HIGGINBOTHAM, Circuit Judge:

This case, along with Nos. 94-30419 and 95-60228, concerns the past actions and future fate of Marine Shale Processors, Inc., a hazardous waste treatment facility. The cases involve multiple aspects of each of the federal environmental laws as affecting disputes between the Environmental Protection Agency and MSP. We provide a brief explanation of the three cases in this opinion before discussing the specific issues raised by this appeal.

I

In 1985, Marine Shale Processors, Inc. opened a facility in Amelia, Louisiana purporting to recycle hazardous waste through its newly acquired rotary kiln, a mechanism 275 feet long and 11 feet in diameter with the capacity to heat materials to temperatures in excess of 2000 degrees Fahrenheit. MSP's treatment process began with placement of materials in its kiln. From there, most material traveled through oxidizers and slag boxes. The process generated significant quantities of smoke, flue gases, and air particles. Carcinogenic heavy metals tended to concentrate in these air particles. The air pollutants passed through baghouses, which collected some of the material in the form of caked dust. The dust dropped off the bags to the bottom of the baghouse, where it was collected, run through the oxidizers and slag boxes, then combined with the rest of the material produced from the primary process. The nature of MSP's operation made it subject to federal and state laws limiting pollution of water, air, and land. See The Resource Conservation and Recovery Act, 42 U.S.C. §(s) 6901-92k; The Clean Air Act, 42 U.S.C. §(s) 7401-7671q; The Clean Water Act, 33 U.S.C. §(s) 1251-1376. These laws required MSP to obtain permits specifying the type and amount of pollutants that it could discharge into the environment.

RCRA regulations divide facilities using heat to process hazardous waste into three basic types, incinerators, boilers, and industrial furnaces. See 40 C.F.R. Section(s) 260.10 (defining all three terms). From 1980 to 1991, the regulations required only facilities engaged in incineration to obtain permits before operating. See Final Rule, Burning of Hazardous Wastes in Boilers and Industrial Furnaces, 56 Fed. Reg. 7134, 7138 (1991); 40 C.F.R. pt. 264 subpt. O. In 1991, EPA amended the regulations to require all facilities using thermal processes to treat hazardous waste to obtain one of two types of permits. 56 Fed. Reg. at 7134. Incinerators needed Subpart O permits, and boilers and industrial furnaces were required to obtain BIF permits. See 40 C.F.R. pt. 266 subpt H. Since opening operations in 1985, Marine Shale has claimed that its kiln system constitutes an industrial furnace under the RCRA regulations. When EPA amended the regulations to require all thermal treatment facilities to acquire permits, MSP filed a six volume permit application with EPA Region VI. Four years later, EPA denied this permit application. Invoking our authority to set aside final agency action under the Administrative Procedures Act, 5 U.S.C. § 706(2), MSP appealed the permit denial. In number 95-60228, we address this denial.

In 1990, the United States sued MSP under RCRA, alleging that MSP was an incinerator of hazardous waste operating without the required Subpart O permit and was illegally disposing incinerator ash on the ground. The United States later amended its complaint to allege violations of the CWA, the CAA, and other provisions of RCRA. Southern Wood Piedmont Company, the entity sending the largest volume of hazardous waste to Marine Shale, intervened and sought a declaration that all material resulting from the processing of its waste was exempt from RCRA regulation. The Louisiana Department of Environmental Quality sought to intervene as a plaintiff; the district court allowed LDEQ to intervene but prohibited it from asserting claims other than those brought by the United States.

Early in the litigation, District Court Judge Haik granted the United States' motion for a preliminary injunction prohibiting MSP from transporting the material resulting from its process away from grounds owned by MSP or its sister corporation, Recycling Park, Inc. After former representatives of MSP attempted to bribe Judge Haik, Fifth Circuit Chief Judge Politz ordered the case transferred to Judge Adrian Duplantier, who has presided since.

Judge Duplantier divided the litigation into phases. In the first phase, the United States and SWP tried their RCRA claims to a jury. After a five-week trial, the jury was unable to agree to answers to four of thirteen interrogatories. Judge Duplantier declared a mistrial on the claims prosecuted by the United States and granted SWP's motion for partial judgment under Fed.R.Civ.P. 54(b). Dissatisfied with the scope of this judgment, SWP appealed to this court. Contending that the district court erred in entering the Rule 54(b) judgment, the United States cross-appealed. In number 94-30419, we address the appeals from this judgment.

After this first attempt to resolve RCRA issues, Judge Duplantier proceeded to the later phases of the case. The court conducted a bench trial on the CWA and CAA issues. It also resolved certain outstanding RCRA claims by summary judgment. The sum of the district court's rulings was that MSP had violated several provisions of all three environmental statutes. The district court fined MSP for each violation and granted the United States' request for injunctive relief. Judge Duplantier then stayed all injunctions pending appeal, and as a condition for this stay, enjoined MSP from distributing dividends to its shareholders. Judge Duplantier entered a second Rule 54(b) judgment incorporating all matters decided at the later phase of the litigation. In this case, number 94-30664, we address issues arising from this second Rule 54(b) judgment.

Because of the complexity of the issues involved, we detail the facts corresponding to each district court ruling with the relevant legal discussion. We begin with CWA issues, continue with RCRA questions, and then consider CAA disputes. We conclude with a discussion of the district court's injunctions.

II

The district court fined MSP for two types of CWA violations, thermal pollution and stormwater discharges. MSP appeals both fines. MSP concedes that it violated the CWA; it argues only that the fine was too high.

A

MSP used water to cool the material produced from its kiln. MSP pumped the water from Bayou Boeuf through a series of pipes, and the water absorbed the heat through the pipes without coming into direct physical contact with MSP's materials. Water used in this manner is called "non-contact cooling water." Since opening operations in 1985, MSP has discharged non-contact cooling water heated to temperatures at times exceeding 100 degrees Fahrenheit back into the Bayou Boeuf. The alternative to this type of heat discharge system is the construction of an expensive form of heat removal, such as a system of cooling towers.

Shortly after opening, MSP first applied to EPA Region VI for a National Pollutant Discharge Elimination System permit. 33 U.S.C. § 1342(a). This application did not mention MSP's need for non-contact cooling water. In July of 1986, MSP received an NPDES permit that did not include allowances for non-contact cooling water. MSP continued its discharge of heat into the Bayou Boeuf.

On February 25, 1987, MSP applied for a revision in its NPDES permit to allow it to discharge hot water into the Bayou. In response to MSP's application, Region VI issued a series of three draft permits, each allowing MSP to discharge non-contact cooling water at temperatures below 100 degrees Fahrenheit. In July of 1991, Region VI denied MSP's request for a revision to its NPDES permit and announced that it would terminate the original 1986 permit on the ground that MSP had intentionally included misinformation in its permit application. Two months later, the Environmental Appeals Board reversed, holding that Region VI could not terminate MSP's original permit or deny its request for revisions on misinformation grounds without granting MPS a hearing. Marine Shale Processors, Inc., NPDES Appeal No. 91-22 (EPA Sep. 12, 1994). Region VI has not yet scheduled a hearing regarding MSP's NPDES permit.

After the bench trial on the Clean Water Act issues, the district court found the following facts:

MSP has discharged large quantities of non-contact cooling water containing heat (by law a pollutant) into Bayou Boeuf almost daily since it began operating, all without a permit . . . . Despite the frequency and duration of MSP's unpermitted discharges, there is little, if any, evidence of actual damage to Bayou Boeuf. This is especially so with respect to the discharge of the non-contact cooling water.

Although the unpermitted discharges did not have a significant impact on the environment, these water violations are serious. MSP's actions in discharging pollutants into Bayou Boeuf have been willful and flagrant . . . . [T]here is not doubt that MSP knew it needed an NPDES permit . . . and simply decided to operate without [one]. MSP not only discharged non-contact cooling water knowing it did not have a permit to do so, but in order to increase capacity and thus maximize profit, substantially increased the amount of such water that it was discharging.

Such discharges demonstrate a callous disregard for the regulatory scheme and the purposes of the Clean Water Act. The court cannot countenance such activity. No one is above the law. MSP's water violations were profit driven, and for the most part were purposeful, not accidental.

I consider as a mitigating circumstance the fact that since February 1987, MSP has attempted to obtain a modified NPDES permit, but note that the request for modification did not come until more than a year after MSP began operations without the necessary permit.

MSP's operation also resulted in a number of stormwater discharges which exceeded the limits specified in its NPDES permit for oil, grease, and chemical oxygen demand. On appeal, MSP concedes that four of 430 of its oil and grease readings and thirty-three of its 431 chemical oxygen demand measurements, during the period from 1988 to 1992, were higher than its permit limits. The district court considered MSP's thermal pollution together with its unlawful stormwater discharges in levying a $3,000,000 fine for all Clean Water Act violations.

B

MSP attacks the portion of the stormwater discharge fine on several grounds. First, MSP argues that the district court insufficiently reduced the penalty in light of EPA's seven-year delay in ruling on MSP's NPDES permit amendment application. Second, MSP urges that the district court erroneously excluded evidence of measurement error in the process used to assess stormwater discharges; MSP contends that two-thirds or more of its stormwater exceedences were within measurement error range of its NPDES permit limits. Third, MSP draws our attention to the fact that all of the exceedences were "first flush" readings, readings taken at the beginning of a storm event where pollutant levels are highest, and suggests that in 1992 EPA amended its stormwater regulations so that compliance is now measured according to readings taken over the entire storm event. Fourth, MSP points out that the Office of Water, United States Environmental Protection Agency Training Manual for NPDES Permit Writer EPA 833-B-93-003, Section(s) 3.3.1 (1993), states that "[I]n any single monitoring observation, a discharger running a properly operated and maintained treatment facility has a 95 to 99 percent chance of complying with its permit limits," and that its compliance for oil and grease was within this range. Fifth, MSP contests the fine on the ground that the district court did not differentiate and define the number or severity of the various violations.

Regarding thermal pollution, MSP contends that the $3,000,000 penalty was too high because the district court based its fine in part on a clearly erroneous finding of fact, namely, that the increase in MSP's discharges of non-contact cooling water were profit-driven. MSP points to evidence in the record supporting a different motivation for its increase in hot water discharges.

C

When considering MSP's argument that the district court fine was too harsh, we begin by noting that the court had statutory authority to impose a fine of around $45,000,000 for CWA violations. The $3,000,000 fine represents less than eight percent of what the court might have imposed. We note that the evidence amply supports the district court's finding that MSP knew it needed a permit for its thermal discharges and simply decided to operate without one for several years. The record also discloses that the district court, at least initially, found that MSP violated the CWA in a third way by allowing rainwater and waves to wash into and out of barges on the Bayou that MSP used to hold hazardous waste, but that the district court apparently later decided not to fine MSP for these discharges. Finally, we note that when imposing penalties under the environmental laws, courts often begin by calculating the maximum possible penalty, then reducing that penalty only if mitigating circumstances are found to exist. See, e.g., Atlantic States Legal Foundation, Inc. v. Tyson Foods, Inc., 897 F.2d 1128, 1142 (11th Cir. 1990). Under such circumstances, we suggest that MSP may act at its own peril in seeking to upset this fine. See United States v. B W Investment Properties, 38 F.3d 362, 368 (7th Cir. 1994) ("In considering fines under the [Clean Air] Act, courts generally presume that the maximum penalty should be imposed."), cert. denied, 115 S.Ct. 1998 (1995).

We find unconvincing MSP's contention that the district court insufficiently weighed EPA's delay in ruling on the NPDES permit amendment application when imposing the fine. MSP concedes that the district court considered this mitigating factor in its analysis. We note also that MSP never sought the aid of the federal judiciary in compelling EPA to act more expeditiously "despite the clear availability of this remedy." United States v. General Motors Corp., 876 F.2d 1060, 1068 (1st Cir. 1989), aff'd, 496 U.S. 530 (1990); see, e.g., Ingalls Shipbuilding v. Asbestos Health Claimants, 17 F.3d 130, 132-34 n. 9 (5th Cir. 1994) (considering remedies for undue agency delay); Atlantic Gulf Stevedores, Inc. v. Donovan, 274 F.2d 794, 802 (5th Cir. 1960); Telecommunications Research Action Center v. FCC, 750 F.2d 70, 79-81 (D.C. Cir. 1984). "There was a simple and straightforward way for [MSP] to avoid paying civil penalties for violations of the Clean Water Act: After purchasing the plant, [MSP] could have ceased operations until it was able to discharge pollutants without violating the requirements of its NPDES permit." Atlantic States, 897 F.2d at 1141-42.

With these concepts in mind, we reject MSP's argument that the district court insufficiently mitigated its fine in consideration of EPA's delay in ruling on MSP's NPDES permit amendment application. In General Motors Corp. v. United States, 496 U.S. 530, 541 n. 4 (1990), the Supreme Court held that in the CAA context courts should respond to EPA's undue agency delay by reducing penalties in an enforcement action in order to counteract any incentive the agency might have to place itself in a superior litigating position. The district court considered EPA's delay in its calculations and reduced the fine accordingly. We find no abuse of discretion.

Next, we reject MSP's argument that the district court abused its discretion by relying too heavily on MSP's stormwater discharge exceedences in calculating its fine. Initially, we note that MSP's statement that the district court excluded evidence of measurement error contradicts the record, which shows that the district court heard evidence from MSP expert John Wagner on this matter. While MSP correctly pointed out that many of its stormwater exceedence levels were within the range of measurement error of its NPDES permit, the district court correctly responded that measurement error is "plus or minus," and thus that many of MSP's discharge samples reading below permit limits could have been exceedences. In addition, we find unpersuasive MSP's reliance on the fact that all of its exceedences were first flush samples, i.e. samples taken at the beginning of a storm when pollutant readings are highest, when EPA's current regulations may rely on readings from the entire storm event to determine if an exceedence has occurred. MSP admits that the regulations in force at the time of its discharges rendered illegal any stormwater discharge in excess of permit limits, even first flush samples. It does not argue that these regulations were invalid. Under these circumstances, the district court did not abuse its discretion by relying on MSP's stormwater exceedences.

For similar reasons, we reject MSP's reliance on EPA publications stating that a properly operated facility should achieve a 95-99% rate of compliance with its NPDES permit limits. MSP does not suggest that EPA's enforcement authority in some way turns upon this figure. Moreover, MSP did not achieve this compliance figure.

We disagree with MSP that the district court's findings of fact and conclusions of law were insufficiently specific to support the fine. The district court's findings were not as detailed as some. See, e.g., Hawaii's Thousand Friends v. City County of Honolulu, 821 F. Supp. 1368, 1394-97 (D. Haw. 1993). Nevertheless, calculation of discretionary penalties is not an exact science, and few courts could comply with MSP's request that the importance of each factor be precisely delineated. See Tull v. United States, 481 U.S. 412, 427 (1987) (observing that "highly discretionary calculations that take into account multiple factors are necessary in order to set civil penalties under the Clean Water Act"); see also B W Investment Properties, 38 F.3d at 368.

Finally, we find some merit in MSP's attack upon the district court's findings of fact, and as a result, we reverse the damage award and remand for recalculation. MSP argues that the district court's penalty calculations were based on clearly erroneous findings of fact. In particular, MSP attacks the district court's conclusions that "MSP's water violations were profit driven, and for the most part were purposeful, not accidental," and that MSP, "in order to increase capacity and thus maximize profit, substantially increased the amount of such water that it was discharging."

Regarding the first finding, we find no clear error. The district court could infer that MSP's water emissions were purposeful and profit-driven from the substantial evidence before it that MSP knew it needed an NPDES permit for its thermal discharges but decided to operate without one. We note that the district court made an explicit finding, a finding the evidence supports, that MSP decided to begin operations knowing that it needed an NPDES permit for hot water discharges that it did not possess.

Regarding the second finding, we agree with MSP. MSP cites to evidence in the record supporting its contention that the increase in hot water discharges resulted from technological improvements in its overall system, and not from a desire to maximize profit by increasing the amount of waste processed. On August 23, 1989, MSP informed LDEQ's Office of Water Resources by letter that it was replacing two of its oxidizers with more efficient machinery and that as a result the volume, although not the temperature, of the hot water discharged would increase. The August 23 letter reminded LDEQ that MSP had previously provided notice of this change in a letter to the Air Quality Division of LDEQ dated April 24, 1989. MSP argues that this evidence shows that MSP's motive for increasing its discharge of heat was a desire to make its operation more efficient. Such self-serving statements are not overwhelming evidence. Nevertheless, the United States cites to no evidence in the record to the contrary, and indeed, does not address this point at all in its brief. Our independent review of the record, which is admittedly large and difficult to decipher, indicates nothing to support the district court's ruling. We note that in the pretrial order at the RCRA stage, the United States announced its intention to prove that MSP had added oxidizers and slag boxes to its facility, and there is some indication in the record that this addition may have increased MSP's capacity to process hazardous waste. Nevertheless, especially with no help whatsoever from appellees, we are unable to connect these additions to MSP's increased discharge of hot water. Under such circumstances, we conclude that the district court's characterization of MSP's motive for increase the volume of its thermal pollution is clearly erroneous.

We hesitate to vacate such a large fine on the basis of such insignificant error. The district court did not, however, differentiate what portion of the fine resulted from each type and quantity of violation. While this failure to differentiate is not in itself reversible error, it does render us unable to determine whether the error was harmless or to dispose of this case by reducing the amount of the fine on our own. Accordingly, we vacate the entirety of the fine and remand to the district court for recalculation.

III

The district court decided all RCRA issues relevant to this appeal by summary judgment. In particular, the district court held that MSP had violated land ban regulations on numerous occasions. In addition, the district court held that MSP had stored K-listed wastes without a permit or interim status, but ruled that MSP had interim status to store F-listed wastes. Both the United States and MSP appeal the district court's RCRA rulings. We consider the land ban violations before discussing listed wastes.

A 1

RCRA land ban regulations prohibit placement on the land of material leaching toxic metals in excess of regulatory limits. 40 C.F.R. pt. 268. The district court granted in part the United States' motion for summary judgment, finding that MSP had violated land ban regulations by placing on the ground materials that would leach lead and other metals in excess of regulatory limits. MSP appeals this finding, disputing the district court's holding as to method by which samples are to be collected and analyzed for testing purposes.

EPA requires facilities like MSP to test a substance's leaching capacity according to the Toxicity Characteristic Leaching Procedure. See, e.g., 40 C.F.R. Section(s) 268.7(a). The parties dispute the method by which material samples are to be collected for TCLP testing and the resulting data analyzed. As we understand them, there are at least three possible methods. In the first, the analyst collects single samples from random sections of a pile of material, a form of collection called "grab sampling;" the analyst then subjects these samples to TCLP testing and compares the raw numbers generated without a statistical analysis to the regulatory limits on leaching. In the second, the analyst collects grab samples, conducts TCLP testing upon them, but then subjects the numerical results to a statistical analysis before comparing the results to the regulatory limits on leaching. In the third method, the analyst combines samples from random sections of a particular pile of material into a single larger sample, a form of collection called "composite sampling," then subjects the composite sample to TCLP testing.

MSP processed hazardous waste through its kiln system. As the material exited the system, MSP took samples every 15 minutes. 24 to 32 samples were combined and thus, in the terminology just defined, became a single composite sample. MSP placed the exiting material on the ground in a pile called a "day pile" and sent the combined samples to an in house laboratory to subject them to TCLP testing. The TCLP testing of these composite samples took approximately 24 hours, during which time the day pile remained on the ground. MSP recorded the results of the TCLP testing on "daily aggregate control sheets." It conducted no statistical analysis of the data on the daily aggregate control sheets. In some cases, the daily aggregate control sheets showed readings exceeding land ban regulatory limits. In those instances, MSP reprocessed the entirety of the day pile from which the offending composite sample was taken through the kiln system and retested the resulting material, generating new daily aggregate control sheets.

Once the daily aggregate control sheets showed that no TCLP readings generated from the composite samples of the day piles exceeded land ban limits, MSP combined several day piles into larger "sale piles." MSP also placed these sale piles on the ground. An independent laboratory then took numerous samples of the sale piles and subjected each individually to TCLP testing. In the terminology defined above, these samples were grab samples. The independent lab subjected the TCLP results of the grab samples from the sale piles to a statistical analysis of the nature outlined in Office of Solid Waste and Emergency Response, Environmental Protection Agency, SW-846, Test Methods for Evaluating Solid Waste, at nine-14 to nine-17 (3d ed. 1986) [hereinafter SW-846].

Thus, MSP placed the material coming out of its kiln system on the ground on two separate occasions, once in the form of day piles, and again in the form of sale piles. MSP used composite samples to test the day piles and grab samples with a statistical analysis to test final sale piles. The parties assume that because RCRA regulates the placement of hazardous waste on the land regardless of the point of a treatment process at which such placement occurs, violations of RCRA could have occurred at either the interim day pile stage or the more final sale pile stage.

The district court granted summary judgment to the United States based on violations of RCRA at the day pile stage. It relied on MSP's daily aggregate control sheets to find that MSP placed material violating land ban regulations on the ground on 27 occasions. It fined MSP $500,000 for these violations of RCRA.

2

MSP argues that the district court's reliance on what it calls "single sample exceedences" constituted reversible error. It argues that because EPA documents, especially SW-846, require that compliance be determined by a statistical analysis of the result of TCLP testing of grab samples, it lacked fair notice that a violation might be proved by TCLP testing of composite samples without a statistical analysis.

3

We assume for the sake of argument that MSP is correct that EPA regulations require a statistical analysis of samples in order to demonstrate compliance with RCRA. The difficulty is that, as MSP itself recognizes in its briefs, any proposed statistical analysis was to be conducted upon the results of TCLP testing of materials collected by grab, not composite, sampling. See 40 C.F.R. Section(s) 268.41(a) (1994), superseded, Final Rule, Land Disposal Restrictions Phase II, 59 Fed. Reg. 47,982, 48,103 (1994). MSP contends that it based its compliance program on grab samples, but concedes that it took grab samples only of sale pile materials, while the district court's findings were based entirely on day pile violations, from which MSP tested only composite samples. Because MSP did not conduct TCLP testing of grab samples from its day piles before combining the grab samples into composite samples, it its now impossible to conduct any relevant statistical analysis. Thus, MSP did not comply with the regulations upon which it now claims it relied. MSP's argument boils down to the contention that its failure to follow EPA's specified procedures, which prevented the generation of the evidence normally relied on to test compliance with RCRA land ban restrictions, should immunize it from a finding of liability.

We disagree. 40 C.F.R. Section(s) 268.40 (1994), superseded, 59 Fed. Reg. at 48,046, stated that wastes like those at issue here "may be land disposed only if an extract of the waste . . . developed using . . . the [TCLP] does not exceed" the values specified in the regulations. At the times relevant to this litigation, 40 C.F.R. Section(s) 268.41(a) stated that "[c]ompliance with these concentrations is required based on grab samples." MSP placed hazardous waste on the land without following these specified testing procedures, procedures the D.C. Circuit labeled "graphically self-defining." Chemical Waste Management, Inc. v. EPA, 976 F.2d 2, 34 (D.C. Cir. 1992), cert. denied, 507 U.S. 1057 (1993).

Moreover, the district court did not presume a violation of land ban regulations because of MSP's failure to comply with applicable testing procedures. Rather, the court below reasoned as follows. The mixing inherent in composite sampling necessarily has an averaging effect and will almost always result in a reading lower than that generated by the most toxic grab sample. Therefore, TCLP tests of composite samples cannot demonstrate compliance with land ban limits in most cases, but they might well demonstrate a violation, if the composite sample's TCLP reading were high enough. MSP's own daily aggregate control sheets showed TCLP results from testing of day piles showed leaching of greater than the land ban standard for certain metals. Accordingly, the district court granted summary judgment to the United States.

Under such circumstances, we are unpersuaded by MSP's argument that it lacked fair notice that a court, because of MSP's own failure to follow graphically self-defining procedures, might focus on reliable, alternative evidence that MSP placed materials on the ground capable of leaching toxins in excess of regulatory limits. Had MSP followed the authority cited in its brief, perhaps it might not have been subject to penalties. Cf. Gates Fox Co. v. OSHRC, 790 F.2d 154, 156-57 (D.C. Cir. 1986) (overturning a penalty on fair notice grounds when an employer followed the most reasonable interpretation of a regulation and received no contrary interpretation from any authoritative source). We express no view on this question. This is not a case of a court construing a regulation "to mean what an agency intended but did not adequately express." Diamond Roofing Co. v. OSHRC, 528 F.2d 645, 649 (5th Cir. 1976). The regulations were clear, MSP did not follow them, and it cannot now be heard to complain on the ground of lack of fair notice that a court would rely on other means of proof. We find no error in the district court's fine.

B 1

Congress passed the current version of RCRA in 1976. 42 U.S.C. § 6925(a) directed EPA to promulgate regulations governing the issuance of permits to store hazardous wastes on the ground. Realizing that EPA could not issue permits to all existing facilities simultaneously with the promulgation of these regulations, Congress created a grandfathering scheme granting interim status to certain facilities. 42 U.S.C. § 6925(e). In order to achieve interim status, a facility must have (1) existed at the time it was rendered subject to a storage regulation, (2) filed a hazardous waste notification form and, (3) filed a permit application with either EPA or the relevant state regulatory body. 42 U.S.C. §(s) 6925(e)(1), 6930(a). RCRA treated facilities that had taken these three steps as though they had been issued permits until EPA had finally resolved the pending permit application.

In 1980, EPA promulgated regulations requiring facilities to obtain permits covering storage before storing F-or K-listed wastes, but not for certain other types of waste. Final Rule, Interim Final Rule, and Request for Comments, Hazardous Waste Management System: Identification and Listing of Hazardous Waste, 49 Fed. Reg. 33,082, 33,123 (1980). At this same time, EPA promulgated the "Mixture Rule," which defined a hazardous waste subject to regulation under RCRA to include "a mixture of solid waste and one or more hazardous wastes listed in Subpart D . . . not excluded from this paragraph . . . ." 45 Fed. Reg. at 33,119. The D.C. Circuit declared the Mixture Rule void ab initio in Shell Oil v. EPA, 950 F.2d 741 (D.C. Cir. 1991), on grounds of inadequate notice. EPA repromulgated the Mixture Rule regulations in 1992. Final Rule, Hazardous Waste Management System; Definition of Hazardous Waste; "Mixture" and "Derived-From" Rules, 57 Fed. Reg. 49,278 (1992).

In 1984, EPA certified LDEQ's state hazardous waste program, allowing LDEQ to regulate the storage of hazardous waste on its own and placing primary responsibility for RCRA enforcement with LDEQ. See Approval of State Hazardous Waste Program, Hazardous Waste Program, Louisiana, 49 Fed. Reg. 2893 (1984); see 42 U.S.C. Section(s) 6926(b). MSP asserts, and EPA does not dispute, that Louisiana's regulations initially allowed recycling facilities to store wastes other than those included in the K- and F-listings without permits. MSP began operations in 1985 without a RCRA storage permit.

Shortly after opening, MSP began accepting D- and U-listed wastes. It also began receiving material manifested as "K001" from two customers, Colfax Creosoteing Co. and Durawood Treating Co. The manifest "K001" refers to the first of the wastes specified as K-listed wastes in 40 C.F.R. Section(s) 260.10. Colfax and Durawood had hired MSP to clean up large wastewater treatment ponds. These ponds contained primarily water, creosote, and pentachlorophenol. Also present were trace amounts of chrome, copper, arsenic, along with unspecified quantities of debris. MSP pumped the water from these ponds and removed it to MSP's rotary kiln site. MSP then added absorbent materials, composed in part of material previously generated from MSP's rotary kiln, to solidify what remained and removed the entirety of the pond site material by bulldozer. Materials arriving at MSP's rotary kiln site from Colfax and Durawood remained there partially on a cement pad.

MSP's treatment of the Colfax and Durawood materials led the United States four years later to file an information alleging that MSP "did knowingly store and cause to be stored hazardous wastes identified or listed pursuant to Title 42, United States Code, Section 6921, namely, bottom sediment sludge from the treatment of wastewaters from wood preserving facilities using creosote and/or pentachlorophenol." The United State's indictment covered MSP's activities only in 1985. MSP pled guilty to this charge.

Sometime in 1985, LDEQ and MSP entered into discussions concerning whether MSP's activity constituted "storage" under the relevant regulations. In these discussion, LDEQ suggested that MSP apply for a storage permit under forthcoming regulations requiring facilities to obtain a permit to store all listed wastes, as opposed to just K- and F-listed wastes. MSP and EPA agree that on January 1, 1986, LDEQ promulgated these storage regulations, which required all facilities to obtain RCRA permits before storing any type of listed waste. In response, MSP submitted to LDEQ a notification form and Part I of a permit application to store U- and D-listed wastes in early January, 1986. On January 31, 1986, MSP amended its application to include a request for permission to store F- and K-listed wastes. Two months later, MSP began accepting F-listed wastes. On June 9, 1986, LDEQ wrote MSP a letter stating its view that MSP had obtained interim status and could store any hazardous waste listed in its application form until LDEQ ruled on the application. LDEQ's letter included the following language:

This letter confirms that the rotary kiln operated by Marine Shale Processors, Inc. . . . has interim status as a storage facility and may receive any hazardous waste for storage prior to reuse and/or recycling if such waste has been listed in the facility's Part I Permit Application or in any subsequent approved revisions to that application.

In this action, the United States alleged that MSP stored K- and F-listed wastes without a permit or interim status. Initially, the district court granted summary judgment to the United States, holding that MSP's activities did constitute storage of hazardous waste. It further held that because MSP was not in existence in 1980 or in 1984, when EPA initially required recycling facilities for storage permits and when EPA certified LDEQ's identical regulations, MSP did not have interim status to store K- and F-listed wastes in 1985, when MSP began to receiving the materials manifested K001. The court rejected MSP's arguments that the materials manifested K001 received from the Colfax and Durawood operations were not pure K-listed wastes but were rather mixtures of K-listed and other wastes covered only by the Mixture Rule, commenting that "the interpretations urged by MSP would produce the ridiculous result that one could receive a listed hazardous waste, add a drop of water to it, and store that waste without having either a storage permit or interim status." The court further stated, "If these wastes fail to qualify as K001 wastes, I cannot envision what wastes would." The district court also rejected MSP's collateral estoppel defense, reasoning that interim status could be granted only by statute, not by regulatory agencies, and that estoppel does not lie against the government. In imposing a penalty, the court counted as one violation each day upon which MSP received either a K-or F-listed waste and counted as two violations those days upon which MSP received both types of waste. See 42 U.S.C. §(s) 6928(g). In passing, the district court found that even after EPA repromulgated the Mixture Rule in 1992, MSP stored F-listed wastes on 185 occasions and K-listed wastes on 49 occasions. The court's minute entry also found that MSP had stored K-listed wastes on 107 occasions before receiving LDEQ's June 9 letter.

Shortly thereafter, the district court sua sponte reversed its decision regarding F-listed wastes and granted summary judgment to MSP on that issue. In this second ruling, the court stated that MSP's January, 1986 notice form and application gave it interim status to store D- and U-listed wastes. The court further reasoned that the RCRA regulations allowed MSP to piggyback interim status to store F-listed wastes upon this previously achieved D- and U-listed interim status by amending its permit application to include F-listed wastes. The court, however, refused to modify its ruling that MSP had no interim status to store K-listed wastes, reasoning that the piggybacking theory did "not apply where, as here, a facility stores a regulated waste at a time when it lacks interim status to store any waste." The court then calculated the number of violations for K-listed wastes alone. Apparently on double jeopardy grounds, the court excluded violations occurring in 1985, the period covered by MSP's prior guilty plea. It then found that MSP had stored K-listed wastes on 354 occasions.

The court concluded by readdressing the issue of whether MSP's activities constituted "storage." It found no evidence that MSP either fed the K-listed wastes directly into its kiln or kept the wastes in a holding container for brief period of time before placement into the kiln. In spite of this finding, the district court went on to comment that "if there were such evidence, those days would not be counted as violations" because "storage" under RCRA excluded "a reasonable period of time between the unloading of the wastes and their placement in the kiln."

In a later order, the district court fined MSP $1,000,000 for storing K-listed wastes without a permit. In assessing this fine, the court considered as a mitigating factor LDEQ's communications to MSP in 1986 and thereafter, finding,

On June 9, 1986, LDEQ represented to MSP in writing that the facility was an interim status storage facility. MSP's reliance on that representation was reasonable. Moreover, because LDEQ never disavowed its previous confirmation, MSP's reliance on LDEQ's representation continued to be reasonable even after LDEQ expresses concern about whether MSP had interim status as a storage facility.

2

MSP urges that the district court erred in reading the ban on the storage of K-listed wastes to cover the Colfax and Durawood materials, and that in fact these materials were waste mixtures covered only by the Mixture Rule regulations declared invalid in Shell Oil. MSP also argues that the court should have accepted its estoppel defense.

The United States, in its cross appeal, claims that the district court erroneously granted MSP summary judgment on the question of illegal storage of F-listed waste. The United States argues that RCRA prevents a facility from obtaining interim status once it has illegally stored hazardous waste. Accordingly, the district court should have granted summary judgment to the United States on its claim that MSP stored F-listed waste without interim status or a permit. MSP admits, the argument runs, that it had no permit when it began to receive the K001 manifested materials; its guilty plea establishes the status of these materials as K-listed wastes, and in any event these materials were K-listed wastes as a matter of law. Accordingly, MSP could not, by amending its permit application, piggyback interim status to store F-listed wastes onto its interim status for D- and U-listed wastes because it lacked such status in the first place. Finally, the United States, with support from the State of Louisiana, claims on cross-appeal that the district court erroneously interpreted the word "storage" to include a reasonable amount of time between the unloading of the waste and its placement in processing machinery.

3

We organize our discussion of RCRA issues in four steps. We begin with MSP's Mixture Rule defense before considering the interim status dispute. We then discuss MSP's estoppel defense, and conclude with the storage issue. On the merits, we hold that both of MSP's arguments lack merit, that the district court erroneously interpreted RCRA in holding that MSP had interim status to store F-listed wastes, and that the current appeal provides us with no reason to reach the storage issue.

a

We reject MSP's contention that the materials manifested K001 were waste mixtures subject to regulation only under the Mixture Rule invalidated in Shell Oil. MSP contends that the material ultimately stored at MSP as a result of the Colfax and Durawood cleanups included soil, debris, creosote, copper, chrome, arsenic, wastewater, and the absorbent material it added to the bottom of the pond before bulldozing. Accordingly, MSP argues, these materials contained matter not included in K001 definition, and thus constituted matter subject to regulation only pursuant to the Mixture Rule.

Excepting the absorbent material, all other results of the Colfax and Durawood cleanup operations easily meet the definition of a K-listed waste. A K001 waste is a "[b]ottom sediment sludge from the treatment of wastewaters from wood preserving processes that use creosote and/or pentachlorophenol." 40 C.F.R. Section(s) 261.32. The evidence established that the materials in the pond came from the treatment of wastewaters from wood preserving processes that used creosote. For instance, Clyde M. Norton, the Vice-President of the corporation that owned Colfax and Durawood, stated in an affidavit that the materials originally in the pond before MSP began its cleanup had been generated from the two companies' wood treating operations.

MSP's primary argument is that extraneous matter in the materials it received prevented those materials from falling within the regulatory definition of a "sludge." We do not agree. The regulations define a "sludge" as "any solid, semi-solid, or liquid waste generated from a[n] . . . industrial wastewater treatment plant." 40 C.F.R. Section(s) 260.10. Thus, the definition of a sludge, like the definition of a K001 waste, focuses primarily on the origin of the material at issue, not, as MSP contends, on its composition. Moreover, a sludge is a waste generated from a wastewater plant, not as MSP contends, from a wastewater operation. "The word 'plant' denotes an entire facility, a collection of units, machines, land, buildings, and fixtures used in a trade or business, not a single intermediate unit in the treatment process." In re Brown Wood Preserving Co., No. RCRA-84-16-R, 1989 WL 253215, at * 6 (EPA May 3, 1989). MSP's own evidence established that the entirety of the material, except for the absorbent material MSP added to the bottom of the ponds after removing the wastewater itself, came from the industrial wastewater treatment plants located on the Colfax and Durawood sites.

We have more difficulty characterizing the results of MSP's bulldozing, after it had added absorbent materials to the Colfax and Durawood wastewater ponds, but we ultimately agree with the district court that the addition of these absorbent material did not cause the Colfax and Durawood wastes to lose their K-listed character. As the district court noted, MSP's interpretation of the definition of K001 waste leads to absurd results. Under its interpretation, MSP could have transformed the Colfax and Durawood materials into mixtures, regulable only by the Mixture Rule invalidated in Shell Oil, by adding a drop of water or a speck of dust to every barrel of waste it received, so long as the drop or speck did not come from a wastewater treatment facility. Indeed, although MSP repeatedly contends that the K001 listing applies only to "pure" substances of the nature described in that regulation, such purity exists only in theory. Rudimentary chemical principles establish that a liquid absorbs gases from the surrounding air and trace amounts of impurities from the container in which the liquid resides. Thus, had MSP added nothing at all to the soils it bulldozed from the Colfax and Durawood sites, the resulting material would still not have been pure K001 waste within MSP's use of the phrase. Instead, the arriving material would have consisted of a mixture of K001 waste, dissolved gas molecules from the air, trace amounts of whatever metal or ceramic or synthetic housed the waste in transit, and impurities stuck to the inside of the container. MSP's brand of purity exists only in the hypotheticals of chemistry classrooms, and its interpretation of the regulations would render them meaningless.

We hold that a substance does not lose its character as a K-listed waste, and thus does not become regulable only by the previously invalid Mixture Rule, unless the materials added to it change its basic composition in some significant way. We draw support from the D.C. Circuit's decision in Chemical Waste Management, Inc. v. Environmental Protection Agency, 869 F.2d 1526, 1539 (D.C. Cir. 1989). In upholding EPA's "contained-in" policy, the D.C. Circuit rejected the argument that "an agglomeration of soil and hazardous waste is to be regarded as a new and distinct substance" and instead accepted EPA's position that "hazardous waste cannot be presumed to change character when it is combined with an environmental medium." 869 F.2d at 1539. Under the circumstances of this case, we decline MSP's invitation to hold that the addition of an absorbing agent or other inert debris to a K001 waste transforms the waste into a new and distinct substance regulable only through the Mixture Rule. We hold that a K001 waste remains a K001 waste after the addition of a substance that results in no significant change in composition.

We need not specify exactly where the line between a significant and insignificant alteration lies. The addition of the absorbent agent did not cause a significant alternation of the Colfax and Durawood materials for several reasons. First, from the standpoint of their toxic composition, these materials were in the heart of the definition of the K001 listing. As the district court observed, "[i]f these wastes fail to qualify as K001 wastes, I cannot envision what wastes would." Second, evidence in the record strongly suggested that at the time these wastes were being shipped, those in the industry considered them K001 sludges. Huey Stockstill, the MSP officer in charge of the Colfax and Durawood cleanup, repeatedly characterized the material brought to MSP's kiln site as a sludge. The materials were manifested as K001 wastes. The contract between Colfax and Marine Shale described the wastes as "creosote waste that has been generated during wood treating operations at the Colfax wood preserving facility." Third, MSP pled guilty to storing K-listed waste upon its premises without a permit as a result of charges focusing on its storage of the Colfax and Durawood materials, suggesting that it too thought these materials constituted K001 waste, although in the face of litigation it has changed its position. MSP argues that its guilty plea was also based on the Mixture Rule, but the charges to which MSP pled recite the definition of a K001 waste without mentioning this rule. Fourth, the addition of the absorbent was entirely incidental to a cleanup operation, and thus resembles the impurities a waste might absorb from its container during transport and storage.

We reject MSP's contention that our decision places us in conflict with the Seventh Circuit's holding in United States v. Bethlehem Steel Corp., 38 F.3d 862, 865, 868-71 (7th Cir. 1994). In Bethlehem Steel, the defendant had mixed an F006 waste with "other kinds of wastewater," 38 F.3d at 865, before the addition of a thickener allowed a sludge to precipitate to the bottom. The addition of these other wastewaters so changed the basic composition of the substance at issue that EPA resorted to arguments found unpersuasive in Shell Oil in an attempt to place the wastewaters within the F006 listing. We find no conflict between our holding and that of Bethlehem Steel.

b

We hold that a facility may not achieve interim status under RCRA if it has illegally stored listed waste without a permit prior to the time it seeks to achieve interim status. Such a facility was not in existence at the time it was required to have a permit, and the facility has rendered itself subject to the permit requirement. Accordingly, we affirm the district court's K-listed waste ruling and reverse its decision that MSP possessed interim status to store F-listed waste.

The interim status dispute in this case centers on the following statutory language:

Any person who owns or operates a facility required to have a permit under this section which facility . . . is in existence on the effective date of statutory or regulatory changes under this chapter that render the facility subject to the requirement to have a permit . . . shall be treated as having been issued a permit until such time as final administrative disposition of [the permit] application has been made.

42 U.S.C. §(s) 6925(e).

Three concepts from this portion of the statute resolve the case before us. First, section 6925(e) refers to "a permit." Under RCRA, EPA issues a particular facility one permit only. If a facility treats, stores, and disposes of hazardous waste, a single permit covers all of these activities. If it engages in any of these activities with respect to more than one type of waste, a single permit covers all wastes specified in that permit.

Second, section 6925(e) grants interim status to persons operating a "facility." As we will explain, the district court's holding implies that the statute grants interim status on a wastestream by wastestream basis, but the statute's plain language contemplates a grant or denial of interim status on a facility by facility basis.

Third, section 6925(e) focuses on whether a facility was in existence at the time it was "render[ed] . . . subject" to the statutory requirement that it obtain a permit. The crucial point in time under RCRA is the moment at which the law required the facility to have a permit. Section 6925(e) grants interim status only to facilities that were "in existence" at this moment.

With these three concepts firmly in mind, we conclude that MSP's storage of K-listed wastes rendered it unable to achieve interim status to store any type of waste. In 1980, EPA required that all facilities, including recycling facilities, have RCRA permits before storing K-listed waste. When Louisiana took over the administration of its own RCRA program in 1984, it also required facilities storing K-listed waste to obtain a RCRA permit. In 1985, MSP stored the K-listed wastes from the Colfax and Durawood cleanup operations. MSP had no RCRA permit at this time. It could not obtain interim status because it was not in existence in either 1980 or 1984, the promulgation dates for the regulations requiring a permit for the storage of K-listed wastes, and because it had no pending permit application. In 1985, therefore, MSP lacked interim status.

An alternative application of the statute to these facts yields an identical result. The section 6925(e) exception to the permit requirement applies only when "statutory or regulatory changes . . . render the facility subject" to the necessity that the facility obtain a permit. In this case, statutory and regulatory changes did not render MSP subject to the permit requirement; rather, MSP rendered itself subject to this requirement by storing a listed waste. Either way, because the section 6925(e) exception does not apply, MSP needed a permit to store waste, and its failure to procure one prior to its storage activity resulted in a RCRA violation.

Our difficulty with the district court's holding lies in its assumption that MSP could achieve interim status by applying for a permit to store F-listed waste when it became subject to regulations corresponding to that type of waste. This holding presumes that MSP was rendered subject to the requirement that it obtain a permit in 1986. In essence, the district court held that RCRA operates on a wastestream by wastestream basis. But MSP needed a permit to cover its storage of K-listed waste several months before. MSP did not need one permit to store K-listed wastes and a second to store F-listed wastes. RCRA contemplates that a facility will receive a single permit to cover storage of all types of waste, and that this permit will govern the storage at the entire facility. RCRA permitting does not operate on a wastestream by wastestream basis.

The district court rejected the position we adopt here on the grounds that "[i]t is unreasonable to suggest that an operator who has once violated a regulation is thereafter precluded from ever lawfully operating. One could argue with equal force that any permittee who violates an environmental regulation would as a result lose his permit . . ." We do not believe that our holding implies that any operator once violating a RCRA regulation may never lawfully operate. Such a violator may operate after it has applied for and obtained a RCRA permit.

We note that our holding does not conflict with the district court's construction of La. Haz. W. Reg. Section(s) 23.2, superseded, counterpart codified at L.A.C. Section(s) 33:V:4303.A.1, which allows a facility to piggyback interim status to store a second type of waste onto its interim status to store a first by amending its pending permit application to cover the second waste. See 40 C.F.R. Section(s) 270.72(a)(1) (suggesting that EPA allows use of the piggybacking theory as well). MSP argues that it could piggyback interim status to store K- and F-listed wastes onto its status to store D- and U-listed wastes. The piggybacking theory can apply, however, only if a facility has achieved interim status in the first place. See 40 C.F.R. Section(s) 270.72(a) ("[T]he owner or operator of an interim status facility may make the following changes . . . ."). MSP did not have interim status to store any type of hazardous waste when in 1985 it rendered itself subject to the requirement that it obtain a permit, and thus it could not take advantage of the piggybacking theory. We reject MSP's attempt to confuse the issues of when a facility needed and attained interim status with what wastes it was entitled to store.

c

MSP's estoppel defense is no more availing. We agree with the district court that equitable estoppel will not lie against the United States under the facts of this case. We hold that MSP's reliance on LDEQ's June, 1986 letter was not reasonable and therefore that it has failed to prove a traditional element of the estoppel defense. We also agree with the United States that MSP has not shown that LDEQ's letter constituted affirmative misconduct.

The district court denied MSP's estoppel defense on the grounds that Congress, not EPA or LDEQ, grants interim status. See State of New Mexico v. Watkins, 969 F.2d 1122, 1130 (D.C. Cir. 1992). Although the district court's reasoning contains a valuable insight, its conclusion is not technically correct. Equitable estoppel does not rest on the grounds that the claimant is in reality entitled to the benefit or status in question. Rather, equitable estoppel responds to the unfairness inherent in denying the claimant some benefit after it has reasonably relied on the misrepresentations of the adverse party. Thus, at least in its estoppel defense, MSP does not contend that it in reality was entitled to interim status, and LDEQ's June 9 letter may have been no defense had the issue of interim status arisen in a citizen suit under 42 U.S.C. §(s) 6972. Rather, MSP claims that the United States should not be allowed to deny that MSP has interim status because MSP reasonably relied on the representations of government agents.

Nevertheless, we agree with the district court's refusal to estop the government in this situation. Recently, the Supreme Court cast further doubt on the proposition that equitable estoppel runs against the United States. In Office of Personnel Management v. Richmond, 496 U.S. 414 (1990), the Supreme Court emphasized the separation of powers difficulties inherent in an estoppel of the United States. It noted, for instance, that while Congress may create a remedy for a federal employees' issuance of erroneous legal advice, "[j]udicial adoption of estoppel based on agency misinformation would, on the other hand, vest authority in these agents that Congress would be powerless to constrain." 496 U.S. at 429. Were courts to estop the United States readily, the executive branch could use this doctrine strategically to achieve results Congress intended to prevent, thus delivering lawmaking power to the executive in a manner that the first sentence of Article I does not contemplate. Although the Court made its observations in the context of the Appropriations Clause, the principles it articulated are more generally applicable and have particular force when a private party seeks to avoid the force of an otherwise applicable law because of a government agent's representation that the law does not apply. Since Richmond, the circuits have repeatedly acknowledged the importance of separation of powers principles to claims of estoppel against the government. See, e.g., FDIC v. Hulsey, 22 F.3d 1472, 1489 (10th Cir. 1994) (stating that courts should allow an estoppel, if at all, only when such a ruling "would not frustrate the purpose of the statutes expressing the will of Congress"); United States v. Guy, 978 F.2d 934, 938 (6th Cir. 1992); Kennedy v. United States, 965 F.2d 413, 419-21 (7th Cir. 1992); Transohio Savings Bank v. Director, 967 F.2d 598, 622-23 (D.C. Cir. 1992).

In addition, estoppel of the United States implicates the President's power and duty under the Take Care Clause. When a court refuses to enforce the law on the basis of a previous representation from a government official, it renders the current executive unable to enforce the law and thus discharge its responsibilities under the Take Care Clause. Although courts rarely trace this responsibility to its constitutional roots, several cases have articulated this interest and given it substantial weight. See, e.g., Heckler v. Community Health Services, 467 U.S. 60 (1984) ("When the Government is unable to enforce the law because the conduct of its agents has given rise to an estoppel, the interest of the citizenry as a whole in obedience to the rule of law is undermined."); Hulsey, 22 F.3d at 1489 (stating that courts should refuse to apply the estoppel doctrine when to do so would "unduly undermine the enforcement of the public laws").

The threat to both sets of constitutional principles in this case is obvious. We recall the district court's observation that Congress, not EPA or LDEQ, grants interim status. Outside of the context of a citizen suit, an estoppel here would effectively allow LDEQ to grant MSP interim status in sharp tension with the principles of Articles I and II. Finally, we note the threat to the RCRA dual enforcement scheme posed by a decision allowing an estoppel in this case. Allowing state representations to estop the federal government in this case would provide the states with a mechanism for going below the federal floor of regulation required by RCRA. See, United States v. Marine Shale Processors, Inc., No. 94-30419, at 12-13.

Courts have reacted to the danger that estoppel of the United States poses to these values in several ways. First, courts have applied the elements of traditional equitable estoppel against the government rather narrowly. See, e.g., Heckler, 467 U.S. at 61-62 (holding that the estoppel claimant had shown no possibility of detriment in spite of the fact that denying its claim might force it into bankruptcy). Second, courts have insisted that any estoppel against the government result from a representation of an official acting within the scope of her official authority, thus implying that the concept of apparent authority does not apply in the case of a government estoppel. See, e.g., United States v. Walcott, 972 F.2d 323, 325 (11th Cir. 1992). Third, courts have required that a party seeking to estop the United States show some sort of "affirmative misconduct," an element normally not required to estop private parties. See, e.g., Fano v. O'Neill, 806 F.2d 1262, 1265-66 (5th Cir. 1987); Cadwalder v. United States, 45 F.3d 297, 299 (9th Cir. 1995). We rely on the first and third of these principles to reject MSP's estoppel defense. MSP's reliance on LDEQ's June 9 letter was unreasonable as that term is used in the estoppel doctrine, and in addition, MSP has not met its burden to show affirmative misconduct.

Courts considering estoppel claims against the government involving an official's misstatement that a particular statute or regulation does not apply to the claimant have read the element of reasonable reliance in light of the principle that all citizens, especially citizens dealing with the government, are presumed to know the law. See, e.g., Breath v. Cronvich, 729 F.2d 1006, 1011 (5th Cir.), modified by, 734 F.2d 225, cert. denied, 469 U.S. 934 (1984); United States Fidelity Guaranty Co. v. Bass, 619 F.2d 1057, 1077 (5th Cir. 1980). Courts have translated this reasonableness requirement into the rule that a party's reliance on a government employee's misstatement concerning the coverage or application of a law will rarely be reasonable if a clear statute or regulation provided otherwise. See, e.g., Federal Crop Insurance Corp. v. Merrill, 332 U.S. 380, 381-82, 384 (1947) (refusing to invoke estoppel to require the government to insure crops when the relevant statute "by explicit language" did not provide for insurance under the circumstances); United States v. Perez-Torres, 15 F.3d 403, 407 (5th Cir. 1994) (rejecting an estoppel argument when the terms of a statute provide in "clear and unambiguous" terms that the alleged official misrepresentation was incorrect), cert. denied, 115 S. Ct. 125 (1994); see also Guy, 978 F.2d at 937-38; Kennedy, 965 F.2d at 419-21.

42 U.S.C. §(s) 6925(e) provides clear guidance regarding interim status. In the previous section, we went no further than the plain language of the statute to decide that MSP lacked interim status once it illegally stored K-listed wastes. MSP makes no compelling argument excusing its failure to read section 6925(e) in accordance with its straightforward terminology. MSP's reliance on the June 9 LDEQ letter was unreasonable.

Even if MSP's reliance on the LDEQ letter had been reasonable, MSP has failed to show that LDEQ's June 9 letter constituted the kind of affirmative misconduct necessary to estop the United States. Although courts have been less than forthcoming in defining what a government official must do to satisfy the affirmative misconduct element of an estoppel defense, the cases support the conclusion that at minimum the official must intentionally or recklessly mislead the estoppel claimant. In REW, Enterprises, Inc. v. Premier Bank, N.A., 49 F.3d 163, 170 (5th Cir. 1995), for instance, we considered the absence of any evidence that an agent "deliberately" misled a company as a reason not to estop the government. And in Ingalls, we refused to estop the government in part because there was no allegation that an official's misstatement was made with knowledge of its falsity or with intent to mislead. 976 F.2d at 938; see also Fano, 806 F.2d at 1265 (remanding for a hearing on the issue of estoppel in part on the basis of allegations that an agency "willfully, wantonly, recklessly, and negligently" misled a citizen) (internal quotation marks omitted); Kennedy, 965 F.2d at 421 (Affirmative misconduct "is something more than mere negligence.") (internal quotation marks omitted); SIU de Puerto Rico, Caribe y Latinoamerica v. Virgin Islands Port Authority, 42 F.3d 801, 803-04 (3d Cir. 1994); Hulsey, 22 F.3d at 1490 ("[T]he erroneous advice of a government agent does not reach the level of affirmative misconduct.").

Requiring an estoppel claimant to prove that the government agent intended to mislead has a sound basis in policy and constitutional values. As we have explained, a citizen's first defense to misstatements from a government official regarding the scope and applicability of a particular law is self-help, that is, her own research to discover the applicable legal principles. An official bent upon misleading a citizen is more likely to anticipate and seek to neutralize any factors that might put the citizen on notice of the deception, thus undermining the citizen's self-help defense to official error. In addition, intentionally deception is more reprehensible than negligent error and implicates the Due Process Clause value of fair dealing between the sovereign and the citizen. See Richmond, 496 U.S. at 435-36 (Stevens, J., concurring) (discussing a hypothetical highlighting the special harshness inherent in intentionally duplicitous conduct by government officials); see also Oliver W. Holmes, The Common Law 3 (1881) "[E]ven a dog distinguishes between being stumbled over and being kicked.").

MSP cannot prove on these facts that LDEQ's June 9 representations regarding its interim status was the result of anything other than a negligent interpretation of section 6925(e). The circumstances in this case suggest simply that LDEQ made an honest mistake, a mistake that the district court considered as a mitigating factor when fining MSP. We hold that the district court correctly refused to estop the United States from disputing MSP's interim status.

MSP's citation to United States v. Pennsylvania Industrial Chemical Corp., 411 U.S. 655 (1973), is unavailing, even if this case remains good law after Richmond and Heckler. See Richmond, 496 U.S. at 426-27 (noting some dispute over whether Pennsylvania Chemical was an estoppel case). In Pennsylvania Industrial, an agency repeatedly reaffirmed in published regulations that a statute did not apply to a certain type of conduct. The defendant engaged in this conduct during the period in which these regulations remained in force. The agency then reversed its position, published new regulations providing that the statute did apply to conduct of the nature engaged in by the defendant, and sought to hold the defendant criminally liable for actions taken while the old regulations remained in force. The Supreme Court held that the defendant was entitled to an opportunity to prove that it "was affirmatively misled by the responsible administrative agency into believing that law did not apply to this situation" on the grounds that "traditional notions of fairness inherent in our system of criminal justice prevent the Government from proceeding with the prosecution." 411 U.S. at 674.

Pennsylvania Industrial does not contradict our holding in this case. The Pennsylvania Industrial defendant availed itself of its self-help remedy and discovered regulations providing that the conduct was legal, regulations to which courts would in appropriate circumstances defer. See Chevron, U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43 (1984). While a citizen might reasonably rely on the regulations of the agency charged with the administration of the statutory scheme. MSP has cited no case holding that a official's non-adjudicative statements regarding the applicability of a statute to a particular set of facts are entitled to similar deference.

We acknowledge that this case presents several factors favoring estoppel. LDEQ's representation to MSP was written, not oral. Cf. Heckler, 467 U.S. at 65 (stating that courts should be especially reluctant to estop the government of the basis of oral statements). Although the district court made no findings as to intent, the fact that LDEQ issued its letter in the context of negotiations allegedly designed to bring MSP into compliance with RCRA's storage regulations suggest that LDEQ intended for MSP to rely on the letter. Although MSP's extensive storage of K-listed wastes prior to receiving the June 9 letter suggests that it might have continued to behave in the same manner regardless of what LDEQ said, at least one of MSP's customers, Southern Wood Piedmont Company, did in fact rely on LDEQ representations regarding the status of MSP's facilities. Nevertheless, the burden on a party seeking to estop the United States is heavy indeed, and MSP has not met its burden in this case.

4

The United States appeals the district court's definition of the word "storage" as "excluding a reasonable period of time between unloading of the wastes and their placement in the kiln." The district court defined storage in an order granting summary judgment to the United States on the issue MSP's storage of K-listed wastes. Immediately preceding this definition, the district court found that "there was no evidence that the wastes were ever fed into the kiln either directly from the transport vehicle or after only a brief period of containment in an MSP tank for the purposes of conveyance to the kiln. If there had been, those days would not be counted as violations." Indeed, a September 15, 1989 Marine Shale response to an EPA demand for information suggests that hazardous wastes often spend at least ten to 15 days in MSP's blending tanks before being placed into the kiln.

In light of the district court's finding that MSP never kept K-listed waste on site for only a brief period of time, a finding that neither party disputes, we do not see how the district court's definition of storage affects any issue properly before this court. The storage issue did not affect the fine, and the parties may litigate the meaning of storage as the term appears in the injunction in a contempt proceeding. We decline to issue an advisory opinion on this matter.

IV

The United States charged MSP with operating several minor emission sources without a permit in violation of the Clean Air Act and accompanying regulations. It also alleged that MSP had exceeded the limits specified in the permit for its kiln stack on numerous occasions. The district court held for the United States regarding minor emission sources but agreed with MSP that the United States could not enforce the kiln stack permit. Both parties appeal.

A

The Clean Air Act's Prevention of Significant Deterioration Program governs the emission of air pollutants in states that have attained CAA National Ambient Air Quality Standards. The PSD portion of the CAA divides emission sources into major and minor emitting facilities. 42 U.S.C. §(s) 7479(1). The CAA and its regulations classify a source as minor if either of two different forces limit its rate of emission of air pollutants to below a specified amount: first, physical or mechanical limits constraining its rate of emission; and second, legal limits, in the form of legal restrictions on its rate of emission or hours of operation. 40 C.F.R. Section(s) 51.166(b)(4). The industry uses the term "synthetic minor source" to refer to a facility subject to this second type of limit but nevertheless possessing the physical and mechanical potential to emit above the statutorily specified rate. The CAA requires facilities constructing or modifying major sources to obtain a preconstruction permit from agencies administering EPA-approved state implementation programs. 42 U.S.C. §(s) 7475(a)(1), 7479(2)(C).

Since opening for business in 1985, MSP has operated at least one source with the mechanical potential to emit sufficient pollutants to qualify as a major source: its kiln stack. It has also operated several emission sources without such potential. When MSP originally bought the rotary kiln in 1984, LDEQ transferred to MSP the air permit previously governing the facility, Permit 722 M-1, with the stipulation that MSP add a baghouse to its kiln stack. Disputes as to the status of MSP's air emissions led LDEQ to issue a compliance order in December, 1985. MSP and LDEQ sought to resolve their differences and in early 1986 settled. As part of this settlement agreement, LDEQ issued a second air permit, Permit 1036 M-1. Permit 1036 M-1 did not include provisions addressing 29 minor sources that MSP operated at the time. In issuing Permit 1036 M-1, LDEQ did not follow the procedural requirements in its regulations governing the issuance of air permits; in particular, LDEQ did not provide public notice and an opportunity for comment regarding MSP's second permit.

In 1986, LDEQ issued a second compliance order addressing MSP's admitted operation of minor emission sources without a permit. LDEQ and MSP apparently continued discussions until April, 1988, when LDEQ ordered MSP to fill out a questionnaire requiring identification of all emission sources in its facility. MSP's response to the questionnaire identified the 29 unpermitted minor sources, from which MSP was still discharging air pollutants. Shortly thereafter, MSP filed two applications with LDEQ. The first sought to modify Permit 1036 M-1 to legalize MSP's discharges from its currently operating 29 minor sources. The second requested a variance to continue operations until LDEQ ruled on the permit amendment application. Throughout this time, MSP was still emitting from its minor sources.

LDEQ acted on MSP's variance application while requesting further information from MSP on the permit amendment application. After a public hearing, LDEQ denied the variance application on February 1, 1989. The next day, MSP requested an additional hearing on the variance denial, a request LDEQ denied.

MSP appealed the variance denial to the Court of Appeal of Louisiana. After holding that it had jurisdiction over an appeal of a variance denial, In re Marine Shale Processors, Inc., 563 So.2d 278 (La.Ct.App. 1990), the court affirmed LDEQ's denial of MSP's variance application on June 26, 1990. In re Marine Shale Processors, Inc., 566 So.2d 994 (La.Ct.App. 1990). Before the Court of Appeal of Louisiana, MSP argued that the variance denial would "result in a practical closing and/or elimination of a significant portion of its lawful business without corresponding benefit to the people of Louisiana." 566 So.2d at 996. The court rejected this argument, holding that MSP could continue operations in conformity with Permit 1036 M-1 and that the people of Louisiana would benefit from avoiding what LDEQ called "an unacceptable risk to those working and residing in the vicinity of MSP['s] facility." 566 So.2d at 998 (internal quotation marks and emphasis omitted).

MSP's arguments regarding the effect of the variance denial on its operations did not prove prophetic. It continued to operate after the ruling of the Louisiana Court of Appeals, and it continued to discharge from its 29 unpermitted minor sources.

Regarding the application to amend Permit 1036 M-1, MSP responded to LDEQ's request for additional information on several occasions and completed its application on December 7, 1989. Thirteen months later, LDEQ denied MSP's application to amend Permit 1036 M-1. MSP invoked LDEQ's appeals process, and the appellate authority remanded on the grounds that LDEQ had failed to adhere to required procedures in denying MSP's application to amend. On remand, LDEQ denied the application anew. MSP requested reconsideration of this denial. On February 11, 1992, LDEQ noted that the pending motion for reconsideration prevented the denial order from becoming final. Since that time, LDEQ has taken no action on the permit amendment application. MSP continues to operate its 29 unpermitted minor sources.

The district court held that Permit 1036 M-1 was not federally enforceable because LDEQ did not issue it in accordance with its own procedural requirements. As a corollary, it held that MSP could not rely on the pollution control devices specified in the permit to transform its kiln stack into a synthetic minor source, and thus that MSP had operated a major source without a permit. The court found that MSP had operated one major and twenty-nine minor emission sources without a permit.

The district court then articulated several factors relevant to the imposition of a civil fine:

MSP's motion for reconsideration of the denial of the air permit, which the state has notified MSP has the effect of suspending the denial of the permit, has been pending since January 15, 1992, despite my repeated statements to the Assistant Attorney General that the state should take final action on the permit application. While the state has been unreasonably dilatory in acting on MSP's motion for reconsideration, it is significant that MSP has failed to install on [its facilities] the very pollution control devices which, in its amended permit application, it proposed to install.

The court fined MSP $1,000,000 for the major source violation and $2,500,000 for the twenty-nine minor source violations.

B

MSP challenges the fine of $1,000,000 for a major source violation, arguing that while the court below correctly ruled that Permit 1036 M-1 was not federally enforceable, it erred in not allowing MSP to rely on the legal limits imposed by Permit 1036 M-1 in order to classify the kiln stack as a synthetic minor source. MSP further contends that the statute of limitations bars any penalty for the minor source violations because MSP began emitting from these sources more than five years ago. Finally, MSP argues that the district court's fine was too heavy. MSP contends that the district court's reliance on MSP's failure to install certain pollution control devices constituted reversible error because Louisiana law prohibited the installation of these devices without a permit. MSP also claims that the district court failed to place sufficient emphasis on LDEQ's delay in ruling on MSP's application to amend its permit.

In its cross appeal, the United States contends that the district court erred in holding that LDEQ's failure to follow its permit issuance guidelines rendered the federal government unable to enforce Permit 1036 M-1. In the alternative, the United States argues that if it cannot enforce Permit 1036 M-1, it can enforce the predecessor and more restrictive Permit 722 M-1.

C

We resolve the issues in the following order. First, we discuss whether the district court erred in ruling that procedural defects in the issuance of Permit 1036 M-1 rendered the United States unable to enforce the permit. Second, we consider whether the district court properly rejected MSP's statute of limitations defense to the minor source violations. Third, we address MSP's challenge to the amount of the minor source fine.

1

We reverse that portion of the district court's fine dealing with MSP's kiln stack. We hold that the United States may enforce Permit 1036 M-1. We remand for further proceedings consistent with this opinion.

The Permit 1036 M-1 portion of the dispute centers on the definition of "federally enforceable" in the CAA regulations dealing with regions that have attained CAA air quality standards. These regulations state that the United States may enforce "any permit requirements established . . . under regulations approved pursuant to 40 C.F.R. part 51, subpart I." 40 C.F.R. Section(s) 51.166(b)(17); see also 40 C.F.R. Section(s) 51.165(a)(1)(xiv) (providing an identical definition). The regulations in 40 C.F.R. pt. 51, subpt. I, require CAA State Implementation Plans to include a program of review of proposed construction or modification of stationary sources, a process called "preconstruction review." 40 C.F.R. Section(s) 51.160-.166. Louisiana's version of this process led LDEQ to issue Permit 1036 M-1, a "preconstruction permit." MSP argues that LDEQ issued Permit 1036 M-1 without following the procedure contained in the Louisiana SIP for the issuance of preconstruction permits and, thus, that Permit 1036 M-1's requirements were not established under the SIP. The United States responds that the regulations refer not to procedure but to the authority under which the state issued the permit.

Although the parties focus their attention on the phrase "pursuant to" in the regulation, their dispute may require us to interpret the phrase "established under" in section 51.166(b)(17). The parties have not disputed that EPA approved Louisiana's SIP pursuant to the relevant regulations. See 40 C.F.R. Section(s) 52.972. The phrase "pursuant to" does appear in 40 C.F.R. Section(s) 52.23, providing that "[f]ailure to comply . . . with any permit condition or permit denial issued pursuant to approved or promulgated regulations for the review of new or modified stationary or indirect sources . . . shall render the person or governmental entity so failing to comply . . . subject to enforcement action under section 113 of the Clean Air Act." The arguments upon which we rely do not depend on whether the relevant regulatory phrase is "established under" or "pursuant to," and we thus decline to consider whether the relevant provision is 40 C.F.R. Section(s) 51.166(b)(17) or 40 C.F.R. Section(s) 52.23.

We agree with the United States for three reasons. First, the common meaning of the phrases "established under" and "pursuant to" provide little guidance regarding federal enforceability; plain language supports neither party. Second, the CAA provides EPA with broad authority to enforce state air permits. Section 113 of the CAA provides, "The Administrator . . . may commence a civil action . . . [w]henever [a person] . . . has violated, or is in violation of, any requirement or prohibition of an applicable implementation plan or permit." 42 U.S.C. §(s) 7413(b)(1) (emphasis added). Although MSP's argument is based on CAA regulations, not the statute, and although EPA may by rule limit its own ability to enforce state air permits, the broad enforcement powers Congress intended to confer upon EPA aids our interpretation of the ambiguous regulatory language. We are reluctant to construe such language as preventing the agency from exercising the enforcement power that Congress intended it to have.

Third, MSP's argument leads to the conclusion that it has no federally enforceable preconstruction permit, a conclusion inconsistent with the nature of the dual enforcement scheme contemplated in the CAA. Congress gave the United States the power to enforce state air permits in part in order to prevent a destructive race among states to attract industry by adopting the least stringent emissions limits. As the Sixth Circuit has explained,

[S]tandards for purification of the ambient air simply cannot be set along the boundaries of our 50 states. The winds, of course, recognize no such boundaries. The 50 states of this union compete intensely with one another for industry. As Congress has recognized, if state control of ambient air emissions were final, in short order, major shifts of smoke stack industries to states with the most lenient pure air standards would inevitably take place. Absent final authority in the United States EPA, the attainment goal of the Clean Air Act would prove ephemeral.

United States v. Ford Motor Co., 814 F.2d 1099, 1102 (6th Cir.), cert. denied, 484 U.S. 822 (1987); see also Sierra Club v. Indiana-Kentucky Electric Corp., 716 F.2d 1145, 1154 (7th Cir. 1983) (Congress "rather clearly embraced the general proposition that federal action was intended to remedy any problem with a state implementation plan.") (emphasis removed).

In contrast, MSP's construction of the relevant regulations would allow states to undermine the United States' section 113 power to enforce preconstruction permits by issuing such permits in violation of the relevant procedural requirements. Worse yet, sources themselves would have an incentive to insert procedural irregularities into permit processes, since doing so would allow them to avoid a federal enforcement action under section 113. Our concern for the integrity of the dual enforcement scheme that the CAA contemplates leads us to reject MSP's argument.

MSP responds to this third argument by contending that no hole in the regulatory scheme exists because Louisiana can enforce Permit 1036 M-1 in a state court action. Even if MSP's contention were correct, section 113 embodies Congress's decision that state enforcement would not always be sufficient to ensure attainment of CAA ambient air standards. In addition, we are uncertain whether Louisiana law would allow the state to enforce the permit. MSP provides no citations to Louisiana law to support its contention. MSP's arguments to this court, if accepted, might prevent it from defending against an action by Louisiana in state court to collect penalties for violations of Permit 1036 M-1 on the basis of procedural irregularities. But other entities could use this defense in other cases, and some states might construe state law to bar state enforcement of permits issued in violation of the relevant procedural requirements. Such a ruling could cause a situation in which a source operates under and violates a permit that no authority, state or federal, can enforce. We refuse to so endanger the statutory enforcement scheme.

MSP also relies upon a series of quotations to the Federal Register, specifically to Final Rule, Requirements for the Preparation, Adoption, and Submittal of Implementation Plans, 54 Fed. Reg. 27,274 (1989), purportedly illustrating that EPA explicitly recognized that states might choose to issue air permits which are not federally enforceable. These quotations include language like the following: "[P]ermits which do not conform to operating permit program requirements and the requirements of EPA's underlying regulations may not be deemed 'federally enforceable.'" 54 Fed. Reg. at 27,282. Later, EPA added, "States are free to continue issuing operating permits that do not meet the above requirements. However, such permits would not be 'federally enforceable' . . . ." Id.

MSP lifts quotations out of context. The CAA statutory scheme contemplates at least two different types of air permits unhappily named "preconstruction permits" and "operating permits," with confusion easily resulting from the fact that preconstruction permits often include limits upon a source's operations. Preconstruction permits result from a review process that occurs before construction of or major modification to a stationary source. At this stage, the permitting authority must determine whether the proposed construction or modification would violate a state's emissions control strategy or interfere with the attainment or maintenance of CAA air quality standards. 40 C.F.R. Section(s) 51.160(a)(1-2). In contrast, operating permits focus on a source's current emissions, even if the source has not recently undergone construction or major modification. See 40 C.F.R. 70.1(b) ("All sources subject to these regulations shall have a permit to operate. . . ."). The distinction between preconstruction and operating permits is critical. Before 1990, no federal law required states to maintain operating permit programs, but the CAA has always compelled states to administer a program of preconstruction review. See 42 U.S.C. §(s) 7410(a)(2)(C), 7475(a).

In 1986, when LDEQ issued Permit 1036 M-1, the CAA and its regulations conditioned approval of SIPs upon a state's maintenance of an effective program of preconstruction review. 40 C.F.R. Section(s) 51.160(a). At that time, however, the CAA did not require states to maintain an operating permit system. Some states chose to do so; EPA informs us, and MSP does not dispute, that Louisiana was not among those states. Our brief search of the Louisiana Administrative Code has unearthed no operating permit program in force at this time.

All of the quotations upon which MSP relies occurred in the context of a discussion of state operating permits, not preconstruction permits. In the pre-trial order in this case, Marine Shale affirmatively argued that any failure on its part to obtain air permits constituted noncompliance with "a 'preconstruction' permit application requirement," thus in effect conceding that Permit 1036 M-1 was a preconstruction permit. MSP does not argue otherwise on this appeal. MSP cites to no authority supporting the proposition that EPA has limited its ability to enforce state preconstruction permits. Although EPA argues that states cannot render the United States unable to enforce preconstruction permits, we need not address this broad contention to decide this case. We hold only that, in light of the statutory scheme, the phrase "established under" contained in the 40 C.F.R. Section(s) 51.166(b)(17)'s definition of "federally enforceable" refers more naturally to the source of the authority upon which the state relied to issue the permit than it does to conformity with appropriate procedures.

MSP's final argument is that even if the phrases "established under" or "pursuant to" refer to the authority upon which a state relied to issue a permit, Permit 1036 M-1 is still not federally enforceable because the LDEQ did not issue this permit under the authority of the Louisiana SIP. According to MSP, EPA recognized in the federal register portion quoted above that states might employ some portion of their police power other than that embodied in the SIP to issue air permits and that LDEQ's failure to follow the procedural steps for issuing a preconstruction permit illustrates that the Permit 1036 M-1 was an "off-the-SIP" permit.

The source of a state agency's authority to take particular action is a matter of state law. The State of Louisiana refers us to La. Rev. Stat. Ann. Section(s) 30:2011(D)(2), which provided the LDEQ power to issue air permits. The United States and LDEQ assert that LDEQ responded by promulgating L.A.Q.R. Section(s) 6.0, superseded, counterpart codified at L.A.C. Section(s) 33:III:501.C.2., at that time the portion of the Louisiana SIP dealing with the issuance of preconstruction permits, and MSP does not dispute these assertions. While nothing in these sections compels the conclusion that LDEQ lacked power from some other source to issue such permits, MSP has not identified what this other source might be. Most importantly, however, we agree with the United States that Permit 1036 M-1 itself provides the best evidence that LDEQ did not attempt to issue an off-the-SIP permit. Permit 1036 M-1 incorporates by reference a series of Standard Air Emission Permit Conditions, the first of which provides that "[f]ailure to install, properly operate and/or maintain all proposed control measures and/or equipment as specified in the application shall be considered a violation of the permit and regulation 6.0." In other words, Permit 1036 M-1 provided that a violation of the permit constituted a violation of the SIP. This reference provides strong evidence that LDEQ used its powers under the SIP to issue Permit 1036 M-1, and in the absence of state authority to the contrary, we hold that Permit 1036 M-1 was established under LDEQ's regulations approved pursuant to 40 C.F.R. pt. 51, subpt. I.

Our decision here does not conflict with National Mining Ass'n v. United States Environmental Protection Agency, 59 F.3d 1351 (D.C. Cir. 1995), decided shortly before oral argument in this case. In National Mining Ass'n, the D.C. Circuit held that EPA had to consider limits in permits unenforceable by the United States and issued pursuant to effective state permitting programs when deciding whether a source with the physical and mechanical potential to emit pollutants above the floor rate for major stationary sources could avoid preconstruction review as a synthetic minor source. 59 F.3d at 1361-65. In other words, National Mining Ass'n addressed the consequences attaching to federal enforceability. This case concerns the analytically prior question of whether a particular permit is federally enforceable. To the extent that the D.C. Circuit discussed procedural requirements necessary to make a permit federally enforceable, see 59 F.3d at 1362, it spoke of operating permits, not preconstruction permits.

2

We affirm the district court's rejection of MSP's statute of limitations defense to the allegations of minor source violations. The United States concedes that the five-year limitations period in 28 U.S.C. §(s) 2462 bars all fines for minor source emissions occurring more that five years before the filing of this lawsuit. MSP contends, however, that because emissions from each minor source began more than five years before the United States filed suit, section 2462 bars all minor source fines, even those occurring within five years of the filing of the complaint. MSP's argument is frivolous. 42 U.S.C. Section(s) 7413(b) states that the United States may sue to collect penalties of $25,000 "per day for each violation." Section 7413(b) contemplates a fine for each day a minor source operates in violation of law, and section 2462 limits the number of days to five years before the filing of the complaint. The district court properly rejected MSP's statute of limitations defense.

3

MSP argument that the district court's $2,500,000 minor source fine was too harsh rests on two grounds. First, MSP contends that the district court did not sufficiently mitigate the penalty in light of LDEQ's delay in ruling on MSP's application to amend its permit to include these emissions sources. Second, MSP argues that the district court erroneously relied on MSP's failure to install the pollution control devices mentioned in its permit amendment application.

MSP's first argument is, to say the least, unconvincing. MSP began to operate unpermitted minor sources in violation of the CAA when it first opened for business. It continued for years to operate these sources and did not apply for a permit for them until after LDEQ ordered it to complete a questionnaire identifying all emission source on the site. It continued to operate these sources while LDEQ considered its request for a variance to render these emissions legal. Emissions did not cease after LDEQ denied the variance, during the pendency of judicial review of the variance denial, or even after the Louisiana courts affirmed the denial, despite MSP's arguments to the Louisiana judiciary that the variance denial would require MSP to shut down. In light of MSP's disregard for the requirement that it obtain the very permit that has been the subject of unreasonable delay, we do not find persuasive the argument that the district court's refusal to further mitigate the fine constituted an abuse of discretion. We note further that the statutory maximum fine for MSP's minor source violations is $1,560,000,000, and that the maximum fine for the operation of these sources after the variance denial became final is $1,175,000,000. Thus, the district court's fine represents around one tenth of one percent of what it might have imposed, and around two tenths of one percent of what it might have imposed for the time period during which MSP operated in flagrant and willful violation of the permit requirement.

We reject MSP's second contention as well. MSP argues that L.A.C. 33:III:505.A.1, superseded, 21 La. Reg. 878 (July 20, 1995), counterpart codified at L.A.C. Section(s) 33:III:501.C.2, prevented it from installing pollution control devices without permission from LDEQ. Section 505.A.1 applied, however, only to "[a]ny person planning to initiate or increase the emission of air contaminants." The regulation's current incarnation, section 501.C.2, applies only to construction, modification, or operation "which may ultimately result in an initiation or increase in emission of air contaminants." This plain language suggests that MSP could have installed devices resulting in the reduction of air pollutants at any time without permission from LDEQ. We find no error in the district court's reliance on the fact that MSP did not follow its proposed course of action.

V

We consider together MSP's appeal from the district court's permanent injunctions prohibiting MSP from discharging non-contact cooling water into Bayou Boeuf, storing K-listed wastes, and operating major or minor air emission sources, all unless MSP obtained permits for these activities. MSP asserts that all three injunctions suffer a common defect, namely, that the district court abused its discretion by failing to support its decision to issue an injunction with a discussion of traditional equitable factors, such as the balance of the harms and benefits to each party, the adequacy of legal remedies, and the presence of irreparable harm. MSP also argues that the district court failed to articulate findings and reasons with sufficient specificity to support the injunctions. See Fed.R.Civ.P. 65(d). MSP supports its arguments primarily with cites to two Supreme Court decisions, Amoco Productions Co. v. Village of Gambell, 480 U.S. 531 (1987), and Weinberger v. Carlos Romero-Barcelo, 456 U.S. 305 (1982).

We do not agree that Amoco and Weinberger require a court to balance the equities and make findings regarding irreparable harm and adequacy of legal remedies in all cases arising under the environmental statutes. Amoco and Weinberger both hold that a federal statute's authorization of injunctive relief does not remove an equity court's traditional discretion over the decision to issue an injunction. In both cases, the Court was "explicitly reject[ing] the notion that an injunction follows as a matter of course upon a finding of statutory violation." Town of Huntington v. Marsh, 884 F.2d 648, 651 (2d Cir. 1989), cert. denied, 494 U.S. 1004 (1990); see also Natural Resources Defense Council, Inc. v. Texaco Refining and Marketing, Inc., 906 F.2d 934, 939-40 (3d Cir. 1990) (collecting cases). Neither decision directs courts to abandon traditional principles of equity jurisprudence in environmental cases. To the contrary, the Court identified the error in both cases as the lower courts' departure from these traditional principles; while recognizing that Congress could circumscribe a court's reliance upon the traditions of equity in a particular context, the Court held that Congress had not intended to do so in the environmental statutes at issue. Weinberger and Amoco allow a court to issue an injunction without making findings of irreparable harm, inadequacy of legal remedy, or the balance of convenience, provided that traditional equitable principles permit such a course of action.

At least two traditional principles of equity are relevant to this case. First, a court need not balance the hardship when a defendant's conduct has been willful. United States v. Pozsgai, 999 F.2d 719, 736 (3d Cir. 1993), cert. denied, 114 S.Ct. 1052 (1994); EPA v. Environmental Waste Control, Inc., 917 F.2d 327, 332 (7th Cir. 1990), cert. denied, 499 U.S. 975 (1991). This doctrine evolved in part from cases involving willful encroachments onto neighboring real estate, see, e.g., 5 John N. Pomeroy John N. Pomeroy, Jr., Pomeroy's Equity Jurisprudence Section(s) 1922, at 4362-64 (2d ed. 1919), and it remains good law today in a variety of contexts. MSP offers no reason why this traditional principle of equity should not relieve a court of its normal obligation to balance the equities when dealing with a defendant who has willfully and repeatedly violated the environmental laws.

Second, when the United States or a sovereign state sues in its capacity as protector of the public interest, a court may rest an injunction entirely upon a determination that the activity at issue constitutes a risk of danger to the public. United States Steelworkers of America v. United States, 361 U.S. 39, 60-61 (1959 (Frankfurter, J., concurring) (discussing the judiciary's historic use of equity powers, at the request of the sovereign, to enjoin activity found to be a public nuisance); Environmental Waste Control, 917 F.2d at 332; Environmental Defense Fund, Inc. v. Lamphier, 714 F.2d 331, 337-38 (4th Cir. 1983); see also Alabama v. United States, 304 F.2d 583, 591 n. 24 (5th Cir.), aff'd, 371 U.S. 37 (1962). But see United States v. Lambert, 695 F.2d 536, 540 (11th Cir. 1983) (upholding a district court's denial of the United State's request for a preliminary injunction without discussing this doctrine). In Georgia v. Tennessee Copper Co., 206 U.S. 230, 237-38 (1907), for instance, Justice Holmes's stated that when a sovereign state is the plaintiff, "[t]his court has not quite the same freedom to balance the harm that will be done by an injunction against that of which it would have in deciding between two subjects of a single political power." 206 U.S. at 238. This doctrine draws support from the extraordinary weight courts of equity place upon the public interests in a suit involving more than a mere private dispute, see, e.g., Virginian Railway v. System Federation No. 40, AFL, 300 U.S. 515, 552 (1937), and from the deference courts afford the political branches in identifying and protecting the public interest.

In the final analysis, however, unless Congress has narrowed an equity court's flexibility in the context of a particular statutory scheme, the issuance of an injunction remains an exercise of the district court's discretion. See Tennessee Copper, 206 U.S. at 238 (refusing to abandon "the considerations that equity always takes into account" even in the context of a suit by a state to protect the public interest). Weinberger and Amoco remind us that the hallmark of equity is flexibility and that courts should not lightly presume that Congress intended to narrow an equity court's traditional exercise of discretion.

Relying on United States v. Bethlehem Steel Corp., 38 F.3d 862, 868 (7th Cir. 1994), the United States argues that Congress has in fact narrowed judicial discretion in the context of RCRA, if not in the CAA or CWA. In Bethlehem Steel, the Seventh Circuit applied the second of these two doctrines to a suit involving unpermitted disposal of hazardous waste on the land. The court held that because the case involved a suit by the United States to protect the public interest, and because Congress when passing RCRA found that "disposal of solid waste and hazardous waste in or on the land without careful planning and management can present a danger to human health and the environment," 42 U.S.C. §(s) 6901(b)(2), an injunction could issue "without undertaking a weighing of the equities or making a finding of irreparable harm." 38 F.3d at 867. The United States seeks to bolster its reliance on Bethlehem Steel by pointing out that RCRA's interim status requirements are fairly specific, and that 40 C.F.R. Section(s) 265.112(d)(3)(I) requires facilities lacking permits or interim status to close.

Although the United States' argument has some force, we do not read Bethlehem Steel to hold that the United States is entitled to an injunction whenever it proves a violation of RCRA. We find nothing in RCRA which, "'in so many words, or by necessary and inescapable inference, restricts the court's jurisdiction in equity.'" Weinberger, 456 U.S. at 313 (quoting Porter v. Warner Holding Co., 328 U.S. 395, 398 (1946)). To be sure, a court of equity must exercise its discretion with an eye to the congressional policy as expressed in the relevant statute, but some of Bethlehem Steel's language may tread too closely to the view, rejected in Weinberger, that a court is "mechanically obligated to grant an injunction for every violation of law" when the United States is the plaintiff. 456 U.S. at 313.

Applying these traditional equitable principles here is not without difficulty. The district court's Order and Reasons included findings of fact and conclusions of law concerning remedies for MSP's violations of the environmental laws, but in its findings and conclusions the district court discussed only the fines levied against MSP. The district court then attached a Judgment on Main Demand ordering MSP to pay fines and enjoining future statutory violations. We find few indications in either the Order and Reasons or the Judgment on Main Demand that the district court considered the equitable theory supporting its injunctions. The district court did not explicitly make findings as to willfulness, risk of danger to the public interest, balance of the equities, irreparable harm, or adequacy of legal remedies, although it did address facts tantalizingly similar to these factors. Regarding willfulness, for instance, the court found that MSP's violations of the CAA were "serious because they have continued for long periods of time and result[ed] to some extent from MSP's deliberate indifference to the necessity of having a permit before commencing or continuing operations." But it also found that "MSP and LDEQ were of the opinion that no PSD permit was required because of the pollution control devices installed by MSP." In addition, the district court found that, overall, MSP's violations of the environmental laws had as yet caused little perceivable harm to the environment. That finding does not, however, compel the conclusion that MSP's illegal activity could continue indefinitely without causing a risk to the public health. See Amoco, 480 U.S. at 545 ("Environmental injury, by its nature, can seldom be adequately remedied by money damages and is often permanent or at least of long duration, i.e, irreparable. If such injury is sufficiently likely, therefore, the balance of harms will usually favor the issuance of an injunction to protect the environment.")

In short, we are uncertain as to whether and in what manner the district court exercised its equitable discretion in this case. On the basis of the findings and record before us, we lack a sufficient basis for appellate review. We have discretionary power to leave the injunction in place, without or without the stay, while requesting the district court to issue a supplemental opinion, Allied Marketing Group, Inc. v. CDL Marketing, Inc., 878 F.2d 806, 814 (5th Cir. 1989), and we believe it appropriate to do so and remand for further proceedings consistent with this opinion. See Sierra Club Lone Star Chapter v. FDIC, 992 F.2d 545, 552 (5th Cir. 1993). On remand, the district court should exercise discretion as to whether to enjoin MSP from its continuing violation of the environmental laws, and accompany any order issued with an explanation of the relevant facts and legal theories.

VI

We AFFIRM the district court's land ban, K-listed waste, and CAA minor source fines. We VACATE the district court's decision regarding F-listed waste, CWA violations, and Permit 1036 M-1 and REMAND for further proceedings consistent with this opinion. We hold in place the injunctions and the conditions issued below and REMAND to allow the district court to exercise its discretion regarding their maintenance and for entry of explicit findings and conclusions.