Deadly landslide in Washington state wasn’t so predictable, scientists find

“The slope that failed at Oso on March 22, 2014,” USGS says, “had a long history of prior historical landslides at the site, but these had not exhibited exceptional mobility” of the Oso slide, which was “unusually mobile and destructive.”

Nobody happened to be making a home video or flying a drone in the hills above Oso, Washington, when the ground gave way a year ago Sunday, burying 40 structures and killing 43 people.

But if they had, they probably couldn’t have captured the scale and pace of the landslide with any greater clarity than the video clip below.


What you see there is a newly published simulation of the landslide prepared by the U.S. Geological Survey in the course of its yearlong inquiry into what is widely (if erroneously) described as the deadliest landslide in our national history (and more about that ranking later).

The narrative USGS has developed of the events of March 22, 2014, based on what it terms the first published study of the disaster, goes like this:

After a 45-day period of heavy rain – 150 to 200 percent of normal in a western Washington landscape that’s pretty rainy to begin with – a somewhat slide-prone section of bluff along the north fork of the Stillaguamish River gave way. It was a little after 10:30 on a Saturday morning.

Although the slope was relatively gentle, the bluff rose nearly 600 feet above the riverbank, and some 18 million tons of mud and rock began to slide at speeds that soon reached 40 miles an hour, maybe more.

Within a minute, the flow had crossed the river and traveled more than half a mile toward a neighborhood called Steelhead Haven. Within five and a half minutes, the interval covered in the clip above, it had fanned out to cover half a square mile with an estimated 8 million cubic meters of former bluff.

How much mud is that? According to USGS, enough to bury 600 football fields about 10 feet deep, but that’s on level terrain.

Along the Stillaguamish, the depths reached 75 feet in places, damming the river and backing up water to create a temporary lake that was two and a half miles long, and in places 25 feet deep. Mud also covered a mile or so of the only highway in and out of Oso, complicating rescue efforts.

Eventually, eight residents of Steelhead Haven were saved.

The fault-finding reflex

In the days after the Oso landslide – its official name, bestowed by Washington State, is the SR530 Landslide, after the buried highway – much was written about previous slides and instability in the area, about the folly of people living in such a place, the failures of public agencies to protect them.

An oft-cited geomorphology report, prepared for the Army Corps of Engineers in 1999 and intended to guide land-use planning in that part of Snohomish County, refers to a history of landslides in the area, some of them both large and recent. It was cited, for example, by The New York Times’ Timothy Egan, who wrote one week after the disaster:

Don’t tell me, please, that nobody saw one of  the deadliest landslides in American history coming. … Enough with the denial, the willful ignorance of cause and effect, the shock that one of the prettiest valleys on the planet could turn in a flash from a quiet respite in the foothills of the North Cascades to a gravelly graveyard.

The scientists who prepared the new USGS study were of course aware of that history of slide activity, but their central conclusion is quite different: that such slides remain unpredictable, and “subtly different” conditions a year ago might have meant no disaster at all.

“The slope that failed at Oso on March 22, 2014,” USGS says, “had a long history of prior historical landslides at the site, but these had not exhibited exceptional mobility” of the Oso slide, which was “unusually mobile and destructive.”

Indeed, the authors found – and demonstrate in a two-part simulation from which the clip above is taken – that if the soil composition had been slightly different, maybe a little drier or coarser, or the rainfall slightly less, the events of last March 22 might have resulted in a fairly minor mudslide and far less destruction.

Eyewitness accounts and seismic energy radiated by the landslide indicate that slope failure occurred in two stages over the course of about 1 minute. During the second stage of slope failure, the landslide greatly accelerated, crossed the North Fork Stillaguamish River, and mobilized to form a high-speed debris avalanche.

The leading edge of the wet debris avalanche probably acquired additional water as it crossed the North Fork Stillaguamish River. It transformed into a water-saturated debris flow (a fully liquefied slurry of quicksand-like material) that entrained and transported virtually all objects in its path.

Field evidence and mathematical modeling indicate that the high mobility of the debris avalanche was caused by liquefaction at the base of the slide caused by pressures generated by the landslide itself. The physics of landslide liquefaction has been studied experimentally and is well understood, but the complex nature of natural geological materials complicates efforts to predict which landslides will liquefy and become highly mobile.

The limits of prediction

I suppose it may be good news for the various planning and regulatory agencies being sued in the aftermath of Oso that USGS has found this disaster beyond our current forecasting capabilities.

On the other hand, it also means the next big landslide catastrophe is also unpredictable.  But they’re working on that.

Although the anniversary coverage this week typically refers to the burial of Steelhead Haven as the deadliest landslide in U.S. history, USGS awards that title – indeed, deadliest in all of North American history – to a slide in Puerto Rico, not so terribly long ago.

It happened in 1985, following heavy rainfall associated with a tropical storm system, and is credited with “killing at least 130 people in the Mamaeyes neighborhood of barrio Portugués Urbano in Ponce.” The landslide followed flooding driven by a tropical storm.

And you may also be wondering: Well, what about all those landslides after the Alaskan earthquakes of 1964, which killed 139 people?

Classifying landslides

Record-keepers typically place landslides like Oso, which result from rain (and gravity) causing subsurface structural failures, in a different category from those associated with earthquakes or volcanic eruptions (like the ones that followed the eruption of Mount St. Helens in 1980, contributing to an overall death toll of 57.).

From a quick summary of other deadly U.S. landslides, not associated with quakes or volcanoes, prepared for the Weather Channel shortly after Oso by weather historian Christopher Burt, who I guess was forgetting for a moment that Puerto Rico is part of the U.S., too:

  • Largest by volume: The landslide at a Kennecott copper mine in Bingham Canyon, Utah, near Salt Lake City in April 2013 is considered the biggest in “modern U.S. history,” with a slide mass of 55 million cubic meters, or seven times what USGS now attributes to the Oso slide. No one was killed or injured, though.
  • Costliest in terms of property damage: A slide that wiped out the town of Thistle, Utah, in April 1983 with a slide mass of about 15 million cubic meters, which formed a dam and in turn a 160-foot-deep lake where the town had been. Losses were estimated at $200 million to $400 million in 1983 dollars.
  • Deadliest in the U.S. before Oso, in modern times, a fairly small slide that killed 10 people in La Conchita, California, in January 2005 with the movement of just 200,000 cubic meters in a densely populated neighborhood.

But here, too, there is sometimes disagreement about whether to count a series of related slides as separate or single events, as Burt observed:

It is difficult to state what the deadliest “landslide” in U.S. history has been since some were a combination of factors (sometimes of human origin like the collapse of the St. Francis Dam in California in 1928 that resulted in 500 deaths) or a series of mud and debris flows over a wide area as occurred in the Santa Cruz Mountains of California in 1982 killing 30 and in Nelson County, Virginia, in 1969 when the remnants of Hurricane Camille dropped 27” of rain resulting in mudslides that killed 153.

According to the USGS, landslides or mudslides/flows kill 25 in the U.S. each year.

You can also learn about all our free newsletter options.

Comments (1)

  1. Submitted by Neal Rovick on 03/24/2015 - 10:10 am.

    “Have it your way.”The

    “Have it your way.”

    The modern world is all about providing the illusion of control and the ability to direct outcomes.

    The processes of nature and environment and the interrelationship of action and consequence are not well enough understood.

    We know the science that excess rain increases the likelihood of slides, but we don’t know exactly when or where the first failure will occur or whether that failure will propagate into a disaster.

    It reminds of another issue where we know the science, but we don’t know the exact point where the failure will turn into a disaster.

Leave a Reply