Instantaneous Spectral Attributes to Assist 3-D Seismic Interpretation

Jianlei Liu, Kurt J. Marfurt, and Charlotte Sullivan
University of Houston, Department of Geosciences, Houston, TX

Spectral decomposition is widely used to predict reservoir net thickness and assist seismic stratigraphic patterns interpretation. Instantaneous spectral attributes computed by wavelet based time-frequency decomposition give better resolution for the study of frequency anomalies and to detect subtle geological structures. Instead of using short window Fourier transform to do spectral decomposition, wavelet based time-frequency decomposition combines the matching pursuit concept and least square solution to give the best matches between the original seismic trace and modeled trace. The final common frequency sections can be used to detect low frequency shallow zones below gas reservoirs. The comparison of time slices at different frequencies may indicate thin bed thickness variation. Composite plots of peak frequency and peak amplitude may illuminate thin bed channels. From 2-D wedge models, higher peak frequency indicates thinner beds while lower peak frequency represents thicker beds.

Data from a 3-D seismic survey in the Fort Worth Basin are used to test this concept. The Caddo limestone at around 750ms is considered as a thin bed reservoir, which is verified by 25 well logs. Instantaneous frequency and spectral attributes are generated along the Caddo limestone horizon. The map view of instantaneous spectral attributes and instantaneous frequency shows thickness variations. Time slices of peak frequency images tectonic and karst collapse features. The combined interpretation of amplitude, coherence and instantaneous spectral attributes illustrates the details of subtle geological structures within this seismic volume.