Short-period period spectral levels of the 1948 Fukui earthquake (M_J 7.1), which was a buried strike-slip fault, and the 1945 Mikawa earthquake (M_J 6.8), which was a reverse fault with a 28 km-long surface rupture were estimated using the empirical Green's function method to match the JMA seismic intensity scales. Based on the strong-motion prediction recipe, we calculated the areas of outerfaults and the strong-motion generation areas. Then we did grid search calculation to get best location of strong-motion generation areas and best value of the short-period spectral level by changing 0.1 increments centered on the level by Dan et al.'s relation (2001). As a result, the short-period spectral levels of the Fukui and Mikawa earthquakes were estimated to be 0.7 to 0.8 times and 0.5 to 0.6 times that of by Dan et al.' relation, respectively.

The seismic source considered in seismic design have become more diverse and large-scale in recent years, and improving the accuracy of strong ground motion prediction near the seismic source of large-scale earthquakes is an important issue for future seismic design. In this study, we estimated the basement ground motions equivalent to Vs = 2100 m/s at the observation sites near the seismic source of the 2023 Turkey-Syria earthquake based on the diffuse field concept. The PGA and PGV of the estimated basement ground motions are almost within the variation of the ground motion model by Morikawa and Fujiwara (2013), and the seismic motion level is reduced compared to the surface records. The estimated basement ground motions are generally reasonable based on multiple verifications, except for certain sites where unusually large amplification of short-period seismic motion has been confirmed. The estimated basement ground motions and the attenuation characteristics of each period of the ground motion model showed good correspondence within the range of variation, indicating that the ground motion model may be able to evaluate large-scale, near the seismic source ground motion with a certain accuracy.

With the recent development of ocean bottom seismograph (OBS) networks, the use of OBS data in earthquake early warning (EEW) systems for railways has been considered. To apply existing P-wave EEW methods—originally validated using inland seismic data—to OBS data, it is essential to understand the amplification characteristics at the seafloor and the effects of seawater on the seismic motions recorded by OBSs. In this study, we investigated the influence of seawater on seismic motions at the seafloor by analysing spectral ratios between land and seafloor seismic motions. Our result show that troughs in the spectral ratios can be attributed to the transmission of P-waves into the seawater. We also found that the frequencies of these troughs vary according to the quarter-wavelength approximation, depending on the installation depth of each OBS. Furthermore, analysis of time history waveforms from OBS data revealed that the amplitude of the initial P-wave observed at the seafloor is less than twice that of the upgoing wave from the seafloor. Additionally, the phase of the P-wave at the seafloor is affected by reflections from the sea surface, and this effect varies with OBS installation depth. These factors reduce the accuracy of existing Pwave EEW methods when applied to OBS data. However, by taking into account the findings of this study regarding P-wave propagation through seawater, it will be possible to develop EEW systems that utilise P-waves of OBS data.

This paper presents a machine learning model that accurately discriminates whether an image frame contains an anomaly or not. It also demonstrates the feasibility of extracting the anomaly type from the anomaly content. We initially attempted to discriminate the 30 defined physical anomaly types contained in the earthquake images using the same discriminator (a CNN that discriminates 30 classes). However, we found that they could not be discriminated (correct response rate: 0.5%).We constructed an ensemble model combining six types of two-class discriminators, called weak classifiers, that are specialized for the type of abnormality to be discriminated. This allowed us to improve the final accuracy by comparing the five ensemble methods. The model with simply parallel weak classifiers showed the best discriminative performance, with a Hamming loss of 0.0015.

During the 2024 Noto Peninsula earthquake, significant long-period ground motions were observed not only in the Noto region of Ishikawa Prefecture near the earthquake source region, but also over a wide area in the Hokuriku region. To clarify the characteristics of these long-period ground motions, we analyzed the strong-motion records obtained from K-NET and KiK-net of the National Research Institute for Earth Science and Disaster Resilience. We found that large pseudo velocity response spectra with a peak period of 1-3 s were observed at many stations in the Noto region, while pseudo velocity response spectra observed at stations in the Echigo Plain of Niigata Prefecture and the Toyama Plain of Toyama Prefecture were comparable to those near the earthquake source region at periods of 4 s or longer. In the Echigo Plain, the pronounced long-period ground motions peaking around 8 s are inferred to result from Rayleigh waves that were generated and amplified near the source and/or during propagation, and subsequently further amplified by the thick sedimentary layers underlying the plain. Moreover, long-period ground motions in the Toyama Plain are considered to be primarily attributable to the superposition of multiple surface waves arriving from different directions and at different times, which were further amplified and temporally prolonged by the subsurface structure of the sedimentary plain.

Transfer functions from underground to surface with seismic records at vertical array observatories might sometimes be evaluated that cannot be reproduced by the conventional one-dimensional analysis. We evaluated a two-dimensional ground structure model through inversion with the observed transfer function at KiK-net station of Kamaishi, where the surrounding ground structure is complex. A stable and reliable model could be evaluated by using the results of PS logging and microtremor array exploration conducted around the station as constraint conditions. The observed transfer function was well reproduced compared with the model optimized by the conventional one-dimensional analysis. We proved the validity of the inverted model through the additional microtremor observations. Even if the observatories whose ground structure have difficulties to be modeled based on the one-dimensional analysis, we could construct the model which well reproduce the observed ground motion with consideration of the irregularity of the surrounding S-wave velocities.

In this study, we introduce the partial participation factor, an index of vibrational sensitivity at specific parts, defined as an extension of the participation factor. This evaluation methodology is applied to structures with multiple vibrational transmission paths, and we propose a technique for selecting major vibration modes based on only eigenvalue analysis results, even when information on external forces and damping is not enough. Furthermore, we present a case study of structural optimization aimed at minimizing the partial participation factor of the major vibration mode of the specific part, thereby demonstrating the practical utility of this index.

We have identified microtremor Horizontal-to-Vertical (H/V) spectral ratios at numerous grid observation points in southeastern Saitama Prefecture in an area within a rectangle with a side length of approximately 35 km; average interval of about 1 km; 853 grids. Peak frequencies were read from the obtained spectral ratios within a frequency range from 0.5 to 20 Hz, from which their spatial distribution was deduced (i.e., a peak frequency map). When multiple local peaks were observed within the analysis frequency range for reading, the amplitude difference between the peak and the corresponding trough was evaluated. By taking into account this information, we set a peak selection threshold and a weighting criterion and then selected a single peak to read the frequency. The resulting peak frequency map was compared with a pre-existing 3D geological structure model (i.e., the urban geological map) that was constructed based on numerous borehole data (∼7338 boreholes). They are in good agreement in topographical and geological distributions. For example, in areas with thick alluvial deposits ( > 20 m) in the Arakawa and Nakagawa lowlands, the peak frequency averaged a low value of 1.3 Hz, whereas in areas without alluvial deposits, the average peak frequency was much higher at 4.7 Hz. The peak frequency of the H/V spectral ratios can be considered to be an indicator for quantifying qualitative assessments of ground conditions assumed based on topography and 3D geological structure, as well as for forecasting earthquake hazards. The resulting peak frequency maps will be released on the website of the urban geological map to enable detailed comparisons with geological distributions. We have a plan to construct and release similar peak frequency maps for the other regions involved in the urban geological map in the future.