Gypsy Field Project in Reservoir Characterization

The objective of the Gypsy Project was to properly calculate seismic attributes and integrate these into a reservoir characterization project. Significant progress was made on the project in four areas. (1) Attenuation: In order for seismic inversion for rock properties or calculation of seismic attributes used to estimate rock properties to be performed validly, it is necessary to deal with seismic data that has had true amplitude and frequency content restored to account for earth filtering effects that are generally not included in seismic reservoir characterization methodologies. This requires the accurate measurement of seismic attenuation, something that is rarely achieved in practice. It is hoped that such measurements may also provide additional independent seismic attributes for use in reservoir characterization studies. In 2000, we were concerned with the ground truthing of attenuation measurements in the vicinity of wells. Our approach to the problem is one of extracting as time varying wavelet and relating temporal variations in the wavelet to an attenuation model of the earth. This method has the advantage of correcting for temporal variations in the reflectivity spectrum of the earth which confound the spectral ratio methodology which is the most commonly applied means of measuring attenuation from surface seismic data. Part I of the report describes our efforts in seismic attenuation as applied to the Gypsy data. (2) Optimal Attributes: A bewildering array of seismic attributes is available to the reservoir geoscientist to try to establish correlations to rock properties. Ultimately, the use of such a large number of degrees of freedom in the search for correlations with limited well control leads to common misapplication of statistically insignificant results which yields invalid predictions. Cross-validation against unused wells can be used to recognize such problems, but does not offer a solution to the question of which attributes should be used to make predictions other than by exhaustive trial and error. Alternatively, we use the approach of using principle component analysis to reduce the seismic data to a minimum number of significant attributes that explain the variation in the data. These are then correlated to rock properties in order to make predictions. Part II of the report describe our efforts in optimal attributes as applied to the Gypsy data. (3) High Resolution 3D Seismic Processing: When faced with the issue of testing the above methods on the Gypsy dataset, we realized that the 3D seismic data were not processed well and exhibited poor ties to well control. The data was reprocessed with surface consistent predictive deconvolution, muting of wide-angle reflections, min/max exclusion stacking, and F-XY deconvolution. After reprocessing, a good character match with synthetic seismograms was observed. This work was presented at the 2001 SEG Annual Meeting and is included as Part III of this report. (4) Reservoir Characterization Education: The Gypsy project has provided the data for a reservoir characterization module which was added to Depositional Systems and Stratigraphy, a course required for majors in Geology and Geophysics. This module is important because it introduces students to the relevance of sedimentary geology to applied, real-world problems. This work was presented at the Geological Society of America annual meeting (Part IV) and is described on the website for the course (Part V of the report)