The epicenter of the Landers, California, earthquake (28 June 1992; MW = 7.3) was located near the TERRAscope network of broadband seismic stations. The direct Rayleigh wave arrivals, R1, were clipped, and the first two later arrivals, R2 and R3, were contaminated by the waves from a large aftershock, but, as reported by Kanamori et al. (1992a), the amplitudes of R4 and later great circle Rayleigh wave arrivals (fundamental mode spheroidal free oscillations) are about 10 times larger than predicted by synthetic seismograms for a spherically symmetric earth model. We show that, for the moment tensor of the Landers event (predominantly vertical strike slip), the amplitudes of synthetics at the TERRAscope stations for a laterally heterogeneous, rotating, elliptical model are about 10 times greater than those for a spherically symmetric model. Because the anomaly ratio is sensitive to both the source model and the three-dimensional (3D) earth model, we do not attempt to reproduce the exact anomaly ratios recorded by the various stations.
To explain the existence of near-field amplitude anomalies in general, we use the first-order Born approximation to find the perturbation to the synthetic seismogram resulting from lateral heterogeneity, ellipticity, and the earth's rotation. In a coordinate system with the source on the z axis a point-source strike-slip earthquake on a vertical fault plane in a spherically symmetric medium excites Rayleigh waves with azimuthal order ±2 only; these waves have a near-field vertical displacement of zero at the source; the displacement increases with the square of epicentral distance for any given azimuth. Coupling as a result of asphericity allows such a source to excite Rayleigh waves with azimuthal order zero, whose near-field amplitude is independent of epicentral distance, thereby generating large near-field amplitude anomalies. We conduct numerical experiments to study the influence of various parameters on near-field amplitude anomalies.