HR: 11:05h
AN: S31G-04    [PDF]
TI: Viscoelasticity, Postseismic Slip, Fault Interactions, and the Recurrence of Large Earthquakes
AU: * Michael, A J
EM: michael@usgs.gov
AF: USGS, MS 977 345 Middlefield Road, Menlo Park, CA 94019 United States
AB: Since Reid formulated the elastic rebound hypothesis, our view of earthquake occurrence has been based on the idea of uniform loading leading to recurrent failures. This view was reinforced by the discovery of plate tectonics but, recently, many studies have demonstrated the existence of transient, high deformation rates after large earthquakes due to either viscoelastic processes or post-seismic slip. Viscoelastic response of the lower crust and upper mantle to large earthquakes results in temporarily higher deformation rates in the region surrounding the mainshock. These higher deformation rates will result in faster than average reloading of strain energy onto the mainshock fault. If post-seismic slip is a planar downward extension of the mainshock, then the post-seismic slip will increase the stress stored on the mainshock fault plane. Again, this will result in faster than average loading of strain energy onto the mainshock fault plane for some time immediately following the mainshock. Thus, the loading of strain energy onto seismogenic faults is not temporally uniform and, in this study, I consider the effects of transient deformation on the estimation of earthquake probabilities by modifying the Brownian Passage Time (BPT) model of earthquake recurrence. The BPT inter-event time distribution is derived from a process where a state variable starts at 0, evolves by a superposition of a linear trend and Gaussian white noise until reaching a value of 1 when failure occurs and state is reset to 0. For earthquake recurrence the linear trend represents the uniform deformation due to plate motions and the noise represents fault interactions and other unknown perturbations to the process. To approximate the effects of viscoelasticity and post-seismic slip; I add a decaying exponential term to the BPT model's uniform loading term. The resulting inter-event time distributions remain approximately lognormal but the balance between the level of "noise"and the coefficient of variability of the inter-event time distribution changes depending on the shape of the loading function. For a given level of noise in the loading process, transient deformation generally has the effect of increasing the coefficient of variability of earthquake inter-event times. Conversely, the level of noise needed to achieve a given level of variability is reduced when transient deformation is included. This would then increase the effect of known fault-interactions modeled as steps in state due to changes in stress or strain because the steps in state due to the known fault-interactions would be larger with respect to the "noise." If earthquake probabilities are estimated solely by empirically fitting inter-event times, the transient loading effect need not be modeled because it produces a similar-shaped distribution as the standard BPT model. However, if the goal is to estimate earthquake probabilities based on the physics of seismogenic processes, then transient deformation must be included. For example, a transient-loading BPT model for the 1906 San Francisco earthquake produces a 60% greater variability in earthquake recurrence when compared to a uniform loading model and the same level of noise. Thus, if transient deformation is included in a BPT model for the faults of the San Francisco Bay region and the noise level is reduced to maintain the same variability in earthquake recurrence, then the effect of the stress change of the 1906 earthquake is greater than for a constant-loading BPT model. Prior work may therefore underestimate the effect of the 1906 stress shadow.
DE: 7209 Earthquake dynamics and mechanics
DE: 7223 Seismic hazard assessment and prediction
DE: 8120 Dynamics of lithosphere and mantle--general
DE: 8123 Dynamics, seismotectonics
SC: Seismology [S]
MN: 2003 Fall Meeting