In this study, we conduct First-motion Augmented Moment Tensor (FAMT) analysis to estimate the source parameters of small-to-moderate suboceanic earthquakes (MJMA 4.5-5.8) occurring off the Okinawa island, Japan, from October 2007 to April 2020. We calculate Green’s functions for a land-ocean unified 3D structure model that includes non-flat complicated land and seafloor topographies, the sediment layers, the oceanic plate and Moho interfaces obtained by compiling with recent seismic survey results. For the inversion, we use velocity waveform data in two period ranges of 4-40 s and 10-40 s and reproduce both the body and surface waves at K-NET and F-net stations. The results indicate that two of ten events are located in the oceanic mantle of the subducting Philippine Sea (PHS) slab, and eight ones are located in the oceanic crust of the PHS slab. The focal mechanisms of the former two events are primarily strike-slip and/or normal-fault type. The latter crustal events have mechanisms of reverse-fault type with the extension axis of NW-SE direction, which is consistent with the subducting direction of the slab. It is found that the depth difference between the JMA unified hypocenter catalog in travel time data and our estimated FAMT solution is large for events near the Ryukyu trench. Our results show that the offshore events in the trench side are located around the submarine active faults and are the same strike direction as the orientation of the faults.
The coasts of the Japan Trench and the Kuril Trench have been repeatedly damaged by tsunamis in the past. In the 17th century, the 1611 Keicho-Oshu earthquake, the 1677 Aomori-Oki earthquake, the 17th century great earthquake (Tokachi-OKi and Nemuro-Oki) occurred, as well as a tsunami caused by the 1640 Hokkaido Komagatake eruption. Tsunami deposits of these earthquakes have been identified over a wide area along the coast from Hokkaido to the Tohoku region. In previous studies, tsunami deposits from the 17th century great earthquake and the 1611 Keicho-Oshu earthquake have been identified along the eastern Pacific coast of Hokkaido and along the Pacific coast of Tohoku region, respectively. The 1640 Komagatake eruption tsunami left the 17th century tsunami deposits in Uchiura Bay. On the other hand, tsunami deposits in the central part of the Iburi Coast to the northern part of the Hidaka Coast may have been deposited by any of the 1640 Komagatake eruption, the 17th century great earthquake, the 1611 Keicho-Oshu earthquake, or other earthquakes, which have not been identified at present. The purpose of this study is to estimate the source of tsunami that reconstructs the 17th century tsunami deposits from the northern Hidaka coast to Uchiura Bay in Hokkaido, because it is important for future earthquake and tsunami disaster prevention to clarify the origin of the tsunami deposits with unknown sources. First, tsunami heights estimated from historical records, distribution of tsunami deposits, characteristics of coastal tsunami arrival, and tsunami history estimated from geologic records are summarized. Next, tsunami simulations were conducted for the 17th century great earthquake, the 1611 Keicho-Oshu earthquake, the Aomori-Oki earthquake, Hokkaido (2012) model, Cabinet Office (2020) model, and the 1640 Komagatake eruption, and tsunami sources that reproduce (1) the distribution of tsunami deposits, (2) the reproducibility of tsunami arrival characteristics, and (3) tsunami history were examined. Finally, the possibility of (1) a large slip (18 m to 40 m) in Aomori-Oki and (2) coupling of the Tokachi-Nemuro-Oki and Aomori-OKi, as assumed by the Hokkaido (2012), Cabinet Office (2020) and Sato et al. (2022) models, are discussed based on seismological findings, tectonic background of the Japan Trench and the Kuril Trench, and examples of M 9-class great earthquakes occurring in Japan and abroad. As a result, the most reasonable source of the 17th century tsunami deposits distributed from the northern Hidaka coast to the Uchiura Bay is the tsunami caused by the 1640 Komagatake eruption. The results of this tsunami simulations indicate that the height of the tsunami deposits in some areas is considerably higher than that of the previously identified tsunami deposits, and that further detailed investigations are necessary.
In Geophysics, it is often necessary to measure microscopic displacements in order to facilitate research. However, the resolution of displacement sensors currently available on the market is about 0.1 μm, and higher resolution is often required. The authors have therefore developed a displacement enlargement system for borehole observation, solving the problem of insufficient resolution. In normal environments, unlike the case of observation in boreholes with small dimensions, there is no limit on the size and material of the displacement enlargement system, and it is considered possible to develop a system with higher performance. An attempt was therefore made to develop and apply the displacement enlargement system in order to examine the performance and further possibilities of the system. We fabricated a short-span strain seismometer incorporating the displacement enlargement system and compared observed seismograms with those made with an existing long-span extensometer. This revealed that the observation results obtained with a 1.1-meter-long short-span strain seismometer incorporating the displacement enlargement system were almost equivalent to those obtained with a 30-meter-long existing quartz-tube extensometer. Although the size of the incorporated displacement enlargement system was only about 10 cm, it was confirmed that the magnification is 53.5 and the system improved the resolution remarkably, with detection down to 6.2×10－5 μm. The results indicate that a short-span strain seismometer developed is useful for observation of strain seismograms and that it is possible to develop a displacement enlargement system with a higher performance.
We estimated the spatial distribution of site amplification factors for wide frequency bands (0.5-1 Hz, 1-2 Hz, 2-4 Hz, 4-8 Hz, 8-16 Hz) around Aomori prefecture, northern Honshu, Japan, by inversion analysis of the coda wave amplitude of strong motion seismograms. This study used a large number of seismograms obtained by K-NET, KiK-net, and AS-net, the dense seismic network installed covering northern Aomori and southwestern Hokkaido. We found that at lower frequency bands of 0.5-1 Hz, 1-2 Hz, and 2-4 Hz, the estimated logarithmic amplification factors range from －0.8 to 1.0 among locations. In particular, the factors are small in the Kitakami Mountains and large in the Tsugaru peninsula and southeastern Aomori, regardless of the depth of the station. On the other hand, at higher frequency bands of 4-8 Hz and 8-16 Hz, the distribution of the site amplification factors was quite different; the site amplification factors varied in a relatively narrow range, and estimated factors differed significantly between the east and west sides of the volcanic front. This study suggests that a comprehensive analysis using different seismic networks is effective for estimating site amplification characteristics.
During deep earthquakes, it is well known that an anomalously large intensity has been observed in areas farther away from the epicenter than in areas closer to it, which is known as anomalous seismic zone. The deep earthquakes off the western coast of Sakhalin (d＝600 km) and off Cape Soya (d＝334 km) have shown a characteristic distribution of maximum amplitudes in northern Hokkaido, closer to the epicenter, in addition to the areas farther away from the epicenter according to the anomalous seismic zone. The maximum amplitude of the seismic wave at each distance displayed a V-shaped, decreasing with increasing epicentral distance and increasing again above a certain distance. The decrease in the northern portion of Hokkaido was very steep and could not be attributed to a simple geometrical spreading. Thus, the maximum amplitude distribution of deep earthquakes in northern Hokkaido could not be explained by the conventional knowledge of anomalous seismic zones. To explain this observation, we performed numerical simulations of 3-D seismic wave propagation using the deep earthquake off Cape Soya as a model and investigate the intrinsic attenuation structure of the mantle wedge under Hokkaido, Japan. We used a standard 3-D heterogeneous structure model under the Japanese archipelago with a plate geometry based on the latest seismic tomography, and the 3-D domain covered the epicenter and Hokkaido. An anisotropic von Kármán-type short-wavelength inhomogeneous medium was superimposed within the slab, and a seismic velocity gradient of 2% larger at the center than at the edges within the plate was set. We also set up a high attenuation zone of Qs＝20 in the mantle wedge to account for the dehydration of hydrous minerals in the oceanic crust. Under these conditions, we calculated the seismic waveforms and 3-D seismic wavefields at each Hi-net station in Hokkaido. The observed V-shaped amplitude distribution was successfully reproduced in the 1-2 Hz band. Results show that the high attenuation zone in the mantle wedge is necessary to reproduce the observed features, as the V-shaped amplitude distribution was not reproduced when the high attenuation zone in the mantle wedge was removed. To further constrain the location and size of the high-attenuation zone, we performed number of numerical simulations, with adjusting the location of the high-attenuation structure and the Q-value of the mantle wedge, and searched for the 3-D structural model that would best explain the observations by forward modeling. This allowed us to approximate constraint on the range of mantle wedge Q-values around and within which the high attenuation structure is present.
The 1923 Kanto earthquake (M=7.9) is one of the most destructive events in Japan. Many studies on seismology and earthquake engineering have been carried out for this earthquake. In one and half decade from 1990, investigations of its source process and strong ground motions had been especially advanced. This paper reviewed the results of the studies in this period. Records from Ewing seismoscope and Imamura seismograph at Hongo were restored and simulated to presume the long-period strong ground motion in Tokyo during the main shock. Other seismograms involving Imamura seismograph records which completely recorded ground motions from main shock and aftershocks at several stations were also revealed in this period. These seismograms were useful for elucidating the source process of the main shock having two big sub-events, whose locations were about 40 km distant on the fault plane, which occurred with time interval of 10 to 15 s after the initial small rupture duration of 3 to 5 s. This complex faulting process of the main shock caused the regional variations of strong ground motions in the near source region, which were estimated from the descriptions of personal experiences. The aftershock sequence was also clarified from these seismograms. Two M=7 class aftershocks were newly found to have occurred at 3 and 4.5 minutes after the main shock, so that total number of aftershocks with more than M=7 was six for the Kanto earthquake. The number of big aftershocks of the Kanto earthquake is meaningfully larger than the other M=8 class subduction events. The seismic intensity on the JMA scale due to the second shaking caused by the first big aftershock was also estimated to be 6 in Tokyo metropolitan area, though the analysis of the descriptions of personal experiences, which was almost the same as the first shaking by the main shock. Besides above studies, the precise seismic intensity distribution of the Kanto earthquake was obtained in this period, after raveling out many confused damage statistics data.
On April 27, 17th year of Tempyo era in the Japanese traditional calendar (June 1, 745 in the Julian calendar) a large earthquake occurred in central Japan, causing serious damage to the local capital and its surroundings of Mino Province (southern half of Gifu Prefecture at present). In the most authoritative catalog of Japanese damaging earthquakes, its magnitude (M) has been estimated at around 7.9, and an active fault study inferred that the entire Yoro fault system of 55 km long had generated this earthquake. However, I show in this paper that these views are probably overestimation of this earthquake because (A) the catalog’s summary came from (1) Omori’s misreading of Shoku Nihongi (Chronicle of Japan, Continued), the only historical record of this earthquake, in 1913 and (2) Kawasumi’s overestimation of M in 1951, and (B) the active fault interpretation was based on the catalog’s M of 7.9. If the earthquake had been the entire rupture of the Yoro fault system with M 7.9, the disaster should have been devastating not only in Mino Province but also in wide areas of Ise, Owari and Ohmi Provinces (present-day Mie, Aichi and Shiga Prefectures, respectively). However, no description in Shoku Nihongi suggests such disaster. Therefore, the earthquake disaster is considered to have been restricted in and around the Mino Provincial capital. Although an archeoseismological investigation reported that liquefaction had occurred at Jizogoshi ruins in Owari Province due to this event, the judgement of the occurrence time was affected by the overestimated magnitude. Meanwhile, Shoku Nihongi says that aftershocks were felt for about two weeks in the capital, Shigaraki-no-miya (SM) in Ohmi Province, about 65 km southwest of Mino Provincial capital, though the damage in SM is not written. Taking all these features into consideration, I infer that the source region of this earthquake was located around Mino Provincial capital, and estimate its M at roughly 7.0 to 7.3. If we assume that the Yoro fault system, which is reported to have ruptured around the 8th century based on several surveys, generated this event, it is reasonable that only its northern half, 20 to 30 km long, ruptured. The seismic intensity (on the JMA scale) at SM resulting from this inferred source is estimated at about 4 by an empirical attenuation relation of seismic intensity, which harmonizes with the historical record. The seismic intensities at SM due to aftershocks of M 5.0 to 6.0 are estimated at 2 to 3, which also harmonizes with the historical record. However, besides aftershocks in the source region, induced seismic activity may have occurred in Ohmi Province causing seismic intensity of 5 at some limited places near SM, because Shoku Nihongi suggests that liquefaction occurred around SM several times. Interdependency among immature study results of historiographical seismology, active fault study and archeoseismology has erroneously fixed the overestimation of this earthquake. For fruitful interdisciplinary cooperation among these fields, at first, investigation in each field should be carried out as deeply and independently as possible.
The Kanto region, Japan, has often been hit by large earthquakes, such as the 1855 Ansei-Edo Earthquake and the 1923 Kanto Earthquake, and has been damaged by strong seismic ground motion. The distribution pattern of strong ground motion, or seismic intensity, in this region depends on the location of the hypocenter because the three-dimensional seismic wave attenuation structure is complex due to the dual subduction system by the Philippine Sea and the Pacific slabs. Since the S-wave attenuation structure (QS) particularly affects the amount of the ground motion and seismic intensity, the Qs structure is essential for seismic hazards in and around the Kanto. In this study, therefore, we examined the three-dimensional Qs structure in and around the Kanto using the short-period (1-10 Hz) seismic motions. The resultant detailed Qs structure reconfirmed the presence of strong attenuation (low-Qs) zone at the depths of 20-40 km, which lay east to west beneath the border of Ibaraki and Chiba prefectures. Another low-Qs zone was detected beneath the Boso peninsula, reflecting the serpentinization of the easternmost part of the subducting Philippine Sea slab mantle. The three-dimensional Qs structure obtained in this study will improve seismic intensity predictions for earthquakes that occur in and around the Kanto region and will contribute to investigations of hypocenter locations of historical earthquakes.
As the interplate Kanto earthquakes along the Sagami trough, central Japan, three events are known in 1923, 1703 and 1293. For disclosing an event between 1293 and 1703, I investigated two candidates of the Kanto earthquake in the 15th century historiographically. The 1433 Eikyo earthquake, the first candidate, was felt strongly and caused damage in a wide area of the Kanto district, and was probably felt in Kyoto. At Kamakura on the northeastern coast of Sagami Bay, many aftershocks were felt for about twenty days, landslides occurred, and all earthen fences fell down. Hearsay in Kyoto tells that in Kanto buildings fell down and many people died. I estimated the seismic intensity at Kamakura at least at 5 Upper to 6 Lower (on the JMA scale) and that at Yamanashi-shi about 90 km northwest of Kamakura, at 5 Lower to 5 Upper. By adopting these estimations to a ground motion prediction equation, a probable source region has been inferred to be almost the same as that of the 1923 Kanto earthquake and M, around 8. However, there is no reliable record of tsunamis, although hearsay in Kyoto tells that the Tone River emptying into Edo Bay connected to Sagami Bay flowed backward, which suggests a tsunami run-up. If the Eikyo earthquake produced no tsunami, it is considered an inland event between Kamakura and Yamanashi-shi. I checked the possibility that the Isehara fault, the most probable source in terms of geography and its activity history, had generated this earthquake. The possibility turned out low because the predicted seismic intensity at Yamanashi-shi was 4. The 1495 Meio earthquake, the second candidate, has been suspected to be a great Kanto earthquake based mainly on a presumable tsunami deposit dated to the late 15th century found at Usami archeological site on the west coast of Sagami Bay. However, there is no record of this earthquake in contemporary historical documents in the Kanto district except for strong ground motion and a tsunami at Kamakura written in a brief chronological table (Kamakura onikki). It is uncertain whether the earthquake was felt in Kyoto though two contemporary diaries in Kyoto record earthquakes on the same day. Moreover, the event deposit at Usami has problems; marine diatom fossils have not been checked yet, and it is reported that the deposit did not contain ‘Ogama’ potteries widely used since around 1485 even though many fragments of potteries of the 15th century were contained. Therefore, it seems almost impossible that the 1495 Meio earthquake was a great Kanto earthquake. If the 1433 Eikyo earthquake is surely an interplate event, it becomes that a Kanto earthquake recurred only 140 years after the 1293 event, which brings about a significant problem for the present earthquake countermeasures. For unquestionable confirmation of the 15th century’s Kanto earthquake, more paleoseismological investigation is needed. Especially, a further study of the Isehara fault and searching for tsunami deposits in the 15th century are very important. Re-examination of the Usami event deposit is also indispensable.