Fluvial-style migration controls autogenic aggradation in submarine channels: Joshua Channel, eastern Gulf of Mexico Open Access

Supplementary material: S1, Animation of interpreted horizons illustrating the meandering evolution of the Joshua deep-water system; S2, Map view; and S3, Perspective view of the evolution of a forward stratigraphic model based on a stream power law are available at https://doi.org/10.6084/m9.figshare.c.6610896.v1

Submarine channel systems are significant conduits for the delivery of terrigenous sediment to deep-sea fans, which can preserve records of environmental change (Romans et al. 2016). Deposits of submarine channels can host large petroleum resources, and they are potential targets for carbon storage (Kunka et al. 2003; Mayall et al. 2006; Deptuck et al. 2007; Marshall et al. 2016). Submarine channels are also part of the transit system for microplastics and other pollutants (Kane et al. 2020).

Three-dimensional seismic-reflection data have significantly improved the understanding of deep-water channel migration through space and time (Deptuck et al. 2003; Posamentier and Kolla 2003; Deptuck et al. 2007). An early idea was that migration of a sinuous deep-water channel primarily occurred through channel-bend expansion with little translation, followed by near-vertical aggradation, during which channel sinuosity and form remained approximately constant (Peakall et al. 2000; Kane et al. 2008). This evolution was thought to be fundamentally different from the evolution of a meandering fluvial system (e.g. Wynn et al. 2007).

Some researchers have shown that the plan-view evolution of fluvial and deep-water channel systems can be similar; although, deep-water channel systems tend to aggrade more due to the reduced density contrast between the turbidity current and seawater compared to the density contrast of water and air in the case of a river (Imran et al. 1998; Jobe et al. 2016, 2020). For example, researchers have observed patterns of systematic lateral migration by deep-water channel-bend growth, with downstream translation and cutoffs (Abreu et al. 2003; Deptuck et al. 2003, 2007; Posamentier 2003; Posamentier and Kolla 2003; Schwenk et al. 2003; Kolla 2007; Babonneau et al. 2010; Sylvester and Covault 2016; Covault et al. 2020, 2021).

Here, we interpret the Joshua channel system in the eastern Gulf of Mexico, which shows features analogous to meandering fluvial systems. In addition to planform migration, we consider processes and controls that influence channel aggradation in a deep-water setting. Allogenic controls on turbidity current properties have been interpreted to play a role in channel aggradation (e.g. Samuel et al. 2003; Mayall et al. 2006; McHargue et al. 2011). Using a forward stratigraphic model, we evaluate the hypothesis that the increasing sinuosity of a channel results in a concurrent decrease in the channel slope by geometrical necessity, and this decrease in slope promotes deposition from turbidity currents and aggradation of the system. Therefore, autogenic processes intrinsic to the channel system can play a role in determining vertical movements of the channel, i.e. its trajectory (Sylvester et al. 2011; Morris et al. 2022). We show evidence that the processes documented here could be applied to other channel–levee systems such as the Amazon, in addition to giving insights into the processes governing the evolution of deep-water channel systems more generally.

The Joshua deep-water channel–levee system is located approximately 300 km to the SE of New Orleans, Louisiana, in the eastern Gulf of Mexico (Fig. 1a). It is situated between the Mississippi fan complex to the SW, salt domes to the north and the Florida Escarpment to the east (Kramer et al. 2016). The Joshua channel system has been regionally mapped as part of the ‘blue unit’ deposited during the Wisconsin glaciation of c. 71 ka when sea level fell by c. 120 m (Martinson et al. 1987; Donoghue 2011; Bjerstedt et al. 2016). The system was active until its abandonment at c. 18 ka when sea level rose (Bjerstedt et al. 2016). Following abandonment, the sediment pathway previously feeding the Joshua system was rerouted to create younger channel–fan complexes to the SW (Kramer et al. 2016).

Posamentier (2003) suggested that the Mississippi River fed terrigenous sediment to the Joshua deep-water channel system. Subsequent mapping has indicated that the Joshua system was connected to the ancestral Pearl River (Bjerstedt et al. 2016). The Joshua deep-water channel–levee system is >600 km long and it is covered by a hemipelagic mud drape that is c. 10–60 m thick (Fig. 1b) (Kramer et al. 2016). On the seafloor, the channel width measured across levee crests ranges from 650 to 750 m. The channel system is flanked by levee-overbank deposits with total levee widths of >10 km (Bjerstedt et al. 2016).

Posamentier (2003) described the Joshua channel system as comprising a highly sinuous channel on the seafloor sitting atop a larger channel belt flanked by levee-overbank deposits. The Joshua channel–levee system was characterized as primarily aggradational, with levees aggrading in tandem with the channel floor. Levees were documented to be higher at the outer banks of channel bends than along the inner banks of bends due to overspill processes, which also led to the development of sediment waves (Posamentier 2003, fig. 6). Meander cutoffs and a significant component of downstream translation of bends were also observed. Numerous avulsions were documented prior to abandonment, resulting in crevasse splay-like features, which were hypothesized to be linked to a fundamental change in the turbidity current flow parameters (Posamentier 2003, fig. 16).

We used a high-resolution 3D seismic-reflection dataset over the Joshua channel system in water depths ranging from 2860 to 3170 m. The data were processed using a tilted transverse isotropy (TTI) Kirchoff depth migration with a bin spacing of 30 m (inline) × 25 m (crossline) and a vertical sample rate of 5 m. We manually interpreted 13 horizons along a 25 km reach of the channel system using Aspen SeisEarth interpretation and visualization software. We picked flat- to concave-up high negative amplitudes, which are sinuous channel forms in map view. We interpreted these reflections as the bases of sandy channel-fill architectural elements (e.g. Normark et al. 1993; Deptuck et al. 2003, 2007; Posamentier 2003). This suggests that the sands are acoustically hard relative to the surrounding mud because the depositional system has not been buried to the depth where differential compaction of mud would become significant. The set of 13 horizons document the evolution of the channel. We tracked the deepest parts of the channel forms (i.e. the thalwegs) over the 25 km reach. We measured the sinuosity and longitudinal profile of each centreline. There are potentially additional channel forms between the 13 we interpreted (Fig. 2).

We also mapped three regional horizons to ensure consistency in our channel-form interpretations (Fig. 3). The deepest and shallowest horizons correspond to the basal erosive surface and seafloor expression of the Joshua channel–levee system. Their depth difference represents the stratigraphic thickness of the channel system. The middle horizon is a stratigraphic surface overlying a neck cutoff of a channel bend. We created five proportional slices (Fig. 4) from smoothed versions of the three regional horizons mapped in Figure 3. Figure 4 shows semblance (e.g. Marfurt et al. 1998) attribute maps from the base of the channel system to the seafloor. The locations of the semblance maps in the subsurface are shown in Figures 1b and 2b.

Figure 2 shows three cross-sections of seismic-reflection profiles of the Joshua channel system. Figure 2a shows a stack of high-amplitude, concave-up (U-shaped) seismic reflections bounded by wedge-shaped packages of low–moderate amplitude, approximately parallel reflections. Below the uninterpreted profile is the same profile with the basal (purple) and intermediate (blue) regional horizons and 13 channel-form horizons interpreted. To the right of the seismic-reflection profiles are line-drawing interpretations of the stratigraphic architecture of the channel system. We interpreted the stack of high-amplitude, concave-up seismic reflections to be channels and their sandy fill (Normark et al. 1993; Deptuck et al. 2003, 2007; Posamentier 2003; Bjerstedt et al. 2016; Kramer et al. 2016). These channel deposits are bounded by muddy levee-overbank deposits (Posamentier 2003; Bjerstedt et al. 2016). Figure 2b shows a cross-section through a neck cutoff in the upstream reach of the channel system (see also Fig. 3b). Figure 2c shows three stacks of channel deposits, with the north–south profile intersecting the sinuous bends of the Joshua system three times (Fig. 1). The channel deposits typically have a U-shaped base, which partially erodes underlying high-amplitude reflections. In all cross-sections, stacks of channel deposits aggrade vertically by >150 m from the basal erosive surface, and laterally migrate in a systematic way. Lateral migration distances across the entire channel-belt width are more variable, with the lateral offsets of individual stacks of channels ranging from c. 1 to >5 km. The average channel-belt width is c. 2.6 km but varies from c. 1 to 10 km depending on whether sections cross multiple bends of the system (Fig. 2). Aggradation results in an average channel-belt thickness of c. 160 m and the channel deposits being elevated (c. 40 m) above the surrounding seafloor (Posamentier 2003). Overall, it appears that the channel underwent significant lateral migration while it aggraded during its evolution (Figs 2 & 3).

Figure 4 shows five semblance attribute maps through the Joshua channel system (locations shown in Fig. 2b). Figure 4a shows the channel system at the seafloor, and Figure 4e shows the basal erosive surface. Figure 5 shows 13 channel-form horizons and their centrelines (CLs). It is clear from Figures 4 and 5 that channel bends systematically underwent expansion and downstream translation. This style of channel-form evolution was evident throughout the channel system's evolution, including during aggradation (Fig. 2).

The channel grew in sinuosity as it aggraded (Figs 4 and 5). We measured sinuosity as the length of the channel thalweg divided by the length of the reach (25 km) (Fig. 6). The earliest interpreted channel form has a sinuosity of 1.25; this number increases to 2.3 for the most recent channel form on the seafloor. The significant reduction in channel sinuosity between CLs 5 and 6 (Fig. 6) resulted from the neck cutoff (Fig. 7c), which reduced channel sinuosity from 1.8 to 1.35 over the reach. Channel expansion and downstream translation continued post-cutoff during channel aggradation and the sinuosity eventually exceeded its pre-cutoff maximum (Fig. 6). This cutoff had a significant impact on the character of the channel longitudinal profiles (see the following subsection) (Fig. 8).

The neck cutoff occurred as the channel intersected itself along a large bend (Fig. 7d). The next centreline is a chute, which resulted in a sudden change in the channel's path (Fig. 7c). The chute reversed the migration direction of the channel compared to before the cutoff in that locality (Fig. 7b). With no additional cutoffs, the bend-migration direction remained consistent until the abandonment of the channel system (Fig. 7a). The other bends along the 25 km reach continued to undergo expansion and downstream translation in the same general direction throughout the channel system's evolution (Fig. 5).

The longitudinal profiles of the 13 interpreted channel-form horizons are plotted in Figure 8a. The length of the channel increased as the channel aggraded; this is a result of bend expansion increasing the channel sinuosity over the reach (see Fig. 6). The average slope of the channel decreased as the channel sinuosity increased over time (Fig. 8b). However, the chute just post-cutoff at CL 6 shows a notable increase in the average slope compared to pre-cutoff (CL 5) (i.e. from c. 0.095° pre-cutoff to c. 0.13° post-cutoff: Fig. 8b). This increase in slope also corresponds with the reduction in sinuosity post-cutoff (Figs 6 & 8b). Moreover, the three channel profiles immediately post-cutoff (CLs 6–8) clearly cross pre-cutoff profiles (CLs 4 and 5) (Fig. 8a). These data suggest that the average channel slope across the reach is influenced by the developing sinuosity of the channel and the occurrence of cutoffs.

The bend expansion and downstream translation of the Joshua deep-water channel, in tandem with an overall increase in the sinuosity of the channel (with the exception of the neck cutoff), suggest similarities to the planform movements of fluvial systems (Howard and Knutson 1984; Furbish 1988; Sylvester et al. 2019). In contrast, the amount of aggradation is significantly larger in the Joshua deep-water channel system than what is typical of fluvial systems, a characteristic feature of many submarine channels (Jobe et al. 2016, 2020).

We found that channel-bend expansion and downstream translation occurred throughout the Joshua deep-water channel system evolution, including during aggradation. The ability of the Joshua Channel to translate and expand is consistent with the observations of Posamentier (2003). In planform, the channel deposits of the Joshua system appear similar to those of meandering rivers (Fisk 1944; Kolla et al. 2007; Durkin et al. 2017, 2018; Schwenk et al. 2017; Sylvester et al. 2019). Indeed, sinuous channel patterns produced by meandering processes have been recognized on continental margins (Damuth et al. 1983; Flood and Damuth 1987; Pirmez and Flood 1995; Pirmez and Imran 2003). But deep-water channel meandering is not limited to the smooth seafloor of passive-margin continental rises. Covault et al. (2020, 2021) recently demonstrated that downstream translational processes occur in salt basins and other tectonically active settings with a complex topography. Clearly, many examples of deep-water channels meander through systematic migration, resulting in bend expansion, downstream translation and cutoffs as in rivers (Abreu et al. 2003; Posamentier and Kolla 2003; Kolla et al. 2007; Babonneau et al. 2010; Sylvester and Covault 2016; Hansen et al. 2017; Covault et al. 2020, 2021).

We documented a channel stacking architecture in which the initial morphology of the active channel strongly influenced its subsequent evolution. This suggests that the planform of the earliest channel plays a critical role in controlling the subsequent development of the system. For example, Figure 7 shows how the initial large bend in CL 1(Fig. 7e) leads to the neck cutoff in CL 6 (Fig. 7c). In contrast, workers have described apparently random patterns of intersecting channel deposits, especially during early periods of the evolution of deep-water channel systems (e.g. Mayall et al. 2006; Hodgson et al. 2011; McHargue et al. 2011). However, a single meandering channel with little vertical movement could generate numerous erosional channel remnants with only one continuous channel thread preserved (Deptuck et al. 2003; Sylvester et al. 2011). This can also result in an apparently disorganized, random pattern of intersecting channel deposits, particularly if the resolution of seismic-reflection data is a limiting factor (Covault et al. 2016).

Sylvester et al. (2021) showed that cutoffs can change the direction of channel migration in fluvial systems. The bend cutoff studied in this deep-water system resulted in a significant change in the channel morphology where a chute developed (Fig. 7c), leading to a reversal in the active channel's migration direction (Fig. 7) (Seminara 2006; Sylvester et al. 2021). Figure 9 compares the kinematic behaviour of the Ucayali River in Peru to the Joshua deep-water channel system. A neck cutoff develops, leading to a chute with a local reversal in curvature (Fig. 9b). This local reversal in curvature promotes a change in the channel migration direction (Fig. 9a). Cutoffs have been reported in a number of deep-water channel systems (Pirmez et al. 2000; Schwenk et al. 2003; Kolla et al. 2012; Sylvester and Covault 2016; Covault et al. 2020). The chute in the Joshua Channel is relatively small, reminiscent of neck cutoffs in meandering fluvial systems. In submarine channel–levee systems elevated above their surrounding bathymetry (e.g. this study), complex cutoffs with long chute channels are unlikely to form because most levee breaches result in an avulsion. However, neck-style cutoffs are possible, as documented here and on the Amazon fan system (Pirmez et al. 2000). This study shows how submarine-channel cutoffs can resemble their fluvial counterparts. These cutoffs can play an important role in reshuffling a meandering pattern (Figs 7 & 9) (Sylvester et al. 2019).

Many authors have suggested that the controls on channel evolution are likely to be a combination of external factors controlling sediment supply and processes intrinsic to the depositional system (Pirmez et al. 2000; Deptuck et al. 2003; Posamentier and Kolla 2003; Mayall et al. 2006). Sinuosity and bend migration develop intrinsically in channelized systems (Seminara 2006; Lazarus and Constantine 2013; Limaye et al. 2021). As the Joshua deep-water channel system became more sinuous, its reach-averaged slope decreased (Fig. 8), and it did so by geometrical necessity. We suggest that this reduction in slope was likely to be the cause for its coeval aggradation. The erosional and depositional dynamics of turbidity currents are sensitive to changes in slope (Middleton and Hampton 1973; Fisher 1983; Parker et al. 1986; Middleton 1993; Kneller 1995, 2003; Kneller and McCaffrey 1999; Ferry et al. 2005). We propose that the increasing sinuosity and reducing slope of the channel may have enhanced deposition from the turbidity currents flowing through it, driving aggradation. This control is significant because it suggests that processes intrinsic to the channel system can play a role in determining the vertical migration of a channel through time, i.e. its trajectory (Sylvester et al. 2011; Covault et al. 2016; Morris et al. 2022). Each preserved channel element is likely to be the result of numerous sediment-gravity flows (Hubbard et al. 2020; Talling et al. 2022); simultaneous lateral migration and aggradation is unlikely to happen from a single flow.

See Table 1 for the variables used in equation (1). A similar approach has been applied to investigate incision in both fluvial and deep-water channel systems (Finnegan and Dietrich 2011; Sylvester and Covault 2016; Mitchell et al. 2021). To account for deposition as part of the same simple relation, we specifiy a critical slope, S_cr, that defines a bypass condition chosen to fit the behaviour observed in the Joshua channel system. A channel slope greater than the slope required for bypass (S > S_cr) will cause erosion; a slope less than the slope required for bypass (S < S_cr) will lead to aggradation (Fig. 10a). Through time, the channel slope profile (S) reduces in tandem with its increasing sinuosity, and this drives the system towards deposition and aggradation (equation (1)) (Fig. 10) (see Supplementary material videos S2 and S3). This approach links intra-channel deposition to the balance between the driving force per unit area of the turbidity current and the channel slope. In other words, it is a simple and uncalibrated model but it can be used to successfully reproduce an autogenic switch from erosion to deposition in sinuous submarine channels.

To illustrate the potential of this simple model to capture the important characteristics of the Joshua Channel, we have chosen parameters so that the model output is overall similar to the Joshua system. Plots of model output parameters through time illustrate the behaviour of the channel as it builds the channel–levee system (Fig. 10). The channel is initially relatively linear but its sinuosity increases through time, with the exception of a cutoff event at c. 680 years (Fig. 10a). In parallel with the increase in sinuosity, the mean channel slope over the reach decreases; this induces a switch from incision to aggradation (Fig. 10b, c). The Joshua Channel also shows early incision followed by aggradation in tandem with increasing sinuosity (Fig. 6).

The early phase of channel evolution (before CL1), which is likely to have been incisional, is not preserved in the Joshua system. In contrast, every centreline is recorded in the model and can be displayed (Fig. 10). The oldest centrelines, before year 300, are not preserved in the final stratigraphic model either. This explains some differences between measurements of the Joshua channel system and the model, such as the apparent linear relation between mean channel slope and sinuosity in the Joshua Channel (Fig. 8), and the obviously non-linear decrease in mean channel slope with increasing sinuosity in the model (Fig. 10). The trends of mean channel slope as a function of sinuosity are similar, and approximately linear, over the range of sinuosities recorded in the Joshua channel system (1.3–2.25: Figs 8b & 10d). The model predicts a neck cutoff that temporarily lowers the reach-averaged sinuosity (Fig. 10a; see also Supplementary material videos S2 and S3); however, subsequent bend growth results in increasing sinuosity. This also aligns with the observations from the Joshua Channel data (Fig. 6). There is good agreement between the trend in slope sinuosity recorded in the Joshua Channel and that predicted by the model (Figs 8b & 10d).

Longitudinal profiles extracted from the model and from the Joshua system are also similar (Figs 8a & 10e). Both the model and the Joshua Channel data show a reduction in the average channel slope during aggradation. The cutoff occurring in both the model and Joshua Channel data results in several subsequent post-cutoff centrelines crossing the profiles of several pre-cutoff profiles. The cutoff in the model generates a local steepening (Fig. 10c, e), which propagates upstream through time (Fig. 10e). Knickpoints have been modelled to occur in incisional systems (e.g. Sylvester and Covault 2016) but here we suggest they can even occur in net depositional environments as a result of channel cutoff processes.

We demonstrated broad agreement in terms of evolution of channel sinuosity, aggradation and slope between the Joshua Channel data and the model (Figs 8 & 10). The stratigraphic architecture observed in the model outputs is also similar to the Joshua Channel and other channel–levee systems, such as the Amazon channel system (Fig. 11). The model reproduces a highly sinuous channel form that is elevated above the surrounding bathymetry, and neck cutoffs are present (Fig. 11). In cross-section, the model captures near-vertically stacked channel-fill facies that are flanked by muddier levee-overbank deposits (Fig. 11).

Changes in channel slope through time could be an influence on deep-water channel evolution more generally. It could influence architectures ranging from primarily incisional (Babonneau et al. 2002; Sylvester and Covault 2016) to infilled-valley systems such as the Benin-major Canyon (Deptuck et al. 2007), in addition to the channel–levee systems previously mentioned. For example, Babonneau et al. (2002) showed a downdip transition from primarily incisional valley architectures to more aggradational channel–levee systems through the Zaire Canyon–Channel system. A potential key factor differentiating the three types of architecture could be the background (i.e. valley) slope, which can vary on continental margins from relatively steep (>2°), driving erosion (i.e. resulting in incised-valley systems) to relatively flat (<0.1°) driving aggradation (i.e. resulting in channel–levee systems) (Babonneau et al. 2002; Nittrouer et al. 2007). In other words, the initial background slope also influences vertical channel behaviour and the resultant stratigraphic architecture, and can be determined allogenically (e.g. via tectonics, diapirism or position on the continental margin).

We interpreted a 25 km reach of the Joshua deep-water channel system, located on the seafloor of the eastern Gulf of Mexico, using a high-resolution 3D seismic-reflection dataset. We documented channel-bend expansion and downstream translation. These processes progressively increased channel sinuosity, except for an abrupt decrease associated with a neck cutoff. Reach-averaged channel sinuosity grew from an initial 1.25 to 2.3 at abandonment. The initial planform channel morphology was key to setting subsequent bend-migration directions. The plan-view style of migration is analogous to a meandering river. The increase in channel sinuosity led to a corresponding reduction in its average slope. We propose that channel aggradation could be controlled autogenically, with a progressive reduction in slope acting to enhance turbidity-current deposition. We advanced our hypothesis using a simple forward stratigraphic model in which vertical changes in channel position are based on a stream power law. We showed how aggradation can potentially be driven autogenically as channel sinuosity developed through time, generating stratigraphy analogous to channel–levee systems such the Joshua Cannel and the Amazon. Furthermore, we showed how trends in sinuosity, aggradation and slope are in broad agreement between the Joshua channel system and that predicted by the model. We suggest that evolving channel slope profiles because of developing sinuosity could be a factor in the vertical movement of submarine channel systems. This highlights the importance of intrinsic deep-water channel processes as potential controls on the evolution and trajectories of the system.

We would like to thank TGS-NOPEC Geophysical Company for granting permission to use this exceptional 3D seismic-reflectance dataset and Paradigm (Emerson) for the use of their seismic interpretation software. We thank both John Snedden and Mike Hudec for their significant help in obtaining this dataset, and Bryan Stevens and Thomas Bjerstedt at the Bureau of Ocean Energy Management (BOEM) for discussions on the stratigraphic history of the eastern Gulf of Mexico. We thank colleagues at both the Quantitative Clastics Laboratory (QCL) and Mohrig research groups at the UT-Austin Jackson School of Geosciences for insightful discussions that have aided the development of this manuscript. Finally, we thank both Nick Howes and Carlos Pirmez for their constructive reviews that significantly improved the manuscript

The authors declare no known competing interests.

PDM: conceptualization (lead), data curation (lead), formal analysis (lead), investigation (lead), methodology (lead), project administration (lead), resources (lead), software (lead), validation (lead), visualization (lead), writing – original draft (lead), writing – review & editing (lead); ZS: conceptualization (supporting), data curation (supporting), formal analysis (supporting), investigation (supporting), methodology (supporting), project administration (supporting), resources (supporting), software (supporting), supervision (equal), validation (supporting), visualization (supporting), writing – original draft (supporting), writing – review & editing (equal); JAC: conceptualization (supporting), data curation (supporting), formal analysis (supporting), investigation (supporting), methodology (supporting), project administration (supporting), resources (supporting), software (supporting), supervision (lead), validation (supporting), visualization (supporting), writing – original draft (supporting), writing – review & editing (equal); DM: data curation (supporting), formal analysis (supporting), investigation (supporting), project administration (supporting), resources (supporting), supervision (supporting), validation (supporting), writing – review & editing (supporting); DD: data curation (supporting), project administration (supporting), software (supporting), visualization (supporting), writing – review & editing (supporting).

This work was funded by the sponsors of the University of Texas at Austin Bureau of Economic Geology (UT-Austin BEG) Quantitative Clastics Laboratory (QCL) (http://www.beg.utexas.edu/qcl; whose continued support is vital to our research efforts.

The seismic dataset/images thereof is used and published with permission from TGS-NOPEC Geophysical Company. Use/sharing of this seismic dataset will require their permission. All other data are available from the authors upon reasonable request.