Segregation of Source Areas of Slow Slip Events and Asperities of Major Seismic Events on the Subduction Interface Around the Japanese Islands

Following the 1989 Sanriku (Mw7.4) and the 1992 Sanriku (Mw6.9) earthquakes on the subducting interface of the Pacific plate, anomalous tsunamis were excited (Hatori, 1993) . The source areas of the anomalous tsunamis do not overlap with coseismic asperities of nearby large earthquakes, as inferred by Yamanaka and Kikuchi (2002). These observations show that, after major seismic faulting of the 1989 Sanriku and the 1992 Sanriku earthquakes, adjoining area slipped at rates that were one-order slower than coseismic rates and caused significant tsunamis. Following the tsunamis, afterslips with time constants of around 10 days and 1 day, respectively, were recognized by geodetic data. Also, static moments from the geodetic data were a few times larger than values calculated from seismic data (Kawasaki et al., 1995). Due to insufficient resolution of source parameters of the afterslips, it is unknown if these source areas overlap with the source areas of the anomalous tsunamis. We have documented several silent earthquakes that occurred in japan over the last several years, including the 1989 Tokyo Bay (equivalent magnitude of Mw5.9, around 1 day), the 1996 Off-Boso (Mw6, around 5 days), the 1997 Bungo Channel (Mw6.8, around 1 year), the 1999 Off-Choshi (Mw5.6, a few days), the 2000 Off-Choshi (Mw5.6, a few days) and the 2001-2002 Tokai (Mw6.8, 1-2 years, ongoing) events. The silent earthquakes around the south Kanto district occurred along the transition zone at depths of 30-40 km between the coupled and uncoupled regions along the subduction interface of the Philippine Sea plate. Their source areas do not appear to overlap with the major asperity of the 1923 great Kanto earthquake (Wald and Somerville, 1996) nor with the source area of the 1703 historical Kanto earthquake (estimated from inversion of tsunami data, Murakami and Tsuji, 2002), which were the dominant events in the past several centuries in the Kanto district. Summarizing these observations, (1) slow slip events of diverse time constants from hours, days, months to years have occurred, (2) the total amount of interplate moments released by the slow slip events is too small to fill gaps in the moment budget, and (3) faulting areas of the silent slip events do not overlap with the asperities of the coseismic slip of large earthquakes, suggesting that the controlling factor of the slip modes is a frictional property.