Beicip-Franlab will present the following talk at IMAGE 2022, Houston, Texas.
Reservoir characterization using multi-seed stochastic inversion
Vincent Thomas, Nicolas Desgoutte and Nathalie Lucet
In this paper, we compare inversion results obtained with deterministic and stochastic approaches.
After short description of the inversion techniques, results are presented on real data case, in a shale/sand clastic environment where sand layer thickness is way below seismic resolution.
Next, two seismic reservoir characterization tools, namely discriminant analysis and multi-variate regression, have been applied on the two sets of inversion results and are compared.
Finally, 4 runs of stochastic inversion, using 4 different seeds enable to dispose of a large set of simulated impedance cubes, leading to a more reliable analysis of shale volume, and to the associated variance cube that gives a good insight on uncertainty attached to reservoir parameter estimates.