Accurate ground‐motion simulations are essential for seismic hazard assessments and engineering practices. Herein, we propose a novel method combining conditional generative adversarial networks (cGANs) and the generalized inversion technique (GIT) to generate site‐specific and variability‐controlled strong‐motion seismograms. The cGANs calculate synthetic seismogram without amplitude scales. The GIT is to separate the source, path, and site characteristics from the Fourier amplitude spectrum (FAS) of the observed seismograms. This method is applied to plate boundary earthquakes off the Pacific coast of Tohoku, Japan. It successfully generates a set of strong‐motion seismograms at a given magnitude, distance, and observation station. The output waveforms reproduce the P and S waves as well as coda waves. We validate the method through a quantitative comparison with observed seismograms in terms of both time‐domain duration and frequency‐domain amplitude characteristics, using metrics of peak ground acceleration (PGA), peak ground velocity, FASs, response spectra, and waveform duration time. The validation results show that the variation in the PGA of the observed seismograms and the synthetic seismograms has a standard deviation of 0.643, and the duration of the seismograms has a standard deviation of 0.346, comparable to the standard deviations seen in the previous studies. Our approach offers high accuracy in stochastic finite‐source modeling for a period of 1 s or shorter. The two features of the method, site‐specificity and variability control, can contribute to further improvements in seismic hazard assessment by incorporating empirical information based on observed seismograms.
精确的地动模拟对于地震灾害评估和工程实践至关重要。在此，我们提出了一种结合条件生成对抗网络（cGANs）和广义反演技术（GIT）的新方法，用于生成特定场地和变异性控制的强震地震图。cGANs 可计算无振幅标度的合成地震图。广义反演技术（GIT）是从观测地震图的傅立叶振幅谱（FAS）中分离出震源、路径和场地特征。该方法应用于日本东北太平洋沿岸的板块边界地震。它成功生成了一组给定震级、距离和观测站的强震动地震图。输出波形再现了 P 波和 S 波以及尾波。我们使用峰值地面加速度 (PGA)、峰值地面速度、FAS、反应谱和波形持续时间等指标，在时域持续时间和频域振幅特征方面与观测到的地震图进行定量比较，从而验证了该方法。验证结果表明，观测地震图和合成地震图的峰值地面加速度的标准偏差为 0.643，地震图持续时间的标准偏差为 0.346，与之前研究的标准偏差相当。我们的方法为 1 秒或更短周期的随机微小震源建模提供了高精度。该方法的两个特点，即场地特异性和变异性控制，通过纳入基于观测地震图的经验信息，有助于进一步改进地震灾害评估。
{"title":"Site‐Specific Ground‐Motion Waveform Generation Using a Conditional Generative Adversarial Network and Generalized Inversion Technique","authors":"Junki Yamaguchi, Yusuke Tomozawa, Toshihide Saka","doi":"10.1785/0120230209","DOIUrl":"https://doi.org/10.1785/0120230209","url":null,"abstract":"Accurate ground‐motion simulations are essential for seismic hazard assessments and engineering practices. Herein, we propose a novel method combining conditional generative adversarial networks (cGANs) and the generalized inversion technique (GIT) to generate site‐specific and variability‐controlled strong‐motion seismograms. The cGANs calculate synthetic seismogram without amplitude scales. The GIT is to separate the source, path, and site characteristics from the Fourier amplitude spectrum (FAS) of the observed seismograms. This method is applied to plate boundary earthquakes off the Pacific coast of Tohoku, Japan. It successfully generates a set of strong‐motion seismograms at a given magnitude, distance, and observation station. The output waveforms reproduce the P and S waves as well as coda waves. We validate the method through a quantitative comparison with observed seismograms in terms of both time‐domain duration and frequency‐domain amplitude characteristics, using metrics of peak ground acceleration (PGA), peak ground velocity, FASs, response spectra, and waveform duration time. The validation results show that the variation in the PGA of the observed seismograms and the synthetic seismograms has a standard deviation of 0.643, and the duration of the seismograms has a standard deviation of 0.346, comparable to the standard deviations seen in the previous studies. Our approach offers high accuracy in stochastic finite‐source modeling for a period of 1 s or shorter. The two features of the method, site‐specificity and variability control, can contribute to further improvements in seismic hazard assessment by incorporating empirical information based on observed seismograms.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"94 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141777162","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ground‐motion models (GMMs) are vital in assessing probabilistic seismic hazards and uncertainties. This study develops new GMMs benefiting from nonparametric machine learning algorithms, including artificial neural networks, kernel ridge, random forest, and gradient boosting regression techniques for small‐to‐moderate potentially induced earthquakes in central and eastern North America (CENA). As part of this study, we evaluate the performance of different machine learning models in estimating peak ground acceleration (PGA) and 17 spectral accelerations based on the moment magnitude (Mw), hypocentral distance (Rhypo), and the timed‐average shear‐wave velocity of the upper 30 m of soil (VS30). To train the algorithms, we have utilized a database of nearly 31,000 ground motions with small and moderate moment magnitudes ranging from 3.0 to 5.8, recorded within a hypocentral distance of less than 200 km in CENA. Typically, for GMM development, analysts employ linear regression‐based models with predefined functional forms. The requirement for predefined functional forms can restrict the use of complicated and nonlinear equations to improve performance. Although the conventional regression model is more interpretable, machine learning can achieve a better result given sufficient training data. The results of error metrics reveal that gradient‐boosting regression provides a better performance. Furthermore, a machine learning ensemble method is used to combine the regression results of four machine learning algorithms. The ensemble method improves the GMM performance and provides smoother results.
{"title":"Ground‐Motion Model for Small‐to‐Moderate Potentially Induced Earthquakes Using an Ensemble Machine Learning Approach for CENA","authors":"Najme Alidadi, Shahram Pezeshk","doi":"10.1785/0120230242","DOIUrl":"https://doi.org/10.1785/0120230242","url":null,"abstract":"Ground‐motion models (GMMs) are vital in assessing probabilistic seismic hazards and uncertainties. This study develops new GMMs benefiting from nonparametric machine learning algorithms, including artificial neural networks, kernel ridge, random forest, and gradient boosting regression techniques for small‐to‐moderate potentially induced earthquakes in central and eastern North America (CENA). As part of this study, we evaluate the performance of different machine learning models in estimating peak ground acceleration (PGA) and 17 spectral accelerations based on the moment magnitude (Mw), hypocentral distance (Rhypo), and the timed‐average shear‐wave velocity of the upper 30 m of soil (VS30). To train the algorithms, we have utilized a database of nearly 31,000 ground motions with small and moderate moment magnitudes ranging from 3.0 to 5.8, recorded within a hypocentral distance of less than 200 km in CENA. Typically, for GMM development, analysts employ linear regression‐based models with predefined functional forms. The requirement for predefined functional forms can restrict the use of complicated and nonlinear equations to improve performance. Although the conventional regression model is more interpretable, machine learning can achieve a better result given sufficient training data. The results of error metrics reveal that gradient‐boosting regression provides a better performance. Furthermore, a machine learning ensemble method is used to combine the regression results of four machine learning algorithms. The ensemble method improves the GMM performance and provides smoother results.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"58 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141777163","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Seismic near‐fault ground motions with large amplitudes and long‐period velocity pulses can cause significant damage to structures. In this study, a new stochastic model has been developed to simulate pulse‐like ground motions for specified earthquake scenarios. Wavelet packet transform is employed to extract and model the velocity pulse. The parameters of the proposed model are fitted to Next Generation Attenuation (NGA)‐West2 pulse‐like ground‐motion database. A group of predictive equations are established to predict occurrence time, frequency content, and total energy of the velocity pulse in terms of earthquake magnitudes, rupture distances, site conditions, and directivity parameters. The correlation relationships among the model parameters are also estimated to jointly simulate pulse and residual components based on specified earthquake scenarios. The model’s capacity to stochastically simulate pulse‐like motions is demonstrated by systematic comparison with real pulse‐like recordings and existing ground‐motion prediction equations. The proposed method can find applications in seismic analyses of key infrastructures in the near‐field region.
{"title":"Stochastic Simulation of Pulse‐Like Ground Motions Using Wavelet Packets","authors":"Zhuo Wang, Duruo Huang","doi":"10.1785/0120230190","DOIUrl":"https://doi.org/10.1785/0120230190","url":null,"abstract":"Seismic near‐fault ground motions with large amplitudes and long‐period velocity pulses can cause significant damage to structures. In this study, a new stochastic model has been developed to simulate pulse‐like ground motions for specified earthquake scenarios. Wavelet packet transform is employed to extract and model the velocity pulse. The parameters of the proposed model are fitted to Next Generation Attenuation (NGA)‐West2 pulse‐like ground‐motion database. A group of predictive equations are established to predict occurrence time, frequency content, and total energy of the velocity pulse in terms of earthquake magnitudes, rupture distances, site conditions, and directivity parameters. The correlation relationships among the model parameters are also estimated to jointly simulate pulse and residual components based on specified earthquake scenarios. The model’s capacity to stochastically simulate pulse‐like motions is demonstrated by systematic comparison with real pulse‐like recordings and existing ground‐motion prediction equations. The proposed method can find applications in seismic analyses of key infrastructures in the near‐field region.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"81 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141777164","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
We present a data‐driven framework for ground‐motion synthesis that generates three‐component acceleration time histories conditioned on moment magnitude (M), rupture distance (Rrup), time‐average shear‐wave velocity at the top 30 m (VS30), and style of faulting. We use a Generative Adversarial Neural Operator (GANO)—a resolution invariant architecture that guarantees model training independent of the data sampling frequency. We first present the conditional ground‐motion synthesis algorithm (cGM‐GANO) and discuss its advantages compared to the previous work. We next train cGM‐GANO on simulated ground motions generated by the Southern California Earthquake Center Broadband Platform (BBP) and on recorded the Kiban–Kyoshin network (KiK‐net) data, and show that the model can learn the overall magnitude, distance, and VS30 scaling of effective amplitude spectra (EAS) ordinates and pseudospectral accelerations (PSA). Results specifically show that cGM‐GANO produces consistent median scaling with the training data for the corresponding tectonic environments over a wide range of frequencies for scenarios with sufficient data coverage. For the BBP dataset, cGM‐GANO cannot learn the ground‐motion scaling of the stochastic frequency components (f > 1 Hz); for the KiK‐net dataset, the largest misfit is observed at short distances (Rrup<50 km) and for soft‐soil conditions (VS30<200 m/s) due to the scarcity of such data. Except for these conditions, the aleatory variability of EAS and PSA are captured reasonably well. Finally, cGM‐GANO produces similar median scaling to traditional ground‐motion models (GMMs) for frequencies greater than 1 Hz for both PSA and EAS but underestimates the aleatory variability of EAS. Discrepancies in the comparisons between the synthetic ground motions and GMMs are attributed to inconsistencies between the training dataset and the datasets used in GMM development. Our pilot study demonstrates GANO’s potential for efficient synthesis of broadband ground motions.
{"title":"Broadband Ground‐Motion Synthesis via Generative Adversarial Neural Operators: Development and Validation","authors":"Yaozhong Shi, Grigorios Lavrentiadis, Domniki Asimaki, Zachary E. Ross, Kamyar Azizzadenesheli","doi":"10.1785/0120230207","DOIUrl":"https://doi.org/10.1785/0120230207","url":null,"abstract":"We present a data‐driven framework for ground‐motion synthesis that generates three‐component acceleration time histories conditioned on moment magnitude (M), rupture distance (Rrup), time‐average shear‐wave velocity at the top 30 m (VS30), and style of faulting. We use a Generative Adversarial Neural Operator (GANO)—a resolution invariant architecture that guarantees model training independent of the data sampling frequency. We first present the conditional ground‐motion synthesis algorithm (cGM‐GANO) and discuss its advantages compared to the previous work. We next train cGM‐GANO on simulated ground motions generated by the Southern California Earthquake Center Broadband Platform (BBP) and on recorded the Kiban–Kyoshin network (KiK‐net) data, and show that the model can learn the overall magnitude, distance, and VS30 scaling of effective amplitude spectra (EAS) ordinates and pseudospectral accelerations (PSA). Results specifically show that cGM‐GANO produces consistent median scaling with the training data for the corresponding tectonic environments over a wide range of frequencies for scenarios with sufficient data coverage. For the BBP dataset, cGM‐GANO cannot learn the ground‐motion scaling of the stochastic frequency components (f > 1 Hz); for the KiK‐net dataset, the largest misfit is observed at short distances (Rrup<50 km) and for soft‐soil conditions (VS30<200 m/s) due to the scarcity of such data. Except for these conditions, the aleatory variability of EAS and PSA are captured reasonably well. Finally, cGM‐GANO produces similar median scaling to traditional ground‐motion models (GMMs) for frequencies greater than 1 Hz for both PSA and EAS but underestimates the aleatory variability of EAS. Discrepancies in the comparisons between the synthetic ground motions and GMMs are attributed to inconsistencies between the training dataset and the datasets used in GMM development. Our pilot study demonstrates GANO’s potential for efficient synthesis of broadband ground motions.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"58 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141776914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Chiara Civiero, Sergei Lebedev, Yihe Xu, Raffaele Bonadio, François Lavoué
The unprecedentedly dense current sampling of the upper mantle with seismic data offers an opportunity for determining representative seismic velocity models for the Earth’s main tectonic environments. Here, we use over 1.17 million Rayleigh‐ and 300,000 Love‐wave, fundamental‐mode, phase‐velocity curves measured with multimode waveform inversion of data available since the 1990s, and compute phase‐velocity maps in a 17–310 s period range. We then compute phase‐velocity curves averaged over the globe and eight tectonic environments, and invert them for 1D seismic velocity profiles of the upper mantle. The averaged curves are smooth and fit by VS models with very small misfits, under 0.1%, at most periods. For phase‐velocity curves extending up to 310 s, Rayleigh waves resolve VSV structure down to the shallow lower mantle. Love‐wave sampling is shallower, and VSH and, thus, radial anisotropy profiles are resolved down to 375–400 km depth. The uncertainty of the VS models is dominated by the trade‐offs of VS at neighboring depths. Using the model‐space‐projection approach, we quantify the uncertainty of VS in layers of different thickness and at different depths, and show how it decreases with the increasing thickness of the layers. Example 1D VS models that fit the data display the expected increase of the lithospheric seismic velocity with the age of the oceanic lithosphere and with the average age of the continental tectonic type. Radial anisotropy in the global and most tectonic‐type models show a flip of the sign from positive (VSH>VSV) to negative at 200–300 km depth. Negative anisotropy is also observed in the shallow mantle lithosphere beneath oceans down to 45–55 km depth. We also compute a global model with the minimal structural complexity, which fits the data worse than the best‐fitting one but does not include a sublithospheric low‐velocity zone, providing a simple reference for seismic studies.
目前地震数据对上地幔的采样密度空前，这为确定地球主要构造环境的代表性地震速度模型提供了机会。在这里，我们使用了自 20 世纪 90 年代以来通过多模波形反演数据测量的 117 万多条雷波和 30 万条爱波基模相位速度曲线，并计算了 17-310 秒周期范围内的相位速度图。然后，我们计算了全球和八个构造环境的平均相位速度曲线，并将其反演为上地幔的一维地震速度剖面。平均曲线非常平滑，并通过 VS 模型拟合，在大多数周期的误差非常小，低于 0.1%。对于延伸至 310 秒的相位速度曲线，瑞利波解析了下地幔浅层的 VSV 结构。爱波取样较浅，VSH以及径向各向异性剖面可解析到375-400千米深度。VS 模型的不确定性主要来自邻近深度的 VS 权衡。利用模型-空间投影方法，我们量化了不同厚度和不同深度地层中 VS 的不确定性，并显示了它是如何随着地层厚度的增加而减小的。与数据拟合的一维 VS 模型示例显示，岩石圈地震速度随着海洋岩石圈年龄和大陆构造类型平均年龄的增加而增加。在全球和大多数构造类型模型中，径向各向异性在 200-300 千米深度显示出从正向（VSH>VSV）到负向的翻转。在海洋下方深度为 45-55 千米的浅地幔岩石圈中也观察到了负各向异性。我们还计算了一个结构复杂度最小的全球模型，其拟合数据比最佳拟合模型差，但不包括岩石圈下低速带，为地震研究提供了一个简单的参考。
{"title":"Toward Tectonic‐Type and Global 1D Seismic Models of the Upper Mantle Constrained by Broadband Surface Waves","authors":"Chiara Civiero, Sergei Lebedev, Yihe Xu, Raffaele Bonadio, François Lavoué","doi":"10.1785/0120230295","DOIUrl":"https://doi.org/10.1785/0120230295","url":null,"abstract":"The unprecedentedly dense current sampling of the upper mantle with seismic data offers an opportunity for determining representative seismic velocity models for the Earth’s main tectonic environments. Here, we use over 1.17 million Rayleigh‐ and 300,000 Love‐wave, fundamental‐mode, phase‐velocity curves measured with multimode waveform inversion of data available since the 1990s, and compute phase‐velocity maps in a 17–310 s period range. We then compute phase‐velocity curves averaged over the globe and eight tectonic environments, and invert them for 1D seismic velocity profiles of the upper mantle. The averaged curves are smooth and fit by VS models with very small misfits, under 0.1%, at most periods. For phase‐velocity curves extending up to 310 s, Rayleigh waves resolve VSV structure down to the shallow lower mantle. Love‐wave sampling is shallower, and VSH and, thus, radial anisotropy profiles are resolved down to 375–400 km depth. The uncertainty of the VS models is dominated by the trade‐offs of VS at neighboring depths. Using the model‐space‐projection approach, we quantify the uncertainty of VS in layers of different thickness and at different depths, and show how it decreases with the increasing thickness of the layers. Example 1D VS models that fit the data display the expected increase of the lithospheric seismic velocity with the age of the oceanic lithosphere and with the average age of the continental tectonic type. Radial anisotropy in the global and most tectonic‐type models show a flip of the sign from positive (VSH>VSV) to negative at 200–300 km depth. Negative anisotropy is also observed in the shallow mantle lithosphere beneath oceans down to 45–55 km depth. We also compute a global model with the minimal structural complexity, which fits the data worse than the best‐fitting one but does not include a sublithospheric low‐velocity zone, providing a simple reference for seismic studies.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"30 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141258827","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Gianmarco Del Piccolo, Brandon P. VanderBeek, Manuele Faccenda, Andrea Morelli, Joseph S. Byrnes
Underdetermination is a condition affecting all problems in seismic imaging. It manifests mainly in the nonuniqueness of the models inferred from the data. This condition is exacerbated if simplifying hypotheses like isotropy are discarded in favor of more realistic anisotropic models that, although supported by seismological evidence, require more free parameters. Investigating the connections between underdetermination and anisotropy requires the implementation of solvers which explore the whole family of possibilities behind nonuniqueness and allow for more informed conclusions about the interpretation of the seismic models. Because these aspects cannot be investigated using traditional iterative linearized inversion schemes with regularization constraints that collapse the infinite possible models into a unique solution, we explore the application of transdimensional Bayesian Monte Carlo sampling to address the consequences of underdetermination in anisotropic seismic imaging. We show how teleseismic waves of P and S phases can constrain upper‐mantle anisotropy and the amount of additional information these data provide in terms of uncertainty and trade‐offs among multiple fields.
欠确定性是影响地震成像所有问题的一个条件。它主要表现为从数据推断出的模型的非唯一性。如果摒弃各向同性等简化假设，转而采用更现实的各向异性模型，这种情况就会加剧。要研究欠确定性和各向异性之间的联系，就需要使用求解器来探索非唯一性背后的所有可能性，并对地震模型的解释得出更明智的结论。传统的迭代线性化反演方案带有正则化约束，可将无限可能的模型折叠成唯一的解，但无法对这些方面进行研究，因此我们探索了跨维贝叶斯蒙特卡洛采样的应用，以解决各向异性地震成像中判定不足的后果。我们展示了 P 相和 S 相远震波如何约束上幔各向异性，以及这些数据在不确定性和多场权衡方面提供的额外信息量。
{"title":"Imaging Upper‐Mantle Anisotropy with Transdimensional Bayesian Monte Carlo Sampling","authors":"Gianmarco Del Piccolo, Brandon P. VanderBeek, Manuele Faccenda, Andrea Morelli, Joseph S. Byrnes","doi":"10.1785/0120230233","DOIUrl":"https://doi.org/10.1785/0120230233","url":null,"abstract":"Underdetermination is a condition affecting all problems in seismic imaging. It manifests mainly in the nonuniqueness of the models inferred from the data. This condition is exacerbated if simplifying hypotheses like isotropy are discarded in favor of more realistic anisotropic models that, although supported by seismological evidence, require more free parameters. Investigating the connections between underdetermination and anisotropy requires the implementation of solvers which explore the whole family of possibilities behind nonuniqueness and allow for more informed conclusions about the interpretation of the seismic models. Because these aspects cannot be investigated using traditional iterative linearized inversion schemes with regularization constraints that collapse the infinite possible models into a unique solution, we explore the application of transdimensional Bayesian Monte Carlo sampling to address the consequences of underdetermination in anisotropic seismic imaging. We show how teleseismic waves of P and S phases can constrain upper‐mantle anisotropy and the amount of additional information these data provide in terms of uncertainty and trade‐offs among multiple fields.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"15 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141165443","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Asaf Inbal, Tudor Cristea‐Platon, Jean‐Paul Ampuero, Gregor Hillers, Duncan Agnew
Because of a printing error, the x‐axis labels are missing from figures 8–10 in the reply article of Inbal et al. (2023) to the comment article of Hutchison et al. (2023). This erratum contains the corrected figures labeled here as Figures 1–3.The authors acknowledge that there are no conflicts of interest recorded.
{"title":"Erratum to Reply to “Comment on ‘Sources of Long‐Range Anthropogenic Noise in Southern California and Implications for Tectonic Tremor Detection’ by Asaf Inbal, Tudor Cristea‐Platon, Jean‐Paul Ampuero, Gregor Hillers, Duncan Agnew, and Susan E. Hough” by Allie Hutchison, Yijian Zhou, and Abhijit Ghosh","authors":"Asaf Inbal, Tudor Cristea‐Platon, Jean‐Paul Ampuero, Gregor Hillers, Duncan Agnew","doi":"10.1785/0120230133","DOIUrl":"https://doi.org/10.1785/0120230133","url":null,"abstract":"Because of a printing error, the x‐axis labels are missing from figures 8–10 in the reply article of Inbal et al. (2023) to the comment article of Hutchison et al. (2023). This erratum contains the corrected figures labeled here as Figures 1–3.The authors acknowledge that there are no conflicts of interest recorded.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"42 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140313220","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jen Andrews, Yannik Behr, Maren Böse, Frédérick Massin, Anna Kaiser, Bill Fry
Immediately after a significant earthquake, rapid scientific information is critical for response decision‐making and estimating secondary hazards, and is a key component of advisories and public communication. Characterization of the fault rupture extent is especially valuable because it strongly controls ground‐motion estimates, or tsunami forecasts in offshore settings. The Finite‐fault rupture Detector (FinDer) is designed to rapidly estimate location, extent, and orientation of earthquake fault rupture by matching spatial distributions of high‐frequency seismic amplitudes with precomputed templates. Under a large public initiative to better prepare for and respond to natural disasters, FinDer is being implemented in New Zealand for rapid source characterization. Here, we report on implementation and performance, including offline and real‐time testing using configurations modified for the New Zealand setting. Systematic testing is used to inform guidelines for real‐time usage and interpretation. Analysis of rupture parameter recovery when using national network GeoNet stations demonstrates that for moderate (M 6+) onshore earthquakes FinDer can resolve magnitude and location well, and the rupture strike is also well determined for large (M 7+) onshore earthquakes. For near‐offshore earthquakes (within 100 km), FinDer can provide reasonable magnitude estimates but cannot determine the location or strike. Real‐time testing shows reliable detection for onshore earthquakes of M 4.5+, with reasonable location and magnitude accuracy. First detection times range between 7 and 65 s of earthquake origin, and stable solutions even for large (M 7+) magnitude events are delivered within 2 min. Although the GeoNet seismic network is not optimized for earthquake early warning, this provides a first exploration of network‐based capability for New Zealand. Offline testing of significant M 7+ historic earthquakes demonstrates that FinDer’s rupture solutions can be used to improve rapid shaking predictions, and may be used to infer additional directivity and tsunami hazard even for complex events like the 2016 M 7.8 Kaikōura earthquake.
{"title":"Rapid Earthquake Rupture Characterization for New Zealand Using the FinDer Algorithm","authors":"Jen Andrews, Yannik Behr, Maren Böse, Frédérick Massin, Anna Kaiser, Bill Fry","doi":"10.1785/0120230213","DOIUrl":"https://doi.org/10.1785/0120230213","url":null,"abstract":"Immediately after a significant earthquake, rapid scientific information is critical for response decision‐making and estimating secondary hazards, and is a key component of advisories and public communication. Characterization of the fault rupture extent is especially valuable because it strongly controls ground‐motion estimates, or tsunami forecasts in offshore settings. The Finite‐fault rupture Detector (FinDer) is designed to rapidly estimate location, extent, and orientation of earthquake fault rupture by matching spatial distributions of high‐frequency seismic amplitudes with precomputed templates. Under a large public initiative to better prepare for and respond to natural disasters, FinDer is being implemented in New Zealand for rapid source characterization. Here, we report on implementation and performance, including offline and real‐time testing using configurations modified for the New Zealand setting. Systematic testing is used to inform guidelines for real‐time usage and interpretation. Analysis of rupture parameter recovery when using national network GeoNet stations demonstrates that for moderate (M 6+) onshore earthquakes FinDer can resolve magnitude and location well, and the rupture strike is also well determined for large (M 7+) onshore earthquakes. For near‐offshore earthquakes (within 100 km), FinDer can provide reasonable magnitude estimates but cannot determine the location or strike. Real‐time testing shows reliable detection for onshore earthquakes of M 4.5+, with reasonable location and magnitude accuracy. First detection times range between 7 and 65 s of earthquake origin, and stable solutions even for large (M 7+) magnitude events are delivered within 2 min. Although the GeoNet seismic network is not optimized for earthquake early warning, this provides a first exploration of network‐based capability for New Zealand. Offline testing of significant M 7+ historic earthquakes demonstrates that FinDer’s rupture solutions can be used to improve rapid shaking predictions, and may be used to infer additional directivity and tsunami hazard even for complex events like the 2016 M 7.8 Kaikōura earthquake.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"5 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140578545","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jennifer L. Harding, Leiph A. Preston, Miles A. Bodmer
The Rock Valley fault zone (RVFZ), an intraplate strike‐slip fault zone in the southern Nevada National Security Site (NNSS), hosted a series of very shallow (<3 km) earthquakes in 1993. The RVFZ may also have hydrological significance within the NNSS, potentially playing a role in regional groundwater flow, but there is a lack of local hydrological data. In the Spring of 2021, we collected active‐source accelerated weight drop seismic data over part of the RVFZ to better characterize the shallow subsurface. We manually picked ∼17,000 P‐wave travel times and over 14,000 S‐wave travel times, which were inverted for P‐wave velocity (VP), S‐wave velocity (VS), and VP/VS ratio in a 3D joint tomographic inversion scheme. Seismic velocities are imaged as deep as ∼700 m in areas and generally align with geologic and structural expectations. VP and VS are relatively reduced near mapped and inferred faults, with the most prominent lower VP and VS zone around the densest collection of faults. We image VP/VS ratios ranging from ∼1.5 to ∼2.4, the extremes of which occur at a depth of ∼100 m and are juxtaposed across a fault. One possible interpretation of the imaged seismic velocities is enhanced fault damage near the densest collection of faults with relatively higher porosity and/or crack density at ∼100 m depth, with patches of semiperched groundwater present in the sedimentary rock in higher VP/VS areas and drier rock in lower VP/VS areas. A relatively higher VP/VS area beneath the densest faults persists at depth, which suggests percolation of groundwater via the fault damage zone to the regionally connected lower carbonate aquifer. Potentially, the presence and movement of groundwater may have played a role in the 1993 earthquake aftershocks.
岩石谷断层带（RVFZ）是内华达州国家安全地点（NNSS）南部的板内走向滑动断层带，1993 年发生了一系列很浅（<3 公里）的地震。RVFZ 在 NNSS 内也可能具有重要的水文意义，有可能在区域地下水流中发挥作用，但当地缺乏水文数据。2021 年春，我们采集了 RVFZ 部分地区的主动源加速重力落差地震数据，以更好地描述浅层地下的特征。我们手动选取了 17,000 个 P 波行进时间和 14,000 多个 S 波行进时间，并在三维联合层析反演方案中对其进行了 P 波速度 (VP)、S 波速度 (VS) 和 VP/VS 比值反演。地震波速度成像区域最深处达 700 米，总体上与地质和构造预期一致。在测绘和推断的断层附近，VP 和 VS 相对减小，在最密集的断层群周围，VP 和 VS 较低的区域最为突出。我们拍摄到的 VP/VS 比率范围在 1.5 至 2.4 之间，其极端值出现在深度为 100 米的断层上。对成像地震速度的一种可能解释是，在 100 米深处孔隙度和/或裂缝密度相对较高的断层密集区附近，断层破坏加剧，在 VP/VS 较高区域的沉积岩中存在成片的半陡峭地下水，而在 VP/VS 较低区域的岩石则较为干燥。在密度最大的断层下方，VP/VS 相对较高的区域在深度上持续存在，这表明地下水通过断层破坏带渗入到区域相连的下碳酸盐岩含水层。地下水的存在和流动可能在 1993 年地震余震中发挥了作用。
{"title":"Hydrologic Impacts of a Strike‐Slip Fault Zone: Insights from Joint 3D Body‐Wave Tomography of Rock Valley","authors":"Jennifer L. Harding, Leiph A. Preston, Miles A. Bodmer","doi":"10.1785/0120230081","DOIUrl":"https://doi.org/10.1785/0120230081","url":null,"abstract":"The Rock Valley fault zone (RVFZ), an intraplate strike‐slip fault zone in the southern Nevada National Security Site (NNSS), hosted a series of very shallow (<3 km) earthquakes in 1993. The RVFZ may also have hydrological significance within the NNSS, potentially playing a role in regional groundwater flow, but there is a lack of local hydrological data. In the Spring of 2021, we collected active‐source accelerated weight drop seismic data over part of the RVFZ to better characterize the shallow subsurface. We manually picked ∼17,000 P‐wave travel times and over 14,000 S‐wave travel times, which were inverted for P‐wave velocity (VP), S‐wave velocity (VS), and VP/VS ratio in a 3D joint tomographic inversion scheme. Seismic velocities are imaged as deep as ∼700 m in areas and generally align with geologic and structural expectations. VP and VS are relatively reduced near mapped and inferred faults, with the most prominent lower VP and VS zone around the densest collection of faults. We image VP/VS ratios ranging from ∼1.5 to ∼2.4, the extremes of which occur at a depth of ∼100 m and are juxtaposed across a fault. One possible interpretation of the imaged seismic velocities is enhanced fault damage near the densest collection of faults with relatively higher porosity and/or crack density at ∼100 m depth, with patches of semiperched groundwater present in the sedimentary rock in higher VP/VS areas and drier rock in lower VP/VS areas. A relatively higher VP/VS area beneath the densest faults persists at depth, which suggests percolation of groundwater via the fault damage zone to the regionally connected lower carbonate aquifer. Potentially, the presence and movement of groundwater may have played a role in the 1993 earthquake aftershocks.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"40 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140313157","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Valentin Kasburg, Jozef Müller, Tom Eulenfeld, Alexander Breuer, Nina Kukowski
The gradual densification of seismic networks has facilitated the acquisition of large amounts of data. However, alongside natural tectonic earthquakes, seismic networks also record anthropogenic events such as quarry blasts or other induced events. Identifying and distinguishing these events from natural earthquakes requires experienced interpreters to ensure that seismological studies of natural phenomena are not compromised by anthropogenic events. Advanced artificial intelligence methods have already been deployed to tackle this problem. One of the applications includes Convolutional Neural Networks (CNN) to discriminate different kinds of events, such as natural earthquakes and quarry blasts. In this study, we investigate the effects of ensemble averaging and fine‐tuning on seismic event discrimination accuracy to estimate the potential of these methods. We compare discrimination accuracy of two different CNN model architectures across three datasets. This was done with the best models from an ensemble of each model architecture, as well as with ensemble averaging and fine‐tuning methods. Soft voting was used for the CNN ensemble predictions. For the transfer learning approach, the models were pretrained with data from two of the datasets (nontarget regions) and fine‐tuned with data from the third one (target region). The results show that ensemble averaging and fine‐tuning of CNN models leads to better generalization of the model predictions. For the region with the lowest numbers of one event type, the combination of ensemble averaging and fine‐tuning led to an increase in discrimination accuracy of up to 4% at station level and up to 10% at event level. We also tested the impact of the amount of training data on the fine‐tuning method, showing, that to create a global model, the selection of comprehensive training data is needed.
{"title":"Cross‐Regional Seismic Event Discrimination via Convolutional Neural Networks: Exploring Fine‐Tuning and Ensemble Averaging","authors":"Valentin Kasburg, Jozef Müller, Tom Eulenfeld, Alexander Breuer, Nina Kukowski","doi":"10.1785/0120230198","DOIUrl":"https://doi.org/10.1785/0120230198","url":null,"abstract":"The gradual densification of seismic networks has facilitated the acquisition of large amounts of data. However, alongside natural tectonic earthquakes, seismic networks also record anthropogenic events such as quarry blasts or other induced events. Identifying and distinguishing these events from natural earthquakes requires experienced interpreters to ensure that seismological studies of natural phenomena are not compromised by anthropogenic events. Advanced artificial intelligence methods have already been deployed to tackle this problem. One of the applications includes Convolutional Neural Networks (CNN) to discriminate different kinds of events, such as natural earthquakes and quarry blasts. In this study, we investigate the effects of ensemble averaging and fine‐tuning on seismic event discrimination accuracy to estimate the potential of these methods. We compare discrimination accuracy of two different CNN model architectures across three datasets. This was done with the best models from an ensemble of each model architecture, as well as with ensemble averaging and fine‐tuning methods. Soft voting was used for the CNN ensemble predictions. For the transfer learning approach, the models were pretrained with data from two of the datasets (nontarget regions) and fine‐tuned with data from the third one (target region). The results show that ensemble averaging and fine‐tuning of CNN models leads to better generalization of the model predictions. For the region with the lowest numbers of one event type, the combination of ensemble averaging and fine‐tuning led to an increase in discrimination accuracy of up to 4% at station level and up to 10% at event level. We also tested the impact of the amount of training data on the fine‐tuning method, showing, that to create a global model, the selection of comprehensive training data is needed.","PeriodicalId":9444,"journal":{"name":"Bulletin of the Seismological Society of America","volume":"10 1","pages":""},"PeriodicalIF":3.0,"publicationDate":"2024-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140325826","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：