SYSTEMATIC IDENTIFICATION OF SEQUENCE STRATIGRAPHIC UNITS FROM WIRELINE LOGS
GEOFFREY C. BOHLING, JOHN H. DOVETON, W. LYNN WATNEY, 1996. "SYSTEMATIC IDENTIFICATION OF SEQUENCE STRATIGRAPHIC UNITS FROM WIRELINE LOGS", Stratigraphic Analysis Utilizing Advanced Geophysical, Wireline and Borehole Technology for Petroleum Exploration and Production, Jory A. Pacht, Robert E. Sheriff, Bob F. Perkins
Download citation file:
Sequence stratigraphy has revolutionized geological ideas concerning the genesis and geometry of sedimentary packages. Consequently, sequence stratigraphic analyses have important genetic information that can refine geophysical interpretations in an analogous way that synthetic seismograms supply explicit geological data for seismic surveys. Unfortunately, because of the limited availability of core, sequence stratigraphic interpretations from well bores rely heavily on the use of wireline logs. Geologists have tended to apply vague generalizations and questionable criteria in their log recognition of bounding surfaces and genetic units that contrast starkly with more laterally extensive outcrop studies.
Contemporary logs record a wide variety of measurements controlled by mineral composition and geochemistry and can be used for systematic sequence stratigraphic predictions when they are properly calibrated to core. Links between units recognized in core to complex clouds of log data in multivariate space are the basis for predictions in logged but uncored wells. A procedure was implemented in computer software as a “shingled block lattice” classification, using concepts developed in robotics. Predictions are made using Bayesian probability methods that use log responses of the zone (“locus measures”), but also incorporate the unit identification of adjacent zones (“context measures”). The method is demonstrated in the log prediction of paleosols, flooding units, condensed sections, and late highstand deep water, peritidal, and subtidal carbonate units in a Kansas Pennsylvanian succession, using lithodensity-neutron and spectral gamma-ray logs. While the method is highly successful, the use of probability is advantageous in contrasting “obvious” units from those whose prediction is more ambiguous.