Crustal Reverberations Research Articles

Distinguishing Unique Earthquakes with Overlapping Signals in Oklahoma

AbstractDuring routine operations monitoring Oklahoma earthquakes, we found that certain earthquakes occurred closely both in space and time and had overlapping phases at the recording stations. Through further scrutiny and analysis, we determined that rather than being distinctly different earthquakes, some of the earthquakes exhibited multiphase arrivals and longer than expected coda due to unique ray paths that encounter impedance contrasts such as at the sedimentary rock-basement. Of course, some of these events truly were distinct events, which we term overlapping earthquakes, for which perceived coda duration overlaps and obscures the phase arrivals of the second event due to the source proximity in both time and space. We detail our classification scheme to separate the local earthquakes in Oklahoma as single, overlapping earthquakes, or those associated with multiphase arrivals. We forward model seismic wave propagation in a 2D crustal model and develop a methodology that utilizes waveform correlation to distinguish phases from overlapping earthquakes to those from crustal reverberations. Duration analysis shows a more elongated duration, qualitatively similar to the duration produced by overlapping earthquakes, at the sites where multiphase arrivals are observed.

Modelling regional-scale attenuation across Italy and the Tyrrhenian Sea

Modelling regional-scale attenuation of seismic waves at ~1 Hz is challenging, especially when these waves propagate across both continental and oceanic crust. Recent developments in seismic imaging and modelling have provided us with the computational tools necessary to reconstruct these mixed settings using deterministic (coherent) and stochastic (coda) information. Here, we present new tomographic maps of coda-attenuation for both the Italian peninsula and the Tyrrhenian Sea. Kernel-based coda attenuation imaging in the diffusive approximation is tested in the oceanic environment, highlighting a non-diffusive behaviour across the Southern Tyrrhenian Sea. Joint deterministic and Radiative Transfer forward modelling of coherent and scattered waves (Radiative3D) is then performed for the portion of this area showing the lowest coda attenuation. The corresponding parametric study shows that coda attenuation is an efficient marker of variations of Moho depths and crustal reverberations, such as those we observe in a transitional area (characterized by continental crust and thicker Moho) between two oceanic basins (Vavilov and Marsili). Diffusive tomographic imaging informed by Radiative3D modelling reconstructs high-attenuation patterns along the Italian peninsula in agreement with the thick sediments cover across the Po Plain and the Adriatic coast and the magmatic systems in the Central Campanian province. A combined approach of coda-attenuation imaging and modelling can constrain Moho depth and spatial variations of seismic attenuation at regional scale and in mixed continental-oceanic settings.

Hales Discontinuity in the Southern Indian Continental Lithosphere: Seismological and Petrological Models

AbstractWe model the depth and Vs structure of the Hales discontinuity (H‐D) beneath Eastern Dharwar Craton (EDC) and Southern Granulite Terrain (SGT) using P wave receiver function (P‐RF) analysis and joint inversion with Rayleigh wave phase velocity dispersion. We calculate P‐RFs at higher frequency (fmax = 0.46 Hz), compared to previous studies, to show that the H‐D P‐to‐S converted phase (Phs) is distinct from crustal reverberations. Phs at stations Hyderabad (HYB) and Gauribidanur (GBA), in the EDC, arrive at ∼11.5 s and ∼12 s, respectively. From joint inversion, the H‐D is modeled at 107 ± 5 km and 113 ± 4 km depth, with ∼3% and ∼4% Vs increase, beneath HYB and GBA, respectively. For station Kodaikanal (KOD), in SGT, the Phs destructively interferes with the negative midcrustal reverberation at most ray‐parameters, which explains its apparent absence in previous studies. We isolated P‐RFs where Phs is distinct at ∼11 s and model it at 102 ± 3 km depth. Common conversion point stack profiles constructed by depth migrating P‐RFs through the Vs model show an undulatory nature of the H‐D. From data of mantle xenoliths (Wajrakarur kimberlite field), we calculate Vs of mantle peridotite and eclogite, using published bulk rock compositions. At the H‐D depth and temperature derived from Indian shield geotherm, we observed a good match to the Vs structure of the H‐D. Our results support the geodynamic model of the H‐D being an interface of paleosubducted eclogitic oceanic crust embedded within the upper mantle peridotite. Global observations of mantle reflectors within the continental lithosphere, at depths similar to H‐D, have been related to relict subduction and independently support our model.

25-Second Determination of 2019 Mw 7.1 Ridgecrest Earthquake Coseismic Deformation

ABSTRACT We have developed a global earthquake monitoring system based on low-latency measurements from more than 1000 existing Global Navigational Satellite System (GNSS) receivers, of which nine captured the 2019 Mw 6.4 Ridgecrest, California, foreshock and Mw 7.1 mainshock earthquakes. For the foreshock, coseismic offsets of up to 10 cm are resolvable on one station closest to the fault, but did not trigger automatic offset detection. For the mainshock, GNSS monitoring determined its coseismic deformation of up to 70 cm on nine nearby stations within 25 s of event nucleation. These 25 s comprise the fault rupture duration itself (roughly 10 s of peak moment release), another 10 s for seismic waves and displacement to propagate to nearby GNSS stations, and a few additional seconds for surface waves and other crustal reverberations to dissipate sufficiently such that coseismic offset estimation filters could reconverge. Latency between data acquisition in the Mojave Desert and positioning in Washington State averaged 1.4 s, a small fraction of the fault rupture time itself. GNSS position waveforms for the two closest stations that show the largest dynamic and static displacements agree well with postprocessed time series. Mainshock coseismic ground deformation estimated within 25 s of origin time also agrees well with, but is ∼10% smaller than, deformation estimated using 48 hr observation windows, which may reflect rapid postseismic fault creep or the cumulative effect of nearly 1000 aftershocks in the 48 hr following the mainshock. GNSS position waveform shapes, which comprise a superposition of dynamic and static displacements, are well modeled by frequency–wavenumber synthetics for the Hadley–Kanamori 1D crustal structure model and the U.S. Geological Survey finite-rupture distribution and timing. These results show that GNSS seismic monitoring performed as designed and offers a new means of rapidly characterizing large earthquakes globally.

Crustal structure of north Peru from analysis of teleseismic receiver functions

In this study, we present results from teleseismic receiver functions, in order to investigate the crustal thickness and Vp/Vs ratio beneath northern Peru. A total number of 981 receiver functions were analyzed, from data recorded by 28 broadband seismic stations from the Peruvian permanent seismic network, the regional temporary SisNort network and one CTBTO station. The Moho depth and average crustal Vp/Vs ratio were determined at each station using the H-k stacking technique to identify the arrival times of primary P to S conversion and crustal reverberations (PpPms, PpSs + PsPms). The results show that the Moho depth correlates well with the surface topography and varies significantly from west to east, showing a shallow depth of around 25 km near the coast, a maximum depth of 55–60 km beneath the Andean Cordillera, and a depth of 35–40 km further to the east in the Amazonian Basin. The bulk crustal Vp/Vs ratio ranges between 1.60 and 1.88 with the mean of 1.75. Higher values between 1.75 and 1.88 are found beneath the Eastern and Western Cordilleras, consistent with a mafic composition in the lower crust. In contrast values vary from 1.60 to 1.75 in the extreme flanks of the Eastern and Western Cordillera indicating a felsic composition. We find a positive relationship between crustal thickness, Vp/Vs ratio, the Bouguer anomaly, and topography. These results are consistent with previous studies in other parts of Peru (central and southern regions) and provide the first crustal thickness estimates for the high cordillera in northern Peru.

Generalization of the H‐κ stacking method to anisotropic media

AbstractWe investigate the effect of anisotropy on estimates of crustal thickness H and average bulk VP/VS‐ratio κ. Here we extend the stacking approach of Zhu and Kanamori (2000) to include all 20 P‐to‐S converted waves and their crustal reverberations that are generated in the anisotropic case—instead of only five phases in the isotropic case. The ray‐based algorithm of Frederiksen and Bostock (2000) is used to calculate the amplitude and arrival time of each phase. Synthetic tests are performed to investigate the feasibility and robustness of the stacking approach. For simplicity, we assume hexagonal symmetry and a horizontal symmetry axis, but more general anisotropy may be considered. The tests reveal that the estimates of H and κ can be significantly affected by the presence of crustal anisotropy. We verify the feasibility of the stacking approach for real data by applying the method to examples from three different tectonic regions. The results show that the anisotropic stacking scheme presented here can provide a much better constraint on the estimation of H and κ than is achieved using isotropic stacking. Anisotropic stacking can also help resolve the ambiguity in determination of H and κ that arises when several maxima from stacking amplitudes of receiver functions occur in the case of complex crustal structure.

Receiver functions and crustal structure of the northwestern Andean region, Colombia

AbstractWe used the receiver function technique to deduce crustal thickness beneath the northwestern Andean system, using data from the permanent seismic network of Colombia, combined with some of the IRIS and CTBTO stations in Colombia and Ecuador. The estimation of crustal thickness was made using the primaryPtosconversion and crustal reverberations. The bulk crustalVP/VSratio was constrained using a crustal thickness versusVP/VSstacking method, in addition to estimations using a time to depth conversion technique based on results of a modified Wadati diagram analysis. We observed a wide range of crustal thicknesses, including values around 17 km beneath the Malpelo Island on the Pacific Ocean, 20 to 30 km at the coastal Pacific and Caribbean plains of Colombia, 25 to 40 km beneath the eastern plains and foothills, 35 km beneath the Western Cordillera, 45 km at the Magdalena River intermountain valley, 52 to 58 km under the northern Central Cordillera, and reaching almost 60 km beneath some of the volcanoes of the Southern Cordilleran system of Colombia; crustal thickness can be slightly greater than 60 km beneath the plateau of the Eastern Cordillera. The values ofVP/VSare particularly high for some of the stations on the volcanic centers, reaching values above 1.79, probably related to the addition of mafic materials to the lower crust, and in the plateau of the Eastern Cordillera near Bogota, where we speculate about the possibility of crustal seismic anisotropy associated with shear zones.

A ubiquitous low‐velocity layer at the base of the mantle transition zone

AbstractGlobal stacks of receiver functions clearly exhibit the upper mantle stratification. Besides the most prominent seismic discontinuities, such as the Moho and the 410 and 660 km discontinuities, a negative discontinuity is detected at a depth of ~600 km, indicating a low‐velocity layer at the base of the mantle transition zone. The slant‐slack technique helps to identify the primary conversions from the multiple reverberations. Presence of the negative 600 km discontinuity underneath both continent and ocean island stations, where the crustal thickness significantly differs, also precludes the possible cause of crustal reverberations. We conclude that the negative 600 km discontinuity could be a global feature, possibly resulted from accumulation of ancient subducted oceanic crust. The X‐discontinuity at ~300 km depth is also observed in our global stacks, which can be explained by the coesite‐stishovite phase transformation.

Estimating a Continuous Moho Surface for the California Unified Velocity Model

Online material : Simple Matlab script to plot Moho profiles; continuous Moho surface; data points used to estimate the Moho surface. The Mohorovicic discontinuity (Moho) is a globally identifiable boundary between the crust and the uppermost mantle (Dziewonski and Anderson, 1981). It is detected using seismic techniques, such as reflected energy from local earthquakes (e.g., Richards-Dinger and Shearer, 1997) or active-source experiments (e.g., Christeson et al. , 2010) or from crustal reverberations of waves from distant earthquakes (e.g., Burdick and Langston, 1977; Zhu and Kanamori, 2000). Locally, the Moho surface may exhibit complexity in terms of both the magnitude and length scale of its variations in depth; furthermore, the impedance contrast across the Moho may also vary in terms of both the magnitude and length scales (Fig. 1). Body waves and surface waves from both teleseismic and local earthquakes, as well as seismic waves from active-source experiments, may be sensitive to the variations in the Moho. These seismic waves can be used within tomographic inversions to improve the characterization of Earth’s structure in the transition from crust to upper mantle. The Moho surface constitutes an integral part of any Earth model, from regional to global scales. The objective of this paper is to estimate a detailed Moho surface, from all available data, for onshore and offshore California; currently, no such map exists. Our motivation is to obtain a continuous, smooth surface that can be implemented within 3D structural models that are used for simulations of seismic-wave propagation (e.g., Komatitsch et al. , 2004; Tape, Liu, et al. , 2009), such as the California Community Velocity Model (CVM-H) of the Southern California Earthquake Center (SCEC; Suss and Shaw, 2003; Plesch et al. , 2011). We first document the available datasets for estimating the Moho surface in California …

Frequency-dependent effects on global S-wave traveltimes: wavefront-healing, scattering and attenuation

We present a globally distributed data set of ∼400 000 frequency-dependent SH-wave traveltimes. An automated technique is used to measure teleseismic S, ScS and SS traveltimes at several periods ranging from 10 to 51 s. The targeted seismic phases are first extracted from the observed and synthetic seismograms using an automated time window algorithm. Traveltimes are then measured at several periods, by cross-correlation between the selected observed and synthetic filtered waveforms. Frequency-dependent effects due to crustal reverberations beneath each receiver are handled by incorporating crustal phases into WKBJ synthetic waveforms. After correction for physical dispersion due to intrinsic anelastic processes, we observe a residual traveltime dispersion on the order of 1–2 s in the period range of analysis. This dispersion occurs differently for S, ScS and SS, which is presumably related to their differing paths through the Earth. We find that: (1) Wavefront-healing phenomenon is observed for S and to a lesser extent SS waves having passed through very low velocity anomalies. (2) A preferred sampling of high velocity scatterers located at the CMB may explain our observation that ScS waves travel faster at low-frequency than at high-frequency. (3) A frequency-dependent attenuation q(ω) ∝q0×ω−α, with α∼ 0.2, is compatible with the globally averaged dispersion observed for S waves.

Moho depth variation in the Baltic Shield from analysis of converted waves

Teleseismic data recorded at 52 permanent broad-band seismic stations in the Swedish National Seismic Network have been used to map Moho depth variation in the Baltic Shield. We use the receiver function technique to isolate signals associated with P to S conversion at the base of the crust. The delay times of this phase, and of crustal reverberations, are used to constrain Moho depth and crustal average V P /V S . For a majority of the stations in the study, well constrained results could be obtained and these are discussed in comparison to previous estimates of Moho depth in the area. Our results are found to be in very good agreement with previous estimates where these exist. For large parts of the area, estimates of Moho depth have not been available until now and the results of the present study serve as a valuable complement to constrain crustal thickness variation in the Baltic shield.

Receiver function study of the crustal structure of the southeastern Caribbean plate boundary and Venezuela

We have investigated crustal thickness and composition across the southeastern Caribbean plate boundary with the receiver function technique. We used teleseismic data recorded by a temporary broadband array deployed under the BOLIVAR project and the permanent national seismic network of Venezuela. We used the primary P‐to‐S conversion and crustal reverberations to estimate crustal thickness and average crustal VP/VS ratio over the region. We observe large variations in crustal thickness and Poisson's ratio. Estimated Moho depth ranges from ∼16 km beneath the southeastern Caribbean Sea to ∼52 km beneath northeastern Venezuela and the Venezuelan Andes. There is a good correlation between crustal structure and tectonic terranes. Data from the Precambrian Guayana Shield suggest that the underlying crustal structure is relatively uniform with a moderate thickness (∼37 km) and an intermediate composition. A thick crust is found below the foreland basins. The two mountain systems in northern Venezuela, the Serrania del Interior and the Serrania del Falcon, have a thin crust with arc composition and are likely dynamically supported by elastic rebound or underthrusting of the oceanic plateau that characterizes the southern Caribbean. On the other hand, the Venezuelan Andes and Perija Range on the western side of the country are probably isostatically balanced by thick crustal roots.

Analysis of waves converted from S to P in the upper mantle beneath the Baltic Shield

Teleseismic data recorded by broadband stations of the Swedish National Seismic Network have been used in a study of waves converted from S to P in the uppermost mantle. S to P converted waves are recorded as precursors to the direct S arrival. This enables us to study seismic discontinuities at depths of 100–300 km where the traditional receiver function technique based on P to S conversion fails due to disturbances from crustal reverberations. Clear signals of S to P conversion are seen for several horizons in the 50–200 km depth range. This is interpreted as a layered lithosphere with alternating high and low velocity bodies. A sharp contrast is imaged at depths around 160 km and continuous for almost 1000 km in the southern part of the profile. This structure coincides with the velocity contrast interpreted by several authors as the lower limit of a prominent low velocity zone in the 95 to 160 km depth range. A feature which we identify as the lithosphere–asthenosphere boundary is clearly imaged at depths around 200 km.

Dispersion Measurements of P Waves and their Implications for Mantle Q p

We analyze the anelasticity of the earth using group delays of P-body waves of deep (>200 km) events in the period range 4–32 s for epicentral distances of 5–85 degrees. We show that Time Frequency Analysis (TFA), which is usually applied to very dispersive surface waves, can be applied to the much less dispersive P-body waves to measure frequency-dependent group delays with respect to arrival times predicted from the CMT centroid location and PREM reference model. We find that the measured dispersion is due to: (1) anelasticity (described by the P-wave quality factor Q p ), (2) ambient noise, which results in randomly distributed noise in the dispersion measurements, (3) interference with other phases (triplications, crustal reverberations, conversions at deep mantle boundaries), for which the total dispersion depends on the amplitude and time separation between the different phases, and (4) the source time function, which is dispersive when the wavelet is asymmetrical or contains subevents. These mechanisms yield dispersion ranging in the order of one to 10 seconds with anelasticity responsible for the more modest dispersion. We select 150 seismograms which all have small coda amplitudes extending to ten percent of the main arrival, minimizing the effect of interference. The main P waves have short durations, minimizing effects of the source. We construct a two-layer model of Q p with an interface at 660 km depth and take Q p constant with period. Our data set is too small to solve for a possible frequency dependence of Q p . The upper mantle Q 1 is 476 [299–1176] and the lower mantle Q 2 is 794 [633–1064] (the bracketed numbers indicate the 68 percent confidence range of Q p −1). These values are in-between the AK135 model (Kennett et al., 1995) and the PREM model (Dziewonski and Anderson, 1981) for the lower mantle and confirm results of Warren and Shearer (2000) that the upper mantle is less attenuating than PREM and AK135.

Crustal structure of the Kaapvaal craton and its significance for early crustal evolution

High-quality seismic data obtained from a dense broadband array near Kimberley, South Africa, exhibit crustal reverberations of remarkable clarity that provide well-resolved constraints on the structure of the lowermost crust and Moho. Receiver function analysis of Moho conversions and crustal multiples beneath the Kimberley array shows that the crust is 35 km thick with an average Poisson's ratio of 0.25. The density contrast across the Moho is ∼15%, indicating a crustal density about 2.86 gm/cc just above the Moho, appropriate for felsic to intermediate rock compositions. Analysis of waveform broadening of the crustal reverberation phases suggests that the Moho transition can be no more than 0.5 km thick and the total variation in crustal thickness over the 2400 km 2 footprint of the array no more than 1 km. Waveform and travel time analysis of a large earthquake triggered by deep gold mining operations (the Welkom mine event) some 200 km away from the array yield an average crustal thickness of 35 km along the propagation path between the Kimberley array and the event. P- and S-wave velocities for the lowermost crust are modeled to be 6.75 and 3.90 km/s, respectively, with uppermost mantle velocities of 8.2 and 4.79 km/s, respectively. Seismograms from the Welkom event exhibit theoretically predicted but rarely observed crustal reverberation phases that involve reflection or conversion at the Moho. Correlation between observed and synthetic waveforms and phase amplitudes of the Moho reverberations suggests that the crust along the propagation path between source and receiver is highly uniform in both thickness and average seismic velocity and that the Moho transition zone is everywhere less than about 2 km thick. While the extremely flat Moho, sharp transition zone and low crustal densities beneath the region of study may date from the time of crustal formation, a more geologically plausible interpretation involves extensive crustal melting and ductile flow during the major craton-wide Ventersdorp tectonomagmatic event near the end of Archean time.

Linearized inverse scattering of teleseismic waves for anisotropic crust and mantle structure: 2. Numerical examples and application to data from Canadian stations

The objective of this study is to characterize elastic properties, including anisotropy, at the base of the crust and uppermost mantle using the Moho PMs phase recorded in the teleseismic P coda at Canadian stations. We use linearized inverse scattering and singular value decomposition to identify those parameter combinations to which idealized teleseismic data sets are most sensitive. Five to seven independent parameter combinations are likely to be resolvable, one of which is sensitive to isotropy, whereas the remainder quantify different harmonic orders (1θ, 2θ, 3θ) of back azimuthal response. Aside from the isotropic component which is resolved only by P‐SV interactions, P‐SV and P‐SH conversions exhibit a redundant sensitivity to model parameters for uniform back‐azimuthal sampling. We use parameter combinations from the idealized, uniform back‐azimuthal data distribution to compare Moho anisotropy at 25 broadband stations on the Canadian landmass. The isotropic component dominates at all stations and corresponds to shear velocity contrasts ranging between 10 and 35%. Perturbations to anisotropic material property parameters are more modest, generally between 3 and 7% when consistent between SV and SH responses and in many cases suggest an anisotropic lower crust. Inconsistent responses may manifest contamination by lateral heterogeneity, upper crustal reverberations, or pervasive crustal anisotropy leading to shear wave splitting.

Attenuation structure beneath the volcanic front in northeastern Japan from broad-band seismograms

Anelastic structure in the asthenosphere beneath the volcanic front in northeastern Japan arc is estimated by using the spectral amplitude ratio data of P and S waves from about 100 events which occurred in the subducting Pacific slab below Japan. These earthquakes occurred within a 90 km radius centered about the station Sawauchi (SWU), with focal depths ranging from 60 to 200 km. Waveforms were recorded by the Carnegie broad-band three-component seismograph and were corrected for instrument responses, crustal reverberations, corner frequencies, and superimposed noise. Ray paths and travel times of P and S waves are calculated using a three-dimensional velocity model [Zhao, D., Hasegawa, A., Horiuchi, S., 1992. J. Geophys. Res. 97, 19909–19928]. We find a low- Q region ( Q S∼70) extending down to 55 km depth from the lower crust beneath the volcanic front. Using Q-temperature laboratory results [Sato, H., Sacks, I.S., Murase, T., Muncill, G., Fukushima, H., 1989. J. Geophys. Res. 94, 10647–10661], this implies a temperature of about 130°C higher than the eastern forearc region and about 30°C higher than the western backarc region, in good agreement with the tomographic results of Zhao et al. [Zhao, D., Hasegawa, A., Horiuchi, S., 1992. J. Geophys. Res. 97, 19909–19928]. This suggests that low velocities in the crust and uppermost mantle beneath SWU may be explained by a subsolidus temperature increase without partial melting.

Mantle stratigraphy and evolution of the Slave province

A data set of 1033 three‐component, P wave seismograms from five broadband stations at the Yellowknife Array is assembled to investigate mantle structure below the southern Slave province in Canada's Northwest Territories. Following wave field decomposition, seismograms are source‐normalized through simultaneous deconvolution to estimate the near‐receiver impulse response as a function of epicentral distance and back azimuth. Images of impulse response reveal a well‐developed mantle stratigraphy, anisotropic in part, extending from the Mohorovicic discontinuity to the transition zone, A layer of depth‐localized anisotropy (±5%), termed H, is situated between ∼70 and 80 km depth with an average shear velocity comparable to that of the ambient mantle and a sharp upper boundary less than 100 m in transition width. The absence of free surface crustal reverberations on the transverse component affords a window into the upper mantle between 100 and 200 km depth. A sequence of at least two layers between 120–150 km depth, collectively termed X, is most clearly evident to the north and is underlain by a second structure L which dips from 170 km in the west to 230 km into the center of the Slave province. The deepest interface above the transition zone W marks a shear velocity inversion near 350 km depth whose signature is restricted to the SV component signalling a dominantly isotropic response. Consideration of these observations in light of data acquired in a recent LITHOPROBE seismic reflection traverse and in petrological studies of kimberlite xenoliths prompts speculation into the role of subduction in craton stabilization. It is suggested that the proto‐Slave craton was assembled through processes of shallow subduction resulting in a near‐horizontal mantle stratigraphy (i.e., H, X) both compositional and rheological in nature. Interpretation of L as the continuation of dipping reflectors on the seismic reflection profile argues for a final phase of craton assembly involving oblique underplating of subducted lithosphere in the Proterozoic. Subsequent modification of the lithosphere, as manifest by Phanerozoic kimberlite volcanism, may be related to W if an interpretation as the top of a layer containing a dense silicate melt fraction is invoked.

Transition zone velocity gradients and the 520‐km discontinuity

Stacks of long‐period Global Digital Seismograph Network (GDSN) seismograms at 110° to 180° epicentral distance reveal precursors to SS that result from underside reflections off upper mantle seismic discontinuities. The 410‐ and 660‐km discontinuities are obvious in these stacks, but identification and modeling of other transition zone discontinuities are complicated by sidelobes from the 410‐ and 660‐km reflections. These sidelobes result from the limited bandwidth of the GDSN instrument responses and the effect of crustal reverberations on the SS reference phase. The crustal effects can be minimized by restricting the records to oceanic bounce points where the ∼6‐km‐thick crust has little effect on the long‐period waveforms. Over 2000 long‐period, transverse‐component seismograms with oceanic SS bounce points recorded by the GDSN from 1976 to 1991 are manually edited, aligned on SS, and then stacked using a new procedure that weights the records by data quality. The resulting image shows a clear reflection from a 520‐km discontinuity that cannot be explained as a sidelobe artifact, confirming earlier results of Shearer [1990, 1991] and Revenaugh and Jordan [1991]. By stacking along the expected travel time curves for discontinuity phases, the time versus range image of the precursor wave field is reduced to a single trace that measures upper mantle reflectivity versus time. The features in this reflectivity profile are sensitive to the brightness and depth of the transition zone discontinuities and to the steepness of the velocity gradients between the interfaces. Using geometrical ray theory and assuming a constant velocity versus density scaling relationship, I fit this reflectivity profile with velocity models of the upper mantle using both forward modeling and direct inversion. The inverse problem is addressed by performing a deconvolution of the profile with the SS reference phase (after a correction for attenuation), followed by a direct mapping of reflectivity versus time into velocity versus depth. Velocity‐depth profiles resulting from these procedures are roughly in agreement with standard upper mantle velocity models, except that the SS precursor data require a minor discontinuity near 520 km and a steeper gradient just below the 660‐km discontinuity. Estimated discontinuity shear impedance changes are 6.7 ± 1.1% at 420 km, 2.9 ± 0.7% at 519 km, and 9.9 ± 1.5% at 663 km. The impedance change near 520 km is consistent with current mineral physics results for the olivine β to γ phase change and places constraints on the fraction of olivine in the transition zone.

A Review of Receiver Function Techniques for Estimation of One-Dimensional Velocity Structures

Two receiver function techniques both broadly used by seismologists to estimate one-dimensional shear wave velocity structures of the crust and upper mantle beneath seismic stations have been evaluated. One employs a deconvolution filtering, which is directly accomplished in the time domain (Vinnik, 1977; Kosarev et al., 1987). The other is completed through a source-equalization, which is performed by spectra division in the frequency domain (Langston, 1977; Owens et al., 1984). In this study, the performance of these techniques is examined from two sets of synthetic seismograms that are computed from two one-dimensional models by using Thomson and Haskell’s method (Haskell, 1962). The results suggest that both techniques can mostly recover the assumed models very well when the energy of multiples (or crustal reverberations) in the direction of P-wave propagation is minor. If the model is more complicated or the multiple energy becomes stronger, however, the results from Vinnik-Kosarev technique appear to be better than the other in both modeling the converted P-SV phases and inverting the structures. Combining the results of the synthetic tests and theoretic comparisons, it is concluded that the differences resulting from both techniques are primarily caused by the processes through different domains. Besides, both techniques consistently indicate that the inversion results are dependent upon the incident angles. For the nearly vertical incidence, neither technique could resolve the models very well.