Ninety minutes. That is how long the Mother Kuskokwim Tribal Coalition says residents of Crooked Creek, Alaska would have to escape a wall of mining sludge if the tailings storage facility at the proposed Donlin Gold mine ever let go.
A newly released, court-ordered draft supplemental environmental impact statement from the US Army Corps of Engineers and the Bureau of Land Management has now put numbers to that scenario. It models a failure releasing about 27% of the mine's tailings capacity — enough, the document finds, to bury Crooked Creek in waste and likely render it uninhabitable, with metal contamination downstream persisting for more than a century, according to MINING.COM.
That the US government felt compelled to study this at all is the story. Donlin sits on one of the largest and highest-grade undeveloped gold deposits anywhere, holding approximately 40 million measured and indicated ounces grading 2.22 grams per tonne. It is a project worth building, in other words — which is exactly why the risk calculus around its waste dam matters so much. And it arrives at an odd moment: the tools to catch a failing dam weeks before it fails have moved from research paper to proven method, and nothing yet compels a mine like Donlin to use them.
A river, a coalition, and a warning
For the Organized Village of Kwethluk, the dispute is not abstract. "Our primary concern all along has been the Kuskokwim River, which has sustained our customary and traditional practices for generations,," said its president, Mary Nicolai. Dr. David Chambers of the Center for Science and Public Participation frames the underlying hazard more clinically: "A tailings dam failure is a low-probability, but high-consequence event, and a threat which exists at every mine with a mine waste facility,," he told MINING.COM. Tailings storage facilities are among the largest structures humans build and among the least stable — an estimated 10 percent of them have experienced stability problems historically, according to Sciencedirect. Over the past century, more than 250 tailings dam collapses worldwide have released as much as 250 million cubic meters of waste and caused approximately 2,650 casualties. Brazil supplied the grimmest recent entries: the Feijão "Dam I" failure in 2019 killed at least 260, and the Fundão facility failed in 2015 with 19 fatalities.
The Army Corps is now taking public comment on the Donlin scenario from Sept. 23 through Oct. 23, with hearings scheduled in Alaska and online in October.
What the satellites already saw
Here is what makes the timing sting. Researchers studying five major mining structures across continents — Cadia in Australia, Çöpler in Turkey, Feijão in Brazil, Jagersfontein in South Africa, and Żelazny Most in Poland — built a satellite radar forecasting method and ran it backward against dams that had already failed. At Cadia, a low-confidence alert would have surfaced up to two months before collapse, sharpening to high confidence twelve days out, according to Sciencedirect. At Feijão, the same method produced a moderate-confidence alert 39 days ahead of the disaster. Retrospective analysis of Çöpler found measurable surface displacement before that dam gave way — the kind of slow creep and accelerating deformation InSAR is built to catch, per Miningdoc.
The mechanics are almost mundane. By comparing radar images of the same patch of ground over time — interferograms, in the jargon — InSAR picks up subsidence, uplift, and lateral movement that can precede a failure once separated from ordinary tailings settlement, according to Geofem. The European Sentinel-1 constellation, which supplied the imagery for much of this research, revisits a given site every 6 to 12 days, per Sciencedirect — frequent enough, in several of these retrospective tests, to have bought weeks of warning.
It is not a silver bullet. A 2024 academic study using more than 400 Sentinel-1 images to study the Żelazny Most dam in Poland concluded satellite InSAR "can serve as a hazard-screening tool" but cautioned that most potential failure modes "may not exhibit InSAR-detectable accelerations" for real-time prediction, according to Springer. Displacement monitoring has conventionally relied on field observations, ground instrumentation, LiDAR and photogrammetry; satellite radar is best understood as a complementary layer, cheap and wide-reaching but not omniscient, per Springer.
The gap nobody has closed
That caveat is honest, and it is also not the point. Nobody is claiming InSAR would have caught every failure mode at every dam. What the Cadia and Feijão retrospectives show is narrower and harder to wave away: in cases that mattered most, the signal was there, sitting in freely available satellite data, weeks before the wave hit. No agency reviewing Donlin's tailings design has required that signal be watched.
The comment window on the Donlin draft closes Oct. 23. Whatever regulators decide about the mine itself, the question the satellites have already answered — whether a slow-motion failure can be seen coming — is no longer in doubt. Whether anyone is made to look is a different matter entirely, and one that will outlast this single mine, this single river, this single village with ninety minutes to run.
