By Alexandra Witze, Science News
Japan’s catastrophic March 11 earthquake was the biggest ever recorded in Japan and the costliest natural disaster in history, but in a way it was nothing new. Three other quakes of magnitude 8.6 or greater have struck worldwide in the past seven years—after a gap of nearly five decades.
Two U.S. Geological Survey scientists contend that the Japan quake bolsters their idea that the planet is experiencing a spasm of great earthquakes, its second since 1900. Other scientists say that any apparent bunching is a statistical fluke that disappears if the data are analyzed in other ways. Researchers presented their opposing views on April 14 in Memphis, Tenn., at the annual meeting of the Seismological Society of America.
On the face of it, large quakes certainly do seem to have been popping off lately. The magnitude 9.1 Sumatra quake in 2004, which caused the disastrous Indian Ocean tsunami, was followed by a nearby magnitude 8.6 the following year. A magnitude 8.8 hit Chile in February 2010 and the Japan quake struck on March 11 of this year, leading many to question whether big quakes are becoming more frequent.
Scientists say there’s not enough evidence to support that idea. But big quakes may at least be coming in groups.
Some earthquake clusters happen naturally; after a large quake, aftershocks continue to rattle the area. And large quakes can also trigger separate tremors, smaller and close to the original quake (SN Online: 3/28/11).
But Charles Bufe of the USGS office in Golden, Colo., argues that Sumatra, Chile and Japan represent an unusual grouping.
In a 2005 paper, Bufe and his colleague David Perkins argued that a similar cluster occurred between 1952 and 1964, anchored by a magnitude 9.0 in Kamchatka, Russia, at its beginning and a magnitude 9.2 in Alaska at its end. The Kamchatka quake was the first magnitude 9 or greater since 1900, when scientists began recording seismic activity worldwide. This cluster also included the largest quake ever recorded, a magnitude 9.5 monster that struck Chile in 1960.
Bufe and Perkins used the same analysis—simulating the probabilities that the earthquake patterns could be explained by random events—to argue that the 2004 Sumatra quake kicked off a similar period. The chance of this particular series of events—an active period from 1952 to 1964, followed by a lull and then a start-up again in 2004—occurring randomly is less than 2 percent, Bufe says.
Why quakes would cluster isn’t entirely known. Perhaps shaking from the biggest quakes sets off a large-scale ringing within the Earth that affects tectonic activity, Bufe proposes. “Maybe on the very large scale we have a weakening of these very long fault zones that are on the verge of failure.”
Over the next six years, Bufe calculates, the chance of another magnitude 9 quake occurring somewhere in the world is 63 percent—compared to 24 percent if the planet were not experiencing a cluster. “We think we’re in an increased hazard situation from these very large quakes,” he says.
Other researchers say Bufe has gone too far out on a statistical limb. Andrew Michael, a USGS researcher in Menlo Park, Calif., has done a separate analysis. “We don’t see clustering,” he says.
Rather than study only the highest magnitude quakes, Michael probed different magnitudes over different time periods. Among other tests, he looked at whether there was a statistically meaningful increase in seismicity worldwide after big earthquakes, and whether the time between big quakes followed a nonrandom pattern. He found that the tests could produce what looked like earthquake clusters, but that the clusters disappeared when different magnitudes and time periods were included. “If we use a range of magnitude cutoffs we find that the data are very well explained by the random model,” he says.
Bufe counters that because clustering may happen only for very large quakes, the pattern wouldn’t be seen when looking at lower magnitudes. But Michael argues that the number of magnitude 9 earthquakes is too small to draw any statistically significant conclusions. “At 9 you just don’t have enough data,” he says.