Glacial Lake Missoula was born from a plug of ice. As the last Ice Age tightened its grip, the Purcell Trench lobe of the Cordilleran Ice Sheet, also called the Clark Fork lobe, pushed south into the Clark Fork River valley of what is now northern Idaho and dammed it near the site of modern Lake Pend Oreille. The river had nowhere to go, so it began to fill the valleys behind the ice.

At its highest, the lake was enormous. It stood roughly 300 meters deep against the ice dam, reached about 300 kilometers eastward across the mountain valleys of western Montana, and held on the order of 2,500 cubic kilometers of water. That is a body of water the size of a large inland sea, held back by nothing but a wall of ice.

The proof that the lake existed is still written on the hillsides above Missoula, Montana. More than 30 horizontal shorelines, or strandlines, are cut into the flanks of Mount Sentinel and Mount Jumbo, each one a bench marking a level where the lake surface once stood. Stacked one above another, they record a lake that filled and drained again and again rather than a single time.

That is exactly what happened. The full cycle, ice dam, filling lake, dam failure, catastrophic drainage, then a fresh dam as the ice pushed south again, repeated dozens of times over a span of roughly 2,000 to 3,000 years, between about 18,000 and 15,000 years ago. Every turn of the cycle ended in a flood.

When those floods reached the Columbia Plateau of eastern Washington, they carved a landscape unlike anywhere else on Earth. The geologist J Harlen Bretz named it the Channeled Scablands: an expanse of roughly 15,000 to 40,000 square kilometers of deep dry canyons, abandoned waterfalls, and basalt scraped clean of its soil, threaded by dead channels and streamlined hills. It looks like the bed of a river that has been switched off.

The largest of the carved channels are the coulees. Grand Coulee runs about 80 kilometers long, between 1.5 and 10 kilometers wide, and as much as 300 meters deep in places, a canyon cut not by a river over millions of years but by floods over days. Moses Coulee is another of the great dry channels in the same system.

The single most dramatic feature is Dry Falls, a horseshoe-shaped cataract 120 meters high and 5.5 kilometers wide. It is the scar of a waterfall that has fallen completely silent, a curving cliff line with no river above it, standing over empty plunge pools.

The floods left their signature in more than the big canyons. Across the plateau they stripped away the loess, the wind-blown soil, to expose bare Columbia River basalt, the raw scabland that gives the region its name. They left hanging valleys, side channels whose floors now sit tens of meters above the scoured main channel they once fed. And they left streamlined hills, teardrop-shaped remnants of loess sculpted by water that flowed around them, sometimes called loess islands.

At their peak, the largest floods discharged an estimated 10 to 17 million cubic meters of water per second. That is the greatest freshwater discharge known anywhere on Earth, roughly 10 times the combined flow of every river on the planet today, all of it forced through a handful of channels in eastern Washington.

Where the flood squeezed through constrictions it accelerated. Flow velocities in those narrows are estimated at 20 to 30 meters per second, somewhere between 45 and 67 miles per hour, fast enough to keep the giant boulders it left behind in motion and to grind its channels straight into solid basalt.

The water was not only fast but deep. In the major channelways it ran 100 to 250 meters deep. Where it met a bottleneck it backed up: at Wallula Gap, the narrow notch the Columbia River threads through, the flood ponded against the constriction and rose to about 370 meters above sea level, drowning the land upstream until it could force its way through.

In 1923, the geologist J Harlen Bretz published a paper in the Journal of Geology titled 'The Channeled Scablands of the Columbia Plateau,' arguing that the whole scarred region had been cut by a sudden, catastrophic flood.

In January 1927, Bretz laid out his evidence before the Geological Society of Washington. The established geologists in the room rejected it almost to a person, on the principle that catastrophic floods had no place in modern geology. Their criticism came with a telling gap: not one of them offered any alternative explanation for how the Scablands had actually formed.

The critics' strongest point was a fair one: where could a flood that big possibly have come from? The answer came from another geologist, Joseph Thomas Pardee, who had been studying an ancient lake in the mountains of western Montana. Pardee documented Glacial Lake Missoula and even found its own giant ripple marks on the old lake bed, and in doing so he handed Bretz the missing water source his critics had demanded: a vast lake, held by ice, ready to fail.

The turning point came in 1965, when the International Association for Quaternary Research brought a field conference into the Scablands themselves. Standing in the dry coulees and beneath the abandoned falls, with the evidence laid out in the ground around them, the visiting geologists were largely won over.

The final vindication came in 1979, when the Geological Society of America gave Bretz its highest honor, the Penrose Medal. He was in his mid-nineties by then, most likely 97, though the sources differ and some give his age as 96. More than half a century after he was rejected, the discipline that had refused his flood accepted it.

When spacecraft returned images of Mars showing enormous dry channels, features like Ares Vallis and Kasei Valles, geologists needed an earthly landscape to compare them against. The Channeled Scablands were the closest thing on Earth. Beginning in the 1970s, Vic Baker and others used the Scablands to argue that the Martian channels had themselves been carved by catastrophic megafloods early in the planet's history. Bretz's once-heretical flood became a working tool for reading the surface of another world, and part of how catastrophic processes won back a legitimate place in the earth sciences.

The first is the simplest to ask and the hardest to answer: exactly how many floods were there? Estimates range from about 25 to more than 80. Careful counts of the stacked flood layers point to at least 40, but the true number is genuinely uncertain, because the smallest floods may have left no lasting deposit, and later floods scoured away the traces of earlier ones. The best current reading is a flood every 35 to 55 years on average, though the real intervals surely varied.

The second question is who, if anyone, was there to see them. There is limited but real evidence that Clovis or pre-Clovis people were present in the Pacific Northwest during the last of the flood cycles. The floods reshaped the human landscape either way, scouring the uplands while laying down the fertile bottomlands of the Willamette Valley and the Portland basin. No direct archaeological trace of anyone caught in a flood has ever been found, but repeated inundation of the Columbia corridor would have been a deadly hazard for anyone living in its path. On that point, the evidence is simply silent.

The third is whether the floods left a fingerprint on the world's climate. The floodwaters drained to the Pacific, not the Atlantic, which makes a direct effect on North Atlantic circulation unlikely, and for a long time that was where the question rested. But a 2020 study by Praetorius and colleagues in Science Advances reopened it, modeling how meltwater from the Columbia megafloods could have spread into the North Pacific and, under the right conditions, been routed north and around into the Atlantic within about a decade, reaching the very regions that drive global ocean circulation. It is a far more roundabout path than anyone had assumed, a peer-reviewed model rather than a settled result, and very much an open research question today.

It is tempting to wonder whether the memory of the floods survived in the oral traditions of the region's first peoples. Some scholars, notably Clark in 1953, have pointed to flood stories told by the Kalispel, the Spokane, and other Plateau peoples and suggested a connection. It is an evocative idea, but the last floods were roughly 15,000 years ago, more than 600 generations back, and no oral tradition can be traced that far with any confidence. This one has to stay speculation.

One nuance is worth being honest about, because it is easy to flatten. Bretz's original 1923 idea was a single, one-time catastrophic flood. The modern consensus, built from the rhythmites, the strandlines, and the dating, is that there were dozens of separate floods, each catastrophic in its own right. So the version of the story that finally won is not quite the version Bretz first told. The catastrophic-flood origin he fought for is thoroughly vindicated; the single-flood detail is not, and has been superseded.

And no, the Missoula Floods are not evidence for Noah's Flood, despite being cited that way. They were genuinely catastrophic, but they were many separate events, not one; they were confined to the Pacific Northwest, not global; and they played out over thousands of years at the end of the Ice Age, not in a single worldwide deluge. The very things that make them scientifically real, the multiple floods, the local geography, the long timespan, are exactly what rule them out as one global flood.