Offshore extension of the Kakuda-Yahiko fault and its fault-propagating fold are investigated based on combination of high-resolution seismic surveys and sedimentary analysis to evaluate their Holocene activity. High-resolution multi-channel seismic surveys with the boomer sonar (13 lines) and high-frequency seismic survey with the chirp sonar (3 lines) ran across the fault line and its extension, and provided us sharp imageries of stratigraphic succession formed after the last lowstand. We can clearly recognize progressive deformation at the eastern-wing of the fault-propagating fold in the observed succession. To determine actual ages of the seismic strata, we measured radiocarbon dates from sedimentary cores obtained by 2 offshore borings and 2 vibro-coring surveys. Based on deformation of strata between 10.8 ka and 4.7 ka, averaged vertical slip rate was figured out to be 2.1 m/ky. On the other hand, we investigated temporal changes of accumulation rates at both the upthrown and the downthrown sides of the fault. Dynamic changes of the accumulation rates indicate that there were approximately 3 events at least during the last 10.8 ky and the averaged vertical slip rate was 3.0 m/ky during the period. Another focus of our study is on the latest activity of the fault. On the fault-propagating fold, we distinguished the uppermost strata younger than 0.9 ka without any deformation from these older than 2.1 ka with obvious fault deformation based on the profile obtained by the chirp-sonar seismic survey. These observations affirm that the latest event should have occurred between 0.9 ka and 2.1 ka. Furthermore, structural analysis on the deformed strata of 2.1 ka indicates that throw of the fault at the latest “single” activity was ca. 7.5 m. This study successfully demonstrated that set of parameters about the active fault including average slip rate, activity history, timing of the latest event, and associate surface displacement per activity of the fault could be elaborated from series of offshore active fault surveys.

Before the 2011 Tohoku-Oki earthquake, the Off-Kamaishi M∼5 repeating earthquake sequence occurs with a nearly constant recurrence interval of ∼5 years. The seismicity in the Kamaishi area becomes active in the later part of their M∼5 earthquake cycles. Just after the 2011 Tohoku-Oki earthquake, however, the recurrence interval has become as short as 9 days and small earthquakes have become undetectable. The decrease of detectability is mainly due to successive occurrence of large aftershocks. It has not been verified that there is no missing event even for the relatively large Off-Kamaishi mainshock sequence (5≦M≦6). In this study, we collect continuous waveform data of the NIED Hi-net and the microearthquake observation network of Tohoku University near the Off-Kamaishi sequence. We used the repeating sequences as template events and applied Matched Filter (MF) method to detect earthquakes for ∼1 month period (March 11, 2011 to April 13, 2011) after the Tohoku-Oki earthquake. We detected 25 earthquakes including 18 earthquakes that are not listed in the Japan Meteorological Agency (JMA) catalog. These detected earthquakes are highly likely to be located in the vicinity of the Off-Kamaishi mainshock sequence. Further, we found that the sequences are almost completely detectable from two minutes after the Tohoku-oki earthquake. This enabled us to evaluate immediate afterslip in Off-Kamaishi region from the cumulative slip of the mainshock sequence. The result shows the afterslip amount was 1.0∼1.4m in a period from 1 hour to 34 days after the Tohoku-Oki earthquake. For the post Tohoku-oki smaller events down to M2, almost all earthquakes occurred in the latter part of the two Off-Kamaishi mainshock earthquake cycles, which is a similar pattern observed before the Tohoku-oki earthquake. The investigation of undetectable time of earthquakes also supports that this tendency is not an artifact. This result indicates that the temporal change of microseismicity is preserved even when the repeating interval is extremely short (10 to 23 days) due to the afterslip of the Tohoku-Oki earthquake and suggests similar process is occurring in the earthquake cycle of the Off-Kamaishi sequence before and after the Tohoku-Oki earthquake.

After the Russo-Japanese War, too little information about earthquakes was published by the Central Meteorological Observatory, and it has made us left in difficulty to examine earthquakes of the Taisho era precisely. Among such earthquakes, the M5.7 destructive earthquake on August 6th, 1916, in Ehime Prefecture was re-examined by the excavated same age documents. Irazu weather station and other stations, which were operated by the company of Besshi copper mine, reported the detail of this earthquake. Identification of reported places about cracks and falling rocks was done, in addition to the examination of the seismic intensity distribution left in newspapers and the Official Gazette. It is proposed to move the epicenter of this earthquake to (33.95°N, 133.4°E), where is six-km south from that in Utsu-catalogue, and closer to the Ishizuchi fault, which belongs to the Median Tectonic Line active fault zone. It is also found that this event has a felt foreshock a half day before, and a felt aftershock an hour later. This event is the first candidate of destructive earthquakes of the MTL active fault zone in the written history.

Receiver function analysis is an effective method to estimate subsurface structure beneath a seismograph. To apply this method, responses of all three components of a seismograph should be almost the same. However, due to various observation conditions and/or deterioration of the sensors, the response of each component tends to differ. To check the effects of the differences in the sensor response between each component to the receiver function analysis results, we investigated the teleseismograms observed at the N.TSMH station of NIED Hi-net as an example. At this station, the NS component showed a lower response than the other components for frequency ranges lower than 1 Hz. As this feature was resolved after seismograph replacement, we set two analysis periods, before and after the seismograph replacement. Comparing backazimuth dependence of the receiver functions, we first confirmed that the Ps converted phase amplitudes were slightly smaller in specific directions before the replacement than those after replacement. However, there were no clear differences for the arrival times of these phases. Next, we applied the harmonic decomposition analysis to two datasets and estimated the plunge azimuth of the subducting oceanic Moho beneath the station. To evaluate the stability of the estimation results, 1000-time bootstrap tests were conducted for each dataset. Using the dataset before the replacement, the results showed that the oceanic Moho dipped to the west. Based on the dataset after the replacement, the estimated plunge azimuth was in the west-northwest direction. Previous studies show the slab might be inclined to the northwest or west-northwest direction beneath the Bungo channel. This means that the dataset before the replacement was inconsistent with the previous results because of the lower response of the NS component.

Several types of long period velocity seismograph systems installed at the Tono Research Institute of Earthquake Science (TRIES) to record long-period components went off the scale and failed to record large amplitude during the 2011 off the Pacific coast of Tohoku earthquake (M_W 9.0), which occurred on March 11. Most of the seismograph systems of the Full Range Seismograph Network of Japan (F-net) also went off the scale and were therefore not able to record large amplitude. However, all the continuously recording stress meters and strain meters, developed by the TRIES, successfully recorded up to the largest amplitude. We examined the observable ranges of the borehole stress meters for continuous observations and found that the stress meters and strain meters that we developed have a wide observable range even for highly sensitive components, responded to direct current (DC) components of frequency, and could record even an earthquake several times as large as the Tohoku Earthquake without going off scale. Because the borehole stress and strain meters can respond to DC components, unlike a seismometer, and can record changes associated with the final change in an earthquake source fault, permanent changes observed at the Toshinomori (TOS) borehole observation site of the TRIES were obtained. The obtained permanent change was compared with permanent change calculated from the changes at the three GNSS-based control stations operated by the Geospatial Information Authority of Japan. Further, the seismograms of a foreshock (M_W 7.3) that occurred on March 9 recorded by a stress meter, the STS-1 seismometer, the Guralp seismometer, and the servo velocity seismometer, which are capable of recording long-period ground motion, were compared. This revealed that the three long-period seismometers installed at different place and depth recorded almost the same waveform. Although waveform from the initial motion to the arrival of the surface wave was different for a stress meter, the subsequent phases were in fairly good agreement. This study revealed that stress meters are useful not only for studies on normal earthquakes but also for studies on strong long-period ground motion and highlighted their importance for future studies including studies on the forecasting of wave height caused by giant tsunamis.

This study analyzes Japanese newspaper coverage of seismology from 1990 to 2016 and investigates how newspaper media portray the scientific ideas of seismology. In all, 19360 articles that contained references to scientific stories on seismology published by the Asahi Shimbun, the Yomiuri Shimbun, and the Mainichi Shimbun are collected. The content of newspaper coverage is classified into 14 topics using the cluster analysis with the distributed representation of articles which is generated by the machine learning method, Paragraph2Vec. The time-series analysis of articles reveals that a very large number of seismological reports are released immediately after big earthquakes occur but this number rapidly decreases within a few months. This means that the fleeting intense coverage greatly elevates the level of public concern about the seismic risk, while it fails to constitute the enduring forum for public debate of seismic risk issues. The content analysis with use of the natural language processing shows that the “earthquake prediction research” topic is covered most frequently and plays a pivotal role in seismological coverage. However, reports on “earthquake prediction research” topic are recently declining and obviously taking on a scientific research aspect. Results argue that the manner in which newspaper media report seismically related issues is undergoing major changes after the 2011 off the Pacific coast of Tohoku Earthquake. Specifically, media pay a great deal of attention to the issue of preparation measures for the future Nankai Trough earthquake and more news related to probabilistic assessments of earthquakes are conveyed as well. Such coverage would develop a public perception that the risk of coming earthquakes can only be described as the probabilistic assessment. In the societal debate about probabilistic seismic hazard maps, accountability and communication responsibilities of experts for the probabilistic assessments of seismic risk must be more strongly demanded from mass media and public. Moreover, evolving digital technologies create various types of information platforms such as digital media and SNS, some of which may provide a distorted depiction of seismic risk. Therefore, in order to realize effective risk communication in a rapidly changing media landscape, seismologists should not solely corroborate with legacy media organizations, but also support communication organizations which improve links between scientific community and media and enhance the accuracy of scientific coverage.

An active fault map of Japan Sea was compiled based mainly on bathymetric data and seismic profiles that cover shelves to slopes between 4 to 150 km from the coasts of Japan Islands. The seismic profiles using air guns as seismic sources revealed active faults that have slipped during the last 104 to 106 years. In addition, high-resolution seismic profiles using a boomer as the seismic source were obtained along seaward extensions of onshore active faults in shallow sea areas less than 150 m below sea-level, and activity of the faults during the last 104 years was identified. In northeastern Japan Sea, to the northeast of the Noto Peninsula, many reverse faults accompanying large hanging wall anticlines (>750 m uplift) are concentrated in two N-S trending fault zones along the Okushiri and Sado ridges. Earthquakes larger than M 7.5 in 1940, 1964, 1983 and 1993 occurred in these fault belts. Three NE-SW trending fault zones cross the N-S trending fault zones and disrupt the structure of the N-S fault zones. In the offshore area from the Noto Peninsula to the Tango Peninsula, active reverse faults accompanying smaller hanging anticlines (<375 m uplift) are identified in a zone trending subparallel to the coasts. In addition, NW-SE to N-S trending strike-slip and reverse faults extend from onshore to offshore. In the offshore area to the west of the Tango Peninsula, E-W and NW-SE trending active strike-slip faults are identified. The former faults developed in about 40 km wide zones sub-parallel to the coast, and the later faults are located landward of the E-W trending fault zones. Some of the later faults are extensions of onshore active faults. Unknown active faults may exist in shallow sea area along coasts where have not been thoroughly investigated. Displacements of the faults during the last 106 years are large in northeastern Japan Sea and decrease to the southwest, while slip rates of these faults during the last 104 years are inferred to have smaller differences. These faults have the potential of future earthquakes, while there is not enough data to evaluate the activities of these faults.

We examine the source, path and site effects to strong ground motions during the 2018 northern Osaka earthquake (M_J6.1, M_w 5.5-5.7) by spectral inversion and ground motion prediction equations (GMPEs) using strong motion records. It is found that the Q for paths is modeled using frequency f as 38f ^1.34 by spectral inversion, which almost agrees with Q estimated by a previous study. The observed attenuation gradients for 5%-damped acceleration response spectra S_A and peak ground velocity PGV are consistent to GMPEs for crustal earthquakes in western Japan. These results show that path effects are average as crustal earthquakes. The site amplification factors from the seismic bedrock for PGV and Fourier spectra at periods of 1 to 2 s are large in the northeast and southwest directions from the source. This result shows that one of the causes of large instrumental seismic intensity in these directions is site amplification factors. The S_A at periods of 0.1 to 4 s after the correction by site amplification factors in the GMPEs are 1.3 to 1.6 times larger than the GMPEs for M_w 5.5 and 1.0 to 1.3 times larger than GMPEs for M_w 5.7. On the other hand, the S_A at periods of 4 to 5 s is the average level. The flat level of acceleration source spectra called as short-period spectral level A is estimated to 5.1−5.2×10¹⁸ Nm at periods of 0.2 to 2 s by spectral inversion. The estimated A is larger than A for previous crustal earthquakes with the same seismic moment M₀ and the empirical M₀-A relations. From both the spectral inversion and the comparison with GMPEs revealed that the generation of strong ground motions from the source at periods of 0.1 to 4 s are larger than the average of crustal earthquakes with the same M₀ in Japan. It was pointed out by previous studies that the northern Osaka earthquake was composed of a strike-slip fault and a dip-slip fault and that the rupture mainly propagated to the southwest direction on the strike-slip fault. Therefore, we examine the effects of near fault rupture directivity and the radiation pattern to strong ground motions by correcting the path and site effects using spectral inversion results. It is found that the spectra become larger at periods longer than about 0.5 s at stations in the southwest direction from the source. Especially the effects of transverse components are larger than radial components and transverse components have clear velocity pulses. On the other hand, the spectra at stations in the northeast direction become smaller. The effects of near fault rupture directivity and the radiation pattern for S_A are almost reproduced by the previous empirical model on the average by using both the strike-slip and dip-slip faults. However, the effects at periods of 0.5 to 1 s are noticeably stronger than those predicted by the model and are interpreted to be the other cause of large instrumental seismic intensity from 6 lower to 5 lower in the southwest direction.