Long-period surface wave inversion for source parameters of the 18 October 1989 Loma Prieta earthquake

Abstract

We performed simultaneous seismic moment tensor inversions of long-period (157–288 s) fundamental mode Rayleigh and Love waves from the 1989 Loma Prieta earthquake to further constrain the source process of the event. Utilizing a two-step spectral inversion technique, we explored the model dependence and centroid location sensitivity of the long-period surface wave analysis to assess the confidence bounds on the results. We found that estimates of the source duration and depth are highly dependent on the choice of propagation, attenuation and source velocity structure models. Including centroid location parameters in the inversion stabilizes moment tensor estimates but yields a biased location away from the epicenter due to model inaccuracies. Our source duration estimate is 11 ± 5 s using a recent velocity model, MPA, with the centroid time of 6 s being significantly less than earlier surface wave studies (centroids of 10–22 s) and hence more compatible with both body wave and strong motion duration estimates. An unconstrained moment tensor inversion at the optimum centroid location yields a stable major double-couple solution (strike = 124 ± 6°; dip = 67 ± 6°; rake = 126 ± 7°) and a seismic moment estimate (3.0 ± 0.2 × 10 19 Nm; M w = 6.9) similar to earlier long-period studies and body wave and geodetic results. The surface wave centroid depth estimate is 22 ± 11 km, which overlaps the body wave estimates (13 ± 5 km). Thus, surface wave source parameters for the Loma Prieta event, allowing for plausible model dependence, are fully compatible with body wave determinations, and there is no evidence for any anomalous coseismic long-period source process.