Journal of Seismology最新文献

Methodological aspects relative to the conversion of probabilistic hazard evaluations in terms of ground motion intensity into hazard in terms of macroseismic Intensity. As a first step, the probabilistic relationships between ground motion intensity and macroseismic intensity are critically re-examined to account for the different formal properties of these two dimensions of the earthquake intensity. Then these relationships are coherently considered in the conversion of hazard curves by accounting for the inherent binning of macroseismic intensity values and of their inherent probabilistic nature. It is shown that approximate procedures currently used to provide this conversion may provide biased outcomes when the dispersion of values relative to ground motion intensity corresponding to macroseismic Intensity is dismissed or not properly accounted for. The impact of this possible bias is evaluated by considering the seismic hazard at reference site condition at the city of L’Aquila in Central Italy.

Understanding clustered earthquake sequences is essential for seismic hazard assessment, as it involves constraining faulting styles and nodal planes of potential ruptures. This study investigates the nature of a dense earthquake sequence (~ 3000 events) initiated on January 27, 2025, in the Santorini-Amorgos region of the Southern Aegean Sea (SAS), a tectonically active Volcanic Island Arc (VIA). We analyzed 23 shallow crustal earthquakes (Mw ≥ 4.5, depth ≤ 10 km) that occurred between February 2–9, 2025, using full-waveform, low-frequency Centroid Moment Tensor (CMT) inversion from regional seismograms. The inversion was complemented by high-resolution Bouguer gravity anomaly data derived from the EIGEN-6C4 satellite gravity model to assess subsurface density variations. The focal mechanisms consistently indicate NE-SW striking, high-angle (≥ 45°) normal faults with NW- and SE-dipping planes and centroid depths ≤ 10 km. Integration of CMT results with gravity anomalies (90–100 mgal) suggests a migrating zone of shallower extensional magmatism (SEM) driving the sequence. These findings reveal a Precursory Seismic Cluster (PSC) and provide new constraints on the seismotectonic and magmatic processes shaping seismic hazard in the region.

In a natural disaster, intelligent Internet of Things (IoT) systems can be utilized to respond appropriately. Recently, the application of IoT technology in seismology, particularly in earthquake detection, has garnered much attention. This approach’s attractiveness lies in its simplicity of installation, minimal processing power requirements, cost-effectiveness, and expansive coverage, even in areas lacking Internet connectivity. However, the locality of installed sensors brings variations in seismic and noise data, making the earthquake detection task very challenging because of the false alarms. Network-based systems connecting multiple IoTs can resolve the issue by running highly computation-intensive algorithms on a powerful server or cloud and aggregating the data sent from those sensors. On the other hand, Standalone IoT devices operate independently, making decisions locally using both traditional and machine learning methods to manage false alarms. However, these techniques struggle to handle diverse noise patterns and often fail to detect low-magnitude earthquakes in noisy environments. While deep learning models can enhance earthquake detection in such conditions, their high computational cost makes them impractical for resource-constrained devices. To address these challenges, this article introduces a lightweight deep learning model incorporating a transfer learning approach for standalone devices. The proposed model outperforms traditional machine learning methods in earthquake detection using IoT sensors while significantly reducing computational demands. Designed to operate without internet connectivity, the Multi-headed Convolutional Neural Network (MCNN) model achieves 99% accuracy without incurring additional processing costs. Furthermore, it demonstrates high adaptability and the ability to update rapidly with minimal configuration changes.

The rapid and accurate identification of natural earthquakes and artificially blasted earthquakes is crucial for effective earthquake monitoring and early warning. We used waveform data from 5480 natural earthquake events and 4482 blasting events recorded by 110 seismic stations in a quarry in Utah, USA from January 2013 to August 2017, to construct a deep machine learning based CNN-ECA model and accurately and efficiently identify and verify these two types of earthquakes. Firstly, these data were preprocessed by removing mean, trend, instrument response removal, resampling (100 Hz), and bandpass filtering (1–20 Hz). Afterwards, the Fast Fourier Transform (FFT), Continuous Wavelet Transform (CWT), and Short Time Fourier Transform (STFT) methods were used to transform the time-domain data of 1453 natural earthquake events and 1103 quarry blasting events in 2013, obtaining four different types of training sample data: time-domain, frequency-domain (FFT results), and time–frequency domain (CWT and STFT results). Next, the four types of sample data were trained and tested using the Efficient Channel Attention Convolutional Network (CNN-ECA) and traditional Convolutional Neural Network (CNN). The results showed that the CNN-ECA model outperformed the CNN model in all four test samples. Especially when using time–frequency data converted through STFT and FFT as input, the recognition performance of the network model is more significant, with test set accuracies reaching 97.94% and 97.80%, respectively. Finally, the trained CNN-ECA model was used to validate and analyze the natural earthquakes and quarry blasting events recorded between 2014 and 2017. The results indicated that the combined use of FFT and STFT/CWT input data to jointly discriminate seismic events further improved the accuracy of earthquake type identification.

In Cenderawasih Bay, Papua, we improved the assessment of earthquake hazards by integrating seismic parameter analysis with gravity anomaly data. This region, which is known for its seismic activity and complex tectonic structures, offers an ideal setting for disaster risk-mitigation studies. In eastern Indonesia, tectonic plate convergence involves several microplates, creating deformation zones with arc-continent collisions, subduction, shear faults, strike-slip faults, and extensional faults. Our study examined the relationship between earthquake activity and gravity anomalies to enhance the understanding of seismic activity distribution in the Cenderawasih Bay region. Seismicity parameters were analyzed using the IRIS Earthquake Catalog, with aftershocks and foreshocks removed through Reasenberg declustering to focus on independent events. The average values for the seismicity parameters, specifically the magnitudes of completeness (M_c), a, and b, were 4.5, 7.3, and 0.95, respectively, indicating significant tectonic activity. Gravity anomalies were analyzed using a two-dimensional radial power spectrum and three-dimensional Euler deconvolution, which resulted in an estimated depth range for the gravity anomaly source between 15 and 33 km. The results demonstrate that seismic activity is primarily concentrated in the northern region and along the main fault lines of Cenderawasih Bay. There is a correlation between the seismicity parameter b-value and variations in gravity anomalies, with low b-values associated with low to moderate gravity anomalies in regions of high crustal pressure. In contrast, elevated b-values are associated with significant gravity anomalies in low-pressure regions. The seismicity parameter a-value was low to moderate, suggesting tectonic activity. High values correlate with significant gravity anomalies, and the inverse is true. Integrating seismicity analysis with gravity anomaly characterization provides insights into the earthquake distribution in Cenderawasih Bay. This methodological approach not only improves the accuracy of earthquake hazard assessments in the region, but also aids effective disaster risk mitigation.

We present a seismic tomography study of the subsurface structure beneath Hachijojima Island, one of the volcanic fronts in the Izu-Bonin Arc, Japan. Seismic observations were conducted over two 7-month periods in 2019 and 2021, utilizing 55 densely installed stations on the island. During these periods, a total of 179 local earthquakes were recorded — 119 in 2019 and 60 in 2021 — resulting in 4671 P-wave arrival times and 3927 S-wave arrival times. The 3-D tomography, derived using the double-difference technique, revealed a shallow low-velocity region between the island’s two main volcanoes, Nishiyama and Higashiyama, suggesting the presence of volcanic sediments near the surface. Additionally, a high-velocity anomaly was identified at a depth of 4–5 km, extending vertically from deeper regions beneath Nishiyama. This feature is interpreted as a magma pathway from past volcanic activity, with high P-wave velocities and elevated Vp/Vs ratios indicating possible fluid presence. At greater depths, low P-wave velocity perturbations and elevated Vp/Vs ratios suggest a magmatic plumbing system comprising a mid-crustal magma chamber at approximately 8–12 km depth and lateral magmatic pathways at 10–20 km depth. Furthermore, a distinct zone characterized by reduced P-wave velocity and increased Vp/Vs is interpreted as a shallow magma chamber with H₂O-saturated magma accumulation. These findings provide valuable insights into the subsurface magmatic processes beneath Hachijojima Island, which are crucial for improving volcanic hazard assessment.

We investigate seismic velocity changes in the rock mass related to mining induced seismic events and ore exploitation by computing a one-month long 4D elastic model of Kiirunavaara mine (Sweden). We focus on a specific mine sector, where a single (varvec{M_W})=2.0 event occurred on May 22 (02:31 local time), damaging the infrastructure. We make use of P- and S-first-arrival times obtained from the permanent seismic system for computing the full 4D (continuous 3D volume in time) seismic velocity model of Kiruna mine using a trans-dimensional Monte Carlo sampling. The trans-dimensional approach guarantees that the resolution, both in space and in time, is strictly data-driven. Our results give the following insights into the velocity differences at the mining levels and at different time-length scales. (a) We observe a striking correlation between spatial variations of (varvec{V_P}) and ore-body geometry, confirming the robustness of the velocity model. Clay zones appear as a low (varvec{V_P/V_S}) ratio zones, as seen in previous tomographic studies. (b) High-frequency (hourly) fluctuations of the rock mass (varvec{V_P}) around the ore-passes are highly correlated with seismic sequences in the same rock volumes. In particular, (varvec{V_P}) increases rapidly when ore-passes are seismically active and (varvec{V_P}) values keep a high value for few (1-4) hours after the end of the seismic sequence. (c) The smoothed velocity model, computed as averaged model over a 2-days moving window, suggests that low-frequency (varvec{V_P}) fluctuations can be compared to stress cell measurements located closely.

In this study, we conduct a probabilistic seismic hazard assessment of induced seismicity in northeast British Columbia, Canada, where fluid injection related to oil and gas activity has caused a significant increase in seismicity rate over the last 40 years. Considering several sources of natural seismicity (based on the 6th generation of seismic hazard map of Canada) as the background and a time-variable induced seismicity source from an earthquake catalogue prepared in this study, we assess the seismic hazard for several time periods at a location in the city of Fort St. John from earthquakes within a radius of 300 km. Seismic sources are characterized based on minimum and maximum magnitudes, Gutenberg-Richter parameters (a-value and b-value), and earthquake focal depth. Following the Monte Carlo sampling, earthquake catalogues are synthesized for different realizations of seismic sources and ground motion is estimated (for peak ground acceleration, PGA, and peak ground velocity, PGV) at the target location from each earthquake. Considering a logic tree to account for epistemic uncertainty in sources of seismicity and ground motion estimation, we calculate hazard curves for different investigation periods of 1980–2002, 2003–2012, 2013–2024, and yearly periods between 2013 and 2024 (inclusive). Our results show that both PGA and PGV increase over time. However, the increase is higher for PGA than PGV. For example, at the exceedance probability of 2% in 50 years (return period of 2475 years), PGA increases by ~ 12 times from the background level to its maximum in 2022, whereas PGV increases by ~ 5 times. These results have important implications for risk assessment, particularly as injection activities, such as hydraulic fracturing and wastewater disposal, continue to influence the seismicity rate. Additionally, emerging technologies like enhanced geothermal systems and geological CO₂ storage further underscore the need for understanding seismic hazard from induced seismicity.

Paliki Peninsula, on Cephalonia Island, is one of the most seismically active areas in Europe, east of the major Cephalonia transform fault. In 1953, a series of three major earthquakes, Mw 5.9, Mw 6.6, and Mw 7.0, produced extensive damage in Cephalonia. More recently, in 2014, a sequence with two main shocks (~ Mw 6.0) caused considerable damage. We integrated a geological and geotechnical dataset from existing sources and used it together with ambient-noise records to study the subsurface structure of Paliki Peninsula along two E-W profiles, crossing the central (13 sites) and southern (seven sites) parts of the peninsula. We combined fundamental frequency microtremor horizontal-to-vertical spectral ratios (MHVSR) and shear-wave velocity, as derived empirically from borehole geotechnical parameters data and, in some cases, from literature, with forward modelling of MHVSR curves, to obtain 2D images of the subsurface structure along the profiles. The depth to the bedrock (Eocene limestones) reach maximum values of 300–450 m to the eastern end of the two profiles, with three overlying soil formations on top of the bedrock: (i) Holocene deposits 2–4 m thick, (ii) Marine deposits, with thicknesses of 4–30 m, and (iii) Marls of varying thicknesses increasing from West to East, with steeper slope in the central profile near the coast. This preliminary image of the subsurface structure of Paliki Peninsula will contribute to a better understanding of local tectonics, earthquake sources, local/path propagation effects, and for improved local seismic hazard assessments and risk mitigation plans.

This study addresses the critical need for accurate prediction models of seismic activity in Turkey, focusing on the main earthquakes and the aftershocks that follow them. The complex geological structure of Turkey, controlled by major fault lines such as the North Anatolian Fault Line and the East Anatolian Fault Line, requires robust analysis to understand seismic hazards better and to implement effective preventive measures. This research aims to fill the gap in the predictive modeling of integer-valued seismic data by comparing the effectiveness of first-order INteger-valued AutoRegression (INAR(1)) models with the more commonly used AutoRegressive Integrated Moving Average (ARIMA) models. To achieve this, we analysed the occurrence of mainshocks and aftershocks on a monthly basis from January 2011 to December 2020. The INAR(1) models were specifically applied to this integer-valued time-series data, and their forecasts were compared with those produced by ARIMA models. Our results indicate that the INAR(1) models provide forecasts closer to the observed values than the ARIMA models for both the mainshock and aftershock datasets. In particular, the INAR(1) models showed superior performance in terms of accuracy, with numerical results showing a reduction in forecast error of about 15% compared to ARIMA models. These results have significant implications for earthquake preparedness and risk reduction in Turkey. Through the use of INAR(1) models, we can improve the accuracy of the prediction of seismic activity and thereby increase the ability to implement safety measures in a timely and effective manner. This study highlights the importance of better understanding and mitigating earthquake risk by using appropriate statistical models tailored to the specific characteristics of seismic data.