T13A-1335 1340h
Geologic structure of Middle Mountain within the San Andreas Fault zone near Parkfield, California
Knowledge of the geometry and history of motion of rock bodies within fault zones such as the San Andreas fault (SAF) is essential input into mechanical models of earthquake rupture dynamics and fault evolution. The Parkfield segment of the SAF is the focus of significant geophysical characterization and borehole studies. In order to enhance the geologic information about the SAF structure in this area, we undertook an intensive high-resolution geologic mapping effort (1:6000 scale) of the Middle Mountain area (about 40 km^2). The geologic structure differs dramatically across the San Andreas fault zone. The northeast side contains numerous sub-parallel faults that likely accommodated significant strike slip motion. These high-angle faults bound granite, marble, and sedimentary rock slivers. The density and complexity of these faults increases toward the center of the fault zone. The Gold Hill reverse fault on the northeast side of the SAF is a low-angle southwest-dipping fault that locally displaces the older Tertiary Monterey Formation over the younger Tertiary Etchegoin Formation. Folds with axes trending parallel to the strike of the Gold Hill reverse fault are present within the hanging wall. The Plio-Pleistocene Paso Robles Formation dominates the southwest side of the SAF and is a formidable cover. Fault-bounded granitoid slivers are also present within the southwest terrain. One fault striking nearly normal to the SAF cuts rock units near the mid-section of Middle Mountain. To the northwest of this fault, older Tertiary formations are present. The folds within the hanging wall of the Gold Hill reverse fault and the reverse fault itself indicate SAF-normal shortening near the SAF zone. The Gold Hill fault most likely cuts the numerous high-angle sub-parallel faults at depth. With the northeastward-verging nature of this fault, the cross-section on the northeast side is a roughly hewn half-flower structure. The sedimentary basin into which the Paso Robles Formation was deposited has been inverted due to subsequent deformation adjacent to the SAF and now comprises the eroding apex of Middle Mountain. Cross-sections of the southwest side of the fault vary remarkably and increase in complexity to the northwest. In the southeastern half, a pronounced syncline within the Paso Robles Formation trends sub-parallel to the SAF. In the northwestern portion, the steeply dipping Buzzard Canyon fault repeats some Tertiary units. Apatite fission track dating of the granitoid blocks will provide us with the exhumation history of the basement blocks and thus an indication of their motion and vertical offset within the fault zone. Intensive field mapping is the essential tool that allows scientists to link varying thicknesses and other intrinsic characteristics of a fault zone to the mechanical properties of the actual rock bodies. This high-resolution study significantly constrains the geology and geometry of the crustal blocks of this area and allows us to better understand the tectonic history of the Middle Mountain uplift feature. This in turn provides us with a greater knowledge of how mechanical properties of these blocks influence the observed geophysical and geodetic properties within the area.
T13A-1336 1340h
Paleoseismology and Tectonic Geomorphology: Results From the Parkfield, CA Segment of the San Andreas Fault
The Parkfield segment of the San Andreas Fault (SAF) is the transition zone between the creep-dominated segment of the SAF to the northwest and the locked segment to the southeast. Geodetic studies indicate ~10 mm/year of the SAF 35 mm/year slip-rate is accommodated by creep at shallow depths of the SAF near Carr Hill, suggesting the remaining slip may occur in moderate to large earthquakes. Foreshock intensities of the 1857 M7.8 Fort Tejon earthquake were similar to 20th century Parkfield events, suggesting that the 1857 event may have nucleated in the Parkfield area. Paleoseismic investigations near Parkfield have achieved limited success in exposing useful fault zone stratigraphy. Our goals were to investigate the style and timing of late Pleistocene and Holocene faulting along this segment of the SAF. Specifically, did the 1857 or similar large prehistoric earthquakes rupture through the Parkfield segment? In addition, is it possible to distinguish deformation caused by these large magnitude earthquakes from deformation associated with the moderate 1966-style earthquakes and aseismic creep? Geomorphic mapping from Middle Mountain to Carr Hill revealed numerous tectonic landforms that define the fault trace and indicated possible excavation sites, including the sites used for this study. Two fault-perpendicular excavations were cut in a late Pleistocene fluvial terrace of Little Cholame Creek just north-northeast of Carr Hill. A thirty-meter excavation across an elongate sag pond, containing an apparently right-laterally offset fluvial channel and several small springs, revealed four fault zones. The outer fault zones were parallel with the regional trend of the SAF (313°), while the inner fault zones trended obliquely at ~350°. The three easternmost fault zones show apparent dip-slip deformation including truncation and down warping of sag and terrace deposits. The westernmost fault zone was characterized by upward splaying sub-vertical clay shear bands, which juxtaposed the youngest sag deposits against heavily bioturbated units. The orientation and styles of deformation are consistent with local extension due to a small right step in the SAF trace. Apparently, the two central, obliquely oriented, fault zones transfer slip across the step. A second ~35 m excavation across a ~2m pressure ridge, just southeast of the sag excavation, revealed two fault zones. The primary fault zone was characterized by upward splaying clay shear bands, most of which terminated within an undifferentiated organic unit associated with a spring line midway up the scarp. Higher on the scarp a zone of dip-slip faulting, characterized by apparent normal offset and stratigraphically lower terminations, is older than the primary fault zone. Strong indicators of coseismic ground rupture such as filled fissures were not observed. However, the lack of such features does not preclude the possibility of large earthquakes in Parkfield. Several upward terminating fault strands provide possible evidence for seismic events. It is likely that much of the deformation in these exposures (the down-warped sands in particular) is due to aseismic creep. Forthcoming radiocarbon ages will provide information about deformation rates and the development of the fault related structures exposed here.
http://activetectonics.la.asu.edu/Parkfield/
T13A-1337 1340h
Microstrutural analyses of an exhumed part of the San Andreas fault near the SAFOD site, California
We examine the microstructures and textures of a suite of samples from an outcrop of the main trace of the San Andreas fault ~2.4 km north of the SAFOD drill site, in the area where the fault changes in slip behavior from locked to creeping. The fault at this exposure juxtaposes highly sheared, probably Jurassic to Cretaceous serpentinite, consisting of altered peridotite and tectonized antigorite within the fault, against Pliocene(?) brecciated sedimentary deposits on both sides of the fault. Serpentinite samples within 20 cm of the fault contact exhibit two subvertical foliations superposed on an earlier metamorphic foliation. Foliated zones 1-5 mm wide are subparallel to the SAF contact, and wavy to folded antigorite grains are cut by thin faults that consist of ultracataclasite. Foliation orientations are consistent with a right-side-up sense of shear. Samples from the sedimentary deposit exhibit Fe-filled brittle fractures and calcite veins in a random cataclastic texture within 50 cm of the fault. Within 10 cm of the fault the sedimentary rocks have a moderate foliation, and ultracataclasite fragments lie in a matrix of sheared sandstone. Beyond 50 cm of the fault the sedimentary rocks exhibit variable amounts of deformation ranging from regions of intense intragranular fracture to nearly undeformed grains. These observations suggest that at this site the SAF consists of a narrow zone of highly localized shear within the foliated host rock, and that seismic ruptures created a brittle damage zone on the SW side of the fault. Further analyses of these, and other exhumed parts of the SAF near SAFOD will provide constraints on the nature of the fault zone that will be encountered in the SAFOD drill hole.
T13A-1338 1340h
Parkfield Unified Visualization Project: A Repository of Geospatial Data and Portable Toolset for its Visualization
Extensive geophysical studies have been conducted over the past decades in and around Parkfield, CA. These studies, while significant on their own, often fail to take into account the full body of pre-existing geophysical data for the Parkfield region. This is due in large part to the difficulty in integrating much of this data, which in turn results from a lack of a centralized, standardized geospatial repository in which all the available geophysical and geological data are stored in consistent units and in a consistent, georeferenced coordinate system. We are developing the Parkfield Unified Visualization Project (PUVP) which will be just such a repository, with the added benefits of easy, WWW-driven access and an in-house Virtual Reality Markup Language (VRML) generator to allow visualization of multiple data sets over the web. The data are stored georeferenced, and queried using the GRASS GIS. The PUVP database currently contains 30-meter USGS DEM coverage over a 65x85 km study area, with USGS DOQQ coverage over approximately 70% of this area. Additional contributed data includes landslide data, a digitized trace of the 1966 rupture, seismic p-velocity model results, and historical seismicity data from the Northern California Earthquake Data Center. The user queries this data via a Java web form which runs a series of Perl scripts to extract this data from the GRASS database in either GRASS or ARC/ESRI format. The scripts also convert this data into VRML directives that can be displayed using any web browser with a VRML plugin, or on a GeoWall. We chose VRML as our visualization environment for several reasons. Foremost, it is quite portable and allows a wide variety of users to bring the different datasets into a common visualization environment so that non-intuitive relationships between different observations or model results can be made and interpretations better illustrated. In addition, unlike several other 3D visualization environments like Vis5d+ or the NVIZ tool available within GRASS, VRML has no single intended use, and is therefore much more adaptible to the needs of the project. For example, because VRML has no native resolution or axis orientation, multiple datasets with diverse resolutions can be displayed in multiple orientations in the same scene, without aliasing or over/undersampling. This project is completely open source, and we hope will be built up by contributions of both data and code by the user community. The repository will continue to grow, residing on a GEON data node at ASU with continued user input and contributions. Simultaneously, the programs that run queries and visualization of the data will continue to expand in capability and grow in sophistication, thanks to the continuing efforts of the user community.
http://activetectonics.la.asu.edu/Parkfield/vis/
T13A-1339 1340h
Structure, Kinematics and Recurrence of Microseismicity in the SAFOD Target Region
Large numbers of small earthquakes recorded over 2 decades and analyzed with advanced techniques are used to characterize the detailed structure and kinematics of the fault and the recurrence interval scaling properties of repeating micro-seismicity in a 4 x 4 km lateral and 6 km deep crustal volume surrounding the SAFOD experiment's deep drilling zone. The characterizations reveal that the seismically active San Andreas fault in the vicinity of SAFOD's repeating magnitude 2 target earthquakes is composed of two sub-parallel fault strands that are creeping at comparable rates and that one of the strands lies between the SAFOD drilling platform and SAFOD's target events. In the study region, $\sim$ 50 % of the earthquakes are members of 52 characteristically repeating earthquake sequences. The recurrence properties of the two repeating target event sequences are also characterized and are found to be consistent with the recurrence scaling of the other sequences in the study zone. This scaling, however, is contrary to that expected from standard constant stress-drop theory. SAFOD's 'in-situ' measurements of fault zone processes and its near-field seismic observations of repeating M2 earthquake sequences should provide the information needed to resolve the apparent discrepancy between the observed and expected microearthquake recurrence behavior.
T13A-1340 1340h
Scattered Wavefield Within the San Andreas Fault System, California
Since transient aseismic deformation at seismogenic depth is one of the key phenomena related to earthquake occurrence, it is important to estimate the physical characteristics of such stress/strain transients within the deeper structure of the fault zone. Analysis of coda (scattered) waves has the potential for identifying such transients because the scattered wavefield is attributed mainly to small-scale heterogeneity that is likely formed by these transient events. In addition, the sampling area of scattered waves is concentrated within the fault zone, compared to the area sampled by direct waves. We have begun a program to map the spatial distribution of fault-zone scatterers and their time dependence within two regions of San Andreas Fault system: the Hayward Fault and the Parkfield segment of the San Andreas Fault. In order to most reliably evaluate the scattered wavefields, we limited our analysis to records from borehole seismographs recorded by the Hayward Fault Network (HFN) and the High-Resolution Seismic Network (HRSN) in each area. For the Hayward fault, we mapped the spatial distribution of scatterers by analysis of the S-wave coda amplification factor (CAF) in a manner similar to Taira and Yomogida (2003). CAF is defined as the amplitude ratio of coda waves among different stations after corrections for source, station, and overall propagation effects (e.g., coda Q). This parameter allows for a statistical characterization of the distribution of scatterers. The station effect for each station and the coda Q averaged over all the seismograms in this area were estimated by the coda-normalization and maximum likelihood methods, respectively, using five regional earthquakes (epicentral distance $>$ 50 km). We evaluated the CAF value of each source-station pair for the transverse-component, using 294 seismograms for 39 local earthquakes (epicentral distance $<$ 50 km) recorded by 14 stations of the HFN. A map of CAF values for stations near the Hayward Fault reveals that they are relatively large ($>$ 2.1 times) along the north-western part of the fault. This result suggests that heterogeneities are concentrated in the north-western segment of this fault, manifested locally as a large degree of scattering. This spatial variation of the scattered wavefield along the Hayward Fault may be related to variations in the stress/strain field. In the case of the Parkfield segment, the high density of borehole seismographs permits the identification of individual scatterers. We applied the imaging method of Taira and Yomogida (2004) incorporating the source array technique presented by Spudich and Bostwick (1987). We selected seismograms from a cluster of 23 aftershocks of the December 20, 1994, M=4.7 Parkfield earthquake, recorded at 10 stations of the HRSN. First, we relocated these aftershocks, using the earthquake catalog data (location and phase) produced by the Northern California Seismic Network and the double-difference earthquake relocation algorithm introduced by Waldhauser and Ellsworth (2000). We then analyzed the coda of observed seismograms in the frequency range 2 - 4 Hz where the signal level was highest, and found what we interpret as a strong S-to-S scatterer located in the south-eastern part of this area at a depth of between 6 - 12 km. We test the hypothesis that this scatterer is related to the 1993 Parkfield aseismic transient event (Gao et al., 2000) by assessing possible time dependence in scatterer properties, through the use of a larger suite of events that spans the time period of the transient.
T13A-1341 1340h
A search for temporal variations in the scattered wavefield associated with the 1993 Parkfield aseismic transient event: a calibration between borehole and surface instruments
Extracting subsurface deformation information from temporal changes in the seismic properties of the crust has been a long-time pursuit of seismologists. One of the more promising approaches to this problem has been the comparison of the codas of similar events occurring at different times and recorded at the same station. If one forms a cross correlation, {\it C(t)}, within a time window centered on elapsed time {\it t}, the lag time {\it $\tau$} at the maximum of the cross correlation $C_{m}(t)$ and the decorrelation index {\it D(t)} (defined as {\it 1-$C_{m}(t)$}) are both measures of the temporal change in the medium, if the contribution from differential event location is negligible. The variation can be due either to a temporal change in velocity, or a change in the location or amplitude of scatterers that produce the coda. This approach has been successfully applied to similar events that have followed major earthquakes. The change appears to reflect the gradual post-seismic healing of cracks in the shallowest crust. We have previously applied this approach to a repeat-earthquake dataset recorded by the High Resolution Seismic Network (HRSN) along the Parkfield segment of the San Andreas Fault ({\it Niu et al.}, 2003). We found a systematic difference in the S-wave coda between earthquakes that occurred before and after the onset of the 1993 Parkfield aseismic transient, which was observed on a variety of geodetic and seismic instruments in the region. We additionally found that the difference can be explained by the motion of scatterers located very close to the initiation region of the aseismic transient. Given the close correspondence between the change in the medium and the aseismic event, we hypothesized that the observed structural change was due to a change in the stress field induced by the aseismic event. We are presently expanding the scope of this study to include a larger section of the San Andreas fault system. While low-noise borehole networks are particularly well suited to the task of detecting small variations in the scattered wavefield, their spatial distribution is presently limited. We have thus begun to evaluate the utility of the noisier, but far more numerous surface stations that would provide greater density and spatial coverage. To address this issue, we are conducting a calibration experiment at Parkfield between the borehole (HRSN) and surface network (NCSN) stations to examine whether the temporal change observed with the HRSN can be successfully performed with data recorded by surface instruments. We find that in general, the signal-to-noise ratio of the NCSN data is about 2 orders of magnitude lower than that of the HRSN data, even in the case where the NCSN and HRSN stations are very close to each other ($<$ 0.1 km). The resulting decorrelation index between repeat seismograms for the surface stations is $\sim$0.1, which is about a factor of 20-100 higher than those calculated from the borehole stations ($\sim$0.001-0.005). Despite the high noise level of the NCSN data, however, we are nevertheless able to detect the changes between the pre- and post 1993 seismograms. This promising result suggests that our technique can be applied to any region of the San Andreas fault system that possesses both repeat earthquakes and a reasonable distribution of surface stations.
T13A-1342 1340h
Combined Teleseismic and Local Earthquake Tomography of the SAFOD Drill Site
Using data from the PASO array in Parkfield, California, we derived an image of P wavespeeds and Vp/Vs ratios in the crust and upper mantle beneath the region of the SAFOD drill site though a combined teleseismic and local arrival time inversion. Relative teleseismic arrival times are determined using a waveform correlation technique similar to that described by VanDecar and Crosson (1990). In all, our teleseismic data set consists of about 14,000 arrivals from 320 events, which we combine with about 60,000 P and S arrival times from 636 local earthquakes and 103 shots. The broader ray coverage afforded by the teleseismic arrivals allows us to resolve structure below the upper 8-10 km sampled by local events to depths of about 30 km, and the combination of data sets reduces the potential for streaking of the strongest heterogeneities near the upper parts of the fault zone to deeper depths. Nevertheless, we find that inversions using teleseismic arrivals alone are capable of resolving the first order features apparent in the local earthquake tomography results of Thurber et al. (2004) and Roecker et al (2004).
T13A-1343 1340h
Controlled Source $P$ and $S$ Wave Tomography at SAFOD
Combined analysis of seismic $P$ and $S$ velocity information provides a basis for petrologic interpretation of seismic cross-sections. The methodology presented here allows extraction of rock-body composition and shape from well-defined $P$ velocities and Poisson's ratios that are derived from tomographic $P$ and $S$ velocity models. Refraction/reflection seismic data were acquired along a 46-km long line perpendicular to the San Andreas Fault centered on the SAFOD drill site. A fixed array of 912 three-component stations at a 25-50 m spacing recorded 62 explosive shots (25-100 kg) at a 0.5-1 km spacing. Over 45000 $P$-wave and 20000 $S$-wave travel times could be picked. P- and S-velocity models were determined independently by travel-time tomography of direct refracted phases, resulting in high-resolution images of the upper crust down to 5-km depth. The 2-D models contain strong velocity anomalies for both $P$ and $S$ waves. Resolution tests have been carried out to evaluate the quality of the tomographic models. $V_p$/$V_s$ ratios, calculated from the $P$ and $S$ wave velocity are generally high, exceeding values of 2. Errors in forming the $V_p$/$V_s$ ratio from the $P$ and $S$ wave models are low in the shallower parts of the model and become larger at greater depth. Cluster analysis procedures have been applied to the $P$- and $V_p$/$V_s$ models. First, prominent clusters were identified in plots of $V_p$/$V_s$ versus $P$-wave velocity. Secondly, the identified clusters were mapped back into spatial distributions in the seismic depth section. These clusters are interpretable as specific rock types; they constrain rock types better than $P$ or $S$ velocities alone. Prominent features are the sedimentary bodies and the high-velocity granitic rocks of the Salinian terrane.
T13A-1344 1340h
Depth-Dependent Low-Velocity Structure of the San Andreas Fault near the SAFOD Drilling Site at Parkfield from Fault-Zone Seismic Waves
Coordinated by the SAFOD PIs, we used 96 PASSCAL short-period three-component seismometers in linear arrays deployed across and along the San Andreas fault (SAF) near the town of Parkfield and the SAFOD drilling site in 2002 and 2003, respectively. The data recorded for near-surface explosions detonated in the experiments (Li and Vidale), PASO project (Thurber and Roecker) and refraction profiling (Hole), and local earthquakes show fault-zone trapped waves clearly for the source and receivers located close to the fault. The time duration of the dominant trapped energy after S-arrivals increases with the event-to-array distance and focal depth progressively. Using a finite-difference code, we first synthesize fault-zone trapped waves generated by explosions to determine the shallowest 1 or 2 km fault zone structure with the velocity constraints from seismic profiling of the shallow SAF at Parkfield [Catchings et al., 2002]. We then strip shallow effects to resolve deeper structure of the fault zone, and synthesize trapped waves from earthquakes at depths between 2.5 and 11 km to complete a model of the SAF with depth-variable structure in 3-D. We also use the P-first arrivals and polarity as additional information in modeling of velocities and location of the material interface with the structural constraints from seismic tomography at Parkfield [Thurber et al., 2004] to the bed-rock velocities. In grid-search modeling, we tested various values for fault zone depth, width, velocity, Q, and source location. The best-fit model parameters from this study show evidence of a damaged core zone on the main SAF, which likely extends to seismogenic depths. The zone is marked by a low-velocity waveguide ~150 m wide, in which Q is 10-50 and shear velocities are reduced by 30-45% from wall-rock velocities. We also find some seismic energy trapped partitioned in the branching faults that connect to the San Andreas main fault at a shallow depth near Parkfield.
T13A-1345 1340h
The Drill Bit Seismic Project at SAFOD
In July 2002, scientists from Duke University successfully installed a downhole seismic array at the SAFOD site in collaboration with SAFOD Principal Investigators Bill Ellsworth (USGS), Steve Hickman (USGS), and Mark Zoback (Stanford University). As part of ongoing seismic studies at SAFOD, the vertical array of three component seismometers is currently being used in a drill bit seismic experiment in conjunction with two surface geophone arrays donated by Schlumberger, Inc. The purpose of this experiment is to collect an inverse vertical seismic profile from signals generated by the drill bit while drilling the SAFOD borehole. These signals are being simultaneously recorded by the three geophone arrays and accelerometers installed on the top drive of the drill rig. The drill bit itself acts as a downhole seismic source capable of illuminating its surrounding. The energy generated by the drill bit is then recorded by the geophone arrays. This type of data collection has been referred to as the "drill bit seismic method". The goal of the joint project between Duke University and Schlumberger is to make use of the borehole drilling operation itself as a source of seismic energy for imaging the complex subsurface geologic structure of the San Andreas Fault while the drilling progresses. An additional goal of the experiment is to record fault zone guided waves at the location of the Buzzard Canyon fault and other unmapped faults between the drilling rig and the San Andreas Fault. In this paper, we present an overview of the DBSeis processing architecture, initial results of our experiment, and preliminary subsurface images of the SAFOD site.
http://www.icdp-online.de/sites/sanandreas/public/dbseis.html
T13A-1346 1340h
High-Resolution Fault Zone Monitoring and Imaging Using Long Borehole Arrays
Long borehole seismic receiver arrays are increasingly used in the petroleum industry as a tool for high--resolution seismic reservoir characterization. Placing receivers in a borehole avoids the distortion of reflected seismic waves by the near-surface weathering layer which leads to greatly improved vector fidelity and a much higher frequency content of 3-component recordings. In addition, a borehole offers a favorable geometry to image near-vertically dipping or overturned structure such as, e.g., salt flanks or faults. When used for passive seismic monitoring, long borehole receiver arrays help reducing depth uncertainties of event locations. We investigate the use of long borehole seismic arrays for high-resolution fault zone characterization in the vicinity of the San Andreas Fault Observatory at Depth (SAFOD). We present modeling scenarios to show how an image of the vertically dipping fault zone down to the penetration point of the SAFOD well can be obtained by recording surface sources in a long array within the deviated main hole. We assess the ability to invert fault zone reflections for rock physical parameters by means of amplitude versus offset or angle (AVO/AVA) analyzes. The quality of AVO/AVA studies depends on the ability to illuminate the fault zone over a wide range of incidence angles. We show how the length of the receiver array and the receiver spacing within the borehole influence the size of the volume over which reliable AVO/AVA information could be obtained. By means of AVO/AVA studies one can deduce hydraulic properties of the fault zone such as the type of fluids that might be present, the porosity, and the fluid saturation. Images of the fault zone obtained from a favorable geometry with a sufficient illumination will enable us to map fault zone properties in the surrounding of the main hole penetration point. One of the targets of SAFOD is to drill into an active rupture patch of an earthquake cluster. The question of whether or not this goal has indeed been achieved at the time the fault zone is penetrated can only be answered if the rock properties found at the penetration point can be compared to the surrounding volume. This task will require mapping of rock properties inverted from AVO/AVA analyzes of fault zone reflections. We will also show real data examples of a test deployment of a 4000 ft, 80-level clamped 3-component receiver array in the SAFOD main hole in 2004.
http://www.paulsson.com
T13A-1347 1340h
Accurate measurement of P and S wave travel times under controlled laboratory conditions over long time scales
The measurement of small changes in travel time associated with changes in stress, crack, or fluid distribution has been the goal of many seismological studies in regions of seismic risk. Several recent studies have shown that properties of the seismic wavefield recorded in the same region over a period of years can reveal temporal changes, probably associated with transient deformation in the fault region. Other studies, however, have shown that seasonal climate changes can also influence seismic properties. We seek to develop a laboratory-based quantitative relationship between changes in stress and wavefield properties that also accounts for the possible influence of environmental effects. For this purpose we are utilizing a small triaxial rig with pore fluid connections, P and two perpendicularly polarised S wave transducers in each piston to study the effect of small controlled changes in deviatoric stress, confining pressure and pore fluid pressure on ultrasonic traces under isothermal conditions. Our objective is to obtain velocity precision (fractional change in travel time) of order 10-6, which is the precision needed to monitor stress changes of order 1KPa. At this level it is possible detect stress changes due to tidal stresses and atmospheric loading (two useful calibrating signals in field applications), as well as tectonic activity. Achieving this level of precision requires a variety of procedures that maximize both signal-to-noise ratio and resolution. For example, we are developing a procedure for the massive stacking of waveforms on a digital oscilloscope with a high sampling rate of 2.5GS/s (0.4ns resolution). We are also experimenting with the characteristic frequency and waveform characteristics of the source. We have begun a series of experiments on thermally cracked Westerly granite cores (38mm diameter and 80mm length). The samples are being subjected to temporal variations in deviatoric stress, confining pressure, and pore fluid pressure over a variety of timescales to establish the stress sensitivity of seismic wavefield properties.
T13A-1348 1340h
Airborne Gravity Gradiometer Survey Over the San Andreas Fault
An airborne gravity gradiometer survey is being conducted over the San Andreas Fault drill site. The survey is over a 10km x 10km area. Forty lines will be flown in a direction parallel to the strike of topography; the lines, 250 meters apart, will each be 10km in length. Ten lines 1km apart will be flown in a perpendicular direction. The lines will be flown on a drape surface at a nominal ground clearance of 200 meters, although the ground clearance will be substantially greater in many parts of the survey because of the presence of steep topography. The principal objective of the survey is to detect density inhomogeneties associated with the San Andreas Fault which lie at a shallow depth (less than 1km). Terrain corrections are very large in the area of the survey and correct application of the corrections is essential for detection of sub-surface density variations. The survey will be carried out by Bell Geospace using its six component airborne gravity gradiometer. We expect to be able to present some preliminary results at this meeting.
T13A-1349 1340h
Imaging the Deep Roots of the San Andreas Fault and the Dead Sea Fault with Magnetotelluric Measurements
We give details of a planned new magnetotelluric (MT) experiment at Parkfield and present results of a similarly designed investigation of the Dead Sea Transform fault. MT studies have always played an important role in the investigation of the San Andreas Fault (SAF) but due to short profile lengths, they have generally focused on the geometry and nature of the upper crustal fault. Continental transform faults separate plates of lithosphere but it is still debatable if these faults cut rigid blocks as narrow shear zones or if the lower crust decouples and deforms continuously over a wide area. A geophysical transect across the Dead Sea Transform (DST) in Jordan and Israel reveals a narrow, approximately 3-5 km wide, sub-vertical zone of high electrical conductivity penetrating the lithosphere to a depth of at least 30 km. This image strongly suggests that the DST is spatially confined even in the ductile lower crust. However, the situation may be different at the SAF. Results from the brittle crust show some similarities in these major transform fault systems, but also marked differences. Field work in Parkfield will start in spring 2005 with a high resolution MT survey following the 50 km reflection / refraction seismic profile of Hole & Ryberg. Additional MT sites will be deployed along and off the main 50 km profile for areal coverage. Acquiring seismic and magnetotelluric data along a coincident profile gives the opportunity to utilize the combined imaging power of the two methods
T13A-1350 1340h
Determination of Thermal Properties at the SAFOD Site Through Cross-Hole Temperature Monitoring
The USGS is investigating the thermal state of the San Andreas fault (SAF) at the site of the San Andreas Fault Observatory at Depth (SAFOD) near Parkfield, California, through a series of detailed heat-flow measurements in both the pilot hole, drilled in 2002, and the main fault-crossing hole, spudded in the summer of 2004. The proximity of the existing SAFOD pilot hole to the new, fault-crossing main hole (ranging from 5 to 10 m in the upper 1100 m) provides the opportunity to utilize pilot hole temperature measurements to monitor the thermal effects of the drilling disturbance in the fault-crossing hole. The time evolution and magnitude of the resulting temperature changes in the pilot hole reflect the bulk thermal properties of the rock mass between the two holes. Analysis of these temperature changes addresses a significant limitation on the accuracy of heat-flow measurements in both holes. This is the absence of whole-rock thermal conductivity determinations in the sediments from the upper 770 m of the drilled section, for which there are no cores or open-hole wireline logs. Although grain density and thermal conductivity measurements on drill cuttings from the pilot hole provide values for the rock matrix, significant uncertainty remains regarding porosity and whole-rock conductivity. The time series of pilot-hole temperature measurements have been compared with a two-dimensional numerical model of the radial propagation of the temperature disturbance from drilling the fault-crossing hole. Measurements of the temperature disturbance are ongoing as drilling continues, but preliminary results indicates that the thermal diffusivity of the sedimentary section averaged over 150 m depth intervals ranges from 5.5 to 8.5x10$^{-7}$ m$^{2}$ s$^{-1}$. Analysis of these estimates with the sediment mineral composition determined from drill cuttings yields a range of thermal conductivity from 2.2 to 2.6 W m$^{-1}$ K$^{-1}$ and indicates that heat flow in the sediments is approximately equal to the value of 91 mW m$^{-2}$ measured in the basement below 770 m.
T13A-1351 1340h
Frictional heterogeneity and heat flow
The magnitude of frictionally generated heat varies as a function of the frictional strength of a fault. The heterogeneity of natural faults suggests that realistic models of frictional heat generation should consider variations in frictional strength. Here we use numerical models to explore the effects of faults with spatially and temporally heterogeneous frictional strength on the spatial distribution of surface heat flow. Lateral variations of friction along strike combined with the lateral displacement of the blocks by a strike-slip fault requires a non-linear solution to heat flow equations and can produce heat flow patterns that are asymmetric across the fault and along-strike. This asymmetry has implications for conclusions about fault strength drawn from existing heat flow measurements. We explore a range of slip rate-asperity size combinations to determine the limit in which a heterogeneous fault is indistinguishable from a fault with uniform frictional properties. We then compare predictions from models with available heat flow data in the vicinity of the SAFOD drill site and surrounding areas near the creeping section of the San Andreas fault. Because the creeping section does not produce large earthquakes, it is insensitive to frictional weakness caused by dynamic mechanisms. We can directly test the importance of dynamic weakening by treating the section as a frictional heterogeneity.
T13A-1352 1340h
Rock Deformability, Brittle Fracture, and Strength Criteria in and Adjacent to the San Andreas Fault Zone; Applications to Tectonic Stress Estimation
A research project is underway at the University of Wisconsin-Madison, the principal objective of which is to investigate deformability, brittle fracture, and rock strength criteria adjacent to and inside the San Andreas Fault Zone. This laboratory study is being carried out on core extracted during the drilling of SAFOD. The rock mechanical properties characterized through this study are fundamental in efforts to understand the mechanisms that bring about earthquakes along the Fault. A unique true-triaxial apparatus is used to obtain realistic empirical strength criteria for the rocks represented by the extracted core. The true triaxial strength criterion is crucial to assessing the tectonic stress conditions when used in conjunction with the dimensions of borehole breakouts logged along the hole. These tests also provide necessary information on deformational properties such as elastic moduli, volumetric strain, dilatancy onset, as well as brittle fracture mechanism, of the exhumed fault and country rocks. The study seeks to answer questions such as how these rock properties vary across the fault zone and beyond; how they depend on environmental factors like in situ stress, mineralogy, pore pressure, and temperature; and how mechanical properties measured on core samples compare with properties inferred from borehole logs and surface-based geophysical observations. This comprehensive study will provide critical input of fundamental importance to SAFOD's scientific objectives of understanding fault zone processes and earthquake physics.
T13A-1353 1340h
Dynamics of Chi-Chi earthquake rupture: Discovery from Seismological Modeling and Taiwan Chelungpu-fault Drilling Project (TCDP)
Taiwan Chelungpu-fault Drilling Project (TCDP) for understanding earthquake physics for a large slip was commenced in Feb., 2004. This project plans to continue coring from the depth of 500 m to 2000 m. The drilling was expected to hit the Chelungpu fault, which ruptured during the 1999 Chi-Chi, Taiwan, earthquake, at around 1000 m. In Aug., 2004, the drilling came to the ChinShui Shale formation around 1025 m to 1300m. Several possible fault zones were identified at the depths of 1111, 1153, 1222, and 1241m. As identified by on-site geologists from the geometry and setting of the fault zone, the depth at 1111m is the most possible fault zone associated with the recent ruptured Chi-Chi earthquake. The identified fault zone has a dip angle of about 20 degree with slip direction of 70 degree, consistent with the value obtained from kinematic waveform inversion for spatial slip distribution. The identified fault zone from the core shows a thickness of about 2-4 cm. Although detail experiment on the core is necessary for further justification, several dynamic parameters can be estimated at the first hand. Considering the amount of slip at the slip zone of 8m in dip direction of 30 degree, the scale of the slip thickness to the final slip is in the scale of 0.5-1x10-2. It yields the temperature rise due to fault slip to a value of about 400-800C for initial stress of 100 bars. If we consider a total release of stress drop of 300 bars obtained from dynamic modeling as the initial stress, the temperature rise for this event will be up to 1200-2400C. For the high temperature, possible fluid pressure, melting and lubrication could take place in the fault zone. These dynamic processes can be further justified from experiments on core. Other further parameters as fracture energy and scaled energy of this event can be also derived through the direct evidence of the core. These numbers can be compared with the dynamic waveform modeling results. The TCDP, indeed, provides the unique opportunity to really see the recent ruptured slip zone to confirm the hypothesis on rupture behavior of faulting.
T13A-1354 1340h
Preliminary Summary Of Current Fault Zones In The Hole-A Of TCDP
Hole-A of Taiwan Chelungpu Drilling Project (TCDP) has drilled through four thick fault zones at the depth of 1111m, 1153m, 1222m and 1241m, respectively. Here we will note preliminary observations of the structures, lithology, bedding dip, and fault architecture for these fault zones. Four structure groups with different senses of shear are identified from the Hole-A of TCDP. The dip direction and dip of bedding between 500-1300m is 105/20-40. For the structures, thrust fault of 355-035/20-60 is very commonly observed. The FZA1111 and FZA1153 are belonged to the thrust-fault group. Strike-slip faults with normal shear are also frequently observed. Sinistral fault is orientated as 075-105/30-80, and calcite-filling is common as calcite-steps on these fault planes. The FZA1222 is belonged to the sinistral fault with normal shear. The attitude of dextral fault is 275-305/30-80. Backthrust of 215-155/30-70 is limited to FZA1241 and few locations of hanging walls of FZA1111 and FZA1153. Normal faults of 355-035/5-10 with distinct slickenside are observed below FZA1222. Lithology between 500-1300m is mainly sandstone, siltstone or their combinations of alternation or interbeds and different degrees of bioturbation. However, the four fault zones are within the siltstone-dominant region. For FZA1111 and FZA1153, the grain size of hanging wall is finer than that of foot wall. The feature of lithology is opposite in the FZA1222, i.e. the grain size of hanging wall is coarser. For FZA1241, the lithology of hanging wall and foot wall are the same. The distribution of bedding dip is influenced by the fault zones. This variation of bedding dip with depth is related with the distribution of the FAZ1111, FZA1153 and FZA1241. Bedding dip gradually changes from 30 to 20 when approaching the FZA1111. Then it quickly became steeper around 35 again after we pass through the fault zone. The same kind of variation of bedding dip was observed in the FZA1152 as well. This variation of the bedding dip could be explained by the drag effect of thrusting. However, the disturbance of bedding dip around FZA1241 is different. There is no distinct change of bedding dip when approaching the fault zone but the dip becomes sub-vertical within the fault zone. The architectures of the FZA1111 and FZA1153 show some characteristic patterns. From top to bottom, the architectural patterns change from fractured zone to intensely fractured zone, breccia zone, gouge zone, black material (carboneous?), and at last, thin layer of breccia zone or fractured zone. Therefore, the damaged zone is thicker above the fault core than below, and thus, the lower boundary is appears sharper compared to the broad upper boundary. Another notable feature of the fault zones is that there usually is a thin layer of coal-like black material at the bottom part of the fault core. The architecture of the FZA1222 is opposite to the shallower fault zones. That is, the gouge and coal-like material is at the top and breccia zone and fractured zone are at the bottom of the fault zone.