Abstract

The Red Fork sandstone is a channel system within the Anadarko Basin in west-central Oklahoma. It was deposited throughout the Desmoinesian and contains multiple target zones. Red Fork channels have earned their description of being “invisible” over many years of dry holes. Even after surprise production out of the Red Fork is encountered, attempts made to go back to seismic and track channel systems have been unsuccessful. Why the Red Fork channels are so difficult to detect is the driving force behind this study.

Attributes are a valuable tool used by interpreters to extract information contained within seismic traces. Often the information extracted by attributes is too small to be recognized by the unaided human eye. Specific attributes such as sobel filter and spectral decomposition have been used with success in the past to identify channel systems, and the recent development of enhanced curvature algorithms give a positive outlook to Red Fork channel detection through attributes.

Our attribute-focused study was aided by other methods as well. The possibility of finding an empirical relationship for differentiating between sand-filled and shale-filled channels led to a petrophysical analysis of rock properties extracted from logs. The trends uncovered in that analysis were enhanced by comparing results from a non-producing Red Fork well to a producing Red Fork well. Amplitude variation as a result of the stratigraphic content was a question that the petrophysical comparison strove to answer.

The results of the petrophysical research suggested that a seismic inversion would be helpful in differentiating bed lithologies, specifically, the 40-ft-thick producing Red Fork sandstone interval from Red Fork shale intervals identified in gamma-ray logs from producing wells. By overlaying the gamma-ray logs over the model-based impedance inversion, some correlation was made linking the gamma-ray-identified sandstone with a small but visible low-impedance feature. This interpretation led to the generation of a horizon, made with difficulty, tracking the correct interval. This horizon was then used to extract volumetric attributes.

Our results aim to explain possible reasons why the detection of invisible Red Fork channels has remained difficult at best, and suggest tools that might be valuable in overcoming this issue in the future.