Four surface ruptures associated with the 2016 Kumamoto earthquake that extend across the alluvial plain of Kiyama River in Mashiki Town, Japan, are investigated via a seismic reflection survey to elucidate their fault subsurface geometries and their interrelationship with each other. Three of these surface ruptures are subparallel, ENE-WSW-striking faults: a right-lateral fault along the southern margin of the plain (fault Fb), a normal fault on the flank of the southern mountainous area (fault Fa), and a right-lateral fault along the northern part of the plain (fault Fc). The faults Fa and Fb correspond to the previously mapped Futagawa fault that are identified by fault landforms. However, the fault Fc is not topographically defined. The fourth surface rupture (fault Fx) is highly oblique to the ENE-WSW-striking faults and cuts across the alluvial plain left laterally, extending from fault Fb to fault Fc. The processed seismic reflection profiles image a basin structure beneath the alluvial plain that is bounded by the faults Fb and Fc. The near-vertical right-lateral fault Fb and a north-dipping normal fault Fa converge at 350 m depth. The fault Fc may be the Kiyama fault due to its location and south-dipping structure. A comparison of the coseismic displacement during the 2016 earthquake and the subsurface geometry of the fault traces highlights that the oblique slip on the north-dipping source fault of the Kumamoto earthquake may be partitioned into the three subparallel faults at the surface. Therefore, it is important to consider the presence of subparallel slip-partitioned faults in a given area and comprehensively evaluate the amount of slip along each fault when conducting seismic hazard assessments via active fault investigations.

There are large discrepancies between the epicenters of the 1933 Showa Sanriku earthquake (M_W 8.4) reported by the Japan Meteorological Agency (JMA) and those reported by other studies. The JMA hypocenter (39.13˚N, 145.12˚E, with depth fixed at 0 km) is more than 30 km east of the other reported epicenters. The JMA location is based on local P and S wave arrival data, whereas the others are based on global arrival data. I examined the cause of the discrepancies and found that the earthquake has a multiple process and consisted of a preliminary rupture and a main rupture that occurred about 4-6 s apart; the JMA location represents the epicenter of the preliminary rupture, whereas the other studies used both preliminary rupture and main rupture arrivals to locate epicenters that made the shift of epicenter to the west and nearer the Japan Trench. I carefully selected the arrival data and estimated hypocenter of the main rupture at 39.2˚N, 145.0˚E, with depth fixed at 0 km. This result shows that the mainshock originated in the outer rise area more than 60 km east of the trench axis which is consistent with the fact that the locations of all recent M>7 normal fault earthquakes in the outer rise area are similarly distant from the trench axis in the study area. The 2011 Tohoku earthquake was also a multiple-shock, with the JMA epicenter located by using wave arrivals of the preliminary rupture, whereas the epicenter of the international agencies, ISC and NEIC, which use global arrival data, were located using arrivals both the preliminary and the main rupture, causing a significant shift of the epicenter to the west. If I also refer to the results of other studies that take into account 3D structures and other factors, the true epicenter is still presumed to be in the vicinity of the JMA epicenter.

Although it is now clear that the main shock of the 1933 Showa Sanriku earthquake occurred in the outer rise area outside of the Japan trench, it is difficult to explain the preceding seismic activity on the northern side of the aftershock region in January 1933, the epicenter of the main shock, the distribution of aftershocks covering both sides of the trench, the estimated tsunami source area extending west of the trench, and the upward tsunami observed on the coast in a unified manner using only normal faulting of the main shock; the aftershock activity suggests the existence of a hidden thrust earthquake along the plate boundary. However, the upward tsunami on the coast cannot be explained by a thrust earthquake, but may be explained by the high angle splay fault that was activated by the thrust earthquake. From the above, we do not know which occurred first, but it is highly likely that the 1933 Showa Sanriku earthquake was a doublet, consisting of a normal fault main shock and a thrust earthquake with slip of splay faults.

In this study, we analyzed the utilization status of observation data of earthquakes, tsunamis, and volcanoes based on data DOI citation information. The target is the nationwide observation network for earthquakes, tsunamis, and volcanoes over land and sea (MOWLAS) in Japan, which has been maintained and operated by National Research Institute for Earth Science and Disaster Resilience (NIED) and consists of seven observation networks. Since February 2019, NIED has progressively assigned data DOIs to its research data, with six of these data DOIs being allocated to the observation networks of MOWLAS. We collected information on the publications citing data DOIs of MOWLAS from the Web of Science database and examined the citation status from 2019 to 2022. The results show that more than 160 publications citing the MOWLAS’s data DOIs have been issued during the survey period (2019-2022), indicating that the MOWLAS’s observations have been used in various research studies. They are most frequently cited in the order of [K-NET, KiK-net], [Hi-net], [F-net], [S-net], [DONET], and [V-net]. The number of times cited of the publications exceeded 600 and has been increasing annually along with the number of publications. We conducted surveys on the number of publications by journal, the affiliations of the first authors, their countries, and multiple citations of data DOIs. We also analyzed the titles of the publications by observation network using quantitative text analysis, clarifying how the MOWLAS data have been used. The observations from MOWLAS are important data in various research fields including seismology and earthquake engineering. We look forward to further use of MOWLAS data, and we kindly ask our readers to cite the data DOIs properly.

We have developed a giant magnetostrictive seismic source (GMSS), which is a small, high-frequency seismic source for constantly monitoring temporal changes in subsurface physical properties with high precision. The GMSS generates a single force by vibrating the weight in the vertical direction synchronized with GPS time, and excites P and SV waves of arbitrary waveforms with very high reproducibility in the subsurface. The GMSS is intended for constant monitoring of approximately 1 km area of underground, and is essentially maintenance-free even during long-term continuous operation spanning several years. In this study, the GMSS was installed in the Mizunami Crustal Movement Observation Vault (Mizunami-City, Gifu Prefecture) and constantly transmitted elastic wave signals in the 100-200 Hz band using 5 kg vibrating weight. In the experimental site, Toki Granite is widely distributed beneath the Mizunami Group, which is composed of Miocene sedimentary rocks. The GMSS was located in the Mizunami Group, and observations were conducted using the seismometers of borehole observation sites, TGR350 (GL-350 m, distance 353 m from the GMSS) and TRS (GL-505 m, distance 690 m from the GMSS), installed in the Toki Granite. The data stacking period was set to 1 day in order to obtain transfer functions and Green’s functions with sufficiently high signal-to-noise ratio at both stations. During the Kumamoto Earthquake (April 16, 2016, Mj7.3), travel time delays of 60 µs (TRS, travel time 0.185 s) were observed for the P wave and 166 µs (TGR350, travel time 0.148 s) for the SV wave. Since the travel time delay of the P wave propagating almost directly beneath the GMSS was only 23 µs (TGR350), it was estimated that a high-angle crack had opened and reduced the seismic velocity. The travel time delays continued even after the earthquake, reaching approximately 100 µs for the P wave and approximately 250 µs for the SV wave about 1.5 months later, and then both gradually became faster. During the earthquake, the pore water pressure in the Toki Granite (depth 200 m in the Mizunami Underground Research Laboratory) and the groundwater level (TGR350) rose suddenly and continued to rise gradually for about 1.5 months, after which they began to decrease. It is presumed that the earthquake caused an increase in groundwater flow, which increased pore water pressure and porosity, resulting in a decrease in seismic wave velocity. The subsequent decrease in pore water pressure led to a recovery in seismic wave velocity. This paper reports the results of an experiment conducted over approximately seven and a half months, but the continuous transmission test has been ongoing for five and a half years, demonstrating that underground monitoring is possible over long periods with almost maintenance-free.

In pre-modern Japan, time was expressed using the 12-division system, which is based on that in China. This was generally an irregular time system, but each division can be approximately 2 hours. The divisions were called as times of animals and others. Among them, Time of Rat refers to midnight and 1 hour before and after that. This implies that Time of Rat on a certain day includes the period from around 23:00 to 24:00 on that day and the period from around 0:00 to 1:00 on the next day. In addition, in pre-modern Japan, the sense of date change was different from that in modern Japan. According to the book of Hashimoto, in the Heian period (794-1185), the date was changed between Times of Ox and Tiger, which is around 3:00. In the Edo period (1603-1868), there was a sense that the date changed at Time of Rabbit, which corresponded to the sunrise. When using historical earthquakes for scientific research, we need their dates in the modern sense. We thus reexamined the dates of historical earthquakes that were listed in the 2013 book of Usami et al. (Book) with Times of Rat, Ox, and Tiger. A simple method for reexamination is to consider an earthquake that occurred on that day if historical documents contain words related to the morning, and consider it to occur on the following day if they contain words related to the night. A more reliable method is to consider an earthquake on that day if its descriptions in historical documents come before any descriptions of human events or actions, and conversely consider it in the following day if they come afterwards. The results from these methods were compared with the dates in the Book, and the Database (DB) of Historical Documents on Japanese Earthquakes and Eruptions in the Ancient and Medieval Ages (up to February 1607), or the DB of materials for the history of Japanese earthquakes (since March 1607). Of the 56 earthquakes in total, there were four for which it was determined that the dates in the Book should have been changed, 22 for which it was determined that the dates in the DB should have been changed, and for the remaining 20, it was confirmed that the dates in the Book and DB were correct.