Cascadia's Fault

by Jerry Thompson

  Woodward-Clyde, meantime, had gone ahead, drawn its own conclusions and submitted its final draft of the safety analysis to the Nuclear Regulatory Commission, suggesting that Cascadia might indeed be a problem. Thus the tone of that next meeting was fairly poisonous, according to Schwartz’s memory. Two rival groups of engineers faced off against each other with their client—representatives from the Washington Public Power Supply System—clinging to a ringside seat.

  “Looking back at this, I have to appreciate the chutzpah they had in coming to our tank and trashing us in front of the client,” said Schwartz. “They simply said we were wrong in every assumption we made that the subduction zone could produce large earthquakes, that they could demonstrate that this wasn’t the case, that mention of even the possibility of seismic subduction would kill the project with the NRC.”

  The implication seemed to be that some other company should take over the entire project (removing Woodward-Clyde from the equation) and “obviously this didn’t go down well with us,” Schwartz continued. “It turned into a year of hell working with those guys!” In 1982 the NRC decided to call in outside consultants to help resolve the issue. They contacted a young geophysicist named Tom Heaton at Caltech and asked him to review the earlier preliminary safety report, the one concluding that Cascadia was not likely to generate large quakes.

  Heaton told me that after reading the first, more optimistic WPPSS document, he concluded that “the reasoning was weak” and decided to consult a more senior Caltech professor to help him write a response. Dr. Hiroo Kanamori had already done extensive studies of other subduction zones around the world. Kanamori was, in effect, Heaton’s mentor and together they had just completed a draft of the paper that would eventually be credited with turning the tide against the aseismic subduction hypothesis.

  In essence Heaton and Kanamori compared Cascadia (back then it was still being referred to as the Juan de Fuca Subduction Zone) to all the other 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. Bottom line: if giant ruptures could happen there—in Chile, Alaska, or Japan—the same would probably happen here, in the Pacific Northwest. Their paper, published in the Bulletin of the Seismological Society of America in June 1984, argued that the absence of quakes in recent history didn’t mean they wouldn’t happen in the future.

  Like the other seismic danger zones around the Pacific Rim, Juan de Fuca’s spreading ridge was so close to the edge of the continent that the new slab of ocean floor pumped out from the bowels of the earth was still relatively warm, thin, and buoyant. Put another way, the seafloor plate had not had much time to cool before it got jammed underneath the landmass of North America. Because it was warm and buoyant, the plate was in all likelihood scraping under the continent at a very shallow, almost horizontal angle. And because it was relatively smooth it was probably sticking to the upper plate the same way wide, flat-surfaced tires known as racing slicks build friction and stick to pavement.

  Heaton and Kanamori found that the biggest megathrust events were directly related to young, buoyant plates being strongly coupled to the overlying landmass at shallow angles—which fit the description of the Juan de Fuca Subduction Zone perfectly. All the other extreme danger zones had shallow trenches full of thick sediments that had been piled up against the outer coast and compressed, causing folds, faults, and uplift of the overriding plate. Same with Juan de Fuca.

  They found that weakly coupled subduction zones like the Marianas Trench had much steeper angles of dip and much deeper trenches, which caused the down-going oceanic plate to melt sooner and therefore not get stuck to the upper plate. Without being “strongly coupled” to the upper plate, the subduction process does occur smoothly, creeping along aseismically, with no big jolts. But the Juan de Fuca Subduction Zone was nothing like the Marianas Trench. Therefore, Juan de Fuca was probably not aseismic.

  At least that’s how Heaton and Kanamori saw it. Compared to all the other places around the Ring of Fire, the Juan de Fuca zone looked just as capable of deadly temblors as the worst of the lot. “This 500-kilometre gap in seismic activity is one of the most remarkable to be found anywhere in the circum-Pacific seismic belt,” they wrote. Their concluding paragraph must have sounded like a call to arms for other geologists: “The Juan de Fuca and North American plates appear to be converging at a rate of between 3 and 4 cm/yr. The Juan de Fuca subduction zone shares many features with other subduction zones that are strongly coupled and capable of producing very large earthquakes. Although the shallow part of this subduction zone shows little present-day seismicity and no significant historical activity, we feel that there is sufficient evidence to warrant further study of the possibility of a great subduction zone earthquake in the Pacific Northwest.”

  The problem, as Tom Heaton explained it to me, was that he did not have direct physical evidence of earthquakes. All the comparison studies in the world could not prove unequivocally that Cascadia’s fault had ruptured in the past. The compression evidence from Jim Savage’s survey markers near Seattle, like those squeezed-together mountaintops Herb Dragert was studying on Vancouver Island, were symptoms of stress, not proof that big quakes had already happened here. And that’s what Heaton was urging others to find—the proof.

  CHAPTER 11

  Quake Hunters: Finding Cascadia’s Ghost Forest

  Fog drifted over the hissing surf and followed us as we paddled in tandem, two canoes across a swollen winter tide up the Copalis River toward a grove of trees that had no business being where they were. In the distance, beyond the glassy green water, stood a thick fringe of marsh grass the color of wheat. Behind that, on higher ground, the dark outlines of heavily timbered hills.

  As we got closer I guessed the marsh grass might be waist deep if the ground were solid enough to hold a human’s weight. The tan thatch might have been the river’s opposite bank in some previous lifetime. Now it was a saltwater maze with a Medusa-like braid of channels meandering through and around it half a dozen different ways to the sea.

  The tide marsh looked like a big unmown pasture with these hulking dead cedars scattered across the flat ground, standing in defiance of time, weather, and gravity. How could fully grown cedars several centuries old be standing in knee-deep salt water, their storm-battered trunks naked of bark, bleached gray by the sun and draped in moss and lichens, in the middle of a tidal marsh on Washington’s west coast? Western red cedars don’t grow in salt water.

  One of two things could explain the wrongness of this picture. Either the level of the sea had risen far enough to drown the trees or the land had dropped, slumping far enough below mean high tide to turn a forest meadow into a salt marsh. The two men in the canoe up ahead were about to show us how they had bored and scraped and dug the answer from the damp coastal muck.

  Decked out in gumboots, faded orange rain gear, and a green life jacket that was never zipped shut and wearing a bright red tuque against the cool ocean mist, paleogeologist Brian Atwater had packed a folding army shovel and his trusty old Grumman canoe to give us a first-hand look at the Washington coast evidence of monster quakes. The dead cedars had become a vital clue in the ongoing mystery of Cascadia’s fault. Atwater’s colleague David Yamaguchi, from the University of Washington in Seattle, led the investigation that established the time of death. Together they were a forensic team worthy of their own CSI spinoff.

  Atwater’s discovery of this ghost forest on the Copalis River in March 1986 did not come about by accident. He’d been driving to the coast at every opportunity for months, specifically in search of proof that subduction earthquakes had left their marks on the Washington shoreline. His journey through the tentacles of saltchuck, sand, and river mud had begun in Seattle in October 1985 when his employer, the U.S. Geological Survey, organized a seminar on seismic hazards in the Pacific Northwest and invited all those doing active research in the region to attend.

&
nbsp; If the two plates were sliding past each other smoothly, at a constant rate, and without getting stuck together, then according to Ando and Balazs there should be a slow, continuous, and irreversible rise in land levels along the outer coast. And that was something Atwater figured he could probably measure and verify—or disprove. It sounded like an interesting research project.

  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 back to the west, toward its original shape.

  But the clear signal—the geodetic fingerprint—of each individual earthquake would be the abrupt lowering of land behind the beaches when the upper plate got stretched like taffy and then sank below the tide line as the upper plate snapped back to the west. That’s precisely what George Plafker had found in Alaska and Chile. It’s also what John Adams thought the mountain-tilting data predicted for Cascadia.

  Although journalists and members of the public who attended the Seattle meeting saw little of the ferment and discord behind the scenes, 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,” Atwater confirmed. “There were people saying, ‘Oh, there’s too much sediment. And there’s too much water. The zone has all this high-pressure water that’s keeping the fault from sticking tightly.’ Or people said, ‘Oh, it’s too warm to give you big earthquakes.’ And then there was quite a bit of discussion about the geodetic evidence of the time—that mountaintops were moving closer together in the Seattle area.”

  The key study by Jim Savage and his colleagues was hotly debated. Did the data really prove that mountain peaks on opposite sides of Puget Sound were being squeezed closer together? The increments of movement measured were quite small and the Geodolite—the best available technology at the time—was being pushed to the limits of its accuracy. Was the signal Savage detected “robust,” or was it just a few spikes in the white noise?

  “Savage concluded that the mountaintops were probably moving closer together in the direction of plate convergence,” Atwater said, “and he took that as evidence that the subduction zone really is locked. Some people dismissed it. And others said, ‘No, it’s a real signal.’”

  Atwater told me he decided to focus personally on the vertical motion predicted by Ando. “When they said the Pacific Coast was rising three millimeters a year relative to Puget Sound, I said, ‘Ah ha! Three meters per thousand!’ I think: You know, those are large amounts—a thirty-foot difference in the level of the shoreline.” That was the kind of motion he thought he’d be able to see in the geology. He would go out to the coast and find out whether a three-thousand-year-old shoreline was now thirty feet (9 m) above sea level, simple as that.

  Making the debate more interesting, however, were the seismic data that the University of Washington’s own Robert Crosson had published more than a decade earlier, in 1972. His fault plane solutions for small to moderate temblors occurring in the upper plate near Seattle suggested a north–south compression—directly contradicting Savage’s east–west motion. Crosson had concluded that the eastward subduction of the Juan de Fuca plate had stopped.

  Although not as widely circulated, there was also the January 1983 PhD thesis of Garry Rogers, who would go on to become a leading seismologist at the Pacific Geoscience Centre in Sidney, British Columbia. Rogers had constructed a theoretical model to explain how there could be north–south ruptures (roughly parallel to the Washington coastline) and east–west compression at the same time. He noted that the underlying oceanic plate seemed to be deforming as it turned a corner beneath Puget Sound. Instead of continuing toward the east as it had for millions of years, the Juan de Fuca plate had apparently started rotating at an oblique angle to the coast, shifting to a northeasterly movement. As it turned, it created a bulge in the overlying plate that became the Olympic Mountains in Washington State. Stress in the lower plate increased until the rock began to collapse in on itself. This partial collapse of the oceanic slab did not, however, relieve all of the stress caused by making the turn. Resistance to movement along the fault plane caused rocks to shear in some places. Imagine a deck of cards spread horizontally by the force of your hands. Once the stress exceeded the strength of the rock, it fractured in cracks that ran parallel to the coast: the north–south fault planes that Crosson had detected.

  Meantime, the continued movement of the lower plate perpendicular to the coast would still cause northeasterly compression, and either the strain would be stored in the rocks until they failed in very large earthquakes or the plates would creep past each other aseismically. The trouble for Rogers and the others who wanted to see this question answered once and for all was that the seismic data seemed to suggest the aseismic option. No earthquakes in recorded history.

  The only evidence of horizontal strain building up had come from Jim Savage and company. Until further geodetic measurements were completed—a parallel study was already underway by Herb Dragert and his team on Vancouver Island—there would be no resolution. “In the meantime,” wrote Rogers in the final line of his cautious conclusion, “large subduction zone earthquakes must be considered a possibility.”

  Then, to complicate the contradictions, new data were released later in 1983 from Craig Weaver and Stewart Smith’s new seismic array, installed after the Mount St. Helens eruption. It revealed a fracture zone in which “maximum compression is northeast, approximately parallel with the direction of plate convergence.” Like the Savage data, this was interpreted as “evidence for locked subduction.” It seemed a tipping point might soon be reached.

  But it was Tom Heaton’s talk at the USGS conference in Seattle about the bigger picture—Cascadia’s similarities with other dangerous subduction zones—that drew sparks and generated most of the attention. “He was out there voicing some views that were somewhat unpopular,” said Atwater. “It did serve as something of a lightning rod.” Thus the quest for proof, for convincing evidence, for resolution, became all the more enticing.

  The conference made news all over the Pacific Northwest. At the Pacific Geoscience Centre about a week later, when the CBC camera crew and I showed up to shoot our very first Cascadia documentary, Dieter Weichert, who was running the seismology lab at the time, took the opportunity to go public with his own bold statement. He stuck his neck out and told us on camera that Canada’s team of experts, just returned from Mexico City, had decided the possibility of a great subduction earthquake was real.

  The Geological Survey of Canada was revising the risk assessment for Vancouver and Victoria and the west side of Vancouver Island from fifty–fifty odds of a magnitude 9 disaster, to seventy–thirty in favor of a Mexico City- or Alaska- or Chile-type quake.

  Brian Atwater drove west from Seattle in March 1986 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. “I went to Neah Bay with Ando and Balazs firmly in mind,” he admitted. “I really went out there looking for elevated shorelines.” That’s not at all what he found.

  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, no matter which way they went—up or down. To the north, the bay opens onto the Strait of Juan de Fuca. Across the Strait is Canada’s Vancouver Island. To the south a narrow green valley runs from the back of the bay all the way down to the Pacific Ocean beaches. Standing there in the cold mist and rain it was easy to imagine how seawater might have filled the entire valley at some point in the past, connecting the Pacific with the Strait, cutting off Cape Flat
tery from the rest of the Olympic Peninsula and making it an island.

  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. No need for an earthquake or even for the chronic uplift of Ando and Balazs to explain this stuff. 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 had studied in San Francisco Bay: 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 salt water. This subsidence of the landscape had apparently happened very quickly because 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. Which is why the plants were still recognizable hundreds of years later.

  This was no gentle or gradual transition zone from one geologic era to another. The peat had a sharp upper boundary caused by an almost instantaneous and probably cataclysmic change in the level of the land and sea. Was it physical proof that the ground had slumped during an earthquake, that the plants of a marsh or forest meadow had been drowned quite suddenly by the incoming tides and possibly buried under the sands of a huge tsunami? Could this finally be Cascadia’s real smoking gun?

  Geologists would need to ask this question at many more places than Neah Bay to know for sure. A few samples at a single bay would not be enough to prove the case. But Atwater remained there long enough to jot a note.