Distribution of seismic coupling on the subducting plate boundary in northeastern Japan inferred from GPS observations

Many Global Positioning System (GPS) stations of the Geographical Survey Institute and Tohoku University in northeastern Japan observed clear coseismic and postseismic deformations associated with the M7.5 1994 Far Off Sanriku earthquake. Applying the geodetic inversion algorithm given by Yabuki and Matsu'ura (Geophys. J. Int., 1992, 109, 363–376) to the observed GPS data, we estimate in this paper the spatial distributions of the coseismic and postseismic slip on the boundary between the subducting Pacific plate and the overriding continental plate. It is noted that the coseismic slip estimated in this paper denotes the slip for a period between the day before and the day after the main shock. The inversion for coseismic slip shows that the main shock was thrust faulting typical for interplate earthquakes, which is consistent with the results of other seismological studies. The short-term postseismic deformation (STD) observed for a period of ca. 10 days between the main shock and its largest aftershock shows that the initially high deformation-rate was lowered rather rapidly in comparison with the subsequent long-term deformations (LTD) lasting longer than 1 year, which decreased at an almost constant low rate. The observed data of STD can be explained by postseismic slip distributed approximately on the same region as the faulting area of the main shock, where the moment release during this period is ca. 16% of that estimated for the coseismic slip. The region of the sliding causing LTD is found to have significantly expanded southward and downward (or westward) on the plate boundary. The moment released by the postseismic sliding for nearly 1 year amounts approximately to 80% of that released by the coseismic slip in addition to the contribution from uniform steady sliding due to the relative plate motion. Furthermore, we find from the inversion analysis a locked region to the south of the long-term postseismic sliding region on the plate boundary. The amount of nominal back-slip estimated is so large that the locked region of the plate boundary is considered to have been tightly coupled for the year of 1995. This study demonstrates that the GPS observation only for ca. 1 year can distinguish the decoupled region due to the postseismic sliding around the source area of the main shock from the coupled one of the plate boundary.