The Robeson Channel is a narrow sea water passage between Greenland and Ellesmere Island in the Arctic. It is a pathway of sea ice from the central Arctic and out to Baffin Bay. In this study, we used a set of daily synthetic aperture radar (SAR) images from the Sentinel-1A/1B satellites, acquired between September 2016 and April 2017, to study the kinematics of individual ice floes as they approach and then drift through the Robeson Channel. The tracking of 39 selected ice floes was visually performed in the image sequence, and their speed was calculated and linked to the reanalysis 10 m wind from ERA5. The results show that the drift of ice floes is very slow in the compact ice regime upstream of the Robeson Channel, unless the ice floe is surrounded by water or thin ice. In this case, the wind has more influence on the drift. On the other hand, the ice floe drift is found to be about 4–5 times faster in the open-drift regime within the Robeson Channel and is clearly influenced by wind. A linear trend is found between the change in wind and the change in ice drift speed components, along the length of the channel. Case studies are presented to reveal the role of wind in ice floe drift. This paper also addresses the development of the ice arch at the entry of the Robeson Channel, which started development on 24 January and matured on 1 February 2017. Details of the development, obtained using the sequential SAR images, are presented. It is found that the arch's shape continued to adjust by rupturing ice pieces at the locations of cracks under the influence of the southward wind (and hence the contour kept displacing northward). The findings of this study highlight the advantage of using the high-resolution daily SAR coverage in monitoring aspects of sea ice cover in narrow water passages where the ice cover is highly dynamic. The information will be particularly interesting for the possible applications of SAR constellation systems.
One of the exit gates for sea ice flux from the Arctic Basin to southern
latitudes is through the Robeson Channel. The Robeson Channel is located
between Greenland and Ellesmere Island (Canada), with its northern location
around 82
Map of the Robeson Channel and its surrounding areas. The Robeson
Channel is located between Greenland and Ellesmere Island (Canada), with its
northern location around 82
Oceanographic measurements in the Robeson Channel have not been routinely
performed. Herlinveaux (1971) found the dominant surface current to be from
north to south in April and May, with an average velocity that increased
from about 0.36 km h
The sea ice cover in the Robeson Channel comprises a combination of seasonal
(first-year ice, FYI) and perennial ice (multi-year ice, MYI), both imported
from the Arctic Basin through the Lincoln Sea. The only locally grown ice is
found in narrow strips adjacent to the land at the two sides of the channel.
Based on earlier results for ice thickness and motion retrieved from optical
satellite sensors and reconnaissance flights in the 1970s, Tang et al. (2004) estimated the ice flux transiting the Robeson Channel to be around
Sea ice drift is influenced by wind forcing, ocean currents and internal
stresses within the pack ice. Internal stresses, which are caused by the
interactions between ice floes and determined by the ice types and
concentrations within the pack, reduce ice momentum. Other minor factors
include the Coriolis force and sea surface tilt. The dynamics of the ice
motion can be assessed at a variety of spatial and temporal scales (McNutt
and Overland, 2003), namely individual-floe (
Tracking individual ice floe motion from a sequence of satellite images is potentially feasible if the temporal resolution of the satellite coverage is sufficient (at least daily). An early attempt was reported in Sameleson et al. (2006) for ice in the Nares Strait using coarse-resolution satellite data (tens of kilometres) from a passive microwave radiometer at 6.5 GHz. The authors tracked the motion using only three to five locations of the same ice floe in a sequence of satellite images. Vincent et al. (2001) used Advanced Very High Resolution Radiometer (AVHRR) data (with frequent passes in the polar region) to track the motion of the ice floes in the Nares Strait. However, due to the coarse resolution of the sensor, the results did not demonstrate the motion of individual ice floes.
Due to their fine resolution (tens of metres), sequential SAR images are the best tool to monitor sea ice kinematics, particularly if available at a short timescale. However, SAR data have a limited spatial coverage. The earliest studies to estimate sea ice displacement using SAR data were presented in Hall and Rothrock (1981) and Leberl et al. (1983), using sequential SeaSat SAR images. Later, making use of the more frequent coverage of RADARSAT-1 in the western Arctic, the RADARSAT Geophysical Processing System (RGPS) was developed and produced gridded ice motion and deformation data, tracked every 3–6 d from 1998 to 2008 (Kwok and Cunningham, 2002). A more recent ice tracking operational system (also gridded) was described in Demchev et al. (2017), using a series of Sentinel-1 SAR images.
While the Robeson Channel is covered most of the year by an influx of ice from the Lincoln Sea, the flow of the ice is sometimes blocked in the winter at the entrance of the channel by the formation of an arch-shaped ice configuration that spans a transect between two land constriction points at Greenland and Ellesmere islands. The arch usually collapses in early summer, allowing continuation of the ice flux. Kwok et al. (2010) pointed out that no arch was formed in 2007, leading to a major loss of Arctic ice, which was equivalent to about 10 % of the average annual amount of ice discharged through the much wider Fram Strait (400 km versus 30 km width). This signifies the fact that the entire Nares Strait could represent a major route to the Arctic if the ice arch ceases to form in the future due to the thinning of Arctic ice. Moore and McNeil (2018) addressed the collapse of this arch in relation to the recent trend of sea ice thinning. In this study, the arch formation started on 24 January 2017 and continued until May 2017. The paper includes a detailed description of the mechanism of the arch's development.
The objective of the study was to utilize the daily Sentinel-1A/1B SAR coverage of the Robeson Channel area during a full freezing season (September 2016 to end of April 2017) to examine two sea ice process mechanisms in the Robeson Channel. The first was the drift of individual ice floes, in terms of speed and direction. For this purpose, 39 ice floes were selected and each floe was tracked manually in the series of available Sentinel-1 images. The motion information was linked to the 10 m wind reanalysis data to explore the influence of wind on the ice drift. When wind did not explain the ice motion, an explanation in terms of other factors, namely ocean current, surrounding ice concentration, tidal forces and to a much lesser extent the sea surface height (SSH), was considered. Knowledge about how wind and ice drift are related enables the improvement of sea ice–atmosphere dynamic models (Leppäranta, 2011). The second process was monitoring the development of the ice arch at the inlet of the Robeson Channel during its development until maturity. The advantage of using the daily coverage of the fine-resolution SAR data in retrieving this information in such a narrow channel is expected to instigate further operational applications of SAR constellation systems (e.g. the recent Canadian RADARSAT Constellation Mission (RCM)) with their finer temporal resolutions. It should be noted that operational ice drift products are generated based on cross-correlation between sequential images, which is a statistical approach, while the current approach is based on the manual identification of individual ice floes. Admittedly, this task is laborious, but the motion product has a superior accuracy and can be linked to a detailed dataset of reanalysis wind.
Sentinel-1A and 1B are two satellites developed within the satellite
constellation of the European Space Agency's (ESA) Copernicus programme. They
were launched on 3 April 2014 and 25 April 2016, respectively. Both carry a
carbon-copy C-band SAR sensor (with a central frequency of 5.405 GHz) with a
selection of single or dual polarization. Image acquisition is performed in
one of four operation modes: stripmap (SM), interferometric wide swath (IW),
extra-wide swath (EW) and wave (WV). The IW (swath width 250 km at a spatial
resolution of 5 m
Deviation of the reanalysis wind speed from the speed measured at
the Arctic Alert weather station (reanalysis wind minus station wind) for
the period 1 October 2016 to 30 April 2017. The reanalysis data are from
NCEP/NCAR, NCEP/DOE, ERA-Interim and ERA5. Note the increasing
underestimation of the reanalysis data as the measured wind increases. Units
of the
Information about the available wind data for the study area from the Arctic Alert weather station and the four sources of reanalysis data.
Reanalysis of 10 m level wind is available from a few sources. Four sources
were examined in this study: (1) the US National Centers for Environmental
Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis
project (Kalnay et al., 1996), (2) the NCEP/Department of Energy (NCEP/DOE)
reanalysis model (Kanamitsu et al., 2002), (3) the European Centre for
Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA-Interim) (Dee et al.,
2011) and (4) its successor ERA5 (C3S, 2017). The specifics of each source
are presented in Table 1. The difference between the estimated speed from
each source and the speed from the Arctic Alert station is plotted in Fig. 2
for the period from 1 October 2016 to 30 April 2017. The Arctic Alert
weather station (Canada) is located at 82.52
Figure 2 reveals the overestimation of the reanalysis wind when the
station's wind measurement is
Daily and monthly mean maps of ocean current (vertical coverage at 50 levels
from
The daily fine-resolution images of Sentinel-1A/1B (50 m) were used to generate tracking of the selected sea ice floes and to detect the temporal evolution of the ice arch from its commencement on 24 January until it matured on 1 February 2017. In total, 39 ice floes were tracked between 26 September 2016 and 11 April 2017. A total of 32 ice floes moved mainly southward, crossing the inlet of the Robeson Channel, and seven moved mainly northward within the Robeson Channel. Among the seven ice floes, five moved in the drifted ice regime before and shortly after the formation of the ice arch, and two moved into a polynya-like regime after the formation of the arch. Each ice floe was visually identified in a sequence of daily SAR images (between 11 and 54 sequential scenes), and then the ice floe displacement, drift speed and direction were calculated. The displacement was determined from the subjectively estimated locations of the same ice floe in two successive daily images, using a code to convert latitude–longitude pairs to distance according to the World Geodetic System (WGS84) coordinate system.
Three sources of error were implied in the estimation of the displacement.
The first was the geolocation error of the SAR imagery. The Sentinel-1
product specification (Bourbigot et al., 2016) mentions that the absolute
pixel location accuracy is less than 7 m for the IW mode, but no figure is
given for the EW mode, which was used in this study. The second source
of error was the assumption of a linear path (as opposed to a curvilinear or
meandering path) for the ice floe between 2 successive days. This
assumption had to be employed because the temporal resolution of
Sentinel-1A/1B is not finer than 1 d. The pattern of ice floe motion
depends primarily on the changing wind direction and the mechanical
properties of the surrounding ice floes. The third source of error was the
subjective estimate of the centroid of the same ice floe in successive
images. This was also estimated to be within a few pixels. Assuming that
these errors were independent and normally distributed, the error in the
estimated ice drift speed would be roughly 0.2 km d
The ice floe speed was calculated using the travelled distance, as mentioned above, and the period between the two successive image acquisitions, which varied between 16 and 33 h. To link the ice floe speed to the wind at any ice floe location, the reanalysis wind values from ERA5 for the four grid points closest to the floe were averaged. This was done to avoid the inclusion of wind data from grid points far from the relevant ice floe location, which can often be very different. The 3 h wind vectors that acted on the given ice floe during its transition in the period of the two satellite passes were produced in the form of polar maps, to qualitatively explore their influence on the ice floe drift. In addition, the 3 h wind speeds from the four grid points around each ice floe at each location were averaged to quantitatively link to the drift speed. Statistics were then generated to quantify the wind influence.
The evolution of the ice arch from its onset of formation on 24 January
until its maturity on 1 February 2017 was manually delineated in each image.
The daily displacement of the two end points of the ice arch was calculated
using the latitude–longitude coordinates. To investigate the role of the
wind in the progress of the arch shape, as well as the location and
displacement of its terminal points, daily wind data from the ERA5
reanalysis were averaged from the 3 h intervals at the coloured grid
points from lines 3, 4 and 5 (corresponding to latitudes 82.5,
82.25 and 82
ERA5 grid points of the wind reanalysis data used in the present study. The background is the Sentinel-1B image acquired on 30 January 2017.
The ice floe motion and arch formation are addressed in two subsections. The ice floe motion is considered in two separate regimes: north of the Robeson Channel near its entrance and within the Robeson Channel. In the first regime, the ice floes approach the Robeson Channel in a convergent path forming pack ice cover, which is a term used when ice concentration exceeds 70 %. On the other hand, the ice regime within the Robeson Channel falls into the category of drifted ice, a term used to denote an ice concentration of less than 60 %. The two regimes feature different ice floe drift patterns, and the influence of the wind also differs, as explained later. Case studies of the ice floe motion are presented for each regime to reveal the quantitative and qualitative information about the wind influence. In the second subsection, the role of the wind in the evolution of the ice arch at the inlet of the Robeson Channel is addressed.
The ice flux transiting the Robeson Channel encompasses ice floes of different ages and sizes. The typical dimensions of the ice floes examined in this study ranged from 2 to 16 km. Some ice floes were aggregates of smaller floes, which disintegrated during their journey. The 39 ice floes selected for motion tracking in the Sentinel-1 images were numbered, and the numbers are used in the following analysis, although the order does not carry any significance. The tracks of 12 ice floes are shown in Fig. 4, with the floe numbers and the dates at each position attached. These ice floes were mostly heading southward, but with a few interruptions to this dominant direction. The high ice concentration in the convergent path upstream of the Robeson Channel caused reduction of the ice motion and induced meandering paths. However, in areas with less ice concentration, the ice floe motion accelerated and became more influenced by wind, as will be demonstrated in case study 1. Once the ice floes crossed the bottle neck at the entrance to the Robeson Channel, they became partially relieved from the stresses induced by the surrounding ice and more responsive to other factors such as wind and current. Thus, the speed increased greatly by a factor of 1.5–5, and the drift direction followed mostly the north–south extension of the Robeson Channel, which coincided with the dominant wind direction. This direction also coincided with the dominant ocean current direction. Figure 4 also shows that the ice floes did not enter any of the fjords at the sides of the channel. In fact, many fjords become filled by locally grown landfast ice early in the freezing season.
Trajectories of 12 selected ice floes, obtained from the daily Sentinel-1 images, as they approach and pass through the Robeson Channel. Note the slow motion upstream of the channel and the faster motion through the channel. The entrance to the channel is marked by the solid line in the top left panel.
Figure 5 shows the average drift speed of each ice floe (regardless of drift
direction) during its entire observation time in the SAR time series, either
upstream or within the Robeson Channel. Upstream of the Robeson Channel, the
drift speed varied within a narrow range (4–10 km d
The situation was different for the ice floes that drifted within the
Robeson Channel. Here, the ice floe speed was much higher, typically between
14 and 45 km d
Average speed of individual ice floes during the periods upstream and within the Robeson Channel. The last seven floes (nos. 33–39) drifted within the Robeson Channel with highly variable speeds.
Maps of ocean current near the surface
While several studies have confirmed the coherency between wind forcing and the large-scale motion of pack ice, the results presented in this section are focused on examining the influence of wind on the drift of individual ice floes. For the large-scale motion of the pack ice in the Nares Strait, Kwok et al. (2013) confirmed that this was triggered by both ocean current and wind, which pushed the ice from the Lincoln Sea southward to the Robeson Channel. As mentioned in Sect. 1, ice dynamics at the individual-floe and multiple-floe scales is triggered by a combination of wind, ocean current, internal stress within the pack ice, the Coriolis force, SSH and tidal forces. In the following presentation, when the wind is not found to be linked to the observed ice motion, other factors are considered. A few points about some of these key factors are reviewed here.
Ice concentration is used as a proxy indicator of internal forces within the
pack ice. In this study, this parameter was visually estimated from the SAR
images. In close pack ice (ice concentration
Tide data in this region are not generated regularly. A few datasets are
available, which were acquired during expeditions to measure other oceanic
parameters. For example, Münchow et al. (2007) and
Münchow and Melling (2008) measured ocean currents in the
Nares Strait using mooring buoys, most of which were located at the southern
end of the Kennedy Channel. They found that the tide impacted the dominant
component of current in the Nares Strait. However, Münchow et al. (2007) indicated that the amplitude and phase of the tidal
constituents varied substantially both along and across the strait.
Meanwhile, Johnson et al. (2011) indicated that Petermann Fjord, at
81
Synoptic information about the ice regime north of the Robeson Channel is
presented using the six selected scenes shown in the sequence of Sentinel-1
images in Fig. 7. A bulge-shaped area of consolidated ice appears to be
attached to the coast of Greenland. It is delineated by the dotted lines in
all the images, except in the image for 26 October, although it is still
just visible in this image. This may possibly be a large extent of landfast
ice, although it was exposed to cracking, as can be seen in the image for 7 November. An arch-like crack is visible in the image for 13 November, with
its boundary coinciding with the bottom boundary of the landfast ice area.
This was probably instigated by the strong southward wind (20–60 km h
Figure 7 shows that the ice entering the Robeson Channel follows a path coming around the north of Ellesmere Island, as shown by the arrow in the image for 1 December. No ice appears to be coming along the coast of Greenland. This large-scale motion is likely driven by the strong southward wind, which is channelled down the atmospheric pressure gradient from the Lincoln Sea to Baffin Bay (Gudmandsen, 2000), and the ocean current. However, since the ocean current is very weak in this area (Fig. 6), it is possible that the ice motion around the tip of Ellesmere Island is driven by the west–east gradient of the SSH (Fig. 6). Wekerle et al. (2013) mentioned that the SSH difference between the Arctic Ocean and Baffin Bay not only leads to a net outflow from the Arctic Ocean, but its variability also drives the variation in the Canadian Arctic Archipelago (CAA). As the ice cover approaches the entrance of the Robeson Channel, the ice floes demonstrate erratic motion (Fig. 4). This nullifies the possible influence of ocean current or SSH. The ice floe motion immediately upstream of the Robeson Channel appears to be mainly determined by the interactions between neighbouring ice floes in the closed pack ice. This observation is illustrated in the following case study.
Sequence of Sentinel-1A/1B images for an area north of the Robeson Channel. The dotted curve marks an area of consolidated ice (still visible in the 26 October image). Ice cracked in this area on 7 November and an ice arch was formed on 13 November. The star in the middle panel of the top row marks Ellesmere Island. Ice floes that made their way to the Robeson Channel originate from the west (not north), following the path shown by the arrow in the 1 December image.
Drift speed of ice floe no. 2 and no. 3 (shown in Fig. 8) during the period between the acquisitions of the two successive daily Sentinel-1 images. The period is shown in the first column.
Figure 8 shows sequential Sentinel-1 images (26 September to 7 October 2016) of a segment just upstream of the Robeson Channel, where two ice floes appear. Ice floe no. 2 is marked by the grey dot, a natural low-backscatter area, and floe no. 3 is marked by the star. The corresponding maps of the 3 h ERA5 reanalysis wind vectors are presented in Fig. 9. The daily speeds of each ice floe are listed in Table 2, along with the wind and qualitative concentration data. This information helps in defining the impact of the wind on the ice floe drift, as explained below.
The image for 26 September shows the two ice floes surrounded by open water
and thin ice. The wind between the two satellite overpasses on 26 and 27 September (averaging 33 km h
Sequential Sentinel-1A/1B images (dates are shown) showing the advancement of two ice floes. Floe no. 2 is marked by a grey dot (a natural low backscatter area), and floe no. 3 is marked by a star. The ice concentration surrounding each floe is visible and can be qualitatively estimated.
Maps of the 10 m level wind vector (km h
Prior to the formation of the ice arch on 24 January 2017, the channel was
filled with ice floes transported from the north. Shortly after the
development of the ice arch, the channel became covered with thin ice and
open water, which is typical of polynya cover. In both situations, the
direction of the ice floe motion was mainly north–south, following the
dominant wind and current directions, although the wind was occasionally
reversed (Fig. 4). The low ice concentration (
The trend of the data points in Fig. 10 is defined by the linear regression
equation
Scatter plot of wind versus ice floe speed components along the Robeson Channel direction. Positive and negative values pertain to wind or ice motion heading north or south, respectively. Data from 39 floes drifting within the Robeson Channel are shown. The open circles pertain to 37 ice floes that originated north of the Robeson Channel and then formed part of the drifted ice regime in the Robeson Channel (where many ice floes existed). The closed circles represent data from two floes that originated inside the Robeson Channel and then drifted in the polynya formed after the ice arch was formed. The dashed line is the linear regression for the data of the 37 floes.
Results from the multivariate regression analysis showing the contributions of wind and ocean current to ice floe motion.
Isolating the wind and ocean current contributions to ice motion can be
better achieved using a modelling approach, as presented in Thorndike and
Colony (1982) and Kimura and Wakatsuchi (2000). However, in order to achieve
this task using the present data of daily gridded wind and ocean current, we
performed multivariate regression analysis, with the wind and current data
as the independent variables and ice floe speed as the dependent variable.
Only the components along the Robeson Channel extent were considered. The
results are shown in Table 3. The standardized coefficient is a measure of
how much an independent variable explains the dependent variable. In this
case, the wind and current speed can explain 0.729 and 0.165 of the ice floe
motion, respectively. Statistical significance is the probability of
rejecting the null hypothesis, which is no significant difference between
the contribution of wind and current, in this case. The Pearson correlation
coefficient is a statistic that measures the linear correlation between two
variables. Here, the wind shows a better linear correlation with the ice
floe motion. The partial correlation coefficient is a measure of the
correlation between the dependent variable and one independent variable, in
the absence of other independent variables. Once again, the results show the
more significant contribution of the wind. The variance inflation factor
(VIF) should be
Figure 11 shows a sequence of daily Sentinel-1 images from 14 to 19 November, where many ice floes originating from the north of the Robeson
Channel can be seen. The path of the ice floe marked with the asterisk (floe
no. 29) is linked to the coincident wind vectors in Fig. 12. This ice floe
moved southward at a speed of 27.0 km d
A sequence of daily Sentinel-1 images showing the path of a number of ice floes. The floe marked by the asterisk (floe no. 29) is the subject of the comments in the text. Dates of the images are shown, as well as the speed of the marked floe.
Maps of the 10 m level wind vectors (km h
Figure 13 shows the track of ice floe no. 38, which broke off from landfast
ice at the Greenland side and drifted north and then south in the polynya
regime. The trajectory covers the period from 8 to 22 February 2017, after
the arch formed. The daily wind vector maps associated with the selected
floe location are presented in Fig. 14. Between 10 and 11 February,
northward wind dominated, although this never exceeded 20 km h
Trajectory of an ice floe (floe no. 38) that separated from landfast ice and drifted in the polynya regime downstream of the ice arc. The track is shown from 8 to 22 February 2017.
Maps of the 10 m level wind (km h
The ice arch phenomenon is a necessary condition for polynya formation downstream. Polynyas can be driven by wind action that removes newly formed ice (latent heat polynya) and/or warm upwelling ocean water that melts the ice as soon as it is forms (sensible heat polynya) (Smith et al., 1990). However, if the flux of ice from a nearby source continues to feed into the area that would become a polynya, then the polynya can only be formed if a natural obstacle develops to block the flux. This obstacle could be an ice arch, which is a mechanically strong formation that can withstand the massive dynamic load of the advected sea ice. Clearly, this factor is irrelevant to coastal polynyas as they are backed by land. This is more common in the Antarctic region (Nihashi and Ohshima, 2015). In the case of the Robeson Channel, an ice arch commonly forms at the inlet of the channel, blocking the ice flux from the Lincoln Sea into the Robeson Channel. The ice arch may collapse a few weeks after formation or persist as late as mid-August (Samelson et al., 2006). More historical context about the ice arches that form at the inlet of the Robeson Channel can be found in Kwok et al. (2010), Ryan and Münchow (2017), and Moore and McNeil (2018). The ice arch observed in the present dataset started its development on 24 January, matured on 1 February and collapsed in May 2017. The mechanism of arch development is described below. After its initial formation, chunks of ice continued to detach from the arch's contour under the action of the southward wind. This altered the arch's shape and the location of its terminal points along the two constriction points at the Greenland and Ellesmere Island sides. The sequence of the development is revealed in the set of Sentinel-1 images shown in Fig. 15.
The white dashed line that appears in some panels represents the arch's contour of the following day. For example, the dashed line in the image for 24 January represents the arch's contour that appears in the image for 25 January, and so on. No line is presented if the contour remained unchanged in the following day (e.g. the cases of 25 and 26 January). The difference between the visible arch and the dotted line in the image of any given day identifies the ice that was detached by the wind action on that day. The 3 h wind vectors during the period between the two daily overpasses of Sentinel-1, obtained from the ERA5 reanalysis, are presented in Fig. 16. Data were obtained from the grid points located on lines 3, 4 and 5 in Fig. 3 and are shown in the same colour as that figure.
Daily Sentinel-1 images showing the development of the arch formation from 24 January 2017 until it matured on 2 February 2017. The dotted line marks the arch shape and location in the following day.
Wind speed (km h
After 7 d of persistent northward wind, southeastward wind returned
for a few hours between 22 and 23 January. A wide rupture of ice cover, not
forming an arch shape, can be observed in the 23 January image (Fig. 15).
Between 23 and 24 January, the dominant southward wind (about 30 km h
The arch legs, which is an engineering term that refers to the end parts of the arch, can be observed to be perpendicular to the land contour (see the images for 1 and 2 February in Fig. 15). As all the forces exerted on the arch's contour are transferred as compression forces, the arch legs must be perpendicular to the surface, in order to provide a robust way to transfer the load directly to the rock base at both sides. Otherwise, the end point of the arch may continue to slip at the surface, leading to eventual failure of the arch (Karnovsky, 2012). This feature, along with the curvature of the arch, will be of interest to the ice mechanics community.
Sequential Sentinel-1 images showing the breakup and drift of an ice piece that was labelled floe no. 39 in this study. This modulated the ice arch. The arrow indicates the dominant wind direction between the acquisition times of the images for 4 and 5 March.
The above discussion highlights the mechanism of ice arch formation. To reiterate, strong southward wind plays an important role in modulating the arch's contour as it may cause detachment of pieces of ice at the locations of fractures. Northward wind, on the other hand, has virtually no effect on the arch's shape and location. On two occasions, ice pieces were observed to detach in the presence of a light northward wind, which suggests the possible influence of the sea surface current. The modulation of the arch's shape abates when the upstream ice becomes too compacted to allow crack formation, and hence further detachment of ice pieces occurs in response to a southward wind. Moreover, the mechanically strong structure of the arch cannot fail under the dynamic force of the incoming ice flux. The structural properties of the arch were not addressed in this study, except for the observed configuration of its terminal points (arch legs) being perpendicular to the land surface.
In this study, a series of daily Sentinel-1A/1B images were used to study the sea ice motion at the scale of individual ice floes in the Robeson Channel, which is located between Greenland and Ellesmere Island, and the process of ice arch formation at the northern entry of the channel. The study period spanned the autumn and winter seasons of 2016/2017. Wind data from the ERA5 reanalysis were used to explore the role of wind on ice floe drift and the arch formation process. Daily gridded data of ocean current were also used. In total, 39 floes were visually tracked in the sequential daily images and their velocity vectors were calculated. Qualitative and statistical data of the ice drift were obtained in two regimes, upstream and within the Robeson Channel. Case studies showing links between the drift of selected ice floes and their driving forces were presented. The local reanalysis wind was obtained from the closest grid points to the ice floe at each location on its trajectory.
Sea ice that approaches the Robeson Channel follows a path around the
northern section of Ellesmere Island. No ice drift was observed along the
coast of Greenland in this case. In the convergent zone that leads to the
channel, ice floes drift at a fairly constant speed of around 5 km d
Once an ice floe crosses the entry to the channel and becomes released from
the stresses engendered by the surrounding ice, it starts to accelerate.
While inside the channel, the ice floe drift speed varies between 15 and 45 km d
Ice arch formation and development were monitored using the daily Sentinel-1 images over a 9 d period from 24 January to 1 February 2017. During this period, pieces of ice along the arch's contour continued to crack and detached under the action of southward wind. Northward wind had no role in this process as it closed and tended to stabilize the arch. The process continued until the pack ice upstream of the arch became fully consolidated and the arch took on a mechanically strong concave shape.
The findings of this study will provide clues to enhance the dynamic modelling of ice by identifying conditions that accentuate the role of wind in ice motion. The study has also demonstrated the possibility of generating non-gridded drift vectors of individual sea ice floes by tracking their motion in a sequence of daily SAR images. Such a product would be important for operational ice mapping as it can identify the distribution of hazardous floes. Daily SAR images covering a limited number of geographic regions have become available from the Sentinel-1 system. They may soon be available from the recently launched RADARSAT Constellation Mission (RCM) (a fleet of three satellites) but only upon request. More SAR constellation systems are expected in the future from several national and commercial agencies. The challenge of generating maps of ice floe drift for sequential SAR images resides in developing an automated identification method for ice floe contours, considering their deformation, rotation, breakup and amalgamation while drifting.
Sentinel-1 data are available free of charge from the Copernicus Open Access Hub (European Space Agency, available at
MES prepared the manuscript, designed the experiments and performed the data analysis. ZHW collected and processed the data. TTL contributed to the data analysis and supported the writing and editing.
The authors declare that they have no conflict of interest.
We are grateful to the following organizations for providing the data used in this study. The European Space Agency (ESA) provided the Copernicus Sentinel-1A/1B product. The ECMWF provided the ERA5 and ERA-Interim reanalysis products. NCEP, NCAR and DOE provided their respective reanalysis products. The CMEMS provided their GLORYS12V1 reanalysis product. The authors would also like to thank the two anonymous reviewers, whose comments have improved the manuscript greatly.
This work was supported in part by the National Key Research and Development Program of China (no. 2018YFC1406102), the fund of the Key Laboratory of Global Change and Marine-Atmospheric Chemistry (no. GCMAC1806), and the National Natural Science Foundation of China (nos. 41676179 and 41941010).
This paper was edited by Yevgeny Aksenov and reviewed by two anonymous referees.