Full opinion text
WIDENER, Circuit Judge: This action is brought under § 509(b)(1) of the Federal Water Pollution Control Act [Act] and seeks review of certain regulations promulgated by the Environmental Protection Agency (EPA) pursuant to its authority under §§ 301, 304, 306 and 316(a). These regulations establish limitations on the discharge of heat from steam electric generating plants into navigable waters. The petitioners in this action include seventy members of the United Water Act Group [UWAG], who include both publicly and privately owned companies and collectively own and operate over 50% of the nation’s electric generating capacity. The regulations here under review implementing §§ 301, 304, and 306 were first issued in preliminary form on March 4, 1974. The proposed regulations under § 316(a) of the Act were issued on March 28, 1974. Following a comment period, EPA issued its final regulations on October 8, 1974. Prior to that time, however, the first of the petitions in this case was filed with the Court. The Act establishes as the national goal the elimination of discharges of pollutants into navigable waters by 1985. Among the substances defined as a pollutant by Congress was heat. It was recognized, however, that a basic technological approach to water quality control could not be applied in the same manner to the discharge of heat as to other pollutants since the temporary localized effects of thermal discharges might, in certain instances, be beneficial. Thus, Congress included within the Act § 316(a) which modifies the requirements of both sections 301 and 306 as they pertain to thermal discharges from point sources. In the instant case, the petitioners challenge the regulations generally and specifically. For the sake of clarity, we will first consider the general challenges and then discuss those applying to specific regulations. STANDARD OF REVIEW Before addressing the issues raised by the Industry, it is important to note that the standard of review imposed upon the court is narrowly prescribed by Section 706(2)(A) of the Administrative Procedure Act. That section requires a finding that the administrative action in question is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law” before it may be set aside. The scope of review thus permitted was discussed by this court in Appalachian Power Co. v. EPA, 477 F.2d 495 (4th Cir. 1973). There, we stated; “Courts require that administrative agencies ‘articulate the criteria’ employed in reaching their result and are no longer content with bare administrative ipse dixits based on supposed administrative expertise. . . . While an agency may have discretion to decide, ‘[Discretion to decide does not include a right to act perfunctorily or arbitrarily’; and, in order for a Court to make a critical evaluation of the agency’s action and to determine whether it acted ‘perfunctorily or arbitrarily,’ the agency must . ‘explicate fully its course of inquiry, its analysis and its reasoning’.” 477 F.2d at 507. While the court may not substitute its judgment for that of the agency, the grounds upon which the agency acted must be clearly disclosed in, and substantiated by, the record. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971); FTC v. Sperry and Hutchinson, 405 U.S. 233, 249, 92 S.Ct. 898, 31 L.Ed.2d 170 (1972); duPont v. Train, 541 F.2d 1018, No. 74-1261, et al. (4th Cir. 1976). In evaluating the course of conduct employed by the EPA, we must, of course, bear in mind that Congress vested that agency with the power to choose among alternative strategies. South Terminal Corp. v. EPA, 504 F.2d 646, 655 (1st Cir. 1974). Nevertheless, because of the “drastic impact” of the statute, see Appalachian Power Co., p. 503, and the anticipatory review provisions of § 509(b)(1) of the Act, which require review of the EPA’s regulations under §§ 301, 302 and 306 within 90 days of their issuance rather than at the time of application to a particular point source, EPA must ordinarily be held to a standard of at least literal compliance with the provisions of the Act. THE REGULATIONS Under the final regulations promulgated by EPA, all existing generating plants of 500 megawatts or more which came on line on or after January 1, 1970 must backfit closed-cycle cooling systems by July 1,1981. All existing units, regardless of size, that began or will begin operation on or after January 1, 1974, are likewise subject to the backfit requirements. Limited exemptions are provided, however, based upon land availability, salt drift impact, and interference with commercial aviation. Finally, all new plants are made subject to narrowly limited no-discharge thermal limitations without exception. The regulations also approve of the use of existing cooling lakes by existing but not new generating units. Cooling ponds but not lakes are deemed acceptable for all units. By definition, cooling water impoundments which impede the flow of a navigable stream are cooling lakes while those which do not are cooling ponds. There are presently three basic methods of closed-cycle cooling systems which may be employed to meet the requirements set forth in EPA’s regulations. These are: (a) evaporative cooling systems such as wet cooling towers and spray systems; (b) cooling ponds and lakes; and (c) dry cooling towers. EPA itself has indicated, however, that this latter method of cooling, which employs huge radiator-like devices, cannot generally be applied to large electrical generating units due to the significant loss of plant efficiency which results. The most commonly used form of evaporative cooling is the wet cooling tower, either natural or mechanical draft. Natural draft cooling towers are enormous concrete cylinders, which may be 350 to 550 feet in diameter and 300 to 600 feet tall. The bottom one-tenth of the tower is filled with slats and baffles to break up the water and expose a larger surface area to the air flow so as to increase evaporation. Warm water from the condenser is pumped to the top of the tower, there discharged, and cooled by moving air as it falls to the bottom. It is there collected and returned to the condenser. Mechanical draft evaporative cooling towers are composed of modules (each a miniature tower) approximately 70 by 40 feet, placed side by side for distances up to 600 feet. Large top or side mounted fans on each cell provide the air movement for a forced draft to aid evaporation as the warm water drops to the bottom of the tower. Spray ponds are also used for cooling. They consist of artificial ponds or canals into which water is sprayed. The water is cooled by evaporation resulting from the contact with the natural air currents during the spraying and after collection in the pond. Cooling ponds and lakes represent the other practical systems. They normally consist of artificially constructed bodies of water built by damming a natural watershed. The condenser water is fed into the cooling lake or pond where it is cooled through evaporation. It is then recycled to the condenser. REGULATORY CHALLENGES (a) The Act’s Regulatory Scheme — Flexibility v. Uniformity Petitioners argue that the steam electric regulations are excessively rigid in that EPA has prescribed nationally uniform effluent limitations rather than “guidelines for effluent limitations” as required by § 304 of the Act. In essence, they contend that there is no authority under the Act for the issuance of uniform effluent limitations. In large part, this controversy centers around whether the regulations in question were issued under § 301 or § 304. This issue was raised and considered by this court in the case of duPont v. Train, 541 F.2d 1018, Nos. 74-1261, et al. (4th Cir. 1976). In that case, we concluded that the EPA was permitted to establish limitations under § 301 which would be generally applicable to point sources by categories. P. 1026 et seq. We further concluded that the agency could validly impose limitations in terms of single numbers rather than a range of numbers. P. 1027. We went on to state, however, that “[f]or all sources, both existing and new, we believe that the solution which most nearly satisfies congressional intent is recognition that the regulations are presumptively applicable to permit applications.” P. 1028. Thus, the issuer of a permit under § 402 may consider whether a particular applicant is to be held strictly to the confines of the agency’s regulations. The burden of proof remains upon the applicant, however. Only after he has established the inappropriateness of the regulations as applied to him, for example, employing the generic factors of §§ 304, 306 or any specific variance clauses promulgated thereunder, need the permit issuer go beyond the regulations. Of course, the permit issuer does not have unreviewable discretion in determining whether limitations prescribed under a particular regulation should be deemed inapplicable to any individual point source. Under § 402(d)(2), EPA may prevent the issuance of any permit to which it objects. Thus, through the exercise of this veto power, the agency may insure that the permit grantors give proper recognition to the need for uniformity in the application of the Act while at the same time recognizing variables which may exist from location to location and plant to plant. Because we are of opinion that the regulations are only presumptively applicable, we think they are not so rigid as to compel an inflexible application of the Act. Accordingly, we reject this portion of the Industry’s argument. (b) The Variance Clause The petitioners have also pressed a number of complaints about the adequacy of the variance clause in the steam electric regulations.3 **************** That provision allows for modification of the 1977 effluent limitations applicable to a particular point source upon a showing that factors relating to that point source are fundamentally different from those considered in the establishment of the applicable single number limitation. Specifically, Industry contends that: (1) the variance clause is too narrow because it is activated only by “fundamentally different factors;” (2) it should apply to 1983 and new source performance standards; and (3) it should include consideration of economic impact. As we noted in duPont, 541 F.2d 1018, Nos. 74-1261, et al., provisions for variances are appropriate to the regulatory process. This is particularly so in the case of regulations having presumptive applicability throughout the nation. The factors to be taken into consideration, however, ought ordinarily to be at least as broad as the factors relied upon in establishing the limitation if the provision is to have meaning. In the instant case, EPA has said that a variance from the 1977 standards will be granted only where “the factors relating to the equipment or facilities involved, the process applied, or other such factors related to such discharger are fundamentally different from those factors considered in establishing the guidelines.” Thus, only technical and engineering factors, exclusive of cost, may be considered in granting or denying a variance. Based upon the Act taken as a whole, we are of opinion that such a variance clause is unduly restrictive and, accordingly, 40 CFR § 423.12(a) must be set aside and remanded for further consideration. Thus, it would appear that, unlike the case in duPont, the administration of these regulations is not a matter of speculation. As such, they are properly the subject of review at this time. We begin with the observation that § 301(c) of the Act provides that EPA may grant a variance from the 1983 standards to any point source upon a showing that “such modified requirements (1) will represent the maximum use of technology within the economic capacity of the owner or operator; and (2) will result in reasonable further progress toward the elimination of the discharge of pollutants.” Clearly, the Act, in its regulatory plan, contemplates increasingly stringent control measures for existing and new sources culminating in the elimination of the discharge of all pollutants into navigable waters by 1985. We are of opinion that the initial phase of these regulations, the 1977 standards and the subsequent new source limitations, were not intended to be applied any less flexibly than the final Phase 11-1983 requirements. Thus, if such factors as the economic capacity of the owner or operator of a particular point source is relevant in determining whether a variance from the 1983 standards should be permitted, they should be equally relevant when applied to the less stringent 1977 standards as well as the new source requirements. In addition, we note that both § 304 (1977) and § 306 (new sources) provide that the factors to be taken into consideration in determining the 1977 and new source standards include not only the engineering aspects of the various types of control technology, but also (1) the total cost of application of such technology (cost of achieving such effluent reduction) and (2) the resulting non-water quality environmental impact (including energy requirements). The EPA has offered no reasoned explanation for limiting the variance clause to considerations of technical and engineering factors only. Certainly the adverse non-water quality environmental impact which may result from the strict application of the agency’s effluent limitations to a particular plant is as significant as the technological difficulties which may be encountered. The same may be said for a consideration of energy requirements. Upon reconsideration, then, EPA should come forward with a meaningful variance clause applicable to existing as well as new sources, taking into consideration at least statutory factors set out in §§ 301(c), 304(b)(1)(B) and 306(b)(1)(B). (c) Section 315 Report Industry next argues that § 301(b)(2)(A) precludes a no-discharge limitation as a 1983 standard absent achievability findings based upon available information including that developed pursuant to § 315(a). This latter provision creates a National Study Commission to review and report on “all of the technological aspects of achieving, and all aspects of the total economic, social, and environmental effects of achieving or not achieving the effluent limitations and goals ... for 1983. . ” Since this report was not published at the time of the promulgation of the regulations, it is Industry’s position that EPA has improperly promulgated the thermal limitations in question. The legislative history of § 315 indicates, however, that Congress did not intend that the Commission report be a condition precedent to the establishment of the 1983 standards. It appears, rather, that the report was intended to provide an independent evaluation of the economic, social, and environmental aspects of the regulatory plan to be obviously used in Congressional review of the agency’s functions. A part of the explanation of Senator Muskie filed with his filing of the Conference Report of the joint House-Senate conference concerning the Act is directly in point: “The House amendment provided a study by the National Academy of Sciences of social, economic and environmental implications of ‘best available demonstrated technology’ and of any effluent limits which would require the ‘elimination of the discharge of pollutants’. Under the House amendment, such a study would have been completed in two years and would have been a condition precedent to any requirements beyond January 1, 1976. “The Conference agreement does not require a subsequent action of Congress to trigger those aspects of the program which are commonly referred to as Phase II and beyond. The requirement to achieve effluent limitations based on the best available technology and the elimination of discharge of pollutants are automatic on enactment.” These remarks are supported by the joint Explanatory Remarks by Conference Committee which indicates that the House version of § 315 conditioning implementation of Phase II standards upon receipt of the Commission report was rejected. We are accordingly of opinion that EPA properly promulgated no-discharge limitations under § 301(b)(2)(A) prior to the submission of the § 315 report. (d) Thermal Backfit Requirements (1) 1983 Requirements — Reduction for . Existing Units Industry contends that EPA’s 1983 thermal backfit requirements for existing units are invalid because the agency failed to balance the overall social benefits to be derived from its regulations against their social costs. In essence, these regulations, which purport to establish effluent reduction levels attainable by the application of the best available technology economically achievable, require all existing generating units placed in service after December 31, • 1973, as well as all units of 500 megawatts or greater coming on line after December 31, 1969, to backfit closed cycle cooling. Petitioners argue that the 1983 effluent limitation standards set forth in the Act reflect an intent on the part of Congress that social benefits of pollution control be measured against their costs in choosing among alternative strategies. EPA, on the other hand, takes the position that the language of the Act pertaining to the 1983 standards requires no balancing of social benefits against social costs. Moreover, the agency asserts that even if the Act were held to so require, it has, in its rulemaking, analyzed the benefits of the challenged regulations and found them to be worth the associated costs. It further asserts that it agrees with its environmental contractor, Energy Resources Company (ERCO), when it states that “benefits cannot be properly assessed within the present state of the Art.” We disagree with EPA’s (and partially with Industry’s) assertions, and, accordingly, set aside and remand for further consideration § 423.13(l), (m). In duPont, 541 F.2d 1018, Nos. 74-1261, et al., we rejected Industry’s contention that benefits derived from a particular level of effluent reduction must be quantified in monetary terms, and such contention is rejected here. This reflects the simple fact that such benefits often cannot be reduced to dollars and cents. Nevertheless, EPA is under a statutory duty to determine whether, in fact, its regulations for 1983 will “result in reasonable further progress toward the national goal of eliminating the discharge of all pollutants . . .” 33 U.S.C. § 1311(b)(2)(A). Accordingly, the agency must consider the benefits derived from the application of its effluent reduction requirements in relation to the associated costs in order to determine whether, in fact, the resulting progress is “economically achievable,” and whether the progress is “reasonable.” EPA argues, however, that it has, in fact, assessed the benefits to be derived from its regulations and has therefore satisfied the requirements of 33 U.S.C. § 1311(b)(2)(A). In support of this contention, EPA refers to the table below. According to that table, EPA calculated the incremental production costs, capital costs, fuel consumption and capacity reduction which would result from the application of closed-cycle cooling at existing plants of various ages. These incremental values were then compared to the amount of heat kept out of the water by such systems. Based upon the resulting cost-effectiveness ratios, EPA concluded that the thermal control technology there analyzed resulted in more favorable effluent reduction benefits when applied to newer and larger generating units. The analysis also indicated the cost per unit of electricity which accrued as a result of applying closed-cycle cooling to the various generating units. Based upon this study, EPA contends it limited coverage of its backfit regulations to units having baseloads of 500 megawatts or more and coming on line after 1970. Smaller units put into operation after 1973 were likewise required to adopt closed-cycle cooling systems. Industry challenges EPA’s reliance upon these figures on the ground that they do not indicate whether the regulations will result in reasonable further progress toward the national goal. We agree. EPA’s study merely establishes the cost-effectiveness of installing mechanical draft cooling towers at individual plant sites. It in no way indicates whether, in light of the associated costs, application of such systems will result in reasonable effluent reduction levels. In response to this, EPA cites the report of its environmental contractor, ERCO, which it claims satisfactorily analyzes the environmental benefits and risks associated with the various alternatives which were before it. The foundation of this report was apparently a random sample taken of various power companies throughout the country. Based upon this sample, the report concludes that if a generating unit uses less than 30% of a stream’s flow, there will be no ecological danger. If as much as 70% of the stream flow is used more than five percent of the time, however, there will be a high risk of such danger. Between these two extremes, the report concludes that there is a medium risk of danger. Yet, nowhere in the report does ERCO state upon what basis it reached this conclusion. Moreover, ERCO goes on to assume that § 316(a) exemptions will issue to all low risk, half of the medium risk, and none of the high risk installations. Again, however, the reason why the exemption would apply to the assumed proportion of installations is not stated. Perhaps it is a result of ERCO or EPA site studies; perhaps not. We are left to guess. By the same token, we note that the low risk category, for example, does not take into consideration the effect of increases in the temperature of the water discharged into the river by the individual generating units. Is this an oversight or do the results of site by site studies justify the exclusion of the very element, temperature, with which the statute is concerned? Again, the record contains no answer. Simply stated, then, the ERCO report upon which EPA now seeks to rely fails to explain and document the basis for the numerous assumptions made and relied upon in its analysis. As such, we cannot say, based upon this report, whether, in fact, EPA’s effluent reduction regulations will result in reasonable further progress toward the elimination of artificial sources of heat from the nation’s waters. Moreover, we further note, finding it to be of some significance, that the record is replete with allusions to the effect of heat upon aquatic life; damage to eggs; different effects on adults and juveniles; the growth of algae; interrupted migration; the thresholds of aquatic communities; differences between streams, lakes and estuaries; are to mention but a few. Yet, despite agreement that the literature is full of learned papers on the subject at hand, EPA contends that the state of the art is not such that the incremental benefits of heat removal from the discharge of generating plants can be predicted. The references throughout the record to the effects of heat on aquatic life, not unreasonably, cause us to consider this position with a certain degree of caution. The references are simply too numerous and are stated by too many people on both sides who are ostensibly qualified to speak. An illustration is a colloquy between Dr. Brooks of the United Water Act Group and Dr. Schneider of EPA: “DR. BROOKS: I would like then, as part of what could have been an answer to that question — I heard earlier that there were numerous shortcomings in existing data available from operating thermal discharges. I think most of us will agree that it is certainly pointed out in the 104 T studies that these deficiencies do exist. Does EPA believe that there is sufficient data to arrive at any predictions of thermal benefits or benefits at all? DR. SCHNEIDER: Is this with respect to specific sites or in general? DR. BROOKS: Both. DR. SCHNEIDER: That is a very broad question. We could write text books on the subject.” Assuming that EPA’s conclusion is correct, that the state of the art is such that the incremental effects of heat are not known with any degree of certainty, the least EPA could have done would have been to articulate what the state of the art was and why, according to scientific opinion, predictions could not be made. Even assuming that it might not be possible to articulate with reasonable certainty the achievability of the benefits to be derived from a specified amount of heat removal, it seems to us that the expectancy might be stated, for if there is no expectancy of benefits to aquatic life, is the expenditure of billions of dollars justified under any standard? We, of course, recognize that EPA may fairly assume that a reduction in the amount of heat discharged will, as a general rule, result in some benefit to the environment. The question is, however, whether the reduction results in reasonable further progress toward the elimination of pollutants. This is ultimately a matter within the sound discretion of EPA itself. Yet, in determining whether a particular reduction level is, in fact, reasonable, EPA must compare the cost of achieving that level of reduction (which it has done) and the ecological benefits to be derived therefrom (which it has not done) with the benefits and costs associated with alternative levels of heat reduction. At the very least, on the best information available, the ecological benefits expected from the ordered reduction should be stated, and, if impossible so to do, EPA should state why. It may well be, for example, that a 90% reduction in thermal pollution at a cost of $5 billion is entirely reasonable even when it is shown that an 80% reduction would cost but $2 billion, for it is possible that the elimination of the additional 10% in total heat discharged would have positive environmental effects which would far outweigh the additional $3 billion in cost. By the same token, if no tangible environmental benefits will accrue by increasing the thermal reduction level from 80% to 90%, the additional expenditure of $3 billion might be considered unjustified. Thus, in choosing among alternative strategies, EPA must not only set forth the cost of achieving a particular level of heat reduction but must also state the expected environmental benefits, that is to say the effect on the environment, which will take place as a result of reduction, for it is only after EPA has fully explicated its course of conduct in this manner that a reviewing court can determine whether the agency has, in light of the goal to be achieved, acted arbitrarily or capriciously in adopting a particular effluent reduction level. Were this not the case, EPA would be free to impose ever more stringent and costly control measures which, while incrementally reducing the level of thermal discharge, would not result in any reasonable improvement in the “chemical, physical, and biological integrity of the Nation’s waters.” 33 U.S.C. § 1251. It should be made clear, however, that our remand here is very narrow in scope since we do not disapprove the general principle of requiring installation of cooling devices on a part of the planned and existing electrical generators in the country. EPA’s conclusion that the size of the generator and the year of its first service operation offers the best means of determining which units will be required to backfit seems to us to be reasonable and subject to no infirmity now apparent calling for reconsideration. Moreover, we are not now prepared to say the particular sizes of generators and dates of service which EPA has adopted as the breaking points for ascertaining the necessity of backfitting cooling devices on existing or planned equipment are unreasonable on the record before us. Yet, while we are unable to say that EPA has not acted reasonably, neither are we able to say it has not acted “perfunctorily or arbitrarily,” Appalachian Power v. EPA, supra, p. 507, for the “criteria used in reaching . . [the] result” has not been stated other than by way of assumptions rather than by the reasoning of those qualified in the field. As was the ease in Tanner’s Council of America v. Train, 540 F.2d 1188, No. 74-1740 (4th Cir. 1976), there is simply no evidence in the record that would reveal the reasonableness of EPA’s conclusions. To sustain these regulations on the present record, this court would have to trust completely EPA’s conclusions. Judicial review must be based on something more than trust and faith in EPA’s expertise, however. duPont v. Train, 541 F.2d 1018, at p. 1036, Nos. 74-1261, et al. (4th Cir. 1976). On remand, then, EPA must state the benefits especially to aquatic life, for the various alternatives considered if that can be done. If these benefits cannot be stated with any degree of certainty, EPA will state the expected benefits according to whatever scientific opinion it relies upon, fully explicating the basis, including the opinion, upon which it relies. If no expected benefits can be stated, EPA must state why they cannot be and the scientific opinion which supports that conclusion. 2) Economic Achievability of EPA’s Thermal Backfit Requirements Industry also attacks EPA’s backfit requirements on the ground that they will impose a staggering burden on the power industry. According to EPA, its thermal controls will require $5.2 billion in additional funding. Industry, on the other hand, estimates the cost at close to $7.4 billion. The agency contends that even if this higher figure is accepted, capital cost to electric utilities will increase by only 7% between now and 1983. In support of its position that Industry can, in fact, secure the necessary funding to meet the backfit requirements, EPA relies primarily upon an economic analysis of its final effluent standards done by Temple, Baker, and Sloane, Inc. (Temple). Temple’s final report, which was submitted in December of 1974, noted that prior to the 1973 Arab oil embargo, the electric utility industry was planning to spend more than $205 billion for capital equipment. Due to a decline in growth, it found that current estimates of such expenditures should have been revised downward to $179 billion. Temple then took the position that if Industry could have raised the additional $25 billion required to finance its original growth estimates, it “should experience little trouble meeting the added requirements to comply with the final guidelines.” In addition, EPA suggests that the unique relationship that exists between government and the utilities lessens the degree to which Industry must contend with the vagaries of the marketplace. The agency points out that certain factors governing the industry’s future growth such as cash flow, return on capital, and demand growth, are subject to direct manipulation by governmental agencies. EPA further asserts that: “[Bjecause the industry plays an important role in determining the actions of state and federal regulatory commissions, to the extent that they themselves promote such innovative practices as peak power pricing to flatten peak demand, they can be a controlling factor in determining their own future capital needs. If the electric utility industry were genuinely doubtful about its ability to meet its future capital needs, it would now be promoting rate structures conducive to a flattening of peak demand curves, and a reduction in the rate of growth, rather than the type of rate structure currently in widespread use, which encourages greater consumption and the wasting of energy.” EPA Br. at 66. Industry disputes both of these assertions. First, it contends that rate regulation is often, in practice, a constraint rather than an aid in efforts to compete for scarce capital. Moreover, Industry argues that despite EPA’s contentions to the contrary, a reduction in demand brought on by the energy crisis and a general downward turn in the economy does not make available additional capital. Since the EPA’s backfit regulations have been set aside for further consideration, we find it unnecessary to pass on the validity of the agency’s economic analysis. Before any revised regulations are promulgated, however, we direct that EPA fully consider any economic changes which may have occurred in the money market or in the demand projections for the electric utility industry since these regulations were issued, and afford Industry an opportunity to comment upon its analysis. (e) Backfit Requirements for AEG Approved Nuclear Power Plants Industry next contends that EPA’s summary rejection of an Atomic Energy Commission proposal that fifty-five nuclear power units which had completed environmental impact statements be exempted from the backfit requirements was arbitrary and capricious. The National Environmental Policy Act requires every federal agency to take into account the environmental impact of certain proposed actions. Pursuant to this mandate, the AEC’s (now Nuclear Regulatory Commission) licensing process entailed, at every stage of review, an analysis of the economic, social, and environmental costs and benefits of proposed nuclear generating units. Following its NEPA review, the Commission concluded that there were approximately 70 units for which the environmental impacts of thermal effluents were not significant nor sufficiently serious to warrant the cost of closed-cycle cooling. Fifteen of these units were found not to come under EPA’s backfit requirements due to either age or size. The remaining fifty-five units would, however, have been required to install closed-cycle systems unless exempted. Accordingly, Dr. Dixy Lee Ray, Chairman of the AEC, wrote EPA requesting: “[A] special class in the steam-electric power plant category of point sources be established under section 304(b)(1) and 304(b)(2) of the FWPCA. As contemplated by the statute, the class would consist of steam-electric power plants of a certain type (nuclear power reactors) and age (those for which final environmental impact statements had been prepared by the AEC Regulatory Staff pursuant to section 102(2)(C) of NEPA prior to the effective date of the guidelines).” In support of this request, Dr. Ray pointed out that the capital expenditures required to backfit closed-cycle cooling systems at these plants would approximate $2.1 billion in 1974 dollars. Escalation would increase that figure by 50% by the year 1983. In response, Russell E. Train, Administrator of EPA, noted that a similar proposal was considered in preparing the preliminary guidelines but was rejected because “in the opinion of [EPA’s] General Counsel, it would not be legally defensible.” This is the only apparent explanation for EPA’s action. The preamble to EPA’s regulations does not discuss the question nor does EPA’s Development Document or Economic Analysis. Certainly, the bald assertion that AEC’s proposal is “not legally defensible” is not a sufficient articulation of the criteria employed by EPA in reaching its decision. It does not permit the court to in any way satisfy itself that the agency engaged in reasoned decision-making. See, e. g., Appalachian Power Co. v. EPA, 477 F.2d 495, 507 (4th Cir. 1973). Accordingly, EPA is directed upon reconsideration of its backfit regulations to fully evaluate the AEC’s proposal and set forth its reasons for rejecting or accepting the proposed category in its revised regulations. In so directing, we do not imply that EPA must abdicate its authority over effluent reduction to the AEC or NRC. As the parties themselves point out, such abdication of authority has been repeatedly held invalid. See Calvert Cliffs Coordinating Committee v. AEC, 146 U.S.App.D.C. 33, 449 F.2d 1109 (1971); Greene Co. Planning Bd. v. FPC, 455 F.2d 412 (2d Cir. 1972). We do require, however, that EPA “explicate fully its course of inquiry.” Appalachian, supra, at 507. (f) Cooling Lakes as Best Available Technology Under 40 CFR § 423.15(1) and 40 CFR § 423.25(1), the discharge of heat by steam-electric power plants classified as new sources is permitted only in “blowdown from recirculated cooling water systems,” (e. g. cooling towers) and in “blowdown from cooling ponds.” Existing units under 40 CFR § 423.13(1), on the other hand, are permitted to employ a variety of cooling techniques, including existing cooling lakes. Industry, as well as the State of Texas, contends that EPA’s restrictions upon the use of existing and new cooling lakes is both arbitrary and capricious in light of the resulting increase in water consumption. EPA has defined “cooling lakes” as any “manmade water impoundment which impedes the flow of a navigable stream and which is used to remove waste heat from heated condenser water prior to recirculating the water to the main condenser.” 40 CFR § 423.11(n). Cooling ponds, by way of contrast, include “any manmade water impoundment which does not impede the flow of a navigable stream and which is used to remove waste heat from condenser water. . . . .” 40 CFR § 423.11(m). This distinction takes on meaning when considered in light of EPA’s interpretation of the term “navigable waters.” According to the agency, all “[i]ntrastate lakes, rivers, and streams which are utilized by interstate travelers for recreational and other purposes; intrastate lakes, rivers and streams from which fish or shellfish are taken and sold in interstate commerce; and intrastate lakes, rivers and streams which are utilized for industrial purposes by industries for interstate commerce” are navigable waters. At oral argument, counsel for EPA took the position that this includes not only the waters of natural streams, lakes and rivers, but also surface waters which diffuse themselves over the ground and follow no defined course or channel, gathering into no more definite body of water than a wet weather creek. We reject this assertion, being of opinion that, where possible, the collection and use of such surface waters as opposed to natural rivers, lakes and streams furthers the purpose of the Act in restoring and maintaining the “chemical, physical and biological integrity of the Nation’s waters.” We nevertheless recognize that virtually every natural stream falls within this narrower view as to what constitutes “navigable waters.” According to EPA’s regulations, any impoundment which impedes the natural flow of any such stream would constitute a cooling lake and thus preclude it from use for cooling purposes. Despite EPA’s restrictions upon the use of cooling lakes, the agency’s own Development Document specifically identifies such lakes as a form of closed-cycle cooling. It states: “The technological basis for best available technology economically achievable, and new source performance standards consist of closed-cycle evaporative cooling towers and cooling ponds, lakes and canals.” Moreover, that document lists cooling lakes as one of the available technologies for achieving waste heat removal in closed or recirculated cooling systems. As is there noted, such lakes “are similar in principle to open, once-through systems, but . are closed inasmuch as no significant thermal discharge occurs beyond the confines of the lake.” So we see that EPA has itself recognized that cooling lakes represent an achievable method of closed-cycle cooling. In addition, the agency has deemed them to be the best practicable technology for existing generating units presently employing such lakes. Despite the foregoing, EPA attempts to justify its limitations on the use of cooling lakes on the ground that it is obliged under the Act to ban all new discharges into existing or new lakes because they are navigable waters. We disagree. The Act contains no blanket requirement that all effluent discharges into navigable waters be eliminated. Rather, § 306(b)(1)(B) of the Act merely requires that EPA propose and publish regulations establishing “standards of performance” for “new sources” within a list of “categories” of point sources, giving due regard at least to non-water quality environmental impact and energy requirements. The term “standard of performance” is defined as “a standard for the control of discharges of pollutants which reflects the greater degree of effluent reduction . including, where practicable, a standard permitting no discharge of pollutants.” Thus, where no-diseharge limitations are not practicable in light of the non-water quality impact or other such factors, the “best available” technology may, in fact, entail some discharge of heat into the Nation’s waters. The EPA has itself recognized this in the case of existing generating units employing cooling lakes. Due to the fact that “the addition of recirculating systems to [existing] lakes would substantially increase water consumption,” the agency concluded that cooling lakes were the best practicable technology. Industry contends that this same reasoning holds true for new units on existing lakes as well as units on newly constructed lakes. We agree. EPA’s § 104(t) report supports Industry’s assertion that the use of cooling towers will result in a dramatic increase in water consumption. It concludes that “the evaporation resulting from a given thermal addition to a [cooling] lake is generally less than 50% of the evaporation that will result from similar thermal loading of a cooling tower.” In California alone, the use of cooling towers rather than lakes at all additional power plants projected to be on line by 1990 will create an annual water deficiency of nearly 2.2 million acre-feet; more than double the 1970 prediction. In the Brazos River Basin of Texas, the consumptive use of water for cooling purposes resulting from the installation of cooling towers would increase by one and one-half to two times, according to some estimates. An unanswered argument at this stage of the proceeding is Governor Briscoe’s (of Texas) letter to EPA: “The level of consumptive use of water that would be necessitated by implementing these proposed regulations is not merely unacceptable; Texas simply does not have the water resources available to comply.” EPA dismissed these concerns during the rulemaking process, however, on the ground that “much of the evaporated water would precipitate [again] through the natural water cycle.” This is obviously not a sufficient answer and demonstrates a serious lack of concern for a balanced consideration of the total environmental impact of the regulations. In Arizona and New Mexico, for example, almost all precipitation is from tropical storms which originate in the Caribbean or Mid-Pacific, while in California almost all precipitation originates in the North Pacific. Little, if any, of the water which evaporates in these States returns in the form of rain. Thus, any new use of water in these areas results in a net reduction in the water supply remaining available for other uses. It is clear that Congress intended such reductions in water supply to be taken into consideration by EPA in determining the best available technology to abate effluent discharges. The basic directive of the entire Act is set forth in § 102(a) of the Act and provides, in part, that EPA must give due regard to “the withdrawal of such waters for public supply, agricultural, industrial and other purposes.” In addition, § 104 provides for continuing comprehensive studies of the effects of the control of thermal discharges. In so doing, it requires that EPA “[i]n evaluating alternative methods of control . . . consider . the total impact on the environment, considering not only water quality but also air quality, land use, and effective utilization and conservation of fresh water. ” This Congressional concern over the conservation of our limited water resources is further reflected in §§ 304 and 306 which require consideration of various enumerated factors including the non-water quality impact of any regulations on the environment in determining what constitutes the “best” technology. We are of opinion, therefore, that EPA’s ban on the use of new and existing cooling lakes is clearly not in accordance with the Congressional directive regarding the conservation of our water resources, probably in most areas of the country, and particularly as that ban applies to regions where fresh water is in short supply. It is evident from an examination of the record that EPA’s regulations will result in needless water consumption and, thus, impede effective utilization of our fresh water resources. Accordingly, 40 CFR §§ 423.13(1), 423.15(1), and 423.25(1) are set aside and remanded to EPA for further consideration with directions that it fully evaluate the total environmental impact of any subsequent regulations which it may issue, particularly with reference to water usage and its effect on the more arid regions of the Nation. On remand, EPA may find it difficult to avoid the logic of the statement of the General Counsel of the Department of Commerce (see, footnote 49, supra) where he concludes that subcategorization of the industry by locality should have been considered taking into account the availability of water for consumptive use. It is difficult for us to see how any program could receive approval which did not, at the very least, balance the consumption of water against the water resources and its effect. Along this line, since the use of water is less in most lakes used for cooling and also for other purposes, their exclusion without taking into account water consumption may only be considered arbitrary and capricious. We repeat for emphasis that EPA must consider the almost immeasurable impact of needless increases in water consumption in any part of the country, especially the arid areas. (g) Sea Water Cooling Towers as Best Available Technology for Open Ocean Dischargers EPA’s regulations also prohibit open ocean thermal discharges and require closed-cycle cooling at power plants located along the nation’s coastlines. Because fresh water towers would exacerbate water shortages in many coastal areas, EPA has recognized that cooling towers in such areas must, of necessity, employ sea water. Industry first challenges the forced construction of such cooling towers at new coastline plants on the ground that sea water cooling towers for full-sized power plants are not “currently available” and, thus, are not “demonstrated” as is required under § 306. We need not reach this question, however, since 40 CFR § 423.13(1), as amended by 40 Fed.Reg. 7095-96 (1975), has previously been set aside. Upon reconsideration of that portion of its effluent limitation guidelines, EPA should, nevertheless, allow industry the opportunity to comment upon any revised regulations insofar as they relate to sea water cooling towers and reconsider its position if industry’s objections have merit. We especially note that the B. L. England Station may well be the only power station presently in existence which approaches the required technology. On remand, particular attention should be paid to data relating to the performance of that station. By emphasizing the B. L. England Station, we do not mean to exclude from consideration on remand other relevant information from legitimate sources. (h) EPA’s Implementation of § 316(a) Section 316(a) of the Act provides that EPA may impose less stringent effluent limitations at any point source than might otherwise be required under either § 301 or § 306 “whenever the owner or operator of any such point source . can demonstrate that any effluent limitation proposed for the control of the thermal component . . . will require effluent limitations more stringent than necessary to assure the projection and propagation of a balanced, indigenous population of shellfish, fish, and wildlife in and on the body of water. . . ” Petitioners contend that in implementing this section, EPA must provide that a plant’s compliance with existing water quality standards constitutes prima facie evidence of compliance with the requirements of § 316(a). Accordingly, industry argues that EPA has not properly implemented § 316 and, thus, the applicable regulations must be set aside and remanded for further consideration. We disagree. Industry cites nothing in the Act or its legislative history which requires EPA to adopt its suggested test under § 316. We, of course, recognize that under the Act each state adopts water quality standards for inter- and intrastate water so as to “protect the public health or welfare, enhance the quality of water and serve the purpose of this chapter.” In addition, such standards must take into consideration “[the receiving water’s] use, and value for public water supplies, propagation of fish and wildlife, [and] recreational purposes. . . ” Moreover, EPA is required to review all such standards and revise them where they are deemed inadequate. So we do not agree with Industry that the § 316(a) “protection and propagation” standard must be deemed satisfied if the thermal discharge in question meets the applicable water quality standards as distinguished from effluent standards. EPA points out that state water quality standards typically apply to an entire waterway or a relatively large segment of it. By way of contrast, EPA views § 316(a) as providing for consideration of specific site conditions in the setting of thermal limitations for individual power plants. Thus, while a greater level of thermal effluent by a generating unit might well fall within the general requirements of an approved state standard, EPA takes the position that such discharge might nevertheless cause serious harm to a particular spawning ground, for example located just below the plant’s discharge point. It is such specific site conditions to which EPA contends § 316(a) is directed. While both the position of Industry and EPA have force in logic, we are of opinion that Train v. NRDC, Inc., 421 U.S. 60, 95 S.Ct. 1470, 43 L.Ed.2d 731 (1975), is controlling here. There, the court indicated that while an agency’s interpretation of its statutory authorization need not be “the only one it permissibly could have adopted . [if] it was at least sufficiently reasonable it should [be] accepted by the reviewing courts.” 421 U.S. at 75, 95 S.Ct. at 1480. Moreover, the court went on to reiterate that where an agency’s interpretation of a particular Act is not unreasonable, Courts of Appeals ought not substitute their judgment for that of the agency. Id. at 87, 95 S.Ct. 1470. Thus, in the absence of statutory language or legislative history indicating that compliance with state water quality standards should be deemed to satisfy the requirements of § 316(a), we think EPA’s position is reasonable. Accordingly, we decline to set aside 40 CFR Part 122 implementing § 316(a) of the Act. (i) Rainfall Runoff Regulations EPA’s regulations also limit suspended solids in rainfall runoff from areas disturbed by construction activity or used for “material storage” to 50 mg./l. Industry challenges these limitations on the grounds that (1) they are impermissibly vague and apply to nonpoint source runoff; (2) EPA did not adequately consider control costs; and (3) the record does not support the 50 mg./l limitation. (I) Applicability to Nonpoint Sources Congress consciously distinguished between point source and nonpoint source discharges, giving EPA authority under the Act to regulate only the former. Section 502(14) defines a “point source” as: “[A]ny discernible, confined and discrete conveyance, including but not limited to any pipe, ditch, channel, tunnel, conduit, well, discrete fissure, container, rolling stock, concentrated animal feeding operation or vessel or other floating craft from which pollutants are or may be discharged. Industry agreed throughout the rulemaking that contaminated runoff discharges from coal storage and chemical handling areas fell within this definition and should be subject to reasonable controls. It does not contest such application of the regulations. Contaminated runoff from these limited areas is, according to Industry, ordinarily collected into a “point source” and can reasonably be treated so as to remove any pollutants. Industry asserts, however, that EPA’s material storage runoff regulations purport to cover not only point sources such as coal storage areas, but also nonpoint sources as well. Rainfall runoff from areas such as those used to store construction material is, according to Industry, not normally routed into a “point source” collection system, and it is Industry’s contention that EPA has no authority to compel such collection so as to transform a nonpoint source into a point source. EPA, on the other hand, argues that runoff from material storage sites, where channeled into a settling pond or other such collection system, is clearly subject to regulation. Thus, it takes the position that to exempt uncollected runoff from regulation would be to permit pollution by indirection which would otherwise be barred. There is some logic in EPA’s position, and we do not dismiss it lightly. Yet, Congress has limited the definition of “point source” to “any discernible, confined or discrete conveyance.” Broad though this definition may be, we are of opinion that it does not include unchanneled and uncollected surface waters. Industry also argues that while EPA’s construction runoff regulations do purport to limit themselves to “point source” application, they fail to define the discharges to which they apply. This difficulty is allegedly brought about by the regulations’ failure to define the area which they cover other than to say they include point source rainfall runoff from “any construction activity and any earth surface disturbed by such activity” related to power plants. We agree. It is impossible to determine from the regulations whether the construction, for example, of chemical treatment plants, sewage lines, fuel storage and transportation facilities or other such units are subject to control. The only indication as to the breadth of these regulations is found in EPA’s Development Document which states: “Rainfall runoff waste water sources include material storage drainage and runoff from construction activities. Construction activities include only those in the immediate vicinity of the generating unit(s) and related equipment. Runoff from other parts of the site (land and future generating units, construction of access roads, cooling ponds and lakes, visitor centers, etc.) is not intended to be covered by these limitations.” This limitation upon the applicability of the rainfall runoff controls is, however, not found in EPA’s published regulations. While the regulation is limited to point sources, the all-embracing aspect is not supported by the development document, the limits in which cannot be deemed to provide the guidance necessary for either the enforcing authorities who must apply these regulations on a case-by-case basis or the dischargees who must conform their activities accordingly. We are of opinion that EPA’s rainfall runoff regulations must be set aside and remanded with directions that EPA clarify the scope of their applicability and properly limit any subsequent controls to “point sources” only. (2) EPA’s Consideration of Costs Industry also argues that EPA, in arriving at its limitations, did not adequately consider the cost of controlling ash pile and construction site runoff. It appears from the record that in estimating the costs associated with such control, EPA relied entirely upon data relative to coal pile runoff. The agency concedes this fact but argues that, insofar as the regulations apply to ash piles, the omission of the data is immaterial due to their rarity. If the agency by regulation controls the runoff from ash piles (however rare) and construction activity (concededly common), it must consider the cost. That is the command of the statute. §§ 304(b)(1)(B); 304(b)(2)(A); 306(b)(1)(B). EPA’s cost analysis is also deficient for another related reason. EPA’s new source standards for disposal of fly ash effectively require the abandonment of the method of hydraulic transport of fly ash to ponds for settling by prohibiting any “discharge of TSS or oil or grease in fly ash transport water.” The practical effect of this standard on coal-fired generating units, according to Industry, is to require all such plants to utilize dry fly ash transport systems. Such systems produce large amounts of ash which will require storage — over 1000 tons per day at a single plant. Given EPA’s observation that most new generating units will be either coal or nuclear facilities, it would appear that the costs of the required controls applicable to ash piles will be significant. Without the use of hydraulic transport for fly ash, it is obvious that the ash pile which EPA now describes as a rarity will become common in the immediate future, especially at coal burning installations. Indeed, it is mandatory that something be done with the ash. And so long as it is an acceptable alternative (as it must be since EPA wishes to control it), the cost of its maintenance and control must be considered. Industry further contends that, in relation to the construction site runoff limitations, a 3,000 MW plant will require, exclusive of any coal storage area, 100 to 400 acres of land, much of which will be the location of some construction activity. Collecting and treating runoff from such large areas, according to Industry, will cost far more than EPA’s coal pile cost estimate. Even if we were to assume that Industry has over-estimated the area affected by the regulations as well as the associated costs, we are nevertheless of the opinion that EPA has failed to offer any data to support its conclusion that the cost of treating construction site runoff is comparable to that required in controlling coal piles. Accordingly, upon reconsideration of its rainfall runoff regulations, EPA is directed to evaluate the associated cost of controlling ash pile and construction site runoff (as well as the maintenance of ash piles which is later discussed) before establishing any new discharge limitation. (3) Evidentiary Basis for the 50 mg./l Limitation Finally, Industry takes the position that the record offers no support for EPA’s conclusion that the 50 mg./l limit on suspended solids in rainfall runoff from construction activity or material storage areas (other than coal piles) is attainable by the treatment method identified by the agency as “best available.” We agree. As has been noted, EPA did not evaluate any “controlled area” relative to construction site or ash pile runoff. While there is data in the record concerning control techniques for coal pile runoff, there is no indication that such technology is feasible in the context of construction site or ash pile runoff. EPA states “[t]he rainfall runoff limitation was derived from runoff studies done for the Agency” citing a March 1975 study at a Pennsylvania strip mine. Although the study shows a 92.8% efficiency in removal of solids, it is not available support for the regulations because it was made after the regulations were promulgated. Tanners Council, 540 F.2d 1188, p. 1191, Nos. 74-1740, et al. (4th Cir. 1976). Industry additionally argues that even with 92.8% efficiency, if the runoff level of 10,000 mg./l TSS were encountered as was demonstrated at a Washington mine, the TSS discharged would be 720 mg./l, or more than 14 times the acceptable level of the regulations. EPA, of course, had no opportunity to reply to this argument, yet enough has been shown to require a remand for reconsideration of the whole problem. We add that no data is in the record on TSS concentrations from ash pile runoff or control of that problem. We express no opinion as to whether, in fact, the limitation is ap