Turbidity currents, and other types of submarine sediment density flow, redistribute more sediment across the surface of the Earth than any other sediment flow process, yet their sediment concentration has never been measured directly in the deep ocean. The deposits of these flows are of societal importance as imperfect records of past earthquakes and tsunamogenic landslides and as the reservoir rocks for many deep-water petroleum accumulations. Key future research directions on these flows and their deposits were identified at an informal workshop in September 2013. This contribution summarizes conclusions from that workshop, and engages the wider community in this debate. International efforts are needed for an initiative to monitor and understand a series of test sites where flows occur frequently, which needs coordination to optimize sharing of equipment and interpretation of data. Direct monitoring observations should be combined with cores and seismic data to link flow and deposit character, whilst experimental and numerical models play a key role in understanding field observations. Such an initiative may be timely and feasible, due to recent technological advances in monitoring sensors, moorings, and autonomous data recovery. This is illustrated here by recently collected data from the Squamish River delta, Monterey Canyon, Congo Canyon, and offshore SE Taiwan. A series of other key topics are then highlighted. Theoretical considerations suggest that supercritical flows may often occur on gradients of greater than , 0.6u. Trains of up-slope-migrating bedforms have recently been mapped in a wide range of marine and freshwater settings. They may result from repeated hydraulic jumps in supercritical flows, and dense (greater than approximately 10% volume) near-bed layers may need to be invoked to explain transport of heavy (25 to 1,000 kg) blocks. Future work needs to understand how sediment is transported in these bedforms, the internal structure and preservation potential of their deposits, and their use in facies prediction. Turbulence damping may be widespread and commonplace in submarine sediment density flows, particularly as flows decelerate, because it can occur at low (, 0.1%) volume concentrations. This could have important implications for flow evolution and deposit geometries. Better quantitative constraints are needed on what controls flow capacity and competence, together with improved constraints on bed erosion and sediment resuspension. Recent advances in understanding dilute or mainly saline flows in submarine channels should be extended to explore how flow behavior changes as sediment concentrations increase. The petroleum industry requires predictive models of longer-term channel system behavior and resulting deposit architecture, and for these purposes it is important to distinguish between geomorphic and stratigraphic surfaces

setting where flows were expanding due to loss of confinement or a decrease in slope gradient. The resultant reduction in flow thickness, Reynolds number, shear stress, and capacity promoted suspension fallout and thus CRCL formation. CRCL in the New Zealand study area was deposited both outside of and within channels at an inferred break in slope, where flows were decelerating and expanding. In the South Africa study area, CRCL was deposited due to a loss of flow confinement. In the Magnolia study area, an abrupt decrease in gradient near a basin sill caused flow deceleration and CRCL deposition in off-axis settings. Sedimentation rate and accumulation time were calculated for 44 CRCL sedimentation units from the three areas using TDURE, a mathematical model developed by Baas et al. (2000). For T c divisions and T bc beds averaging 26 and 37 cm thick, respectively, average CRCL and whole bed sedimentation rates were 0.15 and 0.26 mm/s and average accumulation times were 27 and 35 minutes, respectively. In some instances, distinct stratigraphic trends of sedimentation rate give insight into the evolution of the depositional environment.CRCL in the three study areas is developed in very fine-to fine-grained sand, suggesting a grain size dependence on turbidite CRCL formation. Indeed, the calculated sedimentation rates correlate well with the rate of sedimentation due to hindered settling of very fine-and fine-grained sand-water suspensions at concentrations of up to 20% and 2.5%, respectively. For coarser grains, hindered settling rates at all concentrations are much too high to form CRCL, resulting in the formation of massive/structureless S 3 or T a divisions.3

9Submarine and fluvial channels exhibit qualitatively similar geomorphic patterns, 10 yet produce very different stratigraphic records. We reconcile these seemingly 11 contradictory observations by focusing on the channel-belt scale and quantifying the 12 time-integrated stratigraphic record of the belt as a function of (1) the geometric scale and 13 (2) the trajectory of the geomorphic channel, applying the concept of stratigraphic 14 mobility. By comparing 297 submarine and fluvial channel belts from a range of tectonic 15 settings and time intervals, we identify channel kinematics (trajectory) rather than 16 channel morphology (scale) as the first order control on stratigraphic architecture and 17show that seemingly similar channel forms (in terms of scaling) have the potential to 18 produce markedly different stratigraphy. Submarine channel-belt architecture is 19 dominated by vertical accretion (aggradational channel fill deposits), in contrast to fluvial 20 systems that are dominated by lateral accretion (point bar deposits). This difference is 21 best described with the channel-belt aspect ratio, which is 9 for submarine systems and

Changes in sediment supply and caliber during the last ~130 ka have resulted in a complex architectural evolution of the Y channel system on the western Niger Delta slope. This evolution consists of four phases, each with documented or inferred changes in sediment supply. Phase 1 flows created wide (1,000 m), low-sinuosity (1.1) channel forms with lateral migration and little to no aggradation. During Phase 2, the Y channel system began to aggrade, creating more narrow (300 m) and sinuous (1.4) channel forms with many meander cutoffs. This system was abandoned at ~ 130 ka, perhaps related to rapid relative sea-level rise during MIS (Marine Isotope Stage) 5. Phase 3 flows were mud-rich and deposited sediment on the outer bends of the channel form, resulting in the narrowing (to 250 m), straightening (to a sinuosity of 1.22), and aggradation of the Y channel system. Renewed influx of sand into the Y channel system occurred with Phase 4 at ~ 50 ka, during MIS 3 sea-level fall. The onset of Phase 4 is marked by the initiation of the Y′ tributary channel, which re-established sand deposition in the Y channel system. Flows entering the Y channel from the Y′ channel were underfit, resulting in inner levee deposition that is most prevalent on outer banks, acting to further straighten (1.21) and narrow (to 200 m wide) the Y channel. The inner levees accumulated quickly as the flows sought equilibrium, with deposition rates > 200 cm/ky. Marked by the presence of the last sand bed, abandonment occurred at ~19 ka in the Y channel and ~15 ka in the Y′ channel and is likely related to progressive abandonment due to shelf-edge delta avulsion and/or progressive sea level rise associated with Melt Water Pulse 1-A. The muddy, 5-meter-thick Holocene layer has thickness variations that mimic those seen in the sandy part of Phase 4, suggesting that dilute, muddy flows continue to affect the modern Y channel system. This unique dataset allows us to unequivocally link changes in submarine channel architecture to variations in sediment supply and caliber. Changes in the updip sediment routing system (i.e. the channel "plumbing") are shown to have profound implications for submarine channel architecture and reservoir connectivity.