concurring:
I join most of the majority’s opinion, but I write separately because the statutory treatment of offset requirements is more ambiguous than part II.B. might suggest.
After directing the Administrator of the EPA to “establish requirements to control air pollution from Outer Continental Shelf sources located offshore of the States”, Congress specified that the requirements for OCS sources located within 25 miles of a state’s seaward boundary “shall be the same as would be applicable if the source were *1183located in the corresponding onshore area”. 42 U.S.C. § 7627(a)(1). This mandate obviously calls on the Administrator to determine the requirements that would apply if a source that is not located within a particular onshore area actually were located within that area. But there are two possible ways of approaching this task. One method is to imagine that the OCS source stays where it is, while the onshore area’s boundaries expand to encompass it. The other method is to imagine that the onshore area’s boundaries stay where they are, while the OCS source moves to some point within them.
Many of the state and local requirements that Congress had in mind have a uniform impact throughout the entire onshore area; that is, their application depends only on whether a source is located within the covered area, not on exactly where the source lies. Requirements of this sort (which include offset requirements that do not call for “distance discounting”) can be extended to OCS sources without ambiguity, because their effect will be the same regardless of whether one imagines the onshore area’s boundaries expanding to include the OCS source or the OCS source moving to within the boundaries. If an onshore area limits emissions of a certain pollutant, then nearby OCS sources face the same limitation. If onshore sources of the pollutant must file reports, then so must the OCS sources.
But offset requirements that call for “distance discounting” are different. The application of such requirements to any particular source depends on the source’s precise location within the onshore area, and it is not immediately obvious how to apply such requirements to sources that are not really within the onshore area at all. For requirements of this sort, the distinction between the two possible approaches to § 7627(a)(1) becomes crucial.
Under the first approach, the critical distance is that between the proposed OCS source and the source that will provide the offsetting reductions in pollution. Since “distance discounting” creates an incentive to minimize this distance, the operator of the OCS source might well seek to obtain his offset from another offshore source rather than a land-based one. In addition, he would be indifferent about whether the offset came from the seaward or the landward side of the proposed source.
Under the second approach, by contrast, the proposed OCS source would be treated as if it were really located somewhere in the corresponding onshore area — at Point X, say. Because the operator of the proposed source would thus have an incentive to find offsets from sources as close to Point X as possible, this approach would encourage pollution reductions at sources in or near the onshore area (where pollution is of the greatest concern) rather than near the proposed offshore source.
The EPA — the agency that Congress charged with implementing § 7627(a)(1)— seems to favor the second set of incentives over the first. Offset requirements, after all, are designed to improve air quality within nonattainment areas like Santa Barbara County, and the EPA was concerned that “application of onshore offset requirements [to OCS sources] might unintentionally provide an incentive for an OCS source to obtain offsets far from the nonattainment area”. In promulgating the rules at issue here, the agency asserted that they would respect “the underlying goals and technical rationale” of offset requirements by encouraging operators of OCS sources “to obtain their offsets from the landward side of the OCS source”. 57 Fed.Reg. 40792, 40796 (Sept. 4, 1992).
Still, the regulations do not reflect any coherent interpretation of the EPA’s statutory mandate to treat close-in OCS sources as if they were “located in the corresponding onshore area”. In the area seaward of the OCS source (“zone 1”), the EPA takes the first approach, applying local distance discounts as if the onshore area had expanded to include the Outer Continental Shelf. But when it comes to the area extending inland from the imaginary fine 3 miles off the California coast (“zone 3”), the EPA effectively takes the second approach, applying local distance discounts as if the OCS source had moved to a point on that imaginary line.1 *1184(To judge from comments at oral argument, this line apparently marks the boundary of the “onshore” area.) As for the intermediate area (“zone 2”), the EPA does not apply local distance discounts at all; this treatment— which has the same effect as pretending that both the proposed and the offset sources have been moved to a single point within the onshore area — does not seem to fit either approach.
While we owe deference to the EPA’s reasonable interpretation of § 7627(a)(1), see Chevron USA v. Natural Resources Defense Council, 467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984), the EPA must make such an interpretation; it cannot adopt an amalgam of reasonable but conflicting interpretations, nor can it simply ignore its statutory mandate in favor of what it considers a more rational policy. I therefore agree with the majority that the EPA’s treatment of offset requirements for OCS sources is invalid. But I do not think that the statute requires the EPA to extend its treatment of zone 1 to the entire area; in my view, the agency could equally well extend its treatment of zone 3 (or some similar means of pretending that the OCS source has moved within the corresponding onshore area). Both approaches to the statute seem permissible to me, and it is up to the EPA to choose between them.
. Santa Barbara points out that there may be more than one such point. This observation *1184raises a question about how the regulation would be applied, but it does not prove that the regulation is invalid.