New York v. U.S. Nuclear Regulatory Commission

`United States Court of Appeals
         FOR THE DISTRICT OF COLUMBIA CIRCUIT




Argued February 22, 2016             Decided June 3, 2016

                       No. 14-1210

               STATE OF NEW YORK, ET AL.,
                      PETITIONERS

                            v.

    U.S. NUCLEAR REGULATORY COMMISSION AND UNITED
                  STATES OF AMERICA,
                     RESPONDENTS

       COMMONWEALTH OF MASSACHUSETTS, ET AL.,
                  INTERVENORS



       Consolidated with 14-1212, 14-1216, 14-1217



            On Petitions for Review of an Order
    of the United States Nuclear Regulatory Commission



    Andrew W. Amend, Senior Assistant Solicitor General,
Office of the Attorney General for the State of New York,
argued the cause for petitioners State of New York, et al.
With him on the briefs were Eric T. Schneiderman, Attorney
                             2
General, John J. Sipos, Kathryn M. DeLuca, Laura E. Heslin,
Assistant Attorneys General, Barbara D. Underwood,
Solicitor General, Anisha S. Dasgupta, Deputy Solicitor
General, Monica Wagner, Deputy Bureau Chief, Maura
Healy, Attorney General, Office of the Attorney General for
the Commonwealth of Massachusetts, Seth Schofield,
Assistant Attorney General, Joseph F. Halloran, George
Jepsen, Attorney General, Office of the Attorney General for
the State of Connecticut, Robert D. Snook, Assistant Attorney
General, William H. Sorrell, Attorney General, Office of the
Attorney General for the State of Vermont, and Kyle H.
Landis-Marinello, Assistant Attorney General. Melissa A.
Hoffer, Assistant Attorney General, Office of the Attorney
General for the Commonwealth of Massachusetts, entered an
appearance.

    Kevin W. Bell was on the brief for amicus curiae The
California State Energy Resources Conservation and
Development Commission in support of petitioners State of
New York, et al.

     Geoffrey H. Fettus argued the cause for petitioners
Natural Resources Defense Council, Inc., et al. With him on
the briefs were Diane Curran and Mindy Goldstein.

     Wallace L. Taylor was on the brief for amicus curiae
Sierra Club in support of petitioners.

    Andrew P. Averbach, Solicitor, U.S. Nuclear Regulatory
Commission, argued the cause for federal respondents. With
him on the brief were John C. Cruden, Assistant Attorney
General, U.S. Department of Justice, John E. Arbab,
Attorney, Robert M. Rader, Senior Attorney, U.S. Nuclear
Regulatory Commission, and Michelle D. Albert, Attorney.
Charles E. Mullins, Senior Attorney, entered an appearance.
                              3

    David A. Repka argued the cause for intervenor-
respondents. With him on the brief were Ellen C. Ginsberg,
Jonathan M. Rund, Brad Fagg, Jay E. Silberg, and Kimberly
A. Harshaw.

   Before: KAVANAUGH, Circuit Judge, and EDWARDS and
SENTELLE, Senior Circuit Judges.

   Opinion for the Court filed by Senior Circuit Judge
SENTELLE.

     SENTELLE, Senior Circuit Judge: Several states, a Native
American community, and numerous environmental
organizations have filed petitions for review of a rule and
generic environmental impact statement promulgated by the
Nuclear Regulatory Commission (the “NRC”), concerning the
continued, and possibly indefinite, storage of spent fuel from
nuclear power plants in the United States. The petitioners
argue that the NRC fails to comply with its obligations under
the National Environmental Policy Act (“NEPA”), 42 U.S.C.
§ 4321 et seq. Specifically, the petitioners contend that the
NRC did not consider alternatives to and mitigation measures
for the continued storage of spent nuclear fuel, miscalculated
the impacts of continued storage, and relied on unreasonable
assumptions in its environmental impact statement. Because
we hold that the NRC did not engage in arbitrary or
capricious decision-making, we deny the petitions for review.

I.   BACKGROUND

    The United States has committed to the development of
nuclear energy, yet to-date it lacks a permanent solution for
one consequence of that commitment—the generation of
spent nuclear fuel, which “poses a dangerous, long-term
                               4
health and environmental risk.” New York v. NRC (New York
I), 681 F.3d 471, 474 (D.C. Cir. 2012). This case is not the
first, nor even the second, time that concerned parties have
petitioned this Court to address the spent-nuclear-waste
problem. See, e.g., Minnesota v. NRC, 602 F.2d 412, 413,
418-19 (D.C. Cir. 1979) (remanding the NRC’s decision to
expand “on-site capacity for the storage of spent nuclear fuel
assemblies” in light of “[t]he complex and vexing question of
the disposal of nuclear wastes”); New York I, 681 F.3d at 483
(vacating the NRC’s rule governing the temporary storage of
spent nuclear fuel); see also In re Aiken Cnty., 645 F.3d 428,
430 (D.C. Cir. 2011) (considering a challenge to the
Department of Energy’s attempt to withdraw its application
for a permanent repository for spent nuclear fuel); Ind. Mich.
Power Co. v. DOE, 88 F.3d 1272, 1277 (D.C. Cir. 1996)
(requiring the Department of Energy to fulfill its contractual
obligations to dispose of spent nuclear fuel generated by
operators of civilian nuclear power plants).

     In light of this extensive history, we provide only an
overview of the spent-nuclear-fuel issue. The so-called
“nuclear fuel cycle” consists of three primary phases. See
Blue Ribbon Commission on America’s Nuclear Future,
Report to the Secretary of Energy 9-11 (2012) [hereinafter
BRC Report]. First, “uranium is mined and processed into
fuel for use in a nuclear reactor.” Id. at 9. Second, nuclear
plants use the uranium fuel. Id. Third, spent fuel, even if
reprocessed, ultimately must be sent for disposal. Id. The
term “nuclear fuel cycle” is therefore somewhat of a
misnomer; “every foreseeable approach to the nuclear fuel
cycle still requires a means of disposal that assures the very
long-term isolation of radioactive wastes from the
environment.” Id. at 11. And “virtually all spent fuel[]
remain[s] radioactive for thousands of years . . . .” Id. at 14.
                                5
     Congress passed the Nuclear Waste Policy Act of 1982
for the purpose of “establish[ing] a schedule for the siting,
construction, and operation of repositories that will provide a
reasonable assurance that the public and the environment will
be adequately protected from the hazards posed by high-level
radioactive waste and . . . spent nuclear fuel . . . .” Pub. L.
No. 97-425, § 111(b)(1), 96 Stat. 2201, 2207 (codified at 42
U.S.C. § 10131(b)(1)). In 2008, after nearly two decades of
regulatory and political discord, the Department of Energy
sought construction authorization from the NRC to establish a
repository at Yucca Mountain in Nevada. See In re Aiken
Cnty., 645 F.3d at 431-32. But a change in the presidential
administration brought with it a shift in nuclear energy policy,
and in 2010 the Department of Energy withdrew its
application. Id. at 432. Our characterization in New York I of
the nation’s spent-fuel-storage policy still rings true today:
“[a]t this time, there is not even a prospective site for a
repository, let alone progress toward the actual construction
of one.” 681 F.3d at 474.

     Absent a permanent repository, the majority of spent
nuclear fuel remains stored on-site at reactors. BRC Report,
supra, at 14; see also New York I, 681 F.3d at 474. After
removal from a reactor, “spent fuel is transferred to a deep,
water-filled pool . . . for at least five years” in order to cool.
BRC Report, supra, at 11. Once the spent nuclear fuel has
“cooled sufficiently in wet storage [i.e., a pool], it may be
transferred to dry storage[,]” which “generally consist[s] of a
fuel storage grid placed within a steel inner container and a
concrete and steel outer container[,]” also known as a “dry
cask.” Id. “Most [spent nuclear fuel], however, will remain
in spent-fuel pools until a permanent disposal solution is
available.” New York I, 681 F.3d at 474.
                               6
     From 1984 until this Court’s decision in New York I, the
NRC relied on a “Waste Confidence Decision” in order to
assess the risk of on-site storage of spent nuclear fuel and the
likelihood that a permanent off-site storage solution will be
available. Id. at 474-75 (citing Minnesota v. NRC, 602 F.2d at
418). In New York I, we vacated the 2010 update to the
NRC’s Waste Confidence Decision and its Temporary
Storage Rule governing the storage of spent nuclear fuel. Id.
at 483. In support of the Waste Confidence Decision and the
Temporary Storage Rule, the NRC prepared an environmental
assessment (“EA”) with a finding of no significant impact.
Id. at 476. We held that the NRC’s analysis was deficient
because: (1) the Waste Confidence Decision “did not examine
the environmental effects of failing to establish a repository”;
(2) the NRC “failed to properly examine the risk of [pool]
leaks in a forward-looking fashion”; and (3) the NRC “failed
to examine the potential consequences of pool fires” in
addition to the probabilities that such fires might occur. Id. at
478-79.

     In response to our New York I decision, the NRC altered
its approach to the continued storage of spent nuclear fuel.
Instead of relying on an EA with a finding of no significant
impact, the NRC prepared a Generic Environmental Impact
Statement (“GEIS”) and proposed a Continued Storage Rule
(the “Rule”) to codify its analysis of the effects of continued
on-site storage of spent nuclear fuel. See 79 Fed. Reg. 56,238
(2014) (Continued Storage Rule); 79 Fed. Reg. 56,263 (2014)
(notice of GEIS); J.A. 263-1560 (GEIS). The stated purpose
of the Rule “is to preserve the efficiency of the NRC’s
licensing process by adopting into the NRC’s regulations the
Commission’s generic determinations of the environmental
impacts of the continued storage of spent nuclear
fuel . . . beyond the licensed life for operations of a
reactor . . . .” 79 Fed. Reg. at 56,239. The Rule incorporates
                               7
the findings of the GEIS into all future reactor licensing
proceedings and precludes reconsideration of those findings
absent a waiver under 10 C.F.R. § 2.335. See 10 C.F.R.
§ 51.23(b); 79 Fed. Reg. at 56,243.

     The petitioners in this case, a group of states and a Native
American community (collectively, the “States”) along with a
group of environmental organizations (collectively, the
“NRDC”), submitted comments to both the GEIS and the
Rule. The petitioners now challenge the Rule and the GEIS
on the basis that the NRC failed to comply with NEPA. Cf.
42 U.S.C. § 4332(C) (detailing NEPA’s requirements for an
environmental impact statement). They request that we
vacate the Rule and the GEIS and remand to the NRC for
further proceedings.

    Because we hold that the NRC did not engage in arbitrary
or capricious decision-making, see 5 U.S.C. § 706(2)(A), we
deny the petitions for review.

II. ANALYSIS

     The States and the NRDC raise a panoply of challenges
to the NRC’s Rule and the GEIS. First, the petitioners
contend that the Rule is a major federal action that requires
consideration of alternatives and mitigation measures to
reactor licensing. Second, they dispute the NRC’s assessment
of the environmental impacts of the continued storage of
spent nuclear fuel, asserting: (a) failure to employ
conservative     bounding     estimates;      (b)     inadequate
determination of the probability of failure to site a permanent
geologic repository; (c) insufficient assessment of the
cumulative impacts of the continued storage of spent nuclear
fuel; and (d) unjustified dismissal of the risks of short-term,
high-volume pool leaks. Relatedly, the petitioners challenge
                               8
as “illusory” the NRC’s process for granting a petition for
waiver of the Rule in site-specific licensing proceedings.
Finally, the petitioners characterize several of the NRC’s
underlying assumptions in the GEIS as unreasonable. We
hold that none of these arguments is persuasive and deny the
petitions.

       A. THE NRC APPROPRIATELY CHARACTERIZED ITS
          RULE AND CONSIDERED ALTERNATIVES AND
          MITIGATION MEASURES

     The parties disagree over the proper characterization of
the NRC’s Rule. According to the NRC, the Rule “codif[ies]
its generic determinations regarding the environmental
impacts of continued storage of spent fuel at-reactor, or away-
from-reactor sites beyond a reactor’s licensed life for
operation.” 79 Fed. Reg. at 56,241. The NRC contends that
“the Rule is not a licensing action . . . .” NRC’s Br. 16. The
States and the NRDC respond that the federal action at issue
is reactor licensing. See States’ Br. 44; NRDC’s Br. 20. And
because licensing is indisputably a “major Federal action[]”
under NEPA, 42 U.S.C. § 4332(C), the States and the NRDC
argue that the NRC was required to prepare a complete
environmental impact statement (“EIS”), including a
consideration of alternatives and mitigation measures for the
continued storage of spent fuel. See 42 U.S.C. § 4332(C)
(“[M]ajor Federal actions significantly affecting the quality of
the human environment” require an EIS or its equivalent.);
see also NRDC v. NRC, -- F.3d --, No. 14-1225, 2016 WL
1639661, at *1 (D.C. Cir. Apr. 26, 2016) (same). We agree
with the NRC and hold that, while the Rule is a “major
Federal action” under NEPA, the NRC complied with its
NEPA obligations by preparing the GEIS. Because the Rule
is not a licensing action, the NRC need not have considered
                                9
the alternatives to licensing in the GEIS. We therefore deny
the petitions for review on this issue.

     Under NEPA, an agency must consider both the
environmental impacts of a proposed action and alternatives
to that action. See 42 U.S.C. § 4332(C). Part of the
alternatives analysis includes review of measures available to
mitigate adverse effects. See 40 C.F.R. §§ 1508.25(b),
1502.14(f). “[W]e review both an agency’s definition of its
objectives and its selection of alternatives under the ‘rule of
reason.’ . . . That is, as long as the agency ‘look[s] hard at the
factors relevant to the definition of purpose,’ we generally
defer to the agency’s reasonable definition of objectives.”
Theodore Roosevelt Conservation P’ship v. Salazar
(Theodore Roosevelt II), 661 F.3d 66, 73 (D.C. Cir. 2011)
(quoting Citizens Against Burlington, Inc. v. Busey, 938 F.2d
190, 196 (D.C. Cir. 1991)) (alteration in original).
Furthermore, “NEPA does not require agencies to discuss any
particular mitigation plans that they might put in place, nor
does it require agencies—or third parties—to effect any.”
Theodore Roosevelt Conservation P’ship v. Salazar
(Theodore Roosevelt I), 616 F.3d 497, 503 (D.C. Cir. 2010)
(citation and internal quotation marks omitted).

     Our decision in New York I compels the result that the
NRC’s Rule is a major federal action requiring the
preparation of either an environmental assessment with a
finding of no significant impact or an environmental impact
statement. See 681 F.3d at 476. Like the NRC’s prior Waste
Confidence Decision and Temporary Storage Rule, the NRC’s
Rule in this case “ha[s] a preclusive effect in all future
licensing decisions . . . .” Id. But unlike in New York I, the
NRC has done exactly what NEPA requires for major federal
actions; it prepared an environmental impact statement. See
id.; 42 U.S.C. § 4332(C); 40 C.F.R. § 1502.3. So long as that
                              10
environmental impact statement complies with NEPA, and we
hold that it does, no more is required.

     The face of the NRC’s Rule also makes it clear that it is
not a licensing action. To the contrary, the Rule “codif[ies]
[the NRC’s] generic determinations regarding the
environmental impacts of continued storage of spent fuel at-
reactor, or away-from-reactor sites beyond a reactor’s
licensed life for operation.” 79 Fed. Reg. at 56,241. “[T]he
rule does not authorize the storage of spent fuel at any site
[and] . . . reflects only the generic environmental analysis for
the period of spent fuel storage beyond a reactor’s licensed
life for operation and before disposal in a repository.” Id. at
56,243. Because the GEIS is only an input for future site-
specific reactor licensing and does not itself impose
regulatory requirements on reactors, the NRC need not have
considered the alternative of ceasing licensing in the GEIS.
The NRC instead analyzes that alternative during site-specific
licensing proceedings. See J.A. 1040 (“The alternative of not
issuing or not renewing a nuclear power plant license is
considered during the site-specific review of an individual
license application.”). The NRC did consider alternatives for
the only action it took in the Rule—i.e., incorporating the
GEIS into future licensing proceedings. See J.A. 338-43.

     Furthermore, contrary to the petitioners’ claims, the GEIS
discusses mitigation measures for pool fires, J.A. 1240-41,
1284-85, and pool leaks, including short-term, high-volume
leaks, J.A. 838, 1394-96. It also evaluates measures such as
the expedited transfer of spent fuel to dry storage casks, J.A.
973-74, 1454-55, limiting the use of high-burnup fuel, J.A.
912-19, 1246, 1258, 1339, and implementing hardened on-site
storage, J.A. 1458. We find nothing in the GEIS to indicate
that the NRC went astray of NEPA’s rule of reason.
Regardless, because mitigation is equally relevant during the
                               11
life of a licensed reactor as it is during decommissioning, the
NRC can defer consideration of such measures to site-specific
review. See Public Utils. Comm’n of Cal. v. FERC, 900 F.2d
269, 282-83 (D.C. Cir. 1990) (“[T]he Commission’s deferral
of decision on specific mitigation steps until the start of
construction, when a more detailed right-of-way would be
known, was both eminently reasonable and embraced in the
procedures promulgated under NEPA.”). Regardless, “NEPA
does not require agencies to discuss any particular mitigation
plans that they might put in place.” Theodore Roosevelt I,
616 F.3d at 503 (citation and internal quotation marks
omitted).

     Our holding with respect to this issue is consistent with
our decision in New York I. In that case, we held that the
NRC’s prior Waste Confidence Decision was “a major federal
action requiring either a [finding of no significant impact] or
an EIS.” 681 F.3d at 476. Although we described the Waste
Confidence Decision as “a pre-determined ‘stage’ of each
licensing decision,” id., nowhere did we conclude that the
NRC undertook licensing with its waste confidence
rulemaking. The Rule in this case is likewise a major federal
action because it has a preclusive effect on future licensing
proceedings. See 10 C.F.R. § 51.23(b). But the proposition
that all licensing actions are major federal actions does not
imply its converse. When the NRC does make a licensing
decision in partial reliance on the GEIS, it must at that time
ensure that it has fully complied with NEPA. See 42 U.S.C.
§ 4332(C); cf. 40 C.F.R. § 1502.14 (delineating the
requirements for including alternatives in the EIS); Ctr. for
Sustainable Economy v. Jewell, 779 F.3d 588, 599-600 (D.C.
Cir. 2015) (noting that the obligation to comply with NEPA
“do[es] not mature until . . . there [has] been an irreversible
and irretrievable commitment of resources” by the agency
(citation and internal quotation marks omitted) (first alteration
                               12
in original)). The NRC acknowledges as much. See Oral
Arg. Rec. 40:43-41:40 (statements by the NRC that the
decision whether to issue a license is site-specific and that the
agency will consider mitigation measures and alternatives at
that time). At this stage, we take the NRC at its word. But
should the agency fail to consider a necessary aspect of the
problem during site-specific proceedings, the parties might be
able to challenge the final licensing decision. See, e.g.,
Massachusetts v. NRC, 924 F.2d 311, 315 (D.C. Cir. 1991)
(adjudicating consolidated petitions for review of “the
[NRC’s] licensing of Seabrook Nuclear Power Station”); York
Comm. for a Safe Env’t v. NRC, 527 F.2d 812, 813 (D.C. Cir.
1975) (considering a challenge to “a final decision . . . to
grant a license . . . for operation of a light-water-cooled
nuclear reactor to be used for generating electricity”).

    We therefore deny the petitions for review on this issue.

       B. THE GEIS SUFFICIENTLY ANALYZES THE
          IMPACTS OF CONTINUED STORAGE OF SPENT
          NUCLEAR FUEL

               1. The GEIS Thoroughly Considers
                  Essentially Common Risks to Reactor
                  Sites
     The States argue that the NRC could not generically
analyze the impacts of the continued storage of spent nuclear
fuel because it failed to employ “conservative bounding
assumptions” in the GEIS, particularly with regard to
estimating the risks of pool fires and pool leaks. Specifically,
the States contend that the NRC based its environmental
impact determinations on data from two reactor sites—one in
Surry, Virginia, and another near Lake Michigan. According
to the States, neither plant captures the full range of risks
across the country because the population density near the
                              13
Surry plant is 300 people per square mile, and the density near
the Lake Michigan plant is 860 people per square mile. See
J.A. 862-63, 868, 870. Because the GEIS ignores population-
wide effects and the impacts at atypical sites, the States posit
that the NRC must consider these impacts on a site-specific
basis.

     We noted in New York I that “[b]oth the Supreme Court
and this court have endorsed the [NRC’s] longstanding
practice of considering environmental issues through general
rulemaking in appropriate circumstances.” 681 F.3d at 480.
We also stated that “we see no reason that a comprehensive
general analysis would be insufficient to examine on-site risks
that are essentially common to all plants.” Id. Furthermore,
“whether the analysis is generic or site-by-site, it must be
thorough and comprehensive,” id. at 481, and we are “most
deferential” to the “NRC’s technical judgments and
predictions . . . [,]” Blue Ridge Env’tl Def. League v. NRC,
716 F.3d 183, 195 (D.C. Cir. 2013) (citation and internal
quotation marks omitted). While we acknowledged in New
York I that a generic analysis of impacts is “particularly”
appropriate when the NRC utilizes “conservative bounding
assumptions and the opportunity for concerned parties to raise
site-specific differences at the time of a specific site’s
licensing,” we did not make those factors essential. 681 F.3d
at 480. Instead, the cornerstone of our holding was that the
NRC may generically analyze risks that are “essentially
common” to all plants so long as that analysis is “thorough
and comprehensive.”

     In this case, we are convinced that the NRC has met that
standard. True, the NRC’s analysis is not “bounding” in a
strict sense. For example, in assessing the risks of pool fires,
the GEIS relies on seismic data that covers “about 70 percent”
of reactor sites. J.A. 870. This data therefore does not
                              14
“bound” the environmental impacts of spent fuel storage but
instead approximates the variance in harms. For pool leaks,
the NRC provides a high-level analysis of spent fuel
discharges but neglects any estimate of the expected errors for
its input variables, instead averring to specific “low” values
for these parameters. See J.A. 849. Furthermore, the GEIS
attempts to justify its reliance on data from the Surry and
Lake Michigan plants by noting that the average risks to
individuals are independent of population density. See J.A.
868. However, the NRC admits that this data covers only
“the 90th percentile population density” and that “the accident
consequences could be greater at higher population sites.”
J.A. 868; see also J.A. 1367 (conceding that values in the
GEIS “do not represent worst-case values”).

     Nonetheless, according deference to the NRC’s technical
decision-making, see Blue Ridge, 716 F.3d at 195, we find
nothing in the GEIS to undermine the NRC’s conclusion that
the identified risks are “essentially common” to all reactor
sites. The GEIS incorporates research demonstrating how the
risk analysis for pool fires is conservative, see J.A. 1348,
1366-67, and analyzes the variance in seismic risks, see J.A.
870.      The NRC also considers “typical hydrologic
characteristics at nuclear power plant sites” when assessing
the impacts of pool leaks. J.A. 1054. Furthermore, the GEIS
“explain[s] qualitatively the factors that may cause the risk to
be lower or higher than” at the Surry and Lake Michigan
plants. J.A. 1367. Regardless, the NRC need not provide a
perfect analysis, only one that is “thorough and
comprehensive . . . .” New York I, 681 F.3d at 481. We hold
that the GEIS meets this requirement.

    The States rely on Limerick Ecology Action, Inc. v. NRC,
869 F.2d 719, 738 (3d Cir. 1989), for the proposition that the
NRC cannot generically analyze the site-specific
                             15
consequences of reactor accidents, and hence, we are told,
also the impacts of continued storage of spent nuclear fuel.
However, not only is Limerick non-binding on this Court, but
we recognized in NRDC v. NRC that the Third Circuit’s dicta
in Limerick “did not foreclose the possibility that [reactor
accident mitigation alternatives] could be dealt with
‘generically’ through a subsequent rulemaking.” 2016 WL
1639661, at *2; see also id. at *2 n.2.

     Accordingly, we deny the petitions for review on this
issue.

              2. The NRC Evaluated the Probability of
                 Failure To Site a Repository
     The NRDC argues that the NRC fails to quantify the
probability of failure to site a repository. Because we hold
that the NRC adequately considered both the probability and
consequences of failure to site a permanent repository for
spent nuclear fuel, we deny the petitions on this issue.

     Under its regulations, the NRC need only quantify “the
various factors” in the GEIS “to the fullest extent
practicable . . . .” 10 C.F.R. § 51.71(d). However, “[t]o the
extent that there are important qualitative considerations or
factors that cannot be quantified, these considerations or
factors will be discussed in qualitative terms.” Id. The NRC
complied with these obligations. The agency provided a
qualitative analysis of the likelihood of failure to site a
repository, see J.A. 290, 770, and considered the reasonably
foreseeable impacts of that scenario, see J.A. 458, 461, 469-
70, 472-73, 476, 480, 487, 496, 501, 509, 511, 517, 521, 523-
24, 550, 570, 572, 577, 580, 583, 585, 587-89, 591, 593, 596,
602-03, 605, 607, 610-11, 616, 618, 621. The NRDC
provides no indication of how the NRC can or should
otherwise assess the risk of failure to site a repository. Nor
                             16
does our decision in New York I require the NRC to do so. Cf.
681 F.3d at 478-80 (noting only that “an agency must look at
both the probabilities of potentially harmful events and the
consequences if those events come to pass”). The NRC’s
analysis was therefore sufficient to comply with NEPA.

              3. The GEIS Assesses the Cumulative
                 Impacts of the Continued Storage of
                 Spent Nuclear Fuel
     The NRDC argues that the GEIS fails to discuss the
cumulative impacts of continued storage of spent nuclear fuel
“when added to other past, present, and reasonably
foreseeable future actions . . . .” 40 C.F.R. § 1508.7. We
disagree.

     While it is true that NEPA requires an agency to consider
“cumulative or synergistic environmental impact[s]” of
related, concurrently pending proposals, Kleppe v. Sierra
Club, 427 U.S. 390, 410 (1976), “the purpose of the
cumulative impact requirement is to prevent agencies from
dividing one project into multiple individual actions each of
which has an insignificant environmental impact, but which
collectively have a substantial impact,” Theodore Roosevelt I,
616 F.3d at 514 (citation and internal quotation marks
omitted). In this case, there are no concurrently pending
proposals before the NRC because the NRC is not licensing
any reactors. Instead, the NRC has codified the GEIS for use
in future licensing proceedings. The GEIS also includes a
detailed discussion of the cumulative impacts of continued
storage of spent fuel over the lifetime of a licensed reactor.
See J.A. 628-93. Pursuant to its “tiered” approach to
assessing environmental impacts, see 40 C.F.R. § 1502.20, the
NRC also considers the environmental impacts of waste
disposal through 10 C.F.R. § 51.51, Table S-3, prior to any
licensing action. See also J.A. 351, 1297. Because there is no
                              17
indication that the NRC has improperly segmented its
environmental impact analysis, we deny the petitions on this
issue.

               4. The NRC Did Not Ignore Short-Term,
                  High-Volume Leaks
    The States argue that the NRC unreasonably “assumed”
that short-term, high-volume pool leaks have no
environmental consequences. While styled as a challenge to
the NRC’s assumptions in the GEIS, the crux of the dispute is
with the NRC’s assessment of the probability and
consequences of short-term, high-volume leaks. Because we
hold that the NRC adequately considered the risks of short-
term, high-volume leaks, we deny the petitions.

     The GEIS extensively analyzes the impacts of short-term,
high-volume leaks in addition to historic data on spent fuel
leakage. See J.A. 839-55. In particular, the NRC notes that
“[s]pent fuel pool leaks, while unpredictable, seldom occur.”
J.A. 839. Furthermore, NRC regulations require plant
licensees to monitor reactor sites, thereby increasing the
likelihood of high-volume leak detection. See, e.g., 10 C.F.R.
§§ 20.1501, 50.65; see also J.A. 836-37, 840, 1397-98. We
therefore find nothing in the record to suggest that the NRC
arbitrarily or capriciously disregarded the risks of short-term,
high-volume leaks.

               5. The NRC’s Waiver Process Ensures
                  Consideration of Site-Specific Impacts
    Finally, we note that the NRC’s regulations already
provide a means by which the petitioners can raise site-
specific    challenges    during    licensing      proceedings.
Specifically, under 10 C.F.R. § 2.335(b), “[a] participant to an
adjudicatory proceeding [before the NRC] . . . may petition
                               18
that the application of a specified Commission rule or
regulation or any provision thereof . . . be waived or an
exception be made for the particular proceeding.” The
standard by which the NRC will grant such a petition “is that
special circumstances with respect to the subject matter of the
particular proceeding are such that the application of the rule
or regulation (or a provision of it) would not serve the
purposes for which the rule or regulation was adopted.” Id.
We hold that the NRC’s waiver provision provides an
adequate mechanism by which the petitioners can challenge
the GEIS in site-specific proceedings.

     The petitioners raise two objections to the NRC’s waiver
provision. First, they argue that the waiver provision shifts
the burden of NEPA compliance from the NRC to the party
requesting waiver. Second, the petitioners characterize the
waiver process as “illusory.” States’ Br. 34. Neither
argument is persuasive. First, for the reasons stated above,
see supra Part II.B.1-4, the GEIS fulfills the NRC’s NEPA
obligation to analyze the impacts of the continued storage of
spent nuclear fuel. The NRC, in the GEIS, has therefore
presented sufficient evidence to carry its burden of persuasion
under NEPA that the impacts of continued storage of spent
nuclear fuel are generic to all licensed reactors. The burden
of production therefore necessarily shifts to the parties raising
objections to provide substantial evidence demonstrating that
the GEIS neglects those site-specific considerations, thereby
obstructing the GEIS’s purpose “to preserve the efficiency of
the NRC’s licensing process . . . .” 79 Fed. Reg. at 56,239.
Of course, the NRC always retains the burden of persuasion
under NEPA to consider fully the environmental impacts and
alternatives for its proposed action. See 42 U.S.C. § 4332(C);
40 C.F.R. § 1502.1.
                              19
     Second, the NRC conceded during oral argument that we
have jurisdiction to review its decision to deny a waiver
petition under 10 C.F.R. § 2.335(b). See Oral Arg. Rec.
48:11-:40; see also NRDC v. NRC, 2016 WL 1639661, at *12
(considering whether the NRC properly denied a waiver
petition); cf. Massachusetts v. NRC, 708 F.3d 63, 74 & n.17
(1st Cir. 2013) (same). Although we have stated that the
NRC’s decision whether to grant a waiver petition “is entitled
to deference,” that deference extends only so far as the NRC’s
decision is not arbitrary or capricious. NRDC v. NRC, 2016
WL 1639661, at *12. Therefore, we expect that the NRC will
give due consideration to waiver petitions raising non-
frivolous site-specific challenges to reactor licensing. Cf. 79
Fed. Reg. at 56,242 (stating that “concerned parties who meet
the waiver criteria in 10 C.F.R. § 2.335 will be able to raise
site-specific issues related to continued storage at the time of
a specific license application” (emphasis added)).
Furthermore, the petitioners retain the ability to petition the
NRC for a rulemaking to amend the GEIS. Cf. NRDC v.
NRC, 2016 WL 1639661, at *5, *12. “Although rulemaking
is far from the fastest route, it has transparency, extensive
public input, and broad application to recommend it.” Id.
at *12. We believe these protections are sufficient to prevent
the NRC’s waiver process from becoming “illusory.”

    Accordingly, we deny the petitions for review.

       C. THE NRC’S ASSUMPTIONS ARE NOT ARBITRARY
          OR CAPRICIOUS

     The States and the NRDC contend that the NRC utilized
several unreasonable assumptions, including: (1) that spent
nuclear fuel will be removed from spent-fuel pools within
sixty years of reactor decommissioning; (2) that after the
sixty-year period, spent fuel will be stored in dry casks that
                             20
are replaced every one hundred years; and (3) that
institutional controls over spent nuclear fuel will exist into
perpetuity. We hold that none of these assumptions is so
unreasonable as to render the NRC’s decision-making
arbitrary or capricious. We therefore deny the petitions for
review on this issue.

     An agency does not engage in arbitrary or capricious
decision-making by making “predictive judgment[s]” or even
by relying on “[i]ncomplete data.” New York v. EPA, 413
F.3d 3, 31 (D.C. Cir. 2005). To the contrary, such judgments
are “entitled to deference,” id., and a challenge to the
agency’s assumptions must be more than “an effort by [a
petitioner] to substitute its own analysis” for the agency’s,
Transmission Access Policy Study Grp. v. FERC, 225 F.3d
667, 737 (D.C. Cir. 2000). In this case, the NRC’s
assumptions in the GEIS are ably supported by the record.

     First, NRC regulations already mandate removal of spent
nuclear fuel within sixty years of the expiration of a reactor
license. See 10 C.F.R. § 50.82(a)(3). Furthermore, as the
NRC noted in its responses to comments, “(1) there is no need
to cool spent fuel in a pool for more than 60 years after a
reactor stops operating; (2) operational costs associated with
pool storage exceed dry cask storage costs; and (3) experience
with decommissioning of nuclear power plants indicates that
spent fuel pools are decommissioned before the end of the 60-
year period.” J.A. 1093. According deference to the NRC’s
predictive judgments, we hold that the agency’s assumption
regarding the timeframe for the removal of spent nuclear fuel
is reasonable.

    Second, the NRC’s assumption about the timeframe for
dry cask storage and replacement is conservative. The NRC
concluded that “the 100-year replacement period provides a
                                21
reasonable timeframe for the routine replacement of dry
storage systems, and that actual storage facility replacement
will be needed less frequently than assumed in the GEIS.” Id.
The agency also noted the “low degradation rates for dry cask
storage systems.” J.A. 1056. Furthermore, the NRC analyzed
the costs of dry cask replacement. See J.A. 397-98. This
assumption in the GEIS is therefore reasonable.

     Third, the record demonstrates that assuming the
continuation of institutional controls is both reasonable and
necessary. The NRC acknowledged that the impacts of a
failure in institutional controls would be “catastrophic.” J.A.
794, 798-99. Despite that conclusion, the agency also found
that the probability of institutional controls failing is
“remote.” J.A. 794; see also J.A. 796 (noting that it is
unlikely that the government would abandon continued
storage facilities and that those facilities are “highly visible”).
Furthermore, this assumption facilitates the assessment of
foreseeable environmental impacts from the continued storage
of spent nuclear fuel. See J.A. 794-95; see also J.A. 1094-
1100.

       We therefore deny the petitions for review on this issue.

III.       CONCLUSION

     We acknowledge the political discord surrounding our
nation’s evolving nuclear energy policy. But the role of
Article III courts in this debate is circumscribed. “The scope
of review under the ‘arbitrary and capricious’ standard is
narrow and a court is not to substitute its judgment for that of
the agency.” Motor Vehicle Mfrs. Ass’n v. State Farm Mut.
Auto. Ins., 463 U.S. 29, 43 (1983). To the extent that the
petitioners disagree with the NRC’s current policy for the
                             22
continued storage of spent nuclear fuel, their concerns should
be directed to Congress.

     For the reasons stated herein, the Court denies the
petitions for review.

                                                  So ordered.