Abstract In spite of the simplistic submarine-fan models (emphasizing turbidite channels and lobes) that have dominated the literature since the late 1960’s, deep-water processes and their deposits (facies) are quite complex. Recognition of deep-water facies using detailed process sedimentology is critical in reservoir characterization because depositional processes are the primary controls on the dimensions, geometries, and ultimately the quality of deep-water reservoirs. Classification of sediment-gravity flows into Newtonian flows (turbidity currents) and plastic flows (debris flows), based on fluid rheology and flow state (turbulent and laminar), is still the most practical and meaningful approach. This is because the boundary between Newtonian and plastic flows can reasonably be established using sediment concentration values of about 20 to 25 percent by volume. In general, low-concentration, sandy turbidity currents tend to emplace fan-shaped deposits of finegrained sand in unconfined environments by suspension settling. In contrast, high-concentration, sandy debris flows tend to emplace tongue-shaped deposits of fine- to coarse-grained sand in unconfined environments by freezing. The tongue-shaped deposits of sandy debris flows may be surrounded by fan-shaped deposits of muddy turbidity currents because of surface-flow transformation of basal laminar flows into upper and frontal turbulent flows. Sandy debris flows are considered to be the dominant process transporting and depositing reservoir sands into the deep sea. The concept of “high-density turbidity current” is confusing because the high density (that is, high sediment concentration) of these flows tends to damp the turbulence, the very property that defines turbidity currents. If the existing three turbidite models (Bouma Sequence, Lowe Sequence, Stow-Shanmugam Sequence) are meaningful, then a complete turbidite bed should contain a total of 16 divisions. However, no one has ever documented such a complete turbidite bed. The plethoric family of “traction carpets” has proliferated into nine models (flowing-grain layers, inertia-flow layer, laminar sheared layer, fluidized flowing grain layer, avalanching flow, etc.). In spite of this multiplicity of models, many fundamental problems still remain in recognizing deep-water facies. Recognition of units deposited by deep-water bottom currents (also known as contour currents) is difficult. Bottom-current-reworked sands are thin and lenticular at core scale but may also exhibit sheet-like geometry on seismic scale. Submarine-fan models with turbidite channels and lobes have controlled our thinking for nearly 30 years, but many of us now know that these models are obsolete. The suprafan lobe concept was influential in both sedimento-logic and sequence stratigraphic circles because it provided a basis for constructing a general fan model and for linking mounded seismic facies with sheet-like turbidite sandstones. However, this concept recently was abandoned by its proponent on the grounds that a suprafan lobe is not a discrete mappable unit. This has left the popular sequence stratigraphic fan models, based primarily on seismic geometries, with a shaky foundation. Although depositional processes cause the development of different seismic geometries, the notion that depositional facies can be inferred from seismic geometries is not fully supported because a single facies (sandy debris flows) can generate multiple seismic geometries (mounded/bidirectional downlap, mounded/hummocky, mounded/chaotic, sheet/parallel-continuous, etc.) and multiple wireline log motifs (upward-fining, upward-coarsening, and blocky), as seen in examples from the North Sea, Gulf of Mexico, and Equatorial Guinea. As a counterpart to turbidite-dominated fan models suited for basinal settings, a slope model is herein proposed that is representative of debris-flow-dominated systems. The strength of this model is that it includes a variety of processes, such as slumping, debris flows, turbidity currents, and bottom currents, that are common to the slope settings. Deposits of sandy debris flows, analogous to turbidite fan deposits, are capable of developing sheet-like geometries in the rock record. The conventional notion that sandy debris-flow reservoirs do not have good reservoir properties is not true because the lower Eocene sands of the Frigg Formation (Frigg field, Norwegian North Sea), which are interpreted to be of sandy-slump and sandy-debris-flow origin, exhibit extremely high porosities (27 to 32 percent) and permeabilities (900 to 4,000 mD). In contrast, sands deposited from turbulent turbidity currents in deep-water environments are poorly sorted and include large amounts of silt and clay. In the 21st century, a paradigm shift is in order. This shift will involve the emergence of a new paradigm that will be more inclusive in terms of slope processes and products than just basinal turbidity currents and fan models. Science is a journey, whereas facies models are the final destination.

ABSTRACT The Pliocene-Pleistocene sequence cored in the Ewing Bank 826 Field in the Gulf of Mexico provides an example of sand distribution and reservoir quality produced by reworking by deep-marine bottom currents. A distinctive attribute of reworked sands is their traction bedforms. Common sedimentary features of traction currents include small-scale cross-bedding, starved current ripples, horizontal lamination, sharp upper contacts, and inverse size grading. The sands also exhibit internal erosional surfaces and mud-offshoots indicating oscillating current conditions. Presumably, the Loop current, a strong wind-driven surface current in the Gulf of Mexico, impinged on the sea bottom, as it does today, and reworked sand. A depositional model based on the integration of core, wireline log, and 3-D seismic data suggests that the reworked sediment package may be thick and continuous, but individual sand layers within the package may be thin and discontinuous. This model, which depicts the distribution of bottom-current reworked sand in interchannel slope areas as a distinctly different facies from channel-levee facies, has the potential for general application to other deep-water plays outside the study area. In the Ewing Bank 826 Field, the Type 1 (L-l) reservoir with 80% sand exhibits higher permeability values (100-1800 mD) than the Type 2 (N-l) reservoir with 26% sand (50-800 mD). The increased permeability in the Type 1 sand has been attributed to high sand content, vigorous reworking, and microfractures. The clean, porous and well-sorted Type 1 sands with good communication between sand layers have produced at higher rates and recovery efficiencies than the Type 2 sands with numerous interbedded mud layers.