DI13A-1672
Stratification of Seismic Anisotropy of the Crust and Upper Mantle Beneath East Africa From Joint Inversion of SKS and Receiver Functions
East Africa and the west side of the Arabian plate have been affected by a large number of remote and recent tectonic episodes. The structure of the crust and upper mantle in this area is likely to be complex. Fundamental constraints on stress/strain field can be obtained from the analysis of seismic anisotropy. So far, previous studies in this region have led to apparently inconsistent results depending of the type of data used. In particular, although observations of SKS splitting at most places are usually accounted for by models with a single homogeneous layer, 3D tomographic models derived from surface waves exhibit substantial vertical variations of anisotropy in the crust and upper mantle. These methods have poor vertical and lateral resolution, respectively, and telling which one provides the most robust results is not straightforward. Here we address the issue of the stratification of anisotropy by performing joint inversion of receiver functions (RF) and SKS waveforms at several permanent broadband stations where RFs and SKS waves are clearly indicative of anisotropy. In each case, a satisfactory fit of both RFs and SKS data is achieved by modeling a stack of distinct anisotropic layers, the sequence of which is in good agreement with recent tomographic images obtained from surface waves. In particular, the lithosphere-asthenosphere transition seems to be linked to the vertical discontinuities detected in the anisotropic properties. As previously highlighted by several studies performed in a wide variety of geological and tectonic settings, our analysis emphasizes the need of considering the existence of stratification in anisotropy to obtain realistic models of the crust and upper mantle and thus to achieve a better understanding of local geodynamics.
DI13A-1673
Distribution of Seismic Anisotropy Beneath Central Italy and Geodynamic Implications for Northern Apennines
Competing geodynamic scenarios proposed for northern Apennines make very different predictions about the likely orientation of strain in the upper mantle. Constraints on the actual strain are commonly sought in observations of seismic anisotropy. Previous study of the seismic anisotropy distribution beneath Northern Apennines (Italy) used birefringence of core-refracted shear waves (SKS phases), and demonstrated the presence of two domains. The Tuscany domain (Tyrrhenian side) showed relatively homogeneous NW-SE fast direction in the upper mantle; the Adria domain (Eastern side of the Apennines) showed a mix of N-S to NE-SW fast directions, with measurements being strongly influenced by back-azimuth and justifying the presence of a multi-layers structure beneath it. To better understand the lithosphere-asthenosphere structure beneath the Adria region we analyze seismographic data recorded by a set of seismic stations located in the outer part of the Apennines belt and in the Adria terranes. We combine two complimentary techniques of probing the anisotropic seismic velocity at depth. We use directionally distributed sets of SKS phase records to invert for layered anisotropic structures. Also, for each site, receiver function sweeps are modeled in terms of layered anisotropic structure. These results are used as a constraint layered inversions using shear wave splitting observations, at least when crustal anisotropy seems significant. A combined model of mantle and crustal anisotropy is "assembled", and tested with synthetic seismogram computations to assess the similarity between observed and predicted patterns of shear wave birefringence and P-S conversion. In terms of ongoing mantle deformation, we suggest an hypothetical sketch foreseeing a differential evolution of the trench retreat process along the Northern Apennines orogen. Compared to the anisotropy pattern of the typical slab retreat process, seen in southern part of the Northern Apennines, in the northernmost one the anisotropy would suggest an oblique trench retreat occurrence.
DI13A-1674
Shear Wave Splitting Beneath Western Paraguay: Evidence of Lithospheric Deformation From Mesozoic Rifting
Many shear wave splitting studies of South America have focused on the Andean subduction zone rather than the stable South American platform. In this study, we perform a shear-wave splitting analysis to broaden our understanding of intracontinental deformation, lithospheric structure, and lithosphere-asthenosphere interaction beneath two intracontinental seismic stations (SAML in western Brazil, CPUP in western Paraguay). Results for SAML are inconclusive due to predominance of noise in the seismograms, resulting in only 4 acceptable results of poor quality that do not correlate with each other. Results for CPUP are of high quality and indicate strong anisotropy (1.4 seconds lag time) with the fast direction in a WNW orientation (at an azimuth of 289 degrees). We address several possibilities to explain the results for CPUP. A large NW trending Mesozoic dyke swarm in the area could be responsible for crustal anisotropy, as could microcracks opened by compressional stresses caused by current Nazca/South America plate convergence. Anisotropy may also arise from the shearing of the mantle between the lithosphere and asthenosphere in the direction of plate motion. However, results for CPUP do not support the above possibilities: observed lag times are too high to be attributed to crustal anisotropy alone and observed fast directions do not correlate with expected shear between the lithosphere and asthenosphere. We propose that the observed anisotropy at CPUP is a signature remaining from the last major tectonic episode to affect the area. Rifting of the study area associated with the early stages of the formation of the Atlantic ocean in the Mesozoic would have weakened and heated the lithospheric mantle, making it more likely to form a strong anisotropic fabric. When this rift arm failed, tensional forces rotated into a WNW orientation that eventually successfully rifted South America away from Africa to form the Atlantic Ocean. Due to the correlation of CPUP results and the orientation of extension during the second stage of rifting, we propose that asthenospheric convection during the early stages of Atlantic rifting imposed shear stress on the weakened and hot lithosphere beneath CPUP, resulting in a strong anisotropic fabric in the lithosphere with a fast seismic direction oriented parallel with the spreading direction. This interpretation of a long term, post-event anisotropic fabric provides support for frozen-in anisotropy within the stable South American platform that is strongly connected to past tectonic events.
DI13A-1675
Kinematic Models of Mantle Flow and Fabric Beneath the San Andreas Plate Boundary Region and Implications for Seismic Anisotropy
Seismic anisotropy measurements from shear-wave splitting at the San Andreas fault (SAF) system show fault-parallel fast-polarization directions near the fault and east-west (E-W) orientations away from the fault in northern California. In southern California, a cryptic near-fault region of fault-parallel fast-polarization directions are observed within a broad region of E-W directions. The variation in near-fault splitting parameters in northern California has been previously interpreted in terms of a layered anisotropic structure, a dipping symmetry axis of anisotropy, or a single layer with depth-dependent anisotropy. In both northern and southern California, the regional and/or deeper layer with E-W orientation of fast polarization directions is uncorrelated with the surface kinematic field and suggests asthenospheric flow that is decoupled from the overlying plate. To better understand these observations, we have developed kinematic flow models for sub- lithospheric material beneath the SAF that is influenced by both plate motions and deeper mantle flow. We idealize the SAF system in our models as a pure strike-slip plate boundary and assume that the deep mantle flow field in the region is oriented perpendicular to the fault. Using published relationships between mantle flow, strain, and expected fabric development due to dynamic recrystallization, we predict the expected seismic anisotropy beneath the SAF for modeled kinematic scenarios. We show that at shallow depths, sub- lithospheric flow is strongly influenced by plate motions and produce a focussed, shallow, near-fault region of fault-parallel anisotropy that is distributed asymmetrically around the fault. Away from the fault and at greater depths, flow is influenced by the deeper mantle and yields a "background" fault-oblique anisotropy that surrounds the shallow region of fault-parallel anisotropy. We use isoviscous and layered Newtonian fluids, and find that a high viscosity upper layer overlying a lower viscosity layer produces larger regions of fault-parallel anisotropy. Using a 3D finite-difference approach, we propagate seismic waves through our modeled anisotropic structures to predict the expected shear-wave splitting for various kinematic scenarios. By comparing the predicted splitting to the observations in California, our goal is to better understand mantle deformation beneath the SAF system.
DI13A-1676
Azimuthal Anisotropy From Surface Waves in the Great Basin
The Great Basin, Nevada, is a region of widespread crustal extension and magmatism. It is also characterized by a thin lithosphere (less than 100 km) and a high average heat flow, except in central Nevada where the heat flow is significantly reduced. In this study, we modeled seismic velocities and anisotropy in this region in order to understand how mantle dynamics relate to these geophysical and geological observations. We employed a two-station method to measure fundamental mode Rayleigh wave dispersion between periods of 16 and 170s using USArray Transportable Array (TA) data. We employed 22 events with high quality dispersion curves measured between stations aligned approximately along a common great-circle path (within 3 degrees). The excellent lateral and azimuthal coverage provided by TA seismic stations enabled us to produce azimuthally anisotropic phase velocity maps between periods of 16 and 100s, which gives constraints on shear-wave velocity structure down to depths of 150 to 200 km. Measurement uncertainties were large at periods greater than 100s, and no significant deviation from our reference model (a slightly modified Tectonic North America (mTNA) velocity model) were found. The isotropic part of the maps displays lateral changes in phase velocities between periods of 16 and 28s, but they become generally lower than predicted by mTNA at larger periods over the entire study region. Given that the sensitivity of 28s Rayleigh waves to shear wave velocities peaks at about 40 km depth, these results suggest a reduction in shear wave velocities compared to mTNA located below a thin lithospheric lid of about 60 km. In addition, we found that the data favor no or very little azimuthal anisotropy in central Nevada between 16 and 28s. For periods between 33 and 100s azimuthal anisotropy is required to significantly improve the data fit. This change appears thus to coincide with a lithosphere-asthenosphere transition. Other constraints on azimuthal anisotropy in the region include shear wave splitting, which exhibits a marked reduction in azimuthal anisotropy in central Nevada, comparable to our results at short periods. This indicates that the origin of the low shear-wave splitting signal lies at least partially in the lithosphere. In addition, this reduction in azimuthal anisotropy is coincident with a columnar zone of increased seismic velocities imaged by relative delay time tomography, which, combined with other geological and geophysical constraints, was interpreted as a lithospheric downwelling. The requirement by surface waves of upper asthenospheric anisotropy suggests that this may be a region of lateral flow feeding the lithospheric downwelling. In addition, the absence of azimuthal anisotropy found in the lithosphere could be the signature of past lithospheric delamination and subsequent downwelling, in which case it is very likely associated with the presence of radial anisotropy, such as the one found by others in the crust using ambient noise tomography.
DI13A-1677
Upper Mantle Anisotropy Under Fast Spreading Mid-ocean Ridges: 2-D Whole Mantle Convection Model With Subduction
Analyses of seismic anisotropy caused by spatial alignments of anisotropic minerals (e.g., olivine) have been widely used to infer mantle flow directions in the upper mantle. Deep seismic anisotropy beneath fast spreading mid-ocean ridges (e.g., East Pacific Rise) has been recently observed at depths of 200-300 km and even down to the transition zone, with polarization changes in radial anisotropy from VSH < VSV (shallow) to VSH < VSV (deep). We investigate the origin of the observed deep seismic anisotropy and polarization changes beneath the EPR in 2-D Cartesian numerical models using both kinematically (prescribed velocity) and dynamically (negative buoyancy) driven ridge spreading. Because subduction is thought to be an important controlling factor in the style of ridge spreading and mantle convection, we consider a subduction zone developing at the prescribed weak zone. A whole mantle domain expressed by a one by four box (2890 by 11560 km) is used to minimize the boundary effects on the subducting slab. For the upper mantle rheology, we consider composite viscosity of diffusion and dislocation creep for dry olivine to evaluate the effects of lateral variation of mantle viscosity and the rheological changes from dislocation to diffusion creep under the mid-ocean ridge. For the lower mantle rheology, we use diffusion creep for dry olivine by increasing grain size to match relevant lower mantle viscosity. We also consider the 660 km phase transition with density and viscosity jump as well as Clapeyron slope. Anisotropy is evaluated using finite-strain ellipses based on the assumption that a-axes of olivine crystals are parallel to the major axes of the finite-strain ellipses. Our preliminary results show 1) in general, the development of VSH < VSV anisotropy is confined only in a narrow region under the ridge axis at depths of 200- 300 km; 2) strong VSH > VSV anisotropy can be found in the 'asthenosphere' beneath the entire spreading oceanic lithosphere; and 3) the dominate creep mechanism changes from dislocation creep to diffusion creep at depths of 300-400 km; indicating a more isotropic lower upper mantle. We conclude that our geodynamical modeling in a passive ridge spreading system does not produce the deep seismic anisotropy recently observed beneath the EPR. However, we do not consider partial melting, dynamic recrystallization and anisotropic viscosity which would change seismic interpretation and mantle flow, and thus further study is required.
DI13A-1678
Thin anisotropic layer in the mantle wedge beneath Northeast Japan
A wide variety of seismic anisotropy has been observed in subduction zones, including trench-parallel and trench-perpendicular shear-wave splittings (e.g., Nakajima and Hasegawa, 2004; Long and van der Hilst, 2005). Geodynamic modeling has been proposed to explain such complex anisotropy in the upper mantle of subduction zones (e.g., Wiens and Smith, 2003; Kneller et al. 2005): however, their origin remains still controversial because of the poor vertical resolution afforded by these seismic data. One of the key questions that might resolve this problem is the origin of the anisotropic signature within the upper mantle. We tested the distribution of seismic anisotropy in the mantle wedge beneath Northeast Japan based on the deformation mechanisms: a lattice-preferred orientation and seismic anisotropy are generated by deformation via dislocation creep in the upper mantle, but not by diffusion creep or frictional sliding. Based on the thermal structure and stress field of the upper mantle beneath Northeast Japan, deformation throughout most of the mantle wedge is inferred to be controlled by diffusion creep, and the region of dislocation creep is limited to a thin layer of 10-20 km thickness within a region of relatively high stress and low temperature located above the subducting slab and beneath the island arc crust. The relatively short delay time recorded in Northeast Japan is consistent with the occurrence of a thin anisotropic layer within the mantle wedge. We therefore conclude that the seismic anisotropy observed in the subduction zone upper mantle is largely caused by a very thin (~10-20 km thick) but strong anisotropic layer. Since the mantle flow in this region is mainly coupled with the downgoing plate, being approximately parallel to the down-dip direction of the subducting slab, a change in the dominant olivine LPO arising from spatial variations in temperature and water content is a plausible mechanism to create the complex anisotropy observed in the upper mantle beneath Northeast Japan.
DI13A-1679
Subduction Zone Anisotropic Patterns Produced by Faulting and Hydration of the Slab.
Normal faulting and hydration of the subducting oceanic plate is widely observed at the outer rise of subduction zones. In order to investigate these processes, we performed 2D numerical models of a spontaneously bending oceanic plate using I2ELVIS code that account for visco-elasto-plastic rheologies. At the outer rise, bending-related slab faulting occurs and provides a pathway for water percolation in the slab. Faults generally deep trenchward, but antithetic faults are also common. As the slab subducts, serpentinized faults acquire a sub-vertical position. The upper part of the slab is, hence, formed by a sub-vertical layering of rocks with different elastic properties (dry and hydrated portions) that, for long wavelength SKS waves, appears as transverse isotropic body with a horizontal axis of symmetry (SPO). Furthermore, hydration of ultramafic rocks leads to the formation of sheet silicates such as lizardite, antigorite, talc and chlorite. Such hydrous minerals are highly anisotropic (80-100%) and tend to orient parallel to the shear zone walls. Hence, along the hydrated normal faults these minerals will tend to align parallel to the faults (LPO). Calculations of the slab anisotropy lead to ~1 second time delay for a 30-40 km thick anisotropic layer, depending on the aspect ratio of the hydrated fractures. Indeed, in medium to very old slabs, the vertical distance between the upper and lower planes of the Double Benioff Zone is around 30-45 km. The spatial distribution of fault sets at the outer rise is consistent with the common trench-parallel orientation of the fast SKS component measured in the forearc. We also found that in areas where the fast SKS component is normal to the trench, fault sets are oriented at high angle respect to the trench. Therefore, slab anisotropy produced by localized hydration of the slab can significantly contribute to the observed splitting patterns at subduction zones.
DI13A-1680
Mantle flow around the slab edge: Numerical simulations and their implications for the trench-parallel flow in the sub-slab mantle
A simple semi-dynamic numerical model of the subduction zone, incorporating the plate velocity and a moving plate boundary as a priori constraints, has been constructed. In order to obtain the feature similar to the subduction, the flow velocity on the top and the small region around the shallow plate boundary is kinematically given and the flow in other part is dynamically calculated. We applied this model to the study of mantle flow around the slab-edge. The plate configuration mimics the region around the junction of Aleutian Islands and Kamchatka, that is, the convergent-transform fault plate boundaries. For the case with the overlying plate being almost stationary, the lateral flow from the sub-slab mantle under the subducting slab to the mantle under the neighboring plate hardly exists, once the slab penetrates into the high viscosity layer where the downward flow encounters the resistance. Similar result was obtained for the case with advancing trench, that is, the trench moves toward the overlying plate. For the case with retreating trench, that is, the trench moves toward the subducting plate, such sub-slab mantle flow exists even after the penetration of slab into the high viscosity layer. However, its speed is significantly smaller than that of plate velocity. A significant lateral flow is observed when the high temperature anomaly, that is, buoyant and low viscosity block is carried toward the slab. These results may have important implications for the understanding of trench- parallel anisotropy of seismic velocity in the sub-slab mantle.
DI13A-1681
Subducted Oceanic Asthenosphere and Upper Mantle Flow Beneath the Juan de Fuca Slab
Typical oceanic lithosphere is underlain by a well developed asthenospheric low velocity zone from ~100-250 km depth. However, the fate of the oceanic asthenosphere at subduction zones is poorly understood. A simple test of the effect of subduction on the asthenosphere can be performed using two earthquakes which occurred in the Juan de Fuca slab. The Cascades subduction zone is suitable for our purpose, since it is short but significantly curved, changing strike by some 90°, so thus the potential for three dimensional upper mantle flow can be assessed in a relatively small volume. The July 3, 1999 magnitude 5.8 and February 28, 2001 magnitude 6.8 Nisqually earthquakes, located 45-50 km deep allow measurement of source-side shear wave splitting since S waves arrive at distant recording stations with good signal-to-noise, uncontaminated by surface interaction phases like pS and sS. Source-side splitting was then isolated from splitting of S waves at stations whose upper mantle splitting parameters, φ (fast polarization azimuth) and δ t (time lag between fast and slow orthogonally polarized split phases) are well known. Currently available data from the IRIS DMC yields 41 good measurements of splitting beneath the Juan de Fuca slab. Resulting split S waves vary systematically depending on source-receiver take-off azimuth: splitting along paths that sample the region immediately below the Juan de Fuca slab in the direction of slab dip (eastward azimuths) and along the slab strike (azimuths to NW and S-SE), is systematically parallel to the subduction trench. For paths that sample the sub-slab region along W and WSW azimuths (i.e., at high angle to the slab strike and ~180° from slab dip), splitting parameters are close to the current velocity azimuth of the Juan de Fuca plate with respect to assumed-fixed hotspots). The variation of apparent splitting with source- receiver geometry can be explained by a two-layer model of upper mantle fabrics beneath the Juan de Fuca slab: a layer immediately below the slab has predominant anisotropy with fast shear parallel to the local trench strike; and a deeper layer beneath the western, as-yet unsubducted portion of the Juan de Fuca plate, is characterized by fast shear in the Juan de Fuca absolute motion direction, i.e., ENE. Thus, observations of the asthenosphere beneath oceanic plates are consistent with simple asthenospheric flow, but near the Juan de Fuca subduction zone, flow apparently becomes 3-dimensional and a trench-parallel flow layer develops. Given the geometry of sampling by S waves leaving the study region, the sub-slab trench-parallel flow layer cannot be more than around 120 km thick, although the oblique ray paths traveled by the downgoing waves are much longer, consistent with a nominal 3-4% anisotropy to yield the long delay times observed (mean of 2.9 s). This trench-parallel layer may be a manifestation of strain partitioning in the sub-slab upper mantle due to interaction of the subducted slab with the overriding plate: compressional stresses across the subduction zone apparently yield a reorientation of sub-slab LPO.
DI13A-1682
The Temporal Evolution Of A Subducting Plate In The Lower Mantle
It is now widely accepted that some subducting slabs may cross the lower/upper mantle boundary to ground below the 660 km discontinuity. Indeed, geophysical data underline long and narrow traces of fast materials, associated with subducting slabs, from the upper mantle transition zone to mid-mantle depths that are visible beneath North and South America and southern Asia (Li et al, 2008). Furthermore, seismic tomography data (Van der Hilst et al., 1997; Karason and van der Hilst, 2000, 2001)) show a large variety of slab geometries and of mantle flow patterns around subducting plate boundaries (e.g. the slab geometry in the lower mantle in the Tonga subduction zone). However, seismic tomography does not elucidate the temporal evolution of the slab behaviour and geometry during its descent through the upper and lower mantle. In this work, we therefore propose to study the deformation of a thin plate (slab) falling in a viscous fluid (mantle). The combination of both analogue and numerical experiments provides important insights into the shape and attitude evolution of subducting slabs. Models bring information into the controls exerted by the rheology of the slab and the mantle and other physical parameters such as the density contrast between the slab and the surrounding mantle, on the rate at which this deformation takes place. We show that in function of a viscosity ratios between the plate and the surrounding fluid, the plate will acquire a characteristic shape. For the isoviscous case, the plate shape tends toward a bubble with long tails: a jellyfish form. The time necessary for the plate to acquire this shape is a function of the viscosity and density contrast between the slab and the mantle. To complete our approach, we have developed a semi-analytical model based on the solution of the Hadamar-Rybinski equations for the problem of a dense, yet isoviscous and thus deforming sphere. This model helps to better describe flow processes around the downgoing plate and, simultaneously, to characterize its deformation. In this way, we were able to calculate the velocities in the mantle, the forces exerted by the fluid on the plate, and the dissipated energy in the surrounding fluid. Experimental results will be correlated with geophysical data.
DI13A-1683
Preliminary Evidence For The Presence Of The Hainan Plume From Shear Wave Splitting Analyses Of A Temporary Seismic Array
We have analyzed SKS and SKKS data recorded from a temporary seismic array in Leizhou Peninsula, just to the north of the Hainan Island, to constrain anisotropic seismic structure in the upper mantle for the evidence of a Hainan mantle plume that had been suggested from global seismic tomography studies. The array consisting of 14 seismic stations is distributed over the entire Peninsula, which was mostly outcropped by Cenozoic basalt, and extended into the stable South China block. While shear-wave splitting is seen at most of the stations the fast polarization directions are divided into two different groups. At northern stations of the South China block, anisotropy with a NWW-SEE fast direction and about 0.9s delay time was observed, which is consistent with the overall movement of the South China block. However, at stations within the peninsula, anisotropy of a slightly smaller delay time of 0.7s was observed but most of the fast directions are aligned with a NE-SW direction, which is about 100 degrees from the fast direction at those northern stations. Assuming the observed anisotropy be resulted from the asthenospheric flow, the preliminary results of the difference in the fast direction between the stations within the peninsula and stations in the South China block indicate a different mantle flow pattern beneath these two regions. A possible explanation is the presence of a Hainan mantle plume beneath the Leizhou Peninsula, which could cause the mantle flow beneath the peninsula different from the mantle flow beneath the South China block, and therefore, the difference in the fast directions of the anisotropy of these two regions.
DI13A-1684
Investigating Global Radial Anisotropy in the Earth's Upper Mantle and Transition Zone
A number of recent global tomographic studies have modelled three dimensional variations in the parameters of radial anisotropy. As yet there is limited agreement among such studies, suggesting significant uncertainties in the models, which could lead to divergent geodynamical interpretations. In this study we assess the robustness of lateral variations in radial anisotropy globally in the upper mantle and in the transition zone to determine the extent to which anisotropic parameters are constrained by a massive dataset of over 10,000,000 fundamental and higher mode surface wave dispersion measurements. We carry out a large number of global tomographic inversions for isotropic and radially anisotropic shear wave velocity, systematically changing regularisation and using three different crustal models to remove the effects of the crust on the data. Using crustal corrections from different crustal models has an impact on the data fit comparable or larger than that obtained by including lateral variations of radial anisotropy in the modelling. At a depth of 100 km the images of radial anisotropy obtained using varying crustal corrections are quite different, particularly beneath Eurasia and North America. In addition, we show that including lateral variations of compressional wave velocity in the modelling produces a fit to the data comparable to that obtained by allowing for lateral variations in radial anisotropy. We discuss the reasonableness of the models we obtain and the implications of our findings for future work toward more robust seismic anisotropic models.
DI13A-1685
Geodynamically relevant surface wave tomography: Synthetic testing of nonlinearity and resolvability
Models of seismic anisotropy that seek to resolve both the strength and orientation of the best-fitting symmetry axis have potential to more directly illuminate mantle convection flow patterns than more common models which limit their focus assumptions of vertical or horizontal symmetry axes. In recent years, proposals have been made to invert for models which allow more general models of anisotropic symmetry orientation for shear wave splitting (e.g. Chevrot, 2006; Abt and Fischer, 2008) and surface wave measurements (Panning and Nolet, 2008). Such approaches hold great promise for better observing the dynamic flow patterns of the mantle; however, they also come with a penalty of nonlinearity. Inverting for both strength and orientation of anisotropy leads to partial derivatives with respect to model parameters that depend upon the choice of starting model. To determine the effect such non-linearity may have on resulting models, as well as exploring the best methods to robustly determine the best model choice, we perform tests with synthetic data. Surface wave data are numerically modeled through geodynamically reasonable models of anisotropy (e.g. Becker, 2008), and are then inverted for anisotropic models using the finite frequency theory of Panning and Nolet, 2008. Estimates of resolvability of anisotropic structure, as well as approaches for choosing starting models are presented.
DI13A-1686
Toward Global Models of Lowermost Mantle Azimuthal Anisotropy Using Long-Period Body-Wave Data
The lowermost mantle (a.k.a. D") is an enigmatic and important region of the Earth. As the interface between the core and the mantle its properties place significant boundary conditions on the dynamics and evolution of the two systems. One of the most striking features of D" is its seismic anisotropy (the variation of seismic wavespeed with direction), the magnitude of which is only surpassed in the mantle by the uppermost layers. Anisotropy is an indicator of long-range order in a material, for example the preferential alignment of crystals, grains or inclusions by deformation. As such it can be used to make both mineralogical and dynamical inferences, which are an important area of study for the lowermost mantle. For nearly two decades the anisotropy of this region has been studied using shear-mode body wave phases. The most common technique applied is the measurement of shear-wave splitting. Until recently this was done primarily by measuring differential travel times between radial and transverse components of phases such as S, ScS and Sdiff. For this to be valid, the anisotropy must be assumed to be transversely isotropic (TI). This has been fairly successful in mapping broad anisotropic structure in D". More recently some groups have developed techniques of to allow for more general forms of anisotropy (e.g., Garnero et al, Science, 2004; Wookey et al, GJI, 2005; Wookey and Kendall, EPSL, 2008). These techniques have also had to address the issue of anisotropy in upper mantle overprinting that from D", which is generally done using a reference phase which doesn't transit the lowermost mantle (for example, differential S-ScS splitting; Wookey et al, GJI, 2005). This also permits the use of shallow events. These studies have found anisotropy which, to various degrees, differ from simple TI. While these studies represent an important step forward, they are all limited to fairly small regional datasets. In order to extend this to a broader scale requires some significant challenges to be overcome. Travel-time tomography has enjoyed significant success by using long-period seismic data (e.g., Houser et al, GJI, 2008). This has the advantage of removing the 5-8s microseismic noise band, resulting in higher signal-to-noise ratio data than unfiltered broadband, and greatly increasing the number of usable waveforms. These data, however, present some limitations for the analysis of shear-wave splitting, primarily stemming from the relative size of typical time lags compared to the dominant period. We explore this effect with synthetics, as well as a solution: stacking of shear-wave splitting results for events with different source polarisations. For reasonable SNR data as few as 3-5 events with different polarisations are sufficient to make a high-quality estimate of the splitting. We use the database of Houser et al (GJI, 2008) ScS-S differential travel-time measurements from the LP dataset to identify high-quality waveforms, and demonstrate the coverage available to such a model. The best coverage is in the northern hemisphere, particularly beneath Asia and the Pacific Rim. This includes a significant number of bins which are imaged from multiple azimuths, which can be used to better distinguish between mechanisms of anisotropy (Wookey and Kendall, EPSL, 2008). We demonstrate the methodology on a subset of the LP ScS-S dataset, and compare the results to reasonable models of lowermost mantle anisotropy. In the future we aim to extend the methodology to other phases (such as Sdiff), and also integrate the models with those derived from surface wave and normal mode measurements.
DI13A-1687
A model to quantify mixing and anisotropy in three dimensional, incompressible flow fields
We present a method to calculate the stretching and thinning of the principal axes of passive, infinitesimal strain ellipsoids subject to an arbitrary incompressible, three dimensional flow field. This method can be used to study both the development of fabrics relevant to the origins of anisotropy and to study mixing. Starting with the advection equation, we derive explicit relations between the length of a principal axes and the associated normal strain along that axes. We also derive explicit relations between the orientation of the principal axes, and the associated shear strain and vorticity. We analyze the advective stretching of the principal axes of an aggregate of strain ellipsoids as a proxy for mixing. Their orientation is used as a proxy for development of Lattice Preferred Orientation (LPO), one source of anisotropy. We first apply our method to study the stretching and mixing behaviour in a dynamically driven model of subduction. The flow is driven by a slab in a cartesian box that subducts at an imposed fixed rate. We compute the orientation of the principal axes of strain ellipsoids to simulate the development of LPO around a subducting slab. We also apply our method to calculate rates of mixing and the underlying Lagrangian structures associated with two plate driven flows. The plate driven flows are designed to emulate a ridge-transform system applicable to the mantle. We find that bounded particle paths correspond to regions of rapid stretching which should be associated with efficient mixing. We also find large variations in stretching even in steady kinematic flows.
DI13A-1688
"LPO Lite" : Representing Lattice Preferred Orientation and its Evolution Using Structured Basis Functions
Current methods for calculating the evolution of flow-induced seismic anisotropy in the upper mantle describe Lattice Preferred Orientation (LPO) using ensembles of 103-104 individual grains, and are nowaday too computationally expensive to be incorporated into three-dimensional time-dependent convection models. We propose a much faster (by a factor ~ 103) method wherein LPO is described by a small number of 'structured basis functions' (SBFs.) The number of SBFs required is equal to the number of active slip systems (= 3 for olivine), and each SBF represents the 'virtual' LPO that would be produced by the action of just one of those systems. Analytical expressions for the SBFs are obtained using a simple 'single-slip' (SS) model, and are then tested against the predictions of the second-order (SO) self-consistent model of Ponte-Castaneda (J. Mech. Phys. Solids 50, 737-757, 2002) in which several slip systems act simultaneously. Remarkably, the SS model reproduces exactly (99.9% variance reduction) the orientation- dependence of the slip rate ·γ predicted by the SO model for each active slip system, once the overall amplitude of the SS expression for ·γ has been determined by least-squares fitting to the SO prediction. Having thus demonstrated that the analytical SBFs are physically realistic, we develop a scheme for representing an arbitrary LPO as a superposition of the SBFs and for determining the evolution equations satisfied by the expansion coefficients. We illustrate the method both for simple uniform deformations (uniaxial compression, simple shear) and for more geophysically realistic nonuniform deformation histories.
DI13A-1689
Deformation Microstructures of Olivine and Pyroxene in Peridotite, Spitsbergen, Svalbard and Implications for Seismic Anisotropy
Seismic anisotropy in the upper mantle can be explained mainly by lattice preferred orientation (LPO) of olivine and pyroxene. To understand deformation mechanism and seismic anisotropy in the upper mantle under Spitsbergen, Svalbard near Arctic, deformation microstructures of olivine and pyroxene in peridotite, Spitsbergen were studied. Samples are well foliated and LPOs of olivine and pyroxene were determined using electron backscattered diffraction (EBSD) with the HKL Channel 5 software. Dislocation microstructures of olivine were observed using SEM after oxygen decoration of specimens. Water contents of olivine and pyroxene were measured using a Nicolet 6700 FTIR with Continuum FTIR Microscope. Eight specimens out of ten showed that [100] axis of olivine is aligned subparallel to the lineation and (010) plane is subparallel to the foliation, which is a type-A LPO (Jung and Karato, 2001). Two specimens showed that [100] axis of olivine is aligned subparallel to the lineation and both [010] and [001] axes are distributed in a girdle nearly perpendicular to the lineation, which is a type-D LPO. [001] axis of orthopyroxene is aligned in most cases sub-parallel to the lineation, and (100) plane is sub-parallel to the foliation. Dislocation density of olivine in the samples showing type-D LPO was higher than that in the samples showing type-A LPO. Result of FTIR study showed that both type-A and -D samples are dry. These observations are in good agreement with previous experimental study (Jung et al., 2006): samples showing type-D LPO of olivine were observed at high stress condition and both type-A and -D samples were observed in dry condition. Observations of strong LPOs and dislocations in the specimen indicate that the peridotites studied were deformed by dislocation creep. Seismic anisotropy calculated from the LPOs of olivine can be used well to explain seismic anisotropy of shear wave splitting in the lithospheric mantle under Spitsbergen, Svalbard.
DI13A-1690
Pressure-Induced Fabric Transitions in Olivine and Implications for Seismic Anisotropy in the Upper Mantle
Seismic anisotropy in the Earth's upper mantle is widely observed and considered to be mainly caused by the lattice-preferred orientation (LPO) of olivine. Although seismic anisotropy in the mantle wedge at shallow depth can be easily explained by the LPO of wet olivine, the origin of trench-parallel seismic anisotropy in the upper mantle beneath the lithosphere near trenches at higher pressure and temperature is not well understood. Using a modified Griggs apparatus, we deformed a synthetic harzburgite under low differential stresses (< 200 MPa) and dry conditions (< 100 ppm H/Si) at P = 2.5 - 3.6 GPa and T = 1573 K at constant strain rates of 2 - 6 x 10-5/s. The LPO of olivine was determined using electron backscattered diffraction (EBSD) with the HKL Channel 5 software. Water content was measured using a Nicolet 6700 FTIR with Continuum FTIR Microscope. The LPO of olivine changes with increasing pressure. Type-A LPO of olivine, a conventional fabric reflecting the (010)[100] slip system, was observed at P = 2.5 GPa. However, type-B LPO of olivine, implying the (010)[001] slip system, was observed at P = 3.1 GPa and 3.6 GPa. Seismic anisotropy calculated from the high pressure fabric is significantly different from that at lower pressure. Consequently, the relationship between seismic anisotropy and flow geometry in high pressure regions is expected to be different from that in low pressure regions under the conditions investigated. Some of the seismic anisotropy including the trench-parallel seismic anisotropy of shear wave splitting in many subduction zones can be attributed to the type-B LPO of dry olivine at high pressure and temperature, implying entrainment of asthenosphere with lithosphere descent.
DI13A-1691
Combined in-situ laboratory measurements of rheology and electrical conductivity on olivine and melt-bearing olivine aggregates
Magneto-telluric investigations highlight the presence of highly conductive regions in the oceanic asthenosphere. The conductivity in such portions of the mantle is additionally about one order of magnitude higher in the direction parallel to the spreading direction compared to the direction parallel to the ridge axis. Two major hypotheses have been considered to account for the magneto-telluric data: hydrogen dissolved in olivine or the presence of melt. To examine the role of relatively small weight fractions of melt (0.01-0.04) on both rheology and electrical conductivity of olivine aggregates, a novel experimental assembly has been used to perform experiments in a high temperature, high pressure Paterson-type internally heated pressure vessel. The experiments have been carried out at temperatures between 1273 K and 1473 K and at constant confining pressure of 300 MPa. As samples were used a natural dunite from Aaheim (Norway) and synthetic aggregates of San Carlo olivine both without melt and containing low weight fraction of basalt. The electrical conductivity measurements have been performed both in static conditions and during deformation in simple shear (strain rates between 10-6s-1 and 3x10-5s-1). The measurements have been carried out with a two electrodes configuration with the impedance being radially measured. The inner electrode (Nickel) was placed centrally in the cylindrical specimens while a Nickel foil on the external surface served as second electrode. The conductivity of melt-free samples is about 3x10-4 S/m and increases to 1x10-3 S/m in presence of 4 wt.% melt at 1200 K. The variation of conductivity with respect to the inverse of temperature is Arrhenian with a lower activation energy for the melt bearing samples (1.05 eV) compared to the melt-free samples (1.45 eV). Preliminary results show that the conductivity of melt-bearing samples (up to 4wt.% melt) decreases with increasing strain. The measurement of electrical properties during deformation has the potential to help constraining the factors generating the anomalies present in the oceanic asthenosphere and their anisotropy.