Kenji Satake arranged to meet me in the seaport city of Tanabe, southernmost of the six communities where records of Cascadia’s wave had been discovered. In 1997 Tanabe had a population of 72,000. From a distance the harbor and skyline, set against a range of low, blue-green hills and small mountains, looked to me a lot like the coast of southern British Columbia, where the orphan tsunami was born.
Tanabe was badly shaken in 1946 by the magnitude 8.1 subduction rupture along the Nankai Trough, less than sixty miles (100 km) away. The tsunami that hit the harbor only a few minutes later swept 145 homes away, damaged or destroyed 1,200 more, and killed 69 people. Smudgy black and white photos of the aftermath look similar to those from Phuket and Banda Aceh in 2004—jumbled piles of splintered lumber and debris, fishing boats tossed high and dry, stunned survivors looking lost, wandering in the rubble.
A group of city officials took us up a stone stairway leading to an ancient temple on high ground. Twenty steps up the mountain we saw a monument to the dead, engraved with a notch in the marble tablet to indicate the high-water line of the 1946 tsunami. I tried to picture a wall of water high enough to touch the eaves of a two-story building, because that’s how far up it looked to me. Satake and his team compared the 1946 inundation zone to descriptions of the 1700 Cascadia waves and concluded that they had probably reached this height as well.
Tanabe’s mayor in 1700, a member of the Tadokoro family, lived in the merchant district, north of and slightly uphill from the old walled castle. The city’s population then was roughly 2,600 people. The mayor himself escaped the worst effects of the rising water because the flood stopped roughly five hundred feet (150 m) below his home. As an eyewitness, Tadokoro collected and supervised the writing of the official damage reports.
With brushes and black ink faded only slightly after three hundred years, he or his scribe told the scary story in fluid, artistic strokes. When we arrived to film the evidence, Japan’s seismic historians had already searched thousands of pages of the city’s official ledger as well as the Tadokoro family’s “diary of ten thousand generations” to find the orphan tsunami. Complex ideograms flowed in neat, vertical columns on washi, a traditional acid-free paper strengthened with fibers of bark. Over the course of more than three centuries the smooth, durable pages had survived water damage and were perforated here and there by bookworms. But the essential details remained.
The rising Pacific surge flooded the castle moat and ruined nearby paddies and fields of wheat. It soaked and destroyed bales of rice stored in several government warehouses. In those days rice was almost the same as money. On behalf of the shogun, samurai collected taxes in the form of rice. Therefore it was very likely that the shogun’s rice got wrecked by the orphan tsunami from Cascadia. The documents describing what happened to the shogun’s rice are now considered proof that Cascadia is a dangerous subduction zone.
But there was still one more question to answer. “Kenji’s story seemed to hang together very well,” David Yamaguchi agreed, “but it was still very much a story that was hypothetical. He could not prove conclusively that that wave came from here,” the Pacific Northwest coast of America.
To do that something a little more tangible was needed. And somewhere along the way another eureka moment happened. If radiocarbon would never be precise enough by itself, and if the outer wood of those trees in the ghost forest was too chewed up to give the exact year of death, what about digging deeper? Why not dig up the roots of those standing dead cedars and look for that final year’s growth ring? Would the roots be preserved well enough underground to show a ring for 1699? If so, wouldn’t that suggest the trees did not live long enough to generate a growth ring for 1700? That ’s the question Brian Atwater asked himself, and he called on David Yamaguchi’s expertise one more time.
“Brian asked me for leads,” Yamaguchi recalled. “He asked me to choose my top-ten list of trees that I thought gave me the best tree-ring dates—that I was most confident in. And then he went out with a team—a young team—and excavated these trees down to roots where they found that they had intact bark.”
At the ghost forest, Atwater packed his pickup truck full of muddy roots and a gang of exhausted diggers with aching muscles and drove home. Back in Seattle he sanded the samples in his backyard, checked them with a hand lens, and then packed them off to the Lamont-Dougherty lab in New York. “Trees only put on rings between about May and September,” Yamaguchi explained, “and so if they were killed by an earthquake in January of 1700, their outermost rings should be 1699.”
And that’s exactly what they found. Yamaguchi matched the root rings against “the barcode” of rings from the witness trees and confirmed—beyond any reasonable doubt—that the cedars of the ghost forest were killed in the winter of 1700. “Radiocarbon dating can’t do it. Chemical dating can’t do it. It was up to tree rings,” Yamaguchi proclaimed with a wide smile.
All that wet and dirty shovel work had finally paid off. “This finding here at this estuary,” said Atwater, standing on the bank of the Copalis River, “and at the other three estuaries in Washington to the south of us—all having trees dying between August of 1699 and May of 1700—that kind of long reach of coast all dying together at the time of the solitary tsunami in Japan makes a very big earthquake at our Cascadia Subduction Zone more plausible than it was before.”
For Atwater it was pretty clear from the beginning that a rupture had happened here and that a big tsunami had been generated. The evidence had always been there in the geology along the west coast. Cascadia’s fault had indeed written its own history in rock and mud. “It just becomes reinforced by having a human record in Japan that says there’s something corresponding over there—that people observed, and wrote about and that got preserved in those wonderful records.”
Together Atwater, Yamaguchi, Satake, and their colleagues had carried the investigation to what seemed like a logical conclusion. They tracked down the parent earthquake of Japan’s orphan wave of 1700. It had been an exciting, frustrating, and tedious journey that ended with deadly implications.
How did it feel to learn something so dire? “It’s mixed feelings,” admitted Yamaguchi. “As a scientist, it was sort of a eureka feeling—wow, we’ve worked this out! As a resident of Seattle, it’s also sort of a humbling and scary kind of feeling because you realize this is historical evidence for an earthquake of a size this region—and the world—has scarcely seen.”
“Once we admit this earthquake happened in the past,” Satake told me, “we also have to accept that the same earthquake will happen in the future. Because earthquakes in subduction zones—we know that they repeat. So the most significant thing—the most important thing that comes out of this research—is that now we know that this earthquake can happen in the future.”
CHAPTER 18
Episodic Tremor and Slip: Tracking Cascadia with GPS
At four o’clock in the morning on August 31, 1983, the ramp crew at Anchorage International Airport finished refueling a Boeing 747 operated by Korean Air Lines. A few minutes later KAL flight 007 took off on the final leg of its journey from New York to Seoul. As the sun rose the big jet roared across the International Date Line and headed south and west toward its home port. Then, for reasons never verified, the plane allegedly strayed off course over the Sea of Japan west of Sakhalin Island and was shot down by fighter jets from the Soviet Union.
All 269 passengers and crew aboard KAL 007 were killed and for several weeks the Cold War threatened to escalate into something worse. When flight data recorders were recovered from the wreckage, the Soviets refused to release them to international aviation authorities investigating the crash, so it was impossible to prove or disprove the conflicting stories of why the Korean jet had drifted into prohibited airspace. But after the heat of the moment faded, one positive change did come about as a result of the tragedy.
To reduce the odds that a navigational error could ever cause something like this again, President Ronald
Reagan ordered the U.S. military to make its NAVSTAR Global Positioning System (GPS) available for civilian use. GPS did not change the world of aviation overnight because there still weren’t enough satellites in orbit in 1983 to make the system fully functional. The precision of the new and still-evolving technology would eventually have a significant effect on everything from tracking animal herds in the wilderness to finding a freeway exit to the nearest pizza parlor—and measuring the drift of continents.
By figuring its distance from at least four of the NAVSTAR satellites, a civilian-made GPS receiver could calculate its position anywhere on the surface of the earth to within roughly a hundred feet (30 m). The system was based on the same principles of trilateration that Jim Savage and Herb Dragert and others had used to measure the shift of mountain peaks in Puget Sound and on Vancouver Island. Their Geodolites had calculated the distance between two brass survey markers by measuring how long it took a laser beam to travel from one peak to another, bounce off a set of reflectors, and return to the starting point. GPS positioning did pretty much the same thing.
By 1992, when Mike Schmidt of the Geological Survey of Canada climbed Mount Logan with one of the slightly clunky but then state-of-the-art units, he was able to measure latitude and longitude with an accuracy of a few millimeters. They could nail the altitude within about ten or fifteen millimeters (0.4–0.6 inches). As Schmidt and Dragert and their geodesy team from the Pacific Geoscience Centre built their array of permanent tracking stations to measure the movement of Vancouver Island, enough new GPS satellites were being launched to make measurements far more accurate than laser beams bounced between mountain tops had been. The enhanced precision would also create one of the most baffling new mysteries Cascadia watchers had ever seen—a mystery once solved that would stun even those who thought they knew what to expect from an active subduction zone.
“We have now established that this site,” said Dragert, pointing at the Albert Head GPS monument (near Victoria on Vancouver Island) behind him, “that particular point right there, is moving at a rate of about six millimeters per year towards Penticton, roughly in a sixty-degree azimuth direction.”
While six millimeters of horizontal creep didn’t sound like much—less than a quarter of an inch—a dose of context changed the picture. Given that the last major earthquake to relieve stress on Cascadia’s fault happened more than three hundred years ago, and given that stress has been rebuilding ever since, that tiny increment of eastward squeezing had been accumulating in the rocks of Vancouver Island and all along the west coast for three centuries. Dragert did the math for us. “The six millimeters times three hundred years is 1,800 millimeters. So 1.8 meters [6 feet] will be released at this particular location during the course of the next large earthquake.”
In other words, if Cascadia had ruptured right then and there—after three hundred years of stress accumulation—the entire city of Victoria would have rattled, rumbled, and slid sideways nearly six feet (1.8 m) back toward the west as the strain came out of the rocks. The longer it takes the zone to fail, the more stress there will be—ready to snap in the next great rupture. “We proved that the margin was deforming,” said Dragert emphatically. “The mountains were indeed being squeezed landward. So it’s not a hypothetical problem; it’s a real problem. We are gaining strain energy all to be ultimately released by a large megathrust earthquake.”
The switch from lasers to GPS technology made it possible to track tectonic movement 24/7 without having to wait for budget approval to hire another helicopter. “You didn’t have to fly to the mountaintop—you just set up the instrument,” Dragert enthused, “a totally automated instrument that told you what its position was day after day after day. It worked under any weather conditions. Much cheaper. I mean we basically had to give up [laser] trilateration because it turned out to be too expensive,” he explained. “With this new technology we could measure even longer baseline distances—not just fifty kilometers, hundreds of kilometers—to a precision of one or two millimeters.”
The Pentagon’s new technology gave scientists a close-up view of tectonic motion that had been inconceivable thirty years earlier, when the debate about continental drift began in earnest. GPS proved not only that plates were shifting but how fast, in which directions, and how high the outer coastlines were rising, bending, and buckling. Before another year had passed, the satellites would also reveal to Dragert and his colleagues the next big secret of subduction.
“Without GPS technology we would never, never have observed ‘silent slip,’” he said, with fingers putting quotation marks around the term. The first time he spotted a tiny “backward movement” of Vancouver Island, it was so subtle, so apparently insignificant, it just had to be a glitch. At least that’s what Dragert thought at first.
He remembered the bafflement as if it were yesterday. On a desktop computer screen he opened a file that showed the data from the station nearest Victoria. “We were kind of looking back over the last four years of data in ’96. And we said, ‘Gee, there’s this funny offset,’ right around October ’94. So, what’s going on?” His finger traced the steady, almost straight-line movement of the GPS antenna at Albert Head.
As expected, it had been creeping relentlessly east day by day for months, when all of a sudden there was a sawtooth zigzag in the data. The Albert Head antenna had apparently switched direction and doubled back in the opposite direction for about ten days. A concrete tower built into the solid bedrock of Vancouver Island had inexplicably stopped moving toward Penticton and was temporarily sliding back west. After the ten days, it started moving east again.
Because only one of the four existing stations had displayed this maverick behavior, Dragert convinced himself the data had to be wrong. “So I said, ‘Okay, most likely our monument is unstable.’ For some reason, even though this is in concrete—we have rebar drilled into the bedrock, so there’s a very good coupling to the bedrock, and it ’s very competent bedrock, it ’s not fractured or weathered—so, strange as it may seem, maybe we just didn’t see something. Maybe there’s a fracture zone somewhere that we were unaware of and it’s tilted our monument.”
To find out what had gone wrong, Dragert and his colleagues drove back out to Albert Head, set up a laser transit, and resurveyed the antenna. “We did exactly the same survey as we did in’92 [when the system was first installed]. And according to both surveys—‘Hey, the monument hasn’t moved at all!’ Less than 0.3 millimeters was the difference between the ’92 survey and the ’96 survey.” The GPS tower itself was locked solidly in place, so it had to be the ground that was moving back and forth. In essence, Vancouver Island was being shoved to the east most of the time, but every fourteen months or so it would slip backward as if the underground stress had somehow been temporarily released or reversed.
How could that happen? Dragert laughed heartily. “We couldn’t explain it. We simply said, ‘That’s life,’ and we went on.”
But Dragert and his colleagues kept watching the incoming data, determined to solve the mystery. Then one day—there it was again. The same apparent glitch, the same displacement. But now they had fourteen continuously monitored GPS stations in the Pacific Northwest, giving them much more precise data.
So Dragert started calling around to find out if any of the other research teams had seen anything this weird. Sure enough, when asked to take a closer look at their data plots, several of them saw a similar kind of reversed movement. Six other GPS antennas in both Canada and the United States had “jumped backward.” In all a cluster of seven adjacent sites strung out across southern Vancouver Island from Ucluelet to Nanaimo and to Victoria and down Puget Sound as far south as Seattle had suddenly slipped backward in what looked like a slow, silent earthquake that took anywhere from six to fifteen days to happen.
And it was definitely a geographic cluster rather than a random scatter. The antenna up at Holberg, on the north end of the island, didn’t move at all. Neither did the towers at Willi
ams Lake or Chilliwack, British Columbia, or Linden, Washington. Only the stations in the middle moved. The stations at the extreme north and south ends of the GPS array didn’t flinch.
Not only that but the slippage had started a few days earlier down in Washington State and then moved gradually toward Vancouver Island—almost like the slow-motion unzipping of a fault. Dragert beamed. “We were saying, ‘Holy crap, this is great! This is absolutely great!’ This not only told us that the signal was real, it told us the signal was constrained to a given area. And it took time to travel from the south to the north.” Basically the signal looked like a ripple moving through rock.
“It was like something migrating underneath our feet,” he said. There was still no logical explanation, however, for the odd, backward-jumping movement of a handful of GPS antennas, so a professional skeptic’s first response was to say it still was probably some kind of mistake. If the GPS monuments are locked in solid rock, then something else must be wrong. Before publishing their data, Dragert and his colleague Kelin Wang had to rule out every conceivable analytical glitch and recalculate all the GPS orbits, just to be absolutely sure what they were seeing was real. Eventually they arrived at the conclusion that the silent backward slip was not a fantasy.
On his computer screen Dragert again traced the upward-slanting line of data with his finger. His hunch was that somehow, way down in the lower part of the subduction zone, a small measure of tectonic stress was being temporarily relieved. The deepest part of the zone had—in his words—come unsprung. It had slipped.
When he expanded the timeline to display several years of continuous movement, the zigzag pattern became even more obvious. The reversals put spikes in a straight-line graph that made it look like a saw blade with evenly spaced, sharp teeth. The thing that struck me about it was the regularity. How could anything in nature be that punctual? Why would it keep coming back every fourteen months?
Cascadia's Fault Page 23