Cascadia's Fault Read online

  At 8:32 a.m. on Sunday, May 18, 1980, the volcano started coming apart. A magnitude 5.1 earthquake caused the bulging north side of the mountain to collapse where a new lava dome had been growing. The collapse caused the largest landslide of rock and ice and volcanic mud ever recorded in the continental United States—9,600,000 cubic yards (7,340,000 m3) of boulders, muck, trees, and other debris was swept 17 miles (27 km) downhill into the Columbia River. With the face of the mountain suddenly exposed to cool air, the volcano exploded, flattening or burying more than 230 square miles (595 km2) of forest and farmland under a blanket of mud and ash that shot 12 to 16 miles (19–26 km) into the sky.

  Even though most residents had fled the area, the explosion still killed 57 people, destroyed or severely damaged more than 250 homes and businesses, wrecked 185 miles (298 km) of highway and 15 miles (25 km) of railway track, punched out 47 bridges, and killed more than 7,000 big game animals (deer, elk, and bear) and an estimated 12 million fish at a nearby hatchery.

  Before the eruption Mount St. Helens had a spectacular, nearly symmetrical, cone-shaped peak that stood 9,677 feet (2,950 m) high—a stratovolcano. It collected 140 inches (356 cm) of rain and up to 16 feet (5 m) of snow every year, making it look a lot like those famous pictures of Mount Fuji in Japan. After the explosion, the top thousand feet of the mountain had vanished, leaving a horseshoe-shaped crater two miles wide and a half-mile deep (3.2 km by 0.8 km). As the eruption continued for nine hours, the ash plume drifted east at an estimated 60 miles (100 km) per hour, dumping a thick layer of abrasive grit across eastern Washington and Oregon, coating cars as far north as Edmonton, Alberta, as far east as the Dakotas, and as far southeast as Colorado and New Mexico.

  The first warning signs had come as early as March 20, when a mild tremor (magnitude 4.2) rattled the mountain. Steam vents began to spew a week later. By the second week of April, scientists had alerted the media. Walter Sullivan of the New York Times touched on the explanation of Mount St. Helens’ deep tectonic origin in his story “The West Is Alive with the Sound of Volcanoes.”

  The violence of volcanoes like this, according to the Times story, was a direct result of the collision of North America with the Pacific Ocean floor. It was the Juan de Fuca and Gorda plates grinding down along Cascadia’s fault that had created the Cascade Arc of eighteen major volcanoes from Mount Shasta and Lassen Peak in California to Mount Adams and Mount St. Helens near Portland, to Mount Rainier near Seattle, Mount Baker, about fifteen miles (25 km) south of the Canada–U.S. border, and to Mount Garibaldi, north of Vancouver. Sullivan’s feature explained what would probably happen—and why—a full month before Mount St. Helens blew: “The Cascades, part of the Pacific Ocean’s necklace of volcanoes, its ‘ring of fire,’ are the product of ‘sea floor subduction’ . . . Typically, the sea bed bends down as it nears a continent, forming a trench. A sloping zone of earthquakes marks its path into the earth’s interior. When the sea floor slab reaches a depth of about 75 miles [120 km], part of it apparently melts and, lighter in weight than the overlying material, forces its way up to produce volcanoes.”

  When I thought about the timing and content of Sullivan’s story, it struck me that the eruption of Mount St. Helens should have been Cascadia’s smoking gun—clear, unequivocal, physical evidence that continental drift was real, that the Gorda and Juan de Fuca plates were on the move and dangerous. This should have been the big wake-up call or tipping point for everyone involved in the geophysical sciences and emergency preparedness, a stark statement that the coast of northern California, the Pacific Northwest, and southwestern British Columbia were every bit as threatened by megathrust subduction disasters as were the coasts of Alaska and Chile. I was wrong. For reasons unclear to me, the alarm bells did not ring.

  The Mount St. Helens disaster happened right in the middle of the investigation of those thrust faults that threatened the nuclear reactor at Humboldt Bay, yet it had no discernible impact on the reluctance of some scientists to accept the Cascadia subduction story. Experts at the top of their fields would still doubt the seismic potential of Cascadia for another eight or nine years.

  The spectacle of Mount St. Helens was riveting, no doubt about that. It was also a distraction, such a stunning assault on the senses that few, if any, stopped to think about what this eruption might mean in a larger perspective. Rereading Walter Sullivan’s story, I then found a clue to why some scientists were able to separate the volcano itself from the much wider threat a magnitude 9 megathrust quake spread across five major cities would pose.

  The scientists Sullivan consulted for his pre-eruption feature story had told him how the Cascadia Subduction Zone was thought to be a special case, different somehow from all the other continental collisions around the Pacific Rim. Cascadia’s volcanoes do form a line roughly parallel to the coast about a hundred miles (160 km) inland, and the mountain cones are spaced about forty-five miles (70 km) apart, like most other subduction zones with volcanic arcs. But in the minds of skeptics, that’s where the similarities ended.

  Cascadia is not typical, Sullivan wrote, because “no coastal trench cuts in to the sea floor” at the point where the two tectonic plates converge and “no sloping zone of earthquakes” marks the descent of a seafloor slab beneath the coast. According to Sullivan’s sources, the Cascade volcanoes seem to have been created by an east-driving portion of the Pacific floor that had somehow run out of steam. The subduction process, wrote Sullivan, “is no longer vigorous enough to sustain a coastal trench and cause frequent earthquakes.”

  So Cascadia’s smoking gun had run out of ammunition. No deep trench offshore and no deep quakes along the plate boundary—all because the movement of the eastbound Juan de Fuca plate had slowed down or even stopped. At least that’s what some experts thought at the time.

  Trying to imagine how these huge plates float and slide over the curved surface of an imperfectly spherical planet, geophysicists came up with a sequence of events—a long geologic history—that seemed to fit the observable facts. As the floor of the Pacific Ocean spread apart along the Juan de Fuca Ridge, pushing the Juan de Fuca plate eastward, the rest of the Pacific plate (out on the western side of the ridge) was not moving due west but rotating in a more northerly direction toward Alaska.

  At the same time, the North America plate was pushing westward and riding up over top of the eastbound Juan de Fuca slab. Around ten million years ago, so the theory went, the Juan de Fuca Ridge and plate started rotating clockwise, almost as if it were being spun by the angular movement of the two larger plates on either side of it. Think of a car going eastbound through an intersection when a northbound car passes just behind it, clipping the back fender. That northbound motion would make the eastbound car spin to the right, just as it got hit head on by a big westbound truck.

  Five million years later, the Explorer plate had broken off the northern end of the Juan de Fuca, and some thought it might have fused or welded itself to the larger continental plate just north of Vancouver Island. There was further speculation that the Olympic Peninsula, on the northwest corner of Washington State, might also have been a piece broken off the Juan de Fuca plate and that it too had been fused to the continent, pressed against the outer edge of Vancouver Island, forcing up the Olympic Mountains in the process.

  Two and a half million years later, on this hypothetical timeline, the North America plate had “disposed of ” (subducted or recycled) a huge portion of the original Juan de Fuca plate. Down at the southern end, meantime, the smaller Gorda plate was breaking away as well and the spreading ridge offshore—the entire undersea mountain range—had rotated or been spun even farther to the right. At some point, according to this scenario, the relentless westward movement of North America would completely override what was left of the Juan de Fuca plate—and its spreading ridge offshore—just as it had apparently already done farther south in California.

  Eventually, with the Juan de Fuca ridge and plate system gone, the boundary between th
e North America and Pacific plates would become a much simpler structure, an almost straight-line fault starting with the San Andreas in southern California, extending north along the coast, and connecting with the Queen Charlotte fault system all the way to Alaska. Then the dominant tectonic force in the Pacific Northwest would be a more straightforward, northerly compression caused by the northbound drift of the Pacific plate—just the way it is now along the San Andreas in southern and central California. The eastward subduction of the Juan de Fuca and Gorda plates underneath the continent would become ancient history and Cascadia’s tectonic threat would be rendered harmless.

  In October 1972, Robert Crosson, a seismologist and professor at the University of Washington, wrote a paper suggesting that this had probably already happened. Based on data from the Pacific Northwest Seismograph Network, a newly installed, six-station, high-sensitivity telemetry system capable of pinpointing even the smallest tremors, Crosson and his colleagues had shown that nearly all the jolts in the Puget Sound region around Seattle and Tacoma were the result of north–south compression. There was no evidence of eastward pressure from the Juan de Fuca plate at all, as far as they could tell.

  The vast majority of the recent Puget Sound earthquakes had been relatively shallow ruptures in the upper crust of the North American continental plate. The absence of a down-sloping zone of much deeper shocks (known as a Benioff zone) along the eastward-dipping plate boundary and the lack of any recent volcanic activity in the Cascades (in 1972) could be seen as further evidence that the Juan de Fuca plate had stopped moving or was in its final phase of subduction. Without that constant eastward shove from the Juan de Fuca Ridge, the dominant tectonic pressure would have become northerly—and that’s indeed what the new seismographic data in Washington State seemed to confirm.

  Just across the border in British Columbia, however, two scientists working for the Geological Survey of Canada had looked at their data and come to exactly the opposite conclusion. Robin Riddihough and Roy Hyndman argued in August 1976 that subduction was still happening. They pointed to the “significant eastward dip” of the ocean floor and to the layers of sediment—two and a half miles (4 km) thick, lying on top of the Juan de Fuca plate—that had been dragged sideways into a shallow, less obvious trench at the edge of the continental shelf where they were crumpled, folded, and fractured, all relatively recently.

  The continental shelf itself had been recently deformed and uplifted, just like those terraces (former beaches) hoisted up near Cape Mendocino in California. They cited a higher than normal heat flow from the inland Cascade Range that was probably caused by upwelling magma from the melting oceanic slab. All of these were classic symptoms of active subduction, according to Riddihough and Hyndman, although they agreed it would be hard to tell if and when a plate had stopped moving in the recent past. So there remained a degree of uncertainty about whether or not Cascadia posed a clear and present danger.

  Another possible explanation for the lack of large earthquakes came in June 1979, when Masataka Ando of the U.S. Geological Survey and Emery Balazs of the National Geodetic Survey suggested that the Juan de Fuca plate was still subducting, but doing it aseismically—without earthquakes. Given the lack of large ruptures over the 140 years since white settlers had arrived and written records had been kept, two things were possible. Silence along the boundary zone could either mean the two plates were now locked together by friction and strain was building up for a major rupture, or that big temblors simply didn’t happen in this subduction zone. Which brings us back to the idea that Cascadia is somehow a special case.

  “In some subduction zones, such large earthquakes do not occur,” wrote Ando and Balazs. They had a hunch that friction between the Juan de Fuca and North America plates was too low for the rocks to get stuck together. If, for whatever reason, they don’t get stuck—because of a slower than normal rate of motion, perhaps, or a shallow angle of subduction—then movement could keep happening without major quakes. Strain might build up enough to compress and bend rocks in the overlying plate and still not cause a rupture. And they figured the only way to find out for sure would be to measure the rate of deformation along the highways of Washington State.

  The first precise leveling survey of Washington’s roads had been done back in 1904. By 1974 new surveys had been carried out on ten different sections of highway, some of which ran east–west across the Coast Range mountains. In the time between the first and second surveys, the surveyors’ data showed that the outer coast had been lifted upward and the inland areas east of the Coast Mountains had subsided. In other words, the entire mountain range was tilting slightly toward the east and this had to be a result of active, ongoing subduction because it had happened in the past seventy years, not millions of years ago in geologic time.

  However, another important detail made Cascadia different, according to Ando, who had recently studied strain accumulation in the Shikoku area along the east coast of Japan. There the Philippine Sea plate is thrusting under the Asian plate—beneath the islands of Japan—along the Nankai Trough. The geologic setting is very similar to the Juan de Fuca Subduction Zone. The dip angle of both subducting plates is a shallow twenty degrees and both oceanic plates are relatively thin. The significant difference is that the outer coastal landmass in Japan is tilting down toward the ocean rather than leaning inland as it appears to be doing in Cascadia.

  Bending the outer edge of the coast downward as the ocean floor scrapes underneath it is a sure sign the plates are locked together by friction and building strain for a large quake, according to Ando’s analysis. Once the rocks along the locked portion reach their breaking point—when friction between them is no longer enough to keep the two plates stuck together—the strain is released in a massive shockwave. As the two plates rip apart in a typical or “normal” subduction zone like the Nankai Trough, the outer coast snaps free from the down-going oceanic plate and springs back upward. The area slightly inland from the coast subsides at the same time. This is exactly what happened in previous large quakes in Japan, Alaska, and Chile.

  In the aftermath of these giant jolts, as the overlying continental plate settled back down to its more or less normal position, some of the coastal uplift remained. In other words, the beach never quite got back to where it used to be because the underthrusting oceanic plate was still down there, still moving below the continent, still causing a certain amount of residual deformation. The three-step sequence, according to Ando and Balazs, starts with coastal down-warping just before the quake, followed by heaving upward during the rupture, and then a certain amount of residual uplift of the beach zones in the aftermath.

  Cascadia, however, seemed to be doing something entirely different. If the aseismic hypothesis were true, then the Juan de Fuca plate would be just creeping down underneath the continent, slowly and continuously, lifting and tilting the Coast Range mountains to the east, and doing so without getting completely stuck and without accumulating enough strain to cause a major rupture. To me this sounded like the good news. The bad news came in the concluding paragraphs.

  Studies of other aseismic zones had revealed that temblors are still possible even if the two plates are not completely locked together. Hiroo Kanamori at Caltech found that if you look at the total distance—how much long-term horizontal movement there had been along the subduction zone in the Kuril Islands, for example—and compared that to the movement that happened during large thrust earthquakes, the ruptures accounted for one-quarter of the total slip. In northern Japan another study showed that quakes accounted for one-tenth of the total subduction rate. Which could mean that even in a mostly aseismic zone—where 75 or even 90 percent of the plate motion is slow, smooth, quake-free creeping—the plates can still get locked together and megathrust events do eventually happen. So Cascadia is not completely off the hook for damages, even if the aseismic theory is true.

  The tip-off, according to Ando and Balazs, should occur whenever we see the outer coast of
the Pacific Northwest start to dip down and get pulled under by the Juan de Fuca plate. Not surprisingly, they recommended constant vigilance by a team of surveyors with state-of-the-art equipment to spot any change along the beach. In the meantime, because the Coast Mountains are now tilting eastward instead of toward the sea, they assured us that a large thrust earthquake from Cascadia’s fault is “not expected in the near future.”

  Then along came the eruption of Mount St. Helens one year later. How could a violent explosion like this not be the sign that convinces all and sundry that Cascadia is still very active and that a tectonic disaster is looming? I suppose the first and simplest explanation was that the debate about Cascadia as “a special case” was happening mostly within the confines of the science community. The general public was not reading these new technical papers, not attending the scientific meetings, and therefore they did not know, for the most part, that Cascadia’s fault even existed.

  But why did so many scientists who had read the new literature still hesitate?

  Gary Carver, who was still mapping thrust faults in the rumpled hills around the Humboldt Bay nuclear plant at the time Mount St. Helens blew, knew that the majority of scientists were skeptical of the Cascadia disaster scenario. Why would the volcanic eruption not have been seen as proof positive of active subduction? He told me that an eruption could still happen even after subduction had stopped.

  Presumably, if the Juan de Fuca plate stopped moving tomorrow, the segment of the down-going slab that had already been pulled toward the earth’s hot interior would have begun to melt. Plumes of magma would already be rising up beneath the arc of volcanic mountains. So there could be a lag of who knows how many years—hundreds, maybe thousands—between the end of Cascadia’s plate motion and the final eruption of Mount St. Helens or one of its neighbors. From that perspective, the St. Helens blast didn’t prove anything.