Bowers v. Pollution Control Hearings Bd. , 13 P.3d 1076 ( 2000 )

13 P.3d 1076 (2000) 103 Wash.App. 587

Gregory H. BOWERS, Appellant,
v.
POLLUTION CONTROL HEARINGS BOARD; Southwest Air Pollution Control Authority; Transalta Centralia Generation LLC, Respondents.

No. 44838-2-I.

Court of Appeals of Washington, Division 1.

December 4, 2000.

Michael W. Gendler, Bricklin & Gendler, Seattle, for Appellant.

Andrew M. Kenefick for Southwest Air Pollution Control Authority.

Jean Marie Wilkinson, Asst. Atty. General, Olympia, Richard Lynn Griffith & Kenneth Odza & Deborah Elvins, John Phillips, Seattle, for Respondents.

COLEMAN, J.

Appellant Gregory Bowers seeks to overturn an order issued by the Southwest Air Pollution Control Authority (SWAPCA) and affirmed by the Pollution Control Hearings Board (PCHB). The order directed the Centralia Power Plant to reduce its emissions of four pollutants, including sulfur dioxide and nitrogen oxide. SWAPCA issued the order pursuant to the Washington Clean Air Act, which requires all existing sources of air pollutants in the state to control their emissions by using "reasonably available control technology" (RACT).

Bowers raised a host of challenges before the PCHB. The crux of his argument was that SWAPCA did not conduct its RACT review properly, resulting in emission levels that were too high and control technologies that did not adequately protect human health. The PCHB rejected Bowers' claims and affirmed SWAPCA's decision.

We hold that Bowers did not meet his burden in challenging the PCHB's findings of fact and conclusions of law. The PCHB did not erroneously interpret the Washington Clean Air Act as applied to the requirements for setting RACT. Substantial evidence supported its findings that addressed the scientific underpinnings of the emissions limits and the cost effectiveness and impact of the various control technologies. And the PCHB's decision to affirm SWAPCA's order was not arbitrary or capricious. We therefore affirm.

BACKGROUND

Centralia Plant

The Centralia Power Plant (Plant) is a coal-fired electric generating plant located near Centralia, Washington. The Plant's two units, or boilers, generate electricity capable of meeting the energy needs of a city the size of Seattle. The first unit became operational in 1971 and the second in 1972. Until May 2000, the Plant was owned by eight utilities, including the respondent PacifiCorp (collectively "Owners").^[1] The Plant employs approximately 160 workers.

The Plant's primary source of coal is an adjacent mine, which employs approximately 510 workers. Supplemental coal is supplied by rail from out of state.

The Plant is the largest single point source of sulfur dioxide (SO2) and nitrogen oxide (NOx) emissions in Washington. In 1997, the Plant emitted 63,773 tons per year (tpy) of SO 2, and 16,098 tpy of NOx.

Reasonably Available Control Technology

The Washington Clean Air Act requires all existing sources of air pollutants in the state, including the Plant, to control their emissions by using "reasonably available control technology" (RACT).^[2] RCW 70.94.154; see also WAC 173-400-040; SWAPCA 400-040. RACT is defined as:

the lowest emission limit that a particular source or source category is capable of meeting by the application of control technology that is reasonably available considering technological and economic feasibility. RACT is determined on a case-by-case basis for an individual source or source category taking into account [1] the impact of the source upon air quality, [2] the availability of additional controls, [3] the emission reduction to be achieved by additional controls, [4] the impact of additional controls on air quality, and [5] the capital and operating costs of additional controls.

RCW 70.94.030(19); see also SWAPCA 400-030(76); WAC 173-400-030(66). The Washington Clean Air Act directs the Department of Ecology and local air pollution control authorities (e.g., SWAPCA^[3]) to undertake "RACT reviews" to set RACT limits for specific sources or source categories. RCW 70.94.154(4)-(5). This case involves a source-specific limit.

The first step in the RACT review process is to determine which air pollutants emitted by the particular source are "of concern." RCW 70.94.154(4)-(5).^[4] Once the agency has identified the pollutants of concern, it evaluates different control technologies for each pollutant against the five RACT factors set forth above, namely:

(1) air quality impacts;

(2) availability of additional controls;

(3) emission reduction to be achieved by additional controls;

(4) impact of additional controls on air quality; and

(5) capital and operating costs of additional controls.

RCW 70.94.030(19). The agency is also required to take into account RACT determinations and guidance made by the Environmental Protection Agency (EPA) and other states and local authorities for similar sources. RCW 70.94.154(5). The agency must also consider "other relevant factors." RCW 70.94.154(5).

Once the agency has made its proposed RACT determination, it must provide notice and an opportunity for public comment. RCW 70.94.030(19); see also SWAPCA 400-171(1)(c); WAC 173-400-171(1)(c). The agency reviews the comments and then makes its decision and issues the final RACT order, including a schedule for implementing the RACT emission limits.

SWAPCA's RACT Order

On February 26, 1998, SWAPCA issued RACT order 97-2057R1, which set emissions limits at the Plant for four pollutants: sulfur dioxide (SO2), nitrogen oxide (NOx), particulate matter (PM), and carbon monoxide (CO). The order was accompanied by a technical support document that set forth the regulatory, scientific, and technical bases for SWAPCA's decisions. The primary focus of the RACT order was the reduction of SO2 emissions to 10,000 tpy annually, but the order also set a two-tiered RACT limit to reduce NOx levels to approximately 12,900 tpy and established limits for PM and CO.^[5]

The starting point of the SO2 and NOx emissions limits was a negotiated agreement, announced on December 3, 1996, between the owners and federal, state, and local regulators. The parties of this voluntary "collaborative decisionmaking" (CDM) process^[6] met for approximately one year, with the objective of reducing the Plant's emissions and protecting visibility and the environment, without jeopardizing the Plant's continued existence. The negotiated agreement received bipartisan political support. Implementation of the agreement was predicated upon a tax relief package, which was enacted by the Washington Legislature in 1997.

With the CDM proposal as its starting point, SWAPCA reviewed the proposed SO2 emissions limit and the control technologies needed to achieve that limit. Because the Owners agreed to accept an annual limit of 10,000 tpy for SO2, the RACT determination considered only control technologies that would achieve at least this result. SWAPCA evaluated the technologies and concluded that the 10,000 tpy cap, based on limestone forced oxidation scrubber technology, would meet or exceed RACT. SWAPCA followed a similar procedure in setting RACT for NOx.

Procedural History

On April 3, 1998, Gregory Bowers appealed the RACT order to the Pollution Control Hearings Board (PCHB). The principal focus of his appeal was whether the Plant's emissions presented a public health risk under current emissions levels or when complying with the RACT order. The PCHB dismissed a number of issues on summary judgment. Following a hearing^[7] on the remaining issues, the PCHB issued a decision upholding the RACT order.

Bowers appealed the PCHB decision to King County Superior Court. SWAPCA and the Owners, later joined by Bowers, petitioned for direct review with this court. The PCHB certified its decision for direct appellate review pursuant to RCW 34.05.518(3), and we granted review.

DISCUSSION

A. PROCEDURAL ISSUES^[8]

1. Standard of Review

The Administrative Procedures Act (APA) governs judicial review of the PCHB's decision. RCW 43.21B.180. Bowers bears the burden of showing that the PCHB's action was invalid. RCW 34.05.570(1)(a). Because Bowers asserts the invalidity of an agency order resulting from an adjudicative proceeding, RCW 34.05.570(3) applies. This subsection sets forth nine standards for granting relief from the PCHB's decision, of which three apply here:

(d) The agency has erroneously interpreted or applied the law;

(e) The order is not supported by evidence that is substantial when viewed in light of the whole record before the court...;

. . .

(i) The order is arbitrary or capricious. RCW 34.05.570(3).^[9] We apply these standards of review based upon the record before the PCHB. Nguyen v. Dep't of Soc. & Health Servs., 99 Wash.App. 96, 101, 994 P.2d 216 (1999), review granted, 141 Wash.2d 1001, 10 P.3d 404 (2000).

With respect to issues of law under RCW 34.05.570(3)(d), we review the PCHB's legal conclusions de novo. Substantial weight is accorded the agency's interpretation of the law where the agency has specialized expertise in dealing with such issues, but we are not bound by the agency's interpretation of a statute. City of Redmond v. Cent. Puget Sound Growth Mgmt. Hearings Bd., 136 Wash.2d 38, 46, 959 P.2d 1091 (1998).

In reviewing challenged findings under RCW 34.05.570(3)(e), substantial evidence is a sufficient quantity of evidence to persuade a fair-minded person of the truth or correctness of the order. Redmond, 136 Wash.2d at 46, 959 P.2d 1091 (internal quotation and citation omitted). We neither weigh credibility nor substitute our judgment for that of the agency. Nguyen, 99 Wash.App. at 101, 994 P.2d 216 (citing U.S. West Communications, Inc. v. Wash. Utils. & Transp. Comm'n, 134 Wash.2d 48, 61-62, 949 P.2d 1321 (1997)).

A decision is arbitrary or capricious under RCW 34.05.570(3)(i) if it is a "`willful and unreasoning action, taken without regard to or consideration of the facts and circumstances surrounding the action.'" Redmond, 136 Wash.2d at 46-47, 959 P.2d 1091 (quoting Kendall v. Douglas, Grant, Lincoln & Okanogan Counties Pub. Hosp. Dist. No. 6, 118 Wash.2d 1, 14, 820 P.2d 497 (1991)). If there is room for two opinions, a decision is not arbitrary or capricious if it is made honestly and upon due consideration, even though we think a different conclusion might have been reached. Buechel v. Dep't of Ecology, 125 Wash.2d 196, 202, 884 P.2d 910 (1994).

Bowers claims that our review should be one of "substantial inquiry," and directs our attention to Lead Industries Ass'n v. EPA, 647 F.2d 1130, 1145 (D.C.Cir.1980). We find the Lead Industries decision to be of limited applicability in our interpretation of our state APA, as it sets forth the standard of review for federal regulations. Lead Indus., 647 F.2d at 1145 (scope of judicial review delineated by 42 U.S.C. § 7607(d)). Even so, our reading of Lead Industries does not persuade us that the "substantial inquiry" standard is any more exacting than our substantial evidence test. While the Lead Industries court states that its review of the record must be "searching and careful," it also is quick to observe that its review is not designed to second guess the expertise of the agency entrusted with the responsibility of making scientific judgments. Lead Indus., 647 F.2d at 1146. The court further explains:

[A]fter our careful study of the record, we must take a step back from the agency's decision. We must look at the decision not as the chemist, biologist or statistician that we are qualified neither by training nor experience to be, but as a reviewing court exercising our narrowly defined duty of holding agencies to certain minimal standards of rationality. "Although [our] inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one."

Lead Indus., 647 F.2d at 1146 (quoting Ethyl Corp. v. EPA, 541 F.2d 1, 36-37 (D.C.Cir. 1976)). There is thus little difference between the substantial inquiry standard as articulated by Lead Industries and ours. We, too, review the entire record carefully, giving due deference to the agency's expertise, to determine whether the PCHB's decision is rational and supported by the evidence.

2. Burden of Proof

Bowers contends for the first time on appeal that the PCHB erroneously assigned him the burden of proof under WAC 371-08-485(2). RCW 34.05.554 precludes appellate review of issues not raised below. See also King County v. Wash. State Boundary Review Bd., 122 Wash.2d 648, 670, 860 P.2d 1024 (1993) ("[T]here must be more than simply a hint or a slight reference to the issue in the record."). In addition, Bowers assumed the burden of proof without objecting, even though the respondents addressed the issue in their trial brief. His belated challenge is therefore without merit.

Nonetheless, the PCHB did not err in assigning the burden to Bowers. WAC 371-08-485(2) provides: "The issuing agency shall have the initial burden of proof in cases involving penalties or regulatory orders. In other cases, the appealing party shall have the initial burden of proof." Bowers argues that this case involves a "regulatory order," so SWAPCA has the burden. He is mistaken.

WAC 371-08-485(2) typically applies when an agency issues an enforcement order alleging certain violations or seeks penalties for violations, and the party to whom the order was issued appeals.^[10] The instant appeal, however, is not strictly an enforcement order. It is more akin to appeals of emission or effluent limits found in permits, where the burden of proof is placed on the party contending that the limit does not satisfy statutory or regulatory requirements.^[11]

Bowers' appeal of SWAPCA's decision before the PCHB can also be analogized to his appeal of the PCHB's decision currently before this court. As stated above, the APA allocates the burden of proof in a judicial review action to the party asserting the invalidity of the agency's decision. RCW 34.05.570(1)(a). The same can be said for his appeal below; that is, the burden appropriately rested on Bowers as the party seeking to invalidate SWAPCA's action. The PCHB did not err in concluding that Bowers was required to prove his case by a preponderance of the evidence.

B. AIR QUALITY ISSUES

As discussed above, RCW 70.94.030(19) sets forth five factors that must be considered in any RACT review, the first of which is the impact of the source upon air quality. SO2 and NOx, two of the pollutants emitted by the Plant, can convert in the atmosphere to sulfates and nitrates, both of which are components of fine particulate matter, or PM2.5.^[12] One of Bowers' principal concerns is how the Plant's SO2 and NOx emissions, once they convert to PM2.5, contribute to public health risks under current emissions levels and after the Plant complies with the RACT order. The PCHB determined that the Plant's current emissions have a minor impact on air quality, and that the Plant's emissions after installation of control technology will have an even lesser impact on air quality and human health.

Bowers raises three issues that relate to this first RACT factor. First, he challenges the reliability of the CALPUFF model, which calculated ambient PM2.5 concentrations. Second, he claims that the PCHB erroneously relied on the National Ambient Air Quality Standards in drawing its conclusions about the health effects of ambient PM2.5. Third, he disputes both the methodology and results of the risk assessment undertaken to estimate mortality risks associated with the Plant. Each of these issues will be considered in turn.

1. CALPUFF

After the CDM agreement was announced, the Owners commissioned a risk assessment of the health impacts from the Plant. The Owners hired Kirk Winges, an atmospheric scientist, to model the air quality for the risk assessment. The purpose of the modeling study was to estimate ambient concentrations of air pollutants produced by the Plant within a 150-mile radius.

Of the air dispersion models available, Winges chose the CALPUFF model. CALPUFF is a mathematical "puff" model that tracks puffs of pollutants emitted from a source (i.e., the Plant). Depending on the wind speed, the source is modeled as emitting a puff of pollutants every few seconds to only a few puffs per minute. The model uses information on weather conditions and topography to simulate how the emissions are transported across the region. This type of model is particularly useful for single sources and sources where long range transport (i.e., over 50 kilometers), combined with complex wind fields and topography, need to be modeled. In contrast, other available models either had limited accuracy for distances beyond 50 kilometers (e.g., they assumed that airflow was steady and terrain was flat) or were more computationally intensive or demanding when applied to only a single source. Because of these features, Winges and the air dispersion modeling experts concluded that CALPUFF was the most appropriate model available, notwithstanding certain limitations.^[13]

One of CALPUFF's limitations is its ability to evaluate the rate of conversion of gaseous SO2 to sulfates (SO4). As emissions are transported away from the Plant, atmospheric conditions such as relative humidity, temperature, and wind speed change the originally emitted SO2 to SO4. Because of the Pacific Northwest climate, there can be considerable variability in the conversion rate. For example, SO2 converts to SO4 faster in wet conditions, so the conversion rate could be as high as one hundred percent.

CALPUFF's aqueous phase conversion rate remained constant at three percent per hour. The respondents' experts acknowledged that the conversion rate was a limitation of the CALPUFF model, but explained that changing the conversion rate was not as simple as inputting a different number, and would require modifying the computer code, which would create other concerns about the model's integrity.

Before the PCHB and again in this appeal, Bowers argues that the use of a constant rate (i.e., three percent) is more than a mere "limitation" of the model because it results in an underestimation of ambient PM2.5 levels. Bowers asserts that because it is often cloudy but not rainy in Seattle, he calculates PM2.5 concentrations as eight to sixteen times higher than that estimated by the Owners.^[14] Contrary to Bower's claim, however, the record supports the PCHB's finding that a higher aqueous phase conversion rate would not necessarily lead to the conclusion that there would be an increase in ambient PM2.5 conversions attributable to the Plant.

Clinton Bowman, an atmospheric physicist, testified that a number of additional factors contribute to the conversion rate, including the amount of SO2 emitted, the amount of time a puff spends at different relative humidities, and whether the conversion occurs at day or night. He also explained that although there are often foggy or cloudy conditions in the Pacific Northwest, there is also a strong likelihood that sulfates become captured in cloud vapor which then "precipitate out" in the form of raindrops. Both the SO2 gas and SO4 aerosol can be removed from the ambient air by dry or wet deposition. Dry deposition is caused by contact with the surface; wet deposition is caused by precipitation. The amount of conversion also depends on the travel time for each puff, so the longer the travel time, the greater the likelihood that SO4 will be "rained out" before reaching Puget Sound. Finally, Robert Bachman, an air management specialist, testified that even though moist conditions favor more rapid conversion of SO2 to SO4, the wind conditions in winter are generally more turbulent, which helps with air dispersion and lower SO4 rates.

Bowers did not refute the expert testimony about the "rain out" effect. Nor did he present an alternative or more accurate model supporting a higher estimate of PM2.5 concentrations than that predicted by CALPUFF. The only evidence of potential underestimation was the written assessment of John Vimont, a meteorologist with the National Park Service, who stated that because of the frequent occurrence of clouds and fog in the Pacific Northwest, the analysis likely underestimates the ambient concentration of sulfates. That said, however, Vimont concluded that the "air quality modeling was conducted properly" and the CALPUFF model is "an appropriate model."^[15]

Substantial evidence thus supports the PCHB's finding that even if one of the CALPUFF model's inherent limitations is its conversion rate, and the Plant's contribution to ambient PM2.5 levels may in fact be underestimated, the CALPUFF model provided the most accurate information available and was therefore the most appropriate model to use as the basis for the risk assessment.

2. NAAQS

The PCHB also considered the current National Ambient Air Quality Standards (NAAQS). Section 109 of the federal Clean Air Act directs the EPA to write primary NAAQS that adequately protect public health, with an "adequate margin of safety," without reference to feasibility and cost. 42 U.S.C. § 7409(b)(1); Lead Indus. Ass'n v. EPA, 647 F.2d 1130, 1148 (D.C.Cir.1980). In setting NAAQS, the EPA is required to protect not only average healthy individuals, but also "sensitive citizens," such as children or people with asthma, emphysema, or other conditions making them "particularly vulnerability to air pollution." See Am. Lung Ass'n v. EPA, 134 F.3d 388, 389 (D.C.Cir.1998).

The NAAQS for PM2.5 was first established by the EPA in 1997.^[16] EPA set the average PM2.5 standard at 15 μg/m3 annually and 65 μg/m3 for a 24-hour period. At issue here is PCHB's decision to compare the Plant's PM2.5 contribution with the NAAQS annual PM2.5 standard of 15 μg/m3.

The CALPUFF model estimated that the Plant's current annual average PM2.5 contribution is 0.15 μg/m3, or 1.7 percent of the average annual ambient concentrations in various Seattle locations. After implementation of controls, the Plant's PM2.5 contribution would be 0.04 μg/m3. The PCHB compared the Plant's contribution of PM2.5 to the NAAQS, finding that after controls, the Plant's contribution to ambient air concentrations would be less than 0.2 percent of the national PM2.5 standard. Based on this information, the PCHB concluded that the Plant's current and projected contributions to ambient PM2.5 levels are negligible.

Bowers asserts that the PCHB exclusively and "blindly" relied on the federal standard, which he alleges has since been struck down by American Trucking Ass'ns v. EPA, 175 F.3d 1027 (D.C.Cir.1999), cert. granted, Browner v. American Trucking Ass'ns, ___ U.S. ___, 120 S. Ct. 2003, 146 L. Ed. 2d 954 (2000), as not protective of human health. American Trucking was issued after the PCHB issued its decision below. See Aviation W. Corp. v. Dep't of Labor & Indus., 138 Wash.2d 413, 427, 980 P.2d 701 (1999).^[17] Even if we were to consider American Trucking, our reading of the decision is contrary to that of Bowers'.

The thrust of the opinion is that the EPA's construction of the Clean Air Act in promulgating the NAAQS was an unconstitutional delegation of legislative power. In rejecting the EPA's arguments in defense of its ozone and PM standards, the court did not take issue with the basic criteria the EPA used in evaluating the public health concerns associated with the pollutants regulated by the NAAQS—i.e., the nature and severity of health effects, the type of health information available, and the degree of uncertainty involved. Am. Trucking, 175 F.3d at 1034-35 (citing National Ambient Air Quality Standards for Ozone, 62 Fed.Reg. 38,856, 38,883/2 (July 18, 1997)). Rather, the court found that the EPA failed to articulate an "intelligible principle" to constrain the agency's discretion. Am. Trucking, 175 F.3d at 1034. The court's holding is inconsistent with Bowers' interpretation that the EPA set the NAAQS at an unsafe limit.^[18]

Moreover, American Trucking's constitutional defect does not diminish the role of the NAAQS as appropriate guidance in evaluating air quality impacts as part of the RACT review process. See RCW 70.94.154(5) (directing SWAPCA to consider EPA guidance for similar sources in setting RACT). To this end, the PCHB deemed the NAAQS an "important consideration" in evaluating public health risks associated with PM2.5. The NAAQS were not, as Bowers contends, exclusively and "blindly" relied on by the PCHB. Other equally important considerations included the results from the CALPUFF model, which we discussed previously, and the results of the risk assessment, which we discuss next. We conclude that the PCHB's level of reliance on the PM2.5 NAAQS was reasonable at the time it was made. See Aviation W., 138 Wash.2d at 427, 980 P.2d 701.

3. Risk Assessment

Another consideration in evaluating air quality impacts was the risk assessment conducted by Dr. Jonathan Samet, an epidemiologist with Johns Hopkins University.^[19] Samet prepared the risk assessment in response to the CDM group's request that he evaluate how much pollution in the area comes from the Plant and how much the proposed control technology (scrubbers and low NOx burners) will benefit human health. Accordingly, the project had three goals: (1) to estimate the Plant's contribution to ambient concentrations of SO2 and fine particles (i.e., nitrates and sulfates) in areas where emissions may affect air quality; (2) to estimate the impact on ambient concentrations of SO2 and fine particles of reducing the Plant's emissions of SO2 and NOx with scrubbers; and (3) to assess the health impacts, if any, that would remain after reducing the Plant's emissions.

Samet's risk assessment relied on CALPUFF modeling results to determine the extent of the Plant's contribution to PM population exposure. The assessment also relied on risk coefficients from the same epidemiological studies used by the EPA when it adopted its 15 μg/m3 annual average PM2.5 NAAQS in 1997. By combining the exposure estimate with the risk coefficient, Samet estimated the likelihood of three adverse health effects that could result from increased exposure to airborne emissions from the Plant: (1) hospitalizations of the elderly; (2) emergency room and outpatient visits for asthma sufferers; and (3) premature mortality. The mortality estimates are at issue in this case.

The risk assessment made two primary conservative assumptions. First, the dose-response relationships^[20] were based on a linear model without a threshold; that is, any increase in pollution exposure, no matter how small, was assumed to convey some risk. Second, the risk model assumed sulfates and nitrates formed from the Plant's emissions contribute to mortality risk, even though the particular components of PM2.5 that cause increased mortality risk are unknown. Because of these conservative assumptions and uncertainties, Samet cautioned that the estimates should be used only for comparative risk purposes—i.e., "for gauging the potential extent of the public health consequences of plant emissions and of the potential benefits of implementing controls"—and not as predictions of actual deaths.^[21]

The risk assessment concluded that the Plant's emissions do not pose a significant health risk either with or without emissions reductions. In reaching this conclusion, the assessment used both short-term (daily) and long-term (annual) exposure studies. The short-term studies showed the correlation between health effects and increases in PM pollution on the same day or the day before, whereas the long-term studies measured annual exposure. Based on these studies, two different estimates of premature mortality risk were calculated. Using the short-term exposure studies, the assessment estimated a daily risk of five premature mortalities before RACT controls and one premature mortality after controls are in place. For long-term exposure, the assessment estimated an annual risk of 50 premature mortalities before implementation of RACT controls and 13 premature mortalities after controls.

The PCHB considered both the short-term and long-term estimates. The PCHB found that the Plant is a minor contributor to population exposures and projected health risks without scrubbers and will be even less of a contributor once scrubbers are added.

Bowers claims that the PCHB should not have considered the short-term estimates, and he raises two issues to this effect. First, he contends that Samet underestimated the short-term PM2.5 estimates by incorrectly using a dose-response coefficient from the PM10 short-term studies to estimate PM2.5 health risks. Second, he contends that long-term estimates are better predictors of health impacts. We conclude that the record contains ample support for the PCHB's decision to consider both the short- and long-term estimates.

a. Use of PM10 Short-term Studies

Bowers contends that Samet incorrectly used a dose-response coefficient from the PM10 short-term studies to estimate PM2.5 health risks.^[22] This process of "mixing metrics," argues Bowers, erroneously resulted in an underestimation of the short-term mortality estimates, which the PCHB then should have ignored when making its conclusions about the health risks associated with the Plant's emissions.

Bowers points to EPA documents, including the 1997 NAAQS Final Rule,^[23] that appear to rely on short-term PM2.5 studies. Samet, in contrast, testified that "[t]here have been virtually no studies of PM2.5" because there is "little PM2.5 data available." He continued, "Presumably, the variation in PM10 that is relevant to health that drives these mortality associations is thought to come from variation of PM2.5 component. So... these [PM10] studies have been widely interpreted as indicative of the effects of fine particles" and are considered to be "surrogates."

On a cursory level, Bowers' argument seems to have some merit. The respondents do not refute the existence of PM2.5 studies in their brief, except to point to Samet's testimony. And they tacitly acknowledge the issue by revising their short-term mortality estimates based on the EPA's PM2.5 risk coefficient.^[24] The question then remains whether Samet's decision to use the PM10 coefficient to determine PM2.5 health risk renders the use of the short-term estimates any less valid. We conclude that it does not.

First, the risk assessment was peer reviewed by a number of experts.^[25] Some of the reviewers questioned Samet's use of PM10 studies to estimate risk from PM2.5 emissions and asked Samet to explain his method for doing so.^[26] But these same reviewers generally approved of the assessment's method and conclusions.^[27] Second, use of the short-term PM2.5 studies relied on by the EPA in its NAAQS determination would have resulted in nearly the same results as the estimate in the risk assessment itself.^[28] Finally, to the extent the evidence addressing the appropriateness of using PM2.5 as opposed to PM10 studies may be inconsistent, it is not the role of this court to second guess the way in which the PCHB weighed the evidence or resolved the dispute. See Aviation W. Corp. v. Dep't of Labor & Indus., 138 Wash.2d 413, 429-30, 980 P.2d 701 (1999) ("`[S]ubstitution of our judgment for that of the administrative agency in factual matters is not authorized by the APA and... we will not try facts de novo on review.'") (quoting Franklin County Sheriff's Office v. Sellers, 97 Wash.2d 317, 325, 646 P.2d 113 (1982)). Based on the record before the PCHB, we conclude that it was permissible for the PCHB to consider the short-term estimates based on Samet's application of the PM10 dose-response coefficients.

b. Use of Long-term Mortality Estimates

Bowers also claims that the PCHB should have refrained from considering the short-term estimates^[29] because the studies on which the estimates are based do not completely describe the impact of PM on mortality. He acknowledges that the EPA placed greater emphasis on the short-term exposure studies in setting the annual PM2.5 NAAQS, but claims that the PCHB "confused EPA's use of short-term studies [in setting NAAQS] with EPA findings as to what is the best indicator of total health impacts." Bowers asserts that while the short-term studies provide "greater certainty" that impacts are occurring, the long-term studies better predict total mortality.

Even if his statement were true,^[30] Bowers is unable to substantiate his claim that the long-term studies provide the better estimate of mortality risk. He asks this court to take judicial notice of an EPA report that was released after the PCHB issued its order.^[31] Judicial review of agency action is confined to the agency record, with certain exceptions inapplicable here. RCW 34.05.558, .562.^[32] Because none of the provisions allows us to consider the evidence submitted by Bowers on appeal, we will limit our review to the evidence presented to the PCHB.^[33]

Samet explained that "neither type of estimate is considered to have greater validity," due to the uncertainty regarding the "degree of mortality displacement" in the short-term studies.^[34] In other words, the extent to which daily exposure affects long-term studies and vice-versa is unclear. We also observe that the peer reviewers did not question Samet's use of both types of studies. And as acknowledged by Bowers, the EPA relied more heavily on the short-term studies in setting the PM2.5 NAAQS.

We conclude that Bowers did not meet his burden of showing that the short-term studies were any less reliable, or provided any less insight into the impact of air pollution on mortality, than the long-term studies. The PCHB's decision to consider both sets of estimates presented by the risk assessment was supported by substantial evidence and was not arbitrary or capricious.

C. THE RACT BALANCING PROCESS

1. Balancing RACT Factors

The statute governing RACT requires that in setting RACT limits, SWAPCA "consider" and "tak[e] into account" the five enumerated factors as well as "any other relevant factors." RCW 70.94.030(19),.154(5). The PCHB interpreted this statutory provision as requiring "use its best professional judgment, considering all the information available to it" and then "balance the need for cleaner air (including minimizing adverse health impacts) against the capital and operating costs of additional technologies for controlling emissions." The PCHB also concluded it was appropriate to consider potential economic and social impacts as "other relevant factors" in setting RACT, including the impact on the regional economy from a proposed RACT limit that resulted in closure of the Plant and/or mine. In disagreeing with the PCHB's conclusion, Bowers insists that the RACT statute precludes SWAPCA from balancing human health impacts against economic considerations.^[35] In particular, he argues that other costs, such as potential job loss and regional economic impacts, are not among the enumerated factors that should have been considered by SWAPCA in its RACT determination or the PCHB when it reviewed the soundness of the RACT order.^[36]

This is the first case interpreting RACT requirements under the Washington Clean Air Act. RACT is defined as "the lowest emission limit that a particular source or source category is capable of meeting by the application of control technology that is reasonably available considering technological and economic feasibility." RCW 70.94.030(19) (emphasis added). "Reasonably available" is not defined in the statute. However, in a case interpreting the "reasonableness" of "technological and economic feasibility," even where specific factors were not enumerated, our Supreme Court chose not to second-guess the agency's balancing process:

The agency's discretionary enforcement power is further circumscribed by the requirement of "reasonableness." Under this limitation, a control system must be both economically and technologically feasible. SWAPCA may not require a system which would impose an unreasonable financial burden on the applicant because of excessive initial outlay or annual operating costs. Nor may SWAPCA impose restrictions which are unreasonable in other respects. For example, a system moderately more efficient in controlling emissions might be "unreasonable" if it substantially impeded plant production.... Many other factors might be relevant in determining the reasonableness of available systems. These are matters for the expertise of SWAPCA and the PCHB with ultimate review in the courts pursuant to the administrative procedures act.

Weyerhaeuser v. Southwest Air Pollution Control Auth., 91 Wash.2d 77, 82, 586 P.2d 1163 (1978). Accord Puget Soundkeeper Alliance v. Dep't of Ecology, 9 P.3d 892 at 896-897 (Wash.Ct.App., 2000) (approving of the Weyerhaeuser analysis). Similarly, a Minnesota administrative law judge observed: "At bottom, though phrased in legal terminology, a RACT determination is a policy judgment which must be made by the MPCA Board, weighing its commitment to the environment with the cost to ratepayers of achieving a particular SO2 emissions level." In re Application of N. States Power Co., No. 2-2200-7921-2, 1993 WL 543990, at 27 (Minn. Pollution Control Agency Dec. 30, 1993). Here, too, we accord substantial weight to SWAPCA and the PCHB's interpretation of the law. See City of Redmond v. Cent. Puget Sound Growth Mgmt. Hearings Bd., 136 Wash.2d 38, 46, 959 P.2d 1091 (1998).^[37]

The RACT statute does not state that one factor is more important than any other or state how the factors are to be weighed. It only states that each of the factors is to be taken into account. One of these factors is the impact of the source upon air quality, both before and after controls. Another is the capital and operating costs of additional control technologies. Still another is the requirement that the agency consider "any other relevant factors," which here included the socio-economic costs associated with potential Plant or mine closure. The agency's evaluation of these factors is further circumscribed by the statutory directive that RACT be the "lowest emission limit ... that is reasonably available considering technological and economic feasibility." RCW 70.94.030(19). In this case, the PCHB determined that it was appropriate to evaluate the evidence qualitatively, and where appropriate quantitatively, weighing the environmental and economic impacts associated with implementing the various control technologies. The PCHB's interpretation of the Washington Clean Air Act is neither unreasonable nor untenable, and we decline Bowers' invitation to find otherwise. We therefore conclude that the PCHB did not misapply the law in sanctioning the balancing process used by SWAPCA in carrying out its RACT review responsibilities. RCW 34.05.570(3)(d).

2. Economic Viability of Lewis County

Bowers next claims that even if the Washington Clean Air Act permits a balancing of health impacts with economic considerations, the PCHB failed to evaluate the "magnitude" of job losses because it could not adequately assess how many workers would be laid off as a result of each proposed RACT measure.^[38] We disagree.

The PCHB found that implementing the RACT measures proposed by Bowers, such as converting the Plant to natural gas, would "have severe potential negative employment and regional economic impacts" including job losses and a loss of tax revenues to Lewis County. The respondents presented evidence projecting direct and indirect job losses^[39] of 1,800 to 2,300 workers should the Plant and mine be forced to close. The respondents also provided evidence that there would be a considerable loss of tax revenue should the Plant close, as the Plant is the largest single taxpayer in Lewis County.

Bowers does not identify any evidence in the record supporting his argument, choosing instead to reiterate the PCHB's finding that mine and Plant workers would likely find new employment, either in or out of Lewis County. Thus, says Bowers, "[s]ince laid off `miners' are in-demand heavy equipment operators, PCHB could not (and did not) define what, if any, impact closure of the mine would have to the state's economy." Nothing in his statement discredits the respondents' evidence of potential job loss or negative economic impacts.

Substantial evidence supports the PCHB's finding that Bowers' proposal would jeopardize the Plant and mine's continued viability and that closure of the Plant and/or mine would severely impact employees and the regional economy.

3. Cost-benefit Analysis

Bowers argued before the PCHB that, as part of the balancing process, SWAPCA was required to conduct a quantitative cost-benefit analysis of the health benefits created by the various control technologies. He presented the testimony of economist Robert Halvorsen, who applied a monetary value to selected mortality risk estimates culled from the risk assessment, and then compared these values to the costs of additional controls. The PCHB rejected Bowers' approach, citing the considerable difficulties that exist when attempting to quantify the health benefits associated with air quality. The PCHB's findings on this issue were neither arbitrary nor capricious, nor were its conclusions based on insufficient evidence.

In evaluating this issue, we are mindful of two different sets of considerations: the uncertainty in estimating mortality and the uncertainty in placing a monetary value on these estimates.

First, a necessary precursor to the quantification of health benefits from RACT is the ability to determine actual mortality risks. Here, Bowers' cost-benefit analysis used the results of Samet's risk assessment. Bowers applied the higher mortality estimates from the long-term exposure studies, as opposed to the lower estimates from the daily studies.^[40]

As discussed previously, the Plant's mortality estimates are too uncertain to be relied on as actual health risks or predictions of actual deaths. See discussion supra note 21. It is also unclear whether the daily or annual risk coefficients more accurately predict mortality risk for PM2.5. See discussion supra Section B.3 (discussing importance of considering both sets of risks). Finally, Bowers' expert on cost-benefit analysis reviewed neither the RACT order nor Samet's risk assessment, nor was he qualified to offer an opinion of the health risks associated with the Plant.^[41]

Second, even if the PCHB could determine the actual mortality risks associated with the Plant, the PCHB would still need to quantify the monetary value of preventing each mortality. The monetary value imputed to premature mortality is known as the "value of a statistical life" (VSL). Halvorsen recommended a VSL of $4.8 million per mortality in 1990 dollars, which he adjusted upward to $6.0 million to account for inflation. Halvorsen's VSL figure was based primarily on labor market studies measuring risk aversion of an individual from accidental death on the job, as opposed to the risk aversion of an elderly individual or someone in poor health. The respondents' expert, Alan Krupnick, explained that he considers studies that adjust VSLs from accidental death data to be inaccurate because they are "based on the wrong commodity, the wrong context, and the wrong population." Krupnick also testified that the EPA has used the $4.8 million figure as well as a lower estimate of $500,000.^[42]

It is evident that the divergent VSL figures would create an equally wide discrepancy in calculating the health benefits due to air quality improvements in PM2.5. The PCHB concluded that it was "not persuaded... that the art or science of calculating health benefits due to air quality improvements in PM2.5 has yet reached the level where we may rely upon the VSL estimates." This decision was neither "willful" nor "unreasoning," nor was it "taken without regard to or consideration of the facts and circumstances surrounding the action." City of Redmond v. Cent. Puget Sound Growth Mgmt. Hearings Bd., 136 Wash.2d 38, 46-47, 959 P.2d 1091 (1998).

Based on the uncertainty in estimating mortality and in the reliability of VSL estimates, it was permissible for the PCHB to conclude that assigning a monetary value to mortality risk is not sound practice at this time.

D. RACT CONCLUSIONS

SWAPCA considered a number of control technologies for reducing SO2 and NOx emissions, eventually identifying limestone forced oxidation scrubbers as the appropriate control technology for reducing SO2 emissions and low NOx burners for reducing NOx emissions. Bowers challenged the appropriateness of both control options before the PCHB. He offered three alternatives to scrubbers: as a long-term alternative, he urged conversion to natural gas; as short-term options, he identified "environmental dispatch" and/or the increased use of low sulfur coal. As an alternative to low NOx burners, Bowers proposed the use of selective catalytic reduction technology. The PCHB rejected all of these alternatives. Bowers appeals the findings and conclusions associated with these issues.

1. Low Sulfur Coal

Bowers argues that the Plant should be required to increase its importation of low sulfur coal to two million tons per year until final RACT controls are implemented. The Plant currently acquires most of its coal from the adjacent mine operated by the Centralia Mine Company. The PCHB says little about the use of low-sulfur coal, except for finding that neither the Plant nor the railroad on which it relies has the equipment necessary to haul such a quantity of coal to Centralia. The respondents presented testimony to this effect:

[E]ven though it may be theoretically possible that the current rail siding could unload up to two million tons of coal a year, this level of coal unloading has not been done....

. . . .

... [O]btaining an adequate supply of additional external coal could take up to six to nine months and/or a large capital expenditure to purchase railcars. Currently, the Centralia plant relies on Burlington Northern Santa Fe (BNSF) railroad to supply coal hauling equipment. In preliminary discussions, BNSF has indicated that it does not have available, at costs even close to the current level, the equipment necessary to haul two million tons of coal per year to Centralia. A large capital expenditure necessary for the Owners to purchase their own equipment to supply this interim fuel source likely would be considered unreasonable.

Interestingly, Bowers directs us to this same testimony for his proposition that the respondents' witness "implies" that the "necessary equipment is indeed available," albeit at some unknown cost. Yet Bowers presented no evidence below supporting his argument that railroads are available or that the costs of importing low sulfur coal are less than cost-prohibitive. Given the testimony in the record, we conclude that substantial evidence supports the PCHB's finding and that it was not arbitrary or capricious.

2. Environmental Dispatch

Bowers also advocated the use of environmental dispatch as an interim RACT measure. Environmental dispatch involves a process of prioritizing available power generating resources (e.g., coal, natural gas, nuclear, solar, wind) based on the environmental impacts from those resources. It assumes a pool of power plants, ranked according to their environmental impacts, such as tons of air pollutants emitted or gallons of water discharged. Those plants with the least environmental impact are selected to generate electricity first. Once all of the lowest environmental impact units are operating at full capacity, then the type of generation with the next lowest environmental impact is selected.

The PCHB properly rejected environmental dispatch as an interim RACT measure. First, Bowers failed to present evidence that environmental dispatch is available as a control technology. Neither Bowers nor the expert testifying on his behalf was aware of any power plant that imposed a technology-based emissions limit based on environmental dispatch. In fact, Bowers' expert testified only that it would be "highly desirable for the Owners to investigate [it]."^[43]

Second, Bowers failed to present evidence that the use of environmental dispatch would be economically feasible. Bowers testified that if environmental dispatch were used, the Plant could operate only when the price of electricity exceeded $50 per megawatt hour (MWh). Plant manager Richard Wooley testified that the Plant is only marginally profitable^[44] without scrubbers when the market cost of electricity is $20.83/MWh. Bowers' proposal thus would require the Plant to close until the price rose to $50/MWh or the Plant converted to natural gas. The record thus supports the PCHB's decision that environmental dispatch is neither a reasonably available nor economically feasible control technology.

3. Natural Gas

The PCHB rejected the natural gas alternative because it was not economically feasible, a decision we find is amply supported by the record. First, although the estimated capital cost of converting to natural gas is less than the cost of installing scrubbers, the total annualized cost of natural gas conversion is significantly higher. The total annualized cost of natural gas is approximately $100 million per year, as compared to approximately $25 million for scrubber technology. Second, an analysis of the "cost effectiveness" of both technologies yields a similarly disparate result. The owners presented testimony that the cost effectiveness for natural gas conversion is estimated at between $954 and $1,134 per ton, whereas the cost effectiveness for full scrubbing is estimated at between $256 and $310 per ton.^[45] These economic forecasts led the Plant manager to testify that the costs of converting to natural gas would likely result in closure of the Plant.

Bowers does not offer any evidence to the contrary, but simply asserts that the PCHB arbitrarily refused to evaluate and place a value on the health benefits that would result from conversion to natural gas. As previously discussed, the PCHB appropriately considered the qualitative benefits of reduced mortality and morbidity risk, but concluded that there was too much uncertainty associated with placing a monetary value on the health benefits associated with emissions reductions.

There was an adequate basis in the record for the PCHB to reject natural gas conversion as a reasonably available control technology.^[46]

4. Selective Catalytic Reduction

Bowers argues that reductions in NOx emissions should be based on selective catalytic reduction (SCR) technology, not low NOx burners. In rejecting his argument below, the PCHB's findings again are supported by substantial evidence. The respondents provided evidence that the estimated cost effectiveness per ton for low NOx burners is between $113 and $233, whereas for SCR, it is between $1,186 and $1,837. The total annualized cost of low NOx burners is approximately $1.6 million; for SCR, it is approximately $24 million. This evidence alone is sufficient to uphold the PCHB's finding that SCR is not economically feasible.^[47]

In challenging the PCHB's decision, Bowers maintains, as he has throughout this appeal, that the PCHB "arbitrarily ignored the health costs" of NOx when it approved of low NOx burners. But he presents no evidence in the record to support his assertion.^[48] Our review of the evidence leads us to conclude that the PCHB's decision cannot be characterized as one that ignored human health concerns merely because the PCHB declined to assign a precise monetary value to human life. The PCHB considered air quality issues, cost, and guidance by other states and the EPA, all of which it is required to do under the statute governing RACT determinations. Based on this evidence, the PCHB reasonably concluded that low NOx burners, not SCR, represent RACT.

E. SUMMARY JUDGMENT ISSUES

Bowers appeals two issues that were decided on summary judgment. In reviewing a summary judgment order, we engage in the same inquiry as the PCHB. Clay v. Portik, 84 Wash.App. 553, 557, 929 P.2d 1132 (1997) (applying standard of review directly to record before administrative agency). Summary judgment is appropriate if the pleadings, depositions, answers to interrogatories, admissions on file, and affidavits show that there is no genuine issue as to material fact and that the moving party is entitled to judgment as a matter of law. CR 56(c).^[49]

1. Requirement that Every Air Contaminant be Addressed

The PCHB held that SWAPCA was not required to "analyze, measure and address the health impacts of every air contaminant emitted by the Centralia Plant" when setting RACT limits for SO2, NOx, CO, and PM emissions. Bowers maintains that SWAPCA should have considered the effect of the control alternatives on "all air quality," not just that of SO2, NOx, CO, and PM. Had the agency done so, says Bowers, it would have learned that using the natural gas alternative to reduce SO2 would have had the incidental benefits of eliminating other hazardous air pollutants as well. We reject Bowers' argument for two reasons.

First, the statute governing RACT does not require SWAPCA to address every air contaminant. SWAPCA is directed to set RACT limits, "where practicable," for all contaminants deemed to be "of concern." RCW 70.94.154(5). In this case, SWAPCA explained how it considered which contaminants might have been "of concern" and which were "practicable to address." In determining which pollutants were "of concern," SWAPCA considered the quantities of pollutants emitted, how the quantities compared against certain screening criteria, relevant ambient air quality standards, past violations of emissions limits, potential visibility impacts, and studies of health concerns associated with the pollutants. In considering the "practicability" of setting emission limits, SWAPCA evaluated the availability of additional controls, how other sources have controlled the pollutants, and impending federal standards for the source category. SWAPCA also considered its internal resources and the potential for delay in issuing the final RACT order if a RACT review were required for every pollutant emitted. Based on these considerations, SWAPCA decided not to establish RACT limits for all but the four pollutants (SO2, NOx, PM, CO) because other pollutants were not contaminants of concern, were not practicable to address, or both.^[50]

Second, and more important, Bowers failed to create a genuine issue of material fact by not offering any evidence of other air contaminants that were "of concern" or "practicable" to address. In response to SWAPCA's interrogatory asking him to identify air contaminants and impacts not addressed by SWAPCA's RACT order, Bowers stated, "At this time appellant is unable to identify any specific `related air contaminant' for which it would be better for RACT to now address." The PCHB properly granted summary judgment on this issue.

2. Visibility Issues

With respect to visibility issues, Bowers assigns error both to the PCHB's order granting partial summary judgment and to the PCHB's findings and conclusion.^[51] He appears to argue that the PCHB erred in refusing to consider the Plant's impacts on regional air clarity, but the analytical framework of his argument is not entirely clear. He seems to claim that visibility must be taken into account, as it is a component of air quality. But the record is clear that SWAPCA did consider visibility impacts in setting RACT limits.^[52] At summary judgment, the PCHB considered this and other evidence submitted by SWAPCA. The PCHB also considered a single report (Nothstein/Brown) submitted by Bowers to support his contention that SWAPCA did not appropriately take visibility issues into account. Our review of the report supports the PCHB's finding that the report only analyzed the improved visibility that would be achieved by reducing the Plant's emission to 10,000 tpy of SO2 (i.e., RACT), but did not provide evidence that visibility would improve below the RACT order's emission levels. Absent such evidence, the PCHB correctly held that Bowers failed to create a genuine issue of material fact.

F. CONCLUSION

The PCHB did not abuse its authority or act irrationally in approving SWAPCA's RACT order. Nor did the PCHB base its determination on insufficient evidence. We thus affirm the PCHB's modified findings of fact, conclusions of law, and order, and its order granting partial summary judgment.

NOTES

[1] On May 4, 2000, the Plant and mine were sold to TransAlta Centralia Generation LLC, which was substituted for the Owners as the respondent in this appeal.

[2] In comparison, new and modified sources are required to comply with the more stringent "best available control technology" (BACT) or "lowest achievable emission rate" (LAER) standards. RCW 70.94.030(6), (13); see also WAC 173-400-030(10), (38); WAC 173-400-112, -113.

[3] The Southwest Air Pollution Control Authority has since been renamed the Southwest Clean Air Agency. For ease of reference, we will continue to refer to the agency as SWAPCA.

[4] RCW 70.94.154(5) provides: "In establishing... RACT requirements, ecology and local authorities shall address, where practicable, all air contaminants deemed to be of concern for that source or source category." Robert Elliott, SWAPCA's executive director, testified that because the statute does not explain how the agency is required to determine the pollutants for which RACT is required, SWAPCA uses its discretion and expertise to determine which emissions are "of concern" and then evaluates whether it is "practicable" to establish RACT for those contaminants of concern. We address Bowers' challenge to SWAPCA's interpretation of RCW 70.94.154(5) infra at Section E.1.

[5] The RACT order's PM and CO limits are not challenged in this appeal.

[6] The CDM process involved members of the U.S. Environmental Protection Agency, the USDA Forest Service, the National Park Service, the Washington Department of Ecology, SWAPCA, and the Puget Sound Air Pollution Control Agency (since renamed the Puget Sound Clean Air Agency).

[7] The PCHB considered both prefiled written direct testimony and live cross examination and questioned the witnesses themselves as needed.

[8] The respondents assert that this court may dismiss Bowers' appeal for violations of three RAP provisions: RAP 10.4(c) (requiring that appellant type or submit material portions of challenged statutes or regulations); RAP 10.3(a)(5) (providing that issues raised but not argued on appeal are waived); and RAP 10.3(h) (requiring separate, concise statements of each alleged error). The respondents' RAP 10.4(c) claim is without merit. As for the latter two RAP violations, Bowers concedes that he failed to argue certain errors, but there are other errors he failed to argue as well. We thus confine our analysis to the issues that are clearly set forth and argued in his brief.

[9] Bowers also asserts that the following three additional standards apply: the order is outside the statutory authority or jurisdiction of the agency conferred by any provision of law; the agency has not decided all issues requiring resolution by the agency; and the order is inconsistent with a rule of the agency unless the agency explains the inconsistency by stating facts and reasons to demonstrate a rational basis for inconsistency. RCW 34.05.570(b), (f), (h). Absent any explanation or argument by Bowers as to how these standards apply to the issues raised, however, we limit our review to whether the PCHB's actions were arbitrary and capricious, not supported by substantial evidence, or based on an inaccurate interpretation of the law.

[10] See, e.g., AAA Affordable Asbestos Abatement v. Spokane County Air Pollution Control Auth., PCHB Nos. 99-193 & 00-044, 2000 WL 1289615, at *2 (Aug. 25, 2000) (SCAPCA bears burden of proof that violations of asbestos control standards occurred and penalties are reasonable); Fengel v. Dep't of Ecology, PCHB No. 99-033, 1999 WL 1095004, at *3 (Oct. 7, 1999) (Ecology bears burden of proof to establish that violation(s) occurred and regulatory order was lawful); Vanderhouwen v. Dep't of Ecology, PCHB Nos. 94-108, 94-146, & 94-231, 1997 WL 233297, at *6-7 (Mar. 25, 1997) (Ecology established that appellant violated regulatory order by failing to cease and desist from withdrawing groundwater without permit).

[11] See, e.g., Port Townsend Paper Corp. v. Dep't of Ecology, PCHB No. 98-77, 1999 WL 1611276, at *3 (Aug. 17, 1999) (burden of proof on company challenging opacity limit in permit conditions); Marine Envtl. Consortium v. Dep't of Ecology, PCHB Nos. 96-257 et al., 1998 WL 934931, at *21 (Nov. 30, 1998) (burden of proof on environmental group challenging effluent limit in NPDES permit based on "all known, available and reasonable methods of treatment" (AKART)); Save Lake Sammamish v. Dep't of Ecology, PCHB No. 95-141, 1996 WL 379222, at *3 (June 27, 1996) (burden of proof on environmental group challenging AKART determination); Univ. Mech. Contractors v. Puget Sound Air Pollution Control Agency, PCHB No. 87-56, 1987 WL 55714, at *14 (appellants have burden of proof in challenge to emission limit based on best available control technology).

[12] There is a difference between PM2.5, discussed here, and the PM emissions that SWAPCA regulated in its RACT order. Particulate matter is classified based on whether it is primary (i.e., emitted directly from a stack) or secondary (i.e., formed in the atmosphere). The Plant emits primary PM, which was one of the pollutants regulated by SWAPCA. Particulate matter that forms through atmospheric conversion of SO2 and NOx to PM2.5, which is at issue here, is secondary PM. In addition, as discussed in a later issue, there is more than one size of particulate matter. PM2.5 refers to "fine" particulate matter of less than 2.5 microns (μm) in diameter. PM10 refers to "coarse" particulate matter of less than 10 μm in diameter. PM2.5 is a subset of PM10.

[13] In addition to Winges, the respondents presented the testimony of Clinton Bowman, an atmospheric physicist with the Washington Department of Ecology with expertise in air dispersion modeling, and Robert Bachman, a physical scientist and air resource management specialist with the USDA Forest Service, who has expertise in modeling and monitoring visibility conditions.

[14] In many instances such as this, it is Bowers alone who supplies the evidence and analysis challenging SWAPCA's RACT order. The PCHB is entitled to take into consideration the fact that the evidence supporting Bowers' criticism of the CALPUFF model is limited to his own testimony and exhibits, and that his expertise in the area of air dispersion modeling is minimal. See Davidson v. Kitsap County, 86 Wash.App. 673, 680, 937 P.2d 1309 (1997).

[15] On appeal, Bowers does not direct this court to any evidence in the record, except to point out that he cited "verifiable data" and "several credible studies" in his testimony below. In reviewing his testimony, we find it difficult to discern the relevance and/or applicability of the studies he cites, due in part to the manner in which he presents the evidence. To the extent that Bowers' evidence may support an inconsistent conclusion, it "does not prevent an administrative agency's finding from being supported by substantial evidence.'" Aviation W. Corp. v. Dep't of Labor & Indus., 138 Wash.2d 413, 429, 980 P.2d 701 (1999) (quoting Am. Textile Mfrs. Inst., Inc. v. Donovan, 452 U.S. 490, 101 S. Ct. 2478, 69 L. Ed. 2d 185 (1981)).

[16] The 1997 rule also set NAAQS for ozone and amended the NAAQS for PM10.

[17] In Aviation W., the court was asked to consider a federal district court case that "vacated" portions of an EPA report relied on by the Department of Labor and Industries in regulating smoking in private workplaces. In rejecting appellant's request, the court explained:

The standard under our state's APA is whether the choice to rely upon the EPA report was rational at the time it was made. "The court must scrutinize the record to determine if the result was reached through a process of reason, not whether the result was itself reasonable in the judgment of the court."

Aviation W., 138 Wash.2d at 427, 980 P.2d 701 (quoting Neah Bay Chamber of Commerce v. Dept. of Fisheries, 119 Wash.2d 464, 474, 832 P.2d 1310 (1992)).

[18] If anything, the decision implies that the agency violated the delegation doctrine by articulating NAAQS that were too stringent. In fact, in its petition for certiorari, one of the issues raised by the EPA was that the decision took away the EPA's authority to implement more stringent standards. See DOJ Appeals D.C. Circuit's NAAQS Ruling, Florida Environmental Compliance Update (Sept. 1999). We also note that the EPA came under considerable criticism after it issued the NAAQS in 1997, primarily because the raw scientific data upon which the standards are based was not made available to the public for review. Congressional members, for example, criticized the EPA Administrator for promulgating stringent standards that would "result in a significant economic impact" without a demonstrated scientific basis that the standards would have "significant beneficial health effects." See A. Tina Batra & Marcia C. Sugrue, Note, EPA's Not So Final Rules: Congress' Attack on EPA's New Ozone and Particulate Matter Rules, 4 Envt'l Law. 611, 615 (Feb. 1998) (footnotes omitted).

[19] Samet is Chair of the Department of Epidemiology at Johns Hopkins University. He holds an M.S. degree from the Harvard School of Public Health. He has conducted numerous studies on the health effects of air pollution and spent four years conducting research on the relationship between PM and mortality risk. His research was performed in coordination with the EPA's 1997 adoption of NAAQS for PM2.5. The only testimony offered by Bowers to contradict Samet's was his own. Bowers' experience appears to be limited to the human health risk assessment he performed in connection with his challenge to the RACT order. While it is not the role of this court to weigh credibility, the PCHB is entitled to take into account the background and relevant experience of witnesses when determining the appropriate weight to place on the evidence. See Davidson v. Kitsap County, 86 Wash.App. 673, 680, 937 P.2d 1309 (1997).

[20] Dose-response relationship characterizes how risk varies with dose (or exposure).

[21] In his reply brief. Bowers maintains that the respondents misrepresent Samet's testimony that projected deaths were not estimates because Samet "now confirms that estimated deaths are projections of fatal casualties among the public resulting from Plant operation." To the contrary, it is Bowers who misrepresents Samet's testimony. In his initial draft of the risk assessment, Samet incorrectly calculated the risk for King and Snohomish counties as 2,000 premature mortalities, which he then revised to 680 mortalities. There is no evidence in the record to support Bowers' claim that once Samet corrected his estimate, he "never again asserted that his risk assessment results do not represent actual deaths." In fact, in the same declaration in which Samet acknowledges the error (submitted at the time the PCHB was considering SWAPCA's motion for summary judgment), he reiterates his cautionary note about not likening mortality estimates to projections of actual deaths. Samet made subsequent statements to the same effect in his prefiled testimony ("The estimates should not be considered as actual numbers of deaths, but as offering a risk management tool[.]"), and in testimony before the PCHB. Other witnesses, including Naydene Maykut, a senior air quality scientist for the Puget Sound Air Pollution Control Agency, and Tim Gould, a SWAPCA air quality engineer, testified to the same. The EPA also concluded that its risk assessment for the PM2.5 NAAQS included significant uncertainty and should not be viewed as demonstrated health impacts, but as reasonable estimates of risk. See National Ambient Air Quality Standards for Particulate Matter, 62 Fed. Reg. 38,652, 38,656 (July 18, 1997) (codified at 40 C.F.R. pt. 50). There was thus a sufficient basis for the PCHB to find that the numeric estimates should be used for comparative risk purposes only.

[22] Recall that PM2.5 is "fine" particulate matter less than 2.5 microns in diameter, and is a subset of PM10, which is "coarse" particulate matter less than 10 microns in diameter. The risk assessment relied on 31 short-term studies to derive the dose-response coefficient used to calculate mortality risk; 29 of the studies measured PM10 and 2 measured PM2.5.

[23] Ironically, this is the same NAAQS that Bowers challenges as having been overturned by American Trucking Ass'ns v. EPA, 175 F.3d 1027 (D.C.Cir.1999), cert. granted, Browner v. American Trucking Ass'ns, ___ U.S. ___, 120 S. Ct. 2003, 146 L. Ed. 2d 954 (2000) See discussion, supra Section B.2.

[24] According to the respondents, the revised short-term mortality risk estimates would increase from five to seven before controls and from one to two after controls.

[25] Peer reviews were performed by Dr. Harriett Ammann, toxicologist, Washington Department of Health; Dr. C. Arden Pope, III, professor, Brigham Young University; Dr. Suresh Moolgavkar, member, Fred Hutchinson Cancer Research Center; Joseph Scire, air modeler, Earth Tech, Inc.; Dr. Bart Ostro, research scientist, Office of Environmental Health Hazard Assessment—Air Toxicology/Epidemiology, CalEPA; and John Vimont, meteorologist, National Park Service.

[26] See, e.g., Dr. C. Arden Pope, III: "There needs to be more care in the use of different measures of PM. For example, the different daily time-series mortality studies used various measures of PM. What conversion methods were used to get a common PM effects estimate?" Dr. Harriett Ammann: "[U]nder Findings PM2.5 was the metric for annual mortality. PM10 was the metric for daily mortality????" Dr. Suresh Moolgavkar: "The report keeps jumping back and forth between different PM metrics... What assumption was made regarding the relationship of the two indices of PM?"

[27] See, e.g., Dr. C. Arden Pope, III: "In general, I find this to be an impressive report.... It gives an extensive review of the relevant air pollution epidemiology literature. It reasonably uses the empirical results that have been reported in this literature to calculate health effect estimates." Dr. Harriett Ammann: "The study ... constitutes a reasonable approach to evaluating potential health effects from emissions of the power plant, as well as predictions of future effects with and without the application of scrubbers to the plant's stacks." Dr. Suresh Moolgavkar: "This is a reasonable approach to assessing the health risks due to air emissions from the Centralia plant."

[28] See supra note 24.

[29] Bowers appears to argue that the PCHB affirmatively chose to rely on the short-term rather than the long-term estimates. There is no record of such a statement in the findings and conclusions, except the PCHB's recognition that the EPA used short-term studies in developing its PM2.5 NAAQS.

[30] The corollary of Bowers' argument is equally plausible; that is, because mortality estimates based on long-term studies are less certain than that of short-term studies, the risk may be considerably less (or greater) than estimated.

[31] Environmental Protection Agency, The Benefits and Costs of the Clean Air Act, 1990 to 2010 (Nov. 1999) [hereinafter 812 Prospective Study].

[32] RCW 34.05.558 provides: "Judicial review of disputed issues of fact shall be conducted by the court without a jury and must be confined to the agency record for judicial review as defined by this chapter, supplemented by additional evidence taken pursuant to this chapter."

RCW 34.05.562(1) provides that we may receive additional evidence only if it relates to the validity of the agency action at the time it was taken and is necessary to decide disputed issues regarding: (a) improper constitution as a decision-making body or grounds for disqualification of those taking the agency action; (b) the unlawfulness of the procedure or decision-making process below; or (c) material facts not required to be determined on the agency record. We may also remand for further fact finding if new evidence becomes available "that one or more of the parties did not know and was under no duty to discover or could not have reasonably been discovered until after the agency action." RCW 34.05.562(2)(b)(i).

We also refer to RCW 34.05.518, which permits this court to accept review of an administrative agency's final decision upon certification by either the superior court or an environmental board. For direct review upon certification by the superior court, judicial review is "limited to the record of the agency proceeding." RCW 34.05.518(2). Here, we accepted direct review upon certification by an environmental board (i.e., the PCHB), which does not set forth the same limitations on judicial review. RCW 34.05.518(3). However, it appears the principle is equally applicable.

[33] Even if the EPA's 812 Prospective Study had been before the PCHB, we conclude that it does not preclude the appropriateness of using short-term studies as a tool for estimating the effects of air pollution on mortality. The Study states: "Researchers have found significant correlations using both types of studies; however, for this analysis, we rely exclusively on long-term studies to quantify PM mortality effects, though the short-term studies provide additional scientific evidence supporting the PM/mortality relationship." 812 Prospective Study, supra note 31, at 58 (emphasis added).

[34] "Mortality displacement" refers to "an advancement of the time of death" (i.e., by days or weeks) as compared to an actual increase in the overall mortality rate.

[35] Bowers states in his brief, for example: "By `balancing' the loss of lives against the `sever[e] impact' of `the loss of hundreds of direct and thousands of indirect jobs', the PCHB permitted fatal emissions rather than demand lower RACT limits."

[36] In his reply brief, Bowers appears to abandon his argument that the PCHB "should never have considered job losses" in evaluating the economic feasibility of a particular control option. We nonetheless address this issue as it is relevant to the RACT review process.

[37] "Where an administrative agency is charged with administering a special field of law and endowed with quasi-judicial functions because of its expertise in that field, the agency's construction of statutory words and phrases and legislative intent should be accorded substantial weight when undergoing judicial review." Redmond, 136 Wash.2d at 46, 959 P.2d 1091 (quoting Overton v. State Econ. Assistance Auth., 96 Wash.2d 552, 555, 637 P.2d 652 (1981)).

[38] Bowers also argues that the PCHB failed to evaluate the "magnitude" of health impacts. The only evidence he presents to discredit the PCHB's results is his assertion, which we previously rejected, that the CALPUFF model's aqueous phase conversion rate underestimated ambient PM2.5 levels. See discussion supra section B.1.

[39] Direct losses are those that result from job losses at the Plant or mine, while indirect losses stem from those who lose jobs as a result of Plant employees who can no longer afford to purchase goods and services.

[40] Recall that Samet estimated 13 mortalities after controls based on the long-term studies, as opposed to 1 from the short-term studies.

[41] Bowers states that Halvorsen's knowledge of the health impacts associated with the Plant is irrelevant because he was not called to testify about health impacts, but about the reasonableness of using VSL methodology. While this may be true, Halvorsen implies in his prefiled testimony that health impacts from the Plant are known and measurable. He states:

An accurate analysis of the economic feasibility of additional pollution control equipment at the Centralia power plant would consider the economic value of expected changes to mortality and morbidity caused by the plant as a result of emission reductions. In cases such as this, where substantial mortality and morbidity impacts are forecast, a proper and complete economic feasibility analysis cannot be done without assessing the magnitude in dollars of these impacts.

(emphasis added).

[42] Bowers again impermissibly asks this court to consider additional evidence, in the form of the EPA's 812 Prospective Study, that postdates the PCHB's order. RCW 34.05.558, .562; see also supra note 32. Even if we were to consider the Prospective 812 Study as permissible evidence, we find the following observation made by the study to be noteworthy:

Our analysis clearly shows that uncertainty in the measurement of the value of statistical life dominates the quantifiable uncertainty in our overall estimates. In addition, there remain several important non-quantified uncertainties in the use of labor market studies to value avoidance of environmental risks from air pollution.

812 Prospective Study, supra note 31, at I-6 to I-7.

[43] Bowers' expert later added: "Clearly the fact that something hasn't been done is not an excuse for never having to do it." While it is true that in certain circumstances it may indeed be worthwhile to implement an as-yet-untested technology, it is less the decision than the decisionmaking process that we are concerned with in reviewing the PCHB's decision. Here, the statute requires the control technology options to be evaluated against a variety of factors, one of which is the guidance provided by other states and local authorities. The PCHB was presented with no information that environmental dispatch is used in the industry. We cannot then say that, in rejecting what has been to date a theoretical concept, the PCHB acted in a willful and unreasoning fashion without considering the facts or circumstances surrounding the issue. See City of Redmond v. Cent. Puget Sound Growth Mgmt. Hearings Bd., 136 Wash.2d 38, 46-47, 959 P.2d 1091 (1998).

[44] Wooley explained that "marginally profitable" in this context means that power generated by the Plant can be sold for slightly more than its running costs.

[45] In unchallenged findings, the PCHB highlighted the cost effectiveness estimates for other plants undergoing analogous control technology determinations. For plants in NAAQS nonattainment areas of the East Coast, the cost per ton of SO2 removed is approximately $300 to $400. In a Minnesota RACT determination, $338 per ton of SO2 reduced was determined cost effective. These findings lend further support to the PCHB's decision to reject the natural gas alternative as not economically feasible.

[46] Bowers again impermissibly asks this court to consider evidence not before the PCHB. See RCW 34.05.558, .562; see also supra note 32. We nonetheless find that these submissions lend no additional support to Bowers' argument. In Detroit Edison Co. v. Michigan Dept. of Envt'l Quality, 39 F. Supp. 2d 875 (E.D.Mich. 1999), the court did not "order" the power plant to switch to natural gas fuel. Instead, the company, which restarted its power plant after a decade of nonuse but without obtaining environmental permitting under the CAA, "proposed to convert the Plant to a natural gas-fired facility." Detroit Edison, 39 F.Supp.2d at 876. The interim remedial order simply memorialized the permit conditions agreed to by the parties. Similarly, the State of Florida press release indicates that Tampa Electric Company voluntarily agreed to convert to natural gas; the company was not "ordered" to switch from coal to natural gas, as maintained by Bowers. "DEP and Tampa Electric Company Enter Into Historic Agreement," External Affairs Communications, (Dec. 7, 1999), at http://www.dep.state.fl.us/comm/release/1999/99-167.htm.

[47] In addition, the respondents presented evidence that the NOx RACT emissions limit is more stringent than that of many other states as well as the EPA's Acid Rain NOx Emission Reduction Program, lending further support to PCHB's conclusion that low NOx burners meet RACT requirements.

[48] The only evidence Bowers presents is information that postdates the PCHB's decision. He cites to a final rule adopted by the EPA, 65 Fed.Reg. 2674 (Jan. 18, 2000), and to Michigan v. EPA, 213 F.3d 663 (D.C.Cir.2000), petition for cert. filed, 69 U.S.L.W. 3297 (U.S. Oct. 20, 2000) (No. 00-632), both of which we may not consider. See RCW 34.05.558, .562; see also supra note 32. As respondents point out, however, both the EPA final rule and the Michigan decision address issues of NOx emissions in eastern and midwestern states that are in ozone nonattainment areas under NAAQS, and are thus inapplicable to the instant case.

[49] The PCHB applies the civil rules, which includes CR 56(c) governing summary judgment orders. WAC 371-08-305(2).

[50] The only authority Bowers cites for his proposition that SWAPCA is required to consider incidental air quality benefits is inapposite here. In reviewing the federal statute governing BACT (not RACT), the EPA Administrator ruled that the consideration of incidental benefits is permissive, not mandatory. In re N. County Res. Recovery Assocs., Applicant, PSD Appeal No. 85-2, 1986 WL 80843 (U.S. Envt'l Appeals Bd. June 3, 1986) ("EPA may ultimately choose more stringent emission limitations for a regulated pollution than it would otherwise have chosen if setting such limitations would have the incidental benefit of restricting a hazardous but, as yet, unregulated pollutant.") (emphasis added).

[51] At summary judgment, the PCHB concluded that Bowers had not created a genuine issue of fact with respect to certain visibility issues. Bowers moved for reconsideration. The PCHB deferred ruling on the issue until it issued its modified findings of fact and conclusions of law, in which it affirmed its decision on summary judgment.

[52] Bowers in fact lists in his opening brief the sections of the RACT order's technical support document discussing air clarity.