Cascadia's Fault Read online

  I didn’t know it at the time, but John Adams had already been studying this mystery for five years. He had published three papers on faults that lie deceptively silent for hundreds of years, including an alpine fracture zone in his home country of New Zealand. Now, in one long breath, he had just warned us that what we saw in Mexico City might also happen in the Pacific Northwest. And Adams wasn’t the only one to say it.

  Garry Rogers was the young PGC seismologist whose job it was to monitor the two hundred or so temblors that rattle through southwestern British Columbia every year (only two or three of which are strong enough to be felt). He told us the historical lack of huge megathrust events off the west coast of Vancouver Island could be very misleading.

  “The implication,” said Rogers with focused intensity, “is that the possibility for very large earthquakes—the kind that occurred in Mexico just recently—does exist on the west coast of Canada. The problem is that in the 150 short years that we’ve been here, we have not seen any examples of earthquakes on our subduction zone. Not even small ones.”

  He explained that those two hundred rumbles occur because of stress within the overlying crustal plate, relatively close to the surface. The much larger shock—if it does happen—would occur almost forty miles (65 km) below ground along the length of the subduction zone. Like Adams, Garry Rogers thought the absence of deep Juan de Fuca quakes put seismologists in a quandary.

  “At the moment, we just don’t know,” he said. “It’s a subject of scientific debate. But if we compare other areas around the world that are very similar to our subduction zone, we find that we are the only one that has not had large earthquakes.”

  For seismologists in 1985 it was hard to imagine why the Juan de Fuca plate (or the Cocos plate in Mexico) would be special—the only place on the planet where two plates glide past each other trouble free. How could this not be like the dangerous and deadly subduction faults off the coasts of Alaska, Chile, and Japan? Although Rogers didn’t seem like a gambler, he was willing to speculate.

  “A more likely scenario, comparing it with other zones, is that we are capable of large earthquakes but with very long intervals in between them,” he said. The long quiet history of Juan de Fuca could mean “it’s stuck and one of these days we’re gonna have a monster earthquake like Mexico had.”

  If the fault were “stuck,” I wondered, could the build-up be measured and—if you could see the stress increasing—would it be possible to predict the next quake? “It may be,” answered Rogers. “And, in fact, one rather suspects it should be, because before such a large earthquake a tremendous amount of strain is stored up. We might be able to detect a deformation like that. In fact, they can see this kind of thing in Japan since their last big earthquake—deformation going on.”

  Evidently rocks bend and tilt under stress and there are changes in electrical signals coming from the earth, all of which could be monitored. Rogers described prediction as a dark art that was still many years away from success, but his point was that there are things we could and should be doing to confirm or deny the possibility of large subduction earthquakes off the Pacific Northwest coast.

  It turned out that John Adams was already doing exactly that kind of research. Less than a year earlier, while working at Cornell University in New York State, he had published the first in a series of papers with new data showing that the coastal mountains of Washington and Oregon were in fact being bent and tilted landward, probably by the force of plate tectonics.

  A magnitude 8 or higher, here on my West Coast—really? I’d been living in Vancouver nearly ten years at that point and had never heard anything about a monster shockwave. Not a word of it. How could I, a working journalist covering British Columbia for most of a decade, have missed a blockbuster story like that? Well, it turns out the banner headline was being written in the present tense at that very moment. This news had not escaped the confines of laboratory walls until now.

  With a quickening pulse, I turned back to Dieter Weichert and asked for context. He recited what sounded like a well-rehearsed list of the most recent moderate-size temblors in the Pacific Northwest: “For ten years, we’ve always warned people that there are earthquakes—in Seattle-Tacoma [in 1965], under Pender Island in 1976, central Vancouver Island in 1946.” It was true, he conceded, “We have never talked about this big subduction earthquake. We knew about the possibility, but certainly with a fifty–fifty chance, you’re not going to say there is a big earthquake waiting for us.”

  First, I asked myself, who else knew there was even a fifty–fifty chance of a magnitude 8 rupture? Probably nobody except the scientists. Then it occurred to me—okay, so the senior seismologist at the Canadian government’s West Coast geoscience laboratory is a cautious man who doesn’t want to alarm the public without reasonable and probable cause. I understood that. Yet now, in the aftermath of Mexico City, he was apparently ready to raise the biggest, reddest warning flag I’d ever seen.

  “Now you’re saying it?” I prompted.

  Weichert took the plunge: “We’re saying yes, we have to come to grips with this problem. The chance has increased, in our minds, from a fifty–fifty chance to something like a seventy–thirty chance for the earthquake to happen within, say, the next two hundred years.”

  As a scientist, he really couldn’t say for sure when the megathrust might happen—two hundred years from now, or tonight—so Weichert had erred on the side of caution. That’s what responsible government scientists do. Kaufman and I, however, figured Weichert, Rogers, and Adams had given us a clear signal that the risk level was sufficiently high to justify front-page treatment of the issue.

  On Sunday, November 3, 1985, I flew from Vancouver to San Francisco en route to the U.S. Geological Survey laboratory at Menlo Park, California. First thing Monday morning we shot an interview with USGS seismologist William Bakun, who not only reinforced what the Canadian team had told us the previous week but made an even more ominous prediction. He said the Juan de Fuca plate could generate a disaster even larger than the one in Mexico.

  “We have to take seriously the possibility that a great earthquake—a very great earthquake, such as the 1960 Chilean earthquake—might occur along the Washington, Oregon, and British Columbia coast,” said Bakun. “We’re talking about as big an earthquake as has occurred in historic time—in the world.”

  Knowing almost nothing about what happened in Chile, twenty-five years earlier, I again asked for clarification. “Where would that be on the Richter scale?”

  “Off it,” he laughed weakly, and then quickly followed with an explanation. A moderate earthquake is defined as magnitude 5.0 to 5.9; strong is 6.0 to 6.9; major is 7.0 to 7.9; and a great earthquake registers 8.0 or higher on the Richter scale.

  Because the scale is logarithmic, there is a tenfold increase in the amplitude of the shockwaves with each higher whole number on the scale. If a magnitude 4 caused rocks to vibrate and move less than half an inch (1 cm), a magnitude 5 would cause them to move four inches (10 cm). Some studies have estimated that this tenfold increase in the amplitude of the shockwaves would require thirty-two times more energy. So a magnitude 9 would generate thirty-two times more energy than a magnitude 8.

  The Mexico City quake was an 8.1 and the 1960 Chilean disaster was a 9.5, the largest temblor ever recorded by modern instruments. That means the Chilean rupture generated more than thirty-two times the energy of the Mexico City event. And here was William Bakun of the USGS telling us to expect the same in the Pacific Northwest.

  We had come to Menlo Park primarily because Bakun and his colleague Allan Lindh had recently launched the first high-profile earthquake prediction experiment on U.S. soil. The Chinese and Japanese had both been running prediction studies for several years already, but given their spotty results and the controversial nature of spending money to forecast disaster, this was a bold leap for the USGS. As a journalist I figured the first thing people living in any hazard zone would want to know was: w
hen will the Big One finally happen? Now some of America’s top scientists were trying to provide an answer.

  “We can predict earthquakes, in one sense,” Bakun said, cautiously. “We can identify sections of plate boundaries that will eventually fail in large, damaging earthquakes.” Figuring out where the San Andreas fault might break again, or being pretty sure that the Juan de Fuca plate will rip loose from the North America plate some day, sounded like important science to me, although I’m pretty sure that’s not what most people think of as prediction. Bakun agreed. “We still do not know how to predict earthquakes on a short-term basis. That has turned out to be a very difficult problem, and it’s a focus of our ongoing research.”

  Bakun’s coauthor in the prediction study, seismologist Allan Lindh, told us they were studying a stretch of the San Andreas near the farming town of Parkfield, California, that had ruptured five times since 1857—each event a magnitude 6 temblor that seemed nearly identical to the one that came before, as if the same punch were being thrown over and over again. Bakun and Lindh had convinced themselves the next in this series of “characteristic earthquakes” was due in about three years. According to their calculations, the fault would build up enough stress to break again as early as January 1988.

  In August 1985, only a few months before our visit, Bakun and Lindh published the first official seismic prediction ever issued by the USGS: “The next characteristic Parkfield earthquake should occur before 1993.” Even with a five-year fudge factor, they had stuck their necks out by putting the prediction in writing in Science, one of the most prestigious and high-profile research publications in the world.

  Like Bakun, Lindh seemed to be a cautious man. Still, there was enthusiasm in his voice as he talked about trying to trim the fudge factor and “narrow down the time from a few years to months, to a few days.” He told us, “I think we’ve got a fighting chance,” asserting that the way to refine the prediction was to concentrate as many instruments as possible along one small segment of the fault—the same fifteen-mile (25 km) rupture zone that had moved in each of the previous Parkfield punches—and monitor every little creep and twitch in the earth, day and night, until the next rupture. With luck, they might spot some kind of precursor that would allow them to issue a warning to the public.

  When we wrapped the USGS shoot late that afternoon, my team and I drove four hours south on Highway 101 from Menlo Park, through the rush hour of San José, to a wine-country town called Paso Robles, where we spent the night. Early next morning we headed east into ranch country across dry brown hills on bumpy two-lane blacktop in search of a wide spot in the road called Parkfield. Our map showed it smack in the middle of nowhere about halfway between San Francisco and Los Angeles.

  Roughly thirty miles (50 km) east from Paso Robles we passed a road sign that told us we had found what we were looking for. Parkfield had a population of thirty-four, not counting cattle, and stood 1,520 feet (463 m) above sea level. We pulled up in front of what looked like an old ranch house made of square-cut timbers the color of creosote with a wide veranda, a corrugated metal roof, and a big stone chimney. Out front, in a tidy patch of unnaturally green grass, stood a tall wooden cowboy carved from a log and bolted to a stump with a small wooden dog at his knee.

  In a gravel parking lot stood the rusty iron hulk of what used to be a water tower. In a curvy flourish of creamy white letters, a hand-painted sign read “The Parkfield Cafe.” Under that, in slightly smaller print, was the proclamation “Earthquake Capital of the World. Be Here When It Happens.”

  Farther down the road we found the man we were looking for. USGS technician Rich Lichtie was waiting for us beneath a one-lane bridge that spanned a gully where Little Cholame Creek trickled west toward the sea. Lichtie fit the landscape in his baseball cap, blue chambray work shirt, jeans, and cowboy boots. His windburned complexion and red walrus moustache allowed him to blend in with the surroundings even better. No labcoated scientist from the big city, Lichtie was the guy in charge of all the USGS equipment that was jammed into the ground along both sides of the fault, and he clearly spent a lot of his time outdoors.

  Of all the places he might have arranged to meet us, he had chosen this bridge for a reason and wanted us to see it from below. So we unpacked our gear and trudged down into the gully for a closer look. That’s when Lichtie explained that this big ditch was part of the San Andreas, that the bridge literally crossed the fault, and that the last Parkfield event, back in 1966, had torn the old bridge off its foundations.

  We were looking at a replacement span that already showed signs of stress. Doug, my cameraman, got a telling close-up of one big bolt holding two heavy steel girders together by no more than a few threads. The two main sections of bridge deck had already been pried apart far enough for sunshine to burn through a gap between the beams. It was a crude yet graphic display of creep along the fault.

  Lichtie took us up the road to a cow pasture, where we hiked across the dun-colored grass toward a dry gulch with a storm culvert dug vertically into the earth. When he removed the cover, we could see a metal platform bolted to the corrugated wall of the culvert with a cable-and-drum contraption that looked like something a kid might build with an erector set.

  Halfway down the culvert was a circular hole cut into the earth several feet below the surface, where a horizontal plastic drainpipe extended toward the far side of the gulch. Inside the pipe was a pencil-thick braided steel cable that looked like a buried trip wire. Lichtie called it a creepmeter and explained that it was pretty much what it looked like—a wire sixty-five feet (20 m) long, stretched across the fault and connected to a strain gauge (in a toolbox at the bottom of the culvert) capable of measuring even a few fractions of an inch of slip along the plates.

  Next Lichtie took us to a nondescript shed in a grove of walnut trees, halfway up the side of the gulch. When he unlocked the door we saw what looked like a high-end amateur telescope, with a white steel barrel the size of a small cannon, mounted on a high-tech tripod anchored to a concrete pad with a small spotting scope bolted on top like a rifle sight. He switched on the power and a cherry-red laser shot a beam across the valley toward another tiny shack so far in the distance we could see only a smudge through waves of dusty heat rising in the noonday sun.

  Lichtie used the rifle scope to line up the laser with a parabolic reflector in the other little shed three miles (5 km) away. “We’re shooting the beam across the fault to a reflector, which brings it back here. And we can measure to within a half a millimeter how far that reflector has changed in relation to this building,” said Lichtie. As expected, the laser device had already documented right-lateral motion along the fault—the Pacific plate creeping north toward Alaska.

  As part of Bakun and Lindh’s experiment, the USGS was in the process of installing a cluster of these and other instruments at various points along the fifteen-mile (25 km) rupture zone in Parkfield. The data were being beamed continuously by microwave to a real-time processor in Menlo Park, where members of the research team were keeping constant watch. They even wore pagers that would wake them in the dead of night or ruin a perfectly good dinner if the fault started to creep or warp or bend itself out of shape.

  When producer David Kaufman and I realized the fault ran right up the middle of this gulch, it was impossible to resist the temptation to straddle the fracture and take a picture. Not that you could really see a crack or crevice in the ground; there was nothing more to look at than the V-shaped bottom of this little gully, swathed in straw-colored pasture grass in dire need of rain.

  A better way to see the fault was from the air. Aerial pictures showed that one side of the fault had been thrust up slightly higher than the other side, enough to cast a distinct line of dark shadows that ran for miles and miles in the early morning light. There it was, plain as day—two tectonic plates grinding past each other at the blistering speed of two inches (5 cm) per year, roughly the same speed as your fingernails grow.

  Far abo
ve the valley floor, it was also easier to see why the prediction experiment was being conducted along this particular segment of the fault. The San Andreas is not a straight line. Far from it, especially in Parkfield, where the shadow line zigzags ever so slightly and in a stretch northwest of town actually kinks. There’s a five-degree bend along a segment 1.2 miles (2 km) long that was the epicenter of the 1966 quake.

  In their paper, Bakun and Lindh referred to this as a “geometric discontinuity” and suggested the bend probably controls how much of the fault moves when it ruptures. I imagined it as a kind of plug or doorstop jammed in the crack, causing the fault to slow down or temporarily stop moving. Not only that, with all this zigzagging, there are rough patches in the rocks—seismologists refer to these as asperities—that create friction and also slow the creeping motion along the fault. While the rest of the San Andreas is ripping along at almost 2 inches (5 cm) per year, the Parkfield segment is lagging behind at only 1.4 inches (3.5 cm) per year.

  With tectonic plates as big as these, however, it’s obvious the little snags—those Parkfield asperities—can’t hold up progress forever. The stress builds to a point where the rocks fail. The rough spots finally shear away. In the span of less than a minute, roughly twenty-two years of “lost motion” along the fault is recovered as the Parkfield segment catches up with the rest of the San Andreas in a shuddering leap to the north. An earthquake.