A new geostatistical model for shear wave velocity profiles

A consistent procedure is required to deal with uncertainties in seismic hazard studies. In particular, uncertainties in shear wave velocity (VS) profiles are important for 1D numerical simulations of site response conducted within a probabilistic framework. This work proposes a new geostatistical model for shear wave velocity profiles. The main characteristic of the model is the separation of the random variables space and time. The model is calibrated using a database of surface wave tests compiled for this purpose. The flexibility of the model is then demonstrated by presenting a first prototype version for down-hole tests. The proposed geostatistical model is validated through an application to a real case study at Mirandola (Italy), one of the sites included in the InterPACIFIC project. The results show a significant improvement in the management (i.e., treatment) of uncertainties for ground response analyses, compared to the methods usually adopted for this purpose. The new geostatistical model allows for a rigorous quantification of the uncertainties introduced in the hazard study; these uncertainties depend on both the characteristics of the investigated site and the performed test. It is also shown that the randomization procedure provides a set of profiles which are fully consistent with the independent experimental “site signatures” available at the site. •The proposed model is based on a separation between the fundamental quantities of space and time that avoids the randomization of parasite uncertainties.•The model is experimentally-based in terms of both its architecture and its suggested parameters.•The model is site- and test-specific according with the IQM (Identification, Quantification, and Management) methodology proposed.•The proposed model is calibrated on the specifically compiled database and it is flexible in its application.•The new geostatistical model was validated in the paper for a real case-study. The validation gave excellent results. The new geostatistical model showed results in accordance with all the independent experimental site signatures, including the cumulated travel times, the dispersion curve, the transfer function, and the natural frequency of the site.