U41B-0402
Imaging Lower Mantle Heterogeniety With Differential Dispersion and Attenuation of Core- Diffracted Waves
We investigate global differential travel-time dispersion and attenuation of core-diffracted phases from large, deep earthquakes. This technique aids in constraining radial velocity structure at the core-mantle interface in a manner analogous to surface wave observables constraining upper mantle structure. We confirm that there is noticeable differential dispersion and attenuation caused by diffraction on a global basis for both Pdiff and Sdiff. Variations in differential dispersion and attenuation are observed with both geographic location and between Pdiff and Sdiff along the same azimuth suggesting lateral variations in Vp, Vs and Vp/Vs ratio in the lowermost mantle. We attempt to utilize dispersion and attenuation characteristics to put bounds on the magnitude and distribution of large-scale velocity perturbations in the lowermost mantle and draw comparisons to variations found in several 3D whole-mantle models. Our dataset consists of broadband records available from the IRIS DMC for deep (>180 km), large (>5.6 mb) teleseismic events. Preprocessing of the records includes deconvolution of the instrument response, rotation of horizontal components, filtering using a set of bandpass filters, and sample-rate decimation (5 sps). Relative arrival times and amplitudes are found by computing cross correlegrams in the frequency domain, detecting and removing poor recordings with cluster analysis, and iteratively converging on a stable low-variance solution with a weighted least-squares inversion while automatically remediating phase-skips utilizing a database of potential relative arrivals. Raypath-approximated corrections for reciever-side differences in ellipticity, mantle, and crust are applied for the derivation of phase velocites in the lowermost mantle as a function of azimuth and frequency. Following previous studies of diffracted signals, we limit our analysis to station pairs located in narrow azimuthal windows spread over a considerable distance while attempting to quantify the acceptable range of aspect ratios. This method has the advantage of removing source-side effects, averaging out minor timing errors, and, for our analysis, averaging out receiver-side frequency-dependent upper mantle and crustal biasing. Comparison with 1D reflectivity and 3D SEM synthetics facilitates our quantitative analysis of the lateral and vertical variations in the seismic velocity structure near the core-mantle boundary region.
U41B-0403
Mantle Plumes: Seismic and Dynamic Models
Global-scale seismic tomography is a good tool to test the plume hypothesis and mantle models. We attempt to objectively characterize coherent, plume-likeslow/hot anomalies in recent maps of P and S velocity throughout the mantle, combining the following methodological improvements: (i) A "plume-detection" algorithm, inspired by geodynamic studies. This method identifies a variety of vertically coherent features, with similar properties, in all considered tomographic models. (ii) Quantification of the similarity (e.g. correlation) between tomographic and geodynamic plume-conduit models. Our results favor the idea that only a subset of known hotspots have a lower-mantle origin. Most of those that do can be associated with the large low-velocity zones in the lowermost mantle. From tests of different geodynamic forward models, we find that plume conduits that are distorted in mantle flow yield much better correlations with slow tomographic anomalies than vertical conduits. We complement a formal statistical analysis with Monte- Carlo tests and show that the plume-match is highly significant for several tomographic models. This finding provides further evidence that our geodynamic models adequately describe the nature of hot upwellings rising in mantle flow. We also find that models that allow for moving plume sources in the lower mantle match tomography better than those that have the plumes origin fixed. This is consistent with a thermal boundary layer source, and implies that anchoring by thermo-chemical piles is not required. Our flow-based results are complicated by a circularity issue: Plume conduits depend on mantle flow, which is driven by density anomalies, and density anomalies are in turn inferred from seismic tomography. We address this issue by modeling plume conduits from different tomography models, and show that the match between predicted plumes and tomography is a robust feature of all models.
U41B-0404
Constraints on shear-velocity reductions in the lowermost mantle by using ScP postcursor data
Lateral distributions of the ultralow-velocity zones (ULVZs) at the base of the mantle are constrained by ScP waves reflected and converted at the core-mantle boundary (CMB). The amplitude of postcursor to ScP, which is excited at the ULVZ surface, is strongly dependent on the shear velocity contrast and has been proven to be a very powerful probe for detecting the localized ULVZs in previous studies. In this study, we have assembled a large amount of waveform data of ScP recorded at short-period seismic stations deployed in the Japan Islands and world-wide seismic stations of International Monitoring System. Clear arrivals of postcursor are detected beneath Philippine-Kalimantan and East of Australia, indicating shear velocity reductions of 15% to 40%. On the other hand, no postcursor is found beneath North Pacific and the Asian continent. The estimation of noise levels of waveforms bounced in North Pacific and the Asian continent indicates that there is no significant boundary with shear wave velocity reduction over 5% in the lowermost mantle under those regions.
U41B-0405
Structure of the Edge of the Pacific Low-Velocity Feature Investigated Using Philippine Source Array
The D region of the lowermost mantle has been much studied in recent years, since understanding of its structure will shed light on mantle convection systems and core-mantle interactions. In particular, the role of the post-perovskite phase transition in creating the seismic discontinuities in D has been the focus of much interest, along with the suggestion that the large scale seismic velocity reductions under Africa and the central Pacific may be due to piles of chemically and thermally distinct material. We investigate D structure under the western Pacific, north-west of the Marshall Islands, by using earthquakes in the Philippine Sea recorded in Hawaii to construct source arrays. These events have core-mantle boundary bounce points which sample across the edge of the central Pacific low-velocity region found in seismic tomography of the lowermost mantle. Selection of groups of earthquakes with suitably similar source mechanisms and depths allows us to perform migrations of both P- and S-waves to show variations in the seismic velocity structure of D across the edge of the low-velocity feature. This provides insight into the role of post-perovskite, thermal and chemical heterogeneity in the behaviour of the lower mantle.
U41B-0406
Inverse scattering with SKKS code waves: imaging the D" from the core side
In our previous studies we developed a method for imaging heterogeneity at and near the core mantle boundary (CMB) with a generalized Radon transform (GRT) of (transverse component, broadband) ScS data, and we developed a statistical model for producing images of the D" discontinuity with variable confidence levels. In these applications the background is smooth and perturbations are represented as contrasts. Here we extend the theory to allow (known) discontinuities, such as the CMB, in the background model. The resulting imaging operator, which is formally not a GRT, can be exploited to scan the CMB discontinuity from the core side using the (radial component) wavefield of the underside SKKS reflection. We demonstrate furthermore that the SKKS phase can be used, either alone or along with ScS, for the imaging of lowermost mantle structure and, in particular, the D" discontinuity. Synthetic seismograms calculated with the SEM method are used to test the performance of our approach. As a proof of concept, we transform ~18,000 radial component SKKS waveforms into image gathers of a CMB patch beneath Central America. The SKKS image gathers and image traces are in good agreement with the image traces obtained from the GRT transform of ScS data. We furthermore do the joint inversion of ScS and SKKS data and obtain images with better resolution, in particular of interfaces that are difficult to detect with topside ScS reflections.
U41B-0407
Seismic Velocity Structures of a High-Velocity Area Near the Core-Mantle Boundary Beneath Eastern Eurasia
The shear and compressional velocity structures of a high-velocity area near the core-mantle boundary (CMB) beneath the eastern Eurasia are studied based on waveform modeling and differential travel time analysis of ScSH-SH and PcP-P phases. Our seismic data are obtained from the F-net in Japan, the Global Seismographic Network and several PASSCAL arrays. The observed ScSH-SH and PcP-P differential travel time residuals show similar patterns, but also a change of P/S velocity ratio across the region. The seismic data suggest that the average shear and compressional velocity increases reach 3% and 2% in the lowermost 300 km of the mantle, respectively. The waveform data from the shear and compressional waves sampling the region can be explained by a model with a D" discontinuity with a large variability of its structure. The data sampling southwest and southeast of the high-velocity area can be explained by simplified one-dimensional models with a D" discontinuity that are characterized by a shear velocity jump of 5% to 3.4% and a compressional velocity jump of 2.6% at about 220 km to 260 km above the CMB followed by a negative gradient toward the CMB. But, the shear wave data sampling the region between would exclude a D" discontinuity with a shear velocity jump greater than 1%. We will discuss effects of two-dimensional wave propagation on the inferred seismic structures and the D" discontinuity, and implication to the origins of the inferred seismic structures in the region.
U41B-0408
Probing two low velocity regions with PKP b-caustic amplitudes and scattering
Knowledge of the lowermost mantle region (D") is important in understanding the dynamics of the mantle and the core-mantle interactions. In several regions reflections from the D" discontinuity are observed, however, the imaging in regions that show slow velocities in tomographic models is still hampered by the source-receiver combinations currently available. Some of the regions with slow seismic velocities have been found to exhibit strong scattering. The scatterers are thought to be close to the core-mantle boundary, but scattering throughout the mantle has also been reported. We are probing low velocity regions of the lower mantle beneath the southwest Pacific using PKP waves recorded at the temporary EAGLE array (Ethiopia) and the GRSN (Germany) with epicentral distances between 130 and 143 degrees. The earthquake locations are in the South Pacific for both arrays. The PKP waves to Ethiopia show strong scattering of PKPab waves which arrive as precursors to PKPdf and in addition show large PKP-b caustic diffracted amplitudes. In contrast, the events recorded at the GRSN show scattering of PKPab as precursors to PKPdf but smaller PKP- b diffracted amplitudes. To understand the observed amplitudes we test a large set of models with velocity and density anomalies in both mantle and core. Explanations for the observed waveforms and amplitudes are low velocity regions just above the core-mantle boundary, on the order of 50 km thick for the South Pacific to Ethiopia path with velocity reductions of around 5% for P and larger reductions for the S-wave velocity. The best fitting models require a change in attenuation in the outer 30 km of the inner core (Qp around 100). The low velocity regions act as waveguides for the PKP-b caustic diffracted waves. They cannot be smooth layers but must exhibit a certain roughness to produce the high-frequency scattering observed in the data. The path to Germany also requires small scatterers but the low velocity region is less pronounced compared with that for the South Pacific - Ethiopia path. Cumulatively, these results suggest that low-velocity-zones at the base of the mantle are regions of small-scale heterogeneity populated with low-velocity inclusions.
U41B-0409
Is the double crossing phase transition in D" ubiquitous?
By using waveform inversion to study the seismic velocity structure of D'', we found that the upper half of D'' beneath Central America has S-wave velocity significantly in excess of PREM, while the lower half has S-velocity nearly equal to PREM [Kawai et al., GRL, 2007]. We interpret this as evidence for a double crossing phase transition (a reverse transition from post-perovskite=ppv to perovskite=pv), the existence of which was suggested by Hernlund et al. [2005, Nature]. In addition to the D'' layer beneath Central America, we have now also inverted seismic body- wave waveform data for the vertical dependence of (isotropic) shear-velocity in D'' beneath Asia and the Arctic, also using the transverse components of relatively long period broadband waveforms (20-200 s) as data. We found that these data also suggest the existence of high S-velocity in the upper half of D'' and low velocity in the lower half, consistent with the existence of a double crossing phase transition within D''. Our results for these three separate regions, when taken together, suggest that the existence of a double crossing phase transition is widespread and possibly global or nearly so. This has important implications for studies of the mineral physics, temperature profile, convection and material transport in D''. In the near future, we plan to study D'' beneath Africa and the Pacific in order to further determine the extent to which the double crossing phase transition is (or is not) ubiquitous.
U41B-0410
D" Shear Wave Velocity Structure Under the Cocos Plate: Discontinuity Sharpness and Double Crossing
The possible occurrence of a perovskite to post-perovskite (Pv-pPv) phase transition in the deep mantle may account for complex seismic wave velocity structure in the D" region overlying the core-mantle boundary (CMB). A rapid increase in shear wave velocity several hundred kilometers above the CMB is commonly associated with the Pv-pPv phase change, while velocity decreases within 100 km of the CMB have been attributed to the reverse transformation (double-crossing) resulting from rapid temperature increase in the CMB thermal boundary layer. Establishing the sharpness of the velocity increase is important for evaluating the phase boundary interpretation, as mineral physics experiments suggest the likelihood of a significant pressure range for the two-phase domain. Analysis of stacked broadband shear waves sampling the D" region beneath the Cocos Plate is used to establish constraints on the sharpness of the 1.1-2.2 percent shear velocity increase 232-296 km above the CMB in this region. Separate stacks aligned on direct S or ScS are modeled, with lower mantle models being constrained to match ScS-S differential times. Bounds on the inferred discontinuity sharpness will be discussed in the context of Pv-pPv properties. Evidence for and implications of a rapid velocity decrease found 47-75 km above the CMB in this region will also be presented.
U41B-0411
More Seismic Evidence for a Sharp Edge to the Large Low Shear Velocity Anomaly Beneath the Pacific Anomaly
A number of seismic investigations have well documented a sharp transition between the large low shear velocity province (LLSVP) beneath the Pacific and Africa and the surrounding lower mantle. Using USArray recordings of Fiji-Tonga and Kemadec events, we study the northern portion of the LLSVP beneath the Pacific. For this geometry, normal and multi-pathed SH phases are apparent, along with rapidly changing arrival times consistent with a sharp boundary there. Wave paths sample the implied edge somewhat obliquely, resulting in strong waveform broadening in data that sample within a few hundred kilometers of the transition. A rapid change in SH- wave travel times of up to 8 s relative to the PREM reference model are observed, compared with little change relative to PREM for P wave arrival times. Such acute variations in waveforms and travel times for signals with comparable wavelengths are quite consistent with SH multipathing. Our analyses suggest a sharp lateral transition from a broad slow region to a relatively fast region in the lowermost mantle. We observe a different scaling between P and S velocities within and outside of the LLSVP, implying chemical differences between the LLSVP and the surrounding lower mantle, as well as providing a constraint on deep mantle mineralogy.
U41B-0412
Global Structure of the D" Region Using Diffracted Waves and Finite Frequency Kernels
The base of the mantle serves as a thermal, chemical, and mechanical boundary layer, separating the liquid iron core and the solid silicate mantle. Studies of the region reveal a large number of unusual features, such as the D" discontinuity, anisotropy, ultra-low velocity zones, high amplitude anomalies, and anti-correlation of bulk sound speed and shear velocity. However, the structure at the base of the mantle is not well understood, due to poor coverage by standard seismic phases, particularly in the southern hemisphere. Using a variant of cluster analysis, we have created a new data set of long period diffracted S and P times, consisting of 20,000 diffracted S and 31,000 diffracted P measurements, which significantly improves coverage at the CMB. Finite frequency kernels, made using adjoint methods, for diffracted phases are quite different from ray theory kernels and tend to have increased sensitivity to structure at depths above the CMB. Differences in travel time anomalies predicted for a model using the two different theories can often exceed one second, which is a significant fraction of the observed signal. However, preliminary modeling of Pdiff using finite-frequency kernels, in conjunction with P, PP-P, PKPab, and pP-P data sets, gives models that are very similar to ray theory models, although the finite frequency models tend to have somewhat larger amplitudes. Shear and compressional models derived from our data show large, slow regions at the base of the mantle underneath the Pacific and Africa, circumscribed by fast regions, although the patterns of the shear and compressional models are not well-correlated. To model bulk sound speed, we create bulk sound speed rays by combining our S and P data for common source-receiver pairs and invert directly for bulk sound speed. The now well resolved bulk sound speed pattern is anti-correlated with shear velocity over most of the base of the mantle, confirming the results of previous studies. This anti-correlation suggests the presence of chemical or phase heterogeneity at the base of the mantle.
U41B-0413
Searching for Seismic Signatures of a Plume Source at the Base of the Mantle Below the Galapagos Island Hotspot
This study analyzes SKS and SKKS waveforms recorded on the BOLIVAR array in Venezuela and the BANJO array in South America from earthquake sources located in Tonga and Alaska regions to characterize the lower mantle beneath the Galapagos Islands. The data analysis applies two independent methods, residual differential SKKS-SKS travel times and anisotropy measurements, to examine the historically unsampled region. The residual differential travel time observations were performed using 21 earthquakes from the Tonga trench with magnitudes greater than 5.5 Mw that were recorded on the Bolivar array. Only data that was deemed to have a high SNR for both the SKS and SKKS phases were retained for analysis. Significant positive values of differential travel time that indicate low velocity along the SKKS raypaths are detected east of ~\m270° longitude. The anisotropy data set consists of 31 intermediate and deep focus earthquakes from the Tonga and Aleutian trenches recorded on the BOLIVAR and BANJO arrays respectively. The anisotropy fast axis angle and time lag of the two phases are calculated using the 1-layer cross-convolution method of Menke and Levin (2003) with a maximum time lag window of 3 seconds. We retain results with an amplitude normalized squared L2 norm value of 0.6 or less for analysis. Because the raypaths of the SKS and SKKS phases are similar in the upper mantle and sample different regions of the lower mantle, we attribute inconsistencies between the two anisotropy to difference of the mantle structure near the CMB. We define significant difference in the azimuth of the fast axis as any difference between the SKSac and SKKSac measurements greater than 15 degrees. The dataset is dominated by inconsistent fast axis azimuth measurements between the SKSac and SKKSac phases, but does not isolate a single geographic region. Comparison of the splitting time measurements yields that inconsistency between the two phases is more significant, greater than 0.5 s, in the Northeast portion of the sampled region bounded to the south and west at approximately \m-3°S and \m267° longitude. While the residual differential travel times and the anisotropy measurements do not conclusively show that there is a mantle plume source at the base of the mantle in this region, the data does indicate there the lower mantle beneath the Galapagos Islands has significant structure meriting further study.
U41B-0414
Toward understanding Ultra Low Velocity Zone dynamics
The Ultra Low Velocity zone (ULVZ) is characterized by a sharp drop in seismic velocities in certain regions of the lowermost mantle. Furthermore, recent seismic observations indicate that the ULVZ may have a morphology of small kilometer-scale packets of anomalous material with a very high density increase. Potential explanations for the cause of the ULVZ include the presence of partial melt and/or fine-scale chemical heterogeneity (possibly due to iron enrichment). A fundamental question regarding both hypotheses pertains to whether mantle convection is dynamically capable to support such small-scale, high-density structures. We perform high-resolution, whole-mantle thermochemical convection calculations to test the feasibility of this hypothesis. We find that convection acts to segregate and support very thin layers of ultra-dense material into shapes that resemble seismic models. In addition, we test whether ultra-dense material, over the lifetime of the Earth, created in the lowermost few kilometers of the mantle can be collected into volumes similar to that of proposed thermochemical piles beneath Africa and the Pacific. Our results indicate that this is possible if the lowermost 2-5 km of the mantle is continually being injected with higher density material, possibly iron from the core.
U41B-0415
The relationship between mantle models and core-mantle boundary topography
We numerically model convection (thermochemical and isochemical) and core-mantle boundary (CMB) topography, in an effort to determine its usefulness in constraining mantle models. From our study, the most negative CMB topography for an isochemical mantle strongly correlates with regions of subduction, whereas topography beneath regions of subduction in a thermochemical mantle is not exclusively negative. In fact, inclusion of thermochemical piles leads to an overall reduction in magnitude of CMB topography, relative to isochemical models, and flat to slightly positive topography beneath thermochemical piles. Topography maps yielded from numerous investigations do not exhibit the strong relationship between topography and subduction seen in the isochemical mantle model. This, in conjunction with geochemical data, may further indicate that Earth's mantle is likely a thermochemical mantle and pushes us to further investigate the nature of the seismically observed anomalies beneath the central Pacific and Africa. Moving beyond our previous work, we investigate these mantle models and the resulting CMB topography in three-dimensions and compare them to existing topography maps for Earth's CMB.
U41B-0416
Numerical simulations of mantle convection with a basal stagnant layer bearing high internal heating: implications on Earth's thermal and magnetic evolution
In this contribution we present the results of numerical simulations of the thermal evolution of the Earth assuming a 200-km layer with high internal heating at the top of the core. Our models are inspired from recent isotopic studies that suggest the presence of a region with high internal heating residing in D' ' as an early enriched reservoir in incompatible elements (Boyet and Carlson, 2005, Tolstikhin and Hofmann, 2005). In addition, a chemically dense, distinct layer may be present in the lowermost mantle due to deep subduction (Hansen and Yuen, 1988) or possible iron enrichment (e.g. Tateno et al., 2007). The models consist of a spherical 2D axi-symmetrical model for convection in the mantle, underlain by a theoretical layer in which the heat is transferred through conduction (Lassiter, 2006). The numerical model of the mantle is coupled to a parameterized model for the thermal evolution of the core. The effects on the magnetic history and the age of the inner core are analyzed using the energy and entropy balances in the core. We compare our results with a previous study in which a high internal heating layer at the base of the mantle was allowed to participate in the whole mantle convection (Costin and Butler, 2006). We find that the presence of the stagnant layer reduces the amount of heat flow from the core and increases the age of the inner core. For models bearing more than 2 TW internal heating the heat flow from the core cannot sustain a magnetic field at early times. Our preferred model is represented by an enriched layer bearing no more than 0.5 TW radiogenic heating, for which the magnetic field exists throughout the entire evolution and predicts an inner core about 2.6 Ga old. In addition, we consider a mixed scenario in which the basal layer may be entrained in the whole mantle convection after 2.5 Ga of evolution (Davaille, 2007). For this case, the magnetic field is present over the entire geological time and the predicted age of the inner core is around 1.5 Ga.
U41B-0417
Formation of Steep-Sided Topography From Compositionally Distinct Dense Material at the Base of the Mantle
In this study we use 3D, numerical convection experiments to explore the development of a thin, compositionally distinct, dense layer at the base of the mantle. We investigate the conditions under which steep-sided, flat- topped topography will form on an initially ubiquitous layer as a result of convection in the mantle. Previous authors have found it necessary to introduce compressibility into calculations or employ a large viscosity increase in the lower layer in order for discrete, steep-sided piles of material to form. Our calculations are based on incompressible convection with layers of constant viscosity and we find that, as long as the lower layer is thin compared with the convection wavelength, it is pushed away beneath downwellings leaving steep-sided, flat- topped structures beneath upwelling regions. Experiments involving simple convection planforms allow determination of the factors which encourage or inhibit this topography style. Comparison of the results with previous 2D analytic predictions for thin layer boundary topography enables detailed physical understanding of the deformation process. When convection within the lower layer is coupled viscously with the overlying mantle convection, deformation in the form of discrete, steep- sided piles is observed. Whereas, when the convection systems are coupled thermally, deformation tends to take the form of cusp-like peaks beneath upwelling regions with a relatively small deflection beneath downwelling regions. The dependence of the interface topography style on various system parameters (Rayleigh number, density contrast, initial layer thickness, convection wavelength) can be largely explained by a scaling derived from a balance of buoyancy derived stress with viscous flow stress. The results also explain why the addition of compressibility or a significant viscosity increase in the lower layer has enabled this topography style in previous studies. Calculations with plausible convection planforms are used to estimate parameter ranges which generate interface topography consistent with that observed in seismic studies.
U41B-0418
Controls on Thermal and Dynamic Characteristics of Descending Slabs in the Lower Mantle
Global seismic tomography studies show that subducted slabs become significantly less "visible" in the lower mantle [Masters and Stixrude, 2006], although they are one of the dominant structures in the upper mantle and mid-mantle. On the other hand, the upwelling plume-like structure, while relatively weak in the upper mantle, becomes significant in the lower mantle, particularly above the CMB. The dominant slab structure in the upper mantle and mid-mantle is often suggested to be consistent with the fact that slabs transfer most of the surface heat flux of the Earth and are the main driving force of plate motion. Recently, Leng and Zhong [2007] have demonstrated that upwelling plumes dominate energy flux in the bottom part of the lower mantle, thus providing an explanation for the increased seismic "visibility" of plumes in the lower mantle. Using 3-D regional spherical finite element code CitcomCU with extended-Boussinesq approximation, we studied the thermal and dynamic characteristics of descending slabs and their implications for mantle geotherm (i.e., subadiabatic temperature) and seismic observations. Our main results can be summarized as following: 1) Slabs are being heated by adiabatic heating and diffusive heating effects as they descend, which lead to the great reduction in slab temperature anomalies, slab mass flux and slab heat flux. 2) Subadiabatic temperature is strongly related with slab deficit temperature and slab heat flux rather than with internal heating rate of the mantle. The reduction of slab heat flux from the surface to the CMB and the slab temperature anomalies near the CMB are both proportional to subadiabatic temperature. Although these results are derived from models with extended Boussinesq approximation, their validity will be verified in fully compressible convection models that we have recently developed.
U41B-0419
Mantel Convection Models of the D" Region
We have investigated heterogeneity in the bottom D" region of Earth's lower mantle by means of numerical mantel convection modelling using a 2-D cylindrical geometry. The finite element model used, is based on the extended Boussinesq approximation, and includes the 660 km phase transition in the transition zone and the postperovskite transition in the D" region. We apply variable conductivity and phase dependent composite rheology in our model to account for increased conductivity and a higher tendency towards non-linear dislocation creep of postperovskite. We used particle tracers to reveal compositional variations related to compositional layering associated with a subducted oceanic lithospheric slab. The results show lens-like structures of postperovskite directly above the core-mantle boundary with an underlying thin layer of hot perovskite, similar to the structures related to the Cocos subduction, recently resolved in seismological studies (Van der Hilst et al.,2007). Oceanic basaltic crust and a harzburgitic layer become strongly folded, generating small scale structures at the bottom of the mantle. We show model predictions in terms of seismic reflectivity of the complex structures resulting from variations in temperature, phase and composition that will aid in the interpretation of (image-processed) seismic reflectivity cross sections of the D" region.
U41B-0420
Chemical equilibrium between molten iron and mantle minerals
The molten outer core may be in chemical equilibrium, at least with the bottom thin layer of the mantle, which may be comprised mainly of (Mg,Fe)SiO3-rich perovskite or post-perovskite and (Mg,Fe)O ferropericlase. The element partitioning data between molten iron and these mantle minerals at high pressure and temperature are important to constrain the chemical compositions of the liquid core and the lowermost mantle. The element partitioning between perovskite/post-perovskite and molten iron and between ferropericlase and molten iron has been studied at high pressures (e.g., Takafuji et al., 2005 GRL; Sakai et al., 2006 GRL; Asahara et al., 2007 EPSL). Here we examined the chemical equilibrium between three phases, perovskite/post-perovskite, ferropericlase, and molten iron to the core-mantle boundary condition by using the laser-heated diamond-anvil cell techniques. A starting material was prepared as a powder mixture of iron metal and gel with a composition of (Mg0.9Fe0.1)2SiO4. Mineral assemblage was confirmed by in-situ synchrotron X-ray diffraction measurement at high pressure and temperature at BL10XU of SPring-8. The chemical compositions of coexisting perovskite/post-perovskite, ferropericlase, and quenched liquid iron were determined with analytical transmission electron microscope. A thin section of the recovered sample was obtained parallel to the compression axis by Ar-ion milling method using Ion Slicer (JEOL EM-09100 IS). Results demonstrate that the dissolution both of oxygen and silicon into a liquid metal is enhanced with increasing pressure. We will discuss the chemical compositions of the outer core and the bottom of the mantle based on the present experimental results.
U41B-0421
Siberian Superplumes: Did They Really Exist?
Siberian traps can be regarded as a unique natural object because of their structure, composition and voluminous lava shields, which are rarely found in other LIPs. As is known, Perm-Triassic traps of the Western Siberian Plate (WSP) and Siberian Platform (SP) are due to multi-level magma sources and are of a certain geochemical and petrochemical zoning (Sharma 1997; Reichow et al 2005). In Russian geodynamic and petrogenetic literature of the last decade, the formation of Mesosoic granitoids of Central Asia and Siberian Platform is explained due to the influence of 'superplumes' on the continental lithosphere. However, no details are given concerning the mechanism for this. This work is a generalization of a series of numerical experiments using a quantitative model for the intraplate mantle-crust magma system, where all the conditions for the melting zones of the convecting upper mantle, interacting with the multi-layered lithosphere were taken into account. For this we used a modified version of MAiX 2D convection codes for a complex 7-phase diagram (Perepechko, 2003). The three-layered structure for the profile of melting materials has been taken into account (Sharapov et al, 2006). Both the depth and the horizontal dimensions of melting zones, as well as their detailed structure and evolution can be computed using the convection codes. The results of our modeling showed that the idea of the existence of 'superplumes' explaining SP and WSP trap magmatism is disputable; both systems rather seem to be independent and have different features for multilevel melting zones development. The virtual voluminous magma shields coextensive with what we find in reality can be obtained due to a series of scattered hotspots. For the scheme of trap formation development, two-level melting zones with tholeiites in the metasomated lithosphere appear virtually over hotspots. Sometimes over hotspots, the appearance of two successive melting cycles is possible. This work was supported by the Russian Ministry for Science and Education (Grant DSP.2.1.1.702) and by RFBR Grant # 07-05-00685.
U41B-0422
How can the double crossings of the Post-Perovskite transition constrain the heat-flux from the core ?
There are now accumulating seismic and mineral physical evidences supporting the dominating phase-change nature of the D" region. Therefore any local information on D" topography may provide a global constraint on the thermal boundary layer at the core- mantle boundary( CMB). For instance, it has been proposed that some pairs of discontinuities above the CMB deduced from seismic imaging may result in a double-crossing of the perovskite/post-perovskite phase transition by the geotherm at two different depths. Double-crossings may occur only when the CMB temperature, T-cmb is higher than the temperature of the phase transition at the pressure of the CMB. Thus, the confirmation of this observation would fix a lower bound for T-cmb, completely independent from the one provided by the melting curve of iron. There exists also the opportunity to put bounds on the thermal boundary layer thickness and then of the core heat flux with double-crossing.Using synthetic D" layer simulated by 3D spherical mantle convection, we show that double-crossing data can be explained by various models. We show that there exists a quantitative relationship between the fraction of basal heating used in the model( the ordinate ) and the Clapeyron slope ( the abscissa ) used to simulate the D" layer that fits the double-crossing data. The relationship follows a power-law dependence, y =x**beta, where beta is around - 0.3. This dependence is not sensitive to T- cmb in the range of 3500K to 4000K. However, this power-law relationship is more sensitive to the value of the thermal conductivity at the base of the mantle. Thus it appears possible to measure the core heat flow by the Clapeyron slope of the perovskite/post-perovskite phase transition. Lastly, we demonstrate from many simulations that, if the T-cmb does not exceed the temperature of the phase change at the pressure of the CMB by more than 200K, then we can fit the double-crossing data.
U41B-0423
Phase Stability and Equation of State of Fe2SiO4 in the Earths Mantle
It is well known that the bulk mantle is much richer in magnesium than in iron. However a possible reaction at the mantle/core interface would increase the Fe/Mg ratio in surrounding material. The goal of this research is to expand our understanding of the phase relations and equation of state of the Fe2SiO4 system. Experiments were performed at the Advanced Light Source (beamline 12.2.2) in laser heated diamond anvil cells to explore the high end of the stishovite and wustite stability field. Samples were Rockport fayalite, in some cases thermally insulated between layers of NaCl, with a pressure marker of ruby, Au, or Pt. Several sets of experiments were performed generally in pairs either buffered to reducing conditions with an iron gasket, or buffered to oxidizing conditions with hematite. Several heating cycles were performed on each sample at progressively higher temperatures. Pressures of up to 53.9 GPa and temperatures up to 2500 K were achieved. Postmortem analyses of samples were performed with a scanning electron microscope. X-ray diffraction patterns show a transition from alpha-olivine structured Fe2SiO4 to the spinel structure Fe2SiO4. At 14.4 GPa we show a transition to SiO2 stishovite and FeO wustite, in agreement with earlier work. At the highest pressures and temperatures additional as-yet unidentified phase(s) are observed. In addition, X-ray diffraction patterns allow measurements of the density of the phases in the Fe2SiO4 system at mid mantle conditions. We present these results on the thermal equations of state of spinel-structured fayalite and the stishovite + wustite assemblage.