In Figure 13.5 the best-fitting combined-data inversion result is compared to double-difference relocated aftershocks (Hardebeck personal communication, 2004). The slip model shows an asperity located close to the hypocenter and another in the 12 to 5 km depth range about 20 km to the NW. The two deep patches of slip are well constrained by the seismic waveform and GPS data. The slip shallower than 5 km improves the fit to the GPS data, but is not well resolved.
Aftershocks tend to locate along the edges of the large slip patches in areas where the shear stress increased due to the fault rupture.
As shown in Figure 13.6 the kinematic slip model results in a very good level of fit to the seismic waveform data. The fit to the GPS data (Figure 13.4) is also good.
A surprising result is the NW rupture directivity. The directivity is well constrained by the regional seismic data. Inspection of the waveforms at the PKD and SMM stations (Figure 13.6) reveals significant differences in S-wave pulse width and amplitude that can only be explained by a NW propagating rupture. GPS data and InSar inversion results (Ingrid Johanson, personal communication, 2005) similarly support the result that most of the mainshock slip was located NW of the epicenter. Interestingly strong motion stations to the SE have the largest peak values suggesting a component of SE-ward directivity (Shakel et al., 2005) that is not seen by the regional seismic waveform and the geodetic data sets. It is possible that the elevated strong ground motions SE of the epicenter are due to unaccounted for source process, and/or amplification effects due to the three-dimensional velocity structure of the San Andreas fault in this region (e.g. Michael and Eberhart-Phillips, 1991; Eberhart-Phillips and Michael, 1993). Our ongoing work is investigating both possibilities.
The slip in 2004 fills in the Parkfield segment that ruptured in 1934 and 1966, and therefore on this level the 2004 event is a typical repeat in the sequence of moderate Parkfield earthquakes. The 2004 rupture differs significantly however in that it nucleated at the SE end of the segment rupturing NW, while the two earlier events nucleated at the NW end rupturing SE (Bakun and McEvilly, 1979). This difference indicates that the actual triggering mechanism is very complex, and presently poorly understood. The 2004 event was possibly affected by a series of small events (M4-5) between 1992 to 1994, as well as the stress change caused by the nearby Mw6 Coalinga earthquake (Toda and Stein, 2002).
Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: email@example.com
© 2005, The Regents of the University of California