Evidence for fault asperities from systematic time-domain modeling of teleseismic waveforms

A simple method for determining which events prior to a main shock may be "true" foreshocks, events which are caused by the same failure process as that which triggers the main shock, is proposed. An event is regarded as a "true" foreshock if it takes place within a certain time period of, is at least half a magnitude unit smaller than and occurs within the aftershock zone of a main shock. The time periods of potential foreshock occurrence computed from local and regional seismicity rates for four events were calculated to range from several days to several weeks prior to the corresponding main shocks. A detailed analysis of two foreshocks (FS-1, M = 7, and FS-2, M = 6-1/2), the main shock (M, M = 7-1/2) and two aftershocks (A-1, M = 6-3/4, and A-2, M = 7) from the August 11, 1965 New Hebrides Islands earthquake sequence is presented. Focal mechanisms, depths, moments, time function durations and directions of rupture (if they could be inferred) for the events have been found using time domain synthetic seismograms of the far-field body waves and surface waves. The focal mechanisms of all the events except A-2 are consistent with faulting on the interface between the subducting Indian plate and the overriding Pacific plate. A-2 was an event on a steeply-dipping fault which ruptured into the underthrusting plate. The observed radiation patterns for the Rayleigh and Love waves from these events are consistent with the results of the body-wave analysis. The theoretical static vertical surface displacements computed from the teleseismic source model for M are much smaller than the observed coastal uplift indicating that very long-period deformations accompanied the earthquake. The seismicity during the sequence migrated first from northwest to southeast and then toward the southwest and northeast. A detailed source study of the short-period P waves from the Borrego Mountain earthquake in Southern California is reported. The short-period waveforms at different stations show good coherence, indicating that the seismograms contain reliable information from the source region. From simultaneous long-period-short-period deconvolutions the sP phase was found to consist of two separate pulses. Synthetic seismograms computed from the long-period source model of Burdick and Mellman (1976) did not match the data very well while synthetics with two high-frequency point sources did. The results of a waveform inversion analysis indicate that both sources were located at a depth of 8 km, had similar focal mechanisms, had time function durations of approximately 2 seconds and occurred about 2.2 seconds apart. Synthetic displacement, velocity and acceleration records, computed from a smoothed version of the teleseismic, short-period source model, fit both the amplitudes and waveforms of the SH wavetrain from the strong-motion data from El Centro, California. The existence of asperities on the fault zones in the two source regions is inferred. In the New Hebrides three asperities are proposed--one at the northern end of the 1965 seismic zone, one between the islands of Santo and Mallikolo, and one near the southern end of the main shock fault plane. The sequence of events reflects a pattern of the loading and breaking of asperities on the fault. For the Borrego Mountain earthquake the short-period sources represent the breaking of two asperities. The stress drops of the two events were several hundred bars each while the average stress drop for the entire event was about 20 bars. For both the New Hebrides events and the Borrego Mountain earthquake, the area of the asperities which ruptured during the main event was no more than 15% of the total fault area.