Earthquake prediction has been the Waterloo of many a scientist: Each tremor is unique, foiling attempts to zero in on advance warning signs. But Vladimir Keilis-Borok, a geophysicist at UCLA, and his team may have finally cracked the code. By monitoring past and present activity in seismic hot spots, they have issued dead-on predictions of major quakes that occurred in San Simeon, California, and Hokkaido, Japan, during the second half of 2003.
Keilis-Borok bases his work on the tip-of-the-iceberg premise that patterns of small seismic disturbances hint at the onset of much larger quakes. He looks for chains of minor earthquakes, then examines the seismological history of the area where they took place. If he finds such telltale symptoms as frequent small, clustered quakes during the preceding five years, he concludes that there is a high likelihood of a major quake in that region within nine months. If no historical signs are present, he classifies the miniquakes as probable anomalies. Keilis-Borok calls his backward-looking technique “tail wags the dog,” illustrating how minor “tail” quakes that barely disturb a seismograph can predict a massive “dog” quake. “The ‘tail’ is a potential precursor,” he says. “We then look in the larger time interval to see whether there is preparation or whether it’s a false alarm.”
Although he has an enviable track record so far, Keilis-Borok is the first to admit that his forecasting method will prove its worth only through further testing. “We need to predict more earthquakes and accumulate the errors and successes,” he says. His next prediction? A magnitude 6.4 quake will rock the Californian Mojave this fall. Desert dwellers, be warned.