68 million ton landslide in Alaska: Mount La Perouse

La Perouse Landslide debris 2014
Using this imagery captured by the Operational Land Imager aboard the Landsat 8 satellite on February 23, 2014, scientists have confirmed that a large landslide occurred in southeastern Alaska on the flanks of Mount La Perouse on February 16, 2014. The landslide debris, which slid in a southeasterly direction, appears light brown compared to the snow-covered surroundings. / Courtesy NASA Earth Observatory, image by Jesse Allen and Robert Simmon using Landsat data from the U.S. Geological Survey

Laura Nielsen for Frontier Scientists

February 16, 2014– A roar sounded unheard somewhere in the vicinity of Glacier Bay National Park and Preserve in remote southeast Alaska. Stone and debris, long part of Mount La Perouse, suddenly bowed to gravity as one of the mountain’s near-vertical flanks collapsed.  The colossal landslide carried an estimated 68 million metric tons down Mount La Perouse. Rock, soil, snow and ice mixed and flowed, some of it tracking its way nearly 4.6 miles [7.4 kilometers] from where the landslide began.

How big was it? 100 million tons [1011 kilograms] is about as heavy as all the cars in the United States combined, so 68 million tons is over half the weight of all U.S. cars combined. The landslide flowed toward the southeast, in some places depositing rock, ice and sediments in a layer 40 feet [12 meters] deep, nearly the height of a five-story building.

The landslide may have been too remote for witnesses, but the shaking of all those tons of debris traveling violently down a mountainous incline were certainly felt. Elsewhere, seismographs (devices used to record the motion or shaking of the ground) recorded the rumblings that resulted from the Mount La Perouse landslide. The readings told the story of a massive natural landslide.

Detecting a landslide

Seismic sensors are usually used to detect earthquakes, but their data is also useful for pinpointing landslides. Colin Stark, Göran Ekström and Clément Hibertof the Lamont-Doherty Earth Observatory at Columbia University interpreted the seismic data from February 16th. Their work suggested that a large-scale landslide had occurred in southeast Alaska near Glacier Bay National Park and Preserve.

They shared the information with Dave Petley, Wilson Professor of Hazard and Risk in the Department of Geography at Durham University. His blog, The Landslide Blog, is featured on the American Geophysical Union’s Blogosphere. The Landslide Blog broke the news that a landslide had struck Alaska on February 16th.

Glacier Bay National Park Mount La Perouse landslide Google Earth map
Approximate location of the landslide on the flank of Mount La Perouse in Glacier Bay National Park, southeastern Alaska at 58.542 -137.01 / Google Earth image acquired March 3, 3014

Imaging a landslide

Following the rough coordinates estimated using seismic readings, helicopter pilot Drake Olson located the landslide visually on February 22nd. The landslide’s simulated location was pinned somewhere in the the region of coordinates 58.68, -137.37 and Drake Olson located the debris slide at 58.542, -137.01. The collapse occurred high on Mount La Perouse’s mountain slope at 3000 meters elevation, and the debris sediment sprawled below had already begun to gather a dusting of snow.

Drake Olson shot aerial photographs as well as close up photographs. Those images helped Dave Petley of The Landslide Blog make informed estimations about the landslide’s measurements and materials.

“Drake even landed at the toe of the landslide to get a feel for the constituent materials. This image shows that they consist of ice and rock, suggesting that the landslide has entrained a large amount of snow and ice as it traveled downslope.” – Petley

Then on February 23rd Landsat 8 passed overhead, acquiring satellite images of the location on the flank of Mount La Perouse. The extensive and open Landsat record allows scientists to examine before- and after- images of natural events like landslides. The NASA Earth Observatory website coverage of the landslide features an interactive of the two landscapes below; it allows you to drag a vertical line to cover or uncover the before- and after- images.

La Perouse Landslide comparision 2014
Before (acquired May 27, 2013) and after (acquired February 23, 2014) images of the Mount La Perouse Landslide location. The landslide occured on February 16, 2014. / Courtesy NASA Earth Observatory, image by Jesse Allen and Robert Simmon using Landsat data from the U.S. Geological Survey

Remote and snowy

Even though this landslide moved an estimated 68 million metric tons, it still took a coalition of landslide hunters to track it down. In this case Dave Petley noted that without the coordinated effort, snow could have covered the Mount La Perouse landslide before it was ever seen.

Landslides aren’t always easy to see. Many occur at high elevation in locations that are so steep or cold that few human settlements are nearby. Giant landslides like February 16ths might go entirely unnoticed.

Because there are so few direct observations, we don’t have a terribly firm grasp on the evolution of big landslides. What sparks catastrophic landslides? How, precisely, do the rocks and other materials break, fall, slide, interact, and flow? How swiftly do catastrophic landslides move, and what guides how their movements evolve before the debris finally comes to rest?

Wrangell-St. Elias National Park landslide 2013 satellite image
In July 2013, the Operational Land Imager aboard the Landsat 8 satellite captured this image of Wrangell-St. Elias National Park A landslide had occured there at 61.978, -143.168, near the center of the image. The landslide site is outlined in black, and the top edge of the outline is indicated by a red arrow. This image illustrates the difficulty of spotting landslides using satellite imagery alone. / Courtesy NASA Earth Observatory, image uses Landsat data from the U.S. Geological Survey

The more data we can collect on big landslides, and the more landslide events we can record, the greater our knowledge and preparedness when it comes to catastrophic landslides.

Seismic networks

Enter a new remote sensing approach based on seismic data.

Seismographs or seismometers are set up all over the world, mostly to measure the motion of the ground that occurs during an earthquake. Therefore, a global network of seismic monitoring instruments is available to provide data about the earth shaking from many different locations.

Seismometers record vibrations that travel through Earth’s crust due to the motion of tectonic plates, moving rock or ice or debris, and even flowing magma. Seismic data can detect very large landslides because landslides cause the ground to shake. Ground tremors are recorded by seismometers as waves. While earthquakes are more likely to cause extremely short bursts of high-frequency waves, landslides more often produce long low-frequency waves. The falling, tumbling, sliding mass of debris grinds against the ground during a landslide, accelerating and slowing, creating a sustained undulating seismic signal that can last minutes.

In 2013 geophysicist Colin Stark and seismologist Göran Ekström, both of Columbia University’s Lamont-Doherty Earth Observatory, published a paper proposing a new method for using seismic data to detect catastrophic landslides. The paper was titled “Simple Scaling of Catastrophic Landslide Dynamics” and published in Science Volume 339. “It’s especially critical that we monitor ‘catastrophic’ landslides—fast-moving slides that involve more than a million tons of debris,” Stark believes. “These are the most dangerous landslides, but they often go undetected.”

Their method, the Global Centroid-Moment Tensor Project, systematically analyzes seismic readings. It closely examines low-frequency waves. If a seismic event produces low-frequency waves but not high-frequency waves then it is likely to be a landslide as opposed to an earthquake.

The method uses an analytical model, a highly complex algorithm run by supercomputer, to assess low-frequency readings that have been collected by the global seismic monitoring network. This is important because it’s automated: a rapid detection tool sifts through data and makes note of any signal indicative of a significant landslide event. The Global Centroid-Moment Tensor Project detection tool is what helped the researchers at Columbia University detect the landslide that occurred in Alaska on February 16th.

Lituya Bay Alaska
An aerial photograph showcasing the magnificent scenery of Lituya Bay, Gulf of Alaska. An earthquake here in 1958 caused a huge landslide at the head of the bay that generated the largest recorded wave in history. Trees were sheared off up to 1,720 feet above sea level on the north side of the bay. / Courtesy Alaska ShoreZone Program NOAA/NMFS/AKFSC & Mandy Lindeberg, NOAA/NMFS/AKFSC

Hidden in the signal

Those seismograph readings can tell us quite a bit about landslides. Stark and Ekström’s method not only detects landslides, it also lets scientists learn about their characteristics.

The ‘seismic fingerprint’ of a large landslide collected from multiple seismometers can be used to calculate: the time when a landslide occurred, its direction, its duration, and the landslide’s magnitude.

Ekström said, “The vibrations in the Earth actually record what these forces are. We are not blind to them. Hopefully, that will help us understand landslides and use that knowledge to predict where and how they will happen.”

While seismic data can tell us much, readings from the global seismometer network cannot precisely pinpoint the location of a landslide. So far they can only give a range as small as 10 square-miles [25 square-kilometers] or as large as 40 square-miles [100 square-kilometers] depending on the data available.

The landslide must be found. Satellite images can be used to confirm that the recorded event was in fact a landslide. It’s also important to get visuals because when the researchers combine seismic data with satellite images they can learn more. With both seismic data and satellite imagery, they can calculate the mass of the debris that fell and the length of the landslide’s path, which can be used in further calculations: trajectory, velocity, and acceleration data. Taken together, all of this information allows researchers to create landslide simulations.

The data can even reveal what terrain the debris traveled over; landslides that move over ice can slip along the slick surface, traveling far. The ice temporarily softens the seismic signal, acting as as insulator, and then the vibrations reappear enforce as the debris slams into stony earth once again.

Black Rapids glacier landslide 2012
A landslide flowed over the Black Rapids glacier in the Alaskan Range in 2002, following a 7.9 earthquake on the Denali Fault. The rock debris from the landslide contacted the glacier, and flowed (left-to-right) across the 1 mile width of the Black Rapids glacier. / Courtesy Dennis Trabant, U.S. Geological Survey

Working with satellites

The landslide-detection method depends on interplay with satellite images to confirm the occurrence of a landslide and to provide more data. Happily we live in an era of astonishing satellite capabilities.

Landsat 7 and Landsat 8 are Earth-observing spacecraft; each orbit the Earth once every 99 minutes and cover the same track every 16 days. Working together, the two satellites image every region on Earth every 8 days. “The system has been returning over 500 scenes or Landsat images per day to the 41-year free-and-open Landsat data archive that’s maintained by the U.S. Geological Survey,” noted Jim Irons, Landsat 8 project scientist at NASA Goddard. The National Aeronautics and Space Administration (NASA) and the U.S. Geological Survey (USGS) work together on the Landsat initiative. Irons added: “The Landsat program benefited the U.S. economy by 1.79 billion dollars in 2011. I’m confident that Landsat 8, with its excellent performance, will increase that annual return on investments both in terms of the economy and in terms of the scientific return from analyses of the outstanding data that we’re getting.”

After the automated computer algorithm noticed a likely landslide in southeast Alaska on February 16th, Stark informed NASA staff and requested the acquisition of higher-resolution imagery. The Operational Land Imager aboard NASA’s Landsat 8 satellite took the image you can see at the head of the article.

Stark and Ekström’s seismic analysis joins information from the before- and after- satellite imagery in an effort to understand and model the motion of the landslide.

Detecting historical landslides

We’re always seeking to gain a better understanding of how landslides behave. Stark and Ekström have also applied their technique to historical data, analyzing old seismograph data. Their innovative approach was able to detect about 29 large landslides that occurred between 1980 and 2012 which had not registered on standard earthquake-monitoring systems. What’s more: in some cases they might have learned more information about historical landslides using their new methodology than was on record before.

Seward Alaska before and after tsunami earthquake 1964
On March 27, 1964, Alaska was rocked by an earthquake. Coastal Pacific towns were devastated by earthquake-generated underwater landslides, surge- waves, and tsunami waves. These two pictures show the waterfront at Seward, Alaska. On the left is the before- picture (looking south), while on the right is the after- picture (looking north) The small boat harbor, the railroad yards, the large docks, and other water facilities were removed by the underwater landslides. Note the severed tracks in the railroad yard and the tumbled heaps of railroad cars and other debris thrown up by the tsunami waves. / Courtesy U.S. Geological Survey

Major advantages

“There are regions that have a lot of landslides, like the Himalayas and Hindu Kush, where it would be beneficial to know when things have happened up in the mountains,” Ekström said. Even if a landslide doesn’t strike a populated area, landslides high in the mountains or in far remote areas can cause dangerous flooding by temporarily damming a waterway. Once the water builds up it can break through the debris dam, rushing violently downstream toward inhabited areas.

The automated sensing system gives us a tool; we know better where to look for landslides. Early detection and warning warning can give communities time to evacuate. Also, city planners can use the information gained about high-risk areas and landslide dynamics to practice natural-hazard mitigation and disaster management.

“We used to think they were pretty rare,” Dave Petley said about large landslides, “But we’ve probably just missed them in the past.” Perhaps six catastrophic landslides happen in a typical year –separate from large earthquake events, which can trigger multiple landslides.

Warming temperatures?

The Mount La Perouse landslide, February 16, 2014, involved a massive mountain slope failure.

Lewis Sharman, a Park Service ecologist at Glacier Bay, told Kurt Repanshek of National Parks Traveler that geomorphologists anticipate landslides “Becoming a little more frequent as we have warmer than usual temperatures, we have more melting events, and glaciers and surface sediments become better lubricated on surface terrain at elevation.”

Colin Stark noted in 2013: “Most of the big landslides that I’ve worked on in Alaska from 1999 to now have been south-facing, summer-time failures.” He added that a 2013 “Landslide was probably caused by sustained daytime warming and progressive melting of rock permafrost.”

It could be that unusually warm temperatures in Alaska this winter helped to spur the Mount La Perouse landslide, 2014.


Be Prepared: The Landslide Handbook—A Guide to Understanding Landslides (A free resource) by Lynn M. Highland, United States Geological Survey, and Peter Bobrowsky, Geological Survey of Canada

See Alaska: Pilot Drak Olson provided photographs he took of the landslide on Mount La Perouse’s flanks, Fairweather Range, Glacier Bay National Park, southeastern Alaska. See his images posted on Dave Petley’s The Landslide blog: Mount La Perouse: Sunday’s rock avalanche in Alaska has been found and additional images here.

Frontier Scientists: presenting scientific discovery in the Arctic and beyond


  • ‘Breaking news: a very large landslide in Alaska on Sunday’ Dave Petley, The Landslide Blog (February 20, 2014)
  • ‘Mount La Perouse: Sunday’s rock avalanche in Alaska has been found’ Dave Petley, The Landslide Blog (February 22, 2014)
  • ‘Large Landslide Detected in Southeastern Alaska’ NASA Earth Observatory (February 25, 2014)
  • ‘Simple Scaling of Catastrophic Landslide Dynamics’ Göran Ekström and Colin P. Stark in Science Vol.339 (March 22, 2013) p.1416-1419 DOI: 10.1126/science.1232887
  • ‘Hunting Landslides with Landsat’ NASA Earth Observatory (August 30, 2013)
  • ‘Landslides detected from afar’ Erin Wayman, Science News Magazine (March 21, 2013)
  • ‘Landslide dynamics from seismic wave inversion, satellite remote sensing, and numerical modeling’ National Science Foundation Award Abstract #1227083 (September 1, 2012)
  • ‘Listening with Seismology Could Predict Landslides’ Becky Oskin, OurAmazingPlanet, LiveScience (March 21, 2013)
  • ‘Massive Landslide Coats Glacier Bay National Park’s Johns Hopkins Glacier Like Chocolate Frosting’ Kurt Repanshek, National Parks Traveler (July 11, 2012)
  • ‘Some of Earth’s Biggest Landslides Are Surprisingly Hard to Find’ Adam Mann, WIRED (February 28, 2014)
  • ‘Taking Landsat to the Extreme PressConference’ American Geophysical Union Fall Meeting 2013
  • ‘The Shaky Side of Landslides’ Sid Perkins in Science NOW (March 21, 2013)