STRESS EVOLUTION DURING THE MEGATHRUST EARTHQUAKE CYCLE AND ITS ROLE IN TRIGGERING EXTENSIONAL DEFORMATION IN SUBDUCTION ZONES (Invited Presentation)

Great (moment magnitude Mw ∼8.0+) subduction megathrust earthquakes are commonly followed by increased rates of normal faulting seismicity. We focus here on extensional aftershocks in the forearc, which are triggered primarily by the largest magnitude (Mw 8.5+) events. The total offsets on normal faults at the surface are too large to have occurred in a single event, suggesting that permanent extension accumulates over multiple earthquake cycles. To better understand these observations, we develop a 3D earthquake cycle model with a curved slab geometry (comparable to most subduction zones globally) and stresses that are in balance with plate interface slip and bulk viscous relaxation. The inter-seismic stress field is dominated by compression resulting from convergence across a locked seismogenic zone. Co-seismic slip during the earthquake causes approximately trench-normal extension relative to the pre-earthquake stress field. Post-seismic processes do not appear to play a major role in further promoting extensional deformation. The pre-earthquake state of broadly compressive stress is slowly reimposed by the re-locking of the plate interface and continued plate convergence.

Our models show that the magnitude of compressive inter-seismic stresses in the upper plate (defined by the maximum shear stress) depends strongly on the plate interface locking configuration, from 25 MPa for a fully locked seismogenic zone to 5 MPa for discrete locked patches on an otherwise unlocked plate interface. However, in both locking configurations, the maximum co-seismic stress changes are up to 5 MPa. This implies that the inter-seismic compression can be widely reset to zero by the earthquake if (a) the co-seismic slip magnitude is sufficiently large and (b) the megathrust contains regions that are persistently weak. In our models, the earthquake cycle stresses never become significantly tensile, suggesting that co-seismic extension is the trigger for, but not the primary driver of, extensional seismicity. The heterogeneous orientations of T-axes from observed normal faulting aftershocks in the upper plate are compatible with this interpretation, and suggest that spatially variable, local stress conditions are important in defining the kinematics of extensional deformation after a great earthquake.