地震信号处理中相位匹配滤波器的设计

相位匹配滤波器是一组线性滤波器,它的傅里叶相位与被检测信号的傅里叶相位相等。本文阐述了相位匹配滤波器的基本原理,并根据已获得的特定路径的频散曲线完成了理论设计,实现了对小震级事件面波信号的检测。使用该相位匹配滤波器对内华达地区的18次核爆事件的面波信号进行检测并计算了面波震级,与ISC公报给出的面波震级比较,相对误差很小。证明了本设计对面波信号的检测是有效的,并且可提高小震级面波信号的信噪比。     

A Surface Wave Dispersion Study of the Middle East and North Africa for Monitoring the Comprehensive Nuclear-Test-Ban Treaty

—?We present results from a large-scale study of surface-wave group velocity dispersion across the Middle East, North Africa, southern Eurasia and the Mediterranean. Our database for the region is populated with seismic data from regional events recorded at permanent and portable broadband, three-component digital stations. We have measured the group velocity using a multiple narrow-band filter on deconvolved displacement data. Overall, we have examined more than 13,500 seismograms and made good quality dispersion measurements for 6817 Rayleigh- and 3806 Love-wave paths. We use a conjugate gradient method to perform a group-velocity tomography. Our current results include both Love- and Rayleigh-wave inversions across the region for periods from 10 to 60 seconds. Our findings indicate that short-period structure is sensitive to slow velocities associated with large sedimentary features such as the Mediterranean Sea and Persian Gulf. We find our long-period Rayleigh-wave inversion is sensitive to crustal thickness, such as fast velocities under the oceans and slow along the relatively thick Zagros Mts. and Turkish-Iranian Plateau. We also find slow upper mantle velocities along known rift systems. Accurate group velocity maps can be used to construct phase-matched filters along any given path. The filters can improve weak surface wave signals by compressing the dispersed signal. The signals can then be used to calculate regionally determined M _S measurements, which we hope can be used to extend the threshold of m _b :M _S discriminants down to lower magnitude levels. Other applications include using the group velocities in the creation of a suitable background model for forming station calibration maps, and using the group velocities to model the velocity structure of the crust and upper mantle.     

Microtremor Measurements for the Microzonation of Dinar

v--vThe geotechnical site conditions in Dinar town located in western Turkey were investigated after the 1995 Dinar earthquake based on borings, in situ penetration tests, seismic wave velocity measurements, and microtremor records. The variation of damage distribution within the town was evaluated with respect to 23 district damage ratios calculated, based on the detailed damage survey conducted by the General Directorate of Disaster Affairs. Site amplifications were estimated from microtremor spectral ratios and microzonation was performed using a GIS methodology. The results of in situ penetration tests and seismic wave velocity measurements as well as the damage distribution were compared with the amplification zonation obtained from microtremor records. The results indicate the applicability of microtremor spectral ratios for assessing the local site conditions and site amplifications.     

Optimized Seismic Threshold Monitoring – Part 2: Teleseismic Processing

v--vThis second paper (Part 2) pertaining to optimized site-specific threshold monitoring addresses the application of the method to regions covered by a teleseismic or a combined regional-teleseismic network. In the first paper (Part 1) we developed the method for the general case, and demonstrated its application to an area well-covered by a regional network (the Novaya Zemlya nuclear test site). In the present paper, we apply the method to the Indian and Pakistani nuclear test sites, and show results during the periods of nuclear testing by these two countries in May 1998. Since the coverage by regional stations in these areas is poor, an optimized approach requires the use of selected, high-quality stations at teleseismic distances.¶To optimize the threshold monitoring of these test sites, we use as calibration events either one of the nuclear explosions or a nearby earthquake. From analysis of the calibration events we derive values for array beamforming steering delays, filter bands, short-term averages (STA) lengths, phase travel times (P waves), and amplitude-magnitude relationships for each station. By applying these parameters, we obtain a monitoring capability of both test sites ranging from m_b 2.8-3.0 using teleseismic stations only. When including the nearby Nilore station to monitor the Indian tests, we show that the threshold can be reduced by about 0.4 magnitude units. In particular, we demonstrate that the Indian tests on 13 May, 1998, which were not detected by any known seismic station, must have corresponded to a magnitude (m_b) of less than 2.4.¶We also discuss the effect of a nearby aftershock sequence on the monitoring capability for the Pakistani test sites. Such an aftershock sequence occurred in fact on the day of the last Pakistani test (30 May, 1998), following a large (m_b 5.5) earthquake in Afghanistan located about 1100 km from the test site. We show that the threshold monitoring technique has sufficient resolution to suppress the signals from these interfering aftershocks without significantly affecting the true peak of the nuclear explosion on the threshold trace.     

Sedimentary Basins and the Blockage of Lg Wave Propagation in the Continents

v--vThe phenomenon of "Lg blockage," where Lg is strongly attenuated by crustal heterogeneities, poses a serious problem to CTBT monitoring because Lg is an important seismic phase for discrimination. This paper examines blockage in three continental regions where the Lg blockages may be caused by large, enclosed sedimentary basins along the propagation path. The Barents Sea Basin blocks Lg propagation across the Barents Sea from the Russian nuclear test sites at Novaya Zemlya to Scandinavian stations. Also, "early Lg" waves are observed in Sn codas on NORSAR, NORESS, and ARCESS recordings of Novaya Zemlya explosions where direct Lg is blocked. Early Lg waves may have resulted from Sn-to-Lg mode conversion at the contact between the Barents Basin and the Kola Peninsula. The Northern and Southern Caspian Sea Basins also block Lg waves from PNEs and earthquakes, perhaps due to thick, low-velocity, low-Q sediments replacing the granitic layer rocks in the crust. Lg blockage has also been observed in the Western Mediterranean/Levantine Basin due to low-Q sediments and crustal thinning. A "basin capture" model is proposed to explain Lg blockage in sedimentary basins. In this model, shear waves that reverberate in the crust and constitute the Lg wave train are captured, delayed, and attenuated by thick, low-velocity sediments that replace the "granitic" layer rocks of the upper crust along part of the propagation path. Sn waves, which propagate below the basin, would not be blocked and in fact, the blocked Lg waves may be diverted downward into Sn waves by the low velocity sediments in the basin.     

Parallel PSM/FDM Hybrid Simulation of Ground Motions from the 1999 Chi-Chi,Taiwan, Earthquake

-- A new technique for the parallel computing of 3-D seismic wave propagation simulation is developed by hybridizing the Fourier pseudospectral method (PSM) and the finite-difference method (FDM). This PSM/FDM hybrid offers a good speed-up rate using a large number of processors. To show the feasibility of the hybrid, a numerical 3-D simulation of strong ground motion was conducted for the 1999 Chi-Chi, Taiwan earthquake (M_w 7.6). Comparisons between the simulation results and observed waveforms from a dense strong ground motion network in Taiwan clearly demonstrate that the variation of the subsurface structure and the complex fault slip distribution greatly affect the damage during the Chi-Chi earthquake. The directivity effect of the fault rupture produced large S-wave pulses along the direction of the rupture propagation. Slips in the shallow part of the fault generate significant surface waves in Coastal Plain along the western coast. A large velocity gradient in the upper crust can propagate seismic waves to longer distances with minimum attenuation. The S waves and surface waves were finally amplified further by the site effect of low-velocity sediments in basins, and caused the significant disasters.     

Converted T Phases Recorded on Hawaii from Polynesian Nuclear Tests: A Preliminary Report

v--vWe present a preliminary study of T waves from Polynesian nuclear tests at Mururoa, recorded on digital stations of the Hawaii Volcano Observatory network, following their conversion to seismic waves at the southern shore of the Island of Hawaii, and subsequent propagation to the recording stations. We show that seismograms are composed of several packets, which can be interpreted as resulting from TMP and TMS conversions, and which feature distinct spectral characteristics. As the distance from the shoreline to the station increases, the relative importance of the several wave packets changes; a prominent shadow for TMP is found at 8-12 km from the shore. This pattern is affected by the local crustal structure; in a favorable case, propagation in deep, low-attenuation layers resulted in a clear record as far as 76 km from the shoreline. While these results are generally robust, they can be moderately affected by a change of location of the source inside Mururoa Atoll.     

Surface wave attenuation based polarization attributes in time-frequency domain for multicomponent seismic data

In the paper, we propose a surface wave suppression method in time-frequency domain based on the wavelet transform, considering the characteristic difference of polarization attributes, amplitude energy and apparent velocity between the effective signals and strong surface waves. First, we use the proposed method to obtain time–frequency spectra of seismic signals by using the wavelet transform and calculate the instantaneous polarizability at each point based on instantaneous polarization analysis. Then, we separate the surface wave area from the signal area based on the surface-wave apparent velocity and the average energy of the signal. Finally, we combine the polarizability, energy, and frequency characteristic to identify and suppress the signal noise. Model and field data are used to test the proposed filtering method.     

Optimized Seismic Threshold Monitoring – Part 1: Regional Processing

v--v Continuous seismic threshold monitoring is a technique that has been developed over the past several years to assess the upper magnitude limit of possible seismic events that might have occurred in a geographical target area. The method provides continuous time monitoring at a given confidence level, and can be applied in a site-specific, regional or global context.¶In this paper (Part 1) and a companion paper (Part 2) we address the problem of optimizing the site-specific approach in order to achieve the highest possible automatic monitoring capability of particularly interesting areas. The present paper addresses the application of the method to cases where a regional monitoring network is available. We have in particular analyzed events from the region around the Novaya Zemlya nuclear test site to develop a set of optimized processing parameters for the arrays SPITS, ARCES, FINES, and NORES. From analysis of the calibration events we have derived values for beam-forming steering delays, filter bands, short-term average (STA) lengths, phase travel times (P and S waves), and amplitude-magnitude relationships for each array. By using these parameters for threshold monitoring of the Novaya Zemlya testing area, we obtain a monitoring capability varying between m_b 2.0 and 2.5 during normal noise conditions.¶The advantage of using a network, rather than a single station or array, for monitoring purposes becomes particularly evident during intervals with high global seismic activity (aftershock sequences), high seismic noise levels (wind, water waves, ice cracks) or station outages. For the time period November-December 1997, all time intervals with network magnitude thresholds exceeding m_b 2.5 were visually analyzed, and we found that all of these threshold peaks could be explained by teleseismic, regional, or local signals from events outside the Novaya Zemlya testing area. We could therefore conclude within the confidence level provided by the method, that no seismic event of magnitude exceeding 2.5 occurred at the Novaya Zemlya test site during this two-month time interval.¶As an example of particular interest in a monitoring context, we apply optimized threshold processing of the SPITS array for a time interval around 16 August 1997 m_b 3.5 event in the Kara Sea. We show that this processing enables us to detect a second, smaller event from the same site (m_b 2.6), occurring about 4 hours later. This second event was not defined automatically by standard processing.     

Calibration of the Regional Crustal Waveguide and the Retrieval of Source Parameters Using Waveform Modeling

v--vRegional crustal waveguide calibration is essential to the retrieval of source parameters and the location of smaller (M < 4.8) seismic events. This path calibration of regional seismic phases is strongly dependent on the accuracy of hypocentral locations of calibration (or master) events. This information can be difficult to obtain, especially for smaller events. Generally, explosion or quarry blast generated travel-time data with known locations and origin times are useful for developing the path calibration parameters, but in many regions such data sets are scanty or do not exist. We present a method which is useful for regional path calibration independent of such data, i.e. with earthquakes, which is applicable for events down to M_w = 4 and which has successfully been applied in India, central Asia, western Mediterranean, North Africa, Tibet and the former Soviet Union. These studies suggest that reliably determining depth is essential to establishing accurate epicentral location and origin time for events. We find that the error in source depth does not necessarily trade-off only with the origin time for events with poor azimuthal coverage, but with the horizontal location as well, thus resulting in poor epicentral locations. For example, hypocenters for some events in central Asia were found to move from their fixed-depth locations by about 20 km. Such errors in location and depth will propagate into path calibration parameters, particularly with respect to travel times. The modeling of teleseismic depth phases (pP, sP) yields accurate depths for earthquakes down to magnitude M_w = 4.7. This M_w threshold can be lowered to four if regional seismograms are used in conjunction with a calibrated velocity structure model to determine depth, with the relative amplitude of the P_nl waves to the surface waves and the interaction of regional sPmP and pPmP phases being good indicators of event depths. We also found that for deep events a seismic phase which follows an S-wave path to the surface and becomes critical, developing a head wave by S to P conversion is also indicative of depth. The detailed characteristic of this phase is controlled by the crustal waveguide. The key to calibrating regionalized crustal velocity structure is to determine depths for a set of master events by applying the above methods and then by modeling characteristic features that are recorded on the regional waveforms. The regionalization scheme can also incorporate mixed-path crustal waveguide models for cases in which seismic waves traverse two or more distinctly different crustal structures. We also demonstrate that once depths are established, we need only two-stations travel-time data to obtain reliable epicentral locations using a new adaptive grid-search technique which yields locations similar to those determined using travel-time data from local seismic networks with better azimuthal coverage.     

Calibration of Regional S/P Amplitude-ratio Discriminants

v--vS/P amplitude ratios have proven to be a valuable discriminant in support of monitoring a Comprehensive Nuclear Test Ban Treaty. Regional S and P phases attenuate at different rates and the attenuation can vary geographically. Therefore, calibration is needed to apply the S/P discriminant in new regions. Calibration includes application of frequency-dependent source and distance corrections for regional Pn, Pg, Sn, and Lg phases.¶Jenkins et al. (1998) developed Pn, Pg, Sn, and Lg amplitude models for nine geographic regions and two global composite models, stable and tectonic. They determined frequency-dependent source and attenuation corrections from a large data set obtained from the Prototype International Data Center (PIDC). We use their corrections to evaluate calibrated S/P discriminants.¶Our discrimination data set includes >1000 amplitude ratios from earthquakes, industrial explosions, chemical explosions, and nuclear explosions from Lop Nor, India and Pakistan. We find that the calibrated S/P ratio is largest for earthquakes and smallest for the nuclear explosions, as expected. However, the discriminant is not universally valid. In particular, the S/P ratio for the Pakistan nuclear explosion fell within the normal range for the earthquakes. This event was recorded by only a few stations at far-regional distances and appears to have an anomalously high Sn amplitude. The industrial explosions overlap with the earthquake population, however the buried chemical explosions generally register lower S/P ratio than earthquakes.     

利用多道信号测定短周期面波频散

使用适配滤波频时分析技术处理面波信号,获取滤波后的时域包络,并对具有相同群速度的各道信号包络的振幅进行叠加,取叠加所得极大值对应的走时为该频率的群延时,以此来提高群速度的测量精度对P-τ,（相慢度-时间）域的波场进行傅氏变换,然后在P-ω（相慢度-频率）平面叠加多道信号来测定相速度．利用这两种方法处理合成的短周期瑞利波地震图,结果表明：经过振型分离以后,利用基阶振型信号所求出的相速度、群速度曲线更加接近理论频散值．对地震勘探资料进行处理,取得了5-40Hz内的相速度、群速度频散,且曲线较为平滑     

Regional Seismic Wavefield Computation on a 3-D Heterogeneous Earth Model by Means of Coupled Traveling Wave Synthesis

-- I present a new algorithm for calculating seismic wave propagation through a three-dimensional heterogeneous medium using the framework of mode coupling theory originally developed to perform very low frequency (f < ~0.01т.05 Hz) seismic wavefield computation. It is a Greens function approach for multiple scattering within a defined volume and employs a truncated traveling wave basis set using the locked mode approximation. Interactions between incident and scattered wavefields are prescribed by mode coupling theory and account for the coupling among surface waves, body waves, and evanescent waves. The described algorithm is, in principle, applicable to global and regional wave propagation problems, but I focus on higher frequency (typically f S ~0.25 Hz) applications at regional and local distances where the locked mode approximation is best utilized and which involve wavefields strongly shaped by propagation through a highly heterogeneous crust. Synthetic examples are shown for P-SV-wave propagation through a semi-ellipsoidal basin and SH-wave propagation through a fault zone.     

A New Methodology for Decreasing Uncertainties in the Seismic Hazard Assessment Results by Using Sensitivity Analysis. An Application to Sites in Eastern Spain

v--vIn this study a sensitivity analysis has been carried out by means of the seismic hazard results obtained using the non-zoning methodology (Epstein and Lomnitz, 1966) and the extreme value distribution functions proposed by Gumbel (1958), via a logic tree procedure. The aim of the sensitivity analysis is to identify the input parameters that have the largest impact on assessed hazard and its uncertainty. The research findings from the study of these parameters can serve as a useful guide to facilitate further research studies on seismic hazard evaluations because it allows us to identify parameters that have little or no effect on the seismic hazard results as well as parameters that have great effects on them. In this way, using the obtained results, we have proposed objective criteria in assigning probabilities to the different logic tree branches in a more objective way. It should be noted that, although the sensitivity of the logic tree branches depends on the site, it does not always do so in the same way. Finally, re-evaluation of seismic hazard using the proposed methodology applied to eastern Spain leads to a reduction of uncertainty from 52% to 27% of the expected acceleration with 10% probability of exceedence, at the site with the highest value of seismic hazard (Site 1: Torrevieja).     

A Scheme for Initial Beam Deployment for the International Monitoring System Arrays

v--vThe International Monitoring System (IMS) includes a diverse set of seismic arrays with different configurations. These configurations have apertures ranging from less than 1 to more than 25 km and minimum interelement spacings varying from 0.1 to 3.6 km. This paper presents a scheme for initial beam deployment for this variety of seismic arrays. Beamforming is equivalent to a spatiotemporal bandpass filter of which passband is defined by the minimum and maximum wavenumbers, which are functions of the geometry configuration of the array. Deployment of steered-beams for signal detection is based on the wavenumber resolution of the array, slowness and frequency distributions of seismic phases, and coherence properties of seismic signals and noises among sensors. Within the wavenumber passband, all possible slowness values are determined by the resolution for each frequency band, and those that are outside the range of seismological interest are excluded. The appropriate azimuthal distribution for each selected slowness is determined from the azimuthal resolution. Using this approach, detection beams for each array are rationally deployed in the slowness-azimuth and frequency domain.     

基于背景噪声初步研究河北及邻区的剪切波速度结构

本文根据2010年1～12月河北及邻区的83个宽频地震仪12个月连续噪声记录, 分析了河北及邻区瑞利面波的群速度频散曲线并反演了主要分区内的典型路径剪切波速度结构。 首先采用多重滤波方法提取了台站对5～50 s的面波群速度频散曲线, 然后用Herrmann线性反演方法反演了剪切波速度结构。 结果表明, 群速度频散曲线及剪切波速度分布特征与地表地质和构造特征表现出较好的相关性, 清晰地揭示了地壳内部的横向速度变化。 在短周期(8～20 s), 拥有较厚的沉积层的平原地区表现为明显的低速特征, 而隆起地区则表现为较高的群速度分布特征: 随着周期的增加(>20 s)速度的特征有所改变, 30 s之后由于受地壳厚度的影响, 两者的速度差异逐渐缩小, 在中下地壳波速度随深度增加而增大。 同时, 研究发现在研究区域内西南方向的噪声源占主导作用。     

Time-domain Pure-state Polarization Analysis of Surface Waves Traversing California

-- A time-domain pure-state polarization analysis method is used to characterize surface waves traversing California parallel to the plate boundary. The method is applied to data recorded at four broadband stations in California from twenty-six large, shallow earthquakes which occurred since 1988, yielding polarization parameters such as the ellipticity, Euler angles, instantaneous periods, and wave incident azimuths. The earthquakes are located along the circum-Pacific margin and the ray paths cluster into two groups, with great-circle paths connecting stations MHC and PAS or CMB and GSC. The first path (MHC-PAS) is in the vicinity of the San Andreas Fault System (SAFS), and the second (CMB-GSC) traverses the Sierra Nevada Batholith parallel to and east of the SAFS. Both Rayleigh and Love wave data show refractions due to lateral velocity heterogeneities under the path, indicating that accurate phase velocity and attenuation analysis requires array measurements. T he Rayleigh waves are strongly affected by low velocity anomalies beneath Central California, with ray paths bending eastward as waves travel toward the south, while Love waves are less affected, providing observables to constrain the depth extent of anomalies. Strong lateral gradients in the lithospheric structure between the continent and the ocean are the likely cause of the path deflections.     

SEISMIC MODEL STUDIES ON DIFFRACTION OF WAVES BY EDGES OF VARYING RADIUS OF CURVATURE AND DEPTH *

Results of studies carried out with the help of a three-dimensional seismic model on waves diffracted from edges of varying radius of curvature R and depth h with respect to wave length λ are described. The amplitude decay, travel time, and apparent velocity of the wave diffracted from a sub-surface edge of semi-infinite length are found to depend on the parameters R, h, and distance from the edge on the surface provided the ratio of the parameters to λ are less than some limiting values. The nature of the amplitude decay is independent of R when the depth exceeds 2λ, and independent of h when R exceeds 1.5λ. When these are below the limiting values (h= 2λ and R= 1.5λ), the nature of the decay depends appreciably on R and h. The apparent decay in amplitude on the surface due to geometrical spreading by the diffracting edge is less than that of a cylindrical secondary wave source and decreases with increase in depth of the edge. The nature of the travel time curves of the diffracted waves near the edge depend on R/λ when the depth is within about one λ. Apparent velocity of the wave depends largely on R/λ in the zone of diffraction up to a distance of about one λ from the edge on the surface. Beyond this distance the velocity is almost the same irrespective of R/λ and depend only on h/λ. The width of the zone of diffraction caused by an edge of finite length comparable to λ is more and more narrow as the ratio of the distance of the edge on the surface to its depth increases.     

Optimization of Surface Wave Identification and Measurement

—?Accurate and reliable measurement of surface waves is important to Comprehensive Nuclear-Test-Ban Treaty (CTBT) monitoring because the M _s :m _b discriminant and its regional variants can in many cases unambiguously identify events as earthquakes or explosions. Surface wave processing at the International Data Center (IDC) is designed to be completely automated and is performed using the program Maxsurf. Maxsurf searches for surface wave characteristics in the expected surface wave arrival time window for all continuous long-period and broadband data in the IDC processing stream. The Prototype IDC GSETT3 Reviewed Event Bulletin (REB) now contains a very large and growing data set of surface wave measurements. Users of this data set need to be aware of processing changes and calibration errors in the GSETT3 experimental bulletin. The prototype International Monitoring System (IMS) surface wave detection threshold is approximately one magnitude unit lower than the detection threshold of other global networks that use visual identification of surface waves. Surface wave identification and measurement can be improved through development of regionalized earth models, phase-matched filtering and the use of path corrected spectral magnitudes in place of M _s . Regionalized earth models are developed through tomographic inversion of a very large data set of phase and group velocity dispersion measurements. Discrimination capability can be improved through the use of maximum likelihood magnitudes and maximum likelihood upper bounds.     

Surface Wave Group Velocity Tomography Imaging from Ambient Noise for Fujian Province and Its Adjacent Areas

Two-month continuous waveforms of 108 broadband seismic stations in Fujian Province and its adjacent areas are used to compute noise cross-correlation function (NCF). The signal quality of NCF is improved via the application of time-frequency phase weighted stacking. The Rayleigh and Love waves group velocities between 1s-20s are measured on the symmetrical component of the NCF with the multiple filter method. More than 5,000 Rayleigh wave dispersion curves and about 4,000 Love wave dispersion curves are obtained and used to invert for group velocity maps. This data set provides about 50km resolution that is demonstrated with checkerboard tests. Considering the off great circle effect in inhomogeneous medium, the ray path is traced based on the travel time field computed with a finite difference method. The inverted group velocity maps show good correlation with the geological features in the upper and middle crust. The Fuzhou basin and Zhangzhou basin showed low velocity on the short period group velocity maps. On the long period group velocity maps, the low velocity anomaly in the high heat flow region near Zhangzhou and clear velocity contrast across the Zhenghe-Dapu faults, which suggests that the Zhenghe-Dapu fault might be a deep fault.