HOW DOES AN ASPERITY BREAK - NEW ELEMENTS FROM THE WAVE-FORM INVERSION OF ACCELEROGRAMS FOR THE 2319-UT, OCTOBER 15, 1979, IMPERIAL-VALLEY AFTERSHOCK

The 16 accelerograms obtained at less than 30 km during the 2319 UT, October 15, 1979, Imperial Valley aftershock were analyzed, after having been corrected for the crack-induced anisotropy in the shallow sediments. Both P and S waveforms are used to study the earthquake source process and to infer an estimate of attenuation and site amplification effects on the waveforms at frequencies congruent-to 0.6-1 Hz and congruent-to 2-4 Hz. The comparison between the 0.6-1 Hz P and S wave seismic moment measures and their dependence on distance has provided an estimate for this parameter (1.9(+/- 40%) 10(23) dyn cm) and the bound limits for attenuation (Q(S)(f = 0.6 - 1 H z) congruent-to 20-25) and near-site amplification of shear waves at low frequency. The inversions of the 2-4 Hz S velocity waveforms for different values of rupture velocities and attenuation parameters have been preliminarily performed to investigate the variability of static slip and static stress drop estimates. A circular growth of the rupture at a constant velocity and a final elliptical shape are assumed. The bipolar character of the velocity waveform excludes the asperity models consisting of a small, highly stressed area inbedded in a large stress-free crack. Therefore, only the slip distribution of a quasidynamic crack with healing phases has been considered. The dynamic stress release has been independently measured from the amplitude level of the first acceleration pulses for various rupture velocities and attenuation. The constraint that the static stress drop does not significantly exceed the dynamic stress drop suggests relatively high Q(S) values (greater-than-or-equal-to 200) at f = 2-4 Hz and small rupture velocities v(R) congruent-to 0.5-beta-0.7-beta), beta is the shear wave velocity). The elliptical source is about 1.2 km long and 0.7 km wide (+/- 0.2 km), elongated in the NW-upward direction. The nucleation point location indicates a dominant propagation toward the NW top fault direction. For the inferred values of attenuation parameter and rupture velocity the minimum values of the dynamic and static stress drop are congruent-to 300-700 bars and are associated with about 1 m of average slip on the fault. The nucleation of the rupture is explained as an effect of the high loading stress on a larger size barrier (left unbroken by the mainshock) in which the aftershock fault area is embedded. The hetereogeneous stress and rock strength distribution on the barrier region would therefore have controlled the rupture development and arrest. An estimate of the S amplification in the considered frequency bands has been obtained through the seismic moment computed from the P wave records, which are less sensitive to site amplification and attenuation effects for the sites under study at frequencies below 5 Hz. The results show a differential amplification of S waves (congruent-to 3 at 1 Hz and congruent-to 2 at 2-4 Hz), which is attributed to the propagation in the very shallow, low-velocity sediments of the Imperial Valley.