Reservoir Characterization Using Frequency-Enhanced Data: Case Study of the Eagle Ford Shale Formation for a field in South Texas
De Lilla, Daniel Ubaldo 1990-
MetadataShow full item record
The main goal of this study is to overcome certain limitations related to seismic data inversion due to the lack of information outside the seismic data’s frequency band. This study shows how we can greatly improve seismic inversion by using seismically-derived geological-features to guide the low-frequency well property interpolation. Using sparse-layer reflectivity inversion to extend the high frequencies at the other end of the spectrum, it is possible to obtain superior resolution of thin layers that translates into better volumetric definitions such as thickness and extent. The discussed methodology is applied to a seismic survey from the Maverick basin in South Texas, in particular for the Organic Eagle Ford formation. Using this data-driven approach provides a visual and measurable improvement on the modeled background impedances. The error between the generated low-frequency model and the filtered well log at validation well locations was reduced up to 12%. Furthermore, the data-driven method allows the background model to improve the correlation coefficient with validation logs up to 0.05 from 0.94 to 0.99. At the other end of the spectrum, the central frequency of the seismic data was enhanced from 30 Hz to roughly 60 Hz allowing the Organic Eagle Ford target to be resolved for a much larger portion of the survey, while averaging a correlation coefficient of over 0.87 between the seismic data and the high-frequency well-derived synthetics. The frequency-enhanced data allowed for a much better inversion result for both impedance estimations and localized analysis. Using the data-driven background model for seismic inversion, reduced the average impedance error by up to 5%, and in some cases it reduced the error up to 9%. The high-frequency enhancement provided localized estimations of impedance that were previously smeared across layers. After the seismic-inversion process on the frequency-enhanced data, the previously sub-resolution layers were resolved, and divided into independent impedance layers that allowed for a more precise analysis of the Organic Eagle Ford target. This will directly impact geometry definitions of the evaluated reservoir. The improvement provided by this method allowed for over 20% more accurate rock-property estimations in the organic-rich target. Ultimately, this study provides an approach to improve on seismically-derived acoustic-impedance results that were otherwise affected due to the band-limited nature of the data.