Lateral heterogeneities and microtremors: Limitations of HVSR and SPAC based studies for site response

• We present a site effect study carried out in a small volcanic basin.
• Microtremors are analyzed for site response and structure using HVSR and SPAC.
• We suggest that complexity of the basin structure may limit the usefulness of HVSR.
• Heterogeneities may force dispersion curves among stations to be incompatible.
• We show that dispersion may be estimated using SPAC even when medium is far from 1D.

We report a site effect study carried out in a small alluvial valley, Nari basin, in Ulleung island, Republic of Korea. In that small valley a permanent, broad-band array of five stations records ground motion continuously. In addition, we performed microtremor measurements using two arrays of stations; a small aperture array (60 m) and a medium aperture one (500 m). Ambient vibration was also recorded at 38 points inside the basin. The records were analyzed using horizontal-to-vertical spectral ratios (HVSR), to estimate dominant frequency and maximum amplification, and the spatial autocorrelation (SPAC) method, to estimate the subsoil structure via phase velocity dispersion curves. Nari basin is clearly a 3D structure but it was expected that a laterally irregular model might be built from the interpolation of several 1D models. Dominant frequency maps based on results from HVSR suggest a simple sedimentary structure with mostly smooth lateral variations. In contrast, maximum amplification distribution is very irregular and shows no correlation with dominant frequency. In the case of Nari, it seems that the complexity of the basin structure cancels the usefulness of HVSR. We could estimate a phase velocity dispersion curve for 81% of the station pairs recording simultaneously ambient vibration in our arrays. In all the cases, the resulting dispersion curve had a simple shape, suggesting that the subsoil structure consists of a single layer over a half space. However, the dispersion curves are incompatible among them, even for different station pairs covering the same path, something impossible if the subsoil structure were 1D. It seems that the passive methods we have used in Nari basin are not adapted to its structure. This makes a tomography unfeasible and we are unable to propose a subsoil structure for Nari basin. Our results show that surface wave dispersion may be estimated using the SPAC method even when the medium is far from 1D. We believe that this problem is enhanced by the small size of this basin, which amplifies the effect of lateral heterogeneities.