Possible criteria for predicting earthquake locations and their application to major plate boundaries of the Pacific and the Caribbean

This study presents several possible criteria for forecasting the locations of large shallow earthquakes of the near future along major plate boundaries and for assigning a crudely determined rating to those forecasts. These criteria, some of which were proposed by other investigators, are based on the past space-time pattern of large earthquakes, the lateral extent of their rupture zones, and the direction of rupture propagation. The criteria are applied in two stages. Application of the first set of these criteria to major plate boundaries along the eastern, northern, and northwestern margins of the Pacific from Chile to Japan and also to the Caribbean loop east of about 74°W results in delineation of several areas of special seismic potential along each of the boundaries. The phrase ‘special seismic potential’ is used in this paper only to indicate those segments of plate margins that fulfill certain specific criteria. However, if the criteria are valid, at least some and perhaps most large shallow earthquakes of the near future within the zones examined will occur near these locations. At present the validity of the criteria is not firmly established, and profound social changes based on these predictions are uncalled for, but the forecast presented here can, at the very least, serve as a guide in selecting areas for intensive study and instrumentation prior to the occurrence of a major earthquake. The criteria have greater usefulness in those regions where the rupture zones of large earthquakes have nearly covered the seismic zone in recent years and only a few gaps remain. In certain areas where additional information is available the subsequent application of a second set of supplementary criteria focuses special attention on certain of the areas delimited by the first set of criteria. Rupture zones of large shallow earthquakes as determined from aftershock locations, intensities, tsunami descriptions, and coastal uplift tend to cover plate boundaries without significant overlap. Because of this tendency to fill in the plate boundaries and because large earthquakes account for such a high percentage of the total seismic moment released in a seismic zone, segments of the zone that have not experienced a large earthquake recently are likely locations for future shocks. Along the Kurile-Kamchatka area, the Alaska-Aleutian arc, and much of western South America, the ‘large’ earthquakes tend to be great earthquakes with rupture zones sometimes extending many hundreds of kilometers. Along western Middle America, however, the large earthquakes have rupture zones no larger than 100–200 km. Although great earthquakes with ruptures hundreds of kilometers in length have occurred along some segments of the Caribbean loop, certain other segments have no known history of great earthquakes. Thus there is no recognizable class of large earthquakes around the Caribbean loop. Nearly all the extensive parts of the plate boundaries examined in this study tended to be covered by rupture zones of large shallow earthquakes. Therefore tectonic strain along segments of plate boundaries considered here appears to be relieved in most cases by periodic large earthquakes, regardless of the possible presence of fault creep and small-magnitude earthquakes. Thus fault segments characterized by aseismic creep should still be considered potential sites for large earthquakes unless persuasive evidence to the contrary is forthcoming. For all island arcs examined, the epicenters of the main shocks tended to be located near the inner or landward side of the aftershock zone. Since most large earthquakes near island arcs occur on dipping fault planes, this phenomenon suggests that, during large thrust earthquakes, rupture initiates at depth and propagates upward and outward along the plate interface. This phenomenon occurs not only for shocks of moderate.