A large earthquake on Mount Hood several hundred years ago may have triggered a massive landslide in the Columbia River Gorge that temporarily dammed the Columbia River and created a miles-long lake that killed a forest of trees, new research suggests.
Two Northwest scientists presented their findings last fall during the Geological Society of America conference held in Portland.
Portland-based U.S. Geological Survey research geologist Scott Bennett had studied the earthquake, and Patrick Pringle, professor emeritus at Centralia College in Washington, had studied trees that had died in the landslide or drowned in the lake created by it. Bennett had determined the earthquake had happened within the same time frame as the landslide, and Pringle had narrowed down the date of the landslide to a single year less than 600 years ago. During his presentation, Bennett posed the question: Could the earthquake have been the trigger for the landslide?
The landslide dammed up the Columbia River at what is now Cascade Locks, and today, the “Bridge of the Gods” cantilever toll bridge spans the river at the same location, connecting Oregon and Washington.
The event was so significant that legends formed around it, and Native Americans told of stories of being able to walk from one side of the river to the other without getting their feet wet.
The new research further expands our understanding of the Columbia River Gorge’s violent history that has left it the idyllic place that it is today.
Dig Provides Solid Evidence of Big Quake, Timing
Scientists have had evidence for a massive landslide at Cascade Locks for years. There have also been several hypotheses to what triggered the more than billion dump-truck loads of earth that broke off from Table Mountain on the Washington side of the gorge and tumbled into the river.
Learn More About the Landslide in this Video:
One of the prevailing hypotheses is that a megathrust earthquake on the Cascadia Subduction Zone off the coast could have caused enough shaking in the gorge to trigger the landslide. Another is simply that the area became so waterlogged that eventually gravity pulled it apart.
But Bennett’s research suggests a strong earthquake on Mount Hood could have been the trigger.
In September 2020, Bennett, a couple of colleagues, and several graduate students at Portland State University set out on a hike deep into the Mark O. Hatfield Wilderness Area on the north side of Mount Hood on a mission to investigate a big earthquake on the mountain that had occurred sometime in the past.
Among the trees they came upon a fault scarp, which is a surface manifestation of a large earthquake. It looks like a small cliff or “step” in the landscape, but over time it has eroded, leaving a gentle slope snaking across the land for miles.
In 2018, Ian Madin, a geologist at the Oregon Department of Geology and Mineral Industries, along with Ashley Streig, a professor at PSU, announced the existence of four major crustal faults on the mountain in what became known as the Mount Hood Fault Zone. They identified them through LIDAR images -- a laser-mapping technology used from airplanes.
Bennett had used those images to guide his group to one of the scarps. And now at about 4,000 feet in elevation near the crest of the Cascades, they came to what he would later describe to colleagues as an “absolutely geo-morphically gorgeous fault scarp running through the forest” that was “immaculately preserved.”
Indeed, a gently sloping hill, maybe four to five feet high snaked through the forest for miles on what scientists named the Gate Creek Fault.
With picks and shovels, Bennett and his crew of graduate students dug a trench up to six feet deep across the scarp. And within that trench they could see where the earth had fractured and dropped from the energy of the earthquake.
Measuring how far the earth had dropped, or been displaced, and using estimates of how long the scarp, or rupture, traveled across the surface of the earth, Bennett and other scientists determined the earthquake that occurred there was somewhere between magnitudes 6.5 and 7.
And from the trench they extracted bits of charcoal they would send to a lab for analysis. Using those lab results, they determined the earthquake on the Gate Creek Fault happened about 530 to 660 years ago.
Which was interesting, because a little over 10 miles to the west was the site of the great Bonneville landslide that scientists had determined happened during the same range of time.
Dating the Bonneville Slide
When the earth broke apart from Table Mountain and slid into the Columbia River, it formed a nearly 250-foot-high dam that temporarily blocked the river. The water behind the dam created a 260-foot-deep lake that stretched for about 150 miles, submerging a forest of trees, drowning them. Scientists think it wasn’t long before the water overtopped or burst through the dam.
Lewis and Clark noted the existence of the “submerged forest” as they made their way down the Columbia River in 1805 during their journey to the Pacific Ocean. And starting in the 1930s, before the building of the Bonneville Dam, botanist Donald Lawrence extensively studied and mapped the snags of the submerged forest, and by radiocarbon dating his samples, reported a landslide date sometime between 742 to 772 years ago. Other scientists provided other wide-ranging dates.
Today, scientists have used some of Lawrence’s tree samples to re-analyze and narrow down when the landslide happened.
A group of them, working independently of Bennett, published a paper this year in “Quaternary Research,” placing the age of the landslide sometime between A.D. 1421 to 1455, or 567 to 601 years ago, which is within the time frame of the earthquake Bennett studied.
The scientists used three trees in their analysis. One tree, entombed in landslide debris, was dug up in the 1970s when the U.S. Army Corps of Engineers began work on the foundation for a second powerhouse at the Bonneville Dam. The other two trees were collected by Lawrence from the submerged forest near Wyeth and Perham creeks upstream of the landslide site along the Columbia River. They were drowned in the rising water caused by the landslide.
Using the science of dendrochronology (the analysis of tree rings to determine ages of trees and the interpretation of past events) the scientists found matching patterns in the rings of each tree, which allowed them to determine the trees died during the same year. Then they radiocarbon dated the samples to determine their range of dates, 1421 to 1455.
But last October, Pringle, who was one of the authors of the paper, presented results of further analysis of the tree samples during the geological conference in Portland that narrowed the date of the landslide to a single year, and even more precise than that -- late October 1446 to April 1447.
In an email, he said he accomplished such precision dating by matching the tree rings of the samples to “master chronologies” from trees in Washington and Oregon and from an old growth tree sampled a few miles northwest of the landslide site. He also cross-dated the “victim” trees of the landslide to trees that died in what is known as the Electron Mudflow, about 100 miles to the north that originated on Mount Rainier.
In his email, Pringle said he was working on a new paper to report the more precise date of the landslide.
Causation vs. Correlation
Whether the earthquake caused the landslide highlights two important concepts in science: causation and correlation. At this point, the scientists haven’t determined if the earthquake caused the landslide, but they have correlated the two events in time. But correlation does not mean there is a causal link between the events.
“We know all these things happened,” said Bennett. “We know the earthquake happened, we know the landslide happened, we know the flood happened, so if they don’t actually are related to each other, there was just a period of time here, maybe a few decades, or less than a century, where there’s a lot of really significant natural hazards happening in a short period of time.”
He's dubbed that period as the “decades of terror.”
The Makings of the Bonneville Landslide and Potential Trigger
The ingredients to create a landslide are readily found in the Columbia River Gorge, especially at the site of the Bonneville landslide, which is part of a larger area called the Cascade Landslide Complex.
It rains a lot in the Northwest, and fractured basalt within and around the complex sits atop sedimentary rock, which has turned to clay from water seeping through the cracks in the basalt. Additionally, the hills and mountains on the Washington side of the gorge are slightly tilted toward Oregon.
“The Bonneville landslide is kind of the youngest and potentially biggest of a very long history of landslides that are shedding off the Washington side into the Columbia River,” Bennett said. “And so, with that in mind, you can imagine the moment before the Bonneville landslide occurred, the material downhill of Table Mountain, and even the material up in the headlands was probably weakened due to previous fracturing and failure due to older landslides.”
The setup, then, was ripe for a massive landslide, triggered, perhaps, by a big earthquake.
Alex Grant, a research civil engineer for USGS based in Seattle, did computer ground-motion modeling based on what Bennett and other scientists found in the trench they dug across the Gate Creek Fault. The idea was to see if an earthquake there could cause enough shaking at the Bonneville site to trigger the landslide.
Grant ran about 10 simulations that modeled earthquakes happening on the fault ranging from magnitudes 6.5 to 7.1. They predicted moderate to strong shaking at the Bonneville landslide location more than 10 miles away. His models also showed that the shaking could be strong enough to trigger landslides.
But that does not mean that is what happened.
“We can predict the kind of normal, small landslides. Bonneville’s such a mega thing that there’s only a few of them in the country. We don’t really have good models for that,” Grant said. “We’ve talked about how we should really do a rigorous study and figure out, could this earthquake on this fault have triggered this landslide.”
Such studies would include engineering to determine the strength of the rocks in the slide area that would help develop more sophisticated simulations.
In Bennett’s view, Grant’s simulations suggested a Mount Hood quake was more likely to be the trigger of the landslide than a megathrust quake along the Cascadia Subduction Zone off the coast.
“The ground motions from the subduction zone earthquake, the models predict much smaller ground accelerations here [at the landslide site] than you would get from a Mount Hood Fault Zone earthquake,” he said. “So that’s one reason why the Mount Hood Fault Zone might be a source, or reason why it’s a little more appealing than the earthquake on the subduction zone as being the reason [for the landslide].”