S21A-1798
On the Accumulation of Lithospheric Material in the Pannonian Basin Mantle Transition Zone
Subduction zones are the most efficient consuming boundaries in plate tectonics. But where and how deep does the lithospheric material go afterwards? This question is of great importance as it feeds studies and debates in mantle dynamics. Our present understanding of the fate of descending slabs is essentially based on tomography studies of active subduction zones, such as the Circum-Pacific. These images show three main scenarios (e.g. Bijwaard et al., 1998): direct penetration into the lower mantle, flattening and stagnation in the mantle transition zone (MTZ), and flattening followed by penetration. Here we look at an active and geodynamically complex zone, the Pannonian Basin and its surroundings: oceanic subductions at surface are now completed in the region, but accumulated slab material has been imaged in the MTZ by tomography (Wortel and Spakman, 2000; Piromallo and Morelli, 2003). Our approach is to compute receiver functions from a new, high-resolution deployment as well as permanent station data in order to characterise the structure of the MTZ. We discover that while the top of the MTZ is at its expected depth (410 km); its bottom is clearly depressed below 660 km and shows important undulations. We argue that these are instabilities arising from the uneven accumulation of cold material from successive subduction episodes, and that the observed depressions might develop into viscous drops initiating penetration into the lower mantle.
S21A-1799
Transition zone S velocity structure beneath the East Pacific Rise
Models of seismic velocity as a function of depth through the upper mantle provide some of the strongest constraints on the mineralogy and composition of the mantle. Although receiver function studies have provided new information on the depths of upper mantle discontinuities, they do not provide as much information on gradients and velocities in the upper mantle. The waveforms and travel times of upper mantle turning waves provide the strongest constraints on vertical variations in upper mantle velocity, although in the past they suffered from lack of dense profiles of data sampling a single part of the upper mantle that would minimize effects of 3D variations in velocity. Here we model a dense profile of triplicated upper mantle broadband S waves recorded by USArray and NARS-Baja stations in the western United States and Mexico. Earthquakes along the East Pacific Rise were recorded along profiles within 5 degree backazimuth windows with stations at a maximum of .50 separation. The distance range covered is from 150 to 550 and thus the waves sample the mantle from the lithosphere to depths near 1000 km. The data were inverted using a conjugate gradient algorithm that utilizes the reflectivity synthetic technique. Preliminary results show a much smaller gradient within the transition zone than PREM with larger jumps in velocity at the 410 and 660 km depth discontinuities. These results are more consistent with velocities predicted for a pyrolite composition than the PREM model.
S21A-1800
Transition zone structure beneath Ethiopia from 3-D fast marching pseudo-migration stacking
Several models for the origin of the Afar hotspot have been put forth over the last decade, but much ambiguity remains as to whether the hotspot tectonism found there is due to a shallow or deeply seated feature. Additionally, there has been much debate as to whether the hotspot owes its existence to a 'classic' mantle plume feature or if it is part of the African Superplume complex. To further understand the origin of the hotspot, we employ a new receiver function stacking method that incorporates a fast-marching three- dimensional ray tracing algorithm to improve upon existing studies of the mantle transition zone structure. Using teleseismic data from the Ethiopia Broadband Seismic Experiment and the EAGLE (Ethiopia Afar Grand Lithospheric Experiment) experiment, we stack receiver functions using a three-dimensional pseudo- migration technique to examine topography on the 410 and 660 km discontinuities. Previous methods of receiver function pseudo-migration incorporated ray tracing methods that were not able to ray trace through highly complicated 3-D structure, or the ray tracing techniques only produced 3-D time perturbations associated 1-D rays in a 3-D velocity medium. These previous techniques yielded confusing and incomplete results for when applied to the exceedingly complicated mantle structure beneath Ethiopia. Indeed, comparisons of the 1-D versus 3-D ray tracing techniques show that the 1-D technique mislocated structure laterally in the mantle by over 100 km. Preliminary results using our new technique show a shallower then average 410 km discontinuity and a deeper than average 660 km discontinuity over much of the region, suggested that the hotspot has a deep seated origin.
S21A-1801
Global Transition Zone Discontinuities and Seismic Heterogeneities From Body-Wave Travel-Times and Long Period Surface Waves
The tomographic inversion of 100779 Rayleigh waveforms including higher modes has recently allowed us to
constrain the SV-wave heterogeneities in the top of the upper mantle with a lateral resolution of a few
hundred kilometers and a vertical resolution of a few tens of kilometres. The TOMOGLOB project aims to
extend this model to the transition zone and lower mantle and to include the topography of the major seismic
discontinuities at 410-km and 660-km depths. We present preliminary results of this work.
We have constrained the topography of the 410 and 660 km discontinuities using a global database of SS
precursors completed by Pds converted waves measured at 167 globally distributed stations. Both SS
precursors and Pds results favour significant lateral variations of the MTZ thickness (±35 km) with
patterns in overall agreement. 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. We do
not find clear evidence for a thin MTZ beneath hotspots. However, converted waves favour a deep 410-km
discontinuity beneath hotspots, with topography variations compatible with 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.
We are currently working to improve our 3D S-wave velocity model by increasing the number and the
accuracy of higher modes measurements and by adding long period body-waves. We have developed an
automated scheme to measure travel-times of long period S phases in several frequency bands. The
approach has been tested for S, SS ScS and SKS waves and has been applied to 30 years of data from the
Global Seismological Network (GSN). This dataset opens the way to a global finite frequency S-wave
tomography of the mantle, including the topography of its major seismic discontinuities.
http://eost.u-
strasbg.fr/benoit/research.html
S21A-1802
Elastic and anelastic 1-D structure of the mantle transition zone around Japan: visualization of the dataset
Recent progress in development and application of quantitative and objective inversion for body-wave
waveforms has allowed us to obtain fine 1-D (vertical) structure for several regions (e.g., D" layer studies, Kawai et al., 2007ab, GRL). The inverse problem is written as:
A δ m = δ d_OBS-INIT,
[eqn:observation]
where A is the matrix of partial derivatives, δ d_OBS-INIT is the residual of observed
waveforms and synthetic waveforms for an initial model, and δ m is the perturbation to the initial
model. We minimize the residual of the observed and the synthetics to obtain a modified model.
Utilizing a dataset of many waveforms taken as a whole, we can extract robust information on the Earth's
structure, even though there might be a relatively large difference between the observed data and the initial
synthetics. The robustness of the inversion results is confirmed in several ways.
In our inversion we use the SVD (singular value decomposition) for the matrix A:
A = U Λ VT
and
VT V = [ δ m1 δ m2 δ m3 ·s ].
We represent the modified model δm as:
δ m = M1δ m1 + M2δ m2 + ·s + Mnδ mn.
Here we introduce a factor which we call "SVD sensitivity for each datum," which is the inner product of each
observed waveform and the modified waveform for an individual model change. Plotting these factors helps
us to visualize the distribution of the dataset and to extract robust information on regional structure.
As we have mainly studied the mantle transition zone beneath Japan and environs, we find some structure
which might represent the cold slab stagnating into the transition zone near Hokkaido, or lower Qμ for
all the regions of our interest than the global average PREM model, which might be connected to thermal or
compositional structure there.
S21A-1803
Constraints on the kinetics of mantle phase changes from seismic modes data
In a system where two or more mineral phases are in equilibrium, a pressure perturbation associated with seismic waves and normal modes can disrupt the equilibrium and induce phase changes. If the kinetic rate of the phase change is finite the transformation induces an attenuation of the waves that can be quantified in terms of a compressional quality factor~q. Therefore, we first review the theory of attenuation due to a phase change yielding an equivalent standard linear solid model, and we compute associated bulk attenuation factors. As in-situ kinetic rates of phase changes within the mantle are not well known, seismic attenuation data can be used to constrain them. We show that for each mode the attenuation curve as a function of kinetic time reaches a maximum q=qm at τ=1/ω0 where ω0 is the frequency of the mode. The predicted contribution due to the phase change to the overall mode attenuation cannot however be greater than the observed q value. Therefore it is possible to exclude all the range of kinetic times that give predicted q values larger than the observed one. Using material properties of major mineral phases present in the transition zone, we apply our approach to estimate the kinetic rate of phase transitions taking place at the depths of 410 and 660~km. In the case of olivine-to-wadsleyite phase change, the attenuation curve for the mode 0S0 implies that the actual kinetic time has to be either shorter than 1~min or larger than 10~min. Upon the 178 measurements mode measurements, we obtain the range of excluded kinetic time between 10~s and 10~min. In the case of mantle transitions, these results are affected by the actual width of the phase change. We therefore evaluate the influence of the transition width on our results.
S21A-1804
Characterizing the 410 km discontinuity low velocity layer beneath the western U.S.
Previous seismic study of the 410 km discontinuity beneath the western U.S. commonly found a 25 km thick low velocity layer with an 8 percent shear velocity reduction that resides atop the olivine-wadsleyite phase boundary. New TA-based body-wave tomograms (and previous regional array tomograms) find that high velocity region in the mantle resides about the 410 km discontinuity beneath northern Nevada and eastern Utah in the western U. S. These high velocity extend through the transition zone and are consistent with being <30 ma old subducted slabs. This observation provides a test of the 410 km water-filter model (Bercovici and Karato, 2003). Assuming that the high velocity material beneath this region is cold and sinking, then the 410 km water-filter should be inoperative. Therefore, no 410 low velocity melt layer should be present. To test this hypothesis, data from the TA and the PASSCAL Ristra1.5 array is processed to isolate P to S scattered waves. To isolate RF signals, azimuthal quadrant stacks (i.e., NW,SE,SW) for each of the stations are formed and a subset of the high-quality quadrant stacks are then culled for modeling. The modeling consists of calculating the fit (L2 or cross-correlation norm) of synthetic seismograms using a double-gradient velocity model to the high quality RF quadrant stacks. From the five-dimensional model search, the a posteriori model probability density functions are calculated to characterize model parameter uncertainty and trade-offs. Results will show whether the 410 low velocity layer is absent beneath northern Nevada and eastern Utah where high velocity cold slab material is presumably sinking downwards.
S21A-1805
Upper Mantle Discontinuity Structure From Wavefield Migration of Precursors to SS and PP
The depth and sharpness of upper mantle discontinuities at approximately 410 and 660 km depth are attributed to solid-state phase changes sensitive to both mantle temperature and composition, and regions of thermal and/or chemical heterogeneity produce topography on these boundaries. Seismic mapping of this topography provides an important constraint on the thermal and compositional state of the mantle. We apply a seismic migration approach to image the depth, sharpness, and topography of the upper mantle discontinuities, as well as other possible reflectors. In our migration, we utilize seismic waves that reflect off the underside of a mantle discontinuity and arrive several hundred seconds prior to the SS and PP seismic phases as precursory energy. Migration techniques, which are used frequently in exploration seismology, focus energy at viable true reflection locations, and have the ability of detecting out-of-plane arrivals and hence reflectors. In this study, we utilize a high-quality broadband dataset of SS and PP precursors recorded by the EarthScope USArray Seismic Network, which samples regional discontinuity structure beneath the Pacific Ocean and South America. Migration of the SS and PP precursor data allows high-resolution detection of localized structure on the discontinuity boundaries as well as additional horizons. Such structures are likely related to significant thermal and/or compositional heterogeneity at the discontinuities, and other phase changes within the mantle.
S21A-1806
High Resolution Seismic Imaging of Transition Zone Beneath the Hawaii Volcano Chain: Evidence for Deep-rooted Mantle Plume
A generalized Radon transform (GRT) is applied to SS waves reflected at the underside of seismic discontinuities in order to detect, image, and characterize transition zone interfaces under the Hawaii volcano chain. The GRT makes use of scattering theory and extracts structural information from broad band data windows that include precursors to SS (which are the specular reflections at the discontinuities that form the main arrivals) as well as non-specular scattered energy (which is often discarded as noise). More than150,000 seimograms (from the IRIS Data Management Center) are used to form a 3-D image of the transition zone beneath the central Pacific. In addition to clear signals near 410, 520, and 660 km depth, the data also reveal scatter interfaces near 370 km dept and between 800-1000 km depth, which may be regional, laterally intermittent scatter horizons. Our preliminary results reveal a conspicuous thinning of the transition zone due to uplift of the 660 discontinuity and downwarping of the 410 discontinuity west of Hawaii. This observation may suggest the presence of a deep-rooted mantle plume (with a lower-mantle origin) underneath Hawaii hotspot. The part with the largest topography difference from the ambient mantle is located several degrees west of the active volcanoes of Hawaii, which may imply that the presumed plume conduit is tilted, for instance by large scale mantle advection. This observation may put (local) constraints on "mantle wind" and have important implications for our understanding of mass flux across the transition zone and the geological evolution of the Hawaii-Emperor seamount chain.