Insights into bedrock surface morphology using low-cost passive seismic surveys and integrated geostatistical analysis

- The increase in available geo-environmental data requires ad-hoc data analysis techniques.
- We perform a geostatistical analysis of 116 HVSR single station measurements.
- The trend/residual decomposition permits to gain insight into the bedrock morphology.
- The interpolation approach uses auxiliary information for improved mapping.
- The potential heteroscedasticity of HVSR resonance frequencies should be considered.

The HVSR (Horizontal to Vertical Spectral Ratio) technique is very popular in the context of seismic microzonation and for the mapping of shallow seismic reflectors, such as the sediment/bedrock transition surface. This easy-to-deploy single station passive seismic technique permits the collection of a considerable amount of HVSR data in a cost-effective way. It is not surprising that some recent studies have adopted single station micro-tremor analyses in order to retrieve information on geological structures in 1D, 2D or even 3D reconstructions. However, the interpolation approaches followed in these studies for extending the punctual HVSR data spatially are not supported by a detailed spatial statistical analysis. Conversely, in order to exploit the informative content and quantify the related uncertainty of HVSR data it is necessary to utilize a deep spatial statistical analysis and objective interpolation approaches. Moreover, the interpolation approach should make it possible to use expert knowledge and auxiliary information. Accordingly, we present an integrated geostatistical approach applied to HVSR data, collected for retrieving information on the morphology of a buried bedrock surface. The geostatistical study is conducted on an experimental dataset of 116 HVSR data collected in a small thermal basin located in the Venetian Plain (Caldiero Basin, N-E Italy). The explorative geostatistical analysis of the data coupled with the use of interpolation kriging techniques permit the extraction of relevant information on the resonance properties of the subsoil. The utilized approach, based on kriging with external drift (or its extension, i.e. regression kriging), permits the researcher to take into account auxiliary information, evaluate the related prediction uncertainty, and highlight abrupt variations in subsoil resonance frequencies. The results of the analysis are discussed, also with reflections pertaining to the geo-engineering and geo-environmental context.

544