T21A-0438
Preliminary Observations of Stress and Fluid Pressure in and Near the San Andreas Fault at Depth in the SAFOD Boreholes
A variety of observations in the SAFOD pilot hole, drilled in 2002, and the first two Phases of the main SAFOD borehole allow us to place preliminary constraints on the orientations and magnitudes of principal stresses as well as pore pressure at depth. It will be possible to improve these preliminary results once detailed data on the shape of the main hole acquired during Phase 2 (i.e., through the San Andreas fault zone) are fully processed. In addition, the core holes to be drilled during Phase 3 in 2007 will be used to make measurements of the least principal stress via hydraulic fracturing. Stress orientation has been determined from the orientation of wellbore breakouts in both the vertical pilot hole to a depth of 2.2 km and in the deviated portion of the Phase 1 SAFOD hole down to a vertical depth of 2.5 km. These indicate that the maximum horizontal principal stress rotates with depth, attaining a high angle to the San Andreas Fault at depth and consistent with the hypothesis that there are low shear stresses acting on the fault. A similar conclusion was reached based on the direction of shear velocity anisotropy determined from cross-dipole sonic logs conducted in SAFOD during Phase 1 (see abstract by N. Boness and M.D. Zoback). Three hydraulic fracturing tests were carried out to constrain the magnitude of the least principal stress along the trajectory of the SAFOD borehole, at vertical depths of 1.5 km, 2.5 km, and 3.2 km. The 1.5- and 3.2-km-deep tests, which were conducted at distances of about 1.1 km SW and 0.5 km NE, respectively, from the two seismically active traces of the San Andreas identified at this location, indicate that the least principal stress is the vertical stress. In conjunction with modeling of wellbore failure and estimates of rock strength, these tests further indicate a transitional strike-slip/reverse faulting stress state, which is consistent with our earlier stress analyses from the pilot hole. The test at 2.5 km, carried out within only about 100 m of the closest seismically active fault trace, indicates that the magnitude of the least principal stress is 20 MPa (or more) higher than the calculated overburden stress. While such a result is quite unusual (one principal stress normally corresponds to the lithostat) several theoretical models of the state of stress within the active fault zone predict such high stress magnitudes for the case of a fault with low frictional strength. Observations made to date indicate subhydrostatic fluid pressures both in the pilot hole and at the bottom of SAFOD Phase 1. In contrast, elevated pore pressures are indicated in the deep sedimentary section drilled during Phase 2 by the influx of gas into the well at times when drilling stopped. As the density of the drilling mud in the hole was approximately 40% greater than fresh water, these gas influxes indicate pressures appreciably in excess of hydrostatic pressure. However, it is not yet clear whether these elevated pressures influence the mechanics of faulting or are simply related to hydrocarbon maturation (and other processes) in the low permeability shales encountered at depth. This uncertainty will be addressed when fluid pressure observations are made within the active strands of the San Andreas in 2007 when the Phase 3 core holes are drilled.
T21A-0439
Spectral Analysis of Localized Stress Variations, the Spatial Distribution of Faults, and the Scaling of Physical Properties near the San Andreas Fault
Statistical characterization of stress-induced wellbore failures and rock property heterogeneity from well logs offers potential insight into the scaling properties and mechanisms of stress heterogeneity. Wellbore breakouts identified in acoustic wellbore image data obtained adjacent to the San Andreas Fault, from both the San Andreas Fault Observatory at Depth (SAFOD) and the Cajon Pass Scientific Borehole, reveal multi-scale rotations in the direction of maximum horizontal compressive stress (SHmax) as a function of depth. Similar breakout rotations are frequently observed in other deep wellbores and, in most cases, reflect small variations in the directions and/or magnitudes of the in situ principal stresses superimposed on a relatively uniform regional stress state. To determine possible physical causes for these rotations, we employ spectral and statistical methods to investigate the relationships between the breakout rotations observed in our study wells and stress drops associated with slip on faults in highly fractured crust adjacent to a major fault zone. We also address the possible role of rock property variability as a controlling mechanism, taking into account drilling and data acquisition artifacts. We find that physical property heterogeneity in the SAFOD Pilot Hole behaves as self-similar, flicker noise (i.e., 1/f) over wavelengths from one meter to one kilometer, a result that agrees with similar investigations at Cajon Pass and a variety of other locations throughout the world. The stress orientations in both wells, however, exhibit behavior between that of flicker noise and Brownian motion over wavelengths from one decimeter to several kilometers, which is similar to how earthquake frequency has been shown to scale with fault size. The fractal scaling of observed stress heterogeneity appears to be more closely related to the distribution of faults in the crust adjacent to the study wells than to heterogeneity of elastic or other in-situ physical properties. In addition, we are able to model specific breakout rotations at SAFOD as resulting from stress perturbations created by slip on existing fractures in response to the current stress field. This result further supports our hypothesis that stress heterogeneity near the San Andreas Fault is controlled by slip on secondary, active faults at a variety of scales.
T21A-0440
A Borehole Fiber-Optic Strainmeter
The SAFOD project provides an opportunity to study the basic mechanics of a fault that undergoes relatively frequent events. Magnitude 2 earthquakes within a few km of the borehole occur several times per year and produce strain signals on the order of a nanostrain. We installed a vertical fiber-optic strainmeter in the annulus of cement between the inner and outer casings of the borehole. The sensor consists of an optical fiber cable stretched from the surface to a depth of 864 m. The optical length of the fiber is monitored with a laser interferometer sampled at 50,000 samples per second. The strainmeter noise is about 0.01 nanostrain at 1 Hz, which provides clear records of coseismic offsets.
T21A-0441
Upper-Crustal Reflectivity of the Central California Coast Range Near the San Andreas Fault Observatory at Depth (SAFOD), USA
We describe a new method of extracting seismic reflections that are visually evident in shot gathers but which may or may not come into focus in an image processed using conventional CDP reflection processing. This method has proven extremely useful in the central California Coast Range, near the San Andreas Fault Observatory at Depth (SAFOD), where conventional CDP processing has thus far produced an image that has few to no clear reflections, although special processing has imaged a couple of steeply dipping reflectors. The image described here includes both gently and steeply dipping reflections that combine to produce an interpretable image of the subsurface. Our data was recorded along a 46-km profile centered on SAFOD and perpendicular to the San Andreas Fault (SAF), with 62 shots and 912 recorders (shot spacing 0.5 to 1 km; receiver spacing 25 to 50 m). Although conventional CDP processing produced an image with few to no clear reflections, reflections are definitely visible in shot gathers. Using our new method, coherent energy (reflections and other phases) are picked on shot gathers and converted automatically to line drawings, and then the line drawings are migrated in a tomographic velocity model. The final image has clear reflectivity, including both gently and steeply dipping events. We see subhorizontal to gently west-dipping reflective bands within the granitic Salinian block at depths of 6 to 14 km, beginning at approximately the SAF. Within the Franciscan melange east of the fault, we see diffuse gently west-dipping reflectivity at depths of 4 to 10 km. Near the Coast Range (or here Waltham Canyon) fault (CRF), we see a sharp, steeply east-dipping reflector that begins approximately 2 km below the surface and 2 km west of the surface trace of the CRF. At approximately 4 km depth this reflector bends to become gently east dipping. A short but clear zone of west-dipping reflectors connects the top of this curved reflector to the surface trace of the CRF. The reflective zone within the Franciscan block to the west of this system of sharp reflectors terminates at this system. A preliminary interpretation of the CRF region is as follows: the Great Valley sequence to the east of the Franciscan, consisting of folded Mesozoic and Cenozoic sedimentary rocks, is disposed as a blunt wedge that indents the Franciscan block, with the back thrust, or the west-dipping reflectors at the top of this interpreted wedge, now exposed at the surface as the CRF.
T21A-0442
Testing new Empirical Relations Between Elastic Wavespeeds in the Earth's Crust Using SAFOD Main Hole Logs
I test new empirical relations between compressional-wave (Vp) and shear-wave (Vs) velocity for the Earth's crust using SAFOD Main Hole logs from Phase 1 and 2 drilling. In a recent paper, Brocher (BSSA, Dec. 2005) compiled thousands of observations of Vp and Vs for a wide variety of common rock types to define new empirical relations between these properties. The new nonlinear relations are intended to provide a useful average fit between Vs and Vp, much the way that the Nafe-Drake curve provides an average fit between density and Vp. One of the most useful of these relations is Vs (km/s) = 0.786 - 1.2344Vp + 0.7949Vp2 - 0.1238Vp3 + 0.0064Vp4, for Vp between 1.5 and 7.5 km/s. This relation yields a strongly non-linear Poisson's ratio versus Vp function that can be approximated as a linear decrease in Poisson's ratio from 0.50 to 0.24 for Vp between 1.5 and 4.1 km/s, and a linear increase in Poisson's ratio from 0.24 to 0.27 for Vp between 4.1 and 8.5 km/s. The empirical relation is a regression of data from Vs and Vp borehole logs in Quaternary alluvium and sedimentary rocks in California, Salinian granites from the SAFOD Pilot Hole, as well as global averages of laboratory measurements of crystalline rocks over a 6-km/s-wide range in Vp. SAFOD Main Hole logs sampling Salinian granite in the upper 2/3 of as borehole, as well as sedimentary rocks (Great Valley Sequence?) in the lower 1/3 of the borehole, provide an independent test of the new empirical relation. Using the new empirical relation to calculate Vs from the Vp log over predicts the observed Vs log for the Great Valley Sequence (?), having an average Vp of 4.46 km/s, by an average of 0.10 km/s, and over predicts the entire SAFOD Main Hole log by an average of 0.12 km/s. This close fit indicates that the new empirical relation provides a useful approximation for Vs in sedimentary rocks as well as for a wide-variety of other common rock types. The fact that Vs observations match the empirical relation in the SAFOD Main Hole, which penetrates at least two active strands of the San Andreas fault, implies that there is nothing particularly exceptional about Vs and Vp relations near or within active fault zones.
T21A-0443
Borehole Array Observations of Non-Volcanic Tremor at SAFOD
We report on the observation of non-volcanic tremor made in the San Andreas Fault Observatory at Depth in May, 2005 during the deployment of a multi-level borehole seismic array in the SAFOD main hole. The seismic array consisted of 80 levels of hydraulically-clamped 3-component, 15 Hz omni-directional geophones spaced 15.24 m apart along a 1200 m section of the inclined borehole between 1538 and 2363 m below the ground surface. The array was provided by Paulsson Geophysical Services, Inc. (P/GSI), and recorded at a sample rate of 4000 sps on 24-bit Geode digital recorders provided by Geometrics, Inc. More than 2 TB of continuous data were recorded during the 2-week deployment. Selected local earthquakes and explosions recorded by the array are available at the Northern California Earthquake Data Center, and the entire unedited data set is available as assembled data at the IRIS Data Management Center. Both data sets are currently in the industry standard SEG2 format. Episodes of non-volcanic tremor are common along this reach of the San Andreas Fault according to Nadeau and Dolenc [2004, DOI: 10.1126/science.1107142], with many originating about 30 km southeast of SAFOD beneath the southern end of the Parkfield segment and northern end of the Simmler segment of the fault. We identified tremor episodes using spectrograms routinely produced by the Northern California Seismic Network (http://quake.usgs.gov/cgi-bin/sgrampark.pl) on which they appear as periods of elevated noise relative to the background. A particularly strong tremor episode occurred on May 10, 2005 between 19:39 and 20:00 UTC. In SAFOD, tremor spectral levels exceed the instrumental noise floor to at least 40 Hz. The spatially unaliased recording of the tremor wavefield on the P/GSI array reveal individual phases that can be tracked continuously across the array. The wavefield is composed of both up- and down-going shear waves that form quasi-stationary interference patterns in which areas of constructive interference recur at the same locations along the array. Such a pattern could arise from a spatially stationary source radiating an extended duration time function into a complex medium.
T21A-0444
The Branching Pattern of Low-Velocity Structure on the San Andreas Fault near the SAFOD Site at Parkfield from Fault-Zone Guided Waves
We use fault-zone guided waves recorded at surface seismic arrays deployed across and along the San Andreas fault for explosions and local earthquakes combined with the guided waves recorded at borehole instrument in the SAFOD main hole to delineate the low-velocity structure of the SAF at seismogenic depths. The data from events occurring within the fault zone at different depths and with the raypath incidence angles to the surface array smaller than 30o from vertical show that the length of fault-zone guided wavetrains after S-arrivals at stations within the fault zone increases progressively from ~1 to ~2 s as the event depth increases from ~3 to ~10 km, with greater increasing at depths shallower than 5 km. In contrast, the data recorded at away-fault stations for the same events and the data recorded at on-fault stations but for events occurring away from the fault zone show shorter length of wavetrains after S-arrives and a much milder depth-dependent trend. We also observed guided waves partitioning on the secondary fault at station located at 875 m southwest of the SAF surface trace. The data from seismometers in the SAFOD main hole show fault-zone guided waves channeling on this secondary fault, which passes the borehole station at the depth of 2.7 km and ~400 m southwest of the nearly vertical SAF, for the events occurring on the SAF at depths deeper than ~3 km (Shalev and Malin, 2005). Observations and 3-D finite-difference simulations of fault-zone guided waves show a branching pattern of the low-velocity structure on the SAF at shallow depth near the SAFOD drilling site while the low-velocity zone likely extends across seismogenic depths with the smaller velocity reduction within the deeper potion of fault zone due to larger confined stress at greater depths. Shear-wave splitting analyses of the data suggest slight depth dependence to the measured delays of slow S waves for source depths between 2 and 7 km.
T21A-0445
A Simultaneous Imaging Method of Multiple Scattering Modes for Detecting a Fault-Zone Heterogeneous Structure of the San Andreas Fault, Parkfield, California
One approach to understanding the generation process of earthquakes is to image fault-zone heterogeneity through the use of single-point scatterers. We present an imaging methodology for imaging multiple scattering modes (P-P, P-S, S-P, and S-S) to assess the relative amplitude of heterogeneity in the bulk and shear modulus in the fault zone. This method is designed for a network of three-component seismic stations and a source array produced from an aftershock sequence. Scattering modes and scatterer locations are determined by the following procedure. For each station, the wave-type and slowness (i.e., propagation) vector for the source-to-scatterer part of the path are estimated by performing a semblance analysis. For the scatterer-to-station segment of the path, the wave-type is constrained by comparing the observed polarization vector, inferred from particle motion, with the predicted propagation vectors from candidate scatterer locations, assuming a half-space velocity model ($V_p$ and $V_s$ are 6.40 km/s and 3.45 km/s, respectively in this study). Candidate scatterer locations and allowable scattering modes are then evaluated by comparison of the observed slowness and polarization vectors with a probability density function based on the 95 per cent confidence levels for these two parameters, in addition to the travel time residual between the observed and predicted travel times. We apply the method to borehole seismograms from 10 relocated aftershocks of the October 20, 1992, M=4.7 Parkfield earthquake recorded by eight stations of the High Resolution Seismic Network. To examine the spatial resolution of the image sections and the ability of our data set to distinguishing among scattering modes, we perform a simple numerical experiment with synthetic seismograms in the frequency range of 8-16 Hz where the signal level was highest, adding 20 per cent of Gaussian random noise to the average signal level at each station. We place three scatterers of each scattering mode at various locations along the San Andreas Fault. We find that the scatterers are generally well recovered so that we expect to resolve each of the scattering modes. From the Parkfield data, we obtain image sections for P-P, P-S, S-P, and S-S scattering modes in the frequency band used in the synthetic test. We find a well constrained region for S-S scattering mode that is located about 10 km south-southeast of the epicenter of the 1966 M=6 Parkfield earthquake at 5 km depth. The size is estimated to be 300 m. We observe no other scattering modes with this data set in this region. The strength of the S-S scattering, combined with the absence of P-P, P-S, and S-P scattering modes implies that structural heterogeneity in the region is dominated by variations in the shear modulus. As such, we hypothesize that the S-S scatterer is associated with fluid-filled cracks or fractures (O'Connell and Budiansky, 1974).
T21A-0446
Seismic Reflection and Diffraction Imaging of the San Andreas Fault at SAFOD
A 2D seismic reflection and refraction survey across the San Andreas Fault (SAF) near Parkfield at the location of the San Andreas Fault Observatory at Depth (SAFOD) provides a detailed characterization of upper crustal structure. We present results from prestack migration and from waveform tomography of these data. Three-component stations at 50m spacing recorded 63 explosive shots along a 46 km long line perpendicular to the surface trace of the SAF. P-wave velocities derived by first-break travel time tomography were used in a Kirchhoff prestack migration that included steep dips. The resulting seismic reflection section shows several steep reflectors down to ~5 km depth. A strong reflector ~10 km to the northeast of the SAF is interpreted to mark the faulted transition from the Franciscan accretionary sediments to the Great Valley Sequence. It dips steeply to the southwest down to 2 km depth, where it overturns to dip to the northeast to at least 5 km depth. Its unusual shape is interpreted to be the result of transpressional forces related to the plate motions and backthrust faulting. Weaker, but still prominent steep reflections are imaged at the SAF, and from the steep edge of the Salinian granite 1.6 km southwest of the SAF. A vertical reflection between 0.5 and 1.0 km below the surface trace of the SAF is interpreted to be a direct image of the fault. A second vertical reflection from 1.0 to 2.5 km depth is offset 300 m towards the northeast, and coincides with the mapped location of the Gold Hill Fault (GHF). This reflector either marks an offset to the northeast of the active SAF or marks the inactive GHF. At ~2.5 km depth, the SAF/GHF steps at least 500 m to the southwest, where two active fault strands are constrained by seismic activity. The reflections from the steep edge of the Salinian granite and from below the surface trace of the GHF bound a vertical wedge of sedimentary rocks constrained from previous seismic and magnetotelluric explorations and by SAFOD drilling. In order to improve the spatial resolution of the velocity model, waveform tomography is being applied to the data. Preliminary results from this ongoing work will be presented.
T21A-0447
Application of Fresnel-Volume-Migration to the SAFOD2003 Data set
We present the application of 3D Fresnel-Volume-Migration (FVM) to a shot gather of the SAFOD2003 active seismic reflection survey. FVM is an extension of Kirchhoff prestack depth migration (KPSDM). The basic idea is to restrict the imaging procedure to the physical relevant part of the subsurface within the final step of transforming the seismic data from time to depth. Conventional KPSDM distributes the recorded wavefield along the corresponding two-way traveltime isochrons. An image is generated by constructive interference of these isochrons along the actual reflector elements. This method is considered as a state-of-the-art technique in obtaining high-quality images of the subsurface. However in the case of sparse sampling or limited aperture the resulting image is mostly affected by significant migration noise due to the less constructive interference of the backpropagated wavefield. Here we use the concept of Fresnel-Volumes to restrict the migration operator in a physically frequency-dependent way. The emergence angle at the receiver is determined from a local slowness analysis. Using this emergence angle as the starting direction a ray is propagated into the subsurface and the backpropagation of the wavefield is restricted to the vicinity of this ray according to its approximated Fresnel-Volume. This procedure resolves the spatial ambiguity and results in strongly reduced migration artefacts. Furthermore, steeply dipping events as well as flat reflectors are treated in the same way and no a-priori knowledge or processing is required to enhance either of them. Along with an introduction to the principles and the implementation of FVM this talk presents mainly the results of applying FVM to a single shot gather of the SAFOD2003 survey. Compared to the image obtained by KPSDM the FVM clearly shows a bunch of steeply dipping reflectors, which probably correspond to the San-Andreas-Fault system. In addition horizontally oriented structures are also visible at intermediate depths. The obtained image may serve as a basis for a more detailed future geological interpretation.
T21A-0448
Seismic Evidence for Rock Damage and Healing on the San Andreas Fault Associated with the 2004 M6 Parkfield Earthquake
We deployed a dense seismic array of 45 seismometers deployed across and along the San Andreas fault near Parkfield immediately after the M6 Parkfield earthquake on September 28, 2004 to record fault-zone seismic waves from aftershocks and explosions detonated within the rupture zone. Seismic stations and explosions were co-sited in our previous experiment in the fall of 2002. The data from repeated shots in the two surveys show ~1.0-1.5% decreases in seismic velocity within the ~150-m-wide zone along the fault trace and less changes beyond this zone, most likely due to the coseismic damage of rocks during dynamic rupture of the latest M6 event. The width of this zone characterized by greater velocity changes is consistent with the low-velocity waveguide model on the San Andreas fault, Parkfield we derived from fault-zone trapped waves (Li et al., 2004). The damage zone is not symmetric with the main fault trace but broader on the southwest side of the fault. Waveform cross-correlations for repeated aftershocks in 21 clusters, total ~130 events show ~0.7-1.1% increases in S-wave velocity within the fault zone in 3 months starting a week after the mainshock, indicating that the damaged rock has been healing and regaining the strength with time, most likely due to the closure of cracks opened in the mainshock. The healing rate was greater in the earlier stage of post-mainshock healing. We estimate the velocities within the fault zone decreased by up to ~2.5%, most likely associated with the 2004 M6 event at Parkfield. The magnitude of fault healing is not uniform along the ruptured segment; it is slightly larger beneath Middle Mountain in accordance with larger slip mapped there. The fault healing is seen at depth above ~6-7 km and is likely to be depth-dependent, too. The damage and healing progression observed on the SAF associated with the 2004 M6 Parkfield earthquake are consistent with our previous observations at rupture zones of the 1992 M7.4 Landers and 1999 M7.1 Hector Mine earthquakes. However, the fault-zone damage degree is smaller and the healing process is shorter on the SAF at Parkfield than those on at Landers and Hector Mine.
T21A-0449
Crack Damage in Core Samples From the San Andreas and Nojima Faults
Crack densities, grouped by crack length, have been measured from thin sections of granodiorites from the SAFOD pilot hole (2200 m depth) adjacent to the San Andreas fault near Parkfield, California and the NIED scientific drillhole crossing the Nojima fault (Kobe 1998 earthquake) on Awaji Island, Japan (1279 m depth). Crack densities were determined for crack lengths ranging from 0.046 to 10 mm in an attempt to quantify the mechanical integrity of these rocks taken from active fault zones. As much as possible, only broken, unhealed cracks were counted. Results were compared to crack densities measured in undeformed Eureka quartzite and in both undeformed and fractured Westerly granite. A simple but apparently robust measure of the critical crack density needed for significant crack interaction is the ratio S/L, where S is the distance between crack centers and L is crack length. Because crack-induced stresses decrease as (distance)$^{-2}$, S/L=1 is a good indicator of strong crack interactions. For all samples, the smallest cracks (0.046 to 0.100 mm) had the highest crack densities. Yet these cracks were not interacting with each other since their densities were only 20 to 40 percent of the densities needed for significant stress field interactions. By contrast, for both the SAFOD and Nojima borehole fault zone samples, cracks longer than 0.2 mm had crack densities sufficiently high to result in strong crack interactions leading to mechanically weak grain structures. In other words, both samples were in a state of structural failure for scales larger than the grain size. Both the SAFOD and Nojima samples can be considered representative of off-fault damage zone material from their respective faults. The samples look like 'typical' granodiorites to the naked eye, with no evidence of permanent shearing. Yet both samples have been shattered and have essentially lost cohesion. Both samples show repeated episodes of fracture and hydrothermal healing. We interpret these properties as resulting from periodic passage of high-amplitude dynamic stress waves associated with earthquake rupture.
T21A-0450
Lithologic Characterization of the Deep Portion of the SAFOD Drillhole
Characterizing rock types encountered in the SAFOD drill hole is a fundamental component for developing a geologic model for the observatory, and helps guide subsequent geophysical experiments and later coring. We examined thin sections of grain mounts and loose, washed grains from cuttings from 10,100 to 12,900 feet measured depths (md), and used X-ray diffraction analyses to decipher the rock types in the deeper part of the borehole. The most significant lithologic change in the deep part of the drill hole occurs at 10,300 to 10,700 ft md, where the dominant rock type encountered in the hole changes from arkosic sandstone and shale above to dark green and grey siltstone and shale below. Locally in the lower section there are zones of sheared siltstone and shale. The lithologic transition is nearly complete by 10,800 ft md. Serpentine grains are evident starting at approximately 10,900 ft, some with mesh texture (pseudomorphic after olivine), and some pseudomorphic after pyroxenes. Minor amounts of serpentine are found at the 10,970 to 11,000 ft interval, which is associated with an increase in gas in the borehole. At depths >11,000 ft md, shale and siltstone fragments containing quartz, plagioclase, biotite, and K-feldspar are common, and calcite is abundant, occurring both as separate grains and as veins, cement, and alteration minerals in other rock fragments. Fossil fragments (Inoceramus) are observed in several of the samples, indicating a minimum Cretaceous age for these rocks. At some intervals the washed cuttings have 10 to 30 % green, sheared, lozenge-shaped grains of phyllosilicate-rich rock. Deeper in the hole (11,500 to 12,500 ft) some grains exhibit sheared phyllosilicate surfaces, and some sandstone fragments contain plagioclase, biotite, and K-feldspar. Serpentine grains are found at 12,700 to 12,900 ft md, along with shale and siltstone fragments. Rock types determined from this examination are consistent with the rocks being part of the Great Valley Sequence (or sedimentary rocks of the Franciscan Complex). This suggests the borehole successfully crossed the San Andreas fault, and that the fault zone may contain slivers of Great Valley and/or Franciscan rocks at depth. Petrographic, mineralogic, and geochemical analyses of sidewall and drill core are anticipated to help resolve the structure of the fault zone at depth.
T21A-0451
Structural and Lithologic Characterization of the SAFOD Pilot Hole and Phase One Main Hole
Petrological and microstructural analyses of drill cuttings were conducted for the San Andreas Fault Observatory at Depth (SAFOD) Pilot Hole and Main Hole projects. Grain mounts were produced at ~30 m (100 ft) intervals from drill cuttings collected from the Pilot Hole to a depth of 2164 m (7100 ft) and from Phase 1 of the SAFOD main hole to a depth of 3067 m (10062 ft). . Thin-section grain mount analysis included identification of mineral composition, alteration, and deformation within individual grains, measured at .5 mm increments on an equally spaced, 300 point grid pattern. Lithologic features in the Quaternary/Tertiary deposits from 30 - 640 m (100-2100 ft) in the Pilot Hole, and 670 - 792 m (2200 - 2600 ft) in the Phase 1 main hole, include fine-grained, thinly bedded sediments with clasts of fine-grained volcanic groundmass. Preliminary grain mount analysis from 1920 - 3067 m (6300 - 10062) in the Phase 1 main hole, indicates a sedimentary sequence consisting of fine-grained lithic fragments of very fine-grained shale. Deformation mechanisms observed within the cuttings of granitic rocks from 914 - 1860 m (3000 - 6100 ft.) include intracrystalline plasticity and cataclasis. Intracrystalline plastic deformation within quartz and feldspar grains is indicated by undulatory extinction, ribbon grains, chessboard patterns, and deformation twins and lamellae. Cataclastic deformation is characterized by intra- and intergranular microfractures, angular grains, gouge zones, iron-oxide banding, and comminution. Mineral and cataclasite abundances were plotted as a function of weight percent vs. depth. Plots of quartz and feldspar abundances are also correlated with XRD weight percent data from 1160 - 1890 m (3800 - 6200 ft.) in the granitic and granodioritic sequences of the Phase 1 main hole. Regions of the both of the drill holes with cataclasite abundances ranging from 20 - 30 wt% are interpreted as shear zones. Shear zones identified in this study from 1150 - 1420 m (3773 - 4659 ft.) in the Pilot Hole occur in the same location as shear zones recognized by Boness and Zoback (2004) using borehole geophysical data. These shear zones may possibly be correlated to shear zones identified in the Phase I main hole from 1615 - 2012 m (5300 - 6600 ft). If this is the case, it can be explained by steeply dipping subsidiary fault zones, likely associated with the San Andreas Fault system.
T21A-0452
Elemental and Stable Isotope Chemistry of Cuttings and Core Samples From SAFOD Drill Hole
We have analyzed the major and minor element chemistry and stable isotope composition of samples from the SAFOD drill hole in order to document fluid-rock interactions in the San Andreas and associated fault zones. To date, we have analyzed samples from three sections of the drill hole that contain faults, mineralogical changes, drilling breaks, and/or anomalous gas shows. The elemental chemistry of samples from 10450 to 10530 ft interval (measured depth, MD) does not vary significantly. Only magnesium, yttrium, and barium increase and calcium decreases moderately down hole across this interval. The yttrium and barium variability is potentially due to variable contamination of barite-bearing mud in the cuttings. In samples from 10860 to 11000 ft (MD), iron, magnesium, and potassium increase and calcium decreases moderately down hole. In sample interval 11400 to 11540 ft (MD), iron increases and calcium decreases slightly down hole. The other major and minor elements in the XRF analyses do not vary much across these three intervals. These variations in elemental chemistry correspond to mineralogy changes across these intervals. Carbon and oxygen isotope values of carbonate veins and breccias from the 10450 and 11400 ft intervals range from 0 to +8 per mil (carbon) and 14 to 20 per mil (oxygen), consistent with carbonate precipitation from fluids that had isotopically exchanged with silicates. More work is being done to elucidate the origin of the fluid and the extent of fluid-rock interactions in the fault zones.
T21A-0453
Helium Isotope Measurements on Matrix Fluids From the SAFOD Drillcore
Matrix fluids from the SAFOD main hole at ~3060m have been isolated from subcores, core catcher and core fragments for noble gas analysis. Because SAFOD cores are under (hydrostatic to lithostatic) compressive stress, matrix fluids within this material can be released over time by a series of microcracks, diffusive transport, etc. Preliminary results indicate that this release occurs on a time scale of months for a transport scale of 1-3 cm. By placing subcores, core catcher bits, etc. in high vacuum containers as soon as possible after becoming available at the surface (on site), flushing with ultra-pure nitrogen and evacuating, the noble gases from matrix fluids have been measured in several SAFOD samples. $^3$He/$^4$He ratios from these samples ($^3$He/$^4$He = 4.9-5.5 e-7) are identical to the $^3$He/$^4$He ratios recovered from fluid samples captured during an open hole flow test (Kennedy, et al. pers. comm.) and indicate a mantle derived He component of ~4%. Assuming $^{36}$Ar is at saturation, open hole flow tests suggest $^4$He concentrations of 4-6 e-5 cc$^4$He/g-water, which is largely in agreement with matrix fluid concentrations calculated from the measured mass of volatile fluids (evaporated water) and $^4$He amounts (4-7 e-5 cc$^4$He/g-water). These data are preliminary, because at this time water and He have not been quantitatively released from the rocks. The results suggest that matrix fluids collected from core material is representative of fracture-filling fluids and represents a viable means for fluid collection when flow tests are restricted by limited permeability and/or drilling schedules. Our preliminary results indicate that our method might be suitable for constraining the long-term dynamics of fluids in the San Andreas Fault.
T21A-0454
Mesoscale Structure and Lithology of the SAFOD Phase I and II Core Samples
Core was taken from the SAFOD drill hole during Phase I and Phase II drilling. Almost complete recovery was obtained from the Phase 1 cored intervals of 4798-4824' (1476-1484 m) and 10025-10063' (3085-3096 m) measured depth, MD, and 50% recovery was realized during Phase II from 13090-13116' (4028-4036 m) MD. Although fractured and locally fragmented during the coring process, the core samples are fairly continuous across each interval. After removal of the protective sleeves, we prepared wrap-around 1:1 map tracings of the core and characterized the lithologies and mesoscopic deformation features visible on the core surfaces. The 4798-4824' cored interval consists of a medium-grained hornblende-biotite granodiorite with leucocratic phenocrysts and lenses that are weakly foliated in some places. Small shears, fractures, and veins that record both high- and low-temperature deformation are present in this cored interval. The dominant brittle deformation features are a series of subvertical fractures and moderately dipping shears, both of which contain some secondary mineral fill and display centimeter(s)-thick halos of low-grade alteration and staining of the granodiorite host. The upper section of the 10025-10063' cored interval is composed of pebble conglomerate to coarse-grained arkosic sandstone with lithic fragments of granite, sandstone, siltstone, and volcanic clasts. The beds are massive, well-cemented, and contain rare cobble-sized clasts. The lower section is a fine-grained, well-cemented arkosic sandstone that grades down-hole into a fine- to very fine-grained siltstone. Bedding is indistinct throughout this section. This interval is crosscut by numerous fracture sets that record multiple stages of deformation and fluid infiltration. The dominant brittle features include irregular, somewhat diffuse cataclastic bands, up to two cm thick, that are oriented at high angles to the core axis, and thinner, dark colored shear fractures that are up to several mm thick. Shear fractures are common in the pebble conglomerate and coarse sandstone, and display consistent kinematic indicators and preferred orientations. Two minor faults are present in this depth interval. On the basis of juxtaposition of different rock types, these faults likely have displacements of at least several meters. The 13090-13116' interval consists of shale with a few thin beds of siltsone and very fine-grained sandstone. Graded bedding, fossils fragments, and bioturbated sections are present. Deformation features include numerous small veins, scaly fabric, and some polished slip surfaces. Overall, the deformation intensity of the cored intervals is consistent with the inferred location of the sampled sections along the margins of the San Andreas fault zone. These cored sections are being compared to FMI logs to determine the true orientation of fabric elements.
T21A-0455
The Clay Mineralogy Of Fracture Coatings Observed In Mudrock Cuttings From The SAFOD Borehole
A number of mudrock drill cuttings have been investigated from shear zones at 1500 m and 3066 m depths showing abundant polished surfaces with occasional slip striations. Electron microscopy (SEM, TEM) and X-ray mineral identification study reveal the occurrence of thin clay film coatings containing smectite, chlorite, illite, quartz, albite and a possible serpentine mineral. This type of mineral assemblage was also observed along adjacent microveins and a freshly opened fracture. In order to compare these surface minerals with the drilling mud smectite, they were sequentially removed by ultrasonic treatment. The extreme sensitivity of this clay material under the TEM electron beam confirms the presence of abundant smectite on the polished mudrock surfaces, which shows compositional differences from that of the drilling mud. It contains notably more Fe, Mg and K and has a lower Si-Al ratio than the drilling mud montmorillonite. Additional differences between these two types of smectite could be recognized after hydration and chemical treatment (ethlylene glycol and glycerol solvation) whereby the natural smectite phase develops less hydrated states. In an attempt to examine the natural hydration states of these swelling smectite minerals, onsite impregnation (LR White resin) techniques have been applied. We suggest that these well oriented smectite coatings are partly precipitated from percolating solutions during fault creep.
T21A-0456
Phyllosilicate mineral assemblages, elemental compositions and microstructures from the SAFOD Main Hole
Mineral transformations and fabrics, particularly those involving phyllosilicates, can critically affect the mechanical behavior of shallow faults and may be more important than generally assumed. Thus, detailed phyllosilicate mineralogy, mineral transformation history and fabric development are central to the study of fault rocks and our understanding of fault properties. The SAFOD hole provides the opportunity to study the in situ change from undeformed protolith to fault rock, thereby establishing a reference for the characterization of the type and magnitude of changes in active fault zones. The SAFOD hole also allows for direct comparison with exhumed SAFOD analogs, which have been used as representatives for fault zones at depth. In exhumed faults several populations of discrete and mixed-layer phyllosilicates were observed, including a protolithic population (chlorite and mica), a syn-faulting population (chlorite-rich chlorite-smectite and illite-rich illite-smectite), and a post-faulting population (smectite-rich chlorite-smectite). In the SAFOD Pilot Hole multiple populations of phyllosilicates were also observed, including mixed-layer clays in the shallow sedimentary rocks and chlorite in deeper granitic rocks. These findings indicate that there are variations in phyllosilicate mineralization with (1) depth (2) strain energy/deformation mechanism/position in fault zone and (3) time. There are several preliminary zones of phyllosilicates in the SAFOD Main Hole, including: (1) a zone at 7800-8100 ft MD, marked by an increase in the amount of illite, and the appearance of a mixed-layer illite-smectite (I-S) phase; (2) a zone at 8400-8800 ft MD that includes at least one large shear zone marked large a increase in the amount of I-S and illite relative to surrounding protolith; (3) a zone at 11050-12400 ft MD marked by a large increase in the amount of chlorite that is fairly constant to the bottom of the hole, and the appearance of a mixed-layer clay. Three significant problems with the use of cuttings are (1) mixing of cuttings as they travel from the drill face to the surface, (2) "dilution" of phyllosilicate signatures from shear/alteration zones that are thinner than the 10 ft sampling interval and (3) possible contamination from montmorillonite in the bentonite drill mud. These difficulties can be overcome in part through the analysis of plucked grains of fault rock and protolith, guided by analyses of bulk cuttings. Our current research on SAFOD main hole samples consists of a suite of techniques than can be applied to the relatively small cuttings (less than a few mm). We will report on preliminary observations on cuttings, including mineralogical and elemental data (XRD and ICP), and electron beam (SEM and TEM) characterization of mineralized surface polishes in fault horizons, collectively offering a beginning framework for (1) interpreting fault zone mineral transformations and precipitation in seismically active settings, (2) testing the relevance of processes inferred from the study of nearby exhumed fault rocks, and, ultimately, (3) providing constraints on the thermo-mechanical behavior of faults.
T21A-0457
The San Andreas Fault at SAFOD: Seismic Imaging and Borehole Comparisons
We acquired three high-resolution (2.5-m CDP spacing) seismic reflection and refraction profiles across the SAFOD site in 1998 and 2003. In 1998, we acquired a 5-km-long, high-resolution seismic reflection and refraction profile from ~2 km southwest of the SAFOD drill site to about 1.5 km northeast of the surface trace of the San Andreas fault. In 2003, we acquired two additional high-resolution seismic reflection and refraction profiles that were centered on the SAFOD drill site. All three profiles provide velocity images to maximum depths of 800 m and reflection images to maximum depths of 5 km, which correlate with structures and lithologies observed in the SAFOD borehole. Our seismic images show that the San Andreas fault is located within an approximately 1.7-km-wide, southwest-dipping low-velocity zone at about 1 km depth. On the basis of velocity data, we previously interpreted traces of the San Andreas fault to lie within a wedge of sediments (Catchings et al., 2002; BSSA), a result that was confirmed by SAFOD borehole data. Small repeating earthquakes that were the principal target events for SAFOD drilling reside within the wedge of sediments and are probably related to low-strength sedimentary rocks. However, our reflection images show that additional strands of the San Andreas fault extend into more competent rock, well northeast of the SAFOD target faults. The relatively wide fault zone, inferred from our observations of numerous fault strands on seismic reflection images, suggests that SAFOD sampled only part of the San Andreas fault zone, which is wider and more structurally complex than previous believed.
T21A-0458
Fault Zone Structure of Middle Mountain, Central California
The Middle Mountain uplift feature in Central California provides an exceptional laboratory in which to study the characteristics of an active fault zone. This complex geological terrain is not distinguished by a single master fault, but rather a 1-3 kilometer wide zone with numerous sub-parallel faults and folds. In order to further understand this deformation, detailed 1:6000 scale geologic field mapping was undertaken during the 2004 and 2005 field seasons in order to establish the shallow fault zone structure. From this information, we can address questions such as the fault geometries, distribution of deformation, fault zone development and migration, and fault strength. Offset along the San Andreas fault zone has juxtaposed contrasting rock units. The northeastern side is Miocene to Pliocene marine sedimentary units and Plio-Pleistocene terrestrial sedimentary units. Dominant structures include several high angle faults striking sub-parallel to the main SAF trace that bound marble, granites, and Tertiary sedimentary units. Fault density increases near the active SAF trace. The Gold Hill fault is a reverse fault of varying southwest dip that surfaces along the eastern margin of Middle Mountain and places the Miocene Monterey shale over Pliocene Etchigoin sandstone. Alternating synclines and anticlines with axes trending parallel to the strike of the Gold Hill fault are present within the hanging and foot walls. An 800 meter thick package of the Plio-Pleistocene Paso Robles Formation dominates the southwestern terrain and is deformed by en echelon folds and secondary faults. Some of the faults strike nearly normal to the SAF, offset Tertiary and Quaternary units, and tend to be northwest-side up. Several SAF-parallel striking faults slice Tertiary and Quaternary sedimentary units and granitic bodies. A fault on the southwestern side of Cholame Creek juxtaposes Tertiary rhyolitic rocks (Pinnacle-Neenach equivalent) against Salinian granite. SAF sub-parallel fault density increases three-fold within the overlying sedimentary units, thus confirming that faults bifurcate near the surface. Folding of the mid-Tertiary units on the northeast side of fault probably results from SAF-Normal compression and movement along the Gold Hill reverse fault. The en echelon folding of the Paso Robles formation on the southwest side of the SAF is consistent with simple shear parallel to the main SAF. The transverse faults on the SW-side of the SAF are most likely high-angle normal faults either developed by SAF-parallel simple shear and later rotated to their current position, or possibly created by slip along a bend in the SAF at depth. The fault that offsets rhyolite against granite is a remnant SAF trace, and indicates ~3 km northeastward migration of the SAF trace. However, southwestern units are found on the northeastern side of the active trace, thus indicating the ability of the active trace to jump 10s to 100s of meters back to the southwest.
T21A-0459
Accessing SAFOD data products: Downhole measurements, physical samples and long-term monitoring
Many different types of data were collected during SAFOD Phases 1 and 2 (2004-2005) as part of the National Science Foundation's EarthScope program as well as from the SAFOD Pilot Hole, drilled in 2002 and funded by the International Continental Drilling Program (ICDP). Both SAFOD and the SAFOD Pilot Hole are being conducted as a close collaboration between NSF, the U.S. Geological Survey and the ICDP. SAFOD data products include cuttings, core and fluid samples; borehole geophysical measurements; and strain, tilt, and seismic recordings from the multilevel SAFOD borehole monitoring instruments. As with all elements of EarthScope, these data (and samples) are openly available to members of the scientific and educational communities. This paper presents the acquisition, storage and distribution plan for SAFOD data products. Washed and unwashed drill cuttings and mud samples were collected during Phases 1 and 2, along with three spot cores at depths of 1.5, 2.5, and 3.1 km. A total of 52 side-wall cores were also collected in the open-hole interval between 2.5 and 3.1 km depth. The primary coring effort will occur during Phase 3 (2007), when we will continuously core up to four, 250-m-long multilaterals directly within and adjacent to the San Andreas Fault Zone. Drill cuttings, core, and fluid samples from all three Phases of SAFOD drilling are being curated under carefully controlled conditions at the Integrated Ocean Drilling Program (IODP) Gulf Coast Repository in College Station, Texas. Photos of all physical samples and a downloadable sample request form are available on the ICDP website (http://www.icdp-online.de/sites/sanandreas/index/index.html). A suite of downhole geophysical measurements was conducted during the first two Phases of SAFOD drilling, as well as during drilling of the SAFOD Pilot Hole. These data include density, resistivity, porosity, seismic and borehole image logs and are also available via the ICDP website. The SAFOD monitoring program includes fiber-optic strain, tilt, seismic and fluid-pressure recording instruments. Seismic data from the Pilot Hole array are now available in SEED format from the Northern California Earthquake Data Center (http://quake.geo.berkeley.edu/safod/). The strain and tilt instruments are still undergoing testing and quality assurance, and these data will be available through the same web site as soon as possible. Lastly, two terabytes of unprocessed (SEG-2 format) data from a two-week deployment of an 80-level seismic array during April/May 2005 by Paulsson Geophysical Services, Inc. are now available via the IRIS data center (http://www.iris.edu/data/data.htm). Drilling parameters include real-time descriptions of drill cuttings mineralogy, drilling mud properties, and mechanical data related to the drilling process and are available via the ICDP web site. Current status reports on SAFOD drilling, borehole measurements, sampling, and monitoring instrumentation will continue to be available from the EarthScope web site (http://www.earthscope.org).