S11B-0550
Distinct Velocity variations around the Base of the Upper Mantle beneath Northeastern Asia
Both the global and regional P wave tomographic studies have revealed low-velocity anomalies beneath the high velocity descending slabs in the mantle transition zone and uppermost lower mantle under a number of subduction zones. The limited resolution at large depths and possible trade-off between the high and low velocities, however, makes it difficult to substantiate this feature and evaluate accurately the vertical extent of the low-velocity structure. Using broadband waveform modeling on the triplicated phases near the 660-km discontinuity for three deep events, we constrained both the P and SH wave velocity structures around the base of the upper mantle in northeastern Asia. For the two events beneath the south Kurile, the rays traveled through the lowermost transition zone and uppermost lower mantle under the descending Pacific slab. Our preferred models consistently suggest normal-to-lower P and significantly low SH wave velocities above and below the 660-km discontinuity extending to about 760-km depth compared with the global IASP91 model, corroborating previous observations for a slow structure underneath the slab. In contrast, both high P and SH wave velocity anomalies are shown in our preferred model for the region beneath the Japan back arc, likely reflecting the structural feature of a slab stagnant above the 660-km discontinuity. The velocity jumps across the 660-km discontinuity were found to be on average 4.5% and 7% for P and S waves under the south Kurile, and 3% and 6% under the Japan back arc. The relatively small velocity contrasts in the latter are also consistent with the stagnant slab scenario in which the amplitude of the discontinuity is reduced by the high velocity of the slab. In addition, our synthetic modeling indicates that an apparent uplift or depression of the discontinuity (>10 km) could not expain the data reasonably well, suggesting that temperature variations might be small in both regions. While the limited temperature effect may be sufficient to induce the high velocities of the slab beneath the Japan back arc, the low velocity anomalies, especially the apparently different P and SH velocity gradients around the 660-km discontinuity beneath the south Kurile are likely indicative of lateral chemical heterogeneities existing at the upper and lower mantle boundary. Given the complex morphology of the subducting Pacific slab in this region, we speculate that the distinct velocity structure retrieved here may represent mantle materials that are trapped at around the base of the upper mantle between the surrounding subducting Pacific slabs, and are probably different from the slabs and ordinary mantle both thermally and chemically.
S11B-0551
A depleted and destabilized continental lithosphere near the Rio Grande Rift
Seismic waveform diffraction patterns reveal a 120 km thick slab-like anomaly with about 4% increases in both compressional and shear wave speed, extending down to nearly 600 km between the Rio Grande Rift and the western Great Plains in the southwestern United States. Monte Carlo simulations suggest that the most probable interpretation is a temperature anomaly of about -400 K with additions of olivine (11%), orthopyroxene (12%) and losses of clinopyroxene (-14%), garnet (-9%) relative to the adjoining mantle asthenosphere beneath the Rio Grande Rift. This sesimic anomaly, probably continental lithosphere, is cold and is as depleted as the Archean continental lithosphere. It was likely formed in the Proterozoic age but destabilized during the Cenozoic rifting of the Rio Grande Rift
S11B-0552
Radial seismic anisotropy as a constraint for upper mantle rheology
Seismic shear waves that are polarized horizontally (SH) generally travel faster in the upper mantle than those that are polarized vertically (SV), and deformation of rocks under dislocation creep has been invoked to explain such radial anisotropy. Convective flow of the upper mantle may thus be constrained by modeling the textures that progressively form by lattice-preferred orientation (LPO) of intrinsically anisotropic grains. While azimuthal anisotropy has been studied in detail, the radial kind has previously only been considered in semi-quantitative models. Here, we show that radial anisotropy averages and anomalies from improved seismological maps can be explained to a large extent by novel mantle-flow computations. Our models incorporate LPO formation and a laboratory-derived olivine creep law and were recently shown to produce LPO heterogeneity as observed in mantle xenoliths. In accordance with seismology, we predict the strongest an\-iso\-tropy under oceanic plates. Continental anisotropy at shallow depths is under-predicted, which we interpret as the signature of past tectonic episodes. Previously identified anomalous vSV regions beneath the East Pacific Rise and relatively fast vSH regions within the Pacific basin at ~150~km depth can be linked to mantle upwellings and shearing in the asthenosphere, respectively. Radial anisotropy amplitudes potentially provide a strong constraint for mantle rheology and lateral viscosity variations, which are difficult to infer with other methods.
S11B-0553
Small Scale Heterogeneity in the Uppermost Mantle and the Dynamic Topography on the West Coast of Africa
Tomographic models provide a snapshot of the Earth's velocity structure, and, when combined with other geological and geophysical information we can begin to make further inferences on the dynamics of the system. In Africa, much of the focus has been on the large slow velocity anomaly (African superplume) in the lower mantle and its relationship with dynamic topography. A new surface wave tomographic model of the uppermost mantle beneath Africa is presented here, and discussed in relation to the topography. To improve the reliability of the tomography we use multiple path-specific starting models for the waveform inversion, rather than a starting model based on a global reference model. From the tomographic inversion, well recognised differences in velocity structure are observed between the cratonic regions and active areas such as Afar. There are also a number of smaller scale features within the model, and we investigate the relationship between the pattern of negative wavespeed anomalies and the uplift and possible mantle dynamics along the western coast of Africa. The results from the tomographic model are compared with estimates of uplift derived from seismic reflection profiles along the shelf of the West African Salt Basin, and also with long wavelength gravity anomalies. At the shallowest depths in the model, the areas of slowest seismic velocities show a good correlation with positive gravity anomalies, and with the areas of largest recent uplift. From the combination of results we can therefore infer that the slow wavespeeds are related to small-scale convection in the upper mantle which has given rise to localised uplift in the past 1My.
S11B-0554
A Low-velocity Finger from Iceland beneath Southern Scandinavia - the Key to Understanding Neogene Uplift?
A model of upper mantle S-wave velocity beneath northwestern Europe is presented, based on a tomography of regional surface wave observations. Data from international and, more importantly, regional data archives (including temporary deployments) were used to measure group velocities for both Love and Rayleigh surface waves. The procedure for data selection, group velocity measurements and inversion for group velocity 2-D maps follows closely the one described by Levshin et al. (GJI, 170, 441-459, 2007). Our new set of group velocity maps differs significantly from global reference maps, enhancing many details and amplitudes of group velocity variations in the study region. We then apply a linear inversion scheme to invert for 1-D shear wave velocity profiles which are assembled to a 3-D model. By choosing conservative regularization parameters in the 2-D inversion we ensure the smoothness of the group velocity maps and the resulting 3-D shear wave speed model. To account for the different tectonic regimes in the study region, we compare inversions with 3 different reference models (pure 1-D, 3-D crust / 1-D mantle and pure 3-D) to investigate the sensitivity of the 1-D inversions to inaccuracies in crustal models. We find that all three models are consistent at depths below 90 km and the resulting models deviate only slightly from each other, mostly in amplitudes. We image an intriguing low-velocity anomaly extending from the Iceland plume domain across the north Atlantic beneath southern Scandinavia between 70-150 km depth. Beneath southern Norway, the negative perturbations reach a maximum of up to 13 % w.r.t. ak135 and a shallowing of the anomaly is indicated. This observation could explain the sustained uplift of southern Scandinavia in Neogene times, but the mechanisms are yet undetermined. Furthermore, our upper mantle model reveals good alignment to ancient plate boundaries and first-order crustal fronts around the triple junction of the Baltica-Avalonia-Laurentia collision in the early Paleozoic.
S11B-0555
Transition Zone Structure Beneath the Tibetan Plateau From P-Wave Triplications
We use upper-mantle triplication waveforms to determine upper mantle velocity structure beneath the Tibetan plateau. The data were recorded from 1500 earthquakes in the distance range between 12 and 30 degrees by more than 300 temporary and permanent seismic stations around Tibet since 1990. To eliminate source signatures in the waveforms and isolate structural responses, we developed a new method to estimate the source time function (STF) using records at teleseismic distances. A time-domain iterative deconvolution algorithm was used to remove the estimated STF from the data. The source-free triplicated P waveforms were then stacked and modeled using generalized ray theory. Preliminary results show that the 660-km discontinuity beneath the plateau is deeper than the global average, and the velocity contrast across the discontinuity is smaller in central Tibet.
S11B-0556
Reconstruction of 3D Slab Geometry from Seismicity Using Neighborhood Algebraic Surface Patch Generation and Moving Least Squares Blending
The shape of slabs in the mantle combined with mechanical analysis (analytic or numerical modeling) can provide information on the state of stress in the slab (e.g., compressional versus tensional), the history of subduction (e.g., rate and direction), interaction of the slab with mantle structure (e.g., layers with different material properties) or with large-scale mantle deformation driven by heating from within. Because the primary source of information about slab location is seismicity, the shapes of slabs are often reconstructed from inhomogeneously distributed and noisy observations. Previous attempts to reconstruct three-dimensional slab shapes have relied on approximating the entire slab structure using the analytic solution for a thin elastic or viscous sheet, a surface with Gaussian curvature, or smooth splines. However, these approaches enforce a long wavelength smooth shape on the slab structure that can miss shorter length-scale features, or can introduce short wavelength undulations that are unrelated to the actual slab shape. As the slab structure is often used to create input for analytic or numerical models (e.g., thermal and mechanical structure for deformation models), errors in the slab structure can lead to errors in the predicted deformation. We present a new method for generating 3D slab morphology from seismicity that generates neighborhood surface patches from algebraic surfaces of specified degree (quadratic to quintic) and then blends adjacent patches using a moving least- squares algorithm to create a smooth, continuous surface. The new procedure provides options for pre- processing noisy data to remove outliers, to systematically combine different data sets with user defined weighting, and for defining the 3D edge of the slab surface. Generation of the 3D slab surface using algebraic surface patches and blending allows for smooth, continuous assignment of properties (e.g., temperature or viscosity) that are spatially-related to the location of the slab surface and its edges (e.g., by Euclidean distance) to a numerical model grid without aliasing.
S11B-0557
Array Observations of Short Period Pdiff Coda Waves
There have been several reports of short-period, coherent coda waves following Pdiff at distances of approximately 95° -120°. Early interpretations favored topographical scattering at the core-mantle boundary or strong volumetric scattering in the lowermost mantle (D"). However, a recent study suggested these observations can be explained by relatively mild, uniform scattering throughout the lower mantle. In this study, we present a new data set that can be used to test the competing hypotheses for the generation of short-period Pdiff coda waves. We downloaded data for 824 earthquakes with magnitudes of 5.7 mb - 7.9 mb, depths shallower than 100~km, and at distances of 50° - 120° from the Yellowknife array (YKA) in Western Canada. YKA is a medium-aperture (20~km) array, with 19 short-period, vertical-component seismometers arranged in a cross. By using an array we can increase signal-to-noise ratio and infer the directions from which the coda waves are arriving. We included earthquakes at distances smaller than 90° so that we can test the vertical change in heterogeneity strength as waves approach D". The earthquakes are clustered in three regions: Asia, South America, and Tonga-Fiji. For each event, we used a sliding window slowness analysis technique to estimate the beam power as a function of time. Then we normalized each beam envelope by PP amplitude and corrected the amplitudes for radiation pattern effects, eliminating those data in which P, Pdiff, or PP was close to a nodal plane. We then stacked the data in 2° bins, and fit a log-linear relationship between energy and time to determine the coda decay rate (CDR). The CDRs show a clear distance dependence. Asia and South America have the same pattern above D": high CDR from 50° to 70°, low CDR in the range 72°-80°, followed by a CDR increase. At distances after 90° the coda decay rates in the three clusters show significant variations, with events from Asia having the highest CDR. We are currently experimenting with various methods for simulating Pdiff coda waves including a frequency-wavenumber integration technique for 1D stochastic models and a single-scattering ray theory based technique, modified to account for diffraction induced amplitude decay.
S11B-0558
Constraints on Mantle Heterogeneities Inferred From Forward Modeling of Seismic Wave Scattering
One of the open questions in geophysics is how small-scale heterogeneities in wave speed are distributed within the Earth's mantle. These heterogeneities can be caused by temperature and/or compositional variations, and are assumed to have a profound influence on mantle dynamics. Conversely, insight in the characteristics of the heterogeneities would yield valuable clues on aspects of mantle dynamics such as mixing. The aim of this study is to utilize scattering of seismic waves by heterogeneities to constrain their characteristics, such as size, directivity, strength and location. This is accomplished by adding heterogeneities to a smooth global velocity model, which is then used as input for the global spectral-element software SPECFEM3D_GLOBE. This code produces synthetic seismograms of the full wave field, where phases such as PKP arrive in a scattered manner because of the presence of the heterogeneities. In this study, different distributions of heterogeneities are created, with varying dimensions, strength and depth range. Comparison of the synthetic seismograms of these models with actual data includes analysis of precursors to PKP waves. This procedure allows us to determine if and how well we can constrain the characteristics of small-scale heterogeneities in the Earth's mantle.
S11B-0559
Detection of Diffraction Patterns at the edge of the African Superplume
A metastable thermal-chemical convection model has been developed to explain the African low velocity structure with its dome-like morphology. The model that best fits the seismic waveform data is referred to as the High Bulk Modulus Structure (HBMS), which has sharp walls. A recently developed WKBJ-type code (DWKM) can be used to generate approximate 3D synthetics for such structures. This method approximates 3D effects by adding out-of- plane contributions from virtual receivers at neighboring azimuths with two related to the inner Fresnel zone and two longer-period contributors sampling the outer Fresnel zone. The four responses are weighted by diffraction operators that are defined by the source duration and travel time from the edges of the structure. Here we use the HBMS model to generate 3D synthetics displaying waveform complexity patterns predicted for possible geometries. We show that these synthetics can be simulated from PREM synthetics where the shifts (δ ti) between the four operators are determined by a grid-search. Contours of these (δ ti) values obtained, from a grid of stations, produces a diffraction pattern footprint that suggests that array data can be processed directly to reveal 3D structure. We demonstrate the usefulness of this technique by processing both P and S wave data from the Kaapvaal array in Southern Africa. The difference between the patterns produced by P and S wave data for the same events (same ray paths) further validates the nature of these chemically sharp boundaries.
S11B-0560
An Analysis of Small-scale Heterogeneity in the Mantle with PKP Precursors Recorded at IMS Arrays using a seismic phonon method
It has been known that PKP precursors are direct evidence for the existence of small-scale, non-radially symmetric structure in the Earth's deep mantle since their re-interpretation in the 1970s. However, the scattering strength and depth distribution of the small-scale heterogeneities are still not in good agreement. There are two possible reasons for the discrepancy. One is the data used in different studies. Due to uneven data sampling and different data processing schemes, different authors may obtain different PKP precursor envelopes. The other reason is differences in forward modeling theory. For the same data set, single-scattering theory may yield different results for heterogeneity strength than a multiple-scattering or diffusion based theory. In this study, we take advantage of the globally distributed international monitoring system (IMS) seismic arrays to assemble a large, geographically diverse data set of PKP precursor envelopes. An advantage of using IMS arrays over single stations is that the recordings of all elements at one array can be coherently stacked to suppress noise, and thus get high signal-to-noise (SNR) ratio PKP precursors, even for relatively small earthquakes. We sorted out more than 13,000 events recorded at over a dozen IMS arrays, and selected 2,218 events with high SNR and clear PKIKP arrivals, regardless of precursor strength. The average distance dependence and timing of the precursor envelopes is consistent with previous studies, and there is evidence of coherent large-scale lateral variations in scattering strength. For example, beneath offshore South America the lower mantle has generally weak scattering. However, there are also many small-scale variations in heterogeneity strength that may be artifacts of the source-receiver side scattering ambiguity. We have begun modeling our data set with a multiple-scattering, phonon based approach and initial results indicate that (1) confining the small-scale heterogeneities in the bottom 200~km of the mantle does not fit the data well, and (2) the best scattering models tend to have root-mean-square velocity perturbations less than 0.5% throughout the lower mantle (with a scale length of 8~km).
S11B-0561
Towards Developing Velocity-Attenuation Models of Earth's Mantle Incorporating Mineral Physics Observations
Seismic attenuation structure is important for understanding the thermal and compositional state of the mantle. In general, the anelastic structure of the mantle is far less constrained compared to the elastic (velocity) structure, mainly because seismic amplitudes can be affected by both anelastic attenuation (Q) and elastic focusing/defocusing due to lateral heterogeneities. In our work, we investigate the thermal effects on seismic amplitudes based upon a simple two-layer mantle rheology model. The mantle temperature profile is constructed assuming an oceanic lithosphere of 60 Ma lies on top of an adiabatic mantle with a potential temperature of 1300°C -- 1400°C. To constrain the range of thermal activation parameters (activation volume and activation energy) and the Grüneisen parameters for the upper and lower mantel respectively, we calculate synthetic seismograms using mode summation for the resulting attenuation models and measure amplitude perturbations with respect to PREM synthetics for body and surface wave phases. These rheology parameters obtained from our calculation provide important references in constructing direct relationship between perturbations in velocity and attenuation (Q) for purely thermal heterogeneities. The velocity-attenuation (Q) relationship based upon our rheology parameters can be used to distinguish between compositional heterogeneities and thermal heterogenities by using 3-D seismic velocity and Q tomographic models.
S11B-0562
Velocity and Anisotropy Structures and Thermal and Compositional Models Beneath Eastern Asia and North America
The upper mantle velocity and anisotropy structures are important to the understanding of mantle composition and dynamics. We constrain fine seismic SH and SV velocity and anisotropy profiles in the upper mantle beneath eastern Asia and North America, and explore mineralogical and compositional models and crystal alignments based on mineral physics modeling. We use the triplicated phases recorded in the epicentral distance range of 13°- 40° for events occurring in eastern Asia and North America to constrain the SH and SV velocity models in the upper mantle beneath the two regions. Directly comparing the SH and SV data in the triplication distance range provides information of anisotropy changing with depth. For an eastern Asia event, SH waves arrive at the same time as SV waves for those traveling in the transition zone, but earlier than SV waves for those traveling below the 660-km discontinuity. Such observations indicate two possibilities: 1) the upper mantle in this region is isotropic and the uppermost lower mantle is anisotropic with an SH velocity larger than the SV velocity; and 2) these shear wave splitting difference may indicate existence of lateral variation of lithosphere anisotropy. In joint modeling of the mineral physics and seismic data, we explore a variety of compositional and thermal models and crystal alignments by comparing their predicted velocity and anisotropy profiles using the mineral physics modeling with the seismic velocity and anisotropy structures inferred from the seismic data. We will present best-fitting thermal and compositional models and the crystal alignments with their predicted velocity and anisotropy structure best explaining the seismic data.
S11B-0563
Whole Mantle Vp/Vs Tomography
Interpreting seismic tomographic models in terms of geodynamics requires knowledge of the physical causes of velocity anomalies. A common approach is to compare VP and VS (or Vκ and Vμ) models though it is generally difficult to tell how much of the differences between VP and VS anomaly patterns originate from real differences in Earth's structure or from the difference in resolution and accuracy of the models. To circumvent this problem, we derive the three-dimensional distribution of the VP/VS ratio by directly inverting S-P differential travel times. This approach can bypass some of major difficulties associated with the comparison of independent VP and VS models. Use of S-P differential travel times is advantageous not only because they link directly VP to VS anomalies through common ray paths, but they also are little affected by the uncertainty of origin time or source time function. The resulting VP/VS tomographic model can be used to determine if the VP and VS anomalies in the lower mantle can be consistently interpreted as due to temperature anomalies or not. We converted the resultant VP and VP/VS perturbations to temperature perturbations dTP and dTS-P (= dTS-dTP), using mineral physics data. We find that lateral temperature variation can be the primary cause of the velocity anomalies through most of the lower mantle, though chemical heterogeneity is clearly needed in some regions near the base of the mantle.
S11B-0564
Mantle Anisotropy in Northern Alaska: Analysis of ARCTIC Seismic Data Using Shear Wave Splitting
Shear wave splitting observations from teleseismic SKS phases are used to infer the anisotropy under northern Alaska. SKS waveforms were examined from the ARCTIC PASSCAL experiment, which consisted of 15 broadband seismic stations. These stations were deployed between 2004 and 2007 with twelve in a 550 km north/south line between Fairbanks and the Arctic Ocean and three distributed in a sparse east/west line from Fairbanks to the Bering Sea. Results indicate that fast directions are mostly oriented in a slightly north of east direction for all stations, except the furthest west station at Nome which showed a nearly north/south fast direction. The time delays range from approximately 0.5 to 1.5 seconds, however stations in the Brooks Range yielded somewhat more northerly fast directions with higher splitting delay times. These fast directions reflect the anisotropy in the upper mantle and likely indicate that there is a strong regional east-northeast/west-southwest direction of the flow in the upper mantle under central and northern Alaska. This direction is generally parallel to crustal structures, such as mountain ranges and major faults. This orientation is also similar to the fast directions observed in the Alaska Range from the BEAAR PASSCAL experiment, but orthogonal to fast directions south of the Alaska Range.
S11B-0565
Testing Models of North American Seismic Velocity Structure With Traveltimes From Well-Validated Sources
Surface wave tomography has provided 3D models of the seismic structure of the continents, down to depths of several hundred kilometers (e.g. Van der Lee and Nolet, 1997; Nettles and Dziewonski, 2007). On the one hand, these models are limited by their low horizontal resolution (hundreds of kilometers) and by providing estimates of shear velocity but not compressional velocity. On the other hand, they have the virtue of having a uniform spatial resolution and are based on data (surface wave dispersion) that are only very weakly dependent on the location and origin time of the underlying earthquake sources. This near independence is relevant, for instance, in developing velocity models and/or traveltime models for earthquake location, because the problems associated with circularity-of-results (e.g. the same earthquakes both determining the model and being located by it) are avoided. We discuss the conceptual and practical problems associated with using surface wave results to calculate continental-scale traveltimes and to locate earthquakes. We build such a model of North America and test it against continental-scale traveltimes from five PNE's and the Early Rise series of chemical explosions. Traveltime residuals are reduced by 50 percent compared to the predictions of a radially-stratified model, but some regionally coherent traveltime residuals remain, indicating that the 3D model derived from surface waves is capturing some, but not all, of the actual variability beneath the continents.
S11B-0566
Detailed Seismic Imaging of the Australian Lithosphere Using a Dense Rolling Array of Seismometers
Over the last decade, a rolling array of seismometers has been sequentially deployed throughout southeast Australia to record passive seismic activity. To date, nearly 350 separate sites have been occupied with nominal station spacings of 40-50 km on the mainland and 15-20 km in Tasmania. Deployment periods for each of the eight arrays installed so far have varied between 4-10 months, and the number of simultaneously recording instruments has ranged between 20-80. The majority of seismic sensors used have been vertical component short periods, although a number of arrays included a mix of short period and three component broadband instruments. Since October 2006, all arrays contain only three component stations due to an upgrade of the short period seismometer pool. The next deployment of 50 instruments, scheduled for late November 2007, will bring the cumulative array size to nearly 400 seismometers. The intra-plate location of the Australian land mass does not readily permit detailed seismic imaging with local earthquakes. However, the frequency and distribution of large earthquakes associated with the surrounding plate boundary regions makes teleseismic tomography an ideal tool for mapping the 3-D wavespeed structure of the crust and upper mantle beneath the seismic array cluster. So far, four separate teleseismic tomography studies have been published using data from different locations. These results have yielded valuable insight into the architecture of the Australian plate. For example, the possible presence of sizable fragments of Proterozoic continental lithosphere beneath the Paleozoic Lachlan Orogen in eastern Australia; and evidence of a diffuse mantle source for the late Tertiary and Quaternary volcanism in Victoria. The shear volume and diversity of passive seismic data recorded in southeast Australia by the rolling array of seismometers makes it a valuable resource for other classes of study, including ambient noise tomography, shallow receiver functions (using the three-component data), and array seismology (e.g. illumination of the core- mantle boundary). One current line of research aims to combine teleseismic data from all arrays in a single inversion for a unified image of the 3-D structure beneath southeast Australia.