Quantifying Nonlinearity Susceptibility via Site-Response Modeling Uncertainty at Three Sites in the Los Angeles Basin

Abstract

The effects of near-surface soil stratigraphy on the amplitude and frequency content of ground motion are accounted for in most modern U.S. seismic design codes for building structures as a function of the soil conditions prevailing in the area of interest. Nonetheless, currently employed site-classification criteria do not adequately describe the nonlinearity susceptibility of soil formations, which prohibits the development of standardized procedures for the computationally efficient integration of nonlinear ground response analyses in broadband ground-motion simulations. In turn, the lack of a unified methodology for nonlinear site-response analyses affects both the prediction accuracy of site-specific ground-motion intensity measures and the evaluation of site-amplification factors when broadband simulations are used for the development of hybrid attenuation relations. In this article, we introduce a set of criteria for quantification of the nonlinearity susceptibility of soil profiles based on the site conditions and incident ground-motion characteristics, and we implement them to identify the least complex ground response prediction methodology required for the simulation of nonlinear site effects at three sites in the Los Angeles basin. The criteria are developed on the basis of a comprehensive nonlinear site-response modeling uncertainty analysis, which includes both detailed soil profile descriptions and statistical adequacy of ground-motion time histories. Approximate and incremental nonlinear models are implemented, and the limited site-response observations are initially compared to the ensemble site-response estimates. A suite of synthetic ground motions for rupture scenarios of weak, medium, and large magnitude events (M 3:5-7:5) is next generated, parametric studies are conducted for each fixed magnitude scenario by varying the source-to-site distance, and the variability introduced in ground-motion predictions is quantified for each nonlinear site-response methodology. A frequency index is developed to describe the frequency content of incident ground motion relative to the resonant frequencies of the soil profile, and this index is used in conjunction with the rock-outcrop acceleration peak amplitude (PGA RO ) to identify the site conditions and ground-motion characteristics where incremental nonlinear analyses should be employed in lieu of approximate methodologies. We show that the proposed intensity-frequency representation of ground motion may be implemented to describe the nonlinearity susceptibility of soil formations in broadband simulations by accounting both for the magnitude-distance-orientation characteristics of seismic motion and the profile stiffness characteristics. The synthetic ground-motion predictions are next used for the development of site-amplification factors for the alternative site-response methodologies, and the results are compared to published site factors of attenuation relations. For the site conditions investigated, currently established amplification fac...

Site- and Motion-Dependent Parametric Uncertainty of Site-Response Analyses in Earthquake Simulations

“…This work is based on a previous study by the authors (Assimaki et al, 2008), who evaluated the soil modeling variability in site-response predictions at three downhole array sites in Southern California using approximate and rigorous nonlinear site-response models and synthetic ground motions. The position coordinates, operating agencies, depth of downhole instruments, and geological description of the sites are given in measurements and laboratory resonant column modulus reduction and damping curves (Anderson, 2003) were available at these locations.…”

“…The compiled shear-wave velocity (V S ), attenuation (Q 1=2ξ, where ξ is the material damping), and density profiles (ρ) are shown in Figure 1 and are used as base profiles of the random soil property fields in this study. Because strong ground motion recordings at these stations were scarce, Assimaki et al (2008) developed a statistically significant dataset of seismic input motions using synthetic records. For this purpose, they used 1D crustal compressional velocity (V P ), shear velocity (V S ), and density models (ρ) from the 3D Southern California Community Velocity Model IV (SCEC CVM IV; see the Data and Resources section) and the hybrid low-/high-frequency dynamic rupture source model by Liu et al (2006).…”

“…Finally, Assimaki et al (2008) conducted site-response simulations at the three sites for the limited number of recorded ground motions and the synthetic seismograms using multiple soil models. They evaluated the deviation of ground surface predictions from the observed time histories and identified the monotonic constitutive law by Matasovic and Mladen (1995) coupled with a modified hysteretic formulation of the model proposed by Muravskii (2005) as the hysteretic soil model that yielded the minimum average error.…”

“…Using Toro's (1993) probabilistic velocity model, random realizations of the V S stochastic fields were next generated using a two standard deviation (2 std) truncation of the lognormal distribution to eliminate potential outliers. Realizations of the V S random profile at the La Cienega downhole array site are shown in Figure 4, where the thick black line corresponds to the median profile at the site (Assimaki et al, 2008), and the thin gray lines correspond to 50 randomized V S profiles.…”

“…More specifically, (1) we use extensive geotechnical data on the variability statistics of soils at three downhole array sites in the Los Angeles (LA) Basin (Darendeli, 2001;Anderson, 2003;Toro, 1993) and develop realistic stochastic fields of elastic and nonlinear dynamic soil properties, (2) we generate synthetic ground motions by means of a finite source dynamic rupture model over a wide range of magnitudes and distances to obtain a statistically significant number of ground motions for the analysis, and (3) we simulate the large-strain response of soils by means of a hysteretic soil model validated by comparison with downhole seismogram recordings (Assimaki et al, 2008(Assimaki et al, , 2010. For each of the three sites, we evaluate the effects of soil parameter uncertainty as a function of the seismic input intensity and frequency content.…”