T23E-01 INVITED
Overview of SAFOD Phases 1 and 2: Drilling, Sampling and Measurements in the San Andreas Fault Zone at Seismogenic Depth
In this talk we provide an overview of on-site drilling, sampling and downhole measurement activities associated with the first two Phases of the San Andreas Fault Observatory at Depth. SAFOD is located at the transition between the creeping and locked sections of the fault, 9 km NW of Parkfield, CA. A 2.1 km deep vertical pilot hole was drilled at the site in 2002. The SAFOD main borehole was drilled vertically to a depth of 1.5 km and then deviated at an average angle of 55° to vertical, passing beneath the surface trace of the San Andreas fault, 1.8 km to the NW at a depth of 3.2 km. Repeating microearthquakes on the San Andreas define the main active fault trace at depth, as well as a secondary active fault about 250 m to the SW (i.e., closer to SAFOD). The hole was rotary drilled, comprehensive cuttings were obtained and a real-time analysis of gases in the drilling mud was carried out. Spot cores were obtained at three depths (at casing set points) in the shallow granite and deeper sedimentary rocks penetrated by the hole, augmented by over fifty side-wall cores. Continuous coring of the San Andreas Fault Zone will be carried out in Phase 3 of the project in the summer of 2007. In addition to sampling mud gas, discrete fluid and gas samples were obtained at several depths for geochemical analysis. Real-time geophysical measurements were made while drilling through most of the San Andreas Fault Zone. A suite of "open hole" geophysical measurements were also made over essentially the entire depth of the hole. Construction of the multi-component SAFOD observatory is well underway, with a seismometer and tiltmeter operating at 1 km depth in the pilot hole and a fiber-optic laser strainmeter cemented behind casing in the main hole. A seismometer deployed at depth in the hole between Phases 1 and 2 detected one of the target earthquakes. A number of surface-to-borehole seismic experiments have been carried out to characterize seismic velocities and structures at depth, including deployment of an 80-level, 240-component seismic array in SAFOD in the spring of 2005. With knowledge of P- and S-wave velocities obtained from the geophysical measurements in conjunction with downhole recordings of the SAFOD target earthquake, it appears that the seismically active main trace of the fault is on the order of 400 m SW of the surface trace, in proximity to several candidate zones of particularly anomalous geophysical properties. Observations of casing deformation to be made over the next several years, as well as monitoring of the microearthquakes using seismometers directly within the fault zone, will pinpoint the exact location of this and other active fault traces prior to continuous coring in Phase 3. As will be elaborated in detail by the presentations of the SAFOD science team at this meeting, the activities carried out as part of Phases 1 and 2 of SAFOD lay the ground work for years of exciting research in earthquake physics, fault-rock geology, rock mechanics and the role of fluids and gases in faulting and earthquake generation.
T23E-02
Real-Time Fluid and Gas Monitoring During Drilling of the SAFOD Main Hole in Parkfield, CA.
Little is known about the role and origin of fluids and gases associated with the San Andreas Fault zone (SAF). To gain information on fluids and gases at depth, we performed real-time mud gas monitoring during drilling of the SAFOD (San Andreas Fault Observatory at Depth) Pilot Hole (PH) and Main Hole (MH). Gas extracted from returning drill mud was piped into a nearby laboratory trailer and analyzed on-line. Permanent gases were detected using a portable mass spectrometer, hydrocarbons with a gas chromatograph, and the $^{222}$Rn-activity with a Lucas-Cell detector. When significant amounts of non-atmospheric gases were detected, off-line gas samples were collected from the gas line for further isotope studies. The SAFOD PH and MH were drilled in only a few meter distance, but in contrast to the straight PH, which penetrates through 768 m of sediments into granites down to 2168 m target depth (TD), the nearby MH is deviated towards the SAF and returns into sedimentary strata below 1930 m. The MH drilled sedimentary rocks down to 3987 m TD, approximately 45 m northeast of the surface trace of the SAF. From surface to 1930 m, the depth distribution of gas is similar for SAFOD PH and MH. Shear zones, identified by geophysical logging, are often characterized by elevated concentrations of CH$_{4}$, CO2, H2, Rn, and He. The same gases were found in the MH below 1930 m, but their concentrations were, with the exception of He, significantly higher: CH$_{4}$, CO2, and H2 sometimes reach several volume percent. Generally, the gas composition is partly controlled by the lithology. Variation in the methane concentration in several depth intervals reflects the changes in lithology from low gas abundance in clays and silts to more gas rich shales, which are the source rocks for hydrocarbons. Highly porous and permeable sandstone yield the highest concentrations of hydrocarbons (up to 15 vol% methane), and may be regarded as reservoir rocks. We interpret high radon activities in mud gas as indicator for circulating fluids entering the borehole via fractures. These fluids are also rich in hydrocarbons, carbon dioxide, and hydrogen, but only low concentrated in helium. Such intervals could be identified in several depth intervals (2675-2750 m, 2825-2900 m, and 3550-3650 m depth, and below 3700 m). The hydrocarbons in the surrounding rocks show a similar composition as those associated with fault zones. In addition to the low helium concentration, these results demonstrate fluid migration from the nearby with only little evidence for gas migration from a deeper source. A striking observation is the high amount of hydrogen found in these intervals. We can exclude a significant contribution of artificial hydrogen (drilling artifact) and mantle hydrogen. From soil gas studies, it is known that fault zones sometimes show enhanced concentration of hydrogen. As a possible source of hydrogen, the interaction of water with freshly ground rock, caused by fault zone movement, is discussed. Isotopic studies on hydrogen in combination with laboratory experiments are ongoing to test hydrogen synthesis by rock-water interaction. First isotopic studies on δ13C of methane indicate mixing of microbial methane with only small amounts of methane generated by thermal degradation of organic matter in the shallower depth (down to ~2500 m). Below this depth, the concentration of heavy hydrocarbons increases. CH$_{4}$/(C2H$_{6}$+C3H$_{8}$) significantly drops from >100 to values <30 towards the bottom of the MH, and, methane becomes isotopically heavier, which is more typical for thermogenic hydrocarbons.
T23E-03
Shear Velocity Anisotropy in and Near the San Andreas Fault: Implications for Mapping Stress Orientations
We utilize a suite of geophysical logs from the SAFOD boreholes, earthquake data recorded on the SAFOD Pilot Hole array and regional seismic data from the Northern California Seismic Network (NCSN) and the Southern California Seismic Network (SCSN) to study the physical processes controlling shear velocity anisotropy in and near the San Andreas fault. Because the direction of maximum horizontal compression is at a high angle to the predominantly northwest-southeast structural trend, it is relatively easy to distinguish stress-induced from structurally-induced velocity anisotropy near the San Andreas fault. Dipole sonic logs in the SAFOD boreholes indicate that the shear-wave velocity anisotropy of the granitic rocks surrounding the wellbore is on the order of 3 to 10% and controlled by the tectonic stress field. The amount of stress-induced velocity in the granite decreases with depth because as confining pressure increases, the seismic velocity becomes less stress sensitive to stress. Within the sedimentary section found at greater depth (and closer to the San Andreas fault), both stress-induced and structurally-controlled velocity anisotropy is observed in the dipole sonic logs. At depth intervals where finely laminated shales are present, cross-dipole sonic logs indicate that structurally-induced anisotropy is dominant. However, in the well-cemented arkosic sandstones (which are not well bedded), the fast direction of the shear waves is controlled by the stress field (as in the granite) and is not consistent with theoretical models of structural anisotropy. These data provide further constraints on the orientation of the maximum horizontal compressive stress at depth. As one approaches the San Andreas fault along the trajectory of the SAFOD main hole, the direction of maximum stress rotates from approximately North-South (at relatively shallow depth) to become more fault-Normal at an angle of about 80 degrees to the strike of the fault within the fault zone at seismogenic depths. This observation supports the hypothesis that the San Andreas fault is a weak fault slipping at low levels of shear stress. An analysis of earthquake seismograms shows that ray paths through the Salinian granite adjacent to the fault exhibit fast shear wave polarizations aligned with the direction of maximum horizontal compression (in agreement with pilot hole stress measurements). In contrast, ray paths along the San Andreas fault (or through fault-parallel sedimentary structures) yield fast directions consistent with the northwest-southeast structural trend. An analysis of shear waves from local crustal earthquakes recorded at regional seismic stations in California shows a similar signature, indicating that seismic anisotropy may be useful in mapping the crustal stress field. We conclude that within the San Andreas fault zone, the structural fabric is the dominant mechanism responsible for velocity anisotropy, whereas in crust without predominant structural trend, the direction of maximum horizontal compression is the most important controlling factor.
T23E-04
P-wave and s-wave imaging from drill bit seismic data at SAFOD
We have used the drill bit seismic technique to develop preliminary images of fracture and shear zones associated with the San Andreas Fault at the SAFOD site, Parkfield, California. Our study included the interpretation of the USGS PSINE surface seismic profile and the drill bit seismic data recorded by three different geophone arrays. Three-component geophones were used in two of the arrays, one of which consisted of a 1.2 km string of geophones in the SAFOD Pilot Hole. The multi-component data allowed us to use both p-wave and s-wave imaging techniques for delineating subsurface structure after updating the velocity models for the site. Our interpretation of the locations and dips of linear features imaged in the migrated drill bit seismic data correlates very well with locations and dips of faults in the PSINE profile. Using the available seismic datasets, we interpret numerous faults to cut the Cenozoic sedimentary cover, the Salinian block, and an apparent wedge of metasediments at the SAFOD site. Based on the structural pattern of downward converging faults and fracture zones, we interpret the fault system to comprise a flower structure that is directly related to the tectonic regime of the San Andreas Fault. The interpretation of a flower structure at SAFOD, which is located 1.8 km SW of the SAF, fits well with the surface geological mapping that has been conducted at the site. This mapping and our data indicates that a pervasive system of fractures and faults trend subparallel to the SAF in a zone up to 3 km SW of the main trace of the fault.
T23E-05 INVITED
Structure and Composition of the San Andreas Fault Zone at Parkfield: Initial Results from SAFOD Phases 1 and 2
The San Andreas Fault Observatory at Depth (SAFOD) is designed to directly measure the physical and chemical conditions under which fault movement occurs during both earthquakes and aseismic creep. The SAFOD borehole begins vertically 1.8 km SW of the surface trace of the fault and then angles through the fault zone until it passes beneath the surface trace at a depth of 3.2 km. During SAFOD Phases 1 and 2, continuous sampling of drilling cuttings and fluids; spot coring; optical mineralogy, XRD, elemental composition and isotope analyses; and a comprehensive suite of downhole measurements provide a profile of physical properties and mineralogy across the entire San Andreas Fault Zone. This talk summarizes the data obtained to date by members of the SAFOD science team, who present their results in more detail elsewhere in this session. Although pre-drilling site characterization studies predicted a relatively simple fault zone, with Salinian granite basement on the west side of the San Andreas and Franciscan meta-sedimentary rocks on the east side, the geology actually encountered during drilling was more complex. The vertical part of the hole penetrated Salinian granitic rocks as expected, but after deviating the hole toward the San Andreas Fault, SAFOD passed through 1.2 km of arkosic sandstones and conglomerates (with some shales) before terminating in over 800 m of shales, siltstones and fine sandstones. The presence of Cretaceous fossils in cuttings and core from the bottom of the hole and the absence of high-grade Franciscan metamorphic minerals in the cuttings indicate that SAFOD passed all the way through the San Andreas Fault Zone, terminating either in rocks of the Great Valley group or relatively unmetamorphosed Franciscan sedimentary rocks. The contact between the predominately arkosic sandstone/conglomerate Salinian rocks and Mesozoic siltstone/shale of the Franciscan and/or Great Valley appears to be the long-term "geologic" manifestation of the San Andreas Fault. This boundary is located approximately 600 m SW of the surface trace of the fault. On both sides of this geologic fault contact, several zones exhibit anomalous geophysical properties that may represent active shear zones. Some of the most dramatic of these are zones about 15 m wide exhibiting low resistivities and low P- and S-wave velocities at measured depths (MD) of 3.07 and 3.16 km. The shallowest of these zones was cored at the conclusion of Phase 1, and contains a 30-cm-thick, illite plus illite-smectite shear zone with low friction and abundant internally sheared surfaces. Other potential shear zones indicated by anomalous geophysical properties, drilling rate, changes in cuttings mineralogy or increases in mud gas content are located at 3.19, 3.33, and 3.41 km MD. The anomalous zone at 3.33 km is particularly interesting, in that it is a 13-m-wide zone associated with the sudden appearance of serpentinite, which is seen along surface exposures of the San Andreas Fault in central California and is thought to be important in controlling frictional strength and the stability of sliding. Which of these zones are currently active will be determined through repeat logging of SAFOD to identify casing shear and locating the SAFOD target earthquakes with seismic instruments placed directly within the fault zone at seismogenic depths.
T23E-06
Imaging the deep roots of the San Andreas Fault zone with magnetotelluric measurements
In spring 2005, a huge volume of new magnetotelluric (MT) data was acquired in the vicinity of the SAFOD site near Parkfield, CA. This experiment addresses the dynamics of inter-plate transform faulting at lower crustal depths and contributes to the characterization of the SAFOD site on a regional scale. In particular, we aim to image the conductivity distribution of the deep roots of the San Andreas Fault zone (SAF). Overall, we recorded at 45 combined long-period / broad-band stations distributed in an area of 50 square km in the vicinity of the SAFOD site and an additional 41 broad-band stations along the 50 km long seismic reflection reflection profile of Hole and Ryberg. Data analysis and 2D modeling results of the new profile data together with existing data from previous experiments of Unsworth et al. (2004) and Park et al. (1991) reveal the conductivity structure of the crust from the top to the bottom. First results indicate the existence of a previously unknown, westward plunging conductor (5 ohm-m) at the transition zone between the electrically homogeneous resistive crust (>1000 ohm-m) in the west and the Franciscan Complex in the east. This conductor is spatially related to the seismicity occuring at maximum depths of 12 km. Though the depth extend of the conductor is not yet well constrained, resistivities lower than 5 ohm-m appear to be confined to the brittle crust. However, a broad zone of moderate resistivities below the SAF persists throughout the ductile crust and indicates structural weakness also at large depths.
T23E-07
Heat Flow Studies in the SAFOD Main Hole
Thermal investigations conducted in association with the San Andreas Fault Observatory at Depth (SAFOD) are focused on characterizing the in situ thermal state of the San Andreas fault (SAF) at seismogenic depths and detecting any thermal anomalies associated with frictional slip along the fault. Heat flow in the near-vertical, 2.2-km-deep SAFOD pilot hole, located 1.8 km west of the SAF near Parkfield, California averages 91 ± 3 mW m$^{-2}$, a value consistent with other measurements in the Coast Ranges northwest of Parkfield (Williams et al., GRL, 2004, DOI 10.1029/2003GL019352). We are acquiring new temperature and thermal conductivity measurements from the SAFOD main hole, which starts at the same location as the pilot hole and then deviates to cross the active trace of the SAF at a depth of 3.2 km. Temperature profiles in the two holes are remarkably consistent, with the temperature at 2.2 km vertical depth in the pilot hole equaling 92.5 $^{o}$C, and temperature at the same depth in the main hole approximately 1 km to the northeast equaling 93.0 $^{o}$C. This difference is well within the temperature measurement uncertainty of ±0.01 $^{o}$C and most likely reflects either variations in thermal properties between the two holes or uncertainties in the directional surveys. Preliminary heat flow measurements in the main hole from the depth intervals 2.24-2.33 km (0.8 to 0.7 km from the SAF) and 2.35-2.53 km (0.7 to 0.5 km from the SAF) equal 86 ± 9 mW m$^{-2}$ and 100 ± 7 mW m$^{-2}$, respectively. This variation in heat flow relative to the average measured in the pilot hole most likely reflects thermal refraction through the sedimentary rocks that occupy the interval between the SAF and the granitic basement penetrated by the pilot hole and the upper 1.8 km of the main hole. Alternatively, the observed variations could reflect a modest increase in heat flow toward the SAF. Such large-scale spatial trends, and their implications for frictional heating along the fault, will be resolved when we have completed heat flow measurements over the entire length of the main hole and across the SAF in late 2005.
T23E-08
Chemical and Isotopic Composition of Water and Gases From the SAFOD Wells: Implications to the Dynamics of the San Andreas Fault at Parkfield, California
To investigate the source of fluids within the San Andreas fault zone, we obtained downhole fluid samples from both the SAFOD pilot well (open hole at vertical depth of ~2.2 km) and the adjacent SAFOD main well, from open holes at depths of 1443-1470 m and 2540-2557 m. Each fluid sampling opportunity followed coring runs, which provided open holes at these depths, enabling formation fluid to enter the wells. Prior to coring, the drilling fluids were tagged with fluorescein and Rhodamine WT tracer dyes to allow for calculation of the contamination effects. We used an evacuated Kuster sampler and positive-displacement Westport samplers, that both allow for accurate determination of the dissolved gas concentrations. Chemical data and water-level measurements in the SAFOD pilot well and the shallower zone of SAFOD main well indicated that no significant amount of formation water was produced. Significant amounts of formation water, however, were produced from the deeper open hole of the SAFOD main well. The water level in the well rose ~60 m from completion of coring (October 1, 2004) to the first fluid sampling (April 13, 2005), when three samples were obtained, and rose ~12 m more by June 8, 2005, when an additional 4 samples were collected. Chemical data show that these samples are a mixture of formation water (75-80%) and the dye-tagged 'KCl' drilling solution. High pH values (9.5-10.5) and high Ca concentrations indicate contamination from the cement used for casing the well. Mixing proportions and geochemical modeling, utilizing the tracer dyes and conservative solutes, are used to calculate the compositions of formation water. Results show a Na-Ca-Cl type water with a salinity of ~20,000 mg/L TDS, very low Mg (0.1 mg/L) and carbonate alkalinity (<1 mg/L), but moderate SO$_{4}$ and very high DOC and organic acid anions. This chemical composition is typical of formation water from sedimentary rocks, such as oil field waters from California. The deepest samples from SAFOD main well are extremely gas-rich, with calculated in-situ gas pressures exceeding 50 bar. The gas composition is also consistent with that obtained from sedimentary rocks, about 80% CH$_{4}$, 5% higher hydrocarbons, and 10% N2. Low O2 (<0.02%) and high N2/Ar ratios (~300) reveal a strong non-atmospheric N2 component. Helium was enriched by a factor of ~1500 over the level in air-saturated water, and 3He/4He ratios yielded R/Ra values of 0.33 to 0.36, indicating a small mantle He contribution to the fluids. Although the absence of CO2 (<0.002%) may be an artifact of high pH, the relatively low R/Ra values compared to the nearby Varian-Phillips well (CO2 ~40% of total gas, R/Ra=1.5) and other wells in the Parkfield area indicate a relatively low flux of CO2 and other mantle volatiles, and a relatively minor contribution of these fluids to the San Andreas system at the SAFOD location and depth.