Abstract

A formal algorithm for earthquake prediction is suggested, producing Zones of Increased Probability (ZIP) of size $2^{\circ}\times 2^{\circ}$ for the occurrence of large earthquakes. It is based on a relative preponderance of moderate-size earthquakes in the 2-year period preceding a large event. The algorithm has been applied to catalogs for several regions (northern and southern California, Central Asia, Caucasus, Kamchatka, Kurile Islands, Japan). The sum ($n+\tau$) of the probability $n$ of a failure-to-predict and the fraction $\tau$ of the time-space volume declared hazardous can reach values of 0.32-0.56 for the best parameter choice, depending on the region concerned. The algorithm can be incorporated into earthquake prediction monitoring systems.