UC Berkeley Press Release
Scientists upgrade Dec. 26 Sumatran quake responsible for deadly tsunami
BERKELEY – The Sumatra-Andaman earthquake that generated a deadly tsunami on Dec. 26 was stronger and slower than most seismologists thought, according to scientists at the University of California, Berkeley, the Wadia Institute of Himalayan Geology in India and the United States Geological Survey (U.S.G.S.).
 This map shows the earthquake motions measured at GPS stations around South and Southeast Asia. The arrows indicate the magnitude of the measured offsets, up to an inch (25 mm) in southernmost India, though less than a centimeter for most of the stations. The red arrow that goes off the map represents a 14 centimeter displacement to the west on the island of Northern Sumatra. The blue lines are plate boundaries; the red dots along the boundary between the India and Burma plates are aftershocks of the Dec. 26 quake. (Graphic by Roland Burgmann / UC Berkeley) |
Using data from global positioning system (GPS) stations around the Pacific and Indian oceans, including previously unavailable records from India, the team modeled the fault motions that would produce the observed static ground movements. They concluded that the quake was probably twice as strong as originally estimated - a magnitude 9.15 instead of 9.0 - and that much of the slippage along the fault probably took place more than half an hour after the initial quake and continued as long as three hours afterward.
"This could explain why the structural damage from the quake was less than expected, as well as some reports from the Andaman Islands that the subsidence from the quake didn't occur until after the tsunami had hit, more than half an hour after the initial quake," said Roland Burgmann, professor of earth and planetary science at UC Berkeley.
Burgmann and colleagues Fred F. Pollitz of the U.S.G.S. in Menlo Park, Calif., and Paramesh Banerjee of the Wadia Institute estimate that average slippage along the entire length of the 1,300-kilometer (800-mile) fault was at least 5 meters (16.5 feet), though the fault may have slipped as much as 20 meters (50 feet) near the southern end where the rupture started. These horizontal displacements, Burgmann said, led to a rearrangement of the Earth's surface, producing measurable deformation as far as 4,500 kilometers (2,800 miles) away.
"The Earth is still ringing like a bell today," nearly six months after the earthquake, Burgmann said. "We've never been able to study earthquakes of this magnitude before, where a sizable portion of the Earth was distorted. Normally, we see deformation of the surface a few hundred kilometers away, but here we see deformation 4,500 kilometers away, and five or six times the deformation we've seen in previous quakes."
The authors published their findings as part of a series of scientific papers on the record-setting earthquake appearing this week on the Science Express Web site, which provides rapid online access to papers to be published later in the journal Science.
The Sumatra-Andaman quake struck Sunday morning, Dec. 26, 2004, at 7:59 a.m. in Southeast Asia, just east of the Sunda trench where the Indian plate ducks under the Burma plate. Starting at the southern end of the subduction zone off Banda Aceh, Northern Sumatra, the ocean floor suddenly plunged northeastward as much as 20 meters under the Burma plate, releasing centuries of strain. As the slippage ripped northwestward for more than 500 kilometers, and perhaps the entire 1,300-kilometer length of the zone, it raised the lip of the Burma plate several meters. The ocean was pushed up a similar amount, generating a traveling tsunami that spread throughout the Indian Ocean and struck the coasts of Sumatra, Malaysia, India and Sri Lanka, killing some 300,000 people.
Seismologists immediately gathered data from stations around the world, and initially calculated the magnitude of the quake at 9.0, the largest since the 1964 Alaska quake and the fourth largest since 1900. Subsequent studies looking at long wavelength, low frequency pressure waves from the earthquake have suggested the magnitude was higher, and one subsequent study proposed a 9.3 magnitude, four times stronger than the original 9.0 magnitude estimate.
Burgmann and Pollitz used GPS data from 41 stations around Southeast Asia, 36 of them more than 1,000 kilometers (620 miles) from the epicenter, to model the slippage at the plate boundary and estimate the true energy released in the quake, and thus the magnitude. With Banerjee supplying data from 12 Indian GPS stations, it became clear that the quake caused a large area of the Earth's surface to move. At one site 4,500 kilometers from the epicenter, the surface moved a millimeter toward the epicenter. The land shifted eastward a couple of centimeters - nearly an inch - in southern India, which is closer to the epicenter, but less than a centimeter in most areas.
Using a model developed by Pollitz that simplifies the Earth as an eight-layered sphere, which is more realistic than standard models that represent the Earth as a uniform body with a flat surface, the researchers estimated the motion near the epicenter and along the rupture that would have produced the measured displacements. Based on a comparison with previous seismologic studies, they concluded that the megathrust quake did not occur all at once, but slipped rapidly at first, then slowed down.
"The earthquake lasted at least an hour and perhaps up to three hours, with much of the deformation happening more than one hour after the main shock," Burgmann said. The researchers estimated that as much as 25 to 35 percent of the deformation took place more than an hour later.
Much of the later slippage may have been along the northern segment of the subduction zone, the researchers noted, which would explain why the northern part of the rupture zone appears to have contributed little to the tsunami. Because tsunamis depend on how fast the uplift occurs, a slow uplift along the northern segment would not produce a big wall of water.
The team also concluded that a later calculation of 9.3 for the magnitude of the Sumatra-Andaman earthquake was an overestimate, probably due to an underestimate of the angle at which the India plate plunges under the Burma plate. Though initially shallow at about eight degrees, the India plate rapidly dips at an angle between 10 and 18 degrees.
"The geodesy suggests a much bigger quake than the original estimate of 9.0 but much less than the later 9.3 estimate, which relied on an eight-degree dip," Burgmann said. "Because our models are sensitive to the total slip of the earthquake, and not only to that part that generated seismic waves that seismologists study, or to the relatively rapid uplift that produces the tsunami, we have more solid constraints on some aspects of the earthquake."
Burgmann said that the Earth's interior took a while to adjust to the shifted surface, and that such adjustments or relaxations in the mantle could have triggered a magnitude 8.7 earthquake on March 28. This quake was situated southeast of the epicenter of the Dec. 26 earthquake, on the adjoining segment of the subduction zone that had not ruptured since 1861.
"The effects of relaxation can exceed the co-seismic deformation," he said. "The Sumatra quake wasn't enough to immediately break the second segment, but maybe the relaxation triggered the second quake."
Based on theories of the Earth's relaxation proposed by Pollitz, Burgmann and UC Berkeley colleague Barbara Romanowicz, director of the Berkeley Seismological Laboratory and professor of earth and planetary science, the effects of the Sumatra-Andaman earthquake will continue to propagate through the region for many years and could trigger more large quakes.