PER CURIAM:
Petitioner Galveston-Houston Association for Smog Prevention (“GHASP”) seeks review of the United States Environmental Protection Agency’s (“EPA”) final rulemaking action approving the Mid-Course Review State Implementation Plan (“MCR SIP”) submitted by the State of Texas for the Houston/Galveston/Brazoria Severe Ozone Nonattainment Area (“HGB area”). Texas submitted the MCR SIP to satisfy one of the enforceable commitments contained in a previously approved State Implementation Plan for the HGB area. GHASP contends that (1) the EPA
acted arbitrarily and capriciously in approving the MCR SIP because it does not demonstrate attainment of specified emissions reductions; (2) the EPA acted arbitrarily and capriciously in relying on weight of evidence analysis to excuse modeled nonattainment; and (8) by approving the MCR SIP, the EPA violated the noninterference or anti-backsliding provision of the Clean Air Act. For the following reasons, we deny the petition for review. GHASP also petitions this court for an award of attorneys’ fees, which we deny as well.
FACTUAL AND PROCEDURAL BACKGROUND
A.
Clean Air Act
The Clean Air Act (“CAA” or “the Act”), enacted in 1970 and revised in 1977 and 1990, establishes a comprehensive program for controlling and improving the nation’s air quality through a combination of federal and state regulation. 42 U.S.C. § 7401(a)(4). Under Title I of the CAA, the EPA is responsible for identifying air pollutants that endanger the public and formulating the National Ambient Air Quality Standards (“NAAQS”), which establish maximum permissible concentrations of those pollutants in ambient air.
U2 U.S.C.
§§ 7408-7409. The EPA Administrator has promulgated NAAQS for various pollutants, including ozone. The standard for the one-hour NAAQS for ozone is .12 parts per million (although EPA often refers to it as 124.5 parts per billion (“ppb”)), and an area attains the standard when maximum measured hourly average ozone concentrations exceed the NAAQS no more than one day per calendar year. 40 C.F.R. § 50.9.
Under the Act, the EPA designates areas of the country as “attainment” or “non-attainment” based on whether the area meets the NAAQS for a particular pollutant.
42 U.S.C. § 7407(d). Nonattainment areas are further classified as “marginal,” “moderate,” “serious,” “severe,” or “extreme” based on the severity and duration of their noncompliance. 42 U.S.C. § 7511(a). Based on an area’s classification, the Act establishes the stringency of the measures that the area must implement to reduce emissions of volatile organic compounds (“VOCs”) and nitrogen oxides (“NOx”), both of which are precursors to the formation of ozone in the ambient air.
See
42 U.S.C. § 7511a. This qualification also establishes deadlines for attainment of the NAAQS depending upon an area’s nonattainment classification. 42 U.S.C. §§ 7410, 7502, 7511(a). If an area fails to attain the one-hour NAAQS by the applicable deadline, the EPA must reclassify it to a higher classification. 42 U.S.C. § 7511(b)(2). Generally, reclassification grants the area a later attainment deadline, but requires it to comply with the more stringent pollution control measures applicable to the higher classification. 42 U.S.C. § 7511(a)(1), (b)(2); 42 U.S.C. § 7511a.
Although the EPA determines the standards of air quality, under the CAA, states have the primary responsibility for ensuring that the NAAQS are met for each identified pollutant. 42 U.S.C. § 7407(a). This responsibility includes a requirement that states adopt State Implementation Plans (“SIPs”), specifying the way in which the state will implement, maintain, and enforce ambient air quality standards in the various regions throughout the state.
Id.
For areas designated from moderate to extreme, the SIP must contain an “attainment demonstration” that shows that the area will achieve the
NAAQS by the area’s statutory attainment deadline. 42 U.S.C. § 7511a(c)(2)(A); 40 C.F.R. § 51.112. The attainment demonstration is based on the state’s control strategy for ozone-precursor emissions, which must “include enforceable emissions limitations, and such other control measures ... as may be necessary or appropriate to provide for attainment of such standard in such area by the applicable attainment date.” 42 U.S.C. § 7502(c)(6). The EPA may approve an attainment demonstration if it is “based on photochemical grid modeling or any other analytical method determined by the Administrator, in the Administrator’s discretion, to be at least as effective.”
42 U.S.C. § 7511a(c)(2)(A). The SIP, and any revisions to the SIPs, are to be adopted by each state after reasonable notice and public hearing and thereafter submitted to the EPA. 42 U.S.C. § 7410(a)(1). The EPA reviews each proposed SIP and must approve the SIP if it meets all of the CAA’s requirements. 42 U.S.C. § 7410(k)(3).
B.
Houston/Galveston/Brazoria Area Plan
The HGB area has not complied with the federal one-hour health standard for ground-level ozone since the inception of the CAA in 1970. After the 1990 amendments to the CAA, the HGB area was classified as a “Severe 17” area. Pursuant to 42 U.S.C. § 7511(a)(2), such areas were required to come into compliance with the one-hour standard by November 15, 2007.
The Texas Commission on Environmental Quality (“TCEQ”) prepared and submitted the Texas SIP, which projected compliance with the one-hour ozone standard by the November 15, 2007 attainment deadline. The EPA approved the SIP in 2001. The 2001 SIP contained an emissions control strategy including federal control measures, state emissions limitations that would reduce NOx from industrial sources by 90 percent, an expanded inspection and maintenance program for vehicles, a “Clean Diesel Program,” speed limit reductions, voluntary mobile emission programs (VMEP), controls on airport ground equipment, control requirement for non-road, large spark-ignition engines, and commercial lawn equipment operating restrictions.
See generally
66 Fed.Reg. 57160.
In 2003, the Business Coalition for Clean Air Appeal Group (BCCA Appeal Group), GHASP, and several other groups challenged the EPA’s approval of Texas’s 2001 SIP before this court.
See BCCA v. EPA,
355 F.3d 817 (5th Cir.2003). In that case, this court concluded that: (1) the EPA was not unreasonable in approving Texas’s photochemical modeling which excluded certain ozone characteristics from the modeling; (2) “weight of evidence” analysis is consistent with the CAA’s requirements and reasonably applied to Texas’s attainment demonstration modeling results; and (3) the use of “enforceable commitments” is consistent with the CAA.
One of the enforceable commitments contained in the 2001 SIP approved by this court in
BCCA
is the subject of the present petition for review, specifically the MCR SIP.
Both parties agree that
the MCR contains several revisions to the 2001 SIP. In particular, the MCR SIP relaxed the controls on industrial NOx emissions from the 90 percent level established in the 2001 SIP to 80 percent and supplemented this action by including controls on the release of highly reactive volatile organic compounds (“HRVOCs”). 71 Fed.Reg. at 52,674/2. The MCR SIP also dropped four requirements that had been included in the 2001 SIP: (1) the requirement for vehicle inspection and maintenance in three rural counties; (2) restrictions on morning operations for commercial lawn services; (3) planned reductions in speed limits; and (4) restrictions on idling for heavy duty diesel vehicles. 71 Fed.Reg. 52,-674/1. GHASP contends that: (1) the EPA acted arbitrarily and capriciously in approving the MCR SIP because it does not demonstrate attainment of specified emissions reductions; (2) the use of weight of evidence analysis, as applied here, to excuse modeled nonattainment of the four days modeling exceedances was unreasonable; and (3) by approving the MCR SIP, the EPA violated the non-interference provision of the Clean Air Act. We will address each of these issues in turn.
DISCUSSION
This court may reverse any action of an EPA Administrator that it finds “arbitrary and capricious, an abuse of discretion, or otherwise not in accordance with law.” 42 U.S.C. 7607(d)(9)(A);
see also BCCA,
355 F.3d at 824. A rule is considered arbitrary and capricious “where the agency has considered impermissible factors, failed to consider important aspects of the problem, offered an explanation for its decision that is contrary to the record evidence, or is so irrational that it could not be attributed to a difference in opinion or the result of agency expertise.”
BCCA,
355 F.3d at 824 (citation omitted).
Where the agency is interpreting a statute it administers, that interpretation must be scrutinized under the framework established in
Chevron U.S.A, Inc. v. NRDC,
467 U.S. 837, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). Thus, where Congress has “directly spoken to the precise question at issue,” this court must give effect to the unambiguous intent of Congress.
Id.
at 842-43, 104 S.Ct. 2778. However, if the statute is silent or ambiguous on a particular issue, the question then becomes whether the agency’s interpretation is based on a permissible construction of the statute.
Id.
at 843, 104 S.Ct. 2778. Reversal is warranted only if the agency’s construction is “arbitrary, capricious, or manifestly contrary to the statute.”
Id.
at 844, 104 S.Ct. 2778. Deference must be given where the agency’s construction is permissible.
Id.
at 843, 104 S.Ct. 2778.
This court has previously acknowledged that photochemical grid modeling is a complex scientific determination.
See BCCA,
355 F.3d at 834. “A reviewing court must remember that the agency is making predictions, within its area of special expertise, at the frontiers of science. When examining this kind of scientific determination, as opposed to simple findings of fact, a reviewing court must generally be at its most deferential.”
Id.
at 832 (citing
Baltimore Gas & Elec. Co. v. NRDC,
462 U.S. 87, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983)). Thus, this court must determine whether the EPA’s conclusions “represent arbitrary or capricious exercises of its authority, not whether they are accurate.”
Id.
“There is a presumption of regularity to the EPA’s choice of analytical methodology, so challenging parties must overcome a ‘considerable burden.’ ”
Id.
I.
An SIP must contain an “attainment demonstration” that shows that the area will achieve the NAAQS by the area’s statutory attainment deadline. 42 U.S.C. § 7511a(c)(2)(A); 40 C.F.R. § 51.112. The attainment demonstration must be “based on photochemical grid modeling or any other analytical method determined ... to be at least as effective.” 42 U.S.C. § 7511a(c)(2)(A). Here, TCEQ selected a nineteen day episode from August 19 through September 6 on which to perform photochemical modeling in order to determine whether the control strategy contained in its MCR SIP would meet the NAAQS. 71 Fed.Reg. 52,688. TCEQ determined that it was necessary to exclude nine of the nineteen days before modeling due to model performance issues. Three of the nine days excluded before modeling contained exceedances of NAAQS.
GHASP argues that the EPA was arbitrary and capricious because it approved TCEQ’s exclusion of those three days showing exceedances from modeling without adequate explanation. Citing
Appalachian Power Co. v. EPA,
249 F.8d 1032, 1053 (D.C.Cir.2001), GHASP argues that although courts routinely defer to agency modeling of complex phenomena, model assumptions must have a “rational relationship to the real world.” This court has previously explained that agency decisions will be upheld if the agency has examined the relevant data and articulated a reasonable explanation for its conclusions.
BCCA,
355 F.3d at 834. Thus, GHASP must show that the Agency’s decision was unreasonable in order for this Court to conclude that the EPA acted arbitrarily or capriciously in approving Texas’s modeling despite the exclusion of those days.
Id.
at 832.
GHASP has failed to offer sufficient support for its assertion that the exclusion of three days showing exceedances necessarily leads to the conclusion that TCEQ’s modeling bears no resemblance to the real world. In its final rule, the EPA explained that it approved TCEQ’s decision to exclude nine days (of which three days showing exceedances were included) from its modeling analysis due to “model performance issues.” GHASP responds that merely citing model performance issues is an inadequate explanation for the exclusion of the three days and that the existence of these issues provides further support for its contention that there are “major flaws” in the MCR SIP.
The EPA determined, however, that the ten remaining days provided a sufficient basis for performing the attainment demonstration modeling, noting that the meteorological conditions present in the HGB area were sufficiently represented in the remaining ten days.
See
Technical Support Document for Approval of SIP Revisions to the One-Hour Ozone Attainment Plan for the Houston/Galveston/Brazoria Nonattainment Area Final Clean Air, at 21 (Sept.2005) (“TSD”). This court has previously acknowledged that modeling in the HGB area is exceptionally complicated due to the unique environmental factors of the area.
Id.
at 823 n. 1 (noting that Texas’s unique land-sea breeze meteorological conditions create a “level of complexity ... not seen anywhere else in the country” (citing 66 Fed. Reg. at 57,164)). Similarly, in the present case, the EPA explained that complexity of the meteorology of the HGB area was difficult to replicate in the modeling,
see
TSD, at 5, 21; for example, the EPA stated that the model performed poorly on the first four days of the extended episode, i.e. August 18-22, because it did not generate enough ozone to correspond to the historic data, TSD, at 20. However, the EPA determined that:
The need for further studies does not mean, however, that the modeling relied upon today was unable to estimate the
amount and type of emissions reductions needed for attainment. [The] EPA believes because the diagnostic/sensitivity tests do not reveal serious flaws in model formulations and the model generally predicts the right magnitude of the peak which is confirmed by the statistical measures and graphical analysis, that the model does provide an acceptable tool for estimating the amount of emissions reductions needed.
72 Fed.Reg. at 52,682/2.
We conclude that the EPA offered a rational explanation during the administrative process for its determination that it was necessary to exclude the three days at issue from modeling. This court has repeatedly stated that we will not substitute our judgment for that of the EPA, especially where, as here, the agency’s decision is based on “its evaluation of complex scientific data within its technical expertise.”
BCCA,
355 F.3d at 824. In light of this deferential standard and given that the EPA specifically considered and rejected GHASP’s argument, GHASP has failed to carry its heavy burden of showing that the EPA was arbitrary or capricious for approving the attainment demonstration that excluded certain days from modeling.
II.
We next consider the EPA’s weight-of evidence analysis. Texas’s photochemical grid modeling did not demonstrate attainment on four of the ten modeled days, namely: August 31, September 2, 4, and 6.
Using weight of evidence analysis, the TCEQ supplemented the modeling with other evidence to show that the HGB area would reach its attainment deadline. Specifically, it determined that: (1) based on certain control strategies which could not be modeled, September 2, 4, and 6 would reach attainment although they modeled nonattainment, and (2) August 31 should be excluded from the attainment demon-
station because of the anomalous meteorological conditions present on that date.
The CAA requires that an attainment demonstration be “based on photochemical modeling or any other analytical method determined ... to be at least as effective,” 42 U.S.C. § 7511a(c)(2)(A); the EPA has interpreted the statute to allow states to supplement their photochemical modeling results with additional evidence to demonstrate attainment due to the inherent uncertainties in air quality modeling. 66 Fed.Reg. at 57,170;
see also BCCA,
355 F.3d at 834. This court has held that where the weight of evidence analysis is set forth in notice-and-comment rulemaking, it is entitled to deference if it is reasonable.
BCCA,
355 F.3d at 835.
GHASP does not dispute the ability to consider weight of evidence analysis in evaluating attainment demonstrations, rather it contends that the manner in which the EPA applied it here was arbitrary and capricious. Specifically, GHASP asserts that the EPA applied the weight of evidence analysis in a “totally qualitative” manner, rather than the quantitative approach approved by this court in BCCA
Petitioner maintains that the EPA’s application of weight of evidence analysis here is not only arbitrary and capricious but violative of EPA guidance, which provides that “the weight of evidence provided from other analyses will need to be very compelling to overcome that resulting from the photochemical grid model.” EPA Office of Air Quality Planning and Standards, Guidance on Use of Modeled Results to Demonstrate Attainment of the Ozone NAAQS, EPA-454/ B-95-007 (June 1996).
As an initial matter, we consider the EPA’s contention that GHASP has waived any argument regarding the application of weight of evidence analysis to modeling results for September 2, 4, and 6 because it failed to object during the administrative proceedings. “[Ojnly in exceptional circumstances should a court review for the first time on appeal a particular challenge to the EPA’s approval of a state implementation plan that was not raised during the agency proceedings.”
BCCA,
355 F.3d at 829 (noting that where a petitioner has not challenged “an obligation expressly imposed by the CAA, [the petitioner] needed to raise its objections during the administrative proceeding [in order to] provide EPA an opportunity to consider the issue before asserting, after the fact, that EPA was arbitrary for failing to do so.”). GHASP does not dispute that it failed to raise an objection to the application of weight of evidence September 2, 4, 6; nor does it assert that it raised a more general challenge to the allegedly “quantitative manner” in which TCEQ applied the weight of evidence. Further, after our review of the administrative record, we were unable to find such an objection. Accordingly, we find that GHASP has in fact waived review concerning the application of weight of evidence analysis to results for September 2, 4, and 6.
This court considers only, therefore, whether the EPA was arbitrary or capricious in approving TCEQ’s exclusion of August 31 under its weight of evidence analysis. We conclude that it was not. While this court has yet to consider a similar circumstance where the EPA has declined to include a date in its attainment analysis because it is “atypical,” in
Sierra Club v. EPA,
356 F.3d 296 (D.C.Cir.2004), the D.C. Circuit upheld a similar action by the EPA. In that case, the D.C. Circuit not only approved the use of supplemental statistical analysis to correct for over-predictions in modeling, it also concluded that the EPA could exclude a day from the attainment analysis that presented unusual conditions.
Id.
at 305. In that case, the EPA determined that the excluded date was “too anomalous to demonstrate nonattainment” after finding that the date was the “13th most severe ozone producing day in 44 years.”
Id.
In upholding the EPA’s decision, the D.C. Circuit explained that this adjustment merely accounted for a “base day that appeared] to be a statistical outlier.”
Id.
at 306.
Similarly, here, the EPA explained that August 31 was excluded because the temperature was a record high and the heat taken together with atypical wind patterns created an unusually high ozone level in the Houston area. 70 Fed.Reg. 58,125/1. Texas submitted the following analysis as weight of evidence: August 31 rare meteorology; additional reductions that were not modeled; comprehensive ozone metrics and ambient trends; alternative design value and addressing short-term excursions; and unusual wildfire activity. 70 Fed.Reg. at 58,124/1. Based on this explanation and the substantial deference given to the EPA on these issues, we are unable to conclude that the Agency was unreasonable for allowing TCEQ to exclude August 31 during its application of the weight of evidence analysis.
III.
GHASP argues that the changes to Texas’s control strategy approved in the MCR SIP violate the non-interference language of Section 110(Z) of the Clean Air Act. In support of this contention, GHASP relies primarily on the fact that the MCR SIP relaxed the controls on industrial NOx emissions from the 90 percent level established in the 2001 SIP to 80 percent and dropped four requirements that had been included in the 2001 SIP including: (1) the requirement for vehicle inspection and maintenance in three rural counties; (2) restrictions on morning operations for commercial lawn services; (3) planned reductions in speed limits; and (4) restrictions on idling for heavy duty diesel vehicles. GHASP argues that these actions by themselves are sufficient to show a violation of Section 110(7). CAA section 110(7) provides:
Each revision to an implementation plan submitted by a State under this chapter shall be adopted by such State after reasonable notice and public hearing. The administrator shall not approve a revision of a plan if the revision would interfere with an applicable requirement concerning attainment and reasonable further progress (as defined in section 7501 of this title) or any other applicable requirement of this chapter.
42U.S.C. § 7410(Z).
“Reasonable further progress” is defined by statute to mean “such annual incremental reductions in emissions of the relevant air pollutant as are required by this part or may reasonably be required by the Administrator for the purpose of ensuring attainment of the applicable national ambient ah* quality standard by the applicable date.”
1$ U.S.C.
at § 7501(1). Because the definition of reasonable further progress is defined within the statute, the question of whether the MCR SIP
violates Section 110(7) turns on the meaning of “would interfere.” The statute itself does not define “would interfere.” Because Congress has not directly spoken on this issue, under
Chevron
step two we will defer to the EPA’s interpretation unless it is arbitrary or capricious.
GHASP contends that the EPA must show that results under the MCR SIP are better than results under the 2001 SIP to comply with Section 110(7 )• However, nothing in the plain language of Section 110 (l) supports this interpretation. The EPA and BCCA Appeal Group in its amicus brief, urge this court to adopt the interpretation of interference accepted by the Sixth Circuit in
Kentucky Resources Council, Inc. v. EPA,
467 F.3d 986 (6th Cir.2006), in which that court held that the EPA may approve a SIP revision “unless the agency finds it will make air quality worse.” According to the EPA, under Section 110(Z), the EPA’s approval of Texas’s MCR SIP does not constitute interference as long as the revisions to the SIP were properly substituted by new measures that would offset the effect of those measures removed from MCR SIP. Thus, changes to a SIP, either dropping measures or reducing measurement requirements, are not by themselves sufficient to prove interference. Rather, one must show that the substitute measures are not at least equivalent to the previous measures in achieving attainment. We find the EPA’s interpretation of Section 110(í) reasonable and thus refuse to substitute our judgment for that of the Agency’s. This court need now determine only whether the EPA’s determination that the new measures would offset changes to the 2001 SIP are also entitled to deference.
GHASP contends that the reduction in NOx controls from 90 percent to 80 percent and the removal of four requirements from the control strategy are alone sufficient to demonstrate violation of Section 110(Z)’s non-interference language. Based on modeling, the EPA concluded that the reduction in NOx controls and the removal of the four requirements were offset by the addition of controls for HRVOCs. 71 Fed. Reg. 52,675. GHASP offers nothing to contradict the EPA’s conclusion, rather it argues that this argument is not compelling because the EPA should have required Texas to compare “how the 2001 SIP controls compare with the MCR SIP controls in achieving the one-hour standard.” However, again, GHASP provides no support for its claim that a particular type of comparison must be made. Further, the statute does not indicate any specific methodology which must be applied to determine interference. The record shows that the EPA based its noninterference determination on modeling based on the 2001 SIP control strategy as compared to the MCR SIP. 71 Fed.Reg. At 52,689/1 (explaining that “the noninterference modeling included the control strategies listed in the December 2000 SIP together with updated inventories and updated methodologies utilizing the 2000 episode.”). Because the EPA’s scientific determinations are entitled to deference, and the Agency’s approval of the MCR SIP was based on modeling supporting this determination, its conclusions are not arbitrary or capricious.
See BCCA,
355 F.3d at 833-34.
Finally, GHASP argues that under the MCR SIP, Texas fares worse with regard to the eight-hour strategy than it did under the 2001 SIP. GHASP correctly points out that analysis of the eight-hour standard showed that a number of monitors showed higher ozone levels under the MCR SIP than under the SIP. 71 Fed. Reg. 52,688. However, even in light of this fact, the EPA determined that overall, the new strategy performed better under the eight-hour standard. 71 Fed.Reg. At
52,689/1. In support of this, the EPA noted that two of the three ozone metrics showed improvement under the new strategy.
Id.
This scientific determination is reasonable and entitled to deference.
IV.
Section 307(f) of the CAA provides that in a proceeding for judicial review of an emission standard promulgated under the Act, a court may award reasonable attorney’s fees “whenever it determines that such an award is appropriate.” Having denied GHASP’s petition for review, we have no choice but to deny its request for costs and attorneys’ fees.
Ruckelshaus v. Sierra Club,
463 U.S. 680, 694, 103 S.Ct. 3274, 77 L.Ed.2d 938 (1983) (holding that “absent some degree of success on the merits by the claimant, it is not ‘appropriate’ for a federal court to award attorney’s fees under § 307(f)”).
CONCLUSION
For the foregoing reasons, GHASP’s petition for review is denied and its request for costs and attorneys’ fees is also denied.