Making reliable predictions about the time and place of major earthquakes is a key challenge in earth science. A more realistic aim, however, is to make probabilistic statements about where and when the next earthquake will take place. For more than half a century the magnitude of earthquakes has been described using the Richter scale, and the probability that an earthquake with a certain magnitude will strike has been given by the Gutenberg–Richter law: the probability is proportional to an inverse power of the magnitude. In other words, the bigger the earthquake, the less likely it is to occur.
Power laws like these, which describe how many events will occur on average in a given time period, date back to 1894 and the work of Fusakichi Omori, and they apply not just to earthquakes, but to many other natural phenomena too. Now Sumiyoshi Abe of the University of Tsukuba and Norikazu Suzuki at Nihon University, both in Japan, have discovered similar behaviour in the spatial distribution of earthquakes. Moreover, they have found that the overall correlation between earthquakes in a particular region is 10 times larger than it would be between events on a random network (Europhys. Lett. 65 581, Physica A 337 357).
In the June issue of Physics World Dietrich Stauffer of Cologne University in Germany describes this work in more detail.
