3-D Seismic Attribute Study For Reservoir Characterization Of Carbonate Buildups Using A Volume-Based Method

Introduction And Geologic Overview The Upper Jurassic (Oxfordian) Smackover Formation is a stratigraphically complex carbonate formation, and a major producer of hydrocarbons in the northeastern Gulf of Mexico (Baria et al., 1982; Salvador, 1991). The main reservoir facies in the Smackover Formation are the microbial reefs, which were formed in a gently sloping to distally steepened carbonate ramp depositional setting (Benson et al., 1996). Although several studies have been carried out to describe and characterize physical properties of the Smackover reservoir interval in the northeastern Gulf of Mexico, their heterogeneity resulting primarily from their stratigraphic complexity is yet to be properly defined (Baria et al, 1982; Benson et al., 1996; Mancini et al., 2000; Parcell, 2000; Hart and Balch, 2000). In this study, we use a volume-based 3-D seismic attribute study to directly image rock physical properties (porosity) in the Smackover carbonate buildups of southwestern Alabama. Seismic attribute studies seek to find empirical correlations between seismic attributes and log-derived physical properties such as lithology, porosity, etc, through methods such as multivariate linear regression (MLR) and artificial or probabilistic neural networks (ANN/PNN; Schultz et al., 1994a&b; Russell et al., 1997; Hampson et al., 2001). By examining images derived using this volume-based method, it may be possible to deduce relationships between the predicting attributes and features of the reservoir that were not readily apparent from using a single data type. Finally, we demonstrate the possibility of using multiattribute results to foster an understanding of depositionally oriented trends in porosity distribution that have been observed in these buildups. The study area for this project encompasses Appleton Field (Fig. 1), a Smackover oil field of the basement ridge play in southwestern Alabama. Numerous studies have been carried out in an attempt to define the spatial distribution of depositional facies, the major factor controlling reservoir heterogeneity, and porosity within the Appleton Field (Mancini et al., 2000; Hart and Balch, 2000; Parcel, 2000). Facies heterogeneity and porosity need to be defined in 3-D space in order to optimize field production and development strategies. Fig. 1: Location Map of study area showing existing structural controls at time of Smackover deposition. Adapted from Mancini (2002, Fig. 1). Database And Methodology Sonic logs of 11 were used to generate synthetic seismograms in order to calibrate log and seismic data (Fig. 2). Six of these were chosen for multiattribute analyses. The subset of seismic data used consisted of an approximately 5 x 3.5 km grid of a post-stack, time-migrated 3-D volume, with a bin spacing of 165 x 165 ft (~50 x 50 m), and a 4 second two-way travel time (TWT) trace length. A data-driven approach was used as described by Schultz et al., (1994) and Hampson et al. (2001). A volume-based method using both multiattribute stepwise linear regression (MLR) and probabilistic neural network (PNN) statistical techniques has been adopted due to the thickness (Fig. 3; 80 230ft / 24 70m) and stratigraphic complexity (rapid facies changes) of this formation (c.f. Russell et al., 1997; Hampson et al., 2001). A probabilistic neural network was trained using the same set of predicting attributes derived from MLR to improve the quality of fit. This is because PNN is a pattern recognition tool (c.f. Liu and Liu, 1998) and so may better capture nonlinear relationships between the attributes and log porosity than MLR. Exclusion testing was carried out to test the effectiveness of the statistical relationship in areas of sparse well control.