NRDC v. EPA ( 2015 )


Menu:
  • 13‐1745(L)
    NRDC v. EPA
    UNITED STATES COURT OF APPEALS
    FOR THE SECOND CIRCUIT
    August Term 2014
    (Argued:  January 30, 2015         Decided:  October 5, 2015
    Amended: December 18, 2015)
    Docket Nos. 13‐1745(L), 13‐2393(CON), 13‐2757(CON)
    NATURAL RESOURCES DEFENSE COUNCIL, NORTHWEST ENVIRONMENTAL
    ADVOCATES, CENTER FOR BIOLOGICAL DIVERSITY, and
    NATIONAL WILDLIFE FEDERATION,
    Petitioners,
    v.
    UNITED STATES ENVIRONMENTAL PROTECTION AGENCY,
    Respondent,
    LAKE CARRIERSʹ ASSOCIATION and CANADIAN SHIPOWNERS ASSOCIATION,
    Intervenors.
    ON PETITION FOR REVIEW FROM THE
    ENVIRONMENTAL PROTECTION AGENCY
    Before:
    SACK, CHIN, and CARNEY, Circuit Judges.
    Four environmental organizations petition for review of a Vessel
    General Permit issued by the Environmental Protection Agency in 2013 under
    Section 509(b)(1) of the Clean Water Act, 33 U.S.C. § 1369(b)(1).  The permit
    regulates the discharge of ballast water from ships, a primary cause of the spread
    of invasive species from one body of water to another.  Petitioners contend that
    the Environmental Protection Agency acted arbitrarily and capriciously in
    issuing the permit, and request that it be set aside.  We agree, in part.
    Accordingly, we grant the petition for review in part and deny it in part, and
    remand to the Environmental Protection Agency for further proceedings
    consistent with this opinion.  We do not vacate the Vessel General Permit, but
    allow it to remain in effect until the issuance of a new Vessel General Permit.
    PETITION GRANTED IN PART AND DENIED IN PART AND REMANDED.
    ‐2‐
    ALLISON M. LAPLANTE, Earthrise Law Center, Lewis &
    Clark Law School, Portland, Oregon, and Deborah
    A. Sivas, Matthew J. Sanders, Environmental Law
    Clinic, Mills Legal Clinic at Stanford Law School,
    Stanford, California, for Petitioners Northwest
    Environmental Advocates and Center for Biological
    Diversity.
    Rebecca J. Riley, Natural Resources Defense Council,
    Chicago, Illinois, for Petitioner Natural Resources
    Defense Council.
    Neil S. Kagan, National Wildlife Federation, Ann Arbor,
    Michigan, for Petitioner National Wildlife Federation.
    MARTIN FRANCIS MCDERMOTT (Sam Hirsch, Acting
    Assistant Attorney General, on the brief),
    Environmental Defense Section, Environmental &
    Natural Resources Division, United States
    Department of Justice, Washington, D.C., and
    Dawn M. Messier, Office of General Counsel,
    United States Environmental Protection Agency,
    Washington, D.C., for Respondent United States
    Environmental Protection Agency.
    MATTHEW D. MELEWSKI, The Boutique Firm PLC,
    Minneapolis, Minnesota, for Intervenors Lake
    Carriersʹ Association and Canadian Shipowners
    Association.
    ‐3‐
    CHIN, Circuit Judge:
    This case arises from the efforts of the Environmental Protection
    Agency (ʺEPAʺ) pursuant to section 402(a) of the Clean Water Act (the ʺCWAʺ),
    33 U.S.C. § 1342(a), to regulate the discharge of ballast water from ships. 1  A ship
    takes on and discharges ballast water to compensate for changes in its weight
    caused by activities such as loading and unloading cargo or consuming fuel or
    supplies.  The amount of water can range from hundreds of gallons to as much
    as 25 million gallons ‐‐ enough to fill thirty‐eight Olympic‐sized swimming
    pools.  More than 21 billion gallons of ballast water are released in the United
    States annually.  See Nw. Envtl. Advocates v. EPA, 537 F.3d 1006, 1013 (9th Cir.
    2008).
    When a ship takes on ballast water, it can inadvertently pick up
    organisms and their eggs and larvae, as well as sediment and pollutants.  When
    the ship discharges ballast water, often in a new place, these organisms and
    pollutants are ejected into the surrounding waterbody, enabling these organisms
    to establish new, non‐native populations.  As a result, ships have become one of
    1    Glossary of Acronyms:  As this opinion discusses the CWA and its
    intricacies, it contains a large number of acronyms.  In addition to their definitions in
    the text, a separate glossary of acronyms is therefore set forth in the Appendix to this
    opinion.
    ‐4‐
    the primary ways that invasive species are spread from one waterbody to
    another.  Id. at 1012‐13 (ʺAll told, more than 10,000 marine species each day hitch
    rides around the globe in the ballast water of cargo ships.ʺ (quoting Nw. Envtl.
    Advocates v. EPA, No. C 03‐05760 SI, 2006 WL 2669042, at *3 (N.D. Cal. Sept. 18,
    2006)).
    Invasive species cause severe economic and ecological harm,
    including by destroying native fish species and shellfish industries, creating
    algae blooms, and devastating tourism.  Zebra mussels are a particularly
    destructive example.  They were first introduced to Lake Erie in the 1980s by a
    freighter from Europe that discharged ballast water containing mussels.2  These
    mussels have wreaked havoc in the Midwest and Northeast by blocking water
    intake and outtake at power plants and other industrial facilities, causing nearly
    $70 million in damage between 1989 and 1995.   Nw. Envtl. Advocates, 537 F.3d at
    1013.  One study estimates the damage caused by invasive species collectively at
    ʺabout $137 billion a year ‐‐ more than double the annual economic damage
    2
    ʺFrom that humble start, the invaders colonized the Great Lakes and
    spread across the country on towed boats.ʺ  Jim Robbins, A Western Showdown, N.Y.
    TIMES, Sept. 8, 2015, at D6.
    ‐5‐
    caused by all natural disasters in the United States.ʺ  Id. (quoting Nw. Envtl.
    Advocates, 2006 WL 2669042, at *4).3
    Ballast water discharge is particularly problematic in the Great
    Lakes.  Vessels that sail exclusively in the Great Lakes, known as ʺLakers,ʺ
    account for over ninety‐five percent of ballast water volumes transferred in the
    Great Lakes.  Unfortunately, Lakers are more likely than oceangoing vessels to
    spread invasive species because the short duration of their voyages allows
    organisms to survive in their ballast.
    In April 2013, EPA issued a Vessel General Permit (the ʺ2013 VGPʺ),
    pursuant to section 402 of the CWA, 33 U.S.C. § 1342, to regulate the discharge of
    ballast water from ships.  In response, four environmental groups filed three
    Petitions for Review (ʺPFRsʺ) alleging that EPA acted arbitrarily and capriciously
    in issuing the 2013 VGP: petitioner Natural Resources Defense Council (ʺNRDCʺ)
    filed a PFR on May 3, 2013 in this Court; petitioners Northwest Environmental
    Advocates (ʺNWEAʺ) and the Center for Biological Diversity jointly filed a PFR
    on May 3, 2013 in the United States Court of Appeals for the Ninth Circuit; and
    petitioner National Wildlife Federation (ʺNWFʺ) filed a PFR on July 3, 2013 in the
    3
    See also Robbins, A Western Showdown, at D6 (discussing damage caused
    by zebra and quagga mussels).
    ‐6‐
    United States Court of Appeals for the D.C. Circuit.4  In an order dated May 24,
    2013, the Judicial Panel on Multidistrict Litigation issued a Consolidation Order
    and assigned final venue for the first two petitions, and any subsequently filed
    petition, to this Court.
    On May 31, 2013, the Lake Carriersʹ Association and the Canadian
    Shipowners Association (the ʺCSAʺ) filed a motion to intervene, which was
    granted on October 7, 2013.  On January 1, 2014, the CSA filed a PFR in this case.
    EPA and the CSA jointly moved to sever the CSA PFR from this case and hold it
    in abeyance; the motion was granted on May 23, 2014.
    We find that EPA acted arbitrarily and capriciously in issuing parts
    of the 2013 VGP, and therefore remand this matter to the EPA for further
    proceedings.
    BACKGROUND
    A.         The CWA
    Congress created the CWA to limit pollution in the waters of the
    United States.  See 33 U.S.C. § 1251(a) (objective of CWA is to ʺrestore and
    4  All three petitions were timely filed within 120 days of the issuance of the
    VGP, as required under 33 U.S.C. § 1369(b)(1).  Accordingly, this Court has jurisdiction
    over the petitions pursuant to section 509(b)(1)(F) of the CWA, 33 U.S.C. § 1369(b)(1)(F).
    ‐7‐
    maintain the chemical, physical, and biological integrity of the Nationʹs watersʺ);
    S. Fla. Water Mgmt. Dist. v. Miccosukee Tribe of Indians, 541 U.S. 95, 102 (2004)
    (same); Waterkeeper All., Inc. v. EPA, 399 F.3d 486, 490‐91 (2d Cir. 2005) (same).
    The CWA thus prohibits the ʺdischarge of any pollutantʺ from a ʺpoint sourceʺ to
    the ʺnavigable watersʺ of the United States, except as permitted by the CWA.  33
    U.S.C. §§ 1311(a), 1362 (emphasis added).  The ʺdischarge of a pollutantʺ includes
    ʺany addition of any pollutant to navigable waters from any point source.ʺ  Id.
    § 1362(12)(A).  A ʺpollutantʺ includes solid, industrial, agricultural, and
    biological waste.  Id. § 1362(6).  A ʺpoint sourceʺ is ʺany discernible, confined and
    discrete conveyance, including but not limited to any . . . vessel or other floating
    craft, from which pollutants are or may be discharged.ʺ  Id. § 1362(14).
    ʺNavigable watersʺ is defined as ʺthe waters of the United States, including the
    territorial seas.ʺ  Id. § 1362(7).  The discharge of polluted water from a vessel
    ballast tank is a point source discharge covered by the CWA.  See Nw. Envtl.
    Advocates, 537 F.3d at 1021.
    A key component of the statute is the establishment of water quality
    standards.  Water quality standards are set by states for waters within their
    boundaries and are then reviewed for approval by EPA.  See 33 U.S.C. § 1313; 40
    ‐8‐
    C.F.R. §§ 131.4, 131.10‐.11; see also NRDC v. EPA, 279 F.3d 1180, 1183 (9th Cir.
    2002) (ʺUnder the CWA, each state sets its own water quality standards, subject
    to review and approval by the EPA.ʺ).  EPA must ensure that the standard
    proposed by the state will comply with the requirements of the CWA before
    approving it.  See 33 U.S.C. §§ 1311(b)(1)(C), 1313(a) 1342(a)(1); 40 C.F.R. §
    122.4(d).
    1.     National Pollutant Discharge Elimination System Permits
    An entity seeking to discharge a pollutant is required to obtain and
    comply with a permit that limits the amounts and kinds of pollutants being
    discharged.  See NRDC v. EPA, 822 F.2d 104, 108 (D.C. Cir. 1987); see also
    Waterkeeper All., 399 F.3d at 498 (discharge allowed ʺwhere . . . permits ensure that
    every discharge of pollutants will comply with all applicable effluent limitations
    and standardsʺ).  This permit, known as a National Pollutant Discharge
    Elimination System (ʺNPDESʺ) permit, establishes enforceable effluent
    limitations, as well as monitoring and reporting requirements.
    NPDES permits, which are issued either by EPA or a state in a
    federally approved permitting system, see 33 U.S.C. § 1342, may be individual
    (issued to a specific entity to discharge pollutants at a specific place) or general
    ‐9‐
    (issued to an entire class of dischargers in a geographic location), see 40 C.F.R.
    §§ 122.21, 122.28(a)(2), 124.1‐.21, 124.51‐.66.  The permit here is a general permit.
    Permits can impose two different types of standards on discharges:
    (1) technology‐based standards and (2) water quality‐based standards.  See 33
    U.S.C. §§ 1311(b)(1)(c) and (b)(2)(a), 1313, 1342(a).  The 2013 VGP imposes both.
    a.     Technology‐Based Effluent Limits
    Technology‐based effluent limits (ʺTBELsʺ) set effluent limitations
    on a point source based on how effectively technology can reduce the pollutant
    being discharged.  See 33 U.S.C. §§ 1311(b), (e), 1314(b); see also PUD No. 1 of
    Jefferson Cty. v. Wash. Depʹt of Ecology, 511 U.S. 700, 704 (1994) (holding that, to
    achieve goals of CWA, EPA is required to ʺestablish and enforce technology‐
    based limitations on individual discharges into the countryʹs navigable waters
    from point sourcesʺ).  Congress designed this standard to be technology‐forcing,
    meaning it should force agencies and permit applicants to adopt technologies
    that achieve the greatest reductions in pollution.  See NRDC, 822 F.2d at 124
    (holding that CWA seeks ʺnot only to stimulate but to press development of new,
    ‐10‐
    more efficient and effective technologies,ʺ which is ʺessential purpose of this
    series of progressively more demanding technology‐based standardsʺ).5
    In determining the standard for TBELs, EPA considers the source of
    the pollution (existing or new) and the type of pollutant.  For nonconventional
    pollutants from existing sources, EPA is required to set effluent limits based on
    the ʺbest available technology economically achievableʺ or ʺBAT.ʺ  33 U.S.C.
    § 1311(b)(2)(A).6  BAT requires the ʺapplication of the best available technology
    economically achievable for such category or class, which will result in
    reasonable further progress toward the national goal of eliminating the discharge
    of all pollutants.ʺ  Id.; see NRDC, 822 F.2d at 123 (CWA designed to progress
    ʺtoward implementation of pollution controls to the full extent of the best
    technology which would become availableʺ).  Because invasive species are a
    5  EPA issues national effluent limitation guidelines (ʺELGsʺ), which
    establish limitations for all types of dischargers within a particular industry and for
    certain types of discharges.  See 40 C.F.R. § 125.3(c)(1).  ELGs are enforceable through
    their incorporation into a NPDES permit.  In this case, no states have established
    numeric water quality criteria for living organisms or aquatic nuisance species.
    6  For conventional pollutants from existing sources, the level of pollution
    control is based on best conventional pollutant control technology.  Id. § 1311(b)(2)(E).
    New sources of pollution must meet new source performance standards, which are
    based on best available demonstrated control technology.  Id. § 1316(a)(1).  Neither
    standard is implicated here.
    ‐11‐
    nonconventional pollutant from an existing source, ballast water discharges are
    subject to BAT.
    EPA considers a number of factors in assessing whether a
    technology is BAT, including:
            the cost of achieving the effluent reductions,
            the age of equipment and facilities involved,
            the process employed,
            the engineering aspects of various control techniques,
            potential process changes,
            non‐water‐quality environmental impacts including energy
    requirements, and
            other factors as EPA ʺdeems appropriate.ʺ
    See 33 U.S.C. § 1314(b)(2)(B).
    EPA can mandate that BAT requires the use of a technology that is
    not currently available within a particular industry when (1) the technology is
    available in another industry, (2) EPA finds that the technology is transferrable
    from that other industry, and (3) EPA can reasonably predict that such
    technology will adequately treat the effluent.  See Kennecott v. EPA, 780 F.2d 445,
    453 (4th Cir. 1986) (citing Tannersʹ Council of Am., Inc. v. Train, 540 F.2d 1188, 1192
    (4th Cir. 1976)).
    ‐12‐
    b.     Water Quality‐Based Effluent Limits
    If the TBELs are insufficient to attain or maintain water quality
    standards, the CWA requires NPDES permits to include additional water
    quality‐based effluent limits (ʺWQBELsʺ).  See 33 U.S.C. §§ 1311(b)(1)(C), 1312(a);
    NRDC, 822 F.2d at 110 (ʺWhenever a technology‐based effluent limitation is
    insufficient to make a particular body of water fit for the uses for which it is
    needed, the EPA is to devise a water‐quality based limitation that will be
    sufficient to the task.ʺ).  WQBELs are designed to ensure that the discharges
    authorized by the permit do not violate water quality standards. See 33 U.S.C.
    §§ 1313, 1342(a)(2).
    The WQBELs, which supplement the TBELs, are based on the
    amount and kind of pollutants in the water.  See id. § 1312(a).  WQBELs are set
    without regard to cost or technology availability.  See NRDC v. EPA, 859 F.2d 156,
    208 (D.C. Cir. 1988) (ʺA technology‐based standard discards its fundamental
    premise when it ignores the limits inherent in the technology.  By contrast, a
    water quality‐based permit limit begins with the premise that a certain level of
    water quality will be maintained, come what may, and places upon the permittee
    the responsibility for realizing that goal.ʺ (footnote omitted)).  WQBELs may be
    ‐13‐
    narrative where the calculation of numeric limits is ʺinfeasible.ʺ  See 40 C.F.R.
    § 122.44(k)(3).
    No permit may be issued when ʺthe imposition of conditions cannot
    ensure compliance with the applicable water quality requirements of all affected
    States.ʺ  Id. § 122.4(d).  Thus, permits must establish limits on discharges that will
    lead to compliance with water quality standards.  See Trs. for Alaska v. EPA, 749
    F.2d 549, 556‐57 (9th Cir. 1984) (holding that permit must translate state water
    quality standards into end‐of‐pipe effluent limitations necessary to achieve those
    standards).
    Because no states have established numeric water quality criteria for
    invasive species, EPA is required to establish WQBELs that ensure compliance
    with narrative criteria, designated uses, and antidegradation policies that
    comprise state water quality standards.  The permit may then mandate ʺbest
    management practicesʺ (ʺBMPsʺ) to control pollution.  See 40
    C.F.R. § 122.44(k)(3).
    c.     Monitoring and Reporting Requirements
    NPDES permits also require both monitoring and reporting of
    monitoring results of TBELs and WQBELs to assure compliance with permit
    ‐14‐
    limitations and facilitate enforcement.  See 33 U.S.C. §§ 1314, 1318, 1342(a)(2); 40
    C.F.R. § 122.44(i)(1)‐(2).
    B.     Regulatory History
    When the CWA was first being implemented in the 1970s, EPA
    regulations exempted discharges that were ʺincidentalʺ to the ʺnormal operationʺ
    of vessels from NPDES permitting requirements.  See National Pollutant Discharge
    Elimination System, 38 Fed. Reg. 13,528, 13,530 (May 22, 1973) (codified at 40
    C.F.R. § 125.4); see also National Pollutant Discharge Elimination System; Revision of
    Regulations, 44 Fed. Reg. 32,854, 32,902 (June 7, 1979) (codified at 40 C.F.R.
    § 122.3(a)).  This exemption included ballast water discharges.
    In 1999, the NWEA and other environmental organizations
    submitted a rulemaking petition to EPA seeking to repeal this exemption, then
    codified at 40 C.F.R. § 122.3(a).  See Final National Pollutant Discharge Elimination
    System (NPDES) General Permit for Discharges Incidental to the Normal Operation of a
    Vessel, 73 Fed. Reg. 79,473, 79,475 (Dec. 29, 2008).  EPA denied the petition.  See
    Nw. Envtl. Advocates, 537 F.3d at 1013.  The environmental groups challenged the
    denial in the United States District Court for the Northern District of California,
    while simultaneously filing a PFR in the United States Court of Appeals for the
    ‐15‐
    Ninth Circuit, in case the district court lacked jurisdiction.  The district court
    issued an order vacating the exemption, see Nw. Envtl. Advocates, 2006 WL
    2669042, at *15, and the Ninth Circuit upheld the decision.  See Nw. Envtl.
    Advocates, 537 F.3d at 1027.   EPA finally repealed the exemption and issued a
    Vessel General Permit in 2008 (the ʺ2008 VGPʺ).  Draft National Pollutant Discharge
    Elimination System (NPDES) General Permits for Discharges Incidental to the Normal
    Operation of a Vessel, 73 Fed. Reg. 34,296 (June 17, 2008).
    1.    The 2008 VGP
    Environmental groups, industry groups, and the State of Michigan
    challenged the 2008 VGP in a PFR filed in the United States Court of Appeals for
    the D.C. Circuit, arguing primarily that the 2008 VGP was inadequate because it
    contained only narrative provisions, not specific numeric limitations on
    discharges.  In March 2011, EPA settled this matter, agreeing to: (1) set ʺnumeric
    concentration‐based effluent limits for discharges of ballast water expressed as
    organisms per unit of ballast water volumeʺ; (2) set numeric effluent limits that
    ʺrepresent the applicable levels of technology‐based controlʺ; and (3) ʺinclude
    more stringent water quality‐based effluent limitationsʺ if needed to satisfy
    ‐16‐
    applicable water quality standards.  Settlement Agreement ¶¶ 9‐13, NRDC v.
    EPA, No. 09‐1089 (D.C. Cir. Mar. 8, 2011), ECF No. 1296922.
    2.    The Creation of New Standards
    To create these new, more specific standards, EPA enlisted the help
    of its own Science Advisory Board (the ʺSABʺ) and the National Research
    Council/National Academy of Sciences Committee on Assessing Numeric Limits
    for Living Organisms in Ballast Water (the ʺNAS Committeeʺ).  EPA posed a
    different question to each scientific body.
    a.     The SAB
    In 2010, EPA asked the SAB to ʺprovide advice on technologies and
    systems to minimize the impacts of invasive species in vessel ballast water
    discharge.ʺ  App. at 599.  Specifically, the SAB looked at four issues: (1) the
    performance of shipboard systems with available effluent testing data; (2) the
    potential performance of shipboard systems without reliable testing data; (3)
    system development for the shipboard systems identified in issues 1 and 2; and
    (4) the development of reliable information about the status of ballast water
    treatment technologies and system performance.  In considering these questions,
    the SAB was to take into account The International Convention for the Control and
    ‐17‐
    Management of Shipsʹ Ballast Water and Sediments (the ʺIMO Standardʺ), adopted
    by the International Maritime Organization in 2004, which set certain
    concentration‐based ballast water effluent limits.  Id. at 610.7
    In July 2011, the SAB issued its report Efficacy of Ballast Water
    Treatment Systems: A Report by the EPA Science Advisory Board (the ʺSAB Reportʺ).
    The SAB identified fifty‐one ballast‐water treatment systems, with five categories
    of shipboard systems that could reliably achieve the IMO Standard.  Id. at 601. 8
    The SAB found that none of the systems could meet standards 100 or 1,000 times
    greater than the IMO Standard.  Id. at 602.  The SAB also found that none of the
    fifty‐one shipboard treatments identified could reliably achieve a ʺno living
    organismʺ standard.  Id.
    7  The Coast Guard proposed the same standard in a rulemaking in 2011
    pursuant to its authority under the National Invasive Species Act.  See Standards for
    Living Organisms in Shipsʹ Ballast Water Discharged in U.S. Waters, 74 Fed. Reg. 44,632
    (Aug. 28, 2009).  In 2012, the Coast Guard finalized the rule, entitled Standards for Living
    Organisms in Ships’ Ballast Water Discharged in U.S. Waters.  77 Fed. Reg. 17,254 (Mar. 23,
    2012).
    8  The five categories were: (1) deoxygenation + cavitation; (2) filtration +
    chlorine dioxide; (3) filtration + UV; (4) filtration + UV + TiO2; and (5) filtration +
    electro‐chlorination.
    ‐18‐
    b.     The NAS Committee
    EPA created the NAS Committee to examine ʺthe relationship
    between the concentration of living organisms in ballast water discharges and
    the probability of nonindigenous organisms successfully establishing
    populations in U.S. waters.ʺ  Id. at 235.
    In its June 2011 report, Assessing the Relationship Between Propagule
    Pressure and Invasion Risk in Ballast Water (the ʺNAS Reportʺ), the NAS
    Committee concluded (1) there was ʺno significant relationship between ballast
    volume and invasions,ʺ and (2) ʺ[t]he current state of science does not allow a
    quantitative evaluation of the relative merits of various discharge standards in
    terms of invasion probability.ʺ  Id. at 363.  Essentially, the Committee said that it
    was unable to establish a reliable numeric limit on discharges that would
    guarantee protection against invasive species, other than zero.
    3.    The 2013 VGP
    On March 28, 2013, EPA issued the 2013 VGP, the permit now before
    us, allowing vessels to discharge ballast water subject to certain limitations on
    the living organisms in the discharge.  Final National Pollutant Discharge
    Elimination System (NPDES) General Permit for Discharges Incidental to the Normal
    ‐19‐
    Operation of a Vessel, 78 Fed. Reg. 21,938 (Apr. 12, 2013).9  This constituted final
    action on the permit pursuant to section 402(a) of the CWA, 33 U.S.C. § 1342(a).
    The 2013 VGP included TBELs, WQBELs, and monitoring and reporting
    requirements.
    a.         TBELs
    As discussed above, TBELs impose effluent limitations on a point
    source based on how much technology is able to reduce the amount of a
    pollutant at issue.  See 33 U.S.C. §§ 1311(b), (e), 1314(b).  In this instance, EPA
    chose to set the TBELs at the IMO Standard, which requires:
    (1)        Limiting discharges of organisms 50 micrometers or larger to
    a concentration of fewer than 10 living organisms per cubic
    meter of ballast water;
    (2)        Limiting discharges of organisms less than 50 micrometers
    and greater than or equal to 10 micrometers to concentrations
    of fewer than 10 living organisms per milliliter (ʺmLʺ)of
    ballast water; and
    (3)        Limiting discharges of three types of pathogen and pathogen
    indicators: (1) Vibrio cholerae: fewer than 1 colony forming unit
    (ʺcfuʺ) per 100 mL; (2) Escherichia coli (ʺE. coliʺ): fewer than 250
    cfu  per 100 mL; and (3) intestinal enterococci: fewer than 100
    cfu per 100 mL.
    9 EPA issued the draft NPDES VGP on November 30, 2011 with a 75‐day
    notice‐and‐comment period.  The public comment period ended on February 21, 2012.
    The 2013 VGP replaced the 2008 VGP, which expired on December 19, 2013.
    ‐20‐
    Vessel General Permit for Discharges Incidental to the Normal Operation of Vessels
    (VGP): Authorization to Discharge Under the National Pollutant Discharge Elimination
    System § 2.2.3.5, at 29 (Mar. 28, 2013), available at http://water.epa.gov/polwaste/
    npdes/vessels/upload/vgp_permit2013.pdf.  The VGP did not set standards for
    other ʺsmallʺ organisms, such as bacteria or viruses.
    b.     WQBELs
    The WQBELs in the 2013 VGP require: (1) oceangoing vessels
    entering the Great Lakes to continue to perform ballast water exchanges, and (2)
    all vessels to control discharges ʺas necessary to meet applicable water quality
    standards in the receiving water body or another water body impacted by [the]
    discharges.ʺ  VGP § 2.2.3.7, at 43, § 2.3.1, at 59.
    c.     Monitoring and Reporting Requirements for TBELs and WQBELs
    As noted above, NPDES permits must contain conditions that
    require both monitoring and reporting of monitoring results of TBELs and
    WQBELs to ensure compliance with water quality standards.  See 33 U.S.C.
    § 1342(a)(2); 40 C.F.R. § 122.44(i)(1)‐(2).
    i)     Monitoring and Reporting Requirements for TBELs
    In the 2013 VGP, EPA established the following monitoring
    requirements for TBELs:
    ‐21‐
    (1)        that vessels monitor the functionality of their ballast water
    treatment systems, if installed; and
    (2)        that vessels monitor the concentrations of the two ʺindicatorʺ
    bacteria, E. coli and enterococci.10
    VGP § 2.2.3.5.1.1.2, at 30, § 2.2.3.5.1.1.4, at 31‐32.
    The first requirement is known as functionality monitoring.  Under
    this requirement, a ballast water treatment program is considered to be in
    compliance if it is ʺoperating according to the manufacturersʹ requirements.ʺ
    App. at 96.
    The second requirement is known as effluent biological organism
    monitoring.  Under this requirement, vessels must collect small‐volume samples
    and analyze them for concentrations of two indicator pathogens.  This is
    required between one and four times a year depending on the treatment system.
    ii)        Monitoring and Reporting Requirements for WQBELs
    The only monitoring required for WQBELs is that ships report the
    ʺexpected date, location, volume, and salinity of any ballast water to be
    10  EPA established effluent limits for Vibrio cholerae, but did not require
    monitoring in this respect because the ʺmonitoring of this parameter would generally
    not result in the detection of the presence of this pathogen.ʺ  App. at 99.  The 2013 VGP
    also contains a third requirement that vessels with treatment systems that add or
    generate biocides, such as chlorine or ozone, to kill organisms must monitor ballast
    water discharges for residual biocides.  Id. at 103.  This requirement does not implicate
    TBELs.
    ‐22‐
    discharged.ʺ  VGP § 4.3, at 72 (emphasis added).  Permittees are not required to
    report actual locations, volumes, or composition of ballast water to be
    discharged.
    d.   Lakers
    The 2013 VGP requires all Lakers to comply with non‐numeric
    technology‐based control measures, like ballast water exchange and other BMPs
    found in VGP § 2.2.3.3.  App. at 85; see VGP § 2.2.3.3, at 27‐28.  Lakers are also
    subject to three ballast water management measures found in VGP § 2.2.3.4: (1)
    conducting an annual assessment of sediment accumulations; (2) minimizing the
    amount of water taken in nearshore environments; and (3) adequately
    maintaining sea chest screens, which keep larger organisms like fish out of
    ballast tanks.  VGP § 2.2.3.4, at 28‐29.  In addition, all Lakers built on or after
    January 1, 2009, must comply with VGP § 2.2.3.5, which sets numeric ballast
    water discharge limits.  VGP § 2.2.3.5.3.3, at 39.
    DISCUSSION
    A.    Standard of Review
    We review a NPDES permit under the Administrative Procedure
    Act to determine whether EPAʹs actions were ʺarbitrary, capricious, an abuse of
    ‐23‐
    discretion, or otherwise not in accordance with law.ʺ  5 U.S.C. § 706(2)(A).  To
    determine whether the agencyʹs actions were ʺarbitrary and capricious,ʺ we
    consider whether the agency
    ʹrelied on factors which Congress has not intended it to
    consider, entirely failed to consider an important aspect
    of the problem, offered an explanation for its decision
    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.ʹ
    Islander E. Pipeline Co. v. McCarthy, 525 F.3d 141, 150‐51 (2d Cir. 2008) (quoting
    Motor Vehicle Mfrs. Assʹn. v. State Farm Mut. Ins. Co., 463 U.S. 29, 43 (1983)).  We
    must be ʺsatisfied from the record that ʹthe agency . . . examine[d] the relevant
    data and articulate[d] a satisfactory explanation for its action.ʹʺ  Id. at 151
    (quoting State Farm, 463 U.S. at 43).  An agencyʹs action is lawful ʺonly if it rests
    ʹon a consideration of the relevant factors.ʹʺ  Michigan v. EPA, 135 S. Ct. 2699, 2706
    (2015) (quoting State Farm, 463 U.S. at 43).  We afford the agencyʹs decision
    greater deference regarding factual questions involving scientific matters in its
    area of technical expertise.  See Balt. Gas & Elec. Co. v. NRDC, 462 U.S. 87, 103
    (1983); Envtl. Def. v. EPA, 369 F.3d 193, 204 (2d Cir. 2004).
    In addition, judicial review of statutory interpretation by an agency
    is governed by Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 842‐45 (1984).  Under
    ‐24‐
    Chevron, we must first determine ʺwhether Congress has directly spoken to the
    precise question at issueʺ in the CWA.  Id. at 842.  If so, we must give effect to the
    unambiguously expressed intent of Congress.  Id. at 842‐43.  ʺ[I]f the statute is
    silent or ambiguous with respect to the specific issue, the question for the court is
    whether the agencyʹs answer is based on a permissible construction of the
    statute.ʺ  Id. at 843.  As the Supreme Court held in Michigan v. EPA, ʺ[e]ven under
    this deferential standard, however, ʹagencies must operate within the bounds of
    reasonable interpretation.ʹʺ  135 S. Ct. at 2707 (quoting Util. Air Regulatory Grp. v.
    EPA, 134 S. Ct. 2427, 2442 (2014)).  We also grant deference to EPAʹs
    interpretation of its own regulations ʺunless that interpretation is ʹplainly
    erroneous or inconsistent with the regulation.ʺ  Chase Bank USA, N.A. v. McCoy,
    131 S. Ct. 871, 880 (2011) (quoting Auer v. Robbins, 519 U.S. 452, 461 (1997)).
    B.    Petitionersʹ Challenge
    Here, petitioners challenge EPAʹs issuance of the 2013 VGP as
    arbitrary and capricious, and not in accordance with law, on a number of
    grounds.
    First, petitioners argue that the TBELs are arbitrary and capricious.
    Specifically, petitioners assert that EPA acted arbitrarily and capriciously and not
    ‐25‐
    in accordance with the law when it: (1) selected the IMO standard as the
    standard for the TBELs; (2) failed to consider onshore treatment, limiting
    consideration to shipboard treatment; (3) failed to include numeric TBELs for
    viruses and protists; and (4) exempted Lakers built before 2009 (ʺpre‐2009
    Lakersʺ) from the numeric TBELs of the 2013 VGP.
    Second, petitioners argue that EPA acted arbitrarily and capriciously
    and not in accordance with the law in choosing narrative WQBELs, rather than
    numeric WQBELs, because, among other things, the narrative standard is too
    imprecise to guarantee compliance with water quality standards.
    Finally, petitioners argue that EPAʹs monitoring and reporting
    requirements for TBELs and WQBELs are not in accordance with the law because
    they were inadequate to guarantee compliance.
    1.     TBELs
    Under the CWA, EPA must apply BAT in establishing pollution
    controls for ballast water discharge.  BAT requires the ʺapplication of the best
    available technology economically achievable for such category or class, which will
    result in reasonable further progress toward the national goal of eliminating the
    discharge of all pollutants.ʺ  33 U.S.C. § 1311(b)(2)(A)(emphasis added).  BAT
    also requires ʺa commitment of the maximum resources economically possible to
    ‐26‐
    the ultimate goal of eliminating all polluting discharges.ʺ  EPA v. Natʹl Crushed
    Stone Assʹn, 449 U.S. 64, 74 (1980).
    We hold that in failing to set TBELs that reflected BAT in the 2013
    VGP, EPA acted arbitrarily and capriciously in a number of respects.
    a.     The IMO Standard
    Petitioners argue that EPA failed to apply BAT when it chose the
    IMO Standard for TBELs in the 2013 VGP.  They allege that EPA chose the IMO
    Standard first, and then worked ʺbackwardsʺ to determine which systems could
    achieve that standard.  NWEA Br. at 36.  In doing so, they contend, EPA
    improperly restricted the SABʹs inquiry to whether certain technologies would
    meet what petitioners describe as ʺexisting international consensus standards.ʺ
    Id.  According to petitioners, EPA should have first considered what ʺavailableʺ
    technology was capable of achieving, and then created standards based on that
    capability.  As a result, petitioners contend, EPAʹs standard did not achieve
    greater reductions in pollution discharges that were achievable with current
    technology.
    We agree.  EPA acted arbitrarily and capriciously when it chose the
    IMO Standard without adequately explaining why standards higher than the
    IMO Standard should not be used given available technology.
    ‐27‐
    In choosing the IMO Standard, EPA overlooked crucial portions of
    the SAB Report.  The SAB identified a number of technologies that can achieve
    standards higher than IMO for one or more organism sizes, including all five of
    the technologies identified as also meeting the IMO Standard.11  The SAB
    acknowledged that ʺthese same five systems have the potential to meet a 10x
    IMO D‐2/ Phase 1 standard [i.e., the IMO Standard] in the near future.ʺ  App. at
    636.  In describing the performance of those systems, the SAB concluded that
    each would require only ʺreasonable/feasible modifications.ʺ  Id. at 629‐30,632.
    Indeed, according to the SAB Report, the Ecochlor, BalPure, and PeraClean
    systems can meet 100 times IMO for medium organisms, and Ecochlor can meet
    10 times IMO for large organisms.
    EPA should not have adhered to the IMO Standard without
    explanation when technologies could have exceeded IMO.  Indeed, seeking to
    find systems that are capable of doing better than the current standard is in
    keeping with the technology‐forcing aspect of the CWA.  See NRDC, 822 F.2d at
    11  The record further demonstrates that existing shipboard technology can
    meet a standard between IMO and 10 times IMO.  For instance, Hyde Marine Guardian
    has tested at 1.4 times IMO for large organisms; Optimarin has tested at 7.7 times IMO
    for large organisms; and Alfa Laval/AlfaWall PureBallast has tested at 4.5 times IMO for
    large organisms, and at 3.7 times IMO for medium organisms.
    ‐28‐
    124.  EPA should have first looked at the available ballast water technologies as
    identified by the SAB Report.  Then, finding that those technologies could exceed
    the IMO Standard, EPA should have adjusted its standard accordingly, or
    explained why it would not.  See Islander E. Pipeline, 525 F.3d at 151 (holding that
    agency must ʺexamine[] the relevant data and articulate[] a satisfactory
    explanation for its actionʺ (quoting State Farm, 463 U.S. at 43)(alterations
    omitted)).
    EPAʹs counterargument that no more was necessary because it did
    not limit the SAB to considering the IMO Standard is unavailing.  EPA insists
    that it gave the SAB a list of potential regulatory limits, and then asked the SAB
    to identify the systems that could reliably meet those limits.  In support, EPA
    points to its Charge Question 1 to the SAB, which asked the SAB to identify
    ʺdischarge standards that the available data [about existing systems] credibly
    demonstrate can be reliably achieved.ʺ  App. at 607.  EPA argues that in response
    to this charge, the SAB Report supports the conclusion that, ʺ[b]ased upon the
    data available, no current ballast water treatment technologies were considered
    likely to meet standards more stringent than the IMO D‐2/Phase Iʺ standards.  Id.
    at 91.
    ‐29‐
    While it is true that EPA did not strictly limit the SABʹs
    consideration to the IMO Standard, EPA is incorrect in suggesting that the SAB
    Report supports the conclusion that no system could meet standards stricter than
    the IMO Standard.  Id.12  To the contrary, the record contradicts EPAʹs assertion
    that treatment systems that exceed the IMO Standard are not ʺavailable.ʺ  In fact,
    as noted above, systems that exceed the IMO Standard are available
    Accordingly, by failing to consider adequately a standard more
    stringent than IMO, EPA failed to set permit limits that reflect BAT.  See 33 U.S.C.
    § 1314(b)(2); Natʹl Crushed Stone, 449 U.S. at 74 (BAT requires ʺa commitment of
    the maximum resources economically possible to the ultimate goal of eliminating
    all pollution dischargesʺ); FMC Corp. v. Train, 539 F.2d 973, 983‐84 (4th Cir. 1976)
    The SAB actually stated that it could not reliably test for standards 100 or
    12
    1000 times more stringent than the IMO Standard:
    The Panel also concludes that the [IMO Standard] . . . [is]
    currently measurable, based on data from land‐based and
    shipboard testing.  However, current methods (and
    associated detection limits) prevent testing of BWMS to any
    standard more stringent than [the IMO Standard] and make
    it impracticable for verifying a standard 100 or 1000 times
    more stringent.
    Id. at 601.  While we agree that we must defer to EPAʹs conclusions regarding the
    technical feasibility of testing for standards 100 or 1000 times more stringent than the
    IMO Standard, there is nothing in the record to suggest that it would not be possible to
    test for twice or even ten times the IMO Standard.
    ‐30‐
    (upholding EPAʹs decision to set BAT based on data from a single pilot plant).  In
    doing so, EPA acted arbitrarily and capriciously and not in accordance with law
    in choosing the IMO standard for the TBELs in the 2013 VGP.
    b.     Onshore Ballast Water Treatment
    Petitioners also argue that EPA arbitrarily and capriciously limited
    its consideration to shipboard treatments, failing to consider onshore treatment.
    Petitioners argue that onshore facilities used in other industries, such as sewage
    treatment plants and drinking water treatment plants, were reasonable
    alternatives to shipboard treatment that should have been considered.  EPA
    concedes it directed the Board to ʺfocus its limited time and resources on the
    status of shipboard treatment systems because such systems were either ʹin
    existence or in the development process.ʹʺ  EPA Br. at 56‐57 (quoting SAB
    Report).  It argues that onshore treatment was not ʺavailable,ʺ primarily because
    no onshore system was yet in existence.
    While it is true that no onshore systems existed then ‐‐ unsurprising
    considering ballast water treatment was not required at all until the effective date
    of the 2008 VGP – the record suggests that such onshore systems were
    technologically possible at that time.  Yet, EPA chose to curtail discussion about
    onshore systems and failed to develop information necessary to evaluate their
    ‐31‐
    availability.  We conclude that by failing to consider onshore ballast water
    systems, EPA acted arbitrarily and capriciously.
    What does ʺavailableʺ mean?  As courts have interpreted the term in
    the CWA context, technologies that could be used for a particular discharge,
    even if they are not currently being used by that industry, are ʺavailable.ʺ  As
    the Fourth Circuit noted,
    The model technology [under consideration] may exist
    at a plant not within the . . . industry [at issue].
    Congress contemplated that EPA might use technology
    from other industries to establish the Best Available
    Technology.  Progress would be slowed if EPA were
    invariably limited to treatment schemes already in force
    at the plants which are the subject of the rulemaking.
    Congress envisioned the scanning of broader horizons
    and asked EPA to survey related industries and current
    research to find technologies which might be used to
    decrease the discharge of pollutants.
    Kennecott, 780 F.2d at 453 (emphasis added) (citation omitted).  This Court held
    similarly in Hooker Chemicals & Plastics Corp. v. Train,
    That no plant in a given industry has adopted a
    pollution control device which could be installed does
    not mean that that device is not ʺavailable.ʺ  Congress
    did not intend to permit continuance of pollution by
    industries which have failed to cope with and attempt
    to solve the problem of polluted water.
    ‐32‐
    537 F.2d 620, 636 (2d Cir. 1976); see also Cal. & Hawaiian Sugar Co. v. EPA, 553 F.2d
    280, 286 (2d Cir. 1977) (technology used ʺʹin other industries with similar raw
    waste characteristicsʹʺ was ʺavailableʺ (quoting Liquid and Crystalline Cane Sugar
    Refining Subcategory, 39 Fed. Reg. 10,522, 10,522 (1974))); Am. Petrol. Inst. v. EPA,
    858 F.2d 261, 264‐65 (5th Cir. 1988) (holding that a process can be ʺdeemed
    ʹavailableʹ even if it is not in use at allʺ because ʺ[s]uch an outcome is consistent
    with Congressʹ intent to ʹpush pollution control technologyʹʺ (quoting Assʹn of
    Pac. Fisheries v. EPA, 615 F.2d 794, 816 (9th Cir. 1980); Weyerhaeuser Co. v. Costle,
    590 F.2d 1011, 1061 (D.C. Cir. 1978))).
    For a technology in one industry to be ʺavailableʺ in a second
    industry: (1) the transfer technology must be available within the first industry;
    (2) the transfer technology must be transferable to the second industry; and (3) it
    must be reasonably predictable that the technology, if used in the second
    industry, will be capable of removing the increment required by the effluent
    standards.  See Kennecott, 780 F.2d at 453 (citing Tannersʹ Council, 540 F.2d at
    1192); CPC Intʹl Inc. v. Train, 515 F.2d 1032, 1048 (8th Cir. 1975); Hooker Chems.,
    537 F.2d at 636 (ʺBut even if technology which is not presently in use can be
    treated as available and achievable, there must be some indication in the
    ‐33‐
    administrative record of the reasons for concluding that such technology is
    feasible and may reasonably be expected to yield the effluent reduction
    mandated when applied to the particular industry.ʺ).  For example, in Kennecott,
    the Fourth Circuit upheld EPAʹs use of manufacturing technology from one
    industry as part of a BAT determination for treating wastewater in a different
    industry.  780 F.2d at 453‐54.
    Here, we cannot evaluate whether onshore technology should be
    considered ʺavailableʺ because the record does not contain a full discussion of
    onshore treatment.  This lack of information about onshore facilities, however, is
    a problem of EPAʹs own making because EPA went to great lengths to foreclose
    discussion of onshore treatment both by expressly limiting the SABʹs mandate to
    studying shipboard treatment technology and consistently opposing any attempt
    by the SAB to consider onshore treatment.
    EPAʹs effort to curtail discussion of onshore treatment is well
    documented in the record.  In a letter dated February 10, 2012, thirteen scientists,
    eight of whom were members of the SAB and six of whom were members of the
    NAS Committee,13 including the Chair of the NAS Committee, stated that the
    13         One person was a member of both the SAB and the NAS Committee.
    ‐34‐
    SAB ʺnever actually addressed the question of what is the best treatment that
    available technology can achieveʺ because EPA limited them to the narrower
    question of ʺwhether shipboard treatment systems could meet certain specific sets of
    standards.ʺ  App. at 740.  Furthermore, the scientists assert that their attempts to
    consider onshore treatments were actively thwarted by EPA:
    During the SAB Panel meetings and discussions, some
    members of the Panel attempted to develop and include
    in the Panel report a more detailed assessment of
    onshore treatment, including its cost impacts, and an
    assessment of the full capability of shipboard treatment
    . . . .  [T]hese assessments would have further
    demonstrated that available technology can achieve
    levels of treatment beyond what the EPA has proposed
    as controls.  The EPA Office of Water, however,
    consistently opposed including such information in the
    report.  As a result, some relevant information and
    analysis that could have been developed by the Panel
    was not, and some of what was developed by Panel
    members was excluded or deleted from the final report.  If
    there was less information developed on these issues and less
    provided in the report than the EPA considers sufficient, it is
    in large part because the EPA Office of Water opposed the
    development and inclusion of such information.
    Id. at 744 (emphases added).
    In light of these facts, we cannot well credit EPAʹs assertion that it
    lacked information to support a finding that onshore facilities were ʺavailable.ʺ
    While EPA states that it was ʺunaware of any onshore treatment facility currently
    ‐35‐
    available in the U.S. that is capable of meeting the VGPʹs § 2.2.3.5 ballast water
    discharge standards,ʺ and that it did not ʺreceive information indicating they are
    or would become available over the term of the VGP,ʺ id. at 544, in fact EPA
    turned a blind eye to significant information about onshore treatment.
    Indeed, the lack of information about the ʺavailabilityʺ of onshore
    treatment is due in large part to EPAʹs arbitrary and capricious decision to
    oppose developing such information.  As a result, the TBELs were based on an
    incomplete record ‐‐ one lacking meaningful discussion of an ʺavailableʺ
    treatment, namely onshore treatment.  See Humana of Aurora, Inc. v. Heckler, 753
    F.2d 1579, 1583 (10th Cir. 1985) (agency action is arbitrary and capricious when
    based on a flawed study); Tex. Oil & Gas Assʹn v. EPA, 161 F.3d 923, 935 (5th Cir.
    1998) (ʺA regulation cannot stand if it is based on a flawed, inaccurate, or
    misapplied study.ʺ); Almay, Inc. v. Califano, 569 F.2d 674, 682 (D.C. Cir. 1977)
    (rejecting regulation produced ʺon the basis of the flawed surveyʺ).  Put another
    way, EPAʹs refusal to consider onshore treatment ʺentirely fail[s] to consider an
    important aspect of the problemʺ and ʺoffer[s] an explanation for its decision that
    runs counter to the evidence before the agency.ʺ  Islander E. Pipeline, 525 F.3d at
    150‐51; see Tannersʹ Council, 540 F.2d at 1191 (ʺ[T]he agency must fully explicate
    ‐36‐
    its course of inquiry, its analysis, and its reasoning.ʺ); see also State Farm, 463 U.S.
    at 43; Hooker Chems., 537 F.2d at 636.  Hence, it is arbitrary and capricious.
    In fact, the SAB Report points out a number of reasonably
    predictable advantages to onshore treatment.  The SAB Report states:
    Use of reception facilities for the treatment of ballast
    water appears to be technically feasible (given
    generations of successful water treatment and sewage
    treatment technologies), and is likely to be more reliable
    and more readily adaptable than shipboard treatment.
    App. at 605; see also id. at 694.  The SAB Report also notes that onshore treatment
    has a number of advantages over shipboard treatment because onshore facilities
    are not subject to problems such as limited space, small and overburdened crews,
    vibrations, weight allowances, limited power, ship instability, and greater
    corrosion rates.  Id. at 678‐80.  Regarding ship crews in particular, studies have
    shown that ʺmany of these crews are already overburdened,ʺ ʺ[o]peration by
    trained, dedicated personnel in reception facilities would likely result in more
    reliable performance,ʺ and ʺ[m]aintenance and repair work are more likely to be
    done reliablyʺ as well.  Id. at 681.  Onshore treatment can also be more effective
    by using superior technologies that are not available for shipboard treatment,
    such as settling tanks, granular filtration, and membrane filtration.  Id. at 680‐81.
    Indeed, EPA cites a number of studies that conclude that onshore treatment
    ‐37‐
    facilities are a technically feasible option.  Id. at 107.  These studies date back to
    1992, and proceed with some regularity thereafter‐‐ 1996, 1999, 2000, 2002, 2007,
    2008.
    Moreover, onshore treatment would not necessarily be slower than
    shipboard treatment to implement.  The SAB estimated that onshore
    implementation would take up to thirty months, while EPA allowed eight years
    to phase‐in shipboard implementation.  See id. at 684.  Nor would onshore
    treatment necessarily be more expensive than shipboard treatment.  Regional
    economic studies suggest that ʺtreating ballast water in reception facilities would
    be at least as economically feasible as shipboard treatment.ʺ  Id. at 694.  In
    addition, the cost of monitoring and enforcement is likely to be lower with a
    smaller number of reception facilities compared with a larger number of
    shipboard systems.  Id. at 605, 694.  The Coast Guard also found that onshore
    treatment was generally less expensive per metric ton of ballast water than
    shipboard treatment.  Id. at 679.
    Of course, onshore treatment has many costs, including the cost of
    retrofitting vessels for onshore facilities, particularly ships from outside the
    United States, and the cost of shipping delays created by the time it takes to
    ‐38‐
    discharge ballast onshore (though presumably shipboard treatment is not
    instantaneous).  Costs alone, however, cannot determine BAT.  See 33 U.S.C.
    § 1314(b)(2)(B).  Furthermore, EPA failed to perform the economic analysis
    required to determine relative costs of the differing technologies in reaching its
    conclusion that onshore treatment was not economically achievable.  See
    Waterkeeper All., 399 F.3d at 516 (ʺ[T]he Administrator is obligated to ʹinquire into
    the initial and annual costs of applying the technology and make an affirmative
    determination that those costs can be reasonably borne by the industry.ʹʺ
    (quoting Riverkeeper, Inc. v. EPA, 358 F.3d 174, 195 (2d Cir. 2004))); Natʹl Wildlife
    Fedʹn v. EPA, 286 F.3d 554, 563 (D.C. Cir. 2002) (ʺAlthough its analysis may be
    general, EPA ʹhas the heaviest of obligations to explain and expose every step of
    its [cost‐benefit] reasoning.ʹ . . . This duty to explain arises out of the need for
    reviewing courts to be able to discern the basis for EPAʹs decision.ʺ (internal
    citations omitted) (quoting Am. Lung Assʹn v. EPA, 134 F.3d 388, 392 (D.C. Cir.
    1998))).
    In light of these observations, the SAB and NAS Committee
    scientists concluded that ʺEPA should conduct a comprehensive analysis
    comparing biological effectiveness, cost, logistics, operations, and safety
    ‐39‐
    associated with both shipboard [treatment] and reception facilities.ʺ  App. at 606.
    If that analysis ʺindicate[d] that treatment at reception facilities is both
    economically and logistically feasible and is more effective than shipboard
    treatment systems, it should be used as the basis for assessing the ability of
    available technologies to . . . meet a given discharge standard.ʺ  Id.  EPA chose
    not to do so because the SAB ʺdid not specify a timetable for that complex
    endeavor or suggest that is was possible to complete such an analysis in time to
    inform the impending VGP.ʺ  EPA Br. at 58‐59.  We do not find that answer
    compelling.  There is no impediment to engaging in further study, and further
    study may advance the goals of the CWA.
    Thus, EPA could have well found that onshore treatment was
    ʺavailable.ʺ  Indeed, EPAʹs failure to consider onshore treatment is inconsistent
    with the CWAʹs mandate that TBELs be technology‐forcing.  Congress designed
    the CWA to force agencies and permittees to adopt technologies that achieve the
    greatest reductions in pollutants.  See NRDC, 822 F.2d at 124 (holding that CWA
    seeks ʺnot only to stimulate but to press development of new, more efficient and
    effective technologies,ʺ which is the ʺessential purpose of this series of
    progressively more demanding technology‐based standardsʺ).  As Judge Starr
    ‐40‐
    noted in NRDC, ʺthe most salient characteristic of this statutory scheme,
    articulated time and again by its architects and embedded in the statutory
    language, is that it is technology‐forcing.ʺ  Id. at 123.
    EPAʹs decision on this issue matters.  As the SAB scientists pointed
    out, EPAʹs choice of system in this permit will have a long‐term impact:
    [S]hipboard treatment and onshore treatment represent
    distinct approaches to ballast water management that
    would each require different large investments in
    infrastructure. . . .  Thus we are almost certain to be stuck
    for a very long time with whichever approach is used as the
    BAT in setting discharge standards in 2013.  It is thus of
    the utmost urgency that a fair and thorough comparison
    of the two approaches be made at this time.
    App. at 744‐45 (emphasis added).  We conclude that EPA failed to give fair and
    thorough consideration to both onshore and shipboard treatment systems in
    setting the standard in the 2013 VGP, and we remand to EPA to give full
    consideration to the issue now.
    c.     Viruses and Protists
    Petitioners also complain about the lack of numeric TBELs for
    viruses and protists (primarily single‐celled organisms).  EPA argues, however,
    that it could not set TBELs for viruses and protists in the 2013 VGP because EPA
    could not yet identify ʺsuitable standardized test organisms and/or surrogate
    ‐41‐
    parameters to determine treatment system performance at removing or
    eliminating viruses and protists and which also can be used in establishing
    technology‐based discharge limitations.ʺ  App. at 486; see also App. at 495 (ʺEPA
    does not believe that there are sufficient data available to establish numeric limits
    for protists or other bacteria.ʺ).
    We agree that it was not arbitrary and capricious for EPA to decline
    to set TBELs for organisms for which it is unable to test and for which it has
    insufficient data to set numeric limits.  See Balt. Gas & Elec. Co. v. NRDC, Inc., 462
    U.S. 87, 103 (1983) (ʺ[A] reviewing court must remember that [where the agency]
    is making predictions, within its area of special expertise, at the frontiers of
    science . . . as opposed to simple findings of fact, a reviewing court must
    generally be at its most deferential.ʺ).  Petitioners have not demonstrated that
    sufficient data are available.  EPA has represented that it will consider including
    numeric TBELs for viruses and protists in the next version of the VGP.  App. at
    486.  This is sufficient.
    d.     Pre‐2009 Lakers
    Petitioners allege that EPAʹs decision to exempt Lakers built before
    January 1, 2009 from numeric effluent limits of VGP § 2.2.3.5 was arbitrary and
    capricious.  EPA based this decision on its finding that there was no treatment
    ‐42‐
    technology ʺavailableʺ for these vessels either onboard or onshore.  App. at 115‐
    16.  EPA expressed concern about the difficulty of finding effective onboard
    systems for pre‐2009 Lakers due to their ʺunique operational and design
    constraints,ʺ such as the large volumes of fresh cold water they require, the short
    duration of their trips, their high pumping rates, and their uncoated ballast
    tanks.  Id. at 116.14  In reaching that conclusion, EPA relied on the SAB Report,
    which advised that ʺspecific constraints can greatly limit treatment optionsʺ for
    Lakers.  Id. at 638.  EPA also cited the costs of implementing these systems.  Id. at
    116.
    We agree with petitioners that exempting the pre‐2009 Lakers was
    arbitrary and capricious.  First, the lack of supply of updated shipboard systems
    is not a legitimate reason to exempt pre‐2009 Lakers from the 2013 VGP, as,
    again, the purpose of BAT is to force technology to keep pace with need.  See
    NRDC, 822 F.2d at 124.
    Second, EPAʹs decision was based on a flawed record that failed to
    consider an important aspect of the problem, namely the possibility of onshore
    14For example, certain treatment methods, such as electro‐chlorination and
    ozonation, may only be effective in salt water, and others that use oxidizing chemicals
    may increase corrosion rates in uncoated tanks.  Id. at 638.
    ‐43‐
    treatment.  See Islander E. Pipeline, 525 F.3d at 150‐51.  EPA should have
    considered the comparable ʹcost of achieving such effluent reductionsʹ through
    onshore treatment versus shipboard treatment, rather than merely dismissing
    onshore treatment.  EPA disregarded the SABʹs recommendation that onshore
    treatment would benefit pre‐2009 Lakers that are ʺengaged solely in regional
    trade.ʺ  Id. at 684.  The SAB points out that the space and power constraints
    posed by pre‐2009 Lakers are ʺlargely absent in reception facilities.ʺ  Id. at 680.
    EPAʹs foreclosure of considering onshore treatments for pre‐2009 Lakers ‐‐ and
    indeed, all Lakers ‐‐ seems shortsighted.  See supra at 31‐42.
    Third, EPA imposed the 2013 VGP on Lakers built after 2009, even
    though post‐2009 Lakers face many of the same challenges and constraints as
    pre‐2009 Lakers, such as their short voyages, high pumping rates, and freshwater
    environment. 15  While it is true that shipbuilders were on notice that post‐2009
    15  Intervenors argue that due to these constraints, ballast water treatment is
    infeasible for all Lakers, regardless of when they were built.  EPA has concluded,
    however, that anyone building a ship designed to enter the market after 2009 was well
    aware of the impending VGP requirements, and could anticipate its impact on
    shipbuilding.  App. at 117.  Intervenors also contend that ships exclusively plying the
    Great Lakes do not pose a threat to water quality because they do not introduce any
    invasive species from outside the Great Lakes.  EPA has properly rejected this
    argument, noting that Lakers can spread or more rapidly distribute invasive species
    already present in the Great Lakes.  Id. at 501.
    ‐44‐
    Lakers would be subject to the 2013 VGP, in reality post‐2009 boats appear to be
    similarly situated to pre‐2009 Lakers.  See Islander E. Pipeline, 525 F.3d at 150‐51
    (agency decision is arbitrary and capricious when agency offers explanation for
    decision that runs counter to evidence before agency).  Thus, distinguishing pre‐
    2009 and post‐2009 Lakers was arbitrary and capricious.
    The SAB Report supports our conclusion.  Although the SAB Report
    acknowledged the limitations in treating ballast water from Lakers, it did not
    declare such treatment impossible.  Instead, the SAB concluded that in light of
    these limitations, ʺ[a] variety of environmental (e.g., temperature and salinity),
    operational (e.g., ballasting flow rates and holding times), and vessel design (e.g.,
    ballast volume and unmanned barges) parametersʺ should be considered in
    determining the treatment standards.  App. at 639.
    EPAʹs exemption of the pre‐2009 Lakers from the 2013 VGP was also
    arbitrary and capricious due to EPAʹs failure to conduct an appropriate and
    factually‐supported cost analysis.  Such an analysis might have shown that the
    cost of subjecting pre‐2009 Lakers to the 2013 VGP was not unreasonably high,
    or, alternatively, that onshore treatment was economically feasible.  For all these
    ‐45‐
    reasons, EPAʹs decision to exempt pre‐2009 Lakers from the 2013 VGP was
    arbitrary and capricious.
    2.         WQBELs
    Under the CWA, NPDES permits must include WQBELs where the
    TBELs are insufficient to maintain water quality standards.  33 U.S.C.
    § 1311(b)(1)(C); 40 C.F.R. § 122.44(d)(1)(vii)(A).  Here, EPA concluded that ʺeven
    at the IMO level of discharge, reasonable potential exists for such discharges to
    cause or contribute to violations of applicable water quality standards pursuant
    to 40 C.F.R. § 122.44(d)(1)(ii).ʺ  App. at 129.  To address this concern, EPA
    established WQBELs to ensure compliance with water quality standards.  EPA,
    however, chose to create narrative WQBELs because it believed numeric
    WQBELs were ʺinfeasibleʺ to calculate.16
    16 Federal regulation permits such limits to be expressed narratively if the
    calculation of numeric limits is ʺinfeasible.ʺ  40 C.F.R. § 122.44(k)(3).
    ‐46‐
    The WQBEL in the 2013 VGP states:
    Your discharge must be controlled as necessary to meet
    applicable water quality standards in the receiving
    water body or another water body impacted by your
    discharges.
    VGP § 2.3.1, at 59.  In defending this standard, EPA relied in part on the NAS
    Report, which stated that ʺ[t]he current state of science does not allow a
    quantitative evaluation of the relative merits of various discharge standards in
    terms of invasion probability.ʺ  App. at 363.
    Petitioners argue that this narrative WQBEL does not ensure
    compliance with water quality standards.  We agree.  This narrative standard is
    insufficient to give a shipowner guidance as to what is expected or to allow any
    permitting authority to determine whether a shipowner is violating water quality
    standards.  By requiring shipowners to control discharges ʺas necessary to meet
    applicable water quality standardsʺ without giving specific guidance on the
    discharge limits, EPA fails to fulfill its duty to ʺregulat[e] in fact, not only in
    principle.ʺ  Waterkeeper All., 399 F.3d at 498.  As this Circuit held in Waterkeeper
    Alliance, NPDES permits ʺmay issue only where such permits ensure that every
    discharge of pollutants will comply with all applicable effluent limitations and
    standards.ʺ  Id.  That is hardly the case here.  EPA itself notes that it only
    ‐47‐
    ʺgenerally expects that compliance with the [TBELs] . . . will control discharges as
    necessary to meet applicable water quality standards.ʺ  VGP § 2.3.1, at 59.  The
    WQBELs, although found by EPA to be required to supplement the TBELs, in fact
    add nothing.  The WQBELs do not state how they will ensure compliance.
    Even if determining the proper standard is difficult, EPA cannot
    simply give up and refuse to issue more specific guidelines.  See Am. Paper Inst.,
    Inc. v. EPA, 996 F.2d 346, 350 (D.C. Cir. 1993) (articulating that, even if creating
    permit limits is difficult, permit writers cannot just ʺthr[o]w up their hands and,
    contrary to the Act, simply ignore[] water quality standards including narrative
    criteria altogether when deciding upon permit limitationsʺ).  Scientific
    uncertainty does not allow EPA to avoid responsibility for regulating discharges.
    See Massachusetts v. EPA, 549 U.S. 497, 534 (2007) (ʺEPA [cannot] avoid its
    statutory obligation by noting the uncertainty surrounding various features of
    climate change and concluding that it would therefore be better not to regulate at
    this time.ʺ).
    Moreover, EPAʹs reliance on the NAS Report is misplaced.  EPA
    concedes that the NAS Committee ʺdid not conclude that it is infeasible to
    calculate water quality‐based effluent limits for ballast water discharges.ʺ  App.
    ‐48‐
    at 563.  Rather, the NAS Committee found that it could not formulate a precise
    standard.  In light of this uncertainty, it recommended further study of the issue.
    But EPA declined to engage in further study.  See App. at 363‐67.  For all these
    reasons, EPAʹs WQBELs were arbitrary and capricious.
    EPAʹs remaining counterarguments are unavailing.  First, EPA
    asserts that petitioners fail to offer examples of ʺmeaningful permit limitsʺ for
    WQBELs.  EPA Br. at 74.  EPA, however, could articulate specific actions that
    vessels would be required to take to protect against site‐specific threats.  For
    example, if EPA or the shipowner became aware of an unusual risk posed by a
    specific port, EPA could require vessels to take action to avoid such risk at that
    port, including not uptaking ballast water or not discharging into other ports the
    contaminated ballast water that was taken up.  Toward that end, EPA has
    included a set of specific best management practices in the 2013 VGP § 2.2.3.3,
    further demonstrating the viability of this approach.
    Second, EPA argues that under 40 C.F.R. § 122.44(k)(3) it may
    employ BMPs instead of ʺ[n]umeric effluent limitationsʺ for WQBELs when
    deriving numeric limitations is ʺinfeasible.ʺ  40 C.F.R. §§ 122.44(k)(3).  BMPs
    include ʺschedules of activities, prohibitions of practices, maintenance
    ‐49‐
    procedures, and other management practices to prevent or reduce . . . pollution.ʺ
    Id. § 122.2.  EPA argues that the narrative WQBEL is a BMP, and therefore it has
    discharged its duty under 40 C.F.R. § 122.2.17
    But EPAʹs narrative WQBEL does not qualify as a BMP, as it is
    neither a practice nor a procedure.  BMPs typically involve requirements like
    operating procedures, treatment requirements, practices to control runoff,
    spillage or leaks, sludge or waste disposal, or drainage from raw material
    storage; they can also be structural requirements including tarpaulins, retention
    ponds, or devices such as berms to channel water away from pollutant sources,
    and treatment facilities.  See NRDC v. Sw. Marine, Inc., 236 F.3d 985, 991 n.1 (9th
    Cir. 2000).  Examples of BMPs that have been accepted as substitutes for effluent
    limits include: nutrient management plans for concentrated animal feeding
    operations, see Waterkeeper All., 399 F.3d at 497, 502, filtration of stormwater
    runoff from ditches before it enters rivers and streams (by timber companies),
    and constructing roads with surfacing that minimizes sediment in runoff (by
    17 EPA also claims general prohibitions can be BMPs, citing 40 C.F.R. § 122.2,
    but does not offer an example of something as general as the WQBEL standard being
    found to be a BMP.  As discussed herein, the EPAʹs characterization is inconsistent with
    regulations that require WQBELs to ʺensure compliance.ʺ  40 C.F.R. § 122.4(d); see Auer,
    519 U.S. at 461 (holding interpretation may not be inconsistent with regulation).
    ‐50‐
    timber companies), see Decker v. Nw. Envtl. Def. Ctr., 133 S. Ct. 1326, 1338 (2013).
    The narrative standard here is nowhere as specific as any of these examples.
    Indeed, it requires nothing more of a shipowner than to meet the TBELs.  This
    interpretation is hardly consistent with the regulations that require WQBELS to
    ensure compliance.  See Auer, 519 U.S. at 461 (holding that courts should defer to
    agencyʹs interpretation of its own regulations if not plainly erroneous or
    inconsistent with the regulation).
    Third, EPA claims that WQBEL standards will be sufficiently
    maintained because EPA can take ʺcorrective actionsʺ after the permittee becomes
    aware of a violation.  App. at 160.  This is not reassuring.  The point of a permit is
    to prevent discharges that violate water quality standards before they happen.  See
    33 U.S.C. §§ 1311(b)(1)(C), 1342(a)(2); 40 C.F.R. §§ 122.4(d), 122.44(d)(1).
    ʺCorrective actionʺ is not an effective remedy in an invasive species context ‐‐ it is
    difficult to eradicate a colony of zebra mussels after they are established.  See,
    e.g., Great Lakes Sci. Ctr., U.S. Geological Survey, Zebra Mussels Cause Economic
    and Ecological Problems in the Great Lakes 2 (rev. 2011) (ʺOnce zebra mussels
    become established in a water body, they are impossible to eradicate with the
    technology available today.  Many chemicals kill zebra mussels, but these exotics
    ‐51‐
    are so tolerant and tough that everything in the water would have to be poisoned
    to destroy the mussel.ʺ); Robbins, A Western Showdown, at D6 (noting that
    officials in Western states have instituted elaborate and expensive inspection
    systems for boats because they ʺwant desperately to keep the mussels out of
    blue‐ribbon trout streams and pristine mountain lakesʺ as ʺonce established [the
    mussels] are impossible to permanently eradicateʺ).  This is all the more
    problematic because a vessel operator is not likely to know it has a discharge
    violation if, as discussed below, there are no monitoring requirements.18
    Accordingly, EPA acted arbitrarily and capriciously in issuing the
    WQBELs because they violate section 1342ʹs requirement that NPDES permits
    ensure compliance with the CWA.  Cf. Waterkeeper All., 399 F.3d at 498.
    Intervenors raise one additional argument.  Under section 401 of the
    CWA, before EPA issues a permit, the state in which the discharge is to occur
    18 EPAʹs response is that petitionersʹ arguments regarding corrective action
    and BMPs are waived because they were not raised by petitioners in the comments to
    the permit.  See, e.g., NRDC v. EPA, 25 F.3d 1063, 1073‐74 (D.C. Cir. 1994).  Arguments
    can be considered, however, even if not raised during the notice and comment period.
    See NRDC v. EPA, 755 F.3d 1010, 1023 (D.C. Cir. 2014) (ʺEPA retains a duty to examine
    key assumptions as part of its affirmative burden of promulgating and explaining a
    nonarbitrary, non‐capricious rule and therefore EPA must justify that assumption even
    if no one objects to it during the comment period.ʺ (internal quotation marks and
    alteration omitted) (quoting Appalachian Power Co. v. EPA, 135 F.3d 791, 818 (D.C. Cir.
    1998)).
    ‐52‐
    must either certify, or waive its right to certify, that the discharge will comply
    with the stateʹs water quality standards ‐‐ commonly known as a ʺ401
    Certification.ʺ  NRDC, 279 F.3d at 1183; see also 33 U.S.C. § 1341(a); 40 C.F.R.
    § 122.4(b).  The intervenors argue that because ʺthe 401 Certifications have been
    upheld, the matter is settled: the VGP will ensure compliance with the state
    water quality standards.ʺ  Intervenorsʹ Br. at 46.  We disagree.  EPA has an
    independent duty under the CWA to ensure compliance with state and federal
    water quality standards and may impose ʺadditional permit conditions necessary
    to meet that end.ʺ  NRDC, 279 F.3d at 1188.  Such additional permit conditions
    may be necessary if state water quality standards are potentially less stringent
    than the CWAʹs,ʺ because ʺthe CWA provides a federal floor, not a ceiling, on
    environmental protection.ʺ  Dubois v. U.S. Depʹt of Agriculture, 102 F.3d 1273, 1300
    (1st Cir. 1996) (citations omitted).
    3.     Monitoring and Reporting Requirements for TBELs and WQBELs
    Under the CWA, NPDES permits must contain conditions that
    require both monitoring and reporting of monitoring results of TBELs and WQBELs
    to ensure compliance.  See 33 U.S.C. § 1342(a)(2); 40 C.F.R. § 122.44(i)(1)‐(2).  The
    regulations provide, in pertinent part:
    ‐53‐
    In addition to the conditions established
    under § 122.43(a), each NPDES permit shall include
    conditions meeting the following requirements when
    applicable.
    . . .
    (i)      Monitoring requirements.  In addition to § 122.48,
    the following monitoring requirements:
    (1)   To assure compliance with permit
    limitations, requirements to monitor:
    (i)     The mass (or other measurement
    specified in the permit) for each
    pollutant limited in the permit;
    (ii)    The volume of effluent discharged
    from each outfall;
    (iii)   Other measurements as appropriate
    including pollutants in internal
    waste streams under § 122.45(i);
    pollutants in intake water for net
    limitations under § 122.45(f);
    frequency, rate of discharge, etc., for
    noncontinuous discharges under
    § 122.45(e); pollutants subject to
    notification requirements under
    § 122.42(a); and pollutants in sewage
    sludge or other monitoring as
    specified in 40 CFR part 503; or as
    determined to be necessary on a
    case‐by‐case basis pursuant to
    section 405(d)(4) of the CWA.
    (iv)    According to sufficiently sensitive
    test procedures (i.e., methods)
    approved under 40 CFR part 136 for
    ‐54‐
    the analysis of pollutants or pollutant
    parameters or required under 40 CFR
    chapter 1, subchapter N or O.
    . . .
    (2)       Except as provided in paragraphs (i)(4) and
    (i)(5) of this section, requirements to report
    monitoring results shall be established on a
    case‐by‐case basis with a frequency
    dependent on the nature and effect of the
    discharge, but in no case less than once a
    year. . . .
    40 C.F.R. § 122.44(i)(1)‐(2).
    Enforcing compliance with a permit is the key to an effective NPDES
    program.  See NRDC v. Cty of L.A., 725 F.3d 1194, 1208 (9th Cir. 2013) (ʺ[T]he
    NPDES program fundamentally relies on self‐monitoring,ʺ and ʺCongressʹ
    purpose in adopting this self‐monitoring mechanism was to promote
    straightforward enforcement of the Act.ʺ (internal quotations omitted)).  We now
    turn to the monitoring and reporting requirements in the 2013 VGP.
    a.     Monitoring Requirements for TBELs
    As previously discussed, the 2013 VGP requires vessels to monitor
    (1) the functionality of their ballast water treatment systems, if installed, and (2)
    the concentrations of the two ʺindicatorʺ bacteria (E. coli and enterococci). . VGP
    § 2.2.3.5.1.1.2, at 30, § 2.2.3.5.1.1.4, at 31‐32.
    ‐55‐
    The first requirement, known as ʺfunctionality monitoring,ʺ
    determines whether a ballast water treatment program is ʺoperating according
    to the manufacturersʹ requirements.ʺ  App. at 96.  A shipowner is required to
    check a measurement that would ʺverify system functionality,ʺ such as how
    much chlorine the system is using each month.  Id. at 1168.  If the measurement
    is correct, it is assumed that the program is in compliance.  If the equipment is
    not operating properly, the ship is not permitted to discharge ballast.  The vessel
    owner is not required to take any measurement of pollutants or significant
    categories of living organisms in ballast water being discharged.  Instead, the
    shipowner relies solely on the functioning of the treatment system to determine
    if the ship is complying with the permit.  Treatment systems are inspected
    monthly.  See VGP § 2.2.3.5.1.1.2, at 30 (ʺTo assess the [systemʹs] functionality,
    monitoring indicators of the [systemʹs] functionality is required at least once per
    month for specific parameters that are applicable to your system.ʺ).
    The second provision, effluent biological organism monitoring,
    requires vessels to collect small‐volume samples from ballast discharge and
    analyze them for concentrations of two indicator pathogens, E. coli and
    enterococci.  VGP § 2.2.3.5.1.1.4, at 31‐32.  The idea is that if there are significant
    ‐56‐
    levels of these two pathogens in the sample, then treatment has not been
    effective.  Vessels are not required to monitor Vibrio cholera or medium or large
    organisms regulated in the 2013 VGP.  Sampling is required between one and
    four times a year, depending on the type of system.
    Petitioners present two arguments about why these requirements do
    not accord with the law.  First, petitioners argue that the 2013 VGP violates CWA
    regulations by not requiring vessels to monitor the concentration of living
    organisms.  The regulations require monitoring of mass, volume, or ʺother
    measurement specified in the permit.ʺ  40 C.F.R. § 122.44(i)(1)(i)‐(ii).  Petitioners
    contend the unit of measurement for living organisms in the 2013 VGP should be
    concentration.  Neither functionality monitoring nor testing for two indicator
    microorganisms measures concentration.  Thus, according to petitioners, the
    monitoring and reporting requirements for TBELs violate 40 C.F.R. § 122.44(i)(1).
    Second, petitioners argue that these monitoring requirements violate
    the requirement in 40 C.F.R. § 122.44(i)(1) that monitoring ʺassure[s] compliance
    with permit limitations.ʺ  Relying on functionality monitoring instead of
    requiring an actual measurement of concentrations of organisms means that
    neither EPA nor the public knows if the permittees are complying with the
    ‐57‐
    TBELs.  Similarly, petitioners argue that monitoring for the presence of the two
    ʺindicator bacteria,ʺ E. coli and enterococci, is not sufficient to monitor
    compliance with TBELs because it indicates only their presence in the water, not
    their quantity.
    We disagree with petitioners and conclude that EPAʹs monitoring
    requirements were not arbitrary and capricious.  The CWA regulations expressly
    allow for monitoring quantities other than mass or volume, namely some ʺother
    measurement specified in the permit[] for each pollutant limited in the permit.ʺ
    40 C.F.R. § 122.44(i)(1).  Functionality monitoring and monitoring for the
    presence of indicator organisms qualify as such ʺother measurement.ʺ
    And while there are potential alternatives, petitionersʹ urged
    alternative of direct monitoring is not required because ʺmore sophisticated
    methods for enumerating living organisms in the larger size classes are not
    currently available for use by permittees.ʺ  App. at 524.  Current technology is
    not capable of adequately monitoring ballast water as it is being discharged
    because, EPA notes, such monitoring requires analyzing large volumes of water
    and is prohibitively expensive and impractical.  According to EPA, testing
    medium and large organisms with shipboard systems can cost between $75,000
    ‐58‐
    and $125,000 per vessel per sampling event.  Moreover, the process would be
    impractical, involving sampling and analyzing large volumes of water in labs
    and requiring ʺdozens of hours to collect and analyze those samples.ʺ  EPA Br. at
    89.
    Given the difficult circumstances, EPAʹs monitoring requirements
    for TBELs were not arbitrary and capricious.  Cf. NRDC v. Costle, 568 F.2d 1369,
    1380 (D.C. Cir. 1977) (ʺ[W]hen numerical effluent limitations are infeasible, EPA
    may issue permits with conditions designed to reduce the level of effluent
    discharges to acceptable levels.ʺ).  In the face of the severe technological
    limitations on monitoring, it was reasonable for EPA to decline to require
    monitoring for parameters for which it is currently impractical to collect and
    analyze samples.  Functionality monitoring and biological indirect monitoring
    are the only feasible options at present to assure compliance with the permit.
    We defer to EPAʹs decision that functionality monitoring and biological
    indicator monitoring, when used in combination, offer an acceptable ʺother
    measurement.ʺ  See Auer, 519 U.S. at 461 (holding that courts should defer to
    agency’s interpretation of its own regulations if not plainly erroneous or
    inconsistent with the regulation).
    ‐59‐
    b.     Monitoring Requirements for WQBELs
    Petitioners also argue that EPA acted arbitrarily and capriciously in
    failing to require that permittees monitor ballast water discharges to ensure
    compliance with WQBELs.  The only monitoring requirement for WQBELs is
    that ships report the ʺexpected date, location, volume, and salinity of any ballast
    water to be dischargedʺ into U.S. waters or at a reception facility.  VGP § 4.3, at
    72 (emphasis added).  There is no requirement to report actual volumes,
    locations, or composition of ballast water discharges.
    We agree that failure to include monitoring requirements for
    WQBELs was arbitrary and capricious.  The regulations require monitoring to
    ʺassure compliance with permit limitations.ʺ  40 C.F.R. § 122.44.  Generally, ʺan
    NPDES permit is unlawful if a permittee is not required to effectively monitor its
    permit compliance.ʺ  NRDC, 725 F.3d at 1207.  Here, the reporting requirement
    provided little information on the quality of the ballast water, requiring only
    information on expected date, location, volume, and salinity.  There is no way to
    derive from that information whether a vessel is actually in compliance with the
    WQBELs.  Thus, because the 2013 VGP does not contain a mechanism to evaluate
    compliance with the WQBELs, the monitoring requirements are arbitrary and
    ‐60‐
    capricious and not in accordance with the law.  See Waterkeeper All., 399 F.3d at
    499 (failure of permit to include any mechanism for evaluating compliance with
    required technical standards rendered agency unable to ensure compliance with
    water quality standards).
    Our conclusion is further supported by the simple, but overlooked,
    options that EPA has in structuring WQBEL monitoring requirements.  One
    possible condition EPA could consider including in the WQBELs would be to
    require shipowners to monitor the actual time, place, and volume of ballast water
    discharge, rather than the expected time, place, and volume.  Another possible
    condition would be to require shipowners to monitor for a particular pathogen
    or pollutant if it became known that such a pathogen or pollutant is a problem in
    a particular port.  Each of these options would provide more significant
    monitoring.
    EPAʹs contentions on this point are unpersuasive.  EPA argues that
    if a vessel is in compliance with the TBELs, it should be ʺgenerally expected to
    already be controlling [its] vessel discharges to a degree that is protective of
    water quality,ʺ rendering additional monitoring to demonstrate compliance with
    narrative WQBELs unnecessary.  App. at 530.  In defense of this position, EPA
    ‐61‐
    also argues that 40 C.F.R. § 122.44(i) does not apply because of ʺpractical
    constraints on the ability to collect and analyze the volumes of ballast water
    necessary to ʹdirectlyʹ detect and quantify such organisms at the levels of
    concern.ʺ  EPA Br. at 98.  According to EPA, it is simply ʺunrealisticʺ to have
    stricter monitoring.
    This, however, is not a valid excuse in the WQBEL context.  See
    NRDC, 859 F.2d at 208 (stating legislative history of CWA ʺstrongly supports
    [the] position that Congress did not intend to tie compliance with water quality‐
    based limitations to the capabilities of any given level of technology,ʺ and ʺa
    water quality‐based permit limit begins with the premise that a certain level of
    water quality will be maintained, come what may, and places upon the permittee
    the responsibility for realizing that goalʺ).  It is inconsistent to say that WQBELs
    are necessary to ensure that water quality standards are met, while specific
    enforcement of such WQBELs is unnecessary.  More importantly, this lack of
    enforcement violates the CWA regulations, which mandate that no permit may
    be issued ʺ[w]hen the imposition of conditions cannot ensure compliance with
    the applicable water quality requirements of all affected States.ʺ  40 C.F.R.
    § 122.4(d).
    ‐62‐
    Accordingly, EPAʹs failure to include monitoring for compliance
    with WQBELs was inconsistent with regulations and thus arbitrary and
    capricious.
    C.    Remand
    Accordingly, we remand this matter to EPA for proceedings
    consistent with this opinion.  The 2013 VGP is to remain in place until EPA issues
    a new VGP.  See Idaho Farm Bureau Fedʹn v. Babbitt, 58 F.3d 1392, 1405 (9th Cir.
    1995) (holding that ʺ[o]rdinarily when a regulation is not promulgated in
    compliance with the APA, the regulation is invalid.  However, when equity
    demands, the regulation can be left in place while the agency follows the
    necessary procedures.ʺ (citation omitted)); see also Allied‐Signal, Inc. v. U.S.
    Nuclear Regulatory Comm., 988 F.2d 146, 150 (D.C. Cir. 1993) (ʺAn inadequately
    supported rule, however, need not necessarily be vacated.ʺ); Fertilizer Inst. v.
    EPA, 935 F.2d 1303, 1312 (D.C. Cir. 1991) (ʺ[W]hen equity demands, an
    unlawfully promulgated regulation can be left in place while the agency
    provides the proper procedural remedy.ʺ); W. Oil & Gas Assʹn v. EPA, 633 F.2d
    803, 813 (9th Cir. 1980) (ʺ[A] reviewing court has discretion to shape an equitable
    remedy, [and so] we leave the challenged designations in effect.ʺ).
    ‐63‐
    CONCLUSION
    For the reasons set forth above, we GRANT the petition for review
    with respect to
    (1)   EPAʹs decision to set the TBELs at the IMO Standard,
    (2)   EPAʹs failure to consider onshore treatment for ballast water
    discharge,
    (3)   EPAʹs decision to exempt pre‐2009 Lakers from the TBELS in
    the 2013 VGP permit,
    (4)   EPAʹs narrative standard for WQBELs, and
    (5)   the monitoring and reporting requirements established by
    EPA for WQBELs,
    and REMAND for further proceedings in these respects.
    We DENY the petition for review with respect to TBELs for viruses
    and protists and the monitoring and reporting requirements established by EPA
    for TBELs.
    The 2013 VGP shall remain in place until EPA issues a new VGP.
    ‐64‐
    APPENDIX
    Glossary of Acronyms
    BACT      Best available demonstrated control technology
    BAT       Best available technology economically achievable
    BCT       Best conventional pollutant control technology
    BMP       Best management practice
    BWTS      Ballast water treatment systems
    CSA       Canadian Shipowners Association
    CWA       Clean Water Act
    ELG       Effluent limitation guidelines
    EPA       Environmental Protection Agency
    IMO       International Maritime Organization
    NAS       National Research Council/National Academy of Sciences
    NPDES     National Pollutant Discharge Elimination System
    NRDC      Natural Resources Defense Council
    NWEA      Northwest Environmental Advocates
    NWF       National Wildlife Federation
    PFR       Petition for Review
    SAB       Science Advisory Board
    TBEL      Technology‐based effluent limitation
    VGP       Vessel General Permit
    WQBEL     Water quality‐based effluent limitation
    ‐65‐
    

Document Info

Docket Number: 13-1745(L)

Filed Date: 12/18/2015

Precedential Status: Precedential

Modified Date: 12/18/2015

Authorities (36)

Roland C. Dubois and Restore: The North Woods v. United ... , 102 F.3d 1273 ( 1996 )

Humana of Aurora, Inc. D/B/A Aurora Community Hospital v. ... , 753 F.2d 1579 ( 1985 )

environmental-defense-v-united-states-environmental-protection-agency , 369 F.3d 193 ( 2004 )

California & Hawaiian Sugar Company v. Environmental ... , 553 F.2d 280 ( 1977 )

Hooker Chemicals & Plastics Corp. v. Russell E. Train, as ... , 537 F.2d 620 ( 1976 )

riverkeeper-inc-natural-resources-defense-council-waterkeeper-alliance , 358 F.3d 174 ( 2004 )

Fmc Corporation v. Russell E. Train , 539 F.2d 973 ( 1976 )

Natural Resources Defense Council San Diego Baykeeper, ... , 236 F.3d 985 ( 2000 )

association-of-pacific-fisheries-new-england-fish-company-peter-pan , 615 F.2d 794 ( 1980 )

Cpc International Inc. v. Russell E. Train, (Two Cases). ... , 515 F.2d 1032 ( 1975 )

Tanners' Council of America, Inc. v. Russell E. Train, as ... , 540 F.2d 1188 ( 1976 )

American Petroleum Institute v. United States Environmental ... , 858 F.2d 261 ( 1988 )

waterkeeper-alliance-inc-american-farm-bureau-federation-national , 399 F.3d 486 ( 2005 )

texas-oil-gas-association-marathon-oil-company-trustees-for-alaska , 161 F.3d 923 ( 1998 )

the-fertilizer-institute-v-united-states-environmental-protection-agency , 935 F.2d 1303 ( 1991 )

natural-resources-defense-council-southeast-alaska-conservation-council , 279 F.3d 1180 ( 2002 )

Northwest Environmental Advocates v. United States ... , 537 F.3d 1006 ( 2008 )

trustees-for-alaska-and-gilbert-m-zemansky-v-environmental-protection , 749 F.2d 549 ( 1984 )

western-oil-and-gas-association-a-nonprofit-corporation-california , 633 F.2d 803 ( 1980 )

idaho-farm-bureau-federation-a-non-profit-corporation-idaho-cattle , 58 F.3d 1392 ( 1995 )

View All Authorities »