It took less than seven minutes.

On the morning of September 20, 2002, the Kolka Glacier sat in a gently sloping valley on the Russia-Georgia border. The glacier had a history of unusually fast surges, which—for the ponderous ice flows that cover 10 percent of Earth’s land surface—meant it sometimes lurched several dozen feet forward in one day. When an American scientific satellite, Landsat 7, passed overhead and imaged it around noon, Kolka seemed unsteady but unremarkable.

At 8:08 p.m., it became remarkable. More than three-quarters of the entire glacier, a goliath chunk of ice more than 1.5 miles long, broke off from the rest of the formation and detached from the soil below. It thundered down the side of the Kolka valley, 130 million cubic meters of ice and rock, accelerating as its own meltwater slicked its path. Within minutes it was traveling 150 miles per hour. A scientific account of the cataclysm says that the detached section then “literally flew” over another glacial bed and ricocheted into a turn.

Now an avalanche of snow, ice, and rock, it approached the village of Nizhniy Karmadon. There was nothing its inhabitants could do. By 8:13 p.m., it entered and overwhelmed the town, killing more than 100 people, including a famous Russian actor named Sergei Bodrov, Jr. The ice kept going. A minute later, it snapped a set of telephone lines downhill. Finally, at 8:15, it came to rest in the Karmadon valley below. In six and a half minutes, it traveled 11 miles.

Experts had never seen a glacier move like Kolka before. It seemed like the icy equivalent of a pyroclastic flow of hot gas and rock that gushed out of Vesuvius and flattened Pompeii. Most incredibly of all, Kolka had achieved high speeds on a surface that was inclined an average of only six degrees above the horizontal.

“You’re not talking about a glacier that fell off a mountain peak,” says Stephen Evans, a professor of geological engineering at the University of Waterloo. He studied the Kolka site in person soon after the event. “I walked up [the hillside] without breaking a sweat. Six degrees is almost flat for all intents and purposes.”

Evans and a number of other researchers proposed that the Kolka detachment and avalanche represented a previously unknown form of glacial movement. Most glaciers advance at a dawdling pace, sliding about 10 meters (32 feet) forward per year. At times, though, glaciers can surge for weeks at a time, moving 10 meters per day. Then they usually return to baseline advancement.

The Kolka glacier had become something more—they termed it a glacier-debris flow. It traveled more than 10 meters per second.

Nobody could study any other incidence of this new type of glacier movement, though. Kolka was the only scientifically documented cataclysm of its type, the only glacier-debris flow ever to be researched on-site by glaciologists, and the only sudden collapse ever to be observed before and after by satellite. Though some researchers believe the same Kolka site experienced a similar failure in 1902, science knew of no other incident for sure.

That is no longer the case. This summer, two more glaciers failed not in months, but in minutes—and they did it under the watchful eye of some of the world’s largest satellite constellations.

On July 17, 2016, in Tibet, a Himalayan glacier gave way in a fashion very similar to the Kolka failure. More than 60 million cubic meters of ice and rock plunged out of its bed and spilled into the valley below, eventually stopping in Lake Aru Co. Nine people died, all herders, as did hundreds of sheep and yak. Here as in Russia, ice seems to have given way with little warning and to have achieved high speeds on relatively flat land.

The aftermath was first noticed by Westerners in an image captured by the European Space Agency’s workhorse Earth-observing satellite, Sentinel-2. Evidence soon quickly circulated around the glaciologist community. But Sentinel was not the only satellite to see the Tibetan glacier.

The scene of the twin glaciers in Tibet before the collapse. A Planet satellite captured this image in January of this year. (Planet)
The glacier before its collapse in early July. Some crevasses, a sign of imminent surging, can be seen at its base. (Planet)
The glacial bed after the total collapse, with the former glacier area highlighted in red. (Planet)

Andreas Kääb, a professor at the University of Oslo, discovered that the constellation of remote-sensing satellites operated by Planet, a San Francisco-based private satellite startup, had also captured the glacier-debris flow. Planet had dozens of images of the Tibetan site, all with a higher spatial resolution than Sentinel-2.

This Planet imagery revealed new aspects of the Tibetan glacier. Before the Aru glacier failed, it looked like it was about to surge. Water pooled on its surface, and crevasses started to appear. Surging, though, is a regular part of a glacier’s life. Why had the Tibetan glacier failed catastrophically and not started surging like it should have?

For now, this question will require further study. More worrying was a feature that Kääb noticed in a Sentinel-2 image, something to the south of the avalanche’s debris field. Another glacier had started exhibiting the same characteristics as the Tibetan glacier which collapsed. It showed large crevasses. Would it collapse as well?

The Sentinel image was captured on September 19. Because of processing delays, Kääb did not see it until September 21. Kääb (and another team which reached a similar finding) notified the Chinese government as soon as they deduced the danger—but both were thwarted by time. The second glacier catastrophically failed on September 21, casting ice and debris into the same valley. Luckily, no one was hurt or injured. A Chinese surveying team is already on the ground in Tibet.

This GIF show the two glaciers spilling into the valley below. In the first scene, from May 18, the valley has only traces of white run-off from the glaciers. In the second, from September 14, one glacier has collapsed, its ice and rock spilled across the valley. In the third, from October 8, the second glacier has also failed. These images were captured by Planet satellites. (Planet)

“Now we have imagery that shows them side by side—the two debris fields looking almost exactly the same,” said Evans. He said further on-the-ground research had to be done, but for now, it seemed like “it’s more or less the same process as the Kolka event.”

He praised the ubiquity of free or cheap remote-sensing data in making so fast a diagnosis. Not only did researchers have access to high-resolution satellite images of the avalanche sites, they also could measure the size of the glacial displacement using radar satellites like ESA’s Sentinel-1. “It’s only comparatively recently that we’ve had this, the last 10 years maybe,” said Evans.

Planet believes that the incident shows the validity of their approach to remote sensing. The available imagery of most disaster areas spike immediately after the event takes place, said Joseph Mascaro, a tropical ecologist who now works at the company. Then, a few months after the cataclysm, imagery slows down.

Planet is one of a number of private satellite companies trying to image the entire planet everyday. “Therefore, when an event occurs, you have prior imagery to help understand how the situation unfolds on the ground,” said Mascaro. “But then you also have a reliable stream of imagery that more importantly doesn’t go away.”

Evans said it was too early to know whether this kind of event is climate-triggered.

“There’s been [worldwide] glacier retreat that’s obvious from the various satellite images, but the mechanism linking glacial instability and regional warming has not been mapped out in detail,” he said. Because two glaciers nearby failed at the same time, the source almost certainly had to be thermal, but researchers lacked a good understanding of the recent dynamics of the climate for this area of Tibet.

Glaciologists did not learn of the first Tibetan collapse until early September, so in the course of several weeks, they went from having one observed case of a theoretical phenomenon to three. These remain the only three documented cases of rapid glacier collapse in the past century. Evans believes there will be more.  “Nature just doesn’t invent a new way of operating,” he told me. “I think it will be a newly recognized process of catastrophic glacier instability.”

And it points to a fact of life in the climate-changed age. The surface of the planet is transforming faster than it has in our entire history as a species. Natural cataclysms now surround us. But even as the planet itself undergoes a transformation, our tools for observing and studying it in real time have never been better.