Nonlinear strain buildup and the earthquake cycle on the San Andreas Fault

Abstract

Two contrasting models of the earthquake deformation cycle on strike slip faults predict significant temporal declines in shear strain rate near the fault, accompanied by a progressive broadening of the zone of deformation adjacent to it. In the thin lithosphere model, transient deformation results from flow in the asthenosphere due to stress relaxation following faulting through most or all of the lithosphere. For an earth model with a thick elastic lithosphere (plate thickness » depth of seismic slip), transient motions are due to postearthquake aseisrnic slip below the coseismic fault plane. Data from the San Andreas fault indicate a long‐term temporal decrease in strain rate that persists for at least 30 years and may extend through the entire earthquake cycle. Observations support a cycle‐long rate decrease and a temporal spreading of the deformation profile only if movement cycles on the northern and southern locked sections of the fault are basically similar. If so, the usually lower strain rates and broader deformation zone currently observed on the southern San Andreas represent a later evolutionary stage of the northern locked section, where a great earthquake is a more recent occurrence. Although the data allow some extreme models to be discarded, no sufficiently strong constraints exist to decide between the thin and thick lithosphere models. Regardless of the appropriate model the geodetic observations themselves indicate that strain buildup is sufficiently nonlinear to cause significant departures from recurrence estimates based on linear strain accumulation and the time‐predictable model.

Coulomb stress accumulation along the San Andreas Fault system

“…Savage [1990] argues for a 30 km locking depth estimate for this portion of the Transverse Ranges using an elastic plate model overlying a viscoelastic half-space, providing a realistic match to our simple elastic half-space model. Thatcher [1983] illustrated a similar comparison for this region, making the valid point that two physically different mechanisms (elastic vs. viscoelastic half-space) produce indistinguishable surface deformation. Again, we note the evident unmodeled velocities to the east of the fault trace that are related to complex deformation patterns of the ECSZ.…”

Influence of anelastic surface layers on postseismic thrust fault deformation

“…While long-term geodetic measurements of the earthquake cycle appear to demonstrate lower crustal viscoelasticity with decadal timescales [e.g., Thatcher, 1983], recent studies including those of the 1992 Landers [Shen et al, 1994], 1994 Northridge [Heflin et al, 1998], and 1994 Sanriku-Haruka-Oki, Japan [Heki et al, 1997], events show transient motions on timescales of several months. In particular, Donnellan and Lyzenga [1998] have found that the M W 6.7 Northridge thrust earthquake was followed by -1 year timescale relaxation, associated with the combined effects of fault plane afterslip and bulk relaxation in the uppermost shallow crust.…”

Inference of Multiple Earthquake-Cycle Relaxation Timescales from Irregular Geodetic Sampling of Interseismic Deformation

“…Models for time-dependent earthquake-cycle deformation have generally focused on either the diffuse deformation of a linear viscoelastic subseismogenic layer (e.g., Thatcher, 1983;Savage, 1990;Dixon et al, 2002Dixon et al, , 2003Ergintav et al, 2002;Segall, 2002;Hilley et al, 2005, Motagh et al, 2007 or localized shear on a down-dip extension of the coseismic fault zone (e.g., Marone et al, 1991;Bürgmann et al, 2002), whereas recently more postseismic models have incorporated both (e.g., Freed et al, 2006;Johnson et al, 2009). Both deformation mechanism models have been extended to predict time-variable surface deformation throughout the entire earthquake cycle under the assumption of periodic earthquake occurrence (Savage and Prescott, 1978;Cohen, 1982;Johnson and Segall, 2004;Hetland and Hager, 2005) and to clustered and nonperiodic earthquake occurrence (Meade and Hager, 2004;Hetland and Hager, 2006).…”