U34A-01 INVITED 16:00h
Stability and Crystal Structure of MgSiO$_3$ Perovskite in the Earth's Deep Mantle
Experiments at high pressures and temperatures indicate that (Mg,Fe)SiO$_3$ perovskite is likely the dominant mineral of the lower mantle. However, many contradicting results have been reported regarding its stability and structure. We have measured X-ray diffraction from MgSiO$_3$ in double-sided laser-heated diamond cells at 50-144 GPa and 1600-2900 K using four different starting materials (glass, MgO+SiO$_2$ mixture, enstatite, perovskite) at the GSECARS sector of APS. The platinum pressure scale was used, and the samples were insulated by an argon pressure medium. We have confirmed the stability of Mg-perovskite to at least 2500-km depth conditions. A new diffraction line appears at 88 GPa and 2000 K together with those of Mg-perovskite. The shift of the new line on compression is consistent with those of the Mg-perovskite lines. These observations may indicate a modification of the perovskite structure at 88 GPa. However, other possibilities still remain to be investigated, such as chemical reactions and impurity. At 107 GPa after heating, we observed a weak line which can be assigned to the most intense SiO$_2$ diffraction line. However, the peak was observed only at 10 microns distance from the laser-heated spot, where extreme thermal gradients exist, consistent with the proposal that observations of the dissociation may have been caused by the presence of large thermal gradients at these conditions. At 144$\pm$10 GPa and above 2500$\pm$200 K, together with the major diffraction lines of Mg-perovskite, we observed new lines which are consistent with the recently proposed post-perovskite phase [Murakami et al., 2004]. Our data confirm that perovskite may transform at the bottom of the lower mantle. If so, this new transition may affect the pattern of mantle convection near the core-mantle boundary, perhaps influencing plume formation and the underlying geodynamo. However, important issues that remain to be resolved include the effects of minor elements on the depth and sharpness of the transition, and the uncertainty in pressure scales.
U34A-02 INVITED 16:15h
Evidence for a plum pudding mantle from deep seismic scattering
Recent analyses of PKP precursors and other seismic phases indicate scattering from small-scale heterogeneities in the deep mantle. These features are almost certainly compositional in origin because thermal perturbations would not last very long at the 5 to 20 km scale length of the anomalies. These results support ``plum pudding'' models of the mantle, in which incomplete mixing permits small blobs of chemically distinct material to persist for many convection cycles. In general, analyses of short-period scattered waves provide a window into deep Earth structure at much finer scales than those resolved by tomographic inversions for three-dimensional seismic velocities. Recent computational advances permit modeling of scattering throughout the Earth, despite large differences in scattering strength. Most observations to date can be roughly explained with about 4% RMS velocity heterogeneity in the upper 200 km of the mantle and 0.5% heterogeneity in the lower mantle, although there are also large geographic variations in scattering strength.
U34A-03 16:30h
A Whole-Mantle Three Dimensional Radially Anisotropic S Velocity Model
We present a 3D radially anisotropic model of the whole mantle obtained using a large three component surface and body waveform dataset and an iterative inversion for structure and source parameters based on Nonlinear Asymptotic Coupling Theory (NACT) (Li and Romanowicz, 1995). The model is parameterized by isotropic $V_S$ up to spherical harmonic degree 24 and $\xi$ ($\xi = V_{SH}^2 / V_{SV}^2$), a measurement of radial anisotropy in shear velocity, up to degree 16. While the isotropic portion of the model is consistent with previous shear velocity tomographic models, the anisotropic portion suggests relationships between flow and anisotropy in a vairety of depth ranges. In the uppermost mantle, we confirm observations of regions with $V_{SH}>V_{SV}$ starting at $\sim$80 km under oceanic regions and $\sim$250 km under old continental lithosphere, suggesting horizontal flow beneath the lithosphere (Gung et al., 2003). We also observe a $V_{SV}>V_{SH}$ signature at $\sim$200-300 km depth beneath major ridge systems with amplitude significantly correlated with spreading rate for fast-spreading segments. In the transition zone (400-700 km depth), regions of subducted slab material are associated with $V_{SV}>V_{SH}$. We also confirm the observation of strong radially symmetric $V_{SH}>V_{SV}$ in the lowermost 300 km (Panning and Romanowicz, 2004). The 3D deviations from this degree 0 signature are associated with the transition to the large-scale low-velocity superplumes under the central Pacific and Africa, suggesting that $V_{SH}>V_{SV}$ is generated in the predominant horizontal flow of a mechanical boundary layer, with a change in signature related to transition to upwelling at the superplumes. We also solve for source perturbations in an interative procedure. Source perturbations are generally small compared to published Harvard CMT solutions, but significantly improve the fit to the data. The sources in the circum-Pacific subduction zones show small but clearly systematic shifts in location due to an improved structural model.
U34A-04 16:45h
The survival of geochemical mantle heterogeneities
The last decade witnessed major changes in our perception of the geochemical dynamics of the mantle. Data bases such as PETDB and GEOROC now provide highly constrained estimates of the geochemical properties of dominant rock types and of their statistics, while the new generation of ICP mass spectrometers triggered a quantum leap in the production of high-precision isotopic and elemental data. Such new advances offer a fresh view of mantle heterogeneities and their survival through convective mixing. A vivid example is provided by the new high-density coverage of the Mid-Atlantic ridge by nearly 500 Pb, Nd, and Hf isotopic data. This new data set demonstrates a rich harmonic structure which illustrates the continuing stretching and refolding of subducted plates by mantle convection. Just as for oceanic chemical variability, the survival of mantle geochemical heterogeneities though mantle circulation can be seen as a competition between stirring and renewal. The modern residence (renewal) times of the incompatible lithophile elements in the mantle calculated using data bases vary within a rather narrow range (4-9 Gy). The mantle is therefore not currently at geochemical steady-state and the effect of its primordial layering on modern mantle geochemistry is still strong. Up to 50 percent of incompatible lithophile elements may never have been extracted into the oceanic crust, which generalizes a conclusion reached previously for $^{40}$Ar. A balance between the buoyancy flux and viscous dissipation provides frame-independent estimates of the rates of mixing by mantle convection: primordial geochemical anomalies with initial length scales comparable to mantle depths of plate lengths are only marginally visible at the scale of mantle melting underneath mid-ocean ridges ($\approx$~50~km). They may show up, however, in hot spot basalts and even more in melt inclusions. Up to 50 percent primordial material may be present in the mantle, but scattered throughout as small ($<$~10~km) domains, strongly sheared and refolded, and interlayered with younger recycled material. The exploration of the fine-scale geochemical structure of the mantle and the quest for preserved remnants of very old mantle arise as the strongest priorities of deep Earth geochemistry.
U34A-05 17:00h
Mass Balance Considerations for Highly Siderophile Elements Between the Upper and Lower Mantle
The highly siderophile elements (HSE) are strongly partitioned into metal relative to silicates. In the terrestrial planets these elements are concentrated in metallic cores. Earth's mantle has sufficiently high abundances of the HSE that it has been hypothesized approximately 0.1-0.8 % of the mass of the Earth was added following the last major interactions between the core and mantle. The "late veneer" likely consisted of both differentiated and undifferentiated planetesimals of moderate size. Uncertainties in the amount of mass added to Earth by late accretion largely stem from our lack of understanding of the distribution of the HSE within the mantle. If the HSE abundances present in the upper mantle are representative of the entire mantle, then the upper end of the mass requirements (~0.4 to 0.8 % of the mass of Earth) must have been added to the Earth subsequent to the conclusion of substantial metal-silicate equilibration. This is considerable mass (roughly 0.3 to 0.6 % the mass of the Moon). Assuming the HSE comprising the late veneer were added to the mantle as additions to the outer Earth, it is conceivable that HSE were vertically stratified in the terrestrial mantle during early Earth history, possibly extending to the present. The latter option, however, is only possible if there has been limited chemical transport between the upper and lower mantles. Thus, independent constraints on the mass of the late veneer may be useful to assess how well the modern mantle is mixed, at least with respect to HSE. Constraints on the mass of the late veneer for the Earth-Moon system can potentially best be generated from studies of the Moon. The giant impact that created the Moon likely marked the last period of major core-mantle interaction for Earth. Hence, the Moon probably experienced the same flux of late accretionary materials as Earth. If HSE are uniformly distributed throughout the entire terrestrial mantle, then the minimum required influx mass is approximately 2 x 10$^{22}$kg. Using the latter mass and a commonly reported Earth/Moon mass influx ratio of 35, suggests a late veneer comprising ~6 x 10$^{20}$ kg for the Moon. Our new estimates for the concentrations of HSE in the lunar mantle suggest that such large amounts of mass were not added to the lunar mantle, and are also likely not present in the lunar crust. The mass of the late veneer added to the Moon was probably no more than 8 x 10$^{19}$ kg. This limited amount of mass may be difficult to reconcile with concentrations of HSE present in the terrestrial mantle. Options include the possibilities that high abundances of the HSE are present in only the upper mantle, indicative of limited whole mantle mixing, or that the estimate of the Earth/Moon influx ratio has been incorrectly estimated.
U34A-06 17:15h
Complexity in the Thermal Structure of Deep Mantle Plumes
Thermal plumes are an intrinsic part of any convection system in which the lower boundary is heated. Recent analyses of the core dynamo conclude that heat flow out of the core could be a significant fraction of total surface heat flow. In this case hot thermal plumes should be an important element of the mantle convection system. Thermal plumes arise because fluid in the lower thermal boundary layer detaches from the boundary and moves upwards, either steadily or as transient bursts of hot fluid. Recent seismic images of plumes by Montelli et al. (Science, 203, 338-343, 2004) show considerable internal structure and depth variation of plume velocity anomalies (interpreted usually as thermal anomalies). Some plumes appear to originate from the base of the mantle, others perhaps from shallower depths. I have carried out simplified numerical experiments of constant-viscosity convection at high Rayleigh number (Ra) in order to obtain model plume structures that can be compared with the seismic images. At Ra = 5,000,000, with isothermal boundaries and a plausible rate of internal heating, complex plume structures are obtained. In general a narrow core of hot fluid extends from lower to upper surface, with the thermal anomaly decreasing with height above the lower boundary (contrary to the seismic images). Plume locations at mid-mantle depths are determined by an irregular network of hot sheets, just above the lower boundary. Upward flow is focussed into plumes at the nodes where these hot sheets are joined. Cross-sections of model plumes at mid-mantle depths typically reveal an elongated thermal anomaly, possibly with two or more plume cores in close proximity. The flow pattern has long-term persistence, but in detail has variable time dependence. In particular, boundary layer instabilities that produce new plumes are most likely to occur close to existing plumes, creating new plume conduits that rise parallel to existing conduits. The overall geometrical similarity between seismic and model plume images inspires confidence in both the seismic images and the numerical simulations. Differences between model and seismic images may yet be exploited to obtain improved models of mantle convection properties.
U34A-07 17:30h
Testing Mantle Flow Models with Joint Seismic-Geodynamic Inversions
Fundamental questions remain regarding the means by which the Earth's mantle convects. Three-dimensional seismic velocity models are instantaneous snapshots of the deep interior and should yield important information about the type of convection that occurs. However, these models have irregular resolution within them which changes amongst different models due to several factors including variation of datasets and methodology. Furthermore, the interpretation of the tomography models remains debatable. Seismic tomography models continue to improve over time; however, it is likely that their interpretation in terms of mantle layering will continue to be debated. Global gravity anomalies, dynamic surface topography, and plate motions are other surface observables that contain mantle flow signatures. With a density and viscosity field, these observables can be predicted in a forward model sense via dynamic response functions. Dynamic response functions relate a given density anomaly to a given geophysical observable and depend on the existence of chemical or phase change boundaries that impede vertical mass transport across the mantle as well as the viscosity of the mantle. Consequently, we may discriminate amongst various flow models that may or may not contain barriers to mantle flow. Using various scalings between seismic velocity and density, we can jointly invert geodynamic and seismic data for seismic velocity. We present some joint 3-D shear-wave seismic velocity inversions with various flow model assumptions using optimized velocity to density scalings and viscosity structures. For the seismic data, we use Sn, sSn, ScSn, sScSn, SKS, and SKKS travel time residuals (n refers to multi-bounce waves such as SS) totaling over 44,000 measurements. We combine these data with free-air gravity, dynamic topography, plate motion, and dynamic ellipticity measurements to generate the joint models for given mantle flow model assumptions. We have determined that we can fit the seismic and geodynamic data equally well as the seismic data alone given a whole mantle flow assumption. However, with the assumption that a barrier to flow exists near the 660km depth range, an equal fit to the seismic data cannot be established with the current data and model parameters. We will also present results from other mantle flow assumptions including impedance to flow near 1000km and 2000km depth ranges as well as alternate velocity-density scalings. Along with mantle flow mechanics, these results will give us an indication of the likelihood that negative velocity-density scalings exist in the deepest mantle as has been proposed for the African anomaly.
U34A-08 17:45h
Explaining Topography of the 410-km Discontinuity: Chemical vs Thermal Heterogeneity in the Mantle
Discussion of $SS$ and $PP$ precursors from the 410-km discontinuity (termed $P410P$ and $S410S$) has largely focused on the long-wavelength topography of this discontinuity. Recent models suggest precursor derived topography of the 410-km discontinuity is dominated by low degree spherical harmonics. However, there remain discrepancies as to the details of the topography and its relation to mantle processes. In particular, global models of discontinuity topography do not show significant correlations with expected regions of thermal anomaly in the mantle. This could partly be explained by the influence of water and other chemical heterogeneities on the olivine phase change which produces the 410-km discontinuity. We present a new global model of topography for the 410-km discontinuity which is constrained using both $P410P-PP$ and $S410S-SS$ observations. We examine the relationship between our observations of 410-km discontinuity topography and seismic velocity anomalies in the transition zone. We find a moderate negative correlation between discontinuity topography and seismic velocity anomalies at low spherical harmonic degrees. This is consistent with the expected relationships between the $\alpha -\beta$ olivine phase boundary position, seismic velocity anomalies, and lateral variations in mantle temperature. However, when shorter wavelengths are included the relationship becomes more complex. In smaller regions beneath Northeast Asia and the Western Pacific we observe positive correlations between seismic velocities and discontinuity topography, which we interpret as evidence for the effects of chemical heterogeneities on seismic velocities and the position of 410-km discontinuity. Our results suggest that long wavelength discontinuity topography is primarily controlled by temperature variation in the mantle, but at shorter wavelengths the influence of chemical heterogeneities becomes important. This separation of the spectra for thermal and chemical heterogeneities suggests that chemical heterogeneities can survive in a convecting mantle.