A series of foreshocks which preceded the 1999 Turkey earthquake may provide clues to increased warning times, say an international team of geoscientists.
It is known that some earthquakes are preceded by foreshocks, but so far scientists have been unable to distinguish foreshocks from the almost constant minor seismic activity at plate boundaries. Until foreshocks are followed by a major quake, there is no way of objectively determining that that's what they actually were.
In research published today in Science lead author, Dr Michel Bouchon of the National Centre for Scientific Research at The Université Joseph Fourier in Grenoble, and his co-authors in Turkey, looked at seismic records from recording stations near the epicentre of the magnitude 7.6 earthquake, which devastated the city of Izmit in northwestern Turkey.
Bouchon says they chose the Izmit earthquake because it was one of the best recorded large earthquakes to date.
"It also occurred on the fault which has been the most seismically active continental fault in the past 100 years, with nine earthquakes of magnitude 7 or greater."
Previous research has shown the earthquake was caused by two tectonic plates sliding along a 150-kilometre section of the fault-line in the brittle upper-crust, which in this region is about 17 kilometres thick. Below the upper-crust, rocks are hotter and more ductile and the two plates move past each other continuously at a rate of about 2.5 centimetres a year. The upper-crust usually remains still, but occasionally it catches up with the movement below, resulting in a jolting earthquake.
Whole lot o' shakin' goin' on
In this study, Bouchon and colleagues have identified a pattern of continuous background vibration in the 44 minutes before the earthquake struck.
During that time there was a long-duration seismic signal consisting of a succession of repetitive seismic bursts.
Bouchon says that the seismic vibration before the quake was caused by slow, irregular slipping of the fault at depth, in the zone where the earthquake began - the hypocentre.
The hypocentre for the Izmit quake is estimated to have been 13 to 17 kilometres below the surface. That puts the zone of nucleation close to the base of the brittle part of the crust.
"The fault is slipping slowly in that area, and a patch of the fault - which has a size of a few hundred metres - is resisting this slipping," he says. "The irregular slip produces the foreshocks."
According to Bouchon and co-authors, laboratory and theoretical models of earthquake nucleation predict that there should be slip instability before earthquakes, but whether that instability is large enough to be detected, and lasts for long enough to be useful, is not known.
The authors say the next step is to re-examine the near-fault seismic records of other large, well-recorded earthquakes and see if there are similar signals.
From tens of seconds to tens of minutes
Professor Phil Cummins, a seismologist at the Australian National University in Canberra, says our knowledge of what is really going on that deep in the crust is limited. But he says the study by Bouchon and colleagues is just one small step to better understanding what happens immediately before a quake.
"[This research] seems very exciting," Cummins says.
"This is the most solid evidence I've seen that there is a phase of accelerated creep prior to an earthquake that potentially could be observable a reasonable time beforehand," he says.
"[Current] earthquake early warning is aimed at giving seconds or tens of seconds warning", he says, "and in that time it's possible to take measures that might reduce fatalities, things like shut down gas mains, stop trains, shut down critical systems. So if you have tens of minutes I would say that's significant."
"I'm sure that many seismologists are going to go back and look at their records ... I'd be disappointed and a little surprised if no one else found this."
"But a lot of work still needs to be done before it can be used as an operational early warning system," he says.