Title: Constraints on rheological structure of the lithosphere from geodetic observations of postseismic transients due to the Landers and Hector Mine earthquakes,
We investigate the transient deformation following the 1992 Mw 7.3 Landers and the 1999 Mw 7.1 Hector Mine earthquakes (Mojave desert,
Southern California) using Global Positioning System (GPS) and Synthetic Aperture Radar Interferometry (InSAR) data collected between
1992 and 2010. We test possible mechanisms of postseismic relaxation using physically-based time-dependent models of deformation driven by coseismic stress changes. Considered mechanisms include viscoelastic flow in the lower crust and upper mantle, afterslip governed by a rate-dependent friction, and poroelastic rebound. We find that both afterslip and viscoelastic relaxation models can explain the post-Landers horizontal GPS displacements equally well. However afterslip gives rise to vertical displacements of opposite polarity to the ones measured by GPS and InSAR [e.g., Pollitz et al. 2001].
Viscoelastic models involving a high-viscosity lower crust and low-viscosity upper mantle gives rise to large wavelength line-of-sight (LOS) displacements which are not consistent with the InSAR observations. Poroelastic models are consistent with wavelength and amplitude of InSAR LOS displacements and campaign GPS vertical data, but alone cannot explain the azimuth and amplitude of horizontal displacements. None of the tested "single mechanism" models can explain all of the available geodetic measurements implying a combination of multiple mechanisms. Assuming that forward models of postseismic deformation are separable in space and time we devise a linear inversion to infer the relative contribution of given relaxation mechanisms. A joint inversion of the post-Landers and post-Hector Mine data favors a combination of poroelastic rebound (with ~3% reduction in the effective elastic moduli) and viscoelastic relaxation in the lower crust and upper mantle as the dominant mechanisms of transient deformation. The data neither exclude nor require a contribution of afterslip on a deep extension of the ruptured faults. While the geodetic data cannot resolve fine details of the viscosity variations with depth, the best-fitting model suggests a weak lower crust and weak asthenosphere, separated by a ~
20km-thick layer of strong upper mantle, consistent with the "jelly sandwich" model [e.g., Burov & Watts 2006]. Our inversions also indicate possible lateral variations in the viscosity structure of the lower crust, with a weak zone extending from the Mojave block towards the San Gorgonio Bend of the San Andreas Fault.