*1 regular and con- the rule’s incon- to demonstrate some evidence provide five cases Further, Endorsing this meth- application. Dixon. sistent application sistent are distin- statis- od, only partial several cases a providing and then argues state they petition- involved guishable because must now live with the state picture, tical which restricted pled guilty, had application ers who Dixon’s the inevitable result: See Cal.Pe- appeal. rights on direct their denials tells percent of all habeas to twelve But the fails § 1237.5. state nal Code consis- nothing almost about rule’s .us affect rights these restricted how explain therefore, and, adequa- application tent applicabili- and Dixon’s petitions habeas oral ar- requested at cy. the state While by also belied claim is ty the state’s —and to allow it to that we remand gument cases several habeas the existence records, missing which compile petition- claims of barred the Dixon where available, we were conceded likely state See, Reyes v. e.g., pled guilty. had ers who yet another give the state see no reason 3734550, CV-13-1248, Cash, 2014 WL No. opportunity a full chance. The state had 2014) (C.D.Cal. (noting that July *1 adequacy and failed prove Dixon’s had his pled guilty had who petitioner a Thus, district we remand for the do so. procedural denied on petition state habeas mer- Lee’s claims on the court to consider Dixon); including by grounds, Gustafson its. CV-13-1737, 2014 WL Long, No. v. AND VACATED. REVERSED 2014) (C.D.Cal. *12 Apr. (same). sum, only with evidence
In we are left applied was to between Dixon bar of all habeas twenty-one percent
seven and surrounding filed in the months
cases statistic, with- Because this
Lee’s default. more, failed to incomplete, the state
out proving the Dixon burden of meet its PARKS CONSERVATION NATIONAL “clear, consistently applied, and bar was ASSOCIATION; Environ- Montana pur- at the time of [Lee’s] well-established Center; Sierra mental Information Collier, F.3d at default.” ported Club, Petitioners,
v. CONCLUSION PROTEC- U.S. ENVIRONMENTAL must suggest that state do not We .the Jackson, and Lisa P. TION AGENCY analysis prove a statistical always use Administrator, Environ- United States any nor do we set adequacy, rule’s Respon- Agency, Protection mental that must be statistical bar precise dents, ap- an if the state does use such reached Here, however, the chose to state proach. LLC, Respondent- Montana, PPL framework, and a statistical just use such Intervenor. that, mandatory, if the Dixon bar is argues Petitioner, LLC, way best analysis would be the statistical claims, argument we find because not address argues that two of her also Lee inadequate state law bar to be an were Supplemental Dixon 6 and Ground Ground remaining should ground. All of her claims appeal sufficiently in her direct raised on the merits remand. We need be considered they Dixon. are barred *2 National Parks Conservation Associa
tion, Montana Environmental Infor Center, Club,
mation and Sierra Inter
venors,
v. Agency
U.S. Environmental Protection Jackson,
and Lisa P.
Respondents. 12-73710,
Nos. 12-73757. Appeals,
United States Court of
Ninth Circuit.
Argued May Submitted 2014.
Filed June *4 Blatt,
Lisa S. LLP, Arnold & Porter Washington, DC, argued the cause and filed the briefs for petitioner PPL Mon- tana. With her on the briefs Jona- were than S. Martel Jaros, and Christopher A. Porter, LLP, Arnold & Washington, DC. Jenny Harbine, K. Earthjustice, Boze- man, MT, argued the cause and filed the briefs for NPCA. Wither her on the briefs was Brimmer, Janette K. Earthjustice, Se- attle, WA. Pinkston,
Daniel Environmental Defense Section, Environment and Natural Re- Division, sources Depart- United States Justice, Denver, CO, ment of argued the cause and filed respondents. the brief for With him on the Ignaeia brief were S. Moreno, Attorney General, Assistant Envi- Division, ronment and Natural Resources Department Justice, United States DC, Washington, and Elizabeth B. Daw- son, Section, Environmental Defense Envi- Division, ronment and Natural Resources Justice, United Department States Washington, DC. Jenkins,
Michael G. Assistant General Counsel, PacifiCorp Energy, Salt Lake City, Rawson, Quin- UT and E. Blaine Ray Nebeker, P.C., ney UT, & City, Salt Lake filed the brief for Amicus Curiae Pacifi- Corp. A F.
Before: DIARMUID BERZON, O’SCANNLAIN, S. MARSHA impair “visibility haze is Regional TALLMAN, Circuit and RICHARD C. dispersed by geographically ment caused Judges. their particles fine emitting sources into, Am. Corn Grow the air.” precursors (D.C.Cir. by Judge 1, O’SCANNLAIN.
Opinion Ass’n v. ers curiam) 2002) Haze (citing Regional (per BERZON. by Judge Concurrence 1, 35,714 (July Regulations, 51)). 1999) (codified Pt. Con at 40 C.F.R. OPINION §§ and 169B of the 169A gress enacted “Act”) (the Air Act “CAA” or Clean O’SCANNLAIN, Judge: Circuit haze. Id. regional problem address the Environ- 3-4; Air Act Amendments must decide whether see Clean We 95-95, § haze 91 Stat. Agency’s regional Pub.L. No. Protection mental (current law- version of Montana regulations for the State Amendments, Pub 7491); Air Act § Clean at certain limits fully prescribe emission 101-549, 104 Stat. L. No. power plants. (1990) (current at 42 U.S.C. version 7492). establish as provisions These *5 I future, any “prevention the goal national any existing, impair remedying and operates PPL Montana and the Petitioner mandatory I visibility in Class hydroelectric ment of partially owns coal-fired from man- impairment which results Montana, areas including plants in power air U.S.C. pollution.”1 made Generating Sta- Electric Colstrip' Steam 7491(a)(1). § Corette (“Colstrip”) and the J.E. tion (“Corette”). Peti- Steam Electric Station" imposes requirements several The Act Asso- Parks Conservation tioners National on EPA relevant on States ciation, Informa- Environmental First, requires pro- Act EPA to case. (collectively, Center, and Sierra Club tion to “assure ... reason- mulgate regulations “NPCA”) conservation or- nonprofit are meeting the national progress toward able enjoy members wilder- ganizations whose haze reduction. goal” regional regional 7491(a)(4). impacted Second, areas in- ness the Act § U.S.C. of Montana. regulations for the State EPA a haze “State vites each State submit (“SIP”) such setting are dissatisfied with petitioners Both Plan” Implementation in measures argues, es- limits and other regulations. forth emission to- sence, stringent; necessary progress too to make reasonable they are they visibility goal. not national See contrary, do ward the argues, to 7491(b)(2). If, 7410(a), like §§ visibility impairment remedy U.S.C. enough do Montana, not to submit a State chooses in various relevant by regional haze caused EPA to requires Act plan, areas. wilderness here, alia, include, Na- 7491(g)(5). § Yellowstone Relevant inter national I areas Class Park, 5,000 Park, Bend exceeding acres in size National UL areas Glacier wilderness tional 7, August parks in existence Lake Refuge, and national and Medicine Wildlife National 6,000 exceeding See 42 acres size. designated man- Area have been Wilderness 7472(a). "mandatory § term class U.S.C. The See 40 C.F.R. datory Class I areas. "may describes those I federal areas" 81.400, §§ 81.417. than class I." Id. designated as other produce a Implementation Regional Plan” “Federal Regulations Haze and Guidelines (“FIP”) for that State. U.S.C. for Best Available Retrofit Technology 7410(c)(1)(A). § (BART) (the Determinations “2005 Regu lations”), which provides Act further revised the text imple- all plans mentation Regulations. must earlier 39,- installation See 70 Fed.Reg. of the “best 2005). available retrofit 6, technology” (July (“BART”) to reduce from cer- published also Guidelines tain emission sources that operation- were BART Determinations Regional Under the al between 1962 and 1977 (“BART-eligible (the Haze Rule “Guidelines”), 40 C.F.R. sources”). 7491(b)(2), (g). Pt. App. Y (Sept. 2005), prescribing Five statutory factors determine which five steps for application of the five statu- type of emissions-reducing technology con- tory BART factors: stitutes BART for such sources: 1)
(a) (Step Identify the costs of all compliance; available retrofit control (b)'the technologies; energy and quality non-air envi- impacts ronmental of compliance; 2) (Step Eliminate technically infeasible (c) any existing pollution control tech- options; nology source; at a 3) (Step Evaluate control effective- (d) the remaining useful life of the emis- ness of remaining control technolo- source; sion gies; (e) degree improvement 4) (Step (identified Evaluate impacts anticipated!.] § 7491(g)(2), see 40 C.F.R. Pt. U.S.C. 7491(g)(2). Y, § App. IV.D.4, 70 Fed.Reg. Act, Pursuant to the EPA promulgated 166) and document results; Regional (the Regulations Haze “Regu 5) (Step Evaluate visibility impacts lations”), which States, asked certain *6 in (measured in “deciviews,” see cluding Montana, to analyze sources of 51.301).2 § C.F.R. emissions within the State and to a develop plan to eliminate all man-made visibility § Id. YApp. IV.D. impacts by 2064. See 35, 64 Fed.Reg. at 714; § 40 C.F.R. 51.308. Regulations The B require any implementation plan to include 2006, In the Montana Department of (1) “reasonable, (2) progress goals”; a cal Quality Environmental notified EPA that culation of baseline and natural visibility it did not produce intend to a SIP trigger- (3) conditions; a long-term strategy for ing obligation EPA’s produce to a FIP for achieving “reasonable progress goals”; the State of Montana: (4) See 42 and additional of monitoring emission 7410(c)(1)(A). § EPA published sources in a pro- Class I federal areas. See 40 posed FIP for 51.308(d)(1)-(4). § C.F.R. Montana April After the D.C. (the Rule”). Circuit “Proposed provisions vacated Approval and Regu lations relating to determinations, Promulgation BART Implementation Plans; Growers, see Am. Corn Montana, F.3d at State of 23,988 77 Fed.Reg. promulgated new BART regulations 2012). (Apr.
2. "Deciview a means eye. measurement visibili- Most people can a change detect ty impairment.” 40 C.F.R. 51.301. "Each visibility at one Proposed deciview.” change deciview equal is an incremental 23,992. Fed.Reg. at change in visibility perceived by the human (the Rule” or the “Final Montana State of required petitioner Rule Proposed The 57,864 “Rule”), (Sept. to various actions to take PPL Montana 52.1396). (codified 2012) at 40 C.F.R. compounds of two emissions
reduce —ni- Proposed implemented the Rule The Final dioxide (“NOx”) and sulfur oxide trogen relevant to respects all Rule in almost plants partially power (“S02”) two—at appeal.3 and Corette. Colstrip operates, and owns im- required to First, was PPL Montana C technologies at Units new several
plement pe- station. Colstrip four-unit both filed 1 and and NPCA PPL Montana 30-day to emissions BART determi- reduce EPA’s To review of titions for NOx lb/mmBtu, Corette, of 0.15 rolling pe- which average Colstrip and nations at to PPL required Montana Proposed purposes Rule consolidated for titions were technologies separated two new install contends that PPL Montana this appeal. — (“SOFA”) non- and selective air overfire forth in the Rule limits set the emissions (“SNCR”) at catalytic reduction 1 and Colstrip Units and for S02 NOx 24,035. 24,027, To Id. at 1 and 2. the BART determinations 2—as well a.30-day rolling reduce emissions essentially too underlying such limits—are S02 lb/mmBtu, Proposed average of 0.08 failed rea- argues that EPA stringent. It imple- PPL Montana to required Rule BART selec- sonably explain the Rule’s technologies at new two ment additional visibility model does tions and injection 2—lime 1 and Colstrip Units im- reasonably anticipate 24,028, Id. “scrubber.” and a fourth re- of the Rule’s as a result provement not re- 24,035. Rule did chal- PPL also quirements. Montana new implement PPL quire Montana Rule in the set forth lenges the emissions and Colstrip Units 3 technologies at for Corette. required Second, Proposed Rule emis- challenges Rule’s also emis- limit BART determinations NOx sions limits Pro- station. The at the Corette sions Corette, con- 1 and Colstrip Units 30-day average roll- imposed Rule posed are not essentially that limits tending limits of ing emission NPCA, lb/mmBtu According to enough. stringent id. and 0.70 lb/mmBtu not to installation S02. EPA’s decision required 24,042, It at these loca- technology of more advanced current using achieve such arbitrary, and ca- unexplained, was tions *7 1 and Colstrip Units technology; unlike Moreover, NPCA, according to pricious. require does 2, Proposed Rule the installation not to EPA’s decision technology new at Corette. of installation 3 technologies Colstrip any of new 24,047. 24,043, that, Id. at requirement satisfy 4 the and fails toward progress make reasonable the Rule and com- NPCA Both improvement. the con- Proposed Rule. At on mented the period, notice and comment clusion of the D FIP final for Montana EPA issued its regional the implementation of and EPA’s 18, Approval 2012. See September 307(d) of by Plans; governed plan haze is Section Implementation of Promulgation and from 0.35 EPA's 0.40 lb/mmBtu Although Final Rule did not alter lb/mmBtu lb/mmBtu, re to 0.57 from or 0.70 for either lb/mmBtu BART determinations 77 Corette, spectively. See Final Rule’s lower did 57,911. and at Corette emission limits NOx
1141
Act,
7607(d).
§
42 U.S.C.
The statute
A
requires that
explain
the basis for its
The heart of
dispute
about EPA’s
decisions, including underlying
bas-
factual
BART selection at Colstrip Units 1
2
and
es, methods of analysis,
legal
poli-
and
is its determination that
of
use
SNCR in
cy
7607(d)(6)(A).
Id.
considerations.
addition to SOFA cost-effective,
is
and that
respond
must also
comments,
use
selective
catalytic
reduction
criticism, and new data submitted during
(“SCR”)
aggressive
more
technology—
—a
period.
comment
Id.
in addition to SOFA is not. Both parties
urge that EPA’s
analysis
cost-effectiveness
When we review an agency ac
fails appropriately to consider the costs of
tion “involv[ing] primarily
fact,”
issues of
compliance and degree of visibility impair-
and where “analysis of the relevant docu
ment,
required
as
by
Act,
see 42 U.S.C.
‘requires
ments
high
level of technical
§ 7491(g)(2), at Step Three of EPA’s
expertise,’ we must defer to ‘the informed
BART analysis under the Guidelines, see
discretion of the responsible federal agen
51,
40 C.F.R. Pt.
§YApp.
IV.D.
”
cies.’ Marsh v. Oregon Natural Res.
Council,
requires
Rule
360, 377,
490
U.S.
1851,
S.Ct.
re
duce
(1989)
L.Ed.2d 377
at Colstrip
(quoting
Units 1
v.
Kleppe
and 2 to
Club,
Sierra
390,
over a
412,
30-day
U.S.
lb/mmBtu
96 S.Ct.
rolling
2718,
average.
(1976)).
L.Ed.2d
C.F.R..
However,
52.1396(c)(1).
EPA concluded that
we do not defer to
EPA actions that are
reduction could be
achieved
“arbitrary,
installing
capricious, an abuse of discre
both SOFA and SNCR technologies.
tion,
or
not in
otherwise
accordance with
Fed.Reg. at
law,”
or that exceed EPA’s statutory juris
contends that EPA offered
jus
insufficient
diction. 42
§§
(C).
7607(D)(9)(A),
tification
rejection
for its
SCR BART
EPA’s actions must be reasoned; EPA
in favor of SOFA and
together,
SNCR
acts in an arbitrary and capricious manner
argues that PPL Montana should
re
if it fails to consider an important aspect of
quired to install SCR
SNCR,
instead of
a decision or if
explanation
contradicts
and in
SOFA,
addition to
at all four Col-
the evidence before it. Motor Vehicle
strip units. PPL Montana makes a relat
U.S.,
Ass’n
Inc. v. State Farm
Mfrs.
point,
ed
but on the other side of the
Co.,
Mut. Auto. Ins.
U.S.
argument, contending that EPA offered
(1983).
S.Ct.
H42 the emissions underestimated EPA
that that 1 and and Colstrip Units of the various baseline the costs EPA identified Colstrip implementing the cost reduction technologies for it miscalculated NOx 1 and as follows: units. at these SCR SCR, as- simply EPA 89-97, to 66-74, respect Rule, With Tables Proposed and SCR of SOFA that the cost 24,032-34. serted 24,024-27, Fed.Reg. at 1) ($3,195/ton Unit together reasoning essentially no Rule offers visibility improve- by the justified “not was and of SOFA EPA’s selection behind deeiviews,” but that “[t]he ment of .404 to either opposed as together SNCR i ($1,564/ton) + SNCR cost of SOFA lower together, and SCR alone or SOFA SOFA visibility improvement justified when the is at Col- to reduce BART Pro- considered.” is deeiviews] .264 [of respect to the 1 and With strip Units 24,027. Rule, Fed.Reg. at EPA posed in addition to SNCR require decision not use a set cost- that it did maintains Proposed Rule SOFA, in the EPA asserted threshold to improvement or effectiveness and SNCR the cost of SOFA only that SCR, suggest do not and we disqualify justified when “is together any expla- But absent it do that must so. Pro- is considered.” improvement cost- EPA determines all of nation at how 24,027. In Rule, Fed.Reg. at posed effectiveness, impossible for Rule’s it is re- comment to PPL Montana’s response SOFA, why EPA ruled reader to determine assertion, explain EPA questing out. in ruled SCR and and SNCR Rule dis- noted that alone and the Re acknowledged SOFA the cost for both closes SNCR, that it it prevent and does not from combined with Haze Rule gional SOFA line” “bright called a consideration implementing as BART what SNCR “selected cost-effectiveness, contends costs, presented but of which were all rule of these require not it to do rule.” Final do proposed regulations that its public our Fed.Reg. at 57,886. But PPL so. See Regulations do sure, that EPA failed To be the Act objection not Montana’s EPA explain specifically between cost differential disclose the through use decisions cost-effectiveness and SOFA its alone SOFA implementing But the law does line” rule. “bright of a It -instead contends along with SNCR. why it explain “cogently require why the the reasons explain EPA failed given in a man has exercised discretion through SOFA achieved improvement Farm, U.S. at See State insufficient, ner.” improve- but was alone Coalition, 2856; Yellowstone Greater SNCR S.Ct. through achieved ment S.OFA (9th Servheen, Inc. v. both sufficient together was deemed *9 Cir.2011) “a rational connection (requiring cost-effective. between the data before agency] [the suggested NPCA has no reason to believe conclusion”); Nw. Envtl. Ctr. v. that PPL Montana would change course Def. Admin., Power Bonneville 477 F.3d and remove the additional combustion con- (9th Cir.2007). unexplained EPA’s as trols it already had installed. EPA gave a that the combination sertions of SOFA and reasoned, response rational argu- cost-effective, SNCR is but that SOFA ment, response to which a reviewing insufficient, alone is and that the combina court defers. See Latino Issues Forum v. tion of SOFA and SCR is not cost-effec EPA, (9th Cir.2009). tive, are unsupported by any explained Second, PPL objects that reasoning. These assertions leave the EPA’s reliance on a dollars-per-ton metric metric, Rule’s reader wondering what if to calculate cost effectiveness is inappro- any, BART, EPA used to determine or if priate because it actually does not measure EPA employed metric, no why not. the improvement achieved in visibility. Therefore, we conclude that EPA’s BART But EPA responded that it has previously determination for emissions at Col defined “cost effectiveness” as pol- “tons of strip arbitrary and is capri lutant emissions removed” compared with cious. “annualized control costs.” Regula- tions, 39,167. 70 Fed.Reg. Indeed, at PPL Montana essentially argu- abandoned this responses to petitioners’ more in reply EPA pointed after ment to its challenges minor to its cost-effectiveness regulation in response brief. analysis make clear that capable it is Third, objects that EPA improp- required rational explanation. erly calculated the cost of installing SCR First, NPCA challenges EPA’s selection Colstrip Units 1 and leading it to years 2008-2010 as the emissions reject SCR as BART. It challenges EPA’s period baseline for calculating BART at cost, calculation of the capital its choice of Colstrip Units 1 and 2. It contends .that rate, interest lifespan and the and mainte- EPA underestimated potential benefit nance costs in assumed EPA’s calculation. using by SCR first underestimating the provided a reasoned response to existing emissions baseline. EPA’s BART NPCA’s comments on questions. these Guidelines it to a represen- choose 57,888-89. See Final Fed.Reg. period tative baseline supplies “a real- NPCA maintains should have istic depiction of anticipated annual emis- looked to different data sources to deter- sions for Guidelines, the source.” BART mine these costs. But it offers a reason 70 Fed.Reg. at EPA chose the that EPA should look to different underly- period 2008-2010 on reasoning ing cost in only data one example case. It Colstrip had installed additional combus- (cid:127) contends EPA should have chosen a 5% tion in controls reducing its emis- interest rate instead of a 7% interest rate sions. See argues id. NPCA that EPA calculating these comport costs to should have chosen periods earlier because requirements set forth the EPA Air Montana is not required to maintain Pollution Cost But Manual. that manual rate of emissions achieved between 2008-2010, instructs industry describing they may actors changes Col- strip as wish use a “unenforceable.” But different interest rate EPA of- fered a reasoned their own response explained, calculations. As EPA NPCA’s com- ment about its 57,888-89, choice—that had see achieved reduced using clearly technolo- EPA’s Manual contemplates that gy plans it has no to deactivate—and EPA will use the interest rate set
1144 at control Col- emissions for here, BART pursuing in did OMB, properly as it S02 the essentially same 1 and for strip Units See id. own calculations. EPA’s BART dispute they reasons NOx the determined EPA determinations. technologies reduc- the various costs of rationality the dispute petitioners Both Colstrip as follows: at emissions ing as a fourth scrubber of EPA’s selection of S02 scrubber ves- 102-111, and an additional 78-87, injection Rule, Tables sel). concluded But EPA then id. 24,036-37. 24,028-29, at Fed.Reg. to find that cost “continuéis] that-it un- that EPA Montana contends visibility benefits[.]” given reasonable a fourth installing of cost derestimated EPA in this case acknowl- Although Id. Colstrip Units at scrubber concerning incremen- the comment edged cost-effec- sufficiently explain to failed cost-effectiveness, still it nevertheless tal require- for this determination tiveness a cost what makes reason- explain failed to NPCA, contrary, to the contends ment. visibility benefits. light potential of able in installation required EPA should have limits Rule’s emission Col- at Col- scrubbers replacement advanced S02 1 and force strip addi- Units just the introduction an strip, not $25,000,000 capital an additional spend responds scrubber. tional $2,210,000annually and an additional costs its calculation explained appropriately of 0.055 visibility improvement to achieve an adding additional costs involved id. sampled location. See at the deciviews decided appropriately that it scrubber may be neces very well improvement This system of entirely new an ultimately cost-effective. But sary and scrubbers, only adding one opposed justifying reasons supplied EPA has no additional scrubber. explana Based on its that determination. analysis with EPA’s cost-effectiveness is left to wonder tion, Rule’s reader at respect to S02 to determine EPA used rationale what as its cost- same defect 1 and suffers the requires Again, the law cost-effectiveness. emis- respect effectiveness question. See to that answer a reasoned analysis improve- is an sions. 2856; Farm, 103 S.Ct. U.S. at State acknowledged respect: in one ment 1028; Servheen, Nw. Envtl. F.3d adding an addition- incremental cost Ctr., Def. ($2,410/ton Colstrip vessel al scrubber ra- contrast, thoroughly and By see Final Unit response to NPCA’s explained its tionally adding a 57,892), scrubber and noted contends point. NPCA objection on this incrementally improve visibil- vessel would requir- considered have injec- that EPA should (using lime ity from 0.225 deciview existing replacement of complete alone) ing lime (using 0.280 deciview tion *11 systems control for emissions at Col- tional controls at Corette were not cost- S02 strip 1 explained Units and 2. EPA effective was and arbitrary capricious. decision not to do so: its BART Guidelines entirety The of EPA’s cost-effec recommend constructing system new reasoning tiveness respect to Corette when a current system control achieves follows: weighed “We have costs against than percent “less 50 removal efficiencies.” anticipated visibility impacts for Cor- Rule, 57,892 77 Fed.Reg. at (quoting Any ette. of the options control would Regulations, Fed.Reg. 39,171) 2005 70 at positive have a impact on visibility; howev (internal omitted). quotation marks er, the cost of justified controls is not current system emissions control at S02 improvement.” 1 Colstrip Units and 2 achieves removal Rule, 24,043. 77 Fed.Reg. at Again, this efficiencies exceeding percent, 50 see Pro reasoning fails to reveal to a reader how Rule, posed Fed.Reg. 24,028, 77 at and EPA determined that the cost of controls therefore EPA is not required to consider justified. Moreover, were PPL Mon replacement technology. Rule, See Final tana is correct that this reasoning appears 57,892. Fed.Reg. at inconsistent with analysis for Col- This kind of explanation elab —not strip Units 1 and units, 2. At those orate, rational, but and thus adequate —is concluded that a cost-per-ton ap rate of missing from EPA’s conclusion that PPL $1,500 proximately emissions con Montana must install a fourth scrubber at justified. trols was supra II.A.1. Colstrip Units 1 and EPA’s determina Corette, Yet at EPA concluded that a cost- tion of BART to control $1,487 per-ton rate of did not justify the suffers the same failure of rational expla potential emissions reductions. See Pro nation as its BART determination for NOx posed 24,040-43. at Fed.Reg. emissions. Because the rule offers no rea Corette, moreover, is closer to Class I soned explanation to support require areas than Colstrip, and emissions controls ment of a fourth scrubber at Colstrip there could improved have visibility at Units and we conclude that such re least as much as they were set to control quirement is arbitrary capricious. and visibility at Colstrip. See id. The seem ing inconsistency in EPA’s BART determi B nations at Colstrip Units 1 and and also, relatedly, contends is, Corette absent explanation, “the hall that EPA’s BART at determinations Col- arbitrary mark of action.” Sierra Club v. Units 1 strip arbitrary and are capri- and (D.C.Cir.1983); 719 F.2d they cious because are inconsistent with also Corp. see Gen. Chem. v. United analysis, EPA’s Corette which does not States, (D.C.Cir.1987) station, additional controls curiam) (per analysis (finding arbitrary see Final capricious it “internally because was Montana points PPL reject- out inconsistent inadequately explained”). implementation ed Corette, SOFA but found similar costs for SOFA installa- C justified Colstrip tion and 2. See NPCA, supra hand, II.A.1. Finally, on the other objects, on two argues EPA’s conclusion that addi- grounds, to EPA’s use CALPUFF4 4. CALPUFF is a model used to estimate an App. § 40 C.F.R. Pt. 51 III.A.3. Y impact visibility. emissions source’s EPA’s use of CAL- object wholesale determining BART at
visibility model visibility measurement. model for PUFF first 2. PPL Montana 1 and Colstrip Units the mod- Instead, the results argues incre- potential the maximum objects that insignificant are too in this case predicts el is below of SNCR visibility benefit mental measure, therefore the model -to and falls within range- perceptibility anticipated as the reasonably cannot be error, meaning such *12 margin model’s of During the notice and com- requires. Act ... “reasonably be cannot improvement argued that period, ment by the Act. See 42 required as anticipated” visibility benefit the maximum incremental challenges It also 7491(g)(2). U.S.C. installing at Col- gained by SNCR to be to the emissions of CALPUFF application 2 be 0.085 deci- 1 and would strip Units of their distance because at issue sources PPL views, improvement an incremental areas, contending the model I from Class within the CAL- contends falls Montana great distances. at such is inaccurate EPA re- margin of error. model’s PUFF that, because CAL- responds EPA regulation that its by explaining sponded in the Guide- approved to re- visibility improvements model was be PUFF permits every appli- visibility impacts fall lines, even when quired need not defend EPA perceptibility. See Rule, below threshold Fed.Reg. 77 at Final cation. See 57,867. Rule, at EPA Fed.Reg. Final that the Re- 57,867. responds also EPA Regula- that its 2005 responded further anticipated and Haze Rule itself gional to model direct use of the CALPUFF tions argument, rea- PPL rejected Montana’s percentile of the 98th estimate to consider less-than- “[flailing soning highest than the data impairment, rather visibility im- to contributions perceptible collected, uncertainty minimize in its to intent ignore the CAA’s to would pairment PPL But calculations. id. to requirements apply BART sources have EPA to discontinue does not ask cause, to, such well as that contribute as below one- of CALPUFF application Final impairment.” threshold, toor run perceptibility deciview 70 Fed. 57,883 Regulations, (quoting 2005 points data than using the model different 39,129). responds further EPA Reg. selected; it asks how CALPUFF the ones acknowledged that there is it “has that additional EPA’s conclusion explains pre- model uncertainty in the CALPUFF anticipa- measures will lead reasonable and “the CAL- visibility impacts,” dicted case, in this visibility improvement tion of underpredict and model can PUFF. both asserts, when, improvement an PPL but visibility impacts,” overpredict “beyond the CALPUFF deeiview is applica- ... is a reasonable “CALPUFF any ability predict with confi- model’s facility such a tion to determine whether dence.” exceeding or 3 deci- impact an [with meaning not response does reasonably anticipated can views] comment, PPL Montana’s as address fully any impairment of cause or contribute Falls Aluminum it must. See Columbia 57,- Fed.Reg. at visibility.” (D.C.Cir. 914, 923 Co. v. F.3d EPA Regulations, 77 Fed. (quoting 2005 Indus., 1998); Inc. v. Eagle-Picher 39,123). Reg. (“[I]f (D.C.Cir.1985) ... F.2d agency must challenged, Montana’s the model is fails understand defense.”).5 It is analytical a full provide does not argument. PPL Montana first not, every chemical to which the ad hoc basis explanation does Requiring an applied, de novo re- when faced data argues, improperly model is even perfect of the CALPUFF model fit.” Chem. indicating view of EPA's use that it not a justify the model on an EPA “need itself. no answer to respond, did, as EPA improvement be “reasonably ... anticipat low levels visibility impairment must be ed” as a result of BART installation. 42 addressed though they even are not per- 7491(g)(2). As the D.C. Circuit ceptible to eye, the human or that meas- explained in vacating portion of the Re ures been have taken to minimize the gional itself, Haze Rule it is arbitrary and margin of error. The issue is not per- capricious for EPA to force an emissions ceptibility proposed improvements, source spend “to millions of dollars for new but the ability model’s to anticipate im- technology that will have no appreciable provements at a allegedly level within its effect on the haze’ in any Class I area.” margin error, whether perceptible or Am. Ass’n, Corn Growers F.3d 7. In not to eye. the human EPA simply of- response to PPL Montana’s contentions fered response no objection. to this that the Final Rule just that, would do EPA’s only detailed defense of the mod- *13 requiring PPL Montana to install SNCR el addressed PPL concern Montana’s at Colstrip and without sufficient about distance —not the margin of error assurance of any improvement all, at EPA issue. See 77 Fed.Reg. at has offered no explanation. reasoned 57,867-68. distance, With to regard noted guidance by issued the Inter- Ill agency Workgroup Air Quality on Model- PPL Montana also challenges ing “provides for the use of the CALPUFF the emissions limitations EPA imposed at model at receptor of distances toup 200 to namely, requirement the Corette — 300 km.” EPA explained then why it be- Corette lower its emissions levels even lieved CALPUFF could be used “cautious- without installing additional technology. ly” for distances in that range, even As explained, EPA’s determination that though puffs the be dispersed would more installation of additional to technology con at greater distances, and stated the trol emissions from Corette was not model cost- was not suitable for “very long- effective (300 suffers the range transport same failure expla km and beyond).” nation Id. at as BART its EPA thus offered a determinations Col- rea- response strip. soned to See supra PPL Montana’s chal- II.B. But the Rule also lenge to the use tightened of CALPUFF at emissions limits identified distances in question. That explanation for in the Corette Specifi Rule. not, however, did also suffice as a reasoned cally, in response to comments (including response regarding how CALPUFF could comments Montana), made upon relied predict to an improvement altered its rule to establish an emission of as little as 0.085 deciviews when PPL rate of 0.57 for lb/MMbtu emissions offered reasons to think that doing so was and an emission rate lb/MMBtu outside the capabilities. model’s emissions, both monitored aon 30- Nox
The predictability day concern important is rolling average. See Final because Act requires that any visibility Fed.Reg. 57,893-94. PPL Montana ar- EPA, Ass’n v. 28 F.3d Evans, inappropriate. tions
Mfrs. See Brower v. (D.C.Cir.1994). But PPL Montana does not (9th Cir.2001) (holding justify ask EPA to Accepting model. CAL- deference agency overcome "complete- where model, applicable PUFF ly failed address factor some consideration asks explains how CALPUFF EPA's selection of which was essential to [making in- an] BART, of an technology additional ques- a (internal formed quotation decision" marks tion EPA has not answered. In a situa- omitted)). and citation tion, deference even methodological ques- top- on this disagreement reasoned authorize not does CAA gues comment reflects PPL Montana’s limits without ic with impose EPA to ques- BART, highly EPA insists that scientific but on a determining its conclusion the BART calcula- comported in emissions analysis variance tion—the rates annualized Guidelines.6 when that occurs tions rolling aver- thirty-day into are translated that, it after argues correctly justi- question kind of ages precisely technol- already had BART found Corette — discretion. Re- to EPA’s by the deference fying authorized it was place, ogy remaining Army Corps “skip the v. U.S. Fed’n Rule Nat’l gional Haze Wildlife (9th section, including the visi- in this 1177-78 analyses F.3d Eng’rs, 384 Regu- step 5.” bility analysis Cir.2004). 39,165. PPL lations, was nev- that EPA IV contention
Montana’s re- with the proceed required ertheless not decision that EPA’s contends challenge to analysis is BART maining emission-reduc- any additional Haze Rule Regional provision alone installation technology, let ing in this chal- itself, asserted properly arbi- 3 and was SCR, Colstrip Units FIP. See U.S.C. lenge to fails because capricious trary rulemaking 7607(b)(1) (challenge to re- progress reasonable satisfy the Act’s days). sixty brought within must be *14 and 4 are Colstrip Units quirements. set emissions properly EPA also be- requirements BART subject to 30-day rolling a aver limits on Corette for after 1977. they were constructed cause prescribes Haze Rule Regional age. The 7491(b)(2), Neverthe- (g). § 42 U.S.C. See Regulations, 70 Fed. See such limits. issue EPA to less, directs the Act still to emissions limits 39,172 (requiring Reg. progress” assuring “reasonable regulations 30-day of a averaging time an “specify I in ar- visibility Class improving toward ad noted and EPA average”). rolling 7491(a)(4). § 42 U.S.C. See eas. about Montana’s comment PPL dressed to consider several requires EPA statute requirement. average rolling 30-day the prog- determining reasonable in factors 57,893-94. Rule, See Final (b) time (a) compliance; the costs of ress: converting that PPL contended Montana (c) energy the compliance; for necessary 30-day rolling to limits emissions current of com- impacts and other environmental the to raise require averages would (d) life remaining useful and the pliance; Rule. EPA Proposed the forth in limits set progress subject to reasonable of a source that, by contrary choos the to concluded 7491(g)(1). See U.S.C. requirements. monthly emission rates ing percentile 99th reasonable Regulations prescribe EPA’s margin for com adding an additional and visibility natural attaining toward progress averages 30-day. rolling pro the pliance, C.F.R. conditions comported Corette’s the rule vided in 51.308(f). id. capabilities. See technological current its, average would rolling limits 30-day to 6. EPA contends .first limits. emissions an failing increase point by objection this on its waived Moreover, this specifically addressed comment to a relevant make Rule, noting that a commen- in the argument, PPL comment Contrary Mon- Rule. in emissions suggested an increase had ter objection to the Corette preserved tana its comments, emissions for to 0.81 limits lb/MMBtu it made limits when emissions emissions. to 0.46 that, current tech- lb/MMBtu urged with Corette's which Fed.Reg. at lim- converting emissions annualized nology, responded to comments con setting Rule limits Col- tending visibility benefits strip Units and Corette and RE- gained from at Colstrip SCR 3 and MAND to EPA for further proceedings justify requirement reasonable consistent opinion. with this Each party progress purposes. See Final 77 shall bear its own costs.7 57,902-03. Fed.Reg. at explained It FOR PETITIONS REVIEW GRANT- the cost of compliance only one of the ED IN PART AND PART; DENIED IN statutory four requirements for reasonable VACATED AND REMANDED. progress analysis. See id. EPA also of fered a better explanation for its cost- BERZON, Circuit Judge, concurring. based point decision than it did for I concur. I write separately to under- cost-effectiveness determination my line that, understanding BART at Colstrip and Part Corette. It rea II.C of soned opinion, lead we are benefits to be not impugning gained from requiring (ranging SCR from EPA’s use of the gener- CALPUFF model deciviews) 0.273 deciviews 0.260 ally. Instead, were we are requiring a suffi- not sufficient considering cost, their be ciently response reasoned particular $4,574 $4,607 tween per ton. See id. comment regarding CALPUFF’s useful- EPA contrasted this effectiveness with the ness in these specific circumstances. Had reasonable progress goals implemented EPA given any reasonable explanation as in North Dakota at Valley Antelope Sta to why just CALPUFF was not “sufficient- tion—a location to which specifical ly reliable to inform the making decision ly urged comparison. See id. EPA ex process” generally, 57,864, 77 Fed.Reg. plained that it had found improvements of 57,868 18, 2012), (Sept. but specifically also deciviews each unit of that location suitable for predicting visibility improve- cost-justified to be for reasonable progress ment in pertinent factual context at purposes when the costs of improve *15 increments as deciviews, small as 0.085 ranged ment from $586/ton $661/ton. then we could no more. See id. comparison of the two The arbitrary-and-capricious standard of provided sources least some review where, authorizes remand inter broad metric for understanding which alia, agency the “entirely has and, failed to cost-per-ton ratios EPA approve will consider an aspect important prob the of which it will not—a explanation rational lem,” by “g[iving] no consideration whatev why the reasonable progress options avail er” to it. Motor Vehicle able at Ass’n Colstrip Units 3 and 4 were not Mfrs. of States, United feasible. Inc. v. State Farm Mut. Co., 29, Auto. 43, 46, Ins. 463 U.S. 103 V 2856, (1983) S.Ct. 77 L.Ed.2d (empha 443 added). For the sis explained, reasons we The “standard of GRANT review ... IN PART and DENY [requires] IN peti- only reasonableness, PART the not perfec review, tions for VACATE portions of tion.” Kennecott Greens Mining Creek 7. Petitioners National Parks Conservation As- Corette Steam Electric Station filed March sociation, 17, Montana Environmental Informa- 2015 is DENIED. The facts contained Center, tion and Sierra Club's Motion for therein generally are neither known within Judicial Standing Notice Declarations filed jurisdiction court's territorial nor accu- 14, March is DENIED as moot. rately readily determinable from sources Montana, Petitioner PPL Unopposed accuracy LLC’s whose reasonably ques- cannot be Motion to Take Judicial Regarding Notice tioned. See Fed.R.Evid. 201(b). J.E. Stations, at 513- U.S. Admin., Television 476 Fox & Health Safety v. Mine
Co.
Bowman
14,
(quoting
(D.C.Cir.2007);
J &
129 S.Ct.
see also
946, 954
F.3d
438)
286,
Truscott,
95 S.Ct.
419 U.S.
Transp.,
Ltd. v.
Sales
G
e
omitted).
Cir.2007).
(9th
th
For
(internal
marks
quotation
“[W]here
1051-52
factors
the relevant
to dis-
there is a basis
perhaps
‘considered
agency
example,
be
rational connection
visibility improve-
articulated a
that a
the assertion
pute
choice
and the
found
the facts
tween
within
fell
0.085 deciviews
ment of
”
made,’
to the “reasonable
must defer
we
it
perhaps
error. Or
margin of
model’s
Dan
Arrington v.
decision.
of that
basis”
predicted
fact that a
that the
significant
Cir.2008)
(9th
iels,
F.3d
margin
a model’s
falls within
improvement
Legal
Action
Ranchers Cattlemen
(quoting
im-
predicted
prove
not
of error does
F.3d
Agric., 415
Dep’t
v.
Fund
U.S.
occur,
it is
just
not
will
provement
Cir.2005)).
(9th
the Su
As
1078, 1099
there is
perhaps
to occur. Or
likely
less
reiterated,
frequently
Court has
preme
thought a 0.085
why
reason
another
of less
a decision
‘uphold
“should
courts
“reasonably
improvement' could
deeiview
may
agency’s path
if the
clarity
than ideal
even
from SNCR
anticipated” to result
be
”
Fox
FCC v.
reasonably be discerned.’
the model’s
fell within
if
increment
such
Stations,
Inc., 556 U.S.
Television
7491(g)(2).
error.
margin of
513-14,
173 L.Ed.2d
129 S.Ct.
put
meant
are
My examples
not
v.
(2009)
Transp., Inc.
(quoting Bowman
(if
can
agency
an
EPA’s mouth
words
Inc.,
U.S.
Freight Sys.,
Arkansas-Best
mouth).
But
said
have a
said to
438, 42
281, 286,
L.Ed.2d
95 S.Ct.
sort,
anything else
or
of either
nothing
(1974)).
com-
PPL CALPUFF
to the
responsive
concludes,
Here,
majority opinion
as the
Instead,
general
just
professed
ment.
received,
did
of the comments
light
model—
in the CALPUFF
confidence
explana-
reasoned
any adequate,
give
not
not
was
may well be warranted —but
which
in-
supported
why
tion as to
CALPUFF
ex-
concerns
particular
to the
responsive
1 and
SNCR
stallation
just
position
not in a
are
pressed. Courts
in its comment
argued
2. PPL Montana
statistically based
data-based or
provide
predicted
improvement
why
reason
which is one
explanations,
the er-
from
was “within
SNCR
deciviews
basis
supply a reasoned
“may not
Court
model,” and
therefore
range
ror
[expert]
action that
agency’s
for the
“reasonably
could
improvement
*16
Motor Vehi-
given.”
itself has not
agency
from
anticipated” to result
SNCR.
Ass’n,
103 S.Ct.
463 U.S.
cle Mfrs.
generic re-
7491(g)(2).
Chenery Corp.,
SEC v.
2856 (quoting
model can
CALPUFF
sponse,
“the
194, 196,
Action 1093). F.3d
Perhaps not, it did. But it has even mini-
mally, provided any indication did in
the comments and analysis accompanying For Rule. alone, that reason and not
because I have any reason question
whether the CALPUFF model is generally
fully adequate to the purposes for which it, I uses concur.
Terry D. BEMORE, Petitioner-
Appellant,
v.
Kevin CHAPPELL, Warden,
Respondent-Appellee.
No. 12-99005.
United States Court Appeals,
Ninth Circuit.
Argued and Sept. 10, Submitted 2014.
Filed June