Exploring new refinements to estimation of Te and mass loading

Abstract:

Effective elastic thickness (T_e) is a measure of integrated lithospheric strength that depends on rheological parameters such as lithology, temperature, water fugacity, and state of stress. It is a useful constraint for resolving in-situ rheology and stress within the lithosphere, and therefore critical to understanding surface deformation and controls on tectonic processes.

T_e is commonly estimated via comparison of observed and model-predicted spectral-domain coherence between Bouguer gravity and topography signals. The T_e producing a model coherence that best fits observed coherence is taken to best represent lithospheric bending strength. However, the solution relies on accuracy of the method’s assumption that loads at the surface and in Earth’s interior are truly uncorrelated; T_e estimates may be biased in regions where this criterion fails. Determining mechanical anisotropy is also ambiguous, as true anisotropy cannot be reliably distinguished from artefacts in gravity and topography data inversion.

We propose a new modeling approach that will refine existing T_e estimation via two innovations. First, we reduce the null-space in estimates of surface- versus subsurface contributions to loading by assimilating independent estimates of subsurface mass variations derived from EarthScope USArray seismic imaging data. This reduces the ambiguity of loading estimates, and provides an independent means of testing for anisotropy of the applied loads. Second, we substitute a finite-element approach to forward modeling spatially-varying flexural response in place of conventional uniform-T_e linear thin plate models. This modification allows T_e to vary and and loads to update iteratively, and improves separation of intrinsic T_e-anisotropy from effects introduced by lateral variations in isotropic T_e. A variable-T_e forward modeling approach also improves spatial resolution of the inversion as it is no longer limited by a need to subsample data through windowing or wavelets.

<< Previous Abstract | Next Abstract >>