The Yellowstone Supervolcano: The Clock Beneath the Park

Yellowstone is a hotspot volcano. Our research file groups it with Hawaii, Iceland, and Reunion: volcanism fed by a fixed, deep-mantle heat source over which a tectonic plate moves. That is a different machine from subduction-zone volcanism, and different again from the mid-ocean ridges. Nothing about a hotspot requires a plate boundary, which is why this track runs across the interior of a continent rather than along its edge.

The track it has left is visible on a map. The Yellowstone hotspot's volcanic trail runs roughly 750 kilometers (450 miles) across the Snake River Plain. Widespread volcanism began about 16.5 to 17 million years ago in what is now northern Nevada and southeastern Oregon, where a large plume head, inferred at roughly 400 kilometers across, is thought to have arrived beneath the North American plate. The volcanism then stepped northeast at roughly 25 to 29 kilometers per million years, essentially the same rate as the plate's own motion in the opposite direction, and reached the present Yellowstone Plateau about 2 million years ago.

At the present Yellowstone location the hotspot has produced three cataclysmic eruptions. Huckleberry Ridge, about 2.1 million years ago, ejected roughly 2,450 cubic kilometers of material and left a caldera 100 by 80 kilometers. Mesa Falls, about 1.3 million years ago, ejected roughly 280 cubic kilometers and left a caldera 16 by 26 kilometers. Lava Creek, about 640,000 years ago, ejected roughly 1,000 cubic kilometers and left the 85 by 45 kilometer basin that lies within the park today.

Two of those three reach the top of the explosivity scale. Our research file's own table puts VEI 8, the mega-colossal class, at more than 1,000 cubic kilometers of ejected material, with a typical recurrence for the class worldwide on the order of once every 100,000 years, and it names Toba (about 74,000 years ago) and Yellowstone (about 640,000 years ago) as its examples. Hold two things about that recurrence figure: it describes a global category, not one volcano's timetable, and Mesa Falls, at roughly 280 cubic kilometers, does not reach the threshold at all.

For scale, and only for scale: Toba's eruption about 74,000 years ago ejected roughly 2,800 cubic kilometers. That is about 1.14 times the ejecta of Huckleberry Ridge, Yellowstone's own largest at roughly 2,450 cubic kilometers, and nearly three times the roughly 1,000 cubic kilometers of Lava Creek, the most recent of the three. Toba has its own file in this wing.

What happened after Lava Creek matters more than the three big events, and it is the part almost nobody hears. In the 640,000 years since that eruption, roughly 30 further eruptions of rhyolitic lava, none of them on the scale of the three, have very nearly filled the basin back in, including a large-volume episode totalling around 600 cubic kilometers between about 180,000 and 70,000 years ago. One caldera did form inside that window, and the map above records it: West Thumb, at about 176,000 years ago and roughly 50 cubic kilometers, well below any of the three. The most recent known lava flow at Yellowstone is the Pitchstone Plateau flow, dated to about 70,000 to 77,000 years ago. Lava flows, not supereruptions, are what this volcano has actually done most often and most recently.

Our research file carries this as a claim under examination and tags it Tier 2, which in that file's convention means the tier the statement would sit at if you took it at face value, not a finding that it is true. The file then answers it in the same section. Everything below is that answer, and it is not close.

The United States Geological Survey states the position without hedging: Yellowstone is not "overdue" for eruption, and volcanoes do not work that way. The cycle figure comes from averaging the only two gaps that three eruptions can provide. Huckleberry Ridge to Mesa Falls is about 800,000 years. Mesa Falls to Lava Creek is about 660,000 years. Three data points cannot establish a statistical period, and volcanic recurrence is not periodic to begin with. The claim sits on the Yellowstone Volcano Observatory's own named list of the top ten misconceptions about Yellowstone volcanism, presented by Michael Poland, the observatory's scientist-in-charge, in a public lecture and media advisory in November 2021.

The number the USGS does put on it is an annual probability of roughly 1 in 730,000 for a caldera-forming eruption at Yellowstone. Our own research file quotes that figure and says plainly, in the same breath, that it does not mean the volcano is due, because the intervals are irregular. A low-probability, very-high-consequence system is a genuinely different thing from an overdue one, and the difference is not rhetorical.

In 2014, Farrell, Smith, Husen and Diehl published a seismic tomography study in Geophysical Research Letters built from 4,520 earthquakes recorded between 1985 and 2013. It imaged Yellowstone's upper-crustal magma reservoir at roughly 90 kilometers long by 30 kilometers wide (about 55 by 19 miles), reaching down to about 16 kilometers (10 miles), with a continuous partial-melt layer at roughly 6 to 12 kilometers depth and an estimated melt fraction of about 9 percent by volume. That is about 2.5 times larger in extent than earlier imaging had shown, and it extends past the caldera's own rim.

In 2015, Huang, Yao, Farrell, Karlstrom and colleagues published a second reservoir in Science, this one in the lower crust at roughly 19 to 45 kilometers (12 to 28 miles) depth, with a volume of about 46,000 cubic kilometers, roughly 4.5 times the volume of the shallower one, and averaging only about 2 percent melt against the upper reservoir's 9 percent. They read it as the conduit connecting the shallow reservoir to the deep mantle plume beneath. Together the two studies gave the first complete image of the Yellowstone magmatic system from plume to upper crust.

The USGS ranks Yellowstone's possible future activity explicitly, from most to least likely, and the ranking is the single most important fact in this file. Hydrothermal explosions are the most likely: small ones occur every few years, typically forming craters only a few meters across, and every few thousand years a larger one can open a crater up to a few hundred meters across. Lava flows stand beside them in the same most-likely band. An explosive, caldera-forming supereruption is named explicitly as the least likely of Yellowstone's possible eruption types, with a probability in any given century or millennium the USGS describes as exceedingly low.

The hydrothermal record backs the ranking. Since the end of the last ice age in this region, roughly 14,000 years ago, more than a dozen hydrothermal explosions have left craters hundreds of feet across. The largest known hydrothermal explosion crater on Earth is at Mary Bay, on the north shore of Yellowstone Lake, roughly 2.5 kilometers (1.5 miles) wide, formed by an explosion approximately 13,000 to 13,800 years ago.

And it happens in the present tense. On July 23, 2024, a hydrothermal explosion at Black Diamond Pool in Biscuit Basin, about 2 miles (3 kilometers) northwest of Old Faithful, threw hot water, mud, and rock roughly 400 to 600 feet (120 to 180 meters) into the air. The boardwalk was heavily damaged and the basin was closed. Nobody was hurt. Because so many visitors were filming, the USGS has called it the best-documented hydrothermal explosion in the park's history.

If a VEI 8 eruption did occur at Yellowstone, our research file's description of the consequences is severe and specific. Surrounding states would be buried under meters of ash. Pyroclastic flows would extend roughly 100 kilometers from the caldera. Measurable ash would fall across the entire continental United States, and 1 to 3 or more years of volcanic winter would follow, with major agricultural disruption. That is the consequence of the least likely scenario, and it should never be read as a forecast of what Yellowstone is expected to do.

The USGS's own framing of those consequences is narrower than the popular one. It describes a giant eruption as having regional effects such as falling ash, plus short-term changes to global climate on the scale of years to decades. The serious hazards are ashfall distribution and the agricultural disruption that follows. Nothing in that framing supports a biosphere-ending reading, or the claim that such an eruption would kill all life in North America. The difference between those two pictures is the difference between a catastrophe and a fantasy.

There is no evidence of any government suppression of Yellowstone monitoring data, and the shape of the refutation is worth noticing: the observatory publishes its monitoring data and its alert-level status continuously and publicly, and issues corrections when false claims circulate. A cover-up is not merely unsupported here. It is refuted by the fact that the data anyone would be hiding is published continuously, in public.

Two dated instances, reported in fact-check coverage in 2024 and 2025, show the pattern in motion. A video circulated claiming to show animals fleeing Yellowstone ahead of an eruption; it was AI-generated, and the real footage showed bison behaving normally on a spring day. A separate viral claim held that the USGS had raised the volcano's alert level to high; the observatory's scientist-in-charge called that one "complete hogwash" in public. Neither claim survived contact with the observatory's own published record.