LESSONS LEARNED FROM CO- AND POST-SEISMIC SEAFLOOR GEODETIC OBSERVATIONS OF THE M 9 TOHOKU-OKI EARTHQUAKE

Some of the breakthrough observations that George Plafker made after the 1960 Chile and 1964 Alaska earthquakes were from small islands offshore. He would love to extend his surveys farther offshore but was constrained by technological reality (telepathic personal communication, 1964-2014). Things have fundamentally changed over the past 50 years. Seafloor GPS observation of the 2011 great Tohoku-oki earthquake has changed our view of how the system works. Seafloor sites, operated by Japan Coast Guard and Tohoku University, underwent seaward coseismic displacements up to 31 m. These data, among other observations, led to the conclusion that coseismic slip exceeded 50 m and breached the trench. After the earthquake, the terrestrial GPS network continues to show wholesale seaward motion. However, seafloor sites nearest to the trench immediately reversed their direction to move landward. These data demonstrate that seaward-landward opposing motion, previously recognized only for decadal postseismic deformation (Wang et al., 2012), begins right after the earthquake. The landward motion (~ 30 – 50 cm in the first year) is much faster than the subduction rate (8.3 cm/year) and cannot be explained by the relocking of the fault. Neither can it be explained by afterslip which would cause the surface to move in the opposite (seaward) direction. Using 3D finite element modeling, we demonstrate that the landward motion is due to viscoelastic stress relaxation in response to the asymmetric rupture of the thrust earthquake. Because of a stiffness contrast, the upper wall exhibits much larger coseismic motion than the foot wall in any shallow and large subduction earthquake. Much greater tensile stress is thus induced in the upper plate landward of the rupture than in the incoming plate. As the mantle undergoes viscoelastic relaxation, this asymmetric stress drives the trench area to move landward. Our results indicate that most previously published afterslip models for subduction earthquakes overestimated deep afterslip and underestimated shallow afterslip (where present) because of the neglect of viscoelasticity. Our knowledge of the slip behavior of subduction megathrust based on these estimates needs to be revised in the context of a viscoelastic Earth.

Wang, K., Y. Hu, & J. He. Nature 484, 327-332 (2012).