A broad view of quake 'prediction'

Wide mapping of stress helps to pinpoint when and where the Earth will crack.

After more than a century of effort, geophysicists still can’t predict when and where earthquakes will occur. Some have suggested that it may even be impossible.

But UK and Chinese geophysicists say they have a new approach to seismic monitoring that should allow forecasting of the time, size and sometimes the location of earthquakes. Their evidence suggests they could give perhaps an hour’s warning before a small quake, or months before a big one.

Stuart Crampin of the University of Edinburgh and his colleagues say that the key is to stop fixating on specific geological faults, and to take a wider view. A fault itself does not contain the information needed to make accurate forecasts, but that information can be gleaned by looking at the stresses within several hundred kilometres of a fault, they say.

"In the past," says Crampin, "earthquake prediction research has either tried to investigate the earthquake source or tried to recognize statistical patterns in earthquake occurrence," he says. "After 120 years of investigation, both attempts have been notably unsuccessful." Since 'prediction' tends to imply one of these old approaches, Crampin and colleagues instead call their method 'stress-forecasting'.

Stress mapping

The new method exploits a phenomenon called shear-wave splitting, in which seismic ‘shear’ waves travelling through rocks are split into two components, which vibrate parallel and perpendicular relative to microscopic cracks. These two shear wave components travel through the ground at different speeds, and so reach detectors at different times.

The alignment of these microscopic cracks reflects the amount of stress in the Earth's crust. The more stress, the more the cracks are aligned, and the bigger the gap between the two waves' arrival. The work is described in Geology1

In the past, researchers have gauged the build-up of strain in earthquake zones indirectly, by looking at satellite maps of surface ground movement. But this gives only a rough idea of the stress changes deeper down, where earthquakes originate. Stresses have also been monitored directly at fault lines, but such localized data doesn't show when and where slippage will occur. “Local effects are chaotic and unpredictable,” says Crampin.

This unpredictability can be eliminated, however, by monitoring stress changes over a much wider area, he says.

Seeing a pattern

In October 1999, seismic stations in southwest Iceland reported an increase in shear-wave splitting time delays at a rate similar to that seen before a magnitude 5.1 earthquake four months earlier.

Crampin and his colleagues forecast that another quake of this magnitude would happen "soon", or a magnitude-6 within three months. There was a magnitude-5 quake three days later.

This offered a hint that shear-wave splitting might be providing important information. After further work, Crampin and his colleagues now say that detailed monitoring can make more accurate forecasts of time, size and place. They list 15 other quakes from California to China that have been preceded by characteristic changes in shear-wave splitting time delays, giving an indication of the timing and size of the coming event.

And they think a quake’s location can be deduced from other factors, including how and where the rate of change in the shear-wave time delays abruptly slows hours to weeks before the event. This may be caused by stress release, as small cracks merge into larger ones before the rupture.

Needs more data

John Rundle, a geophysicist at the University of California, Davis, who works on forecasting earthquakes by spotting statistical patterns in past events, says that any new method will need much testing before its effectiveness is clear.

Rundle believes that there is good reason to think that earthquake forecasting is not hopeless, but he suspects it will be a “big slog” to do it by starting from the basic mechanics of rock. “We know so little about the rock properties involved, and there are so many variables,” he says. Crampin says he gets around this by looking at the mechanics of a whole system rather than the mechanics of specific rocks.

There are also practical obstacles to implementing Crampin’s method: it would need more frequent and consistent data than can be generally supplied by small seismic tremors in quake zones. Underground devices radiating man-made seismic waves, as are used to map oil reservoirs, could provide this, but they are costly. The seismic sources and sensors must also be buried deep underground, adding to the expense.

Crampin says a system could be set up for monitoring all damaging quakes in a region within 400 km of a single three-borehole stress-monitoring site, for around £2-5 m ($4-10 m). Crampin and colleagues have proposed the establishment of a global network of 1,500 such sites2. “It’s very ambitious, certainly, but a huge amount of evidence demonstrates that it would work,” he says.


  1. 1

    Crampin, S. et al. Geology 36, 427-430 (2008). link

  2. 2

    Crampin, S. et al. internal report, School of GeoSciences (2008). link

Download references

Related links

Related links

Related external links

All reprints/preprints from Stuart Crampin

Earthquake prediction at the US Geological Survey

Earthquake forecasting in the QuakeSim project at JPL

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Ball, P. A broad view of quake 'prediction'. Nature (2008) doi:10.1038/news.2008.813

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.