Deep-water turbidite fans: anatomy and reservoir architecture

The anatomy of a fan

Normark (1978) and Mutti and Normark (1987) described fans by their architectural elements, in the order a turbidity current meets them on its way down:

• Feeder canyon. An erosional valley cut into the shelf edge and upper slope. Flows pass through it with little deposition; what stays is a coarse lag, slumped wall material and, once the canyon is abandoned, mud.
• Slope channel and channel–levee complex. Below the canyon the channel builds its own banks. Flows are often thicker than the channel is deep, so their upper, finer part spills over and builds levees: wedges of thin-bedded sand and mud that thin and fine away from the channel. The sand stays in the channel axis, and successive channels stack into a belt between the levees.
• Channel–lobe transition. Where the levees die out, the flow loses its confinement, spreads and slows. Scours and surfaces the flows bypassed mark this zone, and most of the sand is laid down just beyond it.
• Depositional lobes. Sheet-like sand bodies spread from the channel mouth, thick and sandy along their axis and thinner, finer and muddier toward their off-axis areas, fringes and distal fringes. Prélat et al. (2009) found a hierarchy in the Karoo Basin: beds stack into lobe elements, lobe elements into lobes, lobes into lobe complexes, and both lobe elements and lobes offset one another.
• Distal fringe. Beyond the lobes, thin beds of very fine sand and silt alternate with mud and pinch out onto the basin floor.

The beds inside: the Bouma sequence

A single turbidite, the bed one flow leaves, fines upward through the divisions Bouma (1962) described: $T_{\mathrm{a}}$, massive or graded sand; $T_{\mathrm{b}}$, parallel-laminated sand; $T_{\mathrm{c}}$, ripple cross-laminated fine sand; $T_{\mathrm{d}}$, parallel-laminated silt; and $T_{\mathrm{e}}$, mud. Few beds show all five, and down the fan they change in a regular way:

• Channel fills and lobe axes: thick, coarser beds that are mostly $T_{\mathrm{a}}$ and often cut into one another (amalgamated), so sand rests on sand with no mud between.
• Off-axis lobe: medium beds that keep more of the sequence, $T_{\mathrm{a}}$ to $T_{\mathrm{e}}$.
• Levees and fringes: thin beds that begin at $T_{\mathrm{b}}$ or $T_{\mathrm{c}}$, their base cut out, each separated from the next by mud.

The same thin, rippled beds form both a levee and a distal fringe, so a bed alone does not tell you where you are. The geometry does: a levee is a wedge beside a channel, a fringe a sheet far from one.

Exercise, walk the fan

• Start with plate (a). The canyon is cut into the hatched slope, the channels radiate from its mouth with the youngest drawn darkest, and each channel ends where its lobe begins. The lobes are numbered by age: no lobe sits where the one before it did.
• Read (b), section A–A′, 10 km onto the basin floor. Seven channel fills stack in one narrow belt between levees of thin beds and mud; an eighth, older, lies 6 km to the side, where its channel had already turned toward its own lobe, and a lobe laid down when the fan was young lies beneath them. Set the table to A–A′ and put the well in the belt, at 33 km: it crosses 85 m of sand in 109 m. Move it onto a levee, at 42 km, and it crosses 5 m in 35 m. Only 26% of the section is sand: the best single well is here, but the levees fill most of the section.
• Read (c), section B–B′, at 35 km. Five lobes spread across it, each shifted sideways from the one before and each thickest and sandiest along its axis, and the last reaches of three channels cut them. Half the section is sand (51%), the most of the three, and 82% of that sand lies in one connected body, joined where the lobes’ axes eroded the mud between them.
• Read (d), section C–C′, at 70 km. Only the thin fringes of the lobes arrive: 9% of the section is sand, in beds about 3 cm thick, each sealed by mud, and none of it connects.
• Look at (e). Net-to-gross and connected sand rise from the channel belt into the lobes and fall toward the fringe, while sand-on-sand contacts fall steadily away from the feeder. Same fan, same flows: what differs is where each flow lost its confinement and how far it had spread.

Compensational stacking

A lobe builds relief: its axis stands metres above its fringe. The next flow follows the lowest path around it and fills the low beside it, so successive lobes are offset sideways rather than piled on one spot. The pattern repeats at each level of Prélat’s hierarchy, lobe elements within a lobe and lobes within a lobe complex. In Figure 4.5 each new lobe is placed exactly this way, at the lowest part of the fan surface, and nothing else steers it.

Implications:

• Offset stacking. A vertical well crosses only some of the lobes that make up a fan, and a neighbouring well crosses others.
• Amalgamation at the axes. Where a younger lobe’s axis erodes the thin mud above an older lobe, the two sands touch; toward the fringes the mud survives and separates them. Mud between lobes is a baffle or a barrier only where it is preserved.
• Heterogeneity. Each lobe has its own thickness, grain size and extent, so neighbouring lobes can produce differently.
• Prediction. Compensation is not random: a low between two mapped lobes is where the next one is likely to lie.

Seismic recognition

• Channel fills. High-amplitude reflections in a narrow belt, often stacked or offset, with erosional bases that cut the reflections below.
• Levees. Low-amplitude reflections that diverge toward the channel and thin away from it: the gull-wing shape that marks a channel–levee complex.
• Lobes. Sheet-like or gently mounded high-amplitude reflections, continuous for kilometres, whose tops step sideways where the lobes compensate.
• Distal fringe. Thin, parallel, low-amplitude reflections that fade as they onlap the basin floor; many fringe beds lie below tuning and show only as a weak composite.
• Polarity. The sign of the top-sand reflection depends on whether the sand is softer or harder than the mud around it. In young, shallowly buried deep-water basins the sand is usually softer, so the top of a sand is a trough in SEG normal polarity, and hydrocarbons make it brighter. Figure 4.5 uses such a soft brine sand.
• Resolution. A body thinner than about a quarter wavelength, $V/(4f)$, shows its top and base as one composite event. At 30 Hz in sand at 2250 m/s that is 19 m: channel fills and thick lobes resolve, thin lobes and fringes merge.
• Amplitude extractions. A horizon amplitude map over the fan shows the channels as bright ribbons and the lobes as offset bright patches, the plan view of compensational stacking.

Why fans matter as reservoirs

• Sorted sand. Turbidity currents carry sand far from the shelf and drop it in channel fills and lobes, sorted by the flow.
• Reservoir quality. Many deep-water sands are young and lie beneath a near-freezing seabed, so at a given burial depth they are cooler than a sand onshore and have seen less quartz cement. Porosities of 20 to 30% and permeabilities from hundreds of millidarcies to a few darcies are common in clean channel and lobe-axis sand.
• Scale. A fan can cover hundreds to many thousands of square kilometres.
• Seal. Hemipelagic mud drapes the fan and seals it above and at its pinch-outs.
• Source. Organic-rich deep-water mudstones often lie beneath or beside the fan, so migration paths are short.
• Imaging. Wide-azimuth and ocean-bottom surveys and reverse-time migration have made fans beneath salt and in deep water far easier to image than they were two decades ago.

Exploration workflow for deep-water fans

• Identify the fan on regional seismic by its lobate outline and by the radiating channels and lobes of an amplitude extraction.
• Locate the prospect on the fan: in the channel–levee complex, among the lobes, or in the fringe.
• Predict the architecture from that position: a belt of channel fills between levees, offset lobes, or thin beds.
• Estimate the volume: gross rock volume × net-to-gross × porosity × hydrocarbon saturation, with the net-to-gross of the element you expect rather than a fan-wide average.
• Apply rock physics (Part 5): AVO and elastic inversion to separate sand from mud and predict the fluid.
• Risk the prospect: charge, reservoir (position and element), seal, trap and migration. For a stratigraphic trap in a fan the updip seal comes first, because the fan’s sands connect updip to its feeder: an abandoned, mud-filled channel or a pinch-out against the slope must close it.
• High-grade the locations with the thickest net sand, a channel-belt axis or a lobe axis, inside a closure whose seal holds.

Pitfalls

• Assuming uniformity. Apply the net-to-gross of a channel belt or a lobe axis to a whole fan and you can overestimate its reserves several times over.
• Confusing mass-transport complexes with fans. Mass-transport complexes are chaotic, poorly sorted and often mud-rich; fans are ordered and layered. Tell them apart before committing to a fan interpretation.
• Undercounting thin beds. Levee and fringe beds are often thinner than the resolution of the logs, so their sand reads as shaly and their pay as low resistivity. A laminated-sand evaluation finds what a conventional cutoff misses.
• Ignoring compensation. Two wells a few kilometres apart can cross different lobes. Map lobes individually wherever the data allow.
• Ignoring differential compaction. Mud compacts more than sand, so the beds above a sand-rich fan drape over it and can form a closure directly above the sand body.