INDUSTRY

A new method of more accurately forecasting earthquakes could also have applications in predicting tsunamis, changes in the stockmarket and political events, a leading scientist says. Professor John Rundle, the Director of the Centre for Computational Science and Engineering at the University of California, said these were all examples of threshold systems and were subject to similar sets of rules. He told a conference in Perth, Australia that his team had been concentrating on developing a computer-based method of forecasting earthquakes with much greater precision than was previously possible. "Most people would say that earthquakes can't be predicted or forecast, and indeed there have been many notable failures," Prof Rundle said. However, he said his work overcame the main obstacle to earthquake prediction - time. By using a computer to simulate activity over a whole fault system, it was possible to study thousands of years of simulation, Prof Rundle said. The program simulates the plates in a fault moving away from each other at a constant rate to show the point, or threshold, at which the stresses become too great and an earthquake occurs to reduce the stress. "What it does is locate the likely sites of large earthquakes during a specified time window of a number of years, for example five to 10 years," he said. "What we've seen so far using the test we have ... is that most of the significant earthquakes in California did occur in places we have forecast - about 15 out of 17 so far. "The one thing it doesn't do yet is tell us exactly when these events might occur, but we think that by modifying the method ... we may be able to reduce that uncertainty by quite a bit." Prof Rundle told the Sir Mark Oliphant International Conference on Thresholds and Pattern Dynamics that the same principle could be applied to help predict other events, including tsunamis. "This is a type of system that has triggers - so if a stock price goes up to some value and you decide to sell, then that is a threshold," he said. "Lots of other systems have the same types of properties ... so the method we propose can be adapted because the fundamental dynamic is very similar." While people could respond to a prediction in the stockmarket and potentially change the course of events, an earthquake could not be influenced in the same way, making those forecasts more accurate, he said.