Methods for simulating earthquake sequences and rupture dynamics in diffuse fault zones

Abstract:

Earthquakes are typically modelled as a displacement discontinuity across a prescribed fault surface embedded in linear viscoelastic media adhering to several simplifying assumptions: (i) faults are infinitesimally thin, although volumetric failure patterns are observed in well-recorded small and large earthquakes and laboratory experiments; (ii) earthquake models artificially distinguish between on-fault frictional failure and off-fault (in-)elastic rock deformation; (iii) initial and fault interface conditions, such as loading stress and frictional strength, drive faulting dynamics but are themselves poorly constrained.

The TEAR project (https://www.tear-erc.eu) aims to develop holistic earthquake modelling methods using high performance computing (HPC). The overarching goal is to understand the mechanics of slip in deforming fault zones from highly complex, short-term rupture dynamics to long-term seismic sequences.

To this end, we first summarize two novel diffuse faulting approaches for nonlinear dynamic rupture processes in natural fault damage zones: a spectral finite element (continuous Galerkin) method with a non-mesh aligned embedded diffusive discontinuity and a unified first order hyperbolic model that incorporates finite strain elasto-viscoplasticity and viscous fluids (Gabriel et al., 2020).

To model quasi-dynamic long-term earthquake cycles, we explore the introduction of diffusive regulating terms into a bulk reformulation of rate- and state-dependent friction (Herrendörfer et al., 2018) in a novel implementation of staggered-grid finite differences. Lastly, we apply a Discontinuous Galerkin method to seismic sequence modeling. Key features are unstructured simplicial meshes, curvilinear geometries, and sub-cell material resolution, which allows the inclusion of realistic fault geometries and heterogeneous velocity models.

We verify and compare all methods in SCEC community benchmarks (Harris et al., 2018; Erickson et al., 2020).