Cyclic 1D matching of time-lapse seismic data sets: A case study of the Norne Field |

Imperfect survey geometry repetition, non-repeatable ambient noise and other discrepancies generate artifacts that contaminate production-related differences between seismic data sets. Before reservoir property changes can be extracted from time-lapse seismic data, these artifacts must be attenuated. The process of attenuating artifacts from time-lapse data sets is generally called seismic cross-equalization (Rickett and Lumley, 2001). In this paper, two robust post-imaging cross-equalization steps are considered.

Because reservoir depletion changes the stress-states in and around a producing reservoir, seismic travel times and path lengths are different between surveys.
However, in practice, the baseline and monitor data sets are migrated with a single (baseline) velocity, leading to *apparent* vertical and lateral displacements between seismic images.
Because it is expensive to estimate all components of the displacement vector field, it is common practice to only consider vertical displacements.
Instead of this *vertical-only* approximation, displacement components are efficiently computed with a cyclic 1D correlation method (Hale, 2009).
It is shown that in order to obtain accurate estimates of reservoir property changes, all displacement components must be taken into account.

To ensure that only production-related changes within the reservoir are interpreted, differences in seismic signal in non-reservoir regions should be minimal. After aligning all monitor images to the baseline (by removing the estimated apparent displacements), an optimal cyclic 1D match-filtering method is used to attenuate differences in the non-reservoir regions. This cyclic filtering method is an extension of the optimal matching method described by Ayeni and Nasser (2009). Together with the cyclic correlation method described above, this matching method forms a robust and effective seismic cross-equalization scheme.

In this paper, I first briefly discuss the cyclic shifting and match-filtering methods. Next, I apply these to four time-lapse seismic data sets from the Norne field. The Norne study shows that neglecting lateral displacements can cause spurious time-lapse signals. Finally, I show that these cross-equalization methods significantly improve the quality of the time-lapse images and their derivatives.

Cyclic 1D matching of time-lapse seismic data sets: A case study of the Norne Field |

2010-05-19