Journal of Conference Abstracts

Friction During Faulting: Inferring Rupture Dynamics from Source Kinematics

William L. Ellsworth (ellswrth@andreas.wr.usgs.gov)¹ & Eiichi Fukuyama (fuku@bosai.go.jp)²

¹U. S. Geological Survey, Menlo Park, CA, 94025, U.S.A.

²National Research Institute for Earth Science and Disaster Prevention, Tsukuba, 305, Japan

When an earthquake occurs, the frictional resistance of the fault drops from an initial static level to a new level that remains after sliding stops. The nature of this evolution has been extensively studied under laboratory conditions and through theoretical investigations, but has only recently come under study using seismic observations. The seismic approach begins with the construction of a kinematic description of the sliding history through the inversion of near and far field waves. The resulting time history of slip at each point on the fault is then applied as the boundary condition to a elastodynamic model, from which dynamic stresses on the fault plane are obtained. Taken together, the kinematic displacements and velocities, and dynamic stresses describe the evolution path for fault friction during the earthquake. Application of these procedures to the M 6.1 March 26, 1997 Northwest Kagoshima Prefecture, Japan, earthquake yields a frictional evolution path characterized by a displacement weakening distance of approximately 5 cm. Slip velocity jumps within 0.1 s to a high value, and maintains a velocity-weakening dependence during sliding. The kinematics of this rupture are also of interest, as they are dominated by a strong velocity pulse that appears 1.0 s into the event and propagates at the P-wave velocity from west to east across the already broken fault plane. Forward models confirm the high propagation velocity of this pulse, which produces a distinctive waveform in the forward propagation direction. The rupture appears to heal shortly after the passage of the velocity pulse, suggesting that it represents a slip pulse.