European Seismological Commission, 32nd General Assembly, Sept. 6-10, 2010, Montpellier, France
Caccavale M., Landolfi L., Convertito V., D'Onofrio A., Silvestri F.
A set of non linear relationships between the stratigraphic amplification factor (Ss) and peak ground acceleration (P GA) at bedrock (ar) was produced to be implemented in Shakemap®. The relationships represent a data-set of pairs (Ss, ar) referred to the sites selected in the Italian accelerometric archive and previously sorted according to the EC8-NTC2008 prescriptions about soil classes.
The data-set contains pairs obtained by:
comparing the site and reference-site records de-convolution of records at sites where the subsoil is adequately characterized 1D seismic response analysis performed on "virtual" stratigraphic profiles, consistent with the EC8-NTC2008 site classification, and subjected to accelerograms recorded at different Italian sites.
The relationship between Ss and ar is expressed by a stepwise function obtained by meaning the data included in different range of P GA considered. A series of test have been performed to validate the results and Shakemap® has been used to generate the P GA distribution associated to the 23.XI.1980 Irpinia earthquake (MW 6.9). The residuals of the generated values have been analyzed.
The 23.XI.1980 event has been chosen due to the detailed characterization of the closer station’ s sites that allow to back figure the accelerograms at bedrock by a 1D numerical deconvolution of the record. In this way, it has been possible to use two data-set: one containing the observed horizontal P GA (i.e. the surface Recorded values), a second made by the deconvolved values, assumed to be the P GA distribution observable in case of bedrock outcropping motion. Sensitivity analysis had been performed to find the best setup, aiming to isolate the influence of the site corrections.
The test 1a and 1b have been run without site corrections; a distribution of values of P GA have been produced with all stations (1a) and removing one station at time (2a), revealing the capability of the algorithm to generate values of P GA consistent with the real values. The role of the attenuation law and the geographical distribution of the real stations has been underlined plotting the residuals on real values. The test 2a and 2b repeat the previous ones introducing site corrections. A comparison has been done on the residuals produced by performing the test 2a and 2b with other two sets of amplification factors: the first has been proposed by Barani et al (2008) and the second one is currently used in italian shakemap implementation.