The 1966 Parkfield earthquake accelerograms are analyzed with particular attention to clarifying the relationship between the observed motions and the crack propagation. The basic finding is that the crack propagation velocity was supersonic relative to the shear-wave velocity. A consistent interpretation of the accelerograms follows.
A crack propagated south along the E branch of the San Andreas Fault from the initial focus to about km N of Station 2, and terminated there. A second earthquake was triggered near the initial focus, on the W fault branch, 0.55 sec after the beginning of the first fracture. This dislocation propagated south along the W branch past Station 2. With the assumption that the crack propagated along ordinary wave paths, the average Mach number (crack velocity/shear-wave velocity) for the propagation to Station 2 is 1.06.
Within about 0.3 sec after the crack propagated to the ground surface 0.08 km from Station 2, a permanent displacement of 2 cm took place parallel to the fault (NW). At about 1.3 sec after the dislocation arrival, relatively large nonpermanent displacements commenced in the horizontal components (to 9 cm in the parallel and 36 cm in the transverse), directed NW and toward the fault. Approximately equal displacements were occurring on the opposite (NE) side of the fault from Station 2, directed in the opposite sense, the parallel displacement being SE, and the transverse being SW toward the fault. The result of the compression on the fault plane was a sudden local deceleration of the slipping. The maximum accelerations occurred at the time. A Mach wave was formed and is represented on the records by the acceleration maxima. There is also an earlier, lower amplitude Mach wave from the initial earthquake. A third Mach wave followed the second by 0.82 sec. This is from the propagation of another dislocation down the E fault branch, again from the focal region, probably to the same location where the initial crack terminated. This third earthquake was promptly triggered by the second.
The dislocation near Station 2 stopped about 2.9 sec after it began. The crack propagated another 9 km SE of the station before stopping. Two possible stopping events, each with P and S phases, were recorded at Station 2. The permanent displacement calculated from the parallel acceleration component is in good agreement with the observed offset at Station 2. The large nonpermanent horizontal displacements are at present unexplained.