Facies Modeling Accounting for the Precision and Scale of Seismic Data: Application to Albacora Field G. Schwedersky-Neto, Marcella Maria de Melo Cortez, and Marcos Fetter Lopes Petrobras CENPES Research Center C. V. Deutsch, Department of Civil & Environmental Engineering, University of Alberta Abstract Seismic impedance provides information on the relative proportion of different facies types. It is important to integrate such seismic data in the construction of detailed 3-D facies models, which are used for reservoir management. Two critical challenges faced in the integration of seismic impedance data: (1) the seismic data is at a larger scale than the well data / geological modeling cells, and (2) the seismic data provides soft (imprecise) information on the facies proportions within that large volume. A novel block cokriging approach was developed and implemented. This method was adapted for use in sequential indicator simulation to explicitly account for the large scale soft seismic data. Conventional sequential indicator simulation and a popular alternative, SIS with Bayesian updating, were considered for comparison purposes. The key challenge in applying stochastic simulation with block cokriging is the construc- tion of a licit model of coregionalization between the “hard” well data and the “soft” seismic data. A hybrid procedure is presented that can be applied in the common case of limited well data. The Albacora field offshore Brazil consists of deep water turbidite sands, shales, and cemented sands. Good quality seismic data and significant variations in facies proportions make this an excellent example to illustrate the benefit of integrating seismic data in high resolution 3-D facies modeling. KEYWORDS: geostatistical simulation, stochastic modeling, reservoir characterization, Bayesian updating, block cokriging Introduction One goal of reservoir modeling teams is to build high resolution predictive reservoir models of facies, porosity, and permeability that, by construction, honor all available reservoir data. These numerical models provide reliable predictions of future reservoir performance at all stages of the reservoir life cycle. The unavoidable uncertainty in reservoir performance forecasting will be measured and minimized by such reliable numerical reservoir models. 1