Mt. Usu is an active volcano located within the Toya Caldera in Hokkaido. The volcano has erupted about every 20-30 years since the 20th century. The location and type of eruptive activity vary with each eruption. Understanding the eruptive activity requires an understanding of its subsurface structure, which remains a key challenge. We conducted a broadband and audio-frequency band magnetotelluric survey to model the 3-D resistivity structure of Mt. Usu. Additionally, drilling surveys around Mt Usu were conducted to a depth of approximately 1 km in the area affected by the 2000 eruption. In this study, we conducted a 3-D resistivity inversion using existing MT data to investigate the subsurface structure and compared the 3-D resistivity structure with the drilling data. We carried out a 3-D inversion based on the FEMTIC code (Usui 2015) to obtain a 3-D resistivity model. The input data were given an error floor of 5 % of the absolute value for the four impedance components and 0.03 for the two tipper components. We started from the initial model with a uniform resistivity at 100 Ωm (model calculation area: 600 km(NS)×600 km(EW)×780 km(vertical)). The atmosphere and sea water were fixed at 108 Ωm and 0.3 Ωm, respectively. We compared the resistivity structure with the existing drilling data. The geological structure estimated from the drilling survey does not correspond to the 3-D resistivity structure. However, the conductive layer corresponds to the drilling core, which is rich in smectite. The drilling core is also classified as a smectite alteration zone. Therefore, it is highly likely that the conductive layer extending throughout Mt. Usu reflects layers that have undergone diagenetic or hydrothermal alteration. Furthermore, the drilling survey revealed an andesite dyke, estimated to have intruded during the Pleistocene, at approximately 0.3 km BSL. On the other hand, the upper end of the dyke estimated from GPS data during the 2000 eruption, was located about 0.5 km (Okazaki et al. 2002). Both the dyke identified in the drilling survey and the dyke inferred from GPS data correspond to the depth of the conductive layer. The 3-D resistivity structure obtained in this study corresponds the alteration zone and volcanic activity sources, as inferred from the drilling survey and geophysical data. In the future, we expect to gain further insights into the mechanisms of volcanic activity by investigating the relationship between the resistivity structure and volcanic activities such as dyke intrusion and earthquake swarms.

Continuous observation of the total magnetic intensity (F) was started by Kakioka Magnetic Observatory at a station (MEA) on the southern slope of the 96-1 Pommachineshiri Crator of Meakandake Volcano in 2003. Since 2014, when an array was formed with the addition of two stations (ME2 and ME3), array observation with the three F stations, apart by some 100 m from one another, has been continuously made. I use the Extended Kalman Filter method to infer the time evolution of demagnetization source described by four parameters (horizontal position, altitude and intensity of the dipole moment) such as to explain monthly variations of the three F data. This has been enabled by taking a priori information on dipole demagnetization sources from the repeat measurement surveys as an initial model. Although the obtained models depend highly on prior setting of the observation and transition errors, I find that the resolution of the time variation of the moment intensity (including the notable increase of intermittent demagnetization) is generally high. On the other hand, the resolution of the elevation (about 500 m depth) is low, limiting the ability of clearly imaging upward migration of the heat source. In order to obtain a model more reliable for the volcanic activity assessment, it is desirable to add further information to constrain parameters, or to reconsider the station layout for enhancing the resolution of the depth variation of the heat source.

　This study shows the first result of the deep subsurface structure beneath Izu-Oshima by using the data from the joint land and marine MT surveys around Izu-Oshima conducted by Earthquake Research Institute, the University of Tokyo and Japan Agency for Marine-Earth Science and Technology in 2021-2022. The land data were obtained at 11 sites on Izu-Oshima for about 40 days, and the marine data were obtained at 7 sites on the seafloor around Izu-Oshima for about 4.5 months. Time series analysis converts the measurement data to MT response functions and geomagnetic transfer functions in 384 ~ 4.88 x 10-4 Hz frequency range for the land data, and in 3.75 x 10-2 Hz ~ 9.77 x 10-5 Hz frequency range for the marine data. Finally, the 3-D inversion revealed the resistivity structure beneath the Izu-Oshima.

Significant geomagnetic seasonal variations are often observed at magnetic stations on volcano and are known to be noisy in volcanic activity monitoring. It has been pointed out that this seasonal variation is an effect of changes in the magnetization strength of rocks due to temperature changes in the surface layer, and the surface layer temperature can be used to correct for the seasonal variations in geomagnetic total intensity. JMA Kakioka Magnetic Observatory has operated magnetic stations (MIK1 and MIK2) on the north side of the crater of Mt. Miharayama at close intervals of about 40 m for the purpose of contributing to volcanic disaster prevention on Izu-Oshima Island since 2007, and then seasonal variations in geomagnetic total intensity that are not very similar have been observed at both stations. On the other hand, it is pointed out that seasonal variations in geomagnetic total intensity at Izu-Oshima Island include variations that cannot be reconciled with changes in ground temperature measured near the stations, and that one of the causes of these variations in geomagnetic total intensity is rain water permeating into porous scoria. However, the causal relationship between precipitation and geomagnetic total intensity was not clarified, and the role of rain water on the geomagnetic total intensity variation was still unknown. Vertical subsurface temperature profiles of basalt scoria layer estimated from a heat conduction model using hourly temperature and precipitation from AMeDAS show the effects of surface permeation from summer rainfall and winter thermal convection from cold air sinking in the scoria layer. Seasonal variation in geomagnetic total intensity and its correction calculated from the subsurface temperature of the scoria layer by the least-squares method are in good agreement. Factorial components of the corrections show that MIK1 is significantly influenced by the subsurface temperature derived from the rain water and the thermal convection, while MIK2 is significantly influenced by the seasonal variation and the subsurface temperature derived from the rain water. The temperature dependence on MIK1 and MIK2 is thought to be due to the magnetic distribution in the scoria near each station, and the effects of water permeation and thermal convection on rock magnetism. Although the corrected geomagnetic total intensity observations from 2021 to 2024 show a slight increasing trend, this is not due to a clear volcanic activity since there are no signs of an eruption at Izu-Oshima Island. This research revealed that the geomagnetic total intensity variation shows a very pronounced dependence on subsurface temperature at magnetic stations on the north side of the crater of Mt. Miharayama.

This study aims to understand the current magma supply system leading to giant caldera eruptions. The Kikai submarine caldera volcano is located on the Southwest Japan Volcanic Front and lies within southern Kagoshima Prefecture. This volcano is known for its 7.3 ka giant caldera-forming eruption, the most recent giant caldera eruption in Japan. Topographic and petrological studies suggest that a new magma supply led to the formation of the central lava dome even after the giant caldera eruption, which has been constrained to have occurred after 3.9 ka. While geological studies provide insights into past magmatic activity, the present state of magma supply can be constrained by investigating the current structure beneath the caldera volcano using geophysical methods. A seismic structure provided by seismic tomography has identified a low-velocity zone within the mantle wedge beneath this region. In this study, we present the resistivity structure obtained from an MT survey, which provides an additional constraint on the current structure beneath the caldera volcano independent of seismic velocity data. We estimated the MT (magnetotelluric) response function using the BIRRP from the data acquired by Ocean Bottom Electro-Magnetometers at 32 sites. The magnetic data from two land-based stations in Kagoshima Prefecture (Kanoya and Haraigawa) was used as remote reference data. The coordinate system was defined such that the x-axis is oriented parallel to the trench axis. We examined the power spectral density (PSD) of the electric field and found that the PSD in the x-direction was smaller than the y-direction. As a result, the MT response function exhibited the following features: the apparent resistivities in the xx- and xy-components tended to be generally low, and the associated uncertainties tended to be large. These characteristics were observed consistently across all survey sites, rather than being limited to specific areas. We estimated the three-dimensional resistivity structure beneath the seafloor in the Kikai Caldera area using the 3-D MT inversion code FEMTIC. The model domain covered approximately 2700 x 2700 x 2000 km, and a non-conforming deforming hexahedral mesh was utilized to easily and appropriately handle the effect of the seafloor topography. The initial model was based on a 1-D structure estimated from the average MT response function obtained at the sites outside the caldera rim and incorporates the Ryukyu slab, referring to previous research on the regional tectonic setting. Data selection was performed using statistical quality criteria and visual inspection. To assess the robustness of the inferred resistivity structure, we conducted inversions with several different values of the trade-off parameters and compared the resulting resistivity models. As a result, all models corresponding to trade-off parameters near the corner of the L-curve consistently showed two distinct structures: a conductive anomaly (C1) within the mantle wedge and an overlying resistive layer (R1). These resistivity features showed minimal lateral variation along the x-axis (trench-parallel direction). Further interpretations will involve comparisons with seismicity distributions and velocity structures in this region.

We have carried out several campaigns of geophysical explorations in the Goshogake geothermal area, Akita prefecture, northeast of Japan. GPR surveys was conducted five times along the part of walk-trail since 2017. We can find the signals of scattering of the electromagnetic wave around the alteration, although we cannot recognize the outstanding time change of the structure of reflection or scattering. SP measurements was conducted nine times since 2019 in the season of no snow. A calculated source point of virtual electric source is located around the active fumarole zone.

Phreatic eruptions are primarily driven by vapor layers, making the detection of changes in these layers essential for volcanic disaster prevention. Inferno Crater Lake in New Zealand, characterized by its periodic 38-day fluctuations in water level and temperature, is hypothesized to experience vapor layer variations that contribute to these phenomena. To investigate this, a six-month observation campaign was conducted in 2023 using the EM ACROSS method, a geophysical technique sensitive to high-resistivity layers. This method involved continuous transmission of artificial electromagnetic signals, allowing precise monitoring of subsurface resistivity structures. By focusing on frequencies near the transmission frequency, errors in electric field and current measurements were evaluated, enabling observations with a time resolution of one hour. Variations in the amplitude and phase of the apparent resistivity tensor were found to correlate strongly with fluctuations in the lake's water level. However, significant phase variations were not observed below 46.95 Hz, and the phase tensor was undetectable in this frequency range. Resistivity fluctuations at these lower frequencies were attributed to changes at depths of approximately 300 m, suggesting that the sensitivity of the method decreases with depth. To further interpret the observed resistivity changes, a 3-D finite element method was employed to model the subsurface resistivity structure. The results indicate that a vapor layer expanding to a thickness of 180–240 m and rising to 60 m below the surface during high water levels provided the best explanation for the observed phase tensor variations. This finding aligns with previous resistivity surveys that identified a resistivity-altered zone near the lake, although the EM-ACROSS method demonstrated greater sensitivity to deeper regions. These results highlight the potential of the EM-ACROSS method as a highly sensitive tool for monitoring vapor layer dynamics, which are critical to understanding and forecasting phreatic eruption processes. The method’s ability i to provide high-resolution temporal and spatial data makes it particularly valuable for observing phreatomagmatic systems, offering new insights into subsurface resistivity changes and their relationship with surface-level phenomena. Future applications of this method could significantly enhance volcanic monitoring efforts and improve predictive capabilities for eruption-related hazards.

Continuous acquisition of geophysical data and the application of time-lapse imaging are crucial for subsurface monitoring. Standard methods, such as parallel and sequential inversion, are widely used to analyze subsurface fluid dynamics in volcanic, geothermal, and resource exploration areas. However, these techniques are highly sensitive to observational layout variations, often producing apparent temporal changes unrelated to actual subsurface dynamics. To overcome these challenges, we present a novel time-lapse imaging method incorporating structural coupling via Group Lasso, a sparse regularization technique. By imposing temporal constraints, our method can enhance model robustness against observational inconsistencies, suppressing apparent temporal variations. Model calculations demonstrate the superior performance of our approach over conventional methods, improving subsurface monitoring accuracy. Applications to repeated surveys and continuous monitoring data further underscore its potential for high-temporal-resolution geophysical analysis, advancing reliable interpretations of subsurface dynamics.

The project “Seismology TowArd Research innovation with data of Earthquake” (STAR-E Project), led by the Ministry of Education, Culture, Sports, Science and Technology Japan (MEXT), has accelerated developments and applications of artificial intelligence (AI) techniques in Japanese seismological community. Integration of seismology and AI, such as detection of earthquakes/slow earthquakes from seismic waveforms, simulation mitigation, and modeling, is going on rapidly and competitively in the world. Among various topics carried out in the SYNTHA-Seis, which is one of the five research subjects in the STAR-E Project, this paper introduces a neural operator to represent seismic wavefield propagation and a data-driven method to acquire a stochastic differential equation to represent low-frequency tremor waveforms based on deep learning.

Some of the catastrophic earthquakes are accompanied by rich Slow Slip Events (SSEs) activities (Kato et al., 2012; Obara and Kato, 2016). This suggests that SSEs are likely closely related to the nucleation process of large earthquakes. However, the triggering mechanism for SSEs remains largely unclear. Many SSEs sources are in a deep area greater than 30 km (Todd et al., 2018), which significantly increases the difficulty of accurately detecting the structures and their temporal changes in the SSEs source regions. While SSEs were frequently triggered in the Northern Hikurangi Subduction Zone at a shallower depth of 2-15 km (Wallace, 2020), it is an ideal area on the earth to investigate the triggering mechanism for SSEs. As increasing evidence suggests that fluid movement at depth may play a critical role in triggering the SSEs, magnetotelluric (MT) data may be one of the best choices for monitoring fluid migration before, during, and after the SSEs. By collaborating with many scientists from GNS Science, the authors initiated a seafloor MT monitoring experiment on the upper plate of the Northern Hikurangi Subduction Zone in 2023 after modifying ocean bottom electromagnetometer (OBEM) to enable continuous recording on the seafloor for at least one year. This instrument improvement has made long-term seafloor MT monitoring happen for the first time. One-year seafloor MT monitoring data was recovered in October 2024 from three sites above previous SSEs sources in the Northern Hikurangi Subduction Zone.

Minute-means geomagnetic observatories capture short-period geomagnetic variations. The tippers' response, by analyzing the relationship between vertical and horizontal magnetic-field variations, provides insights into the heterogeneous electrical conductivity structure of the crust and upper mantle. A long-time series of minute-means data from 61 geomagnetic observatories (approximately 75–135°E, 15–55°N) in China was collected and analyzed. The estimated tippers were derived for sixteen periods, ranging from 240.48s to 9078.12s. We investigate a cost-effective 3-D resistivity modeling that explicitly incorporates topography, bathymetry, and shoreline data from the ETOPO Global Relief Model to enhance the accuracy of electrical structure recovery beneath China by future inversion analysis. While modeling using a finer mesh is expected to yield more accurate model responses, it also increases model complexity and computational cost. To balance the grid resolution and computational resources, we apply the FEMTIC package (Usui, 2015), which utilizes a non-conforming deformed hexahedral mesh, to construct the model with a nested mesh consisting of a regional mesh and local mesh (cuboids). By comparing the geomagnetic observatory tippers’ responses for different meshes from ETOPO1 data, we found that the local mesh around geomagnetic observatories plays a crucial role. Provided that the local mesh resolution is sufficiently high, the precision of the regional mesh has minimal impact on the results. Based on these findings, we designed a practical mesh configuration for accurately modeling topography and coastlines. The study area, covering 5000 × 5600 km², is represented using a regional mesh with 50 × 56 equal-sized grids (each 100 × 100 km²). Around each geomagnetic observatory, a local mesh of 200 × 200 km² is applied, refined to 3.125 × 3.125 km² at the center using unequal-sized grids. This forward modeling results for geomagnetic observatory tippers’ responses are validated using two test models—one with and without topography. Coastal geomagnetic observatory tippers are primarily influenced by the bathymetry effect, while some inland geomagnetic observatory tippers are affected beyond the typical observational errors by undulating surface topography. The verification shows the necessity of incorporating topography into the modeling process. To quantitatively evaluate the accuracy of tippers calculated from resistivity models that include topography, we employed a simple method based on the arbitrary selection of horizontal coordinate systems in 3D topography over 100 Ohm-m half-space. This method calculates the mean and standard deviation of tipper responses at each geomagnetic observatory, for each frequency and component, by randomly rotating the forward model across ten different azimuthal coordinate systems, including the conventional (XN, YE) coordinate system. Our results indicate that the accuracy of forward modeling is negatively correlated with the roughness of local topography. To account for this, we used high-resolution ETOPO data and locally refined meshes to better capture topographic features near each geomagnetic observatory. For observatories located in regions with complex topography, we employed a mesh based on ETOPO 2022 (30 Arc-Second) data, refined to 0.7812 × 0.7812 km² at the center. For observatories in less complex terrain, a coarser central mesh of 3.125 × 3.125 km² was used. Ultimately, the finalized mesh design for the entire study area consists of a regional grid of 50 × 56 cells (each 100 × 100 km²), combined with local meshes of 200 × 200 km² refined to either 3.125 × 3.125 km² or 0.7812 × 0.7812 km², depending on the surrounding topography. In the future inversion analysis, we will use this local mesh design. Also, we are planning to incorporate the forward modeling uncertainty into the inversion analysis in order not to overfit the data beyond the forward modeling accuracy.

Hakodate-heiya-seien fault zone is an active fault zone approximately 26 km in length, extending from the western margin of the Hakodate Plain to the western shore of the Hakodate Bay. The distribution and characteristics of active faults in the Hakodate Plane have been estimated through various geophysical surveys such as gravity, electrical resistivity, and seismic reflection. In the Hakodate Bay, however, geophysical surveys have not been sufficiently conducted except for acoustic survey. In this study, we attempted to obtain magnetic anomalies caused by active faults by conducting marine magnetic and bathymetric survey in the offshore extension of the Hakodate-heiya-seien fault zone. As a result, we were able to detect the magnetic anomaly that is considered to be caused by the active fault. In addition, we were able to estimate the morphology of the active fault distributed in the study area by comparing the magnetic anomalies calculated by the two-dimensional dike model and the step model with the observed anomalies. These results indicate that marine magnetic survey is effective for the investigation of active faults in coastal waters.

Geophysical methods have been used in the characterization and mapping of metallic ore. This study employed magnetic and Electrical resistivity tomography (ERT) survey to characterize nickel ore in Gwandi Prospect, Tanzania. The high-resolution airborne data was used to generate a target for ground survey. The interpreted 2D images of the results provide detailed information on the magnetic and resistivity properties of subsurface geology. The ground magnetic data reveal the high magnetic body at the center, interpreted to be associated with mafic or ultramafic rocks. The mineralized zone was marked based on magnetic intensity range from negative to positive anomaly and low to high resistivity. The integrated results suggest that the nickel ore zone is characterized by low magnetic and resistivity properties. The depth was estimated from the ERT section, and the mineralized zone was found to be located at a shallow depth from the surface. The results were tested and compared with the results from geological and geochemical analysis to validate the presence of Nickel mineralization in the study area. The geological, geochemical, and geophysical results were integrated to delineate the mineralized zone, structure trend, and depth of occurrence at Gwandi prospect.

This study was conducted at Tsunozuka Kofun, located in the Isawa district of Oshu City, Iwate Prefecture, using electrical prospecting and ground-penetrating radar. Tsunotsuka Kofun is the northernmost forward-rear circular mound in Honshu, the oldest mound in Iwate Prefecture, and the largest mound with an existing mound, located in the center of the Mizusawa terrace on the Isawa alluvial fan. In 1974 and 1975, a survey was conducted to confirm the extent of the mound, and a large number of haniwa (clay figurines) and boulders the size of a human head were confirmed. In 1998, 1999, and 2000, surveys were conducted mainly in the posterior circle, and what appeared to be a row of haniwa terra-cotta tombs lined up without gaps in the posterior circle was confirmed. Ground-penetrating radar surveys were conducted on August 20 and 22, November 10 and 30, 2022, and confirmed what appeared to be burial facilities and thatched stones. In this study, the survey was conducted for a total of three days on July 11, September 20, and December 9, 2024. Previous ground-penetrating radar survey could not confirm shallow areas in detail due to insufficient resolution. Therefore, we thought it would be possible to detect the remains in the shallower mound area of the posterior circle in this research. Ground-penetrating radar frequencies of 400 MHz and 900 MHz were used. An electrical survey was also conducted to view the entire posterior circle. As a result, we were able to confirm a number of thatched stones and to show their exact locations, and thus demonstrated the effectiveness of the electrical survey and GPR in investigating the remains.