T11E-1319 0800h
D$''$ anisotropy from differential {\it S-ScS} splitting
The boundary layers of the Earth have been shown to have the most significant seismic anisotropy. Anisotropy has been observed in the upper mantle, near the 660 discontinuity and in the lowermost mantle. Anisotropy in D$''$ has been variously attributed to the shape-preferred orientation of subducted materials (such as basaltic melt) and the lattice preferred orientation of MgSiO$_3$ perovskite and MgO magnesiow\"{u}stite, aligned by dynamic processes in the region. Observations of anisotropy in the lowermost mantle have been made from normal-mode oscillation data, but evidence comes predominantly from body-wave studies measuring the splitting in {\it S}, {\it ScS}, {\it Sdiff} and {\it SKS/SKKS} phases. A major problem in such studies is accurately accounting for anisotropy in the source region and in the upper mantle below the station. One method by which this can be circumvented is by using {\it S-ScS} differential splitting (in the distance range 65--85 degrees), that is, using the splitting in the {\it S}-phase of a seismogram as a pre-analysis correction for the {\it ScS}-phase. We study the feasibility of using this method to determine shear-wave splitting in the D$''$ layer. One problem we highlight is the complication associated with the phase-shift in the {\it ScS} across part of the distance range studied. This introduces complex particle motion which we show to contaminate the analysis of shear-wave splitting. We develop a correction for this using an estimate of the complex reflection coefficients at the CMB, and show that after this correction we can recover accurate shear-wave splitting parameters in synthetics. We also employ an automated shear-wave splitting analysis algorithm which uses cluster analysis to improve the robustness of the results. We apply this method to earthquakes from the Western Pacific region recorded at Canadian National Seismic Network stations; these sample D$''$ beneath the north Pacific. The residual splitting in {\it ScS}, which we attribute to D$''$, shows lag times between 0.8--3.25\,s. Given the near-horizontal raypath of {\it ScS} in D$''$, we interpret the recovered fast directions as the orientation of the fast shear wave in the vertical-transverse plane. The largest population of results shows a south-easterly dipping fast axis. The level of complexity shown in the results and the possible contributions to D$''$ anisotropy from lower mantle minerals, melt and subducted materials suggest that the current paradigm for simply resolving transverse isotropy is, at least in places, inadequate.
T11E-1320 0800h
Imaging Lower Mantle Structure Beneath the Central Pacific by Stacking S Wave Data
The lowermost $\sim$250 km of Earth's mantle is being revealed as a region of surprisingly complex structure and inferred dynamics. Recent waveform stacking efforts indicate the presence of rapid P and S velocity increases about 230 km above the core-mantle boundary (CMB) beneath the central Pacific. Tangential component S-wave (SH) data from the Tonga-Fiji region recorded by broadband networks in western North America are used to map small-scale variability in the topography and magnitude of this central Pacific D" discontinuity, as well as small-scale structure within the D" layer. Our approach involves deconvolving source wavelets, binning and stacking localized subsets of data, and modeling the stacks with synthetics to provide localized spatial resolution of structure in our study region. The D" discontinuity under the central Pacific is found to vary in depth from 2577 km to 2730 km, with shear velocity increases from 0.3% to 2.1%. If the reflector is continuous, it has topography as dramatic as 100 km over scale lengths of $\sim$150 km. We also determine the shear velocity structure in an ultra-low velocity zone (ULVZ) present just above the CMB beneath the central Pacific. ULVZ's have been detected in numerous regions using high frequency P waves or SPdiffKS phases. Results of such studies indicate P wave velocity reductions of 4% to10% in the ULVZ, but the S wave velocity reduction is poorly resolved due to tradeoffs between the P and S velocities and density in the reflection coefficient at the top of the ULVZ. The ULVZ is generally regarded as a region of partial melt, based on estimates of a 3:1 S velocity to P velocity reduction, but the ratio of velocity decrements remains a critical, and poorly constrained, parameter. Direct measurement of S velocity in the ULVZ, independent of P velocity, is needed to better constrain ULVZ properties. We examine our data set for precursors generated by ULVZ structure by aligning the data on ScS and stacking, finding spatially variable precursors with opposite polarity to ScS. Modeling the ScS precursors indicates shear wave velocity reductions of 0% to 22% in the ULVZ across our study region, with the estimates trading off with any assumed density increase. Evidence for abrupt velocity reductions at shallower depth within the D" layer is also found. The overall shear velocity structure under this central Pacific region will be presented.
T11E-1321 0800h
High-resolution investigation of lowermost mantle anisotropy beneath the Central Pacific
Analysis of lowermost mantle anisotropy is a crucial step in fully characterizing the deep mantle. While the majority of studies of shear-wave splitting in D'' report behavior compatible with vertical transverse isotropy (VTI), some observations indicate more complicated structure. Russell et al. (1999) propose that D'' beneath the Central Pacific exhibits azimuthal anisotropy with symmetry orientation relative to northwest trending ray paths changing from transverse to parallel over a few hundred kilometers. Such a rapid change in symmetry orientation warrants further investigation. Over the past few years a wealth of new data have become available, allowing us to further investigate this region with higher spatial resolution and to characterize the type, lateral variability, and depth extent of lowermost mantle anisotropy. We utilize a large number of Tonga-Fiji events recorded by California stations to assess shear-wave splitting of core reflections [ScS]. Waveforms are deconvolved by average source wavelets estimated by stacking transverse ScS arrivals for each event in order to improve the temporal resolution of arrivals and to give uniformity in signals between events. Corrections are also applied for lithospheric anisotropy beneath the receivers. Shear-wave splits [ScSH-ScSV] and differential travel times [ScS-SDATA - ScS-SPREM] are calculated for over 390 records from 37 events, a nearly five-fold increase in data from previous work in the region. There is trend of increasing ScS travel time delays from southwest to the northeast, suggesting that deep mantle shear velocity decreases in this direction. While ScS splitting is pervasive, it is not as simply organized as suggested in prior work, with substantial intermingling of ScSH or ScSV advances. Evaluation of focal mechanism effects, near-source anisotropy, and azimuthal anisotropy in D'' will be presented, and the relationship, if any, to volumetric shear velocity variations will be assessed.
T11E-1322 0800h
ScS-S Differential Attenuation from Spectral Ratios and Instantaneous Frequencies via the Hilbert and Continuous Wavelet Transform
Measurement of differential t* between ScS and S phases leads to a better understanding of attenuation in the D'' layer. Here, we obtain differential t* between ScS and S from instantaneous frequencies via the standard Hilbert transform and the complex continuous wavelet transform. The waveform is attenuated until the instantaneous frequency at the S envelope peak matches that of the unattenuated ScS envelope peak. Differential t* between ScS and S is then the t* operator required to match the instantaneous frequencies. The instantaneous frequency at the peak of a seismic phase is equal to the average Fourier spectral frequency of the phase weighted by its amplitude. We also compute differential t* between ScS and S from spectral ratios by windowing the S and ScS phases and calculating the ratio of the log magnitude of their spectra. The slope of this ratio is then the differential t* between ScS and S. All methods produce consistent results for high signal-to-noise ratio (SNR) synthetic waveforms with S and ScS phases well separated in time. The instantaneous frequency method gives more stable results than the spectral ratio method for low SNR waveforms or with diminished differential time between ScS and S arrivals. The discrepancy between the methods at low SNR can be mitigated somewhat by careful choice of the bandwidth used in the spectral ratio calculation. The instantaneous frequency matching method is applied to broadband seismic data sampling the deep mantle beneath Central America. Differential t* between ScS and S is calculated from the data permitting the calculation of a regional model of shear wave attenuation in the lower mantle.
T11E-1323 0800h
Evidence for a Sharp Lateral Boundary at the Southern Border of the Pacific Superplume
Recently, sharp lateral transitions in the velocity structure have been reported at the borders of the lowermost mantle low velocity "superplumes". Most of the findings are associated with the African superplume (Ni et al, 2002), although there is also some evidence in the east-central Pacific (Br\'{e}ger and Romanowicz, 1998). We here report that a sharp lateral boundary exists at the southern edge of the Pacific superplume. The set of SHdiff waveforms, which graze the South Pacific, have similar features to those observed previously at the southeastern edge of the African superplume. Both waveform sets show a rapid shift of the arrival time with respect to the azimuth as previously reported in the case of the African superplume. In addition, we report that both waveform sets show a secondary pulse that follows the direct Sdiff phase. The coupled mode/spectral element method (Capdeville et al, 2002), which can handle strong lateral variations of the velocity in D", is used to construct synthetic waveforms. The computationally heavy spectral elements are only used in the target region, which is the bottom 370km of the mantle in this study (Capdeville et al, 2003). This allows one to extend computations to much higher frequencies than if spectral elements are used for the whole earth. The postcursors can be explained by a simple structure in the D'' region with a sharp quasi vertical boundary aligned almost parallel to the ray path. The existence of these pulses indicates that modelling of heterogeneity outside of the great circle path can help constrain the 3D structure at the base of the mantle. The similarity of the two observed SHdiff waveform sets at relatively high frequencies indicates that the low velocity regions in the lower mantle under Pacific and Africa, observed as the strong degree-2 pattern in shear velocity tomographic models, have a similar nature also at finer scales.
T11E-1324 0800h
Periodic Folding of Viscous Sheets: A Model for Subducted Slabs at the CMB
A sheet of viscous fluid falling from a sufficient height onto a surface undergoes regular periodic folding. This instability is easily observed in the home kitchen with cake batter or honey, and may occur in the earth when subducted lithosphere impinges on the core-mantle boundary. Using a numerical model for a 2-D thin sheet that deforms by combined bending and stretching with negligible inertia, I determine how the folding frequency depends on the height of fall $L$, the sheet's initial thickness $H_0$, the volume flux per unit length $q$, and the fluid's viscosity $\mu$ and anomalous density $\delta\rho$. Two distinct folding modes are possible. When buoyancy is weak, ``viscous'' folding occurs with a frequency proportional to $f_V = q/L H_1$, where $H_1$ ($\leq H_0$) is the thickness of the portion of the sheet that folds. When buoyancy is strong, ``gravitational'' folding occurs with a frequency proportional to $f_G = (\delta\rho g q^3/\mu H_1^5)^{1/4}$. The bifurcation from viscous to gravitational folding occurs sharply at a critical value ($= 3.9$) of the parameter $f_G/f_V$ that measures the importance of buoyancy. In both modes, the instability generates a multilayered ``sandwich'' of strongly folded sheet material and intercalated ambient fluid. In the earth, such a structure is likely to be associated with complex seismic anisotropy. I will also present results of a numerical study of the closely related phenomenon of 3-D coiling of a viscous filament. While not directly relevant to subduction, this phenomenon is important in many engineering contexts and has been observed to occur in falling lava streams. Coiling can occur in viscous and gravitational modes analogous to those of 2-D folding, and also in an ``inertial'' mode in which viscous forces are balanced by rotational inertia. Coiling is multivalued in the gravitational-inertial transition regime, where three distinct frequencies are possible for the same values of the flow rate and fall height.
T11E-1325 0800h
Geodynamic Implications of Convection-Related Surface Observables: The Role of Lateral Variations in Mantle Rheology
Over the past decade numerous analyses of convection-related observables, such as geoid or gravity anomalies and dynamic surface topography, have been carried out in the context of tomography-based mantle flow models in an effort to better understand the 3-D density and thermo-chemical structure of the mantle as well as the rheology of this region. With few exceptions, most of the studies have been conducted in the framework of a viscous flow theory which assumes that the mantle rheology may be represented in terms of an effective viscosity which varies with depth only. An ongoing effort to simultaneously invert both shorter time-scale glacial isostatic adjustment data, and much longer time-scale convection data, has recently yielded a new series of radial viscosity profiles which provide good fits to both sets of data. The reconciliation of regional and global constraints on mantle rheology, spanning such wide time spans, suggests that we are able to effectively resolve the mean or horizontally averaged value of viscosity as a function of depth. However, apart from some limited investigations, we have yet to carry out a detailed assessment of the impact of lateral variations in viscosity on our inferences of viscosity and deep mantle thermo-chemical structure. To this end, we present a new series of calculations which explicitly incorporate lateral viscosity variations in the viscous flow theory using both a semi-analytical variational approach and a purely numerical finite-element method. We will report here on results pertaining to the "primary" signatures of convection, such as the mantle flow and strain rate fields, as well as "secondary" or indirect signatures such as the dynamic topography and gravity fields. We also consider the impact of this rheological complexity on studies connecting backward-convection flow simulations with the geological record.
T11E-1326 0800h
Deformation of a Partially Molten D$''$ Layer by Small-Scale Convection and the Resulting Seismic Anisotropy and Ultralow Velocity
Partially molten regions in the D$''$ layer have been demonstrated to exist by the discovery of the ultralow velocity zone (ULVZ). Following Solomatov and Moresi(2002), we regard the D$''$ layer as a thermal boundary layer within which small-scale stagnant-lid convection occurs. If the small-scale convection deforms partially molten regions, they would profoundly affect seismic structures including anisotropies. We therefore calculate the deformation history of partially molten regions at the base of the D$''$ layer which is heated from below, using a 2-D model with a strongly temperature dependent viscosity. The melt fraction is assumed to be proportional to temperature above solidus. An initially isotropic partial melt is strongly deformed by the viscous stress caused by the thermal instability, and becomes anisotropic by shape preferred orientation (SPO) of melt inclusions. The aspect ratio of the melt is of the order of $10^{1-2}$ at the base of the plume and becomes as large as $10^{3-4}$ in the plume head. We calculate the effective elastic constants for such anisotropic media which contain deformed partial melt, and obtain the seismic velocity for a horizontal ray path. We find that the horizontally averaged velocity profile consists of three layers corresponding to the base, conduit and head of a rising plume. An ultralow velocity zone (ULVZ) at the base and a negative shear wave gradient in the convective region form as soon as the thermal instability occurs. The deformation and alignment of the melt, rather than the melting itself, is primarily responsible for reducing the seismic velocity, and the lowermost ULVZ becomes strongly anisotropic (SH $>$ SV). On the other hand, in the conduit, the anisotropy is of SV $>$ SH type because of vertical alignment. In the plume head, the anisotropy is SH $>$ SV type with a magnitude of about 2%. We also show how shear wave anisotropy may be used to infer the temporal evolution of the instability at the D$''$ layer.
T11E-1327 0800h
Thermodynamics Of The Lowermost Mantle : Example Of $SiO_2-MgO$ System
Seismic observations of the region above the core mantle boundary (CMB), including Ultra-Low Velocity Zones (ULVZ) and D'' discontinuities, and recent mineral physics data suggest the bottom of the mantle may be chemically heterogeneous, possibly, on account of the occurrence of the recently discovered, $CaIrO_3$ structure. Moreover, the large decrease in shear wave velocity in selected areas (ULVZ) in the lowermost 40 km of the mantle indicates the likely presence of partially molten material. To provide a complete petrologic and mineral physics model of this region, it is important to define both constituent material properties, and phase equilibria including melting behavior. We expect to describe models that encompass the entire $MgO-FeO-CaO-Al_2O_3-SiO_2$ system and to define complete mineral physics equations of state, and employ thermochemical data to obtain complete phase diagrams. Our mineral physics-based seismic models are to be obtained using global inversion methods. Initial model results are reported for the $SiO_2-MgO$ system. For the $SiO_2$ system, we modeled the high-pressure phase diagram over the $TiO_2$-, $CaCl_2$-, $\alpha$$PbO_2$- structure and liquid phase stability range. The high-pressure liquid $SiO_2$ has an EOS with: $\rho$ = 4.05 $g/cm^3$, $K$ = 180 GPa, $K'$ = 6, $\gamma$ = 1.15, $q$ = 0.8, $Cv$ = 1.55 J/K/g. The melting of $CaCl_2$ structure occurs at $\sim$5000 K at 135 GPa. Also the $\alpha$$PbO_2$-structured $SiO_2$ EOS is described by : $\rho$ = 4.18 $g/cm^3$, $K$ = 270 GPa, $K'$ = 3.6, $\gamma$ = 1.35, $q$ = 2.6, $Cvm$ = 1.215 J/K/g, $\theta_0$ = 1222 K. Examination of the diamond anvil melting data for MgO yields the liquid EOS and a model melting curve. The melting point is $\sim$5200 K at 135 GPa. We also calculate the melting behavior of binary $SiO_2-MgO$ system. For MgSiO3 composition the melting point at 135 GPa is $\sim$5300 K, which combined with the above MgO data indicates melting temperatures of mantle composition at 135 GPa is $\sim$4400$\pm$400 K.
T11E-1328 0800h
EH vs. CI chondrite derived mantle: A geodynamical comparison
It is generally assumed that the Earth's bulk composition is derived from carbonaceous CI chondrites. However, arguments based on stable isotopes and redox considerations favor another type of material from which the Earth could be derived: the enstatite EH chondrites. The latter implies substantial heterogeneities either in minor and major elements within the mantle which is strongly suggested by seismological observations and further reinforced by noble gas constraints. Here we investigate the geodynamical consequences of CI and EH derived Earth's mantle compositions. Using numerical simulations in cylindrical geometry with an appropriate scaling to approximate the spherical Earth, we compare the evolution of mantles derived from EH and from CI chondrites from 4.5 Gyr B.P. to present day. For both EH and CI models the upper part of the mantle (from which continental crust is extracted) has a pyrolitic-like composition, therefore the differences between the two models are shifted to the lower part of the mantle mantle, implying substantial differences in heat producing elements concentrations and Si, Fe and Mg content. Both models consider the extraction of continental crust, heterogeneous internal heating related to local concentrations of heat producing elements, and the presence of chemically denser material in the lowermost mantle, as suggested by tomographic studies. The thermal and chemical evolution of these two models is therefore compared and the implications on present day mantle heterogeneity in both major and trace elements as well as the consequences on seismological observables are investigated.
T11E-1329 0800h
Seismic anisotropy in the lower mantle: a comparison of waveform splitting of SKS and SKKS
Splittings of the {\it SKS/SKKS} waves are caused by seismic anisotropy along the receiver-side ray path from the core mantle boundary to the receiver. Yet, the splittings are usually assumed to be originated from the upper mantle anisotropy formed by tectonic processes, which is supported by both seismological and laboratory observations. Recent laboratory study, however, shows that dislocation creep occurs in $MgSiO_3$ perovskite at pressure-temperature conditions of the uppermost lower mantle. Seismic anisotropy is indeed observed at middle and lower mantle depths in some regions. In this study, we carried out a global investigation on seismic anisotropy in the lower mantle by comparing the waveform splitting of {\it SKS} and {\it SKKS} found at the same seismogram. The two shear waves have ray paths very similar in the upper mantle but different in the lower mantle. Any difference in splitting between the two indicates the presence of seismic anisotropy or dipping structures in the lower mantle. Since {\it SKS} and {\it SKKS} are indistinguishably used in measuring the upper mantle anisotropy to increase azimuth coverage, a systematic comparison of the splitting behavior of the two is also essential to the justification of the usage. We collected a total of 104 {\it SKS+SKKS} waveform data recorded at 76 stations. We chose the data based on the following criteria: (1) both phases are clear shown on the radial component. A signal-to-noise ratio (SNR) $>$ 3 is used for both $SKS_{R}$ and $SKKS_{R}$. (2) Amplitude of $SKS_{R}$ and $SKKS_{R}$ are comparable with each other. We limited the amplitude ratio $SKKS_{R}$/$SKS_{R}$ to the range of 0.5-2. This ensure us to obtain same precisions of the two in splitting measurements; (3) there are no other phases, such as the depth phase of {\it SKS}, arrive at the time window of {\it SKKS}. The {\it SKS+SKKS} data are first matched with a single anisotropic model (2 parameters; the fast polarization direction \phai and delay time $\delta{t}$) and two independent anisotropic models (4 parameters). We then applied the F-test to examine whether the 4-parameter models are really better than the 2-parameter ones in terms of error improvement. We found that the data from most of the stations can be explained by the simple 2-parameter models. While this observation provides the compelling evidence that vast part of the lower mantle below $\sim$1000 km (including the D" region) is transverse isotropic in most regions, it is still arguable that the uppermost lower mantle and the transition may partly contribute the {\it SKS/SKKS} splittings. We also found that the 4-parameter models provide a better fitting to the {\it SKS} and {\it SKKS} splitting at 8 stations, suggesting the existence of transverse anisotropy or anomalous dipping structures in some part of the lower mantle.
T11E-1330 0800h
Application of the Yin-Yang grid to three-dimensional spherical shell convection at infinite Prandtl number
We have developed a new numerical simulation code to solve the thermal convection of a Boussinesq-approximation fluid with infinite Prandtl number and spatially variable viscosity using ``Yin-Yang'' grid [Yoshida and Kageyama, 2004]. The Yin-Yang grid, which has been recently proposed by Kageyama and Sato [2004], is composed of two component grids that have exactly the same shape and size. A component grid of the Yin-Yang grid is actually a low-latitude part of the latitude-longitude grid on spherical polar coordinates. The Yin-Yang grid is suitable to solve the mantle convection problems because it automatically avoids the pole problems, i.e., both the coordinate singularity and grid convergence, that are inevitable in the latitude-longitude grid. The non-dimensional equations of mass, momentum and energy conservation governing the thermal convection are solved by the finite difference discretization with second-order accuracy. Using the Yin-Yang grid, we simultaneously solve these equations for each component grid (i.e., ``Yin grid'' and ``Yang grid''). We use the collocated grid method; all the primitive variables, velocity, pressure, and temperature, are defined on the same grid points. The SIMPLER method is applied to solve velocity and pressure. The Crank-Nicolson method is used in the energy equation for the time stepping. The horizontal boundary values of each component grid are determined by linear interpolation of another component grid. We performed benchmark tests with published numerical codes [e.g., Richards et al., 2001] and confirmed the validity of our code. When $Ra = 10^5$, the convection patterns become weakly time-dependent; the geometrical ``cubic'' symmetry at $Ra = 10^4$ is broken. The corresponding case in which a finite volume scheme on the latitude-longitude grid is used shows a symmetric pattern about equator and appears to remain in a steady state. These observations suggest that convections are numerically affected by pole problems in the latitude-longitude grid. On the other hand, the pole problems are removed in our code by making use of Yin-Yang grid.
T11E-1331 0800h
Thermal convection in a 3D spherical shell with strongly temperature and pressure dependent viscosity
The style of convection in planetary mantles is presumably dominated by the strong dependence of the viscosity of the mantle material on temperature and pressure. While several efforts have been undertaken in cartesian geometry to investigate convection in media with strong temperature dependent viscosity, spherical models are still in their infancy and still limited to modest parameters. Spectral approaches are usually employed for spherical convection models which do not allow to take into account lateral variations, like temperature dependent viscosity. We have developed a scheme, based on a finite volume discretization, to treat convection in a spherical shell with strong temperature dependent viscosity. Our approach has been particularly tailored to run efficiently on parallel computers. The spherical shell is topologically divided into six cubes. The equations are formulated in primitive variables, and are treated in the cartesian cubes. In order to ensure mass conservation a SIMPLER pressure correction procedure is applied and to handle strong viscosity variations up to $\Delta \eta=10^6$ and high Rayleigh-numbers up to $Ra=10^8$ the pressure correction algorithm is combined with a pressure weighted interpolation method to satisfy the incompressibility condition and to avoid oscillations. We study thermal convection in a basal and mixed-mode heated shell with stress free and isothermal boundary conditions, as a function of the Rayleigh-number and viscosity contrast. Besides the temperature dependence we have further explored the effects of pressure on the viscosity. As a general result we observe the existence of three regimes (mobile, sluggish and stagnant lid), characterized by the type of surface motion. Laterally averaged depth-profiles of velocity, temperature and viscosity exhibit significant deviations from the isoviscous case. As compared to cartesian geometries, convection in a spherical shell possesses strong memory for the initial state. At strong temperature dependent viscosity ($\Delta \eta=10^4-10^5$) typically a few upwelling plume structures develop. The large scale structure of the plume stays intact over a long time while the plume geometry varies on a smaller scale. The downflows are generally organized in two-dimensional sheetlike flows. Additional pressure dependence strongly influences the dynamics even if the magnitude of pressure variation is relatively small. For an appropriate combination of pressure- and temperature-dependence we observe a well developed high-viscosity zone in the lower mantle.
T11E-1332 0800h
Structure Of The Ultra-low Velocity Zone In The Lowermost Mantle Derived From CMB Reflected Phases
Analyses of reflected seismic waves from the core-mantle boundary (CMB) enable us to constrain the existence and properties of the ultra-low velocity zone (ULVZ) at the base of the mantle. Here we investigate the fine scale heterogeneity of the lowermost mantle beneath the western Pacific region using precursors and postcursors to PcP and ScP from deep earthquakes in the Philippine-Indonesian region recorded by the Hi-net seismic network in Japan and seismic array stations of the International Monitoring System in the western Pacific region. Clear arrivals of postcursor to ScP are detected for several contiguous events, which have bounce points beneath the Philippine Islands, while no significant precursors to PcP and ScP are detected in the same region. Our one-dimensional forward waveform modelings suggest that these postcursors are associated with an ultra-low shear-wave velocity layer in the lowermost mantle. The size of this ULVZ is about 500 km with velocity reductions of 7% for P-wave and 25% for S-wave, and with no or slight density increase (less than 4%). Garnero (2004) suggests the existence of ultra-low velocity patches with a spatial extent ranging from less than 10 km to hundreds of kilometers beneath the Central American subduction region. The properties of the ULVZ we found in this study are consistent with his result. The physical and/or chemical properties of this anomalous layer could be different from that of the ULVZ under South Pacific because clear precursors to PcP and ScP were reported there by several previous studies. The present results will be compared with other ULVZ studies conducted in various regions, focusing on the wavelength of the heterogeneity, seismic properties and physical constraints.
T11E-1333 0800h
Insights Into the D$''$ Region From Analysis of a Thin Dense Layer Beneath a Convecting Cell
In this study we set up a simple model of the earth's D$''$ layer as a thin dense layer which is compositionally distinct from the lower mantle. The dominant external mechanism governing the flow within the layer and displacement of its upper boundary is assumed to be tractions acting on the upper surface of D$''$ resulting from the convecting mantle above. The Navier-Stokes equation is solved analytically and the resulting solutions analysed. Topography on the layer boundaries is predicted by balancing it against dynamic flow stress. A 2D finite element code is used, not only to confirm the results of the analysis but also to allow investigation of solutions with large boundary deflection. The nature of boundary topography depends on the magnitude of the driving tractions and the density variation within the layer. If we impose a variation such that the layer is most dense beneath areas of mantle downwelling and decreases to a minimum beneath areas of mantle upwelling then the upper boundary of D$''$ builds up into a cusp-like peak beneath the upwelling mantle. The size of this peak can potentially be several times greater than the layer depth. If, however, opposite density variations are imposed we can instead observe solutions where the layer is completely swept away beneath areas of mantle downwelling leaving steep-sided `islands' of dense material. The magnitude of the upper boundary driving tractions compared to the magnitude of density variations within the layer is a crucial parameter in determining the nature of flow in, and consequently boundary topography of, the layer. The deflection of the core-mantle boundary is small compared with that of the top of D$''$, but a change in sign in the ratio of these deflections is observed as the magnitude of the driving tractions changes relative to the magnitude of density variations. Seismic data of core-mantle boundary topography, D$''$ topography and lower mantle wave speed are compared to the predictions of the model and used to constrain model parameters.
T11E-1334 0800h
On the Survival of a Heterogeneous Deep Mantle Reservoir: Constraints from Evolutionary Numerical Mantle Convection Models
The concept of a compositional heterogenous reservoir, in the deep lower mantle, has become increasingly popular in recent years, due to indications from seismology, geochemistry and heat budget calculations. In this study, the dynamic feasibility of this concept was investigated, using numerical modeling techniques. We applied 2-D cartesian and cylindrical geometry and a time window of 4.5 Gyr in our thermo-chemical mantle convection models. The models include the effects of the endothermic phase transition at 660 km depth in the extended Boussinesq approximation. We use finite elements for the solution of the convection equations. Composition dependence of several physical parameters is represented by Lagrangian tracer particles in a particle-in-cell implementation. We studied the dynamic evolution of a compositionally distinct layer at the bottom 20 percent of the mantle volume, with a positive density contrast in the order of 1% and time dependent internal heating, increased with respect to the background mantle. The results show a clearly bounded stability domain in a domain-diagram with density contrast and internal heating rate. Preliminary results suggest that the layer stability is smaller for models with cylindrical geometry due to the different surface/volume ratios for the deep lower mantle. We also show that the parameters of the endothermic phase transition have a large impact on the evolution of the compositional layering. Vigorous mantle convection, once it sets in, controlled by the phase transition, produces cold downwellings which eventually destroy the layering with (sub)critical density contrasts.