I respectfully dissent. I do not believe that the decision of respondent Minnesota Public Utilities Commission (commission) was arbitrary and capricious or unsupported by substantial evidence. Relator Honor the Earth (HTE) contends that the final environmental-impact statement (FEIS) failed to consider the effect of oil spills on the Lake Superior watershed. HTE is wrong. It did.
We review the commission's decision under the Minnesota Administrative Procedure Act (MAPA) to determine whether
the substantial rights of the [relators] may have been prejudiced because the administrative finding, inferences, conclusion, or decisions are:
(a) in violation of constitutional provisions; or
(b) in excess of the statutory authority or jurisdiction of the agency; or
(c) made upon unlawful procedure; or
(d) affected by other error of law; or
(e) unsupported by substantial evidence in view of the entire record as submitted; or
(f) arbitrary or capricious.
Minn. Stat. § 14.69 (2018) ; see Minn. Stat. § 116D.04, subd. 10 (2018) (directing review under MAPA). We "accord substantial deference to the agency's decision." Citizens Advocating Responsible Dev. v. Kandiyohi Cty. Bd. of Comm'rs , 713 N.W.2d 817, 832 (Minn. 2006). We "adhere to the fundamental concept that decisions *37of administrative agencies enjoy a presumption of correctness , and deference should be shown by courts to the agencies' expertise and their special knowledge in the field of their technical training, education, and experience." Reserve Mining Co. v. Herbst , 256 N.W.2d 808, 824 (Minn. 1977) (emphasis added).
The "hard look" analysis relating to environmental review is borrowed from federal cases reviewing agency decisions. See id. at 825 (adopting language from Greater Boston Tel. Corp. v. F.C.C. , 444 F.2d 841, 851-53 (D.C. Cir. 1970) ). In reviewing the adequacy of an EIS, federal courts have applied a "rule of reason." See No Power Line, Inc. v. Minn. Envtl. Quality Council , 262 N.W.2d 312, 327 (Minn. 1977) (collecting federal cases). Describing that standard, one federal court has stated:
This does not mean that the courts are to "fly speck" environmental impact statements. The preparation of such a statement necessarily calls for judgment, and that judgment is the agency's. But the courts can, and should, require full, fair, bona fide compliance with [the National Environmental Policy Act].
Lathan v. Brinegar , 506 F.2d 677, 693 (9th Cir. 1974), quoted with approval in No Power Line , 262 N.W.2d at 327.
An environmental-impact statement (EIS) is an "analytical rather than an encyclopedic document which describes the proposed action in detail, analyzes its significant environmental impacts, discusses appropriate alternatives to the proposed action and their impacts, and explores methods by which adverse environmental impacts of an action could be mitigated." Minn. Stat. § 116D.04, subd. 2a (2018) (emphasis added). As respondent Enbridge Energy Limited Partnership (Enbridge) argues,
The FEIS is not required to exhaustively review every conceivable permutation of the Project and every possible alternative and theoretical effect. Instead, the [Minnesota Public Utilities Commission] "shall" find the FEIS is adequate if it satisfies the standards for adequacy set forth in Minnesota Rules 4410.2800, subpart 4. The FEIS far exceeds the standards for adequacy.
This argument has merit.
"It is of course always possible to explore a subject more deeply and to discuss it more thoroughly. The line-drawing decisions necessitated by this fact of life are vested in the agencies, not the courts." Coal. on Sensible Transp., Inc. v. Dole , 826 F.2d 60, 66 (D.C. Cir. 1987). The FEIS was thoughtfully prepared by the Minnesota Department of Commerce's Energy Environmental Review and Analysis division (DOC-EERA) with assistance from other responsible agencies. It consists of 13 chapters, 21 appendices, and thousands of pages that address the potentially significant issues and alternatives raised in scoping. It was subjected to extensive public hearings and comments. It was thoroughly reviewed and determined adequate both by an experienced administrative-law judge (ALJ) and a unanimous commission.
The majority appears to opine that the FEIS is inadequate because it did not conduct a specific analysis of the Lake Superior watershed or include it as one of the seven modeling sites. The final scoping decision document (FSDD) committed the FEIS to
consider potential impacts to the Lake Superior watershed including potential impacts of oil spills along the proposed Project. Potential impacts to the Great Lakes from incidents involving transportation of crude oil by ship, rail, or other pipelines are existing potential effects and not changed by the construction or operation of the proposed pipeline.
*38See Minn. R. 4410.2800, subp. 4 (2017) ("The final EIS shall be determined adequate if it: (A) addresses the potentially significant issues and alternatives raised in scoping so that all significant issues for which information can be reasonably obtained have been analyzed in conformance with part 4410.2300, items G and H; [and] (B) provides responses to the substantive comments received during the draft EIS review concerning issues raised in scoping [.]" (emphasis added)). While I agree that the FEIS needed to address the concerns raised in scoping, which committed it to "consider potential impacts to Lake Superior watershed," I would nonetheless conclude that the representative-modeling approach sufficiently considered the effect of oil spills on the Lake Superior watershed.
I start my analysis by reviewing a case in which this court unanimously found that an administrative agency acted in an arbitrary and capricious manner. In Builders Ass'n of Twin Cities v. Minn. Dep't of Labor & Indus. , a challenge was raised to an amendment to the Minnesota Residential Code (MRC) promulgated by the Minnesota Department of Labor and Industry, which "required sprinkler systems in all newly constructed townhouses and one- and two-family dwellings, with an exception for one-family dwellings with a floor area under 4,500 square feet (Sprinkler Rule)." 872 N.W.2d 263, 266-67 (Minn. App. 2015). Under the arbitrary-and-capricious standard, we reviewed the record and queried the parties for an explanation of how the 4,500-square-foot threshold was reached. Id. at 268-70. We found that the record reflected "no reasoned determination of how respondent arrived at the indefinite 4,500-square-foot exception." Id. at 270 (emphasis added). We concluded that the Sprinkler Rule's exception for new one-family dwellings under 4,500 square feet was arbitrary because it was unsupported by the record. Id. at 274.
Distinguishable from Builders Ass'n of Twin Cities , the record here reflects a reasoned determination of why the representative-modeling approach was selected to consider the potential impacts to the Lake Superior watershed, as opposed to a site-specific approach. As the FEIS explains, the impact of any particular spill will depend on many variables, including "weather, time of year, water levels, human error, and even what type of wildlife is present." Considering these unpredictable factors, the FEIS does not present an encyclopedic report on the impact of every conceivable oil spill that could occur. Instead, it "provides a general assessment of the probability of a spill occurring, a general evaluation of the behavior of crude oil in the environment, a general evaluation of how spilled oil affects the environment, and an assessment of the type and quantity of resources that are exposed along each alternative." (Emphasis omitted.)
During oral argument, when asked how the FEIS was inadequate in addressing the Lake Superior watershed, relator Friends of the Headwaters (FOH) responded that there was no information projecting the quantity of oil that could reach Lake Superior or the specific impact on the natural resources. The fact remains that it is impossible to model every possible impact.
As the DOC-EERA explained:
The specific impacts of large oil releases are highly dependent on incident-specific factors that are impossible to predict with certainty. For example, as the commenter notes, the largest ever inland spill in Grand Rapids did not produce the damages a model may have suggested from a spill of its magnitude because incident-specific conditions happened *39to include ice cover on the Prairie River.
One purpose of the EIS is to provide decision makers with relevant information for their decision. Detailed modeling of site/situation specific environmental damages is so incident-specific, it does not provide decision makers with particularly actionable information about which alternative is environmentally preferable.
In fact, this sort of impact modeling is likely to provide a false level of precision that is counterproductive. The modeling approach used in the EIS is intended to provide information relevant to a reasoned choice among the alternatives by focusing on several components of spill risk that can be considered to develop a broader understanding of the risks and tradeoffs of different routes.
Forecasting impacts at any particular location would require the FEIS to predict each of these unpredictable variables, or to analyze all of the potential permutations. I would conclude, and the record supports, that the FEIS reasonably took another approach. See Minn. Stat. § 116D.04, subd. 2a(a) ("The environmental impact statement shall be an analytical rather than an encyclopedic document ...."); Minn. R. 4410.2300(H) (2017) (providing that the responsible government unit "shall consider the relationship between the cost of data and analyses and the relevance and importance of the information in determining the level of detail of information to be prepared for the EIS").
During the administrative proceedings, HTE and FOH both challenged the representative-modeling approach arguing that additional or different sites should have been modeled. The ALJ stated,
Enbridge commissioned a modeling analysis of hypothetical crude oil releases on behalf of, and with input from, DOC-EERA, DNR, and MPCA. The analysis modeled the impacts following seven different hypothetical crude oil releases. The computer modeling involved "simulating the chemical and physical behavior of hypothetical oil spills in the selected environments under specified conditions, including weathering processes."1
... The oil spill scenarios were set at different locations along the Applicant's Preferred Route and route alternatives. This study later informed both agency and independent analyses of the behavior of crude oil after a release and the assessment of likely impacts following a release. All of these items are detailed in the FEIS.
... Further the DOC-EERA commissioned a study from a private consulting firm, Ecology and Environment, Inc., to conduct an analysis of previous oil spills. In an effort to "quantify the incremental risk for the Line 3 Project," the report provided "an overview of pipeline spill rates and trends in the inland [United States] as a whole, as well as an analysis of historical data for existing crude oil pipelines in Minnesota."
... Drawing upon these materials, and other items, the FEIS analyzed the *40relationship between the volume of oil that would be transported by the project in relation to the risk of later spills and still broader "cumulative potential effects."
... Among the findings made in the FEIS were:
(a) The average volume of pipeline spills has decreased significantly since the late 1960s, and particularly in the last dozen years. The average spill volume (all oil types) is now less than 50 percent of the average volume 10 years ago, and 12 percent of the volume in the late 1960s.
(b) Overall, half of the pipeline spills that do occur would be expected to involve 1 barrel of oil or less. About 90 percent would involve 100 barrels or less. Only 5 percent would be expected to be 400 barrels or more, and only 1 percent would be expected to be 2,500 barrels or more.
(c) The rate of spillage in Minnesota has been lower than that in the U.S. as a whole, accounting for pipeline mileage and amount transmitted.
(d) DOC-EERA estimated that the volumes of spillage in the seven hypothetical Line 3 spill scenarios-ranging from 8,625 barrels to 16,239 barrels-might be expected once in 26 to 99 years somewhere in the state of Minnesota.
... [FOH] maintains that the FEIS is inadequate because the spill analyses did not include an assessment of a hypothetical discharge of oil into the headwaters of the Mississippi, or other [High Consequence Areas] in Itasca and Hubbard counties. It argues:
Before approving or rejecting this pipeline proposal, the public and the [commission] reasonably want to know what would happen if a Kalamazoo-type spill occurred near the Mississippi headwaters, in the wetlands and wild rice habitat north of Itasca State Park, in the central sands area with its vulnerable aquifers and already-compromised drinking water supplies, into the Straight River, a nationally recognized trout stream, and in other sensitive areas along the route.
... The Administrative Law Judge does not agree that the FEIS is inadequate without the specific modeling sought by [FOH]. The regulatory guidance in this area suggests that if the agency's assessments are broadly representative of the spill impacts likely to be encountered along the pipeline route those evaluations are adequate; even if interested persons would have preferred other, particular areas to have been studied by the government. Measured by this standard, the pyramiding analyses undertaken by the DOC-EERA and its consultants are adequate to inform the [c]ommission of the potential impacts from an accidental discharge of crude oil.
(Emphasis added.)
Specifically, the FEIS devotes significant attention to analyzing the potential resource impacts of a spill at locations along the Applicant's Preferred Route (APR) and alternatives. Section 10.3.4 explains:
The representative sites were selected to favor circumstances that would tend to exacerbate potential impacts and effects of the hypothetical spills along the Line 3 pipeline. In all likelihood, a spill of equivalent volume as the hypothetical scenarios that occurred at another location and time would have an outcome that would be of lesser consequences, or, at most similar consequences to one of the representative scenarios.
*41Section 10.4.2 of the FEIS discusses resources subject to exposure for the APR and alternatives. The FEIS states that, "For this analysis, ROIs [regions of interest] for potential releases were identified ... to reflect the potential extent of a large-volume incident that could occur at any point along each route. " (Emphasis added.) Thus, although the FEIS does not model the expected spread of a spill at all locations, it does assess the resources that potentially would be impacted by a spill along the entire APR and alternatives. This discussion encompasses the Lake Superior watershed because the Superior terminal is the endpoint for existing Line 3 and all but one of the alternatives.
Additionally, during the commission hearing, there was discussion about why Lake Superior was not modeled. When asked by a commissioner why the modeling did not include the Lake Superior watershed as one of the representative sites, a representative for DOC-EERA, which drafted the FEIS, explained that the interagency team chose those seven sites to get a "full range of different kinds of conditions." These seven sites were selected to encompass "things like stream morphology and eco region and flow" so that they could be "vignettes" for what a spill may look like at other locations including Lake Superior. DOC-EERA also explained that conducting a model in Duluth would not provide much information "in terms of informing a decision about what the commission can effect" or how the decision would "change the risk profile" because of the six pipelines that currently cross there. Although this explanation did not appear in the FEIS, it is in the record, which we must review in its entirety. See Minn. Stat. § 14.69(e) (requiring review of "the entire record as submitted"); see also Reserve Mining Co. , 256 N.W.2d at 824 (noting the function of an appellate court is "to make an independent examination of an administrative agency's record and decision"); Builders Ass'n of Twin Cities , 872 N.W.2d at 268 (stating that under the arbitrary and capricious standard, "appellate courts make a searching and careful inquiry of the record to ensure that the agency action has a rational basis" (quotation omitted)).
Furthermore, the majority even concluded that the commission's decision to approve the representative-model approach employed in the FEIS-opposed to a site-specific approach-was reasonable. Specifically, in section II.B., the majority concluded that it was reasonable for the FEIS to focus "on analyzing the potential resource impacts of a spill at all locations along the APR and alternatives ", "[r]ather than attempting to predict the consequences of an oil spill at a particular location" because "the impact of any particular spill will depend on multiple variables, many of which are subject to chance." (Emphasis added.) Nevertheless, in section II.A., the majority still concluded that the FEIS needed to "address[ ] the impact of an oil spill on the Lake Superior watershed " because the scoping document committed the EIS to "consider potential impacts to the Lake Superior watershed." (Emphasis added.) I am at a loss to understand this distinction.
But as the majority concedes, the modeling approach used in the FEIS considered the impact of a spill at all locations along the APR and alternatives, and the Lake Superior watershed is a location along the APR. And yet still the majority holds it was arbitrary and capricious for the commission to approve the FEIS because it does not specifically analyze issues raised in scoping and does not respond to comments received. But Minn. R. 4410.2800 does not state that issues must be addressed "specifically" and the FEIS did address the concerns about the Lake Superior watershed. The FEIS simply addressed *42it in a manner different from how HTE wanted. But as an appellate court, we cannot substitute our own judgment for the judgment of the DOC-EERA, DNR, MPCA, the experts tasked with drafting the FEIS, and finally the commission when that judgment is reasonable and supported by the record. This is not was our standard of review dictates.
I also have concerns that our decision today will have the further unintended consequence of delaying the replacement of Line 3, which could pose a serious threat to our environment including Lake Superior. Existing Line 3 has been in operation since the 1960s, has suffered a high amount of corrosion and long-seam cracking, and must be replaced under a consent decree between Enbridge and the Environmental Protection Agency and Coast Guard.
In conclusion, by according substantial deference to the commission's decision as required by precedent, I would affirm the commission's decision that the FEIS is adequate because the representative-modeling approach sufficiently considers "potential impacts to the Lake Superior watershed" as required by the FSDD and the record reflects a reasoned determination of why that approach was used. Therefore, the commission's decision is not arbitrary or capricious and is supported by the record. Consequently, I must dissent.
This specific finding was modified by the commission. The modified finding states:
Enbridge commissioned a modeling analysis of hypothetical crude oil releases on behalf of, and with input from state and federal agency staff, including DOC-EERA, Minnesota Department of Health, and the DNR and MPCA. Staff from the U.S. Army Corps of Engineers were also involved. The analysis modeled the impacts following seven different hypothetical crude oil releases. The computer modeling involved "simulating the chemical and physical behavior of hypothetical oil spills in the selected environments under specified conditions, including weathering processes."