Marine earthquakes are extremely dangerous. The Qiongdongnan (QDN) segment of the Continental Slope Fault Zone (CSFZ) in the northern South China Sea poses seismic risks that may trigger submarine landslides and tsunamis. Here, using the curved-grid finite-difference method (CGFDM), we established a dynamic spontaneous rupture model for this region southeast of Hainan Island, China. We simulated the wave propagation and strong ground motion resulting from these earthquakes and produced seismic intensity distribution maps. The maximum magnitude achieved across all models was MW7.7. A left-lateral strike-slip fault with a dip angle of 59.5° was used in the simulations, and 26 hypocenters at various positions and depths were selected. We further investigated the seismic waves and strong ground motions generated by these events. The results indicated that the velocity structure had a significant influence on the maximum slip concentration on the fault. Additionally, under the considered initial stress conditions, the earthquake magnitudes varied with depth for certain hypocenters. We analyzed the potential risk of earthquake-induced landslides using the pseudostatic method and introduced the factor of safety (FOS). The results showed that the northern landslide area contained a large section where the FOS was less than one, indicating the increased likelihood of landslides. In addition, for hypocenters at a depth of 8 km, earthquake with the smallest magnitude can generate a significantly stronger event than one with hypocenters at depths of 10 and 12 km. Overall, this fault poses significant risk for a chain of earthquake-landslide-tsunami disasters. Our study can provide a reliable reference for the development of disaster warning systems in the region and further our understanding of seismicity along the QDN segment.
By effectively isolating noise, deep borehole seismic observations offer clear advantages related to signal identification and seismic source research. However, quantitative assessments of noise reduction and signal-to-noise ratio (SNR) improvements in km-scale boreholes are limited. In this study, a deep borehole seismic observation platform in Changde, China, was used to investigate noise reduction and SNR improvement. Two Chinese TDE-120VB very broadband seismometers were deployed at a depth of 1,972.7 m and synchronous observations were made by co-located surface instruments. The results indicate that the background noise power spectral density (PSD) of the borehole system was 20–40 dB lower than that of the surface instruments at 0.001–25 Hz, with decreases of up to 60 dB in certain sub-bands. This corresponds to SNR improvements of 2–4 orders of magnitude. For the first time in deep borehole seismology, spectral entropy was applied to downhole seismic data, and cumulative sum change-point detection within a mean-variance statistical framework was introduced into the entropy series, which enhanced automatic seismic signal identification and noise discrimination. Spectral entropy analysis of the 2025 Dingri earthquake indicates that the borehole station maintained a low-entropy state for ∼4.5 h, which is more than twice the length of that observed at the surface (∼2 h). The findings support the optimization of deep borehole observation systems and provide a valuable reference for underground observation network development and seismic source research.
The MS6.8 earthquake that struck Dingri County on January 7, 2025, has prompted extensive investigations into its post-seismic hazards, focal mechanism, and seismogenic structure. However, the absence of a systematic understanding the sedimentary framework, structural architecture, and other key geological characteristics of the Dingmu Co fault-bounded basin has hindered such efforts. Drawing on a 1:50,000-scale regional geological survey, we examined the sedimentary and structural features of the basin. We further integrated pre-earthquake audio-frequency magnetotelluric (AMT) and soil-radon survey data with optically stimulated luminescence dating of the fault fracture zones. Our results reveal that lacustrine deposits in the basin host abundant seismically induced soft-sediment deformation structures. The northern sector contains an east-west-trending half-graben, and the eastern sector is dominated by several north-south-trending strike-slip normal faults. AMT data have identified multiple concealed faults, and soil-radon measurements indicate heightened fault activity east of the central basin area. Shaped by the combined influence of the north-south-trending Dingmu Co strike-slip-normal fault system and east-west-trending Lhagoi Kangri Detachment System, the basin has undergone at least three episodes of intense tectonic activity since the Late Pleistocene, at ∼53.4, ∼37.1–32.2, and ∼9.2 ka BP, and is presently experiencing renewed deformation. Coseismic deformation associated with the January 7 event was concentrated at the intersections of north-south-, east-west-, northeast-, and northwest-trending faults. These results indicate the need to prioritize such intersection zones in future geological monitoring and disaster risk assessments. Overall, this study offers new insights for advancing seismotectonic research and understanding earthquake hazards in the Himalayan orogenic belt.
With the increase in production depth and pressure, microseismic monitoring and location technology has become particularly important for ensuring the safety of oil and gas production. Therefore, we established a comprehensive well and ground microseismic location network specifically designed to accurately locate microseismic events induced by shale gas extraction. The proposed model, JGF-LocNet, integrates convolutional neural networks (CNNs), transformers, long short-term memory (LSTM) networks, and graph convolutional networks (GCNs) to achieve high-precision event localization with real-time computational efficiency, which closely reproduces the QuakeMigrate reference catalog. The method is velocity-model-free at its inference, but inherits velocity-model assumptions through training labels generated from the catalog. Unlike traditional methods that rely on velocity models, our approach requires only the input waveform, enabling JGF-LocNet to complete seismic event predictions within 0.3 s, with a positioning error of less than 40 m. Blind tests on two unseen arrays confirmed a <10% increase in AE-P90 while maintaining real-time latency. With an adequate number of training samples, JGF-LocNet is expected to deliver rapid, high-precision positioning with minimal errors. We conducted comparative analyses of the computational efficiency and positioning accuracy of our method against those of traditional approaches and other deep learning models. Our results demonstrate that the proposed method most closely reproduces the QuakeMigrate reference catalog in real time while retaining subsecond latency. Given the increased frequency and risk of dynamic disasters in high-pressure, deep, and complex geological structures, the real-time monitoring and precise localization of induced microseismic events provided by our proposed method are critical for ensuring industrial safety.
Owing to Kazakhstan’s transition to new construction standards that align with Eurocode 8, detailed seismic zoning (DSZ) maps of the country are necessary to assess seismic hazards using modern methods. The seismic hazard in construction zones is determined using seismic zoning maps with varying levels of detail. The methodology applied to obtain the DSZ maps was consistent with that used to develop the first probabilistic general seismic zoning maps of Kazakhstan, which were included in the 2017 building regulations for seismic zones and conformed to the Eurocodes. This study met the requirements of the DSZ project and included new data and analytical tools. In addition, the seismic databases and source models were updated. The seismic hazard assessments performed in this study included seismic sources and attenuation modeling, ground motion calculations, and determining the probability of motion exceeding the peak ground acceleration within 50 a, which yielded DSZ maps for the Turkestan region of Kazakhstan. On maps showing a 10% probability of exceedance within 50 a, the obtained values ranged from 0.020 g to 0.370 g, whereas for a 2% probability of exceedance in 50 a, the values ranged from 0.052 g to 0.720 g. The maximum predicted ground acceleration values were located in the southeastern part of Turkestan region within a high-seismicity zone containing active faults. The maps produced in this study facilitate assessments of the probability of destructive earthquakes in the region, thereby enabling the development of risk-reduction and mitigation strategies. The maps also serve as a basis for updating previous building codes.
On February 6, 2023, a catastrophic sequence of earthquakes struck southeastern Türkiye and northern Syria along the East Anatolian Fault System (EAFS), culminating in one of the most destructive seismic episodes in recent decades. This study examined the kinematic behavior, structural segmentation, and tectonic dynamics of the EAFS and its northern splay, the Sürgü-Çardak Fault (SCF). To achieve this, we adopted an integrated geodetic-geophysical approach that integrates small baseline subset interferometric synthetic aperture radar (SBAS-InSAR)–derived surface displacement measurements with gravity-based geophysical interpretations. A time-series analysis of Sentinel-1A acquisitions spanning 2018–2023 was conducted to assess deformation processes along the fault system. The SBAS-InSAR processing enabled us to monitor surface displacement associated with fault activity in regions lacking continuous global navigational satellite system coverage along the extent of the fault. In parallel, gravity data were analyzed to delineate subsurface structural boundaries by characterizing lateral gradients. Fourteen line-of-sight displacement profiles, ten across the EAFS and four across the SCF, revealed pronounced spatial heterogeneity in deformation patterns, which may indicate the coexistence of a partially locked and aseismically creeping fault segment. The mainshock may have nucleated within fault zones that were relatively locked and affected by elevated stress accumulation, where deformation patterns are likely to differ from those observed in other parts of the fault system. The integration of surface displacement inferred from InSAR with structural heterogeneities derived from gravity gradients supports a revised segmentation framework for the EAFS, comprising five distinct structural segments. This integrative geophysical approach advances our understanding of fault system architecture and rupture dynamics, with direct implications for probabilistic seismic hazard assessment. The findings underscore the critical importance of implementing continuous SAR-based monitoring strategies in tectonically active regions to enhance resilience and inform long-term risk mitigation efforts.
Accurate characterization of regional seismic activity is crucial for assessing earthquake hazards. Seismic activity depends on multiple factors, including tectonic loading, fault geometry and distribution, crustal and mantle structures, local topography, physical properties, and global environmental changes. Human influences, such as reservoir impoundment, enhanced geothermal systems, and shale gas extraction, further complicate these relationships. Establishing an integrated theoretical and methodological framework through data to analyze these natural and anthropogenic factors represents a frontier challenge in contemporary seismology and geodynamics. This study introduces three novel visualization methods for earthquake catalogs, efficiently capturing the complex relationships between magnitude, frequency, seismic origin time, and epicentral location. Utilizing these methods with comprehensive heterogeneous geophysical datasets from the Sichuan-Yunnan region, including over 420,000 earthquake records, 160 three-dimensional active fault datasets, high-resolution topography, Moho depth, community velocity models, and crustal deformation data, the seismic characteristics of the region over the past 50 years were systematically analyzed. Results indicate that: (1) High seismic activity and hazard areas in the Sichuan-Yunnan region are primarily concentrated near deep major faults, block boundaries, and brittle transition zones with distinct low- and high-velocity anomalies, showing clear spatial banding, temporal clustering, and cyclicity; (2) Fault segments such as Longmenshan, Lijiang-Xiaojinhe, and Nujiang-Irawaddy likely facilitate internal material exchange within the Qinghai-Xizang Plateau. Significant crustal thickening in their northwestern sections corresponds with lower seismic activity; (3) Crustal strain varies notably along the Xianshuihe-Anninghe-Zemuhe-Xiaojiang and Longmenshan fault zones, which delineate the boundary between regions of high- and low-velocity ratio anomalies. These zones host the majority of the regional earthquakes, requiring intensified monitoring due to frequent events despite moderate mainshock magnitudes. Overall, the proposed methodology provides a new reference for deepening our understanding of regional seismicity and developing an improved earthquake visualization technique.
On March 28, 2025, a large earthquake struck central Myanmar and caused massive damage and casualties. The magnitude of the mainshock is a fundamental parameter that both the general public and scientific communities are concerned about. The coda moment magnitude (Mwo) is a new and reliable method for determining the magnitude of large earthquakes, using more than 10 hours of long-period coda waves. We determined the Mwo moment magnitude (with one standard error) to be 7.86 ± 0.03 for the Myanmar earthquake, which is generally larger than the previous reports. The determination includes a small correction (0.02) from our systematic evaluation for smaller earthquakes MW < 8.3. The Mwo of the Myanmar earthquake is comparable to that of the Kunlun earthquake, and is slightly smaller (0.15 ± 0.03) than that of the 2008 Wenchuan earthquake. The differences are consistent with those from the Global Centroid-Moment-Tensor.
Automated classification of seismic events is critical for earthquake monitoring and explosion detection, particularly in tectonically active regions, such as North China, where the waveform features of earthquakes and explosions are highly similar. This study compared feature-based machine learning (ML) and image-based deep learning (DL) methods in event- and station-level classification frameworks. The dataset consisted of 1,847 events and more than 43,000 vertical-component waveforms with two input types, 40-dimensional feature vectors for ML and spectrogram images for DL. The results showed that the event-level models consistently outperformed the station-level models, achieving over 98% accuracy; the station-level models performed well above 94%. On the test set, the ML and DL models exhibited comparable performance; however, the ML models demonstrated better generalization and lower computational demands. In contrast, DL models required fewer manual interventions. The misclassification analysis revealed distinct error patterns across the model types, indicating potential complementarity. These findings highlight the importance of model choice based on the input type, data granularity, and generalization needs. Although DL models are well suited to automated processing, ML approaches provide more robust and efficient solutions for real-world deployment.
On March 28, 2025, a MW7.7 earthquake struck central Myanmar. In this study, an aftershock catalog recorded by the Seismic Monitoring Center of the Thailand Meteorological Department (TMD) was used to analyze the characteristics of the aftershock sequence. The TMD network detected events to ML3.0 for the Sagaing Fault and Myanmar MW7.7 earthquake, with an additional 0.2 applied as a conservative lower bound. The majority (89%) of ML≥5.5 aftershocks occurred within 16 days of the mainshock. Epidemic-Type Aftershock Sequence modeling revealed that α=2.023, indicating a reduced capacity to trigger secondary aftershocks, while p=1.17, reflecting rapid decay. These slightly high α- and p-values may be due to supershear rupturing, the high slip rate of the Sagaing Fault, and warm crustal conditions. The difference in magnitude between the largest aftershock and mainshock was 1.0, which is consistent with regional historical sequences. The spatial distribution of aftershocks was controlled by the N-S-trending Sagaing Fault, and was closely related to the mainshock slip, fault geometry, and historical rupture zones. The spatial fractal dimension (DC=1.60) indicated a fractal structure, reflecting the complexity of the regional fault structure and the spatial distribution of small earthquakes. After the MW7.7 event, the south-central Sagaing Fault ruptured, except for the Bago segment. However, the northern end of the Bago segment has the potential to produce up to a MW7.0 earthquake.
The subduction and rollback of the Western Pacific Plate and expansion of the Qinghai-Xizang Plateau have significantly influenced the tectonic evolution of the North China Craton (NCC). However, the detailed characteristics of the associated deep dynamic processes remain poorly understood. Seismic anisotropy serves as a key indicator for probing deformation and dynamics in the interior of the Earth. Based on surface wave data recorded by dense seismic arrays, this study applied the Eikonal tomography method to obtain Rayleigh-wave azimuthal anisotropy images at periods of 12-120 s across the central and eastern NCC. We further inverted these data to create a three-dimensional (3D) shear-wave velocity structure and azimuthal anisotropy to a depth of 260 km. Our results revealed a prominent low-velocity anomaly in the upper mantle beneath the central-eastern study region, where the fast direction is primarily E-W. We suggest that this low-velocity anomaly likely originated from the dehydration of the stagnant Western Pacific slab in the upper mantle, which induces partial melting that subsequently ascends to the base of the lithosphere. Furthermore, the E-W anisotropic pattern is likely associated with eastward asthenospheric flow caused by the retreating subduction of the Pacific Plate. East of the Ordos Block, an N-S trending low-velocity anomaly below 180 km depth exhibits weak anisotropy, which may be attributed to deep material upwelling associated with eastward asthenospheric flow and abrupt lateral variations in lithospheric thickness. In the northeastern NCC, the fast direction in the asthenosphere and below shifts to NW-SE, which is distinct from that beneath the North China Plain. This difference may reflect variations in the subduction angle and rollback rate between the Japan and Ryukyu trenches. Our findings provide new observational constraints on lithospheric deformation and deep dynamic processes in this region.
We used the seismic regularity and seismic strain dynamics coefficients to investigate the seismic process preceding the 2025 M7.7 Myanmar earthquake, which ruptured the central portion of the Sagaing Fault. The analysis focused on a seismic cluster located between the Sagaing Fault and Sunda Trench over the ten years prior to the mainshock. In our approach, earthquakes are parameterized by magnitude, elapsed time since the previous event, and epicentral distance to the previous event. These parameters were transformed into equivalent dimensions, ensuring their comparability. The seismic regularity coefficient, which is the mean distance between earthquakes represented by these transformed parameters, acts as a proxy for the strain localization and rupture coherence. The seismic strain dynamics coefficient, which is defined as the logarithm of the ratio between the sum of the cube roots of the scalar seismic moments and the time span over which the events occurred, serves as a proxy for the average inelastic strain accumulation within the fault damage zone. Approximately 29 months before the mainshock, a precursory signal emerged in the seismic regularity coefficient time series. This signal decreased to a pronounced minimum approximately 26 months before the earthquake, followed by an increase to a distinct maximum approximately 10 months before the mainshock. The seismicity responsible for this pattern was distributed across a large area (∼1,550 km × 500 km), with most events occurring on smaller fault structures west of the Sagaing Fault, rather than along the main fault itself. The premonitory behavior of the seismic regularity coefficient prior to the Myanmar earthquake closely matched the patterns observed during other large events. The signal was even more pronounced when analyzed with the seismic strain dynamics coefficient. The joint evolution of both parameters was interpreted within the framework of a model that describes the preparatory process leading to large earthquakes. The obtained results confirm the potential of this approach for long-term earthquake forecasting.
On March 28, 2025, an MS7.9 earthquake occurred in the Himalayan-Myanmar Arc at the junction of the Indian Plate and the Eurasian Plate. Utilizing broadband waveform data provided by the Global Seismographic Network, this study adopted the W-phase method and the P-wave first-motion polarity method to invert the centroid-moment-tensor and fault plane solutions of this earthquake. Additionally, the far-field point source model, global attenuation model, and energy flux density method were used to determine the radiated energy of this earthquake. The energy-moment ratio, apparent stress, stress drop, and radiated energy enhancement factor were also calculated based on the measured seismic moment. The main findings are as follows: (1) this event was a strike-slip earthquake with steeply dipping faults. The focal mechanism solution yielded two nodal planes: Plane I (strike 358°, dip 70°, rake −175°) and Plane II (strike 266°, dip 85°, rake −20°). The seismic moment of 4.94 × 1020 N·m corresponds to a moment magnitude of 7.7. The centroid was located at 21.21°N and 95.92°E and a depth of 35.0 km. The centroid time was 35 s. (2) The radiated seismic energy was 2.6 × 1016 J, which was converted to an energy magnitude of 8.0, higher than the moment magnitude. (3) The energy-moment ratio was 5.3 × 10−5, apparent stress was 1.58 MPa, stress drop was 6.79 MPa, and radiated energy enhancement factor describing the complexity of fault rupture was calculated as 113. (4) In summary, the Myanmar earthquake is a strike-slip earthquake, characterized by steeply dipping faults and highly efficient energy release. Compared with earthquakes of the same magnitude, such earthquakes have a greater destructive power on local buildings and infrastructure, increasing the likelihood of severe disasters.
The 2023 local magnitude (ML) 6.4 Jajarkot earthquake was the most destructive event to occur in the source region of the 1505 MW8.2 Lo Mangthang Mustang earthquake in western Nepal. Here we used local seismic stations deployed in Nepal and southern China, and teleseismic stations with depth phase records to study the earthquake relocation and source rupture process. The Jajarkot earthquake was relocated to an epicenter at 28.828°N, 82.110°E and a focal depth of 13 km. The fault plane solution showed a thrust fault earthquake with a north-dipping angle of 21°. The released seismic moment was 3.88 × 1017 N·m, corresponding to a moment magnitude (MW) of 5.7. The rupture area was 25 km along the strike and 20 km along the dip. The rupture nucleated on the ramp structure of the main Himalayan thrust and propagated to shallower depths, causing severe damage to the surrounding region. This study provides new insights into the local geometry of the main Himalayan thrust within the seismic gap zone in western Nepal.
On March 28, 2025, a MW7.7 earthquake ruptured the Sagaing Fault in Myanmar, producing a more than 400 km of surface rupture. A surveillance camera at the Great Success Energy site, located approximately 124.5 km south of the epicenter, captured the complete rupture process. Using this footage, we extracted displacement time histories and permanent displacements at eight reference points across the fault deformation zone. After video stabilization using StabNet, displacement trajectories were obtained via optical flow tracking, and a geometric transformation model was developed to convert image coordinates into three-dimensional spatial coordinates. The derived displacements were validated against post-earthquake measurements of offset at the plant’s outer drainage fence, confirming their reliability. The earthquake produced a maximum coseismic displacement of up to 2.2 m across the nearby north-south right-lateral strike-slip Sagaing Fault, highlighting the distributed nature of deformation. Surface deformation was spatially concentrated within approximately 20 m, reflecting localized energy release. Time-series analyses showed that the maximum transient displacement was modestly greater than the permanent displacement. Velocity and acceleration time histories derived from the displacement records were used to characterize near-fault strong ground motion and to determine the arrival of seismic waves, the onset of strong shaking, and the onset of the velocity pulse. Comparison of ground motions indicates that the western fault block experienced much stronger shaking than the eastern block, with a peak ground acceleration exceeding 1.41 g and a peak velocity pulse of 3.80 m/s. Slip between the two blocks initiated at the onset of the pulse, resulting in a relative displacement of 2.84 m.
A compilation of databases from Cameroon and neighbouring countries, including seismicity, stress tensor distribution, gravity, magnetic, topography, lithosphere structure and geological data, is used to define its seismotectonic zonation. Based on the quality and quantity of available data, a seismotectonic map was drawn up through the characterization of subunits of concentrations of earthquake foci and, large neotectonic and structural domains. To prepare this map, a homogeneous earthquake catalogue was compiled from the literature and international data centers dated from 1852 up to 2023. Another point of study was to establish links between seismicity and deformation zones. Many faults and/or structures were identified as possibly active, although some of them are not always associated with seismicity. A seismotectonic model for Cameroon was then built from a classification of faults, neotectonic and seismogenic regions. This structured and highly data-driven approach has been developed specifically for the definition of source zones where seismicity is not well known. The results of the seismotectonic analysis allowed characterizing seventeen seismotectonic source zones in Cameroon. Five source zones are defined in the Mount Cameroon region which is the greatest seismicity activity in the study area. The crustal thickness map of Cameroon revealed a thinned transitional zone interspersed between the thickened Congo Shield and thin Pan-African belt favourable for the development of megastructures such as Central Cameroon shear zone and Kribi-Campo shear zone. This region represents the second highest seismicity zone and contains five source zones.
With the increasing use of passive seismic data, developing seismic reflection imaging methods based on passive data is of considerable practical significance. This study presents a waveform-matching reverse time migration for the primary reflected data from local earthquakes. In order to mitigate inconsistencies in frequency band and energy across earthquakes of different magnitudes, we first establish reference seismic waveform with standardized dominant frequency and magnitude. A matching operator is derived for each event by matching its waveforms with the reference waveform. This operator is then applied via convolution to all waveforms, producing standardized seismic waveforms with consistent wavelet features. The reshaped waveforms are then subjected to reverse time migration using an impedance imaging condition for primary reflections. To suppress strong energy interference near the hypocenters, both illumination compensation and three-dimensional Smoothed Spherical Mask centered on each source are used. Numerical tests using both simple two-layer model and fault-containing model demonstrate that the new method is robust and effective. The reverse time migration of primary reflected data of local earthquakes accurately images underground impedance boundaries such as stratum interfaces and fault planes, showing its promise for future application in seismically active fault zones.
This paper proposes a fast quality control strategy for P-wave receiver functions based on AlexNet and wiggle plots. Receiver functions are essential tools in seismology, particularly for analyzing seismic wave propagation and subsurface structures, such as the crust and upper mantle. However, the quality control of receiver functions is often a tedious, time-consuming process. In this study, we transform the time series classification problem of receiver function quality control problem into an image classification task by plotting receiver functions as wiggle diagrams and using the deep learning model AlexNet for binary classification to distinguish between “good” and “bad” receiver functions. The model achieved an accuracy of 92.55% on the testing set and demonstrated strong generalization performance with an accuracy of 89.23% on receiver functions of another seismic network (Sichuan Provincial Permanent Seismic Network). While maintaining strong performance, the model is capable of processing approximately 32 receiver function wiggle plots per second on an NVIDIA GeForce RTX 4050. The results show that the proposed feature mapping strategy significantly improves the efficiency and accuracy of receiver function quality control, making it a valuable tool for practical applications. Future work will focus on expanding the dataset and optimizing model performance for broader seismic data applications.
