Construction of Hybrid Geostatistical Models Combining Surface Based Methods with Multiple-point Geostatistics: Use of Flow Direction and Drainage Area

The use of hybrid techniques aims at constructing new models which enable us to overcome the existing limitations in the reproduction of curvilinear structures. Improvements on a deepwater turbidite reservoir model combining several primary modeling methods are presented. In this approach, multiple points geostatistics (MPS) plays an important role in simulating and conditioning the realizations. Representation of flow direction and calculation of upslope areas on the rectangular grid elevation model are used to determine the drainage basin and MPS simulation area (SA). The SA is then used to relate the previously simulated surface with the current geological event being simulated. The procedure is based on representing flow direction as a single angle taken as the steepest downward slope on eight triangular facets centered at each grid point. For a given anchor point joining the channel to the lobe in the channel-lobe parameterization, we obtain the influence and dependence area in a sequential process. The contour of the newly generated SA along with the cumulative density functions (CDF) retrieved from a process-based model output are employed to introduce variability to the object-based generated training images (TIs). These images are used in the MPS simulation. This procedure is repeated as many times as lobes are to be simulated, only updating the current base topography. The improved model emphasizes using more realistic geological rules, especially on lobe orientation and erosion caused during the deposition of the geobodies. The lobe erosion process is simulated in such a way that flow direction and special topographic features related to the flow erosion power are accounted for. Objective rejection rules are taken into account in the implementation of the model. This gives rise to an automatic and user independent algorithm. This approach is potentially capable of representing the internal features of the geobodies. Finally, strategies on how to condition the model to facies and thicknesses are discussed.