Cascadia's Fault

by Jerry Thompson

  Brian Atwater is fond of saying that “the earthquakes had written their own history” by dropping segments of the coastline a meter or two, by forcing sandy seawater across the freshly sunken land, and by causing the edge of the earth to crack. The problem was that by the mid-1990s researchers had reached an impasse in their attempts to define how large the quakes had been. The answer to this question was important because emergency planners and civil engineers needed to know what they were up against when it came to designing new buildings and reinforcing old ones. “How great an earthquake should a school or hospital be designed to withstand? How large a tsunami should govern evacuation plans on the coast?” asked Atwater in a book he cowrote several years later.

  Atwater’s colleague David Yamaguchi had come closer than anyone else to nailing a more specific date. He went back to the ghost forest at Copalis River several times in search of wood cores that might contain enough growth rings to reveal what year the trees had died. He put together a database, a pattern of rings collected from twenty-three ancient cedars that had been alive at the time of Cascadia’s last violent outburst. Like the Native people who lived above Pachena Bay on Vancouver Island, the trees of Washington State stood on slightly higher ground and had witnessed the disaster unfolding below.

  When they compared the rings of “witness trees” with rings from the ghost forest, they knew the forest floor had sunk beneath the tides some time in the late seventeenth century. “The closest I could get was 1691 as a limiting date for the earthquake,” Yamaguchi told me. “The earthquake had happened some time shortly after then, but I didn’t know how many years afterward.” Insects, rot, and roughly three hundred North Pacific winters had pitted and peeled away too much of the outer bark of these standing dead cedars. The final few years of growth rings—the crucial last clues—were missing.

  A dating technology was being tested at the University of Washington, however, and it might pick up where the tree rings left off. Atwater was keen to try it. “Brian realized that he could use the preliminary 1691 date as a tool to help him do something called high-precision radiocarbon dating,” Yamaguchi continued. Wood samples were locked for weeks inside shielded atom counters and the resulting timeline was accurate to “the 95 percent confidence interval.”

  Like the weather-beaten tree trunks, though, radiocarbon technology fell slightly short of the mark. It moved the goalposts closer together without truly scoring a victory. The window had shrunk a little more, putting Cascadia’s most recent event some time between 1690 and 1720, but there was still no certainty about whether a single, giant rupture or a “swift series of merely great” earthquakes had done the job. Nevertheless, when news of this tighter timeline became public, it created a buzz. And a scientist from Japan had one of those eureka moments that researchers everywhere dream of.

  CHAPTER 17

  The Orphan Tsunami: Final Proof of Cascadia’s Last Rupture

  At a science conference in Marshall, California, in September 1994, Kenji Satake was having lunch with Alan Nelson of the USGS when Nelson mentioned the frustration he and others were having with radiocarbon dating. Nelson and eleven colleagues were putting the final touches on a paper that was meant to collect and focus all the physical evidence for past quakes on Cascadia’s fault in a single volume. Satake’s specialty was subduction zones and tsunamis. He had studied the effects of the Okushiri disaster in Japan the previous year and was very interested to hear the latest news about Cascadia. Especially Atwater and Yamaguchi’s work with tree rings.

  “I did not know much about the calibration technique of radiocarbon dating using tree-ring records,” Satake told me. “I met Alan. He explained to me the basics of the technique and I was very much impressed by the small uncertainty—only a few decades.” Something about those dates—a time span from 1690 to 1720 for Cascadia’s last rupture—caused Satake to pay special attention to this latest study.

  He was working at the University of Michigan as an assistant professor. Before that he’d completed a two-year term as postdoctoral researcher at Caltech, where he had studied Cascadia’s quake and tsunami potential with Tom Heaton. Heaton, Hiroo Kanamori, and Stephen Hartzell were then banging the drum about similarities between Cascadia and Chile and other big subduction zones. But in 1994 there was still not enough unequivocal evidence of past ruptures in the Pacific Northwest for the case to be ironclad.

  Later that same year Satake made a point of meeting Brian Atwater on a field trip to the coast as part of the Geological Society of America’s fall meeting. Satake would soon depart the United States for a job with the Geological Survey of Japan and was keen to see the Cascadia evidence first-hand before he left. Curiosity piqued by Atwater, Satake wanted to see more.

  “He spent some time with me, down in Humboldt County,” confirmed Gary Carver. His recollection of events was that Satake zeroed in right away on the dates. “He latched on to that 1700 plus or minus a few years—ten years was the range—that we thought was statistically important.” But what about 1700 stuck out in Satake’s mind? He had a hunch and followed it all the way home.

  When I met Kenji Satake in a small city on the southeast coast of Japan several years later to interview him, he explained why the date had jumped out at him. “If the earthquake happened in Cascadia, and if it was large enough, it must have generated a tsunami which would propagate across the Pacific. And 1700 is not very long ago for Japan. So if such a tsunami arrived in Japan and caused damage that must have been documented in historic documents.”

  No such written record existed in North America. The only cultural evidence of a huge earthquake around 1700 would be the oral histories of the Yurok or the Makah or the Huu-ay-aht people of Pachena Bay. And their stories did not include exact dates. By then a few European explorers had come and gone but the rest of the world knew little about the northwest coast of America and nothing at all about its earthquakes. Even the geography was a yellowing parchment void on most existing maps. The Spanish had outposts in Chile, Peru, Mexico, and southern California, but Cascadia was still terra incognita.

  By comparison the Japan of 1700 already had a long history of writing things down. The damage reports of earthquakes, tsunamis, and volcanic eruptions were usually kept at temples or in the ledgers of local merchants and public officials. When Kenji Satake returned in 1995, he started looking for catalogs compiled by various teachers and scholars who had been working on a history of ruptures and tsunamis going back more than a thousand years. He knew that some of the tsunamis recorded would have no earthquake listed alongside them because the plate motion that caused the waves would have happened so far away—in places like Chile or Alaska or Siberia’s Kamchatka Peninsula—that local residents would not have felt the shaking. Without “parent” earthquakes, the “orphan” tsunamis should stand out and be relatively easy to spot in the seismologists’ catalog.

  He approached Kunihiko Shimazaki, Yoshinobu Tsuji, and Kazue Ueda, colleagues who had begun studying Japan’s historical earthquakes and tsunamis in the 1970s. Together they searched backward chronologically through the list. “Soon after we started looking,” Satake said, “we found that there were some documents describing a tsunami from an unknown origin in the year 1700.” A series of waves swept down the Japanese coast after midnight on January 27 and into the early morning hours of January 28, to be exact.

  “The people studying the history of Japanese earthquakes—for them it was a strange event. There wasn’t an earthquake, so they didn’t care,” Satake explained. “They didn’t care what the origin was because they were interested only in Japanese earthquakes—and this wasn’t a Japanese earthquake.”

  While this may have seemed like a eureka moment, plenty of work remained before Satake could say for sure this wave of unknown origin had come from Cascadia. Perhaps it was caused by a typhoon or some other big storm. Maybe the orphan wave had come from a rupture in Alaska or Chile. So Satake and his colleagues quickly moved to the next stage of their inve
stigation.

  Fairly easily they ruled out the storm surge idea. “We knew that the tsunami waves were documented in many places along the Japanese coast,” Satake continued, pointing to a map that showed nearly 560 miles (900 km) of affected shoreline, “which is too large an area—too wide—for a meteorological origin.” Add to this the fact that most typhoons hit Japan between August and October. These mystery waves had arrived in the dead of winter. “We examined the weather of that day and we found that the weather wasn’t that bad,” he said, with sunny or cloudy skies reported in most of central and northern Japan on January 27, 1700.

  Having ruled out storms, they were convinced it had to be a tsunami and it must have come from across the Pacific. So they examined another catalog of historical temblors that listed the ones from South America. “What we found was there were earthquakes before and after 1700,” remarked Satake. “For example, in 1687 there was an earthquake recorded in South America and a tsunami from this event was also recorded in Japan. And similarly in 1730, we have an earthquake in South America and following that a tsunami was also documented in Japan. But there’s no earthquake in 1700. So we could rule out the possibility for South American origin.”

  Judging by the reported height of the waves in Japan, the orphan tsunami was too big to have come from Alaska, the Aleutians, or the Kamchatka Peninsula. The way the plate boundaries line up, those subduction zones are almost parallel to Japan rather than perpendicular to it. So movements along those faults generate waves with primary energy vectors that usually miss Japan. The 1964 event in Alaska, for example, sent large and destructive waves to Vancouver Island and California, but the side-angle waves that hit Japan were less than a foot (0.3 m) above the tide.

  To make the much larger waves (some as high as sixteen feet, or 5 m) that ran up the beaches of Japan in 1700, a quake in Alaska or Kamchatka would have had to be magnitude 9 or larger. But there were no written records or geologic evidence for a megathrust event of that size in either region in January of that year. Therefore, by process of elimination, Satake and his colleagues concluded that Cascadia was “the most likely source” of the orphan tsunami.

  To estimate the magnitude of the parent earthquake, Satake compared the damage reports from the orphan tsunami with those from the 1960 Chile wave, which killed 140 people in Japan, and found the wave heights to be similar. To create a tsunami as big as the one that came ashore in 1700, Satake and his team figured the Cascadia earthquake must have been at least magnitude 9 or larger. From orphan tsunami to parent quake, these geologic sleuths had traced the last big failure of Cascadia’s fault. The next step was to figure out the exact time it happened.

  They knew from other research how fast a tsunami could move. “The deep-ocean speed is approximately the same as a jetliner,” Satake told me, “so it would have taken about nine or ten hours from the West Coast to Japan.” Next they took the confirmed arrival times of waves hitting the beach in Japan and calculated backward to figure out when the earthquake must have triggered the tsunami.

  “The earliest documented tsunami arrival time was around midnight on 27 January, Japan time,” wrote Satake and his colleagues in Nature. “Because tsunami travel time from Cascadia to Japan is about 10 hours, the earthquake origin time is estimated at around 5:00 on 27 January GMT or 21:00 on 26 January local time in Cascadia. This time is consistent with Native American legends that an earthquake occurred on a winter night.”

  And so the conceptual or philosophical barrier was finally breached. It would no longer be possible to argue that Cascadia was probably harmless. The doubters could not point to an absence of great earthquakes in “all of recorded history” and make their case for aseismic subduction. Thanks to a handful of local officials in Japan, the written history of North America’s west coast had been extended back more than three centuries—far enough to reveal a new and ominous picture of this hidden plate boundary.

  The evidence that Cascadia’s subduction zone had generated earthquakes big enough to send damaging waves all the way to Japan seemed quite convincing. All of Brian Atwater’s tenacious digging in the mud, together with Kenji Satake’s eureka moment and deductive reasoning, had paid off. “We were really very excited to find this evidence,” Satake said with a modest smile.

  “It used to be that we would say that Cascadia made an earthquake and tsunami, or a series of earthquakes and tsunamis, about three hundred years ago,” Atwater began to tell coastal audiences at tsunami-safety workshops. “Now, we can say that at 9:00 p.m. on the 26th of January in 1700 it made one giant earthquake and a trans-Pacific tsunami. That gives this history a more definite feel than it used to have.”

  Like diligent researchers everywhere, though, they knew the quest for confirmation was not over. Convincing and logical evidence they had. But was there any other way to explain the data? Could any part of their hypothesis be falsified?

  Satake sent a draft of his paper to half a dozen other researchers for reviews. His fellow scientists made suggestions for fine-tuning but gave it a thumbs-up. The editors of Nature liked what they saw enough to publish the breakthrough finding on December 5, 1995. The next phase of the investigation would be a thorough search for the original documents written by samurai warriors, local merchants, and public officials in villages along Japan’s east coast.

  The story of how Brian Atwater and Kenji Satake found a way to work together brings a grin to my face. When filming the story of the orphan tsunami, I got caught up in the blind alleys of a detective story, fascinated by the logic and niggling details, constantly asking myself—how the heck did these guys figure this out? Along the way I forgot to ask how they tackled the research from such different scientific and cultural perspectives while separated by thousands of miles of North Pacific Ocean. Satake dashed off a quick email that gave me the context and the smile.

  “As a traditional and sometimes stubborn geologist,” Kenji wrote of Brian, “he would not be convinced until he examined the data by himself. But the ‘data’ in this case are Japanese historical documents, not coastal geologic sections [in which] he was specialized. A normal person would not step further. Brian, however, was different. He first attended a class at University of Washington to learn Japanese. Obviously, learning a new language at an age of nearly 50 is not an easy task, and Japanese is not an easy language for westerners . . . He spent about a year in Japan [1998–99] to study the historical documents and the social and historical backgrounds of the documents. Our goal was to write a book for the U.S. general audience to introduce the Japanese documents recording the Cascadia tsunami.”

  Satake, Atwater, and their colleagues reexamined original documents in the six main villages and towns that had recorded the tsunami’s progress down Japan’s eastern seaboard. The waters arrived late at night on the north coast and worked their way south as the sun rose. In Kuwagasaki, a seaport village with roughly three hundred houses in 1700 where Cascadia’s wave first made Japanese landfall, the midnight flood and ensuing fire destroyed one-tenth of the homes. As villagers fled to higher ground the frigid water wrecked thirteen homes outright and set off a blaze that burned twenty more. The entry was written by a local magistrate, who noted the exact arrival time of the incoming tsunami.

  At the south end of Miyako Bay the water washed away houses along the shore and then entered Tsugaruishi village just over half a mile (1 km) inland, where it caused panic among the residents and sparked fires that burned another twenty-one homes. An independent report written by a local merchant in his family’s notebook mentioned that there was no ground shaking—which must have seemed odd since this damaging mound of water looked so much like a tsunami.

  From Tsugaruishi the wave continued another half mile inland and across a floodplain to a place called Kubota Crossing, and on to the foot of a hill that has long been capped by a shrine to the Shinto god Inari. This put the high-water mark for the 1700 tsunami at roughly the same place as the train of waves that came from Chile in 1960. That
means Cascadia’s wave was about sixteen feet (5 m) high when it crashed to shore.

  Farther down the coast at the river port of Nakaminato, around eight o’clock on the morning of January 28, high waves—probably strong outflow currents from the ebbing tsunami, churning against ordinary ocean swells at the river’s mouth—held a cargo vessel offshore and prevented it from entering port. From nearby Miho came a report that “something like a very high tide” swept ashore seven times between dawn and ten o’clock that morning. At the end of day the vessel was still offshore when a storm blew up, broke the anchor lines, drove the boat against a rocky beach, and sank it.

  A cargo of thirty tons of rice, bound for the capital in Edo, was destroyed and two crewmen were killed. In what amounts to a police report, the boat’s captain and two villagers wrote descriptions of the sinking and petitioned a district magistrate to certify the circumstances of the accident that ruined 470 bales of rice thought to be owned by two samurai from the north coast. The official port certificate absolving the crew of responsibility for the Nakaminato shipwreck of January 28, 1700, was among the documents dug up and examined by Satake, Atwater, and their colleagues.

  One village headman, writing his eyewitness account, described seven surges of water that moved “with the speed of a big river.” He took the precaution of sending elders and children to the safety of higher ground at a local shrine and then consulted village elders about the puzzling lack of an earthquake.