PGA Semi-Empirical Correlation Models Based on European Data

Spatial modelingof ground motion intensity measures (IMs) is required for seismic risk analysis of distributed systems as lifelines. In fact, when a spatially distributed infrastructure is of concern, site-specifichazard tools, which treat IMs at different locations independently, may be not adequate to assess accurately the seismic risk and the joint distribution of ground motion parameters at different sites may beneeded. This work focuses on semi-empirical estimation of the correlation coefficient,as afunction of inter-site separation distance,of peak groundacceleration(PGA). In particular, subsets of the European Strong-motion Database is employed to evaluate the intra-event correlationbased on multiple earthquakes. Starting from the residuals,with respect to a ground motion prediction equation (GMPE), calibrated on the same dataset, a statistical analysis of correlation based on geostatisticaltoolsis performed andsemi-empirical models,for horizontal and vertical PGA components, are proposed.Finally,possible criteria for including spatial correlation in regionalprobabilisticseismic hazard analysis are discussed