A Statistical Method of Identifying Dependent Events and Earthquake Aftershocks

A statistical method of detecting dependent events and aftershocks using spatial and temporal information was develop- ed and applied to Japan and California earthquake catalogs. On the basis of a statistic s s = 7r 2kt where r is the distance between two events, k is the normal earthquake rate and t is the time interval between the events, a decision was made whether a pair of events were dependent. The theoretical distribution of s for a catalog consisting of only independent events was compared to the actual catalogs. On the basis of the differences between the distributions, the number of inferred dependent events was determined. This discrimination technique was applied to the earth- quake catalogs of Northern Japan (1926-1960) and Southern California (1934-1960). Thirty percent of all events in the Japan catalog and 42 percent of events in the Southern California logs were identified as dependent events. The statistical properties of the catalogs without de- pendent events were examined, in particular with respect to the Poisson process. Some small discrepancies with the Pois- son process still existed using a decision threshold of s=0.02 Clusters of events were found that could not be related to any large magnitude main event.