The Impact of Viscoelastic Earthquake Cycles and Elastic Heterogeneity on Interseismic Coupling

Some of the most commonly used tools for estimating the size and spatial distribution of future megathrust earthquakes are interseismic coupling models. These models estimate the degree to which a fault is interseismically coupled (i.e., not slipping at the full plate convergence rate between earthquakes) and thus accumulating strain to be released in future earthquakes. The accuracy of interseismic coupling models depends heavily on the coverage of the interseismic surface deformation data utilized and on the quality of the Earth model used to relate slip on the fault to surface deformation. For the latter, most interseismic coupling inversions assume a homogeneous elastic half-space model, despite the fact that it has been understood for decades that viscous mantle relaxation contributes significantly to surface deformation in the years and decades following subduction megathrust earthquakes.

We present here updates to previous homogeneous elastic probabilistic interseismic coupling models for Cascadia and Nankai subduction zones. Using a Green’s function approach and the spectral element code, visco3d [Pollitz, 2025], we integrate elastic heterogeneity and viscoelastic earthquake cycles into our boundary inversion framework. The earthquake cycle model consists of imposed period earthquakes on a steady backslip history and the heterogeneous elastic models are based on regional seismic velocity models. Additionally, because postseismic studies suggest that vertical velocities are more sensitive to the contribution of viscous mantle flow than horizontal velocities, we incorporate vertical surface deformation data for both subduction zones.

We find that the spatial coverage and quality of the surface deformation data is the most critical factor in constraining interseismic coupling, as evidenced by the limited change in coupling distribution at the highly instrumented Nankai subduction zone. We also find that viscosity has a first-order effect on modeled surface velocities and coupling estimates, while geometric variations in plate thicknesses, cold wedge depth, and cold wedge angle are secondary. When compared to a homogeneous elastic model, incorporating elastic heterogeneity shifts interseismic coupling landward and reduces overall moment accumulation rates, while incorporating a simple viscoelastic earthquake cycle model has the opposite effect at both subduction zones. Therefore, combining both elastic heterogeneity and viscoelastic earthquake cycles results in coupling and moment accumulation rate estimates not too dissimilar to homogeneous models. Further testing is required to determine if these results hold with higher resolution elastic heterogeneity models and more complex viscosity models.