T31D-01 08:00h
Global Multi-Resolution Models of Surface Wave Propagation: the Effects of Scattering.
How relevant are finite-frequency (i.e., scattering, banana-doughnut...) effects in seismic tomography? We derive tomographic maps of surface wave phase velocity at periods between 35-150 s, accounting for scattering via the Born approximation; maps are parameterized in terms of a grid of pixels of non-uniform density, higher within chosen regions of interest (including, alternatively, North America or the Mediterranean Basin). We compare our finite-frequency maps against ray-theoretical ones derived from the same data and with the same parameterization and regularization; this comparison should help determining to what extent the improvement in theory (from rays to finite-frequency kernels) leads to an improvement in model quality. A similar comparison had been attempted by other authors, using a degree-40 spherical harmonic parameterization; but a parameterization of higher nominal resolution, like ours, is more adequate to reproduce improvements in the effective resolution of the models. In addition to analyzing seismic data by tomography, we test the accuracy of linearized scattering theory, in the forward problem, by comparison with finite-differences calculations on an infinitely thin membrane. Finite-frequency effects should be particularly important at surface-wave periods (intermediate to long), where the Fresnel zone of waves becomes larger; moreover, the propagation of surface waves at a given frequency is a purely two-dimensional problem (as opposed to body waves), and its numerical solution is therefore relatively cheap; for both reasons, the study of surface wave propagation at a given frequency is a covenient scenario to learn more about the general nature of finite-frequency effects.
T31D-02 INVITED 08:15h
Global Upper Mantle Structure from Finite-Frequency Surface-Wave Tomography
We report global shear-wave velocity structure and radial anisotropy in the upper mantle obtained by finite-frequency surface-wave tomography, based on complete three-dimensional Born sensitivity kernels developed by Zhou et al (2004). Because wavefront healing effects are properly taken into account, finite-frequency surface-wave tomography improves the resolution of small-scale mantle heterogeneities using long-period surface waves. The resulting S-wave velocity models fit the dispersion data better, and show stronger small-scale mantle anomalies compare to traditional ray-theory-based tomographic models. Separate inversions of Love wave (SH-type) and Rayleigh wave (SV-type) dispersion provide insight into the radial anisotropy in the upper mantle. In our model, the globally averaged radial anisotropy is positive ($V_{\rm SH} > V_{\rm SV}$) (horizontal flow) in the top 220 km, and becomes negative ($V_{\rm SV} > V_{\rm SH}$) (vertical flow) below 220 km depth. In cratons, both SH and SV velocities show strong fast anomalies down to 250 km depth, and the fast anomalies gradually diminish below 250 km. Radial anisotropy beneath cratons is positive, which largely agrees with a recent global model by Gung et.~al (2003). The old Pacific plate is characterized by strong positive anisotropy with its maximum centered west of Hawaii; this supports an earlier observation on Pacific radial anisotropy by Ekstrom & Dziewonski (1998). The depth extent of mid-ocean ridges and the primary force that drives plate tectonics has been a long-standing question. In our model, ridge anomalies are characterized by strong negative radial anisotropy (vertical flow). Ridge anomalies at fast-spreading centers are stronger than those at slow-spreading centers at shallow depth, but the amount of velocity reduction rapidly decreases below 250 km. However, at slow-spreading centers such as the north Mid-Atlantic ridge (MAR) and East Africa (Red Sea), ridge anomalies extend down at least to the top of the transition zones. The different depth extent of the ridge anomalies indicates that the primary driving force of slow-spreading seafloors may be different from that of fast-spreading seafloors; active upwelling beneath slow-spreading ridges may play a major role in the opening of the slow-spreading sea floor. The deep origin of the slow anomaly at MAR at about $20^{\circ}$N may be responsible for the initial opening of the Atlantic ocean between Africa and North America plate 180 million years ago.
T31D-03 INVITED 08:35h
Global Model of Seismic Anisotropy and Geodynamics
Seismic anisotropy is not a second order effect though it was often neglected due to the inherent heavy mathematical and computational tools needed to describe and model its effects on seismic waves. It turns out that it is a good marker of large scale deformations and that it is reflecting some inherent organization of the matter, contrarily to isotropy. The uppermost mantle down to 410 km is the depth range where the existence of seismic anisotropy is now widely recognized and well documented. Azimuthal variations have been found for body waves and surface waves in different areas of the world. The application of seismic anisotropy to geodynamics in the upper mantle is straightforward, if we assume that fast-polarization axis of mineralogical assemblages (primarily of $\alpha$-olivine) is in the flow plane parallel to the direction of flow. Seismic anisotropy in the mantle is therefore reflecting the strain field prevailing in past (frozen-in anisotropy) for shallow layers or present convective processes in deeper layers. From the global geodynamics point of view, seismic anisotropy makes it possible to define the root of continents, to investigate the coupling between the lithosphere and the rest of the mantle and more generally to gain insight into mantle convection. Oceans are the areas where Plate tectonics applies almost perfectly. In the Pacific ocean, the map of the azimuthal anisotropy at 100 km shows that it is very large along spreading ridges with a large asymmetry for the East Pacific rise. The direction of anisotropy is in very good agreement with plate motion. The anisotropy is large as well in the middle of the Pacific plate, but it can be observed that there is a line of very small azimuthal anisotropy almost parallel to the EPR and another one between EPR and Tonga- Kermadec subduction zone. These linear areas of small anisotropy were coined Low Anisotropy Channel by Montagner (EPSL, 2002). They are presumably related to cracking within the Pacific plate and/or to secondary convection within and below the rigid lithosphere, predicted by numerical and analog experiments. These new features provide strong constraints on the decoupling between plate and asthenosphere. The existence and location of these LACs might be related to the current active volcanoes and hotspots (possibly plumes) in Central Pacific. LACs, which are dividing the Pacific Plate into smaller units, might indicate a future reorganization of plates with ridge migrations in the Pacific Ocean. Below continental lithosphere, it is also observed significant azimuthal anisotropy which is reflecting asthenospheric flow. Some recent results beneath the eastern Africa will be presented and also show that this flow is highly perturbed by the presence of the Afar hotspot upwelling and might induce upwellings (with babyplumes) and downwellings at large distance (several thousands kilometers) from Afar. In conclusion, the scientific potential of seismic anisotropy is enormous and largely unexploited. It provides a new dimension in the investigation of processes of our dynamic Earth.
T31D-04 08:55h
Upper Mantle Anisotropy and Normal Mode Coupling
We show that upper mantle models that account for lateral variations in radial anisotropy offer a better explanation for the coupling of normal mode multiplets of the type ${}_0S_l-{}_{0}T_{l+1}$ than isotropic models. These modes are sensitive to the upper mantle only and, although their coupling is known to be mostly due to the Coriolis force, a large part of the degree 2 signal measured by Resovsky and Ritzwoller [1998] remains to be explained. Here, we compare the effect of isotropy and radial anisotropy on the coupling of these pairs of modes. We test several isotropic mantle models filtered at degree 2, and anisotropic models of the upper mantle previously obtained by Beghein and Trampert [2004] with surface wave phase velocity maps and a model space search approach. We find that most of the signal cannot be explained by Coriolis coupling and isotropic upper mantle structure. On the contrary, degree 2 models including shear-wave radial anisotropy in the upper mantle predict structure coefficients that are significantly closer to the data than any existing isotropic models. We also show that the correlation between predictions and data is much higher when anisotropy is included, especially for multiplets whose sensitivity to elastic parameter $N=\rho V_{SH}^2$ increases in the uppermost mantle and transition zone. Interestingly, coupled mode multiplets that are sensitive to the entire mantle (e.g. ${}_3S_1-{}_{1}S_{3}$ or ${}_3S_7-{}_{5}S_{5}$) can be relatively well explained by isotropic degree 2 structure. However, it should be noted that these modes are sensitive to both shear-wave and P-wave related elastic parameters, as opposed to modes such as ${}_0S_l-{}_{0}T_{l+1}$ which can only see shear-wave anomalies. ${}_nS_l-{}_{n'}S_{l'}$ coupled mode structure coefficients could, therefore, bring some constraints on upper mantle P-wave anisotropy or on anisotropy at larger depths.
T31D-05 09:10h
Global upper mantle azimuthal anisotropy and the peculiar behavior of the Australian plate
We have build a global upper mantle tomographic model of Sv-wave heterogeneities and azimuthal anisotropy as a function of depth, from the analysis of over 100,000 fundamental and higher mode Rayleigh waveforms. The selected waveforms are mostly associated with epicenter-station paths shorter than 6000 km and typical of regional surface wave tomography at the scale of a tectonic plate. They provide a global coverage of the Earth with an original dataset, less likely to be affected by spurious effects such as multipathing or focusing/defocusing, compared to the longer R1 and R2 paths classically used in global tomography. We observe a peculiar behavior of the fast-moving Australian plate, which appears to be the only continent for which basal drag on the lithosphere is sufficient to cause azimuthal anisotropy aligned with plate motion. Beneath other continents, azimuthal anisotropy vanishes near 150 km depth and supports a frozen-in origin within the lithosphere with no evidence for a deeper layer, which would explain the 1 s delay time typically observed in SKS studies and the good correlation between results from SKS splitting and past tectonics for continents other than Australia. The weak azimuthal anisotropy observed at depth greater than 150 km for continents other than Australia is compatible with simple shear leading to anisotropy with a plunging axis of symmetry.
T31D-06 09:25h
Tomographic inversion of amplitudes and phases of Rayleigh waves with 2-D sensitivity kernels applied to southern California
We use 2-D sensitivity kernels for fundamental-mode Rayleigh waves based on single-scattering (Born) approximation to calculate the effects of heterogeneous structure on wave field in a regional surface wave study. The calculated phase and amplitude data using the 2-D sensitivity kernels are compared to phase and amplitude data obtained from synthesized seismic waveforms modeled by pseudo-spectral method for a plane Rayleigh wave propagating over heterogeneous structure. The kernels can accurately predict the variation of the wavefield caused by the heterogeneous structure. An inversion method based on the sensitivity kernels is developed and applied to synthesized cases. The method can almost completely recover anomalies within an array of stations when the size of anomalies is larger than one wavelength of surface waves. The new method is also used to invert for phase velocities in southern California with Rayleigh waves recorded at the Trinet network at periods from 25 to 143 s. The new method has much higher resolution compared to our previous method that uses a Gaussian function with same characteristic length along ray path to represent sensitivity, and the inversion yields about 15 percent variation reduction for short periods. New phase velocity images show larger anomalies in amplitude with high velocity along the coast area in southern California, a low velocity anomaly in Western Basin and Range and a high velocity anomaly beneath Peninsular Range. The well-known high velocity anomaly associated with Transverse Range is also clearly imaged, which can be seen at periods up to 83 s in the eastern Transverse Range. The new methods can also retrieve attenuation structure and station site responses, which will be incorporated as additional information with 2-D phase velocity to better constrain shear velocity beneath southern California.
T31D-07 09:40h
High Resolution Anisotropic Structure of the North American Upper Mantle From Inversion of Body and Surface Waveform Data
Seismic anisotropy provides insight into upper mantle structure as well as paleo and recent deformation processes. To date, our knowledge of the North American anisotropic structure arises mainly from global tomographic models or \textit{SKS} splitting studies which lack horizontal and vertical resolution respectively, and are limited to either radial or azimuthal anisotropy. Our goal is a new high resolution model for the North American upper mantle incorporating both radial and azimuthal anisotropy. We hope to achieve unprecedented lateral and depth resolution by improving both methodology and data coverage. We invert seismic long period waveform data in the framework of normal mode asymptotic theory (NACT). The resulting broad band sensitivity kernels allow us to exploit the information contained in long period seismograms for fundamental mode surface waves, overtones and body waves simultaneously. Until now, this approach has only been applied at the global scale. We have adapted the NACT algorithm for the regional case by implementing a lateral parametrization in terms of spherical splines on an inhomogeneous triangular grid of nodes, with the finest mesh for North America. Moreover, accurate crustal corrections are essential for the quality of high resolution regional tomographic studies, because they prevent the mapping of unresolved shallow features into the mantle structure. Going beyond the linear perturbation approximation, we split the correction into a linear and non-linear part. In this way, we can deal with the large lateral variations over a short distance observed in Moho topography more accurately. The inverted dataset consists of more than 100,000 high quality 3 component body, fundamental and overtone surface waveforms, recorded at broad band seismic stations in North America from teleseismic events and provides a fairly homogeneous path and azimuthal coverage. We use information from \textit{SKS} splitting measurements as additional constraints on our anisotropic model. While we focus here on the first step and present the radial anisotropic structure, we also show preliminary results from a model characterized by a more complete anisotropic parametrization. We discuss the prominent features of the anisotropic upper mantle structure beneath North America, in the light of unresolved geophysical questions such as the nature and strength of lithosphere/asthenosphere coupling, the depth extent of continental sub-regions and the relation of observed seismic anisotropy to present-day asthenospheric flow and/or past tectonic events recorded in the lithosphere.