S43A-0972 1320h
Seismic Microzonation Based on Geotechnical Parameters - Estimation of Site Effects in Bucharest (Romania)
Romania frequently is hit by strong intermediate depth earthquakes occurring in the Vrancea area in the SE Carpathians. During the 20$^{th}$ century four major earthquakes (moment magnitudes between M$_{W}$ = 6.9 and M$_{W}$ = 7.7) occurred in this region which strongly affected the Romanian capital Bucharest (1.9 mio. inhabitants). Due to the large hypocentral distance between the Vrancea area and Bucharest (approx. 160 km) influences of source directivity and travel path effects are assumed to be constant in the entire city. Therefore significant lateral variations in ground motion must be caused by site effects, related to near-surface geology, mainly dominated by large alluvial and diluvial deposits and anthropogenic backfill. The objective of this study, which is part of the German Collaborative Research Center (CRC) 461 `Strong Earthquakes: A Challenge for Geosciences and Civil Engineering', is to quantify the local influence of site effects on earthquake triggered ground motion and to generate a microzonation map of Bucharest. At first a numerical modelling of ground response was performed by using one-dimensional linear-elastic approaches and geotechnical data, derived from the digital geological subsurface model. The analysis of ground motion was carried out at discrete points. For these raster dots the transfer functions were computed and parameterised by characteristic shake parameters (e. g. dominant frequencies, peak amplifications or spectral amplifications at characteristic frequencies). After spatial interpolation of the computed values these parameters could be presented in continuous microzonation maps. Further the surficial ground motion was computed using a novel (visco-) hypoplastic constitutive law that also takes non-linear soil behaviour into account. The results of both approaches were compared with each other and indicate, that the consideration of non-linear effects is required to get more realistic results for the microzonation.
http://www.agk.uni-karlsruhe.de/projekte/projekte_ing/mikrozonierung/
S43A-0973 1320h
Estimation of Gas-hydrate Saturation at the Hydrate Ridge, Offshore Oregon
Multicomponent seismic data were acquired in summer 2002, offshore Oregon, on the Hydrate Ridge of the Cascadia convergent margin. The primary goal of the experiment was to map the gas hydrates and free gas (methane) and understand the mechanism of fluid migration. Our analysis estimated the compressional- (P-wave) and shear-wave (S-wave) interval velocities with the final goal of relating these velocities to the presence and quantification of gas hydrate and free gas. We performed interval velocity analysis in the tau-p (intercept time - ray parameters) domain following three main steps: 1) P-wave velocity analysis, 2) P- to S-wave (converted S-wave) event correlation, and 3) S-wave velocity analysis. We correlated P- to S-wave event for a reflector using synthetic seismograms (generated with reflectivity method) and traveltime tables. A traveltime table contains the arrival times of different wavetypes for a reflector and is calculated using the sonic logs or modeled velocities. Seismic velocities are correlated to gas hydrate saturation using a Modified Wood equation. We found that maximum saturation of gas hydrate is 7% of pore space. The P-wave velocity is lower down to 50m below the sea floor at the south summit, probably because of the presence of free gas within the gas hydrate stability zone (GHSZ). Gas probably migrated from below GHSZ up to sea floor through near-vertical fractures. We identified hydrate-bearing- and free-gas-bearing sediments with anomalous increase and decrease of P-wave velocity, respectively (with a reference velocity calculated in water saturated sediments). However, S-wave velocity does not show anomalous increase in the hydrate-bearing sediments. Thus, we interpret that hydrate does not cement sediment grains enough to affect shear properties. It is more likely that hydrates form within pore spaces.
S43A-0974 1320h
Determination of Low-Strain Site Amplification Factors in the Salt Lake Valley, Utah, Using ANSS Data
Using data from the Advanced National Seismic System (ANSS) network in and near the Salt Lake Valley (SLV), Utah, we measured average, frequency-dependent, low-strain site amplification factors for site response units mapped by others on the basis of geology and near-surface shear-wave velocity. The site amplification factors were determined using distance-corrected spectral ratios between horizontal-component ground-motion recordings from soil sites and reference rock sites. To test various models for the distance correction terms, we measured spectral ratios between recordings at 12 Paleozoic rock sites. These spectral ratios indicate that the ground motions decrease with hypocentral distance, r, at rates of r$^{-1.5}$ in the period range 0.4 to 2.0 sec and r$^{-2.0}$ in the period range 0.1 to 0.5 sec. We calculated the soil/rock spectral ratios using two different reference stations on Paleozoic rock. Geometric mean site amplification terms for three SLV site response units were obtained by combining data from both reference stations. Comparing the resultant site amplification factors to those of previous studies indicates that empirically-based predictions better fit the observed data. Specifically, the empirically-based site amplification factors of Borcherdt (1994) and Boore and others (1997) fit the data better than the theoretically-based factors of Wong and others (2002), even though the latter were developed specifically for the SLV site response units.
S43A-0975 1320h
Monitoring non-linear wave propagation in the 2004 Garner Valley demonstration project
On August 18-19, 2004, we performed a series of prototype experiments near the NEES Garner Valley Digital Array (GVDA) to assess the characteristics of non-linear elastic behavior at a deep soil site. We made several different types of observations designed to quantify modulus reduction associated with non-linear wave propagation. In one, a small-scale (1m station spacing) array of accelerometers recorded strong ground motion, both vertical and horizontal, immediately adjacent to a specially designed vibroseis truck (NEES's "TRex"). In this study we focus on the observations from a related experiment: a small-scale (2m station spacing) multi-channel seismograph array that used a sledgehammer source to estimate wave propagation velocities before and after a period of intense ground shaking. This experiment was designed to test the hypothesis that if strong shaking caused rather long-lived changes in the material properties of the soils at the GDVA site, we would see changes in the propagation velocities of surface waves that traversed the experiment locale. Operational considerations permitted us to make three sets of observations before shaking the site, and only one afterward--about 2 hours after. Our stations were laid out in a "star" pattern with TRex in the center of 4 lines of 6 seismometers each (TRex's vibrator pad is about 2.5 m square). To date we have only examined data from one of the lines. Our preliminary analysis reveals evidence for significant changes in the dispersion of surface waves before and after strong shaking. Seismic velocities 2 hours after strong shaking were on the order of 10% slower than before shaking. The volume of material affected extends at least 6 m from the vibroseis pad. It is unclear at this time how much of the difference in velocities was due to modulus reduction, and how much to compaction-induced density increase. The results will be used to help constrain interpretations of recordings of the strong shaking itself made by the nearby array of accelerometers.
S43A-0976 1320h
Shallow Seismic Reflection Survey at Garner Valley Digital Array
The Garner Valley Digital Array (GVDA) site is a NEES-sponsored facility in a small, sediment-filled, intermountain valley in Southern California, established for the purpose of investigating ground motion site response and soil-structure interaction, in situ. The site has been well-characterized geotechnically, and is thoroughly instrumented with both surface and downhole instrumentation of various types. Nevertheless, a borehole recently drilled into lake bed sediments and deeply weathered granitic rocks that comprise the valley fill at GVDA encountered hard, unweathered bedrock at an unexpected depth, suggesting an apparent 38 meter offset in the unweathered bedrock between two wells 40 meters apart. The apparent offset can be most easily explained either by faulting, or as a buried erosional surface. The Hot Springs fault, a strand of the San Jacinto fault zone, runs through Garner Valley, although its inferred location is several hundred meters east of GVDA. To better characterize the subsurface strata, particularly the existence and configuration of faulting that may disturb them; we conducted a 120-meter long, 12-fold shallow seismic reflection common midpoint (CMP) survey at GVDA using a 24-channel seismograph, vertical 4.5 Hz geophones at 2-meter intervals and a sledgehammer seismic source. Preliminary processing reveals strong refractors and surface waves that may mask reflections, although reflections are visible in some raw shot records. Semi-continuous reflections seen in the CMP section from a shallow reflector may coincide with the water table. There are also deeper, discontinuous reflectors obscured by bands of coherent noise. We plan to present a fully migrated and interpreted CMP record section.
S43A-0977 1320h
Progress in Kappa Estimation at Yucca Mountain
We have been studying attenuation parameter kappa using recordings from the Southern Great Basin Digital Seismic Network in support of site and engineering characterization of the Yucca Mountain area in southern Nevada. Earthquake magnitudes range from 0 to 4.4, with 18 of M$>$=3.0. Kappa is estimated from the residual spectral slope after removing an omega-squared source spectrum. The kappa estimate primarily affects the median expected ground motions; uncertainty in the estimate can strongly increase probabilistic peak ground motion estimates when facility lifespans of 10$^{4}$ years and greater are considered. We find that it is rare to resolve corner frequencies $>$6 Hz, even with high SNR recordings. With the available data, if site kappa is estimated solely from spectra with resolved error minima, the results can be biased toward lower kappa and stress drop estimates. To better constrain kappa, two limiting estimates are made from each spectrum, one assuming a corner frequency (cf) of zero, and the other, that it is infinite. For M$>$3 events, cf=0 corresponds to the original Anderson and Hough (1984) definition of kappa as the asymptotic high-frequency slope parameter. For M$<$1 events, cf=inf exploits the flatness of the displacement spectral slope well below the corner frequency. Any observed slope under these conditions is caused by attenuation and modeled by kappa. Finite corner frequency effects cause these estimates to under- and over-estimate kappa, respectively. Limiting estimates bound kappa to a $\sim$15 msec range, within which are found most of the resolved estimates from fitting. Two lines of evidence suggest that some variability in kappa must originate near the source. First, displacement slope kappa estimates from small earthquakes from a source region of 1-2 km square vary by over 20 msec. The source region is small enough that path or near-station variations are unlikely. Kappa variation may be due to near-source scattering or very low stress drops ($<$0.1 MPa). Second, inversions that include a source term in the kappa model consistently confirm its statistical significance. Estimates of site mean kappa depend on the method of distance correction. Over the distance range considered, the distance correction is weakly-constrained. A bilinear model that fits reasonably well indicates an average kappa for selected Yucca Mountain stations near 30 msec. Improved estimates and analysis considering energy and stress-drop constraints are being developed.
S43A-0978 1320h
Shear Velocity Structure Beneath the Las Vegas Valley, Nevada From Regional and Teleseismic Events
The Las Vegas Valley (LVV), Nevada is located in the central Basin and Range province of western North America. The Valley sits atop a broad sedimentary basin and is susceptible to large earthquakes generated by local and regional faults. During ground motion events, the Las Vegas basin has been found to amplify seismic energy. In addition, the crustal and upper mantle structure of the Valley is poorly understood. Therefore, surface wave data have been used to create shear velocity profiles of the crust and upper mantle beneath LVV using regional and teleseismic events. This project is part of a larger collaborative study, which is characterizing the Las Vegas basin for test site readiness and seismic hazards. Although the frequency of large events is small, the risk associated with such an event is very high for the Valley. As a result, the Las Vegas Valley Broadband array (LVVBB) was deployed in late September 2002 by Lawrence Livermore National Laboratory and the University of Nevada Las Vegas. It consists of a mixture of twelve three-component broadband and short period seismometers deployed in a saw-tooth geometry oriented northeast-southwest across the northeastern and central LVV, above the area estimated to be the deepest portion of the basin. Data examined as part of this study include both regional and global earthquake events recorded within a five-month period on seven of the twelve stations. All seven broadband stations used a three-component Guralp CMG40T sensor with a 40 samples/second sampling rate. Group velocity dispersion curves from Rayleigh waves and Love waves were determined using a multiple filter technique. Rayleigh wave group velocities range from 2.7 to 3.5 km/s for periods from 10 to 30s. Love wave group velocities range from 3.1 to 4.0 km/s for periods from 10 to 100s. In addition, Rg and Lg were examined from local events. 1-D shear velocity profiles of the crust and upper mantle have been produced through inversion along regional source-receiver paths and two-station paths. Shear velocities at depths of 3-5 km have never been directly determined in the Valley; these values will be used to better understand the seismic hazards in the area as well as the tectonic development of southern Nevada in the Basin and Range province.
S43A-0979 1320h
Seismic Wave Amplification in Las Vegas: Site Characterization Measurements and Response Models
As part of a multidisciplinary effort to understand seismic wave amplification in Las Vegas Valley, we conducted geotechnical and seismic refraction field studies, geologic and lithologic interpretation, and geophysical model building. Frequency-dependent amplifications (site response) and peak ground motions strongly correlate with site conditions as characterized by the models. The models include basin depths and velocities, which also correlate against ground motions. Preliminary geologic models were constructed from detailed geologic and fault mapping, logs of over 500 wells penetrating greater than 200 m depth, gravity-inversion results from the USGS, and USDA soil maps. Valley-wide refraction studies we conducted in 2002 and 2003 were inverted for constraints on basin geometry, and deep basin and basement P velocities with some 3-d control to depths of 5 km. Surface-wave studies during 2002-2004 characterized more than 75 sites within the Valley for shear velocity to depths exceeding 100 m, including all the legacy sites where nuclear-blast ground motions were recorded. The SASW and refraction-microtremor surface-surveying techniques proved to provide complementary, and coordinating Rayleigh dispersion-curve data at a dozen sites. Borehole geotechnical studies at a half-dozen sites confirmed the shear-velocity profiles that we derived from surface-wave studies. We then correlated all the geotechnical data against a detailed stratigraphic model, derived from drilling logs, to create a Valley-wide model for shallow site conditions. This well-log-based model predicts site measurements better than do models based solely on geologic or soil mapping.
http://www.seismo.unr.edu/hazsurv
S43A-0980 1320h
Three Urban Transects of Shallow Shear-Velocity Using the Refraction Microtremor Method
Surveys of shallow shear velocity in the Reno, Los Angeles and Las Vegas urban areas give us information assisting in the mitigation of earthquake hazards, and in response to a damaging earthquake. The three transects (16, 60, and 15 km long respectively) were completed quickly and economically using the refraction microtremor method, providing characterizations of over 300 separate sites. Shear-wave velocity averaged over 30 m depth (Vs30) is unexpectedly smooth along all three transects, exhibiting realistic fractal dimensions of 1.5 to 1.8. For the Reno transect, which follows the Truckee River, most of the urban area shows NEHRP class C velocities. Very little of the profile shows velocities in the NEHRP class D range. Neither geological nor agricultural soil map units can serve to predict NEHRP hazard class or shallow velocities in the Reno area, aside from distinguishing between volcanic rock and sediments. Measured along the San Gabriel River, Vs30 values through Azusa and the Santa Fe Dam area are significantly higher than the NEHRP C/D range (270-555 m/s) expected from existing hazard mapping. Vs30 values in the southern 30 km of the Los Angeles transect were well within the previously expected NEHRP class D range (180-360 m/s). Measurements at Alamitos Bay were slightly above the expected NEHRP D/E range (90-270 m/s). In Las Vegas, the lowest velocities in the transect, at the NEHRP C/D boundary, are near intra-basin faults found near I-15 and Lake Mead Blvd. Velocities then rise smoothly to the middle of the NEHRP-C range near Sahara Blvd. to the south. Caliche cementation of alluvium in parts of Las Vegas elevates Vs30 values to 500-600 m/s. Vs100 values for the three transects, averaging to the 100 m depth to which most of our measurements are valid, show trends mimicking the Vs30 trends. Across all three cities our measurements correlate poorly against available USDA soil maps, geologic mapping, and hazard mapping. Surface maps do not predict the conditions of any individual site with accuracy sufficient for engineering application. A detailed stratigraphic model derived from deep water-well logs in Las Vegas predicts Vs30 better than maps, but only in well-measured areas.
http://www.seismo.unr.edu/hazsurv
S43A-0981 1320h
Site Amplification, Scattering and Intrinsic Attenuation in the Mississippi Embayment
Site amplification, intrinsic Q (Qi) and scattering Q (Qs) values are measured from coda of broad-band data recorded by the Cooperative New Madrid Sesimic Network stations. For the frequency range between about 0.5and 1.0 Hz, site amplification factors for the transverse component recorded by stations in the Mississippi embayment are between 1.7 and eighteen, compared to stations outside the embayment. The amplification factor becomes smaller at center frequencies lower than 0.5Hz and higher than 1.0Hz. The radial component shows similar amplification, however, the vertical component shows less amplification compared to the other components. Knowledge of site effect is a useful parameter for seismic hazard assessment of a given region. There is a general belief that the impedance contrast between the basement rocks and the overlying sediments of sedimentary basins is responsible for the amplification of coda amplitudes recorded by stations there. However, we found it very difficult to explain this using a simple theory of plane wave propagation through a layered structure of New Madrid Seismic Zone. The Physical mechanism responsible for the amplification of seismic amplitudes under stations in the Mississippi embayment is not known. For each seismogram used in this study, we computed the average pre-event noise level, which is part of the seismogram between the zero recording time and the onset of the P-wave. As is in the case of the SH coda, the pre-event noise recorded by stations within the embayment are larger than those recorded by stations outside the embayment. Qi and Qs are measured from coda using an energy flux model at 0.5 and 1.0Hz for lapse time bands: between the direct SH-wave arrival time (td) and twice this time (2td), and between 2td and a time where the noise level begins. For band 1, most stations within the embayment show higher Qi, but lower Qs, values than stations outside the embayment. Energy from reverberations beneath stations in the embayment may cause the apparent high Qi values there. For band 2, stations outside the embayment have higher Qi and Qs values than stations in the embayment. Energy from reverberations under stations is assumed to be absent for band 2. Both Qi and Qs values are lower at 0.5Hz than at 1.0Hz, showing a frequency dependence. Partially fluid/gas saturated sediments or fluid flow in pore spaces or cracks in the upper crust may be the possible mechanisms for low Qi (high intrinsic attenuation) found in the embayment. Presence of large lateral velocity heterogeneities in the embayment may cause Qs values to be low there.
S43A-0982 1320h
Comparative Study of Local Magnitude Scales for Central U.S. and Western India
Seismic waveform data from 816 aftershocks of 2001 Bhuj Mw 7.7 earthquake, recorded by CERI/STAR aftershock deployment, including eight K2 6-channel dataloggers with 3-component episensor accelerometers and 3-component L-28 geophones, were used to develop a local magnitude scale for Kachchh basin of western India. Results show that the distance correction curve can be expressed as -logA0=1.8286*(r/100.0)-0.0052*(r-100.0)+3.0, displaying weak distance attenuation. This result is much like that of the local magnitude scale for the Central U.S. Both scales show weak distance attenuation, compared with the local magnitude scales for southern California or Tanzania, East Africa, and display a negative K parameter, implying similar patterns of seismic wave spreading in these two regions. These results for local magnitude scales give support to the assertion that the Kachchh basin and New Madrid Seismic Zone are geological analogs inasmuch that ground motion and other seismological results from one area can shed light on similar problems in the other area.
S43A-0983 1320h
A Stochastic Estimate of Ground Motion at Oceano, California, for the M6.5 December 22, 2003, San Simeon Earthquake, Derived from Aftershock Recordings
The U.S. Geological Survey deployed a digital seismic station in Oceano, California, in February 2004, to investigate the cause of damage and liquefaction from the 22 December 2003 {\bf M}6.5 San Simeon earthquake. This station recorded 11 $M\> 2.8$ aftershocks in almost eight weeks. We use these recordings, together with recordings of the main shock and the same aftershocks obtained from nearby stations in Park Hill and San Luis Obispo, to estimate the mainshock ground motion in Oceano. We estimate the Fourier amplitude spectrum using a generalized spectral ratio analysis that averages the spectral ratios from both stations for all the co-recorded aftershocks. We test three aftershocks as Green's functions by comparing simulated and recorded acceleration amplitude spectra for the main shock at Park Hill and San Luis Obispo. Instead of deconvolving the aftershock recordings from the mainshock recordings to estimate a source-time function, we convolve the aftershock accelerograms with a stochastic operator to simulate the duration and phase of the mainshock accelerograms. These stochastic operators are determined as sets of delta functions whose delays are randomly generated from a gamma distribution with a shape parameter of 1. We choose the scale parameter by fitting Husid plots of the Park Hill and San Luis Obsipo mainshock accelerograms. This stochastic approach allows us to extend the range of aftershocks that can be used as Green's functions to events nearly three magnitude units smaller than the main shock. Our realizations for the mainshock accelerogram at Oceano yield PGAs distributed as 28$\pm$4% {\it g}. We interpret these realizations as upper bounds for the actual ground motion because our analysis assumes that the ground behaved linearly, while the liquefaction and lateral spreading indicates that the ground behaved non-linearly. Geotechnical analysis of the site indicates that a PGA of 25% {\it g} would have initiated the liquefaction.
S43A-0984 1320h
THE PREDICTABILITY OF SITE RESPONSE BASED ON THE DATA FROM YOKOHAMA HIGH-DENSITY STRONG MOTION ARRAY
By determining the site response at 150 stations of the Yokohama high-density accelerometer array we can analyze the spatial variation of site response over distances from 100 m to 20 km. Using accelerograms at 9 borehole stations we have isolated the frequency dependence in the source, path and site response. We use a nonlinear heat bath algorithm to solve for seismic moment, corner frequency, frequency dependent Q and frequency dependent borehole response (Mo, fc, Qofx, Sb(f)) by iteratively comparing a computed spectrum with the observed spectrum (Tsuda et al., 2004). We found Q(f) = 285f0.06 for 24 deep events (depth > 40km) and Q(f) = 74f0.28 for 4 shallow events (depth < 25 km). The seismic moments are generally within the factor of two of values determined by NIED. Seismic moments are roughly proportional to fc-3 with a nearly constant stress drop of 100 bars for the deep events. Knowing Mo, fc, and Q(f), we compute the site response for the surface stations as the difference between the predicted and observed amplitude spectrum. A final site response is found by averaging the difference for all of the recorded events. The surface site responses for 0.5-1.0 Hz correlate with the averaged S-wave velocity for the upper 30m. Having obtained site response for 150 stations we have examined the predictability of site response by analyzing the frequency and spatial dependence variation of the site response among the different sites. We have considered both site-to-site and event-to-event variation. In our analysis we examine the ratio of the site response between two stations: a ratio that is frequency dependent. We divided the 150 stations into 7 groups based on the averaged S-wave velocity for upper 30m. The ratio is calculated for any two sites within 5 km of each other. The results showed that the variance of the site response is greater for different frequencies as compared to the spatial separation. We also analyzed event-to-event variation. Event-to-event variation based on shallow vs deep focal depths is small compared to site-to-site variation.
S43A-0985 1320h
GROUND MOTION CHARACTERISTIC IN THE KAOHSIUNG & PINGTUNG AREA, TAIWAN
Kaohsiung city is the most important harbor in Taiwan. Recently, there are many high-rise buildings and public transportation system are under construction in this area. Therefore, it is very important to know the surface geological conditions for many practical reasons especially after the strikes of 1999 Chi-Chi, Taiwan earthquake. To serve the purpose of earthquake hazard mitigation, it would be better to understand the soil amplification effect of the Kaohsiung-Pingtung area. We then conducted a research to study the site effects of the area, which includes analyze three newly installed borehole seismometer arrays, and perform very dense microtremor measurements in the study area. Most microtremor measurements were done during the midnight to reduce artifacts. After carefully selection, we pick 590 records and use the H/V ratio method to get information of soil amplification. From the result, we found it correlated to the basement depth very well. For the purpose of the earthquake resistant design, earthquake engineers must consider the site response at a specific period. For example, the structure period of a ten-flour building is at about 1 second. If the input ground motion is dominate at 1 Hz, then the building will has a resonant effect. Therefore, in this study, we select several frequencies to plot out the contour map for understanding the frequency responses in this area. For the 0.5 Hz, the contours show that main amplification effects occurred at the southern part of Kaohsiung area. With the frequency increasing to 2.0 Hz, the main amplification area move from the harbor and the southern part of Kaohsiung area to the hill area, which locates at the eastern part of Kaohsiung area. For the higher frequency (3.0 Hz), there are no obvious high contour areas. We pick the dominant frequency of each record and plot out the contour map. At the harbor and city area, the dominant frequency is about 0.5 ~ 0.9 Hz, and the northeastern part is about 1.3 ~ 1.7 Hz. We found that the basement structure can explain the contour very well. Yet, the H/V dominated frequency distribution map reveals more detail features.
S43A-0986 1320h
Elastic Nonlinear Response in Granular Media Under Resonance Conditions
We are studying the elastic linear and nonlinear behavior of granular media using dynamic wave methods. In the work presented here, our goal is to quantify the elastic nonlinear response by applying wave resonance. Resonance studies are desirable because they provide the means to easily study amplitude dependencies of elastic nonlinear behavior and thus to characterize the physical nature of the elastic nonlinearity. This work has implications for a variety of topics, in particular, the in situ nonlinear response of surface sediments. For this work we constructed an experimental cell in which high sensitivity dynamic resonance studies were conducted using granular media under controlled effective pressure. We limit our studies here to bulk modes but have the capability to employ shear waves as well. The granular media are composed of glass beads held under pressure by a piston, while applying resonance waves from transducers as both the excitation and the material probe. The container is closed with two fitted pistons and a normal load is applied to the granular sample across the top piston. Force and displacement are measured directly. Resonant frequency sweeps with frequencies corresponding to the fundamental bulk mode are applied to the longitudinal source transducer. The pore pressure in the system is 1 atm. The glass beads used in our experiments are of diameter 0.5 mm, randomly deposited in a duralumin cylinder of diameter 30 mm and height of 15 mm. This corresponds to a granular skeleton acoustic wave velocity of v | 750m/s under 50 N of force [0.07 Mpa]. The loaded system gives fundamental mode resonances in the audio frequency band at half a wavelength where resonance frequency is effective-pressure dependent. The volume fraction of glass beads thus obtained is found to be 0.63 0.01. Plane-wave generating and detecting transducers of diameter 30 mm are placed on axis at the top and bottom of the cylindrical container in direct contact with the glass beads. The wave signals are detected using a lock-in amplifier, and frequency and amplitude are recorded on computer. Drive frequency is swept from below to above the resonance mode. A typical frequency sweep is 3 kHz in width with a frequency sampling of 6 Hz. Frequency sweeps are applied at progressively increasing drive voltages to test for nonlinear-dynamical induced modulus softening. The resonance frequency at peak amplitude corresponds directly to modulus. We find significant elastic nonlinearity at all effective pressures, manifest by the fundamental-mode resonance curves decreasing progressively, at progressively increasing drive level. This is equivalent to progressive material softening with wave amplitude, meaning the wavespeed and modulus diminish. The wave dissipation simultaneously increases (Johnson and Sutin 2004). For example, at 0.11 Mpa effective pressure the observed change in resonance frequency of about 2.6% corresponds to a material bulk modulus decrease of about 5.2%. Strain amplitudes are 10-7-10-6. Thus, we would predict that surface sediments should have significant elastic nonlinear response beginning at about 10-6 strain amplitude. reference: Johnson, P. and A. Sutin, Slow dynamics in diverse solids, J. Acoust. Soc Am., in press (2004).
S43A-0987 1320h
Liquefaction of Sensitive Soils during Earthquakes
The occurrence of liquefaction during earthquakes of a given magnitude (M) is confined within a particular distance from the earthquake source, i.e., the liquefaction limit (Rmax). We interpreted Rmax as the distance at which the energy of the seismic waves has decayed to a threshold energy required for saturated soils to liquefy, beyond which the seismic energy is too weak to cause further liquefaction. Combining an empirical relation between M and Rmax with Bath's relation between M and earthquake energy (Es) we obtain an empirical relation between Es and Rmax. The relation implies a threshold energy of 0.032 J/m3 for liquefaction and a very small effective quality factor Q. The threshold energy, when interpreted in the light of a threshold shear amplitude for liquefaction, implies an effective shear modulus ~ 0.1 - 10 MPa, which is two orders of magnitude or more smaller than the elastic shear moduli for alluvial sands estimated from shear wave velocity. The atypical shear modulus and Q show that liquefaction of sensitive soils during earthquakes may be preceded by nonlinear behaviors, with drastic degradation of shear rigidity and severe dissipation of seismic energy.
S43A-0988 1320h
Saturation of Maximum Acceleration Near the Epicenter of Large Shallow Earthquakes Caused by Nonlinear Response of Surface Layers
We analyze accelerograms from large events (2000 Tottori (Mw=6.6) and 2003 Miyagi (Mw=6.1)) and small events (their foreshocks and aftershocks) recorded at KiK-net stations operated by National Institute of Earth Science and Disaster Prevention (NIED), Japan. Spectral ratio between accelerograms recorded on the ground surface and in the borehole (about 100m deep) at the same station shows amplification characteristics of surface layers. Spectral ratio (surface/borehole) for the large events at stations near the epicenter is very different from that for the small events, which is caused by the nonlinear response of surface layers. The peak frequency of the spectral ratio for the large events shifts towards the low frequency range. Also, reduction of the spectral ratio in the high frequency range is apparent for the large events. The difference between the spectral ratios for the large and small events becomes unclear at distant stations. The ratio of maximum accelerations (surface/borehole) plotted against the hypocentral distance also shows a difference between large and small events. The ratio of maximum acceleration for the large events decreases with the decrease in hypocentral distance while that for the small events is almost constant independently of the distance. The saturation of maximum acceleration near the epicenter of large earthquakes has been generally explained as the effect of finiteness of the earthquake source dimension. However, the present result suggests that the nonlinear response of surface layers is also one of the causes of saturation of maximum acceleration near the epicenter of large shallow earthquakes.
S43A-0989 1320h
A phase decomposition method of vertical array seismic data
We propose a phase decomposition technique of vertical array seismic data into P-, S-, Love and Rayleigh waves and applied to real data. Two important features of this technique are: (1) it has the ability to resolve mixture of phases, (2) it provides the waveforms which are decomposed from an array data into the phases. We assume a linear model for ground motions at a vertical seismic array. Ground motions are assumed to consists of vertical incident plane P- and S-waves, and laterally incident Love and Rayleigh waves. We assume that ground motions of a phase between a depth and surface are described by a transfer function. The decomposition is done by least-squares method in the frequency domain. Numerical tests show the ability of the decomposition of Love wave and Rayleigh wave and the limitation of the decomposition in low frequency. We analysed the data obtained by a vertical array of 5 accelerometers at CTS located on the Ashigara valley, a middle-sized sedimentary valley. Result of the analysis shows that the later waves mainly consist of surface waves. It seems that the surface waves are the basin-induced surface waves generated at the edge of the valley. Particle motions and the apparent velocity obtained with a horizontal array suppose this suggestion.
S43A-0990 1320h
A method of prediction of strong ground motion by stochastic green function considering 3-D attenuation structure
The 3-D attenuation structure is a big problem to evaluate seismic ground motion when the seismic motion that uses a statistical Green function in the great earthquake such as Tonankai earthquake generated along subduction zone. On the other hand, Q structure analysis results (ex, Nakamura and Uetake,2004) of the short period beneath the Japanese Islands show strong inhomogeneity. Therefore, We developed the strong motion evaluation technique by a statistical Green function considering three dimensions Qs structure, and examined the effect of the 3-D Q structure using 1968 Tokachi-Oki earthquake fault parameter. First, we obtained 3-D Q structure of Hokkaido district in Japan by inversion using a large quantity data of the K-NET and the KiK-net strong ground motion records. Second, we modeled fault plane and its divided elements of the fault, and calculate ground motion using 3-D Q values between observation points and the elements, and summed calculated ground motion in time domain. The result of inversion shows that low-Q area are distributed widely on the north side of volcanic front, that distribute in Hokkaido from east to west. The checkerboard resolution shows no good in this area. It is thought that earthquake is very few in northern area in Hokkaido. Nakamura and Uetake (2004) reported there are strong Low-Q at only volcanic front area in Tohoku inland. They used same data and same inversion method. So, widely distributed Low-Q area obtained in Hokkaido district may distribute in smaller area actually than its calculated. So, we could not discuss detail of Q structure. Next, we tested calculation method in this study comparing two case of using 3-D attenuation structure and homoginious structure (here, Q=100f, f : frequency). The waveform synthesis technique of stochastic green function used Kamae et al (1991) method. The source acceleration spectrum and the envelope from the elements of fault were after Boore(1987). The spectrum gives 1-10Hz. Qs effect between calculation site and each divided elements of fault plane are differently considered in the 3-D case. The result of calculation of ground motion shows that the difference of amplitude between for near field site is very small, but the difference for far field site (x : about 100km) is large. And there is difference in ground motion in case of almost same hypocentral distance sites of far field.
S43A-0991 1320h
A Semi-Empirical Approach for Generating Synthetic Seismograms for use in Matched-Waveform Source Location.
A matched field processor for seismic stations is developed to provide refined locations and source parameters. The concept is to find a synthetic seismogram with the best fit to the observed seismograms. For the seismic location problem, synthetic waveforms are computed on a three dimensional grid. The synthetic waveforms are compared to observed data at each grid point. The best fit between the synthetic waveform and the data in the grid is assumed to be the optimal source location. A limit to this approach is the computation of the synthetic waveform, i.e., errors in the earth model will propagate to errors in the synthetic waveform which appear as errors in the estimated location and source parameters. One method to mitigate this limitation is to insert observed waveforms from "Ground Truth" events into the computation of the synthetic waveforms. The semi-empirical approach is based on the time invariance of the propagation medium, i.e., the seismic waveform from a specific location from one event with a specific source mechanism recorded at a station will be identical to the waveform from a second event at the same location with the same source mechanism recorded at the same station. Typically, seismic events do not have the same source parameters. However, full synthetic waveforms for "Ground Truth" events can be computed. By deconvolving these synthetic waveforms from the observed waveforms, and then convolving with the synthetic waveforms computed for nearby locations with arbitrary source mechanisms, the semi-empirical synthetic waveforms can be generated. Errors in the synthetic waveforms are expected to cancel. The application of this technique to multiple "Ground Truth" events stabilizes the deconvolution by averaging over multiple events. In addition, spectral smoothing is used to further stabilize the deconvolution. Using this approach, semi-empirical synthetic seismograms are computed for the fundamental dislocation 3-D grid. The optimal source parameters (moment tensor) are determined by matching the observed waveforms to the semi-empirical synthetic waveforms at each grid point using an earthquake that was not in the Ground Truth set,. The optimal solution is determined to be the one with the least RMS difference. The overall approach is demonstrated with both simulated data and data from aftershocks of the Hector Mine Earthquake.
S43A-0992 1320h
Investigation of shear wave velocity structures in the western Taiwan using array measurement of microtremors
Western coastal plain in Taiwan is an important economic zone with a lot of population. This area borders on the east by the western foothills which is a fold-and-thrust zone and one of the major seismic zones in Taiwan. The earthquakes from western foothills often cause a lot of disaster in coastal plain, like Chi-Chi earthquake. We conducted array measurements of microtremors at seven sites including Lugang, Erlin, Huwei, Dongshih, Taibao, Yijhu and Jiali in western coastal plain in Taiwan. At each site, we performed four arrays with difference apertures. Based on F-K analysis we obtained the phase velocity dispersion curves. After inversion of dispersion curves, the shear wave velocity structures under all sites were estimated. We could roughly describe the variation of shallow shear wave velocity structures in the western coastal plain of Taiwan. According to the comparison between our results and previous studies, the inversion of dispersion curves in the frequency range from 0.1 to 5 Hz can completely estimate shear wave velocity structures down to the depth of 3 km. The shear wave velocity structures of all sites which we desired correctly reflect the velocities and depths of main interfaces including the tops of Pliocene and Miocene. Peikang high apparently dominates the S-wave velocity structure in western Taiwan. Therefore, the shear wave velocity of Dongshih in the depth about 1 km is higher than that of other sites. And the depths of major interfaces are shallowest near Peikang high and dip to north, east and south. Besides, the variation of S-wave velocity at Yijhu is similar to Dongshih, because the Yijhu hinge fault which is the southern boundary of Peikang high cross this area. According to residual gravities and seismic surveys, there are a lot of complex structures including normal faults and anticlines.
S43A-0993 1320h
STUDY ON SHALLOW VELOCITY STRUCTURE MODELING OF SEDIMENTS FOR STRONG-MOTION EVALUATION
The National Research Institute for Earth Science and Disaster Prevention (NIED) has carried on the special research project National Seismic Hazard Mapping Project of Japan to support the preparation of the seismic hazard map in general view of the whole Japan, which is made by the Headquarters for Earthquake Research Promotion. We have studied that the method of strong-motion evaluation for the scenario earthquakes in any fault zone. However, these strong-motion evaluations are deeper than engineering bedrock. In order to evaluate amplifications of strong-motion in surface soils, we use the empirical formulas according to Matsuoka and Midorikawa (1994) and Fujimoto and Midorikawa (2003). We also adopt a more precise method in which we use the surface soil(shallow) velocity structure models made from many boring profiles and geological data. These models were created every 250m meshes. Shallow velocity structure is shallower than engineering bedrock. These models are creating about Kinki, Hokuriku and Kanto area. Moreover, these mesh size of models were changed and the variation in the creation result of models were examined. In this study, strong-motion evaluation analysis was created in the Hokuriku area. The results of Fujimoto and Midorikawa (2003) show higher correlation with those of the precise method using the shallow velocity structure models than the results of Matsuoka and Midorikawa (1994). This suggests that the tuning of models to evaluate site amplifications using proper boring profiles and geological data is indispensable. Toward more accurate strong-motion evaluations, it is necessary to examine the physical properties of soils from the engineering bedrock up to ground surface, and to perform a precise modeling in which the nonlinear effects of material are took into account.