Recent major discoveries at Mad Dog, Atlantis, and Neptune have opened a major new hydrocarbon province in the Western Atwater Foldbelt of the ultra-deepwater Gulf of Mexico. The hydrocarbons are reservoired in good quality Lower and Middle Miocene turbidite sandstones. There is a clear stratigraphic signature expressed in the ultra-deep water, with the principal reservoir sands strongly partitioned into discrete, high net/gross, laterally extensive bodies, at or close to interpreted sequence boundaries. The upper part of the sequence is dominated by thin-bedded turbidites and mudstones.

Good quality biostratigraphic data allow us to correlate third and fourth order strati-graphic boundaries through all the wells across the region. Seismic data, south of the Sigsbee salt, for these deeper, older reservoirs provides resolution of the third order sequence boundaries. Compensation cycles have been mapped between the third order sequence boundaries, but it is not certain that these directly correspond to the fourth order sequences.

The observed stratigraphic hierarchy starts at second order cyclicity, demonstrated by a systematic change in the distribution of net sand from one third order sequence to the next. The highest resolution sequences that we have been able to map are fifth order. As expected, there is a systematic relationship between the stratigraphic hierarchy and the lateral correlation length of stratigraphic surfaces; ie. condensed sections associated with third order sequences can be correlated over greater lengths than the flood surfaces associated with higher order sequences. Third order sequence boundaries are mapped over a distance of in excess of a hundred miles. Fourth order boundaries have been correlated with confidence over a distance of at least 30 miles, and the fifth order correlation surfaces can only be mapped confidently within a single field, a distance of a few miles.

The nature of the flood surfaces changes with the position in the stratigraphic hierarchy and with the location relative to the sand feeder systems. In the core of the depositional system, we do not observe the foraminiferal marls that are characteristic of significant condensed sections elsewhere in the Gulf of Mexico. Our interpretation is that sediment is being supplied continuously to the basin floor in the form of sediment gravity flows and suspended sediment fall out from turbidity currents. As a result, third order flooding surfaces are characterized by high gamma shales, fourth order flooding surfaces are commonly associated with intervals of thin bedded turbidites; and fifth order flood surfaces have a variety of log signatures, and a ranges of facies association. Away from the focus of sediment input, foraminiferal marls are observed and some are interpreted to represent condensed sections.

The detailed high frequency sequence stratigraphy that has been carried through the three principal discoveries in the Western Atwater Foldbelt has impacted the exploration and development strategy for the area. For example, we have been able to demonstrate the presence of significant paleotopography that affected the lowermost Miocene reservoir distribution. We have proven that the thick sand bodies associated with the third order sequence boundaries are laterally extensive and so have reduced the risk of finding good quality reservoir sands in appraisal wells and nearby exploration wells. We have also demonstrated that not all flood surfaces are created equal and are therefore likely to have different transmis-sibility properties. This could have a significant impact on fluid flow within the reservoir and therefore medium and long-term field development.