Waves caused by a magnitude-8.8 earthquake flattened coastal regions in Chile. Credit: R. Candia/AP/Press Association Images

Moments after a magnitude-8.8 earthquake rocked Chile this weekend, a tsunami began to sweep across the Pacific Ocean at hundreds of kilometres per hour. And within 2 hours, scientists had determined that coastal communities beyond Chile probably had little to fear.

Yet around the Pacific, from Hawaii to Japan, authorities ordered extensive evacuations. To some tsunami experts the response was no more than prudent. Others think that it was a costly overreaction. "The warning system worked well in terms of trustworthiness of forecast, but its implementation had large loopholes," says Costas Synolakis, a tsunami expert at the University of Southern California, Los Angeles.

Events in Chile, where the earthquake has claimed more than 700 lives, showed the value of a timely tsunami warning. Despite the enormity of the quake — the fifth-strongest since 1900 — the Chilean Navy, responsible for tsunami warnings, failed to issue an immediate alert. Within 34 minutes, destructive waves had hit the coastal city of Valparaíso, sweeping away buildings and people. The number of casualties there is unknown — but the Chilean minister of defence, Francisco Vidal, says that without warnings issued by local authorities and harbour officials acting on their own, many more lives would have been lost.

Tsunami forecasters were already trying to work out what would happen next. Burak Uslu, a tsunami modeller with the Pacific Marine and Environmental Laboratory (PMEL) in Seattle, Washington, used a mathematical model called Method of Splitting Tsunami, developed with Synolakis and Vasily Titov of the PMEL, to forecast when waves would arrive, how high they would be and how much dry land would be flooded. It was the first time that the model had been applied to a big tsunami in the Pacific.


Click for larger image

About 2 hours after the quake, the tsunami passed the first of the Pacific's network of monitoring buoys, which recorded a wave one-quarter of the height of the 2004 Indian Ocean tsunami. When Uslu used the buoy data to refine the model's predictions (see 'Catching a wave'), he found that most of the tsunami energy would pass between Tahiti and Hawaii and continue on to Japan, and that Hawaiian beaches could be flooded up to a height of 1.2 metres. Around the same time, however, the US Pacific Tsunami Warning Center (PTWC) in Ewa Beach, Hawaii, issued an ocean-wide tsunami warning.

Synolakis contends that the PTWC was divided on how to respond to the forecasts — after all, a tsunami triggered by the largest-ever recorded earthquake, which struck Chile in 1960, had killed 61 people in Hilo, Hawaii. But this is denied by Charles McCreery, director of the PTWC: "I heard no dissent in our office."

The PTWC decided that tsunami alerts should be kept in place despite the forecasts of relatively little impact outside Chile. Many thousands of people were evacuated from beaches and low-lying land in Hawaii and parts of Japan, and Californian beach-goers and communities were held on standby. "We were not going to gamble with people's lives," says McCreery. He maintains that Hawaii had a good chance of sustaining damage and points out that the Marquesas Islands in the south Pacific were hit by waves that were 4 metres from crest to trough.

In time, the model's predictions would be vindicated by events around the Pacific. When the waves arrived in Hawaii, they were little higher than the normal surf, whereas coastal communities in northern Japan, 17,000 kilometres from the earthquake zone, faced modest flooding — the worst impact outside Chile.

Scientists think that this week's tsunami was milder than the Indian Ocean event because the fault that slipped lay beneath a relatively shallow part of the ocean, and therefore displaced a smaller volume of water. The geometry of the fault and the way it ruptured probably also helped to limit the severity of the tsunami. "Had the ground ruptured a bit further north, Hawaii would had been severely hit," says Synolakis.

McCreery admits that the PTWC took a conservative approach to what the model was telling them. "We'll be looking at this very hard over the next few months, and longer, to see what improvements can be made," he says. As confidence grows in the model, experts may be more prepared to cancel tsunami warnings earlier, he adds. Although serial false alarms can make people complacent about the threat of a tsunami, McCreery says that occasional evacuations can actually increase people's confidence that the system is working.

Jörn Lauterjung, a tsunami expert at the German Research Centre for Geosciences in Potsdam who oversees the German–Indonesian tsunami early-warning system, agrees that the PTWC made the right decision. If an earthquake of similar magnitude strikes the Indian Ocean region, he says, authorities should take the same action. But Synolakis argues that the increasing reliability of tsunami forecasts allows emergency planners to order evacuation only when necessary. "The authorities in charge need to listen to science," says Synolakis. "Every ounce of extra prevention is counterproductive as it reduces the overall credibility of the system."

figure a
figure d

NOAA Center for Tsunami Research