The ice conditions of the Arctic Ocean during Pleistocene glacial maxima are highly controversial. Extreme viewpoints vary from an ice-free condition (Donn and Ewing, 1966) to a thick ice shelf covering the Arctic Ocean (e.g., Mercer, 1970; Hughes et al., 1977; Jakobsson et al., 2016). The direct geological evidence of the latter is the numerous submarine erosional bedforms formed by ice shelf grounding on bathymetric highs in the central Arctic Ocean, such as the Mendeleev Ridge, the Lomonosov Ridge, and the Chukchi Borderland (Niessen et al., 2013; Stein et al., 2010; Jakobsson, 1999; Polyak et al., 2001; Jakobsson et al., 2008, 2016; Dove et al., 2014). These erosional bedforms suggest that the maximum thickness of the ice shelf can exceed 1 km (Jakobsson et al., 2016). An alternative view is provided by Kristoffersen et al. (2004), who suggested that armadas of large icebergs disintegrated from land-based ice sheets could also cause such bedforms. These deep-water glacial bedforms not only constrain the extent of the Arctic ice shelf in geological history, but also provide direct observational data for estimating the maximum ice volume around the Arctic Ocean during glacial maxima.

Along the Chukchi continental margin, thick glacial deposits revealed by multi-channel seismic data indicate a long history of glaciation (Hegewald, 2012; Hegewald and Jokat, 2013). Although there are some early views suggesting almost no ice at the Chukchi continental margin during the Pleistocene (e.g., Ehlers and Gibbard, 2007), many studies have revealed a variety of glacial bedforms in this region (Fig. 1). Borehole dating of glacial bedforms on the southernmost Northwind Ridge reveals two recent phases of glaciations, possibly during MIS 2 and MIS 4 to 5d (Polyak et al., 2007). Some studies have even proposed an ice sheet capping the Chukchi margin during previous glacial maxima, implying an ice shelf extending beyond the Chukchi Shelf break (Polyak et al., 2001; Dove et al., 2014). However, the extent and maximum thickness of such an ice sheet/ice shelf are not well established. In areas shallower than 700 m water depth of the Chukchi Borderland, many glacial bedforms, such as mega-scale glacial lineations (MSGLs) and morainic ridges, have been discovered (Polyak et al., 2001; Jakobsson et al., 2008; Dove et al., 2014). However, in deeper water, only a few glacial erosional bedforms have been found (Jakobsson et al., 2008; Dove et al., 2014). One typical example is the streamlined bedforms found at about 900 m water depth on the crest of a seamount located at about 75.5°N (called the “Baoshi Seamount” in this paper) in the Northwind Abyssal Plain (NAP) (called MSGLs in Dove et al., 2014). MSGLs are representative bedforms beneath ice streams (e.g., Shipp et al., 1999; King et al., 2009) and are widely found in cross-shelf troughs on glaciated continental shelves (Ottesen and Dowdeswell, 2009). Similar streamlined bedforms are observed on the submarine highlands away from the wide continental shelf, attributed to the grounding of ice shelves and icebergs (e.g., Polyak et al., 2001; Kristoffersen et al., 2004) instead of ice streams. Obviously, when restoring the extent and thickness of the possible ice sheet/ice shelf in the Chukchi area, it is vital to identify whether these streamlined bedforms are caused by ice shelf or icebergs. Nevertheless, Dove et al.’s (2014) multibeam data on the Baoshi Seamount only reveal part of the bedforms (Fig. 2b) whose origin is questionable. In this paper, a bathymetric grid that fully covers the Baoshi Seamount using newly acquired and previously published multibeam data (Dove et al., 2014) is presented. With this approach, the objectives of this paper are: (1) to determine the genesis and ice source of these sub-parallel linear glacial scours on the Baoshi Seamount; (2) to provide geological constraints for restoring the extent and volume of ice in the Chukchi area during glacial maxima.

The NAP is a N–S trending flat valley between the Chukchi Plateau-Chukchi Rise in the west and the Northwind Ridge in the east (Hall, 1990) (Fig. 1). It is an extensional basin formed during the late Paleocene (Grantz et al., 2011) or the latest Cretaceous–late Paleocene (Ilhan and Coakley, 2018). The Baoshi Seamount is one of the seamounts in the NAP close to the Chukchi Shelf, and the water depth at its top is less than 1 000 m (Figs 1 and 2).

The multibeam data used in this paper were primarily acquired by an ELAC SeaBeam 3020 ICE echo sounder mounted on the R/V Xuelong during the 9th Chinese Arctic Research Expedition (CHINARE-Arc9) in the summer of 2018. In the Baoshi Seamount area (the white rectangle in Fig. 1), the surveyed water depths range from about 2 200 m to 850 m, and the swath width is about 2 times the water depth. The survey lines are roughly E–W (Fig. 2a). The bathymetric survey of CHINARE-Arc9 is complementary to previous investigations in this area (e.g., Dove et al., 2014) and explores more detailed bedforms on this seamount.

The multibeam data acquired during CHINARE-Arc9 were processed by CARIS HIPS software. After processing was completed, a grid with a cell size of 50 m was generated (Fig. 2c), and the data were exported as Generic Sensor Format (GSF). At the same time, open source software MB-System (Caress and Chayes, 2017) was used to edit the historical multibeam data from four cruises in the study area shared by the US National Geophysical Data Center (NGDC), including HLY0302, HLY0404 and HLY0703, by the US Coast Guard’s Icebreaker Healy in 2003, 2004 and 2007, respectively, and MGL1112 by R/V Marcus G. Langseth in 2011. These open-access multibeam data are combined with the new data from CHINARE-Arc9, and consequently, a compiled grid with a cell size of 50 m is generated (Figs 2a and d). The combined data enable a nearly full illustration of the study area, revealing much more complete glacial bedforms than in the past. In the quantitative topographic analyses, if the bedforms are within the data extent of the CHINARE-Arc9 grid, the CHINARE-Arc9 grid is used with a higher priority. Otherwise, the compiled grid is used.

The compiled multibeam grid shows that the Baoshi Seamount is approximately an ellipse in plain view with its long axis along the N–S direction (Fig. 2). Its orientation is similar to those of the other seamounts in the NAP revealed by the IBCAO V3 grid (Jakobsson et al., 2012) (Fig. 1) and other high-resolution multibeam data (e.g., data from the MGL1112 cruise). The N–S trend of the Baoshi Seamount is also consistent with the trends of faults in the southern NAP revealed by the latest multi-channel seismic profiles (Ilhan and Coakley, 2018), suggesting that the genesis of the Baoshi Seamount may be related to the extension of the NAP.

The Baoshi Seamount has two relatively flat regions, one of which is the flat-topped crest (Fig. 2d). A series of sub-parallel scours are identified on the crest at water depths of 850–1 030 m, with an orientation of about 75°/255° (Figs 2 and 3). The water depth of these scours deepens from north to south. According to their morphology and spacing (Figs 3 and 4), these scours can be roughly divided into two categories, located in the northern and southern zones, respectively (Fig. 3).

The scours that developed on the crest of the Baoshi Seamount are not parallel to the survey tracks; thus, the possibility of the scours being artefacts caused by surveying direction can be excluded. In addition to glacial origin, persistent currents can also form lineations on the seafloor. Jakobsson et al. (2008) described the differences between glacial-derived and current-generated lineations. According to these criteria, lineations on the Baoshi Seamount clearly can be ascribed to glacial-origin, as the ridges and troughs of these scours have similar cross-sectional shapes and dimensions (Fig. 4), and they are located on the crest of the seamount, which excludes the possibility that they were generated by persistent currents.

There are two possible factors that can cause glacial linear scours on the deep seabeds, iceberg ploughing (e.g., Dowdeswell et al., 2010) and grounding of the ice sheet/ice shelf (e.g., King et al., 2009; Polyak et al., 2001). Although single-keel icebergs may sometimes generate linear scours, they usually form irregular and chaotic scours (e.g., Figs 6d and 9b of Dowdeswell et al., 2016), which are unlike the sub-parallel scours in the study area. Such sub-parallel linear scours may also be generated by armadas of large icebergs (e.g., Kristoffersen et al., 2004) or multi-keel icebergs (e.g., Wise et al., 2017).

According to the scour orientation, the direction of ice flow may be either from SW to NE or from NE to SW. The longitudinal topographic profiles of the scours in the northern zone and the northern subzone of the southern zone show that the water depths at both ends are almost identical (Figs 5a and b), while that of the southernmost scour shows a slightly larger depth at the SW end (1 030 m) than at the NE end (1 000 m) (Fig. 5c). However, it is difficult to indicate the ice flow direction using only these profiles. In addition, the compiled bathymetric grid that almost covers the entire Baoshi Seamount does not reveal any glacial bedforms that can indicate ice flow direction (such as grounding zone wedges and moraine ridges). Therefore, both directions of ice motion are possible.

If these scours were caused by icebergs, then, where were the icebergs from? The Chukchi Shelf in the southwest seems to be a possible source. MSGLs have been identified in the bathymetric trough east of the Chukchi Rise, some of which are at about 700 m water depth around the shelf break (Dove et al., 2014). If the ice sheet that produced these MSGLs was very thick, then the calved icebergs of this ice sheet might be thick enough to plough over the crest of the Baoshi Seamount. Another possible source is from the east. MSGLs were found at a water depth of about 900 m on the southern Northwind Ridge, southeast of the Baoshi Seamount, with a NWW trend (J900 in Fig. 1) (Jakobsson et al., 2008). These MSGLs were attributed to ice masses from the Laurentide Ice Sheet on northern Canada (Jakobsson et al., 2008). If the linear scours on the Baoshi Seamount were carved by the Laurentide icebergs passing the Northwind Ridge, these icebergs need to undergo a special path of movement, i.e., go to the southwest or northeast of the Baoshi Seamount, and then drift northeastward or southwestward over the Baoshi Seamount. Although this possibility cannot be ruled out, it requires a very special dynamic oceanographic environment, and the possibility of such scenarios is very low. To the northeast of the Baoshi Seamount, there is an area of MSGLs on the Northwind Ridge whose orientation is consistent with that of the scours in the study area (Dove et al., 2014). The water depth of this site reaches up to 700 m. If the paleo ice cap here is thick enough, it may also produce icebergs that can generate scours on the Baoshi Seamount.

Another possibility is that these linear scours are caused by ice sheet/ice shelf grounding. According to the geomorphologic characteristics and zonation of the scours, those from the northern zone indicate that the bottom of the ice shelf is in complete contact with the seabed, so the scours are similar to MSGLs beneath the ice streams. In the southern zone, the scours are deeper and scattered, indicating that only some downward-protruding deep keels of the ice shelf are partially in contact with the seabed. In the Arctic Ocean, similar glacial scours are also thought to be caused by ice shelf grounding (Polyak et al., 2001). Previous studies indicate a Chukchi Ice Sheet existing on the Chukchi Shelf, which may even be connected to the East Siberian Ice Sheet (Niessen et al., 2013) to form the Eastern Siberian-Chukchi Ice Sheet (Dove et al., 2014). A paleo-ice stream moving from SW to NE was identified southwest of the Baoshi Seamount (Dove et al., 2014). These findings make the East Siberian-Chukchi margin similar to the Ross Sea shelf in West Antarctica (Anderson et al., 2014), indicating that ice supply to the Baoshi Seamount may be sufficient by some ancient ice streams from the Chukchi Ice Sheet, possibly forming a thick ice shelf and carving the seabed. In addition, the Northwind Ridge is believed to be affected by the Laurentide Ice Sheet (LIS) (e.g., Polyak et al., 2001; Jakobsson et al., 2008). Although the predominant direction of the LIS ice flow over the southern Northwind Ridge and NAP is NWW (e.g., Dove et al., 2014; Jakobsson et al., 2014), it seems that some paleo ice streams of LIS (such as the M’Clintock ice stream) can affect the Northwind Ridge and Baoshi Seamount in a local SWW direction, thus causing scours with orientations of NEE–SWW on the Baoshi Seamount. Of course, it is also possible that neither of the above two ice streams in opposite directions are not thick enough to ground on the Baoshi Seamount, but the interaction between the two (e.g., convergence) might thicken the ice in the NAP and make it possible to ground on the crest of the Baoshi Seamount as deep as about 1 000 m.

Regardless of which of the above two geneses is correct, the ice at the Chukchi margin to the southwest of the Baoshi Seamount, the southern NAP or the Northwind Ridge northeast of the Baoshi Seamount must be thick enough to produce such scours. Assuming a sea level drop of approximately 130 m during glacial maxima (such as during the Last Glacial Maximum, Lambeck et al., 2014), the minimum ice thickness during glacial maxima (${H_{{\rm{GM}}}}$) can be calculated using the formula below, which is similar to Eq. (2) of Anderson et al. (2014).

where ${\rho _{\rm{w}}}$ (1 028 kg/m³) and ${\rho _{\rm{i}}}$ (910 kg/m³) are the mean densities of the sea water and ice column, respectively. $D$ is the present-day water depth of the scours on the Baoshi Seamount. $\Delta {D_{{\rm{GM}}}}$ is the water depth at glacial maxima minus the present-day water depth (−130 m according the above assumption). For the deepest scour in the study area, $D$ is 1 030 m; thus, ${H_{{\rm{GM}}}}$ is 1 017 m, which means that during glacial maxima, the Chukchi margin to the southwest of the Baoshi Seamount or the southern NAP or the Northwind Ridge northeast of the Baoshi Seamount was covered by a 1-km-thick ice mass.

(1) New multibeam data on the Baoshi Seamount acquired during the 9th Chinese Arctic Research Expedition are presented in this paper. These data fill almost all the blank areas in the multibeam data of the seamount from previous studies. The compiled data revealed much more complete glacial scours on the Baoshi Seamount than previous data.

(2) A series of sub-parallel scours are identified at a water depth of 850–1 030 m on the crest of the Baoshi Seamount (Figs 2 and 3). The scour water depth generally deepens from north to south. According to the morphology and spacing (Figs 3 and 4), these scours can be roughly divided into two categories, located in the northern and southern zones (Fig. 3). Scours in the northern zone are almost equally spaced, while the spacing between neighboring scours in the southern zone varies significantly.

(3) Based on the morphological characteristics of the scours, it is suggested that they are either caused by multi-keel icebergs/armadas of large icebergs, or ice sheet (ice shelf) grounding. In the above two interpretations of bedform genesis, the ice masses that generate the scours may be from the Chukchi continental margin to the southwest or the Laurentide discharge to the northeast of the Baoshi Seamount. The deepest scour suggests that the thickness of the ice mass generating these scours reaches about 1 km.

We thank Captain Quan Shen and the crew of the R/V Xuelong during the 9th Chinese Arctic Research Expedition. Figures 1–3 were drawn using the GMT package developed by Wessel et al. (2013).