S41B-0977
Seismogenic Structure of the Northern Section of Lijiang Basin in Northwest Yunnan (China) in Quaternary
Located at the middle of Jinshajiang-Red River fault zone and intersected by Xiaojinhe fault zone, Lijiang region in northwest Yunnan, China, is surrounded by a complex dynamic environment. The northern section of Lijiang basin (NSLB) has the features of a zigzag fault, a kind of 'tracing extension' in the shape. Taking the west boundary as example, it can be divided into three segments, among which south segment is in the N-S trending, central segment in the NNE trending and north segment becomes the N-S trending again. Basin width changes regularly along the basin boundary. The central part in the NNE-trending is narrow and has an average width of 3.5km. Both the south and north parts are wide and have an average width of 4.5km, which is 1km larger than that in the central part. It is an important character about a zigzag fault zone with a component of shear displacement. Fault slip is characterized mainly by extension, and sinistral shear. Average sinistral-shear and spreading displacements are respectively 1950m and 1730m. The kind of movement began in middle Pleistocene, which is about 800ka ago. Average sinistral and extensional slip-rates can be acquired, which are 2.44mm/a and 2.16mm/a. Geological evidences at different segments of Lijiang basin demonstrate results of geomorphic analysis, and are in consistent with our knowledge about the zig-zag fault. The field observation at the north segment of the NSLB shows a vertical displacement of 25 to 27m and a sinistral displacement of 12.5 to 13.7m, which happened after 9 to 10ka B.P. Based on the conclusions, it can be acquired that the average vertical slip rate of boundary fault in the NSLB in Holocene is 2.74mm/a and the average horizontal slip rate is 1.38mm/a. Sinistral displacement of ridges and gullies are very typical around Longquan in the middle of the west boundary in the NSLB. 29m sinistral displacement of a gully and 16m vertical offset of alluvial, diluvial and remnant deposits in late Pleistocene can be recognized at the north of the Longquan village. Based on the above analysis about structural geology, two conclusions about kinematic features in the NSLB can be acquired. One is that the tectonic movement in the NSLB is characterized by sinistral shear and extension and the other is that the components of vertical and horizontal movements are different at the different segments of the NSLB. The component of the vertical movement is clearly larger than that of the horizontal component in the N-S-trending segments, but it is contrary in the NE or NNE-trending segments in the middle of the NSLB. The phenomena are identical to the kinematic features of a zigzag with a component of shear displacement and the results concluded from the analysis of tectonic-geomorphology in the NSLB. Realization about sinistral shear and extension of the northern section of Lijiang basin provides direct evidence for the model of clockwise rotation of northwest Sichuan active block and the understanding of dynamic features of Red River fault zone.
S41B-0978
Kinematic 3D Retro-Deformation of Fault Blocks Picked from 3D Seismics; Correlation of Strain Results with Natural Data
Deformation of fault-Near strata is not primarily dependent on the displacement along the fault, but rather on the variation in the morphology of the fault plane (in the direction of fault movement) and on the rate of change in fault-plane parallel displacement. Much of this deformation is below the resolution of reflection seismics (i.e. sub-seismic). Kinematic 3D restoration offers a method to define the magnitude and extent of sub-seismic fault damage, which is extracted from the strain tensor for each point in the model.\par We show examples of Permian Rotliegend faults picked from a $10×15$ km$^2$ 3D reflection seismic block from the North German Basin. The Rotliegend aeolian and fluviatile sandstones form an onshore gas play, which lies at a depth of greater than 4500 m. Evaluation of the surfaces of major Rotliegend faults, using their Gaussian curvature and cylindricity, suggests these are normal faults with dominantly 'dip-slip' movement.\par We demonstrate the retro-deformation of these faults, and show the strain tensor values and orientations which occurred in the surrounding strata. We propose the values of the strain tensor, and thus the magnitude of the strain, are equivalent after retro- or forward deformation, and consequently we are able to compare the modelling results with natural data, such as fracture data from boreholes.
S41B-0979
Three-Dimensional Surface Rupture Maps of the 3 November 2002 M7.9 Denali Fault Earthquake Produced Using Digital Photogrammetry
The M7.9 Denali Fault earthquake produced about 340 km of surface rupture along the Susitna Glacier thrust fault and the Denali and Totschunda right-lateral strike-slip faults. Digital photogrammetric methods were used to create a surface rupture map superior to traditional approaches. The process involved: 1) collecting vertical-incidence aerial photographs at a scale of 1:6000 along with aircraft orientation information and GPS positions of each photograph, 2) high-resolution scanning of the photographs, 3) aerotriangulation of the orientation data, 4) 3-D digitization in ArcGIS using Leica's StereoAnalyst plug in. The resulting digital database is high precision with X, Y, Z datapoints collected along the fault trace at intervals from 2-10 m for the entire surface rupture. The absolute accuracy of points along the fault traces is a few meters, but the relative location accuracy can be less than 0.5 m horizontal and 1.0 m vertical. The fault trace map made from this approach is superior to traditional methods in terms of relative and absolute accuracy, completeness, detail, and as a basis for 3-D visualization of the fault trace in three dimensions. Fieldwork complements the air photo observations in locations of dense vegetation, on bedrock, or where the surface trace is weakly developed. The new fault trace map reveals aspects of the geology of the surface rupture. The overall dip of the Susitna Glacier thrust is about $45 north, based on earthquake focal mechanisms. In the near surface, at one location where the fault intersects a stream, the fault dips about $16 northward between 10-30 m depth, but within 10 m of the surface, it rolls over to almost a flat dip. At several other locations the fault rolls over to southward dips and slid downhill at the toe of the thrust. Normal faults demonstrate extension in the hanging wall of the thrust in at least two locations. The Denali Fault strike-slip rupture is typically left stepping with segments 20- to 50-m long. The fault dips north at about $80 near its western extent, where south-side-Up offsets were observed, which suggests a component of extension. The fault rupture through glacier ice is commonly expressed as a wide shatter zone 50-500 m across. The transfer zone from the Denali to the Totschunda Fault has anastamozing segments or normal fault segments as part of the main fault trace. The Totschunda fault is remarkably linear and generally lacks the left-steps of the Denali fault. The completed rupture maps show the details of a large surface rupture in unprecedented detail.
S41B-0980
Electrical Structure of the Shallow Part of the Atotsugawa Fault, Central Japan: Detecting en Echelon Structure in the Fault Zone
Dense VLF-MT and TDEM surveys were carried out to image the electrical structure of a region interpreted as a creeping segment of the Atotsugawa Fault, central Japan. The Atotsugawa Fault is an active fault with a length of 60-70 km and a strike of approximately N60E. The fault type is a right-lateral strike-slip. The most significant characteristic of this fault is a possible existence of creeping segment. In the central region, the stable slip with a rate of 1.5 mm/year was found by the observation of baseline change (Geographical Survey Institute, 1997). However, such slip has not been found at the southwestern region. Therefore, the central region is considered to be a creeping segment. In the creeping segment, many fault outcrops were found on the right bank of the Atotsu-gawa River that runs along the fault. Strikes of shear planes in outcrops were observed to be N30-47E, which is apparently different from that of the Atotsugawa fault. This observation suggested the existence of en echelon structure, which is the cluster of small shear zones oblique to main fault. Investigation of the nature of the en echelon structure will help us to understand the growth history of the Atotsugawa fault and the mechanisms of creeping phenomenon. Because a fracture zone usually includes much water, we can detect it as a low resistivity zone. In order to image the detailed structure of echelon, we carried out the electromagnetic surveys; VLF-MT and TDEM survey as a preliminary and main investigation, respectively. The results of VLF-MT survey has been reported by Yamashita {\it et al}. (2005), and therefore we don_ft refer to the results here. We acquired data at 10000 points with airborne TDEM survey, and over 4000 data were selectively used for modeling the subsurface structure. Apparent resistivity at each point was modeled assuming 1-D structure that consists of 30 and 70 m thick layers on a semi-infinite basement (three layers in total). Because over 4000 survey points were distributed densely, we could approximate the 3-D image of resistivity structure based on 1-D models. The results of TDEM survey showed that the modeled resistivity ranged widely from 10 to 1000 ωm, which was consistent with the results of VLF-MT survey. A low-resistivity zone with a width of about 200 m was identified in planer projection of the 3-D resistivity structure, whereas its location was significantly different from the geologically determined trace of the Atotsugawa Fault. Outstanding short wavelength component of spatial variation in resistivity was also recognized, indicating the existence of the smaller subsurface structure than the main fault. We extracted the lineaments from the planar distribution of modeled resistivity using the Hough transformation. Although the strikes of the extracted lineaments were widely distributed, the mode of them was consistent with the strikes of shear planes observed in outcrops.
S41B-0981
Near-Surface Seismic Images and Geometry of the San Andreas Fault, Santa Cruz Mountains, California
In July 2005, the US Geological Survey and the University of Nevada-Las Vegas acquired a 305-m-long, high-resolution, combined seismic reflection and refraction profile across the San Andreas fault within the Los Trancos Open Space Preserve in Palo Alto, California. The objective of the seismic investigation was to determine the geometry and seismic velocities of the San Andreas fault along a part of the surface rupture area of the 1906 M 7.9 San Francisco earthquake. Seismic sources (sledge hammer blows) and geophones (40-Hz single-element verticals) were co-located (1-m lateral offsets) and were spaced at 5-m increments along the profile. The data were recorded with two 60-channel seismographs without acquisition filters. From the resulting seismic data, we developed tomographic P-wave velocity models of the upper 80 m and stacked and migrated reflection images of the upper few hundred meters along the profile. Seismic P-wave velocities range from about 800 m/s to 4500 m/s in the upper 80 m. For rocks with velocities greater than 2500 m/s, the San Andreas fault is expressed as a well-defined low-velocity zone that is about 60 to 80 m wide. In stacked reflection images, the near-surface (upper 100 m) San Andreas fault zone includes multiple vertically offset reflectors with varying dips. On the basis of offset reflectors and lateral variations in the velocity structure, we suggest that the active fault zone is wider than previously indicated by some geologic maps. Surface ruptures from earthquakes prior to the 1906 rupture have likely involved some or all of the imaged near-surface splays. Other more populated areas along the San Andreas fault probably have similar near-surface geometries, and land-Use planning in those areas should assume similar complex geometries for the San Andreas fault.
S41B-0982
Seismic Imaging Investigation of the Calaveras Fault in Dunne Memorial Park, Hollister, CA
The Calaveras Fault is a major right-lateral strike-slip fault of the San Andreas Fault system in northern California. The southern Calaveras Fault is an area of active creep, which can be seen in the structural deformation of man-made structures in the town of Hollister. Amplification of the soils may result in significant damage to structures in and around Hollister during large-magnitude earthquakes on either the San Andreas or Calaveras faults. In order to understand the subsurface configuration of the fault we acquired high-resolution, shallow-depth, seismic images of an active strand of the Calaveras Fault along a 156-m-long profile in Dunne Memorial Park, Hollister, California in July 2005. The seismic profile was acquired normal to the strike of the creeping section of the Calaveras Fault where there is evidence of both continuous horizontal displacement and small amounts of vertical displacement, down to the west. The surface expression of the fault includes offset curbs (~ 12 cm), bent retaining walls, swells and cracks in the asphalt pavement, leaning houses, offset fences, and a west-facing scarp. The seismic line consisted of shot points (hammer source) and receivers each spaced every 3 m with 1-m lateral offsets between shot points and receivers. For each shot, we acquired 2 s of data at a sampling rate of 0.5 ms. We developed a 2-D P-wave refraction tomography velocity model along the seismic profile by inverting first-arrival refractions using a modified version of the code by Hole (1992). P-wave velocities range from about 400 m/s near the surface to about 600 m/s at a depth of 10-15 m. We also generated stacked and migrated reflection images of the shallow subsurface, which show vertical offsets of layers and laterally discontinuous layers. Both the velocity model and reflection stack infer multiple east- and west-sloping fault splays. These data suggest a complex three-dimensional geometry for the shallow fault zone along the southern Calaveras fault and displays the fault to be more structurally complex than the single trace seen at the ground surface. This complex geometry and the low near-surface seismic velocities suggest that the damage zone from future earthquakes on the southern Calaveras may be wider than anticipated and seismic waves may be amplified by the low-velocity near-surface soils.
S41B-0983
High-Resolution Imaging of the Deep Structure of the Bear Valley Section of the San Andreas Fault With Joint Analysis of Fault-Zone Head and Direct P Wave Arrivals
Understanding the structure of large faults is an important step towards the understanding of earthquake processes on those faults. The short length scales that are important for earthquake physics can not be resolved at depth by conventional geophysical methods. The utilization of seismic energy trapped within low velocity fault zone layers can yield detailed images of the fault structure. However, recent studies at a number of locations have indicated that trapped waves are typically generated only by approximately the top 3km of the fault zones, above the seismogenic portion of the structures. Since major faults typically juxtapose rocks with different elastic properties, the contrast in materials can lead to the generation of fault zone head waves that spend the majority of their propagation paths refracting along the fault interface. The incorporation of fault zone head wave in imaging studies can thus resolve important small scale elements of the fault zone structure at seismogenic depths. In this study we perform a joint direct P and head wave travel time inversion to produce a separate 1D velocity model for each side of the San Andreas fault in the Bear Valley region. The data comes from a dense temporary array of seismometers deployed by Thurber et al. (1997) and the permanent northern California seismic network stations in the area. We have picked arrival times from 450 events at up to 54 stations, resulting in over 9800 direct and over 2700 head, P wave arrival times. One set of inversions is preformed upon the whole data set, and 5 inversion sets are done on various data subsets to try to understand details of the velocity structure. The results imply a strong velocity contrast of ~50% in the near surface that reduces rapidly to 10-20% below 3 km. The presence of a shallow damage zone around the fault is detected by inversions using subsets of the data made up of only stations close to the fault. The faster (southwest) side of the fault shows the development of a low velocity layer at the surface as instruments closer to the fault (<5km and <2km) are used; such a feature is not present on inversions using only stations at greater distances from the fault. On the slower (northeast) side of the fault the presence of low velocity shallow layer is only detected in the inversions using the stations within 2km of the fault. The asymmetry of the shallow low velocity layer may reflect a preferred propagation direction of earthquake ruptures. Using events from different portions of the fault, the head wave inversions also resolve small scale features of the fault visible in the surface geology and relocated seismicity.
S41B-0984
The gouge structure in the San Jacinto fault zone, southern California
The seismic properties of gouge in fault zone have been limitedly investigated due to its narrow width which makes conventional tomographic approaches inapplicable. In particular, the depth extent of the low-velocity gouge remains uncertain, which is very important information for inference of fault properties, e.g., rheology, brittle-ductile transition, pore pressure, and rupture energy. The spatial reciprocity allows us to use natural seismicity to mimic the situation in which seismic arrays are placed deep in fault zone. This experimental geometry is very difficult to achieve in practice. Under the reciprocity theorem, a seismic record at event location for synchronized impulses on the surrounding stations can be constructed. The travel time between given two event locations is estimated using a waveform cross-correlation technique. As events in fault zone are normally clustered on the rupture plane, the estimated velocities directly reflect the structure of the gouge. The technique is applied to 157 events of 1998 to 2004 around the San Jacinto and Elsinore fault zones, southern California. The low-velocity gouge in the San Jacinto fault zone is persistently observed up to the depth of brittle-ductile transition zone (~16 km). On the other hand, the low-velocity gouge in the Elsinore fault zone is only visible in the northwestern portion at shallow depth. The shear velocity in the gouge at the San Jacinto fault zone is slower by about 20 to 35 % than those in the wall-rock.
S41B-0985
Crustal Heterogeneity in the Source Region of the 2004 Mid Niigata Prefecture Earthquake: Inversion Analysis of Coda Envelopes
The 2004 Mid Niigata Prefecture earthquake (M 6.8) and its aftershock sequences generated complicated, i.e., several conjugate fault planes in their source region. In order to understand the generating process of these earthquakes, we estimated a 3-D distribution of relative scattering coefficients in the source region by inversion of coda envelopes of aftershocks. We analyzed 382 seismograms from 70 events which occurred on Nov. 1 to 7, recorded at 7 telemetry stations. The large slip area during the mainshock rupture seems to be a relatively homogeneous part on the fault, bounded by strong heterogeneous zones with larger scattering coefficients. Hypocenters of the mainshock and major large aftershocks with M 5-6 classes tend to be located close to stronger scattering areas. We found that one of these strong heterogeneities already existed before the occurrence of the M 5.9 aftershock on Nov. 8. We suppose that heterogeneous structures in the source region of this earthquake sequence affected the initiation and growth of ruptures of the mainshock and major large aftershocks. We will also discuss the temporal change and frequency dependency of scattering properties using a large data set of aftershocks.
S41B-0986
Trapped wave observation in fault zone of the Mid-Niigata Prefecture Earthquake in 2004
The Mid-Niigata Prefecture earthquake in 2004 occurred in northeast Japan at 17:56, October 23 (JST). In this study, knowledge of the fault-zone structure provides the information required to understand the physics of earthquake. In order to understand the fault structure, we must observe the many aftershocks that occur near the active fault by array observation. Two days after the mainshock, we started an aftershock observation across the active faults in order to obtain the waveform that propagates thorough the fault plane. We succeeded in locating several of the fault-zone trapped waves. The apparent velocities of the fault-zone trapped waves are about 20 - 30 % lower than those of the S waves, and the peaks of the amplitude spectra of the fault-zone trapped waves have low-frequencies between 4Hz and 7Hz. Fault zone trapped waves observed within about 300m of north-west zone of surface fault trace. The ratio of fault-zone trapped waves and aftershocks is only about 1 %. On the other hand, the hypocenter distribution of the characteristic low-frequency (CLF) seismic event corresponds to the region of the large slip distribution of the mainshock. It may be stated that the asperity of the mainshock generates the CLF seismic events and few aftershocks.
S41B-0987
Large-scale Fault Plane of the Western Nagano Prefecture Region Estimated from Precise hypocentral Distribution
It has been considered implicitly that an earthquake occurs on a pre-existing fault plane. However, it has not been well known how the pre-existing fault planes are distributed in a seismogenic region. It is important to identify whether the pre-existing fault planes are randomly distributed or aligned along a certain direction, which forms a large scale fault plane. We estimated the spatial distribution of large scale fault planes from a precisely determined hypocentral distribution, assuming that a large scale structure exists in the region where a planar hypocentral distribution is seen. A dense seismic array with extremely high resolution has been installed in the aftershock area of the 1984 Western Nagano Prefecture Earthquake (Mj6.8) since June 1995 (Iio et al, 1999). Since the recording system has a sampling frequency of 10kHz, P and S onsets are precisely determined. The hypocenters of 24,656 earthquakes that had been observed in this region for the period from October, 1995 to December, 2004 were relocated by using the Joint Hypocenter Determination (JHD) method (Kissling et al, 1994). Because P wave arrival times are read to an accuracy of 1ms, the relative errors of 82% of the hypocenters are calculated to be within 100m. From precise hypocentral distributions, which are shown in vertical cross sections of widths of 500m, we found that the hypocentral distribution is planar, although it looks rather random in a large scale figure. We found that the estimated large scale fault planes are mainly aligned along two oppositely dipping directions of roughly north and south. These orientations are different from the fault plane of the main shock which is a steeply dipping strike slip fault with an azimuth of roughly E-W.
S41B-0988
Earthquake triggering along a segmented creeping fault: 1951 ML7.3 Hualien-Taitung earthquake sequence in eastern Taiwan
Understanding of fault interactions along active plate boundaries is significant for improving seismic hazard assessment. As the most destructive seismic episode even known in eastern Taiwan, the 1951 ML 7.3 Hualien-Taitung (H-T), Taiwan earthquake series provide a good opportunity to study earthquake triggering processes along an arc-continent collision boundary. This sequence occurred at first on October 21, 1951 with the ML 7.3 Hualien earthquake located at the northernmost segment of the Longitudinal Valley Fault (LVF) and then triggered three ML >6 events with 10-40 km surface ruptures propagating southward. The first triggered November shock did not occur at the nearby Yuli fault segment but occurred at the ~100-km away creeping Chihshang fault. By calculation of static coulomb stress transfer with the rate/state stress transfer model, we modeled the temporal priority of encouraged rupture on four segments of the LVF. We found that rupture of the Hualien mainshock is unlikely to trigger the neighboring Yuli fault prior to the creeping Chihshang fault; doing so requires that both the Yuli and Chihshang segments having background aftershock duration of less than 2.5 years. We also found that the factor controlling the priority of large triggered events varies with static coulomb stress change loading on a given fault segment. When the loaded coulomb stress change is less than 0.5 bar, the priority is mostly governed by the seismicity rate. The segment with higher seismicity rate corresponds to faster triggering, which is consistent with our observations. When the loaded coulomb stress change larger than 0.5 bar, the aftershock duration plays an important role on the priority, the segment with longer aftershock duration leads to faster triggering. Modeling of the jumping behavior of the 1951 H-T sequence suggests that the along-strike rupture directivity depends on the location and magnitude of the mainshock and seismicity rate and aftershock decaying rate of individual fault segment, and can be explained by the rate/state stress transfer model.
S41B-0989
Prediction of Triggered Slip and Aftershocks in the Salton Trough: Which is Better, Dynamic or Static Coulomb Failure Stresses?
We have modeled static (CFS) and dynamic (dCFS(t)) Coulomb failure stresses within a 140 km by 140 km area in southern California for four recent historical M>6 earthquakes (1968 Borrego Valley, 1979 Imperial Valley, 1987 Elmore Ranch and 1987 Superstition Hills events) using a fourth-order finite-difference method and a layered crustal model. The dynamic stresses, quantified as the peak values of the dCFS(t), are computed using slip models estimated from measured surface offsets and extended down to the bottom of the fault. The CFS and dCFS(t) are correlated with aseismic slip recorded on nearby structures in the Salton Trough, as well as aftershocks from the four events. Our simple models suggest that, compared to static Coulomb failure stress patterns, the patterns of peak dCFS(t) show a higher correlation with triggered slip along nearby faults, and for some of the events, with the location of aftershocks that occurred up to four years after the events. This finding is due to the rupture propagation effects, particularly directivity, included in the dCFS(t) (e.g., Kilb et al, 2000), but omitted for the CFS. Future studies should include a 3D crustal model and refined rupture propagation in the dCFS(t) computation.
S41B-0990
Seismicity and Stress Change Along the Central Philippine Fault Zone
Recent findings showed that the Philippine fault zone (PFZ) has a creeping section, a transition zone and a locked portion in its central portion. This part of the PFZ is comprised of the Guinyangan, Masbate, northern Leyte and southern Leyte faults. In this study, an attempt is taken to have a closer look on the seismic pattern and relationship between the creep events, moderate quakes and major events along in Guinyangan, Masbate, and northern Leyte, respectively based on the recent available seismic data and field surveys. Dislocation along a fault that involves no rapid release of energy in a seismic event is called creep. Creep was identified along the central PFZ in early 1990s using GPS records. On the other hand, moderate quakes are earthquakes with magnitude range of M5.5~M7. In central PFZ, moderate quakes were recognized to occur along the transition zone through an event in early 2003 that was accompanied by a large ground rupture. Although moderate events have usually a calculated <M7 magnitude, the extent of the observed ground rupture indicated an earthquake with magnitude greater than M7. The moderate events in this area may have a recurrence interval of as short as 10 years. Major events, on the other hand, are devastating earthquakes at least M7. Historical events in this portion of the PFZ show that major events in the locked portion also have a relatively short recurrence interval that range from 60-100 years. The northernmost part of the central PFZ is the Guinyangan fault. This portion is considered locked and the locus of major quakes in the past. South of the Guinyangan fault is the Masbate fault. A very interesting event occurred in this fault in 2003 along the Masbate fault, a transition zone. South of this portion is the northern Leyte fault which is considered undergoing a creep activity. After the 2003 event, the Guinyangan fault was anticipated to experience some changes in local stress field. However, the temporal and spatial plots of seismicity indicate a west-Northwestward propagation of seismic activity along the Sibuyan fault. Sibuyan fault is an offshore fault traversing the Sibuyan Sea and probably into Taal area and/or branches out into the southern part of Marinduque island. Considering the amount of data available and the peculiar seismic activity along the central PFZ, we try to examine the seismicity and determine if a particular seismic pattern is discernable in this region and correlating them to possible regional and local stress change in a specific portion of the PFZ. Considering the importance of the results of this study and in view of the presence of major cities and towns in southern Luzon, the Sibuyan and Guinyangan faults need more attention for seismic and crustal deformation studies. Concerning the disaster mitigation and preparedness in the southern Luzon regions, finding from such researches would be an important input for local stress change research that could reveal the probability of local earthquake occurrence. Keeping in mind that the San Andreas fault has almost the same features, a comparison would be undertaken to note any similarities or differences between the two structures.
S41B-0991
Rupture geometry of microearthquakes inferred from analysis of multiple events
Complicated waveforms of some swarm earthquakes in West Bohemia, Central Europe indicate complicated rupture history and possible space separation of several subruptures. I obtained the position and timing of the subevents building up the multiple event by waveform modeling with the use of empirical Green's functions. In total 18 multiple events were successfully modeled as double or triple events with separate rupture positions. The separation of subsources reached 100 ms in time and 320 m in space. The relative positions of the subevents with respect to the geometry of the fault indicate that most of them occurred very close to the common fault plane that was activated during the swarm. The space-time separation of the subevents corresponds to a speed of 3.0 km/s, a value typical for rupture propagation of large earthquakes. The later subevents occur farther than the nominal rupture radius of the first subevent, and their mutual distance scales with magnitude. These observations suggest that the analyzed multiple-events share a common fault surface and that their subevents represent individual rupture episodes. The angular distribution of the position vectors of later subevents indicates that many of them result from a slip-parallel rupture growth, while some of the ruptures propagate upwards. The hypocenters of the multiple events are not distributed uniformly on the fault plane; their clustering indicates that some patches of the fault are more likely to generate a stick-slip failure than the others.
S41B-0992
Microseismicity Studies in Northern Baja California: General Results.
Between 1997 and 2003, we installed local seismological networks in northern Baja California with digital, three-component, Reftek instruments, and with 100-125 Hz sampling. Each local network had from 15 to 40 stations over an area approximately of 50 x 50 km2. Surveys have been carried out for the Mexicali seismic zone and the Ojos Negros region (1997), the San Miguel fault system (1998), the Pacific coast between Tijuana and Ensenada (1999), the Agua Blanca and Vallecito fault systems (2001), the Sierra Juarez fault system (2002), and other smaller areas (2001 and 2003). These detailed microseismicity surveys are complemented with seismograms and arrival times from regional networks (RESNOM and SCSN). Selected locations presented here have errors (formal errors from HYPO71) less than 1 km. Phase reading errors are estimated at less than or about 0.03 s. Most of the activity is located between mapped fault traces, along alignments which do not follow the fault traces, and where tectonic alignments intersect. The results suggests an orthogonal pattern at various scales. Depth distributions generally have two maxima, one secondary maximum, at about 5 km; the other, located at 12-17 km. The Agua Blanca fault is essentially inactive for earthquakes with M$_{L}$ > 1.7. Most focal mechanisms are strike-slip with a minor normal component; the others are dominantly normal; the resulting pattern indicates a regional extensional regime for all the regions with an average NS azimuth for the P-axes. Fracture directions, obtained from directivity measurements, show orthogonal directions, one of which approximately coincides with the azimuth of mapped fault traces. These results indicate that the Pacific-North American interplate motion is not being entirely accommodated by the NW trending faults, but rather is creating a complex system of conjugate faults.
S41B-0993
Holes In The Hole-in-the-plate Model: The Late Cenozoic Thermal Regime Of Coastal California
In California, the heat flow shows a distinct pattern that can give insight into understanding of the nature of tectonism in the late Cenozoic after the cessation of subduction. In Coastal Ranges heat flow is 50% above the continental average whereas in the Great Valley and western Sierra Nevada region it is 50% below the average. The two thermal anomaly zones extend over 500 km south of the Mendocino triple point along the transform regime. The thermal transition zone is sharp in the north and becomes broader toward the south. Low heat flow is still preserved at areas where the subduction ceased 20 My ago. Using the linear symmetry of the structure and heat flow along the transform margin, we present 2D heat conduction models for the extinct California forearc for various times after the cessation of subduction. We tested two proposed tectonic scenarios, namely the slab window scenario and the stalled slab scenario, in terms of their consequences on the surface heat flow. The first scenario infers emplacement of asthenospheric material to shallow depth whereas the latter implies the thermal re-equilibration of the oceanic slab under the continent after subduction stops. The modeling shows that the slab window scenario cannot match the present low heat flow regime in the Great Valley and Sierra Nevada region whereas the stalled slab scenario cannot match the high heat flow in the Coastal Ranges. Low heat flow in the Great Valley cannot be due to sediment blanketing since the sedimentation rate is slow. In the southern Sierra Nevada region low heat flow occurs 20 My after the cessation of subduction. These facts preclude the possibility of formation of a slab window in these regions. The steep lateral decrease in the heat flow from the Coastal Ranges to the Great Valley shows that the source of the Coastal Ranges anomaly is in the crust. The slab window model is unable to predict this sharp change in heat flow. In the absence of a slab window, the present broad Coastal Ranges high heat flow can best be explained by dissipative heating in the ductile zone of the transform boundary. We conclude that heat flow distribution in California represents the superposition of the thermal regime of the stalled slab and shear heating in the ductile zone of San Andreas Fault system.
S41B-0994
Potentially Pulverized Granites along the Garlock Fault: An analysis into their Physical and Chemical Properties
We collected samples of pulverized granitic rock along three trench exposures across the Garlock fault, Tejon Ranch, CA. Our primary purpose is to assess the roles of mechanical fragmentation versus chemical weathering in the reduction of grain size displayed by samples collected during this study. In each transect, granitic rocks are finer-grained immediately adjacent to the active trace of the fault. Feldspar and quartz grains range from 1-2 mm in size, and material along the fault is so powdered that it was difficult to collect large, intact samples. Quartz and feldspar grains smear to powder when pressed between the fingers. Laboratory work initiated during this study includes bulk and grain density, porosity, mineral XRD, major and trace element chemistry, thin section, and SEM. Resulting data will resolve whether chemical weathering has played a significant role in the observed reduction of grain size. To date, we have completed major and trace element analyses of 11 samples. Resulting Chemical Index of Alteration (CIA) values range from 46-53, and are like values displayed by unweathered granite in the Peninsular Ranges. This result indicates that the finely-powdered granitic material is not distinguishable from fresh, unweathered granite, and that there is no indication of significant chemical weathering. Initial particle size distribution analysis shows grain sizes <0.2 m. These preliminary observations are consistent with the idea that collected samples are derived solely from mechanical (dynamic) fragmentation during slip along the Garlock fault.
S41B-0995
Physical and Mechanical Properties of the Mozumi Fault, Japan: Petrophysics of a Fine-Grained Fault Zone
The Mozumi-Sokenobu fault, a right-lateral strike-slip fault in north-central Honshu, Japan is intersected by the Active Fault Survey Tunnel. This tunnel allows for direct observation of the fault at a depth of 300-400 m below the ground surface. Within the tunnel, the Mozumi fault cuts Jurassic Tetori Group sandstone and shale. We have characterized microstructures, mineralogy, geochemistry, and elastic properties of fault rock samples from the Mozumi fault. These data can be combined to illustrate the in-situ macroscopic hydro-mechanical structure of the fault. Core samples from the main Mozumi fault zone intersected by the Active Fault Survey Tunnel borehole A were analyzed and compared to wireline logs for a petrophysical study of the fault zone rocks. Microstructures, mineralogy, and geochemistry of Mozumi fault rocks indicate syn-tectonic fluid flow and multiple deformation events. Resistivity and sonic log values are depressed through the main fault zone. Likewise, the seismic p and s wave velocity values are decreased across the main fault relative to the surrounding rock. Calculated values for Young's modulus and Poisson's ratio fall at the top of or above the experimentally derived range for elastic moduli of siltstone, shale, and sandstone. Smaller scale variations across the fault zone itself are also present. Samples of foliated fault rocks containing predominantly muscovite have intermediate values for elastic moduli and seismic velocity relative to other fault zone samples used in this study. Fault rocks significantly depleted in oxides relative to host rock samples and containing mixed clays have higher resistivity than surrounding fault rocks and intermediate permeability values. These variations in physical and mechanical properties throughout the fault zone coincide with the complex fault-parallel combined conduit/barrier permeability structure of the Mozumi fault zone.
S41B-0996
Deformation Band Shear Zones Formed in Unconsolidated Sediment From Repeated Late Holocene Coseismic Deformation Along the 1906 Rupture Trace of the San Andreas Fault
Two trenches were excavated across the 1906 rupture trace of the San Andreas Fault (SAF) at Alder Creek, near Manchester, CA, in Mendocino County for the purpose of structural and microstructural analysis of deformed late Holocene unconsolidated sediment. The site records coseismic rupture in the form of upward terminations of deformation band faults and written historical accounts of the 1906 rupture directly adjacent to the site. Based on the age of the deposits exposed and paleoseismic record of the northern SAF, as many as three to five surface faulting events have occurred, including ~5 m of dextral displacement during the 1906 earthquake. Deformation band shear zones, 2-20mm thick, form an upward-branching 1-2m. wide array of splay faults developed in unconsolidated silt and very fine- to medium-grained sand. Vertical displacement on individual deformation band faults is generally <5cm. Microstructural characteristics (porosity, grain size, grain orientation) of oriented samples collected at ~ 2 m in depth were measured in thin section using image analysis of SEM backscatter images. Porosity estimated from SEM images is slightly lower in deformation bands (39.01.8%) than in sand in the same horizon several m. from the fault (42.90.9%) or in 1-5 mm wide sand lenses bracketed by deformation bands (42.92.0%). Deformation band samples contain significantly more very fine and fine sand, and less medium and coarse sand, than samples collected from the same horizon several m. from the fault. Grain size distributions record grain size reduction from grain fracturing within deformation bands. Also, fractured grains in sand adjacent to deformation bands and angular to acicular small grains within deformation bands attest to grain breakage accompanying development of deformation bands. All horizontal samples show a strong preferred orientation of elongate grains. Clustering of grain long axis orientations is more pronounced in deformation band faults than in sand several m. from the fault, but both display preferred orientation of elongate grains parallel or subparallel to the fault zone. This study demonstrates that deformation bands can form in near-surface unconsolidated late Holocene sediments from repeated coseismic displacement along discrete shear planes. These structures record significant porosity reduction, grain rotation, and grain fracturing as a product of surface-fault rupture on the SAF. We speculate that these microstructural characteristics of fault zones in well-sorted unconsolidated granular deposits may be diagnostic of coseismic rupture. Further study of microstructures and the deformation mechanisms that operate during their formation may provide insight into loading conditions associated with earthquake propagation in unconsolidated surface deposits.
S41B-0997
Seismic Energy Partitioning Inferred from Pseudotachylyte-bearing Faults (Gole Larghe Fault, Adamello batholith, Italy)
Fracture energy {\it E}$_{G}$ (the energy used for expanding a rupture surface area) is the fraction of work during seismic faulting that is required for creation of (1) New surfaces in the slip zone, and (2) Damage zone in the wall rocks. Partitioning of the earthquake energy between {\it E}$_{G}$ and frictional heat {\it E}$_{H}$, determines the features of the rupture propagation and the mechanical behavior of the fault. The cataclastic microstructures associated with pseudotachylyte (solidified clast-laden friction-induced melt produced during coseismic slip) veins might contain information about the partitioning. In this preliminary study we used microstructural observations on pseudotachylytes from the Gole Larghe Fault zone (Southern Alps, Italy) to determine both {\it E}$_{H}$ and {\it E}$_{G}$. The {\it E}$_{H}$ for unit fault surface area {\it A} is estimated from pseudotachylyte vein thickness {\it t}. The energy required to produce friction melt is {\it E}$_{H}$/{\it A}= [(1-$\phi$) {\it H}+ {\it c}$_{p}$ ({\it T}$_{i}$-{\it T}$_{hr}$)]ρ{\it t} where $\phi$ is the volume ratio of lithic clasts within the pseudotachylyte, {\it H} is the latent heat of fusion, {\it c}$_{p}$ is the specific heat at constant pressure, ({\it T}$_{i}$-{\it T}$_{hr}$) is the difference between initial melt temperature and host rock temperature and ρ is the density. The {\it E}$_{G}$ is estimated by multiplying the newly created grain surface per unit of fault area {\it A} by the specific surface energy (J m$^{-2}$). In fact the studied pseudotachylyte vein contains plagioclase clasts displaying a characteristic internal fragmentation not observed in the host rock. This indicates a direct association between newly created grain surfaces and the seismic rupture process via pseudotachylyte production. It follows that pseudotachylytes might yield information on the energy partitioning between {\it E}$_{G}$ and {\it E}$_{H}$.
S41B-0998
Experimental observations of slow aseismic failure due to intra-crystalline plasticity in Carrara marble
Two triaxial compression experiments were performed on Carrara marble under wet and dry conditions respectively. The rock samples, first deformed at high confining pressure in the cataclastic regime, were brought back into the brittle field at constant differential stress by increasing the pore pressure and/or reducing the confining pressure. When returning to the brittle field, both samples exhibited exponential increases in axial strain, which eventually led to tertiary creep and failure nucleation. Very little energy was released in the acoustic frequency range (100KHz-1MHz) during rupture. Slip propagation was slow (60 and 500 seconds in the dry and wet case respectively), although failure was accompanied by stress drops of the order of 150 MPa, and millimetric slips. Even in dry conditions, a continuous acoustic recording over the course of rupture shows that the slip initiated aseismically. In this test, the continous recording shows that fast frictional sliding then induced some acoustic activity, probably due to asperity shearing, thus illustrating the transition from aseismic to seismic slip. Elastic wave velocity recordings demonstrated aseismic damage accumulation prior to failure and elastic wave velocity inversion showed that rupture occurred at crack densities close to one, both in the wet and dry experiments. Microstructural analysis highlighted strong interactions between plastically accommodated shear deformation at the intragranular scale and brittle deformation at the macroscopic level. Microscopically, shear deformation was accommodated by twinning and dislocation glide. Howevere, differential plastic strains due to random crystallographic orientations were necessarily accommodated by the growth of ductile cracks at the macroscopic level, in part due to dislocation pileups. Plastic relaxation and dislocation loops at propagating crack tips enabled ductile aseismic growth and slow aseismic failure was triggered in a way analogous to silent earthquakes observed in the field. To our knowledge, these experiments provide the first clear case of silent, strain dependent failure in rocks as a result from intragranular plasticity.
S41B-0999
Experimental investigation of the effect of temperature on friction in gabbro and granite
High speed (~1 m/s) rotary shear experiments on gabbro samples initially at room temperature have exhibited a complex evolution of the apparent coefficient of friction (Hirose and Shimamoto, 2003, 2005). Friction initially drops with increasing displacement, then rises to a peak value and, finally drops to a residual level as the sample develops a continuous molten layer, and viscosity dominates the frictional behavior. Because friction is a result of the interaction of asperities in contact between the two surfaces, such that most of the normal load is borne by these contacts, the latter experience high local normal and shear stresses. Sliding of the interface leads to momentary contacts between pairs of asperities, which may under certain conditions result in a momentary softening or melting of a thin layer of material at the asperity tips (Rice, 1999). The result would be a weakening of the asperity tips, but no change in contact area. Further sliding ends individual contacts, allowing the layer to again solidify or harden. At high slip velocities typical of earthquakes, this mechanism could act to lower friction initially. However, as sliding continues, heat produced by the high stresses at the asperity contacts will diffuse further into the asperities and to the adjacent area to create a more uniform heating that affects the rock surface along the entire interface. Thermally-activated plasticity or fracture may flatten and broaden the asperities, increasing the total area of contact, thus also increasing the force required to maintain sliding. In an attempt to examine the role of temperature on friction, we conducted a series of direct shear experiments on prepared samples of gabbro, granite and quartzite of uniform roughness under normal stresses ranging from 2-23 MPa and at imposed temperatures ranging from room temperature to 400C. The experiments were conducted at velocities between $10^${-4} to $10^${-1} mm/sec in steps of 1-2 orders of magnitude. Temperature was controlled to within .5C, and normal stress to within .1 MPa. The geometry of the samples allowed 40 mm of sliding contact over a constant area of $32cm^${2}. This experimental setup is unique in that it permits direct shear tests both over longer distances and at higher temperatures than previously reported. Results suggest a temperature hardening in both granitic and gabbroic samples, although the data reveal a considerable scatter. We interpret these data as indicating that the temperature-induced changes in the contact topology that lead to increases in the effective contact area offset the effects of thermal weakening of the asperities.
S41B-1000
Frictional behavior of natural clay-rich fault gouges from ODP Leg 190, Nankai Trough, offshore Japan
The decollement at subduction interfaces localizes within sediments carried by the subducting plate, and therefore understanding the frictional behavior of these sediments provides insight into the initiation of the decollement and the nature of slip along it. Variations in frictional behavior due to the conversion of smectite to illite may affect the strength of subduction interfaces and the updip limit of the seismogenic zone. Recent work has shown that illite exhibits velocity strengthening behavior, and this study expands on previous experimental conditions by testing saturated, intact, natural samples, which are sheared parallel to the incipient plate-boundary fault. Here we present results from a mineralogically-varied suite of intact samples from Ocean Drilling Program (ODP) Leg 190, site 1174, offshore Japan. These samples range in bulk illite abundance from 21-35 wt. % and in total clay content from 45-67 wt. %. Thus our experiments provide a natural test of mineralogically-controlled frictional variability, and act as a pilot study into the role of depositional fabrics in shearing. Samples were prepared as ''wafers,'' where ODP cores were cut parallel to the depositional surface to produce thin, intact, rectangular layers, which were then sheared between Berea Sandstone and Westerly Granite driving blocks. Prepared layers of intact gouge were saturated at a controlled fluid pressure of 1 MPa, and sheared in a triaxial apparatus at normal stresses of 20, 50, 120, and 150 MPa. Initial sample thickness varied from ~1-7 mm, with maximum shear strains of ~1.5-10. Corrections were made primarily for displacement-based area change (contact area is decreased by ~40% at a maximum displacement of 10 mm) and jacket strength (linear increase in measured shear stress with displacement). Preliminary results show 1) values of μ (~0.25-0.4) similar to that previously reported for clay mixtures; 2) velocity strengthening behavior (a-b > 0) in all cases, consistent with the stably-sliding behavior of similar clay mixtures noted by other researchers under a variety of conditions; 3) no clear, systematic variations in strength with either total clay fraction or illite content; and 4) no obvious effect of depositional fabric on frictional behavior.
S41B-1001
Depth Dependence of the Fault Strength in the Creeping Section of the Atotsugawa Fault, Japan
The Atotsugawa fault is located along a highly deformed region in central Japan with 60km long, striking to N60$^o$E and dipping to 90$^o$ 10$^o$. From the laser distance measurement survey, a creeping section (1.5mm/y) was found in the northeastern part [Geogr. Surv. Inst., 1997]. In this section, a low seismicity area down to a depth of 7km was found above the seismically active region down to 15 km [Ito and Wada, 1999]. In order to investigate the depth dependent feature of the fault strength, we conducted tri-axial friction tests of the Atotsugawa fault gouge under the conditions of 1, 3, 5 and 7km depth. The NIED drilled a borehole in the fault zone down to a depth of 350m in this creeping section [Omura et al., 2004] and obtained core samples consisting of fault gouge, fault breccia and fractured host rocks (granitic rocks and hornblende gneiss). The samples are taken in the gouge zone (8.5mm in thickness) located at a depth of 342 m. The samples were disaggregated in distilled water and passed through a 100μm diameter sieve for the friction tests. From the XRD analysis, the gouge sample consists of quartz, feldspar, smectite, kaolinite and micas. The average grain size in the sample was approximately 16.9μm measured by a laser diffraction particle size analyzer. The friction tests were run using a gas-medium tri-axial apparatus at the AIST, Japan [Masuda et al., 2002]. For each run, 0.5g gouge powder was put between 30$^o$ sawcut of an alumina ceramic cylinder (20mm in diameter) and sheared at a constant axial slip rate of 0.1μm/s. Each test was done with pore fluid of distilled water at the temperature-pressure conditions of 1-7 km depths assuming a hydrostatic pore-pressure gradient of 10MPa/km, a lithostatic confining pressure gradient of 26MPa/km and a geothermal gradient of 30$^o$C/km. In all experiments, the friction increases rapidly to an axial displacement of about 0.1mm, and then it gradually increases or becomes steady state. We found a strong depth dependence of friction; it increases from 0.25 - 0.3 at 1km to 0.5 at 7km. We need additional experiments to obtain a physical explanation on this depth dependence. However this result gives us useful information for the creeping motion observed at the Atotsugawa fault. If the creeping motion terminates at a depth around 7km corresponding to the lower boundary of the seismic gap, the frictional strength deeper than 7km should be equal to or more than the shear stress applied to the fault from the tectonic stress field. As far as a linear dependence of stress along the depth is assumed, the ratio of shear stress to effective normal stress (normal stress - hydrostatic pore pressure) on the fault should be constant. At shallow part where the friction is smaller than 0.5, the fault cannot sustain the applied shear stress and it is forced to slip. Slip along the fault results in a decrease in the applied shear stress. We modeled the creeping motion along the fault to balance the applied stress and the depth dependent fault strength obtained by the experiments. The obtained model is that at a depth of 0-2km the fault creeps at a rate of 1.5mm/y and it decreases down to 0.75mm/y at a depth of 6-8km.
S41B-1002
A Laboratory Investigation of Off-Fault Damage: Effects on Rupture Velocity
Rice et al. (2005) formulated an analytical model for dynamic propagation of a slip-pulse on a fault plane. Using earthquake parameters analyzed by Heaton (1990), they found that stress concentration at the rupture front should produce granulation of fault rock to a distance of a few meters and wall rock fracture damage to 10s of meters. This off-fault damage contributes to the fracture energy and therefore affects rupture velocity; an effect not addressed by Rice et al. Our challenge is to quantify this feedback for incorporation into the slip-pulse model. To this end we conducted 35 experiments using photoactive homalite samples (Xie et al., 2004). We measured rupture velocities in samples having off-fault "damage elements" introduced in the form of small slits of different lengths that intersected the fault plane over a range of angles. Most experiments with damage elements 1.0 cm and 0.5 cm in length and oriented 45 degrees to the fault plane temporarily decreased the rupture velocity in the area of the element but did not nucleate new damage. We attribute the transient velocity change to a reduction of the stress intensity factor, K-II, due to blunting of the rupture tip by the damage elements. In these cases the rupture velocity was restored and temporarily augmented for a short distance beyond the damage element until the rupture displacement matched that expected for linear propagation in the absence of the damage element. In a few experiments with shorter slits (0.5 cm and 0.25 cm) oriented 45 degrees to 70 degrees to the fault plane, the damage element nucleated additional damage in the form of a mode-I tensile wing-tip crack. As the rupture approached the damage element a velocity reduction occurred followed by a velocity rebound, but these experiments produced a permanent delay in the displacement-time curves. We attribute this offset to energy dissipation due to frictional sliding and increased fracture energy necessary to create fresh crack surface.