Seismic Reservoir Characterization of Thin Stacked Reservoirs Based on Seismic Inversion and Probabilistic Property Estimation Methods - Cheleken Case Study

Abstract The Cheleken fields, located in southern Caspian Sea, are made of a succession of stacked sand reservoirs over a 6km time window, with significant areal and vertical variations. A seismic inversion was carried out with the aim of improving the understanding and delineation of these complex reservoirs. 3D OBN seismic data, covering the area of interest, was recently acquired. A tailored processing and velocity model building workflow was designed to produce a good quality seismic data volume and a reliable velocity model which were used as inputs for a subsequent reservoir characterization study. Deterministic and stochastic inversions followed by facies and porosity estimation based on statistical methods were applied to predict reservoir occurrence and quality. Due to the large time window of the stacked sand reservoirs, the inversion required separate wavelets and facies/porosity model calibration per interval. An integrated process, including seismic data, well logs, seismic velocity fields as well as the interpreted horizons, was adopted to better characterize the targets and succeeded in predicting reliable rock properties away from well boreholes. The inverted elastic attributes were converted into sand/shale facies using a Bayesian classification approach. Despite the limited seismic resolution and a significant overlap in the elastic domain between facies, the method showed good predictability at the well locations and provided a better understanding of the spatial sand distribution. In addition, a new multi-attribute statistical transform approach applied for reservoir property estimation helped predicting the porosity to further optimize future well planning. One of the benefits of the employed statistical methods is the ability to gauge uncertainties associated to the estimates in the form of facies probability and porosity standard deviation. In particular, the facies probability proved valuable for detecting thin sands and mapping their spatial extension. The main challenge of the Cheleken fields is the presence of thin stacked sands that are difficult to interpret and to map on the seismic data. The statistical methods applied to interpret the inversion results allowed a better mapping of the thin sands across the volume. A new approach for reservoir property estimation based on a multi-attribute statistical transform was successfully tested to provide additional reservoir quality estimation.