Description Of A Framework And Associated Sensitivity Analysis For Recovery Of Rheological Models And Their Key Parameters Using Multi-Cycle Fault Slip Models

Constraining the effective rheology of major faults is crucial to improve our understanding of the physics of plate boundary deformation. Laboratory studies have used analog experiments to propose rheological models based on viscoelasticity or friction that match laboratory-observed behavior under stress-controlled conditions. Such models have since been used to fit real-world observations of deformation near plate interfaces (both for co- and postseismic displacement timeseries), yielding a variety of estimates of key rheological parameters.
However, confidently differentiating between models using purely observations of a single earthquake (coseismic and postseismic deformation) is difficult — especially in the presence of coarse spatiotemporal sampling, inherent observational noise, and the simplifications of our forward models. In this study, we present a framework built on numerical probabilistic simulations aimed at using displacement timeseries across multiple earthquake cycles in a subduction zone, which successfully distinguishes between endmember constitutive models and recovers key rheological properties. Using synthetic Global Navigation Satellite System network datasets, we furthermore investigate the sensitivity of (hyper-)parameters to the recovery of the true underlying rheological models, and present progress made towards using real 3D observations of a megathrust.