State of Texas v. Environmental Protection Agency

499 F.2d 289 6 ERC 1897, 4 Envtl. L. Rep. 20,744 STATE OF TEXAS et al., Petitioners, v. ENVIRONMENTAL PROTECTION AGENCY, Respondent. No. 73-3540. United States Court of Appeals, Fifth Circuit. Aug. 5, 1974. John L. Hill, Atty. Gen., Larry F. York, First Asst. Atty. Gen., Mike Willatt, Philip K. Maxwell, Asst. Attys. Gen., Austin, Tex., for State of Tex. Joe Resweber, County Atty., Charles J. Wilson, Asst. County Atty., Houston, Tex., for County of Harris, Tex. William E. Schweinle, Jr., Jessie P. Luton, Jr., W. B. Edwards, Houston, Tex., for gulf Oil Corp. Thomas M. Phillips, Larry B. Feldcamp, James A. Evans, Houston, Tex., for Shell Oil Co. George W. Rice, Houston, Tex., for Armco Steel Corp. Melvin A. Dow, Brian C. Rider, Houston, Tex., for International Council of shopping Centers, Texas/Oklahoma Environmental Action Committee; and Southway shopping Center. Thomas M. Phillips, Larry B. Feldcamp, Dillard W. Baker, Jess W. Van Ert, Exxon Corp., Houston, Tex., for Exxon Corp. Benjamin H. Powell, Jr., Wayne R. Rodgers, Houston, Tex., for Brown and Root, Inc. Thomas M. Phillips, Larry B. Feldcamp, Houston, Tex., for Petro-Tex Chemical Corp., Amoco Oil Co., William Marsh Rice University and Champion International Corp. Edward J. Landry, Senior Asst. County Atty., Oliver J. Guiberteau, Charles J. Wilson, Asst. County Attys., Houston, Tex., for Harris County Hospital Dist. J. F. Hulse, W. A. Thurmond, El Paso, Tex., for Chevron Oil Co., dba Standard Oil Co. of Texas et al. Thomas M. Phillips, Larry B. Feldcamp, Houston, Tex., Alan Griswood, George W. Gray, III, Dallas, Tex., for Sears, Roebuck & Co. Thomas M. Phillips, Larry B. Feldcamp, William G. Christian, Jr., Houston, Tex., for the Prudential Ins. co. of America. Thomas M. Phillips, Larry B. Feldcamp, Houston, Tex., James R. Coffee, Legal Div., Dallas, Tex., for Atlantic Richfield Co. William S. Clarke, Houston, Tex., for Texaco, Inc. Hugh C. Wilfong, II, Norman D. Radford, Jr., Samuel P. Jordan, Houston, Tex., for the Dow Chemical co. James W. McCartney, Norman D. Radford, Houston, Tex., for Houston Center Corp. Hugh C. Wilfong, II, Norman D. Radford, Jr., Harry M. Reasoner, David R. Knoll, Houston, Tex., for American General Ins. Co. Hugh M. Patterson, Thomas M. Phillips, Larry B. Feldcamp, Robert E. Morse, Jr., Hugh C. Wilfong, II, Norman D. Radford, Jr., Houston, Tex., for Texas Chemical Council. F. William Colburn, Houston, Tex., for Port of Houston Authority of Harris County, Tex. Edward P. Lobman, Curtis R. Boisfontaine, Owen A. Neff, Burt K. Carnahan, New Orleans, La., Walter A. Bossert, Jr., William J. O'Brien, II, Pennie & Edmonds, New York City, for Associated Dry Goods Corp. David J. Toomey, John H. Andrew, Allan D. Shafter, Pennie & Edmonds, New York City, for J. C. Penney Co., Inc. Karl J. Bemesderfer, Chicago, Ill., for Montgomery Ward & Co., Inc. Russell E. Train, Administrator, E.P.A., Wallace H. Johnson, Asst. Atty. Gen., Edmund B. Clark, George R. Hyde, Attys., Dept. of Justice, Washington, D.C., Arthur W. Busch, Administrator, E.P.A., Region VI, Dallas, Tex., John Bonine, E.P.A., Alan G. Kirk, II, Acting Asst. Adm. for Enforcement and Gen. Counsel, William B. Saxbe, U.S. Atty. Gen., Dept. of Justice, Washington, D.C., Charles R. Barden, Executive Director, Texas Air Control Bd., Austin, Tex., for Environmental Protection Agency. Before BELL and CLARK, Circuit Judges, and BOYLE, District Judge. BELL, Circuit Judge: — 1 — This case requires that we resolve a major dispute between the Environmental Protection Agency and the State of Texas, several of that state's political subdivisions, and numerous of its corporate citizens. The difficulties arose after the EPA determined that Texas' own air quality implementation plan for control of photochemical oxidant pollution1 would not assure the attainment of federal ambient air quality standards by May 31, 1975, as required by Section 110 of the Clean Air Act Amendments of 1970, 42 U.S.C.A. 1857c-5 (1974 Supp.). Pursuant to its statutory authority and duty,2 the EPA disapproved the Texas plan and issued proposed regulations which would, in its opinion, insure attainment of the air quality standard. These regulations adopted the measures contained in Texas' plan, but also imposed additional controls. After holding public hearings, the EPA promulgated final regulations which in some respects reflected concerns and priorities revealed at the hearings, but which nonetheless imposed on five of the six involved air quality control regions (AQCR's)3 requirements more severe than those in the Texas Plan. See 38 Fed.Reg. 30633-51 (Nov. 6, 1973). — 2 — Twenty-five separate petitions for review were timely filed, and all have been consolidated into the case before us. On Feb. 28, 1974, a panel of this court granted a stay pending appeal as to those portions of the EPA's regulations which go beyond the state's proposed plan. That panel also ordered that this appeal be heard on an expedited basis. Since the measures in the state's own plan are not in dispute and have not been stayed, this litigation has not delayed their effectiveness. As we will discuss specifically at several points in this opinion, these state controls account for the bulk of the emissions reductions needed to meet the oxidant standard. — 3 — The principal questions presented are whether it was proper for the EPA first to determine that the state's plan was inadequate, and then to promulgate its own plan incorporating more onerous regulations. While a negative answer to the first question would settle the second question, an affirmative answer to the first issue is not similarly dispositive. It is possible that the EPA could properly determine that Texas' plan would not insure attainment of quality standards, but would be arbitrary, capricious, or beyond its authority, in its determination of the nature and quantity of additional control measures. We hold that the EPA determined in a legal and enforceable manner that the state's plan was inadequate. However, we also hold that certain of the EPA's additional regulations are either invalid or must be deferred for further agency consideration. The specific effects of our various holdings, on a region-by-region and regulation-by-regulation basis, are summarized at the close of the opinion. — 4 — The factual issues are numerous and complex. We will therefore first outline the elements of the dispute. We will then address ourselves to the standard of review applicable in this case, and will then proceed to a seriatim discussion of the several contentions raised by petitioners. THE DISPUTE IN OUTLINE — 5 — Determining the measures required to meet air quality standards for photochemical oxidants involves a multistep analytical process. Initially, of course, it is necessary to set the standards. These however may not be challenged in this proceeding,4 and petitioners do not openly do so. The second step, since the control strategy for oxidant pollution relies on control of hydrocarbon emissions, see note 1 supra, is to determine the relationship between the quantity of those emissions and the resultant maximum level of oxidants.5 This relationship can be expressed as a 'reduction model' which shows, for each current maximum level of oxidant pollution, the percentage reduction in hydrocarbon emissions that is necessary to achieve the oxidant air quality standard. These models are normally displayed in graphical form, as illustrated in Figure 1. Petitioners raise as an important issue the EPA's choice of reduction model. — 6 — NOTE: OPINION CONTAINS TABLE OR OTHER DATA THAT IS NOT VIEWABLE — 7 — The next step is determination of where on the reduction model curve a given AQCR is located. That is, in order to derive from the model the required reduction in hydrocarbon emissions, the existing level of oxidant pollution must be determined. We note, with some relief, that the parties appear to agree on this point, so that the reduction model issue is the only source of dispute as to how great are the necessary reductions. — 8 — Finally, there must be a prediction of the effect, in terms of reducing hydrocarbon emissions, of each control measure. As will be discussed below, there is considerable controversy on this point, especially as to the state's proposed requirements. — 9 — To summarize, petitioners and the EPA differ as to the model which determines how great a reduciton in hydrocarbon emissions is necessary, and also as to how effective in achieving the required reductions particular measures will be. In general, the state's calculations indicate that the ambient air quality standards can be attained in each AQCR through its Regulation V, imposing extensive controls on stationary sources, such as factories and refineries.6 The EPA's position is that the state not only underestimates the reductions required to meet air quality standards, but also overestimates the reductions to be obtained by its plan. Thus the EPA found it necessary in five of the six regions to impose varying degrees of control on transportation-related emissions, and in one of those regions also to prohibit the construction of additional stationary pollution sources. — 10 — Finally, petitioners attack various of the EPA's required control measures on grounds more specific than their general claim that the state's calculations are correct and its plan adequate. Part IV of this opinion will deal with these more particularized issues, as well as with the related issue whether the EPA erred in its failure to grant Texas a blanket two-year extension of the target date for attainment of the oxidant air quality standard. I. THE SCOPE OF REVIEW — 11 — While the parties do not seriously dispute the basic standard of review, we think it wise for us to state that standard before applying it to the issues before us. The statute provides for review by the appellate courts of the agency's actions with respect to a state's implementation plan, 42 U.S.C.A. 1857h-5(b)(1) (1974 Supp.), but does not specify the standard of review. In this situation courts have been applying the standard set forth in Citizens to Preserve Overton Park v. Volpe, 1971, 401 U.S. 402, 91 S.Ct. 814, 28 L.Ed.2d 136. See Buckeye Power, Inc. v. EPA, 6 Cir., 1973, 481 F.2d 162, 170-171; Appalachian Power Co. v. EPA, 4 Cir., 1973, 477 F.2d 495, 505-507. — 12 — Overton Park similarly dealt with review of agency action which was based upon neither an adjudicatory hearing nor the rulemaking provision of the Administrative Procedure Act, 5 U.S.C.A. 553.7 Likewise, the statute authorizing review did not specify the scope of review. In these circumstances the Court turned to the generally applicable paragraphs of section 706 of the Administrative Procedure Act, 5 U.S.C.A. 706, characterizing these as requiring 'the reviewing court to engage in a substantial inquiry.' 401 U.S. at 415, 91 S.Ct. at 823, 28 L.Ed.2d at 153. Three particular inquiries were specified: (1) whether the action was within the scope of the agency's authority; (2) whether the agency conformed to procedural requirements; and (3) whether the agency decision was 'arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.' 401 U.S. at 415-417, 91 S.Ct. at 823, 28 L.Ed.2d at 153-154. — 13 — The first two of these requirements are relevant in the case sub judice only to the validity of several particular control measures imposed by the EPA to bring the Texas plan up to its standards. The third, however, is of central importance, being the basis for most of petitioners' complaints. We thus will quote the Court's statement as to what it involves: — 14 — 'To make this finding the court must consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment . . .. Although this inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one. The court is not empowered to substitute its judgment for that of the agency.'401 U.S. at 416, 91 S.Ct. at 823, 28 L.Ed.2d at 153. — 15 — In applying this standard, our review must be based not only upon the agency's explication of 'its course of inquiry, its analysis and its reasoning,' Appalachian Power Co. v. EPA, supra, 477 F.2d at 507; Ely v. Velde, 4 Cir., 1971, 451 F.2d 1130, 1138-1139, but also upon the full record before the agency. Overton Park, supra, 401 U.S. at 419, 91 S.Ct. at 825, 28 L.Ed.2d at 155. Only by our own study of the record can we resolve the factual disputes between the parties, much less hope to 'engage in a substantial inquiry' into the agency's action. — 16 — We do add a caveat, however, while the Overton Park mandate does require that we base our review on the entire record before the agency, we do not interpret it to require that we plunge into the record unaided by the parties. The record's length is on the order of 10,000 pages, and it is both technical and poorly organized. Lest we make of this case a career, we must generally restrict our consideration to the parties' specific citations.8 — 17 — II. HOW GREAT MUST THE REDUCTIONS BE? — 18 — Petitioners first challenge to the EPA's analysis concerns its reduction model. As has been stated, this model determines, for each pre-control level of oxidant pollution, how great must be the reduction in hydrocarbon emissions to attain the ambient air quality standard for oxidants. The implementation plan submitted by the state was based on a reduction model requiring, for a significant range of oxidant pollution, considerably lower emission reductions than does the model used by the agency. The result, as shown in Table 1, is that for four of the six AQCR's the EPA determined that the state's reduction target was too low.9 — 19 — As a general matter, the EPA has required that states use the reduction model contained in Appendix J to 40 C.F.R. Sec. 51.14 (1973). However, in submitting its plan Texas utilized a different model on the grounds that it was employing more sophisticated methodology, based not on total hydrocarbons emitted into the atmosphere, but rather upon hydrocarbon emissions which react with nitrogen oxides to form oxidant pollutants.10 Since it is only reactive hydrocarbons which contribute to oxidant pollution, the Texas approach is theoretically attractive. However, reactive hydrocarbons have not yet been the subject of extensive pollution control study, and the resulting lack of knowledge restricts the usefulness of reactive hydrocarbon pollution control plans. One important unknown is the configuration of the appropriate reduction model. — 20 — The EPA concluded that Texas' reduction curve was not supported by technical data, and was based on impermissible assumptions. However, rather than require that Appendix J be applied to the reactive hydrocarbon approach, it determined that the most reasonable model to use in these circumstances would be the 'straight rollback' model. — 21 — This model is essentially a method of coping with inadequate information by assuming that reductions in oxidant pollutants will be proportional to reductions in reactive hydrocarbon emissions. The agency does not contend that the model is an accurate representation of reality, but it does contend that there is no information to support divergences from it, either in the direction of higher reduction requirements, as with Appendix J and part of the Texas curve, see note 14 infra, or in the direction of lower reduction requirements, as in the case of most of the Texas curve. Basically the EPA's position is that in the absence of contrary indications, and given the statutory necessity of accepting some model, the simplistic proportional rollback model is a reasonable resolution. — 22 — We will first consider the agency decision to reject the model on which Texas based its plan. We will then consider the agency's use of the straight rollback model. A. — 23 — In rejecting Texas' reduction model the agency pointed out defects which have not adequately been answered on this appeal. Initially, for all practical purposes the state supplied no theoretical or empirical support for its novel model.11 Upon further inquiry, the EPA learned that the state had modified Appendix J12 by grasping upon the fact that no reliable data exists connecting hydrocarbon emissions and relatively low maximum levels of oxidant pollution (below .13 ppm but above the national standard of .08 ppm). Faced with this same information gap, the EPA had determined, by a method which is not disclosed in the record, that a maximum oxidant level of .08 would be associated with a nonmethane hydrocarbon concentration of .17 ppm. By contrast, the Texas model assumed the appropriate figure to be in the neighborhood of .20 or .21-- the level of hydrocarbons which Appendix J associates with an oxidant pollution level of slightly below .10 ppm.13 The effect of this assumption, for the four regions below .15 oxidant concentration, was to significantly lower the necessary reduction in hydrocarbon emissions,14 as shown in Table 1, supra. This effect, as well as the basic characteristics of the various reduction models relevant to this litigation, is also apparent from Figure 2. — 24 — We can summarize the situation confronting the agency when it chose to reject Texas' reduction model. The model was novel, yet unsupported by data, theory, or even meaningful explanations. It purported to be appropriate for nonreactive hydrocarbons, yet was explicitly based on data for nonmethane hydrocarbons.15 Finally, it was significantly different from the model the EPA derived from the same data, and even if the EPA model was not known to be accurate, it was part of a promulgated regulation which had been subjected to prior public scrutiny and comment. In these circumstances we cannot conclude that the agency's rejection of Texas' model was arbitrary or capricious. B. — 25 — We must now consider the EPA's use of a straight rollback model, which model it acknowledges to be based upon a simplistic assumption rather than experimental data or proven theory. While we are uneasy that such an unreliable construct should be the basis of wide-ranging government action, we do not consider the use of the straight rollback model to be arbitrary and capricious. — 26 — The statute requires implementation plans which will insure attainment of the national air quality standards. To design such plans it is necessary to utilize some model relating reductions in antecedent emissions to resulting reductions in pollutant concentrations. in the absence of sophisticated information, the EPA has been forced to rely on crude assumptions. We cannot object, for it is not our role to judge whether the EPA's projections are accurate, but only whether they represent arbitrary or capricious exercises of its authority. Necessity, which has mothered the EPA's invention of this model, also protects it from a judicial insistence on greater reliability.16 — 27 — Further, we think it significant that the straight rollback model is, in an important sense, neutral, in that it establishes as a starting point the commonsensical proposition that pollutants will be reduced proportionally to reductions in their chemical precursors. Deviations from this assumption are to be permitted or required only when supported by empirical data or sound theory. Such a stance is not arbitrary and capricious, and is readily accepted by Texas for maximum oxidant concentrations above about .28 ppm (as discussed in note 14 supra, for these high pollution levels the EPA permits states to abandon Appendix J and use the much less stringent straight rollback model; it does so not only to avoid disruptive reduction requirements, but also because no solid data relates oxidants and hydrocarbon emissions for these high levels). — 28 — To summarize, we conclude that the EPA was not arbitrary or capricious in its decisions to reject Texas' proposed reduction model and to substitute a straight rollback model. By this conclusion we support the agency's determination that higher emissions reduction targets are required in Regions 3, 8, 9, and 11, as indicated by Table 1 above. III. REDUCTIONS TO BE OBTAINED — 29 — The second broad source of controversy between the state and EPA concerns the reduction in hydrocarbon emissions expected from the state's control of stationary sources. While the EPA's analysis exacerbates the problem in Regions 3, 8, 9, and 11 (in which the EPA reduction model in any case requires higher reduction targets), its impact is at its greatest in Region 7, the Houston ACQR. In Region 7 both the state and the EPA used a proportional rollback model,17 and they thus agreed that the implementation plan must insure a 75 per cent reduction in reactive hydrocarbon emissions. However, they do not agree on the reduction to be obtained from Texas' Regulation V, which controls stationary sources. While Texas believes this will result in a 75 per cent reduction, the EPA puts the figure at 65.5 per cent. Thus the EPA has required additional control measures aimed principally at transportation-related emissions. — 30 — As with the case of the appropriate reduction model, this controversy has its roots in Texas' reactive hydrocarbon approach. The problem arises in determining the reactive component of total hydrocarbon emissions from each emission category. In general, the agency has determined that reactive hydrocarbons constitute a smaller percentage of total stationary source emissions than has the state, and thus that stationary sources are responsible for a smaller percentage of overall reactive emissions. For example, in Region 7 the EPA calculates that 294,000 tons per year of reactive hydrocarbons are emitted from stationary sources, and 106,000 tons per year from mobile sources. This compares with the state's calculations of 481,000 and 99,000, respectively.18 Translated into percentages, the EPA attributes 73 per cent of total reactive emissions to stationary sources, while the state attributes 83 per cent. The result of the EPA's calculations is that the stat's highly effective Regulation V, unaided by transportation-related controls, does not affect a sufficiently large proportion of the total to achieve the 75 per cent overall reduction. — 31 — As we did in the case of the reduction model controversy, we will approach the question in two stages. We will first consider the agency's rejection of Texas' calculations, and then consider petitioners' challenges to the calculations actually used by the EPA. A. — 32 — The state's plan, as originally submitted, was quite summary in nature. It simply stated existing and projected reactive hydrocarbon emissions. The only explanation was by way of reference to a study entitled 'Hydrocarbon Control Strategies for the State of Texas,' Document 4 of the record. — 33 — The referenced study to some extent does deal with determining the reactive component of each source category's total hydrocarbon emissions. Of critical importance are the chemical processing and refinery categories. The study states, 'For fixed chemical sources the reactive emissions were determined from a review of the inventory, source by source, for the significant emitters. For refinery sources the basis from the Radian Report was used.' — 34 — Bolstering Texas' claim, that it determined the reactive component of chemical sources by separately considering each source, is a rather confusing table which lists plant-by-plant emissions for the Harris County chemical industry.19 For large sources the table contains three columns, with the headings 'Emission Inventory,' 'Adjusted Inventory,' and 'Reduced Inventory.' Nowhere does the study define the significance of each column, although it is clear that 'Reduced Inventory' is the final reactive inventory. While a note to Texas' table seems to indicate that many of the 'Emission Inventory' entries were already reduced for nonreactive hydrocarbons, it is by and large only these entries which show reductions as one moves from the 'Emission Inventory' column to the 'Reduced Inventory' column. — 35 — An additional difficulty with the source-by-source data in the supporting study's tables is that nowhere is it specified whether the tables contain 1970 raw data or adjusted data for 1972.20 The result is to restrict even further the meaning and usefulness of these tables. — 36 — So far as we can determine, Texas never presented to the agency additional material domonstrating how its source-by-source survey was conducted and why it should be accepted as reliable. Further, the results produced by Texas' methodology were suspect by comparison with existing data from other states. Whereas studies of chemical industries in California and Louisiana showed reactive htdrocarbons to constitute about one-half of total hydrocarbon emissions, Texas' Harris County 'Reduced Inventory' equaled 86 per cent of its 'Emission Inventory.'21 — 37 — We conclude that the EPA was not arbitrary or capricious when it rejected Texas' chemical processing reactive hydrocarbon inventory. The figures submitted by Texas were not only apparently quite different from those expected by the EPA, but they were unsupported by intelligible explanations or source material. We recognize the possibility that we are doing Texas an injustice by this decision, for it may be that in fact its calculations can be supported. However, this is a factual matter subject to proof. If Texas is correct we are confident that it can present its supporting data in a sound, orderly and convincing manner.22 No such presentation having been made in this proceeding, we will not order the EPA to accept results which are apparently at odds with existing knowledge of the reactive/nonreactive makeup of chemical plant hydrocarbon emissions. — 38 — When we turn to the state's data regarding refineries, we find that it is even sketchier and more conclusory than that for chemical processing. As quoted above, the study supporting the Texas plan stated that to determine the reactive component of hydrocarbon emissions it had used 'the basis from the Radian Report,' a study presented to the EPA in 1972 by a Texas consulting organization. The use in this context of that report is difficult to understand. As far as we can tell, it deals strictly with total hydrocarbons and provides neither data nor methodology for calculating the reactive component.23 While the supporting study (Document 4) does include a table which apparently purports to list the reactive emissions of each Harris County refinery, there is no explanatory or source material provided.24 — 39 — As with chemical plants, the Texas reactive hydrocarbon inventory for refineries is not only supported by the sketchiest and most conclusory of data, but it yields results quite different from those anticipated by the EPA on the basis of its knowledge of similar sources in other states. Whereas the EPA calculates that all refineries in Region 7 emit 17,000 tons per year of reactive hydrocarbons (12 per cent of total hydrocarbon emissions of 142,000 tons),25 the state apparently claimed some 131,000 tons for Harris County refineries alone. — 40 — We conclude that the EPA was not arbitrary or capricious when it rejected Texas' calculations for refinery emissions. This is not to say that Texas' very high figure is incorrect. Rather, our decision turns on Texas' failure to provide persuasive support for data which was inconsistent with results in other areas. — 41 — Given our disposition of the chemical processing and refinery categories, which are the principal sources of disagreement between the state and the EPA as to reactive hydrocarbon inventories, we think it unnecessary to consider other, minor, EPA objections to Texas' data. We turn now to the issue whether the EPA, having reasonably rejected Texas' inventory calculations, was arbitrary or capricious in its own calculations of the inventories on which it based the promulgated regulations. B. — 42 — Since no studies have inventoried the actual reactive content of specific emission sources in Texas,26 the EPA determined the reactive component by applying a 'reactivity factor' to total hydrocarbons. The reactivity factor is a number, running from 0 to 1.0, which indicates the degree of reactivity of a source's hydrocarbon emissions. As shown in Table 2, the factors vary widely, both between different types of emission sources and between the EPA's and state's estimates for identical sources. In obtaining its reactivity factors, the EPA looked to 'the relationship between reactive and non-reactive components for identical processes and sources in other areas of the country.' 38 Fed.Reg. 30634 (Nov. 6, 1973). — 43 — Before this court petitioners claim that the agency was arbitrary and capricious in its methods of constructing hydrocarbon inventories. Principally, this claim arises from the contention that the EPA displaced the state's source-by-source study of the very emission sources to be controlled, and substituted data based on general reactivity factors derived in other parts of the country and applicable only to broad categories of emission sources; petitioners further contend that not only did the EPA use improper reactivity factors, but it also applied them to inventories which had already been reduced to eliminate many nonreactive hydrocarbons. The initial portion of this contention is nothing more than the claim that the state's reactive inventories should have been accepted, an issue just dealt with and which need not be discussed further. The second portion questions the validity of the EPA's reactivity factors and the appropriateness of the inventories to which the factors were applied, and must now be considered. From the briefs it appears that petitioners genuinely challenge only the resulting inventories for chemical processing and petroleum refining. — 44 — 1. The Chemical Processing Reactive Inventory — 45 — As indicated in Table 2, the EPA applied a reactivity factor of 0.6 to chemical processing emissions. This figure was derived from empirical studies of chemical industries in Los Angeles and Louisiana, studies which yielded factors of .52 and .47 respectively.27 The EPA used a 20 per cent higher figure, which in this context benefits Texas, to give the state 'every benefit of the doubt' and to account for having eliminated two minor sources of methane from its Region 7 total hydrocarbon inventory. See note 30 infra. — 46 — Petitioners' only specific criticism of the 0.6 figure is that because the emissions from individual chemical plants vary considerably in reactivity, 'it is unjustified to characterize an entire industry by a single reactivity factor.'28 Even granting that the reactivity factor for particular plants may be quite different from 0.6, we think it not unreasonable for the EPA to adopt that figure on an industry-wide basis. The figure 0.6 represents an average, and accounts for the fact that the individual figures which go into that average may be higher or lower. So long as 0.6 is applied to the sum of total hydrocarbon emissions from a representative array of chemical plants, its use is proper. Noting that petitioners present no empirical or theoretical material which suggests that typical emissions from the Texas chemical industry are substantially more reactive than those in Los Angeles and Louisiana, we conclude that the agency's use of 0.6 was not arbitrary or capricious. — 47 — However, this determination alone does not settle the propriety of the EPA's final reactive inventory for chemical processing. Under the agency's methodology reactive inventories are derived by applying the reactivity factor to an initial hydrocarbon inventory which should include the nonreactive hydrocarbons sought to be excluded by use of the factor. Petitioners urge that the EPA has applied its reactivity factor to an inappropriate initial inventory; specifically, they contend that a factor designed for use with total hydrocarbon inventories was applied to an inventory that had already been reduced by the most significant nonreactive hydrocarbons. At stake here is a very large portion of the difference between the EPA's Region 7 stationary source reactive inventory of 294,000 tons/year and the stat's inventory of 481,000 tons/year-- by applying the .6 factor to its Region 7 chemical processing initial inventory (429,000 tons/year), the EPA eliminated 172,000 tons/year from an inventory petitioners claim is wholly reactive. — 48 — To resolve this conflict we have as a result of a post-argument conference obtained from the agency a step-by-step explanation of the method by which it developed its initial chemical processing inventory. This explanation was complete with the actual numbers used at each step, and enabled this court to trace back to the origins of the final result. In this way we have determined that the agency's initial inventory of 429,000 tons/year was not, as the state claims, based on reduced, reactive emission dats. The EPA inventory if basically the sum of total hydrocarbon emissions from each Region 7 chemical plant, adjusted to reflect growth since the original data was assembled.29 This total hydrocarbon data was not reduced on the basis of Texas' own reactive inventories for Harris County plants (submitted with the Texas plan), despite EPA statements which quite understandably led petitioners to believe this had been done.30 In short, the agency applied its 0.6 reactivity factor to an appropriate hydrocarbon inventory. Its use of the resulting reactive inventory was not arbitrary or capricious.31 — 49 — 2. The Petroleum Refining Reactive Inventory — 50 — As in the case of chemical processing the EPA derived its refining reactive inventory by applying a reactivity factor to an original, presumably total, hydrocarbon inventory. However, we need only deal with the first element of this procedure to determine that the agency must reconsider this portion of its Texas calculations, for we have concluded that the agency's choice of reactivity factors was arbitrary and capricious.32 — 51 — In choosing its factor the EPA again looked to Lousiana and Los Angeles data. The factors for these areas were .18 and .12 respectively, and the EPA chose .12 for Texas. The agency has not explained why the lower figure was chosen and no 'benefit of the doubt' was given. — 52 — The record reveals that at least the following objections to the .12 factor were made before the agency: (1) that the Los Angeles study was outdated, being based on 1955-58 data and preceding both significant changes in refinery technology and considerable additional emissions inventory research; (2) that Texas refineries have a product mix different from those on the west coast, and are more heavily oriented toward highly reactive substances; (3) the bulk of Texas refinery emissions are gasoline vapors from tank farms, and thus the .12 factor is inconsistent with the .65 factor used by the EPA for gasoline emissions from other sources. — 53 — The only EPA response to any of these criticisms is directed to the third point. The agency denies that there is any inconsistency by explaining that Texas refinery emissions include little gasoline, and that most are low-reactivity emissions from refinery feed stock and products other than gasoline. According to the EPA, the low gasoline content is due to the fact that in Texas gasoline evaporation is already kept at a minimum by the use of storage tanks with floating roofs which raise or lower to accommodate the expansion or contraction of gasoline vapors. — 54 — We believe that the state's objections deserve more consideration, and their rejection more explication, than has been accorded by the EPA. We are strengthened in this view by several further points. First, the state submitted data raising the possibility that the EPA's reactive refinery emission inventories were grossly low. For one thing, although the state generally did not break its inventories down by source category,33 it did supply what apparently purported to be a reactive refinery inventory for Harris County.34 As has already been noted in this opinion, that figure was 131,000 tons/year, compared to the EPA's reactive inventory of 17,000 tons/year for all of Region 7 (Region 7 contains three refineries located outside of Harris County). While the state's inventory was not bolstered by sufficient data or explanation to require agency acceptance for all purposes, see pages 303-304, supra, the magnitude of the difference should have alerted the EPA to the possibility that the state's objections to .12 were well-founded. The same alert should have resulted from Texas data showing that one Region 7 refinery had a reactivity factor of .26--35 given that there are only ten refineries in Region 7, the large difference between the one refinery's factor and the EPA's presumed average factor required consideration and explanation. — 55 — Our consideration of whether the agency has been arbitrary and capricious in developing its refinery reactive inventories is affected by a background factor. This factor is the significance of those inventories to the agency's calculations and resulting control requirements. Our analysis indicates that the principal difference between the agency's and the state's Region 7 stationary source inventories arises right here. That difference is 187,000 tons/year (481,000-294,000). Data submitted with the state's plan shows that the difference between EPA's Region 7 reactive inventory (17,000 tons/year) and the apparent state Harris County reactive refinery inventory (131,000 tons/year) accounts for 114,000 tons/year; presumably a significant additional portion is accounted for by emissions from the Region 7 refineries which are not in Harris County and thus do not contribute to the state's 131,000 figure.36 — 56 — Further indication of the importance of the EPA's choice of petroleum refining reactivity factor is the effect a different factor would have on the additional controls required in Region 7. According to our own calculations,37 the use of .18 would eliminate the need for most, and .26 for all, of the transportation controls; the use of .65 would relieve Region 7 of both land-use and transportation controls.38 — 57 — In these circumstances it was arbitrary and capricious for the EPA to disregard Texas' objections to its refinery reactivity factor. Those objections must be considered and answered before regulations based upon that factor may go into effect. We now turn to the question of exactly which regulations are so based. In making these determinations we have calculated the effect in each region, in terms of reduced need for additional control measures, of higher refinery reactivity factors. Our computation method is discussed in the margin.39 Region 3 (Austin Waco) — 58 — Since the Region 3 reactive hydrocarbon emission inventory contains no refinery emissions, no regulations in this region are affected by the .12 factor. Region 7 (Houston-Galveston) — 59 — We calculate that a .65 reactivity factor would increase the effectiveness of Texas' plan by 2.4 per cent, so measures resulting in at least this great a reduction must be held in abeyance pending agency reconsideration.40 Since land-use and transportation controls are to be employed only when 'necessary,' 42 U.S.C.A. 1857c-5(a)(2)(B), and not in place of available direct emission source limitations, Natural Resources Defense Council, Inc. v. EPA, 5 Cir., 1974, 489 F.2d 390, 406-409, regulations 52.2292 through 52.2297 must supply the first 1.5 per cent. The remaining 0.9 per cent may be supplied, for the time being, by regulation 52.2288 (vapor recovery during vehicle fueling), because we are in any case requiring the agency to reconsider that regulation. See pages 315-316 infra. However, the choice of measures to defer is ultimately for the agency. Should it complete its reconsideration of 52.2288 prior to reconsideration of the reactivity factor, and should it wish to repromulgate that regulation, it may select for deferral other regulations contributing reductions of at least 0.9 per cent. Region 8 (Dallas-Fort Worth) — 60 — Refinery emissions are a small part of total hydrocarbon emissions in this region, and we calculate that even a .65 factor would improve 1977 performance of Texas' plan by only 0.4 per cent. This figure is well under the reduction to be obtained by the least effective of the EPA's additional controls. We therefore think it unnecessary to defer any agency regulations pending final disposition of the reactivity factor controversy. In light of the small figure involved, and of the liberality of a .65 reactivity factor, and in light of our disposition, infra, of the gasoline marketing regulations in this region, it is more appropriate for any effect of a larger factor to be accounted for in Texas-initiated agency proceedings to modify the implementation plan which we herein approve. See 42 U.S.C.A. 1857c5(a)(3). Region 9 (San Antonio) — 61 — Calculations based on a .65 reactivity factor increase the Texas plan's 1977 reductions by 2.6 per cent. The EPA's transportation regulations (52.2293-94, 52.2296-97) are designed to obtain a 4.1 per cent reduction.41 We thus defer the effective date of these controls pending agency reconsideration. The remaining regulations, all of which deal with direct emission sources, need not be deferred. Region 11 (EI Paso) — 62 — A .65 refinery reactivity factor would yield additional reductions from the Texas plan of 6.3 per cent. Transportation controls in the EPA's regulations produce a 3.0 per cent reduction,41 and must be deferred. The EPA plan achieves the balance of the requisite 34 per cent reduction by requiring vapor recovery in gasoline marketing operations (regulations 52.2286, 52.2288). Since a substantial potential for lowered additional requirements exists in this region even after elimination of transportation controls, we believe that the gasoline marketing regulations should also be deferred. In addition, deferral is appropriate for a second reason, that of reconsideration of regulations 52.2286 and 52.2288 in light of the discussion on pages 314-315 and 318, infra. IV. SPECIFIC REGULATIONS — 63 — Our disposition of the refinery reactivity controversy pretermits many but not all of the specific challenges to particular EPA regulations. We will now address the remaining issues of this nature on a region-by-region basis. — 64 — To the extent required by the factual setting and the contentions of petitioners, we will consider two matters for each regulation. The first is whether the agency's promulgation of a particular regulation exceeds its authority. The second is whether the effective date of a particular regulation should have been postponed until May 31, 1977, in accordance with the state's request for the two-year extension authorized by 42 U.S.C.A. 1857c-5(e). — 65 — While subsection (e) does not compromise the Clean Air Act's absolute command that the national ambient air standards be met by 1977, it does provide some flexibility for the period between 1975 and 1977. As we read the subsection it recognizes that technological advances by 1977 might render unnecessary, as of that date, onerous control measures which are needed to meet the standards by 1975. Subsection (e) provides a mechanism for subjecting such measures to a reasonableness test. This serves, in our opinion, to protect the public against wastefully heavy expenditures on control measures which by 1977 would be outdated and unnecessary. — 66 — To obtain an extension a state must submit an implementation plan theoretically meeting the 1975 target, but relying on control of emission sources which in fact are unable to comply because 'the necessary technology or other alternatives are not available' until a later date. — 67 — Further, the state must consider and apply in its plan 'reasonably available alternative means' of attaining the standard by 1975. Before granting the extension, the EPA Administrator must determine that the state's plan satisfies the foregoing requirements, and that it applies by 1975 all control measures included in it for which the technology is available. He must also determine that the plan employs all interim measures that are 'reasonable under the circumstances' for control of those sources whose inability to meet the 1975 target is the basis of the extension. The net effect of these provisions is that the agency may not force42 a state to adopt measures needed during the 1975-1977 period merely because they are available, but only if they also are reasonable. — 68 — In the case before us, the state submitted a plan which would, according to its calculations, attain the standards by 1975, but which placed 'significant emphasis on the reductions to be achieved from Federal controls on motor vehicles.' Letter from the Governor of Texas, dated April 13, 1973, and accompanying the state's plan. Recognizing that the impact of these controls would increase through 1977, see note (6) supra, the governor requested an extension 'in order to take the maximum advantage of the Federal controls on automobiles.' — 69 — The EPA concedes that its automotive emissions standards will result in significantly higher reductions from auto sources by 1977, and that certain of its additional requirements for Texas are needed only if the standard must be met by 1975. We therefore think that the only issue before us with regard to Texas' extension application is whether these 'interim'43 control measures are 'reasonably available.'44 As discussed in Part I of this opinion, we must approach this question in terms of whether it was arbitrary or capricious for the agency to determine that its interim measures were reasonable. Region 3 (Austin-Waco) — 70 — The required reduction in reactive hydrocarbons for this region is 27 per cent. The EPA calculates that the Texas strategy would achieve a 19.4 per cent reduction by 1975. In order to reach 27 per cent the agency has required that Texas Regulation V be extended to additional counties (for a reduction of 3.8 per cent), and that vapor recovery systems be installed to eliminate emissions when transferring gasoline from tank trucks to service station storage tanks (also contributing 3.8 per cent). — 71 — The EPA calculates that the Texas strategy alone would achieve the 27 per cent reduction by early 1976, and that by 1977 it would produce a 30.3 per cent reduction. An agency chart indicates that further reductions would be accomplished in each succeeding year through at least the early 1980's. It is thus clear that the agency's requirements are interim measures which must be subjected to the test of reasonableness. — 72 — In considering this issue we think the starting point must not be simply whether an interim measure is physically available. Rather, it must be cost of each regulation. Otherwise, we cannot understand the statutory distinction between longterm measures, which must be applied before 1977 if 'available,' and interim measures which must be applied only if 'reasonably available.' See 42 U.S.C.A. 1857c-5(e)(1). Cf. Natural Resources Defense Council, Inc. v. EPA, supra, which mandates that economic cost or technical feasibility are always to be subordinate to considerations of public health under the Act, 489 F.2d at 411, but which does not involve the precise issue of interim measures being 'reasonably available' under 1857c-5(e)(1). — 73 — Document 30 informs us that the total cost of vapor recovery in Region 3 would be approximately 3 million dollars (at the $2,000 per station figure given in the commentary to the final promulgation, 38 Fed.Reg. 30637 (Nov. 6, 1973), the cost is close to $5,000,000). Document 30 also states that the extension sion of Regulation V would 'involve a significant addition to capital expenditures by the affected industries.' The agency has considered these costs, and believes them reasonable as compared with costs of physically available alternatives, in terms of emissions reductions obtained per dollar expended. However, we do not think that awareness of total cost and selection of relatively cost-effective measures are sufficient for determining that an interim measure is reasonable. We think that that decision must also involve attention to the fact that the reductions obtained by these expenditures are needed only for a short period of time. The EPA balanced the cost of its regulations against reductions to be obtained throughout the life of a control device. However, it is more appropriate in this circumstance to consider cost in terms of the period of time during which the reductions provided are needed to meet the national standard. — 74 — Viewed in this framework we think it impossible to conclude that the EPA's additional regulations for Region 3 are reasonable, and we conclude that the agency was arbitrary and capricious to deny an extension because of the physical availability of these measures. In reaching this conclusion we are moved not only by the requirement of a substantial investment for a speedup of less than one year in attaining the standards, but by several other factors as well. For one, we take judicial notice of the facts that this is a period in which investment capital is scarce and inflationary pressures are severe. In addition, not only are we currently experiencing a variety of raw material shortages, but the particular control devices necessary to comply with the vapor recovery regulations are themselves in short supply-- indeed, other Texas regions have for this reason been granted an extension until 1976 for implementing vapor recovery.45 — 75 — Our decision does not relieve the agency of the determination required by paragraph (2)(B) of subsection (e). This paragraph specifies that when an extension is granted because adequate control of a particular source is not available by 1975, that source must be subjected to 'such interim measures of control . . . as the Administrator determines to be reasonable under the circumstances.' Here the particular source is motor vehicles. Should the Administrator determine that vehicle controls are reasonable during the post-1975 interim created by the extension, he must promulgate regulations implementing them. The determination of reasonableness should involve consideration of the factors identified above, relating not only to total cost but also to the period of time over which necessary reductions would result from a given expenditure. Region 7 (Houston-Galveston) — 76 — As set forth above, we are deferring the effectiveness of transportation and land-use controls incorporated in EPA's promulgated regulation. We thus need not consider various challenges to them. It is also the case that EPA has granted the two-year extension sought by Texas for Region 7. Therefore, we need only concern ourselves with the emission source regulations which are challenged by petitioners, these being those which require vapor recovery during ship and barge loading and during the filling of vehicle gasoline tanks. Since both of these regulations could go into effect immediately, or at least prior to completion by the agency of its reconsideration of the refinery reactivity factor, see page 311 supra, we will address ourselves to the issues raised. — 77 — The automobile refueling vapor recovery regulation requires, for Region 7, 90 per cent recovery of hydrocarbon vapors. While the regulation itself does not specify, the agency has stated that this goal must be attained for each and every vehicle filled (save for certain types of vehicles which are not compatible with standard control equipment). 'Response to Comments,' April 1974, Document 427, p. 6. This latter requirement precludes reliance on techniques which could meet the 90 per cent goal on as average basis, but not for every vehicle. According to petitioners the effect is to require the use of more complex, expensive and hazardous equipment. — 78 — The agency has not contested that its regulation will likely require the more complex techniques. It does contend that a complex system exists which (1) is virtually ready for on-line use, (2) is safe, and (3) is not excessively expensive. The record is such that the first two determinations are not arbitrary and capricious. However, the third cannot be sustained on the record and EPA explanations. — 79 — The agency's principal counter to petitioners' cost objections is the carbon adsorption system. It considers this system to be only somewhat more expensive than the simple, but less effective, displacement system.46 The agency's technical support document does bolster the proposition that initial capital outlay is not a great deal more ($5,065 versus $6,900 per station for existing stations). Document 30, Part II pp. 5-6 through 5-9. However, at page 5-9 we are told that the cost effectiveness for the simple system is $0.19 per lb. of vapor recovered and 0.14cents per gal. of gas pumped. The carbon adsorption system compares rather poorly, at $0.34 per lb. of vapor recovered and 0.25cents per gal. of gas pumped. Further, the more expensive system is expected to achieve only 3 per cent greater recovery than the displacement system (79 per cent versus 76 per cent in 1977, and 98 per cent versus 95 per cent in 1985). Based on this information Document 30 draws the following conclusion: — 80 — 'From these figures it is obvious that (simple displacement) represents the most efficient technique for motor vehicle vapor recovery. (Carbon adsorption and another complex system) offer additional recoveries but the incremental costs associated with these recoveries is very high.' — 81 — The agency apparently ignored this conclusion, making no reference to it although imposing a regulation which requires a complex system. While such a system may be necessary, despite its high incremental cost, it has not been justified in this rule-making. Further, the agency has also failed to explain why it is unwilling to accept compliance with the 90 per cent target on an average basis-- where the average basis adequate, complex systems apparently would be unnecessary. Thus this regulation must be reconsidered; should the agency choose to repromulgate it without change, there must be satisfying explanations of why an average recovery of 90 per cent is unacceptable, and why some slightly lower target has not been adopted in order to avoid the 'very high' incremental costs of complex systems.47 — 82 — The other regulation which is specifically challenged pertains to the control of hydrocarbon emissions during the loading and unloading of ships and barges using port facilities within Region 7. Certain private petitioners assert that EPA lacks statutory authority with respect to ships and barges. Moreover, they assert, all jurisdiction over the safe operation of marine vessels is vested in the Coast Guard. See 46 U.S.C.A. 391a. — 83 — They also contend that control technology is not available and will not be during the short time allotted for implementation of the requirement. Finally, they allege discrimination in that the Houston-Galveston area has been singled out among all American ports for these emission controls and their possible adverse effect on the use of ports in the region. — 84 — We have carefully considered the language and scheme of the statute as well as the legislative history. This requirement is a part of an implementation plan and such a plan under the Act must assure attainment of ambient standards by employing 'emission limitations, . . . and . . . other measures . . . including, but not limited to, land-use and transportation controls.' 42 U.S.C.A. 1857c-5(a)(2)(B). It is true that vessels were eliminated from the Senate bill by a conference committee, but the elmination occurred in a portion of the bill prescribing the establishment of national uniform standards for new automobiles, aircraft and vessels. See 1970 U.S.Code Cong. & Admin.News, pp. 5356, 5381. This subchapter, appearing in its final form at 42 U.S.C.A. 1857f-1 to 1857f-12, must be distinguished, as a separate element of the overall attack on air pollution, from subchapter I which requires implementation of ambient air quality standards on a region-by-region basis. In particular, the 1857f national standards concern only motive power and associated emissions, not the myriad of emissions which degrade ambient air quality in a particular region, such as those from the loading and unloading of vessels. — 85 — Likewise, there is no merit in the suggestion of a lack of authority due the Coast Guard's responsibility for safety regulations. The commentary to the regulation in question recognizes the necessity of agency/ Coast Guard cooperation in establishing emission controls for marine sources. 38 Fed.Reg. 30636 (Nov. 6, 1973). There is no indication that the EPA will attempt to invade the domain of the Coast Guard or otherwise interfere with or jeopardize its provisions for maritime safety. — 86 — Only a short time was allowed by the EPA for implementing the required controls. The necessary technology is not presently available, although it is suggested by the technology used in controlling evaporative losses during the transfer of gasoline from tank trucks to service station tanks and thence to automobiles. There is no basis whatever in the record for concluding that a compliance date of May 31, 1975 is technologically feasible, and it may well be that this requirement must be postponed. However, we do not believe that this fact makes the regulation arbitrary or capricious, for 42 U.S.C.A. 1857c-5(f) contemplates plans which require control measures of less than certain technological feasability. That section specifies procedures for obtaining a postponement of compliance dates for particular sources, and it is through those procedures that petitioners must seek relief. — 87 — The simple fact that the ship and barge regulation discriminates against Region 7 is irrelevant, for the statutory scheme clearly contemplates that heavily polluted areas will be subject to more burdensome implementation plans. Petitioner's argument does go farther, however, and objects that this discrimination will shut down Region 7's refinery industry. Assuming this objection is relevant, we think it premature. On the present record it was not arbitrary or capricious for the agency to conclude that the region's refineries will continue to attract required shipping resources, albeit at the somewhat higher price required to induce tanker owners to make the necessary adaptations. Should future events indicate that this conclusion is erroneous, or that vapor recovery is technologically impossible, and that Region 7 refineries are threatened with closure, the state may adopt a revision of its plan, under 42 U.S.C.A. 1857c-5(a)(3). Agency approval of such a revision would of course require that other controls be substituted to assure attainment of the national ambient air standard. — 88 — In sum, we do not find the ship and barge regulation to be arbitrary and capricious on the present record. Region 8 (Dallas-Fort Worth) — 89 — For this region a 34 per cent reduction in hydrocarbon emissions is required. According to the EPA's calculations, the Texas strategy would result in a 22.5 per cent reduction by 1975, a 27.7 per cent reduction by 1976 and 33.1 per cent by 1977. According to a chart in Document 30, the progressively more effective federal auto emission controls would by some time in late 1977 boost the reduction under the Texas strategy above 34 per cent, and reductions would continue into the early 1980's. — 90 — To make up the shortfall, the EPA required extension of Regulation V to an additional county (resulting in a 1.8 per cent reduction), and vapor recovery during gasoline marketing operations (contributing 8.1 per cent). However, since equipment for the vapor recovery regulations was not expected to be available for installation by 1975, the region was granted an extension until June 30, 1976. — 91 — Because applying Regulation V to an additional county, plus the Texas strategy, would together produce reductions exceeding the 34 per cent target by 1977, the gasoline marketing regulations would be completely unnecessary were the requested two-year extension granted. In considering whether it was arbitrary and capricious for the agency to deny the extension, we note that denial requires an expenditure in excess of $41,000,000, but that the period during which the benefit of this expenditure is needed is only eleven months (June 30, 1976 to May 31, 1977). — 92 — For the reasons discussed when faced with this issue in Region 3, we conclude that the agency's determination was arbitrary and capricious. The extension should have been granted and Region 8 should not have been subjected to the gasoline marketing vapor recovery regulations. Whether interim vehicle controls should be imposed is a matter for the agency. See 42 U.S.C.A. 1857c-5(e)(2)(B); p. 314-315 supra. The promulgated regulations do contain three interim transportation controls, based on an agency determination of reasonableness and required because of the 13-month extension it granted in this Region. Before these regulations can be effective the agency must reconsider their reasonableness in light of the standards discussed at pages 314-315. — 93 — The remaining additional regulation for Region 8 is extension of Texas Regulation V to Tarrant County. Although it is expensive, and although its effect is needed only for a short period of time after May 31, 1977, the statute's command is absolute that all necessary measures be taken to attain the air standards by that date. Therefore, this court cannot disturb the agency's promulgation of the regulation.48 Region 9 (San Antonio) — 94 — The agency granted the state's request for a two-year extension in this region. Even at the end of that time, and for the foreseeable future, controls in addition to the Texas strategy will be necessary. — 95 — Since we concluded in Part III of this opinion that transportation controls for Region 9 must be deferred pending agency reconsideration of the refinery reactivity factor, we are concerned here only with source controls required in addition to the Texas strategy. These are control of degreasing operations, vapor recovery in gasoline marketing operations, and an inspection and maintenance program for light duty vehicles. The only specific challenges which we have found amongst the myriad of petitioner's briefs are to the regulation requiring vapor recovery during vehicle fueling. This regulation affects Region 9 in exactly the same fashion as it does Region 7. For exactly the same reasons we gave when considering that region, we require similar agency reconsideration. Region 11 (El Paso) — 96 — By virtue of our disposition of the reactivity factor controversy, we deferred the only additional control measures required in this region. As indicated at page 312 supra, these regulations should also be reconsidered in light of our discussion of what constitutes a 'reasonably available' control measure for the purposes of subsection (e). According to EPA calculations, the Texas strategy alone would meet the reduction target by 1977. Further our disposition of the vehicle refueling regulation, in Regions 7 and 9, is applicable in this region. V. OTHER ISSUES — 97 — There remains for consideration a miscellany of matters with which we may briefly deal. — 98 — Certain petitioners have submitted a motion to strike various portions of the record. In particular, their motion is directed at the agency's technical suport Documents 30 and 31, which were compiled and made available only after the regulations were promulgated. If these documents contained new data which was not available when the agency made its determinations, the motion would have merit. However, these documents are statements of agency methodology and compilations of information which was in the record when the determinations were made. For this reason the motion to strike is hereby denied. — 99 — Petitioner Exxon has pressed arguments based upon three facts related to the agency's ozone measurement procedures: (1) that different techniques for measuring photo-chemical oxidant pollution yield, in certain circumstances, widely varying results; (2) that the standard of .08 ppm is not altered for areas which record high background levels of ozone; and (3) that reduction requirements are based on oxidant pollution levels which occur during air stagnation episodes. While the parties present arguments on the merits of these attacks, we need not reach them. Since the challenged procedures comply with, and probably are mandated by, the published national standards, 40 C.F.R. 50.9, petitioner's attack is basically directed at those standards. 42 U.S.C.A. 1857h-5(b) makes it clear that ours is not the proper forum for review of such challenges. — 100 — Realizing the restrictions on raising objections to the national standards, petitioner Exxon has also incorporated its criticisms of the agency's measurement procedures into a somewhat more sophisticated and substantial argument that the agency has been arbitrary and capricious. The argument is that by virtue of the alleged deficiencies, as well as by virtue of the numerous uncertainties involved in determining the appropriate reduction curve and the reductions to be obtained by specific measures, the margin of error of the agency's calculations is so large that it was arbitrary and capricious to impose control measures which contribute only the relatively small reductions specified in note 38 supra. As we have indicated elsewhere in this opinion, we are well aware of the EPA's use of projections, assumptions, and flimsy data. However, we do not think that the margin of error inherent in the agency's calculations precludes its promulgating the regulations which remain in this case after our disposition of the other issues. Our reasons are several. For one, as we have already noted, the agency's use of uncertain data is necessary if it is to perform its statutory duty. In these circumstances we can only require that its data be the best that is feasibly available.49 — 101 — A second reason for rejecting the margin of error argument is that margins of error cut both ways-- the argument provides just as much support for the proposition that the Region 7 target should be 85 per cent as it does for petitioner's proposition that the state's strategy would produce reductions so close to the target that any shortfall should be disregarded. Especially is this relevant when we recognize that at several key points in developing the promulgated regulations the agency accounted counted for the uncertain quality of its data by imposing less stringent requirements than the data might support. For example, the reactivity factor used for chemical processing emissions was .6 rather than .47 or .52, and the agency is satisfied with a 75 per cent target in Houston rather than the much higher target that is suggested by the nonmethane data on which Appendix J is based. See note 14, supra. — 102 — As a final note on the margin of error argument, we point out that our only concern with it is in the context of those control measures which are not deferred pending agency reconsideration of the refinery reactivity factor. That is, we deal with this issue only with regard to control of emissions sources and not as it relates to transportation and land-use controls. Thus we have had no occasion to consider the effect of a margin of error on controls which, after all, are unlike emission source controls in that they may be imposed only 'as may be necessary.' See 42 U.S.C.A. 1857c5(a)(2)(); Natural Resources Defense Council, Inc. v. EPA, supra, 489 F.2d at 406-409. Similarly, we have not had occasion to consider the fact that the agency's reduction targets are apparently parently accurate only to the mearest whole percent, whereas the reductions credited to each measure are calculated to tenths of percent.50 — 103 — The final issues which merit discussion are raised only by Harris County. Several of the agency's promulgated regulations direct the state to institute a particular program, such as an inspection system. This contrasts with the typical regulation which is directed at private individuals, for example, prohibiting them from filling automobile gasoline tanks without employing vapor recovery covery mechanisms. Harris County contends that Congress empowered the agency only to promulgate regulations affecting private parties and to enforce them should the state refuse to do so. It was not intended, claims Harris County, that the agency should have the power to require specific legislative acts by states. The county further contends that even if Congress attempted to delegate such power, doing so would violate constitutional principles of federalism and state sovereignty. — 104 — These are weighty issues. However, they are not properly before this court and we decline to consider them. There are two reasons for this disposition. The first is that these issues were not raised below. The agency was thus accorded no opportunity to evaluate these contentions as to the limitations on its power, and to adapt its regulations accordingly.51 Second, this objection to the regulations has not been raised by the state of Texas. We do not think it appropriate to consider the issue in these circumstances, for we interpret the state's silence as indication of its willingness to adopt the regulations if they are otherwise valid. SUMMARY — 105 — We will now summarize the effect of our various rulings for each region. Where a regulation affects several regions, a statement pertaining to it is relevant only for the region being summarized. Region 3 (Austin-Waco) — 106 — An extension of the date for attaining the national ambient air standard for oxidants shall be granted. The agency's additional regulations, 52.2283 (extension of Texas Regulation V) and 52.2286 (vapor recovery during gasoline marketing operations), are invalid. Region 7 (Houston-Galveston) — 107 — The land-use (52.2292) and transportation (52.2293-52.2297) regulations shall be deferred pending agency reconsideration of its refinery reactivity factor. Additional regulations shall be similarly deferred as necessary to accumulate deferred reductions of at least 0.9 per cent, in accordance with page 311 of this opinion. Subject to the conditions of the preceding sentence, regulations 52.2284 (degreasing operations), 52.2285 (vapor recovery during the filling of service station storage tanks), 52.2287 (ship and barge controls), 52.2290 (vehicle inspection and maintenance), and 52.2291 (retrofit of pre-1968 automobile) are valid and enforceable. Regulation 52.2288 (vapor recovery during vehicle refueling) must be reconsidered before it can take effect. Region 8 (Dallas-Fort Worth) — 108 — An extension of the date for compliance with the national ambient air quality oxidant standard shall be granted for this region. The gasoline marketing regulations (52.2285 and 52.2289) are invalid. Regulation 52.2283, requiring extension of Texas Regulation V to Tarrant County, is valid and enforceable. Retion 9 (San Antonio) — 109 — The transportation regulations (52.2293, 52.2294, 52.2296, and 52.2297) shall be deferred pending agency reconsideration of its refinery reactivity factor. Regulations 52.2284 (degreasing operations), 52.2285 (vapor recovery during the filling of service station storage tanks), and 52.2290 (vehicle inspection and maintenance program) are valid and enforceable. Regulation 52.2288 (vapor recovery during vehicle refueling) must be reconsidered before it can take effect. Region 11 (El Paso) — 110 — The only additional regulations applicable to this Region, 52.2286 and 52.2288 (vapor recovery during gasoline marketing operations) are deferred pending agency reconsideration of its refinery reactivity factor. Whatever may be the result of this process the agency may not repromulgate these regulations prior to reconsidering the state's application for an extension of the date for compliance with the national standard for oxidants. Further, regulation 52.2288 (vapor recovery during vehicle refueling) must itself be reconsidered before it can take effect. — 111 — In Regions 3 and 8 the Administrator shall make the determination required, by 42 U.S.C.A. 1857c-5(e)(2)(B), as a result of our order that extensions be granted. In Region 8 the promulgated interim transportation regulations (52.2294, 52.2296, and 52.2297) shall be reconsidered. — 112 — The Petitions to review are granted in part and denied in part. — 113 — 40 C.F.R. Subject Matter 1977 Reduction -------------------- ------------------------------ -------------------- 52.2284 Degreasing operations 0.6% 52.2285 Vapor recovery during gasoline delivery to storage tanks 1.5 52.2287 Vapor recovery during ship and barge loading 2.7 52.2288 Vapor recovery during automobile refueling 1.3 52.2290 Vehicle inspection and maintenance 0.8 52.2291 Retrofit of pre-1968 automobiles 0.8 52.2292 Restriction on new of modified emission sources 1.2 52.2293 Gasoline rationing 0.1 52.2294 Bus/carpool lanes 0.1 52.2295 Parking supply management 0.1 52.2296 Bus/carpool matching and promotion 0.1 52.2297 Employer-supplied incentives for use of mass transit 0.1 -------------------- 9.5 * — 114 — * Actual total of listed figures differs, due to effect of rounding. — 115 — 52.2284 0.5 52.2285 1.3 52.2287 2.3 52.2288 1.1 52.2290-52.2291 1.3 52.2292 1.0 52.2293-52.2297 0.5 — 116 — CLARK, Circuit Judge, with whom BOYLE, District Judge, joins, specially concurring. — 117 — I concur without reservation in the entirety of Judge Bell's thoroughgoing, able opinion. The sole purpose of this addendum is to highlight the adverse effects flowing from the legislative mandate date that judicial review proceedings be injected into the court system at the appellate level. — 118 — No formal hearing has ever been held in this highly technical, factually complex matter. The administrative 'record' upon which we had to base our review was comprised of only the sparest of documentation, for it essentially evolved from an act of agency rule-making. To accentuate the problem the agency contracted the services of a private firm for the formulation of most of the rule requirements it ultimately adopted here, so that not even intra-agency background for these actions was available. The writing judge was required to hold both pre and post-argument conferences with counsel for the parties to enable the three of us as a court to comprehend the substance of the issues and conduct a minimally meaningful review. — 119 — The subject matter of this action involves the health and welfare of millions of citizens, the continued business vitality of tens of thousands of firms and compliance expenditures costing billions of dollars. — 120 — These extensive rights deserve a more orderly process of judicial reflection. Footnotes [1] Photochemical oxidants constitute one of several classes of pollutants identified in 40 C.F.R. Part 51 for which implementation plans are required, but are the only class at issue in this litigation. They are comprised of various chemical compounds and are responsible for the phenomenon commonly known as 'smog.' These pollutants, which can cause breathing difficulties and eye irritation, derive their name from the process by which they are formed, to wit, oxidation reactions triggered by sunlight and involving nitrogen oxide, hydrocarbons and atmospheric oxygen. Since sunlight and oxygen can't be controlled, and nitrogen oxide is difficult to control, strategies aimed at photochemical oxidants rely on control of hydrocarbon emissions. These emissions occur during the production, storage, transportation and use of a variety of petroleum products. For more detail, see the EPA publication, AP-63, Air Quality Criteria for Photochemical Oxidants [2] 42 USCA 1857-5(c) (1974 Supp.) The structure of the 1970 Amendments, the criteria against which state plans must be considered, and the duties of the EPA have been discussed by this court in Natural Resources Defense Council, Inc. v. EPA, 1974, 489 F.2d 390 at 394-396, (stayed by U.S. Supreme Court June 12, 1974). [3] Texas is divided into 12 AQCR's but we are concerned with only 6. Five of the remainder are classified as Priority III for photochemical oxidants, and thus the state's implementation plan need not assure that they attain the national standard. See 40 C.F.R. 51.3(b)(1)(iii), 51.14. The twelfth region was included in Texas' plan, but was not dealt with in the agency action here under review. The omission occurred because that AQCR also encompasses southern Louisiana, and actions concerning it were delayed until the EPA could consider Louisiana's plan [4] The national ambient air standard for photochemical oxidants is expressed as 0.08 parts per million (ppm) of ozone, this being a maximun one-hour concentration not to be exceeded in more than one hour a year. While ozone is but one of numerous photochemical oxidants, it is a readily measured one and is accepted as a reliable indicator of overall oxidant pollution The standard was promulgated at 36 Fed.Reg. 22384 (Nov. 25, 1971), and is found at 40 C.F.R. 50.9 (1973). Under the provisions of 42 U.S.C.A. 1857h-5(b)(1), petitions for review of a national ambient air quality standard may be filed only in the Court of Appeals for the District of Columbia. [5] For the sake of brevity, we will throughout this opinion refer to 'oxidants' when we mean 'photochemical oxidants,' and to 'maximun' oxidant concentration when we mean the second-highest concentration. The latter refinement reflects the fact that the national standard permits one occurrence per year of a one-hour concentration exceeding 0.08 ppm [6] These calculations assume, as they are permitted to do, 40 C.F.R. 51.14(a) (1), that federal motor vehicle emission standards will result in significant emission reductions independent of any measures contained in the state's plan. Part of Texas' difficulty in achieving the standards by 1975, without relying on transportation-related controls, is that the compliance date for federal automobile emission standards has been extended beyond 1975. Partly as a result of this, and partly because a number of completely uncontrolled (pre-1968) cars will still be operating in 1975, that year is the most difficult one facing Texas. By 1977 a number of the measures required to meet the 1975 deadline will be unnecessary. See table at 38 Fed.Reg. 17800 (July 3, 1973). However, the statute permits an extension of a region's deadline only if the standard can't be met by 'reasonably available alternative means.' 42 U.S.C.A. 1857c-5(e)(1)(B). The EPA considers certain transportation-related controls as being 'reasonably available.' Our own view of this issue is discussed below at pp. 314-315, 317 [7] The rulemaking before us is pursuant to the Clean Air Act Amendments of 1970, 42 U.S.C.A. 1857c-5 (1974 Supp.). Were it APA 553 rulemaking, the applicable standard would be that of substantial evidence. See 5 U.S.C.A. 706(2)(E) [8] To assure that this procedure did not cause us to reject contentions for which support in the record does exist, we accorded counsel for the parties a post-submission opportunity to supply us with relevant citations. This hearing was in addition to oral argument and a pre-argument conference. Cf. Rule 33, F.R.A.P [9] Why the Texas model does not indicate lower reductions for Regions 5 & 7 is explained below in note 14 TABLE 1 ------- PERCENTAGE REDUCTION REQUIRED ----------------------------- Second-Highest EPA Reduction Texas' 1-hour ozone Model (straight Reduction AQCR measurement rollback) Model ---- -------------- --------------- --------- 3 (Austin-Waco) 0.109 27 18 5 (Corpus Christi) 0.184 56 62 7 (HoustonGalveston) 0.320 75 75 8 (Dallas-Fort Worth 0.120 34 24 9 (San Antonio) 0.145 45 40 11 (El Paso) 0.120 34 25 [10] Although the EPA doubts that any hydrocarbon other than methane is completely nonreactive, it accepts that certain other hydrocarbons are sufficiently nonreactive to be disregarded, at least in part, in formulating an implementation plan for Texas 'Reactive hydrocarbons,' which completely excludes the nonreactive component of total hydrocarbons, must be distinguished from 'nonmethane hydrocarbons,' which excludes only the single nonreactive substance methane. While considerable data is available as to the methane component of particular emission sources, and as to the relation between nonmethane hydrocarbons and resultant oxidant pollution, little data exists about reactive hydrocarbons. [11] Material accompanying the state's plan when it was transmitted to the EPA merely stated that, 'A reactive carbon compound control strategy is employed and an oxidant non-methane hydrocarbon relationship was formulated using the publication Air Quality Criteria for Nitrogen Oxide (AP-84) as the besis.' Record, page 4188. It is understandable that this statement, combined with the Texas model, might confuse the EPA for the only relevant data in the cited publication is the very data from which the EPA derived its Appendix J reduction model Texas' brief also cites us to a study entitled 'Hydrocarbon Control Strategies for the State of Texas.' The context of the citation indicates that the study should explain the derivation of the Texas reduction model. We have reviewed this 85-page document, but have found no such material. [12] The state's brief contends that its curve improves on Appendix J because it is derived from data in AP-84 which relates oxidant levels with nonmethane hydrocarbons whereas, according to the state, Appendix J is based on different data in AP-84 relating oxidant levels to total hydrocarbons. The only difficulty with this proposition is that Appendix J also was derived from the AP-84 data dealing with nonmethane hydrocarbons. The state's confusion is excusable, however, since nothing in 42 C.F.R. Part 51 discusses the derivation of Appendix J, and the instructions for its use refer only to 'hydrocarbon emissions reduction,' not 'nonmethane hydrocarbon emissions reduction.' See 42 C.F.R. 51.14(c)(4). Also contributing to the confusion is the EPA's carelessness in distinguishing between total and nonmethane hydrocarbons. For example, in the preamble to the final promulgation of its Texas regulations the EPA states that 'the data base embodied in Appendix J was derived from oxidant and total hydrocarbon areometric data'. 38 Fed.Reg. 30627 (Nov. 6, 1973) We have verified that Appendix J was derived from the nonmethane data in AP-84 by comparing it with that data and also, pursuant to an understanding amongst the parties, by obtaining information from the EPA personnel who actually created Appendix J. [13] On this basis the EPA concludes that the Texas plan in effect sets .10 as the oxidant level to be achieved. This is true only if Appendix J is accurate in setting .17 ppm as the nonmethane hydrocarbon concentration associated with .08 ppm maximum oxidant pollution. On the record before us we are unable to determine whether .17 is accurate, or even more accurate than the state's figure of .20 or .21 (or the .24 used by Harris County in a post-argument submission). However, we need not determine which is most accurate, for our concern is solely whether the agency's rejection of Texas' model was arbitrary and capricious We add a note of caution: our characterization of how Texas derived its reduction model is somewhat speculative, since Texas has not seen fit to explain either its calculations or assumptions. Our understanding of the Texas model is somewhat different from that of the EPA, although it is based on the EPA's version of interviews with state officials, as well as on our own study of the model. It is roughly consistent with the reduction requirements submitted by Texas, and also gives Texas more credit for attention to theoretical niceties than does any alternative. [14] While the difference between .17 and .20 may seem small, it is very significant at the lower hydrocarbon concentrations. For example, if the existing nonmethane hydrocarbon level in a region is .25 ppm, the .17 figure requires a reduction of 32 per cent ((.25-.17) / (.25) = (.08) / (.25) = .32), while .20 requires a reduction of only 20 per cent ((.25-. 20) / (.25) = (.05) / (.25) = .20) That the Texas model does not indicate lower reduction requirements for Regions 5 and 7 may be briefly explained. As the oxidant pollution level increases, the effect of the difference between .17 and .20 becomes less significant, and the Texas curve closely approaches Appendix J (which, after all, is based on the same data). In fact, Texas apparently abandoned any distinction above oxidant levels of .17 ppm. Appendix J requires greater hydrocarbon reductions than does a straight rollback model, and this is especially true above oxidant levels of .15 ppm. The result is that for Region 5 (.184 ppm oxidant level), the Appendix J/Texas model requires a 62 per cent reactive hydrocarbon reduction, while the EPA's straight rollback model requires only 56 per cent. As for Region 7, both models indicate an identical 75 per cent reactive hydrocarbon reduction because, for oxidant concentrations above .28 ppm, both the Appendix J and Texas models adopt the straight rollback approach. The EPA permits this for two reasons: first, little data exists relating these very high oxidant levels to antecedent hydrocarbon concentrations; second, such data as is available, and reasonable extrapolations from data for the lower pollution levels, indicate reduction requirements approaching 100 per cent. [15] As discussed above in note 10, methane is but one of a number of nonreactive hydrocarbons [16] Decisions which are not arbitrary and capricious in the light of existing knowledge may become so by dint of scientific advances. By its use of estimations and sparse data, the EPA creates a continuing responsibility to develop, review and apply updated and more sophisticated information [17] See note 14, supra [18] All data is for 1972, which is the base year for Region 7 [19] While 'Hydrocarbon Control Strategies for the State of Texas' does state that the Harris County source-by-source approach was 'expanded to encompass all affected Regions,' no source-by-source data was included in the study for any other region The record does contain a document entitled 'Report on Hydrocarbon Study for Control Strategy for Industry Point Sources in Region 7' (Document 14). This report is identical to material in 'Hydrocarbon Control Strategies,' except that it employs a source-by-source approach for all of Region 7, not merely Harris County. For some unexplained reason this region-wide material did not accompany Texas' proposed plan and appears not to have been brought to the EPA's attention during its consideration of the region-wide report was prepared some 5 1/2 months prior to the Harris County material submitted with the Texas plan. Our disposition of this case is not affected even if we assume that the EPA should the plan. This is especially strange because have evaluated Texas' plan on the basis of the region-wide report (which apparently was part of the record before it), rather than of the Harris County material actually accompanying the plan. The region-wide report has precisely the same deficiencies as are identified in the text with regard to the Harris County material. [20] The study itself states that Texas obtained its 1972 data by increasing 1970 data 20 per cent (to account for underreporting) and then applying a growth factor. The tables containing the source-by-source information are from a separate report and were appended to the study merely to illustrate the method used by Texas, and do not necessarily state the study's results. Yet these tables, and the explanatory material accompanying them, make no mention of either growth or underreporting adjustments [21] Similar calculations using Document 14, discussed in n. 19, yield 75 per cent. Whether these figures have as much significance as the EPA accorded the 86 per cent figure is open to question. We have already indicated how confusing and poorly-defined are the tables from which they were derived As for the percentage of total chemical processing emissions which Texas actually claimed as reactive, we cannot say. It is indicative of how poorly documented was the state's plan that nowhere do there appear any figures for total hydrocarbons or for reactive hydrocarbons broken down by source category (the data submitted with Texas' plan includes a single, reactive hydrocarbon figure for 'Point Sources' for each region; 'Point Sources' consists of eight source categories, including chemical processing and petroleum refining). In short, it is impossible to determine for each category the percent of total emissions claimed as reactive in the state plan. This is true even when one supplements the state's data with the EPA's own figures for total hydrocarbon emissions from each source category. [22] The method by which Texas may now seek acceptance of its chemical processing inventory is the submission to the EPA of a revision of the plan that results from this litigation. See 42 U.S.C.A. 1857c-5(a)(3) [23] The questionnaire used in the Radian Report did request that companies specify each specie of hydrocarbon emitted. However, none of the material in the Report, at least as it appears in the record, reflects the responses to this portion of the questionnaire [24] Further, we can perceive no consistent correlation between the plant-by-plant total hydrocarbon data in the Radian Report and the supporting study's listing of reactive hydrocarbon emissions for each Harris County refinery-- of the five instances in which a particular refinery appears on each list, in two the study's reactive inventory is slightly larger than Radian's total inventory, and in one it is much larger; in the other two, the Redian inventory is somewhat greater [25] The source of the EPA's figure of 12 per cent is discussed below. See p. 308 [26] This disregards the state's claim that it has performed such a study. As already discussed, it was not arbitrary and capricious for the EPA to reject this data, given the nature and sparseness of supporting material TABLE 2 ------- EPA TEXAS --- ----- Area Sources (dry cleaning, degreasing, painting, etc.) . 2 1.0 Point Sources Chemical processing .6 .86* Petroleum refining .12 ** Ship and barge loading .4 .8 *** Other .6 Transportation Sources Gasoline exhaust .75 .8 Diesel exhaust .45 .8 Aircraft .38 .8 Gasoline evaporation .65 .4 Railroads .45 .8 Vessels .8 .8 Off-highway .45 .8 *Accoring to EPA. But see note 21, supra. As discussed above, the state claims to have used a source-by-source inventory method rather than the EPA's method of applying a reactivity factor to total hydrocarbons. **Neiter stated by Texas nor calculated by the EPA. As discussed above, the state claims to have used a source-by-source inventory method rather than the EPA's method of applying a reactivity factor to total hydrocarbons. ***Accrding to EAP, but not stated as such by Texas. [27] We have inspected these studies to determine that they deal with broadly similar industries. However, we have not inquired into their validity or into how closely matched are the profiles of emissions from the studied industries and those in Texas. While these latter inquiries may be relevant to the reliability of the EPA's calculations, they are not necessary to our determination of whether it has acted arbitrarily and capriciously. These are complex matters which are beyond our ken absent the assistance of the parties. Petitioners have objected only in general terms to the EPA's use of these studies, and the agency has responded appropriately and adequately [28] Statement presented by the Texas Air Control Board, during public hearings on the EPA's proposed regulations, at 10 [29] As stated in note 30, a relatively minor reduction was made for two obvious methane sources. The effect of this was accounted for with the use of a 0.6 reactivity factor, rather than .52 or .47 The post-argument reconstruction of the calculation process yielded a figure 4,900 tons/year higher than that relied on by the agency. Use of the revised calculations would result in a 2,900 tons/year higher stationary sources reactive inventory and about 0.1 per cent lower reduction requirements. Although small, the effect of the error could not ordinarily be disregarded since the EPA seeks to impose control requirements contributing only 0.1 per cent reductions. However, we will not in this instance require agency reconsideration so long as this error is completely offset by another error uncovered during the reconstruction, this one significantly benefiting petitioners. A 36,500 ton/year refinery source was treated as a chemical plant, apparently as a result of carelessness on the part of both the agency and the state. Because the chemical plant reactivity factor was .6, while that for refineries was .12, the error added about 17,500 tons/year to the agency's Region 7 stationary source reactive inventory. [30] The agency's principal original explanatory document contains the following statement: 'The Harris County emissions were updated to reflect the results of a source by source review of petroleum refineries and chemical plants within the county which reduced the inventory for the obvious sources of methane like hydrocarbons, benzene, acetylene and methyl alcohol.' Photochemical Oxidant Control Strategy for Texas (Document 30), part II, p. 2-2 (Record page 3407). Further contributing to the conclusion that the EPA's initial Region 7 inventory was reduced by the most important nonreactive hydrocarbons, at least for Harris County sources, is Document 30's explanation of why a 0.6 reactivity factor was chosen rather than the .52 or .47 figures developed in the Los Angeles and Louisiana studies. 'A reactivity factor of 0.6 was used in this study, since the Texas inventory was adjusted to eliminate obvious methane sources.' Id., at 2-4 (Record page 3409) At the June 12 conference this court received a memo from Jack Divita, Chief of the EPA's Air Program Branch. The memo purports to describe the derivation of EPA's initial chemical processing inventory, and states: 'To ensure completeness this (total hydrocarbon inventory) was compared with the Harris County inventory based on a source-by-source review prepared by the State (Document 4). This inventory was used for purposes of comparison only. The technical support document . . . should not be interpreted to mean that the Harris County inventory included in Document No. 4 was used as the data base by EPA. 'EPA used a reactivity factor of 0.6 for chemical processing. This factor was chosen to give the State every benefit of the doubt . . .. The number of 0.6 was not chosen to account for a non-methane data base.' Subsequently, at the request of this court and pursuant to an understanding amongst the parties, another Divita memo was submitted, this one backed up with worksheets and the computer printout used as the principal source for EPA calculations. The second memo and its supportive material confirm that the agency inventory is essentially total hydrocarbons, although it does modify the first memo in several ways. For one, the agency in fact did use, in place of the printout data, some entries from the source-by-source review in Document 4. However, the entries used were reasonably thought to be total hydrocarbon figures, by virtue of comparisons with the printout's 1970 data; indeed, the net effect of using this data was to increase the agency's chemical processing inventory. Second, in the instance of two carbon black plants total hydrocarbon data was eliminated because it was known to be methane. These reductions, which amount to only 11,000 tons, are offered in the second memo as the explanation for the statement in Document 30 that a 0.6 reactivity factor was used because the inventory 'was adjusted to eliminate obvious methane sources.' Since our review is for the purpose of determining whether an agency determination is arbitrary or capricious, we are constrained to consider the EPA's post hoc statements. After all, they are not the post hoc rationalizations of agency actions which 'have traditionally been found to be an inadequate basis for review.' Citizens to Preserve Overton Park v. Volpe, 1971, 401 U.S. 402, 419, 91 S.Ct. 814, 825, 28 L.Ed.2d 136, 155. Rather, they are statements which establish what the agency did, not why it did it. Were we to disregard them we would be reviewing not the agency's actual action, but rather a different action (further reduction of a reactive hydrocarbon inventory) that it in fact never took. While we are very concerned that both petitioners and this court have been temporarily misled and confused by the agency's original language, we cannot be diverted from our responsibility to review that which the agency actually did. We have of course satisfied ourselves that the impression created by the earlier version is incorrect. Presumably the misleading wording resulted from carelessness, or poor communication between the EPA and the contractor responsible for Document 30. We accept the memos' version because their description is backed up by the actual calculations and worksheets required to derive the EPA's inventory from computerized listings of total hydrocarbon emissions for each Region 7 chemical plant. We have thus been able to determine for ourselves that in formulating its inventory the EPA made no significant reduction to account for nonreactive hydrocarbons. [31] As is apparent from the foregoing text and footnotes, this court has gone to considerable length to determine exactly what the agency did do. We have permitted the agency rather freely to supplement its original technical documents, not to provide additional justifications but in order to prove it used the techniques it claims to have used. In doing so we have recognized that some of petitioners' attacks were not presented to the agency (largely bacause its support documents were made available only after the regulations were promulgated). Further, we recognized that neither the parties to this litigation nor this court have had the experience to know in advance what is required to resolve in court the merits of implementation plan disputes We now have that experience, and it will affect our considerations should future litigants raise an issue not raised by the parties here, that of whether the agency's rule-making is arbitrary and capricious solely because it states its calculations and procedures in the summary fashion of Documents 30 and 31. The controversy over the content of the chemical processing inventory has required scores of hours of this court's time, and could have justified the services of a special master. It has also surely required hundreds of hours from lawyers and technicians on each side. Yet it could have been easily resolved at the outset, or would never have arisen, had the agency originally placed in the record the eight-page memo and four pages of worksheets, plus the printout, which show each step, and each assumption, in the process of compiling a regionwide inventory from plant-by-plant data. Further, the availability of this material is important if the agency and states are to be able to understand how and why their calculations yield different results, so that their differences can be resolved by technical personnel rather than the courts. In addition, worksheet materials would enlist the state's technical expertise in protecting against wasteful errors. We are satisfied that such information is vital in these cases, and can be feasibly provided. It is largely for failure to provide similar information that we have accepted the agency's rejection of the state's plan. The result in dialing with the agency's calculations has been different because its determinations are protected by the arbitrary and capricious standard. However, we are uncertain whether that protection will be so effective in future cases. [32] Petitioner Harris County suggests in its brief that the inventory to which the refinery reactivity factor was applied had already been reduced to eliminate at least some nonreactive hydrocarbons. As we did with the EPA's chemical processing inventories, we have obtained memoranda from Mr. Jack Divita, Chief of the EPA's Air Program Branch, outlining the steps taken by the agency in deriving the figure it used as the Region 7 initial refinery inventory. If this information has not satisfied petitioners' doubts, they should raise the matter before the agency during its reconsideration of its choice of reactivity factor [33] But see the discussion of Document 14 in note 19 supra [34] It is unclear whether this inventory represents 1970 raw data or the state's final adjusted 1972 data. See note 20 supra [35] Statement presented by the Texas Air Control Board on Transportation Controls Proposed for Texas at 10 (Aug. 15, 1973) [36] Document 14, discussed above in note 19, indicates that the state apparently claimed a region-wide total of 173,000 tons/year. As with the Harris County inventory, it is not clear whether this figure is 1970 raw data or the state's final adjusted 1972 data. See note 20 supra [37] See note 39, infra [38] The following chart lists the additional controls required for Region 7 by the EPA regulations, along with their corresponding 1977 percentage reductions in reactive hydrocarbon emissions [39] Our calculations were based on the total refining emissions inventory for the baseline year in each region, as shown in Tables 6-1 through 6-6 of Document 30, Part II. In each instance, the appropriate entry was multiplied by .65, yielding a revised refinery reactive inventory. The figure .65 was chosen as the hypothetical refinery reactivity factor in order to test the effect of agency reconsideration against the very highest factor that has some basis in the record before us. We in no sense are suggesting that .65 is the correct factor, nor are we establishing a presumption that the EPA should adopt such a factor. Rather, we use this figure simply to accord petitioners complete interim protection against the possibility of being required to comply with regulations which eventually prove unnecessary In each region the revised reactive refinery inventory was substituted for the original EPA inventory (based on .12 and shown in the baseline year 'Reactive' column of the above-referenced tables), resulting in an increased baseline inventory in each region. For this purpose, the initial baseline inventories were those shown in Tables 7-2, 8-2, 9-2, and 10-2 of Document 30, Part I. (These are identical to baseline year reactive 'Grand Total' entries in Tables 6-3 through 6-6, when the addition errors in these latter tables are corrected.) The revised baseline inventories were divided into the tons/year reductions expected from each EPA control measure, as listed in Tables 7-2 through 10-2. The result was smaller percentage reductions for each measure (except where the change was so minor as to fall within the rounding off to tenths of percent); this effect of a higher baseline inventory is reflected in notes 40 and 41. To determine the effect of a .65 factor on reductions obtained by Texas' own plan, the revised reactive refinery inventories were multiplied by .9, reflecting the 90 per cent effectiveness of Texas' plan on refinery emissions. This revised reduction (in terms of tons/year) was substituted in Tables 7-2 through 11-2 for the EPA's original reduction estimates based on a .12 factor (these were determined by taking 90 per cent of the baseline year reactive refinery emissions shown in Tables 6-3 through 6-6). The result was a higher reduction, in terms of tons/year, for the 'Present Controls' entries in Tables 7-2 through 10-2. Dividing these figures by the revised baseline inventories produced the percentage reduction expected from the Texas plan in each region. These figures were of course greater than those shown in Tables 7-2 through 10-2. The lessened need for additional controls is measured by the extent to which each region's target is exceeded by the sum of the revised percentage reductions for each measure. [40] For this purpose we calculate that each regulation contributes the reduction shown in the following table. These percentages are derived by dividing the tonnage reduction expected from each regulation by a baseline reactive emission inventory of 476,000 tons/year. The latter figure reflects the 75,400 tons/year increase created by assuming 65 per cent of Region 7 refinery emissions to be reactive [41--1] 41 This figure reflects the fact that use of a .65 refinery reactivity factor increases a region's baseline reactive hydrocarbon inventory [42] While measures determined to be unreasonable may not be forced upon a state, apparently nothing precludes their voluntary adoption-- as might happen if a state preferred such a measure to one validly determined to be reasonable [43] Our use of the term 'interim' is somewhat different from that in paragraph (2)(B) of subsection (e). As it appears in this opinion it refers not merely to measures for the temporary control of those sources for which more effective technology should be available by 1977. It also refers to those measures of permanent effect which are initially required only to meet the standard before 1977. The principal example of such an 'interim,' albeit permanent, measure is the requirement for vapor recovery during gasoline marketing operations. To recover these vapors for a one or two-year period requires a capital investment in equipment that will be available for many years thereafter [44] It could be argued that we are ignoring a technical requirement of subsection (e) by going straight to the merits of the decision to deny an extension. While the agency does not raise this argument, we think we should briefly address it in light of Natural Resources Defense Council, Inc. v. EPA, 1973, 154 U.S.App.D.C. 384, 475 F.2d 968. The statute requires that a state submit a plan implementing the national ambient air standard by 1975. When submitting such a plan the state may seek an extension for the reasons stated in the text. As discussed in Parts I & II of this opinion, in this case the agency determined, validly, that the state plan did not implement the national standard Nonetheless, we will not rule that Texas is ineligible for an extension until it submits another plan which in fact implements the standard. In the first place, the statute is inherently ambiguous, by virtue of its failure to specify whether the plan with which the extension application is submitted must receive EPA approval prior to consideration of the extension. That it would be inappropriate to require such approval as an absolute rule is suggested by the fact that even an approved implementation plan as contemplated in subsection (e) provides only for theoretical attainment of the standard, not attainment in fact. Second, to require another submittal from Texas would be pointless, for Texas' original plan, its request for an extension, and its brief all make it abundantly clear that in meeting the standard Texas wishes to rely as heavily as it can on the effects of federal control of automotive emissions. Further, since the state's strategy has long been clear, persons who object to this approach have had ample opportunity to comment and to have whatever impact they would be likely to have if the state implementation process were repeated. Finally, our failure now to address the merits of this issue could produce either further delay before the final adoption of a plan, or the use of techniques which are contrary to the statutory concern, expressed in subsection (e), that interim measures be forcibly employed only if reasonable. We also decline to order resubmittal on the basis of a second potential technical objection, that Texas apparently did not submit a plan imposing or relying on motor vehicle requirements which would not be met by 1975. See 42 U.S.C.A. 1857c-5(e)(1)(A). Rather, Texas appears to have calculated the additional reductions required as of 1977 based on projected independent reductions in motor vehicle emissions. We think this sufficient, both for the reasons stated in the preceding paragraph, and also because vehicle emissions control is being accomplished independently of Texas' plan, such that its effect is more appropriately treated as a background factor than an integral part of that plan. While subsection (e) is principally aimed at cushioning states against unreasonable measures when the controls they impose cannot be fully effective until after 1975, we think it properly applicable to the kindred situation presented here-- that in which the post-1975 reductions are accomplished by control measures or other factors independent of the state's own plan. [45] 38 Fed.Reg. 30635, 30638, 30640 (Nov. 6, 1973) [46] Both systems are designed to capture vapors displaced from vehicle tanks during refilling. The displacement system simply forces these vapors into the service station tank, from which, theortically, an equal volume of fuel has been removed. The more complex systems seek to also capture excess vapors created by temperature differentials between the storage and vehicle tanks [47] We are told that regulations for vapor recovery during vehicle refueling are under agency review on a nationwide basis. However, the agency wished to stand on the record in this case, on the grounds that the information generating the review was not available when the Texas regulations were promulgated and thus should not affect the initial validity of those regulations [48] If Texas believes that the reductions to be obtained from extending Regulation V in its entirety to all of Tarrant County are not needed to meet the standard as of May 31, 1977, and that expenditures needed only to meet the standard prior to that date are not reasonable in light of this court's discussion of that subject, it may present its position to the agency through a 42 U.S.C.A. 1857c-5(a)(3) revision proceeding [49] As we stated in note 16, this aspect of the agency's power imposes a correlative duty to reconsider and revise its requirements as better data becomes available [50] If the figures in note (38) are rounded off, the 75 per cent Region 7 target is achieved without reliance on the land-use and transportation regulations (1.5 is rounded up because the decimal portion of the actual quotient exceeds .5, but in note (38) was rounded down to that figure; the same is true of the 65.5 per cent reduction which the EPA credits to the state's plan) [51] It should be recognized that according to the EPA's view of its statutory power the effect of these objections goes principally to the form of the regulations and not their permissible substance. Had the agency been impressed by Harris County's contention that it is beyond agency power to require that the state institute a vehicle inspection program, the EPA could have sought the state's 'voluntary' cooperation by promulgating measures which it considered to be well within its power. For example, it could have directed its regulation at individual drivers, prohibiting their use of vehicles which have not obtained a state permit pursuant to an emissions inspection. (This is the format for the Region 7 regulation controlling new emission sources). Alternatively, the agency could have promulgated regulations giving the state the option of instituting an inspection program or having the agency accomplish an equivalent emissions reduction through less desirable measures which it could administer such as gas rationing at the wholesale level. Cf. Common-Protection Agency, 3 Cir., 1974, 500 F.2d 246, on these issues