The Giant, Underestimated Earthquake Threat to North America
The enormous fault off the coast of the Pacific Northwest has been silent for three centuries. But after years of detective work, geologists have discovered that it can unleash mayhem on an epic scale.
By Jerry Thompson|Tuesday, March 13, 2012
The “ghost forest” of dead cedar trees at the
Copalis River on the Washington coast is evidence of a major quake three
centuries ago. Brian Atwate/USGS
Just over one year ago, a magnitude-9 earthquake hit the Tohoku
region of northeastern Japan, triggering one of the most destructive
tsunamis in a thousand years. The Japanese—the most earthquake-prepared,
seismically savvy people on the planet—were caught off-guard by the
Tohoku quake’s savage power. Over 15,000 people died.
Now scientists are calling attention to a dangerous area on the opposite side of the Ring of Fire, the Cascadia Subduction Zone, a fault that runs parallel to the Pacific coast of North America, from northern California to Vancouver Island. This tectonic time bomb is alarmingly similar to Tohoku, capable of generating a megathrust earthquake at or above magnitude 9, and about as close to Portland, Seattle, and Vancouver as the Tohoku fault is to Japan’s coast. Decades of geological sleuthing recently established that although it appears quiet, this fault has ripped open again and again, sending vast earthquakes throughout the Pacific Northwest and tsunamis that reach across the Pacific.
What happened in Japan will probably happen in North America. The big question is when.
On a foggy spring morning just before sunrise, 27 miles northwest of Cape Mendocino, California, a pimple of rock roughly a dozen miles below the ocean floor finally reaches its breaking point. Two slabs of the Earth’s crust begin to slip and shudder and snap apart.
The first jolt of stress coming out of the rocks sends a shock wave hurtling into Northern California and southern Oregon like a thunderbolt. For a few stunned drivers on the back roads in the predawn gloom, the pulse of energy that tears through the ground looks dimly like a 20-mile wrinkle moving through a carpet of pastures and into thick stands of redwoods.
Telephone poles whip back and forth as if caught in a hurricane. Power lines rip loose in a shower of blue and yellow sparks, falling to the ground where they writhe like snakes, snapping and biting. Lights go out and the telephone system goes down.
Cornices fall, brick walls crack, plate glass shatters. Pavement buckles, cars and trucks veer into ditches and into each other. A bridge across the Eel River is jerked off its foundations, taking a busload of farm workers with it. With computers crashing and cell towers dropping offline, all of Humboldt and Del Norte Counties in California are instantly cut off from the outside world, so nobody beyond the immediate area knows how bad it is here or how widespread the damage.
At the U.S. Geological Survey (USGS) lab in Menlo Park, seismometers peg the quake at magnitude 8.1, and the tsunami detection centers in Alaska and Hawaii begin waking up the alarm system with standby alerts all around the Pacific Rim. Early morning commuters emerging from a BART station in San Francisco feel the ground sway beneath their feet and immediately hit the sidewalk in a variety of awkward crouches, a familiar fear chilling their guts.
Then another little rough spot on the bottom of the continent snaps off.
The fault unzips some more.
The outer edge of California snaps free like a steel spring in a juddering lurch—nine feet to the west. The continental shelf heaves upward, lifting a mountain of seawater.
The fault continues to rip all the way to Newport, Oregon, halfway up the state. The magnitude suddenly jumps to 8.6. A power surge blows a breaker somewhere east of town and feeds back through the system, throwing other breakers in a cascade that quickly crashes the entire grid in Oregon, Washington, and parts of California, Idaho, and Nevada. A brownout begins in six more western states. The wire line phone systems crash in lockstep.
Then another fragment of rock deep underneath Newport shears away. The fault unzips the rest of the way to Vancouver Island. The quake now pins seismic needles at magnitude 9.2. High-rise towers in Portland, Seattle, Vancouver, and Victoria begin to undulate. The shock wave hammers through sandy soil, soft rock, and landfill like the deepest notes on a big string bass. The mushy ground sings harmony and tall buildings hum like so many tuning forks.
On I-5, the main north-south interstate highway, 37 bridges between Sacramento and Bellingham, Washington, collapse or are knocked off their pins. Five more go down between the Canada–United States border and downtown Vancouver. Nineteen railway bridges along the north-south coastal mainline of the Burlington Northern Santa Fe railway are wrecked as well. The runways of every major coastal airport from Northern California to Vancouver are buckled, cracked, and no longer flyable.
After 50 cycles of harmonic vibration—skyscrapers swaying rhythmically from side to side in giddy wobbles—dozens of tall buildings have shed most of their glass. In some downtown intersections the cascade of broken shards has piled up three feet deep.
Shock waves have been pummeling the Pacific Northwest for four minutes and thirty-five seconds now, and it still isn’t over. After 64 cycles, enough welds have cracked, enough concrete has spalled, enough shear walls have come unstuck that some towers begin to pancake. The same death spiral everyone saw in New York on 9/11 happens all over again. Smaller buildings, but more of them. Dozens of towers go down in the four northernmost of the affected cities.
In the five major urban areas along the fault, tens of thousands of people have been seriously injured. Hundreds, perhaps thousands, are dead. More than a third of the oncoming shift of police, firefighters, paramedics, nurses, and doctors do not show up for work. They are either stranded by collapsed buildings, bridges, and roadways, injured or dead themselves, or have decided to stick close to home to make sure their own families are OK before going to work. People who survive the collapses must do their own search and rescue for family members, friends, and neighbors still trapped in the rubble. Help will come eventually, but who knows when?
People in the United States and Canada, if they think at all about earthquake disasters, probably conjure up the San Andreas fault in the worst-case scenario. In California, as they wait for “the Big One,” people wonder which city the San Andreas will wreck next—San Francisco or Los Angeles? But if by the Big One they mean the earthquake that will wreak havoc over the widest geographic area, that could destroy the most critical infrastructure, that could send a train of tsunamis across the Pacific causing economic mayhem that would probably last a decade or more—then the seismic demon to blame could not possibly be the San Andreas. It would have to be Cascadia’s fault.
One year after Japan’s devastating Tohoku earthquake and tsunami, scientists are still trying to figure out how the world’s most organized and earthquake-ready nation could have been taken so much by surprise. They were hit by an earthquake roughly 25 times more powerful than experts thought possible in that part of the country. How could the forecast have been so wrong? The short answer is they didn’t look far enough back in geologic time to see that quakes and tsunamis just this big had indeed occurred there before. If they had prepared themselves for a much larger quake and wave, the outcome might have been entirely different.
The Cascadia subduction zone is a crack in the Earth’s crust, roughly 60 miles offshore and running 800 miles from northern Vancouver Island to Northern California. This fault is part of the infamous Pacific Ring of Fire, the impact zone where several massive tectonic plates collide. Here, a slab of the Pacific Ocean floor called the Juan de Fuca plate slides eastward and downward, “subducting” underneath the continental plate of North America.
When any two plates grind against each and get stuck, enormous stress builds up until the rocks fracture and the fault rips apart in a giant earthquake. Two other segments of the Ring of Fire ruptured this way—Chile in 1960 at magnitude 9.5, the largest quake ever recorded on Earth, and Alaska’s horrible Good Friday earthquake of 1964, at 9.2 the strongest jolt ever to hit the continent of North America.
Cascadia, however, is classified as the quietest subduction zone in the world. Along the Cascadia segment, geologists could find no evidence of major quakes in “all of recorded history”—the 140 years since white settlers arrived in the Pacific Northwest and began keeping records. For reasons unknown, it appeared to be a special case. The system was thought to be aseismic—essentially quake free and harmless.
By the 1970s several competing theories emerged to explain Cascadia’s silence. One possibility was that the Juan de Fuca plate had shifted direction, spun slightly by movement of the two larger plates on either side of it. This would reduce the rate of eastward motion underneath North America and thus reduce the buildup of earthquake stress. Another possibility was that the angle of the down-going eastbound plate was too shallow to build up the kind of friction needed to cause major quakes.
But the third possibility was downright scary. In this interpretation, the silence along the fault was merely an ominous pause. It could be that these two great slabs of the Earth’s crust were jammed against each other and had been for a very long time—locked together by friction for hundreds of years, far longer than “all of recorded history.” If that were true, they would be building up the kind of stress and strain that only a monster earthquake could relieve.
In the early 1980s, two Caltech geophysicists, Tom Heaton and Hiroo Kanamori, compared Cascadia to active quake-prone subduction zones along the coasts of Chile and Alaska and to the Nankai Trough off the coast of Japan. They found more similarities than differences. In fact, they found that the biggest megathrust events in these other zones were directly related to young, buoyant plates’ being strongly coupled to the overlying landmass at shallow angles—which fit the description of Cascadia perfectly. Bottom line: If giant ruptures could happen there—in Chile, Alaska, or Japan—the same would probably happen here, in the Pacific Northwest.
The problem, as Heaton explained it to me, was that there was no direct physical sign of earthquakes. All the comparison studies in the world could not prove unequivocally that Cascadia’s fault had ruptured in the past. What everyone needed and wanted was forensic evidence. In the breach, significant doubt and strong disagreement had separated the scientists into opposing camps. “There was plenty of skepticism out there among geophysicists that the zone really was capable of doing this stuff,” confirms paleogeologist Brian Atwater of the U.S. Geological Survey at the University of Washington in Seattle.
The only thing that could put an end to the back-and-forth debate would be tangible signs of past ruptures along the entire subduction zone. If the two plates were sliding past each other smoothly, at a constant rate, and without getting stuck together, then there should be a slow, continuous, and irreversible rise in land levels along the outer coast. On the other hand, if the two plates were stuck together by friction, strain would build up in the rocks and the upper plate would bend down along the outer edge and thicken inland, humping upward until the rocks along the fault failed. In the violent, shuddering release of strain during an earthquake, the upper plate would snap to the west, toward its original shape. The clear signal—the geodetic fingerprint—of a large subduction earthquake would be the abrupt lowering of land behind the beaches when the upper plate got stretched like taffy, snapped to the west, and then sank below the tide line.
That was something Atwater figured he could probably measure and verify—or disprove. “When they said the Pacific Coast was rising three millimeters a year relative to Puget Sound, I said, ‘Aha! Three meters per thousand!’ ” He would go out to the coast and find out whether a 3,000-year-old shoreline was now 30 feet above sea level, simple as that.
In March 1986 Atwater drove west from Seattle toward Neah Bay and Cape Flattery, on the northwestern tip of Washington State, and started searching the beaches, tide marshes, and river estuaries for clues about whether the outer coast had risen or dropped.
Neah Bay was as good a place as any to start because the land all around it is so close to sea level it was highly likely he would be able to spot even slight changes in shoreline elevation. Atwater spent a few rainy days on the marshy floor of this valley. At first he poked holes with a core barrel and came up with nothing unusual, just signs that sand and silt had built the marsh by filling a former bay. But late one afternoon, with the tide down, he tried his luck digging into the muddy bank of a stream that emptied into the marsh. Several swipes of his army shovel exposed something odd a few feet below the top of the bank, beneath a layer of sand from the bay. It was a marsh soil, marked by the remains of a plant he recognized: seaside arrowgrass.
Pretty quickly he recognized what he was looking at—evidence that land formerly high enough above the highest tides for plants to be living on it had suddenly dropped down far enough for the plants to be killed by saltwater.
This subsidence of the landscape had apparently happened very quickly. That uppermost layer of sand, above the peaty soil, had been dumped on top quickly enough to seal off the arrowgrass from the air and keep it from rotting. These plants were hundreds of years old, but they were still recognizable.
Was it physical proof that the ground here had slumped during an earthquake, that the plants of a marsh or forest meadow had been drowned quite suddenly by incoming tides and perhaps buried under the sands of a huge tsunami? Could this finally be a real smoking gun?
The deeper Atwater dug, the more he found. During that summer he and two coworkers uncovered evidence of at least six different events— presumably six different earthquakes—that had each caused about three feet or so of down-drop.
He returned to the coast in 1987 with David Yamaguchi, who had a Ph.D. in forestry from the University of Washington and was working on a project for the USGS to use tree-ring dating to figure out when Mount St. Helens had erupted prior to 1980. Together they found groves of weather-beaten, moss-draped dead western red cedar tree trunks standing knee-deep in saltwater, what became known as ghost forests. Western red cedar doesn’t grow in saltwater; these trees had presumably been killed when forest meadows subsided following an earthquake and were swamped with saltwater.
Yamaguchi’s first effort to use spruce stumps to establish a time of inundation and death had failed because, with all the rot, there were not enough rings left to count. Western red cedar, however, was more durable than spruce. Using live trees for comparison, Yamaguchi was able to establish that the cedars had rings up until the early 1690s. The earthquake that killed these cedars must have happened some time soon after then, and later samples from the roots of these trees confirmed that they were killed in the winter of 1700.
What Brian Atwater had discovered in estuaries along the Washington shore, Alan Nelson of the USGS and a team of international colleagues found as well in Oregon and British Columbia in 1995. He and 11 other scientists invested considerable time and effort—including 85 new radiocarbon-dated samples—to obtain the most accurate time line possible. They found that all the ghost forests and marsh plants along the Pacific Northwest coast had been killed at the same moment in time as the land dropped down and was covered by tsunami sand, roughly three centuries ago. If the coastline had slumped in river mouths and bays that were many miles apart, the quakes must have been very big. Atwater was pretty sure they were bigger than anything that had happened during Washington’s written history.
But across the Pacific, written history extends further into the past. Kenji Satake of the Geological Survey of Japan and colleagues soon discovered another piece of the puzzle. They found records from the year 1700 of a 16-foot-high tsunami that struck the eastern seaboard of Japan—apparently out of nowhere, since there was no mention of a local earthquake. Taken together, the evidence strongly suggested that Cascadia’s fault was the source of the giant wave.
Together, Atwater, Yamaguchi, Satake, and their colleagues had sleuthed out precisely when Cascadia had last yawned open. Atwater’s tsunami sands gave a carbon date some time between 1690 and 1720. Rings from the cedar trees narrowed the date to the winter of 1699–1700. Finally, Satake’s written records of a tsunami hitting villages all along eastern Japan nailed the date: Cascadia’s last monster quake happened on January 26, 1700, at 9 p.m. They had cracked the case—except in this detective story, the culprit would almost certainly strike again.
The evidence amassed since then suggests that in fact, Cascadia has generated powerful earthquakes not just once or twice, but over and over again throughout geologic time. A research team led by Chris Goldfinger at Oregon State University (OSU) used core samples from the ocean floor along the fault to establish that there have been at least 41 Cascadia events in the last ten thousand years. Nineteen of those events ripped the fault from end to end, a “full margin rupture.”
It turns out that Cascadia is virtually identical to the offshore faults that devastated Sumatra in 2004 and Japan in 2011—almost the same length, the same width, and with the same tectonic forces at work. Cascadia’s fault can and will generate the same kind of earthquake we saw last year: magnitude 9 or higher. It will send a train of deadly tsunami waves across the Pacific and crippling shock waves across a far wider geographic area than all the California quakes you’ve ever heard about.
Based on historical averages, the southern end of the fault—from Cape Mendocino, California, to Newport, Oregon—has a large earthquake every 240 years. For the northern end—from mid-Oregon to mid- Vancouver Island—the average “recurrence interval” is 480 years, according to a recent Canadian study. And while the north may have only half as many jolts, they tend to be full-size disasters in which the entire fault breaks from end to end.
With a time line of 41 events the science team at OSU has now calculated that the California–Oregon end of Cascadia’s fault has a 37 percent chance of producing a major earthquake in the next 50 years. The odds are 10 percent that an even larger quake will strike the upper end, in a full-margin rupture, within 50 years. Given that the last big quake was 312 years ago, one might argue that a very bad day on the Cascadia Subduction Zone is ominously overdue. It appears that three centuries of silence along the fault has been entirely misleading. The monster is only sleeping.
Now scientists are calling attention to a dangerous area on the opposite side of the Ring of Fire, the Cascadia Subduction Zone, a fault that runs parallel to the Pacific coast of North America, from northern California to Vancouver Island. This tectonic time bomb is alarmingly similar to Tohoku, capable of generating a megathrust earthquake at or above magnitude 9, and about as close to Portland, Seattle, and Vancouver as the Tohoku fault is to Japan’s coast. Decades of geological sleuthing recently established that although it appears quiet, this fault has ripped open again and again, sending vast earthquakes throughout the Pacific Northwest and tsunamis that reach across the Pacific.
What happened in Japan will probably happen in North America. The big question is when.
On a foggy spring morning just before sunrise, 27 miles northwest of Cape Mendocino, California, a pimple of rock roughly a dozen miles below the ocean floor finally reaches its breaking point. Two slabs of the Earth’s crust begin to slip and shudder and snap apart.
The first jolt of stress coming out of the rocks sends a shock wave hurtling into Northern California and southern Oregon like a thunderbolt. For a few stunned drivers on the back roads in the predawn gloom, the pulse of energy that tears through the ground looks dimly like a 20-mile wrinkle moving through a carpet of pastures and into thick stands of redwoods.
Telephone poles whip back and forth as if caught in a hurricane. Power lines rip loose in a shower of blue and yellow sparks, falling to the ground where they writhe like snakes, snapping and biting. Lights go out and the telephone system goes down.
Cornices fall, brick walls crack, plate glass shatters. Pavement buckles, cars and trucks veer into ditches and into each other. A bridge across the Eel River is jerked off its foundations, taking a busload of farm workers with it. With computers crashing and cell towers dropping offline, all of Humboldt and Del Norte Counties in California are instantly cut off from the outside world, so nobody beyond the immediate area knows how bad it is here or how widespread the damage.
At the U.S. Geological Survey (USGS) lab in Menlo Park, seismometers peg the quake at magnitude 8.1, and the tsunami detection centers in Alaska and Hawaii begin waking up the alarm system with standby alerts all around the Pacific Rim. Early morning commuters emerging from a BART station in San Francisco feel the ground sway beneath their feet and immediately hit the sidewalk in a variety of awkward crouches, a familiar fear chilling their guts.
Then another little rough spot on the bottom of the continent snaps off.
The fault unzips some more.
The outer edge of California snaps free like a steel spring in a juddering lurch—nine feet to the west. The continental shelf heaves upward, lifting a mountain of seawater.
The fault continues to rip all the way to Newport, Oregon, halfway up the state. The magnitude suddenly jumps to 8.6. A power surge blows a breaker somewhere east of town and feeds back through the system, throwing other breakers in a cascade that quickly crashes the entire grid in Oregon, Washington, and parts of California, Idaho, and Nevada. A brownout begins in six more western states. The wire line phone systems crash in lockstep.
Then another fragment of rock deep underneath Newport shears away. The fault unzips the rest of the way to Vancouver Island. The quake now pins seismic needles at magnitude 9.2. High-rise towers in Portland, Seattle, Vancouver, and Victoria begin to undulate. The shock wave hammers through sandy soil, soft rock, and landfill like the deepest notes on a big string bass. The mushy ground sings harmony and tall buildings hum like so many tuning forks.
On I-5, the main north-south interstate highway, 37 bridges between Sacramento and Bellingham, Washington, collapse or are knocked off their pins. Five more go down between the Canada–United States border and downtown Vancouver. Nineteen railway bridges along the north-south coastal mainline of the Burlington Northern Santa Fe railway are wrecked as well. The runways of every major coastal airport from Northern California to Vancouver are buckled, cracked, and no longer flyable.
After 50 cycles of harmonic vibration—skyscrapers swaying rhythmically from side to side in giddy wobbles—dozens of tall buildings have shed most of their glass. In some downtown intersections the cascade of broken shards has piled up three feet deep.
Shock waves have been pummeling the Pacific Northwest for four minutes and thirty-five seconds now, and it still isn’t over. After 64 cycles, enough welds have cracked, enough concrete has spalled, enough shear walls have come unstuck that some towers begin to pancake. The same death spiral everyone saw in New York on 9/11 happens all over again. Smaller buildings, but more of them. Dozens of towers go down in the four northernmost of the affected cities.
In the five major urban areas along the fault, tens of thousands of people have been seriously injured. Hundreds, perhaps thousands, are dead. More than a third of the oncoming shift of police, firefighters, paramedics, nurses, and doctors do not show up for work. They are either stranded by collapsed buildings, bridges, and roadways, injured or dead themselves, or have decided to stick close to home to make sure their own families are OK before going to work. People who survive the collapses must do their own search and rescue for family members, friends, and neighbors still trapped in the rubble. Help will come eventually, but who knows when?
People in the United States and Canada, if they think at all about earthquake disasters, probably conjure up the San Andreas fault in the worst-case scenario. In California, as they wait for “the Big One,” people wonder which city the San Andreas will wreck next—San Francisco or Los Angeles? But if by the Big One they mean the earthquake that will wreak havoc over the widest geographic area, that could destroy the most critical infrastructure, that could send a train of tsunamis across the Pacific causing economic mayhem that would probably last a decade or more—then the seismic demon to blame could not possibly be the San Andreas. It would have to be Cascadia’s fault.
One year after Japan’s devastating Tohoku earthquake and tsunami, scientists are still trying to figure out how the world’s most organized and earthquake-ready nation could have been taken so much by surprise. They were hit by an earthquake roughly 25 times more powerful than experts thought possible in that part of the country. How could the forecast have been so wrong? The short answer is they didn’t look far enough back in geologic time to see that quakes and tsunamis just this big had indeed occurred there before. If they had prepared themselves for a much larger quake and wave, the outcome might have been entirely different.
Courtesy Chris Goldfinger
Exactly the same is true of the Cascadia subduction zone—an
almost identical geologic threat off the west coast of North America.
When it was first discovered, many scientists thought Cascadia’s fault
was incapable of generating giant earthquakes. Now they know they were
wrong. They just hadn’t looked far enough into the past.The Cascadia subduction zone is a crack in the Earth’s crust, roughly 60 miles offshore and running 800 miles from northern Vancouver Island to Northern California. This fault is part of the infamous Pacific Ring of Fire, the impact zone where several massive tectonic plates collide. Here, a slab of the Pacific Ocean floor called the Juan de Fuca plate slides eastward and downward, “subducting” underneath the continental plate of North America.
When any two plates grind against each and get stuck, enormous stress builds up until the rocks fracture and the fault rips apart in a giant earthquake. Two other segments of the Ring of Fire ruptured this way—Chile in 1960 at magnitude 9.5, the largest quake ever recorded on Earth, and Alaska’s horrible Good Friday earthquake of 1964, at 9.2 the strongest jolt ever to hit the continent of North America.
Cascadia, however, is classified as the quietest subduction zone in the world. Along the Cascadia segment, geologists could find no evidence of major quakes in “all of recorded history”—the 140 years since white settlers arrived in the Pacific Northwest and began keeping records. For reasons unknown, it appeared to be a special case. The system was thought to be aseismic—essentially quake free and harmless.
By the 1970s several competing theories emerged to explain Cascadia’s silence. One possibility was that the Juan de Fuca plate had shifted direction, spun slightly by movement of the two larger plates on either side of it. This would reduce the rate of eastward motion underneath North America and thus reduce the buildup of earthquake stress. Another possibility was that the angle of the down-going eastbound plate was too shallow to build up the kind of friction needed to cause major quakes.
But the third possibility was downright scary. In this interpretation, the silence along the fault was merely an ominous pause. It could be that these two great slabs of the Earth’s crust were jammed against each other and had been for a very long time—locked together by friction for hundreds of years, far longer than “all of recorded history.” If that were true, they would be building up the kind of stress and strain that only a monster earthquake could relieve.
In the early 1980s, two Caltech geophysicists, Tom Heaton and Hiroo Kanamori, compared Cascadia to active quake-prone subduction zones along the coasts of Chile and Alaska and to the Nankai Trough off the coast of Japan. They found more similarities than differences. In fact, they found that the biggest megathrust events in these other zones were directly related to young, buoyant plates’ being strongly coupled to the overlying landmass at shallow angles—which fit the description of Cascadia perfectly. Bottom line: If giant ruptures could happen there—in Chile, Alaska, or Japan—the same would probably happen here, in the Pacific Northwest.
The problem, as Heaton explained it to me, was that there was no direct physical sign of earthquakes. All the comparison studies in the world could not prove unequivocally that Cascadia’s fault had ruptured in the past. What everyone needed and wanted was forensic evidence. In the breach, significant doubt and strong disagreement had separated the scientists into opposing camps. “There was plenty of skepticism out there among geophysicists that the zone really was capable of doing this stuff,” confirms paleogeologist Brian Atwater of the U.S. Geological Survey at the University of Washington in Seattle.
The only thing that could put an end to the back-and-forth debate would be tangible signs of past ruptures along the entire subduction zone. If the two plates were sliding past each other smoothly, at a constant rate, and without getting stuck together, then there should be a slow, continuous, and irreversible rise in land levels along the outer coast. On the other hand, if the two plates were stuck together by friction, strain would build up in the rocks and the upper plate would bend down along the outer edge and thicken inland, humping upward until the rocks along the fault failed. In the violent, shuddering release of strain during an earthquake, the upper plate would snap to the west, toward its original shape. The clear signal—the geodetic fingerprint—of a large subduction earthquake would be the abrupt lowering of land behind the beaches when the upper plate got stretched like taffy, snapped to the west, and then sank below the tide line.
That was something Atwater figured he could probably measure and verify—or disprove. “When they said the Pacific Coast was rising three millimeters a year relative to Puget Sound, I said, ‘Aha! Three meters per thousand!’ ” He would go out to the coast and find out whether a 3,000-year-old shoreline was now 30 feet above sea level, simple as that.
In March 1986 Atwater drove west from Seattle toward Neah Bay and Cape Flattery, on the northwestern tip of Washington State, and started searching the beaches, tide marshes, and river estuaries for clues about whether the outer coast had risen or dropped.
Neah Bay was as good a place as any to start because the land all around it is so close to sea level it was highly likely he would be able to spot even slight changes in shoreline elevation. Atwater spent a few rainy days on the marshy floor of this valley. At first he poked holes with a core barrel and came up with nothing unusual, just signs that sand and silt had built the marsh by filling a former bay. But late one afternoon, with the tide down, he tried his luck digging into the muddy bank of a stream that emptied into the marsh. Several swipes of his army shovel exposed something odd a few feet below the top of the bank, beneath a layer of sand from the bay. It was a marsh soil, marked by the remains of a plant he recognized: seaside arrowgrass.
Pretty quickly he recognized what he was looking at—evidence that land formerly high enough above the highest tides for plants to be living on it had suddenly dropped down far enough for the plants to be killed by saltwater.
This subsidence of the landscape had apparently happened very quickly. That uppermost layer of sand, above the peaty soil, had been dumped on top quickly enough to seal off the arrowgrass from the air and keep it from rotting. These plants were hundreds of years old, but they were still recognizable.
Was it physical proof that the ground here had slumped during an earthquake, that the plants of a marsh or forest meadow had been drowned quite suddenly by incoming tides and perhaps buried under the sands of a huge tsunami? Could this finally be a real smoking gun?
The deeper Atwater dug, the more he found. During that summer he and two coworkers uncovered evidence of at least six different events— presumably six different earthquakes—that had each caused about three feet or so of down-drop.
He returned to the coast in 1987 with David Yamaguchi, who had a Ph.D. in forestry from the University of Washington and was working on a project for the USGS to use tree-ring dating to figure out when Mount St. Helens had erupted prior to 1980. Together they found groves of weather-beaten, moss-draped dead western red cedar tree trunks standing knee-deep in saltwater, what became known as ghost forests. Western red cedar doesn’t grow in saltwater; these trees had presumably been killed when forest meadows subsided following an earthquake and were swamped with saltwater.
Yamaguchi’s first effort to use spruce stumps to establish a time of inundation and death had failed because, with all the rot, there were not enough rings left to count. Western red cedar, however, was more durable than spruce. Using live trees for comparison, Yamaguchi was able to establish that the cedars had rings up until the early 1690s. The earthquake that killed these cedars must have happened some time soon after then, and later samples from the roots of these trees confirmed that they were killed in the winter of 1700.
What Brian Atwater had discovered in estuaries along the Washington shore, Alan Nelson of the USGS and a team of international colleagues found as well in Oregon and British Columbia in 1995. He and 11 other scientists invested considerable time and effort—including 85 new radiocarbon-dated samples—to obtain the most accurate time line possible. They found that all the ghost forests and marsh plants along the Pacific Northwest coast had been killed at the same moment in time as the land dropped down and was covered by tsunami sand, roughly three centuries ago. If the coastline had slumped in river mouths and bays that were many miles apart, the quakes must have been very big. Atwater was pretty sure they were bigger than anything that had happened during Washington’s written history.
But across the Pacific, written history extends further into the past. Kenji Satake of the Geological Survey of Japan and colleagues soon discovered another piece of the puzzle. They found records from the year 1700 of a 16-foot-high tsunami that struck the eastern seaboard of Japan—apparently out of nowhere, since there was no mention of a local earthquake. Taken together, the evidence strongly suggested that Cascadia’s fault was the source of the giant wave.
Together, Atwater, Yamaguchi, Satake, and their colleagues had sleuthed out precisely when Cascadia had last yawned open. Atwater’s tsunami sands gave a carbon date some time between 1690 and 1720. Rings from the cedar trees narrowed the date to the winter of 1699–1700. Finally, Satake’s written records of a tsunami hitting villages all along eastern Japan nailed the date: Cascadia’s last monster quake happened on January 26, 1700, at 9 p.m. They had cracked the case—except in this detective story, the culprit would almost certainly strike again.
The evidence amassed since then suggests that in fact, Cascadia has generated powerful earthquakes not just once or twice, but over and over again throughout geologic time. A research team led by Chris Goldfinger at Oregon State University (OSU) used core samples from the ocean floor along the fault to establish that there have been at least 41 Cascadia events in the last ten thousand years. Nineteen of those events ripped the fault from end to end, a “full margin rupture.”
It turns out that Cascadia is virtually identical to the offshore faults that devastated Sumatra in 2004 and Japan in 2011—almost the same length, the same width, and with the same tectonic forces at work. Cascadia’s fault can and will generate the same kind of earthquake we saw last year: magnitude 9 or higher. It will send a train of deadly tsunami waves across the Pacific and crippling shock waves across a far wider geographic area than all the California quakes you’ve ever heard about.
Based on historical averages, the southern end of the fault—from Cape Mendocino, California, to Newport, Oregon—has a large earthquake every 240 years. For the northern end—from mid-Oregon to mid- Vancouver Island—the average “recurrence interval” is 480 years, according to a recent Canadian study. And while the north may have only half as many jolts, they tend to be full-size disasters in which the entire fault breaks from end to end.
With a time line of 41 events the science team at OSU has now calculated that the California–Oregon end of Cascadia’s fault has a 37 percent chance of producing a major earthquake in the next 50 years. The odds are 10 percent that an even larger quake will strike the upper end, in a full-margin rupture, within 50 years. Given that the last big quake was 312 years ago, one might argue that a very bad day on the Cascadia Subduction Zone is ominously overdue. It appears that three centuries of silence along the fault has been entirely misleading. The monster is only sleeping.
