IMPORTANCE OF VISCOELASTIC STRESS RELAXATION IN BOTH POSTSEISMIC AND INTERSEISMIC DEFORMATION (Invited Presentation)

It is beyond any doubt that crustal deformation of timescales of hours to millennia is strongly controlled by the viscoelastic behaviour of Earth’s mantle. The question is how long the viscoelastic effect lasts in different processes. In Glacial Isostatic Adjustment, the effect is known to last for thousands of years. In earthquake cycle deformation, it is often assumed that the effect is important only for postseismic deformation and that interseismic deformation can be explained using an elastic Earth model. Here we emphasize that viscoelastic stress relaxation is important in both postseismic and interseismic deformation.

Immediately after a great subduction earthquake, viscoelastic relaxation of earthquake-induced stresses results in opposing motion, with the dividing boundary of seaward and landward motion located roughly above the downdip end of the rupture zone. As the viscoelastic effect diminishes with time, the locking effect becomes more prominent. The dividing boundary thus migrates landward, eventually leading to wholesale landward motion. The larger the earthquake, the longer it takes to complete the motion reversal. By finite-element forward modeling the postseismic deformation processes of 10 great earthquakes constrained by geodetic observations, we find an approximately linear relation between the time of motion reversal and earthquake size.

The wholesale landward motion in late-stage interseismic deformation, such as observed at Cascadia today, by no means suggests that the viscoelastic effect can be ignored. Stresses induced by megathrust locking are being relaxed at the same time, even when crustal deformation is no longer changing with time. Consequently, interseismic strain occurs quite far inland. If the broad interseismic deformation were to be explained using an elastic Earth model, megathrust locking would have to be assumed to occur at large depths, contradicting what we know about fault rheology. We have derived new models of Cascadia megathrust locking by inverting GPS data using a finite element viscoelastic model. The model results also illustrate that land-based GPS has no information on the locking state of the shallowest part of the megathrust and cannot uniquely resolve along strike variations of locking and creep, accentuating the need for seafloor monitoring.