The March 2014 landslide in Oso, Washington, about 55 miles northeast of Seattle, became the deadliest landslide event in United States history. Forty-three people died and 49 homes and structures were destroyed.

A 91Ě˝»¨ engineer who analyzed the event’s aftermath began to investigate the circumstances that can make landslides so deadly. The resulting study shows that certain human actions increase the chance of surviving a devastating event, and suggests simple behavioral changes could save more lives than expensive engineering solutions.

The open-access , published in the October issue of GeoHealth, suggests key actions that range from opening doors and windows to continuing to move and make noise if you do get buried.

“There are in fact some really simple, cost-effective measures that can be taken that can dramatically improve the likelihood that one will survive a landslide,” said senior author , a 91Ě˝»¨professor of civil and environmental engineering.

Worldwide, landslides cause on average more than 4,000 deaths a year recently, with about 25 to 50 of those deaths occurring each year in the U.S. These events may become more frequent as wildfires fueled by warmer temperatures can leave slopes bare and more vulnerable to slides.

Wartman and a 91Ě˝»¨graduate student compiled and analyzed records of 38 landslides that affected occupied buildings. Most of the data came from the U.S., but it included landslides from around the world for which there were detailed records.

The authors recorded the geologic details of each landslide, as well as the reports from survivors of the events. They used newspaper articles, scientific papers, medical examiner reports and other documents to produce a detailed catalog of fatalities caused by landslides hitting occupied buildings. The events, spanning from 1881 to 2019, included the Oso mudslide and the , as well as events in Bangladesh, the Philippines, China, Malaysia, Australia and New Zealand.

Their analysis showed behavioral factors, such as a having an awareness of local landslide hazards and moving to a higher floor of a building during an event, had the strongest association with survival.

“Simply by being on an upper floor, an individual can increase their odds of survival by up to a factor of twelve. This is a powerful finding that we need to consider when we design the layout and vertical access routes in homes,” said first author , who did the work for his 91Ě˝»¨doctorate in civil and environmental engineering and is now a lecturer in the department.

The researchers found some behaviors, despite being performed by only a small number of people, often save lives. ĚýAccording to their results, those actions are:

Before an event

• Be informed about potential hazards, from hazard maps or other sources
• Talk to people who have experienced these events
• Move areas of high occupancy, such as bedrooms, upstairs or to the downhill side of a building

During an event

• Move away from the threat — don’t approach an active landslide
• Escape vertically by moving upstairs or even on countertops to avoid being swept away
• Identify and relocate to interior, ideally unfurnished, areas of a building that offer more protection
• Open downhill doors and windows to let debris escape

After an event

• If caught in landslide debris, continue to move and make noise to alert rescuers

Many things the authors predicted would be important, including the size or the intensity of landslide events, made little difference to the death toll for landslides below about 20 feet depth. Similarly, the distance between a building and the landslide slope, or an inhabitant’s age and gender, didn’t make a big difference to their survival.

The results suggest practical ways to lower the number of lives lost to landslides in the United States, Wartman said. He hopes the information can be incorporated in education and community awareness programs.

“This is a message of hope,” Wartman said. “What this work suggests is that a modest investment put toward social science, policy and education could have a very marked effect in protecting people from landslides.”

Residents who want to know if they are vulnerable to landslides can contact a local agency, such as the Washington State Department of Natural Resources, to learn more about local risks. Federal is pending to make this information more easily accessible across the United States, Wartman said.

The study was funded by the National Science Foundation.

For more information, contact Wartman at wartman@uw.edu or Pollock at wpollock@uw.edu.

Adapted from an by Jack Lee for AGU Eos.

Researchers at the 91Ě˝»¨, Portland State University and the University of Oregon have shown that deep-seated landslides in the central Oregon Coast Range are triggered mostly by rainfall, not by large offshore earthquakes.

The open-access was published Sept. 16 in Science Advances.

“Geomorphologists have long understood the importance of rainfall in triggering landslides, and our study is simply driving home just how important it is,” said first author , who did the work as part of his doctorate at the UW. “Our results show that more frequent, localized landslide events triggered by rainfall are just as important to consider as less frequent but more far-reaching Cascadia Subduction Zone earthquakes.”

Heavy rains are known to cause landslides that can be disruptive and deadly. A less frequent landslide trigger is a rupture on the geologic fault off the coast of Washington and Oregon that’s known as the Cascadia Subduction Zone — among a long list of concerns after a major earthquake. Landslide risks of all types increase if human development or wildfires remove trees, taking away the roots that stabilize the soil.

Recent research in Nepal and Japan, however, suggests that offshore earthquakes might not trigger as many landslides as previously believed. The new study finds a similar situation in the Pacific Northwest.

“We aren’t suggesting that the landscape had no response to these magnitude-9 earthquakes, but that the deeper-seated landslide deposits and scars out on the Oregon Coast Range hillslopes today were primarily triggered by precipitation events,” said senior author , a 91Ě˝»¨associate professor of Earth and space sciences. “We conclude that past Cascadia Subduction Zone earthquakes triggered no more than a few hundred deep landslides during great earthquake events.”

The researchers used high-resolution aerial laser maps of the Oregon coast to look at 1,000 years of landslide activity. Landslides tended to happen in places with heavier rainfall, they found. But surprisingly, there was no detectable change in the number of deep landslides at the time of the large earthquake that shook the Pacific Northwest in 1700, or for two earlier offshore earthquakes that happened in roughly the years 1150 and 1470.

Duvall, LaHusen and co-author at Portland State University developed a method for dating landslides while studying the site of the deadly March 2014 mudslide in Oso, Washington. In that study, they used high-resolution images to view the surface roughness. Over time, soil settles and exposed rock erodes. The surface gets smoother, so surface roughness can be used to calculate a landslide’s age.

“The central Oregon Coast Range offered a massive, 10,000-square-kilometer natural laboratory to explore patterns in deep-seated landslide events through space and time,” LaHusen said. “It’s 50-million-year-old sandstone and siltstone that was deposited offshore, buried and compacted, and then uplifted to form the mountains we see today.”

Aerial with less than 3-foot resolution revealed 9,938 landslides inside the study area. Researchers narrowed those down to 2,676 landslides that have happened within the past 1,000 years, and then looked at the landslide frequency during that time.

For the new study, they applied their method to a larger area in the central Oregon Coast Range. To study landslide activity related to the Cascadia Subduction Zone, the researchers needed an area near the Cascadia fault zone with a consistent rock type and publicly available lidar imagery.

Researchers caution that the study doesn’t apply to shallow landslides, which frequently occur during earthquakes but leave no long-term evidence and can’t be analyzed with this method, or to different soil types, and so doesn’t necessarily apply to other regions. The research also didn’t consider shallower earthquakes from surface faults.

But the paper does support recent findings in Asia suggesting that offshore earthquakes don’t trigger as many deep landslides as once believed, and that rainfall may be the bigger factor in shaping the landscape over longer timescales.

“These data strengthen the point that we don’t need big earthquakes to trigger large and devastating landslides in Washington and Oregon,” Duvall said. “Seasonal precipitation and large rain events are important to focus on in landslide preparedness planning.”

This research was funded by the National Science Foundation and the Geological Society of America. The team began the work as part of the 91Ě˝»¨, which is studying magnitude-9 earthquakes from the fault that runs parallel to the Washington and Oregon coastlines. Slips along this fault can lead to a so-called “Big One,” which last struck the Pacific Northwest in 1700.

Other co-authors are , , and at the UW; and and at the University of Oregon. LaHusen is now working at the U.S. Geological Survey in Mountain View, California.

For more information, contact LaHusen at seanlah@gmail.com or Duvall at aduvall@uw.edu.

91Ě˝»¨ geologists analyzed woody debris buried in earlier slides and used radiocarbon dating to map the history of activity at the site. The , published online Dec. 22 in the journal , show that a massive nearby slide happened around 500 years ago, and not thousands of years ago as some had believed.

“The soil in this area is all glacial material, so one hypothesis is the material could have fallen apart in a series of large landslides soon after the ice retreated, thousands of years ago,” said corresponding author , a 91Ě˝»¨doctoral student in Earth and space sciences. “We found that that’s not the case — in fact, landslides have been continuing in recent history.”

The study establishes a new method to date all the previous landslides at a particular location. The method shows that the slopes in the area around Oso have collapsed on average once every 500 years, and at a higher rate of about once every 140 years over the past 2,000 years.

“This was well known as an area of hillslope instability, but the question was: ‘Were the larger slides thousands of years old, or hundreds of years old?’ Now we can say that many of them are hundreds of years old,” said co-author , a 91Ě˝»¨assistant professor of Earth and space sciences.

LaHusen had not yet begun his graduate studies when he asked about studying the history of geologic activity at the Oso site. In late summer of 2014, the researchers began their work wading along riverbanks to look for preserved branches or trees that could be used to date previous landslides.

“When you have a large, catastrophic landslide, it can often uproot living trees which kills them and also encapsulates them in the landslide mass,” Duvall said. “If you can find them in the landslide mass, you can assume that they were killed by the landslide, and thus you can date when the landslide occurred.”

The team managed to unearth samples of wood buried in the Rowan landslide, just downstream of the Oso site, and the Headache Creek landslide, just upriver of the 2014 slide. Results from several debris samples show that the Rowan landslide, approximately five times the size of the Oso slide, took place just 300 to 694 years ago. The Headache Creek landslide is within a couple hundred years of 6,000 years old.

Previous 91Ě˝»¨ had shown a history of geologic activity at the Oso site, including previous major landslides and a recent small slide at the same slope that collapsed in 2014. But while the position of past slides and degree of surface erosion can show the order that the older slides happened, it has not been possible to give a date for the past events.

The new study uses the radiocarbon dates for two slides to establish a roughness curve to date other events along a 3.7-mile (6-kilometer) stretch of the north fork of the Stillaguamish River. A roughness curve uses the amount of surface erosion to establish each slide’s age. The two dates put firm limits on the curve, so that other nearby slides can be dated from their roughness characteristics without having to find material buried inside each mass of soil.

“This is the first time this calibrated surface dating method has been used for landslide chronologies, and it seems to work really well,” LaHusen said. “It can provide some information about how often these events recur, which is the first step toward a regional risk analysis.”

Applying the new method for other locations would require gathering samples for each area, they cautioned, because each site has its own soil composition and erosion characteristics.

It’s not known whether the findings for the Oso site’s history would apply to other parts of the Stillaguamish River, Duvall said, or to other places in Washington state. The researchers are still studying debris from other locations. But the results do have implications for the immediate area.

“It suggests that the Oso landslide was not so much of an anomaly,” Duvall said.

She and LaHusen are also working with the UW’s , which is studying hazards from magnitude-9 earthquakes along the Cascadia subduction zone. They would like to learn whether landslides across Washington state coincided with past earthquakes, and use simulations of future shaking to predict which places in the state are most vulnerable to earthquake-triggered landslides.

Other co-authors on the new paper are at the 91Ě˝»¨and at Portland State University. The study was funded by the National Science Foundation, the Geological Society of America, and the UW’s Quaternary Research Center.

###

For more information, contact Duvall at 650-815-5560 or aduvall@uw.edu and LaHusen at seanlah7@uw.edu. Note: Duvall and LaHusen are currently out of town and best reached via email. Duvall is on the East Coast and will be on campus Sunday, Dec. 27, and then doing fieldwork in New Zealand from Dec. 28 to Jan. 16. LaHusen will be back on campus Jan. 4.

More photos are posted at .

NSF grant: EAR-1331412

The research indicates the landslide, the deadliest in U.S. history, happened in two major stages. The first stage remobilized the 2006 slide, including part of an adjacent forested slope from an ancient slide, and was made up largely or entirely of deposits from previous landslides. The first stage ultimately moved more than six-tenths of a mile across the north fork of the Stillaguamish River and caused nearly all the destruction in the Steelhead Haven neighborhood.

The second stage started several minutes later and consisted of ancient landslide and glacial deposits. That material moved into the space vacated by the first stage and moved rapidly until it reached the trailing edge of the first stage, the study found.

The report, released Tuesday on the four-month anniversary of the slide, details an investigation by a team from the , or GEER. The scientists and engineers determined that intense rainfall in the three weeks before the slide likely was a major issue, but factors such as altered groundwater migration, weakened soil consistency because of previous landslides and changes in hillside stresses played key roles.

The extreme events group is funded by the National Science Foundation, and its goal is to collect perishable data immediately in the wake of extreme events such as earthquakes, hurricanes, tsunamis, landslides or floods. Recent events for which reports have been filed include earthquakes in New Zealand and Haiti, the 2011 earthquake and tsunami in Japan, and Hurricane Sandy on the U.S. Eastern Seaboard in 2012.

“Perhaps the most striking finding is that, while the Oso landslide was a rare geologic occurrence, it was not extraordinary,” said , a 91Ě˝»¨ associate professor of civil and environmental engineering and a team leader for the study.

“We observed several other older but very similar long-runout landslides in the surrounding Stillaguamish River Valley. This tells us these may be prevalent in this setting over long time frames. Even the apparent trigger of the event – several weeks of intense rainfall – was not truly exceptional for the region,” Wartman said.

Team co-leader Jeffrey Keaton, a principal engineering geologist with AMEC Americas, an engineering consultant and project management company, said another important finding is that spring of 2014 was not a big time for landslides in Northwest Washington.

“The Oso landslide was the only major one that occurred in Snohomish County or the Seattle area this spring,” Keaton said.

Other team members are Scott Anderson of the Federal Highway Administration, Jean Benoit of the University of New Hampshire, John deLaChapelle of Golder Associates Inc., Robert Gilbert of the University of Texas and David Montgomery of the 91Ě˝»¨.

The team was formed and approved within days of the landslide, but it began work at the site about eight weeks later, after search and recovery activities were largely completed. The researchers documented conditions and collected data that could be lost over time. Their report is based largely on data collected during a four-day study of the entire landslide area in late May. It focuses on data and observations directly from the site, but also considers information such as local geologic and climate conditions and eyewitness accounts.

The researchers reviewed evidence for a number of large landslides in the Stillaguamish Valley around Oso during the previous 6,000 years, many of them strongly resembling the site of the 2014 slide. There is solid evidence, for example, of a slide just west of this year’s slide that also ran out across the valley. In addition, they reviewed published maps showing the entire valley bottom in the Oso area is made up of old landslide deposits or areas where such deposits have been reworked by the river and left on the flood plain.

The team estimated that large landslides such as the March event have happened in the same area as often as every 400 years (based on 15 mapped large landslides) to every 1,500 years (based on carbon dating of what appears to be the oldest of four generations of large slides) during the last six millennia.

The researchers found that the size of the landslide area grew slowly starting in the 1930s until 2006, when it increased dramatically. That was followed by this year’s catastrophically larger slide.

Studies in previous decades indicated a high landslide risk for the Oso area, the researchers found, but they noted that it does not appear there was any publicly communicated understanding that debris from a landslide could run as far across the valley as it did in March. In addition to the fatalities, that event seriously injured at least 10 people and caused damage estimated at more than $50 million.

“For me, the most important finding is that we must think about landslides in the context of ‘risk’ rather than ‘hazard,'” Wartman said. “While these terms are often used interchangeably, there is a subtle but important difference. Landslide hazard, which was well known in the region, tells us the likelihood that a landslide will occur, whereas landslide risk tells us something far more important – the likelihood that human losses will occur as a result of a landslide.

“From a policy perspective, I think it is very important that we begin to assess and clearly communicate the risks from landslides,” he said.

Other study conclusions include:

• That past landslides and associated debris deposited by water should be carefully investigated when mapping areas for zoning purposes.
• That the influence of precipitation on destabilizing a slope should consider both cumulative amounts and short-duration intensities in assessing the likelihood of initial or renewed slope movement.
• That methods to identify and delineate potential landslide runout zones need to be revisited and re-evaluated.

###

For more information, contact Wartman at 206-685-4806 or wartman@uw.edu, or Keaton at 323-889-5316 or jeff.keaton@amec.com.

“You guys have a plane, we have a radar, is there anything we can do for the landslide in Washington state?” was the question by Yuly Margulis, the president of New York-based company

, an engineer at the 91Ě˝»¨Applied Physics Laboratory, normally does aerial imaging over the oceans. But he consulted with 91Ě˝»¨colleagues who are geologists, obtained some discretionary funding from the APL, and obtained permission to fly a survey over the landslide site.

Farquharson and , an oceanographer at the APL, conducted two one-hour surveys, on Thursday, March 27, and Monday, March 31. They crisscrossed the site to take photos and then used image processing software to create a composite image.

“With thermal imaging we can tell where water is, so we can see the groundwater and saturation level, and measure the extent of the new river channel, and we can also measure the impounded water behind the slide,” Chickadel said. “It’s a way to map the changing nature of the slide as it’s going to evolve over the next weeks and months.”

Researchers are now processing their data, which will be made available to the public. They hope the two datasets could offer clues to the slide’s cause and evolution.

“It is critical to image the slide as soon as possible, because the mass will likely undergo rapid changes owing to continued movement, river action and seepage in the massive landslide deposition area,” said , a 91Ě˝»¨professor of civil and environmental engineering.

Wartman is co-leading a of the landslide that will incorporate data from the 91Ě˝»¨aerial surveys.

The team uses a Cessna 172 plane that is owned and operated by Regal Air, a flight school and charter company based at Everett’s Paine Field. Farquharson sat in the co-pilot’s seat to operate the equipment, which takes up the back half of the body of the plane.

Their equipment included an Artemis to measure topography, similar to one used by the 91Ě˝»¨group for imaging of coastal circulation and currents near river mouths. Cameras recorded visible light and thermal infrared wavelengths to map water inundation, groundwater seepage and slide extent.

Detection of movement in the soils near Oso could indicate areas of the slide that are still unstable. Data from immediately after the slide might show clues to what caused the slope to fail.

The Washington State Department of Natural Resources conducted similar aerial surveys using lidar technology to make precise topographic maps of rock and soil. Lidar is more sensitive than radar, but can have trouble with trees or in poor weather, Farquharson said, so the two datasets will be complementary.

The 91Ě˝»¨system is designed to be economical. The equipment is lightweight and the radar has a wide-angle view, both of which allow it to operate on small, low-flying planes. Researchers hope to explore the use of this system in taking more regular surveys of landslide risk areas, to monitor changes in groundwater and study changes in soils.

“I could see this technology might be useful for regular monitoring of landslide prone slopes,” Farquharson said.

###

For more information, contact Farquharson at 206-685-1505 or gordon@apl.washington.edu or Chickadel at 206-221-7673 or chickadel@apl.washington.edu.

The is mobilizing to collect information about the landslide that occurred on a steep slope above the North Fork of the Stillaguamish River near Oso, Wash., more than a week ago, killing more than 20 people with 30 still missing.

The group will try to understand why the slope collapsed in the hope that a similar disaster can be prevented, said , a 91Ě˝»¨associate professor of civil and environmental engineering who is leading the investigation with Jeffrey Keaton, principal engineering geologist at AMEC Americas.

“The purpose is to collect data before the site is changed or altered in rescue and recovery. There is a lot of valuable information about how that landslide occurred in the landscape itself,” Wartman said. “Ultimately, we want to learn from these disasters so we can prevent reoccurrence of future catastrophes.”

Local team members, including Wartman and 91Ě˝»¨geomorphologist , are hoping to visit the landslide site this week, followed by a visit next week from the entire reconnaissance team of six experts from universities, government agencies and industry.

The Geotechnical Extreme Events Reconnaissance Association is a National Science Foundation-funded group of experts that responds quickly when a geologic disaster happens. The association tries to collect technical data within days of a disaster to inform future long-term investigations. Some of its disaster investigations include the Colorado floods in 2013, Hurricane Sandy in 2012 and recent catastrophic earthquakes in Chile, Haiti, Japan and New Zealand.

The investigators aim to piece together what happened in the Western Washington disaster by collecting field measurements and gathering details from people who witnessed the landslide. It’s not intended to be a long-term investigation, but rather a quick-turnaround initiative that could help guide future investigations, Wartman said.

The team plans to document the mudslide by taking photos, talking with witnesses, measuring parts of the landslide and looking at satellite imagery from various points on the landscape.

Data collection will be quick, and the team plans to post its observations and findings within a month on the . The group appears to be the only one currently collecting technical data on the mudslide, though other experts have been involved in making sure the area is safe for rescuers.

The team will not be involved with rescue and recovery work and hopes to strike a balance between not interfering in those efforts while making sure to document the disaster before evidence is lost, Wartman said.

“We will be collecting perishable, important technical data that will soon disappear,” he said. “We hope to directly observe and record the effects of this landslide on the community. As a nation, we don’t have a lot of information about the human and capital losses of a landslide disaster.”

###

For more information, contact Wartman at wartman@uw.edu or 206-685-4806.