Petitioner, Northern Plains Resource Council (the “Council”), 1 seeks a review of the U.S. Environmental Protection Agency’s [“EPA”] decision 2 conditionally granting a “prevention of significant deterioration of air quality” [PSD] permit, as required by the Clean Air Act 3 and by EPA regulations, 4 to a consortium of investor-owned utilities (the “Consortium”) 5 for two coal-fired electric power plants under construction at Colstrip, Montana (“Colstrip Units Nos. 3 and 4”). Two rather technical issues are raised: 6

The first is whether the EPA’s approval of the Consortium’s permit application is improper because the Colstrip Units will not utilize the “best available control technology” [hereinafter BACT] as required by Section 165(a)(4) of the Act, 42 U.S.C. § 7475 (a)(4), and the EPA’s regulations, 40 CFR § 52.21 (j)(2) (1980). Two sub-issues are raised by the Council in its challenge to EPA’s acceptance of the Colstrip plan as BACT. First, the Council argues that, because the projected emissions of particulates exceed the applicable 1978 new source performance standards [NSPS], EPA has violated the dictates of § 165(a)(3XC) of the Act which requires that BACT must not result in emissions in excess of any applicable standard of performance under the Act. *1351 EPA’s response is that the plant would meet the 1971 NSPS standards, and that these are the applicable standards. We conclude that the appropriate standard is the 1971 standard.

The other BACT sub-issue is whether, because more sophisticated techniques (i.e. the “baghouse”) exist, acceptance of the Colstrip plan as BACT was without rational basis. The statute provides that BACT is to be determined on a case-by-case basis, taking into account energy, environmental, and economic impacts. See 42 U.S.C. § 7479 (3). Given the constraints in this case, including the need to protect the Class I SO2 increment on the nearby reservation, the EPA’s action was not irrational.

The second major issue raised by the Council is a challenge to EPA’s refusal to validate its air dispersion model with monitoring data proffered by the Council. 7 Because the data was collected on low terrain, it has no relevance to the model used by the EPA to project SO2 emissions on high terrain, the only area where projections approached the relevant PSD increments. Although the monitored readings differed from those the model would have predicted, the Council does not show that the EPA’s actions were arbitrary and capricious or that the agency ignored a significant threat to the relevant PSD increment, given the lower projected SO2 concentrations on low terrain.

We therefore affirm the EPA.

I. BACKGROUND

A. Facts Relating To The Colstrip Units

Colstrip Units Nos. 3 and 4 are part of a complex of four power plants, the first two of which were completed in 1976. 8 Plans to build Units Nos. 3 and 4 were formalized in 1973. On January 25,1974, the Consortium entered into a contract for the construction of two steam generators for Colstrip Units Nos. 3 and 4. Other additional significant preliminary steps for construction of the units were taken in 1974. 9 Thereafter, however, the Consortium ran into significant delays in obtaining three of the requisite state and federal permits.

Originally, the Consortium took the position that it was not required to obtain a PSD permit under the EPA’s 1974 PSD regulations because the Colstrip Units fell within the grandfather provision, having commenced construction prior to June 1, 1975. That position was initially upheld by the district court in Montana Power Co. v. EPA, 429 F.Supp. 683, 703 (D.Mont.1977). Nevertheless, on July 1, 1976, under protest and reserving its rights, the Consortium applied for a PSD permit. Thereafter, the 1977 amendments to the Act were enacted. The EPA held that the statutory PSD program clearly required that the Consortium obtain a PSD permit. 42 Fed. Reg. 60784 -86 (1977). This court agreed, reversing the district court’s ruling and upholding the EPA’s subsequent determination. Montana Power Co. v. EPA, 608 F.2d 334 , 357-58 (9th Cir. 1979).

During this period, on August 5, 1976, the Northern Cheyenne Tribe, whose reservation is in the vicinity of the Colstrip plants, announced its intention to redesignate its reservation from Class II to Class I. See 42 U.S.C. §§ 7472 , 7474. The Consortium’s PSD permit application was held in abeyance by the EPA pending the outcome of the redesignation process. On August 5, 1977, the redesignation was approved. 42 *1352 Fed. Reg. 40695 (1977). 10 The Consortium’s PSD permit was denied on September 30, 1977, because the EPA’s air quality modeling indicated that the newly applicable Class I sulfur dioxide increments on the Northern Cheyenne Reservation would be violated by Colstrip Units Nos. 3 and 4. 11

On November 14, 1977, the Consortium petitioned the EPA for reconsideration. Although initially issuing a public notice of its proposed approval of the permit, on June 12,1978, the EPA reversed itself and denied the permit after holding public hearings on the matter and conducting further analyses. The Consortium then modified its plans for the pollution abatement system at Colstrip Units Nos. 3 and 4, and on February 15, 1979 submitted a revised application. 12 The EPA issued a public notice of its proposed approval of the revised PSD permit application. At the hearings on the proposed permit, the Council proffered materials purportedly demonstrating that the revised plans did not incorporate the best available control technology and that the air dispersion model used in the validation process was inaccurate. After considering the information presented at the hearings, the EPA issued the PSD permit on September 11, 1979, subject to several conditions. 13 This petition challenging that decision followed.

II. DISCUSSION

A. Do The Colstrip Units Employ The Best Available Control Technology As Required By Section 165(a)(4) Of The Act?

The Council raises two arguments in support of its position that the Colstrip Units do not utilize the requisite BACT. First, it asserts that the pollution control system at Colstrip Units Nos. 3 and 4, which were approved by the EPA, will permit a particulate emission rate which exceeds the standard allowed by the applicable NSPS. This initial argument is based on the premise that the 1978 NSPS regulations for fossil-fuel-fired generating units, 40 CFR §§ 60 .-40a-60.49a (1980), are controlling in the present situation rather than the 1971 NSPS regulations, 40 CFR §§ 60.40-60.46 (1980). Second, the Council argues that the EPA in its approval process improperly ignored the existence of new and improved abatement technology, for removing particulate matter. The abatement technology, which the Council urges should be employed, is a “baghouse control system”. 14

1. The Relationship Between The NSPS Regulations And The PSD Program

The statutory PSD program provides, inter alia, that no construction on a major emitting facility may be commenced after August 7,1977 in an area designated under the PSD classification system unless:

Section 165(a)(3) of the Act, 42 U.S.C. § 7475 (a)(3). The EPA would clearly be acting contrary to the statutory PSD pro *1353 gram in issuing a PSD permit to a facility which would produce emissions in excess of an applicable new source performance standard.

Likewise, Section 165(a)(4) of the Act, 42 U.S.C. § 7475 (a)(4), requires the proposed facility to be subject to the best available control technology for each pollutant subject to regulation under the Act. Under the statutory definition of the term “best available control technology”, it is stated that:

Section 169(3) of the Act, 42 U.S.C. § 7479 (3). 15 Thus, the grant of a PSD permit to a facility that violated applicable NSPS regulations would be contrary to that statutory directive as well.

From the materials presented to this court, it appears that the Colstrip Units Nos. 3 and 4 will fail to meet the 1978 NSPS emission standards for particulate matter and nitrogen oxides, but they do satisfy the 1971 NSPS regulations. 16 The issue then is which set of NSPS regulations is applicable to the Colstrip Units.

The two regulations involved herein are mutually exclusive. As stated in 40 CFR § 60.40 (e) (1980), “[a]ny facility covered under Subpart Da [the 1978 NSPS regulations] is not covered under this Subpart [the 1971 NSPS regulations].” As noted in the Act, the applicability of the various regulations rests on whether construction or modification of the affected facility began after the publication of the regulations or, if earlier, the proposed regulations. Sections 111(a)(2) and (b)(6) of the Act, 42 U.S.C. § 7411 (a)(2) and (b)(6). The 1971 NSPS regulations would apply if construction on the Colstrip Units Nos. 3 and 4 commenced on or before September 18, 1978. See 40 CFR § 60.40 (c) and (e) (1980). The 1978 NSPS regulations would apply if construction commenced after that date. See 40 CFR § 60 .40a(a)(2) (1980).

2. The Definition Of “Commenced” Under The NSPS Regulations And Its Application Herein

Section 111 of the Act, 42 U.S.C. § 7411 , which delineates the NSPS program, does not contain a statutory definition of the word “commenced”. Nor does Part A of the Act in which Section 111 is located. *1354 Part C of the Act, which contains the statutory PSD program, does provide a definition of “commenced” that consists of two component requisites:

Section 169(2)(A) of the Act, 42 U.S.C. § 7479 (2)(A). However, the beginning phrase of Section 169 (“For purposes of this part — ”), seems expressly to apply the definitions delineated therein only to the statutory PSD portion of the Act.

The EPA has issued regulations implementing Section 111 of the Act that include a “General Provisions” subpart containing a definition of “commenced”. That definition has only one requisite component, /. e. that the

40 CFR § 60.2 (1980).

The Council makes two arguments as to the application of the 1978 NSPS regulations to Colstrip Units Nos. 3 and 4. First, it contends that the definition of “commenced” in the PSD portion of the Act should be applied to the NSPS provision, Section 111 of the Act. Second, the Council contends that the EPA failed to articulate adequately its reasons for granting the PSD permit and that there is insufficient evidence in the record that construction “commenced" on Units Nos. 3 and 4, regardless of which definition of that word is used, prior to September 19, 1978 (the effective date of the 1978 NSPS regulations).

a. Is The PSD Definition Of “Commenced” Applicable To The NSPS Program?

The Council bases its argument as to the applicability of the PSD definition of “commenced” to Section 111 on two grounds:

If the PSD definition is used, the Council contends that construction on Colstrip Units Nos. 3 and 4 did not “commence” prior to the effective date of the 1978 NSPS regulations since the Consortium had failed to obtain two of the requisite pre-construction permits as required by Section 169(2)(A), i. e. the PSD permit itself and an amended certificate as required under the Montana Major Facility Siting Act, Mont. Code Ann. §§ 75-20-213 and 75-20-219 (1979). 17 To follow the Council’s line of reasoning, be *1355 cause construction has not commenced, the 1978 NSPS regulations are controlling and therefore the PSD permit was improperly granted by the EPA because the Colstrip Units do not meet the 1978 new source performance standards for particulate matter. 18

i. Statutory Construction

It is an accepted rule of statutory construction that the same word or phrase “is presumed to have the same meaning when used in different parts of a statute.” Ghu-gach Natives, Inc. v. Doyon, Ltd., 588 F.2d 723, 725 (9th Cir. 1978). That presumption, however, may be rebutted if the same word or phrase is used in different parts of a statute with manifestly different intent. See District of Columbia v. Carter, 409 U.S. 418,420-21 , 93 S.Ct. 602, 604-05 , 34 L.Ed.2d 613 (1973).

The Council’s contention that the definition of “commenced” in Section 169(2)(A) of the Act is applicable to Section 111 by reason of the aforementioned rule of statutory construction is unpersuasive. First, the definitions in Section 169 are expressly limited by the introductory phrase “For purposes of this part — ” to Part C of the Act, i. e. the statutory PSD program. Section 111 is located in Part A of the Act, which establishes the general programs for air quality control and emission limitations. From the face of the statute, there is no manifest congressional intent to apply the definitions in Section 169 any further than Part C of the Act.

It is significant that the word “commenced” was used in Section 111 prior to the enactment of Section 169 in the Clean Air Act Amendments of 1977. Section 111(a)(2) of the Act, 42 U.S.C. § 7411 (a)(2), which employs the word in the context of defining the term “new source”, was enacted pursuant to the Clean Air Act Amendments of 1970. See H.R. Conference Report No. 91-1783, 91st Cong.,2d Sess. 9 (1970), U.S. Code Cong. & Admin. News 1970, p. 5356, reprinted in Vol. I U.S. Library of Congress, Environmental Policy Division, Legislative History of the Clean Air Amendments 1970 [hereinafter “Legislative History 1970 Amendments”] 159 (1974). In December of 1971, the EPA promulgated a definition of “commenced” for purposes of the NSPS regulatory program. 36 Fed.Reg. 24877 (1971). The EPA has consistently made decisions based on its definition. 19 It is therefore difficult to accept the Council’s argument that the definition of “commenced”, which was enacted in 1977 as an element of the newly created statutory PSD program in Part C of the Act and given a rather unusual technical meaning, 20 was intended by Congress to be applicable automatically to Section 111(a)(2) in Part A of the Act, which was enacted seven years earlier, when that definition is specifically limited on its face by Congress to Part C.

ii. Differences Between The NSPS And PSD Programs

While the NSPS program and the PSD are both interrelated parts of a comprehen *1356 sive federal legislative effort to protect and enhance this nation’s air quality, the two programs play different roles in achieving that broad general goal. This further undercuts the argument that Congress intended the PSD definitions to apply automatically to NSPS.

The focus of the NSPS program under the EPA’s regulations is upon the “affected facility” component in a stationary source, i. e. the particular apparatus to which a standard is applied. 40 CFR § 60.2 (e) (1979). The NSPS program is therefore equipment oriented. On the other hand, the PSD program covers the whole stationary source, and focuses on where the plant will be located and its potential effect on its environs. The PSD program is therefore site oriented.

Due to the differences in the purposes and structure of the NSPS and the PSD programs, the definition of “commenced” for each of those programs need not be exactly the same. Cf. Alabama Power Co. v. Costle, 636 F.2d 323 , 13 ERD 1993, 2040-41 (D.C.Cir.1979) (EPA can adopt different definition for “source” for PSD than NSPS applications due to differences in program purpose and structure).

iii. Legislative History

The legislative history of the Act does not support the Council’s position herein. In Section 111 the word “commenced” is utilized in two places that have a bearing on the present controversy. As discussed above, the first instance is in subsection (a)(2), 42 U.S.C. § 7411 (a)(2), which was enacted in the 1970 amendments and contained the only reference to the word “commenced” in Section 111 at that time. Nowhere in the legislative history of the 1970 amendments is the meaning of that word discussed, nor is there any suggestion that the word incorporates a preconstruction permit requirement. Indeed, the indications are entirely to the contrary. 21

A second instance of the use of the word “commenced” in Section 111 came with the 1977 amendments to the Act when subsection (b)(6) was added. Conference Report, H.R.Rep. No. 95-564, 95th Cong. 1st Sess. (1977), U.S. Code Cong. & Admin. News 1978, p. 1077, reprinted in 3 U.S. Library of Congress, Environmental Policy Division, Legislative History of the Clean Air Act Amendments of 1977 [hereinafter Legislative History 1977 Amendments] at 398. Subsection (b)(6) instructed the EPA to establish within a year a revised set of NSPS incorporating both emission limitation and percentage reduction criteria. It provided that any fossil-fuel-fired stationary source that commenced construction prior to the date of publication of the proposed revised standards would not be required to comply with the new standards. There is no specific discussion in the legislative history as to the meaning of the word “commenced” as used in this second context.

The EPA’s definition in 40 CFR § 60.2 (i) (1979) of the word “commenced”, for purposes of the NSPS program, is entitled to substantial deference since the EPA is the agency charged with the administration of the Act. See Udall v. Taiiman, 380 U.S. 1, 16 , 85 S.Ct. 792, 801 , 13 L.Ed.2d 616 (1965).

The Council, in support of its argument, cites to a “Summary and Statement of Intent” made by Representative Rogers when he presented certain technical and conforming amendments to the Clean Air Act on *1357 November 1, 1977 after the passage of the substantive amendménts to the Act in August of 1977. 22 In that statement, Representative Rogers stated, inter alia:

123 Cong.Rec. H11958 (daily ed. Nov. 1, 1977), reprinted in [1977] U.S. Code Cong. & Adm. News 3648, 3667. We believe that this passage is consistent with our view that Congress intended to leave to the agency any decision about a change conforming the definition of the word “commenced” in the two programs.

We therefore cannot conclude that the PSD definition of the word “commenced” was applicable to the NSPS program at any time relevant to this litigation. 23

b. When Did Construction Commence On Colstrip Units Nos. 3 And 4 For Purposes Of The NSPS Program?

As this court previously noted in Montana Power Co., 608 F.2d at 339, n.6: 24

The total amount of outstanding contracts as of May 25, 1977, which included primarily the steam generating and related equipment, exceeded $123 million. 25

The “affected facility” which the 1971 NSPS regulations govern include “fossil-fuel-fired steam generating unit[s] of more than 73 megawatts heat input rate,” 40 CFR § 60.42 (a)(1) (1980), such as the generating units built for Colstrip Units Nos. 3 and 4. For purposes of the NSPS program, construction on such generating units may be “commenced” at the point when “an owner or operator has entered into a contractual obligation to undertake and complete, within a reasonable time, a continuous program of construction.... ”

40 CFR § 60.2 (1980)

In this case, the two steam generators provided for in the contract with Combustion Engineering, Inc. comprise the heart of *1358 the generating units and the largest expenditure required to construct the affected facility. That contract does contain provisions of the completion of construction on the units within a reasonable time period. 26 Moreover, the boiler apparatus for Colstrip Unit No. 3 was completely fabricated and delivered to the construction site by September 1978.

We therefore agree with the EPA’s conclusion that the Consortium “commenced construction” on the generating units for purposes of the NSPS regulations prior to September 18, 1978, and accordingly that the 1971 NSPS regulations apply. Colstrip Units Nos. 3 and 4 will satisfy those standards and thus the BACT requirement in the PSD program is not violated on that ground.

The Council argues that the record below does not support the EPA’s conclusion that construction on the affected facilities at Colstrip Units Nos. 3 and 4 commenced prior to September 1978, nor does it disclose the EPA’s reasons for reaching that conclusion. We note that the record did include a January 18,1979 letter from John Peterson, an attorney for Montana Power Company, to C. A. Baldwin at the EPA. That letter contained a list of the various orders and contracts that the Consortium had entered into up through May 25, 1975, relating to the construction of the generating units at Colstrip Units Nos. 3 and 4. It was also noted therein that copies of each contract had been “forwarded to the EPA in the PSD proceeding”. Also in the record below was a March 22, 1979 (erroneously dated March 22, 1978) letter from Lance Vinson (then Director of EPA’s Enforcement Division) to John Peterson that stated:

The EPA is obligated to articulate a “rational connection between the facts found and the choice made”. Burlington Truck Lines v. United States, 371 U.S. 156, 168 , 83 S.Ct. 239, 245 , 9 L.Ed.2d 207 (1962). “While we may not supply a reasoned basis for the agency’s action that the agency itself has not given, SEC v. Chenery Corp., 332 U.S. 194, 196 , 67 S.Ct. 1575, 1577 , 91 L.Ed. 1995 (1947), we will uphold a decision of less than ideal clarity if the agency’s path may reasonably be discerned.” Bowman Transportation. Inc. v. Arkansas-Best Freight System, 419 U.S. 281, 285-86 , 95 S.Ct. 438, 441-42 , 42 L.Ed.2d 447 (1974).

The EPA here did not specifically include a finding in its grant of the PSD permit that construction on the units commenced before September 19, 1978, so that the 1971 NSPS regulations applied. However, the March 22, 1979 letter from EPA Director Vinson was included as one of the attachments to the PSD permit and “Final Application Analysis” that accompanied it. The EPA in that letter stated that it was basing its determination as to the applicable NSPS regulation upon the information set out in Peterson’s January 18, 1979 letter to the EPA and upon other unspecified information in the related district court case. The former information included the list of construction orders and contracts for the Col-strip generating units. As noted above, those contracts supply a sufficient basis for determining whether or not the construction on the units began before September 19, 1978 and therefore a basis for finding the 1971 NSPS regulations applicable herein.

3. Is There A Rational Basis For The EPA’s Approval Of The Colstrip Units’ Pollution Abatement System As BACT?

The Council argues that even if the 1978 NSPS are not per se applicable to Colstrip *1359 Units Nos. 3 and 4, those standards do identify the level of technology currently available for emission reduction, taking into consideration various factors such as costs, non-air quality health and environmental aspects, and energy requirements. Colstrip Units Nos. 3 and 4 will violate the 1978 NSPS for particulate matter. The Council asserts, therefore, that the EPA’s approval of the Units’ pollution abatement system must be arbitrary since the EPA has already determined that a higher level of reduction for particulate matter is presently possible. 27

In support of its position, the Council cites to the EPA’s statements accompanying the publication of the 1978 NSPS regulations. Therein, the EPA stated that the newly promulgated standard for particulate matter was based on the use of:

44 Fed.Reg. 33584 (1979). Colstrip Units Nos. 3 and 4 will utilize a “scrubbing system”. 28 As noted by the EPA:

Id. at 33585.

Section 169(3) of the Act, 42 U.S.C. § 7479 (3), in its definition of the term “best available control technology”, states that a BACT determination is to be made “on a case-by-case basis, taking into account energy, environmental, and economic impacts and other costs .... ” See 40 CFR § 52 .-21(b)(10) (1980). Thus, the Council initially errs in confusing the EPA’s finding that baghouse control systems are the “best adequately demonstrated technological systems” for particulate matter with the conclusion that they are the best available control technology for each and every major emitting facility subject to the PSD program. While baghouse control systems in general may be presumed to constitute BACT for particulate matter, the EPA is nevertheless obligated to consider certain factors that would arise from the use of baghouses in the particular facility immediately involved. Thus, the EPA might decide that for a given facility baghouses are inappropriate for economic or energy or environmental reasons. 29 In the “Final Ap *1360 plication Analysis, prepared by EPA Region VIII staff” [hereinafter “Final Analysis”] and the “BACT Evaluation for S02 and Particulate Controls for Colstrip Units 3 and 4” by PEDCo Environmental, Inc. [hereinafter “PEDCo Report”], both of which accompanied the EPA’s decision to grant the PSD permit herein, various reasons are given as to why the proposed scrubber system for Colstrip Units Nos. 3 and 4 was held to be BACT even though it permits a particular matter emission rate in excess of the 1978 NSPS regulations and is surpassed by known baghouse technology.

It should be remembered that at the time the PSD permit was initially sought by the Consortium (July 1976), there was no question that the Colstrip Units would satisfy the applicable NSPS regulations. The EPA originally denied the permit in September 1977, however, because the units failed to meet the newly applicable Class I sulfur dioxide increments for the redesignated Northern Cheyenne Reservation nearby. Through proposed changes in the pollution abatement system, Colstrip Units Nos. 3 and 4 will now satisfy the Class I criterion and will significantly exceed both the applicable 1971 NSPS and the newer 1978 NSPS regulations for sulfur dioxide. The units will also, on the average, exceed by half the 1971 NSPS for particulate matter (0.05 lbs per million Btu versus 0.10 lbs per million Btu), but will not meet the 1978 NSPS (0.03 lbs per million Btu). 30

One primary concern of the EPA in its evaluation of the Colstrip pollution abatement system was the desirability of maintaining the high level of reduction of sulfur dioxide emissions that was the most positive feature of that system. Such sensitivity was rational since the Class I sulfur dioxide standard was the only applicable emission standard that Colstrip Units Nos. 3 and 4 had initially failed to meet. Working from that constraint, the EPA then considered the feasibility of improving the reduction of particulate matter, taking into account energy, economic and environmental impact.

The Council on this issue has not argued that there exists a pollution abatement system that is able to accomplish the reduction of both the sulfur dioxide emissions to the requisite Class I levels and the particulate matter emission to the 1978 NSPS. 31 Rather, the Council urges the use of a baghouse system either in the place of or as enhancement to the wet scrubber system currently proposed for the Colstrip Units Nos. 3 and 4. (For a description of the proposed emission control system, see the Appendix to this opinion.) The possible improvement of the Colstrip pollution abatement system by various means was considered by the EPA in its Final Analysis and in the PEDCo Report. Both found the alternatives to be infeasible and held the currently proposed system to be BACT. Our duty here requires “enough steeping in technical matters” to determine whether the EPA “has exercised a reasoned discretion” in its decision herein. Portland Cement Ass’n v. Ruckelshaus, 486 F.2d 375, 402 (D.C.Cir. 1973), cert. denied, 417 U.S. 921 , 94 S.Ct. 2628 , 41 L.Ed.2d 226 (1974) quoting Greater Boston TV v. FCC (I), 444 F.2d 841 , 850, (D.C.Cir.1970), cert. denied 403 U.S. 923 , 91 5. Ct. 2229, 29 L.Ed.2d 701 (1971). After reviewing the highly technical material here, we find the data and conclusions reached in the Final Analysis and PEDCo Report to be persuasive.

Evaluations of various alternative abatement strategies for Colstrip Units Nos. 3 and 4 were made from two perspectives. *1361 First, dry phase particulate precollection systems, which include baghouses, mechanical collectors and electrostatic precipita-tors, 32 were considered. These approaches, if employed, would be utilized prior to the passage of the exhaust emissions into the wet scrubber system. Second, consideration was given to upgrading the efficiency of the wet phase particulate collection portion of the currently proposed system.

The primary problem with any dry phase particulate precollection system is that the wet scrubbers would be deprived of the necessary alkalinity provided by the collected fly ash from the burnt coal that is necessary for the removal of sulfur dioxide. See PEDCo Report at 20. It was estimated that 65 percent of the alkali provided by the collected fly ash would be utilized for sulfur dioxide removal that would account for about 30 percent of the total sulfur dioxide removal. To compensate for the absence of fly ash alkali, hydrated lime additives would have to be increased approximately 160 percent. That increase would have several adverse impacts on: 1) capital costs (greater storage and preparation requirements); 2) annual costs (higher variable charges because of fluctuations in consumption, higher fixed charges because of reagent consumption); and 3) secondary environmental impacts (greater quantities of wastes produced and water required). Also, the residual alkali of the fly ash, which is collected but not used in sulfur dioxide removal, stabilizes and hardens the scrubbing wastes over a period of time to produce a clay-like soil material more suitable for final disposal. Deprivation of fly ash and alkali may therefore adversely affect waste disposal. The EPA concluded that the presence of the natural alkalinity in the particulate matter was a necessary element in the design of the appropriate emission control system at Colstrip Units Nos. 3 and 4. The Council does not dispute that conclusion.

In addition, the Council does not dispute, and we find no evidence or data in the materials that are contrary to, the PEDCo Report’s further findings that: 1) the high and unpredictable fly ash resistivity of the “Rosebud” coal to be utilized at Colstrip produces an adverse effect on the continual high collection efficiencies required of an electrostatic precipitator system thus making that control approach inappropriate; and 2) the installation of mechanical collectors ■ would not appreciably improve the overall particulate collection efficiency of the proposed emission control system at Colstrip Units Nos. 3 and 4.

The Council’s primary objection concerns the EPA’s failure to order employment of a baghouse system. We find the EPA’s reason for their decision to be persuasive. As noted above, the use of a bag-house system alone will not meet the Class I sulfur dioxide requirements. Yet the employment of those devices in conjunction with and before the wet scrubber modules would deprive the scrubber system of the necessary fly ash-alkali for sulfur dioxide removal. It also must be noted that this would involve a great deal of additional expense for a very small increase in particulate removal. Given the case-by-case approach appropriate here, we cannot find the EPA’s conclusion that such a technology would not be BACT from an overall perspective to be irrational.

The Council argues that a dry phase particulate precollection system could be used in conjunction with a method to reinject the collected fly ash into the scrubbing system in order to maintain the integrity of the latter’s removal capabilities. However, the PEDCo Report did consider that alternative and dismissed it for several convincing reasons: 1) that strategy has not yet been *1362 developed for a low sulfur subbituminous coal application; 2) there is concern that the reinjected fly ash has a tendency to coagulate, which decreases the surface area of the collected fly ash and results in a decrease in the amount of alkalinity for sulfur dioxide removal; and 3) the only power plant employing a fly ash reinjection system (the Milton R. Young station) had performance problems due in part to the chemistry of the fly ash alkali affecting the reinjection process.

The Council still argues that the Milton R. Young abatement system is viable for the Colstrip Units and is BACT. However, that system utilizes electrostatic precipata-tors to remove the fly ash. As noted above, the fly ash resistivity of the “Rosebud” coal used at the Colstrip Units would prevent the utilization of an electrostatic system here. Thus, the precollection strategy used at the Milton R. Young Station is not directly applicable. Moreover, there is no evidence in the materials presented to this court that current technology exists that would permit the employment of baghouses in such a reinjection system. Since bag-houses would collect the fly ash in fabric filters, the extraction of sufficient ash for purposes of reinjection may be not be possible.

Improvements in the wet phase particulate collection were examined by the EPA in terms of increasing the efficiencies of the venturi scrubber or the mist eliminator. Both approaches were rejected because of: 1) their much higher energy demands (e. g. the additional energy consumption for the former was estimated to be $6.4 million per year); 2) their potential adverse effects on the integrity and operation of the other internal components in the scrubber modules; and 3) the uncertainty as to whether either alternative would prove effective in the proposed abatement system at Colstrip Units Nos. 3 and 4. The Council has not challenged the EPA’s conclusions as to possible wet phase particulate collection improvements.

The EPA’s approval of the Colstrip Units’ pollution abatement system must therefore be upheld.

4. Did EPA err by refusing to validate its air dispersion models with data collected at the Hay Coulee site?

The Council last argues that EPA’s refusal to validate the air dispersion model was arbitrary and capricious. Validation is a process in which real world ambient readings are compared to theoretical model projections in order to test the predictive accuracy of the model. There is no question that validation by monitoring is a valuable tool in the permitting process. 33

The value of monitoring, however, is not at issue in this case. The Council has a heavy burden here, that of demonstrating to this court that EPA action was arbitrary and capricious. If the Council demonstrates that EPA ignored reliable data that so undermined EPA model projections of compliance with PSD increments that reliance on the model was irrational, that *1363 burden would be met. The Council’s argument does not meet that burden. 34

The Council’s argument can be summarized as follows: Available data from a nearby site monitored by the Battelle Memorial Institute 35 differs significantly from what the EPA’s model projects that level should be under existing conditions. Indeed, on one particular day, the observations were 238% greater than what the model predicted. Given that great a disparity, and given that the model projects that emissions with the additions of Units 3 and 4 will come close to exceeding the permissible Class I PSD increment of sulfur dioxide (S02) concentration, EPA refusal to validate the model with Battelle’s data was arbitrary and renders invalid the approval by EPA of the Colstrip permit. See Petitioner’s Opening Brief at 38.

Initially, it must be noted that the EPA employed two models. One was the agency’s Valley Model, used to predict S02 concentrations on the elevated terrain of the Northern Cheyenne Reservation. The other model, the PTMPT model (point-multi-point), was used to predict S02 concentrations for low lying areas of the reservation. The highest concentrations for the high terrain covered by the Valley model occur during meteorological conditions in which the “plume” (pollutant stream emitted from the source) does not disperse. These are labeled “stable” conditions. High concentrations of S02 at the lower levels covered by the PTMPT model, including the Hay Coulee site, occur during “limited mixing conditions,” which are midway between unlimited mixing conditions (i. e. turbulent and unstable) and stable conditions. The two models thus cover different areas of the Class I reservation, and predict highest S02 concentrations at different times and under different conditions.

This leads to the important observation that the Hay Coulee data is thus completely irrelevant to any challenge to the Valley model. Yet is is only on the high terrain covered by the Valley model and under stable conditions that the PSD S02 increment is approached by model predictions.

On the lower terrain covered by the PTMPT model, projected S02 emissions are significantly lower. On February 22, 1978, the date selected by the Council to demonstrate the degree of error in the PTMPT model, that model would predict a concentration of 9.3 ugm/m³ (micrograms per cubic meter) of S02 with the addition of Units 3 and 4 at a point near the edge of the reservation, 26 km downwind from the Col-strip site. , At the Hay Coulee site, the predicted concentration level (with only Units 1 and 2) was 18.5 ugm/m³; the actual reading reported by Battelle was 44 ugm/m³. It is the difference between these two figures that leads to the 238% disparity figure cited by the Council. EPA points out that, accepting a 238% margin of error for the sake of argument, the predicted concentration on the reservation would be (9.3 X 2.4) or approximately 22.3 ugm/m³. This is still less than the maximum allowable PSD increment of 25 ugm/m³.

It is thus difficult to see how the EPA’s action could be labeled irrational. The Council has not demonstrated that the EPA irrationally rejected data that would lead to a conclusion that the EPA model was so deficient that the relevant PSD limits would be exceeded.

The Council responds that, the extrapolation of just one day is insufficient. Yet that day was chosen by the Council, and therefore presumably demonstrates a worst case instance. If there were more extreme examples of model variance from Hay Coulee readings, the Council presumably would *1364 have proffered them. To choose the single worst day and claim that a response based on that day is inadequate is a bit absurd.

Similarly, the Council claims that EPA erroneously used a point 26 kilometers from Colstrip for its extrapolation when the border is only 22 kilometers. Again, this court has to presume that were extrapolated figures at the 22 kilometer point appropriate (in terms of, e. g., terrain height) and in excess of the 25 ugm/m³ figure, the Council would have developed its own more pessimistic extrapolation and presented it to this court.

The Council additionally claims that the argument that the Battelle data is not relevant to the Valley model cannot be raised here as it was not relied on by the EPA. This contention is contradicted by the record. The EPA indicated in Appendix II of the Conditional Permit that validation requires “that the measured data be collected under the same conditions for which the model is predicting,” and that the Battelle data was not appropriate. Comment & Reply 6.A.8. As for the PTMPT extrapolation analysis, this was also found in Appendix II which, as the Council notes, provides the EPA’s reasoning. See Comment & Reply 6.A.I.

III. CONCLUSION

For the reasons stated above, we affirm the decision of the EPA.

APPENDIX

The Proposed Emission Control System At Colstrip Units Nos. 3 and 4

The following Figure 2-1 in the PEDCo Report contains a flow diagram for the scrubbing system to be employed at Colstrip Units Nos. 3 and 4. Joint Appendix at 352.

Exhaust from the burnt fossil-fuel, which contains particulate matter (including fly ash) and sulfur dioxide, is sent through a venturi scrubber that removes a portion of both pollutants. The exhaust then travels in a spray zone for additional sulfur dioxide *1365 removal. The treated gas leaves the spray zone through a wash tray and mist eliminator. This activity occurs in one of eight parallel scrubber modules. From there the “cleaned” gas is reheated, boosted in pressure by an induced draft fan, and discharged through a stack into the atmosphere.

The liquid that is sent through the scrubbers is a water-alkali mixture. Alkaline solutions have been found to be most effective in removing sulfur dioxide. The slurry that is formed by the coalescence of the water mixture and the particulate and sulfur dioxide matter is collected from the scrubber modules into a common recycle tank. A portion of the slurry is withdrawn and put through an oxidation process to minimize the formation of calcium sulfite deposits in the scrubber. To that oxidized slurry, alkali (hydrated lime, calcium and magnesium hydroxide) are added and the regenerated slurry is sent back to the recyle tank. Another portion of the slurry is withdrawn to a common sludge settling pond from which the settled sludge is removed and disposed of at specific sites.

. The Council is a membership organization incorporated in the state of Montana. Many of its members live in or nearby Colstrip, Montana. The air quality in that area will be affected by the EPA’s decision in this case.

. Notice of the EPA’s decision herein was reported in 44 Fed.Reg. 55651 (1979).

. Section 165(a)(1) of the Clean Air Act (sometimes referred to herein as the “Act”), 42 U.S.C. § 7475 (a)(1) (Supp. II 1978).

The Act was recodified from 42 U.S.C. § 1857 et seq. to 42 U.S.C. § 7401 et seq. by the Clean Air Act Amendments of 1977. See 91 Stat. 685 (1977). Because the relevant period here includes times both before and after the recodifi-cation, citations will be both to sections of the Clean Air Act and to the United States Code. Unless otherwise indicated, all citations to the latter are taken from the 1978 Supplement II of the United States Code.

. See 40 CFR § 52.21 (1980).

. The Consortium consists of several electric utilities headed by the Montana Power Company and includes Pacific Power & Light Company, Portland General Electric Company, Puget Sound Power & Light Company, and Washington Water Power Company.

. This court’s jurisdiction over this case extends only to a review of a final action of the Administrator of the EPA “which is locally or regionally applicable.” Section 307(b)(1) of the Act, 42 U.S.C. § 7607 (b)(1). Final actions of the Administrator having nationwide scope or more than purely local or regional impact may only be reviewed in the United States Court of Appeals for the District of Columbia. Id. See Natural Resources Defense Council, Inc. v. EPA, 512 F.2d 1351, 1356-57 (D.C.Cir.1975).

. The use of models for the purpose of analyzing air quality data is provided for in Section 165(e) of the Act, 42 U.S.C. § 7475 (e), and in the EPA’s PSD regulations at 40 CFR § 52 .-21(m) (1980).

. The construction of the first two units fell within the grandfather provision of the 1974 PSD regulations and hence no PSD permit was required for them.

. See Montana Power Co., 608 F.2d at 339 & note 6 (9th Cir. 1979). Those orders and contracts were listed in a letter dated January 18, 1979 from John Peterson, an attorney for Montana Power Company, to an EPA official and submitted as part of the record below of the PSD permit proceeding. In that letter, it is stated that the total amount of the outstanding contracts for the construction of the generating units as of May 25, 1975 was over $123 million.

. This court upheld the validity of that redes-ignation in Nance v. EPA, 645 F.2d 701 (9th Cir. 1981).

. Petitions challenging the denial of the PSD permit were filed with this court in Puget Sound Power & Light Co. v. EPA, Nos. 78-2824 and 78-3234. The parties to that litigation reached a settlement and the case was dismissed in June 1980.

. The Consortium asked the EPA for a determination of which NSPS regulation was applicable to the Colstrip Units. See 40 CFR § 60.5 (a) (1979). In a letter sent on March 22, 1979 (but erroneously dated March 22, 1978), the Director of EPA’s Enforcement Division replied that the 1971 NSPS regulations were applicable to the units as opposed to the 1978 NSPS regulations.

. See 44 Fed. Reg. 55651 (1979).

. A “baghouse control system” utilizes fabric filters placed within an enclosed area through which the exhaust emissions from the generating units are passed and where particulate matter in those emissions is trapped. See National Lime Association v. EPA, 627 F.2d 416, 424 , (D.C. Cir. 1980).

. See the parallel regulatory definition of BACT, 40 CFR § 52.21 (b)(10) (1980).

. The Colstrip Units are expected to emit the following three pollutants (which are subject to regulation under the Act) at the rates listed below:

See the Consortium’s “Application For A Determination By The Administrator Under 40 CFR § 52.21 That Colstrip Units 3 And 4 Will Meet The Regulations For Prevention Of Significant Deterioration” dated July 1, 1976, at 5; EPA’s “Conditional [PSD] Permit To Commence Construction And Operate” granted to Colstrip Units Nos. 3 and 4 on September 11, 1979 at 4; and EPA’s “Final Application Analysis” of the Consortium’s amended application, at 3 — 4.

Under the 1971 NSPS regulations, no owner or operator of a fossil-fuel-fired steam generator, like the ones to be utilized at the Colstrip Units, is permitted to cause the discharge of three pollutants in excess of the following:

Under the 1978 NSPS regulations, more stringent standards are established:

(The 1978 NSPS regulations for generating units revised the criterion by imposing not only an emission limitation on the pollutants but also a “percentage reduction” requirement. See EPA “New Statutory Sources Performance Standards; Electric Utility Steam Generating Units”, 44 Fed.Reg. 33580, 33581-82 (1979).)

. The PSD permit was not obtained until September 11, 1979. The Council argues that significant proposed changes in the construction plans for Colstrip Units Nos. 3 and 4 were made by the Consortium after July 22, 1976 when the Major Facility Siting Act certificate was issued by the Board. Those changes, which were initiated by the Consortium to meet the EPA’s PSD requirements, consist primarily of an increase in the height of the smokestacks and the utilization of hydrated dolomitic lime in the pollution control process.

Given our disposition of the applicable definition of “commenced” under Section 111 below, we need not and do not decide the issue of whether the Consortium’s proposed changes in the Colstrip Units’ construction plans mandate an amended certificate under applicable Montana state law.

. While Colstrip Units Nos. 3 and 4 will apparently violate the 1978 NSPS regulations for nitrogen oxides, the PSD permit granted by the EPA subjects the units to the best available retrofit technological requirements when they are promulgated by the EPA. See 44 Fed. Reg. 55652 (1979). The Council has not challenged the potential nitrogen oxide NSPS violation issue in its petition here.

. In 1974, the EPA modified its definition of “commenced”. As noted by the EPA:

39 Fed.Reg. 9308 (1974).

. By no stretch of the imagination can the preconstruction permit requirement incorporated in the Section 169(2)(A) definition of “commenced” be viewed as merely articulating a plain meaning for that word.

. While the version of Section 111 found in the initial proposed Senate bill contained a requirement for preconstruction review and certification, S.Rep. No. 91-1196, 91st Cong.2d Sess. 17 and 92 (1970), reprinted in Legislative History 1970 Amendments at 417 and 492, the Administration opposed that provision as being unnecessarily burdensome. See “Administration’s Letter to Conference Committee Recommending Certain Provisions” dated November 17, 1970, reprinted in Legislative History 1970 Amendments at 216. The Conference Committee, in submitting the version of Section 111 that was eventually adopted by Congress, indicated its agreement with the Administration’s views. See Conference Report, H.R.Rep. No. 91-1783, 91st Cong.2d Sess. 45 (1970), reprinted in Legislative History 1970 Amendments at 196. It therefore appears unlikely that Congress had any intent to incorporate a precon-struction permit requirement into the definition of “commenced” at the time Section 111 was originally enacted.

. Representative Rogers was one of the sponsors of both the substantive 1977 amendments to the Act and the technical amendments.

. The EPA in fact issued a proposed regulation on May 31, 1979, that amends the NSPS definition of “commenced” in 40 CFR § 60.2 (1980) to make it consistent with the definition in Section 169(2). 44 Fed.Reg. 31596-97 (1979). That revised definition will be effective prospectively only and will not, as currently worded, affect the present case. The Council correctly concedes that a challenge to the EPA’s decision to apply the revised definition prospectively may only be brought in the U. S. Court of Appeals for the District of Columbia under Section 307(b)(1) of the Act, 42 U.S.C. § 7607 (b)(1). However, given our holdings stated above, that may be the only remaining unresolved controversy on this issue.

. In Montana Power Co., 608 F.2d at 357-58, this court held that construction on Colstrip Units Nos. 3 and 4 had not “commenced” for purposes of Section 169(2)(A) and the PSD program before August 7, 1977. However, the issue of when construction commenced on the affected facilities therein (the steam generating units) for purposes of the NSPS program was not considered at all in that case.

. See note 22, supra.

. See “Milestone Dates” on page 11 of “Contract for Construction of Boilers for Colstrip Units Nos. 3 and 4”, reprinted in Addendum D at D-20 in Addenda Brief filed herein by the Consortium.

. The fact that Colstrip Units Nos. 3 and 4 satisfy the applicable NSPS regulations does not automatically mean that they also satisfy the BACT requirement of the PSD program. EPA’s original PSD regulations did define “best available control technology” in such a way that meeting the NSPS also meant that the BACT requirement was fulfilled as well. See 40 CFR § 52.01 (f) (1975). However, that definition was amended and now provides for the maximum degree of reduction for each pollutant considering the matter on a case-by-case basis and taking into account various enumerated factors. 43 Fed.Reg. 26404 (1978).

. A basic “scrubber system” utilizes water to “catch” the pollutant particles in the emissions and to make them easier to control. The system operates by accelerating the velocity of the exhaust gas through narrow venturi-shaped passages where it is then brought into contact at great force with a spray of water. Increasing the degree of contact between particle and water will produce an increase in the reduction of pollutants in the exhaust. The particles thus dampened coalesce to form a “slurry” that can be collected by comparatively simple water-gas separation devices. The separated gas is then released into the atmosphere. The slurry byproduct is deposited in ponds where the collected particles settle out from the scrubbing water. The “clean” water is then reused. See National Lime Ass’n, 627 F.2d at 425 .

. The EPA is promulgating an NSPS is also obligated to take into account “the costs of achieving such emission reduction, and any no-nair quality health and environmental impact and energy requirements.” Section 111(a) of the Act, 42 U.S.C. § 7411 (a). That examination, however, can only be a generalized consideration of the technology and its effects on those factors based on data from many varied sources. Under the PSD program, the BACT determination is supposed to be source-specific. What may be applicable to most plants may not be appropriate for a particular facility. The EPA under its regulations is obligated to assure that an individualized inquiry is made on the matter.

. There has been no suggestion that Colstrip Units Nos. 3 and 4 will violáte the Class I increments or ceiling for particulate matter.

. The EPA, in promulgating the 1978 NSPS regulations, did state that the development of dry sulfur dioxide control technology had progressed rapidly. 44 Fed.Reg. 33582 (1979). However, at that time, the EPA cited to only three full scale operations that were scheduled to start up in the 1981-82 period. Those facilities had design efficiencies of only 50 to 85 percent sulfur dioxide removal, id., greatly below the reduction level required in the present situation. It was noted further, however, that bids were being sought for control systems with a 70 to 90 percent reduction range. The Colstrip Units here will have a reduction efficiency of 95.7 percent. See PEDCo Report at 3.

. In an electrostatic precipitator system, “dust particles are charged [by electrodes] and pass through an electrical field [collector plates] of the opposite charge, thus causing the dust to be precipitated out of the exhaust gas.. . . ” Portland Cement Ass’n, 486 F.2d at 390 . Two preconditions must be met before such a system can be employed: 1) the suspended particle must be able to accept an electrical charge, and 2) the particle must then pass through an electric field of sufficient strength to assure removal of the particulate matter from the gas stream at the desired efficiency. Cf., National Lime Ass’n, supra. 627 F.2d at 424 .

. Indeed, the D.C. Circuit recently held that Congress intended monitoring to be a requirement of the preconstruction permitting process under § 165(e)(2), stating:

Alabama Power, 636 F.2d at 372 , 13 E.R.C. at 2019. The Council does not argue that Alabama Power requires monitoring in this case. The relevant holding in Alabama Power was that EPA PSD regulations that required monitoring pursuant to § 165(e)(2) only to determine whether an applicable NAAQS would be exceeded were inadequate. Rather, monitoring was required to determine whether any of the PSD increments would be exceeded. See 13 ERC at 2018-2019. The EPA proposed regulations issued pursuant to Alabama Power indicate that the additional monitoring requirements will be applied only prospectively. 44 Fed.Reg. 51928 (1979).

. In evaluating this argument, we assume that the Hay Coulee data presented by the Council is scientifically reliable, a matter hotly contested by the parties in great technical detail in the briefs. Because the Council fails to establish irrationality by the EPA, even given this highly favorable assumption, we need not reach this highly technical reliability question.

. The data relied on is monitoring data from a station set up by Battelle Memorial Institute’s Pacific Northwest Laboratories at the Hay Coulee site approximately 12 kilometers southeast of the Colstrip site for purposes apparently unrelated to the permit application.