Spatiotemporal model for crustal deformation around the focal area of the 2008 Iwate-Miyagi Inland Earthquake, northeastern Japan, estimated by GPS and InSAR

Applying the program tDEFNODE [McCaffrey, GRL09] to model elastic lithospheric block rotations and strains, and locking or coseismic slip on block-bounding faults, we model GPS and InSAR data of crustal deformation before and after the 2008 M7.2 Iwate-Miyagi Inland Earthquake (IMEQ). The epicenter of the IMEQ is located in a high strain rate zone along the Ou Backbone Range (OBR) in northeastern Japan, where volcanic front runs subparallel to the Japan Trench. Along eastern and western margin of the OBR, major inland active faults have been growing. In order to clarify detailed strain field around these active faults, Japan Nuclear Energy Safety Organization (JNES) installed 7 new continuous GPS sites with ~5 km spacing across the Dedana Fault (DF), which is a part of the eastern marginal active faults, in October 2007. Because the DF is located at only ~20 km northeast from the hypocenter of the IMEQ, the detailed coseismic and postseismic crustal deformation was obtained at these GPS sites [Ohta et al., EPS08; Iinuma et al., GRL09]. Takada et al. [EPS09] also estimated coseismic faults from InSAR with pixel-offset method. These studies, however, handled each data independently. We unify both GPS and InSAR data to model crustal deformation over inter-, co-, and post-seismic period using tDEFNODE. This code interprets geodetic timeseries data by assuming elastic block rotation, transient phenomena such as slow slip, and coseismic slip, on the block boundary faults. We use timeseries data obtained by continuous GPS networks conducted by JNES, Tohoku University, Geographical Survey Institute (GSI), National Astronomical Observatory (NAO), and IGS from 2006 to June 2009, together with a temporal network for postseismic deformation installed by Japanese University Consortium for GPS Research (JUNCO). InSAR data is obtained by analyzing images of ALOS/PALSAR mission. We presumed three blocks in the study area divided by two boundary faults. We expressed the time evolution of slip rates on sub-faults as the superposition of triangle functions with 5 days-width shifted by 2.5 days. As a preliminary result, we found a set of model parameters to reproduce the timeseries data successfully. The large coseismic slip (~3m) is estimated beneath a GPS site ICNS, where the maximum coseismic offset was observed. For the postseismic deformation, it is suggested that the aseismic fault slip occurred not only on the earthquake fault, but also on the shallow part of the DF. After two months from the mainshock, the slip almost disappeared. It should be noted that the successive modeling with various new geodetic data enables to monitor the process of strain accumulation and release in this area.