We analyze the spatial-temporal-energy distribution of shallow (depth, h < 70 km) seismicity (surface wave magnitude, M-s greater than or equal to 7) occurring in the focal and adjacent regions of very great shallow earthquakes (rupture lengths of the order of hundreds of kilometers) in the circum-Pacific seismic belt for as many decades as possible before and after the occurrence of the main shock. For this purpose we have revised the catalog of data for the world instrumental seismicity (1900-1995) by removing a series of heterogeneities in the earthquake reporting and magnitude determination. This revised catalog is fairly complete for earthquakes with M greater than or equal to similar to 7 since early in the century. Our findings indicate that once a main shock-aftershock sequence occurs the focal region enters a period of relative quiescence for large (M-s greater than or equal to 7) earthquakes. Significant seismic activity M-s greater than or equal to 7) occurs during many decades prior to the next main shock. This activity is fairly continuous or random in time and lasts until the onset of the main event but clusters in space around the ends of the forthcoming rupture and/or in the vicinity of the principal epicenter (except in the case of rare bilateral breaks the main epicenter is located at one of these ends). Thus major portions of the rupture zone experience no events with M-s greater than or equal to 7 during the seismic cycle. Larger aftershocks also cluster near the edges of the rupture and near the principal epicenter. Preshock activity, minutes to weeks before the main shock (if any), occurs, in general, near the main epicenter. We also find that the nucleation and ends of the rupture often coincide in space with recognizable geometrical and/or geological inhomogeneities, ''barriers'' existing in the lithosphere. It appears that while strain energy accumulates, high stress concentrations often occur at the ends of the future rupture zone, which results in the long-term foreshock activity and in the subsequent initiation of the main rupture from within those regions. We suggest a simple mechanical model that explains the observed pattern of earthquake occurrence for both unilateral and bilateral ruptures. In all cases of great earthquakes that have occurred this century far which a long and complete (to a level of 2 to 3 orders of magnitude smaller than the moment of the main shock) catalog of prior seismicity exists, we can not recognize any statistically significant ''quiescent'' or ''doughnut'' patterns in the preceding seismicity at the M-s greater than or equal to 7 level.
Wu876Times Cited:12Cited References Count:69