Combination of passive and active methods towards site characterization of accelerometer stations in Greece

Abstract

Abstract The strong correlation between earthquake damage and the local geological conditions in an area has long been established, and site characterization studies are common practice for realistic estimation of seismic hazard and mitigation of the respective seismic risk. Estimation of the time‐averaged S‐waves velocity of the topmost 30 m ( V S30 ) is recognized as a basic proxy for site characterization and has been incorporated in several building codes. In our work, we examine the inversion of single‐station ambient noise horizontal to vertical spectral ratio (HVSR) curves to estimate the local one‐dimensional (1D) V S( Z ) structure. The inversion is performed based on the diffuse field assumption, which allows for the inversion of HVSR, including the contributions of both body and surface waves. Although ambient noise data acquisition is immensely cost‐effective, HVSR curve inversion is subject to the solution non‐uniqueness issue, offering ambiguous results. To provide a possible solution for this matter, additional near‐surface geophysical methods such as the multichannel analysis of surface waves and the electrical resistivity tomography (ERT) were applied to derive information on the shallow subsurface structure. The acquired 1D seismic velocity profile was implemented as initial model in the HVSR inversion to constrain the velocities of the upper meters. Additionally, the inferred stratigraphy from the ERT electrical profile was utilized to constrain the thicknesses of the layers. Tests were conducted with and without a priori information from complementary techniques to explore how the HVSR inversion procedure is facilitated. The proposed methodology was applied at the locations of six accelerometers in the city of Thessaloniki, northern Greece, under different geological conditions. The reliability of the inverted V S( Z ) models was checked by performing 1D numerical simulations of ambient noise waveforms generated by distributed sources and by direct comparison between synthetic ambient noise HVSR and single‐station earthquake HVSR curves.