Distinctive iceberg ploughmarks on the mid-Norwegian margin: Tidally influenced chains of pits with implications for iceberg drift

ABSTRACT Curvilinear depressions and chains of asymmetrical to circular pits up to a few meters deep and a few tens of meters wide have been mapped on the upper continental slope beyond the mid-Norwegian shelf. These features are interpreted to represent a continuum of seafloor landforms produced by the grounding of iceberg keels at successive low tides. The features occur along a 50 km long stretch of the upper slope in modern water depths between 400 and 500 m. The average distance between pits varies between about 70 and 175 m, representing one tidal cycle of 12 hr 25 min between successive low tides. This yields average iceberg drift velocities of between 5 and 15 m hr−1. The pits are often asymmetrical, with a downstream surcharge or berm, making it possible to determine the direction of iceberg drift and, therefore, the current direction at the time of ploughmark formation. The c. 300 m thick icebergs producing the ploughmarks are likely to have been sourced mainly from the fast-flowing Norwegian Channel Ice Stream to the south and probably formed during the early part of regional deglaciation of the Norwegian shelf after the Last Glacial Maximum around 20,000 years ago. The icebergs appear to follow a similar northeastward flow direction to that of the present Norwegian Atlantic Current offshore of Norway.


Introduction
Iceberg ploughmarks (sometimes referred to as iceberg scours or furrows), produced when the keels of drifting icebergs contact the sedimentary seafloor, have been observed on many formerly glaciated continental margins around the world . Some of the first reports were from Arctic continental shelves and were based on data from single-beam echo sounders and side-scan sonar instruments (Carsola 1954;Rex 1955;Kovacs 1972;Belderson and Wilson 1973;Syvitski et al. 2001). Extensive areas of ploughmarks on the Norwegian shelf were surveyed from seismic and sidescan sonar data by Lien (1983). More recently, iceberg ploughmarks have been imaged using multibeam echosounders and using three-dimensional (3D) seismic data from many shelf and upper-slope areas in Arctic and Antarctic waters (e.g., Dowdeswell and Ottesen 2013;Gales, Larter, and Leat 2016;Lewis et al. 2016;Wise et al. 2017). The submarine keels of occasional "megabergs" have produced ploughmarks to water depths of 800 to 1,200 m (e.g., Kuijpers et al. 2008;Arndt and Forwick 2016;Jakobsson and O'Regan 2016). Ploughmarks produced by icebergs derived from fullglacial ice sheets have been found as far south as 24°N off Florida in the Northern Hemisphere (Hill 2016) and as far north as 43°S offshore of New Zealand (Stewart et al. 2016) and 47°S on the Patagonian margin in the Southern Hemisphere (Lopez-Martinez et al. 2016).
Iceberg ploughmarks can be many tens of kilometers long, up to several kilometers wide, and 30 m deep (e.g., Dowdeswell and Bamber 2007;Lopez-Martinez et al. 2011;Montelli et al. 2018). Some ploughmarks have a comb-like morphology, which indicates that the icebergs forming them had multiple keels (e.g., Batchelor et al. 2016). In addition, iceberg ploughmarks often have a consistent broad-scale orientation, although interspersed with loops and other small-scale deviations (e.g., Bjarnadóttir et al. 2016;Newton, Huuse, and Brocklehurst 2016;Batchelor et al. 2020), suggesting that mean drifttracks are influenced by and therefore record the flow direction of ocean currents (e.g., Montelli et al. 2018;Newton, Huuse, and Brocklehurst 2018). Intermittent contact between iceberg keels and the sedimentary seafloor is also recorded by the presence of grounding pits instead of continuous ploughmarks (e.g., Bass and Woodworth-Lynas 1988; Barrie et al. 1992).
Iceberg ploughmarks with an assumed tidal signal have been reported only very rarely, however. Some spiral-formed iceberg ploughmarks on a palaeo-shelf buried about 350 m deep on the mid-Norwegian margin have been identified from a 3D seismic cube; they date from a buried surface relating to deglaciation at the end of Marine Isotope Stage (MIS) 12 about 430,000 years ago (Newton, Huuse, and Brocklehurst 2016). The spiral or corkscrew pattern of these ploughmarks was inferred to be a result of icebergs drift under the influence of both tidal and geostrophic currents. This is very similar to the drift tracks of modern icebergs in Antarctic waters (MacAyeal et al. 2008). Jakobsson et al. (2011) described regularly spaced ridges on a surface with glacial lineations in Pine Island Bay, West Antarctica. They interpreted these features to have been formed by the trailing edge of mega-icebergs that rhythmically settled into seafloor sediments under the influence of tidal motion. In addition, Bass and Woodworth-Lynas (1988) reported chains of up to eight pits spaced 25-70 m apart on the Labrador Shelf. They proposed the origin of these chains of pits to be a result of an iceberg keel alternately impacting and rotating free of the seabed and did not invoke a tidal mechanism. A morphologically similar series of pits was imaged on the Falklands Plateau by Brown et al. (2017), ascribed to a partially grounded iceberg being driven forward and oscillating up and down driven by ocean swell.
In this article, we examine in detail a number of distinctive iceberg ploughmarks that contain chains of pits set within curvilinear depressions on the mid-Norwegian margin (Figure 1a,b). The dimensions of these seafloor landforms are surveyed using highresolution multibeam bathymetry. The chains of pits are then interpreted in the context of tidally influenced processes of formation. The morphology of and distance between pits allows calculation of iceberg drift rates and an assessment of paleocurrent directions during the Late Weichselian glaciation. The sources and timing of iceberg delivery to the Norwegian margin are also considered. High-latitude shelves in both hemispheres, especially when less than 400-500 m deep, commonly exhibit curvilinear and circular depressions distributed in a fairly chaotic pattern formed by the interaction of the keels of drifting icebergs with the seafloor (e.g., Harris and Jollymore 1974;Woodworth-Lynas et al. 1991;Dowdeswell et al. 1993), but chains of regularly spaced pits related to tidal variations are rare.
The mid-Norwegian shelf and upper slope are very suitable areas for our investigations given that: (i) this is a continental margin that has been strongly influenced by dynamic ice sheets during both the most recent and many earlier Quaternary glaciations (Dahlgren et al. 2005;Ottesen, Dowdeswell, and Rise 2005;Rise et al. 2005;Ottesen et al. 2009;Montelli et al. 2017); (ii) a comprehensive multibeam echo-sounder dataset at 5 m horizontal resolution exists so that seafloor morphology can be examined in great detail (Bøe et al. 2020); (iii) modern water depths on the outer shelf are approximately 400 m in the deeper cross-shelf troughs, and ploughmarks produced by relatively deep-keeled icebergs have been observed on the shelf and adjacent upper continental slope (Lien 1983); and (iv) postglacial Holocene sedimentation in the area has been minor, with iceberg ploughmarks and other glacial landforms often buried by only a thin layer of sediments up to a few meters thick (Rokoengen and Frengstad 1999;Morén et al. 2018).

Data sources and methods
A high-resolution multibeam echo-sounder dataset imaging 29,700 km 2 of the mid-Norwegian continental shelf and upper slope was acquired during the Norwegian MAREANO seabed-mapping program (www.mareano. no, Bøe et al. 2020). A part of the dataset was obtained by the Norwegian Hydrographic Service (NHS, Kartverket) and a further portion by commercial companies using various multibeam echo-sounder systems. NHS processed the raw multibeam echo-sounder data including depth conversion, tidal corrections, and spike removal . The various datasets have been gridded at a horizontal cell-size of 5 m and have a vertical resolution of better than 0.15 m. Finally, the data were merged into a large regional dataset.
Detailed visual inspection of the seafloor imagery from the upper slope on the mid-Norwegian margin identified several unusual chains of pits and associated ploughmarks ( Figure 2). The tracks of these ploughmarks have been digitized using ArcMap software along the center-line of ploughmarks or from pit to pit where the ploughmarks are not continuous. Individual pits have also been digitized. Length-and cross-profiles have also been measured along the continuous ploughmarks or regularly spaced pits. The distance between individual pits has been calculated from the length profiles.
Although individual iceberg grounding pits may be superficially similar to seafloor pockmarks produced by gas or fluid escape (e.g., Hovland and Judd 1988), we distinguished the two on the basis of the following visual criteria (e.g., Brown et al. 2017): pits are usually concentrated in association with linear to curvilinear iceberg  (Table 1). (c) Overview map of the North Atlantic with the modern ocean-surface circulation (Saetre 2007). NAC is Norwegian Atlantic Current. NCC is Norwegian Coastal Current. Bathymetric data: ©Kartverket. ploughmarks; they are sometimes found in chains and may be asymmetrical as a result of iceberg drift, whereas pockmarks are usually, although not always, circular; berms are often present around all or part of iceberg pits but are typically absent from pockmarks.

Results: morphology of tidally influenced iceberg ploughmarks
We identify and describe three types of iceberg ploughmarks and associated chains of pits, imaged clearly in our high-resolution multibeam-bathymetric datasets ( Figure 2) from the upper slope north of Skjoldryggen on the mid-Norwegian margin ( Figure 1a). The three types of landform appear to represent a continuum of decreasing disturbance of the seafloor (Figure 2).
On the mid-Norwegian upper continental slope, in water depth between 410 and 460 m, nine distinctive chains of linear or curvilinear depressions or pits with lengths between 1 and 8 km have been mapped (Figure 1b), and the dimensions of pits within each chain have been measured (Table 1). With one exception, each of the nine linear features contains pits of at least two different forms along its length. In some cases, the elongate depressions are crossed by one or more large iceberg ploughmarks which locally overprint the features we describe (Figure 1b). Three different types of regularly spaced chains of pits have been identified as follows ( Figure 2): (i) continuous curvilinear depressions with regularly spaced pits along the ploughmark center-line; (ii) curvilinear chains of asymmetrical pits; and (iii) curvilinear chains of circular pits. We describe these features and discuss their mechanism of formation.

Description
This first type of feature consists of well-defined and semi-continuous linear to curvilinear depressions with, along their center-lines, regularly spaced pit-like features up to a few meters deep (Figure 2a,b, Table 1).
Six such curvilinear depressions with regularly spaced pits of similar morphology have been observed (Figure 2a), noting that the two longest depressions (IPMs 1 and 5 in Table 1) are subdivided into several segments in Table 1a. The pits have average depths between 0.9 and 2.2. m (maximum depth 2.8 m) and widths of between about 35 and 50 m. The linear depressions typically have a V-shaped form with sides having angles mainly between 2 and 10°, reaching a maximum of 15°. Berms occur on one or both sides of many of the curvilinear depressions and the pits within them occasionally have small berms (sometimes known as surcharges) at one end, orthogonal to the axis of the depression (Figures 2a, 3b).

Interpretation
The linear to curvilinear depressions are interpreted as iceberg ploughmarks (IPMs), produced by the keels of drifting icebergs that have become grounded on the sedimentary seafloor. The regularly spaced pits within the depressions are interpreted to reflect a tidal signal, with the pits formed by increased indentation of the iceberg keels into seafloor sediments at successive low tides; each pit is separated from the next by a distance equal to the amount of iceberg drift between one low tide and the next. In some cases, the regular pattern described above is broken up into segments or chains that are separated by less disturbed seafloor, indicating that the icebergs may be completely ungrounded from time to time (Figure 3a,c). Such intermittent grounding Table 1. Dimensions of iceberg ploughmarks (IPMs; located in Figure 1b) and associated pits.  and ungrounding is likely to be a result of minor undulations in the local seafloor or even variations in tidal range over monthly cycles, extreme swell conditions, or the inverse barometric effect of deep low-pressure areas.

Description
Morphologically asymmetrical pits occur as chains in which one side of each pit is a simple depression whereas the other has a raised berm or surcharge at its edge ( Figure 2c). Sometimes the asymmetrical pits are completely separated from one another, and in other cases they partly merge with the adjacent pit (illustrated on the right and left sides of Figure 2c, respectively). The pits are mainly 1 to 2 m deep (one has been recorded as 5 m in depth) and vary between 30 and 40 m wide (Figure 2c,d, Table 1b).

Interpretation
Chains of asymmetrical pits are interpreted to have formed when iceberg keels start to plow the seafloor on a falling tide, pushing sediments to each side to produce berms. As the iceberg continues to drift on the falling tide, its keel pushes deeper into the sediment until low tide is reached. The iceberg keel eventually lifts off from the seafloor on the rising tide. In this way, a chain of asymmetrical pits is generated with the up-drift side forming a gradually deepening pit-like indentation and the down-drift side being steeper (Figure 2c, d). Berms or surcharges typically form at the down-drift side of each asymmetrical pit before the iceberg keel detaches from the seafloor on the rising tide. This asymmetry therefore provides a useful indicator of palaeo-current direction, given that iceberg drift is a response mainly to currents rather than wind forcing because about 80-90 percent of a given iceberg is below the sea-surface (Woodworth-Lynas et al. 1991).

Description
In a single location on the mid-Norwegian upper slope (Figure 1b), a chain of circular pits, fully separated from one another, has been observed ( Figure 2e, Table 1c). Within this chain, the average pit depth is 1.6 m; the average pit length and width are 47 and 48 m, reflecting the symmetry of these features.

Interpretation
These typically equidimensional shallow and selfcontained pits are interpreted to have been formed where iceberg keels are largely free-floating and only touch the seafloor close to successive low tides as they drift. Hence, iceberg keels impact the seafloor only lightly and for short periods by comparison with the examples illustrated in Figures 2a and 2c. This interpretation is supported by the limited development of berms associated with sediment being pushed up by the plowing action of the drifting iceberg keels (Figure 2e). Bass and Woodworth-Lynas (1988) described pit-chains of similar morphology from side-scan sonar images of the Labrador Shelf, which they related to the keel of a single iceberg that rolls and impacts the seafloor intermittently rather than to tidal motion.

Results: dimensions of tidally influenced iceberg ploughmarks
The dimensions of the three types of curvilinear depressions and chains of asymmetrical or circular pits (Figure 2), which we have interpreted to be a product of intermittent iceberg grounding on the sedimentary seabed under the influence of tidal motion, have been systematically recorded by visual inspection and subsequent digitization of bathymetric data from the mid-Norwegian upper slope (Figure 1a,b). The features were discovered during regional work for the MAREANO program, which is concerned with mapping the seafloor around Norway (Bøe et al. 2020). The apparent rarity of pit-chains, both in our study area and the wider literature, is probably a result of the delicate interplay between iceberg-keel geometry, seafloor morphology, and tidal effects that is required in order for them to form. We have identified and mapped several tidally influenced ploughmarks on the upper slope beyond the mid-Norwegian shelf (Figure 1b). In this area, the upper continental slope between 410 and 460 m in depth has a gradient of about 0.5°, but its surface is irregular because it is composed of a series of glacigenic debris flows (e.g., Laberg and Vorren 1995;King et al. 1998;Laberg and Dowdeswell 2016), giving the slope a mounded appearance (Figure 1b). Ploughmark dimensions are reported as IPMS 1 to 9 in Table 1, which is broken down into measurements relating to ploughmarks of each of the three types illustrated in Figure 2. The two longest ploughmarks, IPMs 1 and 5, are subdivided into several segments because tidally influenced features are not found along their entire lengths (Table 1).
IPMs 1 to 8 are located along an approximately 40 km-long stretch of the upper slope, whereas IPM9 occurs some 20 km farther south (Figure 1b). All of the IPMs are linear or curvilinear features with a SW-NE long-axis orientation and are either continuous pitted ploughmarks or comprise chains of asymmetrical or circular pits along their center-lines (Figure 3). Examples of bathymetric data showing continuous ploughmarks with regular depressions are illustrated for IPMs 6 and 2 in Figure 3a,b,d. An example of a ploughmark with individual asymmetrical pits from IPM 7 is shown in Figure 3e. One segment of IPM6 also shows an area of deepening water where a transition from a ploughmark with continuous depressions to individual asymmetrical pits takes place (Figure 3a). In more detail, the deepest water along the line of the ploughmark has a smooth seafloor, where contact with the iceberg keel has been lost, with individual pits of increasing asymmetry occurring to either side as the water shallows (Figure 3c).
Ploughmarks IPM1, IPM2, IPM3, IPM4, and IPM5 are aligned in almost exactly the same orientation along the upper slope ( Figure 1b). It is possible, therefore, that these ploughmarks may have been formed by the keel of the same iceberg. The total length of these five ploughmarks is 36 km (Figure 1b); the spacing of center-line depressions and pits leads to a calculated mean drift velocity of 10 m hr −1 implying that, if this was a single iceberg, it took about 150 days to produce the 36 kmlong feature.

Variations in pit spacing: implication for iceberg drift rates
Several factors influence the drift velocity of a plowing iceberg, including ocean-current velocity, water depth, sediment geotechnical properties, and the shape and size of the iceberg keel (noting that icebergs can be multikeeled with several protuberances at the base; e.g., Batchelor et al. 2016). We have measured the distances between successive pits and, from this, have calculated the hourly drift velocity of icebergs along different segments of a number of ploughmarks by dividing the distance between sequential pits by 12 hr 25 min; this represents the time interval between successive low tides. To generate relatively robust velocity values, we have averaged distance values across several pits in a chain; the resulting iceberg drift velocities are given in Table 1.
The average velocity for each chain of pits on the upper slope varies between 5.5 and 14.2 m hr −1 , which implies that the icebergs drift between 132 and 340 m per day or 4 to 10 km per month. Generally, the drifting velocities of the plowing icebergs appear relatively consistent; for example, IPM 1 varies between 5.5 and 8.1 m hr −1 along its four segments (Table 1a). On a broad scale, there is some evidence that iceberg drift velocities increase slightly toward the northeast, from 5-8 m hr −1 (IPMs 1, 2, 9) to 7-14 m hr −1 (IPM 3-8), although the sample size is relatively small ( Table 1).
The drift rates of modern icebergs have been observed using satellite and GPS systems (e.g., Schodlock et al. 2006). Modern data show that relatively large icebergs in the Weddell Sea (with keels between 80 and 550 m deep) drift at about 40 m hr −1 . Icebergs in the Canadian Arctic north of 67°N have been monitored at about 130 m hr −1 , whereas smaller icebergs off Labrador may reach speeds of over 500 m hr −1 , although wind may play an increasing part in total drift rates for small bergs. These drift rates are higher than those we calculate from our morphological evidence on the spacing of pits from the mid-Norwegian margin formed during Late Weichselian glaciation. This is to be expected, however, since our data relate to partially grounded icebergs rather than those that are free-floating and can therefore travel greater distances, as in the modern observations.

Sources and timing of iceberg occurrence on the mid-Norwegian margin
The major source of icebergs to the mid-Norwegian shelf and slope is likely to have been fast-flowing ice streams draining from the Late Weichselian Scandinavian Ice Sheet. Ottesen, Dowdeswell, and Rise (2005) have reconstructed the western marine-terminating margin of the ice sheet, identifying fast-flowing ice streams located in crossshelf troughs and more passive ice on the intervening shallower banks (Figure 4a). The largest ice stream was the Norwegian Channel Ice Stream which terminated at the shelf break in west Norway at 62°N. This ice stream has eroded the Norwegian Channel during the last few glaciations (Sejrup et al. 2003;Rise et al. 2004;Ottesen et al. 2016). Farther north, a number of smaller palaeo-ice streams were present on the mid-Norwegian shelf (Figure 4a) (e.g., Ottesen, Dowdeswell, and Rise 2005;Montelli et al. 2017).
Such fast-flowing ice streams operate in Antarctica and Greenland today (e.g., Mouginot et al. 2017) and are known as effective agents for the transport of both ice and sediments. Similar Quaternary ice streams, producing large volumes of icebergs as a result of their high velocities, are exemplified by the palaeo-ice stream exiting through Hudson Strait from a huge interior drainage basin of the Laurentide Ice Sheet and delivering armadas of deep-keeled icebergs into the North Atlantic (Bond and Lotti 1995;Dowdeswell et al. 1995;Metz, Dowdeswell, and Woodworth-Lynas 2008). Relating specifically to the Late Weichselian ice stream draining through the Norwegian Channel, Nygård et al. (2007) calculated the ice flux for the Norwegian Channel Ice Stream between 20 and 19 ka. An estimated ice velocity of 2.7 km yr −1 yielded an ice flux of 135 km 3 yr −1 . This ice volume would produce very large numbers of icebergs which would initially drift north and then northeastwards following the Norwegian Atlantic Current along the Norwegian margin ( Figure 1c). The extensive pattern of iceberg ploughmarks on the seafloor in the outer part of the Norwegian Channel is shown in Figure 4b. The overall direction of ploughmarks in the Norwegian Channel is from S-N, demonstrating that drifting icebergs were produced at the retreating terminus of the Norwegian Channel Ice Stream. The modern water depth at the channel mouth is about 450 m, implying that deep-keeled icebergs could have been generated from this source which is the deepest part of the outer shelf along the entire mid-Norwegian margin. Given that the pits we describe are found in modern water depths of 410 to 450 m, a full-or early-deglacial timing for the production of the parent icebergs suggests keel depths of about 300 m given Last Glacial Maximum (LGM) sea-level depression of about 120 m (e.g., Lambeck et al. 2014).
There are few radiocarbon dates for the early phase of deglaciation on the mid-Norwegian shelf apart from the Norwegian Channel area to the south (Figure 1). Nygård et al. (2007) documented that the last active phase of the Norwegian Channel Ice Stream was between 20 and 19 ka ago. It is likely that many of the icebergs drifting along the mid-Norwegian margin came from the Norwegian Channel Ice Stream during the early-phase of deglaciation of this area. Morén et al. (2018) concluded that the grounding-line of this ice stream began to retreat from the shelf edge about 19,000 years ago and that the inner Skagerrak was ice-free by 17.6 ka. Rokoengen and Frengstad (1999) reported two dates from the outer part of the mid-Norwegian shelf of 15,320 ± 115 and 13,545 ± 85 14 C yr BP. These are maximum ages for deglaciation, suggesting that the entire shelf was deglaciated in a few thousand years. Knies et al. (2007) suggested deglaciation from the shelf edge around 18 ka (c. 15,000 14 C years) and Vorren and Plassen (2002) dated deglaciation in Andfjorden, north of our study area (Figure 1a), at 14.7 14 C yr BP. In addition, the oldest deglacial dates in the coastal areas of mid-Norway are around 17.5 ka ago according to Olsen et al. (2013) and this also suggests the relatively early deglaciation of the mid-Norwegian shelf.

Ocean current directions and iceberg drift rates
The dominant ocean current on the mid-Norwegian margin is the Norwegian Atlantic Current (NAC) which transports warm water from the Gulf of Mexico across the North Atlantic and into the Barents Sea and Arctic Ocean north of Norway (yellow in Figure 1c). A western branch passes between Iceland and the Faroes, and an eastern branch flows north of the British Islands and through the Faroe-Shetland Channel and into the Norwegian Sea west of Norway (Orvik and Niiler 2002). The NAC follows the continental slope west of mid-Norway with an average velocity of 10 to 30 cm s −1 (36 to 108 m hr −1 ). In addition, the Norwegian Coastal Current (NCC) flows northwards on the inner shelf and along the coastal areas outside mid-Norway and the Lofoten Islands toward the Barents Sea (red line in Figure 1c) (Saetre 2007).
The generally SW-NE orientation of iceberg ploughmarks on the mid-Norwegian margin implies that this was the drift direction at the time icebergs were last present in the region during full-glacial and deglacial times about 20-15,000 years ago. The asymmetry of some grounding pits (Figure 2c) confirms that the actual direction of drift was to the northeast and, thus, that the NAC did not reverse under full-glacial conditions (e.g., Montelli et al. 2018;Newton, Huuse, and Brocklehurst 2018).
The present current speed of the NAC (10 to 30 cm s −1 ) is equivalent to a velocity of 36 to 108 m hr −1 , remembering also that the North Atlantic Meridional Overturning Circulation (NAMOC) has been stronger during much of the last glacial cycle (e.g., Rahmstorf 2002;Böhm et al. 2015). The modern speed is considerably higher than the drift rate we have calculated for icebergs which is between about 5 and 15 m hr −1 (Table 1). This difference is likely to be related to the following factors: (i) slower NAC current speeds during deglacial stages of the Late Weichselian glaciation when iceberg production from the marine-terminating ice sheet was at a maximum; (ii) the retardation of icebergs resulting from partial grounding and plowing of the sedimentary seafloor; and (iii) that icebergs will tend to drift at a slower speed than the prevailing currents due to their mass. Our data cannot be used to distinguish quantitatively between these options, but it is certain that increased friction associated with iceberg grounding and plowing must have played some part in slowing the drift of icebergs as they moved northwards along the Norwegian margin.

Conclusions
• In this study, high-resolution multibeam-bathymetric data were used to survey nine iceberg ploughmarks with well-developed chains of regularly spaced pits from the upper continental slope on the mid-Norwegian margin (Figure 1a,b). • These distinctive iceberg ploughmarks and pits are interpreted to have a clear tidal imprint (Figures 2,  3). This allows calculation of the drift velocity of the icebergs that formed the ploughmarks. • The features take three forms, although there is a morphological continuum between them: (i) pitted depressions, and chains of distinctive (ii) asymmetrical or (iii) circular pits up to a few meters deep ( Figure 2, Table 1). The distance between such pits is rather regular and varies between 70 and 175 m. This represents the interval between successive low tides, which gives average iceberg drift velocities of between 5 and 15 m hr −1 (Table 1). This is slower than the drift rate of free-floating modern icebergs, probably as a result of frictional retardation during ploughmark formation. • The asymmetrical pits often have a downstream surcharge or berm (Figure 2c), making it possible to determine the direction of iceberg drift and, therefore, the current direction at the time of ploughmark formation. Iceberg drift to the northeast during regional deglaciation is similar to the direction of the modern North Atlantic Current. • A large proportion of the ploughmarks on the upper slope of the mid-Norwegian shelf probably originated from deep-keeled icebergs produced by the Norwegian Channel Ice Stream during early deglaciation of the outer shelf around 19,000 years ago ( Figure 4).