Orals: Fri, 2 May | Room 0.96/97
Seismic velocities, particularly P-wave (Vp) and S-wave (Vs) are critical for imaging the Earth's interior and understanding geodynamic processes. While the basic spatial relationships between seismic velocity structures and seismogenic zones have been extensively discussed, they are often described in a simplistic and coarse manner. Systematic and statistical investigations of these relationships particularly within the shallow crust and uppermost mantle, remain scarce. Significant challenges for such analyses are that the shallow crust and upper mantle are geologically complex, characterized by heterogeneous lithologies, intricate thermal and mechanical properties, and variable fluid distributions. These factors lead to nonlinear and highly heterogeneous spatial distributions of the Vp and Vs, complicating the interpretation and modeling of their relationship with seismogenic zones. Moreover, multiple interpretive paths for the same phenomenon often introduce subjective biases, making objective quantification of these relationships challenging. This study aims to address these challenges by leveraging machine learning (ML) techniques to explore and quantify the non-linear and complex relationships between seismic velocity structures and seismogenic zones across the Japan Arc. Using two distinct three-dimensional seismic velocity datasets, we employed various ML models to enhance the robustness and reliability of our analyses. Our results demonstrate that variations in the spatial distribution of Vp and Vs—especially the Vp/Vs ratios, vertical gradients, and variance of Vp, Vs—serve as reliable indicators for distinguishing seismogenic zones from non-seismogenic zones across both depth and geographic space even though the tectonic settings vary significantly. To interpret the complex nonlinear patterns revealed by ML models, we employed Shapley Additive Explanations (SHAP), which elucidated the spatial relationship between seismic velocities and seismogenic zones. By examining local seismogenic zones, The results by SHAP found factors influencing seismogenic zones differ with depth: at shallow depths, Vp, Vs, Vp/Vs ratio, and variance of Vp, Vs are dominant, while at greater depths, gradient changes are primary. It may relate to the thermal structure and indicate different triggering mechanisms for earthquakes at various depths. These findings can provide deeper insights into the spatial coupling between seismic velocities and seismicity, thereby advancing our understanding of the factors controlling earthquake generation.
How to cite: Zhang, C., Yoshiya, U., Yeo, T., Kato, A., and Iwamori, H.: Relationships Between Seismic Velocity Structures and Seismogenic Zone Decoded by Interpretable Machine Learning, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5005, https://doi.org/10.5194/egusphere-egu25-5005, 2025.
The aftershock sequence typically consists of numerous seismic events, with their distribution exhibiting clustering characteristics. In geologically complex areas, such as the convergence boundary between the Eurasian and Philippine Plate in eastern Taiwan, it is challenging to explain the relationship between seismic events and regional structures. This area has experienced several disastrous earthquakes in recent years, including the Hualien earthquake (Mw 6.4) in 2018, the Chihshang earthquake (Mw 7.0) in 2022, and the Hualien earthquake (Mw 7.3) in 2024. Here, we aim to explore the relationship between the aftershock sequences of three events and the known active faults. Firstly, we apply the algorithm to cluster aftershocks. We analyze aftershock sequences for three events with local magnitudes greater than 3, spanning 45 days after the mainshock. Secondly, we examine the relationship between these clustered sequences and the 3D fault models developed by National Science and Technology Center for Disaster Reduction (NCDR). The results reveal that the aftershock sequence of the 2018 Hualien earthquake can be divided into five clusters, while the 2022 Chihshang earthquake can be divided into seven clusters. The mainshocks are separately located at clusters which have the largest number of aftershocks within their respective sequences. The aftershock sequence of the 2024 Hualien earthquake can be divided into eight clusters. The mainshock is located at a cluster with minor number of aftershocks, which is distributed along the Lingding Fault. Additionally, 3D visualization is employed to better illustrate the relationship between earthquake sequences and active faults, as well as to study potential earthquake mechanisms.
How to cite: Tu, K.-T., Huang, M.-W., Yang, C.-Y., Ke, M.-C., and Ke, S.-S.: The Relationship between Clustered-Aftershocks and 3D-Fault Models in Eastern Taiwan, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-1767, https://doi.org/10.5194/egusphere-egu25-1767, 2025.
Spatiotemporal variations of the b-value (the slope of the Gutenberg-Richter relation) are commonly associated to many disparate factors, such as critical stress conditions, the tectonic regime, and the incompleteness of earthquake catalogs.
During the 2016/17 central Italy earthquake sequence, several studies reported notable b-value variations, including an unexpected increase prior to the Norcia event (Mw 6.5). Such observations highlighted the complex relationship between b-value changes and seismic activity.
To get a better understanding of this relation, we reanalyze this sequence with a focus on the spatiotemporal evolution of seismicity near the Norcia mainshock. We temporally divided the seismic sequence into subperiods separated by the largest events and employ a combination of three machine-learning algorithms: DBSCAN for performing event clustering, OPTICS for analyzing spatially nested dense zones within clusters and PCA for inferring the planar geometry of those zones as fault surfaces. We identified two specific zones and reconstruct two separate fault planes. Those two zones exhibited asynchronous activity before the mainshock. We used the two-sample Kolmogorov-Smirnov Test to investigate similarity between the magnitude-frequency distribution of earthquakes associated with these planes. The results show that the b-values associated with these fault planes remained stable over time. Yet, their temporal changes exhibited a correlation with spatial variations of seismicity. In particular, the analysis indicated a relationship between the b-value and the geometry of the active fault.
These findings suggest that temporal variations of the b-value during an earthquake sequence may not necessarily reflect changes in underlying stress conditions, but rather the activation of different earthquake sources throughout the sequence, each with different lithological and geometrical properties. This finding highlights the importance of understanding the fine-scale structure of earthquake sources in a sequence for correctly interpreting b-value variations.
How to cite: Piegari, E., Corrado, P., Herrmann, M., and Marzocchi, W.: Investigating the complex relationship between b-value changes and seismic activity in Central Italy, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13370, https://doi.org/10.5194/egusphere-egu25-13370, 2025.
This study investigates the different phases of seismic cycles present in earthquake sequences recorded in Chile and Italy on the basis of variations in the probability model for the magnitude and for the spatial distribution of the epicenters. Chile and Italy are two countries that, despite having different tectonic characteristics, face significant challenges due to seismic activity.
The seismic records are analyzed on sliding time windows with a fixed number of events, which shift with each new earthquake. Two probabilistic models are proposed for earthquake magnitude: one is based on the q-exponential distribution, hereafter referred to as the “q-exponential” model for simplicity, and the other is the exponential distribution, which is well known to be consistent with the Gutenberg-Richter law (Rotondi et al., Geophys. J. Int., 2022). The q-exponential distribution is closely related to Tsallis entropy, a generalized form of entropy that accounts for non-extensive systems, and has been widely applied in statistical mechanics to study complex systems. It is characterized by the parameter q, and as q asymptotically approaches 1, it reduces to the exponential distribution. As for the spatial distribution of the earthquakes, we consider the cell areas of the Voronoi tessellation generated by epicenters and adopt four probability models: the q-exponential, exponential, generalized gamma, and tapered Pareto distributions (Rotondi & Varini, Front. Earth Sci., 2022).
By following the Bayesian approach, the posterior distribution of model parameters is estimated by a Markov chain Monte Carlo method based on the Metropolis-Hastings algorithm, and then the optimal distribution in each time window is selected by comparing the estimated values of the posterior marginal log-likelihood. Given the high computational cost, parallel programming has been chosen to drastically reduce the computational time from days to hours or even minutes.
The results obtained in the study areas agree in associating seismic phase changes to the variations of the estimated q-index in the magnitude case and of the best probability model as for the spatial distribution; this provides useful indications for the implementation of risk mitigation actions. In both study areas, despite their different tectonic behaviors, we obtain similar results for seismic sequences characterized by a strong earthquake preceded by foreshocks. During the foreshock activity (i.e., the preparatory phase leading to the strong event), we observe an increase in the q parameter of the magnitude distribution and a preference for the tapered Pareto model for the spatial distribution of the epicenters provided by the Voronoi cell areas.
This work is supported by: ICSC National Research Centre for High Performance Computing, Big Data and Quantum Computing (CN00000013, CUP B93C22000620006) within the European Union-NextGenerationEU program; Chilean National Agency for Research and Development (ANID), Fondecyt grant ID 1241881; Research Center for Integrated Disaster Risk Management (CIGIDEN), ANID/FONDAP/1523A0009.
How to cite: González, A., Varini, E., Rotondi, R., Nicolis, O., and Ruggeri, F.: Exploring seismic cycle dynamics via variations in probability models: Chile and Italy case studies, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9765, https://doi.org/10.5194/egusphere-egu25-9765, 2025.
The field of earthquake prediction presents significant challenges, with effective prediction techniques are still elusive. Taiwan’s high seismicity is due to the collision of the Philippine Sea plate with the Eurasian plate, leading to frequent earthquakes every year. To address challenges in analyzing pre-earthquake strain transfer, number of radon monitoring stations were established across various tectonic zones. Using open-source software a Real-Time Database was developed for radon earthquake precursory research and tested for some major earthquakes occurred in Taiwan. This database enables faster processing of precursor data, hence, enhancing the efficiency related investigations.
Notably, large earthquakes, such as Meinong earthquake in Southern Taiwan, exhibited precursory signals in radon concentrations, with significant variations observed in soil radon concentrations about two weeks prior. The study suggests that variations in soil radon concentrations at different locations may help in predicting the general area of impending major earthquakes. Finding from research indicates that soil-gas anomalies were associated with few earthquakes with a magnitude of 5 or more that were recorded during the study period. Stress accumulation and changes in strain fields prior to seismic events are likely linked to these variations in soil radon levels.
Overall, soil radon measurements have emerged as a promising and practical tool for investigating earthquake precursors in Taiwan. However, further research and validation are essential to refine these findings and advance the development of reliable earthquake prediction method.
How to cite: Walia, V., Fu, C.-C., Lin, S.-J., and Kamra, P.: Radon Monitoring Data in Taiwan: Statistical Perspective to Investigate Earthquake Precursory Studies, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-16828, https://doi.org/10.5194/egusphere-egu25-16828, 2025.
Analysis of seismic precursors is crucial for understanding the spatio-temporal evolution of seismicity and assessing whether a system is approaching an unstable state. Precursors are indicators that are deemed to be related to the processes leading to crustal rupture. Therefore, their real-time monitoring can provide insights into the imminent occurrence of earthquakes. Precursors yield robust results only when analyzed using appropriate techniques. Specifically, the measurement of real-time precursor parameters and the analysis of their temporal trends is highly sensitive to data processing and depends heavily on the characteristics of the seismic data under study. Therefore, careful data management is essential to avoid inappropriate conclusions.
This study examines two seismic precursors: (1) b-value (i.e., slope of the Gutenberg and Richter law), which characterizes the relative likelihood of small versus large earthquakes within a population of events; (2) Hurst exponent, an indicator of the "memory" in time series and, consequently, of the type of stochastic process underlying them (i.e., random, persistent, or anti persistent). While the b-value has paramount importance in earthquake forecasting since its variation (which is deemed to be related to stress conditions of faults) can act as a first-order discriminator between conventional aftershock sequences and sequences including multiplets (i.e., two or more mainshocks that are closely associated in time and space), the Hurst exponent is widely used in econometrics to detect trends and mean reversion in financial data.
The aim of this study is to determine the optimal data-windowing configuration (window size and overlapping percentages), within the framework of a moving window approach, that produces results in agreement with theoretical expectations and effectively captures the characteristics of the seismic data under study. The methodology involves analyzing these precursors (i.e., b-value and Hurst Exponent) along with additional seismic metrics such as: (1) number of earthquakes above a given magnitude threshold, (2) maximum magnitude, and (3) strain energy. Correlation coefficients (e.g., Pearson, Spearman, Kendall) are computed to evaluate the relationships between these parameters under various windowing configurations, including a novel adaptive window approach. The process can be summarized as follows:
- For a given configuration (e.g., window size = 500 years, overlap = 50%), the window slides over time, and parameters are calculated at each step within the time window.
- Correlation coefficients (between pairs of parameters) are computed using various statistical methods (e.g., Pearson, Spearman, Kendall).
- This procedure is repeated across all configurations to identify the setting that maximizes correlations (i.e., higher correlation coefficients and lower p-values).
Analyzing synthetic time series shows that correlations between parameters are sensitive to the adopted configuration, as different data-windowing configurations may capture distinct seismicity patterns. Therefore, the selection of the most effective configuration is strictly study-specific. To enhance the reliability of the results, application of this methodology to other seismic parameters (e.g., Vp/Vs ratio) requires future consideration, especially to validate results against known seismic sequences. The first step towards this direction is the application of the method to time series associated with natural or induced (or triggered) seismicity (e.g., seismic activity in the Geysers geothermal field).
How to cite: Azhideh, S., Barani, S., Ferretti, G., Taroni, M., Resta, M., and Massa, M.: Optimization of rolling window approach to analyze earthquake time series and identify possible precursors, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17245, https://doi.org/10.5194/egusphere-egu25-17245, 2025.
This study investigates whether the 7.8-magnitude earthquake that occurred on February 6, 2023, in Kahramanmaraş, Türkiye, could have been predicted using advanced machine learning techniques. Additionally, it aims to assess the likelihood of similar devastating earthquakes occurring in this region in the future. Addressing these questions serves as the primary motivation for this research, with the goal of improving our understanding of seismic hazards and enhancing predictive capabilities for better disaster preparedness. By combining existing research findings with innovative predictive features, the study developed a meticulously crafted feature matrix to evaluate the capability of machine learning algorithms in forecasting such high-magnitude seismic events. Accurate earthquake prediction is crucial for developing early warning systems, disaster planning, and seismic risk assessments. The analysis utilized instrumental records of 36933 earthquakes (Md≥1) that occurred within a circular area of a 100-km radius, centered at 37.288⁰ latitude and 37.043⁰ longitude, spanning the period from August 30, 1908, to September 30, 2024. The data were obtained from Boğaziçi University, Kandilli Observatory and Earthquake Research Institute, Regional Earthquake-Tsunami Monitoring Center (KOERI-RETMC). The compiled catalogue includes various magnitude scales (Ms: surface wave magnitude, Md: duration magnitude, MLM_LML: local magnitude, Mb: body wave magnitude, and Mw: moment magnitude), along with origin time, epicenter, and depth information.
To create a homogeneous catalogue, a conversion equation between moment magnitude (Mw and other scales (Md, ML, Mb, Ms, M) was determined using the general orthogonal regression method. Depth parameters were analyzed to exclude artificial events, and the final magnitude range was between 1 and 7.8, with depths ranging from 1 to 40 km. The most reliable conversion equation was obtained for ML and Mw as 1,887 events had both ML and Mw magnitudes. The derived conversion equation is: is Mw*=1.00005*ML+(-0.06440), R2 =0.97473.
Machine learning models-including Linear Regression, Support Vector Machines, Naïve Bayes, and Random Forest-were applied to both the uniform catalogue and inhomogeneous catalogue. The results revealed a significant difference in earthquake patterns for events with magnitudes less than 6 before and after modeling. These findings indicate that in regions with high seismic activity, modeling efforts can provide more reliable insights into the spatial distribution and magnitude of seismicity. Among the machine learning algorithms tested, the Random Forest model demonstrated the best performance, achieving the highest accuracy in predicting the maximum earthquake magnitude category within a 30-day timeframe. While predicting extreme-magnitude seismic events remains a significant challenge, the findings highlight the potential of data-driven approaches to enhance seismic risk management and preparedness. The methodology developed in this study offers valuable insights and practical applications for Turkey's Eastern Anatolian Fault and other seismically active regions.
How to cite: HajiHasanli, M., Polat, G., and Altinoglu, F.: Predicting Earthquakes in The Eastern Anatolian Region Using Machine Learning Algorithms, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19273, https://doi.org/10.5194/egusphere-egu25-19273, 2025.
How to cite: Zhuang, J.: Evaluating Earthquake Forecast with Likelihood-Based Marginal and Conditional Scores, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-21558, https://doi.org/10.5194/egusphere-egu25-21558, 2025.
Probabilistic earthquake forecasting has made significant strides in the past decades, to the degree that government agencies around the world have implemented public, real-time systems. The underlying models are largely parametric and statistical, and comprise variants of the self-exciting Hawkes point process, such as the popular Epidemic Type Aftershock Sequence (ETAS) model. ETAS models have gained trust also as a result of their good relative performance in prospective forecast experiments by the Collaboratory for the Study of Earthquake Predictability (CSEP, cseptesting.org) in various tectonic settings around the globe. In recent years, however, machine learning variants of point processes have become available that offer significant advantages: they are much more flexible in their probabilistic description of earthquake interaction, and they are much faster. In this talk, I will review recent applications of neural point processes to seismicity forecasting around the world, which demonstrate distinct advantages and some (moderate) improvement in predictive skill. I will also argue that a clear community benchmarking process is required to make transparent and robust progress. Finally, I will present ongoing model enhancements of neural point processes and preliminary results from benchmarking in California. Machine learning has the potential of transform earthquake forecasting, but progress must be demonstrated in a robust and transparent manner.
How to cite: Werner, M., Stockman, S., and Lawson, D.: Advancing Earthquake Forecasting with Machine Learning , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19183, https://doi.org/10.5194/egusphere-egu25-19183, 2025.
Posters on site: Fri, 2 May, 10:45–12:30 | Hall X3
This study presents a comprehensive analysis of seismicity and tectonic activity in Muş Province, one of Eastern Turkey's most seismically active regions, due to its position along critical tectonic boundaries. The study provides valuable insights into the region's seismic hazards and tectonic dynamics by examining seismic activity, fault mechanisms, and earthquake recurrence intervals.
Earthquake catalogs from national and international sources—Disaster and Emergency Management Presidency (AFAD), Boğaziçi University Kandilli Observatory and Earthquake Research Institute – Regional Earthquake-Tsunami Monitoring Center (B.U. KOERI-RETMC), and the United States Geological Survey (USGS)—were analyzed to investigate the spatial and temporal distributions of earthquakes. Magnitude-frequency distributions were modeled using the Gutenberg-Richter law to estimate earthquake occurrence probabilities and recurrence intervals. Spatial variations in stress accumulation were visualized through high-resolution b-value maps. Notably, consistently low b-values were observed across the region, except for its northeastern part, indicating high-stress accumulation and an elevated potential for significant seismic activity in these zones.
Historical earthquake data from the European Archive of Historical Earthquake Data (AHEAD), the Share European Earthquake Catalog (SHEEC), and AFAD were incorporated into the analysis to provide a long-term perspective on seismic activity. Focal mechanism solutions were compiled from diverse sources, including AFAD, B.U. KOERI-RETMC, the GEOFON data center of the GFZ German Research Centre for Geosciences, the Global Centroid-Moment-Tensor (GCMT) project catalogs, and moment tensor inversions were conducted within this study. These solutions facilitated a detailed characterization of faulting styles and stress orientations, offering critical insights into the tectonic forces shaping the region.
The findings reveal the region's complex fault dynamics and significant spatial heterogeneities in stress distribution and clustering patterns. These results underscore the importance of enhanced seismic monitoring and targeted preparedness efforts in high-risk areas. By integrating historical and recent seismic data with robust statistical and physical models, this study makes a substantial contribution to seismic hazard assessment and establishes a foundation for future research. Potential extensions include incorporating machine learning techniques, microseismicity analysis, and geodetic data integration to refine hazard models tailored to the unique tectonic environment of Muş Province.
How to cite: Tamtaş, B. D.: Seismicity and Tectonic Insights of Muş Province, Eastern Turkey, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6322, https://doi.org/10.5194/egusphere-egu25-6322, 2025.
Investigating the seismicity of a fault system and its geometric and kinematic characteristics is of paramount importance for mitigating the associated seismic risk. In particular, the characterization of microseismicity can both reveal the presence of unknown or blind fault segments and provide important insights into the evolution of seismicity during the occurrence of a strong seismic sequence.
In this work, we have studied a seismic sequence that recently occurred near Benevento (southern Italy), characterized by a main earthquake of magnitude 3.8, which took place in a region with the highest seismic hazard in Italy. This sequence, characterized by significant activity between November 2019 and January 2020, allowed the identification of a previously unknown fault segment. By installing eight 3-C velocimeter stations in a 12 km radius around the epicenter and applying a template matching technique, we have been able to detect a significant number of events that allowed us to generate an enhanced catalog as compared to those provided by the national and local permanent networks. The augmented catalog consists of hundreds of earthquakes with a minimum magnitude of -0.9. Earthquake relocations of the seismic sequence were achieved by computing differential P- and S-wave travel times and by using a double-difference algorithm.
Our results, combined with the estimation of focal mechanisms for the strongest earthquakes, allow us to identify a fine-scale fault structure consisting of several small segments with strike-slip kinematics between 10 and 15 km depth. Integrating these results with the calculation of source parameters and the analysis of the spatio-temporal distribution of the sequence will enhance our understanding of the mechanical and kinematic characteristics of this complex fault structure. Moreover, the approach followed in our work holds significant potential for analyzing microseismicity and defining complex fault geometries in high seismic risk regions.
How to cite: Brondi, P., Adinolfi, G. M., De Matteis, R., Riccio, R., Scotto Di Uccio, F., Festa, G., Picozzi, M., and Zollo, A.: Revealing a complex blind fault structure by the analysis of 2019-2020 San Leucio del Sannio seismic sequence (Southern Italy), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-21486, https://doi.org/10.5194/egusphere-egu25-21486, 2025.
The identification of clusters is crucial for the statistical analysis of seismicity and the forecasting of earthquakes, because discrepancies in the methods used to identify clusters can lead to inconsistent results. In this work, the seismic activity in Molise, southern Italy, from April to November 2018 is analyzed as a case study. The focus is on how such discrepancies can affect forecasting algorithms such as NExt STrOng Related Earthquake (NESTORE), which are designed to forecast strong aftershocks following a first strong event
A detailed analysis was performed using an improved template matching catalog and a comparative evaluation of clustering methods, including window-based analysis techniques, Nearest Neighbor, and fractal dimension. Probabilistic information was integrated through the Epidemic Type Aftershock Sequence (ETAS) model.
Significant differences in cluster definition required further analysis, including principal component analysis (PCA) and ETAS modeling, to investigate spatiotemporal seismic patterns. The main results show an upward migration of seismicity, an extended duration of the sequence and relative quiescence between stronger events, all suggesting fluid-driven mechanisms. These observations suggest that the presence of fluids plays a crucial role in the sequence dynamics and the discrepancies between clustering methods.
The study highlights the importance of refining approaches to cluster identification, incorporating physical and geological factors, and encourages further investigation of anomalous seismic sequences such as the 2018 seismic cluster in Molise. The results also highlight the influence of fluids on seismicity in the Apennines and call for advanced analytical methods to improve the accuracy of strong events forecasting.
Funded by a grant from the Italian Ministry of Foreign Affairs and International Cooperation and Co-funded within the RETURN Extended Partnership and received funding from the European Union Next-Generation EU (National Recovery and Resilience Plan—NRRP, Mission 4, Component 2, Investment 1.3—D.D. 1243 2/8/2022, PE0000005) and by the and the grant “Progetto INGV Pianeta Dinamico: NEar real-tiME results of Physical and StatIstical Seismology for earthquakes observations, modelling and forecasting (NEMESIS)” - code CUP D53J19000170001 - funded by Italian Ministry MIUR (“Fondo Finalizzato al rilancio degli investimenti delle amministrazioni centrali dello Stato e allo sviluppo del Paese”, legge 145/2018).
How to cite: Gentili, S., Brondi, P., Rossi, G., Sugan, M., Petrillo, G., Zhuang, J., and Campanella, S.: Fluid diffusion and seismic clusters identification: results on the seismicity of Molise (southern Italy) in 2018, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7945, https://doi.org/10.5194/egusphere-egu25-7945, 2025.
The earthquake generation process is a complex phenomenon, manifested in the nonlinear dynamics and in the wide range of spatial and temporal scales that are incorporated in the process. Despite the complexity of the earthquake generation process and our limited knowledge on the physical processes that lead to the initiation and propagation of a seismic rupture giving rise to earthquakes, the collective properties of many earthquakes present patterns that seem universally valid. The most prominent is scale-invariance, which is manifested in the size of faults, the frequency of earthquake sizes and the spatial and temporal scales of seismicity. The frequency magnitude distribution exhibits a decay that is commonly expressed with the well-known Gutenberg-Richter (G-R) law. The aftershock production rate following a main event generally decays as a power-law with time according to the modified Omori formula. Scale-invariance and (multi)fractality are also manifested in the temporal evolution of seismicity and the distribution of earthquake epicentres. The organization patterns that earthquakes and faults exhibit have motivated the statistical physics approach to earthquake occurrence. Based on statistical physics and the entropy principle, a unified framework that produces the collective properties of earthquakes and faults from the specification of their microscopic elements and their interactions, has recently been introduced. This framework, called nonextensive statistical mechanics (NESM) was introduced as a generalization of classic statistical mechanics due to Boltzmann and Gibbs (BG), to describe the macroscopic behaviour of complex systems that present strong correlations among their elements, violating some of the essential properties of BG statistical mechanics. Such complex systems typically present power-law distributions, enhanced by (multi)fractal geometries, long-range interactions and/or large fluctuations between the various possible states, properties that correspond well to the collective behaviour of earthquakes and faults. Here, we provide an overview on the fundamental properties and applications of NESP. Initially, we provide an overview of the collective properties of earthquake populations and the main empirical statistical models that have been introduced to describe them. We provide an analytic description of the fundamental theory and the models that have been derived within the NESP framework to describe the collective properties of earthquakes. The fundamental laws of Statistical seismology as that of Gutenberg-Richter (GR) and Omori law a analysed using the ideas of Tsallis entropy and its dynamical superstatistical interpretation offered by Beck and Cohen.
How to cite: Vallianatos, F.: Gutenberg-Richter, Omori and Cumulative Benioff strain patterns in terms of non extensive statistical physics and Beck-Cohen Superstatistics., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13653, https://doi.org/10.5194/egusphere-egu25-13653, 2025.
Slow-slip events (SSEs) are episodic fault slip phenomena that involve the gradual and aseismic release of tectonic stress, bridging the gap between the rapid rupture of regular earthquakes and the steady sliding along fault interfaces. SSEs are common in megathrusts, having been observed in most of the well geodetically-instrumented subduction margins worldwide, both on the shallow plate interface (less than 10 km depth) and on the deeper plate interface (25–60 km). We explore the relations between the occurrence of SSEs and various subduction parameters along megathrusts at a global scale. Using a parametric approach, we applied three Machine Learning (ML) algorithms to predict the presence (or absence) of shallow and deep SSEs, modeling it as a nonlinear function of subduction variables. The subduction parameters considered include subducting plate age and roughness, sediment thickness, slab dip, convergence rate and azimuth, distance to the nearest ridge or plate boundary, maximum observed magnitude, b-value and earthquake rates, among others. We then employed Shapley Additive exPlanations (SHAP) on the ML outcomes, to identify the most influential factors associated with SSE occurrence. Preliminary analysis and previous studies suggest that plate age, slab dip, and b-value are among the most critical variables. These observations point to the possibility that the frictional properties of the subducting plate, which influence plate coupling and stress levels, may play a key role in controlling the occurrence of shallow, deep, or both types of SSEs. Our study provides valuable insights into the complex, nonlinear processes governing SSEs on a global scale and highlights regions where previously undetected SSEs may be occurring.
How to cite: Arroyo Solórzano, M., Crisosto, L., Jara, J., González, Á., and Cotton, F.: Linking subduction parameters to the occurrence of slow slip events using machine learning on a global scale, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7255, https://doi.org/10.5194/egusphere-egu25-7255, 2025.
This study explores the integration of advanced artificial intelligence (AI) techniques into the seismic monitoring framework of the Kandilli Observatory and Earthquake Research Institute (KOERI), enhancing the accuracy and reliability of seismic event detection, location, and magnitude determination. The implementation leverages graph neural networks (GNNs) for seismic phase association and location problems, alongside pretrained AI models for phase picking. GNN is trained using datasets from both the Marmara and Maraş regions, and the resulting AI-based earthquake catalogs are compared against KOERI's legacy catalogs to assess performance and reliability.
Key innovations include:
- Application of GNNs to capture spatial and temporal relationships in seismic networks for improved event association.
- Enhanced phase picking and hypocenter localization accuracy, reducing uncertainty in earthquake catalogs.
Preliminary results indicate significant improvements in detecting low-magnitude events, reducing processing latency, and generating consistent and reliable earthquake catalogs. These advancements allow KOERI to provide high-resolution, AI-processed seismic data and earthquake catalogs, offering the seismological community access to more comprehensive and reliable seismic information while contributing to global research efforts.
The presentation will discuss the technical challenges encountered during integration and compare the new system's performance metrics to traditional methods used by KOERI. It will also explore the implications for future seismic monitoring practices.
How to cite: Meral Özel, N., Diner, Ç., Ata, E., Turhan, F., Güneş, Y., Aksarı, D., Yılmazer, M., Akça, M. E., Şahin, A., and Kalem, B.: Artificial Intelligence-Driven Seismic Event Detection and Association for Enhanced Monitoring at KOERI, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11172, https://doi.org/10.5194/egusphere-egu25-11172, 2025.
All observed giant (≥ MW 8.5) earthquakes have occurred at subduction margins. Due to their long intermittence times, our instrumental and historical earthquake catalogues only contain a handful of giant earthquake occurrences, with no observations at all for some margins. This raises the question whether giant earthquakes may occur at all subduction margins or whether their nucleation requires a certain set of geological properties, which may be present at only some margins.
Since the 1980s, numerous studies have focused on the search for a subduction margin property enabling giant earthquakes, with parameters such as sediment thickness, subducting plate age and hydration, seafloor roughness, convergence rate, and dip steepness amongst the most debated, many of them with contradicting hypotheses. Recent years have brought the hypothesis that giant earthquake occurrence may depend on a combination of margin properties to the forefront, with several studies taking multivariate statistics approaches to relating the two. These approaches are however limited by the incomplete nature of earthquake catalogues, specifically regarding giant earthquakes.
We present an unsupervised approach to examining the connections between margin properties and seismicity, which allows us to uncover patterns in margin property data, excluding any earthquake occurrence data from the incomplete record. Considering sediment thickness, convergence rate, dip angle, and different measures of seafloor roughness, we “fingerprint” margin segments by applying Principal Component Analysis (PCA) to margin property data. Based on these “fingerprints”, we quantify similarity between the margins’ property combinations, and group them into different hazard groups regarding the possibility of giant earthquake occurrence. Using Kernel-PCA, a non-linear PCA variant, reveals non-linear patterns in margin properties, prompting us to suggest that connections between margin properties and seismicity are non-linear. Finally, we apply this method to characterise the seismic behaviour of subduction zones where seismic activity is less well-documented, such as the Makran, Hellenic, and Lesser Antilles margins.
How to cite: Locher, V., Bell, R., Salah, P., Platt, R., and John, C.: Fingerprinting Subduction Margins: An Unsupervised Learning Approach for Earthquake Hazard Assessment , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9575, https://doi.org/10.5194/egusphere-egu25-9575, 2025.
The Campi Flegrei caldera, a high-risk volcanic region in southern Italy, is currently experiencing an unrest phase characterized by significant ground deformation and increasing seismic activity, including events with magnitudes up to Md 4.2 recorded in September 2023. The continuous availability of seismic data provides a valuable framework for evaluating and developing novel event detection methods. However, in regions characterized by extensive natural and anthropogenic noise, the resulting low signal-to-noise ratio poses a significant challenge to seismic event detection. This limitation is sharpened when analyses are based on data from a single seismic station.
To address this challenge, the present study introduces an innovative methodology designed for single-station analysis that combines the Multiscale Entropy (MSE) algorithm with Self-Organizing Maps (SOM) and the Short-Term Average/Long-Term Average (STA/LTA) technique for seismic signal detection and clustering. Linear Predictive Coding (LPC) algorithm is also employed in conjunction with the SOM map for a preliminary stage to certify the quality of the data and check for anomalies.
The analysis uses continuous seismic data recorded over six months in the Pisciarelli area of the Campi Flegrei caldera, segmented into one-minute windows. Key features, including STA/LTA ratios (computed with 1s and 30s windows) and MSE values (computed over 20 time scales using a coarse-graining operation), are extracted to encode the input vectors for SOM training. The resulting 6x6 SOM map effectively clusters the seismic traces, revealing hidden patterns and distinguishing seismic events from background seismic noise. Notably, approximately 20% of the transient signals within the nodes of the seismic event cluster were identified as uncatalogued events, demonstrating the ability of the method to detect previously unrecorded activity. In addition, the map includes different clusters that highlight the influence of environmental factors, such as precipitation occurrences or volcanic fluid emissions, on the seismic waveforms.
The integration of complexity-based analysis of the MSE alongside conventional STA/LTA techniques enables improved single-station event detection, even in a noisy environment, and hints at the correlation between seismic signal complexity and volcano dynamics. These results highlight the potential of advanced clustering and feature extraction techniques to refine seismic monitoring in active volcanic environments.
How to cite: Grimaldi, A., Amoroso, O., Napolitano, F., Convertito, V., Galluzzo, D., Scarpetta, S., Messuti, G., Gaudiosi, G., Nardone, L., and Capuano, P.: SOM-based approach for seismic data analysis in the Campi Flegrei Caldera using Multiscale Entropy (MSE), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11838, https://doi.org/10.5194/egusphere-egu25-11838, 2025.
The advent of artificial intelligence (AI) has opened new avenues in geosciences, particularly for earthquake prediction. Deep learning models, especially Convolutional Neural Networks (CNNs), offer promising capabilities to analyze complex data and detect subtle patterns indicative of seismic activity. However, geophysical records span a time interval that is shorter that the duration of large eartquakes cycle, creating a major challenge for training these models.
In this study, we use paleoseismological data, which include multiple seimic cycles (Cascadia and Sumatra zones). Paleoseismological data are often represented as barcodes, where each "bar" represents an earthquake in time and space. We reproduce these barcodes using a scaled seismotectonic model mimicking subduction megathrust earthquake cycles. The simulated sequences include both partial and complete ruptures, representing earthquakes of varying magnitudes. A CNN model is then trained with these barcodes to predict the timing, location along the margin, and magnitude of the next earthquake.
Our results show that the CNN model can reconstruct the complex temporal loading history and accurately predict the timing of future earthquakes. This approach overcomes the limitations of conventional methods based on slip deficit and highlights the potential of paleoseismological data to enhance seismic forecasting strategies.
This work demonstrates the application of deep learning techniques to paleoseismological data as a tool for earthquake prediction. It opens promising perspectives for seismic hazard assessment and the understanding of fault cycles in subduction zones.
How to cite: Visage, S., Corbi, F., Guastamacchia, S., and Funiciello, F.: Earthquake Prediction from Paleoseismology: a proof of concept based on CNN analysis of data from scaled sismotectonic models, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6345, https://doi.org/10.5194/egusphere-egu25-6345, 2025.
Predicting earthquakes remains a challenging task due to the implicit and nonlinear nature of seismic activity. Traditional methods often struggle to capture the complex patterns underlying seismic events. This study investigates the application of deep learning techniques, specifically a hybrid model combining Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN), to enhance earthquake prediction accuracy.
The proposed approach involves six key stages: (1) collection of historical seismic activity data, (2) comprehensive data preprocessing, (3) exploratory data analysis to identify underlying patterns, (4) characterization using seismicity indicators, (5) implementation of the hybrid neural network model, and (6) evaluation of the model's performance in recognizing significant seismic trends. The model leverages diverse datasets encompassing geological and seismic characteristics to enhance its robustness.
Experimental results reveal that the hybrid LSTM-CNN model achieves superior predictive accuracy, evidenced by a Mean Squared Error (MSE) of 0.62 and a Coefficient of Determination (R²) of 0.91. The LSTM component effectively captures temporal dependencies in the time-series data, while the CNN component identifies spatial features and seismic patterns. This dual capability significantly outperforms conventional approaches and provides deeper insights into earthquake dynamics.
The findings of this study not only contribute to advancing earthquake prediction methodologies but also highlight the potential for transitioning towards image-based seismic data analysis. Such advancements open new opportunities for integrating machine learning with geophysical sciences to improve disaster preparedness and mitigation efforts.
How to cite: Nurtas, M., Zhumabayev, B., Altaibek, A., and Aigerim, K.: A Hybrid Deep Learning Approach with LSTM and CNN for Enhanced Earthquake Prediction, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15124, https://doi.org/10.5194/egusphere-egu25-15124, 2025.
