Full opinion text
Opinion for the Court filed PER CURIAM. PER CURIAM: The American Iron and Steel Institute and the National Wildlife Federation petition separately for review of the Environmental Protection Agency’s Final Water Quality Guidance for the Great Lakes System. We grant the AISI’s petition and vacate the Guidance insofar as it would eliminate mixing zones for bioaccumulative chemicals of concern (BCCs) and impose water quality based effluent limitations (WQBELs) upon internal facility waste streams. We also vacate the criteria for polychlorinated biphenyls (PCBs) in light of the EPA’s conceded error. Petitioners’ other challenges are either unripe for review or without merit. I. Background Section 118 of the Clean Water Act (CWA) requires the EPA to promulgate Water Quality Guidance for the Great Lakes. 33 U.S.C. § 1268(c)(2). The Guidance must contain “numerical limits on pollutants in ambient Great Lakes waters to protect human health, aquatic life, and wildlife,” as well as “minimum water quality standards, antidegradation policies, and implementation procedures.” § 118(c)(2)(A). Each state in the Great Lakes System — Illinois, Indiana, Michigan, Minnesota, New York, Ohio, Pennsylvania, and Wisconsin — has two years from the date of promulgation to incorporate into its water quality standards and National Pollutant Discharge Elimination System (NPDES) permit programs provisions that are “consistent with” the Guidance, failing which the EPA is to impose such standards, policies, and procedures upon the state. § 118(c)(2)(C). The EPA published Proposed Water Quality Guidance for the Great Lakes System on April 16, 1993. 58 Fed.Reg. 20802. In response the agency received more than 26,000 pages of comments, data, and other information from some 6,000 interested parties. After reviewing those submissions the agency promulgated the Final Water Quality Guidance for the Great Lakes System on March 23,1995. 60 Fed.Reg. 15366. II. Jurisdiction Although all parties agree that we have jurisdiction over the petitions for review, we have independently examined the matter in order to satisfy ourselves that we may proceed. Section 509(b)(1) of the CWA, 33 U.S.C. § 1369(b)(1), grants original jurisdiction to the federal courts of appeals in these terms: Review of the Administrator’s action ... (D) in making any determination as to a State permit program submitted under section 1342(b) of this title, (E) in approving or promulgating any effluent limitation or other limitation under section 1311, 1312, 1316, or 1345 of this title, (F) in issuing or denying any permit -under section 1342 of this title ... may be had by any interested person in the Circuit Court of Appeals of the United States for the Federal judicial district in which such person resides or transacts business which is directly affected by such action upon application by such person. The words “effluent limitation” in § 509(b)(1)(E) are defined elsewhere in the Act to mean “any restriction established by a State or the Administrator on quantities, rates, and concentrations of chemical, physical, biological, and other constituents, which are discharged from point sources into navigable waters, the waters of the contiguous zone, or the ocean, including schedules of compliance.” 33 U.S.C. § 1362(11). EPA’s Guidance contains provisions within that statutory definition. These provisions include the Tier II methodology to derive values for additional pollutants, 40 C.F.R. § 132.4(c), the prohibition in Guidance Procedure 3.C against using mixing zones for new and existing BCC discharges, 60 Fed.Reg. at 15,417, 40 C.F.R. pt. 132, App. F, Proc. 3.C, the mercury criteria and the PCB criteria. 40 C.F.R. pt. 132, Tables 3 and 4. AISI challenges each of these effluent limitations. It is true that § 118 of the Act, the section directing EPA to publish the Guidance, is not one of the provisions listed in § 509(b)(1)(E). However, § 118 directs EPA to promulgate the Guidance “pursuant to this section and the Administrator’s authority under this chapter.” § 118(e)(2)(B). In compliance with this directive, the Guidance states that “[t]he Great Lakes States and Tribes shall adopt requirements applicable to waters of the Great Lakes System for the purposes of sections 118, 301, 303, and 402 of the Clean Water Act that are consistent with” the provisions of the Guidance. 40 C.F.R. § 132.4. Section 301, 33 U.S.C. § 1311, entitled “Effluent limitations,” is specifically mentioned in § 509(b)(1)(E). We therefore have jurisdiction over the portion of the Guidance containing effluent limitations. See NRDC v. EPA, 673 F.2d 400, 405 (D.C.Cir.1982); Crown Simpson Pulp Co. v. Costle, 445 U.S. 193, 196-97, 100 S.Ct. 1093, 1094-95, 63 L.Ed.2d 312 (1980). AISI challenges other parts of the Guidance not specified for direct judicial review under § 509(b)(1). Nonetheless we have jurisdiction to pass upon them. As we held in Morrow v. District of Columbia, 417 F.2d 728, 737-38 (D.C.Cir.1969), “questions ancillary to, or growing out of, the main action ... may be taken cognizance of by the court and determined, since such jurisdiction is in aid of its authority over the principal matter” (internal quotation omitted). The magnitude and technical character of the record in this case strongly militates against splitting the case into pieces. See E.I. du Pont de Nemours & Co. v. Train, 430 U.S. 112, 128 & n. 18, 97 S.Ct. 965, 975 & n. 18, 51 L.Ed.2d 204 (1977). The interest in “assuring a forum capable of treating the case coherently might justify the comparatively modest displacement of the district court.” International Brotherhood of Teamsters v. Pena, 17 F.3d 1478, 1482 (D.C.Cir.1994). And “[national uniformity, an important goal in dealing with broad regulations, is best served by initial review in a court of appeals.” NRDC, 673 F.2d at 405 & n. 15. Since most of the case is within our direct jurisdiction, we can properly assert ancillary jurisdiction over the remainder. III. Statutory Authority Petitioners argue that the EPA exceeded its statutory authority under § 118 of the Clean Water Act. They contend that § 118 authorized the agency to do no more than issue suggestions that the states could then decide whether to follow. They characterize the agency Guidance as “an inflexible regulatory mandate” that is far beyond § 118’s limited grant of authority. Brief for Petitioners at 10. They present this argument in the form of four specific challenges to the Guidance. We reject each of the four. A. Issuing a Binding Regulation The Great Lakes Critical Programs Act of 1990, Pub.L. No. 101-596, 104 Stat. 3000, modified § 118 of the Clean Water Act, adding § 118(e)(2) in its present form. Section 118(e)(2)(A) directs the Administrator of the agency to “publish ... for public notice and comment proposed water quality guidance for the Great Lakes System.” Section 118(c)(2)(B) directs the Administrator to publish “final water quality guidance” in the Federal Register. Section 118(c)(2)(C) directs the Great Lakes States to adopt standards, policies, and procedures “consistent with” the guidance published by the Administrator. The agency responded to these statutory commands by initiating rule-making proceedings. It issued a comprehensive proposed Guidance in April 1993. During the public comment period it received comments from over 6,000 persons and entities. It then issued a final Guidance in March 1995. The final Guidance announced that state and tribal programs would be considered “consistent with” the Guidance if they were “as protective as” the provisions in the Guidance. 40 C.F.R. § 182.5(g). If a state or tribe fails to submit a plan that meets this requirement, the standards listed in the Guidance will then become applicable within that state or federal Indian reservation. Id. at § 132.5(f)(2). Petitioners’ first contention is that the agency did not have statutory authority to issue the Guidance in the form of a binding regulation. They argue that by using the term “guidance” in the statute Congress intended that the agency issue nothing more than an informal advisory document. The agency’s initial response is that petitioners do not have standing to bring this claim. The agency notes that the Guidance itself is not enforceable against any party. It will become enforceable only if it is adopted by a state or imposed by the agency on a state that fails to submit an acceptable plan. Given this, the agency says it is impossible for the petitioners to have suffered any injury from the agency’s choice of procedure in issuing the Guidance. It relies on NRDC v. EPA, 22 F.3d 1125, 1146-47 (D.C.Cir.1994), in which we denied standing to a petitioner to challenge an agency decision to issue a new program in the form of a final rule rather than as an informal advisory document. There is, however, a crucial difference between the case before us and NRDC. In NRDC the statute in question directed the agency to “ ‘[ jpublish ... guidance’ ” for the design of state vehicle inspection and maintenance programs. Id. at 1146 (quoting 42 U.S.C. § 7511a(a)(2)(B)(ii)). The statute then directed the states to ‘“comply in all respects’ ” with the guidance that the agency published. Id. (quoting 42 U.S.C. § 7511a(c)(3)(B)). The form in which the agency published its guidance, then, had absolutely no effect on the path that a particular state would follow. No matter how the agency issued its guidance, the states were required by the statute to “comply in all material respects.” The situation in this case is different in a subtle but significant way. Section 118 does not on its face provide such a clear limitation on the flexibility of the states. In § 118 Congress directed the agency to publish guidance, and then required the states to adopt standards,, policies, and procedures “consistent with” the agency’s guidance. § 118(c)(2)(C). The agency was given authority for deciding whether a particular plan was acceptable. Id. The agency elected to respond to the statute by conducting a rule-making rather than issuing an informal advisory document. The resulting rules provide minimum standards that states must meet. See, e.g., 40 C.F.R. § 132.5(g). The decision to proceed by rule-making rather than informal guidance therefore results in a greater restriction of the states’ flexibility to craft their own water quality programs. This reduced flexibility limits petitioners’ potential achievement in the proceedings before the various states. This is an injury sufficient to confer standing on them to challenge the agency’s decision to proceed by rule-making rather than informal guidance, unlike the claimed injury in NRDC which sprang from the initial decision of Congress rather than the form of decision-making chosen by the agency. However, on the merits of the question, petitioners do not fare so well. It is clear that § 118 gave the agency the authority to proceed as it did. Petitioners protest that the agency had no authority to promulgate a binding regulation. We note first that it is not clear that the Guidance can properly be characterized as a binding regulation. It certainly restricts state flexibility, but it does not immediately impose any requirements. It merely announces the standards by which state submissions will be judged and informs the states of the default rule that the agency will apply if a state submits a nonconforming plan. Id. at § 132.5. Regardless of how the Guidance is characterized, it is clear that Congress gave the agency the authority to issue the Guidance in this form. In previous cases we have held that Congress’s use of the term “guidance” does not preclude an agency from doing what the agency did here. NRDC, 22 F.3d at 1147 (“Congress’ use of the term ‘guidance’ is not inconsistent with a rule, like this one, that incorporates a binding performance standard but leaves the states considerable flexibility in fashioning programs that meet the standard.”); Public Citizen v. NRC, 901 F.2d 147, 154 (D.C.Cir.) (“‘[Gjuidance’ certainly can mean, in ordinary parlance, to give advice, or suggestions. The term is not inconsistent with the notion of mandatory regulations.”), cert. denied, 498 U.S. 992, 111 S.Ct. 536, 112 L.Ed.2d 546 (1990). The statute in this case commands the agency to “specify numerical limits on pollutants” and provide guidance on “minimum water quality standards.” § 118(c)(2)(A). It is unlikely that Congress would have directed the agency to set these limits and standards without also intending to allow the agency to enforce them against the states. The use of the term “guidance” in this statute is analogous to Congress’s use of the term “guidelines” in the Sentencing Reform Act. 28 U.S.C. § 994(a)(1). In the Sentencing Reform Act Congress gave the Sentencing Commission the authority to issue “guidelines.” The guidelines the Commission issued restricted the sentencing flexibility of the federal courts. The structure of the Critical Programs Act is identical. In § 118 Congress gave the agency the authority to issue “guidance.” The Guidance the agency issued restricts the flexibility of the states. We hold that § 118 gives the agency the authority to issue the Guidance in the form of a regulation. The agency asserts in the alternative that it had authority under § 501(a) of the CWA to issue the Guidance in this form. 38 U.S.C. § 1361(a) (giving Administrator authority “to prescribe such regulations as are necessary to carry out his functions under this chapter”). Given that we hold that the agency had the authority to do so under § 118, we need not construe § 501(a). B. Interpretation of “Consistent With” Section 118(e)(2)(C) states that within two years after the agency’s guidance is published, “the Great Lakes States shall adopt water quality standards, antidegradation policies, and implementation procedures ... which are consistent with such guidance.” The Guidance as published provides that the agency will consider state or tribal programs “consistent with” the Guidance if they are “as protective as” the Guidance itself. 40 C.F.R § 132.5(g). Petitioners contend that this is an impermissible interpretation of the language of the statute. We disagree. Because this is a case of an agency interpreting a statute that it administers, Chevron U.S.A., Inc. v. NRDC, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), dictates our standard of review. Under Chevron, the agency’s interpretation of “consistent with” is entitled to deference if the statute “is silent or ambiguous on the matter at issue, and if [the agency’s] interpretation is reasonable and consistent with the statutory purpose.” NRDC v. Reilly, 976 F.2d 36, 40 (D.C.Cir.1992) (quotations and citations omitted). It is clear in this case that both of these conditions have been met. First, the statute is ambiguous on this precise question. It does not specify what level of conformity is required for a state plan to be considered “consistent with” the agency’s guidance. Read naturally, the phrase is susceptible to a number of sensible interpretations. The agency’s interpretation is certainly within this range. Given that the statute does not answer the precise question, we must defer to the agen-ay’s interpretation so long as it is “reasonable and consistent with the statutory purpose.” Id. It is clear that this interpretation passes this test. Other language in the statute requires that the Guidance “specify numerical limits on pollutants” and “provide guidance on ... minimum, water quality standards.” § 118(c)(2)(A) (emphasis added). The agency could reasonably construe this language to suggest that Congress was attempting to create a uniform set of requirements for water pollution in the Great Lakes. This goal would be defeated if the agency approved plans that were not “as protective as” the “minimum ... standards” given in the Guidance. The agency’s interpretation of “consistent with” is therefore “reasonable and consistent with the statutory purpose.” Reilly, 976 F.2d at 40. We reject petitioners’ contrary arguments and uphold the agency’s interpretation. C. Imposition of Uniform Basin-Wide Criteria The Guidance establishes four sets of water quality criteria. These criteria are made uniformly applicable to the “waters of the Great Lakes System.” 40 C.F.R. § 132.3. Petitioners argue that § 118 does not justify this imposition of uniform basin-wide criteria. Chevron also governs this challenge. As noted above, the agency’s interpretation is entitled to deference under Chevron if the statute is silent or ambiguous on the interpretive question and the agency’s interpretation is reasonable. As for the first step, the statute does not answer the question. Section 118 provides no explicit direction on whether the required standards may be imposed basin-wide. As for the second step, the agency’s interpretation is at least a reasonable, if not indeed the most likely, interpretation. The statute calls on the agency to “specify numerical limits on pollutants in ambient Great Lakes waters” and “provide guidance on ... minimum water quality standards ... for the Great Lakes System.” § 118(c)(2)(A). It is not unreasonable for the agency to interpret the statute as contemplating that it will establish uniform standards for the basin as a whole. Establishing uniformity across the Great Lakes seems to have been one of the primary purposes of the statute. The agency was therefore justified in developing and imposing basin-wide standards. Petitioners’ arguments to the contrary are without merit. D. Site-Specific Wildlife Criteria Petitioners also challenge the procedures for site-specific modifications to the wildlife criteria. The Guidance states that “[ljess stringent site-specific modifications to wildlife water quality criteria may be developed when a site-specific bioaccumulation factor (BAF) is derived which is lower than the system-wide BAF derived under appendix B of this part.” Procedure l.A(2)(b). Petitioners claim that the Guidance unlawfully restricts the authority of the states because it allows modifications only where the State has established a lower BAF for a particular site. A BAF is the ratio of the concentration of a contaminant in fish tissue to the concentration of the chemical in the ambient water. Petitioners contend that the agency should allow the development of site-specific criteria on other legitimate grounds. The agency responds that the petitioners have misinterpreted the Guidance. In its brief the agency appears to interpret the regulation to allow the development of site-specific modifications on grounds other than a lower BAF: EPA’s intent is not to preclude development of site specific wildlife criteria on other scientifically-justified grounds besides site-specific bioaccumulation factors, but rather to ensure that, if site-specific criteria modifications are developed on other grounds, (1) calculation of the site-specific criteria will also reflect a site-specific bioaccumulation factor and (2) there will be no off-site impairment of designated uses. Respondent’s Brief at 51. This is a permissible reading of the regulation, and we will hold the agency to it. So long as the agency adheres to this reading, the petitioners’ challenge to these procedures is not ripe. Should the agency ever adopt the interpretation the petitioners describe, this court will of course have jurisdiction to revisit the issue. IV. Tier II AISI challenges the Guidance’s so-called “Tier II methodology” for aquatic life and human health. But it never bothers to tell the court exactly what this Tier II methodology is. Everything is presented in the most abstract form. AISI contrasts Tier II with “Tier I,” and complains about the difference, without pausing to explain exactly how Tier I functions in this complex regulatory system. We are treated to sentences like this: “Evidence shows that Tier II values derived from limited data are poor predictors of Tier I criteria from a complete database.” AISI Brief at 35. And this mouthful: “EPA’s wholesale use of safety factors in place of sound science, without adequate justification based on evidence of risk, further magnifies the Tier II methodology’s inherent inadequacies and results in highly conservative values that are arbitrary and unjustified.” Id. We assume the author of these sentences understands them. Perhaps so do those thoroughly versed in the intricacies of the Clean Water Act, in its regulatory jargon, in mathematics, in toxicology, in biology, oncology and so on. Too bad AISI did not take the trouble to educate the court. The first rule of advocacy is to make your argument understandable. Not until we reach page 101 of EPA's responsive brief do we begin to get an inkling of what is bothering AISI. We will try to explain. The centerpiece of the CWA is the NPDES permitting program. 33 U.S.C. §§ 1311(a), 1342. Pollutants cannot be discharged into the waters of the United States unless the discharger has an NPDES permit. Each permit contains discharge limitations depending on levels of pollution-control technology, and — this is the significant part for our purposes — any more stringent limitations necessary to protect the quality of the receiving waters. § 1311(b)(c). But what is necessary to protect water quality? The answer lies in “water quality standards” which contain, among other things, “criteria” setting forth the legally permissible amounts of pollutants in a particular water segment. See EPA v. State Water Resources Control Bd., 426 U.S. 200, 96 S.Ct. 2022, 48 L.Ed.2d 578 (1976). Now we are getting closer to this case. For it seems that all of the Great Lakes states have at least some of what are called “narrative criteria” in their water quality standards. “No toxic pollutants in toxic amounts” is only the example either party offers us. Here are a few others: waters shall be free of “substances that will cause the formation of putrescent or otherwise objectionable bottom deposits”; waters shall be free of “materials that cause odor, color or other conditions in such a degree as to cause a nuisance”; and waters shall be free from “substances in concentrations or combinations harmful or toxic to humans or aquatic life.” U.S. Environmental Protection Agency, Criteria and Standards Division, Standards Branch, State Water Quality Standards For Toxics 2 (Oct.1984). There is another type of “criterion” in water quality standards — one containing a numerical limitation on the concentration of a particular pollutant in the water. For example, waters shall not contain more than 200 fecal eoliform per 100 milliliters. Because narrative criteria do not specify numerical limitations on the concentration of a particular pollutant in the water, a problem arises when it comes to formulating discharge limitations in permits. We have already mentioned that permits must incorporate discharge limitations necessary to ensure that the water quality standards are met. This requirement applies to narrative criteria as well as to criteria specifying maximum amounts of particular pollutants. And so the problem is how one derives numerical values from general, non-numerical narrative criteria. EPA has addressed the problem in the past. It promulgated 40 C.F.R. § 122.44(d)(l)(vi) to deal with it on a national basis. That regulation sets forth three broadly-defined methods to translate narrative criteria into numerical values. We upheld these methods as “reasonable, authorized attempts at necessary gap-filling in the CWA statutory scheme.” See American Paper Inst. v. EPA, 996 F.2d 346, 348 (D.C.Cir.1993). For the Great Lakes, EPA took a different approach. Section 118(c)(2) required EPA to include in the Guidance “numerical limits on pollutants in ambient Great Lakes waters to protect human health, aquatic life, and wildlife.” The Great Lakes Initiative Committees, on whose work EPA relied, developed things known as translator methodologies to convert narrative criteria into numerical values. 58 Fed.Reg. at 20,835. EPA’s endorsement of this approach resulted in the Guidance’s “two-tiered system,” which is a fancy way of saying that if you cannot use the first method (Tier I) to do the translating because there is not enough scientific data available, then use the second method (Tier II). Both Her I and Tier II use complicated equations to do the translating. We do not pretend to understand the mathematics. The difference between Tier I and Tier II is in the amount of toxicological data. 40 C.F.R. pt. 132, App. A. To use Tier I, there must be a rather comprehensive set of such data. Tier II, the fallback, requires significantly less. EPA’s Guidance contains a Tier I and a Tier II for the protection of aquatic life, and a Tier I and a Tier II for the protection of human life. Wildlife gets only a Tier I for translating words into numbers. Why have this two-tiered system? EPA explained that it would bring about greater uniformity throughout the Great Lakes States. Already, EPA reported, some Great Lakes States were using a similar approach. And “EPA and the States determined that there is a need to regulate pollutants more consistently in the Great Lakes System when faced with limited numbers of criteria.” 69 Fed.Reg. at 15,373. At long last we come to AISI’s complaint. It has no problem with what EPA did to protect wildlife. But with respect to the Tier II methodologies for aquatic life and human health, AISI says EPA exceeded its regulatory authority and had no scientific support for what it promulgated. As to EPA’s authority, AISI’s point is that the Tier II methodologies do not fit within the statutory framework of the national Clean Water Act program. But this seems to us to ignore § 118(c)(2). That provision authorizes EPA to promulgate the Guidance and to “specify numerical limits on pollutants in ambient Great Lakes waters to protect human health, aquatic life, and wildlife _” § 118(c)(2)(A). That is what the Tier II exercise does. If Congress had considered the existing Clean Water Act provisions sufficient to protect the Great Lakes, it is difficult to understand why it enacted § 118(c)(2). The provision was added by the Great Lakes Critical Programs Act of 1990 (“Critical Programs Act”), Pub.L. No. 101-596,104 Stat. 3000-3004. The Senate Report on the legislation states that § 118(c)(2) was meant “to assure the adoption of more uniform, enforceable water quality standards for the Great Lakes.” S. Rep. No. 101-339, at 12 (1990). The House Report said the same. See H.R. Rep. No. 101-605, pt. 2, at 35 (1990). EPA found that States were using “different ‘translator’ methodologies in developing derived numeric criteria for implementing narrative water quality criteria,” 58 Fed.Reg. at 20,835. The Tier II methodology moves the Great Lakes States toward greater uniformity, apparently with their consent, and this is one of the things § 118(e)(2) sought to accomplish. As such, this portion of the Guidance was within EPA’s authority. On the subject of science, AISI tells us that the Tier II methodology is scientifically flawed, and constitutes an unexplained fundamental departure from prior EPA policy. We disagree with AISI on both counts. EPA adequately explained its adoption of the methodology. It stated, in response to comments, “[t]he minimum data required for the Tier II methodology promotes consistency in how [§ 303(c)(2)(B) ] is implemented. The Tier II method provides a standardized process for utilizing available data to derive values for purposes of interpreting narrative standards, thereby achieving greater consistency among the States in this activity.” EPA Response to Comments Doc. No. D2791.103. EPA repeated this in the preamble to the final Guidance. 60 Fed.Reg. at 15,373. EPA thus spelled out its objective, and even though we may not fully grasp exactly how the mechanisms work, AISI gives us no reason to doubt that EPA’s means accomplish its ends. See Motor Vehicle Mfrs. Ass’n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 57, 103 S.Ct. 2856, 2873-74, 77 L.Ed.2d 443 (1983); Pittsburgh Press Co. v. NLRB, 977 F.2d 652, 655 (D.C.Cir.1992). AISI’s “inadequate science” argument is that “[b]y definition, Tier II values are created only where Tier I criteria cannot be calculated because sufficient data are unavailable.” AISI Brief at 32. Inadequate data, however, do not mean inadequate science. AISI ignores statutory requirements. Section 118(c)(2) requires EPA to specify numerical limits on pollutants in the Great Lakes; it does not single out only those pollutants whose toxicity has generated a full set of data so that Tier I is satisfied. NPDES permits for point sources must contain discharge limitations necessary to pro- . tect water quality. Section 301 “imposes this strict requirement as to all standards — i.e., permits must incorporate limitations necessary to meet standards that rely on narrative criteria to protect a designated use as well as standards that contain specific numeric criteria for particular chemicals.” American Paper Inst., 996 F.2d at 350. To the extent AISI supposes that no numerical limits are acceptable in the absence of a full set of data, it is very much mistaken. And it is mistaken too in thinking that EPA’s methodology is “by definition” scientifically inadequate. It is true that a Tier II aquatic health value possibly might be calculated from as little as a single acute test for a single species. See 40 C.F.R. pt. 132, App. A.XII. It is true as well that EPA’s Science Advisory Board expressed concern about a Tier II value being derived from a single acute test fer a single species. AISI Brief at 34; An. SAB Report: Evaluation of the Guidance for the Great Lakes Water Quality Initiative 14 (Dec.1992). But the Board also stated that “[f]rom a statistical perspective, the historical data base is probably scientifically defensible to account for many of the sources of toxicological testing uncertainty.” SAB Report at 14. EPA answered that “[a]lthough a Tier II value may be developed using a single data aquatic toxicity test, EPA believes that few, if any, Tier II values, based on a single data point, would be derived for use in control mechanisms,” since the Tier II approach “requires States and Tribes to use as many acceptable data as exist.” 58 Fed.Reg. at 20,855. Given that the CWA requires that permit limits be set for any pollutant that may contribute to a violation of a narrative criteria, American Paper Inst, 996 F.2d at 350, the best scientific approach that determines a value is permissible. The proper question is not whether using Tier II produces results as good as Tier I. What is the alternative when one cannot use Tier I? AISI lists other options, but it has not even attempted to convince us that these are superior to EPA’s methodology. AISI Brief at 38. AISI has another complaint about Tier II, listed in a few lines of its brief — that EPA’s methodology results in large safety margins and hence conservative numerical values. Id. But it is within EPA’s discretion to decide that in the wake of uncertainty, it would be better to give the values a conservative bent rather than err on the other side. The “uncertainty” factors do not, in any event, appear arbitrary to us. The human health uncertainty factors are created as a function of the available data. The aquatic uncertainty factors are scaled — “[a]s more data become available, the derived Tier II values tend to become less conservative.” 60 Fed.Reg. at 15,373; see also 40 C.F.R. pt. 132, App. A.XII, Table A-l. The fact that permit limits derived by the Tier II methodology may be more stringent than those yielded by Tier I does not mean that EPA’s promulgation of the methodology is arbitrary or capricious. Our decision in Leather Industries of America, Inc. v. EPA, 40 F.3d 392 (D.C.Cir.1994), does not assist AISI. We did not there construe § 118. Instead we held only that “a margin of safety must be rooted in an analysis of risk.” 40 F.3d at 400. The Tier II methodology, by sealing the uncertainty factors to reflect existing data, properly correlates risk with knowledge. Y. Implementation Procedures The AISI also challenges several of the procedures that prescribe in detail the factors that a permitting authority must consider when assessing the merits of a permit application and the terms and conditions an authority must include in any permit that it may issue. We address each challenge in turn. A. The Limits of Quantification The EPA considers the discharge of certain toxic pollutants to be unacceptable even at levels where the presence of the pollutant cannot be measured accurately. Procedure 8.A provides accordingly that “the permitting authority shall designate as the limit in the NPDES permit the [WQBEL] exactly as calculated.” The AISI argues that Procedures 8.A and 8.B, the latter of which details the analytical methods to be used when setting a permit limit, violate due process by mandating a standard that is too vague to warn industry of the scope of the conduct prohibited. A standard with which compliance cannot be assessed — and it is agreed that compliance with an effluent limitation set below the level of quantification simply cannot be assessed— is no standard at all for purposes of due process. For much the same reason, the AISI contends that the non-standards established by Procedures 8.A and 8.B are arbitrary and capricious and therefore unlawful under the Administrative Procedure Act. 5 U.S.C. § 706(2)(A). The EPA responds, first, that it has no choice in the matter:' the CWA requires the agency to set each effluent limitation at a level that will “protect the public health and welfare, enhance the quality of water and serve the purposes of the Act,” 38 U.S.C. § 1313(c)(2)(A), and a pollutant may be so toxic that the discharge of even an immeasurably small amount would have a reasonable potential to endanger the public health and welfare. Nothing in the CWA suggests that the discharge of such a chemical should be treated differently than the discharge of a chemical that is dangerous only at measurable levels. The EPA further contends that these Procedures raise no due process concern because, as matters now stand, there is no danger that an enforcement action will be taken based upon a measurement below the level of quantification. Procedure 8.B(3) provides that measurements are considered accurate only “down to the quantification level.” The EPA has announced that it “is the agency’s policy that any effluent sample analyzed in accordance with the analytical method specified in the permit and other applicable procedures that is found to be below the quantification level shall be deemed in compliance with the WQBEL.” The AISI acknowledges that the Guidance provides that measurements are considered accurate only down to the level of quantification. Procedure 8.B(4) violates due process nonetheless, the AISI contends, because neither the States nor third parties are bound to follow the EPA in considering measurements below the level of quantification to be in compliance with the applicable effluent limitation. In fact, Procedure 8.B(4) explicitly provides that each State “permitting authority may specify in the permit the value to be used to interpret sample values below the quantification level.” We conclude that the AISI’s challenge is both misdirected and unripe. It is misdirected because the petitioners have no grievance with the EPA. Under the regime created by the Guidance, a WQBEL set below the level of quantification, although technically enforceable, is without any genuine force unless and until a State should choose to attach a significant value to a measurement below the quantification level. The EPA cannot be said to have violated the due process rights of the petitioners simply by leaving the States free to work the alleged violation. Nor can the EPA itself be said to have acted arbitrarily and capriciously simply because the Guidance leaves the States with sufficient discretion to act in a way that would purportedly be arbitrary and capricious. The challenge is unripe because no standard with which compliance would be difficult — much less impossible — to measure has yet been given any legal significance. If any state has even suggested that it intends to attach significance to measurements below the level of quantification, the AISI has not so informed the Court. Therefore, the arguments that the treatment afforded measurements below the quantification level is arbitrary and capricious and violates due process are not yet ripe for review. B. Pollutant Minimization Program Procedure 8.D provides that each permit will set out a pollutant minimization program (PMP) for each pollutant for which the permitting authority has imposed a WQBEL that is below the level of quantification. Each PMP must be designed to reduce the discharge of the regulated pollutant at the point source to a level that is at or below the WQBEL by monitoring and regulating the discharge of that pollutant from each source within the facility. The PMP will also require the implementation of “appropriate cost-effective control measures ... consistent with the control strategy.” Procedure 8.D(4). The AISI challenges Procedure 8.D on the ground that the CWA permits regulation only of point sources that discharge into the navigable waters of the United States and not of sources that discharge into a waste stream within a facility. Section 301(a) of the CWA prohibits “the discharge of any pollutant” by any person. The “[discharge of a pollutant” is defined as “any addition of any pollutant to navigable waters from any point source.” § 502(12) (emphasis added). In order to regulate such discharges, the Congress authorized the EPA in §§ 301 and 304 of the CWA to establish “effluent limitations,” defined as restrictions placed upon pollutants “which are discharged from points sources into navigable waters.” 33 U.S.C. §§ 1311 and 1314. The AISI asserts that this statutory scheme authorizes the EPA to regulate pollutant discharge levels only at the point where a pipe leading from the facility discharges effluent into the navigable waters of the United States. By imposing restrictions upon pollutants that are discharged from internal plant sources into internal plant streams, the agency has, according to the petitioner, exceeded its clearly delimited statutory authority. The EPA responds that the Act gives the agency broad powers to impose NPDES permit conditions in order “to assure compliance with” point source effluent limitations. 33 U.S.C. § 1342(a)(2). The EPA has equally broad authority under § 308 of the Act to impose monitoring and reporting requirements. These broad grants of authority, per the EPA, give the agency the power to require monitoring and control of internal sources, so long as those requirements further the goal of ensuring compliance at that point where effluent is ultimately discharged into the navigable waters of the United States. We agree with the EPA that the CWA contemplates the imposition of monitoring and reporting requirements for internal plant sources. Section 401(a)(2) authorizes the Administrator to prescribe conditions for permits “including conditions on data and information collection to insure that the water quality standards are met.” A permitting authority might be able to calculate an otherwise unquantifiable point source discharge level only by adding together the measurable levels of a pollutant contained in internal facility waste streams. For example, with respect to a pollutant for which the WQBEL is 10 ppm but the lowest measurable level is 40 ppm, determining that the pollutant is present in an internal waste stream at a concentration of 80 ppm and that this stream contributes a quarter of the total effluent ultimately discharged at the point source would inform the permitting authority that the pollutant is being discharged at the point source in a concentration of 20 ppm, i.e., at an impermissible level. It is reasonable to conclude, therefore, that § 401(a)(2) permits the monitoring of discharges from internal sources in order to calculate otherwise unquantifiable point source effluent levels. The question whether the CWA contemplates an effluent limitation being imposed upon an internal source is not as easily resolved. Although we have held that the EPA may set a point source WQBEL at a level such that a plant would as a practical matter be forced to change internal equipment and processes in order to comply, see American Paper Institute v. Train, 543 F.2d 328, 341 (D.C.Cir.1976), we have not yet spoken to the question whether the EPA is authorized to require compliance with the WQBEL not only at the point source but also in streams and pools that are inside the facility. We recur therefore to the text of the CWA. The Act provides that when a permitting authority determines that “discharges of a pollutant from a point source ... would interfere with the attainment or maintenance of [applicable] water quality standards, .effluent limitations (including alternative effluent control strategies) for such point source ... shall be established which can reasonably be expected to contribute to the attainment or maintenance of such water quality.” 33 U.S.C. § 1312(a). The statute is clear: The EPA may regulate the pollutant levels in a waste stream that is discharged directly into the navigable waters of the United States through a “point source”; it is not authorized to regulate the pollutant levels in a facility’s internal waste stream. We are apprised of nothing in the policy underlying the CWA that undercuts the plain meaning of the statutory text. To the contrary, by authorizing the EPA to impose effluent limitations only at the point source, the Congress clearly intended to allow the permittee to choose its own control strategy. By imposing water quality based standards upon internal facility waste streams, the EPA seeks to deprive the individual permit-tee of the ability to choose between a control system that meets the point-source WQBEL by means of point source controls and a control system that meets the point source WQBEL by means of internal waste stream purification. As we have just seen, however, the statute does not permit this sort of meddling inside a facility. We conclude, therefore, that although the EPA has the statutory authority to require the monitoring of discharges from sources within a facility, the agency exceeded its authority when it sought to impose effluent limitations upon nonpoint-source discharges. We therefore vacate Procedure 8.D insofar as it would impose the point-source WQBEL upon a facility’s internal waste streams. C. Mixing Zones The AISI challenges Procedures 3.C, 3.D, and 3.E, which eliminate mixing zones for BCCs and establish required characteristics for all other mixing zones. A mixing zone is the area within a navigable water of the United States in which the discharge from a point source is initially diluted. Where a permitting authority has authorized a mixing zone, the permittee’s compliance with the applicable water quality standards is assessed at the edge of the zone. A mixing zone is often extended by the permitting authority to include the area in which there occurs secondary mixing, i.e., mixing that is caused not by the turbulence created by the discharge itself but by naturally occurring currents and eddies. Procedure 3.C phases out mixing zones for BCCs. Under the Procedure, no new mixing zones were to be granted for dischargers of BCCs after March 23, 1997. All existing mixing zones for discharges of BCCs were, moreover, to have been phased out by March 23, 2007. In the interim, existing mixing zones were also to have been brought into compliance with the more stringent requirements made applicable to mixing zones for substances other than BCCs by Procedures 3,D and 3.E. Those procedures limit the size of a mixing zone. Procedure 3.D provides that: “In no case shall a mixing zone be granted that exceeds the area where discharge-induced mixing occurs,” and thus removes any area of secondary mixing from a mixing zone. Procedure 3.E requires that a mixing zone embrace an amount of water no more than 25 percent of the design flow of the receiving body of water. The AISI challenges all three Procedures. 1. Elimination of Mixing Zones For BCCs The AISI asserts that the elimination of mixing zones for BCCs is arbitrary and capricious. Eliminating these mixing zones will not, according to the AISI, significantly reduce pollutant loadings to the Great Lakes. And the EPA, we are told, has conceded as much. ' See EPA, Assessment of Compliance Costs Resulting from Implementation of the Final Great Lakes Water Quality Guidance, Document N6115L (“In terms of pollutant load reductions, the addition of mixing zones results in an insignificant decrease in pollutant load reductions”). Moreover, according to the petitioner, the elimination of BCC mixing zones would inflict upon industry costs that are excessive in relation to the degree of pollution reduction achieved, even if that reduction were not insignificant. For example, if BCC mixing zones are phased out, then the town of Owos-so, Michigan will be forced to spend an additional $300,000 in order to remain in compliance with applicable water quality standards. The sole benefit derived from this expenditure would be the removal of 0.4 pounds of mercury per year from the town’s municipal waste discharge. In view of the Owosso situation, the AISI asserts that the agency simply failed to consider the issue of cost — a serious problem raised in the comments — in formulating the final rule. The EPA responds, first, that significant problems caused in the Great Lakes by the release of BCCs even in small amounts justify eliminating mixing zones for them. Even in a relatively open system, a BCC will persist in the ambient water and underlying sediment longer than will a run-of-the-mill pollutant. The Great Lakes, however, are a more-or-less closed system from which pollutants escape only over a long period of time. See Supplementary Information Document at 1 (“Lake Superior also has the longest retention time — the average time for one molecule of water to exit the system — of 173 years, while Lake Erie has the shortest at 2.7 years.”) The retentive character of the Great Lakes combined with the persistent character of BCCs is what warrants, in the EPA’s opinion, the elimination of BCC mixing zones. The EPA also observes that the characterization of BCC loading reductions as “insignificant” was made by the agency’s economists; the technical and policy staffs, who were aware of the environmental peculiarities of the Great Lakes, characterized those same reductions as significant. The EPA also maintains that it responded to concerns about the unreasonable cost of eliminating some BCC mixing zones by providing that such mixing zones be allowed in special circumstances. A permitting authority may permit a BCC mixing zone provided that it first finds that: The discharger is in compliance with and will continue to implement all applicable technology-based treatment and pretreatment requirements of CWA sections 301, 302, 304, 306, 307, 401, and 402, and is in compliance with its existing NPDES water quality-based effluent limitations, including those based on a mixing zone; and the discharger has reduced and will continue to reduce the loading of the BCC for which a mixing zone is requested to the maximum extent possible. Procedure 3.C(6)(a)(i)-(ii). Although the EPA appears adequately to have explained the environmental justification for its decision, we are unable to accept, based upon the record before us, the EPA’s claim to have considered the ratio of cost to benefit. The agency estimated the total cost of eliminating BCC mixing zones at $200,000. The agency failed to explain— indeed even to acknowledge — the comment estimating at approximately $300,000 the cost to one town of removing mercury from its sewage discharge. If Owosso were the only contributor of pollutants to the Great Lakes, this disparity would be troubling. In light of the number of municipal and industrial point sources in the Great Lakes Basin, however, the EPA appears to have adopted a cost estimate that can only be described as fanciful. We therefore grant the petition insofar as it challenges Procedure 3.C. Although the EPA appears to have shown that eliminating mixing zones is not without some environmental benefit, the agency simply failed to address whether the measure is cost-justified. We remand the matter in order to afford the EPA an opportunity to do just that. 2. Limitations on Other Mixing Zones. The AISI also challenges the requirement of Procedure 3.E that a mixing zone encompass no more than 25 percent of the design flow of the receiving water. The AISI observes that the EPA purportedly relied upon a 1968 recommendation of the National Technical Advisory Committee that “mixing zones in streams should be limited to twenty five percent of the cross sectional area of the river to allow a free zone of passage for aquatic organisms.” The agency’s reliance is misplaced, the AISI contends, because 25 percent of the “cross-sectional area” of a stream — the measure recommended by the Advisory Committee — is not the same as 25 percent of the “design flow” of that stream— the measure adopted by the EPA. We conclude that the EPA adequately justified its decision to establish uniform default dimensions for mixing zones. The EPA concluded in the Supplementary Information Document that “the value of 25 percent of total river flow is a rational estimate of the amount of the river flow in 25 percent of the cross-sectional area,” and it is more easily measured. Any concern that the substitute criterion will produce poorly proportioned mixing zones should be allayed by the fact that a permittee may demonstrate that “an alternative dilution fraction is appropriate.” Id. Therefore, in a river where 25 percent of river flow does not correspond even roughly to 25 percent of cross sectional area, the permitting authority may adjust the mixing zone accordingly. The AISI also challenges Procedure 3.D insofar as it limits mixing zones to the discharge-induced mixing area, “which is the area in the immediate vicinity of the discharge where mixing occurs due to the turbulence created by the momentum of the discharge.” The EPA defines this area, for practical purposes, as one embracing a quantity of water equal to ten times the quantity of effluent. The AISI contends that this approach represents an unjustified departure from past EPA practice. The EPA responds that it adopted the 10:1 rule only after reviewing studies conducted for the Milwaukee South Shore and for the Green Bay Metropolitan wastewater treatment plants. These studies concluded that it was reasonable to limit mixing zones to an area that allows 10 parts receiving water for each part of effluent because such a zone generally corresponds to the area “where the velocity and momentum associated with an effluent being discharged from the end of a pipe was dissipated.” SID at 274. The EPA recognized that other processes would, over time, further dilute the pollutant and that site specific conditions might render the 10:1 ratio inappropriate. The EPA concluded, however, that “these [two] studies provide a scientific basis for default mixing zone assumptions for discharges to open waters of the Great Lakes.” Id. Again we find that the EPA has reasonably decided to rely upon conservative but adequately supportive science. The two referenced studies reveal that most mixing occurs in the area covered by the 10:1 rule. The EPA recognized that additional mixing may occur outside of this area. As a default rule, however, the EPA settled upon that area in which, as the available data revealed, discharge-induced mixing normally occurs. Once again the EPA provided for a party to show that these assumptions are inappropriate in a particular situation: “The final Guidance does allow for recognition of site specific conditions by allowing alternative mixing zones subject to the mixing zone demonstration requirements” of Procedure 3.F. We find nothing arbitrary or capricious in this decision. We therefore deny the petition insofar as it challenges Procedures 3.D and 3.E. D. Intake Pollutants The EPA considers any discharge of a pollutant into the navigable waters of the United States, including the return of a pollutant to the body of water from which it was drawn, to be an “addition” of that pollutant to that body of water within the meaning of the Clean Water Act. Accordingly, the EPA did not require that a permitting authority make any allowance for these so-called intake pollutants. A facility must as a result obtain a permit even if its only discharge is the return of an intake pollutant into the river or stream from which it was taken. The AISI challenges the provisions applicable to intake pollutants on three grounds. First, the AISI asserts that the CWA does not authorize the agency to hold a point source responsible for the intake pollutants it discharges. Second, the AISI contends that to hold a point source responsible for intake pollutants violates the due process rights of the discharger by holding it responsible for pollutants it did not originally add to the water. Finally, the AISI alleges that various aspects of Procedures 5.D and 5.E are arbitrary and capricious. Before responding to these allegations, the EPA asserts that they are all unripe for review. Whether the discharge of an intake pollutant constitutes an “addition” for purposes of the CWA is, the EPA contends, a question that cannot be answered in the abstract. The court must consider a variety of factors, including: how a facility uses intake water; whether and how the intake water is removed from waters of the United States; the quality of the receiving water; the quality of the water discharged; and whether the intake water originates and is discharged into the same body of water. The EPA urges that we therefore postpone ruling on this issue until we have a well-developed and concrete factual situation before us. The AISI maintains that its challenge is ripe for review because it implicates the EPA’s purely legal interpretation of the CWA. The EPA has interpreted an “addition” to include all intake pollutants passing through a discharge pipe. The AISI contends that this interpretation can be adjudged erroneous quite apart from any factual considerations. We agree with the EPA that these challenges are not yet ripe for review. This Circuit has twice held that similar intake pollutant provisions remained unripe for review until they were implemented in a particular factual setting. See NRDC v. EPA, 859 F.2d 156, 204-205 (D.C.Cir.1988); Diamond Shamrock Corp. v. Costle, 580 F.2d 670, 674 (D.C.Cir.1978). We are concerned in the present case, as we were in NRDC v. EPA, that “the broad legal issue [of what constitutes an “addition” under the CWA] may have links to the concededly fact-dependent ones that we cannot now fully perceive.” 859 F.2d at 205. For example, Procedures 5.D and 5.E provide for intake credits and offsets in certain situations; this Court would need to know how those provisions operate in practice in order to address the petitioner’s claim that the EPA has acted in a way that is arbitrary and capricious or violates the petitioner’s due process rights. We concede, as we did in NRDC v. EPA, that the question of statutory interpretation “appears purely legal in character” and that we “might well carve this issue out from the more fact-bound claims and find it ripe.” Id. As in NRDC, however, we continue to think that the legal challenge is bound up in the more fact-dependent claims “enough to tilt the balance in favor of finding not only the fact-dependent claims, hut also the more purely legal one, unripe.” Id. E. Reasonable Potential Procedures An NPDES permit must contain a WQBEL for any discharge that either will cause or has the reasonable potential to cause or to contribute to an excursion above a water quality standard. See 40 C.F.R. § 122.44(d)(1). Under EPA regulations, a permitting authority “must use all relevant available data, including facility-specific effluent monitoring data where available” and employ “procedures which account for existing controls on point and non-point sources of pollution, the variability of the pollutant or pollutant parameter in the effluent, the sensitivity of the species to toxicity testing ... and, where appropriate, the dilution of the effluent in the receiving water” when it determines whether a pollutant discharge has the reasonable potential to cause an excursion above a water quality standard. Id. at § 122.44(d)(l)(ii). The Guidance establishes a new two-step procedure for determining whether there is a reasonable potential for a given discharge to cause an exceedance of a water quality standard. The permitting authority must, first, set a preliminary effluent limitation (PEL) for each point source. Each PEL is calculated so that a discharge below the PEL level would not have a reasonable potential to cause an exceedance. Second, the permitting authority must determine the projected effluent quality (PEQ) of the discharge. The PEQ is set at “the 95 percent confidence level of the 95th percentile based on a log-normal distribution of the effluent concentration; or the maximum observed effluent concentration, whichever is greater.” Procedure 5.B(1). Whenever the PEQ of a particular point source exceeds the PEL for that point source, the permitting authority must set a WQBEL for each pollutant that is the cause of such an exceedance. The AISI first complains that this new procedure, by providing that a PEQ must be no lower than the “maximum observed effluent concentration,” will allow a permitting authority to base a “reasonable potential” finding upon only one effluent quality measurement. By contrast, under the national program, a permitting authority must always consider the variability of the pollutant in the effluent and, therefore, must always take into account effluent quality measurements that will reveal the variability of the concentration of the pollutant. The EPA responds that, although the Guidance does allow for regulation to be based upon only one measurement of a discharge, the procedures are nevertheless consistent with the national program because the permitting authority must, under the Guidance, conclude that the measurement before it is “valid and representative” before it may base any PEQ determination upon that one measurement. See SID at 317 (“Permit-tees should ensure they are reporting valid, representative data. See 40 C.F.R. § 122.41(j)(l). Where the permittee believes a certain effluent measurement to not be representative of the effluent, the permittee should bring this to the permitting authority’s attention.”) The EPA also observes that the permittee is in the best position to remedy any perceived inadequacy in the data set presented to the permitting authority. The EPA adequately responded to the AISI’s concern by effectively mandating that a PEQ must not be based upon a single data set unless the permitting authority first concludes that the measurements in that set are “valid and representative” for the point source. In order to be “valid and representative,” the single data set must account for the variability of the pollutant in the effluent. The EPA concedes that a single data set will not often meet this standard. When one set does suffice, however, it will be one that accounts for the variability with which the AISI is concerned. The AISI next complains about Procedure 5.F, which provides that if a pollutant is present in unacceptable leve