The inversion result in a Bayesian setting is, among others, controlled by a prior model covariance matrix, which contains the expected a priori variances and correlations between the seismic properties. In time-lapse analysis, we expect that significant changes occur only in the reservoir sand units, whereas the changes in the surrounding rocks are expected to be negligibly small. In our extension of the Bayesian time-lapse inversion, we utilize this observation by using a vertically and horizontally varying model covariance matrix. In the parts of the data cube containing sands which may be affected by production, we allow for large changes by assigning large variance values. We use small variance values for the remainder of the data cube, which is most likely not affected by production. In order to control the spatial variation of the model covariance matrix, we exploit additional information. Such information can, for instance, be well-log data and interpreted seismic horizons. In this paper, we describe our experience of including stratigraphic interpretation into a 4D seismic inversion workflow. We demonstrate the benefits of this workflow for quantitative interpretation through synthetic and real data examples.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error