Can Groundwater Depth be a Reliable Proxy for Vertical Site-Response?

We explore the empirical correlation between vertical site-response and groundwater depth (GWD), assuming that vertical site response are affected by upward propagating P waves and hence by the level of saturation. Two approaches are used to calculate amplification: the first, in which amplification is computed with respect to a regional ground motion model (GMM); and the second, in which amplification is computed on a record-by-record base, using a nearby rock site. We also use two different approaches to predict GWD at the site: one, treating GWD as a pseudo-stationary parameter, using regional geostatistical interpolation; the second, treating GWD as time-dependent, and computing it within a limited time and space window with respect to the recorded ground motions (GM). We focus on California, which is rich in both GM and GWD data. We test all combinations of amplification and GWD, in search of correlation between these two parameters. Despite genuine efforts to empirically validate such correlation, as suggested by previous analytical studies, we do not find sufficient evidence for such a dependence. We conclude that GWD cannot be used as a reliable site-characterization proxy for forward prediction and that V_S30 is currently the most powerful single proxy for site-response, including that of the vertical GM.