Estimation of the seismic microzonation amplification factors from the GMMs site-to-site residuals: an application to the Italian strong-motion stations in ITACA 4.0

The use of non-ergodic approaches for calibrating ground-motion prediction models is increasingly well established in the seismological and engineering literature. Specifically, the residuals (logarithmic difference between observed and predicted intensity measures) obtained from the calibration of a parametric model are decomposed to identify repeatable effects related to event, station and any other recurring properties (e.g. source and propagation effects). For the empirical estimation of the site effect, we refer to the repeatable station term, usually called δS2S (site-to-site), which measures the systematic deviation of seismic motion at a station from a reference ground motion prediction for that site category. In order to obtain an amplification function from δS2S estimates, the prediction of a rock site, real or modeled on the basis of a Ground Motion Model, can be taken as a reference prediction. In this work, we discuss our experience in estimating empirical amplification functions from δS2S of permanent and temporary recording stations in Italy, highlighting limitations and advantages of this approach. The results are also discussed in the perspective of the estimation of the amplification factors, useful for the seismic microzonation studies. The values of δS2S (with associated aleatory uncertainty) are published on the ITACA (ITalian ACcelerometric Archive) web page for 916 recording stations for PGA, PGV, 36 ordinates of the acceleration response spectrum (damping 5%) in the range T = 0.01–10 s.