Structural geologists have long noted phenomena such as "basin inversion" and "reactivation" which suggest that deformational strength can be inherited. This is sometimes ascribed to weakening by faults, but faults serve only to release the near-surface strain moments originating from flow in the ductile regime. (This is readily observed, for example, at evolving plate boundaries such as the Marlborough region of South Island New Zealand: New major faults will develop when the locus of deeper deformation shifts by as little as 15 to 20 km). As inheritance sometimes occurs across time scales exceeding that for thermal equilibration of the lithosphere, flow strength should be partially controlled by a less transient field than temperature. The relationship of elastic thickness T_e to heat flow Q_s in the western U.S. appears to confirm that a compositional control on ductile flow rheology is required to explain the differences in strength of stable versus deforming lithosphere [ Lowry et al., 2000]. More rigorous Monte Carlo modeling of that relationship, including measurement uncertainties in T_e and Q_s as well as uncertainties in parameters used to map Q_s to the geotherm and T_e to stress, establish that a compositional control is required at greater than 95% confidence. The effect may derive either from crustal thermal conductivity or compositional parameters of power law creep (e.g., mineral composition or water content).

Geodetic studies of western U.S. postseismic rebound and lake loading rebound, which model the Earth as an elastic layer over a viscoelastic halfspace, typically yield elastic layer thicknesses approximately equal to the crustal thickness (30–40 km) and low (<5X10¹⁹ Pa s) uppermost mantle viscosity. Coherence analysis of gravity and topography at the same locations give T_e a factor of 4 to 6 smaller, however. Part, but not all, of the T_e difference can be explained by the difference in timescale of loading (10⁰–10⁴ years for rebound versus 10⁶–10⁸ years for gravity and topography). The rest of the discrepancy suggests a problem with one or both modeling techniques. I am currently examining the sensitivities and assumptions of the spectral isostatic analysis method of Lowry and Smith [1994] using synthetic tests, in collaboration with Marta Pérez-Gussinyé at the University of Barcelona and Tony Watts at Oxford University, UK [e.g., Pérez-Gussinyé et al., 2004], and Nicola Creati at the OGS, Trieste, Italy. The collaboration with Nicola has resulted in a new and improved methodology to solve for surface and internal loading (and hence T_e) using the load limit projection method described in Lowry and Zhong [2003].

I also intend to examine the effects of more realistic (i.e., temperature dependent and compositionally layered) viscosity structure and pre-existing state of stress on postseismic rebound models, and assess whether the combination of long-term isostatic models and geodetic data can be used to separate effects of viscoelastic, poroelastic and rate-state-dependent frictional response in postseismic rebound models. Our most recent results from the Andaman Islands indicate that most of the near-field postseismic deformation in the first two years following the 2004 Great Sumatra/Andaman earthquake results from fault slip [Paul et al., 2007], suggesting that postseismic rebound will be far more useful for constraining fault frictional rheology than for deeper flow rheology.