Search

On 15 January 2022, an eruption of Hunga volcano generated an air-coupled tsunami that propagated across the Pacific Ocean. The tsunami reached Japan approximately 7 hr after the eruption and was recorded by coastal and offshore stations. We conducted a retroactive study of tsunami early warning using a data assimilation approach. We assimilated water pressure records of 44 offshore bottom pressure gauges (OBPGs) off the Japanese Islands after the volcanic eruption and forecasted tsunamis on the Shikoku Island and the Kii peninsula. The tsunami waveforms at 12 points of interest (i.e., coastal and near-coast stations) were predicted using a data assimilation approach at 15:00 (UTC) with forecast accuracy of 97%. Moreover, we discussed the effects of air-pressure variations generated by the volcanic eruption. The results showed that air-pressure variations had negligible effect on the assimilation results. This is the first time that data assimilation approach was applied to a nonseismic tsunami event for early warning. Our study indicates that data assimilation approach plays an important role in tsunami hazard mitigation. Early warnings of nonseismogenic tsunamis can be successfully achieved by assimilating offshore water pressure records of OBPGs.

Ocean bottom pressure-gauge (OBP) records play an important role in seafloor geodesy, but oceanographic fluctuations in OBP data are a major source of noise in seafloor transient crustal deformation observations, including slow slip events (SSEs), so it is important to evaluate them properly. To extract the significant characteristics of the oceanographic fluctuations, we applied principal component analysis (PCA) to the 3-year Dense Oceanfloor Network System for Earthquakes and Tsunamis (DONET) OBP time series for 40 stations during 2016–2019. PCA can separate several oceanographic signals based on the characteristics of their spatial distributions, although transient tectonic signals could not be clearly confirmed from the observed pressure records. The higher-order modes of the principal component reflected the oceanographic variation along the sea depth, and we interpreted that they were caused by the strength or weakness and meandering of ocean geostrophic currents, based on a comparison to the global ocean model ECCO2 by “Estimating the Circulation and Climate of the Ocean” (ECCO) consortium. In addition, to evaluate the ability of PCA to separate transient crustal deformation from oceanographic fluctuations, we conducted a synthetic test assuming an SSE by rectangular faults. The assumed synthetic tectonic signal can be separated from the oceanographic signals and included in the principal component independently depending on its amplitude. We proposed a transient event-detection method based on the spatial distribution variation of a specific principal component with or without a tectonic signal. This method can detect transient tectonic signals larger than moment-magnitude scale M 5.9 from OBP records.

OBP (Ocean bottom pressure-gauge) is an essential sensor in crustal deformation observation in the marine region. Still, non-tidal oceanographic fluctuations in OBP data are a significant source of noise in seafloor transient crustal deformation observations, including SSEs (slow slip events), so it is essential to evaluate them properly. To extract and assess the significant characteristics of the oceanographic fluctuations, we applied PCA (principal component analysis) to the DONET (Dense Oceanfloor Network System for Earthquakes and Tsunamis, southwest margin of Japan) OBP time series of 40 stations for three years (2016–2019). PCA could separate several oceanographic signals based on the characteristics of their spatial distributions. The higher-order modes of the PC were distributed in a varying manner with sea depth, and we interpreted that they were caused by the strength and meandering of the ocean geostrophic currents based on similar spatial distribution with the global ocean models. Since PCA cannot detect the significant tectonic signal in observed data, we applied a synthetic ramp simulating SSE to the observation to evaluate the performance of PCA in tectonic signal detection. The assumed synthetic ramp could be separated into an independent principal component according to its amplitude (i.e., the scale of the events). We proposed a detection method of a transient event using the spatial distribution change of the principal component with and without the synthetic ramp using their normalized inner product. Our method was able to detect the tectonic signal of events of MW5.9 or larger from the OBP time series., The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)

SUMMARY The offshore part of the North Anatolian Fault (NAF) beneath the Marmara Sea is a well-known seismic gap for future M > 7 earthquakes in the sense that more than 250 yr have passed since the last major earthquake in the Central Marmara region. Although many studies discussed the seismic potential for the future large earthquake in this region on the basis of historical record, geodetic and geological observations, it is difficult to evaluate the actual situation on the seismic activity and structure along the NAF beneath the Marmara Sea due to the lack of ocean bottom seismic observations. Using ocean bottom seismometer observations, an assessment of the location of possible asperities that could host an expected large earthquake is undertaken based on heterogeneities in the microseismicity distribution and seismic velocity structure. Specifically, seismic tomography and precise hypocentre estimations are conducted using offshore seismic data whose recording period is 11 months. About five times more microearthquakes are detected with respect to events recorded in a land-based catalogue. A comparison with previously published results from offshore observation data suggests that the seismicity pattern had not changed from 2014 September to 2017 May. The location accuracy of microearthquakes is greatly improved from only the land-based earthquake catalogue, particularly for depth direction. There are several aseismic and inactive zones of microearthquake, and the largest one is detected using land-based seismic observation, whereas other zones are newly detected via offshore observations. The obtained velocity model shows a strong lateral contrast, with two changing points. The western changing point corresponds to a segmentation boundary, where the dip angle of the NAF segments changed. High-velocity zones from tomographic images are characterized by low seismicity eastward of the segment boundary. To the east of 28.50°E, the high-velocity zone becomes thicker in the depth direction and is characterized by low seismicity. Although the low seismic activity alone could be interpreted as both strong coupling and fully creeping, the high-velocity features at the same can be concluded that these zones are consist of brittle material and strong coupling. From comparison with other geodetic and seismic studies, we interpret these zones as locked zones that had been ruptured by the past large earthquakes and could be ruptured by future ones. These zones might accumulate strain since the main shock rupture associated with the 1766 May Ms 7.3 earthquake, the latest major earthquake in this region.

National Research Institute for Earth Science and Disaster Resilience (NIED) integrated the land observation networks established since the 1995 Kobe earthquake with the seafloor observation networks established since the 2011 Tohoku earthquake and tsunami as MOWLAS (Monitoring of Waves on Land and Seafloor) in November 2017. The purpose of MOWLAS is to provide comprehensive, accurate, and rapid observation and monitoring of earthquake, tsunami, and volcano events throughout Japan and its offshore areas. MOWLAS data are widely utilized for long-term earthquake forecasting, the monitoring of current seismic activity, seismic and tsunami hazard assessments, earthquake early warning, tsunami warning, and earthquake engineering, as well as earthquake science. Ocean bottom observations provide an extension of observations to areas where no people are living and have the advantage of increasing lead time of earthquake early warning and tsunami warning. The application of recent technology advancements to real-time observations as well as the processing of MOWLAS data has contributed to the direct disaster mitigation of ongoing earthquakes. These observations are fundamental for both science and disaster resilience, and thus it is necessary to continue ceaseless operation and maintenance.

The 10 July 1804 Kisakata earthquake occurred offshore Kisakata (Akita, Japan), and widespread felt shaking was reported from Matsumae (Hokkaido) to Ohmi-Hachiman (Shiga Prefecture). The earthquake caused strong ground motions that extensively damaged areas near the epicenter, such as along the coast of Kisakata, and the resultant tsunami caused extensive damage along the coast from Kisakata to Sakata. Furthermore, Kisakata lagoon was uplifted by dislocation during the earthquake, exposing the lagoon floor. Here, we performed a field survey of the uplift distribution based on microtopographic remnants of the former shoreline of Kisakata lagoon and used historical documents to re-evaluate tsunami trace heights. Using ocean-bottom reflection profiles, we estimated a fault model for the earthquake and resultant tsunami. Our model indicates that an average of 5.6 m of slip on the fault (equivalent to an Mw 7.1 earthquake) is required to explain the observed crustal deformation and tsunami height distributions, and back correction of the modeled slip reproduced well the former shoreline of Kisakata lagoon.

We estimated the origin time of the 1854 Ansei–Tokai tsunami from the tsunami waveforms recorded at three tide gauge stations (Astoria, San Francisco, and San Diego) on the west coast of North America. The tsunami signal is apparent in the San Francisco and San Diego records, and the arrival time was 0–1 p.m. Greenwich Mean Time (GMT) on 23 December 1854, whereas the tsunami signal of Astoria is ambiguous, and the arrival time could not be determined from the waveform. The simulated waveforms on the basis of nonlinear dispersive wave theory by assuming an origin time of 0 a.m. GMT on 23 December arrived earlier than the observations. Cross-correlation functions between the observed and simulated waveforms recorded at San Francisco and San Diego showed a time gap between them of approximately 30 min. Based on these results, we concluded that the origin time of the 1854 Ansei–Tokai tsunami was approximately 00:30 a.m. GMT or 09:46 local time on 23 December. Our result is roughly consistent with reports by a Russian frigate anchored in Shimoda Bay, ranging the earthquake between 09:00 and 09:45 and the tsunami between 09:30 and 10:00. The earthquake was also reported in historical Japanese documents ranging from 8 and 10 o’clock in local time.

The spatial variation of azimuthal S-wave phase velocity anisotropies caused by differential horizontal stress along the subducting plate at the Nankai Trough was analyzed to understand the stress ...

The offshore part of the North Anatolian Fault (NAF) beneath the Marmara Sea is a well-known seismic gap for future M > 7 earthquakes in the sense that more than 250 years have passed since the last major earthquake in the Central Marmara region. Here, an assessment on the location of possible asperities to host the expected next large earthquake is done based on the heterogeneities on the seismic velocity structure. Using long-term ocean bottom seismograph (OBS) observation data, seismic tomography and precise hypocenter estimations have been conducted. As a result, about five times more microearthquakes than the events in a land-based catalog has been detected. A comparison with previously published results suggests that the seismicity pattern has not changed during the three years period between Sep. 2014 and Jun. 2017. The obtained velocity model shows strong lateral contrast whose changing points locate at 28.10°E and 28.50°E. The western corner of the area (28.10°E) corresponds to a segmentation boundary where the dip angle of the NAF segments changed. The high velocity zones in the tomographic images are characterized by low seismicity eastward from the segment boundary at 28.10°E. Eastern 28.50°E, the high velocity zone becomes thicker in the depth direction. These zones are interpreted as asperities to be ruptured by the next large earthquake which are possibly accumulating strain since the mainshock rupture associated with the May 1766 Ms7.3 earthquake.

A ship-based seismic survey was conducted close to a fiber-optic submarine cable, and 50 km-long distributed acoustic sensing (DAS) recordings with air-gun shots were obtained for the first time. We examine the acquired DAS dataset together with the co-located hydrophones to investigate the detection capability of underwater acoustic (hydroacoustic) signals. Here, we show the hydroacoustic signals identified by the DAS measurement characterizing in frequency-time space. The DAS measurement can be sensitive for hydroacoustic signals in a frequency range from $$10^{-1}\,\text {Hz}$$ 10 - 1 Hz to a few tens of Hz which is similar to the hydrophones. The observed phases of hydroacoustic signals are coherent within a few kilometers along the submarine cable, suggesting the DAS is suitable for applying correlation analysis using hydroacoustic signals. Although our study suggests that virtual sensor’s self-noise of the present DAS measurement is relatively high compared to the conventional in-situ hydroacoustic sensors above a few Hz, the DAS identifies the ocean microseismic background noise along the entire submarine cable except for some cable sections de-coupled from the seafloor.

Subduction zones may develop submarine spreading centers that occur on the overriding plate behind the volcanic arc. In these back-arc settings, the subducting slab controls the pattern of mantle advection and may entrain hydrous melts from the volcanic arc or slab into the melting region of the spreading ridge. We recorded seismic data across the Western Mariana Ridge (WMR, northwestern Pacific Ocean), a remnant island arc with back-arc basins on either side. Its margins and both basins show distinctly different crustal structure. Crust to the west of the WMR, in the Parece Vela Basin, is 4–5 km thick, and the lower crust indicates seismic P-wave velocities of 6.5–6.8 km/s. To the east of the WMR, in the Mariana Trough Basin, the crust is ∼7 km thick, and the lower crust supports seismic velocities of 7.2–7.4 km/s. This structural diversity is corroborated by seismic data from other back-arc basins, arguing that a chemically diverse and heterogeneous mantle, which may differ from a normal mid-ocean-ridge–type mantle source, controls the amount of melting in back-arc basins. Mantle heterogeneity might not be solely controlled by entrainment of hydrous melt, but also by cold or depleted mantle invading the back-arc while a subduction zone reconfigures. Crust formed in back-arc basins may therefore differ in thickness and velocity structure from normal oceanic crust.

The damage and loss of life caused by tsunamis can be reduced by timely warnings, which predict the arrival time and maximum height of tsunamis, to support evacuations and other mitigating actions. We have developed a real-time tsunami prediction system based on data from the Dense Oceanfloor Network system for Earthquakes and Tsunamis (DONET) that has been implemented in some local governments along the Pacific coast of Japan. The system generates estimates of tsunami arrival times and the height, inundation areas, and worst case using selected fault rupture models. The main objective of this paper is to show the possibility of applying the above system for a complicated topography area, and we report a successful application of the system in Sakaide, a city on the Shikoku coast of the Inland Sea, using a simulated great plate-boundary earthquake in the Nankai Trough. The simulated tsunami propagates to Sakaide by complicated routes between several islands. According to calculated tsunami waveforms of 1,506 cases, waveforms of tsunamis propagating to the Inland Sea have a relatively uniform frequency, regardless of the magnitude of the causative event, after running through the narrow straits in the Inland Sea. At the same time, waves are amplified as they pass between the islands of Shodoshima and Shikoku by an interaction with reflected waves. These effects are compatible with this prediction system, and we confirmed that our predicted tsunami is consistent with the final result from a model of a magnitude 9 Nankai Trough earthquake.

To evaluate the crustal structure of a site proposed for International Ocean Discovery Program drilling, the Japan Agency for Marine-Earth Science and Technology carried out seismic surveys in the Izu rear arc between 2006 and 2008, using research vessels Kaiyo and Kairei. High-resolution dense grid surveys, consisting of three kinds of reflection surveys, generated clear seismic profiles, together with a seismic velocity image obtained from a seismic refraction survey. In this paper, we compare the seismic profiles with the geological column obtained from the drilling. Five volcaniclastic sedimentary units were identified in seismic reflection profiles above the 5 km/s and 6 km/s contours of P-wave velocity obtained from the velocity image from the seismic refraction survey. However, some of the unit boundaries interpreted from the seismic images were not recognised in the drilling core, highlighting the difficulties of geological target identification in volcanic regions from seismic images alone. The geological core derived from drilling consisted of seven lithological units (labelled I to VII). Units I to V were aged at 0–9 Ma, and units VI and VII, from 1320–1806.5 m below seafloor (mbsf) had ages from 9 to ~15 Ma. The strong heterogeneity of volcanic sediments beneath the drilling site U1437 was also identified from coherence, calculated using cross-spectral analysis between grid survey lines. Our results suggest that use of a dense grid configuration is important in site surveys for ocean drilling in volcanic rear-arc situations, in order to recognise heterogeneous crustal structure, such as sediments from different origins.

We constructed a real-time tsunami prediction system using the Dense Oceanfloor Network System for Earthquakes and Tsunamis (DONET). This system predicts the arrival time of a tsunami, the maximum tsunami height, and the inundation area around coastal target points by extracting the proper fault models from 1,506 models based on the principle of tsunami amplification. Since DONET2, installed in the Nankai earthquake rupture zone, was constructed in 2016, it has been used in addition to DONET1 installed in the Tonankai earthquake rupture zone; we revised the system using both DONET1 and DONET2 to improve the accuracy of tsunami prediction. We introduced a few methods to improve the prediction accuracy. One is the selection of proper fault models from the entire set of models considering the estimated direction of the hypocenter using seismic and tsunami data. Another is the dynamic selection of the proper DONET observatories: only DONET observatories located between the prediction point and tsunami source are used for prediction. Last is preparation for the linked occurrence of double tsunamis with a time-lag. We describe the real-time tsunami prediction system using DONET and its implementation for the Shikoku area.

Both the geometry and the depth of the seismogenic zone of the North Anatolian Fault under the Marmara Sea (the Main Marmara Fault (MMF)) are poorly understood, in part because of the fault's undersea location. We recorded 10 months of microseismic data with a dense array of ocean bottom seismographs and then applied double-difference relocation and 3-D tomographic modeling to obtain precise hypocenters on the MMF beneath the central and western Marmara Sea. The hypocenters show distinct lateral changes along the MMF: (1) both the upper and lower crust beneath the Western High are seismically active and the maximum focal depth reaches 26 km; (2) seismic events are confined to the upper crust beneath the region extending from the eastern part of the Central Basin to the Kumburgaz Basin; and (3) the magnitude and direction of dip of the main fault change under the Central Basin, where there is also an abrupt change in the depth of the lower limit of the seismogenic zone. We attribute this change to a segment boundary of the MMF. Our data show that the upper limit of the seismogenic zone corresponds to sedimentary basement. We also identified several seismically inactive regions within the upper crust along the MMF; their spatial extent beneath the Kumburgaz Basin is greater than beneath the Western High. From the comparison with seafloor extensometer data, we consider that these regions might indicate zones of strong coupling that are accumulating stress for release during future large earthquakes.

Around Japan, some oceanfloor network systems were installed, which are dense oceanfloor network system for earthquakes and tsunamis (DONET), seafloor observation network for earthquakes and tsunamis along the Japan trench (S-net), and so on. These systems have seismometers and pressure sensor, respectively, and are indispensable for early detection of earthquakes and tsunamis around huge events with a magnitude of over eight around the subduction zone. We have developed a real-time tsunami prediction system using above oceanfloor network systems and it was already installed on some local governments and private company of the infrastructure. The prediction items are tsunami arrival time, maximum tsunami height and inundation area. The principal of the tsunami prediction system is based on the tsunami amplification. We set total 1,506 models changing a magnitude, depth, and dip along the Nankai trough area from the off the Boso Peninsula to Hyuganada of the off Kyushu, and calculated tsunami waveforms of the DONET observatories and target points of the tsunami prediction near the coast line. Our system has a tsunami database including these calculated waveforms and inundation areas for each target point, and the system extracts the worse cases from the database corresponding to input pressure data observed by DONET. The calculated tsunami waveforms of extracted models are also shown with above predicted information. The accuracy of the tsunami prediction depends on the correlation between the maximum height of the target point and the pressure value observed by DONET. And the correlation is influenced by seafloor topography and shape of the coastal line. When diffracted tsunami waves only propagate to the target point, the predicted tsunami height is relatively small. In cases with the same directions of the bay mouse and the tsunami prediction, the tsunami amplification becomes large. Therefore, it is important to select fault models from the database. We use triggering orders of the seismometers and the pressure data for the selection of models based on the source direction and keep accuracy of the prediction. Considering the simple principle for the prediction, tsunami propagation is one of important elements affecting the accuracy of the prediction. In cases with some complicated propagation routes, the tsunamis traveling different routes are stacked and possible large tsunami is locally generated. Through social implementation of the prediction system, we confirm the effectiveness around the rias coast around the Kii Peninsula, small islands area around the Shima Peninsula, and possible area expected some complicated propagation routes around the inland sea. However, oceanfloor network systems do not cover all seismogenic zones around Japan. For such areas, tentative real-time system using a buoy is effective like the DART system installed in worldwide. Around Japan with very strong current like the Kuroshio, the buoy observation system for tsunami and crustal displacement adopting slack mooring has been developed. The real-time tsunami prediction can be applied for the buoy system. With mobility of the buoy system, the deployment point for the buoy system can be selected as needed for the disaster prevention. We investigated that the tsunami height calculated using above models at a point far from a target point. The tsunami height at the estimated buoy location has strongly correlation for that of the target point.

We applied a real-time tsunami inundation forecast system to a disaster response plan. We developed a standard operating procedure (SOP) for a tsunami disaster response based on a Plan, Do, Check, Action cycle to effectively use tsunami observation and prediction information provided by a real-time tsunami inundation forecast system during an initial response to a tsunami disaster. In the Plan stage, we ran a workshop on the tsunami disaster response to confirm the current tsunami disaster response plan and develop a timeline plan for a tsunami disaster. In the Do stage, we conducted a tabletop exercise (TTX) for a tsunami disaster using a real-time tsunami prediction system. In the Check stage, we ran a workshop on an after-action review of the TTX. In the Action stage, we applied the SOPs of the real-time tsunami prediction system to the tsunami disaster management plan and conducted a second TTX. As a result, we verified the information provided by a real-time tsunami prediction system to apply the system to a tsunami disaster management plan for real municipalities. It was confirmed that the SOP that we developed allows a real-time tsunami inundation forecast system to enable government staff to safely and effectively respond during a disaster.

Beneath the Marmara Sea, Turkey, the Main Marmara Fault (MMF), the offshore part of the North Anatolian Fault (NAF), is a well-known seismic gap for future M > 7 earthquakes. However, its detailed fault geometry and microearthquake activity have been debated for several decades. Using data acquired from long-term ocean bottom seismograph (OBS) observations, we made precise hypocenter estimations based on 3-D Vp and Vs velocity structures and assessed the fault geometry beneath the western and central parts of the MMF. The results indicate a segmentation boundary between the near-vertical western part and the south-dipping eastern part located around 28.10°E. Enriched OBS locations indicate microseismicity along both the inner and outer boundary faults of the Central Basin, especially on the western side. A comparison with previously published results suggests that the seismicity pattern has not changed for at least two years, between 2014 and 2016. Using a combined dataset of this and previous studies, lateral variations in the dip angle along the MMF fault segment from 27.4°E to 28.8°E were investigated. Based on this, we depicted on-fault seismicity along the MMF and defined three inactive areas of microseismicity. Two are located in the western segment, corresponding to the rupture area of the 1912 Ms. 7.4 earthquake, and the other, the largest, is located on the eastern segment. From a comparison of previous seismic and geodetic studies, it is considered that this area is a fully locked zone and has the potential for large earthquakes. Having compared the difference between hypocenter locations determined from OBSs and land-based stations, it is proposed that the epicentral locations of the mainshock and aftershocks of the September 26, 2019, M 5.7 earthquake are located much closer to the MMF than locations reported from only land-based results.