Inversion for the physical parameters that control the source dynamics of the 2004 Parkfield earthquake

Abstract A fully dynamic inversion for the earthquake source process, in which the geometry of the rupture area, the stress conditions, and frictional properties on the fault are obtained, is carried out by inverting displacement records for the 2004 M w 6.0 Parkfield, California, earthquake. The rupture area of the earthquake is modeled using elliptical patches, and seismograms from 10 near‐field digital stations are used. Synthetic tests to investigate the performance of the inversion in retrieving the rupture process demonstrate that we can reliably recover the large‐scale features of the spatiotemporal distribution of slip. To investigate the stress conditions and frictional properties of the fault under which we produce a rupture model that fits the observed data, we explore the parameter space using a Monte Carlo method and find an optimal region where the source models fit the data well. The best fitting rupture process is shown to occur mainly within one horizontal elliptical region, 22 km long along strike and 4 km wide along depth. The seismic moment is 1.2 × 10 18 N m, and the stress drop over the ellipse is ∼4 MPa. The rupture speed, nearly constant during the entire rupture process, is ∼2.9 km/s. The dimensionless quantity κ (roughly the strain energy change per unit fault surface divided by the energy release rate), which includes information on the stress and frictional properties on the fault, is found to be ∼1.4 for the ellipse and strongly controls the rupture process along with the size of the initial circular patch that initiates the earthquake.