Cascadia's Fault

by Jerry Thompson

  With ninety-six new cores going back nearly ten thousand years, there was finally a long enough history in the mud to look for patterns in the timing of these monster quakes. Suddenly, with the offshore landslide samples, the clock could be rolled back much further than before. At least in theory the new turbidite timelines offered a glimpse of long-term fault behavior. And the Cascadia cores did seem to reveal a repeating cycle. Judging by the thickness of ocean mud laid down between the turbidite layers, and with increasingly precise radiocarbon dates, they could tell roughly how much time had elapsed between events.

  The first cycle began with a long quiet period after the Mazama volcanic eruption—more than a thousand years without an earthquake. After the first rupture was another moderately long interval of quiet, followed by two more jolts at shorter intervals—the shorter being only 215 years. The recurrence interval, or gap, between jolts was long, short, short. And that same sequence—long, short, short—had apparently repeated three times in the last 7,500 years.

  Goldfinger and Nelson wrote that “while it is tempting to expound about earthquake clustering,” it was still early days. They had taken “a tantalizing look at what may be the long-term behavior of a major fault system” but were careful to point out that their analysis was preliminary and would require confirmation of the radiocarbon age data.

  Essentially they had encountered the same problem with radiocarbon dating that Brian Atwater did: there was very little biotic material to work with in deep-sea mud, and it had a way of getting moved around by burrowing sea creatures and sloshed out of the tops of the piston cores as they were wrangled onto ship’s decks in heaving ocean swells. Unlike Atwater they could not rely on a ghost forest of ancient cedars conveniently nearby; nor could they use tree rings cored from perfectly preserved roots to help nail down the turbidite dates by other means.

  They did, however, feel confident the wiggle-matching technique borrowed from their oilfield colleagues would eventually overcome these problems to help establish a solid and convincing long-term history for Cascadia. By December 2004 the clustering story had become more refined. Their turbidite studies had been updated, peer reviewed, and republished in several different science journals with enough new details that the media relations department at Oregon State decided to issue a news release. A draft prepared just before Christmas was set aside to be polished and sent out after the holidays.

  Then, when no one was looking, another subduction zone in the Ring of Fire ripped apart, and the entire planet got knocked for a loop by a temblor so big it made the earth wobble slightly on its axis. The great Sumatra quake and tsunami of 2004 happened the day after Christmas and all eyes shifted instantly to the Indian Ocean. So it’s unlikely that any more than a few diligent local reporters paid much attention to the OSU release issued on New Year’s Eve. For those who bothered to read it, the release provided the latest chronology of clustered quakes on Cascadia’s fault, a historical record with “two distinct implications—one that’s good, the other not.”

  Looking at the expanded pattern of turbidite beds off the coast of the Pacific Northwest, Goldfinger and Nelson concluded that the Cascadia Subduction Zone had experienced “a cluster” of four massive ruptures during the past 1,600 years. If the trend continued—if the pattern repeated—“this cluster could be over and the zone may already have entered a long quiet period of 500 to 1,000 years, which appears to be common following a cluster of earthquake events,” noted the OSU release. That was the good news.

  The alternative scenario was that the current cluster might still have one or more jolts left in it. The release pointed out that some clusters had up to five events and that within a cluster the average interval between earthquakes was three hundred years. By now most aware residents of the Pacific Northwest had heard the story that Cascadia’s last megathrust shockwave was January 26, 1700, and therefore the next event might well be imminent.

  “The Cascadia Subduction Zone has the longest recorded data about its earthquakes of any major fault in the world,” wrote Goldfinger in the OSU release, putting his new ten-thousand-year turbidite timeline front and center. “So we know quite a bit about the periodicity of the fault zone and what to expect. But the key point we don’t know is whether the current cluster of earthquake activity is over yet.”

  By December 30, 2004, when the OSU release went out to the media, the total number of Cascadia earthquakes identified in the turbidite cores had advanced from eighteen to twenty-one, at least seventeen of which had ruptured the entire length of the plate boundary from Vancouver Island to Cape Mendocino, causing magnitude 9 shockwaves and major tsunamis almost identical to the frightful scenario the whole world was watching hour after hour as the Boxing Day tragedy played out across the Indian Ocean. What we were seeing on television had happened here too—at least seventeen times.

  When Goldfinger went to sea off the coast of Sumatra in the summer of 2007, the Cascadia turbidite data had been updated and refined yet again. The temblor count had gone up again as well. When I met him to film an interview for a documentary (called ShockWave) comparing Sumatra with Cascadia, Goldfinger revealed an all new and even more ominous summary of North America’s biggest and potentially most violent fault.

  “Right now, using all of the land evidence and all of the marine evidence that we have, we can put together a story going back about ten thousand years that shows a total of thirty-four earthquakes, plus or minus one or two,” Goldfinger intoned somberly. The clustering story was there in the data but still difficult to verify. “It’s something that’s totally dependent on the radiocarbon ages and those have a lot of slop in them,” he continued. “So it appears that events seem to cluster in time and then sort of die back and are quiet for a time—and then cluster again. I would say there’s probably a 70 percent chance that there’s some statistically meaningful clusters in the earthquake record.”

  There was no obvious explanation for why quakes might cluster. “We’re just starting to really investigate the relationship between the Cascadia Subduction Zone and the faults that surround it,” Goldfinger said. “It’s of course connected to the entire Ring of Fire by other faults. We have the Queen Charlotte fault going off into Canada and the San Andreas fault going off into California, and all of these faults are all physically connected. So when you move one, it affects the others.”

  I had read a study by Ross Stein of the USGS in Menlo Park that showed how one earthquake could trigger another. The magnitude 7.5 Landers quake in southern California in 1992, fifty miles (80 km) north of Palm Springs, had triggered swarms of smaller jolts—at least sixty thousand aftershocks—as far away as Mount Shasta at the northern end of the state and even in Yellowstone National Park, more than eight hundred miles (1,300 km) away. Goldfinger told me the same thing was probably happening here in the Pacific Northwest.

  “Even if they don’t trigger each other, Cascadia and the San Andreas have stress relationships so that when one moves, it’ll affect the stress state of the other. And so as the entire Pacific plate moves—and the whole Ring of Fire has its ring of subduction zones—everything that happens in one affects the whole system,” he explained. “So hypothetically the Cascadia Subduction Zone, when it has its large event—it slips something like twenty meters [65 feet] or so—the energy from that is transferred to other places. And one of the places it could be transferred to is the northern San Andreas.”

  The thought of earthquake dominoes was hard to avoid. “It looks like somewhere around 80 to 90 percent of the San Andreas events have a Cascadia event associated with them,” he said. “Almost every event on the San Andreas has come fairly closely in time—within forty years or so—of a Cascadia Subduction Zone event. So we could be looking at a scenario where Cascadia goes off, and then some relatively short time later the San Andreas goes off as well. And if that relationship is true, that has some implications for planning for the future.”

  In the summer of 2010, an even more comprehensive
package of data from Cascadia boosted the earthquake count yet again. Now it’s forty-one events in total, nineteen of which have been full-margin ruptures, and the average length of time between megathrust disasters has been recalculated at roughly every 300 years.

  It sounded to me as though Goldfinger’s turbidite timeline—perhaps combined with those ETS events that come every fourteen months—could eventually help emergency planners by shedding new light on the dark art of earthquake prediction. I would soon learn, however, that there are those who vehemently believe any time or money spent trying to predict a disaster is a complete waste.

  CHAPTER 20

  When’s This Going to Happen? The Problems with Prediction

  Boiled down to its essence, the argument against earthquake prediction is that nature is simply too complex and chaotic. There are too many variables, too many things happening deep under ground where we cannot easily see what’s going on. No matter how closely we study the problem, no matter how many instruments we deploy, we’ll never be able to anticipate a specific earthquake in a specific place at a specific time. Critics of quake prediction believe the considerable sums of money spent trying to warn people about something that’s going to happen anyway would be better spent steeling ourselves for the shock. Put the money into mitigation instead. Build stronger buildings, dams, and bridges. Reinforce schools and hospitals to make sure they won’t fall down.

  The emphasis on mitigation certainly sounds logical and sensible, but scientists and the politicians who hand out research funding also know or suspect that a vast number of ordinary people really do want and expect to be warned. Many assume that predicting the next quake is—or should be—the primary goal and responsibility of all researchers involved in earthquake studies. Impossible as the dream of prediction may sound to skeptics, optimists tend to believe that if enough smart people are given the right tools to do the job, there’s no problem science cannot solve eventually.

  With Chris Goldfinger’s newer, longer timeline, showing at least forty-one Cascadia quakes in the past ten thousand years, one might think enough data have finally been gathered to show a pattern. If there is a pattern, then a forecast of some kind should be possible. No matter how much scorn the doubters may have heaped, no matter how many false alarms have been sounded over the years, the optimists refuse to quit. There’s always the promise of better technology, more data, faster computers that should be able to spot patterns in nature’s chaos. Plus there’s the tantalizing story of one earthquake prediction that was absolutely right.

  Kelin Wang, a scientist working for the Geological Survey of Canada who spends most of his research time today working on Cascadia’s fault, told me the story of the Haicheng earthquake in China in February 1975. It was prediction’s golden moment. Wang had heard the heroic stories of Haicheng as a young student. In 2004 he returned temporarily from Canada to the land of his birth and took advantage of a new openness in China to dig up the details of what happened in the months and days leading up to the now famous prediction.

  He and three colleagues interviewed many of the people involved in China’s official quake prediction program and were given free access to a treasure trove of declassified documents, including thousands of pages of handwritten notes and data logs. The story they found is laced with political intrigue and, for skeptics, a lingering ambiguity. What follows is a condensation of what happened.

  At 8:15 a.m., February 4, 1975, Cao Xianqing walked in to a hurriedly convened meeting of his local Party committee in Yingkou County, about three hundred miles (500 km) northeast of Beijing, and announced that “a large earthquake may occur today, during the day or in the evening.” He urged county officials to “please take measures” to make sure people were evacuated from their homes and workplaces as quickly as possible. Although higher ranking authorities in the Liaoning provincial government would later take credit for orchestrating the prediction, most citizens of Yingkou and Haicheng Counties were aware that Cao—known locally as Mr. Earthquake—was the man behind the crucial first warning on that fateful day.

  The warning itself and the subsequent saving of thousands of lives would become the stuff of seismic legend and folklore: the world’s first and to date only successful earthquake prediction followed by an evacuation. Thousands of lives were saved. Researchers around the world heard reports—both censored and at the same time dramatized—of what happened that day in China. Everybody who felt the least bit optimistic about the ability of science to predict fault failures was extremely keen to learn more. Several foreign delegations were dispatched to study China’s success.

  A quick seven months later the National Academy of Sciences in the United States would publish a massive study entitled Earthquake Prediction and Public Policy suggesting that forecasting should be the “highest priority” because it could clearly save lives. The panel of experts who wrote the report took strong issue with the politicians and the few scientists who thought predictions and warnings might cause panic and economic disruption resulting in more harm than the temblors themselves. The Russians were already quite advanced in their own prediction studies, and the Japanese had launched the first phase of a large-scale prediction research program in 1965. China’s reported success story injected a surge of scientific adrenalin into the hearts and minds of those who saw prediction as the holy grail of the new seismology.

  A vocal few doubted China’s claims, coming as they did at the height of the political turmoil of the Cultural Revolution. Skeptics assumed the claims were at the very least embellished, if not complete hokum based on voodoo science and luck. Chinese experts were slow to publish their findings in the open literature in part because “leaking secrets” to foreigners was a criminal offense, so there was precious little in the way of written documentation for the rest of the world to study. Until things changed in China, outsiders who believed in prediction would simply have to take the Yingkou–Haicheng story on faith.

  Cao was a young carpenter who learned to read and write only after he joined the People’s Liberation Army in 1947. He fought in China’s civil war in 1949, retired from the army in 1954, and was doing “Party work” when he was assigned to help establish the Yingkou County Earthquake Office in September 1974. He lived in the town of Dashiqiao, where the county government was located.

  China had launched a seismic prediction program in 1966 with 17 major centers nationwide, 250 seismic stations, and 5,000 observation points. Guided by scientists, more than 100,000 citizen observers were collecting data. Cao worked enthusiastically at the heart of this campaign and took his duties very seriously.

  In keeping with Chairman Mao’s ideology and his distrust of elites—including, to some extent, scientists—these citizen observers became the backbone of the program. While trained experts monitored seismographs and carefully surveyed markers across fault lines to track the build-up of stress in the rocks, geomagnetic anomalies, and the release of radon gas, “the masses,” led by men like Cao, kept track of the fluctuations in well water and sometimes the strange behavior of animals, which according to tradition were symptoms of a coming disaster. A few local citizens were trained to measure basic changes in electrical currents in the ground.

  After three centuries of relative calm, northern China had been rattled by a series of three large temblors between 1966 and 1969. While geologists in the West were just coming to terms with plate tectonics, earth scientists in China had not yet accepted the theory (and would not until after the Cultural Revolution), so it was only later that they realized the northeastern part of the country was being squeezed from the east and west by colliding plates in the Himalayas and the ongoing subduction of the ocean floor along the Japan–Kuril trench.

  On June 29, 1974, the State Seismological Bureau issued a warning that earthquakes might occur within the next two years in an area that included the Yingkou area, home of Cao Xianqing. This was based on a torrent of new data that showed ground movement along the Jinzhou fault. There were als
o reports (never confirmed) of a rise in sea level in the area.

  By mid-December things began to accelerate. Reports of radon gas in the wells emerged, along with changing water levels. Then snakes started crawling out of their hibernation dens, freezing to death on the snowy ground. And the ground had begun to shake. After a swarm of small tremors there came a series of high alerts predicting magnitude 4–5 shocks over the weeks ahead.

  Less than a month later, at the next national quake prediction conference in Beijing, Gu Haoding, a seismologist from Liaoning’s provincial Earthquake Office, drew attention to the short-line leveling data across the Jinzhou fracture and said that “rocks near the fault had reached the unstable stage of plastic deformation” and were on the verge of rupture. “Therefore, a relatively large earthquake will not be very far and should be in the first half of this year or even January and February.”

  Based on this leveling data from the Jinxian Observatory—in one case a 0.1-inch (2.7 mm) elevation change over a period of only ten days—plus the radon gas reports, a series of magnetic anomalies, and a rash of smaller tremors, Gu used an empirical formula to predict that the coming quake would be a magnitude 6 and would occur somewhere around the southern tip of the Liaodong Peninsula. Because radon gas emissions were reported as far inland as Panjin and other anomalies were indicated in Dandong and Gaixian, Gu decided to expand the potential danger zone to include the entire Liaodong Peninsula and its offshore regions. When challenged about his very short time estimate—within six months, although possibly as early as January or February— Gu dug in his heels and said it could happen “even before the end of this conference.”