Rayleigh-wave dispersion curve: a proxy for site effect estimation?

Abstract

One of the open issues on the effects of surface geology regards the estimation of site response when limited resources are available. In that restrictive context, one solution is to use soil characteristics as proxy. Despite its extensive use, the most common proxy, Vs30, is presently criticized because it cannot carry alone the main physics of site response. We propose here a statistical investigation of the capabilities of another proxy, the Rayleigh-wave dispersion curve, DC. When considered over a broad enough frequency band, it can provide deeper information missing in the single Vs30 parameter. A set of shear-wave velocity profiles measured for more than 600 Japanese KiK-net stations is used to compute theoretical dispersion curves (DC) and theoretical SH transfer functions (SH), while instrumental surface/downhole spectral ratios were calculated in a previous work (Cadet et al., 2011a). Canonical correlation techniques are applied to this large data set to analyze the relationship between DC and theoretical or empirical site responses. The results indicate very encouraging qualitative statistical relationships between DC and site amplification for numerically derived SH transfer functions, showing significant canonical couples with correlations up to 0.95. Results for instrumental surface/downhole transfer functions correspond to lower correlations (up to 0.73) but still allow the development of quantitative relationships.