Froment, Bérénice

Site-effect assessments performed through earthquake-based approaches, such as the standard spectral ratio (SSR), require good quality records of numerous earthquakes. In contrast, the use of ambient noise appears to be an attractive solution for ease and rapid computation of site responses with sufficient spatial resolution (microzonation), especially in low seismicity areas. Two main approaches are tested here: the horizontal-to-vertical spectral ratio (HVSR) and the noise-based SSR (SSRn). The HVSR uses the relative amplitude of the horizontal and vertical components of the ambient noise. Instead, the SSRn defines the spectral ratio between the seismic noise recorded simultaneously at a site and at a rock reference station, similar to earthquake-based SSR. While the HVSR is currently used in hundreds of site-specific studies, the SSRn approach has been gradually abandoned since the 1990s. In this study, we compare the results obtain from these two approaches with those of earthquake-based SSR. This comparison is carried out for two sedimentary basins, in Provence (southeastern France) and in Argostoli (western Greece). In agreement with the literature, the HVSR does not provide more than the fundamental resonance frequency of the site (f0). The SSRn leads to overestimation of the SSR amplification factors for frequencies higher than the minimal f0 of the basin (f0min). This discrepancy between SSRn and SSR is discussed, and appears to be mainly dependent on the local geological configuration. We thus introduce the hybrid standard spectral ratio (SSRh) approach, which aims to improve upon the SSRn by adding an intermediate station inside the basin for which the SSR is known. This station is used in turn as a local reference inside the basin for the SSRn computation. The SSRh provides site transfer functions very similar to those of the SSR, in a broad frequency range. Based on these results, the SSRn (or SSRh) should be further tested and should receive renewed attention for microzonation inside sedimentary basins.

In the last decades, the use of microtremors to assess linear site effects has received a significant attention. Various authors demonstrate the reliability of ambient noise processing techniques to assess the relative amplification generally between distant sites (>1-10 km) in the low frequency range (<1 Hz). In this study we attempt to evaluate the validity of such an approach at the scale of an industrial facility (≈300 m) and up to high frequency (≈10 Hz) on two sites: an industrial site located across a small sedimentary basin in the low-to-moderate seismicity context of Southeastern France (Provence) and another sedimentary basin in the seismically active island of Cephalonia in Greece. On both sites, transfer functions relative to a rock reference deduced from microtremors (SSRn) are compared to those obtained from earthquake recordings with the classical Standard Spectral Ratio (SSR) approach. We find a good agreement between both approaches until intermediate frequencies. We thus propose an approach based on microtremor recordings and on one neighboring available SSR computed for a station located inside the basin (SSRh). This new approach seems reliable to assess the spatial variability of the site amplification up to high frequency (microzonation). The upper frequency limit for both noise approaches (SSRn and SSRh) is discussed according to the inter-station distance and to the stations geological context.