The prior three months before the main shock, finely tuned instruments lining the fault picked up hints of activity. If a new study is right, these faint tremors could be a first signifigant step toward predicting earthquakes .

In a new analysis of seismic data collected before and after the quake, David Shelley of the United States Geological Survey in Menlo Park, Calif., has found what he thinks is a precursor to the quake.

Can earthquakes be predicted?

Shelley's analysis uncovered a swarm of tremors barely more than a whisper above normal seismic noise. They formed a pattern, moving steadily from north to south about 15.5 miles below Earth's surface.

At that depth, some six miles below where the main shock would later hit, faults become mysterious places. They can slip several feet without causing more than a seismic shimmy.

Since 2002, scientists have started finding tiny tremors linked up with these "slow slip events." Such events could transfer stress from the deep parts of a fault toward the surface, where rocks stay locked until rupturing violently.

Tremors have been detected in some of the world's most dangerous and well-studied faults: the San Andreas, the massive Cascadia fault in the Pacific Northwest and the Nankai Trough in southwestern Japan.

Shelley's work, due to be published in the journal Geophysical Research Letters, is a breakthrough, because it provides a tentative link between tremors, slow-slip events and a powerful earthquake.

"This is just one example; it's not clear enough that you could make a prediction based on this," Shelley cautioned. "But the implication is there are probably times of higher and lower likelihood of an earthquake happening, and that we could figure out when those are by looking at tremor activity."

Finish reading here