HR: 0800h
AN: S41A-0231    [Abstracts]
TI: Experimental Study of the Convergence of Two-Point Cross-Correlation Toward the Green's Function
AU: * Gouedard, P
EM: pierre.gouedard@ujf-grenoble.fr
AF: LGIT, BP 53, Grenoble cedex 9, 38041, France
AU: Roux, P
EM: philippe.roux@obs.ujf-grenoble.fr
AF: LGIT, BP 53, Grenoble cedex 9, 38041, France
AU: Campillo, M
EM: michel.campillo@ujf-grenoble.fr
AF: LGIT, BP 53, Grenoble cedex 9, 38041, France
AU: Verdel, A
EM: arie.verdel@shell.com
AF: Shell International Exploration and Production, Postbus 60, Rijswijk, NL-2280, Netherlands
AU: Campman, X
EM: xander.campman@shell.com
AF: Shell International Exploration and Production, Postbus 60, Rijswijk, NL-2280, Netherlands
AB: It has been shown theoretically by several authors that cross-correlation of the seismic motion recorded at two points could yield the Green's Function (GF) between these points. Convergence of cross-correlations toward the GF depends on sources positions and/or the nature of the wavefield. Direct waves from an even distribution of sources can be used to retrieve the GF. On the other hand, in an inhomogeneous medium, recording the diffuse field (coda) is theoretically sufficient to retrieve the GF whatever the sources distribution is. Since none of these two conditions (even distribution of sources or a perfectly diffuse field) is satisfied in practice, the question of convergence toward the GF has to be investigated with real data. A 3D exploration survey with sources and receivers on a dense grid offers such an opportunity. We used a high- resolution survey recorded by Petroleum Development Oman in North Oman. The data have been obtained in a 1x1~km area covered with 1600 geophones located on a 25x25~m-cell grid. Records are 4-seconds long. A unique feature of this survey is that vibrators (working in the [8-120~Hz] frequency band), were located on a similar grid shifted with respect to the receiver grid by half a cell (12.5~m) in both directions. This allows us to compare estimated GF's with measured direct waves (GF's) between the geophones. The shallow subsurface is highly heterogeneous and records include seismic coda. From this dataset, we selected two receiver locations (Ra and Rb) distant from d=158~m. We used both different sets of source locations and time windows to compute the cross-correlation between these two receivers. Then we compared the derivatives of correlation functions with the actual GF measured in Rb (resp.~Ra) for a source close to Ra (resp.~Rb). By doing so, we show the actual influence of source locations and scattering (governed by the records' selected time window) on the Signal-to-Noise Ratio (SNR) of the reconstructed GF. When using direct waves, the result illustrates the dominant contribution of the sources located in the endfire lobes of the receiver pair (as defined from the stationary phase theorem). If sources are located all around the receivers, both direct and coda waves yield very good recovery of the GF, with a comparable SNR. The agreement between actual GF and cross- correlation derivative is then very good for direct surface waves, and extends to later arrivals, which are likely scattered waves. If all sources are located outside the endfire lobes, the GF can nevertheless be retrieved when using the scattered waves contained in the coda. Acknowledgment: The authors would like to thank the Ministry of Oil and Gas of the Sultanate of Oman and Shell Research for their permission to publish these results.
DE: 0910 Data processing
DE: 0935 Seismic methods (3025, 7294)
DE: 7255 Surface waves and free oscillations
SC: Seismology [S]
MN: 2007 Fall Meeting