S21A-0229 INVITED
The Hayward-Rodgers Creek Fault System: Learning from the Past to Forecast the Future
The San Francisco Bay area is located within the Pacific-North American plate boundary. As a result, the region has the highest density of active faults per square kilometer of any urban center in the US. Between the Farallon Islands and Livermore, the faults of the San Andreas fault system are slipping at a rate of about 40 mm/yr. Approximately 25 percent of this rate is accommodated by the Hayward fault and its continuation to the north, the Rodgers Creek fault. The Hayward fault extends 88 km from Warm Springs on the south into San Pablo Bay on the north, traversing the most heavily urbanized part of the Bay Area. The Rodgers Creek fault extends another 63 km, passing through Santa Rosa and ending south of Healdsburg. Geologic, seismologic, and geodetic studies during the past ten years have significantly increased our knowledge of this system. In particular, paleoseismic studies of the timing of past earthquakes have provided critical new information for improving our understanding of how these faults may work in time and space, and for estimating the probability of future earthquakes. The most spectacular result is an 11-earthquake record on the southern Hayward fault that extends back to A.D. 170. It suggests an average time interval between large earthquakes of 170 years for this period, with a shorter interval of 140 years for the five most recent earthquakes. Paleoseismic investigations have also shown that prior to the most recent large earthquake on the southern Hayward fault in 1868, large earthquakes occurred on the southern Hayward fault between 1658 and1786, on the northern Hayward fault between 1640 and 1776, and on the Rodgers Creek fault between 1690 and 1776. These could have been three separate earthquakes. However, the overlapping radiocarbon dates for these paleoearthquakes allow the possibility that these faults may have ruptured together in several different combinations: a combined southern and northern Hayward fault earthquake, a Rodgers Creek-northern Hayward fault earthquake, or a rupture of all three fault sections. Each of these rupture combinations would produce a magnitude larger than 1868 (M~6.9). In 2003, the Working Group on California Earthquake Probabilities released a new earthquake forecast for the Bay Area. Using the earthquake timing data and alternative fault rupture models, the Working Group estimated a 27 percent likelihood of a M?6.7 earthquake along the Hayward-Rodgers Creek fault zone by the year 2031. This is this highest probability of any individual fault system in the Bay Area. New paleoseismic data will allow updating of this forecast.
S21A-0230 INVITED
The M7 October 21, 1868 Hayward Earthquake, Northern California-140 Years Later
October 21, 2008 marks the 140th anniversary of the M7 1868 Hayward earthquake. This large earthquake, which occurred slightly before 8 AM, caused extensive damage to San Francisco Bay Area and remains the nation's 12th most lethal earthquake. Property loss was extensive and about 30 people were killed. This earthquake culminated a decade-long series of earthquakes in the Bay Area which started with an M~6 earthquake in the southern Peninsula in 1856, followed by a series of four M5.8 to M6.1 sized earthquakes along the northern Calaveras fault, and ended with a M~6.5 earthquake in the Santa Cruz Mountains in 1865. Despite this flurry of quakes, the shaking from the 1868 earthquake was the strongest that the new towns and growing cities of the Bay Area had ever experienced. The effect on the brick buildings of the time was devastating: walls collapsed in San Francisco, Oakland, and San Jose, and buildings cracked as far away as Napa, Santa Rosa, and Hollister. The area that was strongly shaken (at Modified Mercalli Intensity VII or higher) encompassed about 2,300 km2. Aftershocks continued into November 1868. Surface cracking of the ground along the southern end of the Hayward Fault was traced from Warm Springs in Fremont northward 32 km to San Leandro. As Lawson (1908) reports, "the evidence to the northward of San Leandro is not very satisfactory. The country was then unsettled, and the information consisted of reports of cow- boys riding on the range". Analysis of historical triangulation data suggest that the fault moved as far north as Berkeley, and from these data the average slip along the fault is inferred to be about 1.9 ± 0.4 meters. The paleoseismic record from the southern end of the Hayward Fault provides evidence for 10 earthquakes before 1868. The average interval between these earthquakes is 170 ± 80 years, but the last five earthquakes have had an average interval of only 140 ± 50 years. The 1868 Hayward earthquake and more recent analogs such as the 1995 Kobe earthquake are stark reminders of the awesome energy waiting to be released from below the east side of the San Francisco Bay along the Hayward Fault. The population at risk from a Hayward Fault earthquake is now 100 times larger than in 1868. The infrastructure in the San Francisco Bay Area has been tested only by the relatively remote 1989 M6.9 Loma Prieta earthquake. To help focus public attention on these hazards, the 1868 Hayward Earthquake Alliance has been formed, consisting of public and private sector agencies and corporations (see their website www.1868alliance.org). The Alliance is planning a series of activities leading up to the 140th anniversary on October 21, 2008. These include public forums, conferences, commemoration events, publications, websites, videos, and public service announcements. http://www.1868alliance.org/
S21A-0231
A Virtual Tour of the 1868 Hayward Earthquake in Google EarthTM
The 1868 Hayward earthquake has been overshadowed by the subsequent 1906 San Francisco earthquake that destroyed much of San Francisco. Nonetheless, a modern recurrence of the 1868 earthquake would cause widespread damage to the densely populated Bay Area, particularly in the east Bay communities that have grown up virtually on top of the Hayward fault. Our concern is heightened by paleoseismic studies suggesting that the recurrence interval for the past five earthquakes on the southern Hayward fault is 140 to 170 years. Our objective is to build an educational web site that illustrates the cause and effect of the 1868 earthquake drawing on scientific and historic information. We will use Google EarthTM software to visually illustrate complex scientific concepts in a way that is understandable to a non-scientific audience. This web site will lead the viewer from a regional summary of the plate tectonics and faulting system of western North America, to more specific information about the 1868 Hayward earthquake itself. Text and Google EarthTM layers will include modeled shaking of the earthquake, relocations of historic photographs, reconstruction of damaged buildings as 3-D models, and additional scientific data that may come from the many scientific studies conducted for the 140th anniversary of the event. Earthquake engineering concerns will be stressed, including population density, vulnerable infrastructure, and lifelines. We will also present detailed maps of the Hayward fault, measurements of fault creep, and geologic evidence of its recurrence. Understanding the science behind earthquake hazards is an important step in preparing for the next significant earthquake. We hope to communicate to the public and students of all ages, through visualizations, not only the cause and effect of the 1868 earthquake, but also modern seismic hazards of the San Francisco Bay region.
S21A-0232
Bounding Ground Motions for Hayward Fault Scenario Earthquakes Using Suites of Stochastic Rupture Models
The next major earthquake in the San Francisco Bay area is likely to occur on the Hayward-Rodgers Creek Fault system. Attention on the southern Hayward section is appropriate given the upcoming 140th anniversary of the 1868 M 7 rupture coinciding with the estimated recurrence interval. This presentation will describe ground motion simulations for large (M > 6.5) earthquakes on the Hayward Fault using a recently developed elastic finite difference code and high-performance computers at Lawrence Livermore National Laboratory. Our code easily reads the recent USGS 3D seismic velocity model of the Bay Area developed in 2005 and used for simulations of the 1906 San Francisco and 1989 Loma Prieta earthquakes. Previous work has shown that the USGS model performs very well when used to model intermediate period (4-33 seconds) ground motions from moderate (M ~ 4-5) earthquakes (Rodgers et al., 2008). Ground motions for large earthquakes are strongly controlled by the hypocenter location, spatial distribution of slip, rise time and directivity effects. These are factors that are impossible to predict in advance of a large earthquake and lead to large epistemic uncertainties in ground motion estimates for scenario earthquakes. To bound this uncertainty, we are performing suites of simulations of scenario events on the Hayward Fault using stochastic rupture models following the method of Liu et al. (Bull. Seism. Soc. Am., 96, 2118-2130, 2006). These rupture models have spatially variable slip, rupture velocity, rise time and rake constrained by characterization of inferred finite fault ruptures and expert opinion. Computed ground motions show variability due to the variability in rupture models and can be used to estimate the average and spread of ground motion measures at any particular site. This work was performed under the auspices of the U.S. Department of Energy by University of California Lawrence Livermore National Laboratory under contract No.W-7405-Eng-48. This is LLNL contribution UCRL- ABS-234222
S21A-0233
Site Characterization and Seismic Response of Dumbarton Bridge
The existing Dumbarton Toll Bridge was built in 1982, connecting the cities of Newark and East Palo Alto, California. The initial vulnerability studies conducted by California Department of Transportation (Caltrans) in 2004 indicated that the performance of the bridge during a maximum credible earthquake was uncertain. With over 18 years of practical experience in geotechnical and earthquake engineering, Earth Mechanics, Inc. (EMI) has been retained by the Bay Area Toll Authority and Caltrans to perform the necessary study for the seismic evaluation of the bridge. An extensive field investigation, consisting of 14 soil borings, six down-hole seismic loggings, seven vane shear tests, 33 cone penetrometer tests and an offshore geophysical survey, was undertaken both on-land and over-water at the site. The investigation resulted in a subsurface profile similar to the one dated 1982 (but with more local zones identified), and provided more reliable data for subsequent engineering analyses. According to probabilistic and deterministic earthquake analyses, a 1,000-year return period spectrum was adopted for the Safety Evaluation Earthquake (SEE), with seven sets of rock motion time histories generated. Site response and kinematic soil-pile interaction analyses were carried out at selected piers to develop the ARS design criteria for the bridge that are site and structure specific, and incorporate state-of-the- art seismic design principles. The axial capacity and lateral push-over behavior of each pier was also examined to ensure its sound performance at SEE levels. The current study provides an illustration of the recent great advances in site investigation and geotechnical analysis. http://www.earthmech.com
S21A-0234
Liquefaction Scenarios in the Northern Santa Clara Valley for a Repeat of the 1868 Hayward Fault (M6.7-7.0) Earthquake
The spatial distribution of the probability of liquefaction in the northern Santa Clara Valley, California, was predicted for a repeat of an earthquake like the 1868 Hayward Fault (M6.7-7.0) earthquake. Probabilities were computed with the methodology for probabilistic liquefaction hazard mapping that was developed by Holzer and others (USGS OFR 02-296, 2006). The methodology relies on field-based plots of cumulative frequency of the liquefaction potential index (LPI) for spatially homogenous surficial geologic units. LPI, which is a scalar parameter that integrates the liquefaction potential of the entire soil column, was computed for 164 seismic cone penetration tests (SCPT) that were conducted in Holocene and Pleistocene geologic units. The plots of cumulative frequency were used to estimate the liquefaction probability distribution for each surficial geologic unit given peak ground acceleration (PGA) and earthquake magnitude. Scenario maps were produced with ArcGIS Model Builder. PGA at each node in a 50-m grid was estimated with the new attenuation relation proposed by Boore and Atkinson (2007, v. 3.04). Regional averages of VS30 values, which were based on the SCPT, were used to account for local site amplification. The probability of liquefaction was estimated at each node using the liquefaction probability distribution appropriate for the surficial geology at the node. For a M7 earthquake and an assumed water-table depth of 1.5 m in the central part of the valley, liquefaction probabilities range from 0.1 to 0.2 along Coyote and Guadalupe Creeks, but are less than 0.05 elsewhere. For an M6.7 earthquake, probabilities remain greater than 0.1 along Coyote Creek but decrease along Guadalupe Creek to less than 0.1. For assumed water-table depths greater than 5 m, liquefaction probabilities are less than 0.05 throughout the valley. The probability of lateral spreading is less than 0.05 throughout the valley for both water table depths and both earthquakes. Liquefaction probabilities are substantially higher for an M7.8 1906-type earthquake on the San Andreas Fault, exceeding 0.3 along Coyote and Guadalupe Creeks for the 1.5-m water table depth. Predicted probabilities are highest in areas where liquefaction and lateral spreading were reported following the 1868 and 1906 earthquakes.
S21A-0235
3D structure effects on local and near-regional seismic wave propagation in the San Francisco Bay Area
In this study we performed 3D waveform modeling of 10 small to moderate events (Mw 4.1-5.0) in the San Francisco Bay Area using the USGS SF06 3D velocity model (Brocher et al., 2005; Jachens et al., 2005). In the simulations we assumed the source parameters reported in the Berkeley Seismological Laboratory (BSL) Moment Tensor Catalog. Broadband seismic data from the Berkeley Digital Seismic Network (BDSN), and strong motion data from the USGS and the California Geologic Survey California strong motion arrays were used in the analysis. We analyzed and modeled the data in three frequency bands, namely 0.03-0.15 Hz, 0.1-0.25 Hz, and 0.1-0.5Hz. Preliminary waveform modeling shows that the USGS SF06 model predicts many important features of observed seismograms including bodywave arrival times, and peak ground velocity. On the other hand, as reported by Rodgers et al. (2007), the model produces late arriving surface waves. While peak ground velocity is generally well modeled there are paths that have significant amplitude mismatches and also poor waveform fit to sedimentary basin generated surface waves. We are identifying which paths need additional waveform modeling in order to further calibrate the 3D structure. We will present the bodywave and surface wave arrival time, and peak ground velocity correlations as well as forward modeling results for the problematic paths. References Brocher, T. M., (2005). Emprical relations between elastic wave speeds and density in the Earthfs crust, Bull. Seism. Soc. Am., 95 No. 6, 2081-2092. Jachens, R., R. Simpson, R. Graymer, C. Wentworth, T. Brocher (2006). Three-dimensional geologic map of northern and central California: A basic model for supporting ground motion simulation and other predictive modeling, 2006 SSA meeting abstract, Seism. Res. Lett., 77, No.2, p 270. Rodgers, A., A. Petersson, S. Nilsson, B Sjogreen, K. McCandless (2007). Broadband waveform modeling of moderate earthquakes in the San Francisco Bay Area and preliminary assessment of the USGS 3D seismic velocity model, submitted Bull. Seism. Soc. Am. http://seismo.berkeley.edu/~ahyi/
S21A-0236
Forward modeling of a transect of ground motion recordings of the 3 September 2000 M5.0 Napa/Yountville earthquake
We model a linear transect of recorded ground motions at five stations of the 3 September 2000 M5.0 Napa/Yountville earthquake in the Northern San Francisco Bay area, focusing on the tangential displacement component. The epicentral distances range from 6 km to 57 km. The waveforms at all five stations are characterized by a direct shear-wave (S0), a midcrustal reflection (S1), a surface-layer multiple (S2), and Love waves. Our one-dimensional (1D) and two-dimensional (2D)simulations have a grid spacing of 100 m and minimum velocity of 800 m/s and accurately model displacements at frequencies up to 0.8 Hz. We compare the fit of our synthetic seismogram to the 1D Berkeley Seismological Laboratory GIL7 model and the three- dimensional (3D) USGS San Francisco Bay Area velocity model. We find that a simple 1D velocity model composed of four layers sufficiently predicts the primary arrivals with the exception of the later arriving Love wave at all five stations. Our modeling indicates that the source layer must be faster than the velocities in the current 1D GIL7 and 3D USGS regional velocity model to account for the arrival time of S0. To explain the delay between the S0 and S2 arrivals, the velocities at depths extending from the surface to 5 km must be slower than the velocities indicated in the other regional models. Adding a 1 km deep trapezoidal basin with a shear-wave velocity of 800 m/s significantly improves the fit of the amplitude, phase, and duration of the surface waves. http://www.tufts.edu/~ethomp04/
S21A-0237
Velocity Contrast along the Hayward Fault From Analysis of Fault Zone Head Waves
The Hayward fault is a major branch of the San Andreas system in northern California. It juxtaposes the Franciscan Complex (fast) to the SW side against the Great Valley Sequence (slow) to the NE side. Previous studies based on 3D seismic tomography suggested ~5-10% seismic velocity contrast in the upper 10 km, consistent with geological observations. Here we systematically investigate the velocity contrast along the entire Hayward fault using fault zone head waves (FZHW) that refract along the fault interface. The FZHW provide the most diagnostic seismic signal for the existence of sharp bimaterial interfaces, and the highest-resolution tool for imaging their seismic properties. A total of 10,952 earthquakes recorded by the Northern California Seismic Network (NCSN) between January 1984 and June 2007 are used in the study. We perform waveform cross correlation for all possible event pairs, and group them into similar event clusters if they are located within 3 km and have a median cross-correlation coefficient of at least 0.85. A total of 250 clusters have been identified between 20 km north and 70 km south of Point Pinole along the Hayward fault strike. The waveforms generated by each event clusters are stacked for each station within 10 km on the slower (NE) side of the fault. Next we align the peak and trough of the direct P waves assuming right-lateral strike-slip focal mechanisms, pick the FZHW arrivals, and plot the waveforms against the along-fault-interface distances. The results at many stations on the slow side of the fault show clear propagation of FZHW from most event clusters, implying a velocity contrast that is geometrically coherent along the entire 90 km of the Hayward fault. The strength of the velocity contrast varies somewhat along strike and with depth. For stations in the central portion, the travel time moveout between the FZHW and direct P waves increases continuously with distance over ~75 km, whereas for stations closer to the ends the moveout is continuous only over distances of ~30 km. The moveout analysis indicates average velocity contrasts of ~5-11%, with higher values for the central region and the upper 6 km. The FZHW do not show clear breaks between the Hayward and Mission faults, suggesting a continuous transition between these two fault zones. The existence of a coherent bimaterial interface in the structure of the Hayward fault can have significant implications for properties of earthquake ruptures on this fault.
S21A-0238
A Three-dimensional Geologic Model of the Hayward-Calaveras Fault Junction
We present a 3D geologic model depicting the Hayward-Calaveras Fault junction as a complex network of faults at the surface and as a single through-going fault below about 5 km. Deep seismicity suggests a simple connection between the southern Hayward and central Calaveras Faults, whereas geologic mapping at the Earth's surface shows a complex zone of deformation between the fault traces. Such a complicated fault junction represents a challenge for 3D geologic modeling. Newly relocated earthquake hypocenters for the San Francisco Bay region have improved the precision of mapping fault planes, and integrated analysis of multiple datasets in two and three-dimensions has improved the characterization of subsurface geologic and geophysical contacts. In particular, geophysical modeling of magnetic mafic and ultramafic rocks, such as gabbro and serpentinite bordering the southern Hayward Fault, allows the complex structure of these units to be mapped in the subsurface where the southern Hayward and central Calaveras Faults converge. The new 3D geologic model can be used as a basis for earthquake process modeling such as finite element modeling and local and regional ground motion simulations, which will aid in further characterizing the potential seismic hazard along the Hayward-Calaveras fault system. The increased length of a combined southern Hayward and central Calaveras Fault rupture could generate an earthquake greater than M7 based on magnitude-area relationships, posing a significant seismic hazard to the San Francisco Bay region.
S21A-0239
Segmentation of the Calaveras-Hayward Fault System Based on 3-D Geometry and Geology at Large-Earthquake Depth
For the purpose of estimating seismic hazard, the Calaveras and Hayward Faults have been considered as separate structures and analyzed and segmented based largely on their surface-trace geometry and the extent of the 1868 Hayward Fault earthquake. Recent relocations of earthquakes and 3-D geologic mapping have shown, however, that at depths associated with large earthquakes (>5 km) the fault geology and geometry is quite different than that at the surface. Using deep fault geometry inferred from these studies we treat the Hayward and Calaveras Faults as a single system and divide the system into segments that differ from the previously accepted segments as follows: 1. The Hayward Fault connects directly to the central Calaveras Fault at depth, as opposed to the 5 km wide restraining stepover zone of multiple imbricate oblique right-lateral reverse faults at the surface east of Fremont and San Jose (between about 37.25°-37.6°N). 2. The segment boundary between the Hayward, central Calaveras, and northern Calaveras is based on their Y- shaped intersection at depth near 37.40°N, 121.76°W (Cherry Flat Reservoir), about 8 km south of the previously accepted central-northern Calaveras Fault segment boundary. 3. The central Calaveras Fault is divided near 37.14°N, 121.56°W (southern end of Anderson Lake) into two subsegments based on a large discontinuity at depth seen in relocated seismicity. 4. The Hayward Fault is divided near 37.85°N, 122.23°W (Lake Temescal) into two segments based on a large contrast in fault face geology. This segmentation is similar to that based on the extent of 1868 fault rupture, but is now related to an underlying geologic cause. The direct connection of the Hayward and central Calaveras Faults at depth suggests that earthquakes larger than those previously modeled should be considered (~M6.9 for the southern Hayward, ~M7.2 for the southern Hayward plus northern central Calaveras). A NEHRP study by Witter and others (2003; NEHRP 03HQGR0098) suggested evidence for large surface ruptures on the northern central Calaveras, but that work is not peer-reviewed and there is little or no other paleoseismic or geodetic data from the stepover zone or northern central Calaveras Fault (all commonly cited data are from the southern central Calaveras Fault), so the sparse surface data neither demands nor precludes our interpretation. The additional segmentation of the central Calaveras Fault proposed here may explain the observation that this segment seems to generate characteristic moderate (~M6.0-6.5) earthquakes rather than the larger ~M6.9 earthquakes that could be generated by rupture of the previously defined longer central Calaveras segment. Better information regarding fault plane geometry and 3-D distribution of rock properties adjacent to the faults at seismogenic depths should help us revise proposed segmentation models of other faults for seismic hazard analyses.
S21A-0240
Linking Faults: Subsurface Creep on a Contiguous Fault Structure Connecting the Hayward and Calaveras Faults
The Hayward Fault is the most likely Bay Area fault to rupture in a major earthquake in the next 30 years. Much of the Hayward fault experiences 3-6 mm/yr of creep at the surface, approximately half of its estimated long-term slip rate (~9mm/yr). The southernmost section of the Hayward fault, however, appears to be creeping at up to 9 mm/yr, before creep abruptly ceases just south of Freemont, CA. East of Fremont, the Calaveras Fault transitions from about 15 mm/yr of slip along its creeping southern section, to about 6 mm/yr in the north. Based on location and slip, it seems very likely that the Calaveras fault directly transfers the deficit 9 mm/yr of slip as creep to the Hayward Fault. The existence of characteristic repeating earthquakes in this region strongly suggests that active creep indeed occurs between the two faults. Seismic studies of the Hayward-Calaveras stepover have suggested that the southern Hayward fault may in fact dip into and merge with the northern Calaveras Fault south of Fremont, CA. Improved GPS-derived surface velocities and consideration of new PS-InSAR range-change rates in the area help verify and constrain this proposed geometry, and allows a reevaluation of slip on these faults. A dislocation model combining seismically derived geometry and GPS velocities confirms that active fault creep occurs on the continuation of the dipping surface of the Hayward Fault effectively linking the Hayward and Calaveras Faults along a contiguous structure. The revised model provides an estimate of the distribution of creep on the three-dimensional fault surface through the stepover region, thus aiding in evaluating the seismic potential and rupture scenarios in the area.
S21A-0241
Geophysical Investigations Along the Hayward Fault, Northern California, and Their Implications on Earthquake Hazards
Geophysical studies indicate that the Hayward Fault follows a pre-existing basement structure and that local geologic features play an important role in earthquake seismicity. The recent creeping trace of the Hayward Fault extends for about 90 km from San Pablo Bay in the northwest to Fremont in the southeast, and together with its northern extension, the Rodgers Creek Fault, is regarded as one of the most hazardous faults in northern California. The Hayward Fault is predominantly a right-lateral strike-slip fault that forms the western boundary of the East Bay Hills and separates Franciscan Complex rocks on the southwest from Coast Range Ophiolite and Great Valley Sequence basement rocks on the northeast. The Hayward Fault is characterized by distinct linear gravity and magnetic anomalies that correlate with changes in geology, structural trends, creep rates, and clusters of seismicity. These correlations indicate the existence of fault-zone discontinuities that probably reflect changes in mechanical properties. These fault-zone discontinuities may play a role in defining fault segments—locations where recurring seismic ruptures may tend to nucleate or terminate. Along the central part of the Hayward Fault, a prominent gravity and magnetic anomaly correlates with an exposed gabbro body, the San Leandro gabbro. Modeling of these anomalies reveals that the San Leandro gabbro is much more extensive in the subsurface than the outcrop pattern suggests, extending to a depth of about 6-8 km. The inferred extent of the San Leandro gabbro, it's geologic setting, and associated seismicity suggest that the Hayward Fault evolved from a pre-existing basement feature, similar to the ancestral Coast Range Fault. Combined modeling and relocated double-difference seismicity data indicate that the dip of the fault surface varies from near vertical in the north to about 75 degrees in the central part to about 50 degrees in the south near Fremont and ultimately connects with the central Calaveras Fault. A seismicity cluster along the western edge of the San Leandro gabbro and a bend in the fault associated with the gravity and magnetic high along the gabbro body suggests that this mafic body influences fault geometry and behavior, and may serve as a nucleation point for large earthquakes on the fault.
S21A-0242
Frictional Strength of Hayward Fault Gouge
A recent 3-D geologic model of the Hayward fault in the San Francisco Bay Region shows that a number of different rock units are juxtaposed across the fault surface as a result of lateral displacement. The fault gouge formed therein is likely a mixture of these various rock types. To better model the mechanical behavior of the Hayward fault, which is known to both creep and have large earthquakes, frictional properties of mixtures of the principal rock types were determined in the laboratory. Room temperature triaxial shearing tests were conducted on binary and ternary mixtures of Great Valley Sequence graywacke, Franciscan jadeite-bearing metagraywacke, Franciscan pumpellyite-bearing metasandstone, Franciscan melange matrix, serpentinite and two-pyroxene gabbro. The gouge samples were crushed and sieved (<150 μm grains), then applied in a 1-mm layer between saw-cut sliding blocks. Each sample assemblage was saturated and sheared at constant pore water pressure of 1 MPa and normal stress of 51 MPa. Coefficients of friction, μ, ranged from a low of 0.38 for the serpentinite to a maximum of 0.85 for the gabbro. While the serpentinite and the Franciscan melange matrix were relatively weak, all other rock types obeyed Byerlee's Law. The friction coefficient of mixtures could be reliably predicted by a simple average based on dry weight percent of the end member strengths. This behavior is in contrast to some mixtures of common gouge materials such as montmorillonite+quartz, which exhibit non- linear frictional strength trends with varying weight percent of constituents. All materials tested except serpentinite were velocity strengthening, therefore promoting creeping behavior. The addition of serpentinite decreased a-b values of the gouge and increased the characteristic displacement, dc, of strength evolution. Because temperature strongly influences the mechanical properties of fault gouge as well as speeding chemical reactions between the constituents, elevated temperature experiments simulating deeper seismogenic regions of the Hayward fault are planned for the near future.
S21A-0243
Hayward Fault rate constraints at Berkeley: Evaluation of the 335-meter Strawberry Creek offset
At UC Berkeley the active channel of Strawberry Creek is offset 335 meters by the Hayward fault and two abandoned channels of Strawberry Creek are laterally offset 580 and 730 meters. These relationships record the displacement of the northern Hayward fault at Berkeley over a period of tens of millennia. The Strawberry Creek site has a similar geometry to the central San Andreas fault's Wallace Creek site, which arguably provides the best geological evidence of "millennial" fault kinematics in California (Sieh and Jahns, 1984). Slip rate determinations are an essential component of overall hazard evaluation for the Hayward fault, and this site is ripe to disclose a long-term form of this parameter, to contrast with geodetic and other geological rate evidence. Large offsets at the site may lower uncertainty in the rate equation relative to younger sites, as the affect of stream abandonment age, generally the greatest source of rate uncertainty, is greatly reduced. This is helpful here because it more-than-offsets uncertainties resulting from piercing projections to the fault. Strawberry Creek and its ancestral channels suggest west-side-up vertical deformation across the Hayward fault at this location. The development of the vertical deformation parameter will complement ongoing geodetic measurements, particularly InSAR, and motivate testing of other geological constraints. Up-to-the-west motion across the Hayward fault at Berkeley has important implications for the partitioning of strain and kinematics of the northern Hayward fault, and may explain anomalous up-on-the-west landforms elsewhere along the fault. For example, geological features of the western Berkeley Hills are consistent with rapid and recent uplift to the west of the fault. On the basis of a preliminary analysis of the offset channels of Strawberry Creek, up-to-the-west uplift is about 0.5mm/yr across the Hayward fault at Berkeley. If this is in fact the long-term rate, the 150 m height of the Hills to the northwest of the Strawberry Creek site was produced during the past about 300,000 years by a significant dip- slip (thrust) component of Hayward fault motion. Rapid and recent uplift of some portions of the East Bay Hills has important implications for fault geometries and slope stability, and should strongly influence the investigation fault hazards in areas that are more complexly deformed.
S21A-0244
Nature of Active Traces of the Hayward Fault at the University of California, Berkeley
The location of the Hayward fault zone at the University of California Berkeley Campus is well defined by geomorphic features including offset stream channels, side-hill benches, and the break-in-slope at the base of the Berkeley Hills, as well as by fault-creep related deformation of curbs, buried culverts and utilities, and structures—most notably Memorial Stadium. Based on the mapped fault traces associated with these surficial features, more than 30 trenches have been excavated at various locations on the campus during the past twenty years to assess the exact location and width of the active fault zone near existing and planned structures. These trenches show that the active fault trace(s) range from well expressed to poorly expressed in various surficial materials as a function of (1) the local geometry of the fault, (2) the stability of the near-surface deposits (e.g., it is poorly expressed where it crosses active landslides), and (3) the age of the deposits (i.e., it is better expressed in older deposits). At locations near the Smyth-Fernwald Housing, Prospect Court, the Greek Theater, and Foothill Housing, trenches showed that the fault is characterized by multiple distinct traces that in many places bound alluvial/colluvial-filled depressions up to 6 meters wide, and are in-filled with Holocene deposits. Quaternary deposits and bedrock units are truncated, indicating that significant lateral and vertical displacement has occurred along these fault traces. The creeping trace of the fault generally coincides with these well expressed fault traces. Trenches also revealed that two sub-parallel active fault traces as much as 40 to 60 meters apart extend along the hillslope directly east of the Greek Theater and north and east of Bowles Hall. It remains uncertain as to how fault creep occurs along the two separate branches. Between Memorial Stadium and Bowles Hall, there is a small right bend or stepover in the fault. The location of the creeping trace is well defined by deformation to Memorial Stadium, curbs, and buried culverts in this area, but trenches reveal that the creeping trace is associated only with a weakly expressed "parting fabric" in homogeneous colluvial deposits and as wide, moderately expressed shear zones within landslide-derived bedrock blocks. Stratigraphic markers in trenches north of Memorial Stadium are disrupted in several places by features that have uncertain origin, yet indicate little cumulative lateral or vertical displacement is associated with these features.
S21A-0245
The McCreary Glade Earthquake Sequence: Possible Reactivation of Ancient Structures Near Lake Pillsbury, Northern Coast Ranges, Mendocino County, California
From 15 April 2000 to 31 August 2007, over 900 earthquakes ( S21A-0246 Implications of Preliminary Gravity and Magnetic Surveys to the Understanding of the Bartlett Springs Fault Zone, Northern California Coast Ranges
We use new gravity and magnetic data in the Lake Pillsbury region to help understand the geometry and
character of the Bartlett Springs fault zone, one of the three main strands of the San Andreas system north of the
San Francisco Bay area. We collected 153 new gravity stations in the Lake Pillsbury region that complement the
sparse regional dataset and are used to estimate the thickness of Quaternary deposits in the inferred Gravelly
Valley (Lake Pillsbury) pull-apart basin. We also collected 38 line-km of ground magnetic data on roads and 65
line-km by boat on the lake to supplement regional aeromagnetic surveys and to map concealed fault strands
beneath the lake.
The new gravity data show a significant northwest-striking gravity gradient at the base of which lies the Bartlett
Springs fault zone. Superposed on this major east-facing gravity gradient is a 5 mGal low centered on Lake
Pillsbury and Gravelly Valley. Inversion of the gravity field for basin thickness assuming a density contrast of 400
kg/m3 indicates the deepest part of the basin is about 400 m and located in the northern part of the valley,
although the inversion lacks gravity stations within the lake. The basin is about 3 km wide and 5 km long and
basin edges coincide with strands of the Bartlett Springs fault zone. Our gravity data suggest that Potter Valley,
which lies between the Maacama and Bartlett Springs faults, is also as much as 400 m deep in the southern part
of the valley, although additional data west of the valley would better isolate the gravity low. Geomorphologic
characteristics of the valley suggest that this structure has been quiescent during the late Quaternary.
Ground magnetic data are very noisy but the data in conjunction with 9.6 km-spaced NURE aeromagnetic lines
suggest that regional analog aeromagnetic data flown in 1962 may suffer from location errors. The regional and
NURE data show a northwest-striking magnetic high that extends across Lake Pillsbury. The northeast edge of
this anomaly, caused by ultramafic rocks, coincides with the Bartlett Springs fault zone for nearly 15 km. Lake
magnetic data indicate as many as three right-stepping strands of the Bartlett Springs fault zone within the gravity-
defined pull-apart basin. Two pairs of magnetic anomalies appear to be dextrally offset along the fault, arguing
for about 8-9 km of cumulative offset on the fault since the passage of the triple junction at about 3.5 Ma. This
estimate is similar to proposed offsets of the Eel River (8.6-10.9 km) at Lake Pillsbury. The minimum long-term
slip rate is thus 2.3-3.1 mm/yr, considerably slower than geodetic rates of 5-8 mm/yr.
Seismicity forms a 5-km-wide diffuse zone along the Bartlett Springs fault zone in the Lake Pillsbury area, with
fewer earthquakes about 5 km northwest of the lake and its associated magnetic anomaly. The McCreary Glade
seismicity lineament, located between Potter Valley and Lake Pillsbury, has been attributed to a dike intrusion at
depth or reactivation of an older structure. These earthquakes coincide with the northeast edge of a 100-km-long
belt of aeromagnetic anomalies and thus appear to have reactivated an older basement feature. The coincidence
of the Bartlett Springs fault zone and significant gravity gradients also argues that the much younger fault zone
has reactivated older basement features. Our analysis shows that a modern, high-resolution aeromagnetic
survey is needed to confirm these preliminary interpretations.