Postseismic Afterslip, Viscoelastic Mantle Flow, and the Earthquake Cycle at the Nankai Trough Subduction Zone

摘要

Historical leveling data and recent GPS measurements in the Shikoku region of southwestern Japan record 120 years of vertical crustal motions spanning the 1944 and 1946 M8+ earthquakes. We develop 2D boundary element models of the earthquake cycle at the Nankai Trough that incorporate a subducting elastic slab with a Maxwell viscoelastic mantle wedge and subducting mantle lithosphere. We impose repeated earthquakes on the shallow subduction interface and model rate-strengthening creep at greater depths. The model parameters are systematically modified using known coseismic rupture zones and thermal models as geometric constraints to numerically fit vertical deformation profiles to the vertical data in the co-, inter-, and post-seismic periods. Previous models neglecting mantle flow (elastic half-space models with rate-strengthening creep) or neglecting the subducting elastic slab (elastic plate models) fit vertical displacement data near the coastline reasonably well for most time periods throughout the earthquake cycle but are unable to account for the observed inland subsidence during the postseismic and mid-interseismic periods. We show that models including both rate-strengthening creep, viscoelastic mantle flow with viscosities of order 1019 Pa s (mantle relaxation times of 10--25 years) capture the observed vertical deformation profile spanning the entire earthquake cycle in central Shikoku and along the west coast of the Kii peninsula. These results indicate that both afterslip and mantle flow contribute significantly to the postseismic and mid inter-seismic periods. However, recognition that our mantle flow model requires a higher viscosity to fit data later in the earthquake cycle suggests the simplified single Maxwell viscosity is not sufficient to explain deformation at all time periods. 3D models with multiple mantle viscosities or nonlinear, stress-dependent viscosity are a likely needed to explain data for all time periods.