Expert Panel Recommendations for Ergodic Site Amplification in Central and Eastern North America

Abstract

The U.S. Geological Survey (USGS) national seismic hazard maps have historically been produced for a reference site condition of VS30 = 760 m/sec (where VS30 is time averaged shear wave velocity in the upper 30 m of the site). The resulting ground motions are modified for five site classes (A-E) using site amplification factors for peak acceleration and ranges of short- and long-oscillator periods. As a result of Project 17 recommendations, this practice is being revised: (1) maps will be produced for a range of site conditions (as represented by VS30 ) instead of a single reference condition; and (2) the use of site factors for period ranges is being replaced with period-specific factors over the period range of interest (approximately 0.1 to 10 sec). Since the development of the current framework for site amplification factors in 1992, the technical basis for the site factors used in conjunction with the USGS hazard maps has remained essentially unchanged, with only one modification (in 2014). The approach has been to constrain site amplification for low-to-moderate levels of ground shaking using inference from observed ground motions (approximately linear site response), and to use ground response simulations (recently combined with observations) to constrain nonlinear site response. Both the linear and nonlinear site response has been based on data and geologic conditions in the western U.S. (an active tectonic region). This project and a large amount of previous and contemporaneous related research (e.g., NGA-East Geotechnical Working Group for site response) has sought to provide an improved basis for the evaluation of ergodic site amplification in central and eastern North America (CENA). The term ‘ergodic’ in this context refers to regionally-appropriate, but not site-specific, site amplification models (i.e., models are appropriate for CENA generally, but would be expected to have bias for any particular site). The specific scope of this project was to review and synthesize relevant research results so as to provide recommendations to the USGS for the modeling of ergodic site amplification in CENA for application in the next version of USGS maps. The panel assembled for this project recommends a model provided as three terms that are additive in natural logarithmic units. Two describe linear site amplification. One of these describes VS30-scaling relative to a 760 m/sec reference, is largely empirical, and has several distinct attributes relative to models for active tectonic regions. The second linear term adjusts iv site amplification from the 760 m/sec reference to the CENA reference condition (used with NGA-East ground motion models) of VS =3000 m/sec; this second term is simulation-based. The panel is also recommending a nonlinear model, which is described in a companion report [Hashash et al. 2017a]. All median model components are accompanied by models for epistemic uncertainty. The models provided in this report are recommended for application by the USGS and other entities. The models are considered applicable for VS30 = 200–2000 m/sec site conditions and oscillator periods of 0.08–5 sec. Finally, it should be understood that as ergodic models, they lack attributes that may be important for specific sites, such as resonances at site periods. Site-specific analyses are recommended to capture such effects for significant projects and for any site condition with VS30 < 200 m/sec. We recommend that future site response models for hazard applications consider a two-parameter formulation that includes a measure of site period in addition to site stiffness.