The way to have another look at earthquake prediction/forecast researches in Japan is discussed from a border between the seismology and the history and philosophy of science. Scientific revolution is often symbolically expressed as the shift of the paradigm, idea of which was introduced by Thomas Kuhn. In the field of earthquake prediction research, here we tentatively define its paradigm as a combination of the following ideas - though these would be too humble :1) plate tectonics, 2) asperity or distribution of frictional strength on a fault surface, 3) constitutive law of friction, 4) numerical simulations of precursory process, and 5) observation system of crustal deformation prior to a large earthquake. The case for the paradigm is that any of convincing precursors to a large earthquake has not been observed yet, and, the distribution of frictional strength on a fault surface, in particular, on the subducting plate boundary, is not precisely estimated. Our history of modern seismology is too short to resolve such observational problems. Given that essence of the progress in science is the repetition of proposal of hypotheses and its verification or its rejection, scientists are strongly obliged to propose new hypotheses based on new findings and discoveries to be tested and discussed. Even if the result does not occur along prediction/forecast, it should be appreciated as the science if their scientific context is acceptable. Progressive theory that prevails over degressive one, in the meaning of “research program” theorem proposed by Lakatos Imre, will stimulate the earthquake prediction research in Japan, for which the authors truly believe in. Individual researchers working on the prediction/forecast researches are responsible only for their scientific context. Administrative organizations of seismologists such as the Headquarters for Earthquake Research Promotion and the Coordinating Committee for Earthquake Prediction should take a responsibility for our society.

Strong seismic ground motions often make stone lanterns in temples and shrines collapse and suffer damages, and the stone lanterns tend to collapse in the direction of strong shaking in recent cases. In historic Japanese documents, descriptions of collapsed stone lanterns as results of strong earthquakes appear frequently, and are used in estimating and mapping the seismic intensity. In this report, we examine whether it is possible to further retrieve the characteristics of historic ground motions from the ages and damages of old stone lanterns at Zenkoji Temple, Nagano. We assume that severely damaged stone lanterns would be removed from the site after the earthquakes and be newly rebuilt thereafter, and this will distort the age distribution of stone lanterns. Less severely damaged lanterns would be rebuilt with the damaged parts, and the damages of stone lanterns could be the records of historical strong seismic ground motions at the site. If the directions of the damages caused by collapses are maintained, these could be used to estimate the direction of collapses, or of strong motion. Scarcity of the stone lanterns which were built before 1710s at Zenkoji Temple as well as the sudden increase of repairs and rebuilts in the mid 19th century are likely the results of strong earthquakes and resultant damages at Zenkoji Temple. Ratios between numbers of the damaged and undamaged stone lanterns do not correlate with the time series of historic earthquakes in Nagano. Damages of the stone lanterns are more often found in the rear of the stone lanterns, and we could not find that damages are concentrated into a particular cardinal direction. Stone lanterns at this site are likely to be so maintained that the damaged parts are rotated into the direction which makes the major damages less visible. On the contrary to the findings of the previously published report, we conclude that it is impossible to estimate the direction of historic strong motion from the statistics of damages of stone lanterns at Zenkoji Temple.

An earthquake cannot be observed near the hypocenter as it occurs at a significant depth underground. Accordingly, an earthquake is observed on the surface of the earth, where signals of the earthquake will be disturbed by artificial noises and other factors. Therefore, it is difficult to detect weak signals, including precursory movements. A possible solution to such problems lies in installing a monitoring instrument in the deep bedrock where S/N ratio is improved much better. Based on such an idea, we have been developing a multicomponent borehole instrument (a comprehensive crustal activity observation instrument) that is capable of observing crustal activity at significant depths for predicting earthquakes. This instrument can be equipped with highly sensitive stress meters, strain meters, tilt meters, seismographs, accelerometers, thermometers, and declinometers, and it is capable of observing multiple components of various types of signals. The stress meters were recently developed and can observe both stress and strain as reported in a previous paper about development of stress meters.

Stress and strain meters recorded seismic waveforms generated by the 2011 Tohoku earthquake (M9.0), which occurred on March 11, 2011, and provided useful data for research. In addition, favorable stress waveforms and strain waveforms of an earthquake that occurred in Chile (M8.7) on April 2, 2014 were also recorded．The epicentral distance is approximately 15000 km from the observation point. The tidal movements of the earth were also well recorded by the stress and strain meters.

This paper examined the reliability of the data observed by these instruments. We obtained invariants in elastic dynamics from earthquake waveforms and the earth tide recorded by the stress and strain meters and checked whether the invariants satisfy elastic invariant theory. The analysis revealed that the invariants obtained by the stress and strain meters were in good agreement with the theory. We concluded from these results that the developed stress and strain meters used for observation have been properly installed in the bedrock and have adequate accuracy and reliability.

We used pop-up ocean bottom seismometers deployed in 2013 and 2014 to investigate the southern limit of seismic activity within the Philippine Sea plate south of the Nankai Trough axis off the Kii Peninsula. The hypocenter distribution we determined included microearthquakes with magnitudes lower than 1.5 that were not detected by the land-based seismic network. Hypocentral depths ranged from 5 to 15 km below sea level and there were few earthquakes more than 100 km south of the axis of the Nankai Trough. We therefore infer that the southern limit of microearthquake activity in this region is about 100 km south of the trough axis.

The Gravity Recovery and Climate Experiment (GRACE) satellite system was launched in 2002, and has been playing important roles in various disciplines of earth and environmental sciences through measuring time-variable gravity field of the earth. It also offers a unique viewpoint to study earthquakes in terms of mass redistribution. We provide a review of earthquake studies with GRACE, e.g. basic facts of the satellite system and available data types, several kinds of non-earthquake gravity changes which may mask the earthquake-related signals. We also summarize past researches about co- and postseismic gravity changes. Two dimensional coseismic gravity changes were first observed with GRACE for the 2004 Sumatra-Andaman earthquake. After that, GRACE has caught coseismic gravity changes of the 2010 Maule, the 2011 Tohoku-oki, the 2012 Indian-ocean, and the 2013 Okhotsk deep-focus earthquakes. Such coseismic gravity changes are due mainly to two factors, i.e., the density changes around the fault edges, and the vertical deformations of boundaries with density contrasts such as the surface and the Moho. Short- and long-term postseismic gravity changes are considered to stem from afterslip and viscoelastic relaxation, respectively, but further studies are needed to quantitatively explain the observations.

The Ecuador-Colombia earthquake that occurred on 31 January 1906 has been considered as one of the largest thrust-fault earthquakes, with magnitudes of M_t 8.7 and M_w 8.8, and the largest event ever recorded in the subduction zone off the Ecuador-Colombia region. The value of M_t 8.7 was derived mainly by the tsunami height data in Hilo, Hawaii where the tsunami was reported as high as 12 feet (3.6 m) on a local newspaper. The earthquake was followed by three large earthquakes in 1942 (M_s 7.9), 1958 (M_s 7.8) and 1979 (M_s 7.7) in the same area. We reexamined tsunami records of the 1906 earthquake from the newspaper articles in Hawaii, wave heights in tsunami catalogues, and tidal gauge records in Japan. Ratios of tsunami heights of Hilo to Kahului, both in the Hawaii Islands, are in a same order for all the tsunamis generated by earthquakes in the western coast of South America since 1920’s. This contradicts with the 1906 values of 3.6 m in Hilo and 30 cm in Kahului. We re-evaluated the tsunami magnitude M_t of the 1906 earthquake to be much less than 8.7. Next, we compared synthetic tsunamis of three fault models (M_w 8.5, 8.6 and 8.8) with the observed tsunami heights at several stations. The model of M_w 8.5 was found to be most appropriate. These results are consistent with the long-period seismic magnitude (M_w smaller than 8.5) because of the lack of major arc wave (G2) in the record of Uppsala, Sweden. The previous seismic estimation of M_w 8.8 may be due to the over-estimation of its fault extent, judging from nearby coastal uplift records, which would include after-slips and/or slow slips near its source region.

We investigated the triggering process for the Mj6.4 Eastern Shizuoka earthquake of 15 March 2011, which occurred 4 days after the 2011 Mw9.0 Tohoku-Oki earthquake and about 4 minutes after the Mj6.2 Fukushima-Oki earthquake. The static Coulomb failure stress change on the fault of the Eastern Shizuoka earthquake from the Tohoku-Oki earthquake, was about 20 kPa, and the largest dynamic stress change by the passing surface waves was about 200 kPa. The largest dynamic stress change from the Fukushima-Oki earthquake and the largest tidal stress change after the Tohoku-Oki and before the Eastern Shizuoka earthquake were about 0.3 kPa and 1.4 kPa, respectively, while those at the onset of the Eastern Shizuoka earthquake were at most 0.01 kPa and −0.5 kPa, respectively. We also analyzed seismicity by detecting earthquakes immediately preceding the Eastern Shizuoka earthquake, which is done using a matched filter technique. A single M1.0 event that occurred about 17 hours before the Eastern Shizuoka earthquake and located about 3 km NNE from the hypocenter was found, however, this event may not be classified as a foreshock if we consider the background seismicity in this region before 2011. We propose that the seismic cycle was possibly advanced by about 10¹ to 10² years and delayed triggering might have occurred for the Eastern Shizuoka earthquake. The eventual earthquake was ready to occur to some extent, when the Tohoku-Oki earthquake occurred. The fault strength had significantly decreased due to imposed large dynamic and static stress changes from the Tohoku-Oki earthquake and probably its large aftershocks, in terms of a rate and state dependent friction law.