This opinion deals with certain additional issues presented in 23 industry petitions covering 32 major power and industrial companies in Ohio objecting to the United States EPA's plan for control of SO 2 pollution in Ohio. These additional issues concern only point sources of SO 2 pollution in Ohio's rural areas or areas with complex terrain. This opinion should be read as supplementary to the opinion of this court dated February 13, 1978, Cleveland Electric Illuminating Co., et al. v. Environmental Protection Agency, et al., 572 F.2d 1150 (6th Cir. 1978).

Our focus herein is upon the following petitioners and the designated facilities belonging to them. These petitioners protest certain features of the United States EPA model (MAXT-24) employed for predicting plant pollution in rural and complex terrain areas:

Cincinnati Gas & Electric Co., all facilities (Hamilton & Clermont Counties).

Columbus & Southern Ohio Electric Co., all facilities (Athens, Coshocton, Pickaway Counties).

Dayton Power & Light Co., all facilities not covered by opinion dated February 13, 1978 (Adams County).

Ohio Edison Co. (Jefferson County) Sammis Plant only.

Ohio Power Co., all facilities (Washington and Morgan Counties).

E. I. du Pont de Nemours & Co., all facilities (Hamilton County).

Shell Oil Co., all facilities (Washington County).

Austin Powder Co., all facilities (Vinton County).

The MAXT-24 model (Second Maximum 24-Hour Dispersion Model with Terrain Adjustments) is designed for use in predicting SO 2 pollution resulting from single sources located in rural areas. Unlike the RAM model employed in urban areas, which we dealt with in Cleveland Electric Illuminating Co., supra, MAXT-24 does not provide estimates of comparative contributions to total SO 2 pollution from a number of point sources. The MAXT-24 model treats each point source as an isolated problem, and only general background SO 2 pollution data are added into the formula.

In other respects the MAXT-24 model strongly resembles the RAM model.1 Thus, like RAM, MAXT-24 starts with a solid ascertainable data base, namely, the established design capacity of the power or steam generating plants in question related to the sulfur content of the fuel used by such plants. Emissions data are developed from these factors. Subsequently, stack height, wind, weather, and terrain data are added. Like RAM, MAXT-24 employs a Gaussian plume formula and assumes vertical and horizontal dispersion of the pollution plume. It employs the Pasquill-Gifford stability classifications and coefficients.

Like RAM, the MAXT-24 model was designed by United States EPA largely as a result of industry criticism of the use of rollback modeling. As was true in relation to the RAM results, the results of use of MAXT-24 were generally less strict than those contemplated by the 1972 and 1974 Ohio EPA SO 2 regulations.2

Indeed, the comments this court made in Cleveland Electric Illuminating Co., supra, in Section 3 of the opinion are largely applicable to EPA's adoption of MAXT-24 and we cite and rely on said Section 3 in holding that in general (and with one exception noted below) the EPA's adoption and use of the MAXT-24 model is not arbitrary or capricious and, like the use of the RAM model, must be affirmed by this court.

Despite the discussion above, we are not certain that any of the petitions we deal with in this opinion seriously disputes the general validity of the MAXT-24 model. What these petitioners clearly do contend is that the MAXT-24 model results are badly skewed to their great economic disadvantage by 1) the Class A assumption employed to estimate pollution dispersion in the least stable wind condition, and 2) the failure of EPA to employ the half ground displacement theory in estimating pollution impact on hilly terrain.

The MAXT-24 model makes use of a set of six coefficients for determining plume dispersal. The classes of coefficients employed were based upon six different weather conditions. The term Class A is employed to describe both the least stable weather condition and the set of assumptions which is based on the most direct and quickest impact of the pollution plume upon ground level with the least prior dispersal.

The six Pasquill-Gifford coefficients employed in MAXT-24 are derived from a Nebraska study made in the 1950's and are referred to by United States EPA as "time-tested." What this defense appears to ignore, however, is that petitioners in this instance (contrary to the general attack upon the six coefficients employed in RAM) are not objecting to the use of the coefficients; they are attacking the accuracy of one set of them the Class A set associated with "gusty winds." Specifically they claim that the Class A assumption is fallacious in that it assumes a longer period of downward draft than occurs in fact and fails to make allowance for the lateral dispersal which would accompany such a vertical wind at the point of impact.

The lead brief for the utilities presents the case thus:

In all modeling of rural power plants, EPA utilized dispersion coefficients under Class A stability conditions which have no support in data, which have been repudiated by most modelers and which are demonstrated inaccurate by this record. As applied to this rulemaking, this seemingly simple assumption is exceedingly important because, for almost 1/3 of the power plants in Ohio, it was the determining factor in establishing emission limits.The meaning of "Class A."

Diffusion models can account for thousands of bits of data. Most important are meteorological data of which stability classes are an aspect.

Specifically, stability classes are categorizations of the atmosphere's ability to disperse plumes. These classes are divided into six categories ranging from extreme dispersion of plumes (Class A) to minimum dispersion (Class F). Under Class A, a plume is assumed to disperse very rapidly to the ground level before there is any substantial dilution. This, in turn, leads to predictions of high ground level concentrations. The fundamental issue, therefore, is whether the Class A assumption describes the manner in which plumes disperse at rural power plants and whether the phenomena it depicts really occur.

Brief of Utility Petitioners at 31-32 (emphasis in original).

Petitioners then detail the results of three separate studies which they claim attack and undermine the validity of the Class A coefficients, and generally urge substitution of Class B coefficients. These studies are the privately financed study by Enviroplan, Inc., a similarly produced study by Smith-Singer Meteorologists, and a strongly critical report resulting from the Specialists' Conference of February 22-24, 1977, sponsored by United States EPA itself through the Argonne National Laboratory.

To this argument the EPA's response is as follows:

EPA properly determined that the "Class A" stability factors should continue to be used until new field data proved them incorrect.

Petitioners argue that EPA should have changed the dispersion coefficients used in the rural MAX (CRSTER) model for analyzing ground level concentrations caused by a source in very unstable weather, known as "Class A" conditions. In the remand comment period, the utility petitioners presented various theories that the model did not accurately reflect the way wind patterns in such weather conditions affect dispersion patterns and that therefore the model might be overestimating ground level concentrations for a 3-hour analysis. Petitioners argue that it was arbitrary or capricious for EPA not to accept theories presented in their comments.

EPA recognized in the STSD (Supplemental Technical Support Document) at 55 that there was a growing concern among atmospheric modeling scientists about the issue. EPA determined, however, that until further studies could be done to substantiate the theories, there was no experimental or field data to justify changing the dispersion curves or to determine how the dispersion equations should be changed. And since petitioners did not submit any data, no change could be made in the equations used. Id.

EPA Brief at 48-49.

We are, of course, aware that decision-making (particularly in this highly technical area) is the primary responsibility of the agency and not the responsibility of this court. See Vermont Yankee Nuclear Power Corp. v. NRDC, 435 U.S. 519, 98 S. Ct. 1197, 55 L. Ed. 2d 460 (1978). As we said in Cleveland Electric Illuminating Co., supra:

Our standard of review of the actions of United States EPA is whether or not the action of the agency is "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law." Clean Air Act Amendments of 1977, Pub.L.No.95-95, § 305(a), 91 Stat. 775 (to be codified as 42 U.S.C. § 7607(d) (9)(A)). Thus, we are required to affirm if there is a rational basis for the agency action and we are not "empowered to substitute (our) judgment for that of the agency." Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416, (91 S. Ct. 814, 28 L. Ed. 2d 136) (1971).

572 F.2d at 1161.

On this present record we conclude that United States EPA's employment of the Class A assumption in determining pollution dispersion under "least stable" wind conditions in rural areas and areas of complex terrain is not a rational decision and is arbitrary and capricious.

As we read EPA's position on this point, it is that no better solution has been proposed. This answer, however, ignores the Enviroplan and Smith-Singer studies. More importantly, it ignores the conclusions of the experts' conference convened by Argonne National Laboratory at United States EPA's own request.3 The report of that conference suggested "elimination of the A curve and the use of the B curve for both A and B stability categories." See Report of the Specialists' Conference on the EPA Modeling Guideline, Feb. 22-24, 1977, Chicago, Illinois, at § 2.7.5: "Vertical Dispersion Estimates."

By pointing out this proposed solution, we do not mandate its acceptance by United States EPA. Our ultimate action on this score is simply to remand this issue to United States EPA for further study. This might result in the writing of a new record which supplies the now missing support for the use of Class A coefficients for the least stable wind condition in rural counties, or it might result in United States EPA's adoption of use of the Class B coefficients for the two least stable wind conditions, or it might result in a new record which supports a wholly new solution.

The original MAXT-24 model assumed that the pollution plume moves downwind in a straight line from a point determined by the height of the smoke stack plus plume rise ("effective stack height"). Under that assumption, if effective stack height is 600 feet and there is a hill 800 feet high downwind, a receptor site located 600 feet up that hill will therefore in theory receive the full impact of the pollutants in the center of the plume.

Petitioners' attack upon this set of assumptions is, we believe, best stated by the Shell Oil Co. brief:

The reference to "the effect of the terrain on the plume" is an easily understood concept. Since the wind which blows the plume toward a hill (terrain) cannot blow through the hill, it must blow up and over the hill. This effect causes all the layers of air above the wind at ground level, and hence the plume itself, to be carried up and over the hill rather than hitting directly into its side. A widely recognized means of accounting for such a situation, and one which EPA has used, is to incorporate mathematical changes in the model which reduce the receptor height by the one-half difference between stack base and receptor elevation and limit the approach of the center line of the plume to ten meters above the receptor.

Shell and its consultant, Enviroplan, recommended this change to U.S. EPA. Moreover, this change was supported in the modeling literature by two other independent experts Briggs and Egan. Also, another consultant, Environmental Research & Technology, Inc. ("ERT"), recommended the same adjustment in a report submitted to EPA during the comment period for Columbus & Southern Ohio Electric Co. See "A Technical Review of the U.S. EPA Ohio State Implementation Plan for Sulfur Dioxide," January 1977, prepared for C&SOE. (App. 210-213.) Indeed, this approach is so well recognized that EPA, Region II, approved its use in sustaining a revision of the implementation plan for Puerto Rico. See 40 Fed.Reg. 52410 (1975).

Brief of Shell Oil Co., at 19-20.

As to this argument, the EPA brief contains this comment and admission:

Contrary to petitioners' claims, EPA utilized available monitoring data wherever possible. As set forth in detail in both the Final and the Supplemental Technical Support Documents, EPA conducted validation studies of the dispersion model used to set emission limitations for isolated, rural power plants. See STSD at 53-55, and the FTSD at 27-34. The validation studies compared model predictions of SO 2 ground level concentrations to actual air quality monitor data. These comparisons indicated that for sources located on flat terrain, the correlation between monitor data and predictions was quite good with the model tending to underpredict, but that for power plants located on hilly terrain, the comparisons showed consistent overpredictions.

EPA Brief at 45-46 (footnote omitted).

The EPA brief then goes on to assert that certain adjustments have been made in the model "so that it could handle dispersion in hilly terrain more accurately," and then cited the Supplemental Technical Support Document at page 55. The STSD material referred to follows:

The validation studies which compared model predictions of SO 2 ground level concentrations to actual air quality monitor data indicated that in certain situations the model overpredicted and needed modification. The problem usually occurred when air quality monitors were at elevations higher than the top of the stack. To correct this, terrain data used in the model was limited in such a way that terrain features were always assumed to be no higher than the stack height of the source stack in question. This was deemed to be an appropriate adjustment because the validation study showed a high degree of correlation between model predictions and sample readouts from monitors positioned on terrain lower than stack height in elevation. When this assumption was mathematically incorporated into the model, the validation studies showed that the model accurately predicted the ground level concentrations observed by the monitors.

Petitioners proposed a different method for modifying the model to account for complex terrain situations, but the proposal is not based on any validation studies of the CRSTER model. The Agency has no way of determining if the proposal is a better modification to the model than the modification made by the Agency after the validation studies. The Agency, therefore, has determined that the model does not need further modifications because of any information presented by the petitioners.

EPA Brief at 54-55.

While the record does not establish conclusively that this adjustment made by United States EPA in the remand period will prove a satisfactory solution to the problem posed, neither does the record offer evidence to the contrary. We note, of course, that United States EPA has disowned the apparent implication in its brief that it had made validation studies of this latest adjustment for hilly terrain. And, in fact, our holding on this issue should not be read as this court's rejection of petitioners' half ground displacement theory in favor of the United States EPA adjustment outlined above. There may well be occasion for the agency to continue to review this issue. All we hold is that on the present record, we cannot find that United States EPA's present terrain adjustment in MAXT-24 is "arbitrary or capricious."

For the reasons indicated above, the petitions of Dayton Power & Light Co. (Adams County facilities only), Ohio Power Co. (all facilities), Columbus & Southern Ohio Electric Co. (Coshocton County facility only), and Austin Powder Co. (all facilities) are remanded to the United States EPA for reconsideration of the employment of Class A coefficients in least stable wind conditions in rural counties.

All petitions referred to at the beginning of this opinion are denied to the extent that they attack the MAXT-24 model as to the terrain adjustment feature.

We also hold that there is no merit to objections based on failure to calibrate the MAXT-24 model (or failure to reject its results because of claims of overprediction as demonstrated by some monitor readings). See Cleveland Electric Illuminating Co., supra, 572 F.2d at 1163-64, numbered paragraph 7.

Shell contends that its emission limitation should be expressed in terms of pounds SO 2 per hour rather than pounds SO 2 per million British Thermal Units. The Shell proposal would require the EPA either to assume that stack gas temperature and exit velocity (the important factors bearing on plume rise and thus ultimately on ground level SO 2 concentration) are relatively constant, or alternatively to monitor stack gas temperature and exit velocity. Clearly, EPA considers policing such a system to be an impossible task. EPA's formula, by contrast, requires only the use of fixed, easily ascertainable date the plant's design-rated capacity. We regard EPA's choice of formula, which minimizes administrative costs while obeying the Clean Air Act's command to " insure attainment and maintenance" of national ambient air standards, 42 U.S.C. § 1857c-5(a)(2)(B) (1970),4 to be within the range of the agency's discretion.

We have considered the other issues raised by Shell and find them to be without merit.

Disputes between petitioners and EPA concerning appropriate SO 2 background levels, emission data, or other fact issues will not be decided by this court until completion of the administrative review of such issues which was suggested by this court and agreed upon by the parties.

Based upon what has been said by this court in Cleveland Electric Illuminating Co. v. EPA, supra, and in this opinion, and finding no other material issues, we dismiss the following petitions in toto: Cincinnati Gas & Electric Co., Shell Oil Co.

Final dispositions in the petitions of Columbus & Southern Ohio Electric Co., Ohio Edison Co., and E. I. du Pont de Nemours & Co. will be entered on resolution of the remaining issues therein.