Earthquake rupture and ground-motion: The 2004 Mw 6.0 Parkfield earthquake

We present a study of the faulting process of the 2004 MW6 Parkfield earthquake and its resulting ground-motion. We use both seismic data alone and a combination of seismic and geodetic data to infer the temporal evolution of slip on the fault plane during the 2004 Parkfield earthquake. The seismic dataset comprises strong-motion accelerograms recorded at 43 sites and the geodetic dataset is composed of 1-Hz GPS displacement records observed at 13 stations. We use the geodetic data to infer the static distribution of slip on the fault. The seismic data serves to retrieve the full space and time evolution of slip. The seismic data is corrected in order to account for site effects. The pattern of ground-motion observed during the 2004 Parkfield earthquake can be partly explained by site effects. Three-dimensional material heterogeneity in the Parkfield region also contribute to the strong shaking observed northwest of the hypocenter (over twice the acceleration of gravity). We employ a non-linear least-squares algorithm to infer the static slip distribution from the geodetic data. We then use the geodetic data to constrain the slip amplitude in a non-linear inversion of the seismic data that yields the complete temporal evolution of slip. The 2004 Parkfield earthquake, as imaged in our model, started with a very strong and localized burst of energy from the hypocenter; here the rupture velocity was very fast. We locate a second region of significant slip northwest of the hypocenter where slip was slower and spread over a region wider than the initial asperity at the hypocenter. Trade-offs between various model parameters result in uncertainties in the faulting models, which we assess using various techniques. We also study the rupture mechanism of the 1966 MW6 Parkfield earthquake. We conclude that the two most recent Parkfield earthquakes do not share identical slip distributions. These two earthquakes actually appear to be more complementary than characteristic. Thus, an extreme interpretation of the characteristic earthquake hypothesis can be rejected for Parkfield based on seismic data.