Search

The spread of misinformation on social media can lead to inappropriate behaviors that can make disasters worse. In our study, we focused on tweets containing misinformation about earthquake predictions and analyzed their dynamics. To this end, we retrieved 82,129 tweets over a period of 2 years (March 2020–March 2022) and hand-labeled 4157 tweets. We used RoBERTa to classify the complete dataset and analyzed the results. We found that (1) there are significantly more not-misinformation than misinformation tweets; (2) earthquake predictions are continuously present on Twitter with peaks after felt events; and (3) prediction misinformation tweets sometimes link or tag official earthquake notifications from credible sources. These insights indicate that official institutions present on social media should continuously address misinformation (even in quiet times when no event occurred), check that their institution is not tagged/linked in misinformation tweets, and provide authoritative sources that can be used to support their arguments against unfounded earthquake predictions. [ABSTRACT FROM AUTHOR]

Earthquakes, as natural phenomena, have consistently caused damage and loss of human life throughout history. Earthquake prediction is an essential aspect of any society's plans and can increase public preparedness and reduce damage to a great extent. Despite advances in computing systems and deep learning methods, no substantial achievements have been made in earthquake prediction. One of the most important reasons is that the earthquake's nonlinear and chaotic behavior makes it hard to train the deep learning method. To tackle this drawback, this study tries to take an effective step in improving the performance of prediction results by employing a novel method in earthquake prediction. This method employs a deep learning model based on convolutional neural networks (CNN), bi-directional long short-term memory (BiLSTM), and an attention mechanism, as well as a zero-order hold (ZOH) pre-processing methodology. This study aims to predict the maximum magnitude and number of earthquakes in the next month with the least error. The proposed method was evaluated by an earthquake dataset from nine distinct regions of China. The results reveal that the proposed method outperforms other prediction methods in terms of performance and generalization. [ABSTRACT FROM AUTHOR]

The effort to identify and comprehend potential earthquake-related phenomena shares a common goal: successful earthquake forecasting. Advancements in science and technology have made this goal multidisciplinary. Currently, possibly earthquake-related anomalies in the Vertical Total Electron Content (VTEC) of the Earth's ionosphere are being investigated. Global Navigation Satellite Systems (GNSS) can be used to calculate this ionospheric parameter. In this research work, GPS VTEC was calculated for periods between the years 2015 and 2019. The selection of this periods considered both seismically active and non-seismically areas in Mexico. The M w ≥ 5 earthquakes under study were registered by the National Seismological Service. Moreover, different geomagnetic storm and solar activity parameters, such as the geomagnetic equatorial Dst index and the F10.7 index, were analyzed. Additionally, the daily average and monthly mean number of sunspots (R, SSN, respectively) were included as a direct, long-term record of the development of the solar cycle. To the periods under study different statistical methods were applied, such as Mean-Square Error (MSE) and cross-correlation. The above aims to apply a machine learning technique capable of classifying between periods with seismic and non-seismic activity. The features were constructed using statistical data and results from the implemented analysis. Furthermore, Principal Component Analysis (PCA) was applied to reduce the feature vector dimensions, and accuracy scores were compute using k-fold cross-validation. The results from the Support Vector Machine (SVM) model indicated an accuracy of 88.9% for the training set, and an accuracy of 80% was obtained for the test set. One of the limitations of the current study was the sample size. However, the present initial approach for classifying seismic events from non-seismic periods using SVM demonstrated promising results when considering the indicated parameters and the days under study. [ABSTRACT FROM AUTHOR]

Predicting the remaining time before the next earthquake based on seismic signals generated in a laboratory setting is a challenging research task that is of significant importance for earthquake hazard assessment. In this study, we employed a mel spectrogram and the mel frequency cepstral coefficient (MFCC) to extract relevant features from seismic signals. Furthermore, we proposed a deep learning model with a hierarchical structure. This model combines the characteristics of long short-term memory (LSTM), one-dimensional convolutional neural networks (1D-CNN), and two-dimensional convolutional neural networks (2D-CNN). Additionally, we applied a stacking model fusion strategy, combining gradient boosting trees with deep learning models to achieve optimal performance. We compared the performance of the aforementioned feature extraction methods and related models for earthquake prediction. The results revealed a significant improvement in predictive performance when the mel spectrogram and stacking were introduced. Additionally, we found that the combination of 1D-CNN and 2D-CNN has unique advantages in handling time-series problems. [ABSTRACT FROM AUTHOR]

The study reported in this paper concerns the temporal variation in the b-value of the Gutenberg–Richter frequency–magnitude law, applied to the earthquakes that struck Italy from 2009 to 2016 in the geographical areas of L'Aquila, the Emilia Region, and Amatrice–Norcia. Generally, the b-value varies from one region to another dependent on earthquake incidences. Higher values of this parameter are correlated to the occurrence of low-magnitude events spread over a wide geographical area. Conversely, a lower b-value may lead to the prediction of a major earthquake localized along a fault. In addition, it is observed that each seismic event has a different "occurrence time", which is a key point in the statistical study of earthquakes. In particular, its results are absolutely different for each specific event, and may vary from years to months or even just a few hours. Hence, both short- and long-term precursor phenomena have to be examined. Accordingly, the b-value analysis has to be performed by choosing the best time windows to study the foreshock and aftershock activities. [ABSTRACT FROM AUTHOR]

The ionospheric anomalies before an earthquake may be related to earthquake preparation. The study of the ionospheric anomalies before an earthquake provides potential value for earthquake prediction. This paper proposes a method for detecting ionospheric total electron content (TEC) anomalies before an earthquake, taking the MS 7.8 earthquake in Türkiye on 6 February 2023 as an example. First, the data of four ground-based GNSS stations close to the epicenter were processed by using the sliding interquartile range method and the long short-term memory (LSTM) network. The anomaly dates detected by the two methods were identified as potential pre-earthquake TEC anomaly dates after eliminating solar and geomagnetic interference. Then, by using the sliding interquartile range method to process and analyze the CODE GIM (Center for Orbit Determination in Europe, Global Ionospheric Map) data from a global perspective, we further verified the existence of TEC anomalies before the earthquake on the above TEC anomaly days. Finally, the influence of the equatorial ionospheric anomaly (EIA) on the TEC anomaly disturbance was excluded. The results show that the ionospheric TEC anomalies on January 20, January 27, February 4, and February 5 before the Türkiye earthquake may be correlated with the earthquake. [ABSTRACT FROM AUTHOR]

In a recent work, we applied the every earthquake a precursor according to scale (EEPAS) probabilistic model to the pseudo-prospective forecasting of shallow earthquakes with magnitude |$M\ 5.0$| in the Italian region. We compared the forecasting performance of EEPAS with that of the epidemic type aftershock sequences (ETAS) forecasting model, using the most recent consistency tests developed within the collaboratory for the study of earthquake predictability (CSEP). The application of such models for the forecasting of Italian target earthquakes seems to show peculiar characteristics for each of them. In particular, the ETAS model showed higher performance for short-term forecasting, in contrast, the EEPAS model showed higher forecasting performance for the medium/long-term. In this work, we compare the performance of EEPAS and ETAS models with that obtained by a deterministic model based on the occurrence of strong foreshocks (FORE model) using an alarm-based approach. We apply the two rate-based models (ETAS and EEPAS) estimating the best probability threshold above which we issue an alarm. The model parameters and probability thresholds for issuing the alarms are calibrated on a learning data set from 1990 to 2011 during which 27 target earthquakes have occurred within the analysis region. The pseudo-prospective forecasting performance is assessed on a validation data set from 2012 to 2021, which also comprises 27 target earthquakes. Tests to assess the forecasting capability demonstrate that, even if all models outperform a purely random method, which trivially forecast earthquake proportionally to the space–time occupied by alarms, the EEPAS model exhibits lower forecasting performance than ETAS and FORE models. In addition, the relative performance comparison of the three models demonstrates that the forecasting capability of the FORE model appears slightly better than ETAS, but the difference is not statistically significant as it remains within the uncertainty level. However, truly prospective tests are necessary to validate such results, ideally using new testing procedures allowing the analysis of alarm-based models, not yet available within the CSEP. [ABSTRACT FROM AUTHOR]

The detection of earthquakes is a fundamental prerequisite for seismology and contributes to various research areas, such as forecasting earthquakes and understanding the crust/mantle structure. Recent advances in machine learning technologies have enabled the automatic detection of earthquakes from waveform data. In particular, various state-of-the-art deep-learning methods have been applied to this endeavour. In this study, we proposed and tested a novel phase detection method using deep learning, which is based on a standard convolutional neural network in a new framework. The novelty of the proposed method is its separate explicit learning strategy for global and local representations of waveforms, which enhances its robustness and flexibility. Prior to modelling the proposed method, we identified local representations of the waveform by the multiple clustering of waveforms, in which the data points were optimally partitioned. Based on this result, we considered a global representation and two local representations of the waveform. Subsequently, different phase detection models were trained for each global and local representation. For a new waveform, the overall phase probability was evaluated as a product of the phase probabilities of each model. This additional information on local representations makes the proposed method robust to noise, which is demonstrated by its application to the test data. Furthermore, an application to seismic swarm data demonstrated the robust performance of the proposed method compared with those of other deep learning methods. Finally, in an application to low-frequency earthquakes, we demonstrated the flexibility of the proposed method, which is readily adaptable for the detection of low-frequency earthquakes by retraining only a local model. [ABSTRACT FROM AUTHOR]

In order to explore the correlation between earthquakes and ionospheric very low-frequency (VLF) electric field disturbances, this article uses VLF data observed by the China Earthquake Electromagnetic Satellite (CSES) to analyze very low-frequency signals before and after earthquakes from January 2019 to March 2023 in terms of the amplitude and signal-to-noise ratio of electric field power spectrum disturbances. Taking 73 earthquakes with a magnitude of 6.0 or higher occurring in the Circum-Pacific seismic belt as an example, comprehensive research on the VLF electric field disturbance phenomenon caused by strong earthquakes is conducted, considering both the earthquake location and source mechanism. The research results indicate the following: (1) there is a strong correlation between earthquakes with a magnitude of 6.0 or above and abnormal disturbances in the VLF electric field, which often occur within 20 days before the earthquake and within 800 km from the epicenter. (2) From the perspective of earthquake-prone areas, the VLF electric field anomalies observed before earthquakes in the Ryukyu Islands of the Taiwan region exhibit small and concentrated field fluctuations, while the Taiwan Philippines region exhibits larger field fluctuations and more dispersed fluctuations. The discovery of this correlation between seismic ionospheric phenomena and seismic activity provides a new and effective approach to earthquake monitoring, which can be used for earthquake prediction, early warning, and disaster prevention and reduction work. [ABSTRACT FROM AUTHOR]

Time series analysis was applied to the continuous radon level, temperature, pressure, and rainfall to find clear earthquake signals. Radon signals appeared a few days after heavy rains, and radon signals associated with events M = 3.8–4.2 were detected 12 up to 36 days before. The events are complete data recorded from 1983 to 1986, giving discussion and conclusion on M with prediction time and radon anomaly detected in the Stivos faulting near Thessaloniki, N. Greece. [ABSTRACT FROM AUTHOR]

This paper is the second of two companion papers that seeks to compare homeowners' expectations of damage and engineering predictions of damage to timber-framed houses before and after undertaking seismic structural strengthening. Part I analyzed the seismic vulnerability of wooden-framed houses located on slopes in Wellington, New Zealand, investigating factors of plan shape relative to the slope, slope variations, and wall distribution, and how they influence the final seismic performance of houses. A structural survey provided data on the form and typical details for the subfloor bracing, and this was then used as the basis for a simple strengthening solution that is numerically investigated here. This companion paper analyzes the improvement of seismic performance after undertaking strengthening to the subfloor structure using a multiple stripe analysis. In the last phase of this work, engineering-based predictions of performance as determined using the numerical models were compared to homeowners' expectations of damage before and after undertaking strengthening. This comparison paper found that although strengthened timber-framed houses located on slopes satisfy the minimum requirement of New Zealand design codes, there is still a probability of damage occurring, which would not satisfy owners' expectations of performance in all scenarios. [ABSTRACT FROM AUTHOR]

Taking the Madoi MS 7.4 earthquake of 21 May 2021 as an example, this paper proposes using time series prediction models to predict the outgoing long-wave radiation (OLR) anomalies and study short-term pre-earthquake signals. Five time series prediction models, including autoregressive integrated moving average (ARIMA) and long short-term memory (LSTM), were trained with the OLR time series data of the aseismic moments in the 5° × 5° spatial range around the epicenter. The model with the highest prediction accuracy was selected to retrospectively predict the OLR values during the aseismic period and before the earthquake in the area. It was found, by comparing the predicted time series values with the actual time series value, that the similarity indexes of the two time series before the earthquake were lower than the index of the aseismic period, indicating that the predicted time series before the earthquake significantly differed from the actual time series. Meanwhile, the temporal and spatial distribution characteristics of the anomalies in the 90 days before the earthquake were analyzed with a 95% confidence interval as the criterion of the anomalies, and the following was found: out of 25 grids, 18 grids showed anomalies—the anomalies of the different grids appeared on similar dates, and the anomalies of high values appeared centrally at the time of the earthquake, which supports the hypothesis that pre-earthquake signals may be associated with the earthquake. [ABSTRACT FROM AUTHOR]

The generation of strong earthquakes is a long‐debated problem in seismology, and its importance is increased by the possible implications for earthquake forecasting. It is hypothesized that the earthquake generation processes are anticipated by several phenomena occurring within a nucleation region. These phenomena, also defined as preparatory processes, load stress on the fault leading it to reach a critical state. In this paper, we investigate the seismicity preceding 19 moderate (Mw ≥ 3.5) earthquakes at The Geysers, Northern California, aiming to verify the existence of a preparatory phase before their occurrence. We apply an unsupervised K‐means clustering technique to analyze time series of physics‐related features extracted from catalog information and estimated for events occurred before the mainshocks. Specifically, we study the temporal evolution of the b‐value from the Gutenberg‐Richter (b), the magnitude of completeness (Mc), the fractal dimension (Dc), the inter‐event time (dt), and the moment rate (M˙0 ${\dot{M}}_{0}$). Our analysis shows that out of 19 moderate magnitude events considered, a common preparatory phase for 11 events is clearly identified, plus other five events for which we can guess a preparatory phase but with different characteristics from the previous ones. The latter result confirms that even within the same tectonic context different possible activation behaviors may exist. The duration of the preparatory process ranges between about 16 hr and 4 days. We observe that also for the retrieved preparatory process a decrease in b, Mc, and Dc, and an increase of M˙0 ${\dot{M}}_{0}$. Finally, we show a clear correlation between events with a preparation phase and the location of injection's wells, suggesting an important role of fluids in the preparatory process. Plain Language Summary: We investigate the preparatory phase of moderate magnitude‐induced earthquakes at The Geysers geothermal field in California by studying the spatiotemporal evolution and dynamic properties of small magnitude events and using an unsupervised machine learning approach. To this aim, we rely on features extracted from seismic catalog information that are used for a K‐means clustering analysis. Our results highlight changes in the seismicity characteristics before the moderate‐induced earthquakes. We find that most of the analyzed target earthquakes present a preparatory phase, and that most of the cases the latter presents common characteristics in their key features. We show that the seismicity clusters in space and time before moderate events also becoming more energetic. We estimate a duration for the detected preparatory phases that ranges from 16 hr to 4 days. Finally, we show that the presence of a preparatory phase is correlated to the proximity of injection's wells, suggesting a significant role of fluids in the earthquake's nucleation process. Key Points: Moderate‐induced earthquakes at The Geysers can be preceded by a preparatory phase detectable using K‐means clustering on catalog featuresThe detected preparatory phase has common characteristics presenting negative trends of b‐value, Mc, and Dc and a positive trend of M˙0 ${\dot{M}}_{0}$The earthquakes with a preparatory phase are located nearer to the injection's wells in the area than the events without it [ABSTRACT FROM AUTHOR]

Large earthquakes in the vicinity of Antarctica have the potential to cause postseismic viscoelastic deformation affecting measurements of displacement that are used to constrain models of glacial isostatic adjustment (GIA). In November 2013, a Mw 7.7 strike‐slip earthquake occurred in the Scotia Sea, 650 km from the Antarctic Peninsula. GPS time series from the northern Peninsula show a change in rate after this event, indicating a far‐field postseismic deformation signal is present. In this study, we use a finite element model with a suite of 1D and 3D Earth structures to investigate the extent of postseismic deformation in the Antarctic Peninsula. Model output is compared with GPS time series to place constraints on the Earth structure in this region. The preferred Earth structure has a thin lithosphere combined with a Burgers rheology with steady‐state viscosity of 4 × 1018 Pa s and transient viscosity one order of magnitude lower. Our study shows that including 3D Earth structure does not improve the fit. Using the best fitting Earth structure, we run a forward model of the nearby 2003 Mw 7.6 strike‐slip earthquake and combine the predictions for both earthquakes. We show that postseismic deformation is widespread across the northern Peninsula with rates of horizontal deformation up to 1.65 mm/yr for the period 2015–2020, a signal that persists for decades. These results suggest that much of Antarctica may be deforming due to recent postseismic deformation and this signal needs to be accounted for when using GPS observations to constrain geophysical models. Plain Language Summary: Antarctica is not well‐known for earthquakes and consequently there are very few studies on the subject. However, large earthquakes that have occurred some 100s km from the coastline have the potential to cause deformation within the continent as the Earth undergoes far‐field postseismic viscoelastic relaxation in response to the large stress changes in the decades following an earthquake. In this study, we examine two earthquakes that occurred in the Scotia Sea around 650 km from the northern tip of the Antarctic Peninsula in 2003 and 2013. GPS observations from the Northern Antarctic Peninsula show a change in deformation rate following the 2013 earthquake. We use the time series to constrain a finite element model of postseismic deformation to constrain the underlying Earth properties. Using this model to predict ongoing deformation due to the 2003 and 2013 earthquakes gives an estimated combined horizonal deformation of up to 1.65 mm/yr during 2015–2020. These results suggest that much of Antarctica may be deforming due to postseismic deformation, a signal that had not yet been accounted for in estimates of Antarctic ice‐mass balance. Key Points: The northern Antarctic Peninsula is undergoing horizontal postseismic deformation up to 1.65 mm/yr during 2015–2020A model with Burgers rheology and an asthenosphere steady state viscosity of 4 × 1018 Pa s best fits the GPS‐observed deformationThis previously unaccounted‐for signal should be considered before using Antarctic GPS to constrain glacial isostatic adjustment models [ABSTRACT FROM AUTHOR]

We use machine-learning and cross-correlation techniques to enhance earthquake detectability by two magnitude units for the earthquake sequence near Musreau Lake, Alberta, which is induced by wastewater disposal. This deep catalogue reveals a series of en echelon ∼N–S oriented strike-slip faults that are favourably oriented for reactivation. These faults require only ∼0.6 MPa overpressure for triggering to occur. Earthquake activity occurs in bursts, or episodes; episodes restricted to the largest fault tend to have earthquakes starting near the southern end (distant from injectors) and progressing northwards (towards the injectors). While most events are concentrated along these ∼N–S oriented faults, we also delineate smaller faults. Together, these findings suggest pore pressure as the triggering mechanism, where a time-dependent increase in pore pressure likely caused these faults to progressively reawaken. Analysis of the 'next record-breaking event', a statistical model that forecasts the sequencing of earthquake magnitudes, suggests that the next largest event would be M L ∼4.3. The seismically illuminated length of the largest fault indicates potential magnitudes as large as M w 5.3. [ABSTRACT FROM AUTHOR]

Earthquake-induced gravity variation refers to changes in the earth's gravity field associated with seismic activities. In recent years, development in the theories has greatly promoted seismic deformation research, laying a solid theoretical foundation for the interpretation and application of seismological gravity monitoring. Traditional terrestrial gravity measurements continue to play a significant role in studies of interseismic, co-seismic, and post-seismic gravity field variations. For instance, superconducting gravimeter networks can detect co-seismic gravity change at the sub-micro Gal level. At the same time, the successful launch of satellite gravity missions (e.g., the Gravity Recovery and Climate Experiment or GRACE) has also facilitated applied studies of the gravity variation associated with large earthquakes, and several remarkable breakthroughs have been achieved. The progress in gravity observation technologies (e.g., GRACE and superconducting gravimetry) and advances in the theories have jointly promoted seismic deformation studies and raised many new research topics. For example, superconducting gravimetry has played an important role in analyses of episodic tremor, slow-slip events, and interseismic strain patterns; the monitoring of transient gravity signals and related theories have provided a new perspective on earthquake early warning systems; the mass transport detected by the GRACE satellites several months before an earthquake has brought new insights into earthquake prediction methods; the use of artificial intelligence to automatically identify tiny gravity change signals is a new approach to accurate and rapid determination of earthquake magnitude and location. Overall, many significant breakthroughs have been made in recent years, in terms of the theory, application, and observation measures. This article summarizes the progress, with the aim of providing a reference for seismologists and geodetic researchers studying the phenomenon of gravity variation, advances in related theories and applications, and future research directions in this discipline. [ABSTRACT FROM AUTHOR]

It has been observed that, before some strong earthquakes occur, the space static electricity near the ground is abnormal, which might be caused by a large amount of radioactive gas released from the Earth's crust. In this paper, the information diffusion technology for optimally processing small samples is used to analyze 30 cases, and the relationship between magnitude and parameters such as abnormality of space static electricity is constructed. Each case is composed of four observation values: abnormality e, epicenter distance d, impending time t and magnitude m. Using the causal relationship constructed in this paper, the magnitude m of an impending earthquake could be approximately inferred from abnormality e. According to the progress of locking the epicenter and the passage of time, the predicted magnitude could be adjusted in a timely manner. The research results provided in this paper do not eliminate the uncertainty of earthquake occurrence, so that the study is a work of analyzing seismic dynamic risk. Integrating the monitoring information from seismic stations and the physical field information in the air will promote impending earthquake prediction, which is a worldwide scientific challenge. [ABSTRACT FROM AUTHOR]

Decoding the variation laws of the deformation field before strong earthquakes has long been recognized as an essential issue in earthquake prediction research. In this paper, the temporal and spatial distribution characteristics of deformation anomalies in the northeastern margin of the Qinghai-Tibetan Plateau before and after the Menyuan MS 6.9 earthquake were studied by using the Fisher statistical test method. By analyzing the characteristics of these anomalies, we found that: 1) The deformation anomalies are mainly distributed in the marginal front area of the Qinghai-Tibetan Plateau, where short-term deformation anomalies are prone to occur due to a high gradient of gravity; 2) The deformation anomalies along the northeastern margin of the Qinghai-Tibetan Plateau are characterized by spatial propagation, and the migration rate is about 2.4 km/d. The propagation pattern is counterclockwise, consistent with the migration direction of MS ≥ 6.0 earthquakes; 3) The time and location of the Menyuan earthquake are related to the group migration of earthquakes with MS ≥ 6.0. Finally, based on the results of gravity field variation and the theory of crust stress wave, the law of deformation anomaly distribution was discussed. We suggest that both the deformation propagation along the northeastern margin of the Qinghai-Tibetan Plateau and the earthquake migration are possibly associated with the variation of the stress field caused by subsurface mass flow. [ABSTRACT FROM AUTHOR]

The temporal distribution of earthquakes provides important basis for earthquake prediction and seismic hazard analysis. The relatively limited records of strong earthquakes have often made it difficult to study the temporal distribution models of regional strong earthquakes. However, there are hundreds of years of complete strong earthquake records in the North China Seismic Zone, providing abundant basic data for studying temporal distribution models. Using the data of M ≥ 6.5 earthquakes in North China as inputs, this paper estimates the model parameters using the maximum likelihood method with Poisson, Gamma, Weibull, Lognormal and Brownian passage time (BPT) distributions as target models. The optimal model for describing the temporal distribution of earthquakes is determined according to Akaike information criterion (AIC),and Kolmogorov–Smirnov test (K–S test). The results show that Lognormal and BPT models perform better in describing the temporal distribution of strong earthquakes in North China. The mean recurrence periods of strong earthquakes (M ≥ 6.5) calculated based on these two models are 8.1 years and 13.2 years, respectively. In addition, we used the likelihood profile method to estimate the uncertainty of model parameters. For the BPT model, the mean and 95% confidence interval of recurrence interval μ is 13.2 (8.9–19.1) years, and the mean and 95% confidence interval of α is 1.29 (1.0–1.78). For the Lognormal model, the mean value and 95% confidence interval of v is 2.09 (1.68–2.49), the mean value exp (v) corresponding to earthquake recurrence interval is 8.1 (5.4–12.1) years. In this study, we also calculated the occurrence probability of M ≥ 6.5 earthquakes in the North China Seismic Zone in the future, and found that the probability and 95% confidence interval in the next 10 years based on the BPT model is 35.3% (26.8%-44.9%); the mean value and 95% confidence interval of earthquake occurrence probability based on the Lognormal distribution is 35.4% (22.9%-49.7%); the mean probability and 95% confidence interval based on the Poisson model is 53.1% (41.1%-64%). The results of this study may provide important reference for temporal distribution model selection and earthquake recurrence period calculation in future seismic hazard analysis in North China. [ABSTRACT FROM AUTHOR]

Seismology is witnessing explosive growth in the diversity and scale of earthquake catalogs. A key motivation for this community effort is that more data should translate into better earthquake forecasts. Such improvements are yet to be seen. Here, we introduce the Recurrent Earthquake foreCAST (RECAST), a deep‐learning model based on recent developments in neural temporal point processes. The model enables access to a greater volume and diversity of earthquake observations, overcoming the theoretical and computational limitations of traditional approaches. We benchmark against a temporal Epidemic Type Aftershock Sequence model. Tests on synthetic data suggest that with a modest‐sized data set, RECAST accurately models earthquake‐like point processes directly from cataloged data. Tests on earthquake catalogs in Southern California indicate improved fit and forecast accuracy compared to our benchmark when the training set is sufficiently long (>104 events). The basic components in RECAST add flexibility and scalability for earthquake forecasting without sacrificing performance. Plain Language Summary: We explore the potential for deep learning in earthquake forecasting. Prior work has relied heavily on statistical models that do not scale to fully utilize the currently available large earthquake data sets. Here we build on recent developments in deep learning for forecasting event sequences in general to create an implementation for earthquake data. The new approach allows us to incorporate larger data sets, potentially with more information about each earthquake. We also avoid a specific functional form, so the method naturally adapts to additional information about events, like magnitude or variations in behavior over time. As we add more data, results show continued improvements. This ability to incorporate and improve continually as training data sets increase indicates that there is more information in the earthquake catalogs than has yet been used for earthquake forecasting. Key Points: We introduce a deep learning model for earthquake forecasting and explore its performance on synthetic and regional earthquake data setsIt is flexible in the sense that a predefined functional form is not requiredIt is scalable in two senses: it is efficient on large data sets, and its performance relative to benchmarks improves with more training data [ABSTRACT FROM AUTHOR]

Earthquake hazard analyses rely on the availability of seismogenic source models. These are designed in different fashions, such as point sources or area sources, but the most effective is the three-dimensional representation of geological faults. We here refer to such models as fault sources. This study presents the European Fault-Source Model 2020 (EFSM20), which formed the basis for one of the primary input datasets of the recently released European Seismic Hazard Model 2020. The EFSM20 compilation was entirely based on reusable data from existing active fault regional compilations that were first blended and harmonized and then augmented by a set of derived parameters. These additional parameters were devised to enable users to formulate earthquake rate forecasts based on a seismic-moment balancing approach. EFSM20 considers two main categories of seismogenic faults: crustal faults and subduction systems. The compiled dataset covers an area from the Mid-Atlantic Ridge to the Caucasus and from northern Africa to Iceland. It includes 1,248 crustal faults spanning a total length of ~95,100 km and four subduction systems, namely the Gibraltar, Calabrian, Hellenic, and Cyprus Arcs. The model focuses on an area encompassing a buffer of 300 km around all European countries (except for Overseas Countries and Territories, OTCs) and a maximum of 300 km depth for the subducting slabs. All the parameters required to develop a seismic source model for earthquake hazard analysis were determined for crustal faults and subduction systems. A statistical distribution of relevant seismotectonic parameters, such as faulting mechanisms, slip rates, moment rates, and prospective maximum magnitudes, is presented and discussed to address unsettled points in view of future updates and improvements. The dataset, identified by the DOI https://doi.org/10.13127/efsm20, is distributed as machine-readable files using open standards (Open Geospatial Consortium). [ABSTRACT FROM AUTHOR]

In this paper, we gather and take stock of the results produced by the Operational Earthquake Forecasting (OEF) system in Italy, during its first 10 yr of operativity. The system is run in real-time: every midnight and after each M L 3.5+ event, it produces the weekly forecast of earthquakes expected by an ensemble model in each cell of a spatial grid covering the entire Italian territory. To evaluate the performance skill of the OEF-Italy forecasts, we consider here standard tests of the Collaboratory for the Study of Earthquake Predictability, which have been opportunely adapted to the case of the overlapped weekly OEF forecasts; then we also adopt new performance measures borrowed from other research fields, like meteorology, specific to validate alarm-based systems by a binary criterion (forecast: yes/no; occurrence: yes/no). Our final aim is to: (i) investigate possible weaknesses and room for improvements in the OEF-Italy stochastic modelling, (ii) provide performance measures that could be helpful for stakeholders who act through a boolean logic (making an action or not) and (iii) highlight possible features in the Italian tectonic seismic activity. [ABSTRACT FROM AUTHOR]

In this paper, we explore the use of seismicity data on a single-station basis in site response characterization. We train a supervised deep-learning model, SeismAmp, to recognize and separate seismic site response with reference to seismological bedrock (VS  = 3.45 km s−1) in a broad frequency range (0.2–20 Hz) directly from single-station earthquake recordings (features) in Japan. Ground-truth data are homogeneously created using a classical multistation approach—generalized spectral inversion at a total number of 1725 sites. We demonstrate that site response can be reliably separated from single-station seismograms in an end-to-end approach. When SeismAmp is tested at new sites in both Japan (in-domain) and Europe (cross-domain), it achieves the lowest standard deviation among all tested single-station techniques. We also find that horizontal-to-vertical spectral ratio (HVSR) is not the optimal use of single-station recordings. The individual components of each record carry salient information on site response, especially at high frequencies. However, part of the information is lost in HVSR. SeismAmp could lead to improved site-specific earthquake hazard prediction in cases where recordings are available or can be collected at target sites. It is also a convenient tool to remove repeatable site effects from ground motions, which may benefit other applications, for example, improving the retrieval of seismic source parameters. Finally, SeismAmp is trained on data from Japan, future studies could explore transfer learning for practical applications in other regions. [ABSTRACT FROM AUTHOR]

Our ability to forecast earthquakes and slow slip events is hampered by limited information on the current state of stress on faults. Ensemble data assimilation methods permit estimating the state by combining physics-based models and observations, while considering their uncertainties. We use an ensemble Kalman filter (EnKF) to estimate shear stresses, slip rates and the state θ acting on a fault point governed by rate-and-state friction embedded in a 1-D elastic medium. We test the effectiveness of data assimilation by conducting perfect model experiments. We assimilate noised shear-stress and velocity synthetic values acquired at a small distance to the fault. The assimilation of uncertain shear stress observations improves in particular the estimates of shear stress on fault segments hosting slow slip events, while assimilating observations of velocity improves their slip-rate estimation. Both types of observations help equally well to better estimate the state θ. For earthquakes, the shear stress observations improve the estimation of shear stress, slip rates and the state θ , whereas the velocity observations improve in particular the slip-rate estimation. Data assimilation significantly improves the estimates of the temporal occurrence of slow slip events and to a large extent also of earthquakes. Rapid and abrupt changes in velocity and shear stress during earthquakes lead to non-Gaussian priors for subsequent assimilation steps, which breaks the assumption of Gaussian priors of the EnKF. In spite of this, the EnKF still provides estimates that are unexpectedly close to the true evolution. In fact, the forecastability for earthquakes for the same alarm duration is very similar to slow slip events, having a very low miss rate with an alarm duration of just 10 per cent of the recurrence interval of the events. These results confirm that data assimilation is a promising approach for the combination of uncertain physics and indirect, noisy observations for the forecasting of both slow slip events and earthquakes. [ABSTRACT FROM AUTHOR]

The Every Earthquake a Precursor According to Scale (EEPAS) forecasting model is a space–time point-process model based on the precursory scale increase (⁠|$\psi $|) phenomenon and associated predictive scaling relations. It has been previously applied to New Zealand, California and Japan earthquakes with target magnitude thresholds varying from about 5–7. In all previous application, computations were done using the computer code implemented in Fortran language by the model authors. In this work, we applied it to Italy using a suite of computing codes completely rewritten in Matlab. We first compared the two software codes to ensure the convergence and adequate coincidence between the estimated model parameters for a simple region capable of being analysed by both software codes. Then, using the rewritten codes, we optimized the parameters for a different and more complex polygon of analysis using the Homogenized Instrumental Seismic Catalogue data from 1990 to 2011. We then perform a pseudo-prospective forecasting experiment of Italian earthquakes from 2012 to 2021 with M w ≥ 5.0 and compare the forecasting skill of EEPAS with those obtained by other time independent (Spatially Uniform Poisson, Spatially Variable Poisson and PPE: Proximity to Past Earthquakes) and time dependent [Epidemic Type Aftershock Sequence (ETAS)] forecasting models using the information gain per active cell. The preference goes to the ETAS model for short time intervals (3 months) and to the EEPAS model for longer time intervals (6 months to 10 yr). [ABSTRACT FROM AUTHOR]

Copyright of Progress in Earthquake Sciences is the property of China Earthquake Administration, Institute of Geophysics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Progress in Earthquake Sciences is the property of China Earthquake Administration, Institute of Geophysics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, the downhill jointed rock slope strengthened by anti slide pile is taken as the research object, and the numerical calculation model of slope is established by particle flow analysis software PFC2D. The failure mode of slope in the natural state and the deformation characteristics after reinforcement are analyzed. Through inputting the real seismic wave, the dynamic response and failure mode of downhill jointed rock slope and anti slide pile under earthquake are systematically studied. The results show that the failure mode of the rock slope with inclined joints in the natural state belongs to the leading edge traction type and the trailing edge vertical tensile crack. The internal tensile shear combination failure of the landslide slides along the direction of the inclined joints, which is consistent with the actual situation; The anti slide pile can reduce the adverse effect of the structural plane and maintain the overall stability of the slope; Under the action of earthquake load, the anti slide pile can ensure that the slope will not be damaged on a large scale, and reduce the adverse impact of the structure on the slope failure mode. The slope safety factor under different working conditions is discussed. This study is of great significance to the prediction of earthquake induced landslide risk and reinforcement design. [ABSTRACT FROM AUTHOR]

Point processes have been dominant in modeling the evolution of seismicity for decades, with the epidemic‐type aftershock sequence (ETAS) model being most popular. Recent advances in machine learning have constructed highly flexible point process models using neural networks to improve upon existing parametric models. We investigate whether these flexible point process models can be applied to short‐term seismicity forecasting by extending an existing temporal neural model to the magnitude domain and we show how this model can forecast earthquakes above a target magnitude threshold. We first demonstrate that the neural model can fit synthetic ETAS data, however, requiring less computational time because it is not dependent on the full history of the sequence. By artificially emulating short‐term aftershock incompleteness in the synthetic data set, we find that the neural model outperforms ETAS. Using a new enhanced catalog from the 2016–2017 Central Apennines earthquake sequence, we investigate the predictive skill of ETAS and the neural model with respect to the lowest input magnitude. Constructing multiple forecasting experiments using the Visso, Norcia and Campotosto earthquakes to partition training and testing data, we target M3+ events. We find both models perform similarly at previously explored thresholds (e.g., above M3), but lowering the threshold to M1.2 reduces the performance of ETAS unlike the neural model. We argue that some of these gains are due to the neural model's ability to handle incomplete data. The robustness to missing data and speed to train the neural model present it as an encouraging competitor in earthquake forecasting. Plain Language Summary: For decades, the Epidemic‐Type Aftershock Sequence (ETAS) model has been the most popular way of forecasting earthquakes over short time spans (days/weeks). It is formulated mathematically as a point process, a general class of statistical model describing the random occurrence of points in time. Recently the machine learning community have used neural networks to make point processes more expressive and titled them neural point processes. In this study we investigate whether a neural point process can compete with the ETAS model. We find that the two models perform similarly on computer simulated data; however, the neural model is much faster with large data sets and is not hindered if there is missing data for smaller earthquakes. Most earthquake catalogs contain missing data due to varying capability in our detection methods, therefore we need models that are robust to this missingness. We then find that the neural model outperforms ETAS on a new catalog for the 2016–2017 Central Apennines earthquake sequence, which through machine learning detection contains thousands of previously undetected small magnitude events. We argue that some of this improvement can in fact be explained by missing data. These results present neural point processes as an encouraging competitor in earthquake forecasting. Key Points: We construct a new machine learning variant of point processes for short‐term earthquake forecasting enhanced catalogsThe neural point process gains higher forecasting performance from the low magnitude data than epidemic‐type aftershock sequence (ETAS) and is faster to trainThis forecasting performance on the 2016 Central Italy sequence motivates continued development in this class of models [ABSTRACT FROM AUTHOR]

Abstract Foreshock detection before mainshock occurrence is an important challenge limiting the short-term forecasts of large earthquakes. Various models for predicting mainshocks based on discrimination of foreshocks activity have been proposed, but many of them work in restricted scenarios and neglect foreshocks and mainshocks out of their scope. In disaster prevention, it is often necessary to change the forecast period and the magnitude of target mainshocks. This paper presents a cluster-based statistical discrimination of foreshocks which is applicable all over Japan and adjustable with respect to forecasting time span and mainshock magnitudes. Using the single-link clustering method, the model updates the expanding seismic clusters and determines in real time the probabilities that larger subsequent events will occur. The foreshock clusters and the others show different trends of certain feature statistics with respect to their magnitudes and spatiotemporal distances. Based on those features and the epicentral location, a nonlinear logistic regression model is used to evaluate the probabilities that growing seismic clusters are foreshocks that will trigger mainshocks within 30 days. The log of odds is estimated between the foreshock clusters and other clusters for respective feature values as nonlinear spline functions from a Japanese hypocenter catalog for the period 1926–1999. Based on the estimated odds functions, foreshock clusters tend to have smaller differences in their two largest magnitudes, shorter time durations, and slightly longer epicentral distances within the individual clusters. Given a potential foreshock cluster, its mainshock magnitude can be predicted by the Gutenberg–Richter law over the largest foreshock magnitude. The timing of mainshock occurrences from foreshocks is also predicted by multiplying the portion of mainshocks within a shorter span from those within 30 days by the evaluated foreshock probabilities. The predictive performance of our model is validated by the holdout method using a Japanese hypocenter catalog before and after 2000. The evaluated foreshock probabilities are roughly consistent with the actual portion of foreshocks in the validation catalog and could serve as an alert for large mainshocks.

Geomagnetic vertical intensity polarization is a method with a clear mechanism, mature processing methods, and a strong ability to extract anomalous information in the quantitative analysis of seismogenic geomagnetic disturbances. The existing analyses of geomagnetic vertical intensity polarization are all based on the 5~100 s frequency band without refinement of the partitioning process. Although many successful results have been obtained, there are still two problems in the process of extracting anomalies: the geomagnetic anomalies that satisfy the determination criteria are still high in occurrence frequency; and the anomalies are distributed over too large an area in space, which leads to difficulties in determining the location of the epicenter. In this study, based on observations from western China, where fluxgate observation points are positioned in areas with frequent, densely distributed medium-strength earthquakes, we refined the frequency bands of geomagnetic vertical intensity polarization, recalculated the spatial and temporal evolution characteristics of geomagnetic disturbances before earthquakes, and improved the crossover frequency anomaly prediction index while promoting the application of the method in earthquake forecasting.

Machine Learning (ML) has developed rapidly in recent years, achieving exciting advancements in applications such as data mining, computer vision, natural language processing, data feature extraction, and prediction. ML methods are increasingly being utilized in various aspects of seismic engineering, such as predicting the performance of various construction materials, monitoring the health of building structures or components, forecasting their seismic resistance, predicting potential earthquakes or aftershocks, and evaluating the residual performance of post-earthquake damaged buildings. This study conducts a scientometric-based review on the application of machine learning in seismic engineering. The Scopus database was selected for the data search and retrieval. During the data analysis, the sources of publications relevant to machine learning applications in seismic engineering, relevant keywords, influential authors based on publication count, and significant articles based on citation count were identified. The sources, keywords, and publications in the literature were analyzed and scientifically visualized using the VOSviewer software tool. The analysis results will help researchers understand the trending and latest research topics in the related field, facilitate collaboration among researchers, and promote the exchange of innovative ideas and methods.

This document is a summary of a special issue of the journal Entropy titled "Complexity and Statistical Physics Approaches to Earthquakes." The issue contains 11 original scientific articles that explore the complexity of earthquakes and the use of statistical physics as a theoretical framework to understand their dynamics. The articles cover topics such as regional seismicity, early warning systems, machine learning approaches, empirical scaling relations, complex network approaches, and the study of earthquakes in relation to other natural complex systems. While progress has been made in understanding earthquakes, there are still many unanswered questions, and statistical physics remains a valuable tool for bridging the gap between microscopic laws and macroscopic behavior. The ultimate goal is to provide efficient earthquake forecasting to mitigate risk for people and infrastructure. [Extracted from the article]

The article introduces a recently released dataset called CSNCD, which stands for A Comprehensive Dataset of Chinese Seismic Network. The dataset is the largest and most comprehensive seismic dataset ever released by the National Earthquake Data Center in China. It includes over 1.3 million events and over 45 million annotations, covering a time span from 2009 to 2022. The dataset provides valuable information for data-driven studies, deep learning algorithms, and geoscientific research. It also offers seismic waveforms and annotation data in HDF5 and JSON formats for easy sharing and usage. [Extracted from the article]

INTRODUCTION: This research paper presents an innovative method that merges neural networks and random forest algorithms to enhance earthquake prediction. OBJECTIVES: The primary objective of the study is to improve the precision of earthquake prediction by developing a hybrid model that integrates seismic wave data and various extracted features as inputs. METHODS: By training a neural network to learn the intricate relationships between the input features and earthquake magnitudes and employing a random forest algorithm to enhance the model's generalization and robustness, the researchers aim to achieve more accurate predictions. To evaluate the effectiveness of the proposed approach, an extensive dataset of earthquake records from diverse regions worldwide was employed. RESULTS: The results revealed that the hybrid model surpassed individual models, demonstrating superior prediction accuracy. This advancement holds profound implications for earthquake monitoring and disaster management, as the prompt and accurate detection of earthquake magnitudes is vital for effective mitigation and response strategies. CONCLUSION: The significance of this detection technique extends beyond theoretical research, as it can directly benefit organizations like the National Disaster Response Force (NDRF) in their relief efforts. By accurately predicting earthquake magnitudes, the model can facilitate the efficient allocation of resources and the timely delivery of relief materials to areas affected by natural disasters. Ultimately, this research contributes to the growing field of earthquake prediction and reinforces the critical role of data-driven approaches in enhancing our understanding of seismic events, bolstering disaster preparedness, and safeguarding vulnerable communities.