DI22A-01 INVITED
An Integrated Seismic and Geodynamic Analysis of Mantle Deformation Beneath the San Andreas Plate Boundary Region
Surface deformation across the western US plate boundary depends on not only the mantle forces driving the deformation, but also on the strength of the crust and mantle lithosphere that transmit those forces. Previous analyses of seismic anisotropy in the region imply that Pacific-NA plate boundary deformation penetrates into the mantle in northern California, while the plate motions and mantle deformation are largely decoupled across the plate boundary in southern California. In this project, an integrated seismic- geodynamic analysis is being developed to better quantify deformation across the plate boundary. We have developed a new cross-correlation procedure to measure frequency-dependent phase delays of surface waves traversing the USArray. Conceptually similar to multi-channel cross-correlation algorithms for body- wave tomography and earthquake location, this analysis exploits the similarity between nearby recordings of the surface wavefield to more precisely estimate relative phase and amplitude. These phase-delay and amplitude measurements in turn provide more precise estimates of wave velocities within the array, including subtle azimuthal variations that are diagnostic of anisotropy, and these observations are inverted for 3D anisotropic structure using mode-coupled 3D Frechet kernels. In order to more directly interpret the anisotropy in the context of plate-boundary deformation, we construct numerical models of mantle fabric that can be directly incorporated as constraints in the seismic modeling. The numerical models simulate upper- mantle flow beneath the plate boundary using a viscous fluid subject to surface and basal boundary conditions derived from geologic and kinematic data, testing a variety of rheologies. Combined with the seismic observations, these models allow us to quantitatively test conceptual models of deformation and rheology across the plate boundary.
DI22A-02
Melt Distribution in the Ethiopian Rift System: Constraints From Seismic Observations and Finite-Frequency Modelling
As part of the Ethiopian Afar Geoscientific Lithospheric Experiment (EAGLE) 79 seismic stations were deployed, for up to 18 months, in the Main Ethiopian Rift (MER). Many indicators of melt were observed leading to the idea that magma was driving the rifting process in this region. Some of the best evidence for melt came from observations of anisotropy in studies of surface waves and shear-wave splitting. The shear- wave splitting shows fast directions which change abruptly from being rift parallel on the rift flanks to magmatic-segment parallel in the rift valley. This was interpreted in terms of melt-induced anisotropy. The abrupt change in splitting parameters over small lateral distances suggests that the source of anisotropy is shallow. To further constrain the location of the anisotropy and study the ability of shear-wave splitting to identify sharp lateral changes in anisotropy, we model finite-frequency waveforms for a suite of model representations of the rift zone. This allows us to determine the lateral and vertical extent of the melt-induced anisotropy. The results show how a simple model with two regimes of anisotropy can explain the variability across the rift, in both delay time and shear-wave polarization, over short length scales of the order 20- 40 km. Our models have enabled us to constrain the anisotropic characteristics beneath the MER. Our best model has a 9% anisotropy on the western rift margin, with fast directions of 30°, a 100 km wide rift zone with fast direction of 20° inside the rift zone and with 9% anisotropy close to the western margin, 7% elsewhere, and 7% anisotropy on the eastern margin with fast directions of 30°. In all regions of the model we constrain anisotropy to begin at a depth of 90 km. The depth of anisotropy co-incides with the proposed depth of melt initiation beneath the region, based on geochemistry. Also the elevated splitting beneath the western margin supports evidence of low velocities and highly conductive bodies from seismic tomography and magneto-tellurics, suggesting melt is more focused along the western margin. This study shows that SKS-wave splitting is a powerful technique that can map sharp lateral changes, and has the potential to constrain the depth of the anisotropy.
DI22A-03 INVITED
Shear-wave Splitting Tomography and Flow in the Mantle Wedge Beneath Costa Rica and Nicaragua
Understanding the pattern of solid mantle deformation in subduction zones is fundamentally important in
constraining its effect on geodynamic processes such as material transport and thermal structure.
Measurements of shear-wave splitting, and perhaps even travel-time anomalies, in both local and teleseismic
(e.g., S,SKS,PKS) body waves allow us to characterize upper mantle seismic anisotropy. When combined
with results from deformation experiments on mantle lithologies, inverting these splitting observations for
crystallographic orientation provides an attractive means of inferring the direction of upper mantle flow.
However, for three-dimensional models that allow for general orientations of a representative elastic mantle
material (i.e., olivine and orthopyroxene) the dependence of shear-wave splitting on model parameters is
non-linear. Therefore, in order to resolve crystallographic orientation, we utilize an iterative, linearized, least-
squares inversion in which partial derivatives are generated and re-generated with an efficient forward
calculation of splitting. This method has been applied to a large local-S splitting dataset collected by the
TUCAN seismic array in Nicaragua and Costa Rica. A predominance of arc-parallel (i.e., plate-motion-
perpendicular) olivine a-axes is found in the mantle wedge down to at least 150 km and well into the back-arc
wedge. The location and coherence of arc-parallel a-axes are inconsistent with B-type fabric which is
predicted only in the cold, shallow wedge corner, nor are they easily explained by aligned bands of partial
melt. Arc-parallel flow combined with fabrics in which olivine a-axes align approximately parallel to flow are
more likely the dominant cause. This interpretation is supported by isotopic variations in arc lavas that
suggest NW arc-parallel transport of wedge material. A significant implication of this interpretation is that flow
throughout much of the mantle wedge is not strongly coupled to the subducting Cocos Plate. In addition, SKS
fast directions are also consistently arc-parallel, but with much larger delay times than local-S splits at arc
stations, suggesting a considerable amount of arc-parallel anisotropy and possibly arc-parallel flow likely
exists beneath the slab. Although the shear-wave splitting tomography method is able to resolve broad
variations in anisotropic structure, in the future, we plan to explore the joint inversion of shear-wave splitting
parameters and travel-time residuals as a means of further constraining crystallographic orientation in
subduction zones.
http://www.geo.brown.edu/People/Grads/abt/
DI22A-04 INVITED
Surface-Wave Anisotropy and Deformation of Continents
Classical controversies in continental dynamics persist to this day, due, in large part, to the persistent lack of observational constraints on 3-D flow at depth. Because finite strain due to the flow makes crustal and mantle rock develop anisotropic fabric, observations of seismic anisotropy can provide the much needed constraints. The importance of anisotropy has motivated vigourous and fruitful research. Commonly used data types, however, have shown their limitations: SKS splitting measurements lack vertical resolution, and Pn waves sample a known but limited depth range. New debates have developed within the field itself: at what depth does SKS splitting originate? how much anisotropy is there in the asthenosphere beneath continents? Observations of surface-wave anisotropy in broad frequency bands can constrain the layering of shear-velocity anisotropy in the entire lithosphere-asthenosphere depth range. Thanks to the deployment of a growing number of dense broadband arrays and the development of suitable surface-wave techniques, it is becoming increasingly feasible to image stratification of seismic anisotropy at the scale of tectonic units. Studies in different tectonic environments have now shown that substantial azimuthal anisotropy is present, as a rule, in both the lithosphere and asthenosphere. Beneath currently stable continents, the layering of anisotropic fabric (frozen into the lithosphere since the last major deformation episodes) gives clues regarding the dynamics of orogens and the post-orogenic lithospheric evolution. In tectonically active regions, observed lithospheric anisotropy reveals the extent and character of the diffuse deformation in the lower crust and mantle lithosphere. Anisotropy in the asthenosphere reflects current and recent mantle flow; beneath stable continental lithosphere, surface-wave-inferred fast-propagation directions typically trend roughly parallel to the absolute plate motion. Beneath actively deforming regions asthenospheric anisotropy shows more complex, 3-D patterns. The magnitudes of SKS splitting and the splitting-inferred fast-propagation directions can be accounted for, at least in a number of regions, by the anisotropy constrained with surface-wave observations.
DI22A-05
Probability of Radial Anisotropy in the Deep Mantle
There is a large consensus that radial anisotropy is present in the uppermost mantle, but the few observations of radial anisotropy in the deeper mantle are more ambiguous. We have developed a robust technique to automatically measure higher mode surface wave phase velocities with corresponding uncertainties. Higher mode surface waves, contrary to fundamental mode surface waves, provide information to and below the transition zone. Using our new data, we made a fully non-linear inversion for radially anisotropic shear wave velocity model down to 1500 km. Because we make a full model space search, the earth parameters are expressed in terms of probability density functions. On average, we find significant (higher than 95 % confidence) radial anisotropy down to the lower mantle, with a sign change in anisotropy around 200 km depth. The uppermost mantle shows a high probability of faster horizontally polarized shear wave speed, likely to be related to plate motion. In the asthenosphere and transition zone, however, we find a high probability of faster vertically polarized shear wave anisotropy. There are clear regional differences in the lithosphere, asthenosphere and transition zone which we will discuss. In the lower mantle (down to 1500 km) we find no significant shear wave anisotropy consistent with laboratory measurements.
DI22A-06
Numerical Simulations of Texture Development and Associated Rheological Anisotropy in Regions of Complex Mantle Flow
The aim of this study is to compare the predictions of different micromechanical approaches that have been employed recently to study mineral alignment during flow in the upper mantle. Computational capabilities are reaching a point where the potential rheological effects of such lattice-preferred orientation (LPO) can be considered as an integral part of determining the flow pattern and evolution. But, in order to have confidence in taking this next step, the detailed behavior of the different micromechanical models needs to be understood. An important consequence of LPO development is the subsequent anisotropy of the mechanical properties. Curiously, most published geophysical studies only address the elastic anisotropy, probably because of its link with the observed seismic anisotropy. The viscoplastic (or rheological) anisotropy has received much less attention, although it may have a notable influence on regional and global convective flow pattern, which in turn controls the LPO development. Micromechanical approaches aim at linking the rheological behaviour at the grain scale, associated with the activate deformation mechanisms (dislocation glide and climb, diffusion creep, …), with the overall rheology at the sample scale, including also other mechanisms such as recrystallization. This is achieved by an evaluation of the internal stress generated by the (strong) mechanical interaction between neighbour grains. All models proposed in the literature (kinematic model, finite strain model, tangent self-consistent model, lower bound model, …) make simplifying assumptions, since the mechanical problem is very complicated. One can distinguish between rather simple models that allow some freedom in deformation of individual grains, and more advanced techniques (and generally more accurate) that require a minimum number (=4) of independent slip systems (or directional deformation mechanisms) for the plastic strain to occur. In respect to this, unlike all other models, the recent 'Second Order' self-consistent model has been shown to reproduce almost perfectly several exact solutions for olivine rheology, and this makes it a solid candidate for future applications on large scale convection in the deep Earth. We address in this presentation the predictive capability of these models with an eye toward their eventual use in coupled LPO-geodynamic flow simulations. We will focus on comparisons in the case of a complex flow pattern, rarely investigated in the literature, such as passively-driven upwelling zone beneath an oceanic spreading center.
DI22A-07
Stress-dependence of dislocation creep rate: Implications for transitions between creep mechanisms in the upper mantle
The basis for understanding the conditions under which dislocation creep is the dominant deformation mechanism in the upper mantle depends strongly on measurements of the stress dependence of creep rate. Experimental data on the creep behavior of olivine single crystals, dunites and peridotites are all well fit with a power law flow law (strain rate proportional to stressn) at relatively low differential stress (stress less than 1/20 of the shear modulus), with values of the stress exponent (n) ranging from 3 to 5 in the dislocation creep regime. Most experiments for which n is best constrained were conducted near the transition between dislocation creep and grain size sensitive deformation mechanisms. By accounting for this complication, Hirth and Kohlstedt (2003, HK03) determined n = 3.5 +/- 0.3, consistent with experiments on coarse-grained dunites (e.g., Chopra and Paterson, 1984) and single crystals (e.g., Durham and Goetze, 1977; Bai et al., 1991). In HK03, stress exponents were determined by fitting data from single experiments to avoid complications with run-to-run variability of deformation conditions. By contrast, recent "global inversions" by Korenaga and Karato (2008, KK08) of these data give a significantly larger value of n = 4.9. This small difference in stress exponent has significant implications for understanding of the rheological behavior of the mantle; as the experiments are largely conducted at mantle temperatures, the only significant extrapolations required are those in stress. At face value, the difference in estimated stress exponent indicates a difference in predicted viscosity in the range of three orders of magnitude (i.e., for extrapolation in stress from 100 MPa to 1 MPa). We will re-iterate our arguments in HK03 used in determining n = 3.5 and discuss mineral physics and kinetic data that support this value – as well as remaining uncertainties. Finally, we note that KK08 do not account for the grain size dependence of creep in the grain boundary sliding regime (see HK03); we illustrate the effect of this omission producing an anomalously high value of n.
DI22A-08 INVITED
From Polycrystal Plasticity to Deep Earth Anisotropy
A major contribution to seismic anisotropy in the deep earth is due to the alignment of anisotropic component crystals. The bulk anisotropy can be estimated by averaging single crystal properties over the orientation distribution. Preferred orientation of crystals in a polycrystalline material develops during deformation by slip and may be modified by dynamic recrystallization. Geodynamic models indicate that strains during slab subduction into the lower mantle and upwelling of plumes are very large. Here we present a polycrystal plasticity approach, modified for large strains, and show results for texture changes along streamlines for subduction (2D) and upwelling (3D). Tracers record deformation gradients in the geodynamic model. These are then used in viscoplastic polycrystal plasticity simulations. Significant parameters are slip systems and their relative importance, the evolution of grain shape and recrystallization. Depending on those assumptions significantly different patterns are obtained that can then be compared with observed anisotropy. Large strain, particularly in simple shear, produces significant heterogeneity with textures strengthening and attenuating with increasing deformation. This is in agreement with the rather local documented seismic anisotropy patterns in the lowermost mantle.