Full opinion text
RANDALL, Circuit Judge: Eleven years ago, a home developer applied for a federal permit to dredge canals in the navigable waters around Galveston Island, Texas. From the beginning, environmental groups and neighboring landowners have fought against the development. They sought first to persuade, and then to compel, the Army Corps of Engineers to prepare a formal environmental impact statement, under the provisions of the National Environmental Policy Act, before issuing the permit to dredge. Their attempts to persuade failed; the developer obtained a permit without preparation of an environmental impact statement. Their attempts to compel, on the other hand, could not have been more successful. The District Court for the Southern District of Texas, principally because the Corps did not adequately consider “cumulative impacts,” enjoined the developer from performing work under the permit and ordered the Corps of Engineers to prepare a comprehensive environmental impact statement on the whole of West Galveston Island. 592 F.Supp. 120 (S.D.Tex.1984). The court also enjoined the Corps of Engineers from granting additional dredging permits for this project or for any similar project on the west end of the island until the environmental impact statement is completed. Both the developer and the Corps of Engineers have appealed. The developer argues that the environmental assessment of the project that has already been prepared is adequate and that a formal impact statement is not required. The Corps of Engineers, on the other hand, argues that, although the environmental assessment may be inadequate, the district court should at most have remanded the case to the Corps for reconsideration of cumulative impacts. Both appellants claim alternatively that determining the scope of an environmental impact statement is an administrative function, not a judicial one, and that, even if it can now be said that an environmental impact statement is required, the district court erred by defining the precise parameters of the statement. We agree with the district court that the Corps has not performed an adequate analysis of the cumulative impacts that may flow from this and other developments on West Galveston Island. We also agree, however, with the developer and the Corps that the district court should not have ordered the preparation of an impact statement at this stage in the process. Because of the Corps’ failure to conduct the proper study, it is too early to tell whether an impact statement is needed. We affirm in part, reverse in part, and remand for further proceedings consistent with this opinion. I. FACTUAL AND PROCEDURAL BACKGROUND The northern shore of Galveston Island, which fronts on Galveston Bay, is riddled with lakes, coves and bayous. The shoreline consists of marshlands and mud flats. From a developer’s point of view, this area is ideal for the construction of ‘Venetian” housing developments which consist of homes built along the shores of manmade canals. There are several developments of this type on the western end of Galveston Island. Homeowners there can store their boats in the canals and have immediate access, for recreational purposes, to the waters of Galveston Bay. Just west of Eleven Mile Road, a small peninsula with Hoecker’s Point at its end extends into the bay from the north shore of Galveston Island. The waters of Dele-hide Cove form the western boundary of the peninsula; Eckert’s Bayou (sometimes called Tucker’s Bayou) lies to the east. Mitchell Development Corporation of the Southwest (“Mitchell”) owns a tract of land at the base of this peninsula, near the southern end of Eckert’s Bayou. The property lies about five miles west of the City of Galveston. At some point during the pendency of this controversy, the city annexed the property. We understand that it now lies within a tax zone that provides incentives for development. For many years, Mitchell has been trying to build a venetian-style housing development, similar to others along the northern shore of West Galveston Island, on this waterfront property. The development will be called Pirate’s Cove Section 6. Eckert’s Bayou, the proposed development’s link to Galveston Bay, is a saltwater estuary. Mitchell’s property, which borders the bayou, contains important saltwater and freshwater wetlands. There are, for example, approximately fifteen acres of freshwater marsh on the property. This area, according to some experts, constitutes the “largest natural fresh marsh complex remaining on Galveston Island.” Important marsh grasses, including spartina altemiflora, grow in the forty-four acres of saltwater wetlands that are located within the boundaries of Section 6. Many species of fish and wildlife live in these wetlands and in the bayou itself. Mitchell’s property also contains a rare coastal woodlot — one of the largest and best of only twenty-two on the Texas Gulf coast — that provides an important rest area for migratory birds. The woodlot is known as Live Oak Grove. The rest of Section 6 consists primarily of grazed grasslands. The property, in addition, houses a Kanankawa Indian archeological site. Construction of Section 6 will require excavation from navigable waters and the discharge of dredged materials. Mitchell must, therefore, obtain a Department of the Army permit before commencing work. See Rivers and Harbors Act of 1899, 33 U.S.C. § 403; Clean Water Act, 33 U.S.C. § 1344. The Secretary of the Army has delegated the authority to grant such permits to the appropriate district engineer of the Army Corps of Engineers (“Corps”). See 33 C.F.R. § 320.1(a)(2). On July 20, 1974, Mitchell applied to the Corps’ Galveston District for a permit (1) to dredge a channel in Eckert’s Bayou; (2) to dredge a canal connecting Eckert's Bayou with Delehide Cove; (3) to dredge on its property a system of canals and to thereby extend the waters of Eckert’s Bayou inland; and (4) to construct bulkheads, piers, and boat launches in the canals. The proposed dredging and construction were part of Mitchell’s overall plan to build a 475-unit venetian-style housing development on a 145-acre tract. As finally submitted, the plans proposed extensive dredging of material from the floor of Eckert’s Bayou and surrounding areas. Mitchell planned to use some of the material as fill during construction and to place the rest in designated spoil areas that will be fortified where necessary by levees. The plans called for the destruction of all of the freshwater marsh on the property, a small area of saltwater marsh, and a significant number of trees in Live Oak Grove. Mitchell also proposed, however, to preserve as a park portions of Live Oak Grove and the archeological site. The Corps notified the public and other federal agencies that the application had been filed. Mitchell met on many occasions with representatives of the United States Fish and Wildlife Service (“USFWS”) and the National Marine Fisheries Service (“NMFS”) to discuss objections that both agencies had to the proposed project. As a result of these discussions, Mitchell agreed to redesign the canals to reduce the amount of saltwater marsh that will be destroyed by the development, to designate an additional spoil site, and to take steps to ensure water quality in the canals. Neither NMFS nor USFWS requested preparation of an environmental impact statement. Both agencies, however, commented on the ecological importance of Section 6 and expressed concern about the proliferation of development in the area. NMFS said: We are confident that the [Mitchell] project alone will not severely affect the productivity of West Bay____ However, we are concerned with the cumulative effects of the proliferation of waterfront housing developments on this estuary. Ten such developments with box-cut canals already exist along West Bay and its tributary bayous, and the Mitchell Corporation has plans for several other sections of waterfront housing developments in the Hoeckers Cove-Tucker Bayou area. We fear that such development may be approaching the level whereby the quality of water and habitat throughout the West Bay would become degraded. We, therefore, further recommend that an Environmental Impact Statement be prepared by the Corps of Engineers ... before granting any permits for future waterfront housing developments in West Bay or its tributary bays and bayous. (Emphasis added.) This recommendation was included in nearly every subsequent communication from NMFS to the Corps. The Corps recognized from the outset that its action on Mitchell’s application triggered responsibilities under the National Environmental Policy Act, 42 U.S.C. §§ 4331 et seq. (“NEPA”), which requires in certain circumstances the preparation of environmental impact statements (“EIS”) prior to federal action affecting the environment. The Corps made a preliminary determination, however, that, on these facts, NEPA did not require the preparation of an EIS, and that the Corps could discharge its NEPA responsibilities by conducting a less comprehensive review known as an environmental assessment (“EA”). This determination, which was included in the public notice, triggered several requests to prepare an EIS and to convene a public hearing on Mitchell’s permit application. In response, the Corps held a public hearing on November 6, 1975. Individuals and environmental groups raised several objections to the project, both at the hearing and in subsequent written statements. The objections were, for the most part, expressed in very general terms by laymen. Their comments focused on the loss of trees in Live Oak Grove, the destruction of wetlands and oyster reefs, hurricane evacuation, and the possibility that the development will make it easier for a hurricane to split the island in two. Many people stressed that Mitchell’s application should not be viewed in isolation, but should be assessed in the light of the cumulative impact that Pirate’s Cove Section 6 and other developments — past and future — will have on West Galveston Island. They expressed frustration with the piecemeal development of Galveston Island and urged the Corps to prepare a comprehensive EIS covering the entire west end of the island. Mitchell responded to the objections and stressed that, although some areas will be destroyed, parts of Live Oak Grove will be preserved and many acres of spartina al-temiflora will be planted. With respect to the need for a comprehensive EIS, Mitchell concluded that, because its consultants had already prepared “extensive and precise” ecological studies, “an EIS will serve no purpose in the processing of the application.” The public comments did not persuade the Corps to prepare an EIS. On May 28, 1976, Don McCoy, then district engineer for the Corps’ Galveston District, issued for Mitchell’s application an EA and a Statement of Findings. McCoy expressly recognized, as a general matter, that the cumulative impact on wetlands from surrounding developments can be devastating. He concluded, however, that, “[i]n the absence of any study concerning Galveston Island, the cumulative effects of this and similar projects cannot be estimated with any degree of success.” McCoy determined that an EIS was not required, despite this void in the data, as long as “future projects in the West Bay area are closely scrutinized for potential cumulative effects on the environment.” The Corps issued a permit to Mitchell on June 28, 1976, almost two years after the initial application. After the permit was awarded, Mitchell dredged a canal connecting Eckert’s Bayou to Delehide Cove. Thereafter, Mitchell acquired additional land adjacent to its existing tract on Eckert’s Bayou and suspended work under the 1976 permit. In February of 1978, Mitchell submitted an application for an amended permit to include in the project approximately forty acres of the newly acquired land and to redesign completely the canal system. Although the amended plans lessened the environmental impacts that can be expected from the project, USFWS and NMFS again expressed concerns. USFWS, for example, proposed extensive changes to the amended plans to protect additional saltwater and freshwater marshes. In addition, a number of individuals and groups again opposed the project. Before their concerns could be addressed, however, the administrative-review process came to an abrupt halt. On August 17, 1978, the plaintiffs, who are individuals owning land near Ec-kert’s Bayou, filed this lawsuit. The plaintiffs sued Mitchell and, through various individuals in their official capacities, the Army, the Corps, and USFWS. The complaint alleges that the Corps violated its duty under NEPA to prepare an EIS before granting Mitchell’s original permit application. It seeks both a preliminary and a permanent injunction preventing Mitchell from resuming work under the original permit and an order requiring the federal defendants to prepare an EIS for Pirate’s Cove Section 6 and otherwise to fulfill their statutory obligations. The plaintiffs requested that the EIS address, among other things, “the cumulative impact of development of West Galveston Bay and West Galveston.” They specifically asked that the EIS answer the three questions posed by NMFS. See supra note 1. Judge Cowan held a preliminary injunction hearing in early September. Expert witnesses called by the plaintiffs testified about various deleterious environmental effects that may flow from the project. They echoed, with greater specificity, many of the concerns expressed at the public hearing, including fears that the development may allow a hurricane-storm surge to split the island in two and will cause the contamination of a freshwater aquifer. In addition, the court heard testimony that the work already performed by Mitchell under the original permit — the dredging of a canal between Eckert’s Bayou and Delehide Cove and the deposit of dredged material in spoil areas — was environmentally unsound. Many experts urged the Corps to prepare a full-blown EIS. The experts, like the individuals who spoke at the public hearing, stressed the need to consider in more depth the cumulative impact that Pirate’s Cove and other bay-side developments will have on West Galveston Island and West Bay. On September 15, 1978, Judge Cowan, granted a preliminary injunction that prevented Mitchell from working on the project. He also ordered that the case proceed to trial on an expedited basis. The parties did not, however, proceed according to the expedited schedule. Instead, Mitchell announced its intention not to resume work under the original permit, and all counsel joined in a motion to continue the case until the Corps completed its review of Mitchell’s application for an amended permit. The Corps’ review process resumed shortly after Judge Cowan issued the preliminary injunction. In late September, the Corps busily began to marshal data and expert testimony to refute the plaintiffs’ claims. Many studies were generated on topics ranging from hurricane-storm surges to the nature of the soil and groundwater in Section 6. Coordination between Mitchell, USFWS, and NMFS, moreover, continued. In September of 1979, Mitchell submitted revised plans to address the concerns that had been raised over one year earlier by USFWS and NMFS. Another major revision of the project resulted which again lessened Section 6’s adverse environmental impacts. In November of 1979, Mitchell formally amended the application to reflect these changes and withdrew the February 1978 application. After the second application for an amendment was filed, Mitchell continued to work with USFWS and NMFS. USFWS, at one point, recommended that the permit application be denied because “canal development of the site is simply not possible without significant environmental damage.” USFWS eventually relented, however, and decided that, if certain conditions were imposed, the Mitchell project should be allowed to go forward. Mitchell eventually agreed to all of the conditions. In July of 1982, the Corps received final notification from other governmental entities that the Mitchell project was not objectionable. Ultimately, the following agencies expressed their approval: (1) The Texas Parks and Wildlife Department; (2) the Texas Department of Water Resources; (3) the Texas Department of Highways and Public Transportation; and (4) the Texas Historical Commission. NMFS, like USFWS, conditioned its approval of the project on Mitchell’s agreement to certain changes. Mitchell agreed to all of them except one, which the Corps eventually determined was not necessary. On December 3, 1982, the Corps finally reached a decision on Mitchell’s application for an amended permit. Alan Laubscher (“Laubscher”), who was by then the district engineer for Galveston, decided on that date to grant the amended permit without preparing an EIS. The decision is supported by an EA, which concludes that Section 6 will not have a significant impact on the environment, and a Statement of Findings, which concludes that granting the permit is in the public interest. With respect to cumulative effects, which is the primary issue involved in this appeal, Laubscher said: In the absence of any other studies concerning Galveston Island by the City or other governmental entity, the Corps of Engineers must evaluate each permit on an individual basis, carefully examining the environment of West Bay with each permit application. Uncontrolled development in an estuarine area can undoubtedly be devastating to the natural environment. However, with existing laws and regulatory constraints, projects such as this proposal can be designed to minimize, by mitigation and/or compensation, environmental degradation. Other projects pending before the Corps of Engineers or proposed in the future, including maintenance dredging projects that would affect West Bay, are currently undergoing these strict regulatory constraints. Laubscher also said of cumulative impacts: Cumulative impacts. This development is not related to any overall plan, by the applicant or other developers, to provide additional waterfront housing on Galveston Island. This project has independent utility, and will not encourage or support other related development. The Statement of Findings concludes, without explanation: “The cumulative effects on the aquatic ecosytem are insignificant.” After the Corps issued to Mitchell the amended permit, the parties again shifted their attention from administrative to judicial review. The plaintiffs strongly suggested in pretrial briefs that the inquiry under NEPA must proceed on two distinct, albeit related, levels: (1) was the Corps obligated to prepare a local, site-specific EIS for Section 6 of Pirate’s Cove? and (2) was the Corps obligated to prepare a comprehensive, regional EIS on the whole of West Galveston? The defendants seized upon this distinction as a way to narrow the issues in the case. They moved jointly for partial summary judgment solely on the issue of whether the Corps was obligated to prepare a regional, as opposed to a local, EIS. They argued that the Mitchell project does not propose regional action and that it is not intimately related to other proposals for federal action on West Galveston Island. Thus, under Kleppe v. Sierra Club, 427 U.S. 390, 410, 96 S.Ct. 2718, 2730, 49 L.Ed.2d 576 (1976), a regional EIS, in the defendants’ view, is not required. The Corps’ conclusion that Section 6 is not related to other projects, moreover, is not arbitrary and capricious, according to the defendants, because the Mitchell project will have “independent utility” and will not “irreversibly commit substantial resources” to future projects or “foreclose consideration of subsequent proposals.” The motion was supported by Laubscher’s affidavit which concludes that, absent a “regional development plan,” “the Corps of Engineers must evaluate each permit on an individual basis, carefully examining the environment of West Galveston Island and West Bay with each permit application.” In response, the plaintiffs denied that they had created two separate issues — a regional-EIS issue and a local-EIS issue— with respect to the Corps’ obligation to prepare an EIS. They filed a cross-motion for summary judgment on the general issue of whether the Mitchell permit application triggered a duty on the part of the Corps to prepare an EIS. The plaintiffs argued that, unlike the plaintiffs in Kleppe, they are not seeking a “regional” EIS; rather, they want “a ‘full’ [EIS], not a ‘regional’ EIS.” The record, the plaintiffs concluded, demonstrates that the Corps’ decision not to prepare an EIS was unreasonable and that they are entitled to judgment as a matter of law. The defendants responded with an additional motion for summary judgment, this time arguing that they are entitled to judgment as a matter of law, not just on the regional-EIS question, but on all of the issues raised by the plaintiffs. The motion relies on the administrative record which, to the defendants, shows conclusively that the Corps reasonably concluded that an EIS was not required. On March 2, 1984, the district court granted the plaintiffs’ motion for summary judgment and denied the defendants’ motions. 592 F.Supp. at 120. The court noted first that “[t]he case file, administrative record, an oral hearing, and the exhaustive briefs filed by the parties indicate the absence of any genuine issue of material fact.” Id. at 121. The court then applied the standard of review we first outlined in Save Our Ten Acres v. Kreger, 472 F.2d 463, 466-67 (5th Cir.1973). Significantly, the court determined that the issue in this case is not whether the Corps should prepare a regional EIS. 592 F.Supp. at 126 n. 6. Rather, “the Corps’ alleged failure adequately to consider the cumulative impact of the Mitchell project on the ecosystem of west Galveston Island and its adjacent waters is the gravamen of plaintiffs’ complaint.” Id. at 124 (footnote omitted). The court then moved to what, in its view, is the ultimate question: in the light of the environmental issues raised by the plaintiffs, was the Corps’ decision not to prepare an EIS “reasonable?” The court noted that NEPA requires an EIS for all “major federal actions significantly affecting the quality of the human environment.” 42 U.S.C. § 4332(2)(C). The court then evaluated the Corps’ finding of no significant impact (“FONSI”) against the CEQ definition of “significantly.” See 40 C.F.R. § 1508.27. The definition mandates consideration by federal agencies of “cumulative impacts,” viz., “the incremental impact of the action when added to other past, present and reasonably foreseeable future actions regardless of what agency (Federal or non-Federal) or person undertakes such other actions.” Id. § 1508.7. The court focused upon the two findings made by the Corps in purportedly complying with this mandate: (1) absent “other studies” by “other governmental entities],” the Corps must evaluate “each permit [for action on Galveston Island] on an individual basis,” and (2) the Mitchell development “is not related to any overall plan, by the applicant or other developers, to provide additional waterfront housing on Galveston Island.” The court determined that the Corps’ consideration of cumulative impacts was inadequate. The court specifically “set[] aside,” because it reflects a misdirected analysis, the finding that the Mitchell project is “not related to an overall plan.” 592 F.Supp. at 125-26 n. 6. To the court, the proper inquiry under the CEQ guidelines is, not whether there is an “overall plan,” but simply whether this project “is related” to other past, present and reasonably foreseeable actions. The court concluded that Section 6 is related to other development on West Galveston Island. This conclusion is based upon “an extensive review” of the record. Id. at 124. The court cited three examples to support its point: (1) Mitchell’s statement in the application for an amended permit that “future project expansion” is possible; (2) the December 1974 letter from NMFS to the Corps warning of the cumulative effects of venetian-style housing developments along West Bay; and (3) a May 1983 letter from USFWS to the Corps, which concerns a Galveston County permit to construct offshore breakwaters in connection with a Homecraft housing development, that notes that “cumulative impact [from developments on West Galveston], especially on wetlands, has been substantial.” Based on this evidence, the court concluded: The Mitchell project is clearly related to other past, present, and reasonably foreseeable developments on west Galveston Island. The Corps therefore abdicated its responsibility under NEPA by failing properly to consider cumulative impacts and instead evaluating the Mitchell permit application commensurate with, and narrowly confined to, its stated policy of evaluating “each permit on an individual basis.” AR 17:296. Accordingly, the Corps’ determination that an EIS is not required under NEPA, because the Mitchell project will not significantly affect the environment, is not sufficiently supported by the environmental record and is hence unreasonable and arbitrary. Id. at 125 (emphasis added; footnote omitted). Having found a violation of NEPA, the court sought to “fashion a remedy to fulfill, as closely as possible, the objectives of the statute.” Id. at 126. That remedy originally consisted of a two-pronged injunction: (1) an order requiring “the Corps to prepare an EIS which addresses the cumulative impact of the proposed Mitchell project on west Galveston Island and West Bay, taking into consideration all past, present and reasonably foreseeable developments” and (2) a continuation of Judge Cowan’s injunction preventing “significant work” on the Mitchell project. Id. Over five months later, in response to motions to amend, the court modified its order to state in detail the scope of the EIS that the Corps must prepare. The EIS must cover “all past, present and reasonably foreseeable projects [similar to Section 6] that may occur within the geographic area [of West Galveston Island] within five (5) years of the date of entry of this order.” Similar projects “include, but shall not be limited to, those developments, including residential developments — whether that development be a single unit or more than one unit — that incorporate dredged canals or lagoons and that are wholly or partially located upon lands subject to the'jurisdiction of the [Corps].” Id. (emphasis added). The court also clarified that the Corps is enjoined from granting permits for all “similar projects” until the EIS is completed. Later, the Corps moved for a partial stay of the court’s injunction. The court modified the injunction pending appeal to delete from the definition of similar projects the phrase “but shall not be limited to,” thus rendering the injunction, during the pend-ency of this appeal, applicable only to permits for venetian-style housing developments. In summary, it is clear that Section 6 has been the subject of much study. There have been three public hearings and many scientific reports. Mitchell, at times, has shown the kind of sensitivity to the environment for which it has been praised in the past. Still, Section 6 is a controversial project. It has come at a time of increasing public concern for the unique environmental problems facing wetlands and barrier islands in general and Galveston Island in particular. It is true, as the defendants are fond of emphasizing, that, after extensive review, no governmental entity has requested that an EIS be prepared on this project. It is also true, however, that many entities have expressed grave doubts about continuing this type of development on Galveston Island. Moreover, although the defendants have failed to point this out, USFWS has now taken the position that the EA that has been prepared for Section 6 is inadequate. In a February 3, 1983 letter to the Corps, USFWS stated: “In general we believe that this document fails to adequately convey the profound changes to wetlands and other fish and wildlife habitats that will occur on the project site.” USFWS specifically complained that the EA does not adequately address the issue of cumulative impacts. It is against this background that we, mindful that our role in the process is indeed a limited one, turn to an analysis of the parties’ contentions on appeal. II. CONTENTIONS ON APPEAL Both Mitchell and the Corps have appealed, but they emphasize different points. A. Mitchell’s Position Mitchell raises four points on appeal— one concerning the district court’s substantive decision that the Corps acted unreasonably and three concerning the scope of the relief ordered by the court. Mitchell’s first claim is simple: the administrative record conclusively shows that the Corps adequately considered cumulative impacts and reasonably concluded that any such impacts from the Mitchell project will be insignificant. To support this conclusion, Mitchell points to specific portions of the administrative record that purportedly demonstrate consideration of cumulative impacts, including the Corps’ findings that: (1) although .13 acres of saltwater wetlands will be destroyed by the Mitchell project, nine acres will be created and, (2) although 13.8 acres of freshwater wetlands will be destroyed, 1.2 acres will be preserved and will “allow the area to continue to serve its most important biological functions.” Moreover, the district court’s opinion, according to Mitchell, betrays a misunderstanding of the nature of the review mandated by Kreger and its progeny. In Mitchell’s view, a district court applying the reasonableness test to a NEPA-thresh-old decision should simply satisfy itself that the agency has taken a “hard look” at the proposed action. The district court, in Mitchell’s view, has gone much further than that and has, in reality, labeled the Corps’ action unreasonable upon a simple finding that cumulative impacts will exist, without the requisite discussion of the potential significance of those impacts. With respect to the scope of relief ordered by the district court, Mitchell makes three arguments. (1) Assuming, arguen-do, that the EA’s discussion of cumulative impacts is inadequate, the court should have simply remanded the case for reconsideration of those impacts, thus leaving to the Corps the decision whether, when properly evaluated, they mandate preparation of an EIS or whether they can again be considered in an EA. (2) Despite the district court’s disclaimer to the contrary, the court has in effect ordered the Corps to prepare a regional EIS. The Corps did not, however, abuse its discretion in determining that the environmental review of Mitchell’s permit application could properly be limited to Section 6, rather than all development on West Galveston Island. Because Mitchell’s project will have independent utility, the district court erred in ordering the Corps to prepare a regional EIS. (3) At any rate, even if a regional EIS is required, the court erred in enjoining Mitchell from proceeding with its project pending completion of the EIS. The environmental ramifications of Section 6 have been adequately reviewed, and the project should be allowed to continue while the comprehensive EIS is being prepared. B. The Corps’ Position The Corps’ brief starts from the premise that the case must, because of an inadequate evaluation of cumulative impacts, be remanded to the Corps for further consideration. The Corps argues, however, that the relief ordered by the district court exceeds the scope of its authority. The district court erred, the Corps maintains, by ordering the preparation of an EIS because, although the court found that the EA was inadequate, it did not expressly find that “there may be significant cumulative impacts from this and other related projects.” Absent such a finding, according to the Corps, the district court was empowered, at most, to remand the case to the Corps for a reassessment of cumulative impacts. In the alternative, the Corps argues that, even if the court properly ordered the preparation of an EIS, it should have allowed the Corps to determine its scope in the first instance. Finally, the Corps argues that, even if the court correctly ordered the preparation of a comprehensive EIS, the injunction against all similar developments on West Galveston Island is overbroad. According to the Corps, the injunction, to the extent that it enjoins activities other than the Mitchell project, does not make sufficiently clear the nature of the conduct that is prohibited, is not adequately supported by a statement of reasons, see Fed.R.Civ.P. 65(d), and does not adequately balance the interests involved. C. The Plaintiffs’ Position The Corps, according to the plaintiffs, did not consider cumulative impacts at all and instead blindly deferred to “other governmental entitpes]” to conduct such an analysis. They argue, moreover, that the district court did not, as the defendants assert, limit its analysis to the mere existence of cumulative impacts. The court, according to the plaintiffs, also found that cumulative impacts may be significant, a finding that is amply supported by the record. The plaintiffs also argue that the district court did not exceed the scope of its authority in ordering the preparation of a comprehensive EIS on the whole of West Galveston Island. According to the plaintiffs, defining the “geographic dimensions of an EIS is an issue traditionally decided by the judiciary.” Finally, the plaintiffs claim, the district court properly exercised its discretion in enjoining the Corps from granting permits for “similar projects” on West Galveston Island until an EIS is prepared. III. JUDICIAL REVIEW OF NEPA THRESHOLD DECISIONS We are of course primarily a reviewing court. As such, we are in every case sensitive to the proper standard of review. In administrative-law cases like this one, where the district court was itself in essence a reviewing court, there are really three standards to identify: (1) the standard to be applied in the first instance by the agency; (2) the standard to be applied by the district court in reviewing the agency’s findings; and (3) the standard to be applied by this court in reviewing the district court’s decision. See Sierra Club v. Sigler, 695 F.2d 957, 964-68 (5th Cir.1983). Because the parties in this case disagree sharply on what the standards are and on how they should be applied, we shall, before reaching the merits, set forth the standards that govern this case at all three levels. A. NEPA’s Implementing Regulations NEPA directs “all agencies of the Federal Government” to prepare impact statements for every “major Federal action[] significantly affecting the quality of the human environment.” 42 U.S.C. § 4332(2)(C). The Act also creates the Council on Environmental Quality. Id. § 4342. The CEQ has promulgated regulations, 40 C.F.R. pt. 1500, designed to “tell federal agencies what they must do to comply with the procedures and achieve the goals of [NEPA].” Id. § 1500.1. These regulations are binding on federal agencies. See Sierra Club v. Sigler, 695 F.2d at 972 (citing Andrus v. Sierra Club, 442 U.S. 347, 356-58, 99 S.Ct. 2335, 2340-41, 60 L.Ed.2d 943 (1979)). The CEQ guidelines provide a broad framework for NEPA compliance, but also direct federal agencies to promulgate their own regulations for implementing the Act. See 40 C.F.R. § 1507.-3. The Corps has complied with this mandate. See Policy and Procedures for Implementing NEPA, 33 C.F.R. pt. 230. The CEQ has directed agencies to, among other things, list (1) those actions which normally require preparation of an EIS and (2) those actions which normally do not require preparation of an EA or an EIS. See 40 C.F.R. § 1501.4. If a proposed action fits into neither category, the agency should prepare an ÉA. Id. § 1501.4(b). The Corps has determined that “regulatory permits” are actions “normally requiring an ... EA but not necessarily an EIS.” 33 C.F.R. § 230.7(e); see also 33 C.F.R. pt. 230, App. B. The purpose of an EA is to “provide sufficient evidence and analysis for determining whether to prepare an [EIS].” 40 C. F.R. § 1508.9(a)(1); see also 33 C.F.R. § 230.9(a). An EA, moreover, will result in one of two findings: (1) a finding that an EIS must, because of potentially significant environmental impacts, be prepared for the proposed action or (2) a FONSI — a finding that the proposed action “will not have significant effects on the environment.” 40 C.F.R. § 1501.4(cHe), 1508.13; 33 C.F.R. § 230.10. An EA should be brief. 40 C.F.R. § 1508.9; 33 C.F.R. § 230.9(b) (an EA “should not normally exceed 15 pages”). An EA should, however, “include brief discussions of the need for the proposal, of alternatives ..., of the environmental impacts of the proposed action and alternatives, and a listing of agencies and persons consulted.” 40 C.F.R. § 1508.9(b); see also 33 C.F.R. § 230.9(e). While the discussion of these factors should be brief, it should also be sufficient to demonstrate reasoned decision making. See Foundation on Economic Trends v. Heckler, 756 F.2d 143, 154 (D.C.Cir.1985) (“[s]imple, conclusory statements of ‘no impact' are not enough”). Moreover, “[a]n [EA] that fails to address a significant environmental concern can hardly be deemed adequate for a reasoned determination that an EIS is not appropriate.” Id. at 154. Before preparing an EA on proposed action, the Corps must consult with other federal agencies. The Fish and Wildlife Coordination Act, 16 U.S.C. §§ 661 et seq., for example, directs federal agencies to consult with USFWS and NMFS before allowing bodies of water to be dredged. Id. § 662(a); see also 33 C.F.R. § 320.4(c) (Corps regulations for coordination under the Act with USFWS and NMFS); 33 C.F.R. § 320.4(b), (c) (Corps regulations for coordination with USFWS and NMFS in cases affecting wetlands). As is readily apparent, the decision whether to prepare an EIS may turn in large part on the definition of the term significantly. The CEQ has defined significantly as follows: “Significantly” as used in NEPA requires considerations of both context and intensity: (a) Context. This means that the significance of an action must be analyzed in several contexts such as society as a whole (human, national), the affected region, the affected interests, and the locality. Significance varies with the setting of the proposed action. For instance, in the case of a site-specific action, significance would usually depend upon the effects in the locale rather than in the world as a whole. Both short- and long-term effects are relevant. (b) Intensity. This refers to the severity of the impact. Responsible officials must bear in mind that more than one agency may make decisions about partial aspects of a major action. The following should be considered in evaluating intensity: (1) Impacts that may be both beneficial and adverse. A significant effect may exist even if the Federal agency believes that on balance the effect will be beneficial. (2) The degree to which the proposed action affects public health or safety. (3) The unique characteristics of the geographic area such as proximity to historic or cultural resources, park lands, prime farmlands, wetlands, wild and scenic rivers, or ecologically critical areas. (4) The degree to which the effects on the quality of the human environment are likely to be highly controversial. (5) The degree to which the possible effects on the human environment are highly uncertain or involve unique or unknown risks. (6) The degree to which the action may establish a precedent for future actions with significant effects or represents a decision in principle about a future consideration. (7) Whether the action is related to other actions ivith individually insignificant but cumulatively significant impacts. Significance exists if it is reasonable to anticipate a cumulatively significant impact on the environment. Significance cannot be avoided by terming an action temporary or by breaking it down into small component parts. (8) The degree to which the action may adversely affect districts, sites, highways, structures, or objects listed or eligible for listing in the National Register of Historic Places or may cause loss or destruction of significant scientific, cultural, or historical resources. (9) The degree to which the action may adversely affect an endangered or threatened species or its habitat that has been determined to be critical under the Endangered Species Act of 1973. (10) Whether the action threatens a violation of Federal, State, or local law or requirements imposed for the protection of the environment. 40 C.F.R. § 1508.27 (emphasis added). B. The Standard for Judicial Review An agency’s decision to act without preparing an EIS is of course subject to judicial review. The circuits are, however, split on how that review should be conducted. In roughly half of the circuits that have decided the issue, a decision not to prepare an EIS will be overturned only if it is “arbitrary and capricious.” Other circuits employ a “reasonableness” standard of review. There are, moreover, considerable variations within these two general categories. See generally Shea, The Judicial Standard for Review of Environmental Impact Statement Threshold Decisions, 9 Envtl.Aff.L.Rev. 63 (1980). “This conflict is not merely semantic or academic.” Gee v. Boyd, — U.S. -, 105 S.Ct. 2123, 2125, 85 L.Ed.2d 487 (1985) (White, J., dissenting). In fact, three justices of the Supreme Court have recently argued that the full Court should review a case like this one to end the “disarray” and “confusion” that exists on this issue. Id. 105 S.Ct. at 2126 (White, J., joined by Brennan, J. & Marshall, J., dissenting from denial of petition for writ of certiorari). Since Save Our Ten Acres v. Kreger, 472 F.2d 463 (5th Cir.1973), we have employed the less deferential of the competing standards — the reasonableness test. The reasonableness test is clearly a “more rigorous standard ... than the rule of arbitrary and capricious review that ordinarily governs agency actions.” State of Louisiana v. Lee, 758 F.2d 1081, 1084 (5th Cir.1985). In Kreger, 472 F.2d at 463, we described the inquiry under the reasonableness test as follows: [T]he court should proceed to examine and weigh the evidence of both the plaintiff and the agency to determine whether the agency reasonably concluded that the particular project would have no effects which would significantly degrade our environmental quality. This inquiry must not necessarily be limited to consideration of the administrative record, but supplemental affidavits, depositions and other proof concerning the environmental impact of the project may be considered if an inadequate evidentiary development before the agency can be shown. Id. at 467 (citations omitted). Admittedly, our decisions applying Kreger have not been entirely “consistent” or “pellucid.” Louisiana Wildlife Federation, Inc. v. York, 761 F.2d 1044, 1054 (5th Cir.1985) (Rubin, J., dissenting). We agree with Mitchell that a court applying the reasonableness test does not have unbridled discretion to substitute its judgment for that of the agency responsible for making the decision in the first instance. It may not be possible, without degenerating to a circular definition, to articulate with more precision what an application of the reasonableness standard entails. It is clear, however, that a court applying the reasonableness test may, in certain circumstances, receive and weigh evidence beyond that in the administrative record, see Hiram Clarke Civic Club v. Lynn, 476 F.2d 421, 425 (5th Cir.1973), and should at least satisfy itself that the agency has taken a “hard look” at the environmental concerns raised by the plaintiffs and the factors made relevant by the various regulations implementing NEPA. See Vieux Carre Property Owners v. Pierce, 719 F.2d 1272, 1282 (5th Cir.1983). It is also clear that a decision to forego preparation of an EIS may be unreasonable for at least two distinct reasons: (1) the evidence before the court demonstrates that, contrary to the FONSI, the project may have a significant impact on the human environment, see, e.g., Lee, 758 F.2d at 1085, or (2) the agency’s review was flawed in such a manner that it cannot yet be said whether the project may have a significant impact, see, e.g., York, 761 F.2d at 1053; Foundation on Economic Trends, 756 F.2d at 154. The appropriate relief, moreover, depends upon which of these findings the district court makes. If the court finds that the project may have a significant impact, the court should order the agency to prepare an EIS. Lee, 758 F.2d at 1085; Kreger, 472 F.2d at 467. If the court finds, on the other hand, that the EA is inadequate in a manner that precludes making the determination whether the project may have a significant impact, the court should remand the case to the agency to correct the deficiencies in its analysis. See York, 761 F.2d at 1053 (“[we do] not order [an] ... EIS because the question of whether the Project may have significant adverse impacts is still an open one”); Foundation on Economic Trends, 756 F.2d at 154 (“until [the agency] completes such an evaluation the question whether the experiment requires an EIS remains an open one”). C. Appellate Review of the District Court’s Findings The scope of our review of district court decisions, of course, depends on (1) the nature of the question that has been decided and (2) the procedural context in which the decision was made. Mitchell argues that, because the district court granted a motion for summary judgment, we should in effect conduct a de novo review of its decision. The plaintiffs and the Corps have not bothered to address this threshold issue. Generally speaking, we owe substantial deference to a district court’s resolution of fact questions, but review de novo its pronouncements on questions of law and mixed questions of law and fact. Mitchell is, of course, correct that, when we review a summary judgment, we apply the same standards that the district court has applied. See, e.g., Galindo v. Precision American Corp., 754 F.2d 1212, 1216 (5th Cir.1985). That is because the district court’s role in the summary judgment context is simply to identify, not to resolve, questions of fact and to apply correctly the law — matters upon which we owe no special deference to district courts. The scope of our review is not, however, necessarily determined by the label that the parties or the district court have used to characterize the proceedings. See John v. State of Louisiana, 757 F.2d 698, 708 (5th Cir.1985). In fact, we may in certain circumstances disregard those labels altogether and apply the standard that is appropriate for reviewing what has really happened in the district court. See, e.g., Marathon Manufacturing Co. v. Enerlite Products Corp., 767 F.2d 214, 216-18 (5th Cir.1985) (“A trial procedure which accomplishes what the parties and court intend does not become erroneous merely because it is misnamed.”). The problem arises most often when the parties, in a case involving an extensive discovery record, administrative record, or the like, file cross-motions for summary judgment. Not infrequently, the parties in such a case do not really mean to suggest that there are no material questions of fact. Rather, they intend to submit the remaining fact questions to the district court for resolution on the existing record. If the parties and the district court intended to resolve the case in that manner, we will, despite any misnomers to the contrary, review the decision according to what has really happened — a trial on a stipulated record. See, e.g., Marathon Manufacturing, 767 F.2d at 216-18; John, 757 F.2d at 706 n. 4 (collecting cases); Vet-ter v. Frosch, 599 F.2d 630, 632-33 (5th Cir.1979). The district court expressly found that the summary judgment proof in this case showed an absence of genuine issues of material fact. Nevertheless, we are convinced that this is a case, like Marathon Manufacturing, in which we must disregard the court’s label and inquire further. As Marathon Manufacturing makes clear, it is appropriate for us to do so when (1) the district court has in reality decided questions of fact and (2) the parties have effectively agreed to that procedure. See also John, 757 F.2d at 706 n. 4. We note first that, following the submission of summary judgment motions, both sides informed the court that they had submitted all of the evidence that they had to offer. Said the court, in a docket entry: The Court heard argument ... concerning cross-motions for summary judgment. Both parties informed the Court that no significant new evidence would be adduced if the case were to proceed to trial. Accordingly, the Court CANCELLED the trial setting on 12/27/83 and indicated that it would rule on the summary judgment motions in the near future. (Emphasis added). Moreover, if ever there was a case in which the parties disagree about material questions, this is it. As our extended discussion of the administrative and judicial record should make clear, the parties, each with evidentiary support, surely disagree over at least the following: (1) whether Section 6 “may” have significant environmental impacts; (2) whether Section 6 is “related” to other projects on West Galveston Island; (3) whether the Corps’ review of cumulative impacts was sufficient; and (4) whether the Corps’ ultimate decision not to file an EIS was reasonable. At least the first two of these are questions of fact. See, e.g., Lee, 758 F.2d at 1081; Save Our Wetlands, 711 F.2d at 642; National Wildlife Federation v. Marsh, 721 F.2d 767, 782 (11th Cir.1983) (applying Fifth Circuit precedent). Because the district court may have answered questions of fact, and because the parties effectively consented to that procedure, we will review the court’s answers under the clearly erroneous test, see Fed.R. Civ.P. 52(a), notwithstanding the district court’s view that this is a summary judgment case. That test allows us to reject findings of fact only if we are “left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948). It does not matter that the district court’s decision is based on a record consisting primarily of documentary evidence. See Anderson v. Bessemer City, North Carolina, — U.S. -, 105 S.Ct. 1504, 1512, 84 L.Ed.2d 518 (1985). We will of course review de novo conclusions of law. IV. DISCUSSION A. Cumulative Impacts: Kleppe and The CEQ Regulations The discussion of cumulative impacts must start with the CEQ definition of significance and the Supreme Court’s decision in Kleppe, 427 U.S. at 398, 96 S.Ct. at 2724. The issue in Kleppe was whether the Department of the Interior and other federal agencies were obligated to prepare a comprehensive, regional EIS before granting leases to operate coal mines on federal land in the Northern Great Plains Region — an area encompassing parts of Wyoming, Montana, North Dakota, and South Dakota. The responsible agencies had already prepared a programmatic EIS of national scope on the coal-leasing program as a whole and individual statements on specific leases for specific tracts. The Department of the Interior, moreover, had announced its intention to prepare comprehensive statements on all coal leases in various regions to be “determined by basin boundaries, drainage areas, areas of common reclamation problems, administrative boundaries, areas of economic interdependence, and other relevant factors.” Id. at 411, 96 S.Ct. at 2731. The only issue, therefore, was whether the Department of Interior had violated NEPA by refusing to prepare an EIS on coal leasing in general in the broad region suggested by the plaintiffs. The plaintiffs did not challenge the Department of Interior’s environmental analysis of any specific coal lease. The Supreme Court began its analysis by noting that the obligation to prepare an EIS is triggered by a “proposal” for major federal action. “[T]he mere ‘contemplation of’ certain action is not sufficient to require an impact statement.” Id. at 404, 96 S.Ct. at 2728. The Court rejected the District of Columbia Circuit’s view that, in certain circumstances, the obligation to prepare an EIS may be triggered at some stage in the development of a contemplated project before it becomes a true proposal. The Court next considered two arguments that had been made for the preparation of a regional EIS on coal leasing in the Northern Great Plains area: (1) the Department of the Interior was contemplating a plan for regional development and (2) there were pending proposals for individual actions within the region that were “intimately related.” The Court rejected the first argument because the agencies had not proposed regional action. Contemplated regional action, as opposed to proposed regional action, cannot trigger the obligation to prepare an EIS. Absent a proposal, “there is nothing that could be the subject of the [kind of] analysis envisioned by the statute for an impact statement.” Id. at 401, 96 S.Ct. at 2726. With respect to the second contention, the Court noted that it could be read two ways: First, it amounts to an attack on the sufficiency of the impact statements already prepared by the petitioners on the coal-related projects that they have approved or stand ready to approve____ It also is possible to view the respondents’ argument as an attack upon the decision of the petitioners not to prepare one comprehensive impact statement on all proposed projects in the region. Id. at 408-09, 96 S.Ct. at 2729-30. The Court held, however, that the first of these possible constructions of the argument was not properly before it because the plaintiffs had not challenged the environmental review of any specific coal lease. With respect to the second construction, the Court agreed with the general proposition that cumulative impacts from distinct proposals pending in a given region may in certain circumstances mandate the preparation of a regional or comprehensive EIS even though a single proposal for comprehensive, regional action is not pending. Said the Court: “Thus, when several proposals ... that will have cumulative or synergistic impact upon a region are pending concurrently before an agency their environmental consequences must be considered together.” Id. at 410, 96 S.Ct. at 2730 (footnote omitted). The Court simply disagreed on the record before it with the plaintiffs’ contention that the responsible agencies had arbitrarily and capriciously determined that the coal-related proposals within the entire Northern Great Plains region were not so closely related in terms of environmental impact that they should be considered in the same EIS. The Court, in footnote 20, placed an important limitation on its recognition, of the rule that cumulative impacts may trigger the need to prepare a comprehensive EIS. Proposed actions with potential cumulative impacts may mandate the preparation of a regional or comprehensive impact statement; contemplated actions with potential cumulative impacts cannot. Said the Court: At some points in their brief respondents appear to seek a comprehensive impact statement covering contemplated projects in the region as well as those that already have been proposed. The statute, however, speaks solely in terms of proposed actions; it does not require an agency to consider the possible environmental impacts of less imminent actions when preparing the impact statement on proposed actions. Should contemplated actions later reach the stage of actual proposals, impact statements on them will take into account the effect of their approval upon the existing environment; and the condition of the environment presumably will reflect earlier proposed actions and their effects. (Emphasis added). See also Piedmont Heights, 637 F.2d at 441. Since Kleppe, new, binding CEQ regulations have been promulgated. They reflect the recognition in Kleppe and the independent-utility cases, see supra note 10, that comprehensive impact statements may sometimes be required. In fact, the regulations mirror the cases. The scoping regulation, 40 C.F.R. § 1508.25, states that actions that are either “connected” or “cumulative” should be considered in the same environmental impact statement. “Connected actions” are defined in a manner consistent with the criteria recognized in the independent-utility cases, see supra note 10. “Cumulative actions,” on the other hand, are defined in a manner consistent with Kleppe. This case must be distinguished from both Kleppe and the independent-utility eases in a manner that has apparently escaped the parties’ attention. Kleppe did not involve an attack on a specific project for which the responsible agency had, on the basis of a FONSI, refused to prepare an EIS. Rather, Kleppe and most of the independent-utility cases relied upon by the defendants involve individual proposals that undeniably will have significant impacts on the human environment and for which individual EISs were prepared. The issue in those cases was whether other actions should be considered in the same impact statements or whether separate, comprehensive impact statements were also needed. As noted, the Supreme Court made clear that, although cumulative impacts may sometimes demand the preparation of a comprehensive EIS, only the impacts of proposed, as distinguished from contemplated, actions need be considered in scoping an EIS. In a case like this one, on the other hand, where an EA constitutes the only environmental review undertaken thus far, the cumulative-impacts analysis plays a different role. See, e.g., Citizens for Responsible Area Growth v. Adams, 477 F.Supp. 994, 1002 (D.N.H.1979) (“different legal consequences flow from decisions to segment a project made prior to the threshold determination than the same decision made after the finding of significant effect”), vacated in part on other grounds, 680 F.2d 835 (1st Cir.1982). This distinction is clearly recognized in the CEQ regulations. Sections 1508.7 and 1508.27 require an analysis, when making the NEPA-threshold decision, as opposed to the EIS-scoping decision, whether it is “reasonable to anticipate cumulatively significant impacts” from the specific impacts of the proposed project when added to the impacts from “past, present and reasonably foreseeable future actions,” which are “related” to the proposed project. The regulation does not limit the inquiry to the cumulative impacts that can be expected from proposed projects; rather, the inquiry also extends to the effects that can be anticipated from “reasonably foreseeable future actions.” Cf 40 C.F.R. § 1508.25(a)(2) (cumulative actions are “proposed actions ... ”). In other words, when deciding the potential significance of a single proposed action (i.e., whether to prepare an EIS at all), a broader analysis of cumulative impacts is required. The regulations clearly mandate consideration of the impacts from actions that are not yet proposals and from actions — past, present, or future — that are not themselves subject to the requirements of NEPA. See 40 C.F.R. § 1508.7 (“past, present, and reasonably foreseeable future actions regardless of what agency (Federal or non-Federal) or person undertakes such other actions ”) (emphasis added). This requirement, moreover, is entirely consistent with Kleppe. Said the court in Adams: Both ... footnote [20] and Kleppe itself refer to cases in which an EIS is prepared. But contemplated actions which have not reached the proposal stage may