Modeling Site Amplification in Eastern North America

Abstract:

A critical component in the understanding and interpretation of earthquake ground motions is the role that site effects play. In many parts of eastern North America, the soil layers which overlie glaciated bedrock produce strong and highly variable site responses. We use horizontal-to-vertical (H/V) response spectral ratios as the indicator variable by which to characterize the salient characteristics of site response in eastern Canada. We show that site response can be modeled using two descriptive variables that are readily obtainable: (i) peak resonant frequency (f_peak), as determined from H/V or depth-to-bedrock; and (ii) overall soil type (or stiffness). We use these variables to create a model of site amplification that can be used in the development of ground-motion prediction equations (GMPEs) and in real-time interactive ground-motion (IGM) map applications.

The key to the site characterization is the relationship between f_peak and drift thickness (depth-to-bedrock), which we derive using H/V data from earthquakes in the region, combined with a detailed digital drift thickness map available online from the Ontario Geological Survey (OGS). The OGS map also provides information on soil type, which is correlated with peak amplitudes (A_peak) of response. We extend the study area to the city of Montreal using similar information from Chouinard and Rosset (2012). H/V spectral shapes may be associated with four main soil categories, which in decreasing order of stiffness are: bedrock, till, sand/clay, and organic soil/fill. The value of A_peak increases as stiffness decreases. We model site response by defining a generic site amplification curve, which is dependent only on f_peak and soil type. The generic curve enables an estimate of site amplification to be made over the entire frequency band of 0.1 to 50 Hz, knowing just the soil thickness and type. These site amplification curves can be applied in the development of regional GMPEs, and in the construction of robust near-real-time IGMs.