By Jane Qiu

Some tremors may be aftershocks of bigger earthquakes that occurred hundreds of years ago, researchers say.

Scientists assess earthquake risks mainly by looking at fault movements and seismic activity along the boundaries of tectonic plates, where most quakes take place. But at the plate interior, "we have little idea where the next fault rupture will happen on the continent and every major quake came as a surprise", says Mian Liu, a geophysicist at the University of Missouri in Columbia, and an author of a study in Nature looking at aftershocks.

Indeed, the magnitude-7.9 Sichuan earthquake in China in 2008 caught scientists off guard because it occurred in an area that had been experiencing little seismic activity (see 'Seismology: The sleeping dragon').

In contrast, the seismic zone near New Madrid in Missouri has seen frequent earthquakes in recent years but studies using Global Positioning System (GPS) measurements have shown little strain accumulation due to ground motion across the fault.

Lasting effects

Liu and Seth Stein, a geophysicist at Northwestern University in Evanston, Illinois, noticed that the frequency and magnitude of these New Madrid quakes had declined since 1811-12, when four earthquakes of magnitude 7-7.5 shook the town.

"These are patterns of aftershocks," says Liu. But aftershocks were thought to be felt for no more than a decade after the initial quake.

Stein and Liu, however, suspected that the duration of aftershocks, which take place as the crust around the fault plane adjusts to the effects of the main shock, could be related to the rate at which the fault moves. At fast-moving faults, such as the San Andreas fault, aftershocks would die out quickly.

But with the New Madrid fault, which moves over 100 times more slowly than the San Andreas fault, it could take hundreds of years for the effects of a major earthquake to fade away.

Using a simple model of the changes in fault property after an earthquake, the researchers verified the relationship. Moreover, they found that data from 16 major earthquakes around the world are consistent with their work.

"It is interesting that, with the same seismic history, we can come up with completely opposed predictions of where the next quakes are most likely to occur." says Ross Stein, a geologist at the US Geological Survey in Menlo Park, California, who was not involved with the work. "We may have been fooled by the seismic activity [along the New Madrid fault], assuming that it represents the build-up of the next major earthquake."

Liu stresses, however, that the study has just looked at one aspect of a very complicated problem. "We are just trying to correct the part of the practice that isn't working," he says. "Recent seismic activity can be misleading for assessing earthquake hazards."

Whacking a mole

Liu says that the study also suggests that seismic models based on studies at plate boundaries may not work for predicting earthquakes in the interior of plates. At the boundaries, tectonic stress builds up at a constant rate, leading to fairly regular cycles of earthquakes. In the interiors, however, such constant stress-loading processes do not occur.

"Predicting big quakes based on small quakes is like the 'whack-a-mole' game: you wait for the mole to come up where it went down," says Seth Stein. "But we now know that big earthquakes can pop up somewhere else."

Indeed, detailed earthquake catalogues in China, which go back more than 2,000 years, show that large earthquakes have never struck in the same place twice. "So the mole never came up from the same hole twice," says Liu.

To further test his hypothesis, Liu is leading a US-China collaboration to study earthquakes in northern China. In the past 700 years, there have been three earthquakes of magnitude 8 or greater -- including the Huaxian earthquake in 1556, the most deadly earthquake in human history which killed more than 830,000 people -- and a dozen of magnitude 7 or more along the Shanxi graben, a depressed block of land between parallel faults, in Shanxi province.

But the Shanxi graben has been dormant for the past 200 years, whereas the seismic activity seems to be increasing in the North China Plain to its east, resulting in large earthquakes such as the magnitude 7.8 Tangshan earthquake in 1976.

Liu and his colleagues now plan to use seismological methods to map the distribution of fault zones in the region as well as their movements and underlying structure; they will use GPS stations to measure strain accumulation along active faults and dig trenches to establish the history of ancient earthquakes.

"Hopefully, these multiple approaches will be able to test our aftershock hypothesis and advance our understanding of earthquakes," says Liu.