Mid States Coalition for Progress v. Surface Transportation Board

Petitioners challenge the decision of the Surface Transportation Board issued January 30, 2002, giving final approval to the Dakota, Minnesota & Eastern Railroad Corporation’s (DM & E) proposal to construct approximately 280 miles of new rail line to reach the coal mines of Wyoming’s Powder River Basin (PRB) and to upgrade nearly 600 miles of existing rail line in Minnesota and South Dakota. They maintain that in giving its approval the Board violated 49 U.S.C. § 10901, the National Environmental Policy Act (NEPA) (42 U.S.C. §§ 4321-4347), the National Historic Preservation Act (16 U.S.C. §§ 470 to 470w-6), and the Fort Laramie Treaty of 1868. Although we conclude that the Board should prevail on almost all of the issues raised by the petitioners, our rulings on a few issues require us to vacate the Board’s decision and to remand for further proceedings not inconsistent with this opinion.

*533I.

Under 49 U.S.C. § 10901, the Board has exclusive licensing authority for the construction and operation of rail lines. This statute provides that the Board shall authorize the construction and operation of a proposed new line “unless the Board finds that such activities are inconsistent with the public convenience and necessity.” Although the Board’s authorizing statute does not define the term “public convenience and necessity,” in reaching its decisions the Board has historically asked whether there is a public demand or need for the proposed service, whether the applicant is financially able to undertake the construction and provide service, and whether the proposal is in the public interest and would not unduly harm existing services. If the Board is satisfied that the proposed project is not inconsistent with the public convenience and necessity, it proceeds to conduct an environmental review as required by NEPA. Once the environmental review is completed, the Board determines whether its original conclusion is still warranted after taking into account the potential environmental effects of the project and the cost of any necessary environmental mitigation.

In this case, the Board made an initial determination that DM & E’s proposal was merited under § 10901. The Board found that there was public demand for the line because it would offer a shorter and less expensive method by which to transport coal from the PRB mines to power plants. It also concluded that the proposed project would benefit existing shippers and that DM & E had demonstrated its financial fitness to carry the project through to completion. Having preliminarily found that the project would not be inconsistent with the public convenience and necessity, the Board instructed its Section of Environmental Analysis (SEA) to examine the potential environmental effects resulting from the construction and continuing operation of the proposed project.

SEA, in coordination with five cooperating federal agencies, then produced a nearly 5,000-page draft environmental impact statement (DEIS) examining the effects both of constructing the rail line extension to the PRB mines and rehabilitating DM & E’s existing lines in Minnesota and South Dakota to accommodate the coal traffic anticipated as a result of the project. SEA initially allowed 90 days for public review of and comment on the DEIS, but later extended this period by 60 days to ensure that the large number of persons and entities who wished to comment had ample opportunity to do so. The environmental review culminated with the issuance of a final environmental impact statement (FEIS), which contained further analysis in response to the comments received on the DEIS. The FEIS also made recommendations to the Board regarding environmentally preferable routing alternatives and mitigation measures. In all, the environmental review process took nearly four years and generated roughly 8,600 public comments.

II.

The NEPA mandates that a federal agency “take a ‘hard look’ at the environmental consequences” of a major federal action before taking that action. Baltimore Gas & Elec. Co. v. Natural Res. Def. Council, Inc., 462 U.S. 87, 97, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983) (quoted case omitted). To comport with this standard, an agency must prepare a “detailed statement” (generally, an EIS), 42 U.S.C. § 4332(2)(C), “from which a court can determine whether the agency has made a good faith effort to consider the values NEPA seeks to protect.” Minnesota Pub. Interest Research Group v. Butz, 541 F.2d *5341292, 1299 (8th Cir.1976), cert. denied, 430 U.S. 922, 97 S.Ct. 1340, 51 L.Ed.2d 601 (1977).

In reviewing the agency’s decision, we are not free to substitute our judgment for that of the agency. Vermont Yankee Nuclear Power Corp. v. Natural Res. Def. Council, Inc., 435 U.S. 519, 555, 98 S.Ct. 1197, 55 L.Ed.2d 460 (1978); Friends of the Boundary Waters Wilderness v. Dombeck, 164 F.3d 1115, 1128 (8th Cir.1999). Our role in the NEPA process “is simply to ensure that the agency has adequately considered and disclosed the environmental impact of its actions and that its decision is not arbitrary or capricious.” Baltimore Gas, 462 U.S. at 97-98, 103 S.Ct. 2246; see 5 U.S.C. § 706.

A.

We begin our review of the agency’s actions under NEPA by addressing the objections raised by the city of Rochester, the Mayo Foundation, and Olmstead County, since their objections are to a large degree overlapping. DM & E’s existing line, carrying an average of three trains per day, runs through Rochester and Olm-stead County. In the FEIS, SEA recommended reconstruction of the existing line as the environmentally preferable alternative for the Rochester area, and the Board in its final decision accepted this recommendation. SEA rejected the proposed construction of a bypass around Rochester and a no-action alternative. Because Rochester is the largest community located on DM & E’s projected route and Mayo is one of the most sophisticated medical centers in the nation, they could be expected to be particularly alert to any environmental degradation that might arise from the project that the Board approved. We therefore address their claims with some degree of specificity.

1. Reconstruction of the rail line through Rochester would result in increased rail traffic through the city. Whereas at the present time approximately three trains pass through the city daily, SEA estimated that upon completion of the project rail traffic could increase to as many as 37 trains per day: This increase in traffic would, among other things, increase noise in the city to what SEA calls “adverse” levels. (SEA considers average noise levels above 65 decibels to be adverse and noise levels above 70 decibels to be significantly adverse.) SEA calculated that the average wayside (engine and wheel) noise level produced from 37 trains per day would be at least 65 decibels at distances within 420 feet of the fine and would be at least 70 decibels at distances within 210 feet of the line. With noise produced from the trains’ horns included, SEA calculated that the average noise level would be at least 65 decibels at distances within 2220 feet of the line and would be at least 70 decibels at distances within 1110 feet of the line. Finally, SEA determined, using aerial photographs, the number of noise-sensitive “receptors” (e.g., homes, schools, hospitals, churches) that would fall within the distances described above.

Based upon SEA’s recommendation, the Board’s final decision required DM & E to mitigate noise for those noise-sensitive receptors that would experience an average noise level of 70 decibels from wayside noise. Specifically, the mitigation required a minimum average noise reduction of 5 decibels to the affected receptors and stated a design goal of achieving an average noise reduction of 10 decibels. The Board did not require mitigation for average wayside noise levels below 70 decibels or for noise caused by train horns.

In its comments on the DEIS, Rochester produced data that purportedly *535showed that 88 residences would experience average wayside noise levels between 75 and 80 decibels and that 8 residences would experience average wayside noise levels above 80 decibels. (Noise levels double every 10 decibels; 80 decibels, therefore, is twice as loud as 70 decibels.) It asserts that SEA misled the Board (in violation of NEPA’s requirement of full disclosure) by aggregating all of these residences into a group described as having an average noise level of at least 70 decibels. If SEA had advised the Board that some residences could experience average noise levels greatly exceeding 70 decibels, Rochester argues, the Board might have determined that additional mitigation was necessary.

There is no meaningful dispute that SEA accurately identified the distances at which receptors would experience an average noise level of at least 70 decibels. The question, rather, is whether SEA was derelict in failing to calculate the extent to which the average noise levels would exceed 70 decibels. Rochester maintains that the Board’s own regulation, which instructs the agency to “quantify the noise increase” for receptors which will experience an average noise threshold of 65 decibels, 49 C.F.R. § 1105.7(e)(6)(h), requires an affirmative answer to this question. Although Rochester’s argument that SEA has not quantified the noise increase is facially attractive, we think it ultimately unavailing. For one thing, Rochester’s own proposal (that SEA determine the number of receptors affected at the 75 and 80 decibel levels) would not actually “quantify” the noise increase associated with the project; it would still yield only an aggregation, albeit at a higher level, of the noise increase in Rochester. One could, of course, understand the instruction to “quantify the noise increase” to require a measurement of the actual noise level experienced by every receptor that would meet the threshold noise level. But we do not think that the Board’s regulations, or NEPA, require that level of precision. Indeed, we doubt that such a determination would be feasible given that this project involves some 880 miles of railroad line.

Having concluded that SEA is not required to measure actual noise levels for potentially affected noise receptors, we consider whether it nonetheless violated NEPA to aggregate all average noise levels above 70 decibels into one category that SEA described as significantly adverse. In considering whether the EIS “adequately sets forth sufficient information to allow the decision-maker to ... make a reasoned decision,” we are guided by the “rule of reason.” See Boundary Waters, 164 F.3d at 1128 (internal quotations omitted). Despite Rochester’s assertions, it does not appear to us that SEA hid the facts from the Board. The DEIS fully disclosed the number of noise receptors that would suffer significantly adverse effects from both wayside and horn noise. It may be true, as Rochester contends, that even after mitigation there will still be residences that are subject to significantly adverse noise levels (indeed, the Board does not claim otherwise). But NEPA does not require that an agency eliminate all adverse affects that might result from a project.

Rochester draws attention to the fact that the Federal Aviation Administration (FAA) gathers data for average noise levels above 65 decibels, 70 decibels, and 75 decibels, and argues that SEA’s analysis is therefore deficient. While it is true that the FAA includes a level of analysis that SEA does not, we do not think that this is dispositive. It would be wrong to conclude that one agency’s selection of a particular methodology necessarily makes another agency’s chosen, but different, methodolo*536gy insufficient under NEPA. On the whole, we find that SEA’s choice of analysis, which was consistent with its past practice and similar in nature to the noise analyses performed by other federal agencies, was not arbitrary or capricious.

2. Rochester submits that the Board violated NEPA by failing to consider and mitigate horn noise. As described previously, SEA calculated the number of noise receptors that would experience average noise levels from train horns of at least 65 decibels and 70 decibels. But unlike the treatment given to wayside noise, SEA’s discussion of the effects and mitigation possibilities for horn noise was relatively perfunctory. The only mention of mitigating horn noise in the FEIS occurred in a footnote explaining that “SEA is not recommending mitigation for horn noise because of potential safety concerns in the absence of Federal Railroad Administration [FRA] standards addressing this issue.” In its argument to this court, the Board further explained that the FRA has recently proposed standards for establishing quiet zones (areas where horns do not have to be sounded), and that absent FRA approval it would be inappropriate for the Board to impose its own limitations on horn soundings. Given the important role that train horns play in reducing traffic accidents, we cannot second-guess the decision of SEA in refusing to limit the use of train horns. We do not believe, however, that this relieves SEA of the obligation to consider mitigation not involving limitations on the use of horns.

SEA required mitigation for receptors subjected to an average noise level of 70 decibels from wayside noise. Rochester maintains that SEA should have considered similar mitigation measures for receptors subjected to comparable levels of horn noise. Such measures might include sound-insulating treatments for buildings within high noise areas. By SEA’s own calculations, horn noise will increase the distance at which buildings may be subjected to average noise levels of 70 decibels from 210 feet (the distance due to wayside noise alone) to 1110 feet. Although it is hard to imagine how insulating affected buildings might pose a safety threat (horns are sounded primarily for traffic safety), it is just conceivable that safety reasons do exist. But without a reasoned discussion of its rationale, we cannot say that SEA has taken a “hard look” at this substantial issue.1 This is not to say that the Board must ultimately mitigate for horn noise, but it must at least explain why mitigation is unwarranted. Even though NEPA’s requirements are predominantly procedural, they do require that SEA “explain fully its course of inquiry, analysis and reasoning.” Minn. Pub. Interest, 541 F.2d at 1299. We conclude that it did not do so here.

3. Rochester’s next objection is to the method used by SEA in calculating ambient (background) noise for use in its noise analysis. SEA used noise levels in rural South Dakota as its baseline for ambient noise. Rochester argues that since the ambient noise levels in an urban area are higher, it was arbitrary for SEA to use the lower rural levels. We disagree. SEA adequately supported its analysis by explaining that noise is not additive; when *537two sounds are of different levels, the higher level predominates and the lower level adds little to the overall noise level. This conclusion is supported by the EPA, which has stated that “adding a 60 decibel sound to a 70 decibel sound only increases the total sound pressure level less than one-half decibel.” See Protective Noise Levels, Condensed Version of EPA Levels Document 3 (1979), at http//www.no-noise.org/hbrary/levels/levels.htm. Even if we credit Rochester’s estimate that its own ambient noise level is 59 decibels, that would add less than one-half a decibel to those receptors that SEA has determined will experience average train noise of 70 decibels. SEA’s decision to forego a separate ambient noise measurement for every community located along the DM & E project was clearly within its permissible discretion.

4. Likewise, we find that SEA did not act arbitrarily in responding to concerns about nighttime noise. Because nighttime noise can lead to sleep disturbance, its effect on the human environment is greater than a similar level of daytime noise. To account for this, SEA employed the accepted practice of counting each nighttime train as ten trains (adding an approximately 10 decibel penalty to each train). For the purpose of its analysis it also assumed that train traffic would be spaced evenly throughout the day, an assumption that Rochester contends was a clear error in judgment. As the basis for this contention, Rochester presented in its comments a statistical model that purportedly shows that more trains will run at night than SEA’s model predicts. Rochester’s model is based upon DM & E’s plan to schedule a block of up to six hours for maintenance (a period where no trains will run) each day. In response to this comment, SEA stated that maintenance was impossible to predict and would vary considerably depending on what particular coal contracts that DM & E obtained. SEA therefore chose not to alter its methodology. Rochester’s model may indeed be a better predictor of night traffic if DM & E actually uses a full six hours per day for maintenance and if DM & E is equally likely to schedule maintenance during the daytime as it is to schedule it at night, but these assumptions are just as speculative as SEA’s assumption that train traffic would be spaced evenly throughout the day. Due to the highly uncertain nature of rail traffic patterns, we cannot say that it was a clear error of judgment for SEA to prefer one set of assumptions over another in conducting its analysis.

5. Rochester argues that SEA failed to make any response to evidence presented in Rochester’s comments that households experiencing both noise and vibration perceive the effect of the noise to be approximately twice the measured value of the noise. Although SEA included analysis for noise and vibration effects separately, we can find no evidence that it considered the synergies between the two in its response to comments or in the environmental impact statements. “Although the agency is not required to include in its final analysis every factor raised by ... a comment and may respond, for example, by explaining why the comment does not warrant [further] agency response,” see Oregon Natural Res. Council v. Marsh, 52 F.3d 1485, 1490 (9th Cir.1995), the Council on Environmental Quality (CEQ) regulations impose upon an agency preparing an FEIS the duty to assess, consider, and respond to all comments, see 40 C.F.R. § 1503.4(a). In this instance, SEA has not met this minimum requirement. On remand, SEA is instructed to fulfill its duty under the applicable NEPA regulations.

*5386. We now consider whether SEA’s use of aerial photographs to identify noise-sensitive receptors led it to under-count the number of receptors eligible for noise mitigation. According to Rochester’s comments, the method employed was flawed because any single building identified by photograph might contain multiple residences (the Board’s regulations designate each “residence” as a separate noise-sensitive receptor). Rochester suggested that a more accurate count could be obtained by using the tax records to determine the number of affected residences. SEA responded to this comment by stating its belief that the number of potentially affected noise receptors was likely overestimated because there were no adjustments for ambient noise or for shielding (by an object between a noise source and a noise receptor) and because aerial photographs do not differentiate between eligible noise receptors (such as homes) and ineligible structures (such as businesses and garages). SEA also explained that any discrepancy between its calculations and the actual number of affected receptors can be corrected by the Board during its oversight period. We cannot say, as Rochester argues, that SEA’s choice of methodology amounted to a clear error in judgment. In a project of this size, the agency is not required to maximize precision at all costs. We view SEA’s decision to use aerial photographs as a sensible way reasonably to approximate the number of affected receptors along the entirety of the proposed project.

7. Rochester submits that SEA improperly failed to consider the environmental impact on the city of passing sidings (locations where westbound trains move onto alternate track so that eastbound trains can pass). Although there are no present plans to build a passing siding in Rochester, the city asserts that since the FEIS acknowledges that “siding locations have not been finalized” SEA should have assessed the environmental impact that would result from locating a siding in Rochester. We believe that Rochester has misconstrued SEA’s statement. DM & E has proposed 45 locations for possible sidings, but it anticipates needing only 35^0 total sidings for efficient rail operation; thus there are several proposed locations that ultimately will not be used. It is not inconsistent for SEA to acknowledge that the locations have not been finalized while at the same time denying that a siding will be built in Rochester (because none has been proposed there). We note, moreover, that SEA’s analysis with respect to the proposed sidings was more than adequate.

8. Rochester maintains that SEA committed two errors in its assessment of the traffic effects that will result from reconstruction of the existing line through Rochester. First, Rochester argues that SEA should have used more current data when determining average daily traffic (ADT) volumes for those streets where train crossings exist. According to Rochester, SEA used data from 1994 when data from 1998 was available.

Our comparison of the data from those years reveals that there has been little change in traffic volumes. In fact, the aggregate volume of traffic on the twelve streets where train crossings exist has actually declined (albeit only slightly) based on data from the Minnesota Department of Transportation. Even if we assume that SEA “erred” in using the older data, we need not remand unless “there is a significant chance that but for the errors the agency might have reached a different result.” Boundary Waters, 164 F.3d at 1129. Given the inconsequential difference in the data, we find remand unnecessary on this issue.

*539Rochester also asserts that SEA used a nonsensical formula in calculating the average delay to vehicles that would result from increased rail traffic. According to Rochester, any formula designed to compute the average traffic delay to all vehicles must include as one of its variables ADT volumes. In lieu of using ADT volumes, however, SEA’s formula calculates average traffic delays for all vehicles as a proportion of the delay for vehicles actually stopped. SEA explains that this method “results in a conservative estimate of vehicle delay.” Indeed, our careful inspection of both the SEA methodology and that proposed by Rochester suggests that SEA’s calculations likely overstate the average traffic delays. We are convinced, in any ease, that SEA’s chosen methodology did not undermine the purposes of NEPA.

9. Contrary to Rochester’s assertions, we believe that SEA’s analysis of ground vibration was adequate in all respects. In discussing the effects of vibration generally, SEA’s analysis determined that there was little risk of damage to structures located 50 feet or more from the tracks and also that there was little risk of disturbance to structures located 100 feet or more from the tracks. SEA also noted that residences within 100 feet (SEA counted 14 such structures) might experience increased disturbances as rail traffic increased. In addition to its general analysis, SEA consulted with the manufacturer of a security fence at a nearby prison to verify its own conclusion that increased rail traffic would not be incompatible with operation of the security fence; the manufacturer assured SEA that if properly maintained and operated, the fence would not be affected by increased rail traffic. Lastly, SEA discussed the possibility that PEMSTAR, a local company that uses vibration-sensitive equipment, might not be able to continue operations at its current facility were traffic levels to increase. SEA noted that if PEMSTAR left Rochester altogether (a proposition that SEA considered unlikely since PEMS-TAR has other facilities in Rochester that are located farther from the tracks), it would result in a loss of about 600 jobs. Given Rochester’s size and the fact that some jobs would be created by DM & E’s expanded operation, SEA concluded that the loss of 600 jobs would not have significant economic effects on the city.

Having thoroughly reviewed SEA’s analysis of vibration, we cannot agree with Rochester’s contention that SEA “buried” the facts. It seems to us that Rochester’s real grievance is that SEA did not adopt Rochester’s proposed mitigation condition that would have prohibited any increase in vibration. This proposal, it seems to us, would have essentially sounded a death knell to any plans to reconstruct the existing Rochester fine. This result would have no doubt been met with Rochester’s approval, but it was not compelled by the substantial body of evidence that SEA amassed on this issue.

10. Mayo asserts that SEA failed adequately to address the possibility of groundwater contamination. SEA acknowledged that both the existing route and the proposed bypass cross areas that are susceptible to groundwater contamination in the event of a rail line accident. After detailing these risks, SEA noted that because rehabilitation of the existing line would improve track that is currently in poor condition, the risks of groundwater contamination would actually decrease. Mayo’s counter-argument is based entirely upon a verified statement of one of its experts that was presented to SEA after SEA had prepared and released the FEIS. SEA does not have an obligation to respond to arguments that were not presented to the agency during the appropriate *540time period, especially when, as here, there is no indication that the information presented was previously unavailable. But even if it were appropriate to consider Mayo’s evidence, we would be unable to say that SEA has failed to take a “hard look” at the possibility of groundwater contamination.

11. Mayo also maintains that SEA did not take a “hard look” at the risk that the project would cause delays to emergency vehicles. Although Mayo acknowledges that SEA analyzed independently the effects of the reconstruction alternative and the bypass alternative on emergency vehicles, it argues that SEA arbitrarily avoided a direct comparison of the alternatives. Essentially, Mayo faults SEA for failing to say explicitly that the existing route would cause more delays to emergency vehicles than the bypass. We think, however, that SEA made this point abundantly clear when it recommended that the Board, if it chose to utilize the existing route, should “require construction of two additional grade separated crossings in Rochester to prevent potential reductions in the quality of emergency response.” FEIS 9-65. In contrast, SEA found that mitigation related to emergency vehicles would be unnecessary for the bypass. The only logical inference that can be drawn from this is that SEA anticipated that reconstruction of the existing route would pose more risk of disruption to emergency vehicles than would the construction of the bypass. For the purpose of complying with NEPA, it was not incumbent upon SEA to state this conclusion in a single explicit sentence.

12. Mayo contends that SEA failed properly to examine the relationship between increased levels of train vibration and the formation of sinkholes. Despite the high level of concern it expresses now, however, Mayo did not raise this issue in its comments on the DEIS. (Mayo did comment on its concerns about the effect of vibration on its facilities, and SEA responded to this comment by undertaking a more extensive vibration analysis.) Mayo seeks to excuse its failure to raise this issue earlier by stating that the issue was obvious (although apparently not obvious enough for Mayo to have raised it before). But even if it is true that increased vibration will hasten the formation of sinkholes, we fail to see how this advances Mayo’s interests. SEA recommended the existing route because it was less susceptible to sinkholes than the bypass alternative. This relative advantage of the existing route would seemingly be magnified if SEA were to find that vibration accelerated sinkhole formation in susceptible areas. In any event, we do not believe that SEA’s failure to respond to a concern that was never raised tainted its analysis.

13. Olmstead County raises alleged deficiencies with respect to SEA’s air quality analysis. First, it asserts that SEA should have used Minnesota’s thresholds for determining whether sulfur dioxide levels were significant instead of the less stringent EPA thresholds. The decision to apply these lesser standards, Olm-stead County argues, led SEA erroneously to conclude that it did not have to undertake more precise modeling to determine the exact scope of the effect of the proposed line on air quality. The County also faults SEA for failing to take background levels of hazardous air pollutants into account when it determined that hazardous air pollutant concentrations caused by train locomotives would be insignificant.

After reviewing the record, we conclude that SEA’s decision to use the EPA’s thresholds was not arbitrary. SEA has an interest in using a standardized measurement to compare and contrast the relative air quality effects across a variety of re*541gions. The EPA thresholds provide a reasonable standard by which to accomplish this. SEA’s decision to forego the testing necessary to determine the background levels of hazardous air pollutants in Olm-stead County was similarly reasonable. Its measurements of hazardous air pollutant concentrations from locomotive exhaust showed that increased rail traffic would result in only a minuscule increase in overall concentration levels. NEPA regulations require agencies to expend the bulk of their efforts on the most pressing environmental issues. In this instance, SEA had evidence that showed that the increase in the concentrations of hazardous air pollutants would be de minimis in comparison to the background levels, whatever they might actually be. Further expenditure of agency resources was therefore not required.

14. Olmstead County maintains that SEA’s environmental justice analysis was inadequate. The purpose of an environmental justice analysis is to determine whether a project will have a disproportionately adverse effect on minority and low income populations. To accomplish this, an agency must compare the demographics of an affected population with demographics of a more general character (for instance, those of an entire state). On the EPA’s recommendation, SEA used 1990 census data (2000 census data were not yet available) to compare data at the census block group level (the smallest geographic unit for which data on both race and income are obtained) to data at the state level.

Olmstead County raises two objections to this approach. First, it argues that SEA should have used projected 2000 census data, which were available for some, but not all, communities. Second, it argues that for some areas, data were available at a level finer than that of census block group (for example, discrete neighborhoods, subdivisions, etc.). In Olmstead County’s opinion, these “corrections” would allow SEA to identify more affected groups. In response to these comments, SEA explained that an environmental justice analysis must use consistent data sets in order for the comparison to be meaningful. SEA, after close consultation with the EPA, used the most current and consistent data that were available to it. It seems to us that it is Olmstead County’s suggested approach (using a medley of assorted data), and not SEA’s, that could more fairly be characterized as arbitrary.

15. We consider next whether the Board’s decision-making process was flawed by the unlawful consideration of ex parte communications from DM & E. The record demonstrates that DM & E officials submitted a letter to SEA after the FEIS was issued (but before the Board’s final decision). In this letter, the DM & E expressed its views on mitigation proposals in the FEIS that called for three grade-separated crossings. Mayo argues that this contact violated the Administrative Procedure Act, see 5 U.S.C. § 557(d)(1)(A), and the Board’s own code of ethics, see 49 C.F.R § 1103.14.

We are not sure that these prohibitions apply to communications, such as DM & E’s letter, that are submitted to the authority in charge of an environmental review and that express comments about that review. We note, moreover, that three United States Senators wrote letters to the Board on behalf of Mayo during the same time period that the alleged improper communications of DM & E occurred, communications to which Mayo does not object. In any event, as we indicate below, the Board did not adopt the view expressed in the letter that Rochester should be required to pay for the proposed *542grade-separated crossings, so we discern no remediable harm here.

Mayo also objects to a discussion between DM & E’s president and the Board’s chairman regarding the jurisdiction of the Board to impose bypass alternatives. As this discussion concerned matters of jurisdiction, neither of the cited authorities would seemingly apply. But, again, the communication was of little consequence since the Board expressly rejected DM & E’s understanding of its jurisdictional authority.

16. We turn our attention to the Board’s rejection of a proposed bypass around Rochester, a decision that Rochester argues vigorously is arbitrary and capricious. According to SEA, the fundamental flaw in the proposed bypass was that it would require the construction of new track through karst areas that are topographically susceptible to sinkhole collapse. Construction in high sinkhole areas requires expensive mitigation to reduce the risk that heavy construction equipment will cause the collapse of underground caverns. Even with expensive mitigation and monitoring, there is the potential that sinkholes could develop at some point in the future, resulting in the derailment of trains, which could, in turn, lead to groundwater contamination. In addition, the necessary mitigation, which could require construction of a cement dam wall underneath the rail line, might itself result in potentially significant alterations in groundwater flow, thereby affecting the region’s ecology and accelerating the formation of other sinkholes.

Rochester does not dispute the fact that construction over karst terrain presents increased risks and costs. Its argument, instead, is that SEA’s treatment of the Rochester bypass was inconsistent with its treatment of other areas where sinkholes were a potential difficulty, primarily the proposed East Staging and Marshalling Yard in Lewiston, Minnesota (Lewiston Yard). SEA chose to approve the construction of Lewiston Yard despite the fact that it was located in an area having the potential for a high concentration of sinkholes. In its analysis, however, SEA points out significant differences between the situations presented by Lewiston Yard and the proposed bypass of Rochester. Lewiston Yard requires only 2.1 miles of construction, and although the site initially proposed was in an area having the highest probability for sinkholes, there was some evidence that shifting the site slightly to the west would avoid the most troublesome topographical features. In contrast, the proposed Rochester bypass, which is 34.1 miles long, would involve 1.4 miles of construction through an area having the highest probability of sinkholes, 6.3 miles , of construction through an area having a moderate to high probability of sinkholes, and 19.4 miles of construction through an area having a low to moderate probability of sinkholes. And unlike Lewi-ston Yard, there is no indication that these areas could be avoided (Rochester did not make such a case to SEA during the comment period).

The Lewiston Yard and Rochester bypass proposals are dissimilar in another important, and we think conclusive, respect. SEA concluded that there was an environmentally and fiscally preferable alternative to the Rochester bypass, namely, the reconstruction of the existing route, and that there was no such alternative in the case of Lewiston Yard. Rochester’s arguments are misplaced in that they focus on demonstrating that the bypass could be built in spite of the existing terrain. But SEA does not contend in its analysis that the bypass could not in fact be built, only that it would entail considerable cost and significant environmental risk to do so. *543While Rochester may prefer those displaced environmental consequences associated with the bypass to the ones associated with reconstruction of the existing line, it is not our place to reallocate those burdens. When the “resolution of [the] dispute involves primarily issues of fact” and “analysis of the relevant documents ‘requires a high level of technical expertise,’ we must defer to ‘the informed discretion of the responsible federal agencies.’ ” Marsh v. Oregon Natural Res. Council, 490 U.S. 360, 377, 109 S.Ct. 1851, 104 L.Ed.2d 377 (1989) (quoting Kleppe v. Sierra Club, 427 U.S. 390, 412, 96 S.Ct. 2718, 49 L.Ed.2d 576 (1976)).

Mayo argues that SEA had an obligation under NEPA to analyze the feasibility of other bypass alternatives to or variations on the rejected bypass alternative. But we note that Rochester had considered five alternatives to DM & E’s proposed reconstruction through the city, and it determined that the 34.1 mile bypass that it submitted was the environmentally preferred alternative. Guided by the “rule of reason” approach, Boundary Waters, 164 F.3d at 1128, we do not think that SEA was under an obligation to examine alternatives that Rochester itself considered environmentally inferior to the alternative ultimately rejected. Nor do we think that SEA has an obligation thoroughly to study new alternatives that were proposed only after it became apparent that Rochester’s preferred bypass alternative would be rejected. “Common sense ... teaches us that the ‘detailed statement of alternatives’ ” required by 42 U.S.C. § 4332(c)(iii) “cannot be found wanting simply because the agency failed to include every alternative device and thought conceivable by the mind of man.” Vermont Yankee, 435 U.S. at 551, 98 S.Ct. 1197.

Mayo and Olmstead County argue that SEA’s rejection of the bypass because of the difficulties with karst topography is arbitrary in light of the fact that the current rail line runs through similar topographical areas. SEA explained, however, that the risk of encountering sinkholes along the existing route was unlikely since surveys had identified only four sinkholes near or within DM & E’s right of way and since the existing route had been in operation for over a century without incident. We think that this provides a reasoned basis upon which to conclude .that the existing route presented fewer topographical challenges and risks than the proposed bypass.

Finally, Mayo maintains that SEA’s conclusion that the bypass would be significantly more expensive to construct and operate when compared with reconstruction of the existing route is unsupported by the evidence. This contention is undermined by the fact that Rochester (Mayo’s partner in interest) calculated that the bypass would cost approximately $37 million more than reconstruction of the existing route. DM & E and SEA, for their part, estimated that the difference in cost could be as much as $90 million. There was thus more than ample evidence to support SEA’s conclusion that construction of a bypass would be considerably more expensive than reconstruction of the existing route.

17. After the period designated for comments on the DEIS had passed, Mayo petitioned the Board to reopen the record to consider concerns caused by a train derailment involving the release of toxic materials in Maryland and the terrorist attacks that took place on September 11, 2001. Mayo asserted that if similar incidents occurred in Rochester, it would be difficult to evacuate its medical facilities immediately. In denying Mayo’s request to reopen the record, the Board explained that the proposed project would actually *544increase safety because it entailed system-wide improvements to existing track. The Board also noted that it was unlikely that DM & E would be involved in the increased shipment of hazardous materials. Finally, the Board did not view the two incidents as posing a threat specific to Mayo.

An agency is required to prepare supplements to an FEIS if “[t]here are significant new circumstances or information relevant to environmental concerns and bearing on the proposed action or its impacts.” 40 C.F.R. § 1502.9(c). This provision, however, has limits, for “an agency need not supplement an EIS every time new information comes to light after the EIS is finalized.” Marsh, 490 U.S. at 373, 109 S.Ct. 1851. “To require otherwise would render agency decisionmaking intractable, always awaiting updated information only to find the new information outdated by the time a decision is made.” Id. We therefore review denials of such requests applying the “rule of reason,”