NEAR-FAULT GROUND MOTION AND ITS RELATION TO THE FAULT RUPTURE PROCESS

摘要

The effect of fault rupture characteristics on near-fault strong ground motions is investigated using a kinematic modeling approach in an attempt to identify physical processes that lead to specific ground motion patterns. The shear-stress distribution on the causative fault plane of well-documented seismic events (i.e., 1979 Imperial Valley, 1985 Michoacan, 1989 Loma Prieta, and 1999 Izmit) is calculated based on fault slip models available in the literature using the methodology proposed by Bouchon (1997) for stress field computations. In order to associate the fault rupture characteristics (i.e., slip, rupture velocity, state of stress) of the investigated earthquakes with near-fault ground motions generated by the events, forward ground motion simulations are performed using the discrete wavenumber representation method and the concept of the S-wave isochrones is exploited. The results indicate that the seismic energy radiated from the high-isochrone-velocity region of the fault arrives at the receiver within a time interval that coincides with the time window of the long-period ground motion pulse recorded at the site. Furthermore, the near-fault ground motion pulses are strongly correlated with large slip on the fault plane locally driven by high stress drop. In addition, the local rupture velocity seems to be inversely correlated to the spatial distribution of the strength excess over the fault plane confirming findings of previous studies. For various events the area of the fault that contributes to the formation of the near-fault pulse encompasses more than one "patch" of significant moment release (subevent) (e.g., 1979 Imperial Valley). This observation explains why a dislocation model with average properties (i.e., slip, rise time, etc) reproduces successfully near-fault ground motions for strike-slip faults and for dip-slip faults with intermediate-to-large earthquake magnitudes (Aki, 1979). However, for very large earthquakes, such as megathrust events on subduction zones (e.g., 1985 Michoacan), the fault region that contributes to the pulse formation encompasses individual subevents and, consequently, crack-like slip functions (rather than dislocation models) may be more appropriate for the simulation of the near-fault ground motions.