Abstract

In order to estimate the potential damage to infrastructure and losses that a seismic event could cause, it is necessary to estimate the ground motion intensities that can occur simultaneously in different sites given the event. In this context, the within-event spatial correlation has an important contribution in the definition of the probability distribution of a field of ground motion intensities. This subject has been widely studied over the last 20 years and several models have been developed for different regions around the world. In this paper, we propose a spatial correlation model for peak ground acceleration and spectral pseudo-acceleration ordinates for the Chilean subduction zone. The Chilean database has several station pairs that share multiple common events. Therefore, non-stationary spatial correlations between station pairs are computed, and a stationary model is then fitted. This model is compared with previously developed ones, concluding that subduction models show higher correlations than models for active shallow crustal events, in particular, for long-period spectral ordinates.