There were significant changes in seismicity preceding a large earthquake, the Urakawa-Oki earthquake of 21 March 1982. This event, a MJMA = 7.1 thrust-fault, crustal earthquake, is the largest from a data set of 35,000 earthquakes beneath the southeastern corner of Hokkaido Island (Japan) during the period July 1976 through December 1986.
There was a quiescence for the 2 years preceding the main shock, surrounding, but not including, the immediate epicenter, and only for magnitudes ≧ 3. During the same time span, there was an increase in the number of small (magnitudes ≦ 2.4) events restricted to the general focal region of the main shock. The magnitude dependence of the precursory seismicity patterns is a departure from self-similarity, one of the currently prevalent assumptions in earthquake source models.
These seismicity anomalies can be seen using conventional methods such as cumulative number versus time for selected magnitudes in a bounded geographical region. However, such displays require a priori knowledge or assumptions about the prospective target region and magnitude range. For this anomaly, there was no reason to target the epicentral region in advance, and the usual assumption that quiescence should be most evident at small magnitudes was not appropriate here. We have developed an efficient, objective procedure for alerting one to the possible existence of a developing anomaly and then analyzing its spatial bounds and its magnitude dependence. This procedure has potential value as a general monitoring tool.