At 5:26 in the morning, the north wall of Tracy Arm let go.
A slab of mountain roughly 80 million cubic yards in volume dropped into the narrow fjord about 80 miles south of Juneau, Alaska.
Three kayakers were camped on Harbor Island near the mouth of the arm, asleep in the pre-dawn dark.
One of them woke to a roar and saw a wall of water rushing past the tents.
Miles behind them, the same wave had already climbed 1,578 feet up the opposite rock face. So how does a falling mountain produce a wave that tall?
What a fjord does to a falling mountain
A landslide in open ocean spreads its energy in every direction, thinning fast as the circle grows. In Tracy Arm the facing wall stands only about a mile from where the rock came down, and that narrowness turns the water into something closer to a piston, with most of the energy having nowhere to go but straight up the cliff opposite.
An initial breaking wave roughly 330 feet tall raced outward at more than 155 miles per hour within the first seconds. That wall of water struck the far cliff with much of its force still behind it, stripping trees and soil from the rock in a single pass and leaving a pale trimline far above the waterline.
Then the fjord kept sloshing. Water moved back and forth, a phenomenon called a seiche, for more than a day after the landslide, like a deep bathtub nobody could reach to steady.
The kayakers on Harbor Island
The three campers had spent 78 days paddling the Inside Passage from Washington and planned two more weeks working toward Glacier Bay. They had taken a rest day beforehand, sitting out about three inches of rain, and were camped on an uninhabited islet in Holkham Bay, where Tracy Arm and Endicott Arm meet.
One of them awoke to the roar and saw a massive surge rushing past the tents. Kayaks and gear were swept away as the water washed over the island, though the campers themselves were unharmed. A charter yacht that heard their call on marine radio picked them up later that morning.
Passengers aboard a small cruise vessel in neighboring Endicott Arm also reported swings in water levels and a strong current. The wave arrived before the tour boats had reached the upper fjord. In summer, Tracy Arm draws an average of three cruise ships a day, plus other vessels that pass within a few miles of where the slope failed.
How the numbers were pinned down
The Alaska Earthquake Center first detected the event through seismic waves consistent with a large landslide at about 5:30 in the morning, and the collapse radiated a signal comparable to a magnitude 5.4 earthquake. Poor weather closed in afterward, and it was three days before a helicopter reconnaissance flight reached the site.
Satellite imagery and elevation model analysis put the wave’s reach on the hillslope opposite at 1,578 feet. The tsunami also registered on a tide gauge in Juneau, where peak height reached about 14 inches above the tide, with waves continuing for hours.
That faint echo at Juneau is a striking contrast. The Kamchatka tsunami showed how far a wave can travel before a distant gauge catches it. Here the pattern ran in reverse: the most violent energy stayed trapped inside the fjord walls, and only a whisper escaped roughly 80 miles to the coast.
What the mountain had been doing before it fell
The landslide was preceded by several days of microseismicity that increased in rate and magnitude in the run up to failure. Nobody read that signal as a warning at the time, and scientists have since said openly that Tracy Arm was not on their list of especially dangerous fjords.
South Sawyer Glacier had been pulling back from the base of that wall, retreating roughly 1,800 feet in the year before the slide alone and removing the ice that had long buttressed the slope. Heavy rain added to the load. Without the glacier pressing against its foot, the rock face above became progressively less supported until the early morning it finally gave way.
That link between a retreating glacier and a destabilized slope is not unique to Tracy Arm. Arctic sea ice trends track the same broad warming, and coastal fjords face increasing risk of slope failures as ice pulls back from valley walls. Monitoring microseismicity on vulnerable slopes is one approach researchers are now weighing.
The second highest ever recorded, and what comes next
The 1,578-foot runup falls just short of the global record set in Lituya Bay in 1958, which the U.S. Geological Survey puts at roughly 1,740 feet. That earlier event was triggered by an earthquake that shook rock into the bay. The Tracy Arm slide had no earthquake behind it at all.
Scientists combined satellite, airborne and ground based observations with eyewitness accounts and computer simulations, publishing their full analysis in Science. It traces the sequence from the first microseismic tremors to the last faint oscillation on the Juneau tide gauge, and notes that the seiche was only the second known to produce an observable global seismic signal.
What the work could not do is say which slope will go next. Tracy Arm was not flagged as high risk beforehand, and that surprise itself is part of what the researchers argue needs addressing in fjords full of summer traffic.
In a narrow arm of the sea, the margin is thin. The mountain let go at 5:26, and the water was in the kayakers’ camp by 5:45.
Read the whole thing?
Get the week's signal, not the noise
Our sharpest reporting on energy, climate and nature — free, once a week.
