Importance of Pre-Processing Before Geostatistical Cosimulation: an Illustration in Reservoir Modelling

Geostatistical cosimulation algorithms provide values for multiple oil reservoir features at unknown locations. Cosimulation can consider primary data (as data sampled in wells) and secondary data (as data readings from seismic) combined in running the simulation. Seismic data provide less precise information than those acquired at wells. However, the first is exhaustively available (at all grid node locations) along the study region. In order to generate more accurate simulations for primary data, it is necessary to transform the secondary data histogram to approximate the primary data histogram along the same region (study region) before running simulations. However, the primary data histogram along the study region, is not known, when there are just a few exploration wells. This paper presents how to transform the secondary data histogram to represent the primary counterpart before combining the two for cosimulation. The results showed that the transformation of the secondary data histogram using the methodology proposed generated simulated scenarios more accurately than if this transformation assumed that statistics of the primary data collected at wells represent the parameters for the entire region where there are only a few wells, and they are not distributed along the entire study region.