T41B-0570
Annealing Versus Deformation: How can we Track Down the Spatial Evolution of Localization Structures?
Shear zones often show long lasting deformation histories incorporating variations in the spatial and temporal distribution of strain. Based on seismic activity of shear zones, we know that localized deformation can occur episodic and may also vary spatially. In case of exhumed natural shear zones, the preservation of the corresponding microstructures might be problematic, since deformation under retrograde conditions and/or post- tectonic annealing often overprint the prior deformation stages. This is particularly the case for monomineralic fabrics, where only slight changes in the physical conditions, allow diminishing older structures. In this study, we present an approach using polymineralic carbonate mylonites with various contents of second phase minerals. Since, in second-phase controlled microstructures the calcite grain size hardly changes during annealing due to pinning, the microstructures in pure calcite aggregates are able to adapt their fabrics to the new conditions. In this way the Zener relation, i.e. the relation between matrix grain size and second phase size and volume fraction, allows an unambiguous discrimination between annealing and deformation structures. Based on the different microstructural evolution within a vertical drill core across one of the major thrust systems in the Helvetic nappes (Doldenhorn nappe, Swiss Alps), we present the variation in localization structures. It becomes evident, that with decreasing temperature, strain continuously localizes into very narrow zones focusing more and more at the thrust contact. At the same time, the older high temperature deformation fabrics evolved at farther distances from the active more localized zone are affected by annealing. In these annealed deformation zones, the calcite grain size and consequently the degree of grain growth does not increase linearly with distance to the thrust contact but rather shows abrupt changes. Particularly this lateral occurrence of finer grained but annealed thrust parallel zones suggest that deformation did localize episodically at different levels within the shear zone before becoming overprinted by annealing.
T41B-0571
Strain Localization on Different Scales and their Related Microstructures - Comparison of Microfabrics of Calcite Mylonites from Naxos (Greece) and Helvetic Nappes (Switzerland)
In the upper crust, shear zones are widespread and appear at different scales. Although deformation conditions, shear zone history, and displacements vary in time and space between shear zones and also within them, in all shear zones similar trends in the evolution of large- to micro-scale fabrics can be observed. The microstructural analyses of calcite mylonites from Naxos and various Helvetic nappes show that microstructures from different metamorphic zones vary considerably on the outcrop- and even on the sample- scale. However, grain sizes tend to increase with metamorphic degree in case of Naxos and the Helvetic nappes. Although deformation conditions (e.g. deformation temperature, strain rate, and shear zone geometry, i.e. shear zone width and rock type above/below thrust) vary between the different tectonic settings, microstructural trends (e.g. grain size) correlate with each other. This is in contrast to many previous studies, where no corrections for second phase contents have been applied. In an Arrhenius-type diagram, the grain growth trends of calcite of all studied shear zones fit on a single trend, independent of the dimensions of localized large-scale structures, which is in the dm to m- and km-range in case of the Helvetic thrusts and the marble suite of Naxos, respectively. The calcite grain size increases continuously from few μm to >2mm with a temperature increase from <300°C to >700°C. In a field geologist's point of view, this is an important observation because it shows that natural dynamically stabilized steady state microfabrics can be used to estimate temperature conditions during deformation, although the tectonic settings are different (e.g. strain rate, fluid flow). The reason for this agreement might be related to a scale-dependence of the shear zone dimensions, where the widths increase with increasing metamorphic conditions. In this sense, the deformation volumes affected by localization must closely be linked to the strength of the affected rocks. In comparison to experiments, similar microstructural trends are observed. Here, however, shifts of these trends occur due to the higher strain rates.
T41B-0572
Experimental Insights on Grain Growth Under Static Conditions Affected by Second Phases
Mineral composition and grain size are two important parameters for the characterization of the microstructure of a rock and they both are closely linked to each other. Under static conditions, the grain size of the matrix phase is mainly controlled by the size, amount and dispersion of the second phases (Zener relation). In such polyphase systems, the second phases can pin or drag the migrating grain boundaries reducing the grain growth kinetics. Due to the fundamental influence of the interaction by different mineral phases and the fact that most natural rocks are of polymineralic composition, it is essential to understand the processes responsible for the formation of a rock's microstructure in such polymineralic systems. To gain new insights, we performed in-situ rock analogue experiments using rigid non-reacting phases (micro glass beads) as second phase to reduce on one hand the complex influence of a growing second phase, and on the other to focus on the behavior of the matrix grains. The second phase was added in different volume percentages (0-33vol%) to the matrix phase (norcamphor) to investigate the microstructural evolution and related growth kinetics in regard of different second phase contents. The results indicate that all mixtures undergo two main stages during the microstructural evolution: (a) a stage of continuous grain growth, which is followed by (b) a stage of stabilized grain sizes. With increasing amounts of second phases a well defined transition between (a) and (b) evolves, which increases with increasing second phase content. Furthermore, the amount of second phases is inversely proportional to the matrix grain size for all microstructural stages mentioned above. For continuous grain growth, this behavior can be expressed in form of a conventional grain growth law: Gn - G0n = k (t - t0), where G is the main grain size at time t and G0 the grain size at t=t0. The grain growth exponent n, decreases from 4.6 to 2.8 for pure and impure (33vol% micro beads) samples, respectively, while the constants k show an increase with enhanced second phases contents. Based on these results, the second phase effect on grain growth of a matrix phase can be integrated into the grain growth law, allowing more realistic modeling of polymineralic systems in nature.
T41B-0573
Constraints on Coupled Microprocesses, Fluid Flow, and Rheological Evolution in a Syn- metamorphic Shear Zone
Microstructural and petrologic data gathered across the sheared margin of the Lincoln Syenite in south-central Maine, USA are used here to evaluate processes involved in shear-zone initiation and development. These data suggest that heterogeneous plastic strain and fluid infiltration enhanced metamorphic reactions, and determined the degree to which textural and chemical reorganization were achieved. Within the study area, a transition from a coarse-grained, granular, igneous rock (Cpx+Opx+Ksp+Bt) to a recrystallized and strongly foliated rock (Act+Ksp+Bt+Qz) is preserved both texturally and petrologically across intermediate domains. Systematic variations in biotite composition track the progress of the bulk reaction, and show highest OH/(F+Cl) ratios in the most strongly foliated domain, where recrystallization is complete. Reaction progress calculations for each of the progressively sheared and recrystallized domains are used to constrain the volume of fluids which moved through the system. Calculated values are incorporated into numerical models that investigate the evolution of permeability and the distribution of stresses within the developing structural fabric, possibly leading to pumping or channeling of fluids during progressive shearing. The effect of fluids on deformation mechanics and metamorphic reactions are of primary importance in understanding rheological changes in crustal rocks during deformation, and are the focus of this study. The textural and mineralogical changes discussed above should have significant implications for the mechanical behavior of the transformed volume. The fine-grained, interconnected product phases, biotite, actinolite, and quartz, within the developing matrix foliation comprise weakened domains between feldspar megacrysts along which strain may be localized. Numerical modeling and experimental studies indicate that these types of processes can cause partitioning of higher strain-rates into biotite-rich, foliated zones, and result in a drastic weakening of the bulk rock. Thus, we suggest that the bulk effective viscosity of the Lincoln Syenite was decreased substantially along its southeastern margin through strain-enhanced reactions and coalescence of elongate biotite and actinolite grains and grain aggregates.
T41B-0574
Viscosity Contrast between Gabbro and Peridotite: A Case Study From the Oman Ophiolite
The viscosity contrast between the lower crust and upper mantle controls key tectonic processes such as lithospheric coupling between the crust and mantle and the origin and transmission of plate driving forces. While experimental data provide a basis to predict viscosity contrast, pertinent geologic observations of synchronously deformed gabbro and peridotite are not well documented. We are investigating this problem by analyzing microstructures in deformed gabbronorite dikes (~ 10-50 cm) and host harzburgite from the Oman ophiolite. Outcrop scale observations demonstrate that deformation is highly localized in 30-50 mm wide mylonitic shear bands within the gabbronorite. We constrained the conditions of deformation using pyroxene thermometry, petrology, grain size piezometry and EBSD analyses of lattice preferred orientation (LPO). Syn-deformation temperature ranged from 800-850 °C, based on Fe-Mg-Ca exchange between recrystallized orthopyroxene- clinopyroxene pairs using QUILF (Andersen et al., 1993), and pressure was ~ 300 MPa based on structural thickness of the section. Interpretation of plagioclase grain size indicates stress was ~ 100 MPa during localized, low T deformation in gabbronorite shear bands. In contrast, olivine grain size in peridotite corresponds to a stress of ~ 10 MPa, probably during earlier, high T, distributed deformation. Plagioclase exhibits a weak LPO throughout the analyzed samples when compared to fabrics observed in mylonitized gabbros from the Southwest Indian Ridge that deformed at similar temperatures (Mehl and Hirth, in press). Plagioclase fabric strength does not vary significantly with variations in grain size (from ~ 20–50 μm) or presence of second phases (e.g. amphibole). These observations suggest that deformation of the gabbronorite involved a significant component of diffusion creep. Olivine fabrics in the harzburgite, indicative of deformation by dislocation creep, are stronger than the associated plagioclase fabrics, and are consistently oriented throughout the field area with the [100] axis at ~ 45° to the shear plane/dike orientation. Several observations indicate that deformation occurred at a relatively high water content. In the finest grained regions, amphibole appears to be a participant in deformation and plagioclase has a high anorthite content (An97) indicative of high water content in the magma. Our observations indicate that gabbroic rocks were weaker than mantle peridotite during low T hydrous deformation.
T41B-0575
Strain weakening versus strain localization in olivine aggregates
Evidence of strain localization in olivine aggregates has been observed in both obducted slices of upper mantle and in high stress experiments (Post, 1977). Most experimental deformation studies of olivine aggregates have been done in gas apparatus at low P (<300 MPa) and thus at low flow stress; both axial compression and shear experiments involving climb-accommodated dislocation creep show steady state flow, with little if any strain weakening. However, experimental studies of quartz and feldspar at high P have shown the existence of a low T, high stress dislocation creep regime in which climb is very limited and creep is accommodated by bulging recrystallization; in this regime extreme strain weakening and/or localization occur. A series of deformation experiments were performed natural and synthetic aggregates using a modified molten salt assembly in a Griggs apparatus at a P of 1.5 GPa, T of 950 to 1200°C, and strain rates of 5x10-5/s to 10-6/s, in both axial compression and general shear. The starting material was either Balsam Gap dunite (d~500 μm) or synthetic aggregates hot pressed from San Carlos olivine powders (~20 μm). Prior to weld sealing in Ni and outer Pt, each sample was dried for 24 hrs at 900°C in a CO/CO2 atmosphere. Shear experiments were performed using three types of shear pistons: Balsam Gap dunite, alumina and cores from single crystals of San Carlos olivine oriented with (010) perpendicular to ƒã1. The peak stresses of experiments ranged from 1380 MPa to 250 MPa and all experiments underwent some degree of strain weakening. Deformation mechanisms observed in samples from axial compression experiments or shear experiments with alumina or San Carlos single crystal shear pistons can be characterized by three behavioral regimes. Microstructures in these samples were homogeneous and no strain was accommodated by the shear pistons. Samples from experiments deformed the highest stresses contain very high dislocation densities, cracks and few recrystallized grains, indicating the dominant deformation mechanism is semi-brittle flow. The samples from intermediate peak stress experiments (350-780 MPa) contained porphyroclasts with rims of recrystallized grains (2-5 μm) and the porphyroclasts have high densities of tangled dislocations and serrated grain boundaries. The fine grains at the edges of porphyroclasts have few to no dislocations. These microstructures indicate the intermediate samples are deforming by recrystallization accommodated dislocation creep. At the lowest stresses, the remaining porphyroclasts contain subgrains, have very low dislocation densities and the recrystallized grains are considerably larger (10-25 μm), indicating deformation by climb-accommodated dislocation creep. In the experiments which had Balsam Gap shear pistons with a fine-grained (20 μm) layer of San Carlos powders, at high stresses (550 MPa) recrystallization accommodated dislocation was the dominant deformation mechanism creep and strain was localized within the fine grained layer. However, at low stresses (250 MPa), both the olivine in the shear pistons and fine-grained layer deformed by climb-accommodated dislocation creep and strain was homogeneous throughout the sample. These results indicate that all olivine aggregates undergo some degree of strain weakening during shear deformation, but strain localization only occurs in olivine aggregates deforming by recrystallization-accommodated dislocation creep when a fine-grained zone exists prior to deformation. Therefore, heterogeneities in the grain size or different deformation mechanisms may be required to form shear zones in the mantle.
T41B-0576
Structural Analysis of the Exhumed SEMP Fault Zone, Austria: Towards an Understanding of the Mechanics of Shear Zone Localization
One of the most exciting frontiers in earthquake science is the linkage between the internal structure and mechanical behavior of fault zones. Little is known about how fault-zone structure varies as a function of depth, yet such understanding is vital if we are to understand the mechanical instabilities that control the nucleation and propagation of seismic ruptures. This has led us to the Oligo-Miocene Salzach-Ennstal-Mariazell-Puchberg [SEMP] fault zone in Austria, a major left-lateral strike-slip fault that has been exhumed differentially such that it exposes a continuum of structural levels along strike. In order to establish the structure of this fault zone, we are studying outcrops at a variety of exhumation levels, from <1 km near the eastern end of the fault, downward through the seismogenic crust, across the brittle- ductile transition, and into the uppermost part of the lower crust in western Austria. Here we present new results and discuss the mechanical implications of these new data from two key outcrops at Gstatterboden and Taxenbach, where the SEMP has experienced 40-60 km of displacement. The outcrop at Gstatterboden has been exhumed from 2-3 km depth. Here the SEMP juxtaposes limestone of the Wettersteinkalk on the south with dolomite of the Ramsaudolomit on the north. Faulting has produced extremely asymmetric damage, extensively shattering and shearing the dolomite while leaving the limestone largely intact. We interpret this brittle damage using both mesoscopic calculations of damage intensity and microscopic grain size distribution analysis, which suggests that shear has localized to a zone approximately 10 m wide. These findings are compared to the brittle-ductile outcrop at Taxenbach, which has been exhumed from depths of up to 10 km. Here, the SEMP juxtaposes Greywacke Zone rocks with carbonate mylonites of the Klammkalk. Microstructural observations of grain size and lattice preferred orientation suggest a marked increase in strain within 100 m of the lithologic contact, although the first-order observation is that most of the 60 km of total displacement has localized along the contact between these units, suggesting that to first order, strain at the base of the brittle-ductile transition is extremely localized.
T41B-0577
Shear Band Genesis and Frictional Properties, McKinleyville Fault, Humboldt County, California
Understanding the genesis of shear bands and fabric development in rock and sediment is essential to understanding fault formation, propagation, and earthquake mechanics. Previous laboratory experiments and field evidence indicate that shear strength, velocity-dependence of friction, and other host rock properties play key roles in shear band formation. However, we lack laboratory data on undisturbed, naturally formed shear bands. Deformation bands in the footwall of the active McKinleyville thrust fault provide an excellent opportunity to study the effects of shear strain and displacement in natural samples. Samples consist of deformation bands roughly parallel to the McKinleyville Fault, conjugate deformation bands and non-banded material. Shear bands are hosted in late Pleistocene nearshore marine sand and are identified as compacted sand within surrounding less-compacted sand. Shear bands range in thickness from 0.1cm- ~10cm and band parallel displacement from 0.4cm-3.35m. We report on laboratory experiments contrasting displacement and shear band thickness versus particle characteristics, shear strength and velocity-dependence of friction. Experiments were conducted in a servo- controlled apparatus using the single-direct shear configuration. Layers were initially 5 mm thick with nominal contact dimensions of 5 cm x 5 cm. Normal stress was varied from 0.25-2.5 MPa. Velocity stepping experiments were done in an increasing fashion at 1, 3, 10, 30, 100 and 300μm/s. Post shear samples are analyzed under transmitted and SEM light microscopy. Through laboratory experiments we aim to resolve the physical alteration of natural material due to shear by comparing natural shear-band material to laboratory simulated shear bands of non-banded material. Preliminary results on non-banded material yield an average angle of internal friction of 30.4° and cohesion of 93.1 kPa.
T41B-0578
Effect of Brucite on the Frictional Strength of Serpentinite
Brucite is a common by-product of the serpentinization of olivine-rich ultramafic rocks. Brucite is considerably weaker than the serpentine mineral antigorite; in the T range 25-400°C, antigorite serpentinite has a coefficient of friction of 0.5-0.6 whereas that of brucite is 0.2-0.3. We investigated the possible weakening effect of brucite on serpentinite, because of the close association of serpentinite with faults in a variety of plate-tectonic settings. Frictional strength experiments were run on mixtures of brucite and an antigorite-rich serpentinite. Experiments were run at 100 MPa effective normal stress, comparing behavior at 25 and 300°C. Adding small amounts of brucite to serpentinite gouge dramatically lowers its residual frictional strength. At both temperatures, 5 wt% brucite reduces the coefficient of friction of serpentinite by about 25% towards that of the brucite end member, and serpentinite containing 50 wt% brucite has essentially the strength of pure brucite. However, at 300°C, gouges containing 65-85% brucite that were sheared at 0.01 μm/s axial displacement rate were stronger for a given amount of displacement and showed more pronounced stick-slip motion than gouges containing 50% and 100% brucite that were sheared at the same rate. Both the size of the stress drops and the shear strength decreased with increasing displacement for the gouges with 65-85% brucite, but strength was still decreasing when the experiments ended at about 3-mm axial displacement. At faster shearing rates of 0.1 and 1.0 μm/s, gouge containing 75% brucite reached its residual strength between 1 and 1.5 mm axial displacement and showed a reduced tendency for stick-slip behavior. The antigorite-rich samples slid stably at 300°C and 0.01 μm/s axial displacement rate. Antigorite and brucite both have platy habits, and shear of serpentinite gouge is highly localized along shears in which the platy grains are oriented parallel to the shear surfaces. SEM examination of gouge samples containing 5-50% brucite indicates that brucite is concentrated on the shear surfaces, particularly the R shears. The shears may therefore form in a manner that maximizes the proportion of the weaker mineral within them. The difference in strength between samples of gouge containing 75% brucite that were sheared at different rates at 300°C is consistent with the relative proportions of brucite incorporated in the shears, but the cause of this shear-rate dependent behavior is not understood. However, further study is warranted, because the gradually diminishing stick-slip behavior at the slow shearing rate resembles in some ways the tremor events that have been reported in subduction zones at depths where serpentinized ultramafic rock may be present.
T41B-0579
Mantle-derived fluids and their potential role in weakening the San Andreas Fault
Many plate boundary faults, including the San Andreas Fault (SAF), have been interpreted to slip at shear stresses considerably less than expected for typical rock and fault gouge friction coefficients and hydrostatic fluid pressures. The lack of a heat flow anomaly from frictional heating as expected near the SAF and maximum horizontal principal stress directions inferred to be at a very steep angle with the SAF imply that the fault is weak in both an absolute sense and considerably weaker than the surrounding crust. One explanation for this fault weakness is the presence of near-lithostatic fluid overpressures localized within the fault zone. Here, we use numerical models of fluid flow and heat transport to investigate the potential role of mantle- derived fluids in generating and sustaining overpressures localized within the fault zone. In cross-sectional models of seismogenic crust perpendicular to the SAF, we consider two types of basal boundary conditions representing the influence of mantle fluids. First, we consider a lithostatic pressure boundary extending 25 km or more along the model base. In a separate set of scenarios, we treat a portion of the base as a flux boundary for fluids produced within the mantle. We consider a large, yet geologically reasonable mantle flux of water of 1.7x10-9 kg s-1 per m2 of basal area, equivalent to a 10 km thick mantle wedge serpentinized up to 50% during subduction prior to the creation of the SAF, dehydrating over 25 Myr. We evaluate the sensitivity of the location of both basal boundary conditions by centering them directly beneath the fault zone as well as at positions further east. We evaluate two different permeability architectures hypothesized to localize overpressures generated from a large, prolonged regional source onto a near-vertical fault. First, we evaluate a permeable fault zone surrounded by low permeability crust, both with pressure-dependent permeability (e.g., Rice, 1992). Second, we investigate an alternative model in which hydraulically conductive, east-dipping faults and fractures within the Franciscan Assemblage, truncated against the SAF to the west, focus overpressures on the near-vertical fault zone through flow focusing (i.e., the centroid effect). This model allows for crust more permeable than the fault zone, and may more easily account for peak mantle sources located east of the fault as suggested by geochemical data. We find that pressure-dependent permeability has little influence on localizing pressures within the fault zone; resulting pore pressures are similar to those for depth-dependent and depth-invariant permeability. Through sensitivity analysis, we show that large overpressures can be localized within the fault zone with both basal conditions as long as the surrounding crust is at least 2 orders of magnitude less permeable than the fault zone. Models that incorporate dipping permeable faults within the Franciscan result in a broad zone of overpressure near the SAF, with pressures greatest near the fault zone if it acts as a barrier to fluid flow. Our results also suggest that for either permeability architecture to be viable, crustal permeability ≤ 10-19 m2 is required in order to prevent a large near-fault heat flow anomaly from advection in simulations with a basal pressure boundary condition. There is little effect on heat flow in our models with a fluid flux boundary condition, but these models require low crustal permeabilities (<10-19 m2) in order to generate large overpressures. Both permeability architectures may provide a mechanism for localizing overpressures within the fault zone, and could act in conjunction with low-friction fault gouge to explain a weak SAF.
T41B-0580
Frictional Behavior of Materials in the 3D SAFOD Volume
Laboratory data for fault zone materials acquired at hypocentral depths are necessary for improving our understanding of the relationship between fault zone properties and earthquake physics. These data will illuminate the processes that determine fault strength and stability, and will allow for better assessment of seismic hazard. Here, we report on experiments designed to investigate the frictional behavior of materials in the 3D SAFOD volume including wall and fault rock. We conducted experiments on fault zone materials recovered from SAFOD drilling, and surface samples of rock formations in the 3D crustal volume surrounding the SAFOD drill hole, selected using recent detailed cross sections. Samples from San Andreas drilling range in depth (MD) from 1.4 km to 4.1 km. Surface samples include country rock, wall rock, and fault rock. We also conducted a set of experiments on synthetic mixtures of clay, quartz, and talc. We conducted shearing experiments in double direct shear, under both room conditions and in a true-triaxial pressure vessel. Layers of pulverized fault and surface rock were sheared at constant effective normal stress between rough rigid forcing blocks. Once steady state friction was obtained, shear velocity was stepped between 1 and 300 microns/s. At room temperature and humidity, normal stress ranged from 5 to 100 MPa. Under saturated conditions in a pressure vessel with pore and confining pressures applied, normal stress ranged between 5 and 20 MPa. Results for SAFOD borehole samples of granodiorite, arkosic metasandstone and siltstone indicate coefficients of sliding friction (we assume zero cohesion for the granulated layers) ranging from 0.52 to 0.66. Surface samples show consistent coefficients of friction ranging between 0.55 and 0.68. A serpentinite melange has a coefficient of friction of 0.25 to 0.30 under dry conditions, and 0.20 to 0.25 under saturated conditions. Velocity stepping tests indicate that steady-state friction values were reached and that the fault rocks exhibit slip- rate and history-dependent friction behavior similar to that documented for simulated fault gouge. A sudden increase in load point velocity results in an immediate increase in friction followed by a displacement-dependent decay to a new steady-state level. Measurements of steady-state friction as a function of slip velocity show velocity-weakening frictional behavior for some SAFOD materials at low normal stress. The serpentinite melange shows velocity strengthening behavior under dry conditions and becomes increasingly velocity strengthening under saturated conditions. Our ongoing work includes a comparison of SAF related materials under dry and saturated conditions to synthetic mixtures in an effort to gain insight into fault behavior. Future work will attempt to determine the effects of temperature and pore/confining fluids on the strength and second-order frictional characteristics of fault zone materials.
T41B-0581
Frictional Behavior and Slip Localization in Simulated Faults of Halite at Sub-seismic to Seismic Slip Rates
Halite exhibits deformation behavior ranging from brittle to plastic at room temperature and at low pressures, and has been used to simulate deformation processes of the brittle-ductile transition zone. However, previous experiments on halite were performed at very low slip rates (10-9-10-3 m/s), requiring friction data at seismic slip rates for a more complete assessment of the applicability of the experimental results to natural earthquakes. We conducted friction experiments on halite at slip rate of 0.02-1.3 m/s and normal stresses of 0.8- 10.0 MPa using a high-velocity rotary shear apparatus. A thin layer (0.6-1.0 mm thick) of halite gouge was inserted between precut rock cylinders jacketed with Teflon sleeve. We found that mechanical behavior and deformation processes of halite gouge are remarkably different depending on slip rate and that frictional melting and dislocation creep can occur simultaneously at seismic slip rates. At sub-seismic slip rates of 0.02 to 0.05 m/sec, peak friction (μp = 0.76-0.85) was followed by steady-state friction (μss = 0.35-0.37). Gouge layer consists of a thin slip localization zone at the halite gouge-rock contact and a thick low slip-rate zone. The low slip-rate zone shows evidence for cataclastic flow with angular fragments set in a fine matrix. In contrast, the slip localization zone consists of very fine gouge with some remnants of fragments. At seismic slip rates of 0.1 to 1.3 m/sec, μp (0.64-0.99) was followed by μss (0.36-0.03). μss decreases with increasing slip rate. The shear zone consists of a thin slip localization zone at the halite gouge-rock contact and a thick low slip-rate zone. The low slip-rate zone consists of polycrystalline halite ribbons highly elongated obliquely to shear zone boundary and the oblique foliation is dragged into the thin slip localization zone. Each ribbon is also defined as a lattice preferred orientation domain by electron back-scattered diffraction (EBSD) analysis. Individual grains within the halite ribbons are also elongated with their long axis subparallel to the ribbons and their grain boundaries are either straight or slightly wavy. The size of the grains within the ribbons gradually decreases toward the slip localization zone. These microfabrics indicate that each halite ribbon results from plastic deformation of the original halite grain, with individual grains within the ribbon representing product of dynamic recrystallization. Their straight or slightly wavy grain boundaries suggest a static adjustment during cooling after deformation. In contrast, the thin slip localization zone consists of very fine euhedral grains (1-3 μm) at the margin of layer. We interpret that the very fine euhedral grains have grown from frictional melt and the plastic deformation in the low slip-rate zone was enhanced by heat conduction from the slip localization zone. Our experimental results suggest that pseudotachylyte and mylonite may develop simultaneously by 'slip-rate partitioning' at seismic slip rates, providing a new insight into the interpretation of their coexistence. Also, in view of preexisting halite friction data at lower slip rates, halite gouge exhibits a slight velocity weakening at low velocity regime (10-9-10-5 m/s), velocity strengthening at intermediate velocity regime (10-5-10-3 m/s), steady state at sub-seismic velocity regime (10-2-10-1 m/s) and remarkable velocity weakening at seismic velocity regime (> 10-1 m/s).
T41B-0582
Localized Versus Distributed Deformation as a Control on the Evolution of Permeability in Anhydrite Rocks
We have taken an experimental approach to understand and quantify the deformation processes and fluid flow within anhydrite-bearing fault damage zones during the seismic cycle. Triaxial loading tests have been performed on borehole samples of anhydrites at room temperature, 100 MPa confining pressure (Pc), and range of pore fluid pressures (Pf). Permeability and porosity development was continuously measured throughout the deformation experiments. The tests were conducted on samples with different grain sizes (10 microns to 1 mm) that were cored in different directions relative to the macroscopic foliation. Static permeability measurements have been carried out to determine the permeability anisotropy and sensitivity of the permeability on the effective pressure (Pc – Pf). Our results show that the brittle-ductile transition occurs for effective pressures (Pe) between 20 to 40 MPa and is almost independent of fabric orientation and grain size. Brittle failure is localized along discrete fractures and is always associated with a sudden stress drop. Conversely, ductile failure occurs by distributed deformation along cataclastic bands. In this case no stress drop is observed. Static permeability measurements show increasing values of permeability for decreasing values of Pe, (k = 10E-20 – 10E-22 m2). During single cycle loading tests, the evolution of the permeability is controlled by the failure mode: permeability begins to increase significantly at 40% and 80% of the max load for samples displaying brittle and ductile behaviour, respectively. The permeability values, immediately prior to failure, are about three orders of magnitude higher than the initial values. Multiple cycling tests, within the ductile field, show that permeability starts increasing at only 40% and 30% of the max load during the second and third loading cycle, respectively. Our results show that the history of deformation and the mode of deformation can control the evolution of the permeability, and that they are more significant than other factors such as fabric and grain size. In natural environments, fluid pressure fluctuations, such as might be experienced during the seismic cycle, can promote a switch from localized (brittle behaviour) to more distributed (ductile behaviour) deformation, leading to complex permeability patterns.
T41B-0583
Strain Localization due to Structural Softening During Pressure Sensitive Yielding
Mohr coulomb non associated flow rule l is able to catch the main characteristic of brittle behavior i.e. pressure sensitivity and relatively small volumetric strain versus shear strain observed in shear bands with only two plastic parameters (φ and ψ). Without any further increase of complexity of rheological model, it is sufficient to explain strain self-localization in granular material with predictions that are compatible with major observations such as the occurrence of localization within hardening regime, orientation of the localized zones and their diversity. For non rate dependant material, the problem is found to be solvable analytically as a system of linear equation for which we derived analytical expression for the maximum stress drop a shear band may produce, for the equivalent tangent modulus and rate of rotation of the stress within the band as a function of shear band orientation. Performing a parametric study, we found a good approximation for the characteristic strain necessary for a shear band to achieve its maximum stress drop. This reveals that pressure sensitivity is not only related to friction but that elasticity also affects the scale of the displacement needed to achieve stress drop. This effect, often neglected when scaling experiments to nature, may lead to effective brittle plastic in the laboratory which does not have to exist in nature (or the opposite with sand box type analogue experiments in the field of tectonics). Within this model, the most favorable orientation for shear band evolves with strain so that, in a system which is not kinematically locked, the MC model predicts changes in the orientation of the strands with strain and part of the complexity of natural shear bands. For rate dependant material, numerical experiments were used to quantify the strain rates at which brittle ductile transition occurs. We show that in that case, the brittle plastic transition occurs as a function of confining pressure and strain rate and that at that level, a small domain exists where the shear zones harden with decreasing the viscosity.
T41B-0584
Fault Zone Weakening and Strain Localization Adjacent to the Alpine Fault in Fiordland, New Zealand: the Roles of Cohesion Loss and Elevated Fluid Pressure
The mechanisms by which continental transforms and major strike-slip faults localize deformation and accommodate large amounts of strain are unresolved problems in continental tectonics. We used fault-slip data, geological observations, and an analysis of paleostress tensors to empirically test the roles of cohesion loss and elevated fluid pressure in localizing strain adjacent to the Alpine Fault in northern Fiordland, New Zealand. A kinematic analysis of fault populations reveals the presence of a narrow, ~10 km wide zone on the southeast side of the Alpine Fault where deformation is dominated by reverse faulting and oblique-dextral slip. Cross cutting relationships and published geochronology suggest that most of these faults record motion since ~11 Ma, when changing relative plate motions resulted in an increased component of compression across the Alpine Fault. Pervasive hydrothermal alteration, veining, and mineralization indicate that this zone was preferentially infiltrated by metamorphic fluids during deformation. Outside of the 10 km wide zone, to the east and southeast, evidence of pervasive fluid infiltration and metasomatism is lacking, and faults record dominantly strike-slip motion. Stress inversions indicate that within 10 km of the Alpine Fault, compression axes are oriented at moderately high (~65°) angles to the dominantly northeasterly strike of the plate boundary. Outside of this 10 km zone, stress axes are oriented at low (~10°) angles. The high angles recorded near the Alpine Fault contrast with a 11-25° angle of oblique plate convergence and suggest that the Alpine fault zone in Fiordland is very weak, and slips in response to very low shear stresses. A quantitative test of this hypothesis using geometric and frictional constraints on stress inversions confirmed that the coefficient of friction on the faults near the Alpine Fault is very low (μ = 0.10). This result suggests that cohesion loss as a result of high fluid pressures, mineral transformations, and a reduction in the coefficient of friction on faults are primary causes of fault zone weakening up to 10 km away from the main trace of the Alpine Fault. A loss of cohesion also appears to contribute to the localization of the contractional component of deformation away from the main trace of the Alpine Fault in Fiordland. This style of kinematic partitioning, where the contractional component of deformation occurs elsewhere and the Alpine Fault itself records strike-slip motion, is unique to the southernmost segment of the plate boundary. Farther north the Alpine Fault records reverse-dextral slip and low degrees of strike-slip partitioning.
T41B-0585
Co-seismic Static Stress Drops for Earthquake Ruptures Nucleated on Faults After Progressive Strain Localization
We estimate the coseismic static stress drop on small exhumed strike-slip faults in the Mt. Abbot quadrangle of the central Sierra Nevada (California). The sub-vertical strike-slip faults cut ~85 Ma granodiorite, were exhumed from 7-10 km depth, and were chosen because they are exposed along their entire lengths, ranging from 8 to 13 m. Net slip is estimated using offset aplite dikes and shallowly plunging slickenlines on the fault surfaces. The faults show a record of progressive strain localization: slip initially nucleated on joints and accumulated from ductile shearing (quartz-bearing mylonites) to brittle slipping (epidote-bearing cataclasites). Thin (< 1 mm) pseudotachylytes associated with the cataclasites have been identified along some faults, suggesting that brittle slip may have been seismic. The brittle contribution to slip may be distinguished from the ductile shearing because epidote-filled, rhombohedral dilational jogs opened at bends and step-overs during brittle slip, are distributed periodically along the length of the faults. We argue that brittle slip occurred along the measured fault lengths in single slip events based on several pieces of evidence. 1) Epidote crystals are randomly oriented and undeformed within dilational jogs, indicating they did not grow during aseismic slip and were not broken after initial opening and precipitation. 2) Opening-mode splay cracks are concentrated near fault tips rather than the fault center, suggesting that the reactivated faults ruptured all at once rather than in smaller slip patches. 3) The fact that the opening lengths of the dilational jogs vary systematically along the fault traces suggests that brittle reactivation occurred in a single slip event along the entire fault rather than in multiple slip events. This unique combination of factors distinguishes this study from previous attempts to estimate stress drop from exhumed faults because we can constrain the coseismic rupture length and slip. The static stress drop is calculated for a circular fault using the length of the mapped faults and their slip distributions as well as the shear modulus of the host granodiorite measured in the laboratory. Calculations yield stress drops on the order of 100-200 MPa, one to two orders of magnitude larger than typical seismological estimates. The studied seismic ruptures occurred along small, deep-seated faults (10 km depth), and, given the fault mineral filling (quartz-bearing mylonites) these were "strong" faults. Our estimates are consistent with static stress drops estimated by Nadeau and Johnson (1998) for small repeated earthquakes.
T41B-0586
True Triaxial Strength and Brittle Fracture of the Granodiorite at the SAFOD Drillhole Wall, and the Potential for Estimating the Maximum Horizontal Principal Stress
Salinian granodiorite core from the 1462-1470m segment of the SAFOD drillhole was used to derive its critical mechanical properties under true triaxial stress conditions, analyze shear localization and brittle fracture characteristics, and establish the strength criterion under dry conditions (Eos Trans. AGU, 87/52, Abstract T32C- 03). Here we report on a series of true triaxial tests on ‘unjacketed' specimens simulating stress conditions prevailing at the drillhole wall and responsible for borehole failure in the form of breakouts. Owing to numerous random cracks inherent in the core, only 11 rectangular prismatic specimens (19×19×38 mm3) were successfully tested, employing the University of Wisconsin polyaxial cell. The two larger principal stresses, σ1 and σ2, were transmitted through metal pistons, while σ3 was applied by confining fluid pressure. Specimen sides facing σ3 were left ‘unjacketed', i.e. in direct contact with the confining fluid, to simulate the condition of drilling-mud pressure applying the principal radial stress (σ3) to the exposed borehole wall. The loading path called for first bringing σ2 and σ3 to preset levels and then increasing σ1 at a constant strain rate (5x10-6/sec) until brittle failure occurred. Invariably, failure occurred at σ1 levels that were only about half as high as those in previously tested dry samples under the same σ2 and σ3 magnitudes. Instead of a shear fracture, or fault, steeply inclined in the direction of σ3, as previously observed in the dry specimens, brittle failure took the form of a localized cluster of through-going extensile cracks parallel and adjacent to the faces subjected to σ3. Since failure occurred at σ1 values close to those at dilatancy onset in dry specimens, we infer that as soon as microcracks reopened, confining fluid rushed into those daylighting at the σ3 faces and extended them along a path of least resistance, i.e. along a plane normal to σ3. Thus brittle failure under drillhole wall conditions is drastically different from that conventionally expected, but is compatible with breakout formation mechanism in granite (Haimson, Int. J. Rock Mech., 2007). All the ‘unjacketed' true triaxial strength data can be fitted by a simple function in the octahedral shear stress versus octahedral normal stress domain, yielding a Nadai-type true triaxial strength criterion. The criterion can be used in conjunction with breakouts that have been located within the cored zone to yield the maximum horizontal in situ stress σH when the other two principal stress are known. Assuming that the state of stress at breakout-drillhole intersections (located for example by BHTV logging) is sufficient to bring about brittle failure (Vernik and Zoback, 1992), one can substitute the known principal stresses there (obtained from the Kirsch solution) for the corresponding values in the criterion. The in situ σv is given by the overburden density, σh is typically obtained from hydrofrac shut-in pressures, breakout width is extracted from BHTV logs, borehole fluid pressure is a function of its density, and the Poisson's ratio is obtained from mechanical lab testing. The only unknown, σH, is thus readily computed. An actual computation was not carried out because data on hydrofrac pressures and breakout dimensions were not available at the time of this submission.
T41B-0587
Localization of Deformation in Elastic-Plastic Analysis of Dynamic Shear Rupture Propagation
Recent studies have allowed for off-fault elastic-plastic deformation in analyses of dynamic earthquake rupture
propagation on slip-weakening faults [Andrews, 2005; Shi and Ben-Zion, 2006; Templeton et al., 2006; Viesca et
al., 2006]. These studies used Mohr Coulomb (MC) or Drucker-Prager (DP) type pressure-dependent yield criteria
for describing the onset of plastic deformation in granulated or cracked rocks.
In plane strain finite element analyses of such dynamically propagating ruptures, for a DP material, we found
features in the strain field that indicate localization into shear-band-like structures. Bifurcation analyses for states
of spatially homogeneous quasistatic deformation have shown that instabilities in the constitutive description for
certain classes of elastic-plastic materials can lead to such localizations [Hill, 1952; Thomas, 1961; Rudnicki and
Rice, 1975; Rice, 1976]. Localization conditions coincide with those for a vanishing propagation speed of elastic-
plastic body waves [Hadamard,1903; Hill, 1962; Mandel, 1963]. Rudnicki and Rice found that localization can
occur even for positive values of the plastic strain-hardening modulus, h, in materials with pressure-dependent
yield and non-associated plastic flow. They determined a critical value, hcr, such that localization will occur
only for h ≤ hcr. For 2D stress fields with the out of plane principal stress equal to the average of the in
plane values, the MC and DP criteria coincide, and hcr>0.
Elastic-plastic analyses of rupture dynamics typically use a plane strain, non-hardening model, h=0; that
means h
T41B-0588
Modeling Spontaneous Generation of Off-Fault Plastic Strain Localization During Dynamic Earthquake Rupture
We extend an elastodynamic finite element method to incorporate off-fault plastic yielding of Mohr-Coulomb form into a 2D spontaneous dynamic earthquake rupture model. For straight faults under uniform stress conditions, we find that rupture-induced plastic strain tends to spontaneously localize into discrete bands if the off-fault material is close to the yielding strength in the initial stress state. While fully regularized, convergent solutions of the strain localization are difficult to be achieved, the numerical simulations can capture the onset of localization. Attempts at regularization via viscoplastic relaxation sometimes result in a smooth, numerically convergent solution , while in other cases the result is simply to delay the onset of localization, without suppressing the smallest, numerically irresolvable scale lengths of localized plastic strain. For non-straight faults, we find that a discrete kink in fault strike also localizes plastic strain into several discrete bands whose orientations are consistent with the accommodation of complex fault geometry at the kink. Apparently convergent numerical solutions (as assessed by numerical element size reduction) can be obtained for the strain localization at the kink. Off-fault plastic yielding and strain localization can have significant effects on spontaneous dynamic rupture propagation and seismic radiation. It lengthens the cohesive zone at the rupture front, such that the width of this zone tends to stabilize with rupture propagation distance, with the fault behavior under slip-weakening friction essentially equivalent to a time-weakening friction law. It also reduces seismic radiation from fault kinks by filtering high-frequency contents above several Hertz.
T41B-0589
Onset of Strain Localization in Sheared Granular Materials
Strain localization plays a key role in determining the frictional stability of brittle shear zones, which in turn influences the rheology and seismic/aseismic behavior of fault zones and deforming glacial till. Recent studies show that seismic, stick-slip motion occurs in dilatant till layers. We report on detailed laboratory experiments to measure the onset of shear localization of Caesar till sampled from the Scioto Lobe of the Laurentide Ice Sheet, collected in central Ohio, USA. Experiments were conducted in a servo-controlled, double-direct shear apparatus with saturated samples at normal stresses of 0.5, 1, and 5 MPa. The nominal frictional contact area is 100 cm2 and remains constant during shear. The layer thickness was either 0.5, or 1 cm prior to shear. Till was sheared under constant velocity or constant shear stress (creep) boundary conditions. Constant velocity experiments were conducted with a series of velocity steps from 10 μm/s to 30 μm/s. Constant shear stress experiments were employed to study frictional creep and the localization of strain. Creep was induced after an initial shear strain ranging from 0 to 5 to investigate the role of shear fabric. In creep experiments, shear stress increments began at ~ 2/3rds of the shear strength and continued until tertiary creep occurred. Stress steps were 5% of the shear strength and we determined the resulting strain rate and layer dilation. Microstructures are analyzed in transmitted light microscopy and under SEM. Both velocity stepping and creep experiments show a transition from distributed deformation to localized deformation at low strains. Velocity steps show that Caesar till is a velocity strengthening material, with the critical slip distance decreasing with strain up to 0.7 then remaining constant. Creep experiments show that dilatancy after a stress step decreases exponentially from ~10 μm at shear strain of zero to ~2 μm at a shear strain of 1. Beyond strains of 1 no variation in dilatancy is observed. Decreasing initial layer thickness decreases dilation by the same factor at low strains, but has no effect at strains of 1 or greater. These results imply that shear becomes more localized over a finite displacement in a velocity strengthening material. Beyond the onset of localization, variations in strain and driving velocity do not change the characteristics of strain localization. Localized deformation in till implies shallow deformation, which does not regulate fast glacial slip.
T41B-0590
Long-Lived Crustal Structures and Their Role in the Tectonic Evolution of the Cascadia Backarc
The backarc region of the Cascadia forearc in Oregon and northern California displays a northeast-trending crustal fabric evident in regional gravity data, seismicity, fault distribution, Cretaceous plutonism, Mesozoic accreted terranes, and Quaternary volcanism. These associations involve rocks of diverse ages and origins and indicate that long-lived crustal structures may have influenced backarc tectonic evolution and may be a natural consequence of clockwise rotation and westward migration of the southern Cascade forearc. A series of northeast-striking elongate gravity lows are aligned to Cascade-arc volcanism and crustal seismicity. We suggest that these lows indicate an internal fabric in the pre-Tertiary basement that has focused Neogene magma transport and faulting occurring in the wake of the migrating volcanic arc. In a broader sense, they aided partitioning of strain associated with east-west Basin-and-Range extension and right lateral shear ultimately related to North American-Pacific plate interactions. This wide region of deformed crust, which we refer to as the Harney Basin deformation zone (HBDZ), appears to be accommodating strain between rigid blocks of the Sierra Nevada to the south and the coherent Oregon forearc to the north. The southern edge of the HBDZ coincides with a young physiographic depression, which in turn marks a major terrane boundary, juxtaposes distinct Tertiary volcanic stratigraphy, coincides with an active tectonic boundary, and parallels a belt of Cretaceous plutonism. The northern boundary of the HBDZ is marked by the Klamath-Blue Mountains lineament that lies along the southern edge of the Blue Mountains province, an assemblage of terranes accreted in Jura-Cretaceous time that itself forms a northeast-elongate block. Taken together, these associations suggest that the HBDZ has an inherited crustal fabric of northeast trending structures that has influenced multiple episodes of magmatic and tectonic activity since terrane accretion, circa Eocene time. One possible model that describes the role these structures play in guiding backarc deformation indicates that the entire HBDZ may have undergone significant CW rotations similar to those observed in the Blue Mountains.
T41B-0591
Effect of strain localization on sub-rift upper mantle rheology and anisotropy
In order to explain a possible weak rheology of the uppermost mantle suggested by work on the focal mechanism distribution in the continental lithosphere, we have investigated microstructures, LPO and water contents using FTIR of peridotites from present-day (Baja California) and ancient (North Pyrenean) rift zones. The aim of this study is link microstructural analysis with the rheology and the seismic anisotropy from small scale to lithosphere-scale processes. Infrared studies reveal a significant amount of water in the San Quintin xenoliths minerals (Peslier et al., 2002) while our infrared analyses done on the Pyrenean peridotites show that coarse olivines and pyroxenes have low water contents. Furthermore, microstructures and LPO orientations suggest that the North Pyrenean rift mantle had overall low water content with higher but modest water content in shear zones. The inferred anisotropy from olivine LPO is crucial to understand the observed shear wave-splitting (anisotropic pattern). In detail, we address the effects of ultra fine-grained shear zones development on the mechanical and thermal evolution of the upper mantle underneath the North Pyrenean Mesozoic rift and the contribution of inferred anisotropy from olivine LPO in Baja California with the shear wave splitting direct observations. Electron Backscatter Diffraction work on Pyrenean mylonites shows that LPOs of fine-grained minerals are scattered in contrast to coarse olivine grains with strong LPOs. These LPOs suggest a dislocation creep mechanism in large grains and grain size sensitive creep in the mylonite. The fine-grained shear zones may initiate as fractures and will weaken the lithosphere (H2O+CO2 infiltration) depending on the spatial distribution of the zones. Shear zone development will significantly reduce the seismic anisotropy of the lithosphere, as higher strains produce more fine-material with weak LPO. Detectable seismic reflections may be produced between host rocks and mylonites if the mylonite zones are wide enough. Both the upper-mantle of Baja California and shear zone of the Pyrenean rifts have wet rather than dry rheology. The fine-grained hydrated shear zones result in a very weak uppermost mantle.