A study of the tendency of deep and shallow earthquakes in the Fiji-Tonga-Kermadec region to cluster in space and time revealed that (a) in general, deep earthquakes do not form either aftershock sequences of swarms of the types commonly observed in series of shallow shocks throughout the world; (b) a small percentage of the deep earthquakes cluster in the form of multiplets, i.e., small numbers of events closely grouped in space and time; and (c) during the 7-year interval studied, clustering of the shallow events in the Kermadec region was markedly greater than that of shallow events in the Tongan region; clustering of shallow events in the Tongan area was, in turn, greater than that of the deep shocks in the Fijian region. The data provide new constraints for hypotheses of focal mechanisms.
Most deep multiplets were doublets, and in no case were more than 12 events observed per multiplet. Resolution of the analysis was sufficient to show that, in the case of at least one multiplet, separation between events was smaller than 2 km, but that in other cases it was larger than 5 km. Separations between events in time were sometimes smaller than 2 seconds, but in no case was the relation in time and space such that a later event of a multiplet could not have been dependent upon an earlier event, i.e., the time between events was greater than that required for seismic wave propagation between the two foci.
A narrow linear feature about 60 km long is defined by a spatially progressive sequence of deep multiplets that were observed during 1965. These deep multiplets are also confined to a narrow range of depths between about 625 and 660 km. This feature seems to form a portion of the lower edge of the narrow inclined zone of deep earthquakes. Patterns of first motions radiated from earthquakes in the linear feature are generally similar to one another and are in agreement with a model of the source corresponding to a double-couple or shear dislocation. Variations among the radiation patterns, although minor, are distinct and indicate some heterogeneity in the detailed distribution of stress. The data of this study do not support the hypothesis of a volumetric phase change as the mechanism of the deep-focus earthquakes. Although there is no existing hypothesis that does predict completely the results determined here, the data appear to agree best with the mechanism proposed by Orowan.
A significant source of data for this work is a new network of five seismograph stations now operating in Fiji and Tonga. This study, under the U.S. Upper Mantle Program, is part of a more comprehensive investigation of deep earthquakes in the region.