S42A-01 10:25h
Topographic Slope as a Proxy for Seismic Site Amplification Correction
As part of an effort to rapidly predict ground shaking and earthquake impact globally (Prompt Assessment of Global Earthquakes for Response, U.S. Geological Survey Project PAGER, also see Earle et al. paper, this meeting), we need at least a first order approximation for seismic site conditions worldwide for input into our ground motion estimation. In many seismically-active regions of the world, information about shallow geology and shear velocity does not exist, varies dramatically in quality, or is not easily accessible. Topographic elevation data, on the other hand, are available at uniform sampling globally, and here we attempt to derive site condition maps directly from this highly accurate and readily available data. For calibration, we use the California statewide topographic data and geologically-based shallow (30 m) site condition map of Wills et al (2000). The Wills et al. map is categorized by NEHRP lettering with grades at intermediate values ranging from hard rock to bay mud (i.e., B, BC, C, CD, D, DE, E). In short, we find that the slope of the topography is well correlated with the site condition map. By taking the gradient of the topography and choosing the range of slope that maximize the correlation with the existing site condition map we can recover, to first order, many of the important aspects of the map. Additionally, we assign class E (bay mud) to all flat regions with elevations between negative 3 and plus 3 meters. This assumption produces reasonable class E boundaries and avoids incorrectly assigning class E to large flat areas such as the San Joaquin Valley and to areas below sea level such as Death Valley. The largest discrepancy in geologically- and topographically-derived site conditions is between soft and hard rock. Fortunately, the differences in site amplification for these types are relatively small. Naturally, this simple assumption will break down for some topgraphic/geomorphic combinations. As an example, in glacial terrains, we cannot distinguish between topographically similar depositional (glacial till) drumlins and erosional (bedrock) roche moutonnees. Nonetheless, this approach should provide adequate first-order estimates of regional site amplification for the entire globe.
S42A-02 10:40h
Uncertainties in Site Amplification Estimation
Typically geophysical profiles (layer thickness, velocity, density, Q) and dynamic soil properties (modulus and damping versus strain curves) are used with appropriate input ground motions in a soil response computer code to estimate site amplification. Uncertainties in observations can be used to generate a distribution of possible site amplifications. The biggest sources of uncertainty in site amplifications estimates are the uncertainties in (1) input ground motions, (2) shear-wave velocities (Vs), (3) dynamic soil properties, (4) soil response code used, and (5) dynamic pore pressure effects. A study of site amplification was conducted for the 1 km thick Mississippi embayment sediments beneath Memphis, Tennessee (see USGS OFR 04-1294 on the web). In this study, the first three sources of uncertainty resulted in a combined coefficient of variation of 10 to 60 percent. The choice of soil response computer program can lead to uncertainties in median estimates of +/- 50 percent. Dynamic pore pressure effects due to the passing of seismic waves in saturated soft sediments are normally not considered in site-amplification studies and can contribute further large uncertainties in site amplification estimates. The effects may range from dilatancy and high-frequency amplification (such as observed at some sites during the 1993 Kushiro-Oki, Japan and 2001 Nisqually, Washington earthquakes) or general soil failure and deamplification of ground motions (such as observed at Treasure Island during the 1989 Loma Prieta, California earthquake). Examples of two case studies using geotechnical data for downhole arrays in Kushiro, Japan and the Wildlife Refuge, California using one dynamic code, NOAH, will be presented as examples of modeling uncertainties associated with these effects. Additionally, an example of inversion for estimates of in-situ dilatancy-related geotechnical modeling parameters will be presented for the Kushiro, Japan site.
S42A-03 INVITED 10:55h
Three-dimensional Liquefaction Susceptibility Using Geostatistics
We have developed a new three-dimensional (3D) geostatistical analysis method to evaluate liquefaction susceptibility. Using 3D allows us to identify continuous volumes of liquefiable soil at depth that can lead to surface manifestations of liquefaction and is an improvement to more traditional one-dimensional and two-dimensional methods. The geostatistical analysis provides an estimate of spatial variability and estimate uncertainty as well as a method for interpolation beyond known values. The resulting interpolation inherently includes a measure of prediction uncertainty. The 3D geostatistical analysis of liquefaction susceptibility accounts for spatial variability at the site and therefore provides a method for informed extrapolation from known values. In order to map the liquefaction hazard using this 3D method, we need to evaluate how much soil at a site must be classified as liquefiable in order to cause liquefaction-induced ground failure. Knowing the volume of theoretically liquefiable soil at a site is not directly useful unless we understand how much soil is needed to cause ground failure. To determine this threshold value, we studied data from several sites in California that were exposed to the October 17, 1989, Loma Prieta Earthquake or the January 17, 1994, Northridge, California Earthquake. The investigated sites included detailed subsurface investigations as well as post-earthquake surveys. The goal of this study was to determine how much theoretically liquefiable soil is necessary to induce a ground failure. To achieve this goal, indicator values were assigned to soil samples. Using the indicator value data and geostatistical interpolation, we created solid three dimensional models for a given probability of liquefaction at each site. We calculated the volume of soil at each site associated with a range of probability levels. We determined the relationship between the percentage of theoretically liquefiable soil at the site and the probability of liquefaction for each site.
S42A-04 11:15h
Nonlinear Soil Response Induced in Situ by an Active Source at Garner Valley
As part of the nonlinear soil studies at Garner Valley, on August 18 we performed a sequence of exploratory experiments using the NEES's vibrator source T-Rex, in order to determine whether or not elastic nonlinear response could be induced and measured. The experiments described here are standing wave (resonance) measurements. This type of experimental approach, known as Nonlinear Resonant Ultrasound Spectroscopy, is commonly applied in the laboratory to characterize sample nonlinear response (Guyer and Johnson, 1999). The method relies on monitoring a resonance mode frequency change as a function of drive level. In the field experiment, the TRex source is step-swept in frequency encompassing the modal frequencies expected theoretically for the stratigraphy at the GVDA, measured independently 10 m distant from the experimental site (Bonilla et al., 2004). We focus on several interfaces determined from that study, and therefore selected a frequency band of 5-30 Hz for the step-sweep. An array of nine 3-component accelerometers located immediately adjacent to the vibration plate recorded motions generated by six progressively larger forces, ranging from 3000-21000 lbs in shear and 6000-54000 lbs in compression. We observed clear modes in both compression and shear. Measured accelerations reached 2G (the instrument clip-level) at full force, corresponding to strain amplitudes on the order of 4.5x10-3. The detected acceleration amplitudes span more than a decade, implying minimum strains are on the order of 10-4 to 10-5. Laboratory experiments on granular material suggest these strains should be sufficient to induce a strong nonlinear response (Guyer and Johnson, 1999). The raw data in both compression and shear show a clear, progressive shift in frequency downward as a function of driving force. References Guyer, R. and P. A. Johnson, The astonishing case of mesoscopic elastic nonlinearity, Physics Today, 52, 30-35, (1999). Bonilla, F., J. Steidl, J.-C. Gariel and R. Archuleta, Borehole response studies at the Garner Valley Downhole Array, southern California, Bull. Seism. Soc. Am. 92, 3165-3179, (2002).
S42A-05 INVITED 11:30h
Shear-Wave Velocities to 100 m Depth From Seismic Reflection/Refraction, ReMi, and MASW Techniques Compared With Borehole Velocities: Implications for Earthquake Ground Motion Assessment
Shallow shear-wave velocity (V$_{s}$) has long been recognized as a key factor in variable ground motion amplification and site response in sedimentary basins. Seismic hazard maps traditionally have not dealt with local geologic site effects in detail; however, hazards-mapping methodology is advancing to more accurately incorporate local V$_{s}$ information into the hazards calculation, particularly in urbanized areas. This trend is expected to accelerate with future expansion of these efforts. Incorporation of scenario earthquakes into future hazards characterization will also depend on reliable V$_{s}$ in the upper few hundred meters. As such, the need to rapidly and inexpensively determine shallow V$_{s}$ over large urban sedimentary basins will become critical to accurately represent site response in future urban hazard maps. Borehole logging is generally considered the standard for obtaining V$_{s}$ data, but drilling and logging to the depths generally required for earthquake ground motion investigations is very expensive, and it is becoming increasingly problematic in heavily urbanized settings. Traditional active-source seismic reflection/refraction has been used extensively for V$_{s}$ characterization. MASW (Multi-channel analysis of surface waves) and ReMi (refraction microtremor) are two of the most recently-developed surface techniques for determining shallow V$_{s}$. We conducted a blind experiment to determine V$_{s}$ depth profiles at four boreholes in the Santa Clara Valley, CA, with these three methods. Average V$_{s}$ estimates at 30, 50 and 100 m depth demonstrate that the surface methods as implemented in this study can generally match borehole results to within 15% at these depths. Spectral amplifications predicted from the respective borehole velocity profiles similarly compare to within 15% or better from 1 to 10 Hz with the surface-method velocity profiles. Overall, no one method was consistently better at matching the borehole velocity profiles or amplifications. These results suggest reflection/refraction, MASW, and ReMi surface acquisition methods are all appropriate for estimating V$_{s}$ in urban settings for hazards assessment.
S42A-06 11:50h
Debunking Urban Legends about Seismic Velocities in Urban Basins Using Oil Industry Sonic and Density Logs
I use compressional-wave (sonic) and density logs from 130 oil industry boreholes to debunk several simplistic myths about velocity-depth functions commonly used in site response estimates and computer modeling of seismic waves in basins. The logs are being used to help construct a new 3-D velocity model for the greater San Francisco Bay Area, and to test tomography and gravity models of the region. The logs sample strata to an average depth of 2 km, some as deep as 5.6 km, in several urban sedimentary basins, mainly in northern California. The logs sample primarily Pliocene to Upper Cretaceous clastic sedimentary rocks including shales, sandstones, and conglomerates: higher-velocity chemical precipitates such as carbonates, dolomites, anhydrites, and halites are notably missing. Myth #1: Velocity increases monotonically with depth. Many logs show abrupt, pronounced, and thick low velocity zones often associated with an abrupt downward reduction of the grain size of the sedimentary strata. These low velocity zones often record regional uplift in which coarse grained and higher velocity strata are deposited above finer grained strata. These regional low velocity zones, with thicknesses of several hundred m, may represent an efficient wave guide for seismic wave propagation. Myth #2: There are only small regional variations in the seismic velocities of the basin strata. In the San Francisco Bay Area, logs from wells extending from the Great Valley to offshore basins show a pronounced westward increase in sonic velocity associated with relatively high velocity Miocene strata. Myth #3: The impedance contrast at the bottom of basins is depth independent. Most wells penetrating basement at depths less than 1.5 km around basin edges exhibit a large (30%) impedance contrast. In the Bay Area, wells penetrating basement at depths greater than 3 km in the middle of basins show only a modest impedance contrast. Thus, the logs predict pronounced basin edge effects and more modest conversions in the geographic center of the basins. Myth #4: Strata of the same age have similar velocities. Miocene strata west of the Calaveras fault have much higher seismic velocities than those east of the fault. Eocene strata in the Great Valley have lower seismic velocities than those in the San Francisco peninsula. Myth #5: Velocities systematically increase with the age of the strata. In the Bay Area, coarser-grained Miocene strata west of the Calaveras fault have higher velocities than finer-grained Upper Cretaceous strata in the Great Valley. Myth #6: Vertical velocity gradients always decrease with depth. Numerous logs from offshore basins on the continental shelf and in the Salinas Valley exhibit a velocity gradient that increases with depth within Miocene age strata. Myth #7: Velocities in the basins are low. In the Bay Area, logs from several basins show sonic velocities up to 4.7 km/s at depths between 2 to 3 km, even for Miocene strata, that are comparable to velocities in the underlying Franciscan complex basement. Myth #8: Sonic velocity-density relations are regionally uniform. Although well logs demonstrate that on average the sonic velocity and density obeys Gardner's rule, strata in Bay Area basins have 4-5% lower densities than predicted by this rule.
S42A-07 12:05h
Using Borehole Vertical Array Data to Determine Local Attenuation and Velocity Structure: A Combined Global-Local Optimization Algorithm for Plane Wave Seismogram Inversion
A seismic waveform inversion algorithm is demonstrated for the estimation of elastic soil properties from one-dimensional downhole array recordings. For a given bedrock motion, scarcity of near-surface geotechnical information, error propagation and limited resolution of the continuum usually result in predictions of surface ground motion that poorly compare with low amplitude observations. This discrepancy is further aggravated for strong ground motion, associated with hysteretic, nonlinear, and potentially irreversible material deformations. Seismogram inversion is a nonlinear multi-parameter optimization problem. Traditional search techniques that use characteristics of the problem to determine the next sampling point (e.g. gradients, Hessians, linearity and continuity) are computationally efficient, yet limited to convex regular functions. As a result, they fail to identify the best fit solution in seismogram inversion problems, when the starting model is too far from the global optimal solution. On the other hand, stochastic search techniques (e.g. genetic algorithms, simulated annealing) have been shown to efficiently identify promising regions in the search space, but perform very poorly in a localized search. The proposed inversion technique is a two-step process, namely a genetic algorithm in the wavelet domain in series with a nonlinear least-square fit in the frequency domain; we thus improve the computational efficiency of the former, while avoiding the pitfalls of using local linearization techniques such as the latter for the optimization of multi-modal, discontinuous and non-differentiable functions. The parameters to be estimated are stepwise variations of the shear modulus, attenuation and density with depth, for horizontally layered media with refined near-surface discretization. Equality constrains are imposed on the vector of unknowns to bound the search space, based on the available soil investigation. For the genetic algorithm, the objective function is defined as the normalized cross-correlation between the observed data and the synthetics. We perform the inversion in the wavelet domain to allow for equal weighting of the information across all frequency bands. Since ground motion is non-stationary in time and frequency, a time-domain inversion would inevitably emphasize the larger amplitude signals. The process is repeated in series for a subset of the available borehole and surface waveform pairs, selected on the basis of signal quality. The mean estimated soil properties from the genetic algorithm are then used as a starting model for the local minimization scheme. The target function in this stage is the empirical transfer function in the frequency domain, estimated using the average spectral ratio between surface and borehole pairs. The global-local inversion technique can efficiently identify the optimal solution vicinity in the search space by means of the hybrid genetic algorithm, whereas the use of nonlinear least-square fit accelerates substantially the detection of the best fit model. The algorithm has been implemented in MATLAB, and inversion results are illustrated for stations in the Japanese strong motion borehole array Kik-Net, as well as for borehole stations in Southern California jointly operated by the California Integrated Seismic Network, the Southern California Earthquake Center, and the University of California at Santa Barbara.