Constraining attenuation with ambient noise correlations of reservoir scale seismic data

Surface waves extracted from the ambient seismic wave field via interferometry can be used for velocity inversion. Bussat & Kugler (2009) adapted this approach to Scholte waves at frequencies up to 1 Hz, extracted from comparatively short broad band Ocean Bottom Seismometer (BBOBS) recordings. Shear wave velocities were obtained by De Ridder and Biondi (2010) for seismic land data at the reservoir scale. By fitting a modified Bessel function to the real part of the stacked cross spectra Prieto et al. (2009) obtained a 1-Dimensional Q profile for Southern California. We applied this approach to two passive seismic data sets and compare the results. The first recorded on land, the second being BBOBS recordings. The survey characteristics are roughly the same considering scale and length of the recordings. The data characteristics however are quite different. The main energy in the OBS data below ∼2 Hz stems from swell noise and ocean microseisms. This energy is expressed as Scholte waves traveling along the seabed. The land passive data set is the result of a survey over a gas storage facility in Central France and the main energy corresponds to microseisms and anthropogenic noise. The land seismic survey was carried out twice. Once for a filled reservoir and once for an empty reservoir. The subsurface geology is well-known. About 200 three-component broadband particle velocity sensors were deployed. Data was recorded for 24-48 hours at 610 surface locations. Station spacing is nominally 500m. The data is cut into time windows of 60 seconds and whitened. Stacked cross spectra are obtained for all station couples. The stability of the cross spectra is examined with respect to (1) the power of the cross correlated wave field and (2) the azimuth of the station couples. In the presence of anisotropy of the noise source geographic distribution, we can average the calculated cross spectra over station-station azimuth. Theory derived for an isotropic source distribution can then be applied to our data. We shall nevertheless evaluate the uniformity of the noise source distribution and attempt to obtain stable cross spectra at all available azimuths. This would allow us to evaluate lateral variations in scattering and intrinsic attenuation. At this scale, this represents a significant step forward in the application of the seismic ambient noise method.