Microtremor Measurements in the City of Palermo, Italy: Analysis of the Correlation between Local Geology and Damage

Abstract

This study presents the results of 90 seismic ambient noise measurements in Palermo, the main city of Sicily (Italy). The dataset has been processed using the horizontal-to-vertical spectral ratio (HVNSR) technique and interpreted in terms of local geology, which is characterized by the presence of alluvial sediments of two riverbeds masked by urbanization since the seventeenth century. HVNSRs show significant variations in the study area: when the transition stiff to soft is crossed, a typical spectral peak appears in the HVNSRs, mostly in the frequency band 1–2 Hz, and exceeding a factor of 3 in amplitude. Using available information on subsurface geological structure, we compute theoretical 1D and 2D transfer functions. The resonance frequencies of soft soils obtained by HVNSR are well reproduced by the fundamental frequencies from numerical modeling. The distribution of frequency peaks of HVNSR and their amplitudes are also compared with the local damage caused by historical earthquakes. Previous studies demonstrated that damage variations in Palermo were controlled more by near-surface geology than building vulnerability. A uniform vulnerability is an ideal condition to test statistical methods and their capability in seeking correlation between HVNSR and potential damage due to local geological conditions. We apply two well-established multivariate statistical methodologies (factor analysis and canonical correlation) to the HVNSR dataset and macroseismic data (damage grades of the European macroseismic scale). Through these analyses we quantify the significance of the correlation between the HVNSR peak in the low-medium frequency range (0.5–3 Hz) and the occurrence of the highest damage grades. This approach allows us (1) to estimate the threshold value in the resulting linear combination of the HVNSR amplitudes, which separates zones of light damage from zones of significant damage, and therefore (2) to improve the spatial definition of potentially high hazard zones through a denser grid of microtremor measurements.