It came ashore without warning. Driven by a mammoth undersea earthquake, a speeding wall of water rushed inland, decimating everything in its path.
The date: Jan. 26, 1700. The place: the west coast of North America. The tsunami devastated a stretch of coastline from California to northern British Columbia, and was so powerful that it wreaked havoc on the other side of the Pacific as well.
Scientists say it will almost certainly happen again. The only question is when. "None of us would be too surprised if it occurred tomorrow," said Alan Nelson, a researcher for the U.S. Geological Survey. On the other hand, the next colossal wave may be centuries away.
The 1700 tsunami was the last really big one to strike the U.S. mainland. It was probably as powerful as the recent Indonesian surge. Both were triggered by gargantuan underwater quakes relatively close to shore.
The North American quake occurred along what is known as the Cascadia subduction zone, which is 150 miles off the coast, and extends 650 miles, from the northern tip of California to Vancouver Island.
Along this stretch, two tectonic plates are slowly smashing into each other: The North American continental plate is moving west, into the Juan de Fuca ocean plate, which is traveling east. The boundary is called a subduction zone because the ocean plate is being pushed down, or subducted, by the continental plate.
Until recently, scientists didn't realize that the Cascadia tsunami had even happened. In 1986, University of Washington seismologist Brian Atwater was nosing around the coast of Washington, looking for geologic deformations that would help pinpoint the size of the Cascadia earthquake.
One day, while canoeing in a tidal creek, he noticed a thick layer of sand in the sediment of the creek bank.
"I'd never seen that," says Atwater. "It was a sharp boundary. I was very surprised." He suspected that the sand had been carried inland by a huge rush of water. He went to other spots along the coast, and found the same sand deposits.
Then he came across even more compelling evidence. In some marshes, he found acres of stumps from dead spruce and cedar trees, or "ghost forests," as he calls them. Atwater realized that the trees, which normally live in freshwater swamps, had been killed by an inundation of salt water. Over time, the dead trees had rotted and fallen, leaving only stumps.
He and several colleagues used radiocarbon dating to find out when the trees had died. They also used the technique to pinpoint the demise of microscopic plants found in the sand layers. Both the trees and the sea plants had perished at the same time: about 300 years ago.
By the early 1990s, news of Atwater's work spread to Japan, where researchers were using historical records to document past earthquakes and tsunamis. These historians had an "orphan" tsunami - a wave that they couldn't connect with any known earthquake in the region. The wave, 15 feet high in some places, had struck the Japanese coast on Jan. 27, 1700, killing a few dozen people. There was no warning, and no earthquake preceded the surge.
But after reading Atwater's work, the Japanese researchers knew they'd found the tsunami's parents. "It was a eureka moment," says Atwater. "They had been wondering where the hell this thing had come from."
Working with Japanese accounts of the tsunami, he and two other scientists used a computer model to reconstruct the Cascadia event. They concluded that the quake occurred at 9 p.m. Pacific time, creating a tsunami that may have reached a mile inland along the Pacific coast.
Scientists found other human reports closer to home. In the 19th and 20th centuries, ethnographers recorded hundreds of Native American oral histories. Earthquake researcher Ruth Ludwin combed through these, and found descriptions of shaking and huge waves. One native told of a "trembling of the earth beneath, and a rolling up of the great waters." Another said canoes ended up resting in treetops.
The key question, of course, is how often these Cascadia earthquakes and tsunamis occur. The best evidence comes from Bradley Lake, a body of fresh water in southern Oregon. The lake sits 600 yards from the ocean, and 15 feet above sea level. Small wave surges, such as those from far-off tsunamis or large storms, do not reach its shores. But a huge wave, such as the 1700 event, would have swamped the lake.
When these tsunamis came, they left deposits of ocean sand and microscopic saltwater plants. Bradley Lake, in other words, offers an exceptionally clear window onto past Cascadia upheavals. "This lake had an amazing record of tsunamis," says Nelson, the USGS scientist.
He and several colleagues drilled into the lake bottom and pulled up cores, or vertical slices of the bottom sediment. They revealed that over the past 4,500 years, Bradley Lake had been deluged 12 times.
So on average, tsunamis seem to occur every 375 years. But it's not that simple, Nelson says. Over the past four and a half millennia, the lake was engulfed as frequently as every 250 years, with gaps of up to 800 or 1,000 years between tsunamis.
"It's not very regular," says one of Nelson's collaborators, Harvey Kelsey, a geologist at Humboldt State University in California.
So which period are we in now? No one knows. Earthquakes are notoriously difficult to forecast, and scientists generally try to avoid specific predictions.
"It may be that we're starting the next cluster," says Kelsey. "Or we may be in a thousand-year gap."