Tsunami Science: Reducing the Risk

Brian Atwater is one of those people whose surname coincidentally fits his or her line of work. As a geologist for the U.S. Geological Survey, he studies earthquakes and tsunamis of the past few thousand years. Doing this often requires him to get wet along shorelines, tidal marshes, and river deltas to investigate the residue of these catastrophes, buried where land meets sea.

For tsunami researchers, witnessing an event as wide-reaching and destructive as last winter’s Indian Ocean tsunami is exceedingly rare. This makes Atwater’s soggy forays into geologic history quite valuable. By unearthing sediment deposits tsunamis leave behind, scientists can study the waves’ origins, extent, and frequency. Such work helps avert surprises from locations that have the geological apparatus to produce a tsunami, but haven’tin written history, at least.

History in Layers

Science Bulletins met up with Atwater at the Copalis River estuary on the Pacific coast of Washington State. The estuary is one of dozens that sit above an enormous fault plane that slants beneath the Pacific coast. Called the Cascadia subduction zone, the fault stretches 1,100 km from Vancouver Island, B.C., to northern California. One tectonic plate descends beneath another here. As they abrade, the overriding continental plate sticks and warps atop the subducting one. Strain builds over time. When the plates suddenly slip free, an earthquake occurs.

At this moment, the continental plate springs upward, and can launch a massive volume of water: a tsunami. Thus “unstuck,” the plate settles, now lower in elevation than it was prior to the earthquake.

Despite the fault’s existence, written records from the Pacific Northwest coast, which began about 200 years ago, are silent on the subject of large earthquakes and tsunamis generated from them. “I first went to the Copalis River in the spring of 1986,” says Atwater. “At that time, very few scientists believed that large earthquakes and tsunamis could happen here, and nobody had demonstrated that they had.” But Atwater became one of the first researchers to find geologic proof.

Atwater explains as he hacks into a marshy bank of the Copalis River with a World War II folding shovel. “What I’m unearthing is a record of a catastrophe from 300 years ago,” he says. Atwater points to the lowest of three distinct bands of sediment stacked in the bank. “Around 1700, this salt marsh, represented by the soil here, was up at the level of the present marsh above us. But then the land abruptly dropped a meter or two during an earthquake.” He traces the 10 cm thick layer of sand above the ancient marsh. “Then comes the tsunami, and lays down a sheet of sand. The sea was free to come in because the continental plate had dropped, so the ocean then laid down this top layer of mud.”

In 1986, Atwater surveyed a sand sheet that he suspected a tsunami washed into Willapa Bay, Washington. Sand deposits had been associated with only one tsunami previously, the 1960 event in the southeastern Pacific that affected Chile and Japan. Jody Bourgeois, a sedimentologist at the University of Washington, investigated further. “We started with little to go on,” she recalls. “We had to show that the sand layer was from a surge of a tsunami wave, and not from a high tide or a storm.” A team of colleagues mapped the sand and land-level changes along coastal Washington State and compared the results with deposits and eyewitness accounts of the 1960 tsunami. The picture began to come together.

Atwater and other researchers later found key clues in eerie stands of Western red cedars bordering the Copalis River and three other estuaries in the state. “We call it a ghost forest because the trees have been standing dead for centuries,” Atwater says. It’s a sign not of a tsunami, but of an earthquake capable of causing one. “After the earthquake drops the continental plate, saltwater can come in at high tide and routinely cover the forest floor, killing the trees,” he explains.

A Restless Record

At the start of World War II, a Japanese geographer looking at municipal documents from Japan’s Pacific coast noted a mention of a destructive tsunami on the evening of January 26, 1700. In 1996, Japanese researchers proposed that this event and the one that had affected the Pacific Northwest were one and the same. Tree-ring dating of the Western red cedars corroborated this date: the trees had all died at once, somewhere between August 1699 and May 1700. Recent analysis of flooding and other damage from the Japanese documents show that the tsunami’s parent earthquake was a gargantuan magnitude 9.

The mounting evidence since 1986 has convinced Earth scientists that the Pacific Northwest’s 1700 earthquake was just the most recent tsunami-generating quake of a surprisingly fitful fault. Additional deposit data have disclosed seven great Cascadia earthquakes over the past 3,500 years, with an average interval of 500 years.

As the geologic record unfolds, other researchers such as Ruth Ludwin, a seismologist at the Pacific Northwest Seismograph Network, are digging up oral traditions of Northwest Coast native communities that existed previous to written records. “It turns out there are stories amongst those tribes that are consistent with historical earthquakes and tsunamis on the coast of Cascadia,” says Bourgeois.

As the continental plate sluggishly gathers strain at Cascadia, there is no doubt that another massive earthquake and tsunami will roil the Pacific Northwest. When is unclear. 

In communities such as Cannon Beach, Oregon, or Crescent City, California, or the Quinault Indian Nation in Washington, bright blue metal signs with a white, menacing ocean wave dot coastal streets. “Tsunami hazard zone,” they warn, or “evacuation shelter,” or “tsunami evacuation route.” A tsunami hasn’t affected the Pacific Northwest coast since 1964, when an earthquake and submarine landslides at Alaska’s Prince William Sound caused a significant one. Still, scientists are certain that these specific communities are at risk. Exhaustive computer modeling--mathematical simulations of where a likely wave will start, travel, and end up--tell NOAA, the U.S. government agency responsible for tsunami warning, where to stake the signs.

Lessons from Sumatra

Vasily Titov, a mathematician for NOAA’s Pacific Marine Environmental Lab in Seattle, is one of a handful of researchers worldwide who can craft a computerized tsunami. His expertise was put grimly to the test when NOAA learned that an earthquake had generated an enormous tsunami off the coast of Sumatra, Indonesia, on December 26, 2004. After eight hours of overnight work, Titov generated the first model to describe the wave’s travel speed, direction, and amplitude (height) in the open ocean.

But the tsunami raced faster than could Titov. It was slamming into Somalia and Kenya by the time the model was complete. “I had to start from scratch,” he explains. “If the event were in the Pacific Ocean, the model could have been done minutes after learning the magnitude and location of the quake.”

The delay was due to lack of data. The Indian Ocean has been studied far less than the Pacific, where the vast majority of tsunamis occur. The first piece of information Titov needed, he hadan initial seismic measurement of the earthquake. The quake’s magnitude roughly relates to how much seawater the shifting Earth’s crust could displace. (The first alert about the quake, after about fifteen minutes, described it as magnitude 8.0. The measure was refined four hours later to 8.9. New calculations published in Nature in March 2005, however, put it at 9.3the second-largest earthquake ever recorded on a seismograph.)

Titov’s model applied equations incorporating Newton’s laws and wave physics to the data about this initial “bump” of water. The calculations described how the tsunami would likely propagate from its source. “But the nature of the tsunami wave is such that underwater topography, or bathymetry, defines the way it propagates,” explains Titov. The shapes of coastlines further transform a tsunami’s speed and size, as the sloping seafloor slows the wave and increases its amplitude.

For the model to predict where and when the wave would hit, and how hard, Titov desperately needed bathymetric data for coastal regions around the Indian Ocean. This information, regrettably, was both scarce and impossible to access in time. When combined with political hurdles and an infrastructure not suited to handle tsunami warning in the Indian Ocean basin, NOAA’s efforts were unable to help avert some of the disaster’s more than 283,000 deaths.

Toward the Future

Still, Titov’s Sumatra model is incredibly useful for mitigating the effects of future tsunamis. Researchers are now comparing its predictions to the wave’s real-life effects on coastlines around the world. “Every tsunami leaves traces on the coast,” says Titov. “So we go to the hardest-hit coasts and measure how high the wave came up the shore, or on trees. That gives us the amplitude data for the tsunami at different coasts.”

Worldwide data on the tsunami is still being collected months after the event. Some of it has been surprising, and illuminating. For example, the tsunami took 30 to 32 hours to reach the most distant coastlines, such as those in Halifax, Nova Scotia, 24,000 km from Sumatra. Curiously, some wide-ranging spots like Halifax and Lima, Peru, recorded waves several times larger than those that hit the Cocos Islands in Australia, only 1,000 km from the source. “Now we realize how a tsunami can import energy into different oceans,” says Titov. It turns out that the energy can be channeled by mid-ocean ridges: long, underwater mountain ranges formed by magma rising up in between plate boundaries. “The mid-Atlantic ridge provided the pathway for the tsunami into the Atlantic,” he explains.

Overall, Titov’s Sumatra computer model closely matches the real event. That means the calculations can be applied to models modified for use in places like the Pacific Northwest, which has a coastal fault line of the same type and length, and with similar bathymetry, as Sumatra. A magnitude 9 earthquake there could produce a tsunami that could affect coasts worldwide. Since local communities would be hit hardest, however, Titov is now working to test dozens of scenarios with his models, trying different combinations of earthquake magnitudes, locations, and local bathymetries in northern California, Oregon, Washington, and British Columbia. If an earthquake occurs at any fault in the Pacific basin, emergency managers could draw from this stockpile of scenarios to predict wave characteristics and time of strike at a particular coastline. 

A prototype system for such model forecasting is already in place, and is now being tested by NOAA and by tsunami warning centers responsible for events in the Pacific, Atlantic, and Caribbean. (As for the Indian Ocean, a warning system is still being set up, negotiated, and organized both technically and politically, says Titov.) The hopes are that within a year, NOAA’s system will be able to generate model forecasts in a matter of minutes. “We definitely don't want the Sumatra case to repeat itself,” he says. “We’re hoping to get our system in good enough shape so that the next big tsunami will not catch us by surprise.”

Hilo Bay faces the Pacific Ocean on the northeastern shore of the island of Hawaii. It is 4,000 kilometers due south of Alaska’s Aleutian Islands, home to one of the most active subduction fault lines in the world. A tsunami launched from the Alaska-Aleutian subduction zone would take just three hours to reach the city of Hilo. From 1900 to 1964, a tsunami with a run-up of at least a meter (inundating all areas below one meter above sea level) struck Hilo an average of once every five years.

On May 7, 1986, warning of a magnitude 7.7 Aleutian earthquake reached the Pacific Tsunami Warning Center in Honolulu. Not all earthquakes produce tsunamis, but based on this seismological information--and nothing else--the state of Hawaii decided to evacuate Hilo and the surrounding area. “It was an absolutely horrendous situation,” says Christian Meinig, head of engineering at NOAA’s Pacific Marine Environmental Laboratory (PMEL) in Seattle, Washington, “because the wave that arrived was very small and not damaging at all.” The false alarm triggered panic, phenomenal traffic jams, and cost the state both credibility and tens of millions of dollars in expenses and lost productivity.

On November 17, 2003, a magnitude 7.8 earthquake rumbled through the Aleutians. This time, the warning center didn’t fear a false alarm. For the first time in history, they could track the wave directly as it traveled. Three special monitoring buoys designed by Meinig’s lab floated near the Aleutian fault, and they detected and measured the wave height as it passed. The trio relayed their data via satellite to the warning center in Honolulu. “The wave was about 2 centimeters high,” says Meinig. “Based on that information we knew it was not going to be a damaging event. The warning was called off within 90 minutes of the earthquake.”

The system, called DART, for Deep-ocean Assessment and Reporting of Tsunamis, was developed by PMEL in the mid-1990’s in response to the relentless threat posed by Aleutian-born tsunamis. It has been fully operational only in recent years. Right now, 11 DART systems reside in the Pacific Ocean: Four float near the Cascadia subduction zone along the Pacific coast, four follow the Aleutians, two are mid-Pacific, and one sidelines Chile. After last winter’s devastating tsunami in the Indian Ocean, extending the DART system is a critical aspect of global tsunami monitoring.

Whither the Water

Each DART buoy is built for accuracy and speed. The two-meter-high buoys are tethered with nylon and wire rope to a 3,107-kilogram anchor on the ocean floor 4,000 to 6,000 meters below. A device that records pressure, called a “tsunameter,” is anchored nearby.

Normally, one psi (pound per square inch) of seawater equals 670 millimeters of water height. If the tsunameter detects a sudden higher-than-normal pressure of the column of water above the tsunameter, and therefore an increased sea-level height, a giant wave is certainly passing. In the open ocean, tsunami wave heights can be anywhere from millimeters to meters, which DART can measure to within a quarter-millimeter accuracy.

The tsunameters measure natural tidal cycles, but these effects are predictable and easily subtracted out. The tsunameters do not register minor surface-wind waves, however. This is because of their wavelength. Wind waves affect the water column as deep as half their wavelength, which can be up to a few hundred meters. Tsunamis, however, have wavelengths approaching 200 kilometers. These monster waves affect the water column clear to the seafloor and the awaiting tsunameters. “We’re measuring pressure in a normally very stable, very cold, very unchanging environment in the deep ocean,” says Meinig. “This gives us a pure energy measurement of the tsunami.”

The Old Wave

Monitoring tsunamis with DART is an astounding improvement over conventional methods, which were used in Sumatra in December 2004 and are still in use in much of the world. As happened during the 1986 Hilo event, scientists sometimes predict tsunamis based only on seismometer readings of an earthquake. Why every subduction-zone earthquake doesn’t cause a tsunami is still unclear. Meinig says, “Seismic information alone isn’t useful, because it’s not a direct measurement of the wave.” He points to a second Sumatran earthquake that followed three months after the 9.0 quake. Although it was a magnitude 8.7, and occurred just 150 kilometers southeast along the fault line from the December 26 epicenter, it produced a much weaker (though still damaging) tsunami. 

Another conventional tsunami-measuring technique is through tide gauges, plastic tubes installed vertically near shorelines. As water fills the tube, a computerized sensor records the sea level. But tide gauges are problematic. Many must be physically accessed to retrieve the data. Also, the topography of shorelines can reduce or amplify a tsunami’s wave height, leaving researchers unable to predict how that same wave may affect coasts far away. Tide-gauge data oftentimes come too late to be useful, and the tsunami can wash away the tubes.

“When you measure offshore, deep-ocean pressures with a tsunameter, you can see what the energy means for specific communities,” says Meinig. “Our goal is to have site-specific forecasts just like we’re able to do for a hurricane. But the challenge is that we don’t have three days to do it. We have three hours.”

The New Wave

After the tsunameter measures the wave, it relays the digital data acoustically to the floating buoy. The buoy transmits the information to a satellite, which beams it to ground-based scientists who incorporate the data into community-specific computer model forecasts. The modelers then send their predictions to local emergency managers. All these steps can be completed within 15 minutes.

Improving tsunami monitoring and warning in the Indian Ocean means installing better seismometers, tsunami-appropriate tide gauges, and of course, DART systems. PMEL is now working to adapt the DART technology for the challenges of the tropical climate. “Alaska has freezing rain and monster waves,” says Meinig. “The tropics have instead tremendous bioproductivity. The buoys will be covered with things like mussels and gooseneck barnacles.” Also, the Indian Ocean is among the most-pirated seas in the world. Vandalism, boat tie-offs, and even gunshots will threaten any buoys placed there.

NOAA is also working on increasing the U.S. monitoring system to 39 buoys by mid-2007. As technology improves and scientific understanding of tsunamis matures, monitoring them will become more akin to weather forecasting, predicts Meinig. “We’re getting better and better weather forecasts all the time,” he says. “We’ll also get better and better tsunami forecasts to benefit people all over the world.”