Affirmed by published opinion. Judge NIEMEYER wrote the opinion, in which Chief Judge TRAXLER and Judge
WILKINSON joined. Judge WILKINSON wrote a separate concurring opinion.
OPINION
NIEMEYER, Circuit Judge:
In connection with a proposed surface coal mine adjacent to Reylas Fork (a stream) in Logan County, West Virginia, the West Virginia Department of Environmental Protection (“WVDEP”) issued Highland Mining Company a permit under the Surface Mining Control and Reclamation Act (“SMCRA”) to do the mining, finding that the proposed mine would not cause material damage to the hydrologic regime. The WVDEP also issued a water quality certification under § 401 of the Clean Water Act (“CWA”), concluding that the proposed mine would not cause or contribute to violations of the State’s EPA-approved water quality standards, as well as a National Pollutant Discharge Elimination System (“NPDES”) permit under CWA § 402, finding that the proposed sediment pond for the mine would not have significant adverse effects. Finally, the U.S. Army Corps of Engineers (the “Corps”) issued a fill permit under CWA § 404, authorizing Highland Mining to place rock overburden into the adjacent valley of Reylas Fork as part of the mining process. The Corps issued the permit without an environmental impact statement, finding that the fill would not have a substantial cumulative impact on the water quality in the relevant watershed.
Four environmental groups (collectively, the “Environmental Coalition”) commenced this action to challenge the fill permit issued under CWA § 404. The Environmental Coalition contends that the Corps, in conducting its analysis for the § 404 permit, “materially misapprehended” the baseline conditions in the relevant watershed, thus corrupting its analysis of [122]*122the cumulative impact that the mine would have on the streams in the watershed. It also contends that the Corps acted arbitrarily and capriciously in determining that the valley fill would not have a significant cumulative impact on the water quality in the relevant watershed.
The district court evaluated the data considered by the Corps, the Corps’ analysis, and the Corps’ conclusions and found that the Corps did not misapprehend the baseline conditions in the relevant watershed. Ohio Valley Envtl. Coalition, Inc. v. U.S. Army Corps of Eng’rs (“OVEC ”), 883 F.Supp.2d 627, 642-44 (S.D.W.Va.2012). It also found that the Corps analyzed a “wide array of evidence about water quality” to reach a reasoned decision that the individual and cumulative environmental impacts of the Corps’ CWA § 404 permit would not rise to the level of significance required to trigger the need for an environmental impact statement. Id. at 645. As the district court concluded, “[t]he Corps has analyzed the cumulative impacts, ‘articulated a satisfactory explanation for its conclusion,’ and thus has not acted arbitrarily or capriciously.” Id. (quoting Ohio Valley Envtl. Coalition v. Aracoma Coal Co., 556 F.3d 177, 209 (4th Cir.2009)).
For the reasons that follow, we affirm.
I
The mining operation proposed at the Reylas mine would involve removing mountaintop rock that covers the coal seams to be mined, placing the rock in the adjacent valley, extracting the coal, and replacing the rock on the mountaintop. See Bragg v. W. Va. Coal Ass’n, 248 F.3d 275, 286 (4th Cir.2001) (describing the process). Because the rock is broken up as it is removed from the mountaintop, its volume swells such that all of the rock would not be needed to restore the mountaintop. The excess rock, or overburden, would accordingly be left in the valley as permanent fill. See Kentuckians for the Commonwealth, Inc. v. Rivenburgh, 317 F.3d 425, 431 (4th Cir.2003). A sediment pond, as was typical, would be constructed below the valley fill to collect water until the valley fill is stabilized. See Aracoma Coal, 556 F.3d at 186. In connection with the Reylas mine, the overburden rock would be deposited in a valley that includes a stream called Reylas Fork. Reylas Fork flows into Bandmill Hollow, which in turn flows into Dingess Run. The Dingess Run watershed was the relevant area for assessing the mine’s cumulative impact.
To operate the Reylas mine, Highland Mining obtained a SMCRA permit from the WVDEP, a state agency that has assumed “exclusive jurisdiction over the regulation of surface coal mining and reclamation operations” on nonfederal lands. 30 U.S.C. § 1253(a). Highland Mining also obtained from the WVDEP a permit under CWA § 401, certifying that the discharge from the mine will be consistent with the State’s water-quality standards, and a permit under CWA § 402, authorizing the discharge of pollutants other than dredged or fill material into navigable waters. Finally, Highland Mining obtained a permit under CWA § 404 from the Corps to allow “the discharge of dredged or fill material into the navigable waters at specified disposal sites.” 33 U.S.C. § 1344(a).
Under guidelines issued by the EPA, the Corps could issue a § 404 permit only after concluding that the mining activity would not cause or contribute to violations of the State’s water-quality standards or to the significant degradation of waters of the United States. 40 C.F.R. § 230.10(b)(1), (c). In making that assessment, it had to comply with the National Environmental Policy Act (“NEPA”) and take a “hard look” at the potential environmental conse[123]*123quences of the activity. See Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 350, 109 S.Ct. 1835, 104 L.Ed.2d 351 (1989). NEPA required federal agencies such as the Corps to prepare an environmental impact statement (“EIS”) for major federal actions that significantly affect the quality of the human environment. 42 U.S.C. § 4332(C). To decide if an EIS was required, the Corps had to prepare an environmental assessment. 33 C.F.R. § 230.10; 40 C.F.R. § 1501.3. If, after conducting the assessment, the Corps found that the proposed project would not, either individually or cumulatively, have a significant impact on human health or the environment, it could make a “finding of no significant impact” (“FONSI”), obviating the need for preparing an EIS. 33 C.F.R. § 230.11; see also 40 C.F.R. § 1508.13 (describing the contents of a FONSI).
Before evaluating Highland Mining’s application for a § 404 permit, the Corps gave public notice of the application and invited comment. The Environmental Coalition — consisting of Ohio Valley Environmental Coalition, Inc.; West Virginia Highlands Conservancy, Inc.; Sierra Club; and Coal River Mountain Watch, Inc.— submitted comments opposing Highland Mining’s application. The Coalition maintained that the disturbance from the Rey-las mine would be ecologically significant, as the “sheer volume of disturbance in the region had already indicated significant degradation and would only be made worse by additional mining.” The EPA also submitted comments stating that “this proposal is likely to cause or contribute to an excursion from the State’s water quality standards downstream resulting in an impairment of the aquatic life use.”- The EPA warned that “the direct and cumulative impacts from this and future mines will be persistent and permanent and can not be sufficiently or effectively compensated through the proposed mitigation.”
After receiving the EPA’s comments, Highland Mining requested that the Corps stay its consideration of the permit application until Highland Mining had an opportunity to allay the EPA’s concerns. Thereafter, the Corps, the EPA, and Highland Mining consulted each other and agreed to modifications to the conditions of the permit. Under the agreement reached, Highland Mining would eliminate approximately 400 feet of impact to Reylas Fork and apply a mitigation plan involving best management practices and monitoring. After the Corps made these changes, the EPA advised the Corps, “We believe that with appropriate permit conditions that the applicant could move forward with the issuance of the permit.”
The Corps released its Combined Decision Document and § 404 permit on March 4, 2011, in which it included a FONSI with respect to the Reylas mine.
The Environmental Coalition shortly thereafter commenced this action against the Corps, its Commander, and its District Engineer, alleging that the Corps’ failure to require an EIS violated NEPA and was arbitrary and capricious “because the Corps failed to take a hard look at the environmental impacts of those projects.” The Coalition also alleged that the Corps arbitrarily and capriciously concluded that the § 404 permit would not result in violations of West Virginia’s water-quality standards. The district court granted Highland Mining’s motion to intervene as a defendant.
On cross-motions for summary judgment, the district court denied the Environmental Coalition’s motion for summary judgment; granted the Corps’ and Highland Mining’s cross-motions for summary judgment; and entered final judgment. OVEC, 883 F.Supp.2d at 645. The court rejected the Coalition’s argument that the [124]*124Corps, in conducting its analysis, “misapprehended” the baseline conditions of Din-gess Run. Id. at 642-44. It also rejected the Coalition’s NEPA challenge, finding that the Corps took the requisite “hard look.” Id. at 644. It noted that the Corps reasonably relied “on the expertise of the WVDEP, the agency with primary responsibility for water quality, in determining that impacts on water quality will be insignificant.” Id. at 645.
The Environmental Coalition filed this appeal and at the same time requested that the district court issue an injunction to prevent Highland Mining from filling Reylas Fork valley pending appeal. When the district court denied the motion, the Coalition filed a similar motion in this court, which we too denied, by order dated September 24, 2012.
II
The Environmental Coalition contends first that the Corps, in determining the baseline conditions of the relevant watershed, which was part of its analysis of the cumulative impact of the proposed mining activity, made “a material factual error” and “misapprehended” the baseline conditions. It explains:
The Corps’ [Combined Decision Document] repeatedly stated that [Dingess Run] had good water quality, healthy aquatic life, and no significant mining impacts.... In fact ... WVDEP has measured Dingess Run as being in “poor” biological condition and listed its entire 7.4-mile length as an impaired water under § 303(d) of the CWA.
This argument requires us to evaluate both the Combined Decision Document, in which the Corps makes its baseline assessment of current conditions in the relevant watershed, and the § 303(d) document, in which the WVDEP listed Dingess Run as among the State’s impaired streams. But our review is limited to deciding whether the Corps considered the “relevant factors” when assessing the baseline conditions of the watershed. See Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983).
In a 2010 report, the WVDEP included Dingess Run on its list of impaired streams for purposes of complying with CWA § 303(d). The WVDEP’s report explained that Dingess Run was included on the list based on measurements of a portion of a stream, using six benthic community metrics — i.e., metrics evaluating the stream bed for biological life — which were combined into a “single multi-metric index.” A stream scoring above 68 using that index was categorized as biologically unimpaired; a stream scoring between 60.6 and 68 fell within a “gray zone”; and a stream scoring less than 60.6 was considered biologically impaired. Although the report found that Dingess Run was “impaired,” it cautioned that the scope of its finding was limited:
Most streams with low biological scores are listed as having an unknown source/ cause of impairment on the 303(d) List and most are listed, by default, for their entire length. It is doubtful that the entire length of every stream is impaired, but without further data, the exact length of impairment is unknown.
J.A. 327. Thus, the WVDEP § 303(d) list represented, based on limited data, the assessment that Dingess Run was “impaired” when measured against the “single multi-metric index” and that the causes for the impairment were unknown.
This impairment finding undoubtedly raises the question of whether Dingess Run would be able to assimilate the effects of the Reylas mine, particularly the effects of the valley fill that was the sub[125]*125ject of the Corps’ § 404 permit. Two distinct analyses conducted in the Combined Decision Document, however, convince us that the Corps addressed the question only after considering the relevant data about baseline conditions and properly assessing them. First, the Corps analyzed the conditions at the fill site itself. And second, the Corps recognized and analyzed the impaired conditions of the streams in the relevant watershed.
In its analysis of the conditions at the site of Reylas mine fill, the Corps exhaustively reviewed the relevant data, reporting:
The streams proposed for impact at the site have perennial, intermittent and ephemeral flow regimes. The perennial and intermittent streams exhibit relatively good water quality based on analytical methods developed and approved by the USEPA and WVDEP.... Based on USEPA’s Rapid Bioassessment Protocol, the streams fall in the middle range of “fair-good” range for habitat features, and have good West Virginia Stream Condition Index (WVSCI scores). The ephemeral streams lack sufficient water flow to conduct these analyses. Within the perennial and intermittent streams, the macroinverteb-rate analysis indicates a good benthic community [representing the vitality of the stream bed] with good diversity with a dominance of shredders. Approximately 6% of the species were comprised of the sensitive mayfly species. The presence of these types of benthic species indicates the stream has the ability to provide functions such as transport of nutrients, decomposition of organic input/detritus, and food web support. Overall, the perennial and intermittent stream channels located on-site are of good quality and do provide several important functions. The ephemeral streams, due to their lack of flow throughout the majority of the year, do not provide these same functions.
J.A. 180-81. The Corps followed up this summary by detailing the supporting data collected with respect to “physical habitat,” “water quality,” “benthics,” and “stream functioning.”
With respect to physical habitat, the Corps found that, overall, five of the eight streams proposed for impact exhibited good physical habitat, despite previous logging and deep mine activities at the site.
With respect-to water quality, the Corps referred to data collected in April 2004 and July 2007. The 2004 samplings were conducted during the rainy season and presented
overall good water quality for the majority of parameters tested. Two sites, located within the perennial segments, had slightly elevated conductivities [dissolved salts], while the site located higher in the watershed had a relatively low conductivity reading.
J.A. 203. The samplings revealed that magnesium was higher than recommended but concluded:
Although these parameters are slightly high, they do not appear to have resulted in adverse effects on the benthic communities at the sample sites as the data indicates thriving benthic communities.
Id. The 2007 sampling found that the conductivities were higher, although magnesium levels remained similar to the 2004 levels. As to the water quality aspect, the Corps concluded: ■
Overall, water quality on-site is good with the exception of elevated conductivity and magnesium readings in the lower perennial and intermittent sections of Reylas Fork. Based on data obtained in downstream segments of Bandmill Hoi-[126]*126low, the receiving stream and Dingess Run magnesium levels fall within acceptable parameters established by WVDEP.
Id.
With respect to benthics, the Corps noted that, based on testing conducted in April 2004 and July 2007, “the benthic community within the perennial and intermittent stream channels contained good benthic communities although there were slight differences in species diversity.” J.A. 203. The Corps stated:
[T]he unnamed first right tributary of Reylas Fork presented the most diverse benthic community however the WVSCI scores indicated very good biological integrity at each sampling location. The remaining streams proposed for impact are ephemeral and did not contain benthic communities. Overall, the intermittent and perennial stream channels at the site contain good aquatic habitat despite the slightly elevated levels of conductivity and magnesium present in several of the streams at the site.
Finally, with respect to stream functioning, the Corps found that “the majority of streams proposed for impacts are fully functional and of high quality based primarily on their position within the landscape, and how the stream interacts with geology, hydrology and soils, which influence biotic and physical processes.” J.A. 207.
With the specific water data evaluated and the impairments noted, the Corps’ Combined Decision Document concluded with a summary that the streams at the impact site
are of good quality with good physical aquatic habitat, good riparian cover, good water quality (with the exception of elevated conductivity and magnesium levels) and viable benthic communities.
J.A. 209.
After looking at site-specific factors, the Corps turned its attention to the Dingess Run watershed as a whole. It recognized, at the outset, that “deep and surface coal mining and timber removal activities have been on-going within the Dingess Fork Watershed ... for the past 150 years.” J.A. 247. Because most of this activity was unregulated, “many areas within the Appalachian Coal Fields have been adversely impacted by these activities.” Id. The Corps stressed the importance of these earlier mining activities:
These pre-SMCRA and pre-CWA impacts are relevant as they have set a baseline condition for water quality and aquatic habitat values within the watershed. They also provide the context for purposes of assessing the significance of impacts in our NEPA analysis.
Id. (emphasis added).
Taking this history into account, the Corps then considered water-quality testing data provided by Highland Mining “from eight baseline surface monitoring sites located on Bandmill Hollow, as well as seven streams which flow into Dingess Run.” J.A. 247.
The Corps’ cumulative analysis thus considered data not only from the impact area, but also from other tributaries. The sampling from Bandmill Hollow, the area that would be directly impacted by the Reylas mine valley fill, indicated elevated levels of conductivity. See J.A. 203; J.A. 254. The samplings from other upstream tributaries of Dingess Run revealed “slightly elevated selenium levels.” J.A. 248; see also J.A. 253. Thus, the Corps acknowledged that the data indicated a level of impairment. But it concluded that this level of impairment did not prevent [127]*127the streams in the Dingess Run watershed from absorbing the impacts of past mining:
[DJespite past unregulated mining, which has impacted the relevant environment and provides the context for this assessment, the watershed is heavily forested and continues to provide a functioning aquatic ecosystem with good water quality.
J.A. 256-57. Taking into account the historical damage as well as current sampling, the Corps found that “[rjecent water quality and biological data has indicated the watershed is sufficiently absorbing the impacts without significant aquatic impairment and/or degradation.” J.A. 256.
These observations and assessments do not support the Environmental Coalition’s claim that the Corps “misapprehended” existing conditions. The Corps’ conclusion that the Dingess Run watershed had good water quality was a contextual judgment made after considering all relevant data. Moreover, the Corps’ discussion of the baseline conditions was more exhaustive than that found in West Virginia’s discussion made in connection with its § 303(d) list and not necessarily in tension with it. Specifically, the Corps recognized that the watershed had elevated levels of conductivity and selenium. See J.A. 203-07 (identifying elevated levels of conductivity); J.A. 248 (acknowledging “slightly elevated selenium levels”); J.A. 253 (“Several streams have exhibited elevated selenium levels during some sampling events”); J.A. 254 (recognizing high conductivity levels). The § 303(d) document, on the other hand, was not pollutant-specific, concluding only that the Dingess Run was “impaired.”
We thus find no merit to the Environmental Coalition’s claim that the Corps “misapprehended” the baseline conditions. The Corps considered the relevant factors, evaluating both the impact site and the entire watershed. Only after this evaluation did the Corps reach its informed judgment as to the baseline conditions.
Ill
For its second argument, the Environmental Coalition contends that the Corps’ finding of cumulative insignificance was “arbitrary and capricious” because the Corps irrationally dismissed the strong correlation between surface coal mining activities and downstream biological impairment. Under the Coalition’s reading of the Combined Decision Document, the Corps failed to take a “hard look” at potential environmental consequences because the Document is “not supported by any reasoned analysis of, or expert opinion about, the science on conductivity and stream impairment.” Stated otherwise, the Coalition challenges the rationality of the Corps’ predictive judgment that the valley fill from the Reylas mine will not have a cumulatively significant impact on the streams in the Dingess Run watershed.
The district court rejected the Environmental Coalition’s argument, observing that the Corps had before it
a wide array of evidence about water quality and the effects of conductivity including the SMCRA permit, the NPDES permit, the § 401 certification, and- the studies provided by the Plaintiffs and the EPA. The [Combined Decision Document] reflects a substantive review of these materials, after which the Corps concluded that the individual and cumulative environmental impacts of this permit would not rise to the level of significance required to trigger an EIS.
OVEC, 883 F.Supp.2d at 645. The court thus concluded that the Corps had “analyzed the cumulative impacts, ‘articulated a satisfactory explanation for its conclusion,’ and thus has not acted arbitrarily or capri[128]*128ciously.” Id. (quoting Aracoma, 556 F.3d at 209).
In assessing whether a project’s impacts will be significant, the Agency must take a “hard look” at potential environmental consequences. Robertson v. Methow Valley Citizens Council, 490 U.S. 332, 350, 109 S.Ct. 1835, 104 L.Ed.2d 351 (1989). “The hallmarks of a ‘hard look’ are thorough investigation into environmental impacts and forthright acknowledgment of potential environmental harms.” Nat’l Audubon Soc’y v. Dep’t of Navy, 422 F.3d 174, 185 (4th Cir.2005).
In this case, the Corps collected the competing views of the Environmental Coalition, the EPA, the WVDEP, and Highland Mining and examined them in some detail, along with the supporting data. Indeed, a substantial portion of the Combined Decision Document is dedicated to addressing water quality and the cumulative effect of the Reylas mine fill on water quality. The Corps devoted four years to the effort, meeting repeatedly with the parties to resolve concerns, prompting the district court to observe:
The administrative record documents the extensive interaction among the Corps, the EPA, the WVDEP and Highland to resolve the EPA’s concerns. Though Highland and the WVDEP disputed the EPA’s position on conductivity and cumulative water quality impacts of valley fills, ■ the Corps focused on site-specific factors which provide at least a rational basis for its decision.
OVEC 883 F.Supp.2d at 644.
In response to the EPA’s concerns, the Corps adopted measures agreeable to the EPA as conditions of its § 404 permit, involving reduction of the fill and post-permit monitoring and other mitigation requirements. These special conditions incorporated a series of best management practices designed to minimize increases in conductivity and total dissolved solids associated with the mining activities of Highland Mining. The conditions also required that if the monitoring showed that the mining activities were resulting in adverse impacts to water quality, Highland Mining would be required to initiate remedial actions, provide additional water quality-based mitigation under the terms of the permit, or both. J.A. 265-66. All of these conditions were incorporated as conditions of the § 404 permit. This accommodation process was significant and, indeed, resulted in a two-year delay of the issuance of the § 404 permit.
In addition to these modifications to the § 404 permit, the Corps’ Combined Decision Document gave a “hard look” at conductivity and water quality more generally. After reciting the history of mining in the Dingess Run watershed over the period of 150 years and its impact, the Document evaluated the data from various sources which indicated that the watershed would still be able to absorb the impacts of the Reylas mine without significant aquatic impairment or degradation.
The Document specifically discussed the elevated levels of conductivity and selenium in the streams. As to the selenium, it concluded that the source of selenium in that area is located “in a stratum or strata located in rock formations higher in elevation than those proposed to be mined as part of this project,” J.A. 248, so that the selenium level would not be increased by the mine. And as to conductivity, the Document compared elevated conductivity levels found in areas downstream of other valley fills with those of streams containing no valley fills. It found that “benthic mi-croinvertebrate communities do become established downstream of valley fills and often are comparable to those communities in un-impacted stream channels.” J.A. 254. Indeed, the conclusions were sup[129]*129ported in part by a study that showed that streams below valley fills. sometimes did better than the streams constituting the reference sites. J.A. 262.
At bottom, the Document reached the conclusion that “the valley fill, sediment pond, and mine through activities, if conducted in accordance with all applicable state and Federal regulations, should not contribute to or result in cumulative significant adverse impacts to the aquatic or human environment within the Dingess Run Watershed.” J.A. 256.
Thus, contrary to the Environmental Coalition’s contention that the Corps failed to take a hard look at conductivity and stream impairment, the record amply shows that the Corps grappled with the issue extensively, rationally finding that (1) the connection between conductivity and stream impairment was not strong enough to preclude a permit and (2) the compromise measures agreed to by the EPA and Highland Mining would successfully mitigate the potential for adverse effects.
With the inability to demonstrate that the Corps failed to take a “hard look,” the Environmental Coalition’s arguments are reduced to no more than a substantive disagreement with the Corps. But our review is limited, and we may not “use review of an agency’s environmental analysis as a guise for second-guessing substantive decisions committed to the discretion of the agency.” Nat’l Audubon Soc’y, 422 F.3d at 185 (citing Robertson, 490 U.S. at 350, 109 S.Ct. 1835).
The Corps’ predictive judgment in this case was based on facts and recommendations, adduced during a lengthy consultation between the Corps, Highland Mining, the EPA, and the WVDEP, and we conclude that this process satisfies NEPA’s procedural requirement to take a “hard look.” See Hughes River Watershed Conservancy v. Johnson, 165 F.3d 283, 288 (4th Cir.1999) (“[A]n agency takes a sufficient ‘hard look’ when it obtains opinions from its own experts, obtains opinions from experts outside the agency, gives careful scientific scrutiny and responds to all legitimate concerns that are raised” (citing Marsh v. Oregon Natural Resources Council, 490 U.S. 360, 378-85, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989))). Because the Corps’ analysis satisfied NEPA’s procedural requirements, the Corps’ finding of cumulative insignificance is neither arbitrary nor capricious. See Aracoma Coal, 556 F.3d at 209.
Accordingly, we affirm the judgment of the district court.
AFFIRMED