The focus of the study is on lithology-fluid inversion from prestack seismic data. The target zone is a 3D reservoir, and the inversion problem is solved in a Bayesian framework where the complete solution is given by the posterior model. The likelihood model relates the lithology-fluid classes to elastic variables and the seismic data, and it follows the lines of Larsen et al. (2006). In order to make allowances to the strong lateral coupling between the lithology-fluid classes, the prior model is defined as a profile Markov random field. To model vertical continuity of the lithology-fluid classes along the profiles, a Markov chain model upward through the reservoir is used. The posterior model is given as the complete set of the full conditional pdf’s in the profile Markov random field model, and a block Gibbs simulation algorithm is used laterally. The profiles are simulated exactly using the efficient upward-downward algorithm defined in Larsen et al. (2006). The inversion model is evaluated on a synthetic 2D reservoir. The lithology-fluid classes in the reference model have strong horizontal continuity with thin layers of shale, and the fully coupled 3D model provides reliable results.


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