Long-term slip events in the Kanto district revealed from GNSS data during 1996—2011 and a relationship with variations in ocean bottom pressure

Recently, Mavrommatis et al. 2014 (MA2014 hereafter) reported that a long-term accelerative transient crustal deformation was occurring in eastern Japan before the 2011 M-9 earthquake. They revealed that the observed deformation could be explained by a fault slip on the plate boundary with depths of approximately 20—60 km where afterslip of the M-9 event was detected. Its characteristic time is at least 15 years, which is much longer than those of precursory deformations usually considered in numerical simulation and observation studies. The acceleration was more dominant below the Tohoku area, where the coseismic slip of the M-9 event was larger. In this study, GNSS data in the Kanto area were analyzed in detail and it was found that the acceleration was not necessarily uniform if seen at smaller spatial scales than in MA2014. The detrended GNSS data with co- and post-seismic deformations due to large earthquakes and published SSEs being removed showed i) a slower acceleration in north of Tokyo Bay during 1996—2000 (period A), ii) a deceleration over the same area during 2002—2006 (period B) and iii) a faster acceleration over northeast Kanto during 2007—2011 (period C). A result of slip inversion indicated that i) a fault slip on the plate boundary with the Philippine Sea Plate was more likely to cause the events during periods A and B because the observed displacement rate was parallel to the relative motion by the PHS to the continental plate. Similarly, a slip on the boundary with the Pacific Plate was more likely to generate the event during period C. If the slip during period A is ignored, our result is consistent with the result of MA2014 because amplitude and spatial scale of the slip during period C is larger. Temporal variations of these slips were well correlated with slip history predicted by a non-linear rock frictional law which considers combined effects of tides and non-tidal ocean bottom pressure (Tanaka et al. 2015). The same model can explain the acceleration in Tohoku qualitatively. A quantitative explanation is possible if it is allowed to assume that "micro SSEs" that cannot be detected with current geodetic observations occur when delta CFS is higher than a critical limit. Our result may indicate that a different-type SSEs can occur beneath eastern Japan, where episodic deep SSEs are not seen as in western Japan.