Towards a 3D Earthquake Cycle Model Powered by Data Assimilation for Northeastern Honshu

In this study, we seek to quantify bulk viscoelastic flow, afterslip and locking, within a rheological framework that ensures a consistent formulation of strain accumulation and release throughout the entire earthquake cycle. To achieve this, we use Bayesian inference in the form of an ensemble smoother with multiple data assimilation (ESMDA) to estimate geodynamic model parameters. In our earlier study, we successfully reproduced both interseismic and postseismic observations for the Tohoku margin including the 2011 earthquake using a 2D model (Marsman et al. 2025). Building on these insights, we extend our analysis to a 3D configuration.

We construct a 3D finite element seismic cycle model. We incorporate a priori information into the model, including a realistic geometry of slab and overriding plate, the temperature field, multiple asperities, and the observed coseismic slip distribution of the 2011 Tohoku-Oki earthquake. The model has a steady-state power-law rheology. Away from asperities, different parts of the megathrust respond by power-law viscoelastic relaxation, simulated by a thin low-viscosity shear zone, or instantaneous slip. By assimilating observations of 3D surface deformation, we constrain power-law flow parameters for both the asthenosphere and the megathrust. Specifically, we estimate the pre-exponent factor and the activation energy of the mantle wedge and oceanic mantle, as well as the pre-exponent factor and stress power of the shear zone using ESMDA.

We assimilate 3D GNSS displacement time series spanning from 1997 onwards. Preliminary results with actual GNSS data indicate that power-law flow parameters can be retrieved remarkably well and are consistent with estimates from laboratory experiments. The trade-off between the pre-exponent factor and activation energy hinders their individual estimation but does result in a well-constrained viscosity structure. Consistent with our 2D models, our 3D results demonstrate that enhanced landward motion near the rupture zone occurs postseismically without the need for a separate low-viscosity sub-slab layer. Instead, the release of elastic stresses accumulated interseismically beneath the oceanic plate significantly contributes to the observed offshore postseismic landward motion near the trench on the overriding plate.