Towards global seismic monitoring of underground nuclear explosions using waveform cross correlation. Part II: Synthetic master events

Waveform cross correlation is an efficient tool for detection and characterization of seismic signals. For the purposes of the Comprehensive Nuclear-Test-Ban Treaty, cross correlation can globally reduce the threshold of detection by 0.3 to 0.4 magnitude units. However, the technique critically depends on the availability of master events. In Part I of this paper, we have demonstrated that in seismically active regions the best master events (grand-masters) replicated over a regular grid allow improving the efficiency of signal detection and event finding. In aseismic areas, there are two approaches to populate the global grid of master events for the International Monitoring System: the replication of grand-masters and calculation of synthetic seismograms for master-events in the global grid nodes. The efficiency of synthetic templates depends on the accuracy of shape and amplitude predictions controlled by focal depth and mechanism, source function, velocity structure and attenuation along the master/station path. Here we test three focal mechanisms (explosion, thrust fault, and actual Harvard CMT solution for one of the April 11, 2012 Sumatra aftershocks) and two velocity structures (ak135 and CRUST 2.0). Sixteen synthetic master events were distributed over a 1° × 1° grid covering the zone of aftershocks. We built five cross correlation standard event lists (XSEL) and compared detections and events with those built using the real and grand master events as well as with the Reviewed Events Bulletin of the International Data Centre. The XSELs were built using an explosion source and ak135 and the reverse fault with isotropic radiation pattern to demonstrate the performance similar to that of the real and grand masters. Here we have proved quantitatively that it is possible to cover all aseismic areas with synthetic masters without significant loss in seismic monitoring capabilities based on cross correlation.     

Perspectives of Cross-Correlation in Seismic Monitoring at the International Data Centre

We demonstrate that several techniques based on waveform cross-correlation are able to significantly reduce the detection threshold of seismic sources worldwide and to improve the reliability of arrivals by a more accurate estimation of their defining parameters. A master event and the events it can find using waveform cross-correlation at array stations of the International Monitoring System (IMS) have to be close. For the purposes of the International Data Centre (IDC), one can use the spatial closeness of the master and slave events in order to construct a new automatic processing pipeline: all qualified arrivals detected using cross-correlation are associated with events matching the current IDC event definition criteria (EDC) in a local association procedure. Considering the repeating character of global seismicity, more than 90 % of events in the reviewed event bulletin (REB) can be built in this automatic processing. Due to the reduced detection threshold, waveform cross-correlation may increase the number of valid REB events by a factor of 1.5–2.0. Therefore, the new pipeline may produce a more comprehensive bulletin than the current pipeline—the goal of seismic monitoring. The analysts’ experience with the cross correlation event list (XSEL) shows that the workload of interactive processing might be reduced by a factor of two or even more. Since cross-correlation produces a comprehensive list of detections for a given master event, no additional arrivals from primary stations are expected to be associated with the XSEL events. The number of false alarms, relative to the number of events rejected from the standard event list 3 (SEL3) in the current interactive processing—can also be reduced by the use of several powerful filters. The principal filter is the difference between the arrival times of the master and newly built events at three or more primary stations, which should lie in a narrow range of a few seconds. In this study, one event at a distance of about 2,000 km from the main shock was formed by three stations, with the stations and both events on the same great circle. Such spurious events are rejected by checking consistency between detections at stations at different back azimuths from the source region. Two additional effective pre-filters are f–k analysis and F _prob based on correlation traces instead of original waveforms. Overall, waveform cross-correlation is able to improve the REB completeness, to reduce the workload related to IDC interactive analysis, and to provide a precise tool for quality check for both arrivals and events. Some major improvements in automatic and interactive processing achieved by cross-correlation are illustrated using an aftershock sequence from a large continental earthquake. Exploring this sequence, we describe schematically the next steps for the development of a processing pipeline parallel to the existing IDC one in order to improve the quality of the REB together with the reduction of the magnitude threshold.     

Locating Seismic Events in the CTBT Context

—?The verification of the Comprehensive Nuclear-Test-Ban Treaty (CTBT) requires the determination of accurate location of seismic events from a fixed network of seismic stations across the globe. The requirements of possible on-site inspections mean that the goal is to place the location estimate in a zone smaller than 1000 km² that includes the true location. Because a defined set of stations will be used, corrections can be refined to represent the influence of departures from the global reference model IASPEI91. The primary stations in the International Monitoring Scheme (IMS) are mostly seismic arrays and therefore the present location scheme is based on minimisation of a misfit function built from arrival time, azimuth and array slowness residuals. The effective network will change markedly with the magnitude of the event and as a result regional information has to be integrated into the location process.     

Slowness Corrections — One Way to Improve IDC Products

—?The first step to identify and locate a seismic event is the association of observed onsets with common seismic sources. This is especially important in the context of monitoring the Comprehensive Nuclear-Test-Ban Treaty (CTBT) at the International Data Center (IDC) being developed in Vienna, Austria. Well-defined slowness measurements are very useful for associating seismic phases to presumed seismic events.¶Shortly after installation of the first seismic arrays, systematic discrepancies between measured and theoretically predicted slowness values were observed, and therefore slowness measurements of seismic stations should be calibrated. The observed slownesses measured with small aperture arrays, some of which will be included in the International Monitoring System (IMS) now being implemented for verifying compliance with the CTBT, show large scatter and deviations from theoretically expected values. However, in this study a method is presented, by which mean slowness corrections can be derived, which show relatively stable patterns specific to each array.¶The correction of measured slowness values of these arrays clearly improved the single array location capabilities. Applying slowness corrections with seismic phases observed by ARCES, FINES, GERES, and NORES, and associated to seismic events in the bulletins of the prototype International Data Center (pIDC) in Arlington, VA, also clearly demonstrates the advantages of these corrections. For arrays with large slowness deviations that are due to the influence of a dipping layer, the corrections were modeled with a sine function depending on the measured azimuth. In addition, the measured values can be weighted with the corresponding uncertainties known from the process of deriving the mean corrections.     

Best Practice in Earthquake Location Using Broadband Three-component Seismic Waveform Data

—?We present an earthquake location algorithm, the Broadband Waveform Regional Earthquake Location Program (BW_RELP), which utilizes phase onset times and wave azimuths recorded by three-component broadband seismic stations and an adaptive migrating grid search algorithm to find the global minimum in an arbitrary normed misfit parameter. The performance of BW_RELP is demonstrated using regional (300–800?km distant) broadband recordings to locate events in the 1995 Ridgecrest, California earthquake sequence. The purpose of this study is to introduce the BW_RELP algorithm in detail and to expand on the previous paper by Deger et?al. (BSSA, 88, 1353–1362, 1998), using one Berkeley Digital Seismic Network (BDSN) station (YBH) and two USNSN stations (ELK and MNV) which span 300–800?km in distance and 55 degrees in azimuth, to further investigate the capability of a sparse broadband network of three-component stations at monitoring a region located outside of the network, as will be the case in the monitoring of the Comprehensive Test-Ban-Treaty (CTBT) for low magnitude seismic events. We assess the capability of this sparse three-station broadband network and we compare locations estimated from phase onset time and wave azimuth measurements to a ground-truth catalog of high-quality earthquake locations derived from data recorded by the Southern California Seismic Network (SCSN). The results indicate that in the regional distance range it is possible, when an appropriate calibration event is available, to obtain absolute event locations to within 18?km as is prescribed by the CTBT.     

Detection,Estimation of Magnitude,and Relative Location of Weak Aftershocks Using Waveform Cross-Correlation: The Earthquake of August 7, 2016, in the Town of Mariupol

The study of the stress-strain state of a medium in seismically quiet areas is difficult because of the absence of strong events. Under such circumstances, each earthquake, even relatively weak, is of high importance. In this case, all possible information on tectonic stresses and their dynamics, e.g., information on time, location, and magnitude of aftershocks, should be obtained from available seismic data. The earthquake near the town of Mariupol which occurred on August 7, 2016, had a body wave magnitude of 4.5–4.9 from the data of the different seismological centers. We detected 12 aftershocks that occurred within 5 days after the main shock using two seismic arrays (AKASG and BRTR) and one three-component station (KBZ) of the International Monitoring System, as well as two array stations of the Institute of Geosphere Dynamics, Russian Academy of Sciences. For six aftershocks, signals were found at three or more stations. The other aftershocks were detected from the data at two out of three nearest stations. Signal detection and association with aftershocks of the main shock, as well as estimation of magnitude and relative location of the found aftershocks, were carried out using the method of waveform cross-correlation (WFCC). The signals from the main shock that acted as the only master event (ME) for the WFCC method were used as waveform templates. To increase the signal-to-noise ratio and to determine the exact onset time of regular seismic waves from aftershocks, we used waveform templates of different length, from 10 to 180 s depending on the wave type and distance to the station, as well as band filtering in narrow frequency bands. The highest sensitivity of the detector and accuracy of the P-wave onset time estimates were reached when a waveform template included all regular waves from P to L _g . Association of signals with aftershocks was based on back projection of signal arrival times to origin times using the travel time from a master event to the station, which was measured with a very low error, being equal to almost half of the digitization step length. To develop a seismic event hypothesis, the origin times at two or more stations should be spaced within a 2-s interval.     

浙江临安地震台阵勘选测试与背景噪声水平分析

在大中城市周边建设地震台阵,可解决城市内地震监测台站建设不便和背景噪声高等问题,增强人口密集区的地震监测能力。杭州湾地区沉积层较厚,建设基岩地震台站不够现实,为适应长三角区域的一体化发展,增强该区地震和非天然地震的监测能力,在浙江临安建设地震台阵。使用GL-PS60一体化宽频带地震计进行测点勘选工作,利用Welch平均周期法,对9个子台进行背景噪声功率谱估计。结果发现,背景噪声功率谱相关性较好,且与全球噪声模型一致性较好。在台阵勘选过程中,于2017年4月12日和13日分别记录到临安4.2级地震及菲律宾远震事件,波形清晰,表明初选台址效果较好,符合地震台阵建设要求。     

The detection and location of low magnitude earthquakes in northern Norway using multi-channel waveform correlation at regional distances

A fortuitous sequence of closely spaced earthquakes in the Rana region of northern Norway, during 2005, has provided an ideal natural laboratory for investigating event detectability using waveform correlation over networks and arrays at regional distances. A small number of events between magnitude 2.0 and 3.5 were recorded with a high SNR by the Fennoscandian IMS seismic arrays at distances over 600 km and three of these events, including the largest on 24 June, displayed remarkable waveform similarity even at relatively high frequencies. In an effort to detect occurrences of smaller earthquakes in the immediate geographical vicinity of the 24 June event, a multi-channel correlation detector for the NORSAR array was run for the whole calender year 2005 using the signal from the master event as a template. A total of 32 detections were made and all but 2 of these coincided with independent correlation detections using the other Nordic IMS array stations; very few correspond to signals detectable using traditional energy detectors. Permanent and temporary stations of the Norwegian National Seismic Network (NNSN) at far closer epicentral distances have confirmed that all but one of the correlation detections at NORSAR in fact correspond to real events. The closest stations at distances of approximately 10 km can confirm that the smallest of these events have magnitudes down to 0.5 which represents a detection threshold reduction of over 1.5 for the large-aperture NORSAR array and over 1.0 for the almost equidistant regional ARCES array. The incompleteness of the local network recordings precludes a comprehensive double-difference location for the full set of events. However, stable double-difference relative locations can be obtained for eight of the events using only the Lg phase recorded at the array stations. All events appear to be separated by less than 0.5 km. Clear peaks were observed in the NORSAR correlation coefficient traces during the coda of some of the larger events; the local stations confirm that these are in fact aftershocks exhibiting very similar waveforms to the main events. Many of the more marginal correlation detections are not made when the calculations are repeated using shorter signal segments, fewer sensors or more distant stations. We demonstrate in addition how these almost repeating seismic sources have been exploited to detect and measure timing anomalies at individual sites within the arrays and network.     

利用地震背景噪声互相关检测流动地震台波形时钟漂移

时钟精度是地震观测中最重要的参数之一, 需要通过卫星授时信号来保障。 当卫星信号被屏蔽或仪器守时部件出现问题时, 地震计内部时钟会逐渐漂移, 给后续数据处理带来极大困扰。 利用地震背景噪声台站对互相关提取的经验格林函数, 不仅可以用于结构成像, 也可以用于检测波形时钟是否存在漂移, 并获得时钟漂移幅度。 使用地震背景噪声互相关方法对2017—2019年云南永胜地区4个流动观测台站的连续波形进行了时钟漂移检测, 结果发现部分流动台站在不同时间段存在不同模式的时钟漂移, 最大幅度可达到1.75 s。 同时, 利用云南宾川气枪重复震源激发的信号进行互相关计算, 对上述结果进行了验证, 发现两种结果具有较好的一致性。 研究表明, 背景噪声互相关对波形的时钟漂移有较高的灵敏度, 能够有效检测出时钟问题, 防止波形被误用, 可为后续波形时钟校正提供参考信息。     

天山中段地区重复地震时空特征研究

基于2009—2017年新疆区域数字地震台网记录的地震波形数据,利用波形互相关技术及主事件定位方法识别并重新定位了新疆天山中段及其周缘的重复地震。以波形互相关系数0.9作为阈值来确定研究区的重复地震事件,统计结果显示3万零181个事件中的1万1 618个为重复地震事件,这些重复地震事件组成了2395组重复地震对和重复地震丛,占总事件数的38.5%。根据重复地震重定位前后地震对之间距离的统计结果推测,该区域的系统定位误差约为5—10 km。进一步结合该区域最新的震源分类结果对不同震源类型重复地震的时空分布特征予以分析,结果显示:重复矿山爆破事件在空间上呈丛集性分布,且其中的93.6%发生于白天,同时呈现季节性发生模式,即爆破多发生于夏季,而冬季较少;而重复构造地震在空间上大多沿断层分布,24小时内呈随机分布的特征,且研究时段内每个月的活动水平相对平稳;重复诱发地震成丛分布于靠近油气田和水库的区域,但其中部分诱发地震的位置与构造地震重叠,发震时间特征与构造地震相似,为随机分布。      

A new stacking technique in ambient noise tomography

We have developed a new stacking technique in ambient noise tomography to obtain high-quality dispersion curves of Rayleigh waves.This technique is used to stack the vertical components of the Estimated Green Functions(EGFs) obtained respectively from cross correlation of the ambient noise data recorded by a remote seismic station and one of the short distance seismic stations of a seismic array.It is based on a phase-matched filter and is implemented by a four-step iterative process:signal compression,stacking,signal extraction and signal decompression.The iterative process ends and gives the dispersion curve of Rayleigh wave when the predicted one and the processing result converge.We have tested the method using the vertical components of synthetic Rayleigh wave records.Results show that this new stacking method is stable and it can improve the quality of dispersion curves.In addition,we have applied this method to real data.We see that the results given by our new technique are obviously better than the ones employing the traditional method which is a three-step process:signal compression,signal extraction and signal decompression.In conclusion,the new method proposed in this paper can improve the signal to noise ratio of EGFs,and can therefore potentially improve the resolution of ambient noise tomography.     

结合波形互相关技术的双差算法在地震定位中的应用探讨

本文致力于探讨在数字化地震监测大规模发展后,如何更好地将大量丰富的数字化波形信息应用到地震定位领域,从而更大程度地提高地震定位精度问题。笔者认为"结合波形互相关技术的双差定位算法"是一种比较有发展前景的方法,因此分别从基本理论、发展历程、台网实用化过程以及推广使用中可能存在的问题等角度进行分析,探讨该技术在地震定位领域中的发展方向及推广使用的可能性。     

地球背景噪声干涉应用研究的新进展

对地震台记录到的长时间的地球背景噪声进行互相关处理,得到的互相关函数可以近似地表征这两个台站之间的格林函数,这种数据处理方法被称为背景噪声干涉.近年来,背景噪声干涉研究在理论和实践上都有了长足进展,并被广泛用于地下波速结构及其演化规律的研究.本文首先简单介绍了背景噪声干涉研究的概念和理论依据,进而重点介绍了针对噪声源性质、互相关函数的波形到时和振幅特性的应用研究.通过对这三个方面研究的回顾,分析总结了该研究领域的新进展及需要解决的问题,并对背景噪声干涉研究的新特点进行了分析和展望.     

Sources of Error and the Statistical Formulation of M S: m b Seismic Event Screening Analysis

The Comprehensive Nuclear-Test-Ban Treaty (CTBT), a global ban on nuclear explosions, is currently in a ratification phase. Under the CTBT, an International Monitoring System (IMS) of seismic, hydroacoustic, infrasonic and radionuclide sensors is operational, and the data from the IMS is analysed by the International Data Centre (IDC). The IDC provides CTBT signatories basic seismic event parameters and a screening analysis indicating whether an event exhibits explosion characteristics (for example, shallow depth). An important component of the screening analysis is a statistical test of the null hypothesis H ₀: explosion characteristics using empirical measurements of seismic energy (magnitudes). The established magnitude used for event size is the body-wave magnitude (denoted m _b) computed from the initial segment of a seismic waveform. IDC screening analysis is applied to events with m _b greater than 3.5. The Rayleigh wave magnitude (denoted M _S) is a measure of later arriving surface wave energy. Magnitudes are measurements of seismic energy that include adjustments (physical correction model) for path and distance effects between event and station. Relative to m _b, earthquakes generally have a larger M _S magnitude than explosions. This article proposes a hypothesis test (screening analysis) using M _S and m _b that expressly accounts for physical correction model inadequacy in the standard error of the test statistic. With this hypothesis test formulation, the 2009 Democratic Peoples Republic of Korea announced nuclear weapon test fails to reject the null hypothesis H ₀: explosion characteristics.     

Use of GSETT-3 gamma data in the Slowness-Azimuth Calibration of IMS primary arrays at regional distances

For the purpose of verifying compliance with the CTBT seismic monitoring is one of the four techniques used by the IDC. In order to improve the accuracy of the automatic and the reviewed bulletin epicenter locations the IDC uses SASC for the IMS seismic stations. SASC determination is a straightforward calculation done by comparing for selected events the azimuth and slowness from the waveform processing using array techniques to the theoretical values based on the event locations and the velocity model.The main problem, however, is to build a set of reference events, whose locations are accurate enough and not based on information from the stations to be calibrated. A reference event list assumed to meet this requirement is the Gamma bulletin, which was collected since 1993 and was compiled in the framework of the GSETT-3. In this work calculation of SASC for regional to teleseismic distances (up to 30 degrees) was performed for 11 IMS primary arrays. The calculation was done using Pg, Pn, P, Sg, Sn, and S phases based on the detection list obtained from the pIDC and the Gamma bulletin for 6 years (1993–1999). The number of Gamma events varies from several hundreds for some arrays (BRAR) to several tens of thousands for others (i.e. ARCES, ILAR). Due to the fact that the Gamma bulletin is purely voluntary, the coverage is non-uniform both in time and in space and the location accuracy is non-uniform. This drawback can be overcome by encouraging signatory states to submit quality Gamma bulletin data to the IDC. The work presented here can be used as a routine procedure for improving IMS array performance, especially at regional distances.     

数字地震台网大震前地脉动信息变化特征研究

通过计算机对数字地震台网产出的大动态、高分辨率、高精度的地震波形数据进行的实时采集跟踪分析，发现了有些台站大震前地脉动波形的周期和幅度有明显异常变化，探索大震前地脉动变化的特征和规律，将为区域性地震预报研究开辟一条新的途径。     

Improving Regional Seismic Event Location in China

—?In an effort to improve our ability to locate seismic events in China using only regional data, we have developed empirical propagation path corrections and applied such corrections using traditional location routines. Thus far, we have concentrated on corrections to observed P arrival times for crustal events using travel-time observations available from the USGS Earthquake Data Reports, the International Seismic Centre Bulletin, the preliminary International Data Center Reviewed Event Bulletin, and our own travel-time picks from regional data. Location ground truth for events used in this study ranges from 25?km for well-located teleseimic events, down to 2?km for nuclear explosions located using satellite imagery. We also use eight events for which depth is constrained using several waveform methods. We relocate events using the EvLoc algorithm from a region encompassing much of China (latitude 20°–55°N; longitude 65°–115°E). We observe that travel-time residuals exhibit a distance-dependent bias using IASPEI91 as our base model. To remedy this bias, we have developed a new 1-D model for China, which removes a significant portion of the distance bias. For individual stations having sufficient P-wave residual data, we produce a map of the regional travel-time residuals from all well-located teleseismic events. Residuals are used only if they are smaller than 10?s in absolute value and if the seismic event is located with accuracy better than 25?km. From the residual data, correction surfaces are constructed using modified Bayesian kriging. Modified Bayesian kriging offers us the advantage of providing well-behaved interpolants and their errors, but requires that we have adequate error estimates associated with the travel-time residuals from which they are constructed. For our P-wave residual error estimate, we use the sum of measurement and modeling errors, where measurement error is based on signal-to-noise ratios when available, and on the published catalog estimate otherwise. Our modeling error originates from the variance of travel-time residuals for our 1-D China model. We calculate propagation path correction surfaces for 74 stations in and around China, including six stations from the International Monitoring System. The statistical significance of each correction surface is evaluated using a cross-validation technique. We show relocation results for nuclear tests from the Balapan and Lop Nor test sites, and for earthquakes located using interferometric synthetic aperture radar. These examples show that the use of propagation path correction surfaces in regional relocations eliminates distance bias in the residual curves and significantly improves the accuracy and precision of seismic event locations.     

青海省新增地震台波形数据质量评价

为了进一步提升青藏高原地震监测能力,中国地震局在青海、西藏和新疆3个地区建设地震台站。为确保高质量产出地震数据,以青海地区新增16个地震台站为例,围绕观测数据的完整性和可靠性,基于数据连续率、台基噪声水平和事件波形记录质量,对地震观测数据质量进行评价,结果发现：青海地区新增地震台站观测环境较好,总体运行稳定,数据连续率较高,适合进行宽频带数字地震观测;地震台站空间分布更加均匀,青海省中西部地区理论地震监测能力从约M_S 4.5提升到M_S 2.0,地震监测能力提升。本项工作的开展,为我国青藏高原地震监测能力提升提供了业务化支撑。     

A Machine Learning Approach for Improving the Detection Capabilities at 3C Seismic Stations

We apply and evaluate a recent machine learning method for the automatic classification of seismic waveforms. The method relies on Dynamic Bayesian Networks (DBN) and supervised learning to improve the detection capabilities at 3C seismic stations. A time-frequency decomposition provides the basis for the required signal characteristics we need in order to derive the features defining typical “signal” and “noise” patterns. Each pattern class is modeled by a DBN, specifying the interrelationships of the derived features in the time-frequency plane. Subsequently, the models are trained using previously labeled segments of seismic data. The DBN models can now be compared against in order to determine the likelihood of new incoming seismic waveform segments to be either signal or noise. As the noise characteristics of seismic stations varies smoothly in time (seasonal variation as well as anthropogenic influence), we accommodate in our approach for a continuous adaptation of the DBN model that is associated with the noise class. Given the difficulty for obtaining a golden standard for real data (ground truth) the proof of concept and evaluation is shown by conducting experiments based on 3C seismic data from the International Monitoring Stations, BOSA and LPAZ.     

远震P波层析成像研究阿尔金断裂带东端及其邻区深部结构

本文利用"中国地震科学台阵探测——南北地震带北段"项目在内蒙古阿拉善西部及甘肃西北部地区布设的80个流动宽频带地震仪及16个固定台站,于2013年10月—2015年6月所记录的787个远震事件,采用波形相关方法拾取了共49052个高质量的P波走时残差数据,并利用Fast-Marching远震走时层析成像方法,反演获取了研究区下方的三维P波速度结构.结果显示:阿尔金断裂带东段、祁连山、北山地区下方地壳结构表现为低速异常特征,具有明显的造山带构造特征;阿拉善地体下方地壳结构表现为高速异常特征,为典型的大陆地壳结构特征;阿拉善地块沿着青藏高原北边界逆冲断裂(NBT)南向俯冲,其在祁连山造山带下与北向俯冲的柴达木岩石圈形成了面对面的碰撞接触关系;阿尔金断裂带的末端并没有北东向延伸到阿拉善块体,而是受到刚性的阿拉善岩石圈阻挡沿着其南缘断裂带继续向东发展.