The NGA-West2 project database expands on its predecessor to include worldwide ground motion data recorded from shallow crustal earthquakes in active tectonic regimes post-2000 and a set of small-to-moderate-magnitude earthquakes in California between 1998 and 2011. The database includes 21,336 (mostly) three-component records from 599 events. The parameter space covered by the database is M 3.0 to M 7.9, closest distance of 0.05 to 1,533 km, and site time-averaged shear-wave velocity in the top 30 m of V S30 = 94 m/s to 2,100 m/s (although data becomes sparse for distances >400 km and V S30 > 1,200 m/s or <150 m/s). The database includes uniformly processed time series and response spectral ordinates for 111 periods ranging from 0.01 s to 20 s at 11 damping ratios. Ground motions and metadata for source, path, and site conditions were subject to quality checks by ground motion prediction equation developers and topical working groups.

The NGA-West2 project is a large multidisciplinary, multi-year research program on the Next Generation Attenuation (NGA) models for shallow crustal earthquakes in active tectonic regions. The research project has been coordinated by the Pacific Earthquake Engineering Research Center (PEER), with extensive technical interactions among many individuals and organizations. NGA-West2 addresses several key issues in ground-motion seismic hazard, including updating the NGA database for a magnitude range of 3.0–7.9; updating NGA ground-motion prediction equations (GMPEs) for the “average” horizontal component; scaling response spectra for damping values other than 5%; quantifying the effects of directivity and directionality for horizontal ground motion; resolving discrepancies between the NGA and the National Earthquake Hazards Reduction Program (NEHRP) site amplification factors; analysis of epistemic uncertainty for NGA GMPEs; and developing GMPEs for vertical ground motion. This paper presents an overview of the NGA-West2 research program and its subprojects.

No abstract

No abstract

Suites of earthquake ground motions play an important role in the seismic design and analysis process. A semi-automated procedure is described that selects and scales ground motions to fit a target acceleration response spectrum, while at the same time the procedure controls the variability within the ground motion suite. The basic methodology selects motions based on matching the target spectral shape, and then fits the amplitude and standard deviation of the target by adjusting the individual scale factors for the motions. The selection of motions from a larger catalog of motions is performed through either a rigorous method that tries each possible suite of motions or an iterative approach that considers a smaller set of potential suites in an effort to find suites that provide an acceptable fit to the target spectrum. Guidelines are provided regarding the application of the developed procedures, and example applications are described.

The time-averaged shear ( S) wave velocity in the upper 30 meters of sediment ( V S30) is a widely used site parameter for ground motion prediction. When unavailable from measurements, as is often the case at accelerograph stations in Central and Eastern North America (CENA), V S30 is typically estimated from proxies. We propose an alternative for CENA based on a theoretical relationship between S-wave velocity and the ratio of radial to vertical components of the compressional ( P)-wave–dominated portion of the velocity time series. This method is applied to 31 CENA accelerograph sites having measured S-wave velocity profiles. Time-averaged S-wave velocities to depth z ( V SZ) from the proposed method agree well with those from measurements. We develop linear relationships between V SZ and V S30 using CENA S-wave velocity profile data. Values of V S30 established from the proposed method (including depth extrapolation) have lower dispersion relative to data ( σln V = 0.43) than do estimates from available CENA proxies.

A method for deriving kappa (κ) scaling factors that can be applied to ground-motion prediction equations (GMPEs) to account for site-specific κ estimates is described. This method relies on inverse random vibration theory as implemented in the computer program Strata (Kottke and Rathje, 2008a,b) to derive a Fourier amplitude spectrum (FAS) that is consistent with the response spectrum from the GMPE. The GMPE host κ values are estimated by fitting the high-frequency FAS with the Anderson and Hough (1984) κ scaling function. The derived FAS are then scaled from their host κ value to a target κ. Random vibration theory (Cartwright and LonguetHiggins, 1956) is then used to convert the κ scaled FAS to response spectra, and κ scaling factors are computed by the ratio of the κ scaled response spectra to the GMPE response spectra. In contrast to the commonly used hybrid empirical method (Campbell, 2003), the proposed approach does not require a full seismological model for the stochastic parameters (stress drop, whole-path attenuation, etc.) of the host and target regions and does not assume that response spectral shape of GMPE is consistent with that of the representative point-source stochastic model for the host region, which can lead to inappropriate response spectral scaling factors. Finally, the effects of the wellknown trade-off between κ and stress drop scaling are reduced. The method, when applied within the frequency limitations discussed in this paper, can be used to incorporate κ scaling into GMPEs.

The reference rock site condition has two important applications for ground-motion prediction in the stable continental region of central and eastern North America (CENA). (1) It represents the site condition for which ground motions are computed using semiempirical ground-motion prediction equations. In addition, (2) it represents the site condition to which site amplification factors, which are used to modify ground-motion intensity measures for softer site condition, are referenced (i.e., site amplification is unity for reference rock). We define reference rock by its shear (S)-and compression (P)-wave velocities, as well as a site attenuation parameter (κ 0 ), which is used in stochastic ground-motion simulation methods. Prior definitions of reference rock conditions in CENA were based mostly on indirect large-scale crustal velocity inversions and judgment. We compile and interpret a unique database of direct velocity measurements to develop criteria for assessing the presence of reference rock site condition based on measured seismic velocities and their gradient with respect to depth. We apply the criteria to available profiles and perform rigorous statistical analysis from which we recommend S-and P-wave velocities of 3000 and 5500 m=s, respectively, for the reference rock condition. We recommend that, for practical applications, use ranges of reference S-and P-wave velocities of 2700-3300 m=s and 5000-6100 m=s, respectively. The ranges are based on a 5% change in amplification using quarterwavelength theory. We do not find evidence for regional dependence of the reference velocities, which are derived principally from three general geographic regions: (1) Atlantic coast, (2) continental interior, and (3) Appalachian Mountains. Our data do not provide reference velocities for the Gulf Coast region. The recommended velocity-compatible reference rock site kappa is 0.006 s.