DI51B-01 INVITED
Insights on Compositional and Thermal Structure of the Earth's Upper Mantle Using Mineral Physics and Seismic Data.
Temperature and composition of the Earth's mantle are key factors to understand the dynamics of Earth's interior. Interdisciplinary studies that profit of the parallel advances on studies of material properties at high pressure (and temperature) and seismology are needed to improve the knowledge of these fundamental physical parameters. We present published and preliminary results from different studies based on such an approach, focusing on global average constraints. First, we show how well an adiabatic pyrolite model fits global seismic data by accounting for the full range of uncertainties on elastic and anelastic properties of mantle minerals. We found a small set of "best-fitting" models, all seismically very similar, that can be used as starting models for seismic inversion. Second, we show the results of a global seismic waveform inversion that adopts one of these physical models instead of commonly used seismic reference models (e.g. PREM) as starting model. Absolute shear velocities and gradients with depth are well constrained by long period seismic waveforms throughout the upper mantle, except near discontinuities. We found a high velocity gradient between 250 and 350km that cannot be explained with any realistic thermal structure, assuming than composition does not change with depth. A gradual enrichment in garnet with depth may help. Third, we use SS and PP precursors in order to provide a better characterization of the seismic structure around the 400 and 660 km discontinuities. Relative amplitudes between the discontinuity reflections and the reference phases SS and PP are evaluated.
DI51B-02 INVITED
Phase relations of cold subducted slab: constraints on the slab temperature and on the chemical heterogeneity in the lower mantle
We will talk about two topics of phase relations in cold subducted slabs. They are stability relations of high- pressure hydrous phases in a peridotite layer and of Ca-perovskite in a mid-oceanic ridge basaltic (MORB) crust on the basis of the thermodynamic calculations and high-pressure experiments. In the hydrous peridotite system MgO-SiO2-H2O, seven high-pressure hydrous phases appear after serpentine dehydration (~150-km depth). These hydrous phases carry water to the deep mantle condition. At the transition zone, a series of dehydration reactions will occur if the slab temperature is above 1300K. In the case of lower temperature, high-P hydrous phases will further carry water into the deep lower mantle. At about 1300-km depth, hydrous phase D will transfer water to high-pressure ice if the temperature is lower than 1300K. After the ice formation, no fluid-forming reaction may occur in the slab except at the core-mantle boundary where the temperature increase is expected. The depth distribution of dehydration reactions in the slab is well consistent with that of seismic event in the subduction zones if the appropriate temperature is assumed. This suggests that the deep-focus seismicity is induced by the dehydration reactions in the slab, and further suggests that the seismic event is an indicator of slab temperature and water transport into the deep mantle. CaSiO3-perovskite undergoes a structural phase transition from tetragonal to cubic symmetry at about 540 K, almost independent of pressure. This transition temperature significantly increases with increasing Al2O3 contents in Ca-perovskite. Unlike in multicomponent peridotite systems, in a mid-oceanic ridge basalt system, Ca-perovskite contains significant amounts of Al2O3 up to about 3 wt% where the structural phase transition may occur at about 1200 K. Seismological studies reported numbers of seismic scatters, reflectors, and low-V layers at a wide depth range (1100-1850-km depth) beneath Mariana subduction zone (e.g., Kaneshima and Helffrich, 2003). A key feature of these observations is a large drop in S-wave velocity without significant anomaly in P-wave velocity. The structural phase transition in Ca-perovskite may be ferroelastic type and therefore easily explain these abnormal seismic features. If these observations are indeed due to the structural phase transition in Ca-perovskite, the temperature of MORB crust of the slab at those depths should be around 1200K. We will further discuss the slab temperature and the chemical heterogeneity in the lower mantle caused by the subduction of the oceanic plate. References, Kaneshima and Helffrich, 2003. JGR, 108, NO. B5, 2272, doi:10.1029/2001JB001596.
DI51B-03 INVITED
Upper Mantle Discontinuity Topography from Thermal and Chemical Heterogeneity
We investigate the depth of upper mantle seismic discontinuities utilizing high-resolution stacks of underside- reflected, horizontally-polarized precursors to the seismic phase SS. Bootstrap-derived histograms of discontinuity depths are used to retrieve detailed topographic information from stacks of a high-quality broadband dataset. This dataset consists of over 20,000 seismograms that sample beneath South America, the Hawaiian hotspot, and the surrounding Pacific and Atlantic Oceans. Beneath South America, our dataset reveals the 660-km discontinuity is 20 km deeper on the down-dip side of subduction, in agreement with cold material descending to this boundary. However, there is also a 10-15 km depression in the overlying 410-km discontinuity that is difficult to explain by thermal effects alone. In the same region, multiple discontinuities are detected on the 410-km boundary where it intersects the slab. We also find the 660-km boundary to be unexpectedly shallow beneath the mid-Atlantic and eastern Pacific ridges, consistent with warmer material rising through these depths. The 660-km discontinuity is shallow beneath the location of the Hawaiian hotspot, supporting a hot mantle plume passing through this boundary. We explore different mechanisms connecting chemical heterogeneity to our observed topography, including the transport and isolation of H2O at the 410-km discontinuity and/or the extraction of Fe from wedge material later entrained into the mantle transition zone. An overlying partial melt layer can explain multiple reflectors near 410-km depth, and phase changes in the ilmenite system can explain our observations of multiple reflectors below 660-km depth. These findings strongly support the idea that the origin of upper mantle heterogeneity has both chemical and thermal contributions, and is associated with deeply rooted tectonic processes.
DI51B-04
Calcium perovskite exsolution from majorite garnet and splitting of the 520 km seismic discontinuity: insights into mantle heterogeneity.
Seismic observations have identified a weak seismic discontinuity in the mantle at approximately 520 km depth which has been attributed to the isochemical phase transformation of (Mg,Fe)2SiO4 wadsleyite to the higher pressure polymorph ringwoodite. Recent seismic observations have shown that in some regions of the mantle two separate discontinuities can be observed within a similar mantle depth range of 500 km to 560 km. A possible cause of the second discontinuity is the exsolution of calcium perovskite from majorite garnet at mid transition zone conditions. An important question to address, therefore, is why this doubling of the discontinuity is only observed in certain regions of the mantle. We performed multianvil experiments to investigate the depth interval of the calcium perovskite forming reaction as a function of temperature and garnet majorite content. Results show the exsolution to be a non-linear function of pressure resulting in formation of a significant proportion of calcium perovskite over a narrow depth interval. The calculated sound velocity impedance contrast for this reaction in a fertile peridotite is comparable to that of the wadsleyite to ringwoodite transition and is in good agreement with seismic observations. The effective Clapyeron slope of the reaction indicates that at high temperatures its depth would converge with that of the wadsleyite to ringwoodite transition, potentially explaining observations of a single discontinuity. However, this depth is much deeper than 520 km, where a single discontinuity is generally observed. Temperature variation alone cannot, therefore, explain the variability in 520 splitting, which is more likely to result from variability in the Ca content of the mantle, either due to varying mantle fertility or due to a varying proportion of recycled oceanic crust. The split in the 520 km discontinuity is therefore a sensitive indicator of mantle heterogeneity
DI51B-05
Seismic observations of mantle discontinuities, and their mineral physical interpretation
We study transition zone discontinuities using a new large collection of stacked receiver functions, in addition to our previous data sets of PP and SS precursors. Such stacks provide a means of investigating hypotheses about the mineral physical nature of discontinuities and constrain the thermal and compositional structure of the Earth's mantle. The 410-km discontinuity is observed in all data types and is characterised by a simple single peak. The 660-km discontinuity, however, shows a much more complex structure. PP precursors only show the 660-km discontinuity in certain regions, while it is always seen in SS precursors and receiver functions. Our receiver function observations reveal a very complicated global structure with single and double reflections ranging in depth from 640 to 720 km. A weaker and possibly non-global discontinuity at approximately 520 km depth is also present in all data types. Confirming previous work on SS precursors, our receiver functions show that this 520-km discontinuity is split in certain regions, while in other regions one discontinuity is observed. So, the different data types are agreeing on the type of complexity observed at each of the transition zone discontinuities. These observations are explained by the presence of multiple phase transitions on a global scale in the transition zone. The combined phase transitions of olivine and garnet in a pyrolite mantle explain the seismic observations of double peaks (or splitting) at the 520 and 660 km discontinuities, while the single olivine phase transition at 410 km explains the simple seismic observations in this depth range. Thus, we can reconcile the complex seismic observations of the transition zone with mineral physical phase transitions. Additional regional variations in the seismic observations imply the need for compositional heterogeneity in minor elements such as water, Fe, Al and Ca. Our three data types also support evidende for reflections from additional upper and lower mantle discontinuities, the most consistent ones being at approximately 220, 800 and 1150 km depth. Discontinuities at 1100-1220 km have been proposed before by some regional studies and would be consistent with tomographic models, particularly in subduction zone areas.
DI51B-06
Mapping thermal and compositional variations in the mantle transition zone: Uncertainties due to the resolution of seismic observations
Seismic observations of the transition-zone discontinuities along with volumetric velocity variations in the transition zone are diagnostic of the thermal and compositional states of the mantle and associated dynamics. As high-pressure experiments and theoretical calculations provide more and more accurate constraints on the mantle phase relations and the elastic coefficients of likely minerals as a function of pressure, temperature, and composition, attempts have been made in recent years to quantitatively estimate thermal and compositional variations in the transition zone by solving the equations in which the discontinuity depths and volumetric velocity perturbations are expressed as a function of mantle temperature and composition. By integrating the constraints on the phase transitions and surrounding mantle, these studies were able to separate thermal anomalies and variations in water content in the transition zone. The estimates are, however, highly sensitive to the thermal and compositional dependence of the seismic parameters as well as the uncertainties of tomographic models and discontinuity depths, whose effects on mapping thermal and compositional variations in the transition zone are yet to be carefully assessed. I will present a re-analysis of the thermal and compositional states of the mantle transition zone beneath southern Africa. A finite-frequency tomography method, which takes into account of wavefront healing of realistic seismic waves, is used to obtain more robust P and S velocity models of the transition zone than in previous studies. The results will be compared to those based on ray theory and tested by comparing observed waveforms and simulations of full wave propagation in the 3D models. The finite-frequency velocity models will be used to re-calculate the discontinuity topography. The propagation of the uncertainties of the tomographic models and discontinuity depths to the estimates of the thermal and compositional variations will be discussed.
DI51B-07
Elasticity of Hydrous Olivine Polymorphs: Implications for Seismic Structure of the Transition Zone
The presence of water in the upper mantle and transition zone has the potential to explain various phenomena such as shear velocity anomalies or uplift and broadening of the 410-km discontinuity. The presence of H2O in the transition zone has also been frequently invoked to reconcile laboratory elasticity data on olivine polymorphs with seismic data for the amplitude of the 410-km discontinuity (Li et al., 2001; Chambers et al., 2005). Recently, we have measured the single-crystal elastic properties of hydrous olivine (Jacobsen et al., 2006) and a suite of hydrous wadsleyites (Mao et al., 2007a) at ambient conditions and one hydrous wadsleyite composition (0.84 wt% H2O) up to 12 GPa (Mao et al., 2007b). These data provide new constraints on elastic moduli and their pressure derivatives for hydrous olivine and wadsleyite. Using this data, we first examine the effect of H2O on bulk sound velocities under transition zone conditions because anelastic effects can be neglected in this case. At 410 km depth (~13.8 GPa, along a 1400°C adiabat), the bulk sound velocity of wadsleyite with 1 wt% H2O is 3.1% lower than for dry wadsleyite. Comparison of the seismic velocity jump across the 410-km discontinuity with the measured velocity contrast between wadsleyite and olivine provides a means to estimate the olivine abundance at 410-km depth. For mantle wadsleyite with 0.1-0.2 wt% H2O (Huang et al., 2005) and using experimentally determined olivine- wadsleyite H2O partition coefficients, the olivine abundance is found to be 40%, much lower than a pyrolite model. In order for a pyrolite composition to satisfy the seismic data, 1.2 wt. % H2O is needed in wadsleyite- a value greater than its maximum solubility under these conditions. The anomalously steep seismic gradient in the transition zone has been another feature of the region that has long defied explanation. We show that the seismic gradient can be matched if there is a gradient in H2O concentration across the transition zone such that the H2O content drops, for example, from 0.3 wt% at 410 km to 0.1 wt% at 520 km dpeth. For compressional and shear wave velocities, 0.1 wt% H2O in wadsleyite would lead to 0.3% and 0.4% reductions in VP and VS, respectively, neglecting any anelasticity. If the water content of wadsleyite was instead 1.0 wt. %, then the corresponding velocity reductions would be 3.3% and 3.6%. Following the work of Karato and Jung (1998), we have implemented a preliminary model accounting for the effect of H2O on anelasticity. This model indicates that 0.1 wt% H2O in wadsleyite could be responsible for reductions in shear velocities up to 1.0%.
DI51B-08
Seismic Velocity Contrasts at the Top Surface of the Subducting Pacific Slab Down to the Deep Upper Mantle ; Implications for Water Flow Into the Mantle Transition Zone
It has recently been proposed that the mantle transition zone is a water reservoir in the Earthfs interior. Several experimental researches have shown that transition zone minerals, such as wasleyite and ringwoodite, can contain significant amounts (~several wt%) of water in the crystal structure, whereas the solubility of water in olivine is less than 0.2 wt%. Hydrous minerals above and inside the subducting slab may play an important role in carrying water into the mantle transition zone. To image seismic velocity contrasts at the upper mantle discontinuities including the subducting Pacific slab, we applied a receiver function (RF) analysis to the Hi-net tiltmeter recordings. Usually, radial RF is stacked to image seismic discontinuities. In this study, we also used transverse RF, which is effective for detecting P-to-S phase conversion at (or P-to-S converted phases from) a dipping layer, by limiting back azimuth of the teleseismic events between 120° and 180°. We combine radial RF with transverse RF in seismic imaging. We applied a band-pass (0.02-0.16 Hz) filter and CCP (common conversion point) stacking for imaging. We present a receiver function image of the subducting Pacific slab with undulations of both the 410 km and 660 km discontinuities beneath the Japanese Islands. Especially, the top surface of the Pacific slab indicated by positive RF amplitudes can be seen down to the mantle transition zone, implying that the subducting slab is faster than the mantle wedge in seismic velocity. In synthetic forward modeling, the seismic velocity contrast at the top slab surface is approximately 8-10 % at depths of 200-300 km. This could be attributed to the existence of hydrous minerals on the slab; these minerals possibly carry water into the mantle transition zone. Furthermore, in addition to the top slab surface, we could also image the uplift of the 410 km discontinuity. This means that we could successfully identify both the top slab surface and the 410 km discontinuity at depths of 300- 400 km.
DI51B-09
Self-consistent Synthetic Mantle Discontinuities From Joint Modeling of Geodynamics and Mineral Physics and Their Effects on the 3D Global Wave Field
Our current understanding of mantle structure and dynamics is to a large part based on inversion of seismic data resulting in tomographic images and on direct analysis of a wide range of seismic phases such as Pdiff, PcP, ScS SdS etc. For solving inverse problems, forward modeling is needed to obtain a synthetic dataset for a given set of model parameters. In this respect, great progress has been made over the last years in the developement of sophisticated numerical full waveform modeling tools. However, the main limitation in the application of this new class of techniques for the forward problem of seismology is the lack of accurate predictions of mantle heterogeneity that allow us to test hypotheses about Earth's mantle. Such predictive models should be based on geodynamic and mineralogical considerations and derived independently of seismological observations. Here, we demonstrate the feasibility of joining forward simulations from geodynamics, mineral physics and seismology to obtain earth-like seismograms. 3D global wave propagation is simulated for dynamically consistent thermal structures derived from 3D mantle circulation modeling (e.g. Bunge et al. 2002), for which the temperatures are converted to seismic velocities using a recently published, thermodynamically self-consistent mineral physics approach (Piazzoni et al. 2007). Assuming a certain, fixed mantle composition (e.g. pyrolite) our mineralogic modeling algorithm computes the stable phases at mantle pressures for a wide range of temperatures by system Gibbs free energy minimization. Through the same equations of state that model the Gibbs free energy, we compute elastic moduli and density for each stable phase assemblage at the same P-T conditions. One straightforward application of this approach is the study of the seismic signature of synthetic mantle discontinuities arising in such models, as the temperature dependent phase transformations occuring at around 410 Km and 660 Km depth are naturally taken into account. We study these a priori known discontinuities by analysing the synthetic global seismic data obtained from full 3D global wave propagation through our geodynamically derived models using the spectral element method (SPECFEM3D, Komatitsch and Tromp 2002a,b). Classical techniques from global seismology as for example SS precursors or receiver functions are possible candidates for this task. http://www.bernhard-schuberth.de
DI51B-10
The Mantle Transition Zone as Seen by Global Pds Phases : no Clear Evidence for a Thin Transition Zone Beneath Hotspots
We present a new global study of the transition zone from Pds converted waves at the 410 and 660 km discontinuities. Our observations complete previous global Pds studies with a larger dataset, especially in oceanic regions where we have been able to measure Pds travel-times sampling the mantle transition zone (MTZ) beneath 26 hotspot locations. We find significant lateral variations of the MTZ thickness. Both the maximum variations (± 35 km) and the long wavelength pattern are in overall agreement with previous SS precursors studies. We observe a negative correlation between the apparent depth of the 410-km discontinuity and the MTZ thickness. This negative correlation is enhanced, if we use recently published S-wave tomographic models to correct our Pds travel-times for the 3D velocity variations above the MTZ. This supports the idea that a significant part of the 410-km and 660-km topography variations are due to the effect of temperature on the olivine phase transformations. The MTZ is generally thick beneath subduction zones and the observed MTZ variations are consistent with thermal anomalies ranging between -100° K and -300° K. In central and North America, we observe a NW-SE pattern of thick MTZ which can be associated with the fossil Farallon subduction. We do not find clear evidence for a thin MTZ beneath hotspots. However, the 410-km discontinuity remains generally depressed after our corrections for 3D heterogeneities above the MTZ, and its topography variations can be explained by thermal anomalies between +100° K and +300° K. The depth of the 660-km discontinuity may therefore be less temperature sensitive in hot regions of the mantle, which is consistent with the effect of a phase transition from majorite-garnet to perovskite at a depth of 660 km.