Earth moved in the Chilean quake where the strain was highest.
Richard A. Lovett 
A detailed analysis of Chile's February quake may help seisomologists to predict the severity of future termors. (LEO LA VALLE/epa/Corbis)
For 100 years, scientists have thought that earthquakes are caused by 'elastic rebound', in which strain builds up across fault lines until something causes the fault to slip in a big, earthquake-producing lurch. Only now — in the aftermath of the magnitude-8.8 earthquake that rocked Chile in February this year — have scientists been able to measure all of these events in action.
In a study in Nature1, a team of geologists from the GFZ German Research Centre for Geosciences, Potsdam, compared earth movements during the Chilean earthquake to a decade's worth of measurements of strain build-up, finding that the greatest slippage indeed happened on parts of the fault that had accumulated the most strain.
The earthquake occurred on a section of the Andean subduction zone that represented a 'seismic gap' in which no major earthquake had occurred since 1835, when Charles Darwin reported one that is now thought to have had a magnitude of about 8.5.
Taking the strain
To determine how much strain had accumulated, the scientists used high-precision data from 232 Global Positioning System (GPS) stations spanning hundreds of kilometres of Chile, ranging from the Pacific coast to the Andes.
These instruments revealed, with millimetre precision, that this portion of Chile has been creeping eastwards at several centimetres per year. It has been forced in that direction by the Nazca plate under the Pacific Ocean, which is colliding with South America from the west.
Based on plate tectonics, the scientists knew how fast the two plates were converging, and by how much that would shove the South American plate eastwards if the two plates were not sliding past each other at all — or locked.
“The similarity is simply amazing.”
In some parts of the earthquake zone, they found, the ground had indeed been moving at precisely the rate consistent with the fault being locked in this way. Other areas, however, were moving more slowly, indicating that the fault was slowly slipping in those places. This slippage, they calculated, would have released part, but not all, of the accumulating strain.
The scientists next mapped the amount by which the ground lurched back (westwards) when the fault slipped, releasing the accumulated stress. Then they compared the amount of movement to the accumulated strain.
"The similarity is simply amazing," says Matthias Rosenau, a co-author on the current study. High-strain areas lurched by as much as 10 metres. Areas of low strain moved considerably less.
Prediction hopes
Although the earthquake rupture began in one of the high-stress regions, it didn't stop when it hit the nearest lower-strain (creeping) section of the fault. Rather, Rosenau says, it "bridged the gap", allowing hundreds of kilometres of accumulated strain to be released all at once.
When it did stop, he adds, it was at regions where the GPS data suggested that the fault was locked and presumably accumulating strain rapidly. But these regions weren't ripe for failure because they had seen relatively recent, large earthquakes: a magnitude-9.5 quake on one side of the fault in 1960, and one of magnitude 7.8 in 1985 on the other side. "It stopped in areas where stresses were relaxed by prior earthquakes," Rosenau says.
Although none of these findings is conceptually groundbreaking, other scientists still find them to be important because the Chilean quake provided a rare opportunity to measure precisely what happened before and after a 'great earthquake' — defined as magnitude-8 or bigger.
The study might also help scientists to come closer to forecasting earthquakes. "The study corroborates that the earthquake process, at least along plate boundaries, is not totally random," says Rosenau.
And, he adds, monitoring other faults with similar precision might allow scientists to track the build-up of stress along them. This would help to produce better estimates of the risk and severity of earthquakes, even if precise forecasts are not possible.
