Search

Peaks in horizontal-to-vertical spectral ratios (HVSRs) of Fourier amplitudes from three-component recordings of ground vibrations without undue local anthropogenic influence are used to identify site resonances, which are an important component of site response. We address two topics: (1) how should HVSR peaks be identified and (2) are there appreciable differences in HVSR derived using different instruments recording microtremors and seismic strong ground motions? We propose identifying peaks by considering peak amplitudes relative to neighboring ordinates and peak width. The procedure incorporates a regression tree algorithm that can be tuned to conform with user preferences toward relatively "conservative" or "liberal" peak identification (producing few or many sites with peaks, respectively). We then investigate the consistency of microtremor-based HVSRs (mHVSRs) derived from seismometers and accelerometers, which show a high rate of false negatives (missed peaks) from accelerometers with a full scale of ± 2g or greater. In contrast, mHVSRs derived from collocated temporary and permanent seismometers (optimized to record teleseismic signals) have about 60%-80% consistency (with no apparent bias in peak attributes). This indicates that mHVSRs from accelerometers having a broad full scale are unreliable but that mHVSRs can be reliably obtained from temporary or permanent seismometers. Finally, we compare seismometer-based HVSR from microtremor and earthquake (eHVSRs) sources. Results are consistent for 60%-70% of sites (i.e., both either do or do not have significant peaks, and when peaks are present, they occur at similar frequencies, <20% change). For sites with an mHVSR peak, the rate of corresponding eHVSR peaks is nearly 50%, whereas for sites without an mHVSR peak the eHVSR peak rate is low (about 20%). The mismatch rate for mHVSR peak sites is sufficiently high that the use of eHVSR to derive site response models is likely too optimistic (overestimates model effectiveness); mHVSR is preferred for consistency with information available in forward applications. [ABSTRACT FROM AUTHOR]

This study uses an ensemble Kalman method for near-surface seismic site characterization of 154 network earthquake monitoring stations in California to improve the resolution of S -wave velocity (VS) and P -wave velocity (VP) profiles—up to the resolution depth—coupled with better quantification of uncertainties compared to previous site characterization studies at this network. These stations were part of the Yong et al. site characterization project, with selected stations based on future recordings of ground motions that are expected to exceed 10 per cent peak ground acceleration in 50 yr. To estimate VS and VP from experimental dispersion data, Yong et al. investigated these stations using linearized (local search and iteration) routines, and Yong et al. later studied a subset of these stations using nonlinear (global search and optimization) routines. In both studies, the selection of model parameters—that is, discretization of the VS and VP profiles with only five fixed thickness layers—was mainly based on trial and error. In contrast, this paper uses an approximate Bayesian method to assimilate experimental dispersion data and sequentially update an ensemble of particle estimates that span the VS and VP parameter spaces. Doing so, we systematically determine the most probable profiles conditioned on the experimental dispersion data, the introduced noise levels, and a priori knowledge in the form of physical constraints. We consider two configurations to discretize the soil depth from the surface to half of the maximum discernible wavelength obtained from the experimental dispersion data, namely refined and coarse models, and two initial models for each configuration to study solution multiplicity. Our results suggest that using the refined model for the top surface layers improves the resolution of near-surface site characteristics and the model's success rate in capturing dispersion data at high frequencies. All models result in similar VS but distinct VP profiles, with increasing uncertainty at deeper layers, suggesting that the fundamental mode of Rayleigh wave dispersion data is not adequate to constrain the P -wave velocity profile and the S -wave velocity close to the resolution depth. [ABSTRACT FROM AUTHOR]

Site response is a critical consideration when assessing earthquake hazards. Site characterization is key to understanding site effects as influenced by seismic site conditions of the local geology. Thus, a number of geophysical site characterization methods were developed to meet the demand for accurate and cost-effective results. As a consequence, a number of studies have been administered periodically as blind trials to evaluate the state-of-practice on-site characterization. We present results from the Consortium of Organizations for Strong Motion Observation Systems (COSMOS) blind trials, which used data recorded from surface-based microtremor array methods (MAM) at four sites where geomorphic conditions vary from deep alluvial basins to an alpine valley. Thirty-four invited analysts participated. Data were incrementally released to 17 available analysts who participated in all four phases: (1) two-station arrays, (2) sparse triangular arrays, (3) complex nested triangular or circular arrays, and (4) all available geological control site information including drill hole data. Another set of 17 analysts provided results from two sites and two phases only. Although data from one site consisted of recordings from three-component sensors, the other three sites consisted of data recorded only by vertical-component sensors. The sites cover a range of noise source distributions, ranging from one site with a highly directional microtremor wave field to others with omni-directional (azimuthally distributed) wave fields. We review results from different processing techniques (e.g., beam-forming, spatial autocorrelation, cross-correlation, or seismic interferometry) applied by the analysts and compare the effectiveness between the differing wave field distributions. We define the M index as a quality index based on estimates of the time-averaged shear-wave velocity of the upper 10 (VS10), 30 (VS30), 100 (VS100), and 300 (VS300) meters and show its usefulness in quantitative comparisons of VS profiles from multiple analysts. Our findings are expected to aid in building an evidence-based consensus on preferred cost-effective arrays and processing methodology for future studies of seismic site effects. [ABSTRACT FROM AUTHOR]

Seismic attenuation and the associated quality factor (Q) have long been studied in various sub-disciplines of seismology, ranging from observational and engineering seismology to near-surface geophysics and soil/rock dynamics with particular emphasis on geotechnical earthquake engineering and engineering seismology. Within the broader framework of seismic site characterization, various experimental techniques have been adopted over the years to measure the near-surface shear-wave quality factor (QS). Common methods include active- and passive-source recording techniques performed at the free surface of soil deposits and within boreholes, as well as laboratory tests. This paper intends to provide an in-depth review of what Q is and, in particular, how QS is estimated in the current practice. After motivating the importance of this parameter in seismology, we proceed by recalling various theoretical definitions of Q and its measurement through laboratory tests, considering various deformation modes, most notably QP and QS. We next provide a review of the literature on QS estimation methods that use data from surface and borehole sensor recordings. We distinguish between active- and passive-source approaches, along with their pros and cons, as well as the state-of-the-practice and state-of-the-art. Finally, we summarize the phenomena associated with the high-frequency shear-wave attenuation factor (kappa) and its relation to Q, as well as other lesser-known attenuation parameters. [ABSTRACT FROM AUTHOR]

The time-averaged shear-wave velocity in the upper 30 m depth from the ground surface, or VS30, is often used as a predictor to describe local site effects in ground-motion models. Although VS30 is typically determined from in situ measurements, it is not always feasible to obtain such measurements due to project restrictions or site accessibility. This motivates the development and use of proxy-based VS30 predictions that leverage more readily available secondary information such as surface geology, topographic slope, or geomorphic terrain classes to estimate the mean VS30 and associated uncertainty. Traditionally, empirical distributions of VS30 have been observed to have long right tails, leading to high levels of associated uncertainty. In this study, we present a physical framework that is grounded in fundamental principles of geostatistics and probability to explain the uncertainty and skewness associated with VS30 measurements. Specifically, by invoking Lyapunov's central limit theorem, we hypothesize that the distribution of VS30 can be theoretically approximated by a reciprocal-normal distribution. We show that a non-normal and skewed distribution of VS30 is to be expected and is not a sign of measurement error or sampling bias, although sampling bias can exaggerate such skewness. Our framework also enables us to propose the mode as a characteristic value of VS30 measurements, as opposed to the mean or median, which can overestimate the most probable value. [ABSTRACT FROM AUTHOR]

One of the biggest challenges in computer vision and deep learning is recognizing the emotion based on facial expression. Besides the challenge of model development in Deep Learning (DL), the dataset is considered one of the main factors that plays a crucial role in producing highly accurate deep learning frameworks for pattern classification, features extraction, and emotion recognition from images. Despite the existence of several dedicated datasets for the FER, the desired accuracy was not being achieved due to the enormous size of the available dataset, where some of such are unfeasible for regular computers or cloud computing services. In contrast, there is a lack of the other datasets' size, generalization, and quality. The objectives of this paper are two-fold. First, reviews the existing datasets. Second, presents a significant new dataset for facial expression called Emot-FE. This dataset contains 276,305 images of facial expressions formulated in a single file in the form of xlsx format. Emot-FE dataset has been filtered, pre-processed, labeled, and classified based on the seven labels (emotion's expressions). Thus, this dataset will be the largest in such format with high-quality images. This new dataset has been evaluated using VGGNet Convolutional Neural Networks where the recognition accuracy outperforms the previously achieved accuracies of the previous studies. This dataset is available for distribution to the researchers. [ABSTRACT FROM AUTHOR]

New seismic stations added to a regional seismic network cannot be used to calculate local magnitude (⁠ML⁠) until a revised regionwide amplitude decay function is developed. Each station must record a minimum number of local and regional earthquakes that meet specific amplitude requirements prior to recalibration of the amplitude decay function. Station component adjustments (⁠dML⁠; Uhrhammer et al., 2011) are then calculated after inverting for a new regional amplitude decay function, constrained by the sum of dML for long‐running stations. Therefore, there can be significant delay between when a new station starts contributing real‐time waveform packets and when data can be included in magnitude determinations. We propose the use of known estimates of seismic site conditions such as the time‐averaged shear‐wave velocity (⁠VS⁠) of the upper 30 m (⁠VS30⁠) and the site dominant frequency (⁠fd⁠) to calculate dML⁠. Previously established dML⁠, measured VS30⁠, and fd data are available for between 126 and 458 horizontal components (east-west and north-south) at 81 seismic stations in the California Integrated Seismic Network; dML data range from -1.10 to 0.39, VS30 values range from 202 to 1464  m/s⁠, and 440 fd values are compiled from earthquake and microseismic records that range from 0.13 to 21 Hz. We find VS30 and dML exhibit a positive coefficient of determination (⁠R=0.59⁠), indicating that as VS30 increases, dML increases. This implies that greater site amplification (lower VS30⁠) results in smaller dML⁠. fd and dML also generally exhibit a positive correlation (⁠R²<0.56⁠), which implies lower dML values are related to site resonance at depth‐dependent frequencies. Using the developed relationships, VS30 or fd measurements can be used to establish a provisional dML for newly established stations. This procedure allows new stations to contribute to regional network ML determinations immediately without the need to wait until a minimum set of earthquake data has been recorded. [ABSTRACT FROM AUTHOR]

In engineering seismology, the time-averaged shear-wave velocity (VS) of the upper 30 m of the crust (VS30) is the primary parameter used in ground-motion models to predict seismic site effects. VS30 is typically derived from in situ recordings of VS, although proxy-based approaches (using geologic and/or geomorphometric classifications) are provisionally adopted when measurement-based VS30 are sparse or not readily available. Despite the acceptance of proxy approaches, there are limited studies that examine the empirical relationships between VS30 and topographic attributes measured from various length scales and different resolutions of the digital elevation model. In this study, we examine the relationships between compiled VS30 measurements from 218 sites in southern California and topographic metrics of slope and relief measured over various length scales. We find that the correlations between topographic metrics and VS30 are weak but statistically significant. The correlations are improved when topographic slopes and relief are measured over length scales longer than typical hillslopes and VS30 sites are separated by different geologic groups. This is likely because VS30, especially on the rock sites, is better reflected in topographic metrics that capture large-scale topographic relief, as well as landscape positions such as hilltops and valley bottoms. [ABSTRACT FROM AUTHOR]

An important predictive variable for site amplification is the site dominant frequency (fd). At seismic monitoring stations, fd can be calculated from the peak of the horizontal-to-vertical spectral ratios (HVSRs) obtained from earthquake recordings (eHVSR). For other sites, fd can be estimated from microseismic (mHVSR) observations. We compare the fd values derived from eHVSR (5% damped response spectra from the Next Generation Attenuation-West2 [NGA-West2] database; Ancheta et al., 2014) with those derived from mHVSR (Fourier spectra from Yong et al., 2013) for seismic stations in California. We show that the logarithm of eHVSR fd scales linearly with the logarithm of mHVSR fd, with a standard deviation of 0.14log10 units for mHVSR fd larger than 0.2 Hz. The relationship holds for microseismic surveys at distances up to 300 m away from the seismic stations. The results of this study have beneficial implications for the characterization of site response in modern ground-motion models as well as in building codes. [ABSTRACT FROM AUTHOR]

We present an approach based on geomorphometry to predict material properties and characterize site conditions using the VS30 parameter (time-averaged shear-wave velocity to a depth of 30 m). Our framework consists of an automated terrain classification scheme based on taxonomic criteria (slope gradient, local convexity, and surface texture) that systematically identifies 16 terrain types from 1-km spatial resolution (30 arcsec) Shuttle Radar Topography Mission digital elevation models (SRTM DEM5). Using 853 V~30 values from California, we apply a simulation-based statistical method to determine the mean VS30 for each terrain type in California. We then compare the VS30 values with models based on individual proxies, such as mapped surface geology and topographic slope, and show that our systematic terrain-based approach consistently performs better than semiempirical estimates based on individual proxies. To further evaluate our model, we apply our California-based estimates to terrains of the contiguous United States. Comparisons of our estimates with 325 VS30 measurements outside of California, as well as estimates based on the topographic slope model, indicate our method to be statistically robust and more accurate. Our approach thus provides an objective and robust method for extending estimates of V~30 for regions where in situ measurements are sparse or not readily available. [ABSTRACT FROM AUTHOR]

Waveform analysis of aftershocks of the Mw7.0 Haiti earthquake of 12 January 2010 reveals amplification of ground motions at sites within the Cul de Sac valley in which Port-au-Prince is situated. Relative to ground motions recorded at a hard-rock reference site, peak acceleration values are amplified by a factor of approximately 1.8 at sites on low-lying Mio-Pliocene deposits in central Port-au-Prince and by a factor of approximately 2.5–3 on a steep foothill ridge in the southern Port-au-Prince metropolitan region. The observed amplitude, predominant periods, variability, and polarization of amplification are consistent with predicted topographic amplification by a steep, narrow ridge. A swath of unusually high damage in this region corresponds with the extent of the ridge where high weak-motion amplifications are observed. We use ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) imagery to map local geomorphology, including characterization of both near-surface and of small-scale topographic structures that correspond to zones of inferred amplification. [ABSTRACT FROM AUTHOR]

We report about a preliminary study to evaluate the use of semi-automated imaging analysis of remotely-sensed DEM and field geophysical measurements to develop a seismic-zonation map of Port-au-Prince, Haiti. For in situ data, Vs30 values are derived from the MASW technique deployed in and around the city. For satellite imagery, we use an ASTER GDEM of Hispaniola. We apply both pixel- and object-based imaging methods on the ASTER GDEM to explore local topography (absolute elevation values) and classify terrain types such as mountains, alluvial fans and basins/near-shore regions. We assign NEHRP seismic site class ranges based on available Vs30 values. A comparison of results from imagery-based methods to results from traditional geologic-based approaches reveals good overall correspondence. We conclude that image analysis of RS data provides reliable first-order site characterization results in the absence of local data and can be useful to refine detailed site maps with sparse local data. [ABSTRACT FROM AUTHOR]

We develop an integrated digital imaging analysis approach to produce a first-approximation site characterization map for Islamabad, Pakistan, based on remote-sensing data. We apply both pixel-based and object-oriented digital imaging analysis methods to characterize detailed (1:50, 000) geomorphology and geology from Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) satellite imagery. We use stereo-correlated relative digital elevation models (rDEMs) derived from ASTER data, as well as spectra in the visible near-infrared (VNIR) to thermal infrared (TIR) domains. The resulting geomorphic units in the study area are classified as mountain (including the Margala Hills and the Khairi Murat Ridge), piedmont, and basin terrain units. The local geologic units are classified as limestone in the Margala Hills and the Khairi Murat Ridge and sandstone rock types for the piedmonts and basins. Shear-wave velocities for these units are assigned in ranges based on established correlations in California. These ranges include Vs30-values to be greater than 500 m/sec for mountain units, 200-600 m/sec for piedmont units, and less than 300 m/sec for basin units. While the resulting map provides the basis for incorporating site response in an assessment of seismic hazard for Islamabad, it also demonstrates the potential use of remote-sensing data for site characterization in regions where only limited conventional mapping has been done. [ABSTRACT FROM AUTHOR]