T33E-01 INVITED 13:40h
Modeling Seismic Anisotropy at Strike-slip Boundaries
Despite similar surface transform faulting behavior, shear-wave splitting in the California (Ca) and New Zealand (NZ) plate boundary regions is markedly different. To better understand the origin of the anisotropy we model mantle flow and strain for a variety of strike-slip plate boundary scenarios. In our first models, simple relations between the flow or strain and elastic anisotropy are assumed to determine the integrated splitting in shear particle motion along teleseismic paths. Of these, strain-controlled models fit the observations in NZ and Ca better than simplified flow-controlled models. Fast shear polarizations are progressively rotated toward the shear plane over time, and even a constant viscosity model provides a good fit to the fast directions in NZ and southern Ca. The constant viscosity implies strong coupling between the surface and the deeper mantle. To fit the lack of decrease in delay times with distance from the fault, the relationship between delay time and strain must saturate at small strains. If we consider only directions, then southern NZ and southern Ca both fit models in which strain is smaller than in nearby regions, equivalent to that achieved along an infinite fault by about 3-10 My of their present motion. Stratified viscosity allows more rapid rotation of fast directions toward fault-parallel than occurs in isoviscous models, and can explain the nearly fault-parallel fast directions in the central South Island. Different aspects of the northern Ca results are fit with different models, but a rapid change in viscosity with depth is needed to produce the full effects of the behaviour previously modelled as two layers of anisotropy noted in the area, suggesting vertical decoupling. We are testing these conclusions with new models for NZ that allow for the evolution of temperature-dependent viscosity and for anisotropy to be determined using polycrystal plasticity theory. Preliminary modelling shows that increased thickness of the crust in the continental compared to oceanic regions localises the strain under the continental landmass, which may help to explain the large width of the shear zone in NZ.
T33E-02 14:00h
Stress and Crustal Anisotropy in Marlborough, New Zealand: Evidence for Low Fault Strength and Structure-Controlled Anisotropy
The major faults in Marlborough and Wellington are of both scientific and societal interest as they accommodate relative plate motion in the upper plate of an oblique subduction zone and pose a high seismic risk to central New Zealand. Studies in California suggest that some plate-bounding strike-slip faults are frictionally weak and that crustal anisotropy is controlled by the ambient stress. Whether these observations are more generally applicable to major strike-slip faults is yet to be determined. We have used inversions of focal mechanism and first motion data to calculate the principal stress directions and relate them to the geometry of the major faults. We have also conducted shear-wave splitting analysis on local S phases to determine the directions of crustal anisotropy and investigated their relationship to the geological fabric and the principal stress directions. The average angle between the axis of maximum horizontal compressive stress ($S_{Hmax}$) and the average strike of the major faults is $60\deg$; this is substantially higher than the $\sim 30\deg$ expected for an Andersonian strike-slip fault. This geometry can be explained, however, by the faults having a moderately low friction coefficient ($\sim$0.35) or moderately high fluid pressure ($\sim$0.7$\times$lithostatic). The anisotropy directions determined using shallow earthquakes reveal that the fast directions are aligned with the NE--SW-striking faults, and we therefore conclude that the anisotropy is mainly controlled by the geological fabric. The observation that faulting occurs at high angles to $S_{Hmax}$ substantiates the hypothesis that the San Andreas fault is not unique in being frictionally weak. Our shear-wave splitting calculations suggest that anisotropy in the crust varies spatially in regions of active faulting but that in Marlborough, at least, it is controlled more by the geological structures than the prevailing stress field.
T33E-03 14:15h
The seismic expression of deformation in the Australian lithosphere
The 3D azimuthal seismic anisotropy of the Australian lithosphere, as seen by multimode surface waves, shows a coherent alignment at depths below $\sim$150 km. Anisotropy above $\sim$150 km is related to the fossil strain field preserved in the relation of gravity anomalies to topography. The region below it, then, might reflect the active deformation of the upper mantle. We discuss four classes of models with increasing complexity. A first model of mantle strain is given by the direction of absolute plate motion. The alignment of the fast axes is better in the hot-spot reference frame than in the no-net-rotation frame. Our second class of models predict local instantaneous velocities from plate motions and/or from driving density anomalies inferred from 3-D tomographic models with radial viscosity structures. Third, we show modeling that uses the flow field to predict finite strain. The latter calculations are carried out in the no-net-rotation frame. We show results of strain calculations by backward advection over 10 Ma (constant time), and for logarithmic strain ratios of 0.5 (constant strain), a textural saturation level. None of the simulations in the no-net-rotation frame improve the fast axes alignment with absolute plate motion, whereas the velocities from the model with assimilated hot-spot referenced plate motions perform only slightly worse than the zeroth-order hypothesis. Such strain predictions, however, are subject to much uncertainty on the constitutive equations. In our last set of models, we focus more on the flow dynamics, by exploring the time-dependence of the alignment between the velocity field and the instantaneous strain field as yielded by three-dimensionsl convection calculations incorporating tomographic density anomalies and the history of plate motion. We are thus able to see (since) when and where upper mantle seismic anisotropy in Australia is likely to express mantle motion, by investigating the degree to which both predict each other. Varying the amount of core heating does not go unnoticed at the top of the mantle. The lithospheric seismic anistropy of our models is able to provide ``feasability'' constraints on such basic geodynamic parameters as the ratio of bottom to internal heating.
http://www.frederik.net
T33E-04 14:30h
Evidence of long-term weakness on faults in western North America from dynamic modeling
Dynamic models of the lithosphere resulting from thin sheet approximations provide estimates of the total strength of the lithosphere but only to a maximum thickness that is governed by the degree of mechanical coupling between rheologically stratified layers. The brittle-plastic transition (BPT) within the crustal portion of the lithosphere divides the crust into two distinct rheologic layers, a mechanically strong elastic upper crust and a mechanically weaker plastic lower crust. Over geologic time, the elastic upper crust is subject to deformation by brittle fracture and faulting while the lower crust beneath the BPT is ductile and flows plastically. The cut-off depth for shallow earthquakes occurring in tectonically active regions of upper crust often delineates the depth at which there is a transition from a brittle, seismically active elastic upper crustal layer to a plastic, aseismic lower crustal layer [\textit{Bonner et al.}, 2003]. These observations suggest that the BPT depth is highly variable, ranging from 7 to 40 km, where the actual depth at a given location is a consequence of the local fault regime and temperature gradient. Due to the large rheological contrast between the upper and lower crust, we regard the upper crust as sufficiently mechanically de-coupled from the lower crust such that the upper crust cannot sustain long-term shear stress at its base. In this study, we seek to match the long-term styles and directions of the deviatoric stress field associated with active faulting within the seismogenic layer of the plate boundary zone of western North America. We use a thin sheet approximation that neglects the small shear stresses at the base of the seismogenic layer as well as stresses due to flexure. We vertically integrate the force balance equations, incorporating spatially varying vertically integrated vertical stresses and stress field boundary conditions [\textit{Flesch et al.}, 2001], to provide estimates of the total strength of a variably thick upper crustal seismogenic layer. In addition, the BPT for these models are assumed to occur at crustal depths where the temperature ranges from 300 to 350$^\circ$C. Our BPT depths are based on interpolated geothermal gradients constrained by heat flow observations. Our modeling process begins by development of a long-term kinematic strain rate and velocity field model based on interpolation of Quaternary strain rates and to a lesser extent GPS velocity vectors. We assume that the relationship between deviatoric stress directions and kinematic strain rate directions is isotropic and the directions and style of the principal axes of kinematically defined strain rates are the appropriate stress field indicators. We assume that over geologic time the upper crustal elastic layer fails by Byerlee's rule [\textit{Kohlstedt et al.}, 1995] and that our estimates of the absolute magnitudes of the total strength of the upper crust, which range from $0.5$ -- $1.5$ $\times$ $10^1$$^2$ Nm$^-$$^1$ are supported by long-term values of the coefficient of friction ($\mu$). Initial results for western North America indicate that long-term values of $\mu$ that support the stresses in the upper crust are low, ranging from 0.05 to 0.45, under hydrostatic conditions. The disparity between $\mu$ values inferred from laboratory based experiments versus those inferred here suggest that $\mu$ may be lowered significantly during the rupture process, when most of the work is done within the seismogenic layer.
T33E-05 14:45h
Active Crustal Deformation in the Western US: A 3D Finite Element Model
Gravitational potential energy has been suggested as the major cause of the diffuse crustal deformation in the western United States through the late Cenozoic, whereas recent geodesy (mainly GPS) measurements indicate strong influence of plate boundary force on the active deformation. Some studies have also emphasized the role of traction at the base of the lithosphere. We have developed a 3-dimensional finite element model to assess the different roles of these driving forces. The model rheology is that of a temperature-dependent power-law fluid. A series of experiments have been conducted to explore the effects of the various driving forces, the heterogeneity of rheology structure, and the boundary conditions. Major constraints for the model include the strain rate field derived from the GPS data, and stress state in the crust from earthquake data and other data from world stress map. Our results indicate that much of stress patterns in the western US are consistent with the dominance of the gravitational buoyancy forces; the effects of plate boundary forces along the San Andreas fault are largely limited to the region within a couple of hundred kilometers of the plate boundary. The present surface velocity field and stress states can be explained by the combined influences of gravitational buoyancy force and plate boundary force without appealing to basal shear. The model predicts concentration of strain energy in the western Basin and Range province, consistent with high seismicity in this region. Fitting the GPS data from the Cascadian region requires strong coupling between the Juan de Fuca plate and North America. With the timescale-dependent crustal rheology, the model predicts short-term (interseismic) NE-SW shortening as shown by the GPS data, and long-term N-S compression in the northwestern Pacific region consistent with earthquake data.
http://www.geongrid.org
T33E-06 15:00h
Two and Three Dimensional Simulations of Thick Skin Continental Deformation : Exploring the Influence of Rheology on Localization Patterns
We present simple two and three dimensional models of extended and sheared thick-skin models of continental lithosphere. We examine the role of rheological layering, yield model (von Mises v. Mohr-Coulomb), and strain softening model (isotropic / anisotropic) on the pattern of surface deformation and the distribution of deformation with depth around localized zones. We examine the conditions required to develop throughgoing "faults" versus decoupled crust/mantle deformation in the large deformation limit and related the evolution of the deformation planform from the pristine state to the fully evolved state to the thermal/mechanical parameters. Models are run using a Lagrangian Integration Point finite element method which tracks material strain and damage at each material through the entire simulation. A variety of materials of distinct rheology can coexist in the mesh with implicit no-slip boundary conditions between domains composed of each material
T33E-07 15:15h
Interpreting Crust and Mantle Stress and Strain Indicators at Yellowstone
Results from several seismic and geodetic studies of the Yellowstone hotspot crust and mantle are integrated to derive a consistent interpretation. Mantle seismic tomography and anisotropy were determined using data recorded at two temporary arrays of IRIS-PASSCAL seismographs deployed in a 400 km (NE-SW) by 500 km (NW-SE) area centered on Yellowstone, combined with data from the U.S. National Seismograph Network and the University of Utah Seismograph Stations' permanent network. Teleseismic tomography reveals a low velocity anomaly in the mantle beneath the 0.63 Ma Yellowstone caldera to a depth of 200 km of up to -2.3% V$_{P}$ and -5.5% V$_{S}$. The low velocity zone is elongated NE-SW, parallel to the direction of absolute plate motion (APM), from the edge of the tomographic model in the SW to 75 km beyond the caldera to the NE. Anisotropic seismic fast directions from teleseismic shear-wave splitting are generally parallel to the direction of APM, but rotate up to 80\deg from APM at stations within and adjacent to the caldera. The splitting fast axes are inconsistent with olivine lattice preferred orientation due to a broad parabolic flow pattern that might be predicted for a hotspot plume at the base of a moving plate. Instead, stress-oriented, partial melt-filled lenses in the lithosphere could be responsible for splitting fast axes that are perpendicular to the plate motion. Focal mechanisms of local earthquakes, which occur in the upper 10 km of the crust, were used to estimate the stress field orientation. Local earthquake data have also been used to make new shear wave splitting measurements. The crustal splitting measurements are consistent with anisotropy due to stress-oriented crustal cracks. The direction of minimum horizontal stress is N-S in the E-W band of high seismicity north of the caldera, but rotates to NE-SW in the center of the caldera. GPS-derived strain is also consistent with this N-S to NE-SW rotation. Yellowstone is at the NE edge of the Basin and Range province, which is undergoing NE-SW extension in that area. Continued NE migration of the hotspot, as indicated by the mantle low velocity zone, may be driving localized expansion of the extensional regime. The rotation of crustal stress and strain at Yellowstone may reflect the NE migration of the edge of the Basin and Range.