T44B-01
Structure and Composition of the San Andreas Fault at Seismogenic Depths: Recent Results from the SAFOD Experiment
The San Andreas Fault Observatory at Depth (SAFOD) was drilled into the San Andreas Fault Zone to study the physics of earthquake nucleation and rupture and determine the composition, physical properties, and mechanical behavior of an active, plate-bounding fault at seismogenic depths. SAFOD is located 10 km NW of Parkfield, CA, and penetrates a section of the fault that is moving through a combination of repeating microearthquakes and fault creep. During Phases 1 and 2 in the summers of 2004 and 2005, SAFOD was drilled vertically to a depth of 1.5 km and then deviated to penetrate the active San Andreas Fault Zone at a vertical depth of about 2.7 km. During Phase 3 in the summer of 2007, cores were acquired from holes branching off the main SAFOD borehole to directly sample fault and country rocks at depth. Geophysical logs and cuttings analyses conducted during Phases 1 and 2 define the San Andreas Fault Zone to be relatively broad (~250 m), containing several discrete, highly localized zones only 2-3 m wide that exhibit very low P- and S-wave velocities and low resistivity. Two of these zones have progressively deformed the cemented casing at measured depths of 3194 m and 3301 m, indicating that they are actively creeping shear zones. These active shear zones were targeted for coring during Phase 3. The 3194 m casing deformation zone lies ~100 m above a cluster of repeating M2 earthquakes that form the southwestern boundary of the creeping and microseismically active San Andreas Fault Zone. Casing deformation is most pronounced across the 3301 m zone; hence this zone is believed to accommodate most of the current creep deformation across the San Andreas Fault at this location. During Phase 3 we have obtained core from just outside the geologically defined San Andreas Fault Zone, at the boundary between the Salinian and Great Valley/Franciscan terranes, and from the active deformation zones at 3194 and 3301 m. The cores obtained from these deformation zones exhibit a variety of features indicating pronounced strain localization and probably marked weakening. These include highly sheared and foliated shales and siltstones, cataclasites, veined serpentinite and chert bodies, and compacted, cohesionless serpentine-bearing fault gouge, all juxtaposed in shear zones only a few meters wide. The occurrence of serpentinite is particularly significant, because serpentine and related minerals are widely regarded to be important in controlling frictional strength and the stability of sliding. The Phase 3 core samples will be extensively tested in the laboratory to study the composition, deformation mechanisms, physical properties and rheological behavior of fault rocks from the active traces of the San Andreas Fault at realistic in-situ conditions. http://www.earthscope.org/
T44B-02
Geodetic and Geologic Constraints on the Distribution of Fault Slip-Rates in Southern California
The spatial partitioning of deformation in the continental crust and, in particular, at plate boundary zones is determined by the distribution of fault slip-rates. Analytic and numerical models of strain accumulation in the elastic upper crust have been divided into those that parameterize faulting as localized on a finite length fault system comprised of relatively few fast slip-rate faults, or as distributed throughout a continuum of relatively slow slip-rate faults. We use both geodetic and geologic data to estimate the frequency distribution of fault slip rates, per unit length, between the Pacific and North American Plates in the southern California fault system. Using these empirically constrained scaling relationship we calculate the partitioning of potency accumulation rate, which determines the distribution and magnitude of slip localization. These model describes the kinematics of both micro-plate and continuum deformation models, and predicts that 97% of the deformation in Southern California is accommodated on faults slipping at >1 mm/yr.
T44B-03
Geometrical and Structural Asperities on Fault Surfaces
Earthquake dynamics are strongly affected by fault zone structure and geometry. Fault surface irregularities and the nearby structure control the rupture nucleation and propagation, the fault strength, the near-field stress orientations and the hydraulic properties. New field observations demonstrate the existence of asperities in faults as displayed by topographical bumps on the fault surface and hardening of the internal structure near them. Ground-based LIDAR measurements on more than 30 normal and strike slip faults in different lithologies demonstrate that faults are not planar surfaces and roughness is strongly dependent on fault displacement. In addition to the well-understood roughness exemplified by abrasive striations and fracture segmentation, we found semi-elliptical topographical bumps with wavelengths of a few meters. In many faults the bumps are not spread equally on the surface and zones can be bumpier than others. The bumps are most easily identified on faults with total displacement of dozens to hundreds of meters. Smaller scale roughness on these faults is smoothed by abrasive processes. A key site in southern Oregon shows that the topographic bumps are closely tied to the internal structure of the fault zone. At this location, we combine LiDAR data with detailed structural analysis of the fault zone embedded in volcanic rocks. Here the bumps correlate with an abrupt change in the width of the cohesive cataclasite layer that is exposed under a thin ultracataclasite zone. In most of the exposures the cohesive layer thickness is 10-20 cm. However, under protruding bumps the layer is always thickened and the width can locally exceed one meter. Field and microscopic analyses show that the layer contains grains with dimensions ranging from less than 10 μ up to a few centimeters. There is clear evidence of internal flow, rotation and fracturing of the grains in the layer. X-Ray diffraction measurements of samples from the layer show that the bulk mineralogy is identical to that of the host rock, although thin section analysis suggests that some alteration and secondary mineralization of the grains also occurs. We infer that the cohesiveness of the layer is a consequence of repacking and cementation similar to deformation bands in granular material. By comparing the thickness of the cohesive layer on several secondary faults in this fault area we found that the average thickness of the layer increases with total slip. The correlation is nonlinear and the thickening rate decreases with increasing slip. We conclude that granular flow decreasing with increasing slip and thus the deformation is continually localized.
T44B-04
Frictional Melting of Syenite in High-velocity Rotary Shear Experiments: `Mingling' of Molten Layers and Slip Localization
We report a mingling behavior of granitic and andesitic melts produced by frictional heating in simulated faults of a syenite using a high-velocity rotary shear testing apparatus at Kyoto University. The syenite (so-called larvikite) is composed of perthitic alkali feldspar (~80%), mafic minerals (~15%), apatite (2 - 3%) and nepheline (1 - 2%). The mafic minerals (clinopyroxene, olivine, titanomagnetite and biotite) and apatite occur as aggregates, although there are some isolated smaller grains of titanomagnetite, biotite and apatite within feldspars. The friction experiments were conducted at seismic slip rates of 0.31 to 1.13 m/s, normal stresses of 5.2 or 18.3 MPa and at room temperature. At the onset of the shearing, the first peak friction (μ = 0.35 - 0.66) was followed by a transient slip weakening. Then with further slip, the friction coefficient increased to a second peak value (0.25 - 0.77) followed by a final weakening with steady-state friction coefficient of 0.15 - 0.35. We also terminated the shearing experiments during the first transient weakening and second strengthening to observe microstructural developments, and found that the first transient weakening and second strengthening correspond to gouge-generating wear and patchy melting, respectively. The final weakening was induced by the development of molten layer(s) along the slip zone. These mechanical and microstructural evolutions are similar to those of high-velocity friction experiments on gabbro (Hirose and Shimamoto, 2005, JGR, B05202). The ˇ®bulk' temperature of the slip zone measured by a radiation thermometer reached up to about 1250°C. However, partially molten olivine (Fo40), clinopyroxene (augite) and titanomagnetite in pseudotachylyte (PT) band suggest that the local temperature was at least about 1500°C. The PT band along the slip zone consists of two parallel glass layers, one granitic (25 - 55 μm thick) and the other andesitic (25 - 100 μm thick) layers, or three glass layers in which the andesitic layer is mantled by granitic layers. The granitic glass was produced probably by friction between feldspars, while the andesitic glass was likely to be formed by friction between feldspars and mafic minerals. With the layered configuration of molten materials with different viscosity, we suspect that slip might be localized along the less viscous andesitic layer. However, it is not clear at present why the two molten layers were juxtaposed rather than mechanically mixed at seismic slip rates.
T44B-05
Extreme grain size reduction in dolomite: microstructures and mechanisms.
Pure dolomite sample were deformed at room temperature and under a variety of confining pressures (0 - 100MPa) to examine the processes of grain size reduction. The dolomite is composed of > 97 vol. % dolomite with accessory quartz, calcite, tremolite, and muscovite and has been metamorphosed to amphibolite facies and subsequently annealed. At the hand sample scale, the rock is isotropic, except for minor, randomly oriented tremolite porphyroblasts, and weakly aligned muscovite. At the thin section scale, coarser grains have lobate grain boundaries, exhibit minor to no undulose extinction and few deformation twins, although well- developed subgrains are present. Growth twins are common, as is the presence of well developed {1011} cleavage. Mean grain size 476 microns, and porosity is essentially zero (Austin and Kennedy, 2006). Samples contain diagonal to subvertical faults. Fractures are lined with an exceptionally fine-grained, powdered dolomite. Even experiments done at no confining pressure and stopped before sliding on the fracture surfaces occurred had significant powdered gouge developed along the surfaces. In this regard, fracturing of low porosity, pure dolomite, with metamorphic textures (e.g. lobate, interlocking grain boundaries) results in the development of fine-grained gouge. As expected the dolomite exhibited an increase in strength with increasing confining pressure, with a maximum differential stress of ~400MPa at 100 MPa confining pressure. At each chosen confining pressure, two experiments were performed and stopped at different stages along the load-displacement curve: just before yield stress and at peak stress. Microstructures at each stage were observed in order to determine the possible mechanisms for extreme grain size reduction. SEM work shows that in samples with little to no apparent displacement along microfractures, extreme grain size reduction still exists, suggesting that frictional sliding and subsequent cataclasis may not be the mechanism responsible for grain size reduction. Within individual dolomite clasts, apparent Mode I cracks are also lined with powedered gouge. Alternative mechanisms for grain size reduction are explored. Austin et al. 2005, Geological Society, London, Special Publications, 243, 51-66.3.
T44B-06
A theoretical model of grainsize evolution during deformation
Lithospheric shear localization, as occurs in the formation of tectonic plate boundaries, is often associated with diminished grainsize (e.g., mylonites). Grainsize reduction is typically attributed to dynamic recrystallization; however, theoretical models of shear-localization arising from this hypothesis are problematic since (1) they require the simultaneous action of two exclusive creep mechanisms (diffusion and dislocation creep), and (2) the grain-growth ("healing") laws employed by these models are derived from static grain-growth or coarsening theory, although the shear-localization setting itself is far from static equilibrium. We present a new first-principles grained-continuum theory which accounts for both coarsening and damage-induced grainsize reduction. Damage per se is the generic process for generation of microcracks, defects, dislocations (including recrystallization), subgrains, nucleii and cataclastic breakdown of grains. The theory contains coupled statistical grain-scale and continuum macroscopic components. The grain-scale element of the theory prescribes both the evolution of the grainsize distribution, and a phenomenological grain-growth law derived from non-equilibrium thermodynamics; grain-growth thus incorporates the free energy differences between grains, including both grain-boundary surface energy (which controls coarsening) and the contribution of deformational work to these free energiesconservation and positivity of entropy production provide the phenomenological law for the statistical grain-growth law. We identify four potential mechanisms that affect the distribution of grainsize; two of them conserve the number of grains but change their relative masses and two of them change the number of grains by sticking them together or breaking them. In the limit of static equilibrium, only the two mechanisms that increase the average grainsize are allowed by the second law of thermodynamics. The first one is a diffusive mass transport from small grains to large grains that captures the essential component of normal grain-growth theories. The second one is the aggregation of grains. With the inclusion of nonstatic/nonequilibrium conditions, the theory predicts two mechanisms for which the thermodynamic requirement of entropy positivity always imposes large grains to shrink and small ones to grow. A first damage mechanism opposite to diffusion, tends to homogenize the distribution of grainsize around its initial average grainsize. A second damage mechanism favors the creation of small grains by division of larger grains. This second mechanism reduces the average grainsize with time. The competition between the two coarsening mechanisms, the damage-induced grain homogenization and the damage-induced grain reduction controls the general evolution of the grainsize population. Under uniform and constant shear, and assuming than grains cannot stick together, the average grainsize is related to the energy dissipated during the deformation (in agreement with Austin and Evans, 2007). The prediction of this theory fits satisfactorily with the available observations and we will discuss the implications of this model for large scale geodynamics.
T44B-07
Transitions in convective behavior as a function of damage: an explanation for the difference between Earth and Venus
The generation of plate tectonics from mantle convection requires shear localization in order to form narrow, weak zones that separate the broad, strong plate interiors. The existence of plate tectonics on Earth and its absence on the other terrestrial planets remains a significant conundrum for geophysicists. Two-phase damage theory provides a theoretical framework to describe the failure and weakening that leads to shear localization by allowing for the development of damage to be manifested in two distinct ways: void generation associated with dilation of the matrix and increasing the fineness of the mixture (e.g. grain size reduction). This work will examine the application of two-phase damage theory in two-dimensional convection simulations to model mantle convection. We find that by varying the healing rate for grain-growth the convective model undergoes significant transitions in convective style; by increasing the healing rate for grain growth the system will transition from stable plate-like behavior to episodic behavior, and further increases in healing rate eventually force the system to stagnant-lid behavior. The various states of convection observed in our simulations are similar to the different modes of convection seen in our solar system: stagnant lid (Mars), episodic (possibly Venus), and stable plate- like (Earth). The numerical experiments are used to test a simple conceptual model whose underlying hypothesis is that the transitions are due to grain size variations controlling subduction lubrication, hence either allowing for unimpeded subduction or the cessation of subduction altogether. Finally, we propose a model for the generation of plate tectonics that explains the convective transitions by coupling a planet's lithosphere and atmosphere/ocean climate system. The model suggests a criteria to determine if a planet will develop Earth-like plate tectonics.
T44B-08
Heterogeneity and Shear Localization in Natural Systems With Emphasis on the Interplay of Inelastic and Plastic Processes
The ubiquity of transient (cyclic), heterogeneous deformation, notably shear localization, in the rock record is such as to characterize it as the anticipated 'normal' behaviour. The corollary is that steady, homogeneous deformation is rare, and where approached must reflect some special set of conditions that are not representative of the general case. An issue central to natural deformation is then not the existance of localized strain, but rather how the extant deformation processes scale over tectonic phenomena and in turn organize to enable a coherent descripion of earth deformation. The strain energy distributions which drive thermo- mechanical responses are in the first instance established at the grain-scale where the non-linear interaction of micromechanical processes introduces fundamental, heterogeneous behaviour described by various gradient theories, and evidenced by the defect microstructures of deformed rocks. Hence, the potential for non-uniform response is embedded within even quasi-uniform, monomineralic materials, as seen in the spatially discrete evolution of dynamic recrystallization. Despite the interest in the latter purely mechanical localization, the more common type of heterogeneity promoting localization comprises mechanically contrasting materials typical of most rocks. Natural shear localization commonly demonstrates a cyclic interplay between inelastic rupture and subsequent plastic material softening resulting from the concomitant introduction of exogenous material in the form of igneous melts, deformation-induced melts and fluid precipitates (veins). Together this 2-stage process reflects localization and stabilization of the shear phenomena. The cyclicity of this behavior indicates that material hardening and non-associated flow over some characteristic time is a precursor to localized instability, with stabliization of localized shear correlated with system softening tied to redistribution of strain energy dissipation within what is effectively a reconstituted material. The resultant spectrum of mechanical responses observed at all lithospheric levels has the effect of masking brittle (seismic?) records. Fluctuations in rock composition, pressure and temperature and fluid activity with crustal level produce 'windows of transience' which are particularly amenable to this behavior.