Testing probabilistic seismic hazard estimates by comparison with observations: an example in Italy.

An evaluation of the actual reliability of probabilistic seismic hazard (PSH) assessments, provided by existing numerical techniques, is mandatory to orientate new researches and improvements. Two procedures devoted to this task are proposed, which are based on the comparison of the hazard estimates with empirical observations (e.g. strong-motion data). These procedures have been applied to the estimates provided by the methodology adopted for most recent seismic hazard evaluations in Italy. The analysis shows that a significant mismatch exists between peak ground acceleration values characterized by an exceedence probability of 10 per cent in 30 yr and what has actually been observed at 68 accelerometric stations located on stiff soil, where continuous seismicity monitoring has been performed in the last 30 yr. Although this finding should be considered with caution, it suggests that a future re-examination of the adopted PSH computational model could be useful to avoid possible underestimates of seismic hazard in Italy