Marathon Oil challenges the validity of an Environmental Protection Agency permit authorizing the discharge of pollutants into waters of the State of Alaska. Marathon argues that the EPA has exceeded its statutory authority by requiring restrictions more stringent than necessary under Alaskan law and that the EPA disregarded relevant data submitted by Marathon. We disagree and deny the petition for review.

Marathon operates two oil processing facilities on the shore of Cook Inlet, a large tidal estuary in south-central Alaska. Cook Inlet is a high-energy water environment with extreme tides and fast currents--tides ranging up to 30 feet and tidal currents up to six or seven knots. Marathon's facilities--referred to by location as the Trading Bay facility and the Granite Point facility--receive a natural mixture of crude oil and water by pipeline from production wells offshore, process the mixture to separate the crude oil from the so-called "produced water,"1 treat the produced water to reduce remaining traces of hydrocarbons, and discharge the treated water into the Inlet. The Trading Bay facility discharges about 2.8 million gallons per day (MGD) of such effluent through a discharge pipe that ends in the intertidal zone between mean high water and low water marks. The Trading Bay discharge pipe is submerged during half of the daily tidal cycle; during the remainder of the day, the effluent flows in a stream across the tidal flats into Cook Inlet. The Granite Point facility discharges about 0.2 MGD of effluent above the high water mark. The Granite Point effluent is discharged above the high water mark and at all times flows across the mudflats to meet the receiving waters of the Inlet.

Both facilities operated for some years under a national pollutant discharge elimination system ("NPDES") permit issued to Marathon's predecessors. In July 1985, the EPA published the draft of a superseding general NPDES permit covering various oil and gas facilities in Cook Inlet, including Marathon's.2 The draft proposed to ban the flow of effluents across the mudflats, requiring instead that Marathon construct discharge pipes with submerged outlets. Marathon protested, both to the EPA and to the Alaska Department of Environmental Conservation ("ADEC"), pointing out that it would cost about $3 million to build the extensions to the discharge pipes, and arguing that there had been no showing that the existing arrangement harmed the environment. In addition, Marathon submitted a study by a consultant purporting to show that Marathon's facilities currently met state water quality standards within reasonable mixing zones.3 The EPA issued the final NPDES permit in October 1986. That permit requires that the discharge pipes at Trading Bay and Granite Point be extended past the intertidal mudflats, so that the discharge is submerged at all times.

Marathon petitions for review of this aspect of the general permit.

We have some difficulty sorting out Marathon's contentions to this Court. Marathon's argument shifts significantly from opening brief to reply brief to oral argument. Mindful of our minimalist review of the EPA's science, Marathon opens its briefing with an observation calculated to perk up judicial ears: "This appeal does not involve technology-based limitations, such as were at issue in American Petroleum Institute v. EPA, 787 F.2d 965 (5th Cir.1986). Rather, the Court need only concern itself with a single provision of the [Clean Water] Act, which addresses state water quality standards." The initial argument is this: The Clean Water Act authorizes the EPA to require compliance to the extent necessary to meet Alaska state standards. See 33 U.S.C. Sec. 1311(b)(1)(C).4 Marathon asserts that the EPA interpreted Alaska law to prohibit absolutely the discharge of effluent across the tidal mudflats, and that this reading was erroneous. Instead, according to Marathon, the EPA should have applied the Alaska Administrative Code standard for contamination of the sediment on the tidal flats around the effluent streams, one that requires only that the contamination not "cause deleterious effect to aquatic life." 18 A.A.C. 70.020.5 On this analysis, Marathon contends that there has been no showing or finding of "deleterious effect" during the two decades of effluent discharge, and that the permit's ban on open discharge is therefore not "necessary" to meet state requirements. It follows, we are told, that the EPA "exceeded its statutory authority" and the permit should be vacated and the matter remanded to the EPA for further review. Marathon's opening brief does not contain any other argument.

The EPA has three responses. First, it correctly points out that Marathon never advanced this particular argument below, and that arguments not made to an agency usually cannot be raised for the first time to a reviewing court. See Brotherhood of Railway Clerks v. St. Louis Southwestern Ry. Co., 676 F.2d 132, 136-37 (5th Cir.1982). Second, it defends its interpretation of the Alaska regulations in dispute. The EPA contends that it is perfectly logical to read the Alaska standards for limiting pollution in the "water column"6 to include streams of waste flowing across the open landscape of the State. Third, the EPA argues that even if it erred in taking the position that discharges across the mudflats were per se violations of Alaska law, the permits must still be upheld because Marathon does not challenge an equally if not more important reason for the EPA's decision: The EPA's computer modeling showed that the discharges must be submerged at all times during the tidal cycle in order to meet state water column standards once the effluent meets the water of Cook Inlet. The EPA contends that its data showed by clear inference that the open effluent streams from Trading Bay and Granite Point into the Inlet could not meet these requirements.

In reply, Marathon shifts ground. It vehemently disagrees with the EPA's assertion that the effluents violate Alaska standards once they reach the water; it argues that "there is absolutely no record evidence" to support the EPA's assertion.7 Marathon asserts that the EPA has never found that its current discharges violate the water column standard once the effluent streams meet the receiving waters, that the EPA decision rests entirely on its allegedly erroneous view that Alaska law forbids the discharge of effluent streams across tidal flats. Marathon charges that the EPA's argument based on the failure of the discharges to meet water column standards in Cook Inlet is a "classic post hoc rationalization of agency counsel" unrelated to the actual ground of decision.

Marathon's oral presentation to us combined the distinct arguments from its opening and reply briefs by characterizing the EPA's decision as a two-step process: First, the EPA (erroneously) determined that the open effluent streams themselves violated Alaska law, and therefore concluded that the outfalls must be submerged at all times. Second, the EPA used computer models to determine not whether but how much the outfalls were to be submerged. Marathon contends that because it had already decided to submerge the outfalls, the EPA's computer models prove nothing about the current compliance of Marathon's outfalls. The EPA has totally failed to model the discharges as they currently exist, we are told; therefore, its decision, being without proper foundation, must be vacated and the case remanded for further consideration.

The record does not square with Marathon's characterizations of it. Our review convinces us that the EPA considered and rejected Marathon's claim that its effluent streams meet state water column standards once they reach the waters of Cook Inlet. Therefore, we need not consider the EPA's other arguments.8A.

In the July 1985 draft permit, the EPA proposed to ban all discharges shoreward of the five meter isobath measured from the mean lower low water mark ("MLLW").9 This proposal meant that Marathon and Shell Western E & P Inc., the owner of an equivalent facility in Cook Inlet, would have to construct discharges pipes whose ends were submerged at least five meters during an average low tide. According to data from the EPA's computer modeling programs PLUME and MPN, the discharges at the five meter isobath would probably still violate state water quality standards if a 100 meter mixing zone was assumed to be proper; but the EPA was satisfied that "removal of the three facilities' discharges to this depth (and possible use of outfall diffusers) will be a significant improvement over the current situation."10 The EPA noted that this aspect of the draft permit was far from final; the ADEC was considering proposals by Marathon and Shell for mixing zones as large as several kilometers; and, if the ADEC approved these proposals, "[s]uch mixing zones would be likely to allow the companies to continue discharging well above the mean lower low water mark."11 There is no hint in the draft permit that the EPA considered the effluent streams themselves to violate Alaska law in their course across the tidal flats.

We pause at this point to explain briefly the crucial notion of a "mixing zone." Environmental agencies do--and under present technology, must--permit polluted effluents to be discharged into natural bodies of water. By definition, the effluent itself does not meet water quality standards; otherwise, it would not be considered polluted. But the receiving water dilutes the effluent, and this dilution increases as the plume of effluent gradually diffuses in the receiving water. The "mixing zone" is simply the area of dispersal in the receiving waters where the pollutants in the effluent are not sufficiently diluted to meet water quality standards. It necessarily follows, then, that the edge or outer circumference of the mixing zone is defined as the boundary at which water quality standards are first met. The size and configuration of the mixing zone is a crucial variable in determining whether or not a given effluent can be discharged. If the permitted mixing zone is tiny--say, one meter in diameter--any effluent whatever will violate water quality standards; if the permitted mixing zone is huge--say, 100 kilometers--a tremendously toxic effluent can be discharged without violating water quality standards. The EPA has never set a general-purpose figure or formula for determining the proper mixing zone; instead, at least in some cases, the Agency appears to use a 100 meter mixing zone as a rule-of-thumb.12 Nor does the ADEC have a general-purpose mixing zone formula.

Soon after publication of the draft permit, Marathon protested the proposal to ban its effluent streams across the tidal mudflats. It submitted to the EPA and ADEC a study by an environmental engineering consultant which estimated that a time-averaged mixing zone of 440 meters for Trading Bay and 260 meters for Granite Point would give the dilutions of approximately 700:1 necessary to meet state water quality standards.13 The analysis in this report differed significantly from the EPA's analysis in at least two key respects. First, the proposed mixing zone was an average zone based on projected dilutions in the course of one complete tidal cycle. At intermediate stages in the tidal cycle, for example, the study projected that at the Trading Bay facility rapid currents and a submerged outfall would create near-field dilutions14 of 200:1 or more, and that far-field dilutions15 of 20,000:1 or more would occur within a mixing zone of 30 meters or less.16 On the other hand, at high tide and low tide the slow current produced all but stagnant receiving water and caused dramatically lower initial dilutions--16:1 or less; and, when coupled with the shallow level of the ambient water at low tide--even when the effluent was falsely assumed for computational purposes to enter the water at a depth of one meter, far-field dilutions were computed to require mixing zones of 10 kilometers or more for as little as an inadequate 300:1 dilution.17 The EPA's study, on the other hand, had considered only the worst case (low tide and zero current) mixing zone with the PLUME model, and had not averaged it with the smaller mixing zones at high tides or swifter currents. Second, as we have already noted, the consultant's study purported to model the near and far-field dilutions of the effluent stream flowing into the water's edge when the tide was below the discharge pipe at Trading Bay, and at all times at Granite Point. However, the computer modeling techniques do not appear to have been designed for this situation; for computational purposes they seem to have required that both the receiving water and the point of initial discharge be modeled at some depth greater than zero. The consultant therefore (arbitrarily) assumed a one meter depth,18 even though the effluent stream's initial contact with the receiving water is theoretically always at zero depth.

The EPA reviewed Marathon's proposal internally. One EPA scientist objected to Marathon's use of the near-field dilution model, apparently on the ground that the model was inappropriate for discharges at water's edge.19 An EPA official informed a Marathon representative by telephone of some of these initial objections.20

The EPA wrote the ADEC with official criticisms of Marathon's request.21 These took two forms: First, the letter itself included several objections but did not directly attack Marathon's mixing zone proposals. We have seen that the "mixing zone" determination is basically a cost-benefit judgment on a given set of environmental facts, rather than any sort of "scientific" determination; and both Marathon and the EPA appear to have assumed that the ADEC had regulatory discretion to approve a time-averaged mixing zone or a mixing zone large enough to permit Marathon's current forms of discharge to continue.22 Consequently, the EPA letter did not focus on the mixing zone requests in the Marathon proposal, apparently because of the realization that there was no "right" answer to the mixing zone question. Instead, the EPA for the first time began to emphasize the problem of the open waste flowing across the Alaska landscape. Still without any suggestion that the effluent streams were per se illegal under state law, the EPA argued that they created "a significant potential for adverse environmental or health effects" in violation of 18 A.A.C. 70.032. The EPA urged the ADEC to require more convincing refutation of these risks from Marathon before it approved the status quo. The letter concluded: "We look forward to continued discussions with you and with Marathon regarding the mixing zone determinations." R. 17433. Second, the EPA letter referred the ADEC to further responses attached as hand-written comments in the margins of a copy of Marathon's submission.23 These comments included a rejection of Marathon's mixing zone methodology, again on the ground that the projected initial dilutions were erroneously optimistic as a result of the false assumption that the discharges would enter the water at a fixed depth of one meter.24

In October, 1985, an EPA official had a telephone conversation with a representative of the ADEC which is memorialized in the record by the official's notes.25 The ADEC representative complained about the pressure from Marathon to remove the five meter isobath restriction, indicating her belief that the ADEC could not make the restrictions imposed by the EPA any less stringent. The EPA official apparently did not recognize that the ADEC thought itself constrained by federal law;26 but she nevertheless made it clear that state water quality standards were the only standards at stake and that the EPA was merely an interested observer of Marathon's request that the ADEC eliminate the ban on above-MLLW discharges.27 There was no hint of the later EPA theory that the discharges were per se violations of Alaska water quality standards.

Marathon wrote the ADEC again in November 1985, summarizing its position.28 It pointed out that state law controlled the water quality issue, urged the ADEC to approve the mixing zone proposed in its consultant's report, and argued that the 100 meter mixing zone chosen by the EPA was erroneously small even under federal standards. Marathon noted that extension of the outfalls at Trading Bay and Granite Point to comply with the draft permit would cost about $3 million. Marathon sent the EPA a copy of this letter along with other comments.29

EPA and Marathon representatives then met.30 According to notes by an EPA official, Marathon sought the EPA's approval of the status quo under section 403(c), 33 U.S.C. Sec. 1343(c), the "Ocean Discharge Criteria." The EPA explained that 403(c) was inapplicable to these discharges in the inland waters of Alaska. The EPA representatives apparently were surprised at Marathon's request, since Marathon had been actively lobbying the ADEC for a change in the isobath restrictions "because they believe it is up to the state, not EPA, to decide what to do."31 According to the notes, "Marathon urged the EPA to use the flexibility in the mixing zone definitions to allow a mixing zone enabling the facilities to remain status quo."32 There were further discussions of the lack of empirical data on harm to the environment, some disagreements on what would count as harm to the environment, and some suggestions for compromise solutions.33 Again, the discussions were about mixing zones; there was no mention of the EPA's subsequent theory that the open effluent streams were per se violations of Alaska law.

The ADEC wrote to Marathon in late November, responding point by point to Marathon's assertions.34 Most of these responses do not concern us; the letter does make clear, however, that the ADEC continued under the mistaken impression that it was bound by federal law to acquiesce in the EPA's mixing zone determination.35 The ADEC apparently still did not understand that the EPA was enforcing only Alaska water quality standards against Marathon's onshore facilities, and that it (the ADEC) had the final say about Alaska law.

In December 1985, the ADEC issued a "Certificate of Reasonable Assurance" pursuant to Section 401 of the Clean Water Act, 33 U.S.C. Sec. 1341(a)(1), as a tentative state certification that the draft permit was in accord with state law.36 The Certificate was conditional: it included various proposed changes to the draft permit to ensure compliance with state law, none of which was related to this case. The Certificate's cover letter stated:

Unless specifically noted through addition or modification of conditions in the accompanying certification, conditions of the proposed final general permit meet the requirements of State law and regulations and we believe cannot be made less stringent without violating the requirements of State law and regulations including Alaska water quality standards.

R. 22025 (emphasis added).

The ADEC wrote Marathon again in early January, ruminating about the lack of specific information on the environmental impact of the produced water discharges, but noting that "our hypothesis is that degradation is occurring in the intertidal area."37 The ADEC then made some rather opaque comments about mixing zones:

For Upper Cook Inlet, where the State Water Quality Standards are the primary regulatory authority, the configuration of mixing zones is a matter of state discretion. In the case of the draft general permit for Cook Inlet, EPA assumed a 100-m[eter] mixing zone.... Our certification of the draft permit indicates our agreement with the EPA's finding that State Water Quality Standards will be met with this mixing zone applied to the 5 meter depth contour.

R. 22031-32.38 Later in the letter, however, the ADEC returned to its erroneous position that "the State may specify conditions more stringent than provided under the NPDES permit but not less stringent." R. 22032. In September 1986, the ADEC issued its final Certificate.39 The cover letter included the same "cannot be made less stringent" boilerplate as the tentative certification.

In October 1986, the EPA issued the final general NPDES permit,40 published in the Federal Register a long commentary on the permit under the title "Supplementary Information,"41 and issued a document entitled "Response to Comments Received on Cook Inlet/Gulf of Alaska Permit."42 The final permit required that Marathon extend the discharge at Trading Bay to the 6.5 meter isobath, and the discharge at Granite Point to the 3.5 meter isobath below MLLW.43 The EPA's commentary on the permit included a brief discussion of the mixing zone problems. The commentary stated that, based on complaints that PLUME (used in setting the requirements of the draft permit) was inappropriate for the high-energy environment of Cook Inlet, a more appropriate computer model UMERGE was substituted for PLUME. The comments referred the reader to a document entitled "Computer Modeling of Dilution of Produced Water Discharges in Cook Inlet."44

The "Computer Modeling" document, dated September 1986, gave an elaborate description of the background of the problem, reported the variables and conditions in the effluent and ambient waters, and described the computer modeling techniques used to attain the final permit's depths and mixing zones of 6.5 meters and 750 meters at Trading Bay and 3.5 meters and 450 meters at Granite Point.45 UMERGE was substituted for PLUME because it permitted the modeling of dispersion into receiving water with currents; PLUME was designed for stagnant waters. This document has three points of interest for our purposes. First, despite having read and commented upon the Marathon consultant's report, the EPA did not use time-averaged modeling; to the contrary, it employed worst-case variables. As examples, depths were adjusted for lowest tide rather than MLLW or an average tide; and current speeds and ambient temperatures were modeled at values corresponding to lowest dilution and largest mixing zones. Second, computer models were produced for discharges in depths as shallow as one meter, and the results were considered unacceptable. Third, there is absolutely no mention of any problems associated with the flow of the effluent streams across the tidal flats--the document is entirely concerned with the fate of the effluent once it reaches the waters of Cook Inlet.

The EPA made the argument of which Marathon complains for the very first time in the "Response to Comments Received on Cook Inlet/Gulf of Alaska Permit,"46 issued with the final permit. In response to comments that there had been no showing of harm to the environment from years of produced water discharges, the EPA gave various reasons why the EPA and the ADEC were concerned about the open streams and the shallow water discharge. The EPA included a new argument for submerging the outfalls:

The shore based facilities in Cook Inlet cannot meet the water quality standards from their existing locations. Dilutions of 660 to 2100 are necessary to meet the 10 ug/1 standard [for aromatic hydrocarbons], depending upon the facility. However, no mixing of the effluent with the receiving water can occur at low tides when the outfalls are exposed and undiluted effluent flows over the intertidal zone.

* * *

* * *

The final permit retains the prohibition on the discharge of produced water to intertidal areas. As discussed above, effluents from the three shore facilities will run undiluted over the beach at times when the tide is lower than the outfall. At such times, the effluent has had no receiving water in which to mix, and a mixing zone does not exist. Thus, the shore facilities will be required to extend their outfall pipes at least to a point where they will be submerged at all times.

R. 22940. In essence, Marathon's whole argument turns on the last quoted sentence. That sentence suggests that the EPA required that the outfalls be submerged because it would not permit the open effluent streams. That one sentence is not enough. The record as a whole shows that this reasoning was not the basis of the EPA's decision to submerge the outfalls.

A fair summary of the record is this: The EPA's original modeling with PLUME showed that the Marathon outfalls should be submerged to five meters below MLLW; this formed the basis for the proposal in the draft permit. Marathon protested vigorously to the EPA and the ADEC, emphasizing to the ADEC that it had regulatory discretion to increase the size of the mixing zone or to choose more liberal techniques for computing it which would allow the current discharges to continue. Recognizing that the ADEC could set the mixing zone as it pleased, the EPA encouraged the ADEC to consider the potential for bad environmental effects from Marathon's open effluent streams. At the same time, the EPA reviewed and criticized Marathon's dilution modeling.47 The ADEC showed concern about the environmental effects of the open effluent streams across the mudflats, complained about the lack of hard data on contamination, and essentially acquiesced (perhaps on the basis of a mistaken view of the law) in the EPA's suggested mixing zones and the extension of the outfalls to the five meter isobath. In response to complaints that PLUME was inadequate, the EPA prepared a more elaborate computer modeling study using UMERGE. The authors of the study modeled outfalls as shallow as one meter--the same depth used by the authors of the Marathon study--and found the resulting mixing zones inappropriately large. The EPA directly incorporated the results of this study into the final NPDES permit. In response to comments noting the lack of perceptible harm from the current arrangements, the EPA argued that the open effluent streams in the intertidal zone were not permitted under Alaska water quality standards. The EPA also used this argument to justify the final permit's requirement that all outfalls be submerged.

The record demonstrates that the EPA did not rely solely on its allegedly erroneous view that state law prohibits Marathon's open effluent streams across the tidal flats. Instead, the EPA's determination was based primarily on computer modeling of mixing zones of the effluent plume in the waters of Cook Inlet. The EPA's argument to this effect is simply not a "classic post hoc rationalization" as Marathon asserts. If anything, the EPA's assertion that the effluent streams were themselves illegal under Alaska law comes much closer to a "post hoc rationalization." This argument did not emerge until the issuance of the final permit, more than a year after the EPA first decided to ban discharges above the five meter isobath. Since Marathon does not directly challenge the validity of the EPA's computer modeling techniques, its final position must be reduced to this assertion: The EPA has not adequately demonstrated that the current discharges from Trading Bay and Granite Point violate state water quality standards because it did not try to model directly the present configuration of the outfalls.48

We reject Marathon's contention that the EPA's failure to attempt to model the configuration of the present outfalls somehow invalidates its decision to require that the outfalls be submerged. The continuing and unavoidable implication of the record is that the EPA did not bother to model the discharges entering Cook Inlet at water's edge because it believed that those discharges would be less diluted and would require larger mixing zones than the worse-case underwater discharges actually modeled by the EPA at depths of two meters and less. That inference seems perfectly reasonable to us; in fact, the contrary inference is unimaginable. Even Marathon's experts did not model the actual configuration of the outfalls;49 moreover, the Marathon study actually confirmed the EPA's belief that the unsubmerged outfalls would result in low dilutions and require large mixing zones at low tide.50

For our purposes, the point is that the EPA actually considered and rejected Marathon's mixing zone methodology, relying instead on its own methodology in deciding that the outfalls at Trading Bay and Granite Point must be submerged. Marathon has not made a direct attack (usually futile, of course) on the quality of the EPA's applied science. That ends the case.

"We must even 'uphold a decision of less than ideal clarity if the agency's path may reasonably be discerned.' " American Petroleum Institute v. EPA, 661 F.2d 340, 348 (5th Cir.1981) (quoting American Meat Institute v. EPA, 526 F.2d 442, 453 (7th Cir.1975)). In this case, we can readily discern that the EPA considered and rejected the mixing zones methodologies proposed by Marathon in favor of a more conservative approach designed to guarantee compliance with state water column standards within a moderately sized mixing zone at any given time. "Mindful that EPA's choice of analytical methodology is entitled to a presumption of regularity," American Petroleum Institute v. EPA, 787 F.2d 965, 983 (5th Cir.1986) (citing United States v. Chevron Oil Co., 583 F.2d 1357, 1363-64 (5th Cir.1978)), we DENY Marathon's petition for review.