ABSTRACT

Sequence stratigraphic well log analysis combines specific gamma-ray/resistivity log relationships and biostratigraphic indices to identify depositional sequence and parasequence boundaries. Visual identification of these well log patterns relies on recognition of long wavelength (>100-1000 ft) variations in signal amplitude. Systems tracts and lithologic facies are distinguished based on more subjective "fining-upward," "coarsening-upward," "arcuate-convex," and "blocky" log patterns. We believe that systems tract and facies recognition can be improved by assessing variations in <200 ft wavelengths contained in well logs.

To test our thesis, several Gulf of Mexico deep-water siliciclastic gamma-ray logs (0.5 ft resolution) sampled at 0.5 ft and 1.0 ft intervals were subdivided using classical sequence stratigraphic analysis. The subdivided sections of the logs were processed using Fourier analysis to examine the importance of wave-lengths <200 ft. Calculated this way, wavelengths are comparable to bed thicknesses.

Fourier spectra in the study record striking, sequence-related differences in <200 ft bed-set thicknesses. For example, strong representations of "thin" bed-sets <3 ft to 6 ft occur in thin fining-upward and coarsening-upward parasequences. These dominant "thin" bed-set thicknesses record probable overbank facies. Where overbank facies transition upward into sandy prograding imbricate and amalgamated fan-lobe facies, log intervals are marked by "blocky" patterns. Such intervals, which include reservoir and reservoir quality wet sands in the test wells, contain both dominant "thin" (<10 ft) and dominant "thick" (ca. 40 ft to >200 ft) bed-sets. Spectral analysis of lithologic well logs can significantly improve stratigraphic well log evaluations in deep-water settings. Also of significance to petroleum reservoir modelers, dominant bed-set thickness data derived from Fourier spectra are directly applicable to the construction of more realistic digital reservoir flow models.