FOR PUBLICATION
UNITED STATES COURT OF APPEALS
FOR THE NINTH CIRCUIT
TRI-VALLEY CARES; MARYLIA
KELLEY; JANIS KATE TURNER,
Plaintiffs-Appellants,
No. 10-17636
v.
U.S. DEPARTMENT OF ENERGY; D.C. No.
4:08-cv-01372-SBA
NATIONAL NUCLEAR SECURITY
OPINION
ADMINISTRATION; LAWRENCE
LIVERMORE NATIONAL LABORATORY,
Defendants-Appellees.
Appeal from the United States District Court
for the Northern District of California
Saundra B. Armstrong, District Judge, Presiding
Argued and Submitted
January 11, 2012—San Francisco, California
Filed February 7, 2012
Before: John T. Noonan, Jr., and Milan D. Smith, Jr.,
Circuit Judges, and Jed S. Rakoff, Senior District Judge.*
Opinion by Judge Milan D. Smith, Jr.
*The Honorable Jed S. Rakoff, Senior District Judge for the U.S. Dis-
trict Court for Southern New York, sitting by designation.
1181
1184 TRI-VALLEY CARES v. U.S. DOE
COUNSEL
Scott Yundt (argued), Livermore, California, and Steven Sug-
arman, Cerrillos, New Mexico, for the plaintiff-appellant.
TRI-VALLEY CARES v. U.S. DOE 1185
Barclay T. Samford (argued), Ignacia S. Moreno, and Jennifer
Scheller Neumann, United States Department of Justice,
Environment and Natural Resources Division, Denver, Colo-
rado, for the defendants-appellees.
OPINION
M. SMITH, Circuit Judge:
This case arises out of Plaintiffs-Appellants Tri-Valley
CAREs’, Marylia Turner’s, and Janis Kate Turner’s (collec-
tively, Tri-Valley CAREs) second challenge to the sufficiency
of the United States Department of Energy’s (DOE) Environ-
mental Assessment (EA) of a prospective “biosafety level-3”
(BSL-3) facility at the Lawrence Livermore National Labora-
tory (LLNL). In an earlier round of litigation, we upheld all
aspects of the DOE’s original EA, except for its failure to
consider the impact of a possible terrorist attack. Following
our remand, on September 30, 2009, the district court entered
summary judgment in the DOE’s favor on the grounds that it
had sufficiently revised its Final Revised Environmental
Assessment (FREA) to adequately consider the environmental
impact of an intentional terrorist attack on the BSL-3 facility
at LLNL. On November 18, 2010, Tri-Valley CAREs timely
appealed the district court’s decision, petitioning us to require
the DOE to prepare an Environmental Impact Statement
(EIS), or, in the alternative, to revise its EA, in light of the
allegations set forth in its original complaint, to determine
whether an EIS is required.
We hold that the DOE took the requisite “hard look” at the
environmental impact of an intentional terrorist attack in the
manner required by the National Environmental Policy Act
(NEPA) and San Luis Obispo Mothers for Peace v. Nuclear
Regulatory Commission, 635 F.3d 1109 (9th Cir. 2011). We
further hold that the district court did not abuse its discretion
1186 TRI-VALLEY CARES v. U.S. DOE
in denying Tri-Valley CAREs’ motion to supplement the
record. Accordingly, we affirm.
I. FACTUAL AND PROCEDURAL BACKGROUND
A. The LLNL Biosafety Level-Three Facility
On December 16, 2002, the National Nuclear Security
Administration (NNSA), an agency within the DOE, autho-
rized the construction of a BSL-3 laboratory at LLNL. Center
for Disease Control (CDC) and National Institutes of Health
(NIH) guidelines divide biosafety lab operations into four
levels: BSL-1 (least hazardous) through BSL-4 (most hazard-
ous). BSL-3 laboratories work with agents that may cause dis-
eases in humans with serious or lethal consequences if
untreated, and which have the potential of airborne transmis-
sion. Common agents found in BSL-3 facilities include West
Nile virus, yellow fever virus, Mycobaterium tuberculosis,
and SARS virus. There are more than 1,350 BSL-3 laborato-
ries in the United States. Common examples of BSL-3 facili-
ties include hospital surgical suites, laboratories associated
with medical schools, and university research laboratories. At
the time of construction, the LLNL BSL-3 facility was the
only BSL-3 facility operating in the same facility as a nuclear
laboratory.
The DOE decided to undertake the construction of an on-
site BSL-3 facility at LLNL because limitations in its BSL-1
and 2 laboratories forced LLNL to conduct its BSL-3 research
off-site. This off-site research was difficult and costly because
LLNL lacked physical control, and shipping and handling
increased the risk of cross-contamination and degradation.
For the new BSL-3 facility, LLNL selected a 1,500 square-
foot, prefabricated building to be constructed next to existing
BSL-2 facilities. The air-handling system comprised a double
High Efficiency Particulate Air-Purifying (HEPA) air filtra-
tion system, as is consistent with CDC guidelines. Each
TRI-VALLEY CARES v. U.S. DOE 1187
HEPA filter removes at least 99.97 percent of bioagents. The
facility also has additional safeguards against pathogenic
breach, including a ventilation system that would draw the
contaminated air back into the facility, a backup power sys-
tem that would enable employees to shut down portions of the
contaminated part of the facility, and a “zone-tight” system
that would prevent any air flow in the facility in the event of
a total power loss.
B. The 2002 Original Environmental Assessment
The DOE performed an EA for the proposed LLNL BSL-3
laboratory, pursuant to NEPA. The EA considered the envi-
ronmental impacts of the BSL-3 laboratory on a wide range
of issues, including human health, ecological resources, trans-
portation, waste management, geology, soils and seismology,
noise, and air quality. The EA also discussed how CDC and
NIH guidelines govern the facility’s operations and mitigate
the risk of infection and accidental release.
In evaluating the public risk potentially caused by the
BSL-3 facility, the DOE relied upon three major sources of
data: (1) statistics from hundreds of other CDC-registered
BSL-3 laboratories; (2) the U.S. Army’s Biological Defense
Research Program (BDRP) laboratories; and (3) LLNL’s
BSL-1 and -2 laboratories. In addition to examining the nor-
mal operations of the aforementioned sources, the EA ana-
lyzed potential abnormal impacts on those sources, using a
“catastrophic release” scenario, modeled upon a “Maximum
Credible Event” (MCE), simulating the outer bounds of
impact caused by a pathogen’s accidental release.
The DOE considered numerous possible methods of assess-
ing the threat of release, but it chose a catastrophic release
simulation (a centrifuge analysis), that the Army used to per-
form a NEPA analysis of its own biological research labs. The
catastrophic release model used by the Army also was an
MCE type of analysis, which simulated a reasonably foresee-
1188 TRI-VALLEY CARES v. U.S. DOE
able event with a low likelihood of occurrence, but with high
risk. In the Army’s simulated catastrophic release model, a
liter of coxiella burnetii (C. burnetii)1 was hypothetically
divided into six centrifuge tubes with loose caps and loose O-
rings. When the centrifuge was activated, some of the tubes’
contents would be aerosolized, resulting in the production of
almost 10 billion airborne pathogens.
The Army then modeled a plume of the airborne pathogens
as it moved through the lab and outside via the ventilation
system. In order to produce conservative results, the Army
simulated only one HEPA filter, operating at only 95 percent
effectiveness. The Army concluded the chance of public
exposure to an airborne pathogen, at a 50 percent rate of con-
tracting the disease, was extremely remote.
Using the Army centrifuge model, the DOE concluded that
the chances of exposure at the LLNL BSL-3 lab were even
more remote than those modeled by the Army because the
Army scenario assumed one HEPA filter that was 95 percent
effective. The LLNL BSL-3 lab, the DOE reasoned, filters all
room air through two HEPA filter banks, each of which is at
least 99.97 percent effective. The Army scenario also
assumed a lab in close physical proximity to the public,
whereas the LLNL BSL-3 lab is one-half mile from the near-
est public area. Finally, the Army assumed lower wind speeds
than are prevalent at LLNL; higher wind speeds would
decrease airborne concentrations more quickly. Based on this
analysis, the DOE concluded that even if a catastrophic
release were to occur, there would be no significant impact on
public health or safety. This conclusion thus led the DOE to
issue a Finding of No Significant Impact (FONSI).
1
Coxiella burnetii is the extremely resistant bacterial pathogen com-
monly found at BSL-3 labs that causes Q fever, a disease with both acute
symptoms (high fever, headache, fatigue, myalgia, night sweats and/or
chills, nausea, vomiting, chest pain) and chronic symptoms (pneumonia,
granulomatous hepatitis, myocarditis, and central nervous system disease).
TRI-VALLEY CARES v. U.S. DOE 1189
C. The First Round of Litigation
In August 2003, Tri-Valley CAREs brought suit in the
United States District Court for the Northern District of Cali-
fornia, challenging the EA on numerous grounds under
NEPA. Tri-Valley Cares v. U.S. Department of Energy, No.
C 03-3926-SBA, 2004 WL 2043034 (N.D. Cal. Sep 10,
2004). On September 10, 2004, the district court granted sum-
mary judgment for the DOE on the ground that the DOE had
satisfied the requirements of NEPA in preparing the original
EA. Id., at 2004 WL 2043034 at *1. On appeal, we affirmed
all aspects of the EA, except for the DOE’s failure to consider
the environmental impact of a terrorist attack. Tri-Valley
Cares v. Dep’t of Energy, 203 F. App’x 105, 107 (9th Cir.
2006). We thus affirmed in part and reversed in part, remand-
ing for the sole purpose of the DOE’s considering whether the
threat of a terrorist act required the preparation of an EIS. Id.
at 106-07.
D. The Draft Revised Environmental Assessment
(DREA)
On remand, the DOE prepared a DREA in March 2007,
addressing the impacts associated with terrorist attacks to
determine whether the threat of terrorist attack necessitated
preparation of an EIS. The DREA was circulated for public
comment from April 11, 2007 through May 11, 2007.
To analyze the threat that terrorist activity posed to the
LLNL BSL-3 laboratory, the DOE was required to take a dif-
ferent approach than it did when analyzing the threat posed by
accidents. Because there are an infinite number of possible
modes of attack, the DOE considered three general types of
terrorist threats: (1) a direct terrorist attack at the LLNL
BSL-3 facility, resulting in loss of containment; (2) the theft
and release of pathogenic material by an LLNL terrorist out-
sider; and (3) the theft and release of pathogenic material by
an LLNL terrorist insider.
1190 TRI-VALLEY CARES v. U.S. DOE
1. Direct Attack Resulting in Loss of Containment
The DOE considered various possible modes of direct ter-
rorist attack on the LLNL facility, including a suicidal plane
crash or an explosive device delivered by vehicle or on foot.
In considering the impact of this type of threat, the DOE used
the centrifuge scenario to determine the bounded or outer lim-
its of any release, and then analyzed factors which could miti-
gate such a release.
The DOE concluded that the outer bounds of dispersion in
a terrorist attack would be the same as those of the cata-
strophic release scenario used in the original EA. The centri-
fuge model analysis supported a finding of no significance for
terrorist threat based upon a direct attack on the LLNL BSL-3
facility.
To further validate its findings using the centrifuge model,2
the DOE also considered several factors that would further
limit the consequences of a direct terrorist attack. Specifically,
(1) very limited quantities of biological agents are generally
in use at LLNL BSL-3, (2) a fire resulting from an airplane
crash or explosive device would likely kill many pathogens
quickly; and (3) in the unlikely event of bioagent release,
microorganisms would generally be rendered innocuous by
exposure to generally occurring environmental conditions.
The DOE concluded that these factors in the aggregate would
substantially reduce the number of pathogens released and
capable of human infection, as the result of a direct attack. It
also concluded that the impact of a facility breach caused by
a direct terrorist act would be no greater than the impacts
2
The DOE’s conclusions in the DREA regarding the lack of any signifi-
cant impact on public health, took into account the fact that all the bioa-
gents used in the BSL-3 facility cause diseases for which treatment or
inoculation is available. Further, LLNL briefed local health care providers,
so if necessary, the consequences of any release could be mitigated by
inoculation and treatment of exposed individuals.
TRI-VALLEY CARES v. U.S. DOE 1191
addressed in the Army’s catastrophic release scenario ana-
lyzed in the initial EA.
2. Theft and Release by an LLNL BSL-3 Terrorist Out-
sider
The DREA next considered the threat of the theft and
release of a pathogen by a LLNL BSL-3 terrorist outsider. It
examined this possible form of attack by evaluating the types
of pathogens available at the LLNL BSL-3 facility in compar-
ison to the types of pathogens available from other sources,
including other BSL-3 facilities and the natural environment.
Comparatively speaking, the DOE reasoned that hundreds of
other BSL-3 facilities in the United States regularly handle
and store the same types of substances as the LLNL BSL-3
facility. Moreover, many substances are available from com-
mon environmental sources. Thus, a terrorist outsider seeking
such materials could find them in many BSL-3 facilities
nationwide, most of which lack the safeguards and security
infrastructure of LLNL.
Unlike the majority of the 1,350 BSL-3 labs nationwide,
most of which are academic or clinical facilities, LLNL
BSL-3 is surrounded by a patrolled security fence with badge-
identification required for entry; it has its own security force,
including an armed emergency response force. Access to the
BSL-3 facility is limited to employees registered with the
CDC and trained and qualified under its guidelines. Access to
individual lab rooms within the building is limited to staff
members approved to work during specific shifts, and all lab
rooms are equipped with motion sensors. Finally, within the
lab, select agents are stored in locked freezers when not in
use. Accordingly, the DOE reasoned that when considered
against the thousands of other sources from which a terrorist
outsider could obtain the same pathogenic materials, the addi-
tion of a single, highly-guarded BSL-3 facility at LLNL did
not significantly alter the status quo.
1192 TRI-VALLEY CARES v. U.S. DOE
3. Theft and Release by an LLNL BSL-3 Terrorist
Insider
Lastly, the DOE considered the potential impact of the theft
and release of a pathogen by an LLNL BSL-3 terrorist insider.
To evaluate this threat, the DOE did not use an empirical
model, but rather, engaged in a two-step probabilistic analy-
sis: (1) assessing the probability that an insider with access to
BSL-3 pathogens would have the motive to commit such an
attack; and (2) assessing the public threat assuming that an
insider had the access and motive to release a pathogen.
With regard to the first prong of the DOE’s insider theft
and release analysis, the DOE found that the likelihood that
an insider would have the motive to commit such an attack
was very low, based upon the small number of persons with
access to pathogens at the LLNL BSL-3 facility (fewer than
ten) and the high number of screenings and protective pro-
grams and security procedures, unique to the LLNL BSL-3
facility. In addition to complying with nationwide DOJ risk
assessment screenings, HHS authorization screenings, and
CDC registration, LLNL BSL-3 requires its employees to also
comply with its own monitoring and training system, “Select
Agent Human Reliability Program.”
Further, even assuming motive, the DOE found that the risk
of effective theft and release by an LLNL BSL-3 terrorist
insider was extremely low because an LLNL BSL-3 terrorist
insider would need to further cultivate and prepare any
covertly removed pathogen before release. The DOE specifi-
cally reasoned that this outcome further rendered the threat of
insider theft and release highly improbable because direct
removal of any significant quantity of a pathogen would be
quickly noticed because (1) material inventories are regularly
audited, and (2) the LLNL BSL-3 facility does not even con-
tain large amounts of “ready-to-use” aerosolized pathogens,
but instead stores them in small, frozen samples, in 2 ML
vials.
TRI-VALLEY CARES v. U.S. DOE 1193
E. The 2008 Final Revised Environmental Assessment
On January 25, 2008, after evaluating public comment, the
DOE found that no significant environmental impact would
result from a terrorist attack on the BSL-3 laboratory. Accord-
ingly, it released a FREA and a FONSI on January 25, 2008.
The FREA was substantially identical to the DREA, except
for a few administrative updates and three substantive
updates. One of the substantive updates addressed two
anthrax shipping incidents in 2005, while the other two dealt
with 2007 subcommittee hearings regarding BSL-3 safety
procedures, following an incident at Texas A&M.
F. The Current Litigation
On March 10, 2008, Tri-Valley CAREs filed a new com-
plaint alleging that the DOE had again failed to prepare an
adequate EA and FONSI, failed to prepare an EIS, failed to
supplement the REA, and failed to publicly circulate the
FONSI. Following the district court’s denial of a motion for
preliminary injunction by Tri-Valley CAREs, each party
moved for summary judgment. In its motion for summary
judgment, Tri-Valley CAREs asserted that the DOE (1) failed
to prepare an adequate FREA and Revised FONSI; (2) failed
to conduct an EA in good faith; (3) failed to prepare an EIS;
(4) failed to supplement the FREA; and (5) failed to publicly
circulate the FONSI. Tri-Valley CAREs sought a judgment
compelling the DOE to prepare an EIS, or alternatively, to
prepare a revised EA and to reconsider whether an EIS is nec-
essary.
Tri-Valley CAREs relied upon three primary arguments to
support its motion: it claimed that (1) the DOE’s reliance on
the same centrifuge analysis used in the original EA to assess
the dangers of a terrorist attack in the FREA violated NEPA;
(2) the DOE’s failure to disclose information related to a 2005
anthrax shipping incident in the March 2007 DREA violated
NEPA; and (3) the DOE’s failure to disclose information
1194 TRI-VALLEY CARES v. U.S. DOE
regarding violations for “restricted experiments” conducted at
the LLNL BSL-3 facility violated NEPA. On September 30,
2010, the district court granted summary judgment in the
DOE’s favor on the ground that the DOE had properly con-
ducted its analysis in revising the FREA, in compliance with
this court’s original mandate.
Tri-Valley CAREs appeals, claiming: (1) that the DOE did
not comply with our mandate in Tri-Valley CAREs v. DOE,
203 Fed. App’x 105 (9th Cir. 2006) (Tri-Valley CAREs I) by
failing to take a “hard look” at the human health, safety and
environmental risks associated with an intentional terrorist
act; (2) that the DOE violated NEPA by failing to supplement
its DREA and FREA with information regarding incidents in
which the DOE violated protocols and policies at LLNL
BSL-3’s biological facilities, depriving decision-makers and
the public of a reasonable opportunity for input; and (3) that
the district court erred in excluding Tri-Valley CARE’s extra
record evidence proving that centrifuge scenarios are inade-
quate to measure risks from an intentional terrorist attack.
II. STANDARD OF REVIEW AND JURISDICTION
We review a district court’s grant of summary judgment on
NEPA claims de novo. Northern Cheyenne Tribe v. Norton,
503 F.3d 836, 845 (9th Cir. 2007). An agency’s action must
be upheld unless it is “arbitrary, capricious, an abuse of dis-
cretion, or otherwise not in accordance with law.” 5 U.S.C.
§ 706(2)(A).
[W]e will reverse a decision as arbitrary and capri-
cious only if the agency relied on factors Congress
did not intend it to consider, entirely failed to con-
sider an important aspect of the problem, or offered
an explanation that runs counter to the evidence
before the agency or is so implausible that it could
not be ascribed to a difference in view or the product
of agency expertise.
TRI-VALLEY CARES v. U.S. DOE 1195
Lands Council v. McNair, 537 F.3d 981, 987 (9th Cir. 2008)
(en banc) (citations and internal quotation marks omitted).
NEPA is a pragmatic device that “ ‘does not mandate par-
ticular results,’ but ‘simply provides the necessary process’ to
ensure that federal agencies take a ‘hard look’ at the environ-
mental consequences of their actions.” Muckleshoot Indian
Tribe v. U.S. Forest Serv., 177 F.3d 800, 814 (9th Cir. 1999)
(per curiam) (quoting Robertson v. Methow Valley Citizens
Council, 490 U.S. 332, 350 (1989)). We look to the evidence
the agency has provided to support its conclusions, along with
materials in the record, to make this determination. Lands
Council, 537 F.3d at 993. Judicial review of agency decision-
making under NEPA is limited to the question of whether the
agency took a ‘hard look’ at the proposed action as required
by a strict reading of NEPA’s procedural requirements.
Bering Strait Citizens for Responsible Dev. v. U.S. Army
Corps of Eng’rs, 524 F.3d 938, 947 (9th Cir. 2008) (citing
Churchill Cnty. v. Norton, 276 F.3d 1060, 1072 (9th Cir.
2001)).
The district court’s decision to exclude extra-record evi-
dence is reviewed for abuse of discretion. Northwest Envtl.
Advocates v. Nat’l Marine Fisheries Serv., 460 F.3d 1125,
1133 (9th Cir. 2006).
We have jurisdiction under 28 U.S.C. § 1291.
III. DISCUSSION
A. Environmental Claims
A court generally must be “at its most deferential” when
reviewing scientific judgments and technical analyses within
the agency’s expertise. Northern Plains Resource Council,
Inc. v. Surface Transp. Bd., ___ F.3d ___, 2011 WL 6826409,
at *3 (9th Cir. Dec. 29, 2011) (quoting Balt. Gas & Elec. Co.
v. Natural Res. Def. Council, Inc., 462 U.S. 87, 103 (1983)).
1196 TRI-VALLEY CARES v. U.S. DOE
We may not impose ourselves “as a panel of scientists that
instructs the [agency] . . . , chooses among scientific studies
. . . , and orders the agency to explain every possible scientific
uncertainty.” Lands Council, 537 F.3d at 988. And “[w]hen
specialists express conflicting views, an agency must have
discretion to rely on the reasonable opinions of its own quali-
fied experts even if, as an original matter, a court might find
contrary views more persuasive.” Id. at 1000 (quoting Marsh
v. Or. Natural Res. Council, 490 U.S. 360, 378 (1989))
(emphasis added).
At a minimum, an agency must support its conclusions with
studies that the agency deems reliable. Lands Council, 537
F.3d at 994. An agency will have acted arbitrarily and capri-
ciously only when “the record plainly demonstrates that [the
agency] made a clear error in judgment in concluding that a
project meets the requirements” of NEPA. Id.
1. Our Mandate in Tri-Valley CAREs I
In Tri-Valley CAREs I, we affirmed the original EA on all
grounds, except for its failure to consider the impact of a pos-
sible terrorist attack. On that ground alone, we remanded “for
the DOE to consider whether the threat of terrorist activity
necessitates the preparation of an Environmental Impact
Statement,” by conducting a comprehensive analysis of the
human health, safety, and environmental risks associated with
a terrorist attack at LLNL’s BSL-3 facility. Tri-Valley CAREs
I, 203 F. App’x at 107 (citing San Luis Obispo Mothers for
Peace v. Nuclear Regulatory Comm’n, 449 F.3d 1016 (9th
Cir. 2006) (“Mothers for Peace I”)). Following that decision,
in Mothers for Peace II, we upheld the NRC’s finding of no
significant impact in a revised EA in which the NRC first ana-
lyzed site the site-specific low risk factors, and then bounded
the maximum impact of an attack using an MCE model. See
San Luis Obispo Mothers for Peace v. NRC, 635 F.3d 1109
(9th Cir. 2011) (Mothers for Peace II) (applying 28 U.S.C.
§ 157(b)(2)(A)).
TRI-VALLEY CARES v. U.S. DOE 1197
[1] An agency has “the discretion to determine the physical
scope used for measuring environmental impacts” so long as
the scope of analysis is “reasonable.” Idaho Sporting Cong. v.
Rittenhouse, 305 F.3d 957, 973 (9th Cir. 2002). If the pro-
posed action does not significantly alter the status quo, it does
not have a significant impact under NEPA. Burbank Anti-
Noise Group v. Goldschmidt, 623 F.2d 115, 116 (9th Cir.
1980). At bottom, an agency need only provide a “convincing
statement” of why the threat did not require an EIS to satisfy
NEPA. See Ocean Advocates v. U.S. Army Corps. of Eng’rs,
402 F.3d 846, 864 (9th Cir. 2005) (internal citations and quo-
tation marks omitted). An agency is not required to consider
every scenario, and further, nothing in NEPA requires it to
rely on purely empirical data. Id.
a. Direct Attack on the LLNL Facility
[2] The DOE utilized its MCE scenario, developed in the
original EA, to evaluate the outer bounds of a pathogen
release. In Tri-Valley CAREs I, we previously upheld the use
of this centrifuge model for a bounding analysis of the
impacts of abnormal catastrophic events, including earth-
quakes and accidental plane crashes. There, the model was
applied not to measure the cause of the triggering event, but
rather the outer limits of its impact.
In selecting the centrifuge model to measure the potential
impact of a direct terrorist attack on the LLNL BSL-3 facility,
the DOE specifically reasoned that the “catastrophic release”
model was analogous to the “direct attack” scenario because
the triggering incidents (earthquake or accidental plane crash
in the former, intentional plane crash or suicide bombing in
the latter) would result in similar structural damage to the
LLNL BSL-3 facility. Moreover, the DOE further refined its
analysis of a terrorist attack using the centrifuge model by
highlighting several key distinctions from the Army’s original
model that would significantly alter the consequences of such
an event at the LLNL facility—specifically: (1) the very lim-
1198 TRI-VALLEY CARES v. U.S. DOE
ited quantities of biological agents generally in use; (2) the
likely destruction of pathogens resulting from the fire caused
by an airplane crash or explosive device; and (3) the likely
destruction of pathogens resulting from general environmental
exposure.
[3] Because the Army’s catastrophic release scenario mea-
sured the effects of a catastrophic event at LLNL BSL-3, it
was reasonable that the model be applied to consider the outer
bounds of a threat of terrorist attack. We read Mothers for
Peace II as supporting the proposition that use of an MCE
model, such as the catastrophic release scenario applied here,
is an acceptable method to simulate the bounds of a direct ter-
rorist attack when the agency decision to use that model is
reasonably supported by agency evidence. See Mothers for
Peace II, 635 F.3d at 1113 (affirming assessment of a terrorist
threat where agency projected pathogen release using maxi-
mum credible event).3 Here, as in Mothers for Peace II, the
DOE applied a general MCE analysis to measure the outer
limits of the impact of a direct terrorist attack. The MCE cen-
trifuge model used in the original EA was an outer bounding
model for a hypothetical maximum credible event—it was not
designed to be unique to only one particular incident, or even
type of incident. We find the DOE’s use of the MCE centri-
3
Tri-Valley CAREs distorts the meaning of Mothers for Peace II in con-
tending that the analysis undertaken in Mothers for Peace II is distinguish-
able from the analysis undertaken here. Specifically, Tri-Valley CAREs
maintains that Mothers for Peace II contained a nebulous level of “addi-
tional analysis” not present in the instant case. We disagree. Although the
DOE did not use the exact same type of additional analysis used by the
NRC in Mothers for Peace II, it nonetheless considered different, yet
equally significant, additional analysis in the form of its site-specific Bio-
logical Risk and Threat Assessment (BRTA), which included an “in-
depth” analysis of potential LLNL BSL-3 vulnerabilities and which
guided the discussion of terrorist threats in both the DREA and FREA.
Thus, we conclude that the DOE’s consideration of site-specific factors in
the BRTA is analogous to the “additional analysis” undertaken by the
NRC in contacting law enforcement and intelligence communities under
Mothers for Peace II. 635 F.3d at 1113.
TRI-VALLEY CARES v. U.S. DOE 1199
fuge model sufficient under NEPA and Mothers for Peace II
because the DOE reasonably justified its selection based upon
record evidence and additional analysis of site-specific fac-
tors.
[4] Whether we agree that a centrifuge model was the best
way to assess the threat of direct terrorist attack is not the
inquiry before us. Lands Council, 537 F.3d at 988 (“[w]hen
specialists express conflicting views, an agency must have
discretion to rely on the reasonable opinions of its own quali-
fied experts even if, as an original matter, a court might find
contrary views more persuasive.”) (internal citations omitted).
Under NEPA, we must restrain from acting as a type of
omnipotent scientist, and instead must restrict ourselves to
inquiring only whether an agency took a “hard look” at the
potential environmental impacts at issue. Muckleshoot Indian
Tribe v. U.S. Forest Serv., 177 F.3d 800, 814 (9th Cir.1999)
(per curiam) (quoting Robertson v. Methow Valley Citizens
Council, 490 U.S. 332, 350 (1989)). When reasonable scien-
tists disagree on appropriate models for analysis, we must
defer to agency experts. Lands Council, 537 F.3d at 988.
Here, the DOE provided ample justification and evidence for
why it used the centrifuge model to assess the impact of a ter-
rorist attack: it analogized triggering events, compared critical
distinctions, and considered uniquely different circumstances.
Accordingly, because of the deference that must be afforded
to the agency, we find that the DOE took the requisite “hard
look” at the threat of direct terrorist attack.
b. Theft and Release by a LLNL BSL-3 Terrorist
Outsider
In assessing the impact of a terrorist threat by the theft and
release by an LLNL BSL-3 terrorist outsider, the DOE used
a comparative nationwide analysis to determine that the
LLNL BSL-3 facility would not be an attractive terrorist tar-
get. Specifically, the DOE explained that the LLNL BSL-3
would not alter the status quo basis because there are hun-
1200 TRI-VALLEY CARES v. U.S. DOE
dreds of other BSL-3 facilities in the United States that regu-
larly handle and store the same substances, and moreover, that
such substances are also available to potential terrorists from
common environmental sources.
Tri-Valley CAREs contends that the DOE’s analysis is
deficient because, pursuant to 40 C.F.R. § 1508.27(a), the
DOE must assess the risk of terrorist theft and release “in the
context of the Livermore locale.” We disagree. Although 40
C.F.R. § 1508.27(a) suggests that site-specific actions are
generally evaluated in the context of a project locale, nothing
in the regulation prohibits the DOE from exercising its discre-
tion to apply a nationwide analysis when appropriate. See
Nat’l Parks & Cons. Ass’n v. Babbit, 241F.3d 722, 731 (9th
Cir. 2001). The “identification of the geographic area” within
which a project’s impacts on the environmental resources may
occur “is a task assigned to the special competency of the
appropriate agencies.” Kleppe v. Sierra Club, 427 U.S. 390,
414 (1976).
[5] We find that the DOE’s determination of the potential
impact of a terrorist theft and release of a pathogen on a
national level satisfies NEPA because the record does not
show any meaningful difference between the materials pres-
ent at the LLNL BSL-3 facility and those present at other
BSL-3 facilities nationwide. Nowhere in the record is there
any proof that the LLNL BSL-3 facility is more prone or
attractive to terrorist theft and release of a pathogen by an out-
sider than any other BSL-3 facility. To the contrary, the
record reveals that LLNL is actually one of the most heavily
guarded federal facilities, in contrast to hundreds of relatively
unguarded BSL-3 facilities nationwide. Given that there are
more than 1,300 other BSL-3 facilities nationwide, many of
which lack the safeguards of LLNL’s BSL-3 facility, and fur-
ther, given that many of the BSL-3 pathogens also exist in the
natural environment, DOE reasonably concluded that the con-
struction of a BSL-3 facility at LLNL did not change the sta-
tus quo, and therefore found no significant impact. See
TRI-VALLEY CARES v. U.S. DOE 1201
Burbank Anti-Noise Group v. Goldschmidt, 623 F.2d 115, 116
(9th Cir. 1980) (holding that where a proposed project does
not alter the status quo then it does not have a significant
impact). Accordingly, we find that the DOE reasonably exer-
cised its discretion in determining no significant impact from
the threat of theft and release by a LLNL BSL-3 terrorist out-
sider.
c. Theft and Release by a LLNL BSL-3 Terrorist
Insider
[6] Lastly, the DOE’s discussion of the impact of the
potential theft and release of a pathogen by an LLNL BSL-3
terrorist insider also satisfies NEPA. Although the DOE did
not use an empirical model, it engaged in a thorough two-step
probabilistic analysis that assessed: (1) the probability that an
insider with access to BSL-3 pathogens would have the
motive to commit such an attack; and (2) the public threat that
would result, assuming that an insider did have the access and
motive to release a pathogen.
[7] Tri-Valley CAREs’ claim that the DOE violated NEPA
because it did not employ empirical analysis fails. Empirical
analysis is not required under NEPA; an agency must only
provide a “convincing statement” of why the threat did not
require an EIS. See Ocean Advocates v. U.S. Army Corps. of
Eng’rs, 402 F.3d 846, 864 (9th Cir. 2005). The DOE laid out
its dual-tiered probabilistic analysis, discretely defined the
scope of its inquiry, and thoughtfully examined the likelihood
of an insider stealing and releasing pathogens from LLNL.
Based upon the facts that (1) a very small number of people
have access to the select agents at LLNL BSL-3, all of whom
are subject to extensive screening procedures from multiple
agencies; and (2) the form and quantities of the pathogens at
LLNL BSL-3 would require significant additional efforts to
bring about a terrorist attack, the DOE concluded that the
threat of a theft and release by an insider was not significant.
The DOE’s methodical inquiry satisfies NEPA’s requirement
1202 TRI-VALLEY CARES v. U.S. DOE
that it provide a “convincing statement” as to why the threat
did not require an EIS. Accordingly, we find that the DOE
reasonably concluded, based upon its discretion and a thor-
ough examination of the evidence in the record, that threat of
terrorist attack by a theft and release from an LLNL BSL-3
terrorist insider was not significant.
2. Adequacy of the FREA
The purpose of an EA under NEPA is not to amass and dis-
close all possible details regarding a proposal, but to create a
“concise public document” that serves to “[b]riefly provide
sufficient evidence and analysis for determining whether to
prepare an environmental impact statement or a finding of no
significant impact.” 40 C.F.R. § 1508.9; League of Wilderness
Defenders Blue Mountains Biodiversity Project v. Allen, 615
F.3d 1122, 1136 (9th Cir. 2010) (NEPA review “must concen-
trate on the issues that are truly significant to the action . . . ,
rather than amassing needless detail.”) (quoting 40 C.F.R.
§ 1500.1(b)). An EA “must provide the public with sufficient
environmental information, considered in the totality of the
circumstances, to permit members of the public to weigh in
with their views and thus inform the agency decision-making
process.” Bering Strait Citizens v. U.S. Army Corps. of
Eng’rs, 524 F.3d 938, 953 (9th Cir. 2008).
a. The 2005 Anthrax Shipping Incident
Tri-Valley CAREs maintains that the DOE did not satisfy
the standard set forth in Mothers for Peace I, which held that
NEPA serves two fundamental purposes: (1) to require
agency consideration of detailed information concerning sig-
nificant environmental impacts; and (2) to ensure that the
public can both access and contribute to that body of informa-
tion via comments. Mothers for Peace I, 449 F.3d at 1034 (cit-
ing Dep’t of Trans. v. Public Citizen, 541 U.S. 752, 768
(2004) (emphasis added)). Specifically, Tri-Valley CAREs
claims that the DOE violated NEPA by failing to fully dis-
TRI-VALLEY CARES v. U.S. DOE 1203
close a 2005 anthrax shipping incident, therefore depriving
the public of the ability to comment.
The 2005 incident involved a professor at Louisiana State
University who owned a collection of anthrax samples used
at LLNL and contracted with a former LLNL employee to
return to LLNL to package and ship the collection to two pri-
vate labs. The former employee returned to LLNL on August
25, 2005, and packaged and shipped 1,065 samples from the
collection to one of the labs, without incident. On September
13, 2005, the former employee returned to the facility to ship
3,108 samples to the other lab. The shipment arrived intact
but the second lab noted discrepancies between the shipment
received and the inventory. Finally, on September 14, 2005,
the former employee sent a second shipment of 1,025 samples
to the first lab. In unpacking the vials, employees at the first
lab discovered improperly sealed vials, and were exposed to
anthrax that had leaked into the interior packaging. The first
lab concluded, however, that “nothing was detected on the
outside of the shipping container and therefore [the leak] was
‘not a public health issue.’ ” As a result of the incident, the
CDC suspended all transfers of select agents from LLNL, and
LLNL voluntarily suspended all work with select agents
pending an investigation. It also established an Incident Anal-
ysis Committee. In December 2005, the Incident Analysis
Committee completed a comprehensive report identifying
areas in need of correction and tracing many of the causes of
the incident to the unique role of the former employee. In
response, LLNL implemented numerous corrective actions,
including expanding the Select Agent Security Plan. In Febru-
ary 2006, DOT examined LLNL’s select agent program, con-
cluded that the new procedures were sufficient and that the
2005 incident resulted from an “unusual” event. On April 18,
2006, the DOE authorized the resumption of select agent
work at LLNL.
In the original EA, the DOE reasoned that the addition of
milliliter-quantity samples from LLNL to the hundreds of tons
1204 TRI-VALLEY CARES v. U.S. DOE
of infectious material already shipped daily would not have a
significant impact on the risk of transportation accidents. In
the DREA, the DOE added a brief discussion of the 2005
shipping incident, without identifying the select agent
involved, but found that a more detailed discussion was not
warranted when considered against the decades-long history
of safe shipments of hundreds of tons of infectious materials.
Following public comment to the disclosure of this incident
in the DREA, including comments from Tri-Valley CAREs
itself, the DOE revised the FREA to include an even more
detailed discussion of the 2005 anthrax shipping incident in
the hope of providing the public with a “better understand[ing
of] why the incident did not add significant information and
did not challenge the conclusions of the document.”
[8] Tri-Valley CAREs’ arguments are unpersuasive
because in the original EA, the DREA, and the FREA, the
DOE specifically and carefully considered the risks of ship-
ping infectious materials to and from the BSL-3 lab and dis-
closed these risks to the public. In the original EA, the DOE
analyzed and found that the risk of fatality from hazardous
waste transportation incidents was less than .11 per million
shipments, and the specific risk from infectious substance
incidents was too low to even be quantified. Tri-Valley
CAREs cannot escape the logical inconsistency of its posi-
tion. How could the DOE’s disclosure of the 2005 shipping
incident be so deficient as to deprive the public of the ability
to respond, when Tri-Valley CAREs itself relied upon that
very document to specifically and publicly comment on the
2005 shipping incident?
[9] The purpose of an EA is not to compile an exhaustive
examination of each and every tangential event that poten-
tially could impact the local environment. Such a task is
impossible, and never-ending. The purpose of the EA is sim-
ply to create a workable public document that briefly provides
evidence and analysis for an agency’s finding regarding an
environmental impact. 40 C.F.R. § 1508.9; League of Wilder-
TRI-VALLEY CARES v. U.S. DOE 1205
ness Defenders Blue Mountains Biodiversity Project v. Allen,
615 F.3d 1122, 1136 (9th Cir. 2010) (emphasizing the parsi-
monious nature of an effective EA) (quoting 40 C.F.R.
§ 1500.1(b)). The DOE has more than met its burden here.
Accordingly, we find that the DOE’s discussion of the 2005
anthrax shipping incident in the DREA and the FREA satis-
fies NEPA.
b. The “Restricted” Experiments
We do not reach the issue of whether the DOE’s decision
not to include information about “restricted” experiments in
the FREA was arbitrary or capricious because the district
court properly found that this argument was waived. Here,
Tri-Valley CAREs did not address the district court’s ruling
that the argument was deemed waived in its opening appellate
brief. Thus, the issue is not properly before us. See Smith v.
Marsh, 194 F.3d 1045, 1052 (9th Cir. 1999) (holding that
arguments not raised by a party in its opening brief are
deemed waived). Claims not made in an opening brief in a
sufficient manner to put the opposing party on notice are
deemed waived. Swierkiewicz v. Sorema, N.A., 534 U.S. 506,
512 (2002) (holding that claims must put parties on sufficient
notice of underlying arguments, or arguments are deemed
waived).
c. The 2008 Security Assessment
[10] NEPA requires supplementation of any NEPA analy-
sis in response to “significant new circumstances or informa-
tion relevant to environmental concerns and bearing on the
proposed action or its impacts.” 40 C.F.R. § 1502.9(c)(1)(ii).
Supplementation is not required “every time new information
comes to light after the EIS is finalized. To require otherwise
would render agency decisionmaking intractable, always
awaiting updated information.” Marsh v. Oregon Natural Res.
Council, 490 U.S. 360, 373-74 (1989). Whether new informa-
tion requires supplemental analysis is a “classic example of a
1206 TRI-VALLEY CARES v. U.S. DOE
factual dispute the resolution of which implicates substantial
agency expertise.” Id. at 376.
[11] Tri-Valley CAREs contends that the DOE violated the
NEPA supplementation rule when it failed to supplement the
FREA to address the results of its Security Assessment (SA)
conducted at LLNL in 2008 by the DOE’s Health, Safety and
Security Office of Independent Oversight. The SA included a
mock attack on the “Superblock,” where special nuclear mate-
rials are stored, and identified several deficiencies in perfor-
mance of LLNL’s protective force. The SA gave LLNL’s
protective force the lowest possible rating, “Significant Weak-
nesses.” Specifically, the SA identified deficiencies in
LLNL’s physical security systems and protection program
management. In July 2008, however, the DOE prepared a sup-
plemental report to determine whether the SA constituted sig-
nificant new information requiring supplementation of the
FREA. There, the DOE examined whether the low rating, and
the deficiencies identified therein, significantly altered the
outcomes of any of the three terrorist attack scenarios (as pre-
viously discussed, (1) intentional airplane crash, (2) inten-
tional theft and release from LLNL outsider, and (3)
intentional theft and release from LLNL insider). Because the
DOE determined in its supplemental report that the SA did
not show a “seriously different picture of the likely environ-
mental harms stemming from the proposed project,” we must
defer to the DOE’s finding that a supplemental REA was not
required. Wisconsin v. Weinberger, 745 F.2d 412, 416-17 (7th
Cir. 1984).
B. Evidentiary Claims
A reviewing court may consider extra-record materials
only: (1) if necessary to determine whether the agency has
considered all relevant factors and explained its decision, (2)
when the agency has relied on documents not in the record,
(3) when supplementing the record is necessary to explain
technical terms or complex subject matter, or (4) when plain-
TRI-VALLEY CARES v. U.S. DOE 1207
tiffs make a showing of agency bad faith. Inland Empire Pub.
Lands Council v. Glickman, 88 F.3d 697, 703-04 (9th Cir.
1996). However, exceptions to the normal rule regarding con-
sideration of extra-record materials “only appl[y] to informa-
tion available at the time, not post-decisional information.”
Rock Creek Alliance v. U.S. Fish & Wildlife Serv., 390 F.
Supp. 2d 993, 1002 (D. Mont. 2005). When considering the
Inland Empire factors, post-decision information “may not be
advanced as a new rationalization either for sustaining or
attacking an agency’s decision” because “it inevitably leads
the reviewing court to substitute its judgment for that of the
agency.” Ctr. for Biological Diversity v. U.S. Fish & Wildlife
Serv., 450 F.3d 930, 943 (9th Cir. 2006) (internal citations
and quotations omitted).
1. Northern District of California Civil Local Rule
7-11(a)
[12] Tri-Valley CAREs contends that the district court
abused its discretion in denying its motion to augment the
record with a report by the National Research Council of the
National Academies of Science (NRC), regarding the alleged
inapplicability and inadequacy of centrifuge models to mea-
sure the environmental impact of terrorist attacks. In this case,
the district court properly denied Tri-Valley CAREs’ motion
to supplement the record because of its failure to comply with
the local rules. Denial of a motion as the result of a failure to
comply with local rules is well within a district court’s discre-
tion. Grove v. Wells Fargo Fin. California, Inc., 606 F.3d
577, 582 (9th Cir. 2010). Northern District of California Civil
Local Rule 7-11(a) requires that any motion for administrative
relief include a stipulation or declaration explaining why a
stipulation could not be obtained. N. Cal. L. R. 7-11(a).
Nowhere in its motion, nor in its additional briefs, did Tri-
Valley CAREs include the requisite stipulation or declaration.
Accordingly, the district court acted well within its discretion
to deny Tri-Valley CAREs’ motion to augment the record
with the NRC Report.
1208 TRI-VALLEY CARES v. U.S. DOE
2. Adequacy of the Record
Even if the district court had abused its discretion in deny-
ing Tri-Valley CAREs’ motion to supplement the record for
failure to comply with the local rules, it nonetheless could
have appropriately denied Tri-Valley CAREs’ motion to aug-
ment the administrative record on the merits. In its complaint,
Tri-Valley CAREs only contended that augmentation was
necessary “to ascertain whether the agency considered all rel-
evant factors or fully explicated its course of conduct or
grounds for decision.” Such an assertion fails to demonstrate
how the NRC report satisfies any of the exceptions under
Inland Empire. The only ground on which Tri-Valley CAREs
asserts that this evidence should be admitted into the record
is that it is relevant to determining whether the agency consid-
ered all relevant factors, or the first exception under Inland
Empire. The post-decision bar, however, renders this excep-
tion inapplicable, as the NRC report was completed over
nearly two years after the commencement of this litigation.
See Ctr. for Biological Diversity, 450 F.3d at 943 (9th Cir.
2006) (explaining the post-decision bar may not be applied to
require augmentation of information used to rationalize,
attack, or even analyze an agency decision post hoc). Accord-
ingly, the district court did not abuse its discretion in refusing
to grant Tri-Valley CAREs’ motion to supplement the record
with the NRC report.
IV. CONCLUSION
For the foregoing reasons, we affirm the decision of the
district court.
AFFIRMED.