Surprising turbidity current activity in a land-detached canyon
Acoustic Doppler Current Profilers (ADCPs) on two deep-water moorings in the eastern branch of Whittard Canyon recorded profiles of water column velocity and backscatter (a proxy for sediment concentration17,25) from June 2019 to August 2020 (Fig 1). A down-looking high-frequency (600 kHz) ADCP positioned 30 m above the seafloor on Mooring M1 (1591m water depth; 26 km downstream of the canyon head) recorded at 1 m intervals, every 5 minutes. Mooring M2 (2259 m water depth; 47 km downstream of the canyon head) included a 75 kHz up-looking ADCP placed 14 m above seafloor that recorded at 26 m intervals, every hour. The ADCP at M2 operated with a blanking distance of 25 m, thus inhibiting detection of flows <39 m thick. A sediment trap was also located at 10 m above the seafloor on Mooring M1 to sample suspended sediment. This trap was programmed to sample for 18 days, after which point a carousel mechanism rotated a new sampling bottle into place. Two surface buoys managed by the UK MetOffice provided hourly meteorological data during the monitoring period (Fig. 1 & 2).
Six turbidity currents were detected within the 1-year monitoring window, with maximum ADCP-measured velocities of 1.1-5.0 ms-1 (Fig. 2; Fig. 3; Table 1). All six flows were recorded at M1, three of which were also clearly identified 21 km downstream at M2. The two slowest (1.1-1.3 ms-1) and thinnest (15-20 m) flows (Flows 3 & 4) were not detected at M2; either dissipating before reaching M2, or were too thin for detection by the lower frequency upward-looking ADCP. It is not possible to discern whether Flow 6 reached M2 as it occurred shortly after M2 stopped recording (Fig. 2). Transit velocities of the turbidity currents estimated between M1 and M2 range from 2.7-8.0 ms-1 (Table 1). The instantaneous (ADCP-recorded) velocities recorded at M1 are similar to the transit velocities of the respective flows, albeit at the lower end; in keeping with previous results that indicate ADCPs often under-record velocities compared to transit speeds19,33. Upon recovery of M1, the first sediment trap bottle and overlying funnel were completely filled with well-sorted sediment with an average grain size of 121 μm, and a maximum of 460 μm (Fig. S2). This significant sedimentation event occurred within the first 18 days of monitoring; a time window that included the first turbidity current, hence we interpret that this, and presumably later flows, were capable of suspending fine sand at a height of at least 10 m above the seafloor. Sediment stocks, and hence the source of the flows, likely derive from extensive sediment wave fields at the Whittard Canyon rim that dominantly comprise fine to coarse sand, for which off-shelf transport has been inferred from high-resolution seafloor surveys34. Plastic fishing line was observed wrapped around the M1 mooring anchor chain (Fig. S1), supporting evidence of active transport of litter as inferred from previous studies13.
A frequency and magnitude of flows equivalent to active land-attached canyons
Dating of cored deposits previously indicated that episodic turbidity currents have reached the Celtic Fan at the distal end of the Whittard Canyon within at least the last 2,000 years, and turbidity current deposits accumulated in the proximal reaches of the system27,30,35. However, this is the first field study to document that these flows can occur on such a short recurrence (i.e. sub-annual frequency), and velocities exceed previous measurements in the Whittard Canyon by an order of magnitude. Down-canyon flows of <0.8 ms-1 were recorded from benthic landers at 1400 and 4200 m water depth, but could not always be confidently differentiated from the background effects of internal tides in the canyon, which regularly attain near-bed velocities of ≥0.5 ms-113,30. Perhaps most surprising, is that the observed frequency (6 in a year) and velocity (up to 5 – 8 ms-1) of flows in Whittard Canyon is on a par with highly active land-attached deep-sea submarine canyons (Table 2), such as the river-connected Congo Canyon (6 flows in 4 months at 2 km water depth; 1-2.4 ms-117) and the littoral-fed Monterey Canyon (15 flows in 18 months, of which only 3 reached 1.9 km water depth; 1-8 ms-119). Indeed, the velocities and frequency reported here are some of the highest yet directly recorded from turbidity currents worldwide (Table 2).
A major trigger is not required for turbidity currents, nor is a consistent trigger in effect
Based on inference from NE Atlantic land-attached canyons (e.g. Nazaré Canyon), storms were inferred to be responsible for triggering previously reported turbidity currents in Whittard Canyon13,30. While Flow 2 coincided with a storm (local minimum in air pressure and maximum in wave height and wind speed), many of the other storms during the monitoring period, including more vigorous events, did not correlate with the occurrence of turbidity currents (Fig. 2). Surprisingly, no turbidity currents were recorded during the European winter storm season, when wave heights and air pressures experience the largest excursions36 (November-March; Fig. 2). Analysis of other triggers proposed for turbidity currents in submarine canyons, including earthquakes, surface tidal and internal tidal currents, and seafloor disturbance by fishing, showed no consistent finding. Five earthquakes of magnitude 2 and over occurred within 1000 km of the canyon head, but none coincided with any of the flow timings, and no larger (> Mw 6) earthquakes occurred within 2000 km during the monitoring period (Table S1). Turbidity currents occur during both spring and neap tides and are not correlated with any particular phase of the semidiurnal surface tidal cycle. Flows 1 occurs when the surface tidal flow is down-canyon, but Flow 2 and Flows 4 and 6 occur when the flow is up-canyon (Fig. 4). Near-bed, currents caused by the trapping of internal tides occur along the canyon13,30,37,38. However, turbidity currents occur during periods of both low and high internal tide magnitude, as well as during both down-canyon and up-canyon phases of near-bed internal tidal flow (Fig. 4). Fishing occurs daily around the head of Whittard Canyon (Fig. 2f) and there is no obvious connection to turbidity current inception when considering all types of fishing. When considering only fishing activities that directly disturb the seafloor, some flows (e.g. flows 1,4,5) appear to coincide with periods of heightened dredge fishing and bottom trawling. However, this relationship is far from equivocal, as flows did not occur on other days that had much higher bottom fishing intensity.
It appears that turbidity currents in Whittard Canyon do not require a major trigger, and instead likely occur following a period of preconditioning (i.e. from sustained or sudden sediment supply); at which point a number of minor perturbations, are capable of initiating a flow. This adds to growing evidence of seasonally-clustered activity first documented in land-attached canyons, where preconditioning during and after periods of heightened sediment supply governs turbidity current timing and frequency, rather than requiring external triggers. For river-fed canyons, seasonal pulses in river discharge are the primary control on sediment supply, while for land-attached canyons fed by long-shore drift, heightened wave energy during the winter storm season explains turbidity current activity26. In Whittard Canyon, however, the storm season is absent of turbidity currents and instead the more meteorologically-quiescent summer months are more active. Sediment transport on the Celtic Shelf is complex, as is the topography of the dendritic branches of the Whittard Canyon, which exerts a strong control on hydrodynamic processes such as internal tides13,30,37,40. We suggest that the ‘switch on’ outside of the storm season may be explained by seasonal variability of cross-shelf transport on the Celtic Margin and from the adjacent Bay of Biscay (to the SE), wherein sediment transport toward the canyon head is enhanced during summer months39-42. We conclude that the lack of consistency in a trigger, and the seasonal clustering of flows, result from this combined complexity of spatiotemporally-variable hydrodynamic processes and sediment supply, which may also be further complicated by anthropogenic disturbances such as shelf-edge and deep sea fishing.
Underestimation of contemporary particulate transport in land-detached canyons
This recognition of frequent highstand turbidity current activity in land-detached canyons is important, as more than 75% of the >9000 submarine canyons worldwide are land-detached (Fig. S3). We cannot infer that all these land-detached canyons will be similarly active. Canyons of the Antarctic margin, for instance, are formed and maintained by dense cascades of cold water43. However, analysis of a global database reveals at least 10% of the world’s canyons have a very similar setting to Whittard Canyon; separated >100 km from shore by a broad shelf, where sediment stocks have accumulated since the Last Glacial Maximum, typically occurring on passive margins1,2,11. We infer a 50% increase in the number of canyons worldwide that may potentially feature active turbidity currents in the present-day highstand, including some of the largest systems on Earth20,44. An additional n=1162 such land-detached canyons can therefore be added as potentially-active systems, to the existing n=2104 land-attached canyons that efficiently convey particulate matter from shelf to deep sea (Fig. S3). While such canyons may not connect directly to terrestrial sources of modern organic carbon, these systems can still be effective conveyors of fresh labile organic carbon to the deep sea. In the Whittard Canyon, phytoplankton blooms can generate elevated quantities of phytodetritus that is rich in organic carbon30. As these blooms occur in the spring and summer30 (i.e. the same period within which we observe most frequent turbidity currents), it is conceivable that the powerful flows we observe play a role in the down-canyon transfer of fresh organic material, as well as anthropogenic material such as the discarded fishing gear found wrapped around our the anchor chain of mooring M1 (Fig. S1). To further quantify particulate fluxes, and constrain the potential for their activity, it is important to better characterise the nature and quantity of sediment transport on the continental shelf adjacent to these canyons26,29,34, the energy and rate of cross-shelf transport relative to the canyon head40, and the relative role of human activities, such as bottom fishing, that can markedly influence sediment fluxes at the shelf-scale41. Future monitoring efforts should thus not focus solely on land-attached canyons and must include land-detached canyons, many of which host well-documented carbon, nutrient, pollution and biodiversity hotspots and intersect routes for new seafloor cables that are vulnerable to the impacts of turbidity currents8,12,45,46.