Recently, a widespread and densely continuous-recording ocean-bottom seismograph network has been deployed in the Japan Trench subduction zone. Utilizing the offshore network data improves azimuthal station coverage for offshore earthquakes in the Japan Trench subduction zone. It has a potential to obtain centroid moment tensor (CMT) solutions more accurately than conventional analyses using onshore networks and a simple one-dimensional seismic velocity structure model. In this study, we conducted CMT inversion for subduction zone earthquakes that occurred between April 1, 2017, and March 31, 2024, with a moment magnitude range of 5.2–7.0. We used seismograms obtained from both the offshore and onshore networks. We calculated Green’s functions using a three-dimensional seismic velocity structure model. Our CMT solutions with thrust-type mechanisms mostly indicated depths and dip angles consistent with the plate interface. For earthquakes in the outer-rise region, our CMT solutions were characterized as normal-fault mechanisms. The joint use of the offshore and onshore networks reduced the estimation errors of the CMT solutions compared with the only use of the onshore network, although the optimal solutions were consistent. The dip angles for the thrust earthquakes determined by our analysis were more consistent with the dip angle of the plate boundary than those determined by conventional CMT analyses. Additionally, we found that the conventional CMT analysis could introduce a systematic bias in depth and magnitude determinations. This finding highlights the importance of an offshore seismograph network and a reliable seismic velocity structure model for CMT inversions.

Tsunamis with maximum amplitudes of up to 40 cm, related to the Mw 7.1 normal-faulting earthquake off Fukushima, Japan, on November 21, 2016 (UTC), were clearly recorded by a new offshore wide and dense ocean bottom pressure gauge network, S-net, with high azimuthal coverage located closer to the focal area. We processed the S-net data and found that some stations included the tsunami-irrelevant drift and step signals. We then analyzed the S-net data to infer the tsunami source distribution. A subsidence region with a narrow spatial extent (~40 km) and a large peak (~200 cm) was obtained. The other near-coastal waveforms not used for the inversion analysis were also reproduced very well. Our fault model suggests that the stress drop of this earthquake is ~10 MPa, whereas the shear stress increase along the fault caused by the 2011 Tohoku earthquake was only ~2 MPa. Past studies have suggested that horizontal compressional stress around this region switched to horizontal extensional stress after the Tohoku earthquake due to the stress change.The present result, however, suggests that the horizontal extensional stress was locally predominant at the shallowest surface around this region even before the 2011 Tohoku earthquake. The present study demonstrates that the S-net high-azimuthal-coverage pressure data provides a significant constraint on the fault modeling, which enables us to discuss the stress regime within the overriding plate around the offshore region. Our analysis provides an implication for the crustal stress state, which is important for understanding the generation mechanisms of the intraplate earthquake.

The development of seafloor seismic observations facilitates reliable estimation of the rupture directivities of offshore earthquakes. We used seismic waveforms obtained by a new seafloor seismic network (S-net) and onland stations to systematically examine the rupture directivities of interplate earthquakes along the Japan trench. We estimated the rupture directions of 206 (M w 3.5-5) events, most of which occurred near the base of the seismogenic zone. We found that most earthquake ruptures (>~80 %) were directional, primarily propagating in the updip direction. This tendency cannot be explained by the effect of the bimaterial interface. The prevalence of updip rupture in the data suggests that deep, steady creep and upward fluid migration along the plate interface affected earthquake ruptures in the subduction zone. The updip ruptures redistributed the accumulated shear stress from the base of the seismogenic zone to the shallow large seismic patches. Furthermore, the updip ruptures may open up ways for deeper fluids to migrate further upward along the plate interface. Both the stress redistribution and the upward fluid migration facilitate the occurrence of shallow megathrust earthquakes.