The American Petroleum Institute, an industry trade association, and four individual companies, Atlantic Richfield Company, Conoco, Inc., Exxon Corporation, and Mobil Oil Corporation (hereinafter collectively "API" or "industry"), seek judicial review of the action by the Regional Administrator for the Environmental Protection Agency's Region 10 ("Region 10") in issuing two permits controlling discharges of pollutants from offshore drilling rigs to the Alaskan Outer Continental Shelf ("OCS") and territorial seas. The two challenged permits, issued on May 30, 1984, are "area-wide" or general permits, authorizing discharges from oil and gas exploration rigs in areas of the Bering Sea and the Beaufort Sea. The permits allow certain discharges, but also set limits and conditions on the discharges. API contends that EPA overreached its statutory authority in imposing five of the discharge limitations and requiring two test methods in the permits, thereby restricting exploration.

The scope of our examination, in terms of both the data presented and the law involved, has been defined in previous decisions, and we dispense with repeating it anew. See, e.g., American Petroleum Institute v. EPA, 661 F.2d 340, 348-49 (5th Cir.1981). Having carefully reviewed the administrative record, it is our conclusion that one of the discharge limitations must be remanded to the agency, but that, enforcing traditional principles of judicial deference, the other features of the permits are approved.

The Clean Water Act ("CWA" or "Act"), 33 U.S.C. Secs. 1251 et seq., prohibits the discharge of any pollutant into the nation's waters unless the discharge complies with its specific requirements.1 Sec. 301(a), 33 U.S.C. Sec. 1311(a). Compliance may be achieved by obtaining a permit issued pursuant to Sec. 402, 33 U.S.C. Sec. 1342. National Pollutant Discharge Elimination System ("NPDES") permits are issued by EPA or, in those jurisdictions in which EPA has authorized a state agency to administer the NPDES program, by a state agency subject to EPA review.2 See 33 U.S.C. Sec. 1342(a)-(d). NPDES permits must incorporate applicable technology-based effluent limitations guidelines promulgated by EPA on a nationwide industry-by-industry basis under Secs. 301(b) and 304 of the Act. 33 U.S.C. Secs. 1311(b), 1314. Where EPA has not promulgated applicable technology-based effluent limitations guidelines, the permits must incorporate, on a case-by-case method, "such conditions as the Administrator determines are necessary to carry out the provisions of the Act." Sec. 402(a)(1), 33 U.S.C. Sec. 1342(a)(1). See Consolidated Coal Co. v. Costle, 604 F.2d 239, 248 n. 46 (4th Cir.1979), rev'd on other grounds sub nom. EPA v. Nat'l Crushed Stone Ass'n; NRDC v. Costle, 568 F.2d 1369, 1378-79 (D.C.Cir.1977).

Sections 301 and 304, which contain the effluent limitations guidelines, are the fundamental technology-related provisions of the Act. Section 301 sets sequential deadlines for the achievement of a series of increasingly stringent "technology-based effluent limitations." Section 301(b)(1)(A) directs the Administrator to establish effluent limitations requiring "the application of best practicable control technology currently available" ("BPT"), which dischargers were to have met by July 1, 1977. Section 301(b)(2)(E) requires the Administrator to establish effluent limitations for conventional pollutants to have been met not later than July 1, 1984, requiring "application of the best conventional pollutant control technology" ("BCT").3 Section 301(b)(2)(A) and (F), elevating the BPT standard even further, requires the dischargers to have begun applying "the best available technology economically achievable" ("BAT") to listed toxic4 pollutants, by July 1, 1984, and to all other pollutants5, by July 1, 1987. Section 304(b) of the Act (which will be discussed more fully infra) sets the technical criteria for determining effluent reductions attainable under BPT and BAT.

In addition to technology-based limitations, an NPDES permit for ocean discharges must also incorporate ocean discharge criteria established by EPA pursuant to Sec. 403(c) of the Act. 33 U.S.C. Sec. 1343(c). Ocean discharge criteria require EPA to ascertain that pollutant discharges will not have a significant adverse effect on the receiving water.

Oil and natural gas exploration in the offshore areas of Alaska began at Cook Inlet, off Anchorage on the south-central coast of the State, in the late 1950's and early 1960's. In the late 1960's and early 1970's, exploratory drilling spread to nearshore areas of the Alaskan Arctic. Federal offshore leasing there began in December 1979 with the lease of joint federal and state areas in the Beaufort Sea. In 1982 and 1984, additional OCS sales, both for the Diapir Field in the Beaufort Sea, were held. The Beaufort Sea general permit at issue in this case covers these lease sales.

The first federal lease sale for the areas offshore the western coast of Alaska was held in 1983 for Norton Sound. Subsequent federal lease sales in the Bering Sea occurred in 1983 for St. George Basin and in 1984 for Navarin Basin. The Bering Sea general permit at issue in this case covers St. George and Navarin Basins federal lease sales.

The offshore areas covered by the Bering Sea and Beaufort Sea general permits are home to a diverse, abundant and/or unique population of marine life. Subsistence fishing (in nearshore waters) plays an important cultural, social, and economic role in the lives of coastal residents. The Bering Sea, one of the world's major fishing grounds for both fish and shellfish, also provides an important feeding and breeding habitat, as well as a migratory pathway, for large numbers of marine mammals, and supports many of Alaska's most important marine and coastal bird populations. The nearshore shallow-water region of the Beaufort Sea is also an important marine and bird habitat.

Alaskan offshore oil and gas operations also possess unique features. Unlike offshore production areas of California and the Gulf of Mexico, the majority of current Alaskan operations involve exploratory drilling rather than development of proven reserves. Only the major oil companies operate exploratory wells in these Alaskan offshore areas, where extensive planning and a large financial commitment are necessary. In the deep waters of the Bering Sea (greater than 230 feet), drilling is conducted from newer semi-submersible drilling vessels built to survive high seas and harsh weather. In order to withstand ice forces in the Beaufort Sea, drilling is conducted from gravel islands (usually manmade), concrete island drilling structures or other special Arctic structures, which rest on the sea floor. At present these technologies can be used only in shallow waters (depths less than approximately 60 feet). The cost of drilling an exploratory well is approximately $40 to $50 million in the Beaufort Sea, not including the island or drilling structure (costing up to $100 million), and roughly $20 to $30 million in the Bering Sea. These figures compare to the cost for an average well in the Gulf of Mexico of $1.5 to $3 million.6

Region 10 issued its first general permits7 in 1983, one covering Norton Sound and another covering the Lease Sale BF in the Beaufort Sea (the "1983 permits"). 48 Fed.Reg. 54,881 (December 7, 1983). These two permits were not challenged. On May 30, 1984, Region 10 issued the Bering Sea and Beaufort Sea general permits which are the subject of API's instant petition for review. 49 Fed.Reg. 23,734 (June 7, 1984).

The permits here were the first in the nation to incorporate case-by-case effluent limitations purportedly based on BAT and BCT for the offshore oil and gas industry. In the absence of promulgated nationwide BAT and BCT effluent limitations guidelines, Region 10 made a best professional judgment ("BPJ") determination of what represented the appropriate BAT and BCT effluent limitations. 40 C.F.R. Sec. 125.3. In addition to these technology-based effluent limitations, the agency took into account factors imposed under Sec. 403(c) of the Act to prevent unreasonable degradation of the marine environment.

The permits authorize discharges from 15 different waste streams generated by offshore oil and gas operations. 49 Fed.Reg. 23,749. The permit conditions challenged by API relate to two of these waste streams, drilling fluids and drill cuttings. Drilling fluids, commonly called drilling "mud," include any fluid that is pumped down the drill pipe and through the drill bit, from the time a well is begun until cessation of drilling at that hole. Drilling muds have numerous functions, including maintaining hydrostatic pressure control in the well, lubricating the drilling bit, and removing drill cuttings from the well. Drill cuttings are the mineral particles generated by drilling into subsurface geologic formations. The drill cuttings and eventually the drilling fluids must be removed from the well and disposed. Petitioners challenge five of the limitations and two compliance test methods that apply to the drilling muds and drilling cuttings waste streams.

Four effluent limitations, designated as BAT limitations, appear in both permits and are designed to control the discharge of potentially toxic substances in drilling muds and cuttings. The limitations (1) limit the concentrations of mercury and cadmium in barite contained in the discharge muds and cuttings, (2) prohibit the discharge of muds or cuttings which have contained diesel oil, (3) allow only the discharge of muds and cuttings that Region 10 has determined to be within established toxicity limits, and (4) require prior authorization for the discharge of biocides.

The effluent limitation based on Sec. 403(c) ocean discharge criteria appears only in the Beaufort Sea permit. It prohibits the discharge of muds and cuttings between the shoreline of the mainland and islands and the two-meter isobath during the brief summer/fall season when nearshore waters are ice free.

The two compliance test methods require use of a "static sheen" method to detect free oil in drilling muds and cuttings, and a gas chromatography analysis to determine if oil present in a mud is diesel oil.

1. Barite Limitations.

Barite, a widely found barium sulfate mineral, is a major drilling fluid compound.8 8] Trace amounts of mercury and cadmium, both of which are CWA-listed "toxic" pollutants, are routinely found in barite. See 40 C.F.R. Sec. 401.15. Region 10 chose to regulate the cadmium and mercury content of barite under Sec. 304(b)(2), by setting best available technology-based limitations ("BAT"), using its best professional judgment ("BPJ"). The BAT limitation for barite was based on the technology of product substitution,9 i.e., using less toxic, but operationally satisfactory, mud components rather than installation of end-of-pipe treatment equipment. 49 Fed.Reg. 23,736-37. The permits provide that a drilling mud may not be discharged if the barite used in that drilling mud contains mercury in excess of 1 mg/kg or cadmium in excess of 3 mg/kg (dry weight basis). 49 Fed.Reg. 23,751. These amounts were drawn from an industry survey of the barite being used by the industry, in the Alaskan waters and elsewhere. The permits also contain an exemption by which Region 10 may allow the discharge if an operator shows he cannot comply with the limitation because of the unavailability of suitable barite.

API challenges the barite limitation on three grounds: that (1) EPA has failed to evaluate the costs of achieving the limitation; (2) the danger exists that EPA may incorporate similar restrictions in other regions; and (3) the limitation is unwarranted and fails to meet the ocean discharge criteria contained in Sec. 403(c).

EPA's duties with regard to the first of these contentions are defined in Sec. 304(b)(2) of the Act:

Factors relating to the assessment of best available technology shall take into account the age of equipment and facilities involved, the process employed, the engineering aspects of the application of various types of control techniques, process changes, the cost of achieving such effluent reduction, non-water quality environmental impact (including energy requirements), and such other factors as the Administrator deems appropriate; ... (emphasis added).

Unlike Secs. 304(b)(1) and 304(b)(4), which define the criteria for BPT and BCT, respectively, Sec. 304(b)(2) does not expressly direct that the Administrator compare costs with effluent reduction benefits in determining BAT limitations.10 API concedes as much, but nonetheless contends that the agency must evaluate the degree of pollutant removal to determine whether the costs associated with that removal are "reasonable." We agree with the Ninth Circuit's comment that, "at some point, extremely costly more refined treatment will have a de minimis effect on the receiving waters." Ass'n of Pac. Fisheries v. EPA, 615 F.2d 794, 818 (9th Cir.1980). Indeed, EPA would disserve its mandate were it to tilt at windmills by imposing BAT limitations which removed de minimis amounts of polluting agents from our nation's waters, while imposing possibly disabling costs upon the regulated industry. See Appalachian Power Co. v. Train, 545 F.2d 1351 (4th Cir.1976); Alabama Power Co. v. Costle, 636 F.2d 323, 360-61 (D.C.Cir.1979). As will be shown, a careful review of the record convinces us that Region 10's efforts to control minuscule trace amounts of mercury and cadmium in barite rest on still unconcluded investigations concerning their undersea toxicity. The point of regulation ad absurdum has not been reached in this case, but only because the costs imposed on industry by these permits alone will not be significant.

EPA's comments accompanying the permits admit that "the bioavailability, biotransformation and chemistry of these metals in discharged mud needs further study," 49 Fed.Reg. 23,744, and "the ultimate fate, transformation and bioavailability of these metals in drilling muds is poorly understood at present." Id. at 23,737. This is the best conclusion the agency could draw from a number of studies performed in the Gulf of Mexico, the Atlantic and the Beaufort Sea. In fact, our review of the record indicates no study which clearly (1) correlated increased cadmium and mercury levels with any sustained change in sediment composition near offshore drill-sites or (2) related any mercury or cadmium changes in marine fauna to the offshore drilling operations.11 Moreover, whether the small amounts of mercury or cadmium discharged in barite are bioavailable to the marine environment at all is questionable, because they ride as passengers in barite in highly insoluble form.12 The studies highlight the likelihood that changes in trace metal concentrations, if attributable to offshore oil operations, have effects that are fleeting, because of the dynamic ocean environment and the small area of their impact, and the ability of most marine organisms to swim through and away from the human activity. We will give EPA the benefit of the regulatory doubt concerning the necessity for imposing limitations on barite, as we must, BASF Wyandotte Corp. v. Costle, 598 F.2d 637, 652 (1st Cir.1979), noting that here, the technological feasibility and economic achievability have a sufficiently small impact on industry to overcome API's arbitrariness challenge.

API's objection to the cost of compliance ultimately fails because the permits here do not impose increased costs. Having decided to limit the mercury-cadmium content of barite used in offshore Alaskan operations, Region 10 examined the barite sources available to and used by the industry and concluded that adequate sources of bedded barite,13 predominantly from the Battle Mountain area of Nevada, would be available at competitive prices to allow the Alaskan operators to comply with the mercury and cadmium limitation. Comments by the oil and gas industry during the administrative proceedings acknowledge that bedded deposits in Nevada currently provide a primary source of barite for Alaskan operations. Testimony presented for the Alaskan Oil and Gas Association, to which petitioners belong, stated that

[t]he preferred source of barite for drilling muds to be used on the U.S. outer continental shelf is the Battle Mountain area of north-central Nevada where there are extensive bedded deposits of relatively pure barite. Based on the limited number of analyses available for cadmium and mercury concentrations in Alaskan offshore drilling fluids ... it is apparent that relatively metal-free barite is being used in offshore Alaska.

Furthermore, four operators (Exxon, Shell, Chevron and Placid Oil Co.) have contracted for Battle Mountain barite to be used in their Alaskan offshore drilling operations. "Clean" barite obviously was the product of choice of Alaska operators with or without EPA permit requirements. EPA thus properly assumed, without painstaking analysis, that for the limited scope of these permits, the cost of compliance would be small.

Further mitigating any adverse cost effect, in order to address the "possibility of unexpected changes in metals' content of 'clean' barite," the final permits included a special provision for approving the discharge of barite that does not meet the limitation. 49 Fed.Reg. 23,751. In such cases, the operator must demonstrate that uncontaminated barite was not available and provide an analysis of the substitute barite. Id. This special provision reinforces the agency's position that the permit terms are economically achievable and technologically feasible.

Ordinarily, a regulated industry would be expected to applaud the regulators' decision to allow the industry to use the type of material already in use. API fears, however, that this limitation, never contained in any other EPA permit covering offshore drilling, may become a national effluent guideline. Then, instead of requiring Battle Mountain-quality barite to be used on operations which accounted for less than 2 percent of all offshore drilling discharges through 1981, EPA's restriction might force the entire domestic drilling industry to compete for less than 50 percent of the available supply of barite. At such time, the economic achievability of such barite restrictions would become a significant regulatory factor. EPA could not conclusorily rest its cost consideration, as it did here, on the fact that the industry already uses the required barite. Moreover, EPA should adduce data that supports a determination of likely toxic effects of the compounds of mercury and cadmium which are found in barite.14

During oral argument, API noted the potentially harsh impact of EPA's "anti-backsliding" regulation, 40 C.F.R. Sec. 122.44(1), on limitations imposed, like the present ones, on the basis of regulatory uncertainty and conservatism concerning BAT requirements. The anti-backsliding regulation provides that, except in limited circumstances, an NPDES permit, once issued, will not be modified to become more lenient. If further scientific research undermines the reasonableness of BAT limitation, EPA should reconsider its terms. The anti-backsliding rule may not be used as a device to keep in place a regulation which, over the march of time and scientific progress, becomes arbitrary and irrational. EPA conceded as much in oral argument. Nor can anti-backsliding regulation modify EPA's statutory duty every five years to review and revise its guidelines, 33 U.S.C. Sec. 1311(d), and to apply specific statutory guidelines in issuing permits. 33 U.S.C. Sec. 1311(b).

We affirm the agency's BAT-based limitations on mercury and cadmium. Accordingly, it is not necessary to discuss API's contention that the limitations may not be supportable under the separate but not preemptive Sec. 403 ocean discharge provisions.

2. Diesel Oil Ban.

In the past, industry has routinely added diesel oil to the drilling fluid (1) as a general lubricating agent to reduce torque and drag and (2) as a "pill" to free pipe stuck in the well bore.15 Beginning with the 1983 permits, Region 10 has all but disallowed this practice by issuing permits that prohibit the discharge of drilling muds that "have contained diesel oil as an additive," and proposes, as a BAT-based control, the substitution of mineral oil in its stead. API concedes that mineral oil would be generally as effective a lubricant as diesel oil in normal drilling operations, but adamantly maintains that it is inferior to diesel oil for freeing stuck pipe. Industry consequently challenges the limitation, contending principally that in banning diesel oil under BAT, Region 10 has violated national EPA regulations, that the prohibition presents serious operational difficulties, and that the required cost analysis has not been performed.16

Our discussion begins with the threshold question, API's contention that Region 10 impermissibly "upgraded" diesel oil from a conventional pollutant to a toxic pollutant in violation of EPA national guidelines. It is elementary administrative law that an agency must operate within the confines of its own regulations. Morton v. Ruiz, 415 U.S. 199, 232-36, 94 S. Ct. 1055, 1073-74, 29 L. Ed. 2d 270 (1974). Region 10 has failed to abide by this maxim in banning the discharge of drilling muds that have "contained" diesel oil.

Region 10's unprecedented designation of diesel oil as a "toxic" was not in accordance with its own regulations and therefore arbitrary. The permits explicitly regulate diesel oil as a "toxic" pollutant subject to BAT limitations.17 Diesel oil is not, however, a toxic duly listed in 40 C.F.R. Sec. 401.15 pursuant to Sec. 307 of the Act. Rather, it belongs among the "oil and grease" family of conventional pollutants identified at 40 C.F.R. Sec. 401.16. In its preamble to the rule classifying oil and grease as conventional, EPA recognized that this pollutant category covers oil and grease "from animal and vegetable origin and those associated with petroleum sources.... [I]t is the entire class of oil and grease which has traditionally been of concern in waste water control." 44 Fed.Reg. 44,502 (July 30, 1979). Not only has EPA previously failed to identify diesel oil as a "toxic," but it has traditionally prohibited only discharges of "free" oil (see, e.g., 40 C.F.R. Secs. 435.12, 435.42 and 435.52) and oil-based drilling fluids. Regions 6 and 9 have each issued offshore drilling permits that contained no general diesel oil ban. See Gulf of Mexico General Permit, 46 Fed.Reg. 20,28 4 (April 3, 1981), and Southern California General Permit, 47 Fed.Reg. 7,312, 7,315 (February 18, 1982). The only regulations that prohibited the use of diesel oil were the 1983 Norton Sound/BF general permits, issued by Region 10 just five months prior to those in this case.

The agency's brief to this court no longer seeks to justify its prohibition of diesel oil as a "toxic." Instead, the agency characterizes diesel oil as an "indicator pollutant," which, because of its "direct correlation" with certain toxics and inherently toxic qualities, deserves BAT regulation. A specific provision of the agency's regulations governs the determination of indicator pollutants. 40 C.F.R. Sec. 125.3(g)(1).18 However, the term "indicator" appears nowhere in the permits; the EPA nowhere followed or referred to this provision in its explanation accompanying the permits. An agency may not justify its action by means of post hoc rationalizations. Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 169, 83 S. Ct. 239, 246, 9 L. Ed. 2d 207 (1962). See Asbestos Information Ass'n/North America v. OSHA, 727 F.2d 415, 425 (5th Cir.1984) ("It is axiomatic that the reasons the Agency gives at the time it acts form the actual basis for the Agency's action"). As stated in Industrial Union Dept., AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 631 n. 31, 100 S. Ct. 2844, 2858 n. 31, 65 L. Ed. 2d 1010 (1980) (plurality):

[W]e have often held [that] the validity of an agency's determination must be judged on the basis of the agency's stated reasons for making that determination. See SEC v. Chenery Corp., 318 U.S. 80, 95, 63 S. Ct. 454, 462, 87 L. Ed. 626 ("[A]n administrative order cannot be upheld unless the grounds upon which the agency acted in exercising its powers were those upon which its action can be sustained"); FPC v. Texaco, Inc., 417 U.S. 380, 397, 94 S. Ct. 2315, 2326, 41 L. Ed. 2d 141; FTC v. Sperry & Hutchinson Co., 405 U.S. 233, 249, 92 S. Ct. 898, 907, 31 L. Ed. 2d 170.

See also Texas Power & Light Co. v. FCC, 784 F.2d 1265 (5th Cir.1986). The vice in post hoc justification is that it deprives regulated parties of a fair opportunity, during the give and take process of policy formation, to comment upon the agency's proposal. EPA recognized, when it defined "oil and grease" as conventional pollutants, that they may in some cases be used as indicators of toxic pollutants. The agency advised commenters at that time to "reserve objections to their use for those regulations in which such an approach is employed." 44 Fed.Reg. 44,502 (July 30, 1979) (emphasis added). The agency thus expressed its intention to identify conventional pollutants as toxic indicators specifically in future proceedings, affording commenters the opportunity to discuss its methodology. In this case, however, the best justification put forth by the agency for declining to identify diesel oil as an indicator pollutant is that it implicitly treated it as such. Based on the preceding analysis of EPA regulations and the terms of the permits, we reject this argument as a post hoc rationalization. This part of the permits must be remanded to the agency for further consideration.

The stakes are high in this controversy over whether diesel oil can be regulated as a toxic, an indicator pollutant or a conventional pollutant. If a toxic, it is subject to control on the basis of BAT; as an indicator pollutant it may be subject to BAT-level control; and if a conventional pollutant it is subject to BCT standards. BCT control requires, inter alia, "consideration of the reasonableness of the relationship between the costs of attaining a reduction in effluents and the effluent reduction benefits derived...." 33 U.S.C. Sec. 1314(b)(4)(B). Blithely assuming that diesel oil could be regulated as a toxic, Region 10 gave rather short shrift to industry's operational, safety and cost concerns, as well as the actual evaluation of the toxicity of the diesel oil when used solely in infrequent application as a "pill." Operationally, industry representatives unanimously contended that the diesel pill is "state of the art" for freeing stuck drill pipe. Insufficient testing has been done to enable them to use mineral oil confidently as a substitute.19 Because EPA did not take these comments at face value, it assumed that the cost of a mineral oil substitute for this type of procedure would be very low. However, industry demonstrated that if the diesel pill continues to be the preferred tool for fishing expeditions, and if its use requires the removal of the entire mud supply for the well, a cost of $200,000-400,000 per well will be incurred. Further, barging to, and disposal of the mud, onshore, given unfavorable climatic conditions in the Arctic, could be risky for personnel aboard the drilling rig. Although no industry spokesman denied that diesel oil contains toxic substances, the amounts of such toxics vary widely with the composition of diesel oils. Moreover, further research is needed to gauge the actual toxicity in the marine environment of residual diesel oil after removal of the diesel pill and appropriate mud buffers below and above it, because of the rapid dispersion of trace substances in the ocean and lack of any prolonged contact with marine flora or fauna.20 The purpose of BCT controls is to eliminate "treatment for treatment's sake." Even under a BAT regime, EPA should engage in a careful analysis of the statutory and regulatory criteria before departing from prior permit terms and depriving the oil and gas exploration industry of its most effective means to recover stuck pipe.21

Because EPA did not act in accordance with its regulations in enacting a ban on "drilling fluids which have contained diesel oil," this portion of the permits is invalid and unenforceable. We remand to EPA for further proceedings consistent with this discussion.

3. Mud Toxicity Limit.

The Bering/Beaufort permits, as well as many previous offshore oil and gas exploratory permits, include an effluent limitation for controlling the discharge of toxic substances (both listed toxic pollutants and non-conventional/non-toxic pollutants) by requiring an operator to certify that it will discharge only drilling muds and additives that Region 10 has authorized for discharge. Over the years, EPA, with the cooperation of the oil and gas industry, has developed a list of eight generic mud types and "approved" specialty additives encompassing most of the compositions used for offshore drilling. Both the Norton Sound/BF general permits, 48 Fed.Reg. 54,889, 54,896, and the Bering/Beaufort permits at issue in this case adopted the list of generic muds and approved additives. In addition, the Bering/Beaufort permits allow operators to discharge new additives by requesting and receiving approval from Region 10 prior to discharge.

API does not challenge these provisions. Industry's complaint pertains only to a special provision allowing operators to discharge muds with new additives prior to Region 10 approval. Pre-approval discharge is allowed if the unapproved additives pass two screening criteria. First, for generic Mud No. 1, there can be no increase in toxicity; hence, no additive may be allowed which will cause that effect. Second, for Muds Nos. 2-8, the unapproved additive may not cause a substantial increase in toxicity, defined in the permits as a 10 percent decrease in the LC-50 of the second most toxic generic mud and a corresponding greater percent decrease for those muds that are less toxic. 49 Fed.Reg. 23,750, Permit Part IIA. (1)(e)(2)(b) EPA justifies both of these limitations as conservative measures designed to implement BAT limitations on effluent discharge and the Ocean Discharge Criteria Evaluation (ODCE). 49 Fed.Reg. 23,750-51.

In the limited context of its challenge, what API has failed to demonstrate is as significant as the arguments it makes before this court. API has failed to identify any single unapproved drilling additive, except mineral oil, which will violate the agency's proposed toxicity limits for pre-approval discharge. It has failed to identify any specific technological constraints on drilling operations that may occur as a result of the limit. It has failed to suggest that the permit terms concerning discharge of non-approved additives will result in inordinately higher costs of drilling. API's only challenge, then, lies in an argument based on the excessive conservatism of the regulation from a scientific perspective. No actual damage to the industry has been shown.

Because the mud toxicity limit enunciated in the permits is based upon the slightly toxic composition of generic Mud No. 1, API first attacks the calculation of toxicity for that mud. Industry points out that if no combination of mud plus additives can have a toxicity greater than that of Mud No. 1, then Mud No. 1 is effectively unusable in operations covered by these permits.22

EPA measured toxicity of Mud No. 1 by using a standard bioassay test known as the "96-hour LC-50" test. Typically, such bioassays are carried out in aquaria containing a range of animal species. The animals are exposed to different concentrations of the drilling mud for a set time, usually 96 hours. Then, by observing mortality rates and by calculation, the concentration required to kill 50 per cent of the test animals in 96 hours is determined. The "96-hour LC-50" is defined as the lethal concentration of a toxicant that will kill 50 percent of the test organisms after a 96-hour exposure. Thus, the lower the LC-50 value, the higher the relative toxicity.23 EPA calculated an LC-50 of 3,000 ppm (parts per million) as the acute toxicity level for generic Mud No. 1. So described, Mud No. 1 falls within the ambit of "slightly toxic" substances, while EPA acknowledges that generally, "drilling muds must exhibit low toxicity." 49 Fed.Reg. 23,735. API characterizes this toxicity level as unduly restrictive and "irrationally overprotective."

The nub of the industry's contention is that the bioassay test, based on a four-day-long exposure of critical marine fauna to drilling mud, creates a completely unrealistic situation which would never occur in nature because of the rapid dispersion of drilling mud from an offshore platform as a result of simple dilution, currents, and storms. In its brief, API proposes an alternate calculation of acute toxicity, using a one-hour LC-50, whose adoption would result in a permissible minimum toxicity of at least ten times higher than that allowed in the permits. Needless to say, the higher toxicity level would permit pre-approval use of a much wider variety of non-listed additive and mud combinations.

API's suggestion that the dynamic ocean environment would never result in the extended exposure to drilling mud contemplated by the 96-hour LC-50 test commends itself to common sense. On the other hand, API concedes that test to be perhaps the most widely accepted benchmark for toxicity evaluations by EPA. Therefore, EPA has not selected a patently irrational methodology to measure the relative toxicity of generic Mud No. 1. Under such circumstances, we are required to resist API's attempts to substitute our judgment for that of the agency, Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416, 91 S. Ct. 814, 824, 28 L. Ed. 2d 136 (1971), and must sustain its choice. Permian Basin Area Rate Cases, 390 U.S. 747, 810-11, 88 S. Ct. 1344, 1382-83, 20 L. Ed. 2d 312 (1968); BASF Wyandotte v. Costle, 598 F.2d at 647-50, 655.

Even if, as API suggests, the effect of setting the general mud toxicity level at that for Mud No. 1 effectively eliminates its availability for use in the area of operations covered by the permit, this is not necessarily an arbitrary and capricious result. EPA justifies its action as a BAT-based effluent limitation. EPA believes that the mud toxicity restriction is technologically achievable by means of product-substitution. That is, if operators are unable to use Mud No. 1 in the permitted areas, they can employ generic muds Nos. 2-8, together with approved additives, and such other combinations as may be allowed under the Section II.A(1)(e) and (f). API did not rebut this conclusion; so it must stand unchallenged in the record.

The toxicity limitations also apply to mineral oil used as a lubricating or spotting agent, requiring EPA's prior approval that (1) the selected mineral oil would not cause the drilling mud to be more toxic than Mud No. 1 or (2) the operator demonstrates that the chosen product is the "least toxic available alternative." 49 Fed.Reg. 23,751, Permit Part II.A(1)(g). An industry representative voiced API's basic concern that because the toxicity of muds containing mineral oil may have LC-50's below 3050 ppm, no prudent operator could certify in advance that the discharged mud will comply. Dr. Robert C. Ayers, Jr. of Exxon Product Research Co., Comments on the Bioassay Text Certificates & Need for the Generic Mud and Approved Additives Approach. API asserts that there are no standards for determining what mineral oil is the least toxic available alternative, thus leaving unbridled enforcement discretion in the agency. We find the agency's responses persuasive. Facing incomplete knowledge on the toxicity of various mineral oils, EPA chose to require advance certification of mineral oils. Further, by allowing use of the "least toxic available alternative," the agency placed the onus on industry to explore the alternatives in advance of discharge. The limitation, as stated, does not prevent the use of mineral oil additives, it only regulates their approval. This was a rational regulatory decision and enables the development of approved mineral oils.

API finally contends that the mud toxicity limit is irrational because it prohibits discharge of muds prior to EPA approval if the toxicity of the mud-additive combination is greater than that of the generic mud by more than 10 percent. Industry points out that, because generic Muds Nos. 2-8 exhibit LC-50 acute toxicity levels that are very low and in some cases untraceable,24 operators would be prohibited from discharging combinations that were far less toxic than the nominally-permitted discharge of unalloyed generic Mud No. 1. Again, the complaint voiced by industry lacks references to specific adverse effects that would result to the technology of drilling operations or their cost.25 EPA's not unreasonable rejoinder is that if any additive not previously approved by Region 10 causes a substantial increase in the toxicity of the generic mud, Region 10 believes that it has a responsibility to evaluate such additive prior to its discharge. The greater the effect of an additive on the toxicity of the drilling mud to which it is added, the more important to carefully evaluate the compound prior to discharge. This logic is not arbitrary and capricious, and we must uphold it.4. Biocide Discharge Restriction.

The permits place biocides within the BAT effluent limitations and prohibit their discharge without prior EPA approval. 49 Fed.Reg. 23,750, Permit Part II.A(1)(e)(2)(c) and II.A(1)(f). API contends that the restriction is unreasonable because biocides are necessary to suppress growth of bacteria in the starch or xanthum gum polymer that must be added to generic Mud No. 1. Without the use of biocides, API contends, bacteria growing in the drilling mud render it useless in a few days, and it must be discarded. EPA claims record support for its finding that there are available alternatives for starch and xanthum gum polymer which do not require the use of biocides.26 The agency does not reject the possibility that biocides may be given pre-clearance for use with Mud No. 1, despite the toxicity restriction previously discussed. Further, under the permits, biocides are treated no differently from other additives. They simply do not also fall under the special exceptions provided for new additives which allow discharge before Region 10 approval if the above-mentioned criteria are met. Because biocides are made and used to kill bacteria, they are potentially damaging to other living organisms, and Region 10 appropriately determined that biocides must be evaluated before discharge. API has adduced no reasons why the use of biocides cannot be planned in advance, as are many aspects of exploratory operations in Alaska, allowing time for EPA evaluation and approval.

API levies several additional challenges which misconstrue the permit terms and applicable law. First, API contends that the approval process is unreasonable because it requires submission of information on the location and duration of the discharge sixty days prior to the discharge. 49 Fed.Reg. 23,750. Permit Part II.A(1)(f). The permit actually says that EPA may take up to 60 days to evaluate a request. Normally, the agency approves or disapproves requests within a couple of weeks. Information on the approximate date of the discharge is needed in order for EPA to determine how quickly it needs to respond.

Second, API argues that the permits do not define "biocide" and do not state whether they are toxic, conventional, or non-conventional pollutants. API contends that most are non-conventional pollutants and are thus not subject to BAT limitations before July 1, 1987. Sec. 301(b)(2)(F), 33 U.S.C. Sec. 1311(b)(2)(F).27 Therefore, the permit term is arbitrary and capricious as a matter of law. API has misread the Act. The statute requires that EPA promulgate BAT effluent limitation guidelines applicable to non-conventional pollutants not later than July 1, 1987. Until such guidelines are promulgated, the EPA may impose BAT limitations case-by-case. Sec. 402(a)(1), 33 U.S.C. Sec. 1342(a)(1).

Third, the industry claims EPA ignored the cost analysis required for BAT limitations. In fact, the permits do not prohibit all discharge of biocides, and the record does not support API's claim that biocides are essential for Mud No. 128, or that without them the cost of operating with Mud No. 1 will be prohibitive.29 EPA also contends that it did not consider costs and concluded the limitations were economically feasible, relying on the standards set forth in the case law.30

API has failed to convince this court that Region 10's limitation ban on biocides is arbitrary or capricious.

5. Ocean Dumping Criteria.

The Beaufort Sea permit prohibits the discharge of drilling muds between the shoreline and two-meter isobath during the open-water season (approximately three months a year). 49 Fed.Reg. 23,757, Permit Part II.A(2)(c). This permit extended a corresponding limitation contained in the 1983 permit to water shallower than two meters off the near-island shoreline (the prior permit having covered depths off the mainland only).

The EPA is required by the Act and implementing regulations to certify that any ocean discharge allowed by its permit will not cause an unreasonable degradation of the marine environment. 33 U.S.C. Sec. 1343(a). The statute states that if EPA is unable to obtain sufficient information on any proposed discharge to make a reasonable judgment as to its environmental effect, "no permit shall be issued ..." 33 U.S.C. Sec. 1343(c)(2). See also 40 CFR Sec. 125 subpart M.

Following the prescribed procedures for evaluating the proposed discharge of drill muds and cuttings, EPA prepared two Ocean Discharge Criteria Evaluations (ODCE), one prior to issuance of the 1983 general permit and one preceding the 1984 general permits. The latter ODCE found there was "insufficient evidence to identify specific impacts on environmentally important areas [including] ... foraging areas in the nearshore zone (0-2m) for fish, birds, and mammals.31 Existing theoretical models for dispersion of drill muds and cuttings, one of which was developed by the Offshore Operators Committee, an industry body, are inapplicable to water less than five meters deep. 49 Fed.Reg. 23,738. Further, the nearshore area "is the most productive area on the Beaufort Sea for fishes and water fowl.... It is a critical habitat for spawning, migration, and/or overwintering of the major subsistence species...."32 The majority of fish species found near shore during the open-water season are anadromous species, i.e., those which migrate from fresh water streams to the sea, and hence find the nearshore area an essential feeding zone. Major commercial and native fisheries also depend on these species.33 The ODCE indicates gaps in our knowledge of overwintering areas for fish, of dispersion of discharges in shallow water,34 of the impact on benthic organisms which are essential food for demersal fish,35 and of biological and life-cycle data for many species.36 EPA therefore concluded that "a potential for irreparable harm may exist for populations depending on near-shore benthos for food," and pursuant to statute, it was precluded from authorizing effluent discharges in the nearshore environment during open-water season.

API argues that EPA ignored evidence in the record showing that shallow water discharges disperse rapidly because of the natural accumulation and dispersion of sediment and would hardly affect, much less irreparably harm the marine environment.37 EPA allegedly also ignored testimony that only larger, more mature, less vulnerable fish enter the shallow waters to feed during open water periods in the summer, and that such fish, being highly mobile and highly tolerant of a wide range of temperatures, salinities, and turbidity, would not be exposed long to effluents.38 Alternatively, API suggests that the permit limitation be modified (similar to a State of Alaska limitation) to forbid discharge of drilling fluid unless the permittee shows the effluents can be adequately dispersed. It also challenges the EPA's conclusion that there are insufficient data to determine that the discharges, taking into account all the technology-based limitations, would not irreparably harm the marine environment. API summarizes that, in making its determinations, the EPA failed to consider any of the testimony and information provided by the industry.

The agency responds that it did consider the above evidence with respect to the 1984 Beaufort Sea general permit, and commented upon it. 49 Fed.Reg. 23,746, Response to Comments 26 and 32. Moreover, industry offered essentially the same data when commenting on the 1983 permit. Because EPA formally responded to industry's challenge concerning the 0-2 meter isobath discharge prohibition, see Comments 26 and 32, we cannot agree that the agency "ignored" the testimony of industry's highly qualified experts, and we are not permitted to substitute our evaluation of the evidence for that of the agency.39

6. Sheen Test.

Since 1979, EPA's BPT effluent guidelines and general permits have forbidden the discharge of "free" oil in drilling muds and cuttings. The permits, then as now, define the limitation to mean that a discharge may not "cause a film or sheen upon or a discoloration on the water or adjoining shorelines .... 49 Fed.Reg. 23,755. API does not question the restriction itself here, but instead challenges the methodology EPA has chosen to monitor compliance with the restriction.

Under permits issued by Region 10 prior to 1983, operators monitored their discharges by observing the surface of the receiving water after bulk discharges of drilling muds to determine if a sheen was present ("visible sheen" test). Region 10 questioned the efficacy of that test in Alaskan operations, where observation of the receiving waters would be hampered by the extended hours of darkness and the receiving waters themselves would be either frozen sea ice or obscured by icy conditions.40 In response to these concerns, for the 1983 Norton Sound/BF permits, Region 10 developed a new method, the "static (laboratory) sheen" test, which involves testing the discharge on the platform itself, prior to depositing it into the receiving water. The material to be discharged is placed in a container of seawater and stirred for a specified time, following which the surface of the liquid in each container is examined. If a "sheen, color, irridescence or an oil slick" is visible on greater than one half of the water in the test container, the fluid may not be discharged.

API contends that the new test is unproven and unreliable, and thus arbitrary and capricious. API had raised this same argument in connection with the issuance of the 1983 Norton Sound/BF permits, but had still not, during the comment period here, offered any supporting data.41 Mindful that EPA's choice of analytical methodology is entitled to a presumption of regularity, United States v. Chevron Oil Co., 583 F.2d 1357, 1363-64 (5th Cir.1978), we note that API has a considerable burden to overcome.

The record shows that prior to the issuance of the Bering/Beaufort permits, EPA performed the static sheen test on 54 samples of drilling muds and cuttings. The samples were mixed in varying amounts with sea water, resulting in 118 separate test formulations. The test formulations were each divided into three separate test containers, which were then visually examined to determine if a sheen appeared on the water surface of the test container. Consequently, 354 separate observations were conducted.

The results of these tests demonstrate that the test method is reliable and reproducible. Not one single false positive observation was made of samples with no added oil. There was almost complete reproducibility among all the samples which were of sufficient size to meet Region 10 guidelines, and where there was a variance, it indicated the presence of trace amounts of free oil that were near the limit of detectability. Significantly, from the dischargers' standpoint, these variances did not result in a false positive result.

API alternatively argues that the static sheen test could give rise to unwarranted enforcement penalties under Sec. 309 of the Act, 33 U.S.C. Sec. 1319(d),42 if the test results erroneously indicate a violation has occurred. Should that happen, API fears that because judicial review of the "action of the administrator" in "issuing any permit" is available under Sec. 509(b), 33 U.S.C. Sec. 1369(b), it is possible that no review of the validity of the sheen test itself would be permitted in the enforcement action, under Sec. 509(b)(2) of the Act.43 API, however, misses a step in its analysis, because the test provides operators with safeguards to avoid violations. By analyzing muds and cuttings prior to discharge, materials that would violate the "no discharge of free oil" limitation would not be discharged. Further, Region 10's guideline provides that if there is some question as to the accuracy of the test as applied to any particular sample, the test may be repeated. (By contrast, under the visible sheen test, violations would be detected only after they occurred, i.e., after discharge.). The speculative nature of this argument is underscored by API's choice of support, Marathon Oil Co. v. EPA, 564 F.2d 1253, 1273 (9th Cir.1977), where the court set aside agency action which would have required permittees to achieve compliance with standards 100 percent of the time. API's reliance on Marathon is misplaced, because that case held only that permits must contain a formal "upset" provision, so that unintentional noncompliance beyond the control of the permittee would not constitute violations of the permit standards. The permits here include such an upset provision, 49 Fed.Reg. 23,753, an additional safeguard against unwarranted penalties.

Based on our review of the record before us, Region 10 did not act arbitrarily or capriciously in requiring the use of the static sheen test in the Alaskan operations.

7. Gas Chromatography Test.

We review API's challenge to another of the analytical test methods required by the permits, gas chromatography, again hampered by the lack of any data in the record to support API's attack. The permits prohibit the discharge of drilling muds and cuttings contaminated by diesel oil and require permittees to monitor compliance with this effluent limitation by gas chromatography analysis. 49 Fed.Reg. 23,749-50. In this method, a gas chromatograph machine would be used to compare the types of hydrocarbons in the drilling mud with those present in the diesel oil stored on the drilling rig for fuel.44 The discharge chromatographs, each with its identifying "fingerprint," are compared with those of the stored diesel fuel to determine if the diesel fuel is present in the sample. Monitoring is required once on each mud system at its greatest well depth and on any sample of drilling mud which is found to violate the static sheen test after discharge.

API does not challenge the validity of the procedure per se, but rather its use in detecting the presence of diesel oil. Specifically, API contends that the method (1) yields erroneous results, (2) requires the use of highly trained specialists to perform and interpret the results, and (3) like the static sheen test, may give rise to apparent violations of permit conditions, thereby exposing permittees to unwarranted penalties. Its sole support for these contentions, however, is a report that is not part of the record and we decline to consider it prior to examination by the agency.45 In the absence of evidence to the contrary, we cannot conclude that EPA, by requiring the use of gas chromatography for determining compliance with the prohibition of discharge of diesel oil, has employed its discretion in an arbitrary and capricious manner.

The Region 10 provisions here challenged, with the exception of the diesel oil limitation, are APPROVED. The diesel oil provision is VACATED and REMANDED to the agency for further proceedings consistent with this opinion.