Cascadia's Fault

by Jerry Thompson

  In principle, if Lynett’s model worked well for Seaside, then it could be reprogrammed and modified with new bathymetric and street grid details for the next town on the coast, and a more realistic appraisal of the inundation zone and specific levels of risk could be had much sooner and at lower cost. At least that was the theory and the reason that people like Dan Cox and Patrick Lynett were eager to see what happened next.

  Chris Goldfinger, back from his research cruise to Sumatra, offered a sobering caution. The numerical simulation of anything as sloppy as moving water is extremely difficult to do. It was hard enough to work on a broad, oceanwide scale as Vasily Titov had done, but even more challenging when you tried to zoom in to detailed street grids and individual buildings in a single town. The tighter the grid, the more exacting the model, the greater the chances for error.

  “The best computer models now are working hard at quantifying the flow [of water] around one or two objects,” Goldfinger explained, “a cylinder, a bridge piling, something like that—a relatively simple case—just because the computational time is enormous.” When the myriad three-dimensional obstacles in a real harbor and town are assigned numerical values—the friction coefficient for water moving over the sandy ocean bottom, a different level of friction and drag once the swell crests, crashes over the seawall, and begins moving over dry ground cluttered with buildings, cars, trucks, trees, and lumpy terrain—it gets a lot more difficult.

  After a quick check by portable radio with the crew standing by in the control room to confirm that the computer and the paddles were ready, Cox turned to his visitors. “So what you’re going to see next,” he explained, “is the rough equivalent of the five-hundred-year-event.” Meaning the full-margin rupture of Cascadia’s fault that takes place on average every five hundred years. “So this is a twenty-centimeter lab scale or [the equivalent of] a ten-meter full-scale tsunami that is coming into Seaside.” Quickly doing the conversion in my head, I tried to picture a surge of water more than thirty feet above the high tide, thundering toward the beach.

  Cox gave the order and moments later the long row of paddles at the far end of the tank thrust forward at the calculated angle and speed. A dark swell began moving, quietly hissing toward Seaside. The wave crashed against the breakwater and shot a slice of foamy spume straight up. In the next heartbeat the on-rushing tide poured across the promenade and churned up Broadway, sweeping up toy school buses, cars, and trucks in a frothy vortex that quickly swamped the entire model.

  As a matter of interest I noticed that one prominent multistory hotel right on the waterfront had been completely overtopped by the wave. The upper deck of the parkade building directly behind the beachfront condominium complex had remained dry. Patrick Corcoran, standing elbow to elbow with the other observers on the pool deck, noticed it too. Anyone who could get to the top of that building would probably have survived, but anyone stranded at street level or anyone trying to escape in a car probably wouldn’t have had a chance.

  When the sloshing finally stopped nobody said a word. Like Corcoran, many of those in the crowd knew the streets of Seaside well. It seemed as though everyone in the room was momentarily stunned, trying to absorb the news that most of the downtown area would be inundated by the pulses of seawater from Cascadia’s fault.

  Patrick Lynett, meantime, had called up a file from his computer model and was explaining how he’d created a numerical duplicate of the objects in the basin. The first frame of his simulation of Cascadia’s wave looked like an animated cartoon of the physical model, a 3D aerial view of the miniature plywood buildings nailed to the concrete floor in front of us. His intention had been to make the layout of the numerical model resemble the built world of the research tank as closely as possible so that when he hit the run button, his computer-generated wave would face exactly the same obstacles. With any luck the digital tsunami would match the behavior of the real wave we had just witnessed.

  As I watched the simulation play on the monitor, it looked pretty convincing to me. After pounding across the seawall, a dozen or more jets of dark-blue liquid surged past the first rank of buildings, pushing straight up Broadway and all the eastbound streets simultaneously, turning corners and twisting together like braided hair as they seemed to amplify themselves in thicker, darker currents along some of the narrow side streets.

  “What we’re looking at here is momentum flux,” said Lynett, “which is a very good measure for the potential force of fluid for the tsunami as it comes in and inundates Seaside . . .”

  It wasn’t clear to me that anyone really heard what he’d said. They couldn’t take their eyes off the screen. This time we could see more clearly what the wave was doing because the computer had slowed it down to something resembling real time. With a click of the mouse, Lynett slowed it down even more and then stopped the action completely.

  “If you look at this little building right here,” he said, hitting pause and pointing to what looked like a small house on a side street several blocks away from the beach, “what actually happens is—you get a wake off of this building.” He clicked play again and advanced the animation a few more frames. A thick wedge of the inky blue water pouring east on one street ricocheted off a larger commercial building across the street from the house. “It bounces off the side of this building—and additionally you get a large wake off of this building.” Now we could see the almost synchronous arrival of another tongue of water on the next parallel street. “And those two line up and just pound this tiny little building,” said Lynett.

  Like pool balls bouncing off padded rails, like Titov’s wave bouncing off the Maldives to hit the back side of Sri Lanka, these two strands of liquid energy careened off two multistory commercial buildings in the digital clone of Seaside, curled together in the middle of a side street, and combined forces. The computer animation changed the color of the flow to bright yellow and red to indicate the magnified amplitude and intensity.

  “And so what you’ll see,” said Lynett, “if you look at it,” and everybody clearly was glued to the screen, “you’ll see that dash of red shoot here and then bounce. And then—just eyeball right on this building.” He paused as if there was nothing else to say, then added quietly, “It would be extremely damaging . . .”

  It would take months of detailed side-by-side comparison to see how closely the computer simulation and the wave-tank model had matched up, but they looked remarkably similar to the untrained eyes of those of us standing poolside that day in Corvallis. In the end some combination of the two approaches would probably emerge from this experiment to create a refined and updated system for predicting the effects of tsunami attacks on other coastal communities. “You sort of have to boot-strap between a physical model in a wave tank and a computer model to validate one against the other for the things that you can test,” Chris Goldfinger explained, “and then go beyond the capabilities of either one by using them together.”

  After watching Lynett’s computer simulation, it was time to play back our HD video. Because the physical model of Seaside had been built on a scale of one-to-fifty, we needed to slow the frame rate of our pictures by the same ratio—and the high-speed camera allowed us to do that—producing a slow-motion image that looked almost identical to those tragic home videos from the Indian Ocean. Now we could see the swelling mound of water as it hurtled toward the surrogate Seaside. At some critical point along the beach where the ocean bottom angled upward, the leading edge slowed down long enough for the back of the wave to catch up with the front. The swell piled up like a surfer’s dream, curled forward, and then broke under the force of gravity in a hissing bore of fast-moving water.

  As it shot across the last stretch of beach toward the base of the promenade a knife-edged geyser of spraylike jets, as if from a thousand vertical fire hoses, rocketed straight up in a perfect replay of the Sumatra waves hitting the wall of a resort in Phuket, Thailand. And just as mesmerized as those poor souls who stood like backlit deer at the foot
of the palms waiting to see what would happen next, we observers at the OSU tsunami basin could not take our eyes off the screen.

  Having crashed over the promenade, the wave continued pounding straight at us. The street-level view from the snorkel lens revealed the Broadway canyon between the seven-story condominium complex and the beachfront hotel across the street as the roiling water lifted a toy school bus, an ambulance, and several other vehicles and swept them away—exactly the way the tsunami did in Banda Aceh. A floating garbage truck crossed the sidewalk and slammed backward like a levitated battering ram into a two-story commercial building. A few frames of video later, the wall of tumbling junk rolled right over us and the picture went dark.

  “Very sobering,” said Doug Barker of the Seaside Fire Department when the video finally stopped. “It was . . .” he paused, searching for the right words. “It was actually a shock. I was—it took me more by surprise than I thought it would to watch the water roll through, between those buildings. And cascade over the buildings. So, yeah, it was very eye opening, even though I’ve been dealing with it for a number of years.”

  Another of the invited observers, Barbara Lence, a civil engineer from the University of British Columbia who had just completed work on a computer model of Cascadia’s wave showing how it was likely to inundate much of the village of Ucluelet on Vancouver Island, was also stunned by the video. “One of the things that I take away is that this decision about whether we stay—or go—during one of these events is a very critical decision,” said Lence. “Do we shelter in place? Or do we focus on emergency exit?” She was also aware as never before of “the importance of debris—vehicles and so on—floating in water, hitting buildings, hitting structures that may be providing safety.”

  “Hopefully we’ll be able to use that videotape or the simulations to maybe wake up some folks in the local area,” said Barker, “to show them—give them an idea of what to expect.”

  “If we could communicate that intensity to everyone,” added Lence, “we might have a better chance at being prepared in these emergencies.”

  Wave-tank models and digital tsunamis were not the only kinds of experiments being conducted to anticipate the effects of Cascadia’s next violent outburst. Coastal inundation zones are clearly not the only concern. By the spring of 2009, a new question had arisen: what would happen to the urban cores of major cities from Victoria and Vancouver to Seattle and Portland? These four cities are built on land that lies well to the east of the main fault, and some experts had suggested the rupture zone was far enough away that damage in the urban areas might not be as catastrophic as first thought.

  When Roy Hyndman and Kelin Wang published their study of the locked part of the subduction zone in 1995, they calculated that the stuck and truly dangerous area of the fault lay nine miles (15 km) underground to the west of Vancouver Island and the beaches of Washington, Oregon, and California. The “landward limit” or leading edge of the locked zone extended “little if at all beneath the coast,” which “limits the ground motion from great subduction earthquakes at the larger Cascadia cities that lie one- to two hundred kilometres inland.”

  With this in mind many emergency planners in the Pacific Northwest have worked on the assumption that Cascadia’s magnitude 9 event would not be the worst scenario. Any of the much shallower faults in the continental crust could generate a magnitude 6 or 7 rupture directly beneath or very near an urban area, and this might actually cause more severe, localized damage. But in 2009 Timothy Melbourne and his colleagues suggested the locked zone was considerably nearer to the big cities, perhaps within 50 miles (80 km) of Seattle, for example. And if that’s true, then we’re right back where we started with Mexico City in 1985: what do we do about heavy shockwaves hitting high-rise buildings?

  From a civil engineer’s point of view, the elephant in the room has always been how an urban forest of tall towers, long bridges, freeway overpasses, and hydro dams would respond to the much longer duration of low-frequency seismic shocks from Cascadia’s fault. The way it was explained to me, going from a magnitude 7 to a magnitude 9 means the intensity of the shaking doesn’t change so much as the length of time it lasts.

  Instead of forty-five seconds or a minute of shaking in a magnitude 6 or 7, the seismic shockwaves from a magnitude 9 could go on for four or five minutes. Like they did in Alaska. And nobody really knew how well tall buildings would stand up to that much horizontal motion, slamming side to side, undulating like metronomes fifty or a hundred times, flexing every beam of steel, stressing every welded joint, every slab of concrete, every pane of glass to the limits of endurance. All anyone could do was speculate about the outcome because there has never been a magnitude 9 in a city full of skyscrapers. Neither the Chile quake of 1960 nor the Alaska rupture of 1964, nor even the Sumatra disaster of 2004, shook a large, modern city with high-rise towers. Mexico City gave us only a hint of what might happen.

  To oversimplify things just a bit, building codes in North America generally require engineers to design tall structures in earthquake country to survive the forces imparted by temblors up to magnitude 7. Specifications for a magnitude 9 don’t exist because there have been so few of these megathrust events—only four in the last century—and so few measurements of the ground motion in real-world circumstances that nobody has a good grasp of how strong the vibrations would be when they hit any given high-rise. The cities of modern civilization have been built taller and taller with not a single full-scale test of what might happen in a megathrust temblor.

  That was the problem Tom Heaton, who heads the Earthquake Engineering Research Laboratory at Caltech, and Jing Yang, one of his doctoral students, decided to tackle next. Heaton and Yang built a computer model to simulate the effects of a magnitude 9.2 Cascadia rupture on downtown Seattle. They began with data from the 2004 Sumatran quake and Japan’s magnitude 8.3 Tokachi-Oki rupture in 2003, along with geological data about various soil and rock conditions in and around Seattle.

  They tested a series of hypothetical steel-frame buildings from six to twenty stories tall with older “brittle” and newer “perfect” welds at the critical joints. They ran the model several times, factoring in different possible distances from Seattle to the locked zone of Cascadia’s fault, just in case the real rupture does extend farther inland toward the city. Their digital temblor made the ground shake for four minutes, the dominant shock being the low-frequency kind that caused so much grief and damage in Mexico City in 1985. The deep sedimentary soils in some areas of Puget Sound predictably amplified the waves, just as the dried-up lakebed did in Mexico City. In the new Caltech simulations, the soils increased the duration of shaking as well.

  Heaton and Yang presented their results at the annual meeting of the Seismological Society of America in Monterey in April 2009 and journalists immediately wanted to know the bottom line. A glance at the poster Jing Yang had prepared for the meeting pretty much told the story: “Our simulations show that Seattle high-rise buildings with brittle welds have a significant potential for collapse.”

  When Yang started work on her numerical model, she was forced to simulate steel-frame buildings rather than large concrete structures simply because it was easier for a computer to predict how steel would bend and eventually fail. The fracturing of concrete was much more difficult to model. It was certainly plausible, according to Heaton, that older concrete buildings could be at even greater risk because they were probably even more brittle. But there was no reliable way to re-create that kind of failure in a computer.

  Another reason to study steel was that the Northridge jolt of 1994 had shown scientists that brittle welds in older steel-frame buildings had failed more often than anyone had expected. Amendments to the building code made in the wake of Northridge have changed the way structural joints are welded, presumably giving newer buildings extra strength. And Yang’s model did seem to confirm that newer towers would be stronger—but only up to a point.

  If the rupture of Cascadi
a’s fault happens to extend down below the west coast beaches to some point underneath the Olympic Mountains—closer to Seattle—the shaking would be much worse. And when that scenario was run in the Caltech simulation, all the high-rise buildings in Yang and Heaton’s experimental model collapsed. Even those with “perfect welds.”

  Some of the science writers saw parallels to Mexico City and wanted to know more. “All the crummy little buildings that existed in Mexico City were completely undamaged,” Heaton explained, offhandedly, to one reporter, “but the high-rise buildings, which were the pride of their construction industry, many of them collapsed. It wasn’t just a matter of poor construction. It was a case of the wrong buildings being in the wrong place at the wrong time.”

  Low-rise, low-tech buildings simply did not vibrate or resonate at the same frequency as the big shockwaves generated by a subduction zone. High-rise buildings, on the other hand—even relatively new ones—constructed on thick sedimentary soils, vibrated more than any engineer or any building code had anticipated. They shook to the point of collapse. And what happened in Mexico would presumably happen in Seattle, Vancouver, Victoria, and Portland as well, according to Heaton and Yang’s research.

  “In general, high-rise buildings behave very differently from low-rise buildings,” Heaton said. “They’re primarily designed to be flexible. And in sharp, rapid shaking—during a moderate-size earthquake—high-rise buildings perform extremely well.” But a magnitude 9 was quite obviously a different story. Yang told another reporter that there were approximately nine hundred high-rise towers within striking distance of Cascadia and half of those were built prior to 1994, when the new building code imposed tougher standards.