In Re: Katrina Canal Breaches

JERRY E. SMITH, Circuit Judge: Treating the petition for rehearing en banc as a petition for panel rehearing, the petition for panel rehearing is GRANTED. We withdraw our opinion, In re Katrina Canal Breaches Litig., 673 F.3d 381 (5th Cir.2012), and substitute the following: Decades ago, the Army Corps of Engineers (the “Corps”) dredged the Mississippi River Gulf Outlet (“MRGO”), a shipping channel between New Orleans and the Gulf of Mexico, and levees alongside the channel and around the city. When Hurricane Katrina struck in 2005, MRGO’s size and configuration greatly aggravated the storm’s effects on the city and its environs. Claimants alleging damage from Katrina filed hundreds of lawsuits, many of which were consolidated before the district judge a quo. That court worked with plaintiffs’ litigation committees to identify several categories of plaintiffs and individual “bellwether” plaintiffs. This opinion concerns three groups of bellwether plaintiffs, all suing the United States for flood damages. One group of seven plaintiffs went to trial; three prevailed on all claims, and four did not. Another group was dismissed before trial when the government was found immune. The third hás survived motions to dismiss and is proceeding to trial. All losing parties have appealed; the government has also petitioned for a writ of mandamus to stay the third group’s trial pending issuance of this opinion. We REVERSE each judgment for the plaintiffs, AFFIRM each judgment for the government, and DENY the petition as moot. I. Background. In 1943, Congress requested a report from the Chief of Engineers, Secretary of the Army, investigating ways to make the Port of New Orleans more accessible for maritime and military use. That request led to the authorization of MRGO in 1956. The channel was built to its full dimensions by 1968 and afforded a shorter shipping route between the Gulf of Mexico and New Orleans. As the district court noted, the channel, as originally designed, “was to be 36 feet deep and 500 feet wide, increasing at the Gulf of Mexico to 38 feet deep and 600 feet wide.” In re Katrina Canal Breaches Consol. Litig., 647 F.Supp.2d 644, 717 (E.D.La.2009). MRGO was cut *442 through virgin coastal wetlands at a depth that exposed strata of so-called “fat clay,” a form of soil soft enough that it will move if made to bear a load. The channel’s original designers considered and rejected armoring its banks with foreshore protection, leaving them vulnerable to erosion. During the design and construction of MRGO, the Corps also implemented the Lake Pontchartrain and Vicinity Hurricane Protection Plan (“LPV”). Pursuant to that plan, the Corps constructed, inter alia, the New Orleans East Unit, levees protecting New Orleans East; the Chalmette. Area Unit, levees protecting the Ninth Ward and St. Bernard Parish; and higher floodwalls in the outfall canals at 17th Street, Orleans Avenue, and London Avenue. Separately from MRGO, between 1967 and 1985 the Corps also “lifted” and enlarged portions of the levee paralleling Reach 2 of the channel. Reach 2 of MRGO runs southeasterly from a point near Michoud in eastern New Orleans along the south shore of Lake Borgne and through the marshes to and across Chandeleur Sound to the Gulf of Mexico. Its south shore parallels the Reach 2 levee (later breached during Katrina, resulting in massive flooding). Foreshore .protection was authorized for that shore in 1967 and costs charged to the MRGO project in 1968. From 1968 until 1980, it is hot evident why the protection was not implemented. In March 1980, the Corps scrapped its original design for further study, which it continued for the next two years. In 1982, it began testing foreshore protection along the south shore; the study was completed in early 1983. A contract was awarded in 1985, and the foreshore protection was finished in 1986. For the north shore of Reach 2 — that shore which abuts Lake Borgne and its wetlands — the district court found that, by the early 1970s, erosion of the channel had threatened the wetlands, in particular the land bridge that prevented Lake Borgne from flowing directly into the MRGO. In the early 1980s, the Corps was directed to study the feasibility of protection along the north shore. In 1984 and 1988, the Corps reported studies recognizing that erosion from wave wash had widened the channel and that the north shore was close to being breached, thereby exposing development and inhabitants to the southwest to hurricanes from Lake Borgne. The Corps outlined two erosion-control plans in the 1984 report. The 1988 report concluded that the bank-erosion problem threatened to increase dredging costs sixfold, a problem Corps engineers with the New Orleans Division attempted to address via a supplement to MRGO’s Design Memorandum, which would encompass further studies and bypass any requirements for local cost-sharing. That attempt was rejected by the Lower Mississippi Valley Division, which criticized the modeling used and the estimated costs of dredging. The Corps took the position that design modification was not warranted under the cost-benefit ratio: “[U]ntil the cost of providing foreshore protection proved to be less expensive than the continued need for dredging to maintain the channel’s navigability, the Corps did not actively pursue funding for this protection.” Id. at 662. The Corps refused to undertake the cost of foreshore protection unless there was local cost participation under the Water Resources Development Act, 33 U.S.C. § 2201 et seq. Furthermore, the district court noted, the Corps did not prioritize protecting the north shore, because there was no levee to protect, and the primary mission of the Corps was to keep the channel navigable. In 1994, the Corps issued another report in which it still proceeded under the as *443 sumption that the cost of foreshore protection was greater than it proved to be; the Corps took the position that costs for bank stabilization should be shared by the local population. Finally, in the mid-1990s, the Corps realized that the actual costs of maintaining the foreshore protection were less than estimated in the 1988 report, so a “re-analysis of the benefits and costs based on this new cost information was the genesis for an April 1996 Evaluation Report.” In re Katrina Canal Breaches, 647 F.Supp.2d at 663. That report recommended funding for protection along five sections of the north shore of Reach 2, and that same year, Congress instructed the Corps to use its available operations and maintenance funds to protect the shore to minimize future dredging costs and preserve the wetlands. The Corps’ delay in armoring MRGO allowed wave wash from ships’ wakes to erode the channel considerably, 1 destroying the banks that would have helped to protect the Reach 2 levee (in the Chalmette Area Unit) from front-side wave attack and loss of height. The increased channel width added more fetch 2 as well, allowing for a more forceful frontal wave attack on the levee. MRGO’s expansion thus allowed Hurricane Katrina to generate a peak storm surge capable of breaching the Reach 2 levee and flooding the St. Bernard polder. 3 Separately from MRGO, the hurricane also caused the 17th Street, Orleans Avenue, and London Avenue levees to breach. Over four hundred plaintiffs sued in federal court to recover for Katrina-related damages, many naming the federal government as a defendant. Seven of those (the “Robinson plaintiffs”) went to trial. The court issued its impressive rulings in thorough opinions. 4 The court found that neither the Flood Control Act of 1928 (“FCA”), 33 U.S.C. § 702, nor the discretionary-function exception (“DFE”) to the Federal Tort Claims Act (“FTCA”), 28 U.S.C. § 2680(a), protected the government from suit; after nineteen days of trial, the court found that three plaintiffs had proven the government’s full liability and four had not. Another group of plaintiffs (the “Anderson plaintiffs”) had their cases dismissed on the government’s motion, the court finding both immunities applicable. Still a different group (the “Armstrong plaintiffs”) are preparing for trial of their own case against the government. The government appeals its losses in Robinson-, the losing Robinson plaintiffs cross-appeal. The Anderson plaintiffs have appealed as well. On the theory that a favorable ruling here might moot the pending Armstrong trial, the government petitions this court for a writ of mandamus to order the district court to stay trial until we issue an opinion in Robinson. The *444 three cases have been consolidated on appeal. II. Standard of Review. A district court’s construction of immunity under the FCA and of the DFE is subject to de novo review. See Withhart v. Otto Candies, L.L.C., 431 F.3d 840, 841 (5th Cir.2005). Factual findings are reviewed for clear error. Lehmann v. GE Global Ins. Holding Corp., 524 F.3d 621, 624 (5th Cir.2008). III. The FCA: Construction and Application. Under the Central Green-Graci test, 5 the government enjoys immunity only from damages caused by floodwaters released on account of flood-control activity or negligence therein. Some Katrina-related flooding was caused not by flood-control activity (or negligence therein) but by MRGO, a navigational channel whose design, construction, and maintenance cannot be characterized as flood-control activity. Therefore, the FCA does not immunize the government against liability for that flooding. A. The Scope of FCA Immunity. The FCA “was the Nation’s response to the disastrous flood in the Mississippi River Valley in 1927.” United States v. James, 478 U.S. 597, 606, 106 S.Ct. 3116, 92 L.Ed.2d 483 (1986). The law enacted “a comprehensive ten-year program for the entire [Mississippi River] valley, embodying a general bank protection scheme, channel stabilization and river regulation, all involving vast expenditures of public funds.” United States v. Sponenbarger, 308 U.S. 256, 262, 60 S.Ct. 225, 84 L.Ed. 230 (1939). Staggering in scope, the FCA’s flood-control program was “the largest public works project undertaken up to that time in the United States.” James, 478 U.S. at 606, 106 S.Ct. 3116. The FCA predated the FTCA and the latter’s abrogation of sovereign immunity from tort liability, but Congress nonetheless included Section 702c in the FCA, 33 U.S.C. § 702c, which affirms the government’s sovereign immunity in the flood-control context: “No liability of any kind shall attach to or rest upon the United States for any damage from or by floods or flood waters at any place.” The Supreme Court has read Section 702c’s legislative history as reflecting Congress’s “consistent concern for limiting the Federal Government’s financial liability to expenditures directly necessary for the construction and operation of the various projects” funded by the FCA. James, 478 U.S. at 607, 106 S.Ct. 3116. This court had similarly emphasized Congress’s reluctance to build flood-control projects without a guarantee of immunity: “[T]he immunity from liability for floodwater damage arising in connection with flood control works was the condition upon which the government decided to enter into the area of nationwide flood control programs.” Graci, 456 F.2d at 26 (internal quotation marks omitted). Although the text of Section 702c could not more broadly preserve immunity, in Graci we read the FCA and its legislative history to include a limitation. We determined that it was unreasonable “to suppose that in exchange for its entry into flood control projects[,] the United States demanded complete immunity from liability for the negligent and wrongful acts of its employees unconnected with flood control projects.” Id. We therefore held that the Graci plaintiffs’ claims were not barred by the FCA, because they alleged *445 flood water damage caused by “the negligence of the United States unconnected with any flood control project.” Id. at 27. The government proposes a broader interpretation of FCA immunity that would hold it immune where plaintiffs allege damage caused by flood waters that a federal flood-control project had failed to contain. If the flood waters at issue are connected to a flood-control project, the government argues, the claim is barred. Language in James supports the government’s position: The Court read Section 702c’s “plain language” and defined the phrase “floods or flood waters” to mean “all waters contained in or carried through a federal flood control project for purposes of or related to flood control, as well as to waters that such projects cannot control.” James, 478 U.S. at 605, 106 S.Ct. 3116. The issue in James, however, was whether Section 702c’s reference to “ ‘damage’ encompassed not just property damage, but also personal injuries and death.” Cent. Green, 531 U.S. at 429, 121 S.Ct. 1005. In Central Green, the Court most recently considered the scope of Section 702c immunity and the meaning of the phrase “floods and flood waters.” The Ninth Circuit — purporting to follow James’s definition of “floods and flood waters” — had focused on whether the source of water that caused the complained-of damage was part of a federal flood-control project; that court ultimately concluded that immunity attached “ ‘solely because [the Madera Canal] is a branch of the Central Valley Project.’” Id. at 428, 121 S.Ct. 1005 (quoting Cent. Green Co. v. United States, 177 F.3d 834, 839 (9th Cir.1999)). In so holding, the Ninth Circuit “recognized that the Government would probably not have enjoyed immunity in at least three other Circuits where the courts require a nexus between flood control activities and the harm done to the plaintiff.” Id. The Court granted certiorari to resolve that circuit split and framed the issue as “whether [‘floods or flood waters’] encompass all the water that flows through a federal facility that was designed and is operated, at least in part, for flood control purposes.” Id. at 427, 121 S.Ct. 1005. With the meaning of the phrase “floods and flood waters” squarely before it, the Court rejected James’s definition as “confusing dicta ... that, if read literally, ... sweeps so broadly as to make little sense,” id. at 431, 121 S.Ct. 1005; the Central Green Court set about narrowing the scope of Section 702c immunity by shifting the analytical focus away from the water’s presence in a federal flood-control project. Under the Ninth Circuit’s interpretation of James, waters constituted “floods or flood waters,” and immunity applied where the government was linked to the waters through the mere presence of a federal flood-control project. In Central Green, the Court noted that “the text of the statute does not include the words ‘flood control project,’ ” id. at 434, 121 S.Ct. 1005, so the scope of Section 702c is not determined “by the character of the federal project or the purposes it serves, but by the character of the waters that cause the relevant damage and the purposes behind their release,” id. Waters that constitute “floods or flood waters” within the meaning of Section 702c, therefore, are not all waters that pass through a federal flood-control project, but are instead waters of a certain “character.” Under Central Green, determining whether the water that caused damage had this immune “character” has become the heart of the Section 702c inquiry. The Court exemplified how water attains this “character” by contrasting James with Henderson v. United States, 965 F.2d 1488 (8th Cir.1992). *446 In James, 478 U.S. at 612, 106 S.Ct. 3116, the Court determined that the government was immune from suit. There, the plaintiffs were injured when the Corps opened gates in flood-control projects to release water from reservoirs that had reached flood stage. Id. at 600-01, 106 S.Ct. 3116. Notably, the gates were opened as part of the projects’ flood-control function. Id. The Central Green Court distinguished James from Henderson, in which the plaintiff was injured when the Corps opened gates in a flood-control project to produce hydroelectric power. See Henderson, 965 F.2d at 1490. The water in James was “flood waters,” stated the Court, but the water in Henderson was not. Cent. Green, 531 U.S. at 436, 121 S.Ct. 1005. The only factual difference between James and Henderson is that in James, the Corps acted to control a flood, whereas in Henderson, its activity was entirely unrelated to flood control. Thus, after Central Green, waters have the immune character of “flood waters” if the government’s link to the waters is through flood-control activity. That is to say, the government’s acting upon waters for the purpose of flood control is flood-control activity, and flood-control activity is what gives waters an immune “character.” We therefore reject the government’s interpretation of the scope of Section 702c and conclude, instead, that the United States enjoys immunity under that section only where damages result from waters released by flood-control activity or negligence therein. B. The MRGO-St. Bernard Polder Plaintiffs: No Immunity. Plaintiffs Anthony Franz, Jr., Lucille Franz, Kent Lattimore, Lattimore and Associates, and Tanya Smith allege that the operation and maintenance of MRGO caused the levee along Reach 2 of MRGO to be breached, resulting in the flooding of the St. Bernard polder. The district court found that a combination of erosion caused by MRGO (which could have been prevented through timely foreshore protection) and destruction of wetland vegetation caused by increased salinity levels on account of MRGO’s operation led to the breaches in the Reach 2 levee. The court concluded that MRGO was a navigational, not flood-control, project and, unlike the canals in Anderson, wholly extrinsic to the LPV. The court analogized MRGO to a Navy vessel that, as a result of negligent operation, crashes through a levee, causing a flood: Though the levee would be a flood-control project, the cause of the waters’ release — negligent operation of a Navy vessel — would not be the flood-control project or its negligent operation, but rather the negligence in a wholly extrinsic government action. Thus, the government would be afforded no immunity under Section 702c. Similarly, the court ruled, the negligently maintained MRGO acted upon the levees in a way that caused them to be breached during Hurricane Katrina, and, because MRGO was not a flood-control project and was separate from the LPV, no immunity should attach under Section 702c. But the government maintains that, instead of being distinct from one another, MRGO and the LPV were intertwined. Because the failure to implement timely foreshore protection (as opposed to continued dredging of MRGO) caused the levee breach, the government argues that foreshore protection would have been done solely to benefit the levee system, a flood-control project. Thus, the government argues, the Corps’s actions with respect to MRGO were relevant precisely because of their asserted impact on the Reach 2 levees; the government describes the *447 plaintiffs’ claim as an assertion that “the government did not take adequate measures to ensure that the Reach 2 levee would be able to restrain flood waters.” Consequently, the government says that the claim should be barred by Section 702c. Under the district court’s rule for applying Section 702c — “immunity arises where damage is caused by flood waters emanating from a flood control project,” In re Katrina Canal Breaches, 577 F.Supp.2d at 824 — the negligent maintenance of MRGO arguably does not create Section 702c immunity. As the court held, the lack of foreshore protection was properly allocable to MRGO (a navigational project), not the LPV (a flood-control project). Costs of the foreshore protection were to be charged to the MRGO project and were initially envisioned to protect the channel width and maintain navigability. “Thus,” the district court concluded, “the failures at issue here are extrinsic to the LPV and are not subject to ,§ 702c immunity.” In re Katrina Canal Breaches, 647 F.Supp.2d at 699. Our interpretation of Section 702c and the caselaw, however, provides a rule slightly different from the district court’s. Instead of its strictly categorical approach, which would have immunity attach only where a flood was caused by a project that had the purpose of flood control, we recognize immunity for any flood-control activity engaged in by the government, even in the context of a project that was not primarily or substantially related to flood control. Thus, for example, if the government had attempted foreshore protection inside MRGO, but that protection (whether by design or negligence) caused or exacerbated flood damage, the district court’s rule would grant the Corps no immunity, because MRGO was not a flood-control project. Our rule, by contrast, attaches immunity if the foreshore protection had flood control as its purpose — that is, if installing and maintaining foreshore protection was a flood-control activity regardless of the nature of MRGO, the overall project. To assess the St. Bernard plaintiffs’ argument, then, we must determine whether the Corps’s decision to dredge MRGO instead of implementing foreshore protection constitutes flood-control activity qualifying for Section 702c immunity. The government would contend that, because foreshore protection would have been done instead of dredging for the sole purpose of protecting the levees, the decision to dredge was flood-control activity protected by Section 702c. Even under our formulation instead of the district court’s, the government’s defense fails. The district court found that “the Corps clearly took the position that its primary mission was to keep the shipping channel open to deep draft traffic regardless of the consequences.” Id. at 660. Accordingly, the Corps chose to dredge MRGO to keep it navigable rather than promptly implement costlier foreshore protection, which would have had the dual purpose of keeping MRGO navigable and protecting the levees. Thus, even in the context of MRGO, the Corps took -no action that could be characterized as flood-control activity. The only conceivable flood-control activity is the failure of the government to implement timely foreshore protection, the very omission complained of by the plaintiffs. Such an omission cannot provide the basis for immunity under Section 702c, lest all flood damage caused by government activity be regarded as a result of (a lack of) flood-control activity. The district court’s naval analogy is apt. There, a negligent government activity (operating a ship) wholly unrelated to flood control causes a flood, destroying another *448 government project (a levee). Though one could imagine a plaintiffs saying that the Navy was negligent in not creating an onboard warning system preventing its ship from running into the levee, that hypothetical system does not transform the operation of the Navy vessel into a flood-control activity. To use another analogy, suppose the government builds an Army base on the banks of the Mississippi River. Because of soft soil, the weight of its structures depresses the land and causes flooding to nearby farms during a heavy rain. Although the government might argue that the Army could have built dikes to prevent the flooding, that hypothetical solution does not transform the building of a base into a flood-control activity, or the failure to build dikes into negligence in flood-control activity. In each example, the government is negligent, but in nautical or construction activity, not flood control. Thus, the government cannot claim Section 702c immunity for MRGO’s role in breaching the Reach 2 levee. The dredging of MRGO was not a flood-control activity, nor was MRGO so interconnected with the LPV as to make it part of the LPV. Therefore, the flood waters that destroyed the plaintiffs’ property were not released by any flood-control activity or negligence therein. C. The MRGO-New Orleans East Plaintiffs: Immunity Irrelevant. The district court decided that plaintiffs Norman and Monica Robinson were not entitled to relief for reasons unrelated to the FCA. Specifically, the court reasoned that the Corps had no duty to build a surge-protection barrier and that any negligence in operating or maintaining MRGO did not cause the Robinsons’ injuries, which would have occurred had MRGO remained at its design width. Id. at 660. Because we uphold the district court’s factual findings infra, there is no need to analyze Section 702c immunity as to the Robinsons. D. The Anderson Plaintiffs: Immunity Applies. The Anderson plaintiffs allege that they were harmed by the breaching of the levees along the 17th Street, London Avenue, and Orleans Avenue Canals caused by the negligent dredging of the 17th Street Canal and the levees’ negligent design and construction. The district court, in considering the government’s motion to dismiss, took judicial notice of the fact that those levees were constructed as part of the LPV, and the canals themselves were incorporated by Congress into the overall LPV project. Id. at 637-38. Although the design and plan for those levees and canals may not have been prudent, they were ultimately approved by Congress. And though the dredging of the 17th Street Canal may have been negligent, the canal was part of the LPV. Because those canals were designed to prevent flooding either by creating drainage or by preventing storm surge with levees, and were fully incorporated in the LPV plan, their design and dredging are flood-control activities. Thus, because the waters that damaged the Anderson plaintiffs’ property were allegedly released by negligence in flood-control activity, the Corps is immune under Section 702c. IV. Construction and Application of the DFE. The DFE bars suit on any claim that is “based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a federal agency or an employee of the Government, whether or not the *449 discretion involved be abused.” 28 U.S.C. § 2680(a). The purpose of the DFE “is to prevent judicial second-guessing of legislative and administrative decisions grounded in social, economic, and political policy through the medium of an action in tort.” Spotts v. United States, 613 F.3d 559, 568 (5th Cir.2010) (internal quotation marks omitted). “The Supreme Court has developed a two-part test for determining whether the federal government’s conduct qualifies as a discretionary function or duty.” Freeman v. United States, 556 F.3d 326, 336-37 (5th Cir.2009). First, the conduct must involve “an element of judgment or choice.” Id. at 337 (quoting United States v. Gaubert, 499 U.S. 315, 322, 111 S.Ct. 1267, 113 L.Ed.2d 335 (1991)) (other citation omitted). “If a statute, regulation, or policy leaves it to a federal agency or employee to determine when and how to take action, the agency is not bound to act in a particular manner and the exercise of its authority is discretionary.” Id. (citation omitted). “On the other hand, [t]he requirement of judgment or choice is not satisfied and the discretionary function exception does not apply if a federal statute, regulation, or policy specifically prescribes a course of action for an employee to follow, because the employee has no rightful option but to adhere to the directive.” Id. (internal quotation marks omitted). Second, the DFE “protects only governmental actions and decisions based on considerations of public policy.” Id. (citations omitted). The “proper inquiry” is not whether the decisionmaker “in fact engaged in a policy analysis when reaching his decision but instead whether his decision was susceptible to policy analysis.” Spotts, 613 F.3d at 572 (emphasis removed) (internal quotation marks omitted). Under Gaubert, 499 U.S. at 324, 111 S.Ct. 1267, “the very existence” of a law or regulation allowing a government employee discretion (satisfying Berkovitz’s first prong) “creates a strong presumption that a discretionary act authorized by the regulation involves consideration of the same policies which led to the promulgation of the regulations.” A. Applying the DFE to the Corps’s Failure to Contain MRGO. The Robinson plaintiffs advance three arguments against applying the DFE. First, they claim that the impact-review requirement of the National Environmental Policy Act (“NEPA”) constituted a legal mandate that overrides the Corps’s discretion. Next, they maintain that one or more project authorizations created a non-discretionary duty to armor the banks of MRGO. Finally, they argue that the critical calculation made by the Corps in waiting to armor MRGO was an erroneous scientific judgment, not a decision susceptible to public-policy considerations. 1. NEPA. NEPA mandates that “federal agencies must, except in certain qualifying situations, complete a detailed environmental impact statement (‘EIS’) for any major federal action significantly affecting the quality of the human environment.” O’Reilly v. U.S. Army Corps of Eng’rs, 477 F.3d 225, 228 (5th Cir.2007). “An agency is not required to prepare a full EIS if it determines — based on a shorter environmental assessment (EA) — that the proposed action will not have a significant impact on the environment.” Winter v. Natural Res. Def. Council, Inc., 555 U.S. 7, 16, 129 S.Ct. 365, 172 L.Ed.2d 249 (2008); see also 40 C.F.R. § 1508.9. If the federal action has not been completed, agencies have an obligation to update an EIS by filing a supplemental environmental impact statement (“SEIS”) if there are *450 “significant new circumstances or information relevant to environmental concerns and bearing on the proposed action or its impacts.” 40 C.F.R. § 1502.9(c)(l)(ii). The district court found (without challenge on appeal) that the Corps violated NEPA’s mandates in at least three ways: (1) [Its] 1976 [ ]EIS was fatally flawed; (2) the Corps never filed a SEIS even after it acknowledged substantial changes caused by the maintenance and operation of the MRGO; and (3) it improperly segmented its reporting guaranteeing that the public and other agencies would remain uninformed as to the drastic effects the channel was causing. In re Katrina Canal Breaches, 647 F.Supp.2d at 725. Though the court acknowledged that “decisions to file EISs, SEISs and EAs are committed to the judgment of the agency,” id. (citations omitted), it found the Corps had violated NEPA under the standards of the Administrative Procedure Act: “[A] reviewing court shall ... hold unlawful and set aside agency action, findings, and conclusions found to be ... arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with the law,” 5 U.S.C. § 706. Even assuming the truth of those findings, however, NEPA is “a procedural, not substantive^] environmental statute.” In re Katrina Canal Breaches, 647 F.Supp.2d at 717. It is “‘now well settled that NEPA itself does not mandate particular results, but simply prescribes the necessary process.’ ” City of Shoreacres v. Waterworth, 420 F.3d 440, 450 (5th Cir.2005) (quoting Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 350, 109 S.Ct. 1835, 104 L.Ed.2d 351 (1989)). “Indeed, NEPA does not prohibit the undertaking of federal projects patently destructive of the environment; it simply mandates that the agency gather, study, and disseminate information concerning the projects’ environmental consequences.” Sabine River Auth. v. U.S. Dep’t of Interior, 951 F.2d 669, 676 (5th Cir.1992). NEPA’s procedural mandates require agencies to inform their discretion in decisionmaking. An agency that complies with NEPA gives outside influences (the public, lawmakers, other agencies) more information with which to put pressure on that agency, but the original agency retains substantive decisionmaking power regardless. At most, the Corps has abused its discretion — an abuse explicitly immunized by the DFE. 2. Duty To Armor MRGO. As their second attempt to negate the first DFE prong, the Robinson plaintiffs allege that the Corps had a mandate from Congress to maintain MRGO at a certain size: “The channel was to be 36 feet deep and 500 feet wide, increasing at the Gulf of Mexico to 38 feet deep and 600 feet wide.” In re Katrina Canal Breaches, 647 F.Supp.2d at 702. But MRGO has now eroded to an average of three times its design width. Id. at 671. The district court recognized, however, that the design for MRGO expressly contemplated erosion from wave wash and did not provide for armoring the banks. 6 The court held that these design features were “shielded by the [DFE],” id. at 702, a ruling that the plaintiffs do not challenge on appeal. Logically, therefore, the absence of armoring and the resultant erosion cannot have violated a mandate sufficient to negate the first Berkovitz prong. *451 S. Delay in Armoring MRGO. In their attempt to negate the second DFE prong, the Robinson plaintiffs allege that the critical calculations’ made by the Corps in waiting to armor MRGO were only erroneous scientific judgments, not decisions susceptible to public-policy considerations. If the government’s discretion is “grounded in the policy of the regulatory regime,” Gaubert, 499 U.S. at 325, 111 S.Ct. 1267, the decision is immune under the DFE, even if it also may entail application of scientific principles. If it is susceptible only to the application of scientific principles, however, it is not immune. 7 For the government to enjoy DFE immunity, the deciding agent need not have actually considered any policy implications; instead, the decision must only be “susceptible to policy analysis.” Spotts, 613 F.3d at 572. Under Gaubert, 499 U.S. at 324, 111 S.Ct. 1267, “the very existence” of a law or regulation allowing a government employee discretion (satisfying Berkovitz’s first prong) “creates a strong presumption that a discretionary act authorized by the regulation involves consideration of the same policies which led to the promulgation of the regulations.” The relevance of the grounds for the decision is strictly limited: “Evidence of the actual decision may be helpful in understanding whether the ‘nature’ of the decision implicated policy judgments, but the applicability of the exemption does not turn on whether the challenged decision involved such judgments.” Cope v. Scott, 45 F.3d 445, 449 (D.C.Cir.1995) (emphasis added). As discussed above, there is ample record evidence indicating the public-policy character of the Corps’s various decisions contributing to the delay in armoring Reach 2. Although the Corps appears to have appreciated the benefit of foreshore protection as early as 1967, the record shows that it also had reason to consider alternatives (such as dredging and levee “lifts”) and feasibility before committing to an armoring strategy that, in hindsight, may well have been optimal. The Corps’s actual reasons for the delay are varied and sometimes unknown, but there can be little dispute that the decisions here were susceptible to policy considerations. 8 Whatever the actual reasons for the delay, the Corps’s failure to armor timely Reach 2 is shielded by the DFE. 9 B. Applying the DFE to the Corps’s Dredge Permitting Process. The Anderson plaintiffs fare no better. In appealing the dismissal of their case, they rely on the following regulation: *452 § 320.4 General policies for evaluating permit applications. (a) Public interest review. (a)(1) The decision whether to issue a permit will be based on an evaluation of the probable impact including cumulative impacts of the proposed activity and its intended use on the public interest. Evaluation of the probable impact which the proposed activity may have on the public interest requires a careful weighing of all those factors which become relevant in each particular case. The benefits which reasonably may be expected to accrue from the proposal must be balanced against its reasonably foreseeable detriments. The decision whether to authorize a proposal, and if so, the conditions under which it will be allowed to occur, are therefore determined by the outcome of the general balancing process. That decision should reflect the national concern for both protection and utilization of important resources. All factors which may be relevant to the proposal must be considered including the cumulative effects thereof: among those are conservation, economics, aesthetics, general environmental concerns, wetlands, cultural values, fish and wildlife values, flood hazards, flood plain values, land use, navigation, shore erosion and accretion, recreation, water supply and conservation, water quality, energy needs, safety, food and fiber production, mineral needs and, in general, the needs and welfare of the people. No permit will be granted unless its issuance is found to be in the public interest.[ 10 ] The Anderson plaintiffs argue that the Corps’s issuance of dredging permits for the 17th Street Canal fails both prongs of the Berkovitz test. Prong one, they contend, is not met, because the regulation imposes a “non-discretionary duty” on what the Corps does; it is “replete with language couched in mandatory terms such as ‘will,’ ‘must,’ ‘requires,’ ‘determined,’ and ‘should,’ ” with such terms’ being included in almost every sentence. Prong two, they contend, is not met, because the regulation implicates matters of scientific judgment. The argument fails on both tries. The regulation “leaves it to a federal agency or employee to determine when and how to take action,” meaning that the Corps had discretion, satisfying prong one; “the ostensibly mandatory language,” “when read in light of the broad goals” of the regulation, allowed for the exercise of judgment and choice. Freeman, 556 F.3d at 337, 338 (citation omitted). The regulation instructed the Corps to consider several factors, some technical (e.g., shore erosion and accretion) but also many that concern policy (summarized by the catch-all “needs and welfare of the people”), satisfying prong two as well. The DFE is applicable to the Anderson plaintiffs, because the Corps’s action under the regulation satisfies both prongs of the Berkovitz test. V. Robinson Cross-Appeals. Four Robinson plaintiffs cross-appeal the ruling that the Corps is not liable for some or all of the flooding that destroyed their homes. We review their legal arguments de novo and their factual arguments under the deferential clear-error standard. A. The Robinsons. Norman and Monica Robinson, who reside within the New Orleans East polder, claim that MRGO created a “funnel effect” that increased the power of Hurricane Ka *453 trina’s storm surge such that the Reach 1 levee was breached and the Citrus Back levee was overtopped, exacerbating the flooding of their house. They also contend that the Corps should have installed a surge barrier, which would have prevented any significant flooding in New Orleans East. In re Katrina Canal Breaches, 647 F.Supp.2d at 696. They contend that the Corps was negligent in its maintenance and operation of MRGO and its failure to create a surge-protection barrier. The district court found that the Corps permissibly relied on the “Bretschneider and Collins Report,” which stated that MRGO’s effect on storm surge would generally be minimal. Id. Therefore, the court reasoned, “a duty did not exist to construct a surge protection barrier,” and the Corps was not negligent for failing to construct the barrier or for the effects of a widened MRGO. Id. at 697. In addition, the court found that the complained-of “funnel effect” would have been present in MRGO’s initial design and construction, meaning the Corps cannot be held liable for the plaintiffs’ damages because of the DFE. Id. Plaintiffs contend that the court erred in finding that the Corps reasonably relied on the Bretschneider and Collins report, because later studies called it into question, a fact allegedly ignored by the district court. More specifically, the plaintiffs contend that though the Corps may have reasonably relied on the report when it was released, it was unreasonable to rely on it for the next thirty years while new reports came out calling its conclusions into question. Despite these contentions, the court did not clearly err in finding that the Corps was reasonable in relying on the report. Contrary to the plaintiffs’ contentions, the court did consider the later reports but still found that the Corps’s actions were reasonable and that it had no duty to erect a storm-surge barrier. See id. at 696-97. Moreover, there were several later studies and occurrences that supported the Bretschneider and Collins report’s conclusions, including a 2003 Corps study and the experience of Hurricane Camille. Id. at 678. Thus, the district court did not commit clear error when it determined that the Corps had no duty to construct a storm surge barrier based on its reasonable reliance on the report, both when it was issued and in later decades. In addition to these arguments, however, the Robinsons urge that the maintenance of MRGO itself, regardless of the installation of a surge barrier, worsened the flooding of their area by an additional six feet of water on top of the six feet that would have been there anyway. The district court did not consider that distinct theory of liability in a separate section, but its findings imply that it considered and rejected the negligent-maintenance theory as well. The court intertwined its ruling on the surge-protection barrier with its ruling on the maintenance issue. The court called attention to the fact that, had MRGO remained at its original dimensions through proper maintenance, its original design would have caused the Robinsons’ property to suffer six feet of flooding regardless. “This fact indicates that a substantial portion of the harm would have arisen from the original design and presents substantial causation issues which will not be discussed based on the Court’s finding of no negligence.” Id. at 697. If harm from the original design were legally actionable, then, it would present a substantial causation issue only if it were mixed with another harm the court found non-actionable, namely, the added flooding caused by the widened MRGO. The Robinsons have not shown that the court clearly erred in re *454 jecting their negligent-maintenance argument. B. The Franzes. The Franzes, who lived in the St. Bernard polder, allege that the Corps’s negligent maintenance of MRGO caused the destruction of their home via the breach of the Reach 2 levee. The district court found that the Corps was negligent in maintaining MRGO, which caused the breach of the Reach 2 levee and the destruction of many homes in the polder. Id. The court found, however, that the Franzes’ home was destroyed by flooding from the breaches in the Inner Harbor Navigation Canal (“IHNC”) floodwall (not caused by the Corps’s negligence in maintaining MRGO), not by floodwater from the Reach 2 levee breach, which reached the Franzes’ home some time later. Id. at 735. But, the court found, “without the Reach 2 Levee breaching, the second floor would have not been flooded and the majority of the contents would not have been damaged.” Id. It therefore awarded damages for the loss of possessions on the second floor but not for the loss of the house. The Franzes challenge the award by arguing that the court erred in finding that their home was destroyed by the IHNC waters. Claiming their evidence was unrebutted, they maintain that it was the Reach 2 waters that destroyed the foundation of the house, and the combination of the two waters standing still for three weeks destroyed its structure. Unfortunately for the Franzes, our application of the DFE moots their causation argument. Because the Corps may not be held liable for damages stemming from MRGO’s role in the St. Bernard flooding, we reverse the partial judgment in the Franzes’ favor and instead find for the government in full. VI. Conclusion. Our construction of the FCA leaves undisturbed the district court’s ruling on that issue. Our application of the DFE, however, completely insulates the government from liability. Accordingly, we REVERSE the judgments for Kent Lattimore, Lattimore and Associates, and Tanya Smith and the partial judgment for the Franzes, and we AFFIRM the judgment for the government against the Robinsons. We also AFFIRM the judgment in Anderson 11 and DENY, as moot, the government’s petition for a writ of mandamus to stay the Armstrong trial pending the Robinson decision. Footnotes [1] MRGO eventually reached a total average width of 1970 feet, well over three times its authorized width. [2] Fetch is defined as the width of open water that wind can act upon. The height of waves (such as the storm surge created by Katrina) is a function of the depth of the water as well as the width of the expanse (i.e., the fetch) over which wind impacts the water. [3] A polder is a tract of low land reclaimed from a body of water. [4] See In re Katrina Canal Breaches Consol. Litig., No. 2:05-CV-4182, 2011 WL 651946, 2011 U.S. Dist. LEXIS 16351 (E.D.La. Feb. 11, 2011); In re Katrina Canal Breaches Consol. Litig., 647 F.Supp.2d 644, 699 (E.D.La.2009); In re Katrina Canal Breaches Consol. Litig., 577 F.Supp.2d 802 (E.D.La.2008); In re Katrina Canal Breaches Consol. Litig., 533 F.Supp.2d 615 (E.D.La.2008). [5] See Cent. Green Co. v. United States, 531 U.S. 425, 121 S.Ct. 1005, 148 L.Ed.2d 919 (2001); Graci v. United States, 456 F.2d 20 (5th Cir.1971). [6] See In re Katrina Canal Breaches, 647 F.Supp.2d at 654 (" 'No channel protection [was] included in the overall cost estimate of the project’ ” even though " 'erosion due to wave wash in open areas [was] expected.’ ”) (quoting MRGO Design Memorandum 1-B (revised 1959) ¶ 19, at 5). [7] See Bear Medicine v. United States ex rel. Sec’y of Interior, 241 F.3.d 1208, 1214 (9th Cir.2001) (stating that under Berkovitz, “actions based on technical or scientific standards are not the kind of judgments meant to be protected from liability by the [DFE] because those actions do not involve a weighing of policy considerations”); Ayala v. United States, 980 F.2d 1342, 1349-51 (10th Cir.1992) (applying Berkovitz and concluding that “incorrect and inadequate” technical advice was not discretionary, because it involved "objective principles of electrical engineering”) (internal quotation marks omitted). [8] Cf. Spotts, 613 F.3d at 573 ("Whatever Maldonado’s actual decisionmaking process, it is clear that the health, safety, financial, and other feasibility. concerns implicated by the evacuation decision render that decision susceptible to policy analysis.”). [9] The Robinson plaintiffs and amici point to record evidence suggesting that the Corps flatly failed to gauge the risks posed by leaving MRGO’s banks unarmored. Much, if not all, of the proffered evidence, however, suggests negligence in the original design of MRGO, including the "funnel effect,” and does not support the theory that the Corps’s decision to delay armoring was grounded in a purely scientific misjudgment. [10] 33 C.F.R. § 320.4(a)-(a)(l) (1984). The regulation quoted is the version in effect in 1984. It has since changed in ways not relevant to this litigation. [11] Because we find the government immune in Anderson, we decline to reach its argument that the plaintiffs failed to state a claim under Louisiana law.