MORE THAN JUST PARKS
PART III · WHAT IT WOULD COST CHAPTER 3 Counted
We counted every crossing on the road's own engineering centerline and got 2,932. The federal record agrees on the total, changed twenty of them from pipes to bridges, and plans in its own design documents for culverts blocked by ice.
What this chapter finds
The number that circulates is "about 3,000 crossings." It's close, and it's a shorthand, and nobody who repeats it can say what's in it. So we didn't repeat it. AIDEA, the state public corporation that holds the permits, publishes the engineering geometry for the chosen route as a public map service. We paged every feature in its crossing layer, all of it, and counted. An earlier version of our own map carried 2,000, because the service returns a fixed page size and reports nothing about what it left out. That's the kind of error that survives for years when nobody counts twice.
The total we got is the total the record of decision gets: 2,864 minor culverts, 7 moderate, 12 major, and 49 bridges in the government's arithmetic. Ours is the same 2,932. Where the two part company is in what they call things.
| Structure, as the applicant's file classifies it | Count |
|---|---|
| Minor culvert | 2,869 |
| Moderate culvert | 15 |
| Major culvert | 19 |
| Small bridge | 3 |
| Medium bridge | 15 |
| Large bridge | 11 |
| Total | 2,932 |
AIDEA, Ambler Access Project Alternative A engineering GIS, crossing layer, field "cls", every feature paged, pulled July 2026. 2024 Record of Decision, p. 12, component table for Alternative A.
The record of decision counts 49 bridges. The applicant's engineering file, the one the road would actually be built from, draws 29, and puts the other twenty down as major or moderate culverts. The totals reconcile. The breakdowns don't.
It matters because a bridge spans a stream and a culvert puts it in a pipe. A pipe changes the water's speed, its depth, its temperature and what can swim up it. The federal record's own language on fish passage is written with culverts in mind:
"The road would impact fish habitat and alter free fish passage based on likely changes to channels, flows, and sedimentation, and due to impacts caused by culverts, bridge piers, alteration of surface and subsurface flow patterns." 2024 Record of Decision, p. 14, repeated at p. 23
The 2020 Joint Record of Decision explains the gap. Among the design changes AIDEA adopted in the Army Corps' permit action, "the number of medium and large culverts was reduced by 20 and replaced by single span bridges." The applicant's engineering file still draws the earlier design. Which twenty crossings changed, the record doesn't list, and that's a fair question for the Corps.
Discrepancy documented in our sources sheet on the map. ROD p. 12 against AIDEA GIS field "cls". The map lets you color every crossing by structure and show only the 29 the applicant calls bridges.
Eleven named rivers, twelve crossings, and the applicant's file bridges all but one of them. This is the part of the road most people picture when they hear "three thousand crossings," and it's twelve of them.
| River | Structure | Land |
|---|---|---|
| Koyukuk River, and its side channel | Large bridge, small bridge | ANCSA corporation land |
| Wild River | Large bridge | ANCSA corporation land |
| John River, and a side channel | Large bridge, minor culvert | State of Alaska |
| Alatna River | Large bridge | State of Alaska |
| Malamute Fork of the Alatna | Large bridge | State of Alaska |
| Reed River | Large bridge | State of Alaska |
| Mauneluk River | Large bridge | State of Alaska |
| Kobuk River | Large bridge | National Park Service, Gates of the Arctic |
| Shungnak River | Large bridge | ANCSA corporation land |
| Ambler River side channel | Small bridge | BLM, state selected |
AIDEA Alternative A crossing layer, fields "wname", "cls" and "own". Watercourse identity joined to the USGS National Hydrography Dataset where the applicant left the name blank; AIDEA names only 22 of the 2,932 waters itself. The landowner is the applicant's own attribute, not our spatial join. The twelve river crossings on the map.
The Kobuk is the one to hold onto. The road crosses it inside Gates of the Arctic National Preserve, and the federal review says the chosen route "would cross the Kobuk River and more than 20 of its tributary streams that flow directly into the Kobuk River sheefish spawning grounds." Sheefish, the review says elsewhere, "may be especially vulnerable to population-level effects from mine related dewatering, large scale spills, or leaching of acid rock into waterways," and could be "particularly vulnerable to toxic bioaccumulation" because they mature late and eat other fish. Ninety-four percent of households in the village of Kobuk use them.
2024 Final SEIS, Volume 1, Kobuk River tributaries and sheefish; 2024 Record of Decision, sheefish vulnerability; SEIS Appendix L, Table 5, household use in Kobuk. The metals chain behind "acid rock" is the ground chapter.
Twenty-eight of the 2,932 crossings sit on water the State of Alaska has catalogued as important for the spawning, rearing or migration of anadromous fish. Fourteen of those are named waters, the Alatna, the Reed, the Koyukuk, the John, the Kobuk, the Wild and the Mauneluk among them; fourteen are unnamed streams the catalogue nonetheless lists. Chum and king salmon, Dolly Varden, sheefish and whitefish are the species the catalogue records on them.
The other 2,904 cross water that isn't in the state's catalog, and that doesn't make it water without fish. A missing catalogue entry often means a missing survey, and the federal review makes the same point in its own numbers: it counts 17 crossings of known anadromous water for this route, and 40 once the streams it assumes carry fish are added.
Our count: a crossing is on catalogued fish water if it falls within 60 meters of a stream in the Alaska Department of Fish and Game Anadromous Waters Catalog, 2026. That tolerance absorbs the offset between the applicant's crossing points and the state's stream lines; the method and its limits are in the map's sources sheet. Federal counts: 2024 Final SEIS, Appendix E, Table 16, Alternative A, "known anadromous crossings" 17, "known plus assumed" 40, "essential fish habitat crossings" 13.
The record of decision describes "over 2900 culverts" on the road. The 2020 environmental impact statement explains how they're to be built, and the explanation is an admission:
"Overflow culverts, typically set at higher elevations relative to the primary culvert, would be considered at stream crossings where aufeis formation is probable. The overflow culverts would greatly improve the ability to keep water flowing across the roadway and prevent erosion and damming should flow through the primary culvert become impeded or blocked by ice. Overflow culverts also would be considered at stream crossings where there is a high likelihood of large woody debris (e.g., fallen trees)." 2020 Final EIS, Volume 1, p. 51
Read plainly: the design anticipates that primary culverts will be blocked by ice and by fallen trees, and the answer is a second culvert, "considered" where blockage is "probable." Not at every crossing. Aufeis, the sheet ice that builds where groundwater keeps flowing through the winter, is listed elsewhere in the same document among the region's hazards alongside thaw settlement and thermokarst. This is a road of 2,900 pipes in a place where its own designers expect the pipes to fill with ice.
The Dalton Highway, the road this one would begin from, is the working example. Its culverts clog, it floods, and the state is rebuilding all 414 miles of it. That chapter is here.
2024 Record of Decision, p. 8. 2020 Final EIS, Vol. 1, p. 51 (overflow culverts) and p. 70 (aufeis among regional hazards). 2024 Final SEIS, Vol. 1, p. 242 and Vol. 2, p. 50, Dalton reconstruction "from MP 0 to 414."
A crossing is a structure. Under most of the 211 miles is something simpler: fill. The joint record of decision, as revised, authorises 8,460,218 cubic yards of fill material discharged into waters of the United States, permanently filling 1,431 acres of wetlands, 110 of them inside Gates of the Arctic. The original proposal put the wetland figure at 2,059 acres, and the 2020 statement at 2,079; the revised number is the one the permit carries.
Keep the eight million apart from the fifteen million. Fifteen million cubic yards is the gravel it takes to build the whole road. Eight and a half million is the part of it that goes into water and wetland. They come from different documents and they measure different things.
Joint Record of Decision, as revised, p. 20 and p. 136 (fill volume, wetland acres, acres within GAAR); original figures at p. 135. 2020 Final EIS, p. 60. Gravel volume: 2024 ROD, p. 12. The linear feet of stream affected appear as two irreconcilable figures in the joint record, 77,004 and 250,435, and we do not print either until the agencies say which measures what.
The federal review predicts hazardous-material truck accidents over the life of the road. For the chosen route the range is 316.9 to 2,644.0, roughly six to fifty-three a year, and 121.1 to 1,009.9 of them would be on the new road; the rest are on the Dalton Highway leg. The upper bound is the review's own adjustment to the observed spill rate at the Red Dog Mine between January 2011 and November 2023, applied because the standard national rate, in the review's words, "is underestimated."
"Once contaminants reach unfrozen waterbodies, clean up and removal would be difficult." 2024 Final SEIS, Volume 1
2024 Final SEIS, Vol. 1, hazardous materials transport, Alternative A accident prediction and the Red Dog adjustment; embankment proximity to wetlands and waterbodies. The Red Dog precedent in full, both the success and the record, is in the precedents chapter.
The communities downstream have already lost their salmon to something that has nothing to do with this road, and the chapter isn't honest without saying so. On the Yukon, no Chinook escapement goal was met in 2025, the commercial summer chum harvest has been zero every year since 2021, and the fall subsistence season has been closed six years running. A United States and Canada agreement signed in April 2024 suspends directed Chinook fishing for a full seven-year life cycle "regardless of run abundance." Chum is more than half of the total subsistence harvest in Hughes and nearly half in Huslia, Allakaket and Alatna, villages on the rivers this road would cross.
The federal analysis of the road never engages the marine causes of that collapse; the words "bycatch" and "trawl" appear a combined one time across both volumes of the 2024 SEIS and the record of decision. That is why the road's effect on fish lands on people who have already lost their fish, and why "population level impacts to fish and a reduction in the abundance of harvestable fish," the record of decision's own forecast for named communities, isn't an abstraction here.
ADF&G 2025 Yukon summer and fall season summaries; United States and Canada Yukon River Chinook agreement, April 1, 2024; 2024 Final SEIS Appendix L, Tables 5 and 18 (harvest shares); 2024 Record of Decision, p. 19. Word counts run against the full text of the 2024 SEIS Volumes 1 and 2 and the ROD in our document record, with "caribou" as the control term (777 hits in Volume 1 alone).
What this chapter doesn't say
It doesn't say the road crosses 2,932 rivers. Twelve crossings are rivers. Most of the rest are drainages a person could step across in August. Their number is the point, not their size: every one is a pipe under an embankment in permafrost country, and the design plans for pipes that ice can block.
It doesn't say the unnamed drainages hold fish. Nobody knows. That's the finding.
It doesn't say the road caused the salmon collapse. The collapse is on the Yukon and Kuskokwim and its drivers are in the ocean and upstream in Canada. The road arrives after it, into communities that have already lost the fish the record says the road would affect.
The spill figures are the government's, including the low one. 317 is the review's own floor. We didn't pick the high number and hide the range.
The road crosses water 2,932 times: twelve river crossings, 35 creek crossings, two more on named channels, and 2,883 drainages that carry no name on any map. We counted them ourselves from the applicant's engineering centerline, and the total matches the federal decisions' own figure. Every one is on the map.
AIDEA Alternative A engineering GIS, our own count; 2024 Record of Decision, p. 12. Method on How We Know.By our count, 28 crossings are on water the State of Alaska has catalogued for fish. The federal analysis counts 17 crossings of known anadromous streams and 40 counting assumed ones. The other crossings are on water that isn't in the state's catalog, which isn't the same as water without fish. The federal analysis itself counts 17 known anadromous crossings and 40 when the assumed ones are added.
Alaska Department of Fish and Game, Anadromous Waters Catalog 2026; 2024 Final SEIS.The applicant's file calls 2,903 of the crossings culverts and 29 bridges. The record of decision calls 49 of them bridges. The 2020 Joint Record of Decision explains the gap: in the Army Corps' permit action, AIDEA's design changes replaced 20 medium and large culverts with single-span bridges, and the applicant's file still draws the earlier design. Fish passage is the difference.
The design plans for culverts that ice can block: overflow culverts are to be "considered" at crossings where icing is "probable," in case the main pipe is "blocked by ice," and not at every one of the 2,900.
AIDEA engineering GIS; 2024 Record of Decision, p. 8 and p. 14; 2020 Final EIS, p. 51.The federal review predicts 317 to 2,644 accidents involving hazardous-material trucks over the life of the road, six to fifty-three a year, 121 to 1,010 of them on the new road and the rest on the Dalton Highway, with the upper bound set by Red Dog's own observed spill rate. The ground within 328 feet of the embankment is about 60 percent wetland and water.
2024 Final SEIS. Full citations under How we know on this page.Every crossing on this page is a feature in the applicant's own published engineering geometry, which is why it can be counted, colored and filtered on our map down to the individual culvert. Every federal figure is from the record of decision, the joint record, the 2020 statement or the 2024 supplement, all of which are in our document record and searchable by passage. Where two federal documents disagree we print both and say so.
Twenty crossings are a bridge in one federal document and a culvert in another, and nobody has said which is current. If you're a reporter, that's a one-line question for the Bureau of Land Management and the Army Corps of Engineers, the two agencies that signed the joint record, and it hasn't been asked in print.
If you fish or live on the Kobuk, the Koyukuk, the Alatna or the John, the document to cite is the 2024 Final SEIS, Appendix E, Table 16, and the record of decision at page 14. Our record will pull the passage for you: ask it what the record says about your river.
