Spatial correlation of ground motion is required in modelling hazard for risk assessment of spatially distributed engineering systems. In particular when a portfolio of buildings or a transportation/distribution network is of concern, pseudo spectral accelerations (PSA) correlation models may be considered in order to evaluate the expected  performance loss in case of seismic events. The estimation of an appropriate spatial correlation model is still a research task in earthquake engineering, because of involving of different issues: dataset, statistical approaches and estimation tools, and tests to evaluate the estimated models.

In the presented paper, to investigate these issues, an extensive analysis of the spatial and cross-spectral correlation of PSA  is carried out using the European Strong-motion Database (ESD) and the Italian Accelerometric Archive dataset (ITACA) with respect to different ground motion prediction equations (GMPEs), calibrated on the same datasets considered. Moreover, possible dependency on earthquake source characteristics and regionalization are addressed, and the dependence of the distance decay rate of correlation with structural period is also investigated. Finally, developed correlation models are discussed with respect to possible criteria in regional seismic hazard assessment.