[S12-P] Poster

[S12-P-06] Abnormal seismicity of slow earthquakes on land prior to 2011 Tohoku earthquake

^○Tomoki Tokuda, Hirohiko Shimada (Okinawa Institute of Science and Technology Graduate University, Okinawa, Japan)

Slow earthquake (EQ) has recently gained much attention as a possible precursor of large earthquake. A number of studies so far suggest that slow EQ (or slow slip) may serve as an indicator of stress accumulation/transfer, thus quite useful for predicting a large earthquake. Indeed, it was observed that slow EQs migrated toward the epicenter of 2011 Tohoku earthquake, which suggests a possible nucleation process prior to the occurrence of the megathrust earthquake. As regards Tohoku EQ, the current focus of research tends to be on off-shore areas near the epicenter, exploring relationships among slow tremors, slow slip events and the megathrust earthquake. On the other hand, slow EQs that occurred in remote areas (on land) far from the epicenter receive less attention. Nonetheless, geodetic data suggest that the accumulation of on-land stress in Tohoku region is associated to that of the off-shore one. So, we cannot rule out a possibility of abnormal seismicity in on-land slow EQs prior to Tohoku EQ. In this talk, we discuss such a possibility, examining the relationship between slow EQs in remote areas and Tohoku EQ in a data-driven approach. First, we analyze on-land slow EQ that occurred in northern Japan in the last 20 years (based on the catalogue of Japan Meteorological Agency) by means of cluster analysis. In this analysis, three subtypes of slow EQs are identified, which are statistically characterized by short, mid and long-term intertime between slow EQs. Second, it is found that the short and mid-term slow EQs became quiescent two months and one month before the Tohoku EQ, respectively. Such a phenomenon is not observed in generic earthquakes, and would be identified with a unique characteristic of megathrust earthquakes. Finally, we apply OFC (Olami, Feder, Christensen) model to the slow EQ data in question, giving a theoretical interpretation to the aforementioned phenomenon.