LETTERS PUBLISHED ONLINE: 17 AUGUST 2015 | DOI: 10.1038/NGEO2505

Braiding of submarine channels controlled by aspect ratio similar to rivers Brady Z. Foreman1*†, Steven Y. J. Lai2, Yuhei Komatsu3 and Chris Paola1 The great majority of submarine channels formed by turbidity and density currents are meandering in planform; they consist of a single, sinuous channel that transports a turbid, dense flow of sediment from submarine canyons to ocean floor environments1,2 . Braided turbidite systems consisting of multiple, interconnected channel threads are conspicuously rare1 . Furthermore, such systems may not represent the spontaneous planform instability of true braiding, but instead result from erosive processes or bathymetric variability3–5 . In marked contrast to submarine environments, both meandering and braided planforms are common in fluvial systems6,7 . Here we present experiments of subaqueous channel formation conducted at two laboratory facilities. We find that density currents readily produce a braided planform for flow aspect ratios of depth to width that are similar to those that produce river braiding. Moreover, we find that stability model theory for river planform morphology8 successfully describes submarine channels in both experiments and the field. On the basis of these observations, we propose that the rarity of braided submarine channels is explained by the generally greater flow depths in submarine systems, which necessitate commensurately greater widths to achieve the required aspect ratio, along with feedbacks9,10 among flow thickness, suspended sediment concentration and channel relief that induce greater levee deposition rates and limit channel widening. Alluvial rivers and submarine channels represent the main conduits by which sediment is transported and deposited across Earth’s surface, and are central in the construction of terrestrial and marine topography1,7 . High-resolution imaging of submarine morphology has revealed a striking degree of parallelism between submarine and terrestrial channels, with meandering, dendritic (tributary) and fan-like (distributary) morphologies well developed in both realms. But there is a glaring gap in the set of morphologic parallels: whereas both subaerial and submarine channels systems exhibit single-thread meandering and multithread braided planform morphologies1,2,6,7 , it is not clear whether true braiding occurs in the submarine realm. Meandering and braided planforms are both relatively common in fluvial systems6,7 . Meandering rivers tend to occur on shallower slopes characterized by more uniform hydrographs and with relatively more suspended (finer) sediments6,7 . Braided rivers tend to occur on steeper slopes that transport coarse-grained sediments; these are also environments in which the hydrograph is often highly variable (‘flashy’)6,7 . However, the link between braiding and ‘flashy’ hydrographs is probably not fundamental, as braided rivers are easily produced experimentally under constant water discharge8,11,12 .

Similarly, the braided rivers of the Indo-Gangetic plain are a clear natural counterpoint to these broad generalizations, as they occur on lower slopes, transport large volumes of sand, and, in some cases, do not exhibit flashy discharges13 . Furthermore, meandering and braided rivers commonly differ in total sediment load, with the former exhibiting lower concentrations than the latter6,7 . Finally, it is common for braided rivers to transition into meandering planforms downstream as slopes decrease and coarse sediment is selectively deposited out6,7 . Submarine channels formed by sediment gravity flows (turbidity and density currents) exhibit several key differences from their subaerial counterparts. Meandering planforms are overwhelmingly the most prevalent, if not the sole channel planform morphology in modern submarine settings1 . Documented braid-like submarine morphology is rare in the literature and where identified it is unclear whether the bars that define the pattern represent a dynamic, autogenic instability of the flow–sediment system, as is the case for true braiding8 , or are instead nucleated around seafloor topography (for example, gas and fluid vents)5 or related to erosive events that scour multiple channels around fixed erosional islands1,3,4,14 . If the scant submarine examples are, in fact, true braided systems, they show some opposite trends relative to their subaerial braided counterparts. These multi-thread submarine channels occur on notably shallow slopes, whereas meandering occurs across a wider range of steep and shallow slopes3,4,15 . Multi-thread density currents seem to be associated with more sustained flows as opposed to flashy discharges. For example, such systems occur in response to nearcontinuous saline undercurrents from the Bosphorus Strait into the Black Sea5 and long-duration subglacial outburst floods of the late Pleistocene into the northeastern Labrador Basin4 . In comparison, episodic turbidity currents of varying magnitude and frequency readily construct meandering submarine channels2 . Similar to rivers, multi-thread submarine channels transport coarser grain sizes4,16 compared to meandering submarine channels1 . Finally, multi-thread submarine channels seem to occur downstream of a meandering channel after flow stripping and other overbank processes preferentially remove finer-grained sediment1,3,14 or finergrained sediment is lost to surface plumes4 . In the case of the late Pleistocene submarine channels of the Labrador Basin these processes, combined with sediment lofting, potentially led to multithread turbidity currents characterized by relatively low sediment concentrations4 , the opposite of their subaerial counterparts. Current theory suggests that bar deposition and the development of a braided planform is fundamentally a response to flow instabilities controlled by the depth–width (aspect) ratio of the flow8 . Other conditions such as slope, hydrograph, grain size and

1 Department

of Earth Sciences, St Anthony Falls Laboratory, University of Minnesota, Minneapolis, Minnesota 55414, USA. 2 Department of Hydraulic and Ocean Engineering, National Cheng Kung University, Tainan 701, Taiwan. 3 Japan Oil, Gas and Metals National Corporation, Technology Research, Chiba 261-0025, Japan. †Present address: Department of Geology, Western Washington University, Bellingham, Washington 98225, USA. *e-mail: [email protected] 700

NATURE GEOSCIENCE | VOL 8 | SEPTEMBER 2015 | www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE GEOSCIENCE DOI: 10.1038/NGEO2505 0

c

24 18

20

12

90 100

40

−6 270

0

100 y (cm)

200

360

80 y (cm)

100 120

160

20 40 60 80

80

−18

140

30 20 10 0

60

−12

−24 0

x (cm)

x (cm) 200

0

z (mm)

x (cm)

6 180

d

0

) cm x(

b

0

z (cm)

a

LETTERS

100 120 140

0

20 40 y (cm)

60

0

20

40 y (cm)

60

Figure 1 | Images and topography of braided density current deposits. a, Ortho-rectified image of braided density deposits at the SAFL facility. b, Laser scan of braided topography at the SAFL facility. c, Ortho-rectified laser-scanned image of braided density deposits at the NCKU facility. d, Image of active density current transporting sediment in multiple channel threads draped over laser-scanned topography at the NCKU facility.

sediment flux–water discharge ratio are not directly responsible for the development of braid bars and a braided planform8,11,12 . Instead, these factors and other associative factors (for example, basin geometry, vegetation, cohesion, overbank sedimentation rates) act to influence the depth–width ratio of the flow8,11,12 . Thus braiding develops across a wide range of conditions so long as the flow is sufficiently wide relative to its depth7,8,17–19 . Indeed, the phenomenon of braiding within wide channelized flow is so pervasive that it is considered one of the premier examples of scale independence in nature, with striking similarities between experimental and field dimensionless morphologic measures18,19 . However, despite the ease with which braided fluvial systems develop on experimental scales18,19 , to our knowledge braiding in submarine channels has yet to be produced under controlled conditions. This is particularly surprising given the importance of experiments in the study of sinuous submarine channels. Density currents, particularly turbidity currents, are rare, unpredictable, and difficult to observe in nature; they occur underwater, often at great depth, and submarine channels can reach enormous scales (for example, channel widths can be several tens of kilometres and channel depths several tens to hundreds of metres)1,2 . Thus, for many purposes experiments are the sole means to examine channelized density currents in detail. The majority of these experimental studies impose a static channel form, straight or sinuous9,20,21 . Although these studies accurately capture in-channel flow dynamics and overbank deposition, they are not intended to provide insight into channel initiation and evolution1,9,19,20 . The comparatively few studies that do involve channel initiation on experimental scales have produced single-thread, weakly sinuous channels22 or a distributary network of multiple sinuous channels23 . Here we report experiments explicitly designed to subject subaqueous density currents to conditions known to produce spontaneous channel braiding in fluvial systems. Two experimental series run in parallel at St. Anthony Fall Laboratory (SAFL) and National Cheng Kung University (NCKU) successfully produced braided planform morphologies from high-aspect-ratio, subaqueous saline sheet flows with a constant flow discharge and sediment supply (see Methods). In both experimental series the sheet flow quickly evolved into two to five main channels that were further segmented downstream by a network of braid bar and lobe deposits (Fig. 1a–d). These main channels underwent local avulsion events and braid bars tended to migrate downstream (see Supplementary Movie). This morphology developed on an initially smooth bed of non-cohesive sediment. An initial slope was imposed, but the system was otherwise free to self-adjust to sediment and discharge inputs. The two experimental systems were,

on net, aggradational. Thus, by simply imposing a thin, wide flow as an initial condition, a dynamic, braided planform spontaneously developed. We conclude, assuming scale independence applies to braided turbidity systems as it does to braided fluvial systems24 , that the sole limitation in natural turbidity systems is developing a sufficiently wide flow, relative to depth. The question of how submarine currents self-channelize has been raised by a number of workers1,2,22,23 . There is clear experimental evidence for selfchannelization in net erosional settings22 . The experiments we report here demonstrate robust spontaneous formation of multiple channels without net erosion, under conditions similar to those known to produce multiple channels in rivers. Although the problem is far from fully solved, this work adds to the mounting evidence for similarities in self-channelization between density currents and rivers. Our experiments, in combination with other field and experimental examples, indicate that the aspect ratio of density currents dictates the planform morphology of submarine channels, capturing the essence of controls on the meander-braided continuum in a similar manner to fluvial systems. Figure 2 shows the modelled stability fields for fluvial channel planforms8 along with experimental and field examples of both submarine and subaerial channels (see Supplementary Information). Forty-two of the 46 singlethread submarine channels plot in the appropriate stability field. Moreover, our experimental braided submarine channels, the three modern braided examples for which there are sufficient accurate constraints, and three examples of braided submarine channels from the stratigraphic record all plot within the braided stability field (see Supplementary Information for calculations and assumptions). We conclude from these observations that, as in fluvial systems, the main determinant for a braided versus meandering planform in submarine channels is the depth–width ratio. Why then does braiding seem to be less common in submarine channels than in rivers? We suggest that there are several related factors that inhibit the development of wide, shallow channels in submarine as compared to fluvial systems. First, assuming all other factors equal, the reduced density contrast driving the flow in density currents compared to rivers means that, in general, density currents must be much deeper to produce the same shear stress relative to rivers, so from the outset the width needed to produce the same aspect ratio is much larger for density flows than for rivers. Second, in order for braiding to develop in submarine channels in a similar manner to fluvial systems the braid bars themselves must subdivide the flow, which requires bar forms that scale with flow depth8,11,12,17 . However, bar forms in turbidity currents are rarely commensurate with flow depth1 owing to the substantial

NATURE GEOSCIENCE | VOL 8 | SEPTEMBER 2015 | www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved

701

NATURE GEOSCIENCE DOI: 10.1038/NGEO2505

LETTERS Methods

100

Slope/densimetric Froude number

NCKU 10−1

Methods and any associated references are available in the online version of the paper.

Cambro−Ordovician and Miocene turbidites

Received 13 April 2015; accepted 10 July 2015; published online 17 August 2015

SAFL Northeastern Labrador Basin

10−2

References

Black Sea

br aid

3−

10

O

ve r1

10−3

s

0

br aid

s

Orinoco fan

2−3 braids 1−2 braids

10−4 10−5

10−4

10−3 10−2 Depth/width

Turbidity data (experimental/field data) / Braided / Meandering Straight

10−1

100

Fluvial data / Braided / Meandering / Straight

Figure 2 | Theoretical stability fields of fluvial channel planforms with supporting field and experimental river examples8 , and comparison to submarine examples. Explanation of the sources of the fluvial and submarine data underlying this figure can be found in the Methods and Supplementary Information.

thicknesses of many turbidity currents and the fine-grained nature of the upper reaches of the flow. Third, many submarine systems are exceedingly clay-rich1,2 , implying that cohesive effects in overbank deposits may be generally greater in submarine systems than in fluvial systems, inhibiting channel widening. Fourth, the rates of levee and overbank deposition compared to in-channel deposition are significantly higher in submarine than in fluvial systems9,10,25–27 . If channel widening is viewed as a related rates problem between sediment removed along channel margins and sediment deposition on levees and overbank areas, then the rate of widening should be proportionally slower in turbidity systems. Higher levee and overbank sedimentation rates occur due to overspill processes because many channelized submarine flows are thicker than the conforming channels that they construct10,28,29 . Experiments and seismic data indicate balanced, or equal rates of aggradation between levee tops and the channel base9,10,26–28 ; thus, thicker flow events pass over the channel topography largely unimpeded and contribute to overbank sedimentation9,10,26–28 . This has been proposed to lead to the unusual stability in submarine systems9,10,26 , and may also help explain the paucity of submarine braiding; changes in the thickness of flow events are accommodated by either preferential overbank or in-channel deposition to maintain the channel geometry9,10,26,27 , as opposed to channel widening. Analogously, in some fluvial systems significantly reduced channel mobility and bank erosion have been linked to high overbank aggradation rates due to trapping of sediment by vegetation30 . These observations combined with our experiments indicate that the rarity of braided submarine channels is probably not due to inherent differences between fluvial and density flow behaviour in the channel itself, but rather is a result of the generally greater thickness of density flows combined with greater rates of cohesive overbank deposition in submarine systems. Overbank processes seem to play a major role in controlling the evolution of bathymetry and morphology of the largest sedimentary systems on Earth. 702

1. Wynn, R. B., Cronin, B. T. & Peakall, J. Sinuous deep-water channels: Genesis, geometry and architecture. Mar. Petrol. Geol. 24, 341–387 (2007). 2. Piper, D. J. W. & Normark, W. R. Processes that initiate turbidity currents and their influence on turbidites: A marine geology perspective. J. Sedim. Res. 79, 347–362 (2009). 3. Ercilla, G. et al. New high-resolution acoustic data from the ‘braided system’ of the Orinoco deep-sea fan. Mar. Geol. 146, 243–350 (1998). 4. Hesse, R. et al. Sandy submarine braid plains: Potential deep-water reservoirs. Am. Assoc. Petrol. Geol. Bull. 85, 1499–1521 (2001). 5. Flood, R. D., Hiscott, R. N. & Aksu, A. E. Morphology and evolution of an anastomosed channel network where saline underflow enters the Black Sea. Sedimentology 56, 807–839 (2009). 6. Leopold, L. B. & Wolman, M. G. River channel patterns: Braided, meandering and straight. US Geol. Surv. Prof. Pap. 282-B, 39–85 (1957). 7. Schumm, S. A. Patterns of alluvial rivers. Annu. Rev. Earth Planet. Sci. 13, 5–27 (1985). 8. Parker, G. On the cause and characteristic scales of meandering in braiding in rivers. J. Fluid Mech. 76, 457–480 (1976). 9. Straub, K. M. & Mohrig, D. Quantifying the morphology and growth of levees in aggrading submarine channels. J. Geophys. Res. 113, F03012 (2008). 10. Peakall, J., McCaffrey, B. & Kneller, B. A process model for the evolution, morphology, and architecture of sinuous submarine channels. J. Sedim. Res. 70, 434–448 (2000). 11. Ashmore, P. E. Bed load transport in braided gravel-bed stream models. Earth Surf. Process. Landf. 13, 677–695 (1988). 12. Paola, C. in Gravel-Bed Rivers V (ed. Mosley, M. P.) 11–38 (New Zealand Hydrological Society, 2001). 13. Sinha, R. & Friend, P. F. River systems and their sediment flux, Indo-Gangetic plains, Northern Bihar, India. Sedimentology 41, 825–845 (1994). 14. Cronin, B. T. et al. in Atlas of Deep-Water Environments: Architectural Styles in Turbidite Systems (eds Pickering, K. T. et al.) 84–88 (Chapman & Hall, 1995). 15. Covault, J. A., Fildani, A., Romans, B. W. & McHargue, T. The natural range of submarine canyon-and-channel longitudinal profiles. Geosphere 7, 313–332 (2011). 16. Hein, F. J. & Walker, R. G. The Cambro-Ordovician Cap Eragé Formation, Queébec, Canada: Conglomeratic deposits of a braided submarine channel with terraces. Sedimentology 29, 309–352 (1982). 17. Schumm, S. A. & Kahn, H. R. Experimental study of channel patterns. Geol. Soc. Am. Bull. 83, 1755–1770 (1972). 18. Murray, A. B. & Paola, C. A cellular model of braided rivers. Nature 371, 54–57 (1994). 19. Paola, C., Straub, K., Mohrig, D. & Reinhardt, L. The ‘‘unreasonable effectiveness’’ of stratigraphic and geomorphic experiments. Earth Sci. Rev. 97, 1–43 (2009). 20. Straub, K. M., Mohrig, D., McElroy, B., Buttles, J. & Pirmez, C. Interactions between turbidity currents and topography in aggrading sinuous submarine channels: A laboratory study. Geol. Soc. Am. Bull. 120, 368–385 (2008). 21. Sequeiros, O. E. Estimating turbidity current conditions from channel morphology: A Froude number approach. J. Geophys. Res. 117, C04003 (2012). 22. Métivier, F., Lajeunesse, E. & Cacas, M.-C. Submarine canyons in the bathtub. J. Sedim. Res. 75, 6–11 (2005). 23. Yu, B. et al. Experiments on self-channelization subaqueous fans emplaced by turbidity currents and dilute mudflows. J. Sedim. Res. 76, 889–902 (2006). 24. Malverti, L. E., Lajeunesse, E. & Métivier, F. Small is beautiful: Upscaling from microscale laminar to natural turbulent rivers. J. Geophys. Res. 113, F04004 (2008). 25. Pirmez, C. Growth of a Submarine Meandering Channel-Levee System on the Amazon Fan PhD dissertation, Univ. Columbia (1994). 26. Deptuck, M. E., Steffens, G. S., Barton, M. & Pirmez, C. Architecture and evolution of upper fan channel-belts on the Niger Delta slope and in the Arabian Sea. Mar. Petrol. Geol. 20, 649–676 (2003). 27. Straub, K. M. & Mohrig, D. Growth of constructional canyons via sheet flow turbidity currents: Observations from offshore Brunei Darussalam. J. Sedim. Res. 79, 24–39 (2009). NATURE GEOSCIENCE | VOL 8 | SEPTEMBER 2015 | www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved

NATURE GEOSCIENCE DOI: 10.1038/NGEO2505 28. Hiscott, R. N. et al. Basin-floor fans in the North Sea: Sequence stratigraphic models vs. sedimentary facies: Discussion. Am. Assoc. Petrol. Geol. Bull. 81, 662–665 (1997). 29. Mohrig, D. & Buttles, J. Deep turbidity currents in shallow channels. Geology 35, 155–158 (2007). 30. Smith, D. G. & Smith, N. D. Sedimentation in anastomosed river systems: Examples from alluvial valleys near Banff, Alberta. J. Sedim. Petrol. 50, 157–164 (1980).

Acknowledgements The authors thank the St. Anthony Falls Laboratory Industry Consortium, which includes Japan Oil, Gas and Metals National Corporation (JOGMEC), ConocoPhillips, Chevron, Shell, ExxonMobil, and BHP Billiton, as well as the Ministry of Science and Technology from Taiwan (MOST 103-2221-E-006-215) for funding of this research.

LETTERS S. S. C. Hung, D. Baldus, R. Rosario, A. Sorenson and B. Erickson are acknowledged for assistance in conducting experiments.

Author contributions B.Z.F., S.Y.J.L., Y.K. and C.P. co-wrote manuscript. B.Z.F. and S.Y.J.L. designed the experimental set-up and ran experiments. C.P. and Y.K. conceived of the project. All authors contributed to data analysis.

Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. Correspondence and requests for materials should be addressed to B.Z.F.

Competing financial interests The authors declare no competing financial interests.

NATURE GEOSCIENCE | VOL 8 | SEPTEMBER 2015 | www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved

703

NATURE GEOSCIENCE DOI: 10.1038/NGEO2505

LETTERS Methods The experimental set-up at St. Anthony Fall Laboratory (SAFL) consists of a table, 2.2 m wide and 3.65 m long, with 0.15 m roughened sideboards on either side. This table is submerged within a 5.0 m × 5.0 m × 0.64 m tank. The experimental set-up at National Cheng Kung University (NCKU) is similar to the one at SAFL. An inside tank (0.7 m × 1.4 m × 0.06 m) is submerged into an outside tank (0.9 m × 1.65 m × 0.6 m). The SAFL set-up introduces a saline current as a sheet flow at the proximal end using a simple weir system, whereas the NCKU set-up initially introduces the saline current as a point source that is allowed to immediately expand across the sediment bed. Saline currents have a density of ∼1.2 g ml−1 compared to ∼1 g ml−1 density of ambient freshwater in the basin. Saline current discharges are held constant at 3,000 ml s−1 and 27 ml s−1 at SAFL and NCKU, respectively. The distal end of the SAFL set-up extends to the flat basin floor with a pump system employed to prevent the accumulation of saline bottom waters. In the NCKU set-up an overflow weir is employed to remove saline water and sediment at the distal end. An initial bed of crushed walnut sand with D50 of 1 mm (SAFL) and white plastic sand with D50 of 0.3 mm (NCKU) covers the floor of each experimental set-up. At SAFL walnut sand is hand-fed at a constant rate at the proximal end at a rate of 53 ml s−1 . At NCKU a motor-controlled sediment feeder was used to maintain a constant rate of sediment at 0.73 ml s−1 . At SAFL, sediment was introduced evenly across the sheet flow as best as possible. Walnut sand was soaked in saline water before introduction to the experiment. The wet bulk density of the walnut sand is 1.2 g ml−1 and white plastic sand 1.5 g ml−1 . The sediment flux/water discharge ratio is similar, 0.02 and 0.03, for SAFL and NCKU, respectively. The experiments were initiated with a bed slope of 3.0◦ and 6.0◦ for SAFL and NCKU, respectively. In both cases flow depths of the initial saline current

were 2 ± 1 mm. This results in a width–depth ratio of flows of 1,000:1 for SAFL and 350:1 for NCKU. Runtime for the SAFL experiment was 4 h and for the NCKU experiment it was 2 h. Images were collected every 15–20 min from a drained basin in the SAFL experiments and every 5 s while the NCKU experiment was running. A sheet laser system was used to obtain topography at the end of the run at SAFL. The NCKU laser system is able to scan the submerged surface and obtained topography every 10 min. Data. Fluvial data presented in Fig. 2 was obtained from the compilation of Parker8 . Submarine channel data sets presented in Fig. 2 were obtained from the sources listed in the Supplementary Information; many occur within a recent compilation21 . Depth and width of submarine channels were obtained from base of the thalweg to levee top and levee top to levee top for several of the field-based channels, respectively. Note that this actually underestimates the flow depth, because the thickness of many density currents exceeds the bounding channel. For submarine channels from the stratigraphic record (noted in data table) the maximum width was limited by outcrop exposure in some cases. In a few instances the densimetric Froude number for field submarine channels was not available. In those cases we have estimated this value within the middle range of those documented reported21 unless otherwise noted. Turbidity currents exhibit a restricted range in Froude numbers (one order of magnitude) in comparison to their fluvial counterparts (four orders of magnitude)21 . The uncertainty of this estimate does not affect the overall pattern or interpretations. Slopes and Froude numbers for submarine channels in the stratigraphic record were estimated to obtain the most conservative ratio of the two.

NATURE GEOSCIENCE | www.nature.com/naturegeoscience

© 2015 Macmillan Publishers Limited. All rights reserved