FOR the past year, I have been doing something that many of my fellow seismologists consider foolish -- even ridiculous. I have been issuing regular and frequent public earthquake forecasts for New England, broadcast as probability values in much the same way that weather forecasters predict the probability of rain (see www.bc.edu/westonobservatory). My colleagues shake their heads because they consider earthquake forecasting unreliable. And they are right: to date I have had few successes. But I am determined to continue, and I believe other seismologists should start doing the same for other parts of the US, and ideally the whole world. That's because the only way to improve earthquake prediction is to invest much more time and money in it, and the only way to attract more funding is to raise the level of public interest, and convince people of the need for more accurate forecasts.
That there is a desperate need for such forecasts should be beyond doubt. Since 1990, earthquakes have killed more than 135,000 people around the world and injured many more. At a stroke, they can devastate entire communities. Every year, the cost of earthquake damage to property amounts to many hundreds of millions of dollars. And still we are incapable of predicting when and where they will strike.
A successful earthquake forecasting scheme requires sophisticated monitoring infrastructure as well as a dedicated research programme. In 2000, the US Congress authorised a five-year grant worth around $34 million a year for a programme run by the US Geological Survey called the Advanced National Seismic System. The ANSS is designed to expand and upgrade regional earthquake monitoring throughout the US. It would provide new equipment for the gathering, analysis and communication of seismic data that would vastly improve seismologists' chances of providing more reliable earthquake forecasts. However, Congress voted only $4 million for ANSS in 2002 and the same in 2003, and President Bush's draft budget for 2004 proposes cutting this to $1.9 million.
Why this lack of commitment? It stems, I believe, from the woefully low level of public understanding and interest. Even the most destructive earthquakes are soon forgotten. The Los Angeles quake of 1994 caused damage worth $20 billion, and the Seattle quake of 2001 cost $4 billion in property damage, yet they are no more than dim memories for those who were not affected. Out of sight is out of mind, until the next big one strikes -- and one thing we do know for certain is that there will be a next one.
Hence my proposal for weekly or fortnightly forecasts. Given the infrequency of major earthquakes, I focus primarily on learning to forecast the smaller quakes of magnitudes 2.5 to 4.5 on the Richter scale, which people might feel but which do no damage. To those who say this serves no practical purpose, I would say they are missing the point. Weather forecasters not only tell us about severe storms; they also tell us when things are calm. Forecasts of even minor seismic activity would increase people's awareness of what was going on beneath their feet, and remind them of the need to predict more serious events. At the same time, seismologists would have frequent opportunities to hone their forecasting skills.
Even with the current limited funding, seismologists are learning to make fairly reliable short-term earthquake forecasts of aftershocks. Omori's law -- the decrease in the number of aftershocks with time after a large earthquake -- has a well-established statistical basis and can be used to forecast the probabilities of future aftershocks when a significant earthquake takes place. Recent studies have also shown that changes in static stress around faults that move in large earthquakes dictate the location and rates of aftershocks, and that shock waves from strong earthquakes can trigger small and moderate earthquakes hundreds of kilometres from the main epicentre.
I accept that at the outset of a national or global forecasting programme, most of the predictions that seismologists publish might well be wrong. But I don't believe this will matter, for they will learn from both their failures and their successes. With experience, their forecasting abilities are guaranteed to improve. By concentrating on weaker seismological events, we will accumulate enough forecast cases over a few years to enable us to test in a quantitative, statistical way how well they are working. If we were to focus solely on bigger earthquakes, we might need years or even decades to evaluate a single forecast in a seismically active region.
Only if earthquake forecasts become commonplace will the public start to understand what we know about coming earthquakes and what we don't. People might initially laugh at our mistakes, just as they deride inaccurate weather forecasts. But earthquakes, like hurricanes, are potentially devastating. Better that we suffer insults on the path to better predictions than to remain in perpetual ignorance.