Characterizing decadal transient deformation and aseismic fault slip in the northeastern Japan subduction zone prior to the 2011 M 9 Tohoku-oki earthquake

Postseismic deformation and slow-slip events are well documented by modern geodesy, yet transient deformation not related to postseismic effects spanning decades has not yet been observed. From analysis of GPS data from Japan's GEONET, we previously reported spatially coherent patterns of transient deformation in various regions of Japan, spanning the 15 years prior to the 2011 M 9 Tohoku earthquake. While several of these transients are clearly postseismic effects, an apparent decrease in the rate of deformation in southeastern Tohoku is surprising, since it implies a decadal decrease in the degree of coupling on the plate interface prior to the 2011 M 9 Tohoku earthquake. Given that this inference, if correct, has profound implications for our understanding of the earthquake cycle, it is important to investigate how assumptions in removing postseismic deformation due to the many M 6 and 7 earthquakes during this time period may bias the estimation of residual transient deformation. The ultimate question is whether there is robust evidence for accelerating aseismic slip on the plate interface and, if so, how its spatial distribution compares with known locked ';asperities', including the rupture area of the M 9 Tohoku earthquake. To correct for postseismic slip of M > 6.3 earthquakes, we construct a model of the Japan trench plate interface, which consists of circular asperities that represent rupture areas of interplate earthquakes that occurred after 2003. Coseismic slip distributions are computed assuming uniform stress drops. Afterslip is assumed to be space-time separable, with its temporal evolution represented by the solution for a steady-state, rate-strengthening spring-slider including steady load-point velocity. The spatial distribution of cumulative afterslip is computed such that it fully relaxes coseismic stress increase anywhere on the fault except for the asperities, which remain locked. Surface displacements are then computed using elastic half-space Green's functions. Preliminary results reveal that the patterns of coseismic and time-dependent postseismic surface displacements predicted from this model are in good agreement with GPS observations. Notably, the data require afterslip from two earthquakes before 2011 overlapping the coseismic slip area of the M 9 Tohoku earthquake. Removing postseismic effects as accurately as possible, including the effects of intraplate earthquakes, will provide a robust estimate of any residual transient deformation, and hence inference of as yet unexplained transient slip on the plate interface, in the years leading up to the Tohoku earthquake.