Deep-Water Sediment Bypass

Submarine sediment density flows (herein referred to as submarine flows) comprise a range of processes from dilute turbidity currents to cohesive debris flows (Bouma 1962; Lowe 1982; Mulder and Alexander 2001; Meiburg and Kneller 2010; Talling et al. 2012). They are the principal mechanism for transporting and bypassing large volumes of sediment from the continents to the deep ocean. Sediment bypass concerns the transport of sediment by a flow beyond a fixed area or geographical point. The bypassed sediment, which may include substrate material entrained during the passage of the flow, is subsequently deposited farther down-system. The deposits of submarine flows, turbidites, debrites, and hybrid beds are relatively thin (decimeters to meters) and can be areally extensive (100s km²). Thus, flows must inherently bypass a significant proportion of their sediment load down-system of a fixed geographical location along the flow pathway. This is a fundamental principle behind deep-water facies tract models, which predict sequential deposition of particular facies and bed thicknesses along a flow pathway (Bouma 1962; Lowe 1982; Mutti 1992; Sumner et al. 2012) (Fig. 1A). For example, a hypothetical flow depositing a 1-m-thick bed along a 10 km 2D tract deposits only 0.0001 fraction of its initial sediment load every meter, and bypasses orders of magnitude more sediment at all but its most distal locations. The amount of sediment bypassed is dictated by the longitudinal (Fig. 1B) and vertical (Fig. 1C) structure of a flow, which evolves with distance along the flow pathway (Fig. 1A, B; Points X, Y, and Z). Paola and Martin (2012) have formalized this principle at the system scale by applying mass-balance calculations to ponded deep-water successions, described from seismic datasets, to estimate the proportion of deposition versus bypass at particular locations along the sediment pathway.

Underfilled modern submarine channels with morphologically related submarine fans demonstrate that significant volumes of sediment are bypassed down the continental slope. Amalgamated coarse sands and gravels dominate the channel thalwegs and fine-grained internal levee and terrace deposits occur over their banks (e.g., Paull et al. 2005; Babonneau et al. 2010). Similar facies are found within the axes of ancient channel-fill successions, interpreted to represent the depositional record of large volumes of sediment bypassed into deeper water (e.g., Winn and Dott 1977; Mutti and Normark 1987; Pickering et al. 2001; Hodgson et al. 2011; Hubbard et al. 2014). In addition, fine-grained thin-bedded channel-base drapes found in ancient channel fills have been interpreted to represent high-energy, largely bypassing flows that only deposited material from their tails (Mutti and Normark 1987; Macauley and Hubbard 2013). Farther down the slope profile at the base of slope, channel–lobe transition zones (CLTZs) have been shown to be bypass-dominated areas in modern systems (Normark 1978; Normark et al. 1979; Wynn et al. 2002a) and ancient systems (Mutti and Normark 1987; Ito et al. 2014; van der Merwe et al. 2014). CLTZs are characterized by a range of scour types draped by fine-grained sandstones and mudstones and laterally discontinuous cross-laminated sandstone beds (Cazzola et al. 1981; Mutti and Normark 1987; Chapin et al. 1994; Palanques et al. 1995; Morris et al. 1998; Wynn et al. 2002a; Gardner et al. 2003; MacDonald et al. 2011; Ito et al. 2014). Deposits of dune-scale cross-laminated sandstones have been interpreted to record sediment bypass in basin-floor settings (Amy et al. 2000). The presence of sharp grain-size breaks in individual beds has also been considered as an indicator of sediment bypass (Gladstone and Sparks 2002; Kneller and McCaffrey 2003).

Identification and examination of sediment bypass zones provide insights into the dynamics of submarine flows (Fig. 1), which improves our understanding of a key process by which sediment is moved across the planet. The timing and location of sediment bypass is important in deep-water systems because this process ultimately dictates the extent and thickness of the associated deposits (Fig. 1). Hence, from an applied perspective, recognizing the degree and caliber of sediment bypassed through any particular region is important for the down-system prediction of sandstone bodies in hydrocarbon plays and may also aid geohazard assessments for seafloor infrastructure. Despite many studies that describe or interpret areas of deep-water sediment bypass, the phenomenon, and recognition criteria, have not been systematically described. Here, we bring together theoretical and numerical approaches that mathematically describe the processes that govern bypass. We then synthesize the expression of bypass in the geological record from modern seafloor, outcrop, core, and subsurface datasets. Examining these different types of field data allows the stratigraphic and planform expression of bypass to be described and interpreted across a broad range of scales (centimeters to 100s km; Fig. 2). From these data we address the following questions: (i) what are the key factors that influence sediment bypass?; (ii) What recognition criteria can be established to identify sediment bypass in the stratigraphic record? (iii) How do different zones of sediment bypass vary in character? (iv) Is it possible to quantify the volume of sediment bypassed?

The diversity of specialists working on deep-water systems means that similar terms are often used to mean different things. Such miscommunication is particularly evident between those using theoretical, experimental, or numerical approaches to understand sediment transport and deposition, and those charged with interpreting these processes from the geological record. We therefore outline the key terms we use herein to describe sediment bypass, drawing on input from both process and stratigraphic perspectives:

1. Bypassing flows transport their entire sediment load in suspension or traction, resulting in downslope sediment transfer with no depositional record at the geographic point or area of measurement.

2. Partially bypassing flows deposit a diminutive proportion of their sediment load and transport the remainder downslope, typically resulting in relatively thin deposits across the zone of sediment bypass. This definition covers a broad range of flow states from flows bypassing almost their entire sediment loads across the area of bypass, to more strongly depositional flows that generate thicker deposits across the area of bypass.

3. Depositional flows are in their most depositional state along their flow path, resulting in relatively thick deposits at the area of measurement compared to other locations along the flow pathway.

Sediment bypass zones are an expression of the balance between sediment transport and deposition, and substrate entrainment by numerous submarine flows along a tract. The following stratigraphic terms are employed herein:

1. Complete sediment bypass zones are associated with bypassing flows, resulting in no deposition. The stratigraphic record of this process may be negligible (cf. Stevenson et al. 2013) or solely manifest by erosion surfaces that were generated from the bypassing current (cf. Hubbard et al. 2014).

2. Bypass-dominated zones are associated with bypassing and partially bypassing flows. The stratigraphic expression may be composite and complex, recorded by no deposition and erosion (as above), or with thin-bedded deposits.

3. Depositional zones are dominated by depositional flows, resulting in relatively thick deposits. This term is used in a relative sense to distinguish areas or intervals where deposits are thickest compared to other areas of the sediment flow pathway.

Given the dynamic nature of deep-water systems, these bypass zones are likely to be transitional and transient, and may evolve from one type to another in time and space.

Previous studies have not attempted to link deposits interpreted to be products of sediment bypass with processes of sediment transport and deposition simulated in theoretical, numerical, and experimental models. Consequently, our understanding of the physical parameters that influence the timing, location, and stratigraphic expression of bypass is limited. Here, the processes that drive sediment bypass in submarine flows from theoretical, numerical, and experimental approaches are examined.

A number of theoretical and numerical approaches link the properties of submarine flows with their ability to bypass sediment. We outline these approaches and highlight four common influential parameters: grain size, sediment concentration, slope, and flow size. These common parameters provide a framework to better understand the processes governing sediment bypass in deep-water systems.

Rouse (1937) presented a landmark solution for the distribution of suspended sediment in a shear flow, whereby a nondimensional number balances the shear velocity of the transporting flow with the settling velocity of sediment:

where, v_s is the settling velocity of sediment particles [m/s], κ is the von Kármán constant [0.41], and u* is the shear velocity [m/s]. There is no value for Ro that signifies a threshold for suspension, but significant suspension occurs at Ro <∼ 2 or u*/v_s> ∼ 1, with suspension of grains in the lowest 10% of free surface flows at Rouse numbers as high as 5 (u*/v_s  =  0.5; van Rijn 1993). This simple criterion considers the main controls on sediment bypass to be sediment concentration, slope, and flow size. Of these, concentration, flow size, and slope are the three main variables governing the shear velocity, which can be approximated by:

where g is the acceleration due to gravity [m/s²], ρ is the density [kg/m³] with the indices indicating the interstitial and ambient fluid (i) and bulk flow (f), H_r is the hydraulic radius of the flow [m], and S is the slope. In the Ro criterion, these three parameters have a positive correlation with sediment bypass. In contrast, grain size is the main variable controlling settling velocity (v_s), which has an inverse relation with sediment bypass (Eq. 1).

Leeder et al. (2005) state that Ro expresses the ability of flows to suspend and bypass grains of a certain size (i.e., flow competence) but does not provide information on the amount of sediment that can be kept in suspension (i.e., flow capacity). Hence, Leeder et al. (2005) propose a dynamic suspension criterion that does not contain grain size as a term, instead relating the lift of the vertical Reynolds stress component τ_vv to the downward gravitational forces on the sediment (Bagnold 1966; Leeder 1983; Leeder et al. 2005):

where m is the suspended mass of sediment per unit bed area [kg/m²] with ρ_s the density of solid particles in the flow. With respect to Equation 3 of Leeder et al. (2005), the densities have been reorganized to make Λ dimensionless. Flow size and concentration are precisely cancelled out due to their relation with u* and m, leaving slope (i.e., u* via Eq. 2) as the main controlling variable on sediment bypass.

Pantin and Franklin (2009) investigated the energy budget of a developing flow under the simplifications of averaging the mass, volume, momentum, and turbulent kinetic energy (TKE) over the depth of the flow. From a four-equation model (Pantin 1979; Parker 1982; Parker et al. 1986) they isolate the spatial change in TKE to yield a criterion that can be interpreted to signify diminishing turbulence and depletion of the sediment in suspension for negative values, and increasing turbulence and erosion of sediment for positive values:

where, e_w is the water entrainment coefficient [-], Ri is the bulk Richardson number (Ri  =  g(ρ_f/ρ_i – 1)H/U², H is the flow height [m], U is the average velocity [m/s], K is the TKE per unit mass [m²/s²], and ε₀ is the rate of viscous dissipation of turbulence [m²/s³]. For subcritical flow (Ri > 1) the first term is negative, and the only remaining positive term is the drag coefficient (u*²/U²; of order 10^–3 to 10^–2). In such flows the LHS of Equation 4 is always negative, meaning diminishing turbulence and flow-power downstream. Under supercritical flow conditions (Ri < 1), this criterion develops a division in parameter space between depositing and eroding flows, “the ignition condition.” This is an unstable bypass situation, meaning that small perturbations will grow and the flow will become either depletive with decreasing flow power downstream or erosional with increasing flow power downstream. Grain size appears in the 5^th term, where larger grains will again make the flow less prone to bypass. Flow size, sediment concentration, and slope have a presence in all other terms (e_w, Ri, u*, U, K, ε₀, and H), and the variety of their occurrence makes Equation 4 challenging in terms of understanding how individual factors influence sediment bypass.

“Direct numerical simulation” (DNS) has been used to probe the effect of vertical gradients in the concentration of settling sediment (“stratification”) on turbulence suppression (Cantero et al. 2009; Cantero et al. 2012a; Cantero et al. 2012b; Cantero et al. 2014). According to these contributions, turbulence in suspensions can be maintained when

where the subscript τ denotes the shear Richardson and Reynolds flow scales:

where C is the sediment concentration. The settling velocity (v_s) in Equation 5 has been made nondimensional by

and the weak dependence of the turbulence threshold value on the Reynolds number has tentatively been extended far beyond the reach of DNS simulations as (Cantero et al. 2012b):

The experiments show that turbulence at the base of the flows is rapidly and completely extinguished if the term on the LHS of Equation 5 becomes larger than 1. Although sedimentation is not actually modeled, the inference of a simulation with no turbulence at the base is that sediment suspension cannot be sustained, and that such flows would rapidly deposit their sediment loads. Equation 5 essentially combines a Rouse-type criterion for sediment suspension with a Richardson scale of the flow. Since the latter contains the total amount of sediment in suspension (Eq. 6), this criterion unifies both the competence for suspension and the capacity to suspend a specified amount of sediment. Again, all four parameters of grain size, sediment concentration, slope, and flow size appear in the criterion.

Attempts have been made to test the criteria discussed above individually, with variable success (van Rijn 1984; van Rijn 1993; Pantin 2001; Pantin and Franklin 2011; Cantero et al. 2012b), but none of these entail a comparison with flows that are able to actually bypass sediment. Cartigny et al. (2013) present a series of scaled experiments of high-concentration sand-laden flows from either side of the bypass condition, i.e., depositional or bypassing. The criteria outlined above do not predict the bypass condition in these experiments (Table 1). However, the stratified-suppression criterion produces bypass threshold values that are very close to the experimental observations (within estimated error). This comparison highlights that, although process models of bypass incorporate common factors of grain size, sediment concentration, slope, and flow size, there is not yet a consistent framework of experiments, theory, and numerical models that accurately predicts sediment bypass. Four approaches have been discussed here, and only the stratified-suppression threshold approach yields encouraging predictions of the bypassing nature of sand-laden experimental flows.

The modern seafloor provides a wealth of information on deep-water clastic systems through high-resolution mapping and extensive coring, which has enabled their planform geometry and associated depositional architectures to be constrained in considerable detail over large areas; particularly well-studied systems include the Amazon Fan (e.g., Damuth and Kumar 1975; Damuth et al. 1983, 1988; Flood 1987; Pirmez and Flood 1995), Navy Fan (e.g., Normark et al. 1979; Piper and Normark 1983), the Mississippi Fan (Garrison et al. 1982; Kastens and Shor 1985; Twichell et al. 1991, 1992), the Zaire Fan (Droz et al. 1996; Savoye et al. 2000; Babonneau et al. 2002, 2010) and the Bengal Fan (Curray et al. 2002; Schwenk et al. 2003, 2005; Kolla et al. 2012). Surface mapping generally considers decimeter scale stratigraphy of years to 100s years; near-surface mapping commonly captures 10s of meters of stratigraphy of 1 to 100s kyr; shallow and deep drill cores can provide centimeter scale stratigraphic records over 100s of meters representing millions of years of sedimentation (Normark et al. 1979; Mutti and Normark 1987; Piper and Normark 2001). The high level of planform detail afforded in these modern deep-water systems allows anatomical links to be made between proximal and distal areas. This aids interpretation of sediment bypass zones by linking them down-system with areas of sediment deposition. For example, the Congo Channel extends across the Zaire Fan for 760 km (Babonneau et al. 2002). Within the channel axis, > 5 m of amalgamated, structureless sands are present (Babonneau et al. 2010). The channel passes down-system into a submarine lobe complex, the youngest of which is sand-rich (Savoye et al. 2000), suggesting that the modern channel is an efficient conduit for sediment transport (Fig. 3A). Amalgamated structureless sands in the channel axis can therefore be attributed to partially bypassing flows, where coarse-grained hyperconcentrated near-bed layers deposit the structureless sands, whilst the finer-grained upper parts of the flow are bypassed down-system (Lowe 1982).

However, despite good anatomical links between different parts of the Zaire Fan system and a speculative association between structureless sands and sediment bypass, high-resolution stratigraphic links between the deposits found in the channel and sand bodies documented farther down slope are difficult to constrain. Hence, it is not demonstrable that specific sedimentary facies or stratigraphic surfaces found in the channel actually relate to specific sand bodies found down slope (Fig. 3B). External levees adjacent to modern submarine channels contain a coeval and more complete depositional record. For example, the currently inactive Amazon Channel has thick channel–levee deposits that border and confine the channel system for almost its entire length (∼ 800 km), and yet the channel thalweg itself is underfilled (Hiscott et al. 1997).

An improved understanding of deep-water sediment bypass has been achieved through correlations that connect up-system areas of sediment bypass with strata deposited farther down-system in the Moroccan Turbidite System (Fig. 4A; Wynn et al. 2002b; Talling et al. 2007; Stevenson et al. 2014a). Correlating individual beds along their depositional extent makes it possible to estimate total sediment volumes, as well as the amount of sediment bypassed at various locations along the system (Fig. 4B). From cores, the sedimentary facies associated with areas dominated by sediment bypass can be examined. The deposits of one submarine flow (Bed 5; Fig. 5) reveal: 1) the general sedimentological and stratigraphic expression of two distinct bypass-dominated zones in the Moroccan Turbidite System, and 2) how grain-size breaks within the bed can be used as an indicator of sediment bypass in basin-floor deposits. The Bed 5 event occurred at ∼ 60 ka and most likely originated from the Moroccan Margin in the head of the Agadir Canyon (Wynn et al. 2002b). The flow deposited approximately 52 km³ of sediment (21 km³ sand and 31 km³ mud) along an ∼1500-km-long pathway from the Agadir Canyon, across the Agadir Basin, through the Madeira Channel System, and onto the Madeira Abyssal Plain (Fig. 4) (Wynn et al. 2002b; Wynn et al. 2010).

The Moroccan Turbidite System case study describes two distinct bypass-dominated zones with characteristic facies situated proximally at the base of slope, and distally on the basin floor. During the Bed 5 event, both of these zones were bypassed by large volumes (∼ 52 and 30 km³ respectively) of sediment. The base-of-slope bypass-dominated zone is located close to the mouth of the Agadir Canyon (Fig. 4A, B), where Bed 5 deposits drape scour surfaces with 8–15 cm of very fine sands that grade normally into silt and mud; where no scour surfaces are observed, thin (2–5 cm) gravel layers are present (Fig. 5A) (Talling et al. 2007; Wynn et al. 2010; MacDonald et al. 2011). These deposits grade down slope over approximately 30 km into thicker (50–100 cm) medium and coarse sands, which extend for ∼ 200 km along the length of the Agadir Basin before grading into fine sands (Fig. 5B) (Talling et al. 2007; Wynn et al. 2010; Stevenson et al. 2014b).

Benthic foraminifera species found in Bed 5 show that approximately 20% of the total bed volume (∼ 10 km³) comprises species entrained into the flow via erosion at water depths greater than 2500 m, suggesting that the scours in this bypass-dominated zone were formed and/or modified by the Bed 5 event (Talling et al. 2007). Almost the entire bed volume (∼ 52 km³) is found farther down slope, across the Agadir Basin and the Madeira Abyssal Plain (Wynn et al. 2002b; Frenz et al. 2008; Wynn et al. 2010; Wynn et al. 2012; Stevenson et al. 2014a). Hence, deposits found in the area close to the mouth of the Agadir Canyon represent only a tiny fraction of the total sediment carried by the flow (< 0.8%).

A second bypass-dominated zone is present approximately 250 km downslope, where flows pass from the Agadir Basin with a slope of ∼ 0.02° into the high-aspect-ratio (250:1) Madeira Channel System with a slope of ∼ 0.06 (Fig. 4B) (Stevenson et al. 2013). Deposits of Bed 5 in the Agadir Basin comprise relatively thick (50–100 cm) ungraded coarse to fine sands (Fig. 4A, B), which extend across the entire 100 km width of the basin (Talling et al. 2007; Frenz et al. 2008). The flow was approximately 100 km wide as it approached the head of the Madeira Channel System. Therefore, only parts of the flow were confined by the ∼ 5-km-wide channels, whilst the remainder spread across the channel margins and interchannel highs. In relatively proximal (up to ∼ 50 km from the SW extent of the Agadir Basin) areas of the Madeira Channel System there is no expression of Bed 5 in the channel axes, either as an erosional hiatus or as a deposit of sand or mud (Stevenson et al. 2013) (Fig. 5B). However, along the margins of the channels (i.e., the unconfined seafloor) Bed 5 develops laterally extensive (10’s km), thin (10–15 cm), planar- and ripple-laminated fine sands with inversely graded bases, which are sharply overlain by very thin (1–2 cm) mud caps (Fig. 5B). These deposits correlate with normally graded rippled fine sands present 150–700 km down slope (Stevenson et al. 2013).

Grain-size breaks are common in submarine flow deposits throughout the Moroccan Turbidite System (Sumner et al. 2012; Stevenson et al. 2014a, 2014b). They occur as sharp surfaces between: coarse-grained structureless sands overlain by finer-grained sands, often with planar lamination (Type I); inversely graded sands overlain by finer-grained sands (Type II); iii) planar-laminated sands overlain by finer-grained ripple cross-laminated sands (Type III); sand overlain by mud (Type IV), and; mud-rich sand overlain by mud (Type V) (Fig. 4; cf. Stevenson et al. 2014a). In proximal basin-floor localities, Type I, II, III, and IV grain-size breaks are common, forming composite vertical grading profiles in the deposits (e.g., Fig. 5A; Core 50). In more distal localities, sand deposits generally have simple normal grading profiles overlain by Type IV or V grain-size breaks from sand to mud (e.g., Fig. 5B; Cores P36 and 12). By mapping out individual beds for large distances, several studies de monstrate that grain sizes missing across grain-size breaks in relatively proximal areas are found farther down-system (Amy et al. 2000, 2007; Sumner et al. 2012; Stevenson et al. 2014a). Hence, in proximal areas, these grain sizes must have been bypassed, meaning grain-size breaks represent internal bypass surfaces within a single event bed.

Modern seafloor data provide regional, high-resolution images of the planform morphologies of deep-water systems (Fig. 2A). Facies recorded in cores then allow deposits to be associated with particular fan environments of known slope and confinement, and in rare cases, the spatial evolution of stratigraphically equivalent deposits down slope. This information is valuable to those working in outcrop, where lateral exposure limits our understanding of paleoenvironment, and the spatial variability of deposits down-system. However, modern datasets typically describe fine-grained systems and utilize relatively short (generally < 10 m) cores (Fig. 2A; Core 2), which limits our understanding of coarser-grained facies (sands), lateral facies variability, and the larger-scale stratigraphic evolution of deep-water systems.

Exposures of ancient deep-water systems provide a complementary dataset to the modern sea-floor because: 1) they typically have a sand-rich bias due to the poor weathering of fine-grained rocks, 2) facies can be described laterally over 10s to 100s of meters, and 3) stratigraphic context can be established from successions 100s of meters thick (Fig. 2B). Consequently, there are a range of spatial and stratigraphic scales to be considered, from bed scales (centimeters to meters) and sedimentary-body scales (10s to 100s m) to system and basin scales (10s to 100s km).

Sedimentary facies interpreted as indicators of sediment bypass have been described within slope-channel and base-of-slope (CTLZs) settings. Using high-quality field data, primarily from the Cretaceous Tres Pasos Formation, Chile (Hubbard et al. 2010), we outline the outcrop expression of sediment bypass in these environments.

Scour- and channel-form bodies are defined by an erosional basal surface that truncates underlying strata. Such surfaces are often draped by centimeter to decimeter thick packages of thin-bedded (centimeters thick) fine-grained mudstone, siltstone, and very-fine sandstone (Fig. 6) (Mutti and Normark 1987; Elliott 2000; Gardner and Borer 2000; Grecula et al. 2003; Gardner et al. 2003; Mayall et al. 2006; Barton et al. 2010; Di Celma et al. 2010; Pyles et al. 2010; Brunt et al. 2013; Macauley and Hubbard 2013; Hubbard et al. 2014; Ito et al. 2014). These deposits are typically characterized by horizontal lamination, ripples, or structureless mudstone, and rarely low- to high-angle cross-lamination (Fig 5). In some instances, thin (centimeters) and discontinuous lags of coarse-grained sediment are interstratified with the otherwise fine-grained facies (e.g., Beaubouef et al. 1999). Typically channel base drape units are poorly preserved because of truncation by younger erosion surfaces, which are overlain by sandstones (Fig. 6A; Macauley and Hubbard 2013).

If a flow is erosive (entraining substrate), then its capacity has exceeded its sediment load and de facto must be bypassing as erosion is taking place. Hence, the presence of composite or isolated erosion surfaces in a succession indicates bypass of sediment down-system. Mutti and Normark (1987) considered mud draping erosion surfaces to be the product of longitudinal variations in flow behavior over a fixed geographical point. Their premise was that the head and body of a flow is erosive and bypassing, with deposition from the fine-grained tail; the erosion and mud drape providing the only evidence of its passing.

Lenticular beds or bodies composed of extrabasinal pebble-, granule-, or coarse sand-size grains, and/or intrabasinal sedimentary clasts, are common in axial channel fills (Fig. 7A, B, C, F) (Mutti and Normark 1987; Gardner et al. 2003; Beaubouef 2004). They are sharp-based (often undulatory), are commonly planar laminated or cross-stratified, and can range from 10 cm to many meters thick (e.g., O’Byrne et al. 2007; Jobe et al. 2010). Large-scale cross-stratification (bed sets > 1–2 m thick) in these deposits can accrete upstream (e.g., Pickering et al. 2001) or downstream (e.g., Winn and Dott 1977) (Fig. 7A, B). In rare instances, dune-scale (up to 1 m thick) cross-bedding with mud drapes between foresets characterizes lenticular, coarse-grained turbidite sandstones (Hubert 1966; Keith and Friedman 1977; Mutti 1992; Amy et al. 2000, 2007; Sylvester and Lowe 2004).

Large-scale cross-stratification is interpreted to form through sustained bedload transport of sufficiently coarse-grained sediment beneath flows with relatively low aggradation rates (Allen 1982; Southard 1991), where all but the coarsest sediment is bypassed down-system (Mutti and Normark 1987). Therefore, the presence of cross-stratified coarse-grained sandstone and conglomeratic lag facies in basal slope-channel fills is interpreted to represent bypass, recording the role of these conduits in long-lived sediment transfer (Mutti 1992; Gardner et al. 2003; Hubbard et al. 2010, 2014; Pyles et al. 2010). Mud drapes between foresets indicate repeated intermittent downstream migration of the bedform (Allen 1982), which is interpreted to be the product of multiple flows partially bypassing all but their coarse-grained bed load and then depositing fine-grained sediment from their dilute tails. Therefore, such facies present in base-of-slope (proximal basin-floor settings) indicates repeated bypass of sediment down-system (Mutti and Normark 1987; Mutti 1992; Amy et al. 2000, 2007).

Because conduits for sediment transfer, such as channels, focus submarine flows, significant erosion on the seafloor can occur, removing soft, poorly compacted sediment, and exposing a semiconsolidated or firm substrate (Hubbard and Shultz 2008). This substrate is subject to colonization by burrowing organisms. Upon the death of the organisms, burrows remain open and are subject to filling by sediment derived from partially bypassing flows. In instances where the channel form is mudstone-filled, a record of coarse-grained sediment bypass in the conduit may be present in sandstone-filled trace fossils that subtend from the basal erosional surface (Fig. 7D, E) (Hubbard et al. 2012; Callow et al. 2014). The trace fossils most typically associated with “bypass assemblages” include unlined vertical to horizontal tubes (e.g., Skolithos, Planolites) that often comprise u-shaped structures (e.g., Arenicolites) that are sometimes spreitenated (e.g., Diplocraterion) (Hubbard et al. 2012; Callow et al. 2014).

Channel fill varies from axis, off-axis, to margin (Fig. 6A). Channel-axis stratigraphy is commonly dominated by coarse-grained lags that can include mudstone breccia, which is overlain by amalgamated to non-amalgamated thick-bedded conglomerate- or sandstone-dominated facies (e.g., Normark 1978; Mutti 1992; Gardner et al. 2003; Mayall et al. 2006; Arnott 2007; Kane et al. 2007; DiCelma et al. 2010; Hodgson et al. 2011; Brunt et al. 2013). A progressive transition from amalgamated to non-amalgamated and thinner-bedded sandstone towards channel margins is notable (e.g., Mutti and Normark 1987; Macauley and Hubbard 2013). Fine-grained thin-bedded facies is preferentially preserved at the channel margins (Fig. 6A, B, D, E) (e.g., Mutti and Normark 1987; Beaubouef et al. 1999; Sullivan et al. 2000; Pyles et al. 2010; Macauley and Hubbard 2013; Hubbard et al. 2014; Stright et al. 2014).

Distinguishing between channel-axis and channel-margin drapes is important because channel-axis drapes record sediment bypass from the lower, axial parts of the flows, whilst channel-margin drapes record a lateral depositional record from the upper parts of the flows (Mutti and Normark 1987; Beaubouef et al. 1999; Hubbard et al. 2014). Hence, channel-margin drapes represent a coeval depositional record of sediment bypass within the channel-axis. Numerous erosion surfaces are recorded in the channel-axis fill and are occasionally preserved extending laterally into the channel margins where they are draped by fine-grained thin-bedded facies (Fig. 6A). This indicates repeated and protracted bypass throughout the history of a channel, which is punctuated by depositional events (Hubbard et al. 2014). A reduction in the number of erosion surfaces and the degree of amalgamation upward suggests diminishing levels of bypass as a channel becomes filled (e.g., Brunt et al. 2013). Therefore, channel fills can represent significant amounts of sediment being bypassed down-system, despite local accumulations of thick-bedded coarse-grained facies.

In large-scale outcrops, individual slope channel fills are bounded within larger stratigraphic channel-form surfaces (e.g., Hodgson et al. 2011). Initial incision and sculpting defines the bases of these highly composite surfaces (Mutti and Normark 1987; Beaubouef 2004; Kane et al. 2007; Di Celma et al. 2010; Di Celma et al. 2011; Hodgson et al. 2011; Fildani et al. 2013; Covault et al. 2014). In some instances, these surfaces are overlain by thick chaotically bedded deposits in outcrop, which are attributed to mass wasting (e.g., Crane and Lowe 2008; Macauley and Hubbard 2013). The composite and diachronous nature of these surfaces is direct evidence of repeated and prolonged periods of erosion and sediment bypass. Thus, composite channel-form bodies represent a significant amount of sediment bypass, which would have persisted during repeated phases of channel incision, maintenance, and fill.

Thick successions of fine-grained facies located adjacent to channel bodies are dominated by thin, normally graded turbidites and structureless to laminated mudstone beds. These units are attributed to channel overbank processes and flow stripping on external levees or internal levees, as well as to hemipelagic sedimentation (Winn and Dott 1979; Hickson and Lowe 2002; Kane et al. 2007, 2009a; Hubbard et al. 2008; Hodgson et al. 2011; Kane and Hodgson 2011; Morris et al. 2014).

Although levee deposits are not often considered to represent bypass-related deposits per se, their origins from flow stripping and overspill at channel bends mean that they represent deposition from the fine-grained (upper) parts of flows passing through channels, where the coarse-grained (basal) components of the flows are confined within the channel axes and bypass downslope (Buffington 1952; Normark 1978; Piper and Normark 1983; Piper and Savoye 1993; Fildani et al. 2006; Kane et al. 2010). Hence, internal and external levees are coeval depositional records of sediment bypass within channels.

Outcrops of ancient submarine slope systems that allow detailed, regional correlation of stratigraphic surfaces along slope-parallel, dip-orientated profiles provide a means to consider the expression and impact of sediment bypass at the system-scale (10s to 100s km) (Gardner et al. 2003; Fildani et al. 2009; Flint et al. 2011; van der Merwe et al. 2014). A series of high-relief clinoforms that crop out in the Magallanes Basin of southern Chile preserve segments of slopes that were at least 30–40 km long, from the shelf edge to the lower slope (Fig. 7G, H, I) (Romans et al. 2009; Hubbard et al. 2010). Evidence for sediment bypass in upper to middle slope positions includes lags of extrabasinal clasts (Fig. 7C) and up to 8-m-thick conglomerate units that contain upstream-facing, dune-scale cross-stratification (Fig. 7A). These deposits, prevalent in composite 20–60 m thick siltstone-dominated sedimentary packages that also include isolated sandstone bodies, can be tracked basinward, where they pass into stacked sandstone-dominated channel complexes that comprise 300 m of stratigraphy (Hubbard et al. 2010; Macauley and Hubbard 2013). The shift from more proximal, siltstone- and conglomerate-dominated slope sections that are 10s of meters thick to distal, sandstone-prone units that are 100s of meters thick is interpreted to be the result of long-lived sediment bypass of the slope.

Slope channel–levee complexes in Sub-Unit C2 in the Fort Brown Formation, Karoo Basin, South Africa, record > 50 m of composite erosion with multiple erosion surfaces overlain by claystone- and siltstone-clast conglomerates (Di Celma et al. 2011; Hodgson et al. 2011; Morris et al. 2014). These channel–levee complexes are stratigraphically attached down-system to ∼ 40 km³ of sandstone-rich lobe units (Van de Merwe et al. 2014). However, in the overlying “Unit E” there are basin-floor bypass-dominated zones which generate “sand-detached” lobe architectures; these areas are characterized by thin-bedded (decimeter thick) rippled sandstones with numerous shallow scours mantled by rip-up clasts, and a top surface that includes megaflutes and evidence for remobilized sediments (van der Merwe et al. 2014). These basin-floor bypass zones are related down-system to ∼ 25 km³ of sand-prone lobes.

Stratigraphic context as a means to assess sediment bypass in the outcrop record must be attempted with caution, due to the two-dimensional nature of exposures (Fig. 2B). Although segments of sediment-routing systems might be correlated, longitudinally or across-strike, with confidence, a quantitative assessment of basin-scale down-dip sediment transfer is rarely possible.

Subsurface data, such as core, wireline logs, and seismic, can help address the spatial and stratigraphic limitations encountered by outcrop and modern seafloor datasets (Fig. 2). However, recognizing deep-water sediment bypass is challenging in subsurface datasets because the sedimentological and stratigraphic record can be subtle, and the data are typically of limited vertical and lateral resolution. In this section, a number of subsurface examples are used to illustrate the expression of deep-water sediment bypass in each of these data types; behind-outcrop boreholes and wireline logs, isolated core and wireline datasets, and seismic reflection profiles (Fig. 2C; Core 1). Finally we present a seismic reflection profile integrated with core (cuttings) and wireline data to illustrate how these complementary data can be used to interpret sediment bypass in the subsurface.

Behind-outcrop research boreholes help to constrain stratigraphic successions and sedimentary facies observed in outcrop, providing an important link between subsurface data and field-based observations (Van Wagoner et al. 1990; Beaubouef et al. 1999; Donselaar and Schmidt 2005; Pickering and Corregidor 2005; Luthi et al. 2006; Luthi and Flint 2014). Where behind-outcrop cores penetrate submarine channel fills and basin-floor deposits, sedimentological processes can be deduced, and recognition criteria for sediment bypass can be grounded in 2D and 3D observations from the outcrop belt. These data are invaluable for interpretation of sediment bypass in subsurface datasets that are not characterized by outcrop or reservoir-scale 3D constraints (i.e., that which typically falls below seismic resolution).

The potential for using strategically positioned behind-outcrop boreholes to constrain the depositional record of sediment bypass is realized from analysis of the Permian Upper Brushy Canyon Formation, West Texas (Fig. 8). In this study, a detailed correlation panel built from closely spaced measured sections was constructed to constrain the depositional architecture of a lower-slope channel-complex fill at the “Buena Vista” outcrop (Fig. 8C) (Beaubouef et al. 1999). ExxonMobil EPR drilled a pair of wells behind the outcrop in the late 1990s. At outcrop, stacked channel-margin deposits are characterized by thinly interbedded fine-grained sandstone and siltstone, as well as lenses of coarse-grained sandstone that overlie multiple inclined erosional surfaces (Fig. 8D). Research borehole Buena Vista 1 (EPR BV#1) intersected these channel-margin deposits (Fig. 8A, B) (Rossen and Beaubouef 2008). The facies association found in the core is dominated by inclined thin-bedded fine- and coarse-grained turbidite sandstone and siltstone with rare debrites. Characteristically, the dip of thin beds shallow upward, but these trends are truncated by steeply inclined erosion surfaces overlain by steeply dipping beds (Fig. 8B, between y and z). The change in bed dip, the number of erosion surfaces, and the coarse-grained facies found in this thin-bedded association is interpreted to record largely bypassing flows transporting most of their sediment load basinward; the head and body of the flows bypassing all but their coarsest material and deposition of fines occurring from their tails (Beaubouef et al. 1999; Gardner et al. 2003; Rossen and Beaubouef 2008).

Criteria for sediment bypass can be established in cores that are isolated from nearby outcrop. The presence of erosion surfaces can be used to interpret channel incision and bypass-dominated intervals. These surfaces may be overlain by lag deposits of intraformational or extrabasinal clast-rich conglomerate (Fig. 9A) (Beaubouef et al. 1999; Pickering et al. 2001; Pickering and Corregidor 2005; Luthi et al. 2006; Mayall et al. 2006; Hubbard et al. 2009; Grundvag et al. 2014), or draped by fine-grained sandstone and mudstone (Fig. 9B). Identification of erosion surfaces across the diameter of a core can be challenging (e.g., Fig. 9B). Therefore, particular facies associations can be used to infer the base of a channel and its overlying fill. For example, an erosion surface draped by thin-bedded fine-grained sandstone and mudstone may record the base of a channel fill (Fig. 9C). In this case, the thin beds are attributed to periods of erosion and sediment bypass, as are the overlying structureless sandstone beds with abundant rounded mudstone clasts at their base (Fig. 9C).

Dune-scale cross-bedding intercalated with mudstone drapes and coarse-grained filled trace fossils can also be observed in drill cores (Fig. 9D, E). As in outcrop, these facies can be interpreted to record sediment bypass. The interpretation of sediment bypass from cores recovered from offshore New Jersey (IODP Expedition 313) is supported by reflection seismic data that show that the succession is located at the base of slope, with stratigraphically equivalent thicker sandy turbidites encountered in another core drilled approximately 10 km down-system (Miller et al. 2013).

Seismic reflection data represent a powerful tool with which to investigate the architecture and evolution of deep-water sedimentary systems (e.g., Posamentier and Kolla 2003; Deptuck et al. 2003, 2007; Wynn et al. 2007; Janocko et al. 2013; Ortiz-Karpf et al. 2015). However, a key constraint on the application of these data is their finite vertical and lateral resolution to image a geological feature. Due to increasing velocity and frequency attenuation with depth, seismic resolution decreases (e.g., Brown 2011). At a given depth, vertical seismic resolution has two components: separability and visibility. The limit of separability is the thickness at which reflections from two successive interfaces begin to interfere; typically 1/4 of the dominant wavelength (Widess 1973). The limit of visibility is the minimum thickness at which the reflections from two successive interfaces can be seen; typically 1/30 of the dominant wavelength, depending on the signal-to-noise ratio and the impedance contrast at the interfaces (Brown 2011). These limitations in vertical resolution impact our ability to identify stratigraphic surfaces and units associated with deep-water sediment bypass. For example, a sandstone-mantled erosion surface, which is encased in an otherwise mudstone-dominated succession (e.g., slope-channel coarse-grained lags; Fig. 7A) may not be expressed geometrically (i.e., as an erosion surface truncating underlying reflections) or acoustically (i.e., as a change in seismic amplitude) if the sandstone is too thin or if the contrast in acoustic impedance between the sandstone and mudstone is too small. Only if the sandstone is sufficiently thick will it be expressed via a change in amplitude, and may coincide with an erosion surface truncating underlying reflections (provided that the magnitude of incision is within resolution).

To illustrate the variable and subtle expression of deep-water sediment bypass in seismic reflection datasets we use examples from offshore Colombia and Brazil. The time-migrated 3D dataset collected in 2009, offshore Colombia, has an inline and crossline spacing of 12.5 m, a dominant frequency of 30 Hz, and a vertical and horizontal resolution of approximately 15 m. Figure 10 shows a composite channel-form body, approximately 1.3 km wide and 300 m thick, incised into a packages of low-amplitude, shallow-dipping reflectors that taper away from the channel form (marked 1). The base of the channel form is expressed as a composite surface that comprises multiple incisions (marked 2). Within this there are a number of smaller-scale, laterally offset channel forms, approximately 200–500 m wide and up to 80 m thick (marked 3). The seismic facies characterizing the basal parts of the channel forms vary laterally. For example, high-amplitude reflections may be present on one side of the channel forms (marked 4) whereas low-amplitude reflections occur on the other (marked 5).

Similar composite channel-form bodies are widely observed in subsurface datasets (e.g., Deptuck et al. 2003, 2007; Mayall et al. 2006; Janocko et al. 2013; Ortiz-Karpf et al. 2015). In particular, Deptuck et al. (2003) used 3D reflection seismic data to propose a link between processes and sedimentary products during protracted channelization on continental margins. In their conceptual model, long-lived and multi-phased incision establishes the template for the subsequent focus of channels, and initiates external levee growth. Initial sinuous channels are laterally offset and are highly erosional, the flows cannibalizing early deposits and transferring the exhumed sediment downslope. Subsequent establishment of internal levees promotes channel aggradation, and the final product preserved in the rock record includes a composite, time transgressive, confining surface kilometers wide and hundreds of meters deep (cf. Strong and Paola 2008; Sylvester et al. 2011). Internal to this large-scale feature are smaller-scale channel forms 200–1000 m wide and up to 100 m thick (aspect ratios  =  10–20). Linking observations from outcrop, described earlier, with the larger-scale subsurface architecture supports an interpretation that these subsurface channel forms represent long-lived and multi-phased incision associated with large amounts of sediment bypass.

The time-migrated 3D seismic reflection dataset, collected offshore Brazil, has an inline and crossline spacing of 25 m and 12.5 m respectively, and a vertical sample spacing of 4 ms. Vertical and horizontal resolution are estimated to be 20 m. In the profile shown in Figure 11, a series of flat-lying, subparallel reflections in the west (marked 2a) are truncated by a concave-up, east-facing erosional surface (marked 2b) that is downlapped by clinoforms (marked 3a). In the central part of the profile a series of relatively conformable, gently eastward-dipping clinoforms are observed, which show limited variations in amplitude (marked 4). Down dip of this point, a package of relatively chaotic, variable-amplitude reflections is developed, which has a highly rugose upper surface (marked 3b) and which onlaps and pinches out westward onto a discrete clinoform surface developed in the clinoform-bearing succession. Clinoforms downlap the concave-upward surface in the western part of the profile and the underlying package of chaotic reflections in the east (marked 4).

The observed seismic-stratigraphic architecture is interpreted to document the progradation, collapse, and reestablishment of a shelf margin (cf. Nemec et al. 1988; Dixon et al. 2013). The concave-upward surface (2b) represents the shelf-edge failure scar (cf. Levell and Kasumajaya 1985), which truncates flat-lying shelf strata (e.g., 2a). The clinoforms that downlap this surface (e.g., 3a) record healing of relief associated with scar formation. The chaotic package of reflections (overlying surface 3b) developed down-system at the toe of slope is interpreted as a mass-transport deposit (MTD), sourced from the failure of the shelf edge. The size of the shelf-edge scar and the volume of the MTD together indicate that a significant amount of material was remobilized from the shelf edge during this event. A critical question is therefore: What is the seismic expression of the stratigraphic surface that documents this major slope bypass event?

Based purely on seismic-stratigraphic relationships, in particular (i) the location of MTD pinchout on the lower slope and (ii) the lower slope position of the stratigraphic surface downlapped by shelf-edge clinoforms, we suggest that the surface that records the major slope bypass event is located on the lower slope, between the horizons that define the top and base of the failure scar up-system, and the top and base of the MTD down-system (marked 2c in Fig. 11). This interpretation suggests that the sediment bypass surface is extremely subtle and is represented by neither a pronounced erosional surface nor a change in amplitude (Fig. 11B). Outcrop and core data, such as those described in previous sections, suggest that it is likely that a series of incisions are developed on sediment bypass surfaces (e.g., Figs. 5A, 6A, 8). However, if present in the Santos Basin example, these features are less than a few tens of meters deep and are therefore too subtle to be imaged on these data.

Identification of sediment bypass surfaces in seismic datasets can be augmented by cores and well logs. To highlight this we draw on an example from the Upper Jurassic of the South Viking Graben, northern North Sea (Fig. 12A) (Jackson et al. 2011). In this location, deposition of syn-rift turbidite sandstone and slope mudstone was synchronous with activity on a series of thin-skinned, gravity-driven normal faults (Fig. 12B) (Thomas and Coward 1996). The upper Oxfordian turbidite-bearing succession displays large and abrupt changes in thickness across a seaward-dipping fault (the Gudrun Fault; Fig. 13). Well data indicate that 80–200 m thick sandstone-rich packages are developed in the hanging wall of the fault with intervals of amalgamated sandstones up to 40 m thick (Fig. 12; 15/3-1S and 15/3-8). In contrast, well data (15/3-7) from the footwall indicate that the equivalent succession is relatively thin (< 50 m) and mudstone-dominated, indicating that the thick, sandstone-rich, turbidite-dominated part of the upper Oxfordian succession is restricted to the hanging wall of the Gudrun Fault. This observation prompts the question: What is the sedimentological record of sediment bypass up-system of the thick sandstone packages? Detailed analysis of closely spaced (< 0.2 m) cuttings data in the footwall well (15/3-7) indicate that a number of thin (1–2 cm), sharp-based, very-fine to medium-grained sandstones are developed in the footwall of the Gudrun Fault, and that these are age-equivalent to the upper Oxfordian turbidite-dominated packages identified 1.8 km farther down-system. Furthermore, close inspection of wireline logs up-system of the hanging wall indicate that, despite an overall high gamma radioactivity (generally > 120 API), the succession in the footwall contains a number of low GR (< 80 API) “spikes,” which correspond to the sandstones documented in cuttings data (Fig. 14). These thin sandstones are interpreted as lag-type units deposited beneath partially bypassing sand-rich flows, similar to facies described previously in modern seafloor (e.g., Fig. 5A), outcrop (e.g., 7C), and core (Figs. 8, 9A). This record of sediment bypass is variable both down-system and across-strike with low GR (< 80 API) “spikes” not recorded in well 15/3-5 (located 4.5 km from 15/3-7) (Fig. 14).

The case studies outlined in this paper illustrate the spectrum of recognition criteria that can be used to interpret sediment bypass in deep-water systems. These sediment bypass indicators vary spatially, across different parts of deep-water systems. In this section we place the range of sediment bypass indicators into assemblages found in different positions down the profile of an idealized deep-water system. This multi-scale synthesis of sediment bypass criteria complements gross-scale analysis of sediment bypass, derived from seismic profiles (Paola and Martin 2012).

Slope-channel settings in modern seafloor, outcrop, and subsurface datasets have a range of features indicative of sediment bypass (Fig. 15A). These include: 1) abundant composite erosion surfaces (e.g., Figs. 6A, B, 8C, 10) that are often draped by thin-bedded sandstone and mudstone, preferentially preserved towards channel margins (Figs. 6C, D, E, 8D, E, 9C); 2) sandstone-filled trace fossils that subtend from the bases of mudstone-filled channel-forms (Figs. 7D, E, 9E); 3) lenticular coarse-grained lag deposits ranging in thickness from centimeters to meters thick, which may be characterized by large-scale cross stratification (Figs. 7A, B, 9D); and 4) external levees, which form a laterally equivalent depositional record of sediment bypass within the channel axis (e.g., Fig. 10A).

In outcrop and subsurface, striations at the base of MTDs may aid identification of the sediment bypass surface and the direction of sediment transport (e.g., Pickering and Corregidor 2005; Moscardelli et al. 2006; Dakin et al. 2013; Olafiranye et al. 2013; Ortiz-Karpf et al. 2015). Where the sediment bypass surface is not directly overlain by its genetically related MTD, identifying the surface is challenging. At outcrop and in core, this surface may be cryptic and it may only be possible to infer its presence based on the recognition of high-angle truncation surfaces (erosional cuts) or an abrupt vertical change in facies (e.g., Beaubouef et al. 1999; Brunt and McCaffrey 2007; Rossen and Beaubouef 2008; Di Celma et al. 2011; Macauley and Hubbard 2013). In seismic reflection datasets, MTD-related bypass surfaces can be extremely subtle and challenging to detect (Fig. 11B). Such cryptic sediment bypass surfaces may only be identifiable based on their wider spatial context, such as its location between an upslope slide scar and a downslope MTD (Figs. 11A, 15B).

Several sediment bypass indicators recorded in base-of-slope settings are also found in slope-channel settings (Fig. 15C). It is possible that base-of-slope settings, which are spatially coincident with CLTZs, may have well-developed channels (e.g., van der Merwe et al. 2014), making it difficult to compartmentalize slope-channel and base-of-slope settings based on isolated indicators of sediment bypass. Indicators of sediment bypass in base-of-slope (CLTZ) settings (Fig. 15C) include: 1) isolated and amalgamated scours 10’s m deep and kilometers in areal extent, which are typically overlain by fine-grained thin-bedded drapes (Fig. 5A); 2) where scouring is absent, thin (centimeter to decimeter), coarse-grained sand and conglomeratic lags, often with lenticular bed geometries (e.g., Figs. 5A, 9A); and 3) rarely, lenticular coarse-grained sandstones with dune-scale cross bedding (Fig. 9D). The similarities in facies associations between slope-channel and base-of-slope settings mean that a wider, regional, context is required to distinguish them, and their associated bypass criteria.

Indicators of sediment bypass in basin-floor environments are expressed internally within beds as grain-size breaks (e.g., Fig. 5). Grain-size breaks are most common in proximal localities and become progressively less frequent with distance from source (Fig. 15D, E, G). Basin-floor settings may have relatively distal bypass-dominated zones, where the system changes longitudinally from depositional to bypass-dominated (Fig. 15F). Bypass indicators across these zones include inversely graded rippled sandstones with sharp tops and almost no mud caps (Fig. 5B), and shallow (decimeter) isolated scours (van der Merwe et al. 2014). Certain areas (down-system and laterally) may also be subject to complete sediment bypass, whereby flows leave no trace of their passing.

Similar criteria for sediment bypass can be found in more proximal settings such as slope channels and base-of-slope settings. However, a number of important differences exist at a bed and sedimentary-body scale, which may allow them to be distinguished. First, all of the sandy deposits in a basin-floor bypass-dominated zone will be relatively fine-grained compared to more proximal localities. Second, scouring in basin-floor bypass-dominated zones is more likely to be isolated (van der Merwe et al. 2014), which contrasts with the composite erosion surfaces found proximally. From a wider stratigraphic perspective, basin-floor sheet sands or lobes (e.g., Hodgson et al. 2006; Amy and Talling 2006; Talling et al. 2007; Prélat et al. 2009) should be found both up-system and down-system of the basin-floor bypass zone (e.g., Stevenson et al. 2013; van der Merwe et al. 2014).

Theoretical and numerical criteria struggle to predict sediment bypass across a range of experimental flow conditions. Equations 1–8 assume that turbulence maintains sediment in suspension. This assumption breaks down at higher sediment concentrations, where turbulence is suppressed and grains are transported in the flow via other mechanisms, such as hindered settling and grain-to-grain collisions (Lowe 1982; Postma et al. 1988; Sohn 1997; Cartigny et al. 2013). Hence, sediment concentration can have a positive influence on the ability of flows to transport and bypass sediment (cf. van Maren et al. 2009). This phenomenon is not yet included in criteria that describe sediment bypass and may explain the limitations of the approaches outlined in this paper to predict bypass in the high-concentration experiments of Cartigny et al. (2013). In addition, the criteria do not explicitly link boundary-layer conditions with those in the upper parts of the flow. Poor validation, coupled with a limited ability to relate properties of the lower and upper parts of a flow, means that it is not possible to quantitatively link bypassing flow processes with deposits observed in the field.

The common parameters of grain size, sediment concentration, slope, and flow size, however, are linked through a physical framework, which can be applied in a qualitative sense to support conceptual models and interpretation of bypass facies in the field. For example, Mutti and Normark (1987) proposed varying styles of bypass in base-of-slope settings (CLTZs), which are controlled by flow size and grain size (two parameters highlighted above). Small-volume and coarse-grained flows have a high sensitivity to breaks in slope, whereby they deposit rapidly (low bypass), producing localized wedge or lenticular geometries. In contrast, large flows with a significant amount of fine-grained sediment have a lower sensitivity and largely bypass the break in slope. This results in bypass facies such as mud-draped scours, thin coarse-grained lags, and deposits with large-scale cross bedding (e.g., Fig. 15). Qualitatively associating these parameters (within an established physical framework) with particular styles of bypass and facies in the field is an important step towards quantitatively linking process with product in deep-water systems.

This contribution examines the phenomenon of sediment bypass in deep-water clastic systems from the key physical processes that govern it to the expression left behind in the stratigraphic record.

At the temporal and spatial scale of individual flows, bypass is governed by vertical and longitudinal variations in flow dynamics, which evolve with distance down-system. From a range of theoretical and numerical approaches we outline the principal controls that underpin these dynamics, which in turn govern sediment bypass: grain size, slope, flow size, and sediment concentration. However, these approaches struggle to predict bypassing flow conditions across a range of physical experiments. Poor validation through experiments and a paucity of direct measurements from natural flows (cf. Talling et al. 2015) for comparison means that it is not yet possible to establish quantitative links between deposits in the field and bypassing flow processes. This provides us with inspiration for future work connecting the processes and products of deep-water sediment bypass.

We examined the stratigraphic expression of sediment bypass in modern seafloor, outcrop, and subsurface datasets. Commonalities are found, giving rise to a spectrum of criteria that can be employed to interpret sediment bypass from the geological record at specific locations on the slope-to-basin-floor profile (Fig. 15). These criteria for sediment bypass vary spatially, from proximal to distal and laterally from axis to margin, and with the scale of observation. No single criterion is an effective tool for interpreting sediment bypass; rather, an assemblage of indicators of sediment bypass must be used depending on the paleogeographic context within the system and the scale of observation. We emphasize that an improved understanding of sediment bypass requires a holistic approach, integrating different data types at various scales of spatial and stratigraphic resolution.

Two studies (the Moroccan Turbidite System and Karoo Basin) in which sediment volumes can be mapped down-system of bypass-dominated zones show that similar facies can represent a large range of sediment volumes bypassed. Therefore, it is not possible to quantify volumes of sediment bypassed from isolated criteria. Currently, quantitative estimates can be determined only from good paleogeographic understanding of the system, i.e., mapping of stratigraphically equivalent deposits down-system of the bypass zone.

The establishment of criteria to recognize sediment bypass at multiple scales, linked with a physical framework, is an important step for improving our understanding of submarine flow dynamics and their associated depositional architectures, and key to reducing uncertainty in the down-system prediction of reservoir bodies in hydrocarbon exploration.

We wish to thank the SEPM for sponsoring the “Sediment Bypass in Deep-Water Systems” evening session at the AAPG AGM, Pittsburgh (2013), Bret Dixon for organizing the session, and the numerous scientists who contributed to an engaging and lively discussion. Brian Romans and Zoltan Sylvester are thanked for their thorough and constructive reviews, which have greatly improved the paper. Morgan Sullivan is thanked for editorial handling of the manuscript and for making additional valuable comments. We thank Andrea Ortiz-Karpf for providing the seismic reflection profile in Figure 10.