On the cause of enhanced landward motion in adjacent segments of subduction zones after megathrust earthquakes

Abstract

<p>Secular GNSS velocities on the overriding plate have been observed to change after several megathrust earthquakes (2003 Tokachi-oki, 2007 Bengkulu, 2010 Maule, 2011 Tohoku-oki, 2012 Oaxaca, 2014 Iquique). In particular, landward velocities are larger than before the earthquake in regions laterally adjacent to the ruptured zone, as close as 300–500 km along-trench from the centroid. Previous explanations of this enhanced landward motion invoked either increased slip deficit accumulation or overriding plate bending during postseismic relaxation; the mechanism is relevant for seismic hazard. We investigate whether plate bending alone – driven by afterslip and/or viscous relaxation – is likely to occur and can reproduce the observed patterns of velocity changes. We use 3D quasi-dynamic finite element models with periodic earthquakes.</p><p>We find that rapid afterslip in the shear zone downdip of the brittle megathrust, at depths greater than 40 km, produces only enhanced trenchward surface motion in the overriding plate. Bulk viscous relaxation in the mantle wedge does produce enhanced landward motion, although only when afterslip in the deep shear zone is spatially restricted by imposing a constant shear rate beyond a certain depth (75–150 km in our models). This landward motion indeed is a consequence of in-plane bending and shortening of the overriding plate. However, this modeled landward velocity change is one order of magnitude lower than observed, decays , and , resulting in no net enhanced landward motion. These findings are robust with respect to variations in mantle viscosity, plate elasticity, maximum afterslip depth and earthquake size. We thus conclude that plate bending probably occurs, but cannot explain the observed lateral landward velocity changes. The observed enhanced landward motion in the laterally adjacent regions of the subduction zone consequently reflect true local slip deficit accumulation at a faster rate than before the earthquake, whether because of transient slab acceleration or because of frictional changes on the megathrust. The implication is a greater seismic hazard in those lateral regions.</p>