Sasha Calvey awoke to an ominous sound on August 10, 2025, a deep roar like a waterfall that grew alarmingly louder as it approached her campsite on a remote island in Southeast Alaska. She and two friends were 74 days into a sea kayaking expedition from Washington when the unusual noise jolted them from their sleeping bags. The overnight rain and strong gusts had swayed the towering spruce and hemlock trees, but this new sound, which seemed to vibrate the very ground, signaled a far greater danger. "Everything in my gut told me to check outside," Calvey recounted.

Scrambling from their tent, they discovered that 15-foot waves had already slammed into the beach, pushing water deep into the surrounding forest. Seaweed lay scattered on the moss beside their tent, a stark indicator of the ocean’s surge. Their carefully stowed food and gear were strewn across the gravel beach, with half of it swept away. One kayak was found wedged against a tree, another precariously hung from a bluff, and the third drifted in the distance. The friends were acutely aware of their near-miss; had the tsunami struck just three hours earlier, when the tide was 10 feet higher, the churning water could have inundated their campsite, toppling trees and sweeping them away. "The water was chaotic for hours," Calvey recalled, describing powerful swells and immense whirlpools, a scene that prompted them to radio for rescue and end their ambitious journey.

The source of this cataclysmic wave was a massive landslide, occurring approximately 30 miles away at the terminus of Tracy Arm fjord, a narrow, winding channel nestled between imposing mountains south of Juneau. Scientists believe the slide was exacerbated by the ongoing retreat of the South Sawyer Glacier in that region, a phenomenon linked to glacial melt destabilizing adjacent slopes. This colossal landslide created a visible bedrock scar, stretching an astonishing 1,580 feet up the mountainside, a measure known as the tsunami’s run-up. This event stripped away trees, brush, and soil, marking the second-largest run-up ever recorded globally. As the tsunami propagated outwards from the slide, it scoured the steep shorelines of the 30-mile fjord, with its remnants reaching Calvey’s beach just beyond the fjord’s mouth.

Evidence suggests a disturbing increase in landslide-generated tsunamis across Alaska in recent years. In 2015, a landslide at Taan Fjord, located 400 miles northwest of Tracy Arm, generated a wave with a 630-foot run-up. Even more concerning, a creeping landslide near Barry Glacier in Prince William Sound poses a threat of a tsunami potentially exceeding the magnitude of the Tracy Arm event. Scientists, land managers, and tour operators are urgently seeking solutions to this escalating climate-related hazard, which threatens coastal communities, vital infrastructure, and the numerous vessels that navigate Alaskan waters, including the large cruise ships that bring nearly two million visitors to the state annually.

While researching her trip, Calvey had encountered warnings about tsunamis in Alaska’s fjords, but these were typically attributed to seismic activity, such as the infamous 1958 earthquake that triggered a landslide and a staggering 1,720-foot run-up in Lituya Bay, Glacier Bay National Park. The possibility of giant, tsunami-inducing landslides caused by retreating glaciers, however, was not widely documented. Calvey now hopes her experience will bring much-needed attention to this emerging threat.

The sheer scale of the Tracy Arm tsunami underscores the element of luck involved in Calvey’s survival, given the fjord’s popularity. Each summer, this picturesque waterway hosts over 200 visits from large cruise ships, their passengers eager to witness the dramatic spectacle of the Sawyer and South Sawyer glaciers meeting the ocean. Tracy Arm, situated within the Tongass National Forest, attracts an estimated 500,000 boat-based visitors each year. Though dozens of vessels were present in the area in the days preceding the landslide, the timing of the collapse—in the early morning hours—minimized the risk to large passenger vessels. The closest known vessel, the over-100-passenger National Geographic Venture, was in Tracy Arm but approximately 20 miles from South Sawyer when the mountain gave way, remaining unharmed due to its deep-water location. Had the landslide occurred just a few hours later, the Venture would have been closer to the glaciers, where passengers disembark onto skiffs for closer views. In that zone, waves exceeding 100 feet, laden with icebergs and debris, could have obliterated the tour boat and its smaller craft.

Christine and Jeffrey Smith, owners of the 65-foot tour boat David B, also experienced a fortunate turn of events. They had planned to anchor their vessel, carrying six guests, within five miles of South Sawyer the previous evening. The tsunami would have struck them minutes after the slide, but due to stormy weather, they opted to anchor 45 miles away in the neighboring Endicott Arm. "We’re definitely feeling our mortality," Christine Smith reflected, finding it difficult to comprehend surviving such a cataclysmic event. Even in Endicott Arm, the tsunami’s impact was felt; Smith witnessed 10-foot swells suddenly surge over a shoal, and an 80-foot vessel struggled against an unusually strong current. The Smiths’ immediate action was to send a satellite text to their geologist friend, Jackie Caplan-Auerbach, in Washington, urging her to check seismic stations for signs of a landslide, providing the first external report of the Tracy Arm tsunami.

Caplan-Auerbach, a seismologist and professor at Western Washington University, was enjoying her morning coffee when the Smiths’ urgent message arrived. She immediately began analyzing seismic data from stations near the David B. While much of this data is publicly accessible, the scarcity of seismometers in the vast and remote Southeast Alaska region presents a challenge for comprehensive monitoring. However, a sensor located in the Tlingit village of Angoon, about 60 miles west of the David B, revealed a telltale seismic signature indicating a nearby landslide. Further collaboration with colleagues quickly pinpointed the slide’s location to Tracy Arm. "This is a big deal," Caplan-Auerbach stated, recognizing that the immense volume of water displaced by the landslide was necessary to generate a tsunami that traveled 45 miles.

Intriguingly, Caplan-Auerbach also noted that the Angoon seismometer had registered subtle seismic signals for hours preceding the main landslide event. These "precursory" signals, though faint, were only recognized in hindsight, after the catastrophic slide occurred. Geologists later identified similar signals on other seismometers, all of which ceased abruptly with the mountainside’s collapse. Caplan-Auerbach has been studying these precursory signals for nearly two decades, exploring their potential for integration into future landslide warning systems. However, she cautioned that such systems are still a distant prospect, as landslides are notoriously difficult to predict due to complex factors like rock type, precipitation, and other variables. Furthermore, she noted, precursory signals do not precede every landslide event.

In October, Bretwood "Hig" Higman, an independent geologist working with his nonprofit Ground Truth Alaska, and the author paddled packrafts along the shores of Portage Lake, located about an hour southeast of Anchorage and 500 miles from Tracy Arm. Higman was investigating sensors installed on a slow-moving landslide above the Portage Glacier. Dense fog shrouded the lake as they approached the glacier, and the distant rumble of a rockslide somewhere high in the mist served as a potent reminder of the unstable mountainous terrain. Higman, characterized by his wire-rimmed glasses and worn ballcap, displayed a boyish curiosity towards the sound.

In May 2020, Higman and over a dozen other scientists had issued a public warning in The New York Times about a potentially catastrophic creeping landslide near the retreating Barry Glacier in Prince William Sound. This unconventional approach, while raising awareness, drew criticism from some locals who felt they should have received official notification rather than learning of the threat through national media coverage that caused pre-tourism season anxiety. The Barry Arm landslide is a significant concern; in summer, rocks cascade down its gravelly surface, kicking up dust and splashing into the ocean, while the nearby Barry Glacier calves ice with booming reports. Modeling suggests that a complete slope failure could unleash a massive tsunami, endangering boats, cabins, campsites, and their occupants across a wide area. Even a diminished but still powerful 6-foot wave could impact the small city of Whittier, located 30 miles away, posing a risk to those along its low-lying shores.

Shortly after the researchers’ open letter, state and federal agencies deployed a comprehensive array of high-tech instruments into Barry Arm fjord. These included weather stations, seismometers, live-feed cameras, radio repeaters, and ground-based radar, providing real-time observations of the unstable slope. An experimental tsunami warning system was also implemented, enabling scientists to notify Whittier officials or alert boaters via VHF radio. The research conducted using these instruments has significantly improved scientists’ ability to pinpoint more distant landslides, including the recent Tracy Arm event. This extensive project represents millions of dollars in state and federal investment.

As they paddled across Portage Lake, Higman explained that certain Alaskan glaciers, particularly those in retreat, pose significant hazards. The key ingredients are visibly unstable slopes situated above thinning ice, a body of water large enough to generate a tsunami, and proximity to human populations and infrastructure. Portage Lake exhibits all these characteristics. A Chugach National Forest visitor center on its shore attracts up to 1,000 visitors daily during the summer, many of whom board the 140-passenger Ptarmigan tour boat. A nearby road, railroad, campgrounds, and trails are all situated within a potential tsunami’s path.

They landed their packrafts on a narrow strip of shore and secured them to a cliff to deter curious bears. Shouldering their packs in the cool air, they hiked through dense alder brush and up steep, green tundra. Below them, the fog dissipated, revealing Portage Glacier, a colossal river of ice wedged between bulging mountains. Icebergs floated where its face met the lake, but along its edges, rubble-darkened ice had receded from the mountain walls, giving the glacier a weakened appearance. Higman explained this could be a manifestation of "debuttressing." Healthy glaciers naturally erode and sculpt the surrounding mountains, sometimes undercutting their walls, but they also provide lateral support.

As glaciers thin globally due to climate change, this buttressing effect diminishes, potentially leading to the collapse of steepened slopes. Climate change also contributes by thawing alpine permafrost and intensifying extreme rainfall events, both of which can trigger landslides related to debuttressing. However, the precise mechanisms that dictate whether and when a slope will fail remain elusive. Landslides do not always immediately follow glacial retreat, and sometimes they do not occur at all. While the Barry Glacier has receded from the base of its unstable slope over the past decade or more, the slope itself has not yet slid. "It’s frustrating that we don’t know more about how these systems work," Higman lamented, gazing down at the wasting glacier meeting Portage Lake.

As they ascended higher above the glacier, Higman pointed out thin cracks in the ground, visible evidence of the slope’s instability. These cracks had formed so recently that living heather still clung to their edges, its desiccated roots dangling in the air. More cracks appeared as they progressed, leading them to navigate around seemingly bottomless chasms several feet wide. Far below, Portage Glacier thundered as it shed icebergs into the lake. The tundra abruptly gave way to the landslide’s most active zone, a chaotic expanse of upturned rock and soil stretching as wide as a football field. Leaning spires, shattered boulders, and deep, gravelly sinkholes characterized the landscape. The occasional hard knock of tumbling rocks echoed, and Higman noted that the entire slope was slumping towards the glacier by several meters annually, attributing this movement to the thinning glacier’s removal of support.

Higman and other researchers expressed concern that the slope could collapse catastrophically into the lake in the coming years as the glacier continues to recede. This scenario mirrors what happened at Tracy Arm, where satellite imagery indicates the South Sawyer Glacier rapidly retreated from beneath the doomed mountainside in the weeks and days leading up to its collapse. Higman, Caplan-Auerbach, and other scientists, led by Dan Shugar of the University of Calgary, recently published research in the journal Science detailing how the glacier’s retreat primed the slope for the landslide and subsequent tsunami. However, Dennis Staley, a U.S. Geological Survey scientist leading the multi-agency effort at Barry Arm, acknowledged that multiple factors, including weather and inherent slope weakness, contribute to landslides. While the recent research highlights the connection between retreating glaciers and landslide-induced tsunamis, Staley stated, "It’s hard to know a specific trigger." Higman concurred, carefully noting that while many factors could have caused the Tracy Arm slide, the timing of the glacier’s rapid retreat presented an "extraordinary coincidence" that heightened concerns about the Portage area.

As they traversed the steep, barren slide, dirt and scree sloughed away beneath their boots. Navigating around wobbly boulders, Higman reached over the edge of a sheer column to retrieve an extensometer, a compact instrument he had previously drilled into the rock. This device, described by Higman as an "electronic tape measure," captures hourly measurements of the rock’s movement. Though still experimental, he hopes that its data will eventually be accessible online to visitor center staff, providing insights during periods of heavy rain or rapid glacial retreat. Ideally, Higman envisions a network of more precise instruments at Portage, similar to the array at Barry Arm, connected to satellites and a government office capable of triggering public alerts if movement accelerates. Such a system could also detect critical precursory signals. Higman highlighted the cost-effectiveness of his simpler, homegrown sensors, allowing for broader deployment, but acknowledged their limitations: they must be placed "at just the right crack" to effectively measure movement. For now, these remain the sole sensors at Portage.

They continued to scramble upward to the crown of the slide, reaching firm tundra. For an hour, they traversed the slope, visiting other sensors, each acting as an observational point for what Higman hoped were the crucial cracks.

On a drizzly evening a few weeks after the Tracy Arm tsunami, Staley drove to Whittier for a public meeting. Dozens of companies operating boat and kayak tours in Prince William Sound launch from this town. Captains, guides, officials, and local residents gathered aboard a spacious tour boat moored in the harbor, sharing pizza and discussing the situation. Staley provided an update on Barry Arm, followed by an open forum for questions and shared strategies for safely navigating near glaciers that support their businesses. Many operators had ceased visiting Barry Arm in 2020 following Higman’s initial warning, but some had resumed excursions after monitoring equipment was installed. However, the Tracy Arm event, coupled with a 2024 landslide and minor tsunami at Surprise Glacier near Barry Arm, led some operators to avoid the area again—a recurring pattern for those working near Alaska’s melting glaciers.

Jeff Pedersen, operations manager for Alaska Wildland Adventures in Girdwood, described this as an ongoing assessment. In 2024, his company sustained minor damage from a significant landslide near the retreating Pedersen Glacier in Kenai Fjords National Park, about 60 miles south of Portage. The landslide struck a lagoon during a night of heavy rain, generating a tsunami that reached heights of 50 feet or more on nearby slopes. Further from the slide, at the company’s backcountry lodge, a 3-foot wave overran boardwalks near a cabin where guests were sleeping. Staff reviewed their protocols and have since stopped visiting that particular glacier. Christine and Jeffrey Smith of the David B similarly reassessed their operations following the Tracy Arm slide, engaging in sober discussions about safety and revised anchoring practices. They intend to continue visiting glaciers but will adapt their procedures based on evolving conditions and, they hope, further research.

Major cruise lines, including Carnival, Holland America, and Royal Caribbean, have also taken note. In early April, nearly all major cruise companies operating in Alaska announced they would forgo visits to Tracy Arm in 2026, rerouting over 300 scheduled visits to the neighboring Endicott Arm, a similar fjord where the Dawes Glacier is receding from beneath steep mountainsides. None of the large cruise companies contacted for this story responded to interview requests.

Melenda Lekanof, secretary and treasurer of the Yakutat Tlingit Tribe Council, expressed deep concern. Yakutat, a village of 600 residents situated between Southeast Alaska and Prince William Sound, lies near the massive Hubbard Glacier. Lekanof recently coordinated a community workshop attended by Higman and other scientists, as part of the ‘Á;²utÍ Geohazards Project. This initiative combines Indigenous and Western scientific knowledge to enhance understanding of landslide risks. "We have a lot to learn about our changing landscape," Lekanof stated, referencing the increasing frequency of landslides and melting glaciers.

While cruise ships do not dock in Yakutat, they frequently pass the village en route to Hubbard Glacier in Disenchantment Bay. Scientific models suggest a potential landslide and destructive wave along this route. Lekanof worries that a lack of coordinated emergency planning among cruise ship companies, state agencies, and other entities would place the burden of disaster response on the small village. Furthermore, a cruise ship incident could result in loss of life or environmental damage to areas vital for her people’s fishing, hunting, and sacred lands. Lekanof also advocates for a monitoring array similar to the one at Barry Arm to be established in Disenchantment Bay, a project that would require expanded funding, funding that the Trump administration had previously sought to cut.

Traditional teachings about glaciers and tsunamis, Lekanof explained, embody millennia of ancestral knowledge. "We were taught to not be loud, we were taught to not damage the landscape, to be careful where you go," she recounted. She has, however, observed cruise ships blasting their horns near the glacier, an action she finds "spiritually disruptive and concerning for the birds and mammals there."

A month prior to the author’s visit to Portage Lake, the captain of the Ptarmigan tour boat photographed fresh debris at the base of the slow-moving landslide, in an area recently exposed by the receding Portage Glacier. She shared these images with visitor center staff, who consulted Higman. Subsequently, an intense multiday storm brought 8 inches of rain. Considering the escalating risks, the Forest Service evacuated a school group and staff from the visitor center, according to an agency statement. The Ptarmigan also canceled tours, citing landslide and tsunami hazards. This adaptive response may signal a new operational paradigm for navigating Alaska’s changing glacial environments.

While this adaptability offers a glimmer of hope, Higman pointed out that Tracy Arm raises critical questions about how scientists prioritize monitoring efforts. Visible slope movement, as observed at Portage and Barry Arm, is crucial, but so is the rate of glacial retreat. Higman and his colleagues have gathered preliminary data suggesting a correlation between the length of time a newly exposed slope was covered by a glacier and its likelihood of sliding. They believe that Endicott Arm and LeConte Bay, two busy fjords south of Tracy Arm, warrant further investigation. After all, the Tracy Arm mountain appeared stable, lacking the overt warning signs present at Portage, and crucially, had no sensors. "As scientists, we need to own up to having totally missed the ball on that one," Higman concluded.