In this appeal the State of New York challenges the Environmental Protection Agency’s (EPA) approval of a revision to Illinois’ State Implementation Plan (SIP), which permits the Kincaid Power Station in Christian County to increase its sulfur dioxide emissions.1 New York alleges that the EPA action was arbitrary and capricious because the emission relaxation violates numerous provisions of Section 110 of the Clean Air Act, 42 U.S.C. § 7410. We disagree. We find that the EPA complied with all Section 110 requirements and, thus, uphold EPA approval.
Under the Clean Air Act, as amended in 1977, 42 U.S.C. § 7401 et seq., the federal and state governments share responsibility for air pollution control. The EPA is to establish nationwide air quality standards and the states are responsible for developing implementation plans to attain and maintain these standards. 42 U.S.C. §§ 7408, 7409, 7410. The EPA sets national ambient air quality standards (NAAQS’s) which are maximum limits on open air concentration of numerous pollutants; each state then submits an implementation plan outlining how it will maintain air quality in the state consistent with the NAAQS’s.
The Act requires the EPA to evaluate state plans according to the criteria enumerated in Section 110(a)(2). 42 U.S.C. § 7410(a)(2). Any revision to a previously approved plan must be evaluated by the same criteria. The EPA may not approve a plan that prevents another state from at-*442tainting or maintaining any NAAQS or that interferes with a state’s Prevention of Significant Deterioration Program.
Contending that the increased sulfur dioxide emissions in Illinois are causing or contributing to its nonattainment of the NAAQS’s, New York alleges that the EPA violated its statutory duty because it approved the revision of the Illinois SIP without considering its interstate effects. New York reads the statute to require the EPA to apply the Section 110(a)(2) criteria to the revised SIP as a whole; by contrast, the EPA here considered the isolated effects of increased emissions in the immediate area of an individual plant. Under New York’s analysis the EPA had no authority to approve the revision because it failed to analyze the effects of the Kincaid emissions in conjunction with emissions from all other sources controlled by the Illinois SIP. Further, New York contends that the EPA failed to build an adequate factual record to support its action.
I
New York maintains that the EPA cannot satisfy the interstate pollution provisions of the Clean Air Act simply by assessing the local impact of emissions from a single source. It asserts that the EPA finding that emissions from the Kincaid plant will not interfere with another state’s compliance with the NAAQS for sulfur dioxide is inadequate for several reasons.
First, New York argues that even if the increased emissions do not significantly affect the air in the Kincaid vicinity, these emissions, in combination with sulfur dioxide emissions from other parts of the state, may cause impermissible levels of sulfur dioxide outside the state. For this reason New York maintains that Illinois’ SIP must be considered as an organic entity and that any revision must be reviewed to determine whether a change in emissions from one source, when analyzed against the aggregate emissions from all SIP sources, will adversely affect national attainment of the Clean Air Act’s ambient air quality goals.
New York also contends that the EPA erred by failing to determine the impact of the increased sulfur dioxide emissions on the attainment and maintenance of total suspended particulate NAAQS. New York emphasizes that, because sulfur dioxide emissions convert into sulfate particulates in the atmosphere, the EPA abused its discretion by restricting its air quality analysis to direct emissions from the Kincaid plant while ignoring particulates formed in the atmosphere from those emissions.
Finally, New York contends that the EPA did not carry out its statutory duty to evaluate the long-range interstate effects of the increased emissions at the Kincaid plant because it evaluated the revision by using a short-range, in-state model which can assess the sulfur dioxide impacts of the emissions only within a fifty mile radius. New York strenuously argues that the 1977 amendments to the Act compel the EPA to assess the long-range effect of increased emissions by using the most accurate modeling techniques available and maintains the EPA ignored available long-range models which can assess the interstate effects of the Kincaid emissions.
We agree with New York that the Clean Air Act Amendments of 1977 impose stringent requirements with respect to EPA approval of SIP revisions. We note, however, that both the Second and Sixth Circuits have considered and rejected New York’s contention that these amendments require the EPA to determine the cumulative interstate impact of all sulfur dioxide emission sources within a state when considering a proposed revision involving a single source. New York v. EPA, 710 F.2d 1200 (6th Cir.1983). Relying on Connecticut v. EPA, 696 F.2d 147 (2d Cir.1982), the Sixth Circuit concluded that the focus in evaluating a revision to a SIP should be on emissions from the single pollution source and that “it [is] within EPA’s discretion to determine the scope of its inquiry in connection with a proposed revision so long as the requirements of Section 110 are met.” New York v. EPA, 710 F.2d at 1204 (6th Cir.1983). We agree and adopt that analysis here.
*443A relaxation of the emission limits at Kincaid may mean more than just increased emissions directly pouring out of Kincaid’s chimneys. Under certain atmospheric conditions, these sulfur emissions may combine with other airborne molecules to form sulfates, thereby further polluting the air. New York contends that the statute requires the EPA to model the effects of Kincaid sulfur dioxide emissions not only on out-of-state sulfur dioxide levels but also on out-of-state total suspended particulate levels. New York’s reading of the statute was adopted by the Second Circuit, which held that Section 110(a)(2)(E)2 “require[d] the EPA to consider the effect of a revision on one state’s implementation plan upon all NAAQS’s in other states.” Connecticut v. EPA, 696 F.2d 147, 163 (2d Cir. 1982). Despite this broad reading of Section 110(a)(2)(E), the Connecticut court concluded that the EPA had not violated the Clean Air Act with respect to total suspended particulate levels because it had, to the limited extent possible with current modeling tools, considered the effects of increased emissions on interstate total suspended particulate concentrations.
Similarly, in this case, the EPA stated that it had not yet adopted or approved any models that can accurately predict particulate concentrations resulting from sulfur dioxide emissions. Moreover, the study it did conduct indicated that the ambient air quality effects from the increased emissions at Kincaid would be greatest within the immediate area of the plant, the greatest concentration occurring only 1.7 kilometers from the plant and decreasing sharply beyond that distance. The EPA noted that the marked decrease in concentrations over a relatively short distance indicated that the emissions were quickly dispersed and, thus, unlikely to result in significant ambient concentrations beyond the immediate area. Given Kincaid’s distance from the nearest state border, the EPA concluded that “SO2 emissions from [Kincaid] would not prevent attainment or maintenance of SO2 standards in New York or other downwind states.” Joint Appendix Vol. VIII, p. 1759; Doc. No. 62, 47 Fed.Reg. 8773 (1982) (to be codified at 40 C.F.R. Part 52). Further, it explained that it could not, under the situation here, evaluate the effects on the increased emissions on interstate total suspended particulate concentrations because reliable models for making such an analysis have not yet been developed. Thus, while we agree with the Second Circuit that the statute requires the EPA to consider the effect of a revision to a state plan upon all NAAQS’s in other states whenever possible, we cannot require the EPA to do the impossible.
Along with the proposal to revise its SIP, Illinois submitted an analysis of the impact of the proposed increased emissions which had been conducted by Commonwealth Edison. This study analyzed the impact of the increased emissions by applying the CRSTER model, a modeling technique approved by the EPA. The EPA also conducted independent modeling and evaluation. Although the results of the two evaluations varied slightly, both indicated that the ambient air quality impacts of the increased emissions would be greatest within the immediate vicinity of the plant. Joint Appendix Vol. VIII, p. 1759, Doc. No. 62, 47 Fed.Reg. 8773 (1982) (to be codified at 40 C.F.R. Part 52).
New York challenges the use of the CRSTER model. It notes that the EPA’s Guideline on Air Quality Models recognizes that the CRSTER model is inadequate to measure the impact of emissions beyond fifty kilometers from the source. Thus, New York reasons, the EPA, by its own admission, did not conduct a reliable study. *444New York asserts that because the EPA has a nondiscretionary duty to develop and utilize the most accurate and feasible modeling techniques available, the EPA was required to use the computer modeling techniques suggested by New York or to defer any action on the proposed revision until the EPA developed and approved other long-range models.
New York argues that computer based modeling techniques exist which can analyze the effect of interstate air pollution and devotes a considerable portion of its brief to arguing that the EPA irresponsibly ignored these techniques. Petitioner’s br. at 40-48. It contends that the EPA’s reasons for refusing to use these techniques— that the Agency has not formally approved these models for use and that the models are not sufficiently developed to warrant use for regulatory purposes — are not valid.
Our task is to determine if the EPA’s approval of the revision is “arbitrary or capricious, constitutes an abuse of agency discretion or is otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). Thus, our scope of review is a narrow one and we may not substitute our judgment for that of the agency. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971). Moreover, the EPA’s interpretation of the statute it is charged with administering, particularly in technical areas, must be accorded substantial deference. Baltimore Gas & Electric Co. v. Natural Defense Council, Inc., - U.S. -, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983); Cincinnati Gas & Electric Co. v. EPA, 578 F.2d 660 (6th Cir. 1978), cert. denied, 439 U.S. 1114, 99 S.Ct. 1017, 59 L.Ed.2d 72 (1979).
The statute does not require the EPA to adopt any specific modeling techniques. The EPA chose to use a short-range model rather than one of the modeling techniques preferred by New York because it concluded that despite its limited scope the short-range model was most reliable. This is the kind of technical decision particularly within the realm of Agency expertise. Connecticut v. EPA, 696 F.2d 147,157 (2d Cir.1982).
New York is asking this court to substitute New York’s technical judgment for that of the Agency; we do not have that authority. We must defer to the EPA’s choice of reference models if that choice conforms to Agency guidelines. New York v. EPA, 710 F.2d at 1204 (6th Cir.1983). The Agency guidelines do not mandate the use of any particular model. Indeed, the Agency’s manual emphasizes that its specific recommendations are not “rigid requirements” and that each situation should be treated on a case-by-case basis. Connecticut v. EPA, 696 F.2d 147, 159 (2d Cir.1982). Thus, we cannot say that the EPA acted arbitrarily or capriciously in rejecting the predictive devices favored by New York.
We conclude that the EPA did not violate the Act by approving the revision based on an analysis only of the effect of increased sulfur dioxide emissions in the immediate vicinity, New York v. EPA, 710 F.2d at 1204 (6th Cir.1983), because it: (1) analyzed the impact of the increased emissions on the immediate ambient air quality; (2) explained its inability to measure long-range effects; and (3) supplied an analysis that strongly suggests any long-range effects would be insignificant.
New York also asserts that the EPA erred by failing to perform a prevention of significant deterioration (PSD) analysis to assess whether the Kincaid revision would cause significant air quality degradation in other states. Under Section 110(a)(2)(J), a revision to an SIP may not be approved if the revision would cause serious deterioration of the existing air quality of any state. Impermissible air quality deterioration is measured by the increase of air pollution concentrations over the baseline level of air quality. 42 U.S.C. § 7473. The baseline is the level of air quality existing at the time the first PSD application is submitted. 42 U.S.C. § 7479(4).
The EPA contends that a PSD analysis was not required because no baseline had *445yet been set for Christian County. It emphasizes that no PSD permit application had been filed for Christian County at the time the Kincaid revision was submitted and the revision did not require a permit. Thus, no calculation of baseline concentrations was made. Because baseline concentrations had never been calculated, the EPA maintains that it was impossible to calculate whether the Kincaid revision would cause impermissible increments in pollution over the baseline level.
New York rejects the EPA’s explanation for its failure to conduct a PSD analysis. It argues that a permit application is not the only circumstance which triggers a PSD analysis and maintains that other sections of the statute, as well as EPA’s own regulations, mandate that the revision be analyzed to insure that it will not cause an impermissible increase in pollution.
New York has not cited any case which requires a PSD analysis to assess incremental increases in pollution before there has been a permit application triggering calculation of the baseline concentrations. While we agree with New York that the EPA may approve only those SIP revisions which will not cause an impermissible increase in pollution, an EPA analysis based on the eleven criteria specified in Section 110 accomplishes that goal. Moreover, logic dictates that increments over the baseline concentration cannot be calculated before the baseline concentrations themselves have been calculated. Accordingly, we hold that where baseline concentrations have not been computed, no duty to conduct a PSD analysis exists.
II
New York also alleges that the EPA’s approval of the Kincaid revision is not supported by an adequate factual record. In support of this contention it argues that the record is devoid of data relating to interstate air pollution. It argues that EPA failed to adequately investigate interstate impacts or to employ the proper assessment tools. These are essentially the same arguments which New York raised in support of its contention that the EPA failed to make the required Section 110 determinations. We have already considered and rejected these arguments. Accordingly, we need not repeat them here.
The gravamen of New York’s complaint is that the aggregate air quality impact from many out-of-state sources is adversely affecting its air. New York has filed eight cases similar to this one in the Sixth and Seventh Circuits, each seeking to reverse EPA approval of increased emissions from numerous sources. New York has also filed several petitions under Section 126(b) of the Clean Air Act, 42 U.S.C. § 7426(b), raising essentially the same claims. Section 126(b) provides for an agency hearing to consider interstate pollution impact. The EPA is currently conducting a Section 126(b) proceeding to consider several of New York’s petitions, including one concerning the Kincaid plant.
A Section 126 proceeding “providefs] New York with the ‘consolidated forum’ it [seeks] and provide[s] the EPA with a proceeding at which the cumulative impact of all the claimed violations of interstate limitations on emissions [can] be considered together.” New York v. EPA, 710 F.2d at 1205 (6th Cir.1983). That is the proper forum in which to challenge revisions of emission levels from separate sources to determine if these aggregate emissions create an impermissible interstate impact.
We hold that the EPA did not violate Section 110 of the Clean Air Act in approving the revision on the basis of an analysis limited solely to the effect of increased sulfur dioxide emissions from a single source. Accordingly, the petitions for review are denied.
Denied.
. Commonwealth Edison was permitted to intervene as a party-respondent and filed a brief supporting the EPA’s approval. Peabody Coal Company, and the States of Minnesota and Rhode Island filed amicus curiae briefs. We have considered these briefs in rendering our decision.
. 42 U.S.C. § 7410(a)(2)(E) provides:
(E) it contains adequate provisions (i) prohibiting any stationary source within the State from emitting any air pollutant in amounts which will (I) prevent attainment or maintenance by any other State of any such national primary or secondary ambient air quality standard, or (II) interfere with measures required to be included in the applicable implementation plan for any other State under part C to prevent significant deterioration of air quality or to protect visibility.....
(Emphasis added.)