In the aftermath of this week's catastrophe in Colombia, Ayala Ochert asks, is earthquake prediction possible?

On January 17 1994, an earthquake of magnitude 6.7 struck the Northridge suburb of Los Angeles, causing nearly $15 billion worth of damage. On January 17 1995, an earthquake of magnitude 6.9 struck the town of Kobe in Japan, claiming more than 5,000 lives and causing $200 billion worth of damage.

If anyone had suggested that we brace ourselves on January 17 1996 for another major earthquake, that prediction would have been no worse than any other that scientists could produce. Because, despite advances in predicting the weather and anticipating volcanic eruptions, science has little to offer when it comes to predicting when and where an earthquake will strike.

"Most scientists are not optimistic that we will be able to make skilful predictions any time in the near future. It may even be impossible," says Thomas Jordan of the Massachusetts Institute of Technology and chair of a National Academy of Sciences committee evaluating the state of earthquake science.

³ÉÈËVRÊÓƵ

"In the early 1970s there was optimism that if we had enough information then we would be able to make accurate short-term predictions, in enough time to prepare for a calamity," Jordan recalls. But as detailed observations came in, those early hopes turned sour. Some seismologists now believe that earthquakes are such "chaotic" systems that short-term prediction will remain futile.

Many have switched instead to long-term forecasting. "It is like weather forecasting," says Thomas Henyey of the University of Southern California. "We can say roughly when, where and how big an earthquake is going to be, but within certain probabilities." For example, seismologists forecast that an earthquake of magnitude 6.7 will hit Southern California in the next 30 years with a probability of 60 per cent.

³ÉÈËVRÊÓƵ

But some, like Paul Silver of the Carnegie Institution of Washington, are not yet ready to give up the holy grail of short-term prediction. "Earthquakes have been predicted in the past, so it's not unreasonable to expect that we can predict them in the future," Silver says. In 1975, Chinese scientists predicted an earthquake in Haicheng and managed to evacuate the town. Few details got through to the West because China was in the midst of its cultural revolution, but what did was cause for great hope.

In 1995, the Chinese again predicted a large earthquake in the Yunnan province in enough time to evacuate the town. In both cases, there were "foreshocks" - small earthquakes that preceded the main shock - and other indications, such as changes in the level of the water table.

"The Chinese have taken a more empirical approach," Silver says. "What we are lacking is not theoretical knowledge - we need an observational base. Otherwise it is like trying to predict the weather without satellites. We need to map the plate-boundary zone over many years. We need to know what is going on beneath the surface, but that has never been done in a systematic way and we have missed three opportunities in recent years to look at an earthquake close up."

Ross Stein of the United States Geological Survey in Menlo Park, California, agrees. "What about aftershocks? They are predictable. When we can understand them, we'll be way ahead."

³ÉÈËVRÊÓƵ

Silver remains persistent: "Whether a system is chaotic or deterministic, predictability is another issue. A volcanic eruption is also non-linear and chaotic, yet successful predictions and evacuations are made."

reuters THE RIPPLE EFFECT

Earthquakes, according to Ross Stein of the United States Geological Survey in Menlo Park, California, come in clusters. "The most common thing that you hear from people who study ancient earthquakes is this clustering in space and time on different faults," he says.

"Between the Gold Rush in 1849 and the turn of the century we had (magnitude) sixes and sevens ricocheting across the San Francisco Bay area, but not now. The 1906 San Francisco earthquake put the whole Bay area into a "stress shadow".

"During the second world war there was a series of huge and devastating earthquakes of around magnitude seven in Turkey," Stein says.

³ÉÈËVRÊÓƵ

This evidence, he says, suggests that one earthquake can set up another, even on a different faultline.

While most seismologists have been looking at faults as though they were isolated entities, he believes that faults "communicate with one another".

³ÉÈËVRÊÓƵ

When an earthquake occurs on one fault, it transfers stress through the earth to others. In some places that is enough to trigger another quake or hasten its arrival, while in others the stress is reduced and the next earthquake postponed.