitioners are correct that we construe statutes to give meaning to the disparate inclusion of particular language, that principle hardly compels the interpretation they favor. When interpreting statutes that govern agency action, we have consistently recognized that a congressional mandate in one section and silence in another often “suggests not a prohibition but simply a decision not to mandate any solution in the second context, i.e., to leave the question to agency discretion.” Cheney R. Co. v. ICC, 902 F.2d 66, 69 (D.C.Cir.1990); see also Clinchfield Coal Co. v. Fed. Mine Safety & Health Review Comm’n, 895 F.2d 773, 779 (D.C.Cir.1990) (“[W]here an agency is empowered to administer the statute, Congress may have meant that in the second context the choice should be up to the agency.”). Silence, in other words, may signal permission rather than proscription. For that reason, that Congress spoke in one place but remained silent in another, as it did here, “rarely if ever” suffices for the “direct answer” that Chevron step one requires. Cheney, 902 F.2d at 69 (internal quotation marks omitted); see also Am. Forest & Paper Ass’n v. FERC, 550 F.3d 1179, 1181 (D.C.Cir.2008) (statute’s discrepant inclusion of the modifier “competitive” to describe “markets” renders statutory provision lacking the modifier ambiguous). Undaunted, petitioners insist that the silence here is unambiguously prohibitive. They point out that Congress not only refused to treat PM2.5 like ozone and carbon monoxide but chose an altogether different scheme in subsection (d)(6) — one “based on air quality monitoring data,” § 7407(d)(6)(A), rather than OMB-drawn metropolitan boundaries and a multi-factor test. As indicated above, however, Congress’s mere choice of different standards in subsections (d)(4) and (d)(6) tells us nothing about whether Congress wanted to mandate different approaches or to permit them. Thus, for this argument to succeed, subsection (d)(6) must itself preclude EPA from adopting a test for PM25 like the one that Congress mandated for ozone and carbon monoxide in subsection (d)(4). But nothing in subsection (d)(6) even hints at such a prohibition. Subsection (d)(6)(A) requires states to submit PM25 designations “based on air quality monitoring data collected in accordance with any applicable Federal Reference methods”; subsection (d)(6)(B) in turn requires EPA to promulgate those designations in accordance with subsection (d)(l)’s general provisions. For the sake of argument, we shall assume that subsection (d)(6)’s “based on” language unambiguously applies to EPA even though it appears only in the particular provision governing states. We shall also assume that the language substantively constrains EPA’s discretion in determining nonattainment boundaries. But even given these assumptions, binding precedent, the statute’s purpose, and basic common sense foreclose petitioners’ argument that section 107(d)(6) itself precludes EPA from adopting the C/MSA presumption and nine-factor test. First, as noted above, we have repeatedly held that the words “based on” are unquestionably ambiguous: they neither compel the agency to rest its decisions “solely on” the specified factor nor indicate the extent to which the agency may rely on additional factors. Sierra Club, 356 F.3d at 305-06; accord. Nuclear Energy Inst., 373 F.3d at 1269. Instead, they simply constrain the agency from “abandoning]” or “supplanting]” the specified factor altogether. Sierra Club, 356 F.3d at 306. We need ask only whether EPA’s PM2.5 designations “can still reasonably be described as ‘based on’ ” air quality monitoring data, id., and we have no doubt that the contribution designations meet this modest standard. If nothing else, the contribution designations are “based on” the air quality monitoring data that demonstrate a nearby violation of the NAAQS. Even were this insufficient, EPA’s use of air quality monitoring data as part of its nine-factor analysis — for instance, its use of urban excess and weighted emissions scores based on speciation data — would surely suffice. Second, because subsection d(l) directs EPA to designate some areas as nonattainment despite monitoring data that provides no basis, on its own, to do so, see § 7407(d)(l)(A)(i), we think it quite clear that the statute contemplates reliance on factors other than monitoring data to determine contribution. How could EPA possibly fulfill its statutory duty to determine, for instance, whether emissions in Indiana contribute to monitored violations in Chicago without considering wind and emissions data from Indiana? Obviously it could not. That the statute fails to set forth the additional criteria for EPA to consider in evaluating contribution hardly forecloses EPA from developing such criteria in order to accomplish Congress’s objectives. See e.g., Entergy Corp. v. Riverkeeper, Inc., — U.S. -, 129 S.Ct. 1498, 1508, 173 L.Ed.2d 369 (2009). Indeed, when a statute is “silent ... with respect to all potentially relevant factors, it is eminently reasonable to conclude that [the] silence is meant to convey nothing more than a refusal to tie the agency’s hands.” Id. (emphasis added). Still undaunted, petitioners advert to section 107(d)(3)^-which permits EPA to require a revision to an area designation “on the basis of air quality data, planning and control considerations, or any other air quality-related considerations the Administrator deems appropriate,” § 7407(d)(3) — as evidence that Congress “knew how to authorize a flexible, ‘multifactor’ approach to attainment decisions when it so intended.” Counties’ Opening Br. 20. But the Supreme Court rejected just this kind of argument in Entergy Corp. v. Riverkeeper, Inc., which asked whether EPA may engage in cost-benefit analysis under section 316(b) of the Clean Water Act given that it says nothing of cost-benefit analysis while other sections expressly authorize it. 129 S.Ct. at 1508. Pointing out that section 316(b) “is silent not only with respect to cost-benefit analysis but with respect to all potentially relevant factors,” the Court rejected the claim that the silence reflected a prohibition, for if it did, “then the EPA could not consider any factors in implementing [section 316(b)]' — an obvious logical impossibility.” Id. Here we face just this kind of overwhelming statutory silence. Although logic dictates that EPA must evaluate some factors in addition to monitoring data to determine contribution, the statute says nothing about which factors it should consider. We thus have no difficulty rejecting the claim that the statute unambiguously forecloses EPA from adopting the C/MSA presumption or considering its nine factors in applying it. The legislative history petitioners cite fails to rehabilitate their claim that section 107(d)(6) unambiguously requires EPA to base all PM2.5 designations — including non-attainment designations for contribution— “on air quality monitoring data” alone. Although petitioners are correct that in the Transportation Equity Act for the 21st Century, Congress provided for the deployment of a PM2.5 monitoring network to produce adequate monitoring data for these designations, that does nothing to dispel the ambiguity over what criteria EPA should rely on to assess contribution. Indeed, the particular provision of TEA-21 that petitioners emphasize actually reinforces that ambiguity rather than resolves it. See TEA-21 § 6102(c)(1) (“Only data from the monitoring network ... shall be considered for such designations. Nothing in the previous sentence shall be construed as affecting ... the Administrator’s authority to promulgate the designation of an area as nonattainment, under section 107(d)(1) of the Clean Air Act, based on its contribution to ambient air quality in a nearby nonattainment area.”). In sum, we conclude that neither section 107(d)’s requirement that PM2.5 designations be “based on air quality monitoring data” nor its mandate that EPA apply the C/MSA presumption and a multi-factor test for pollutants other than PM2 5 unambiguously reveals Congress’s intent to prevent EPA from using the presumption and the nine-factor test to determine contribution here. Our rejection of petitioners’ purportedly strongest argument is a fair harbinger of the fate of their remaining statutory complaints. To begin with, petitioners waived two of their statutory arguments by failing to raise them in their opening briefs. See New York v. EPA, 413 F.3d 3, 20 (D.C.Cir.2005) (petitioners waive those arguments that they fail to raise in their opening briefs). Specifically, not until their reply briefs did they present a statutory challenge to EPA’s interpretation that a “nearby” area under section 107(d)(1)(A) may include noncontiguous areas, or to its conclusion that an area’s future reductions in emissions qualifies as a relevant factor for assessing contribution. This leaves just two general challenges: petitioners’ claim that the C/MSA presumption and the nine-factor test run afoul of the statutory term “contribute”; and their claim that the presumption impermissibly encroaches on states’ statutory prerogative to have first-say on area designations within their borders. As to the textual claim, petitioners insist that the verb “contribute” necessarily connotes a significant causal relationship, meaning that EPA may not designate a county as contributing to nonattainment if “corrective measures in [the county] will do nothing to address the problem or help achieve compliance in the nonattainment area.” Counties’ Opening Br. 25. We reject both the major and the minor premise. Although petitioners cite one dictionary that supports the claim that the adverb “significantly” is implicit in the verb “contribute,” EPA cites other dictionaries that define “contribute” without reference to any threshold level of significance. This alone suggests an ambiguity that fatally undermines petitioners’ Chevron step one argument. See Cellular Telecomms. & Internet Ass’n v. FCC, 330 F.3d 502, 509 (D.C.Cir.2003) (“[Djueling over dictionary definitions is pointless, for it fails to produce any plain meaning of the disputed word.”). But even were we to think that “contribute” unambiguously means “significantly contribute,” we still disagree that “significantly contribute” unambiguously means “strictly cause.” Cf. Michigan v. EPA 213 F.3d 663, 667-68 (D.C.Cir.2000) (“significant” is ambiguous). Given that the statute uses the word “contribute” and that a contribution may simply exacerbate a problem rather than cause it, we see no reason why the statute precludes EPA from determining that a county’s addition of PM2.5 into the atmosphere is significant even though a nearby county’s nonattainment problem would still persist in its absence. In fact, a contrary interpretation of “contribute” would effectively preclude a nonattainment designation for any attaining county when the cause of the violation is metropolitan-wide. We may not interpret “contribute” in a way that does such violence to section 107(d)’s very purpose. We also reject petitioners’ argument that EPA violated the statute by failing to articulate a quantified amount of contribution that would trigger a nonattainment designation. Petitioners apparently prefer a bright-line, “objective” test of contribution, see, e.g., Counties’ Opening Br. 28, but it is the statute, not petitioners’ preferences, that constrains EPA’s discretion. And nothing in the statute compels EPA to quantify a uniform amount of contribution below which counties will automatically escape nonattainment designations or to quantify similar thresholds for the nine factors EPA evaluated in making those determinations. Section 107(d) is ambiguous as to how EPA should measure contribution and what degree of contribution is sufficient to deem an area nonattainment, as even petitioners seem to concede, see Counties’ Reply Br. 13 (“EPA was supposed to be defining and deciding ‘contributes.’ ”). Thus, reasonably exercising the discretion that Congress delegated to it, EPA interpreted “contribute” to mean “sufficiently contribute,” and then applied the C/MSA presumption and nine-factor test precisely to identify those areas that meet that definition. Petitioners offer no plausible reason to think that the statute forecloses this approach. Nor do we agree with petitioners that EPA’s failure to quantify its analysis somehow rendered its interpretation of “contribute” arbitrary and capricious and therefore unreasonable under Chevron step two. Cf. Northpoint Tech., Ltd. v. FCC, 412 F.3d 145, 151 (D.C.Cir.2005) (statutory interpretation that is arbitrary and capricious is unreasonable under Chevron step two). An agency is free to adopt a totality-of-the-circumstances test to implement a statute that confers broad discretionary authority, even if that test lacks a definite “threshold” or “clear line of demarcation to define an open-ended term.” PDK Labs., Inc. v. DEA 438 F.3d 1184, 1195 (D.C.Cir.2006) (internal quotation marks omitted). To be reasonable, such an “all-things-considered standard” must simply define and explain the criteria the agency is applying, id. at 1194, something the Holmstead Memo and certainly the Technical Support Document did in spades. Of course, EPA may have applied its nine-factor test inconsistently, resulting in similar counties being treated dissimilarly — a question we address in Part V. EPA may also have applied it so erroneously in a particular case that it could not have reasonably concluded that a county was contributing to nearby violations — an issue we consider in Part VI. But EPA’s use of a flexible multi-factor analysis is not in and of itself unreasonable just because it lacks quantitative standards. See id. at 1194-95. We are equally unimpressed by petitioners’ last general argument: that the C/MSA presumption unlawfully “deprived states of the deference to which their designations were entitled” under section 107(d). States’ Opening Br. 1. To the extent petitioners are claiming that EPA owes the states a measure of procedural deference under section 107(d), we agree that EPA must wait its turn before it makes any individual county designations. Indeed, in contrast to its many ambiguities, section 107(d)(1) clearly provides that states submit their “initial designations” first, § 7407(d)(1)(A), and only then does EPA promulgate or modify the designations as it “deems necessary,” § 7407(d)(l)(B)(ii). Not only is that precisely what happened here, but nothing in section 107(d)(1) prevents EPA from developing general principles to govern its exercise of discretion when the time comes, or from announcing those general principles before the states submit their initial designations. To the extent petitioners think that EPA owes the states a measure of substantive deference under section 107(d)(1) — a claim that seems implicit in their objection that the C/MSA presumption somehow alters states’ “burden” in the designation process, see States’ Opening Br. 24 — we disagree. Though EPA may, of course, go along with states’ initial designations, it has no obligation to give any quantum of deference to a designation that it “deems necessary” to change. See, e.g., Pa. Dep’t of Envtl. Prot. v. EPA 429 F.3d 1125, 1129 (D.C.Cir.2005) (recognizing that section 107(d) gives “deference” to states’ initial designations provided EPA deems no modification necessary). In short, EPA had authority to apply the C/MSA presumption at the time it rejected the states’ submissions. We see no reason why section 107(d)(1) would force EPA to bite its tongue until then. Having rejected petitioners’ arguments that section 107(d) unambiguously precludes EPA from adopting the C/MSA presumption and nine-factor test, we can easily conclude that EPA reasonably interpreted the statute as permitting it to do so. Even if we read section 107(d) favorably to petitioners, it requires only that EPA designate, based on air quality monitoring data, nonattainment areas that either violate or contribute to violations of the PM2.5 NAAQS. Acting on evidence that urban PM2.5 violations usually stem from metropolitan-wide activities, EPA adopted a presumption that designates all metropolitan areas as nonattainment when at least one metropolitan area registers a PM2.5 violation, as well as a specifically defined multi-factor analysis to assess when that presumption fails to reflect the realities of a given metropolitan area. Petitioners give us every reason to think they would prefer another system of analysis— specifically, one that would allow them to escape certain nonattainment designations — but they give us no reason to think the system EPA selected is unreasonable. V. Petitioners next argue that the Designations Rule is arbitrary and capricious because it is riddled with methodological flaws and inconsistencies. They challenge the Rule in four regards: EPA’s designation of noncontiguous townships as nonattainment, the agency’s refusal to account for some potential emissions reductions, the so-called carbon error in some weighted emissions scores, and the way EPA applied the nine-factor test. “[W]e apply the same standard of review under the Clean Air Act as we do under the Administrative Procedure Act (APA), 5 U.S.C. § 706(2)(A),” Allied Local & Reg'l Mfrs. Caucus v. EPA 215 F.3d 61, 68 (D.C.Cir.2000), and will set aside the Designations Rule only if it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law,” § 7607(d)(9)(A). We must affirm the Rule if the record shows EPA considered all relevant factors and articulated a “rational connection between the facts found and the choice made.” Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962). Of particular note in this challenge, we give an “extreme degree of deference to [EPA] when it is evaluating scientific data within its technical expertise,” City of Waukesha v. EPA 320 F.3d 228, 247 (D.C.Cir.2003) (internal quotation marks omitted). Such deference is especially appropriate in our review of EPA’s administration of the complicated provisions of the Clean Air Act. See Nat’l Ass’n of Clean Air Agencies v. EPA 489 F.3d 1221, 1229 (D.C.Cir.2007). In basing its designation decisions on a rigorous analysis of each county’s particular attributes, EPA satisfied the requirements of reasoned decisionmaking. Given our highly deferential standard of review, these four challenges provide no basis to question EPA’s general analysis or to upset the entire Designations Rule. A. Designation of Noncontiguous Townships As we have described above, EPA’s designation analysis starts with two presumptions. See supra at 30-31. First, “if a PM2.5 monitor [is] violating” the NAAQS, “at a minimum [EPA will] designate the entire county where that monitor is located as nonattainment” because the “county boundary” is “the basic jurisdictional boundary for determining the extent of the area reflected by the PM25 monitor.” PM25 Designations Rule, 70 Fed. Reg. at 946-47. Second, all counties within the C/MSA contribute to that violation. See id. at 947; Holmstead Memo Guidance at 4. When these presumptions operate in tandem, a violating monitor within a C/MSA will result in a single contiguous block of nonattainment counties that includes both the county with the violating monitor and the other C/MSA counties that are deemed to contribute to that violation. Likewise, when EPA includes out-of-C/MSA counties in a nonattainment block, those counties typically adjoin violating counties. But EPA also invites states to recommend smaller PM2.5 designation areas on a “case-by-case basis” if they “provide an adequate justification demonstrating that a smaller area would include the full area that is violating the standards and all nearby source areas that contribute to the violation.” Holmstead Memo Guidance at 6. Some states took up EPA’s offer and asked that a township rather than a county be designated as nonattainment when a power plant in the township was the overwhelming source of the county’s contribution to a nearby PM2.5 violation. See PM2S Designations Rule, 70 Fed.Reg. at 947; Technical Support Document § 6.5.4.4 (Discussion). To borrow petitioners’ imagery, this approach created an island of nonattainment — the township — surrounded by a sea of attainment throughout the remainder of the county. Industry petitioners argue that creating such islands of nonattainment arbitrarily deviates from the presumption that designations should be made at the county level. Because emissions from a power plant in a township do not skip over all other parts of the surrounding county and come to rest in a nearby area with a violating monitor, petitioners argue that violating areas designated non-attainment should be contiguous to those areas contributing to the violation. We find no fault in EPA’s conclusion that it would be unreasonable to designate as nonattainment those areas that do not, in fact