Bayesian Stochastic Inversion of Seismic Data in a Stratigraphic Grid

@inproceedings{MoyenRM2007,
 abstract = { Traditional impedances inversion techniques, referred to as deterministic, cannot generate results with a higher frequency content that the input seismic data. This limitation can however be overcome by using a stochastic modelling approach and generating multiple realisations of elastic properties. The first generation of stochastic algorithms, like the one proposed by Haas and Dubrule and extended by Shtuka and Mallet, are based on a trace-by-trace impedance simulation which is accepted or rejected according to its fit to the observed seismic data. This scheme does not allow for a fine user control of the data integration, is computationally expensive and is not adapted to elastic inversion. We propose here a new method called GeoSI, which uses a Bayesian framework and a linearised, weak contrast approximation of the Zoeppritz equation to estimate a log-Gaussian posterior distribution for both P- and S-wave impedances in a fine scale stratigraphic grid, at reservoir scale rather than seismic scale. The posterior distribution is then sampled using a sequential simulation algorithm integrating lateral and vertical continuity constraints through variograms. The sequential nature of the simulation process allows to parallelise the algorithm and invert simultaneously different traces on different processors of a shared memory computer. Moreover, the inversion is entirely controlled from the Gocad interface with dedicated mechanisms to handle long runtimes and avoid locking the user interface. The multiple realisations of P- and S-wave impedances can furthermore be used for cascaded stochastic simulation of petrophysical reservoir properties, lithology classification and uncertainty analysis. GeoSI has been tested on several real data sets in clastic environnement, and results on a model of more than 30 millions grid cells are shown here. },
 author = { Moyen, Rémi AND Thore, Pierre },
 booktitle = { 27th gOcad Meeting },
 month = { "june" },
 publisher = { ASGA },
 title = { Bayesian Stochastic Inversion of Seismic Data in a Stratigraphic Grid },
 year = { 2007 }
}