High Sensitivity Seismograph Network Research Articles

Detection of Deep Low‐Frequency Tremors From Continuous Paper Records at a Station in Southwest Japan About 50 Years Ago Based on Convolutional Neural Network

AbstractSince deep low‐frequency tremors are considered to be associated with large earthquakes that occur adjacently on the same subducting plate interface, it is important to investigate tremors that occurred before the establishment of modern seismograph networks such as the High Sensitivity Seismograph Network (Hi‐net). We propose a deep‐learning solution to detect evidence of tremors in scanned images of paper seismogram records from over 50 years ago. In this study, we fine‐tuned a convolutional neural network (CNN) based on the Residual Network, which was pre‐trained using images of synthetic waveforms from our previous study, using a data set comprised of images generated from real seismic data recorded digitally by Hi‐net to facilitate a supervised analysis. The fine‐tuned CNN was able to predict the presence or absence of tremors in the Hi‐net images with an accuracy of 98.64%. Gradient‐weighted Class Activation Mapping heatmaps created to visualize model predictions indicated that the CNN's ability to detect tremors is not degraded by the presence of teleseisms. Once validated using the Hi‐net images, the CNN was applied to paper seismograms recorded from 1966 to 1977 at the Kumano observatory in southwest Japan, operated by Earthquake Research Institute, The University of Tokyo. The CNN showed potential for detecting tremors in scanned images of paper seismogram records from the past, facilitating downstream tasks such as the creation of new tremor catalogs. However, further training using an augmented data set to control for variables such as inconsistent plotting pen thickness is required to develop a universally applicable model.

Heat flow data and thermal structure in northeastern Japan

New heat flow data corrected for climate change over Northeastern Japan were obtained using the temperature profile of the borehole of the High Sensitivity Seismograph Network (Hi-net). The obtained spatial distribution of heat flow shows low heat flow on the forearc side, high heat flow along the Ou Backbone Range, and low heat flow in the plains on the back-arc side. However, the distribution is not clearly divided into high and low heat flow along the VF front; for example, the low heat flow extends from near the northern Kitakami Mountains on the forearc side to the Ou Backbone Range crossing the VF, while the high heat flow extends to the central Kitakami Mountains and Sendai plain on the forearc side. In addition, a crustal temperature structure model was developed that considers into account the presence of sedimentary layers, the temperature dependence of thermal conductivity, and differences in heat generation due to lithology. There is a good correlation between this temperature structure and the lower limit of the seismogenic layer, which is between 400 and 450 °C. Compared to previous studies, the crustal thermal structure calculation method assumed is a model whose estimated temperature distribution is sensitive to structural differences; however, a more accurate estimation of the temperature structure is possible if detailed structural information is available. On the other hand, it seems necessary to treat fluid behavior in more detail in areas of high heat flow. However, the estimation of crustal temperature structure, especially in regions with thick sedimentary layers, is considered an improvement over the previous study.Graphical

River discharge prediction for ungauged mountainous river basins during heavy rain events based on seismic noise data

Several mountainous river basins in Japan do not have a consistent hydrological record due to their complex environment and remoteness, as discharge measurements are not economically feasible. However, understanding the flow rate of rivers during extreme events is essential for preventing flood disasters around river basins. In this study, we used the high-sensitivity seismograph network (Hi-net) of Japan to identify the time and peak discharge of heavy rain events. Hi-net seismograph stations are distributed almost uniformly at distance intervals of approximately 20 km, while being available even in mountainous regions. The Mogami River Basin in Northeastern Japan was selected as a target area to compare the seismic noise data of two Hi-net stations with the hydrological response of a nearby river. These stations are not located near hydrological stations; therefore, direct comparison of seismic noise and observed discharge was not possible. Therefore, discharge data simulated using a hydrological model were first validated with gauging station data for two previous rain events (10–23 July 2004 and 7–16 September 2015). Then, the simulated river discharge was compared with Hi-net seismic noise data for three recent events (10–23 July 2004, 7–16 September 2015, and 10–15 October 2019). The seismic noise data exhibited a similar trend to the time series of simulated discharge in a frequency range of 1–2 Hz for the selected events. Discharge values predicted from the noise data effectively replicate the simulated discharge values in many cases, especially the timing and amount of peak discharge.Simulated and predicted discharge near NIED Hi-net seismic stations in the Mogami River Basin for the event of October 2019 (Typhoon Hagibis).

Focal mechanisms and the stress field in the aftershock area of the 2018 Hokkaido Eastern Iburi earthquake (MJMA\u2009=\u20096.7)

The tectonic stress field was investigated in and around the aftershock area of the Hokkaido Eastern Iburi earthquake (MJMA = 6.7) occurred on 6 September 2018. We deployed 26 temporary seismic stations in the aftershock area for approximately 2 months and located 1785 aftershocks precisely. Among these aftershocks, 894 focal mechanism solutions were determined using the first-motion polarity of P wave from the temporary observation and the permanent seismic networks of Hokkaido University, Japan Meteorological Agency (JMA), and High Sensitivity Seismograph Network Japan (Hi-net). We found that (1) the reverse faulting and the strike-slip faulting are dominant in the aftershock area, (2) the average trend of P- and T-axes is 78° ± 33° and 352° ± 51°, respectively, and (3) the average plunge of P- and T-axes is 25° ± 16° and 44° ± 20°, respectively: the P-axis is close to be horizontal and the T-axis is more vertical than the average of the P-axes. We applied a stress inversion method to the focal mechanism solutions to estimate a stress field in the aftershock area. As a result, we found that the reverse fault type stress field is dominant in the aftershock area. An axis of the maximum principal stress (σ1) has the trend of 72° ± 7° and the dipping eastward of 19° ± 4° and an axis of the intermediate principal stress (σ2) has the trend of 131° ± 73° and the dipping southward of 10° ± 9°, indicating that both of σ1- and σ2-axes are close to be horizontal. An axis of the minimum principal stress (σ3) has the dipping westward of 67° ± 6° that is close to be vertical. The results strongly suggest that the reverse-fault-type stress field is predominant as an average over the aftershock area which is in the western boundary of the Hidaka Collision Zone. The average of the stress ratio R = (σ1 − σ2)/(σ1 − σ3) is 0.61 ± 0.13 in the whole aftershock area. Although not statistically significant, we suggest that R decreases systematically as the depth is getting deep, which is modeled by a quadratic polynomial of depth.

Improving seismic remote sensing of typhoon with a three-dimensional Earth model.

Typhoon-induced P-wave microseisms can be observed using seismological arrays and analyzed for the seismic monitoring of ocean storms. This paper presents a frequency-domain beamforming (FB) method that integrates a three-dimensional (3-D) Earth model to better capture the heterogeneities in the subsurface structure, and therefore yield more accurate ray-tracing and travel-time predictions. This method is applied to the Super Typhoon Lupit (2009) using seismological array observations from the Northeast China Extended Seismic Array (NECESSArray) and high-sensitivity seismograph network in Japan (Hi-net). The results show that the localized P-wave microseism source regions based on the 3-D model are in better agreement with the theoretical source regions and typhoon centers than those based on a conventional one-dimensional (1-D) model. The significance of using a 3-D model instead of a 1-D model in the FB method is further investigated by comparing the consistency of the localization results for the two different arrays, with the localized source regions being more mutually concordant when using the 3-D model. The results demonstrate that integrating the 3-D model into the FB method improves the accuracy of locating the typhoon-induced P-wave microseism source regions.

Automatic Aftershock Forecasting: A Test Using Real‐Time Seismicity Data in Japan

Real‐time aftershock forecasting is important to reduce seismic risks after a damaging earthquake. The main challenge is to prepare forecasts based on the data available in real time, in which many events, including large ones, are missing and large hypocenter determination errors are present due to the automatic detection process of earthquakes before operator inspection and manual compilation. Despite its practical importance, the forecast skill of aftershocks based on such real‐time data is still in a developmental stage. Here, we conduct a forecast test of large inland aftershock sequences in Japan using real‐time data from the High Sensitivity Seismograph Network (Hi‐net) automatic hypocenter catalog (Hi‐net catalog), in which earthquakes are detected and determined automatically in real time. Employing the Omori–Utsu and Gutenberg–Richter models, we find that the proposed probability forecast estimated from the Hi‐net catalog outperforms the generic model with fixed parameter values for the standard aftershock activity in Japan. Therefore, the real‐time aftershock data from the Hi‐net catalog can be effectively used to tailor forecast models to a target aftershock sequence. We also find that the probability forecast based on the Hi‐net catalog is comparable in performance to the one based on the latest version of the manually compiled hypocenter catalog of the Japan Meteorological Agency when forecasting large aftershocks with M >3.95, despite the apparent inferiority of the automatically determined Hi‐net catalog. These results demonstrate the practical usefulness of our forecasting procedure and the Hi‐net automatic catalog for real‐time aftershock forecasting in Japan. Online Material: Figures and tables showing detailed forecast results for all considered aftershock sequences and all forecast time frames.

Automatic identification of fake patterns caused by short-width wavelets in seismic data

Despite the popularity of using the Haar wavelet filter in many applications, it sometimes introduces fake patterns into the multi resolution analysis (MRA) of seismic data. In this work, we compared different wavelet filters to demonstrate that these patterns are fake and not part of the original waveforms and to show that they are a result of using the Haar wavelet filter as a short-width wavelet. To achieve this, many seismic waveforms from two different sources: the Egyptian National Seismic Network (ENSN) and the High Sensitivity Seismograph Network Japan (Hi-net) are used with different wavelet filters. We propose an algorithm based on an autoregressive (AR) model to detect these patterns automatically and fully.

Complex inner core boundary from frequency characteristics of the reflection coefficients of PKiKP waves observed by Hi-net

Frequency-dependent reflection coefficients of P waves at the inner core boundary (ICB) are estimated from the spectral ratios of PKiKP and PcP waves observed by the high-sensitivity seismograph network (Hi-net) in Japan. The corresponding PKiKP reflection locations at the ICB are distributed beneath the western Pacific. At frequencies where noise levels are sufficiently low, spectra of reflection coefficients show four distinct sets of characteristics: a flat spectrum, a spectrum with a significant spectral hole at approximately 1 or 3 Hz, a spectrum with a strong peak at approximately 2 or 3 Hz, and a spectrum containing both a sharp peak and a significant hole. The variety in observed spectra suggests complex lateral variations in ICB properties. To explain the measured differences in frequency characteristics of ICB reflection coefficients, we conduct 2D finite difference simulations of seismic wavefields near the ICB. The models tested in our simulations include a liquid layer and a solid layer above the ICB, as well as sinusoidal and spike-shaped ICB topography with varying heights and scale lengths. We find that the existence of a layer above the ICB can be excluded as a possible explanation for the observed spectra. Furthermore, we find that an ICB topographic model with wavelengths and heights of several kilometers is too extreme to explain our measurements. However, restricting the ICB topography to wavelengths and heights of 1.0–1.5 km can explain the observed frequency-related phenomena. The existence of laterally varying topography may be a sign of lateral variations in inner core solidification.

Monitoring the instrument response of the high-sensitivity seismograph network in Japan (Hi-net): effects of response changes on seismic interferometry analysis

More than 10 years have passed since observations began to be recorded by Hi-net, a network of high-sensitivity seismometers located in Japan. Several large earthquakes, including the 2011 Tohoku-Oki earthquake, have been recorded by the network during this period. Age-related degradation and the strong ground motion of large earthquakes may change the instrument response of the high-sensitivity seismometers of Hi-net. Thus, we checked the natural frequency f and damping constant h for each Hi-net sensor and monitored the instrument response for 10 years from 2003 to 2013. Most of the sensors showed a stable instrument response over this period. More than 95 % of the sensors whose responses we could well estimate showed small fluctuations in their natural frequencies and damping constants of within 0.05 Hz and 0.05, respectively. We also found that many Hi-net sensors in northeastern Japan showed slight changes in the instrument response as a result of the 2011 Tohoku-Oki earthquake. Based on the assumption that the instrument responses remained unchanged, the fractional velocity reduction in the subsurface structure was reported by seismic interferometry analysis. To investigate how changes in the instrument response can cause errors in seismic interferometry analysis, we conducted a synthetic test. The results indicate that the instrument response did not result in systematic variation in the time delay observed in the interferometry analysis. This confirmed that the velocity decrease observed as a result of the 2011 Tohoku-Oki earthquake was not due to artificial instrument error.

Scattering of teleseismic P-waves by the Japan Trench: A significant effect of reverberation in the seawater column

We detected a scattered wave train in data from the high-sensitivity seismograph network in Japan (Hi-net) following the arrival of the near-vertically incident P-wave generated by the 2009 earthquake (Mw 7.8) off the South Island of New Zealand. The scattered wave train represented predominantly vertical ground motion at a period of 20 to 50 s and with an apparent velocity of 3.5 km/s; it propagated cylindrically westward through the Kanto area of central Japan. Array analysis showed that the scattered wave train developed beneath the Pacific Ocean near the Boso triple junction, southeast of the Kanto area. A 3D finite-difference simulation of seismic wave propagation using a high-resolution model incorporating subsurface structure, topography, and bathymetry revealed that the strong scattered waves that were generated along the Japan Trench and propagated normal to the trench axis represented multiple reverberations of seismic waves between the seafloor and the Pacific plate boundary. In addition, strong reverberation of acoustic waves in the seawater column above the Boso triple junction causes elongated scattered waves, which reasonably explains our observations.

Long-period seismic amplification in the Kanto Basin from the ambient seismic field

Tokyo, like many seismically threatened cities, is situated atop a sedimentary basin that has the potential to trap and amplify seismic waves from earthquakes. We study amplification in the Kanto Basin by exploiting the information carried by the ambient seismic field. We use 375 seismic stations from the high sensitivity seismograph network across central Honshu as virtual sources and 296 seismic stations of the Metropolitan Seismic Observation network shallow borehole seismometers within the basin as receivers to map the basin response. We find a linear relationship between ground motion and basin depth at periods of 2–10 s that could be used to represent 3-D basin effects in ground motion prediction equations. We also find that the strength of basin seismic amplification depends strongly on the direction of illumination by seismic waves.

Two-dimensional thermal modeling of subduction of the Philippine Sea plate beneath southwest Japan

In this study, we constructed a two-dimensional thermal convection model associated with subduction of the Philippine Sea (PHS) plate beneath southwest Japan to estimate the thermal state. To evaluate the validity of the calculated temperature distribution, we compared the calculated heat flows with observed heat flow data such as the High Sensitivity Seismograph Network Japan (Hi-net) borehole, land borehole, marine probe, and bottom-simulating reflectors. Hi-net heat flow data, which have never been used for thermal modeling, have spatially high resolutions on land and enable detailed estimation of the thermal state beneath the Japanese islands. We considered the spatio-temporal change in the age of the PHS plate, change in its plate motion direction at 3 Ma, and the geometry of its upper surface. Calculated heat flows associated solely with subduction of the PHS plate, passing through central and eastern Shikoku and the Kii Peninsula, were not consistent with observations. They tended to be lower on the seaward side of the corner of the mantle wedge and to have much longer spatial wavelengths on the continental side. To explain heat flow data in southwest Japan, frictional heating at the plate interface on the seaward side of the corner of the mantle wedge and temperature changes due to surface erosion and sedimentation on land associated with crustal deformation during the Quaternary must be incorporated into the model. The most suitable pore pressure ratio to explain the heat flow data is estimated to be 0.95. Temperatures on the upper surface of the PHS plate obtained in this study are lower at shallow depths and higher at greater depths than those obtained by Hyndman et al. (1995) , indicating that the seismogenic zone inferred from this study is narrower than that of their model. Microearthquakes within the subducting PHS plate occur at temperatures below 700–800 °C. • By developing a source code, we calculated the temperature field in southwest Japan. • We used current dense heat flow stations to evaluate thermal structure. • We found that erosion, sedimentation, and frictional heating were required. • The seismogenic zone is narrower than that obtained by Hyndman et al. (1995) . • Intraslab microearthquakes within the PHS plate take place below 700–800 °C.

Scattering of high-frequency P wavefield derived by dense Hi-net array observations in Japan and computer simulations of seismic wave propagations

SUMMARY We studied the scattering properties of high-frequency seismic waves due to the distribution of small-scale velocity fluctuations in the crust and upper mantle beneath Japan based on an analysis of three-component short-period seismograms and comparison with finite difference method (FDM) simulation of seismic wave propagation using various stochastic random velocity fluctuation models. Using a large number of dense High-Sensitivity Seismograph network waveform data of 310 shallow crustal earthquakes, we examined the P-wave energy partition of transverse component (PEPT), which is caused by scattering of the seismic wave in heterogeneous structure, as a function of frequency and hypocentral distances. At distance of less than D = 150km, the PEPT increases with increasing frequency and is approximately constant in the range of from D = 50 to 150 km. The PEPT was found to increase suddenly at a distance of over D = 150 km and was larger in the high-frequency band (f > 4 Hz). Therefore, strong scattering of P wave may occur around the propagation path (upper crust, lower crust andaroundMohodiscontinuity)oftheP-wavefirstarrivalphaseatdistancesoflargerthanD = 150km.WealsofoundaregionaldifferenceinthePEPT value,wherebythePEPT valueislarge at the backarc side of northeastern Japan compared with southwestern Japan and the forearc side of northeastern Japan. These PEPT results, which were derived from shallow earthquakes, indicate that the shallow structure of heterogeneity at the backarc side of northeastern Japan is stronger and more complex compared with other areas. These hypotheses, that is, the depth andregionalchangeofsmall-scalevelocityfluctuations,areexaminedby3-DFDMsimulation using various heterogeneous structure models. By comparing the observed feature of the PEPT with simulation results, we found that strong seismic wave scattering occurs in the lower crust due to relatively higher velocity and stronger heterogeneities compared with that in the upper crust. To explain the observed regional difference, the velocity fluctuation model with 3‐4 per cent stronger fluctuation and smaller κ is required at the backarc side of northeastern Japan.

Seismic reflector alignment along the volcanic front in northeastern Honshu, Japan

We found seismic reflectors aligned along the east side of the volcanic front in northeastern Honshu, Japan, by analyzing seismograms of intermediate‐depth earthquakes recorded at High Sensitivity Seismograph Network Japan stations. We identified an unusual high‐frequency later phase on the seismograms, which arrives after the direct S waves and continues for >5 s. The characteristics of the waveforms suggest that this later phase is an S wave incidence. We plotted envelope waveforms and identified later phase arrivals on envelope waveforms. To estimate the location of the source of the later phases, we defined an event group that consisted of six to eight events distributed in a region of ∼50 km3. The source of the later phases was located by a combination of a station and an event group using later phase travel times. If we assume that later phases arriving at a station from an event group are generated by a common source, then we can use the observed travel times in a grid search to obtain the source location. The sources of the later phase were found to be aligned along the east side of a volcanic front at a depth of 20–30 km. The averaged ratio of the amplitudes of the later phases to the amplitudes of the incident waves on the later phase source was estimated to be ∼0.6. This result suggests the existence of unusual reflectors spread in the lower crust beneath the fore‐arc region. It may be possible that the later phase sources are steeply inclined bodies filled with fluid.

防災科研 Hi-net の構築とその成果

After the 1995 Hyogoken-Nanbu earthquake, National Research Institute for Earth Science and Disaster Prevention (NIED) has constructed new nation-wide high sensitivity seismograph network. We call this network “NIED Hi-net” or “Hi-net”. Hi-net stations uniformly cover the Japanese Islands in addition to old high sensitivity seismographs. To reduce the seismic noise near the ground surface, a three-component velocity-type seismograph is installed at the bottom of a borehole. Recorded signal at each station is continuously transmitted to NIED data management center (DMC) at Tsukuba using IP-VPN technology. Hi-net waveform data are sent to the JMA and other related institutes for seismic activity monitoring and seismological research. NIED archives all waveform data generated by the Hi-net and other seismic networks. These waveform data and JMA hypocenter catalog are open to the public via the Hi-net website. Although less than 10 years has passed since the Hi-net data became available, this network has brought us a lot of contributions to geoscience. Stable and long-term observation by this seismograph network may play important roles to mitigate earthquake disasters and to predict earthquakes.

Quantifying early aftershock activity of the 2004 mid‐Niigata Prefecture earthquake (Mw6.6)

We analyze the early aftershock activity of the 2004 mid‐Niigata earthquake, using both earthquake catalog data and continuous waveform recordings. The frequency‐magnitude distribution analysis of the Japan Meteorological Agency (JMA) catalog shows that the magnitude of completeness of the aftershocks changes from values around 5.0, immediately after the main shock, to about 1.8, 12 hours later. Such a large incompleteness of early events can bias significantly the estimation of aftershock rates. To better determine the temporal pattern of aftershocks in the first minutes after the Niigata earthquake, we analyze the continuous seismograms recorded at six High Sensitivity Seismograph Network (Hi‐Net) stations located close to the aftershock distribution. Clear aftershocks can be seen from about 35 s after the main shock. We estimate that the events we picked on the waveforms recorded at two seismic stations (NGOH and YNTH) situated on opposite sides of the aftershock distribution are complete above a threshold magnitude of 3.4. The c value determined by taking these events into account is about 0.003 days (4.3 min). Statistical tests demonstrate that a small, but nonzero, c value is a reliable result. We also analyze the decay with time of the moment release rates of the aftershocks in the JMA catalog, since these rates should be much less influenced by the missing small events. The moment rates follow a power law time dependence from few minutes to months after the main shock. We finally show that the rate‐and‐state dependent friction law or stress corrosion could explain well our findings.

An improved double-difference earthquake location algorithm using s P phases: application to the foreshock and aftershock sequences of the 2004 earthquake offshore of the Kii peninsula, Japan (Mw=7.5)

AbstractWe have useds Pphases to improve the hypocentral locations of the earthquakes that occurred offshore southeast of the Kii peninsula in association with the Mw=7.5 mainshock on September 5, 2004. The earthquakes were more than 100 km from the onshore seismic network and, thus, their focal depths were poorly constrained. Thes Pphases were recorded about 7–11 s after the initialPphases, with both phases having almost the same apparent velocities. The computation of ray-paths revealed that the arrival times of the laters Pphases are sensitive to focal depths. We have recalculated the hypocenters and origin times for 36 events with more than sixs Pphase identifications on seismograms recorded at the High Sensitivity Seismograph Network by includings Pphases in a double-difference earthquake location algorithm, which eliminates the errors introduced by crustal heterogeneity. The relocation results were then compared with those from the Japan Meteorological Agency based on traditional absolute location techniques.

A densely distributed high-sensitivity seismograph network in Japan:Hi-net by National Research Institute for Earth Science and DisasterPrevention

Seismic observations to retrieve various information from the Earth are the basis of seismology. A seismic observation system requires various technologies for vibration sensors, analog-and-digital measurement, data transmission, and computing for mass data analysis, for example. New developments in technology are adopted whenever possible in the construction of seismic observation systems. In Japan, after the disastrous Kobe Earthquake in 1995, a high-density and high-sensitivity seismograph network was constructed. The seismic network, called the National Research Institute for Earth Science and Disaster Prevention (NIED) Hi-net, uniformly covers the Japanese Islands with a spacing of 20–30km. As a result, the detection capability for microearthquakes has been greatly improved, and various research using Hi-net data has indicated that this seismic network has a great potential to resolve the underground structure and various geophysical phenomena as a radar-array oriented toward the Earth. Equipped with modern standard techniques, the Hi-net system is designed to be able to be smoothly upgradeable in the future.

Recent progress of seismic observation networks in Japan —Hi-net, F-net, K-NET and KiK-net—

After the disastrous 1995 Kobe earthquake, a new national project has started to drastically improve seismic observation system in Japan. A large number of strong-motion, high-sensitivity, and broadband seismographs were installed to construct dense and uniform networks covering the whole of Japan. The new high-sensitivity seismo-graph network consisting of 696 stations is called Hi-net, while the broadband seismograph network consisting of 71 stations is called F-net. At most of Hi-net stations strong-motion seismographs are also equipped both at depth and the ground surface. The network of these 659 stations with an uphole/downhole pair of strong-motion seismographs is called KiK-net, while another network consisting of 1034 strong-motion seismographs installed at the ground surface is called K-NET. Here, all the station numbers are as of April 2003. High-sensitivity data from Hi-net and pre-existing seismic networks operated by various institutions have been transmitted to and processed by the Japan Meteorological Agency since October 1997 to monitor the seismic activity in and around Japan. The same data are shared to university group in real time using satellite communication for their research work. The data are also archived at the National Research Institute for Earth Science and Disaster Prevention and stored in their database system for public use under a fully open policy.