The time window for analyzing local shear-wave splitting (SWS) phases significantly affects the quality of measurements, revealing a noteworthy domain influence. In this study, an approach using convolutional neural network (CNN) is applied to determine the end of time window (e), which has a similar idea of the phase-picking CNNs. The start of time window is 0.5 s before e. Our data set contains 803 human-labeled measurements, recorded from three stations located in Ridgecrest, California. These measurements are foreshocks and aftershocks of an M 7.1 earthquake on 6 July 2019. After 21 times shifting on each measurement, 90% of the data set is applied as the training data set, with the remaining 10% as the testing data set. The performance of CNN with the testing data set is compared with a nonmachine learning method, multiple filter automatic splitting technique (MFAST). The results reveal that the CNN yields more similar results with human-labeled outcomes than MFAST, as evidenced by lower absolute error and standard deviation for e, SWS time, the orientation of fast-wave polarization, and more consistent results on the map. The CNN also performs well when applied to data recorded by a station in Parkfield, California. This study shows the outstanding performance of CNN in picking the time window and the reliable automatic determination of this time window, and it is also a crucial step for future development of automatic ranking methodologies.
{"title":"Using Convolutional Neural Network to Determine Time Window for Analyzing Local Shear-Wave Splitting Measurements","authors":"Yanwei Zhang, Stephen S. Gao","doi":"10.1785/0220230410","DOIUrl":"https://doi.org/10.1785/0220230410","url":null,"abstract":"\u0000 The time window for analyzing local shear-wave splitting (SWS) phases significantly affects the quality of measurements, revealing a noteworthy domain influence. In this study, an approach using convolutional neural network (CNN) is applied to determine the end of time window (e), which has a similar idea of the phase-picking CNNs. The start of time window is 0.5 s before e. Our data set contains 803 human-labeled measurements, recorded from three stations located in Ridgecrest, California. These measurements are foreshocks and aftershocks of an M 7.1 earthquake on 6 July 2019. After 21 times shifting on each measurement, 90% of the data set is applied as the training data set, with the remaining 10% as the testing data set. The performance of CNN with the testing data set is compared with a nonmachine learning method, multiple filter automatic splitting technique (MFAST). The results reveal that the CNN yields more similar results with human-labeled outcomes than MFAST, as evidenced by lower absolute error and standard deviation for e, SWS time, the orientation of fast-wave polarization, and more consistent results on the map. The CNN also performs well when applied to data recorded by a station in Parkfield, California. This study shows the outstanding performance of CNN in picking the time window and the reliable automatic determination of this time window, and it is also a crucial step for future development of automatic ranking methodologies.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":" 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141668619","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Newport–Inglewood fault (NIF) is a slowly deforming fault cutting through a thin continental crust with a normal geothermal; yet it hosts some of the deepest earthquakes in southern California. The nucleation of deep earthquakes in such a continental setting is not well understood. Moreover, the deep seismogenic zone implies that the maximum NIF earthquake magnitude may be larger than expected. Here, we quantify the resolution of the Long Beach (LB) and the Extended Long Beach (ELB) dense arrays used to study deep NIF seismicity. Previous study of the regional catalog and of downward-continued LB array data found NIF seismicity extending into the upper mantle beneath LB. Later studies, which analyzed the ELB raw data, found little evidence for such deep events. To resolve this inconsistency, we quantify the array’s microearthquake detectability and resolution power via analysis of pre- and postdownward migrated LB seismograms and benchmark tests. Downward migration focuses energy onto the source region and deamplifies the surface noise, thus significantly improving detectability and resolution. The detectability is also improved with the increase in the array aperture-to-source-depth ratio. The LB array maximum aperture is only 20% larger than the ELB aperture, yet its resolution for deep (>20 km) events is improved by about a factor of two, suggesting that small changes to the array geometry may yield significant improvement to the resolution power. Assuming a constant aperture, we find the LB array maintain resolution with 1% of its sensors used for backprojection. However, the high-sensor density is essential for improving the signal-to-noise ratio. Analysis of the regional and array-derived NIF catalogs together with newly acquired Moho depths beneath the NIF suggests that mantle seismicity beneath LB may be a long-lived feature of this fault.
{"title":"Earthquake Detectability and Depth Resolution with Dense Arrays in Long Beach, California: Further Evidence for Upper-Mantle Seismicity within a Continental Setting","authors":"A. Inbal, J. Ampuero, Robert W. Clayton","doi":"10.1785/0220240035","DOIUrl":"https://doi.org/10.1785/0220240035","url":null,"abstract":"\u0000 The Newport–Inglewood fault (NIF) is a slowly deforming fault cutting through a thin continental crust with a normal geothermal; yet it hosts some of the deepest earthquakes in southern California. The nucleation of deep earthquakes in such a continental setting is not well understood. Moreover, the deep seismogenic zone implies that the maximum NIF earthquake magnitude may be larger than expected. Here, we quantify the resolution of the Long Beach (LB) and the Extended Long Beach (ELB) dense arrays used to study deep NIF seismicity. Previous study of the regional catalog and of downward-continued LB array data found NIF seismicity extending into the upper mantle beneath LB. Later studies, which analyzed the ELB raw data, found little evidence for such deep events. To resolve this inconsistency, we quantify the array’s microearthquake detectability and resolution power via analysis of pre- and postdownward migrated LB seismograms and benchmark tests. Downward migration focuses energy onto the source region and deamplifies the surface noise, thus significantly improving detectability and resolution. The detectability is also improved with the increase in the array aperture-to-source-depth ratio. The LB array maximum aperture is only 20% larger than the ELB aperture, yet its resolution for deep (>20 km) events is improved by about a factor of two, suggesting that small changes to the array geometry may yield significant improvement to the resolution power. Assuming a constant aperture, we find the LB array maintain resolution with 1% of its sensors used for backprojection. However, the high-sensor density is essential for improving the signal-to-noise ratio. Analysis of the regional and array-derived NIF catalogs together with newly acquired Moho depths beneath the NIF suggests that mantle seismicity beneath LB may be a long-lived feature of this fault.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":" 1276","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141669093","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The 17 October 1989 M 6.9 Loma Prieta earthquake was the most significant earthquake to impact the San Francisco Bay region since 1906. It occurred 26 min before the scheduled start of the third game of the World Series between the San Francisco Giants and their across-the-bay rivals, the Oakland A’s. Those watching the pregame TV broadcasts from north of San Francisco saw the impacts of the strong shaking at Candlestick Park well before the shaking hit them—a clear demonstration of the physics underlying earthquake early warning systems: electromagnetic transmission is about 100,000 times faster than the propagation of seismic waves through the earth. Every earthquake is different, with unique features, and the Loma Prieta earthquake was unique in many respects. This article describes some of those unique aspects from the perspective of the Seismology Branch of the U.S. Geological Survey, located in Menlo Park, California, roughly midway between the epicenter and Candlestick Park.
{"title":"Memories of the 1989 Loma Prieta Earthquake","authors":"W. Bakun","doi":"10.1785/0220240236","DOIUrl":"https://doi.org/10.1785/0220240236","url":null,"abstract":"\u0000 The 17 October 1989 M 6.9 Loma Prieta earthquake was the most significant earthquake to impact the San Francisco Bay region since 1906. It occurred 26 min before the scheduled start of the third game of the World Series between the San Francisco Giants and their across-the-bay rivals, the Oakland A’s. Those watching the pregame TV broadcasts from north of San Francisco saw the impacts of the strong shaking at Candlestick Park well before the shaking hit them—a clear demonstration of the physics underlying earthquake early warning systems: electromagnetic transmission is about 100,000 times faster than the propagation of seismic waves through the earth. Every earthquake is different, with unique features, and the Loma Prieta earthquake was unique in many respects. This article describes some of those unique aspects from the perspective of the Seismology Branch of the U.S. Geological Survey, located in Menlo Park, California, roughly midway between the epicenter and Candlestick Park.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":" November","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141669832","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Stretching across New Mexico and Texas of the United States, the greater Permian basin is composed of two subunits—the Delaware and the Midland basins. Induced seismicity in the greater Permian basin has significantly increased since 2008, which has revealed previously unmapped seismogenic structures in several geographic regions. Among them, the Snyder area of northwest Texas has a long history of oil and gas activities, resulting in a higher rate of induced seismicity. In this study, we investigated these previously unknown seismogenic structures using three main approaches: (1) relocated and delineated seismicity, (2) performed waveform moment tensor inversion to determine earthquake source mechanisms, as well as (3) conducted stress inversion to assess the stress state. The results show that the overall depth range of seismicity is 0–5.5 km and concentrated in a range of 2–3 km below mean sea level, in the top portion of the crystalline basement. As we have determined 297 source mechanisms, their collective pattern presents a mix of strike-slip and normal faulting, suggesting an extensional strain field at the edge of the Midland basin. We have identified nine significant seismogenic episodes by distinctive increases of seismic moment release in 2017–March 2024. The results also demonstrate a temporal variation of b-value spanning across the seismogenic episodes, associated with the progression of fault reactivation initiated by fluid injection.
{"title":"Reactivated Seismogenic Faults and Earthquake Source Mechanisms in the Snyder Area of Texas","authors":"Guo-Chin Dino Huang, Alexandras Savvaidis","doi":"10.1785/0220240048","DOIUrl":"https://doi.org/10.1785/0220240048","url":null,"abstract":"\u0000 Stretching across New Mexico and Texas of the United States, the greater Permian basin is composed of two subunits—the Delaware and the Midland basins. Induced seismicity in the greater Permian basin has significantly increased since 2008, which has revealed previously unmapped seismogenic structures in several geographic regions. Among them, the Snyder area of northwest Texas has a long history of oil and gas activities, resulting in a higher rate of induced seismicity. In this study, we investigated these previously unknown seismogenic structures using three main approaches: (1) relocated and delineated seismicity, (2) performed waveform moment tensor inversion to determine earthquake source mechanisms, as well as (3) conducted stress inversion to assess the stress state. The results show that the overall depth range of seismicity is 0–5.5 km and concentrated in a range of 2–3 km below mean sea level, in the top portion of the crystalline basement. As we have determined 297 source mechanisms, their collective pattern presents a mix of strike-slip and normal faulting, suggesting an extensional strain field at the edge of the Midland basin. We have identified nine significant seismogenic episodes by distinctive increases of seismic moment release in 2017–March 2024. The results also demonstrate a temporal variation of b-value spanning across the seismogenic episodes, associated with the progression of fault reactivation initiated by fluid injection.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":" 879","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141669166","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Vincenzo Maria Schimmenti, Giuseppe Petrillo, Alberto Rosso, F. Landes
We present a machine learning approach for aftershock forecasting of the Japanese earthquakes catalog. Our method takes as sole input the ground surface deformation as measured by Global Positioning System (GPS) stations on the day of the mainshock to predict aftershock location. The quality of data heavily relies on the density of GPS stations: the predictive power is lost when the mainshocks occur far from measurement stations, as in offshore regions. Despite this fact and the small number of samples and the large number of parameters, we are able to limit overfitting, which shows that this new approach is very promising.
{"title":"Assessing the Predictive Power of GPS-Based Ground Deformation Data for Aftershock Forecasting","authors":"Vincenzo Maria Schimmenti, Giuseppe Petrillo, Alberto Rosso, F. Landes","doi":"10.1785/0220240008","DOIUrl":"https://doi.org/10.1785/0220240008","url":null,"abstract":"\u0000 We present a machine learning approach for aftershock forecasting of the Japanese earthquakes catalog. Our method takes as sole input the ground surface deformation as measured by Global Positioning System (GPS) stations on the day of the mainshock to predict aftershock location. The quality of data heavily relies on the density of GPS stations: the predictive power is lost when the mainshocks occur far from measurement stations, as in offshore regions. Despite this fact and the small number of samples and the large number of parameters, we are able to limit overfitting, which shows that this new approach is very promising.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":"23 4","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141684105","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The rapidly advancing technology of distributed acoustic sensing (DAS) has profoundly impacted the field of underwater geophysics. Our study investigates the effectiveness of DAS in underwater geological stability monitoring, with a particular focus on microseismic monitoring in the Xinfengjiang reservoir. The 6.2 km long acquisition setup, covering both land and reservoir bottom, was verified using temporary shore-based short-period seismometers to ensure reliable data acquisition in various environments. Higher background noise was observed on the land section compared with the lakebed section during the day, whereas both sections exhibited similar noise levels at night. We confirmed that the DAS system was capable of detecting distant microseismic events, some of which were previously unreported. These detections exhibited temporal and phase consistency with neighboring seismometers. Comparison of signal-to-noise ratios indicates that the lakebed section demonstrates higher sensitivity. This system delivers cost-effective performance through natural settling, negating the requirement for costly embedding methods. Moreover, the DAS system identified “comet-like” small-scale signals on the lakebed that had eluded shore-based seismometers. This exemplifies the exceptional high-density and high-resolution capabilities of DAS technology in both aquatic and terrestrial environments. This study underscores the pivotal role of the DAS technology in conducting underwater microseismic monitoring, real-time seismic monitoring, seismic mechanism research, and earthquake hazard assessment.
分布式声学传感（DAS）技术的快速发展对水下地球物理领域产生了深远影响。我们的研究调查了分布式声学传感技术在水下地质稳定性监测中的有效性，尤其侧重于新丰江水库的微震监测。我们使用临时岸基短周期地震仪验证了 6.2 公里长的采集设置，覆盖陆地和水库底部，以确保在各种环境下可靠地采集数据。与湖床部分相比，陆地部分白天的背景噪声更高，而夜间两个部分的噪声水平相近。我们证实，DAS 系统能够探测到远处的微震事件，其中一些事件以前从未报道过。这些探测结果在时间和相位上与邻近的地震仪一致。信噪比比较表明，湖床部分的灵敏度更高。该系统通过自然沉降实现了经济高效的性能，无需采用昂贵的嵌入方法。此外，DAS 系统还在湖床上识别出了 "彗星 "般的小尺度信号，而这些信号是岸基地震仪无法识别的。这充分体现了 DAS 技术在水生和陆地环境中卓越的高密度和高分辨率能力。这项研究强调了 DAS 技术在进行水下微震监测、实时地震监测、地震机理研究和地震灾害评估方面的关键作用。
{"title":"Integrated Amphibious Distributed Acoustic Sensing for Seismic Monitoring in the Xinfengjiang Reservoir","authors":"Chao Li, Xingda Jiang, Min Xu, Haocai Huang, Zhuo Xiao, Yuejin Li, Zehui Lin, Hongxing Cui, Siyuan Cang, Xiaoming Cui, Yong Zhou, Huayong Yang","doi":"10.1785/0220240001","DOIUrl":"https://doi.org/10.1785/0220240001","url":null,"abstract":"\u0000 The rapidly advancing technology of distributed acoustic sensing (DAS) has profoundly impacted the field of underwater geophysics. Our study investigates the effectiveness of DAS in underwater geological stability monitoring, with a particular focus on microseismic monitoring in the Xinfengjiang reservoir. The 6.2 km long acquisition setup, covering both land and reservoir bottom, was verified using temporary shore-based short-period seismometers to ensure reliable data acquisition in various environments. Higher background noise was observed on the land section compared with the lakebed section during the day, whereas both sections exhibited similar noise levels at night. We confirmed that the DAS system was capable of detecting distant microseismic events, some of which were previously unreported. These detections exhibited temporal and phase consistency with neighboring seismometers. Comparison of signal-to-noise ratios indicates that the lakebed section demonstrates higher sensitivity. This system delivers cost-effective performance through natural settling, negating the requirement for costly embedding methods. Moreover, the DAS system identified “comet-like” small-scale signals on the lakebed that had eluded shore-based seismometers. This exemplifies the exceptional high-density and high-resolution capabilities of DAS technology in both aquatic and terrestrial environments. This study underscores the pivotal role of the DAS technology in conducting underwater microseismic monitoring, real-time seismic monitoring, seismic mechanism research, and earthquake hazard assessment.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":"53 s43","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141837731","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Mingye Feng, Ling Chen, S. Wei, U. Muksin, Andrean V. H. Simanjuntak, Yukuan Chen, Chang Gong
Receiver function (RF) imaging using seismic data from dense short-period arrays has gained increasing importance in recent years in investigating fine-scale structures of the crust and uppermost mantle. A crucial step in such studies is to remove the instrument response (IR) to enhance teleseismic signals at ∼0.01 to 5 Hz, thereby simulating broadband records. However, this procedure also amplifies noise within the same frequency band. For weak signals, distinguishing them from noise is often challenging and in some cases is even impossible with traditional denoising methods such as filtering. To address this challenge, we develop a new convolutional neural network model, NodalWaden, using decades of high-quality global broadband teleseismic body waves for training. The broadband data exhibit the characteristics we target to achieve by removing the IR from the short-period records. The applicability of NodalWaden is justified by denoising the three-component short-period records of more than 18 months from 155 nodes deployed in northern Sumatra. We find that NodalWaden substantially improves the signal-to-noise ratio (SNR), upgrading ∼50% of the teleseismic data from the “very-low-SNR” (∼1) to “very-high-SNR” (>10) categories. RFs calculated from the denoised dataset show better separation of merged phases and noticeable enhancement of weak signals, resulting in improvement in the quality of structure imaging. In particular, a positive phase is consistently detected at ~2 s throughout the dataset and interpreted as the Conrad discontinuity, which is unresolvable in the original RFs. This denoising technique would be particularly useful for short-duration (e.g., one month) deployment with limited teleseismic data, both from the past and in the future.
{"title":"Deep Learning–Based Denoising Improves Receiver Function Imaging Using Dense Short-Period Teleseismic Data","authors":"Mingye Feng, Ling Chen, S. Wei, U. Muksin, Andrean V. H. Simanjuntak, Yukuan Chen, Chang Gong","doi":"10.1785/0220240017","DOIUrl":"https://doi.org/10.1785/0220240017","url":null,"abstract":"\u0000 Receiver function (RF) imaging using seismic data from dense short-period arrays has gained increasing importance in recent years in investigating fine-scale structures of the crust and uppermost mantle. A crucial step in such studies is to remove the instrument response (IR) to enhance teleseismic signals at ∼0.01 to 5 Hz, thereby simulating broadband records. However, this procedure also amplifies noise within the same frequency band. For weak signals, distinguishing them from noise is often challenging and in some cases is even impossible with traditional denoising methods such as filtering. To address this challenge, we develop a new convolutional neural network model, NodalWaden, using decades of high-quality global broadband teleseismic body waves for training. The broadband data exhibit the characteristics we target to achieve by removing the IR from the short-period records. The applicability of NodalWaden is justified by denoising the three-component short-period records of more than 18 months from 155 nodes deployed in northern Sumatra. We find that NodalWaden substantially improves the signal-to-noise ratio (SNR), upgrading ∼50% of the teleseismic data from the “very-low-SNR” (∼1) to “very-high-SNR” (>10) categories. RFs calculated from the denoised dataset show better separation of merged phases and noticeable enhancement of weak signals, resulting in improvement in the quality of structure imaging. In particular, a positive phase is consistently detected at ~2 s throughout the dataset and interpreted as the Conrad discontinuity, which is unresolvable in the original RFs. This denoising technique would be particularly useful for short-duration (e.g., one month) deployment with limited teleseismic data, both from the past and in the future.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":" 39","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141681016","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The seismic phases Pn and Sn play a crucial role in investigating the velocity and anisotropic characteristics of the uppermost mantle. However, manually annotating these phases can be time-intensive and prone to subjective interpretation. Consequently, the use of travel-time data for these seismic phases remains limited. Despite the potential of deep learning to address this challenge, the scarcity of extensive training data sets for Pn and Sn presents significant constraints. To address this challenge, our research compiled a global million-scale benchmark data set of Pn and Sn seismic phases, namely Seis–PnSn. The data set is derived from earthquake events with epicenter distances ranging from 1.8° to 18°. The high-quality travel-time data used in this study are all from the International Seismological Centre and span the period 2000 to 2019. The waveform data were sourced from data centers located in different regions of the world under the International Federation of Digital Seismograph Networks. By leveraging the unique attributes of this data set, we trained baseline models and explored the prevailing challenges in deep-learning-based Pn and Sn phase picking as the scope transitions from local to regional epicenter distances. Our results show that the performance of the model is considerably enhanced after training on the proposed data set. Our study is a significant complement to the data foundation for future data-driven Pn and Sn seismic phase-picking studies, which will contribute to enhancing our understanding of the uppermost mantle structure of Earth, for example, the seismic velocity, anisotropy, and attenuation characteristics.
地震相 Pn 和 Sn 在研究最上地幔的速度和各向异性特征方面发挥着至关重要的作用。然而，人工标注这些地震相既费时，又容易产生主观解释。因此，这些地震相的走时数据的使用仍然有限。尽管深度学习具有应对这一挑战的潜力，但 Pn 和 Sn 大量训练数据集的稀缺带来了巨大的限制。为了应对这一挑战，我们的研究编制了一个全球百万规模的 Pn 和 Sn 震级基准数据集，即 Seis-PnSn。该数据集来自震中距 1.8° 至 18° 的地震事件。本研究使用的高质量走时数据全部来自国际地震中心，时间跨度为 2000 年至 2019 年。波形数据来自国际数字地震仪网络联合会下位于世界不同地区的数据中心。通过利用该数据集的独特属性，我们训练了基线模型，并探索了当范围从本地震中距离过渡到区域震中距离时，基于深度学习的 Pn 和 Sn 相位拾取所面临的挑战。我们的结果表明，在拟议的数据集上进行训练后，模型的性能大大提高。我们的研究是对未来数据驱动的 Pn 和 Sn 地震选相研究的数据基础的重要补充，这将有助于增强我们对地球最上地幔结构的理解，例如地震速度、各向异性和衰减特征。
{"title":"Seis-PnSn: A Global Million-Scale Benchmark Data Set of Pn and Sn Seismic Phases for Deep Learning","authors":"Hua Kong, Zhuowei Xiao, Yan Lü, Juan Li","doi":"10.1785/0220230379","DOIUrl":"https://doi.org/10.1785/0220230379","url":null,"abstract":"\u0000 The seismic phases Pn and Sn play a crucial role in investigating the velocity and anisotropic characteristics of the uppermost mantle. However, manually annotating these phases can be time-intensive and prone to subjective interpretation. Consequently, the use of travel-time data for these seismic phases remains limited. Despite the potential of deep learning to address this challenge, the scarcity of extensive training data sets for Pn and Sn presents significant constraints. To address this challenge, our research compiled a global million-scale benchmark data set of Pn and Sn seismic phases, namely Seis–PnSn. The data set is derived from earthquake events with epicenter distances ranging from 1.8° to 18°. The high-quality travel-time data used in this study are all from the International Seismological Centre and span the period 2000 to 2019. The waveform data were sourced from data centers located in different regions of the world under the International Federation of Digital Seismograph Networks. By leveraging the unique attributes of this data set, we trained baseline models and explored the prevailing challenges in deep-learning-based Pn and Sn phase picking as the scope transitions from local to regional epicenter distances. Our results show that the performance of the model is considerably enhanced after training on the proposed data set. Our study is a significant complement to the data foundation for future data-driven Pn and Sn seismic phase-picking studies, which will contribute to enhancing our understanding of the uppermost mantle structure of Earth, for example, the seismic velocity, anisotropy, and attenuation characteristics.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":"33 12","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141688062","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ting Yang, L. Fang, Jianping Wu, Stephen Monna, Weimin Xu
High-precision seismic phase arrivals are a prerequisite for building reliable velocity models with travel-time tomography. There has recently been a growing use of seismic phase arrival data obtained through deep learning techniques in travel-time tomography research. Nevertheless, a significant challenge that has emerged pertains to the assessment of the quality of these automatic arrivals. In this article, we used PhaseNet, a deep learning method, to automatically detect the arrival times of the P wave and S wave of 3086 seismic events recorded by dense seismic arrays, obtaining 87,553 high-quality arrivals. To evaluate the quality of the arrival times subsequently used for travel-time tomography inversion, we applied a weighting scheme that includes both detection probability value and signal-to-noise ratio. This new weighting scheme can effectively reduce the overall travel-time residual by 7%. The weighted data were then used in the double-difference tomography method to invert for the crustal velocity structure of the Anninghe–Xiaojiang fault zone. The resulting new model exhibits a lateral resolution of up to 0.25° and reveals velocity anomalies that exhibit a strong correlation with major geological features and block boundaries. Notably, the presence of low-VP and low-VS in the middle crust of the Ludian–Qiaojia seismic zone suggests the existence of hot and weak felsic rocks, as well as possible fluid presence beneath the seismogenic layer of this area. This study not only validates the practicality of using deep learning-based phase picking arrivals in travel-time tomography but also proposes a new weighting scheme to refine the tomographic velocity models.
高精度地震相位到达数据是利用走时层析技术建立可靠速度模型的先决条件。最近，通过深度学习技术获得的地震相位到达数据越来越多地被应用于走时层析成像研究。然而，在评估这些自动到达数据的质量方面出现了重大挑战。在本文中，我们使用深度学习方法 PhaseNet 自动检测了密集地震阵列记录的 3086 个地震事件的 P 波和 S 波到达时间，获得了 87,553 个高质量到达时间。为了评估随后用于走时层析反演的到达时间的质量，我们采用了一种包括检测概率值和信噪比的加权方案。这种新的加权方案可以有效地将整体旅行时间残差降低 7%。加权后的数据被用于双差分层析反演法，以反演安宁河-小江断裂带的地壳速度结构。新模型的横向分辨率高达 0.25°，并揭示了与主要地质特征和区块边界密切相关的速度异常。值得注意的是，鲁甸-巧家地震带中层地壳中存在低 VP 和低 VS，这表明该地区存在热弱长英岩，并可能在成震层下存在流体。这项研究不仅验证了在走时层析成像中使用基于深度学习的相位选取到达的实用性，还提出了一种新的加权方案来完善层析速度模型。
{"title":"Double-Difference Tomography with a Deep Learning–Based Phase Arrival Weighting Scheme and Its Application to the Anninghe–Xiaojiang Fault Zone","authors":"Ting Yang, L. Fang, Jianping Wu, Stephen Monna, Weimin Xu","doi":"10.1785/0220230362","DOIUrl":"https://doi.org/10.1785/0220230362","url":null,"abstract":"\u0000 High-precision seismic phase arrivals are a prerequisite for building reliable velocity models with travel-time tomography. There has recently been a growing use of seismic phase arrival data obtained through deep learning techniques in travel-time tomography research. Nevertheless, a significant challenge that has emerged pertains to the assessment of the quality of these automatic arrivals. In this article, we used PhaseNet, a deep learning method, to automatically detect the arrival times of the P wave and S wave of 3086 seismic events recorded by dense seismic arrays, obtaining 87,553 high-quality arrivals. To evaluate the quality of the arrival times subsequently used for travel-time tomography inversion, we applied a weighting scheme that includes both detection probability value and signal-to-noise ratio. This new weighting scheme can effectively reduce the overall travel-time residual by 7%. The weighted data were then used in the double-difference tomography method to invert for the crustal velocity structure of the Anninghe–Xiaojiang fault zone. The resulting new model exhibits a lateral resolution of up to 0.25° and reveals velocity anomalies that exhibit a strong correlation with major geological features and block boundaries. Notably, the presence of low-VP and low-VS in the middle crust of the Ludian–Qiaojia seismic zone suggests the existence of hot and weak felsic rocks, as well as possible fluid presence beneath the seismogenic layer of this area. This study not only validates the practicality of using deep learning-based phase picking arrivals in travel-time tomography but also proposes a new weighting scheme to refine the tomographic velocity models.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":"56 8","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141688580","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Zhengya Si, Jiancang Zhuang, Stefania Gentili, Changsheng Jiang, Weitao Wang
We introduce a Bayesian algorithm designed to integrate earthquake magnitudes of the same type reported by various seismic networks, aiming to create unified and standardized catalogs suitable for widespread use. The fundamental concept underpinning this algorithm is the utilization of the inherent consistency within each individual network’s magnitude determination process. Assuming that the magnitudes for an earthquake measured by all networks conform to a Gaussian distribution, with a linear function of the unknown true magnitude serving as its mean, we derive the posterior probability distribution of the true magnitude under four different assumptions for the prior distribution: the uninformative uniform distribution, the unbounded Gutenberg–Richter (GR) magnitude–frequency law, the GR magnitude–frequency relationship restricted by the detection rate, and the truncated GR law as priors. We assess the robustness of the method by a test on several synthetic catalogs and then use it to merge the catalogs compiled by five seismic networks in Italy. The results demonstrate that our proposed magnitude-merging algorithm effectively combines the catalogs, resulting in robust and unified data sets that are suitable for seismic hazard assessment and seismicity analysis.
我们介绍了一种贝叶斯算法，旨在整合不同地震台网报告的同类型地震震级，从而创建适合广泛使用的统一标准化目录。该算法的基本概念是利用每个地震台网震级确定过程中固有的一致性。假定所有地震台网测得的震级都符合高斯分布，并以未知真实震级的线性函数作为其均值，我们推导出了在四种不同先验分布假设下真实震级的后验概率分布：无信息均匀分布、无约束古腾堡-里克特（GR）震级-频率定律、受探测率限制的 GR 震级-频率关系，以及作为先验的截断 GR 定律。我们通过对几个合成目录的测试来评估该方法的稳健性，然后用它来合并意大利五个地震台网编制的目录。结果表明，我们提出的震级合并算法有效地合并了震级目录，得到了稳健而统一的数据集，适用于地震灾害评估和地震度分析。
{"title":"A Bayesian Merging of Earthquake Magnitudes Determined by Multiple Seismic Networks","authors":"Zhengya Si, Jiancang Zhuang, Stefania Gentili, Changsheng Jiang, Weitao Wang","doi":"10.1785/0220230404","DOIUrl":"https://doi.org/10.1785/0220230404","url":null,"abstract":"\u0000 We introduce a Bayesian algorithm designed to integrate earthquake magnitudes of the same type reported by various seismic networks, aiming to create unified and standardized catalogs suitable for widespread use. The fundamental concept underpinning this algorithm is the utilization of the inherent consistency within each individual network’s magnitude determination process. Assuming that the magnitudes for an earthquake measured by all networks conform to a Gaussian distribution, with a linear function of the unknown true magnitude serving as its mean, we derive the posterior probability distribution of the true magnitude under four different assumptions for the prior distribution: the uninformative uniform distribution, the unbounded Gutenberg–Richter (GR) magnitude–frequency law, the GR magnitude–frequency relationship restricted by the detection rate, and the truncated GR law as priors. We assess the robustness of the method by a test on several synthetic catalogs and then use it to merge the catalogs compiled by five seismic networks in Italy. The results demonstrate that our proposed magnitude-merging algorithm effectively combines the catalogs, resulting in robust and unified data sets that are suitable for seismic hazard assessment and seismicity analysis.","PeriodicalId":508466,"journal":{"name":"Seismological Research Letters","volume":"20 3","pages":""},"PeriodicalIF":0.0,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141687000","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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：