During summer 2014 and summer 2015 two autonomous Seagliders were operated over several months close to the ice edge of the East Greenland Current to capture the near-surface freshwater distribution in the western Greenland Sea. The mission in 2015 included an excursion onto the East Greenland Shelf into the Norske Trough. Temperature, salinity and drift data were obtained in the upper 500 to 1000 m with high spatial resolution.
The data set presented here gives the opportunity to analyze the freshwater distribution and possible sources for two different summer situations. During summer 2014 the ice retreat at the rim of the Greenland Sea Gyre was only marginal. The Seagliders were never able to reach the shelf break nor regions where the ice just melted. During summer 2015 the ice retreat was clearly visible. Finally, ice was present only on the shallow shelves. The Seaglider crossed regions with recent ice melt and was even able to reach the entrance of the Norske Trough.
The data processing for these glider measurements was conducted at Alfred Wegener Institute (AWI). The first part consists of the Seaglider Toolbox from the University of Each Anglia; the second was exclusively composed for the data from the Greenland Sea.
The final hydrographic, position and drift data sets can be downloaded from
The Nordic Seas are shaped by a strong near-surface salinity contrast arising from the northward flow of saline Atlantic Water along their eastern rim in the Norwegian Atlantic Current (NwAC) and West Spitsbergen Current (WSC) and the southward flow of fresh Polar Water and sea ice along their western rim in the East Greenland Current (EGC) (Fig. 1). Due to strong cooling in winter, the Nordic Seas are one of the few regions in the world ocean where convection normally reaches depths of 500 to 2000 m (Nansen, 1906; Rudels et al., 1989; Budéus and Ronski, 2009). With this intermediate water ventilation the Nordic Seas contribute substantially to the Atlantic Overturning Circulation (Schmitz and McCartney, 1993; Lumpkin and Speer, 2003). The convective overturning depends on the density stratification, which in the cold Nordic Seas is mostly set through salinity.
The map shows the Nordic Seas. Topographic contours are given on the basis of RTOPO2 (Schaffer et al., 2016): the 1000 m contour is marked in red, the 3000 and 300 m contours in black, and the 2000, 500 and 200 m contours in gray. The inlet marked by the full blue line shows the area of Fig. 2; the inlet marked by the dashed blue line shows the area of Fig. 3. Red to yellow arrows indicate the cooling of the warm and saline Atlantic Water as it flows through the Nordic Seas and Arctic Ocean. The blue arrows indicate the flow of cold and fresh Polar Water through the Nordic Seas. EGC – East Greenland Current; WSC – West Spitsbergen Current; NwAC – Norwegian Atlantic Current; FS – Fram Strait; GS – Greenland Sea; NT – Norske Trough.
Freshwater leaves the Arctic with the EGC in liquid form and as sea ice. For
the liquid export de Steur et al. (2014) estimated 2100 km
During late summer, low salinities were frequently observed in the near-surface layer of the deep basin of the Greenland Sea (GS) (Latarius and Quadfasel, 2016). However, this seasonal signal shows large inter-annual variability in magnitude and vertical extension. A fresh surface layer stabilizes the water column and may reduce wintertime convection in the GS (Latarius and Quadfasel, 2016). Oltmanns et al. (2018) observed the effect of slashed convection by a fresh surface layer for the Irminger Sea in winter 2010/11. Very likely, the freshwater in the inner western Nordic Seas originates from the EGC, yet the explicit sources as well as the transport mechanisms are still unclear. De Steur et al. (2015) revealed local ice melt as the primary source for freshwater in the GS for summers 2011 and 2013. Dodd et al. (2009) found that a significant amount of sea ice leaves the EGC into the Nordic Seas whereas the liquid freshwater remains in the EGC up to the Denmark Strait. However, it is also possible that liquid freshwater from the EGC reaches the inner western Nordic Sea. The transfer to the interior Nordic Seas may take place by eddies shedding off the Polar Front (Spall, 2011; Lherminier et al., 1999).
To investigate the spreading of freshwater from the western rim into the convection regions of the inner Nordic Seas, the project “Variation of freshwater in the western Nordic Seas” was conducted in the framework of the Research Group FOR1740: “A new approach toward improved estimates of Atlantic Ocean freshwater budgets and transports as part of the global hydrological cycle”, funded by the German Research Association (DFG). The goal of this project is to capture and analyze fluctuations of freshwater in the western Nordic Seas and understand related processes. To achieve the necessary observations over several months with high spatial resolution, two missions were conducted in the summer months of 2014 and 2015 in the western GS using autonomous gliders (Fig. 2). The gliders were operated in ice-free regions, but close to the ice edge. As glider navigation in shallow waters is difficult, the sections were limited to areas with water depth greater than 300 m.
Maps of the Seaglider missions in 2014
General information on the two glider missions in summer 2014 and summer 2015. SN denotes serial number.
In this paper we describe the details of the two missions in the challenging region near the ice edge and the shallow shelf east of Greenland (Sect. 2), the processing of the data, and appendant uncertainties and error estimates (Sect. 3). In the last section, we give a brief description of the observations.
The hydrographic and drift data of both glider missions were published in the World Data Center PANGAEA.
The development of the ice cover in the western Greenland Sea during
the time span of the glider missions in summer 2014 and summer 2015. Left
column for summer 2014, right column for summer 2015. Month and day of the
individual ice concentration maps are given in the upper left corner. The
maps are based on ice concentration data made available by Drift+Noise
(
The glider missions in summer 2014 and 2015 took place in the GS, which is the northernmost basin of the Nordic Seas (Fig. 1). The GS basin is up to 3600 m deep and flanked to the west by the steep continental slope east of Greenland. The EGC flows along the shelf break and the slope from the Arctic Ocean to the North Atlantic, transporting water masses of Arctic origin and sea ice. The West Spitsbergen Current, the northward extension of the Norwegian Atlantic Current, flows to the north along the eastern shelf break and slope, transporting mainly water of Atlantic origin. In the Fram Strait part of the flow continues to the Arctic Ocean, while another part of the Atlantic Water recirculates and joins the EGC, thereby subducting below the Polar Surface Water (Quadfasel et al., 1987; Hattermann et al., 2016; von Appen et al., 2015).
Sea ice is transported with the EGC from the Arctic Ocean to the Nordic Seas. During winter, local sea ice formation also takes place in the western Nordic Seas. The ice primarily covers the western shelf and only in extreme winters reaches the deep GS (Comiso et al., 2001, 2008; Comiso and Hall, 2014).
Seagliders are buoyancy-driven autonomous underwater vehicles that move through the water in a sawtooth pattern between the sea surface and a prescribed dive depth (Davis et al., 2003; Rudnick et al., 2004). Data are recorded during the dive and climb (down- and upward motion) and transmitted via satellite to the base station during every surfacing. At that time, the glider can receive commands concerning its flight behavior and direction and its data sampling scheme. Typically, the glider is instructed by a target file, containing waypoints, about the planned courses of the mission, and by a science file about the sampling frequencies for the different sensors (see Table 1 and Sect. 3.1 with Table 2). New command files are sent if tuning of the flight behavior is needed.
Origin of the gliders 127 and 558 and their setup during the two missions in the Greenland Sea in summer 2014 and summer 2015. SN denotes serial number
For a given dive depth and dive time the glider's internal flight model calculates the needed buoyancy change and trim of the instrument for a sawtooth pattern of down- and upward motion in direction to the next waypoint. The hydrodynamic shape and the small fins of the glider support the steering. The flight model additionally calculates the vertical velocity of the glider during dive and climb, which is used in the post-processing of the data. During every surfacing, the flight model compares the calculated position with the real one determined by GPS. From the discrepancy the depth-averaged current is calculated. If requested, these depth-averaged currents can be used during the following dives in the flight model for an advanced calculation of the course to the next waypoint. If water depths less than the prescribed dive depth are expected, information from altimeter bottom tracking can be used for the ending of the downward motion.
The buoyancy of the glider is changed by changing the volume through inflation/deflation of an oil bladder (similar to profiling floats). The pitch (downward/upward orientation of the instrument) is changed by repositioning the center of mass by moving the battery pack forward/backward. Buoyancy and pitch together determine the angle of the downward or upward motion. To control the roll of the instrument an additional weight is fixed axial asymmetrically at the battery pack. As gliders behave like an “underwater sailplane”, turning the battery to the right or left forces the glider to turn horizontally to the right or left accordingly.
Deep and slow dives need less energy and thus allow longer missions than shallow and fast dives; furthermore, they allow better steering between waypoints. Conversely, shallower dives increase the horizontal resolution and faster dives allow us to capture sections in shorter time. Seagliders were used in the described missions because high energy supply is characteristic for this type of glider. The used instruments are restricted to work in ice-free water.
During summers 2014 and 2015, Seaglider missions were carried out in the western GS. The goal was to capture the spreading of freshwater from the western rim into the inner Nordic Seas. For this goal we run the glider(s) along a section between the deep GS basin and the EGC. Repeating the section in 2015 allowed observation of the variability of the spreading both during the course of the individual summers as well as between the two summers.
In 2014, the measurements started with an east-to-west section. Because of
the ice coverage in the early summer, the mission had to be changed later to
a southeast-to-northwest section (Fig. 2) perpendicular to the isobaths. For
comparability, the southeast-to-northwest section was carried out in 2015 too
(from 75
The focus of the project was on the near-surface hydrography and thus high
horizontal resolution in the upper part of the water column was of foremost
interest. Nevertheless, because of its unstable flight behavior during summer
2014 glider 127 was set to dive nearly always to 1000 m depth to achieve
good steering. Glider 558, which had a more stable flight, was set for more
than 70 % of the dives to 500 m depth. On 21 August, glider 127 stopped
diving because the battery was almost empty (voltage cutoff). The remaining
energy was used to continuously send positions ashore. The instrument was
recovered by MV
Until the middle of July, the ice situation at the observation site was quite similar in 2014 and 2015 (Fig. 3): the broad shelf east of Greenland was covered with ice and the ice reached at least the position of the 1000 m depth contour. In the months after July, however, the ice coverage evolved differently in the two years.
In 2014, the ice between the perennial fast ice east of Greenland (Schneider and Budéus, 1995) and the central Fram Strait decreased continuously until September. Nevertheless, during the whole summer an ice tongue remained above the shelf break and the upper slope. Thus the ice edge, located above deep waters, was always reached by the gliders at the northeasternmost position of the glider sections. However, the ice tongue prevented the glider from being operated across the EGC and to the shelf.
In 2015, a similar gap between the fast ice east of Greenland and the ice coverage in the Fram Strait developed. However, in that year the shelf break and slope were completely ice-free from mid-July to mid-September. A large part of the shallow shelf was also ice-free, but was already ice-covered again in the beginning of September. This situation gave the opportunity to extend the sections to the shelf. With altimeter bottom tracking, a number of dives were carried out at around 300 m bottom depth. In August a more northern line across the shelf break to the inlet of the Norske Trough (Arndt et al., 2015) was executed. However, since the navigation of Seagliders in shallow water is problematic, the ice edge was never reached in 2015.
The glider missions in summer 2014 and summer 2015 give insight into the distributions of temperature and salinity in the upper part of the water column. In summer 2015, the distribution was also observed in regions where the ice coverage just disappeared. The observations can be interpreted in relation to the different ice coverage (see Sect. 4).
The gliders were equipped with sensors for temperature, conductivity,
pressure, oxygen (glider 127 only) and optical parameters (Table 2). However,
during the missions, only temperature, conductivity and pressure data were
recorded. Temperature and conductivity sensors have been calibrated by
Sea-Bird (
During every surfacing, the data of the gliders were sent ashore. On the base station, the data of each dive were decoded and transformed into two files containing the scientific data (eng file) and the technical data as well as information about the setting of the piloting parameters (log file). These files were the basis for the real-time analysis of the glider performance. If changes in the flight behavior were necessary, the pilots submitted new command files to the base station, which were transferred to the glider during the next surfacing. Sometimes new target files were also sent for changes in the planned track, or new science files for changes in the sampling of the sensors. In addition, the data processing is based on the eng and log files.
A glider measures temperature, conductivity and pressure while it is moving
vertically and horizontally through the water. The relation between
horizontal and vertical movement during our missions was
For a proper calculation of salinity and density, temperature and conductivity measurements from the same water body are needed. Otherwise spikes occur, especially if the glider was diving through strong gradients. Due to different response times and different placement of the sensors on the instrument, different water bodies were measured by the temperature and conductivity sensors at the same time. This time lag was corrected by an alignment of temperature and conductivity data, which involves a vertical interpolation of the measured conductivity values.
The thermal lag, induced by the different geometry and the heat capacity of the cells, was corrected in a second step. This error lasts over several consecutive measurements. The correction was derived from the original data but it failed if extreme outliers were still present in the data.
Glider CTD data are more distorted than ship-based CTD data because of the following.
The vertical resolution of the glider measurements is low compared to that
of a ship-based CTD. The gliders sample with 0.2 Hz (every 5 s). With a typical vertical
velocity of 0.1 to 0.12 m s A ship-based CTD samples with 24 Hz (every 0.04 s) and is typically
lowered at 0.5 to 1 m s The vertical velocity of a glider is much more variable than that of a
lowered CTD because the change in buoyancy (which accelerates the glider) is
calculated with the flight model using a prescribed vertical density
structure. The latter might deviate from the real local density structure.
Due to the lower sampling rate, information about the vertical velocity of
the instrument is also much coarser for gliders than for ship CTDs. The water is not pumped through the glider's conductivity cell because
of limited energy resources; instead, the cell is freely flushed. Thus, the
calculation of the flushing time depends on the uncertain vertical velocity
(see 2.). The quality of the interpolation for the time-lag correction depends on
the vertical resolution, the information about the vertical velocity and the
flushing time of the cell (see 1., 2. and 3.; and Alvarez, 2018, for
numerical analysis of the performance of unpumped SBE sensors at low flushing
rates). The thermal-lag correction depends on the geometry and flushing time of
the conductivity cell (see 2. and 3.). The sensors are calibrated before mounting them on the glider, which
limits the applicability of the calibration.
For these reasons glider data are much noisier and the time lag and thermal-lag correction are of lower quality for glider CTD data than for ship-based CTD data. With respect to these problems, a data processing was set up at AWI, which consists of two parts (Table 3).
Flow chart of glider data processing with UEA toolbox and AWI
additions. For each individual profile, the eng and log files contain the
scientific and technical data. The sg_calib_constants file contains the
information about the pre-deployment calibration of the individual gliders.
Vbd – vertical buoyancy device; UEA – University of East Anglia;
First, the raw data, as provided by the Kongsberg base station, were transformed to physical units and merged with the pre-mission calibration information. Corrections were applied to the data according to information from the gliders flight model and for the time lag and thermal lag of the sensors. This part was carried out by means of the UEA Seaglider Toolbox.
In addition to that, an extended processing was developed and applied to the glider data to exclude erroneous data, interpolate the data to discrete pressure levels, smooth the derived quantities, and adjust absolute temperature and salinity to data from high-precision ship-based CTD casts close to the glider mission in space and time. An analysis was also made to determine if down- and upcast data showed systematic discrepancies and thereupon it was decided if both could be used or not. This second part of the processing involved knowledge of the regional hydrographic conditions.
A prerequisite for a proper functioning of the thermal-lag correction is the exclusion of erroneous data/spikes. Thus, the first and second parts of the processing are entangled.
The individual steps of Table 3 are described in the following.
The movement, and thus the exact position, of the glider under water was
derived from a flight model. These data were corrected by comparing the
vertical velocity of the glider, calculated as change of pressure with time,
with the flight model vertical velocity. The correction of the flight model
has influence on derived variables such as dive-averaged currents, but also
on the calculation of conductivity. In two steps various parameters were
fitted so as to minimize the difference between the vertical velocities
(Frajka-Williams et al., 2011). For details of the UEA toolbox see
Also, all unrealistic data were deleted. The limits were temperatures lower
than The mean vertical velocity (
Finally, pressure, temperature with time-lag correction, conductivity with
thermal-lag correction, and the derived quantities salinity and density were
saved. In the final data set the variable NOBS gives the number of observations from
which 2 dbar means were calculated. If NOBS is empty for a certain line of
data, values for temperature, conductivity, salinity and density were
interpolated.
Sea-Bird temperature sensor and free-flushed conductivity sensors, referred to as the CT sail, were installed on Seagliders 127 and 558. The CT Sail consists of three parts, the “CT” temperature sensor and conductivity cell, the temperature circuit board, and the conductivity circuit board. These parts were disassembled and reassembled each time the CT Sail was calibrated and afterwards installed into the glider again. The calibration was conducted in advance for both missions. As the process of disassembly and assembly was not within the control of Sea-Bird, the applicability of the calibration is limited.
The specifications of the CT sail are as follows:
initial accuracy: conductivity typical stability: conductivity (comparable to SBE 37, Sea-Bird Scientific).
Exemplarily, three temperature, conductivity and salinity profiles from glider 127 during summer 2015 demonstrate how spikes in salinity show up in the depth of the highest vertical gradient between the surface layer and the water masses below. These spikes are generated by an insufficient time-lag correction (see Sect. 3.3.2 for details).
Errors in the form of spikes in salinity occurred when the gliders moved
through sharp gradients. The spikes were produced by insufficient alignment
between temperature and conductivity in relation to pressure. Negative spikes
were expected if the conductivity measurement was before the temperature
measurements and positive spikes if conductivity lagged behind temperature
(Sea-Bird Scientific software documentation:
We decided to leave the decision how to deal with the spikes to the users of the data set. To help identification of affected profiles we list them in Appendix A. The spikes will possibly level out during gridding or averaging routines in further processing. For example, Queste et al. (2016) developed a method to deal with glider measurements across sharp gradients. They built composite profiles from the downcasts between the surface and the thermo-/halocline and from the upcasts between maximum depth and thermo-/halocline and combined these in a gridded data set.
By visual inspection of all individual profiles at different steps of the
processing, several individual faulty values or profiles are detected.
Spikes in salinity in the depth range of the thermo-/halocline were
removed, if they exceeded 0.1 (see Sect. 3.3.2). Wrong values during the apogee, which were not removed by the criterion
Outlier profiles of conductivity, which are considerably
separated from the entity of profiles of a mission, were removed. Profiles with large gaps in the depth of the largest gradient are as follows. If the
gaps exceeded a depth range larger than the typical depth range of the
thermo-/halocline ( Incomplete profiles are as follows. When the dive was aborted by the glider-intrinsic
software after an uncommanded change in the bleed counts of the vertical
buoyancy device, these profiles were removed.
No individual temperature, conductivity or salinity values were removed, but complete data lines or even the whole profiles were always removed before the interpolation to 2 dbar levels took place. This results in a reduction of the original data sets between 2 % and 5 % (Table 4).
Summary of the data quality and effects of the different steps of the data processing for the three individual data sets (glider 127 mission 2014, glider 558 mission 2014 and glider 127 mission 2015). References to the specific steps of the data processing are given in the left column.
As described above (Sect. 3.2) the original measurements were averaged over 2 dbar to reduce the noise of temperature and conductivity. The 2 dbar mean values were on average based on three or four original data points. Figure 5 shows the number of values for the averages and the standard deviations, exemplarily for glider 127 during the mission in 2015 (for gliders 127 and 558, mission 2014, the figures look similar).
There was a small difference in the number of records between the down- and
upcast due to slightly different velocities. The reduced numbers at the
maximum depth of the profiles and at the beginning of the downcast reflect
the rejection of data during the start of the dive and during the
apogee/bottom dead point, where the vertical velocity was below
5 cm s
Standard deviation
The standard deviations of temperature and conductivity below 200 dbar depth
are small (0.01
To quantify the noise reduction resulting from step B.5, where the criterion of stable density was applied, we calculated the variability of a profile before and after the step. The variability is defined here as the difference between consecutive values of salinity in a profile. Figure 6 shows the variability for all individual salinity profiles; again, glider 127 during the mission in 2015 was used as an example. Above 80 dbar the noise reduction is of the order of 10 %, but between 45 % and 63 % below. The average reduction for the whole depth range is 22 % for glider 127 in 2014, 13 % for glider 558 in 2014 and 33 % for glider 127 in 2015.
Deep sections (0 to 1000 dbar), one for each summer as an example.
Sections: 2014, glider 127, sections 1–5. Shown is the upper 55 m
of the water column. Left column: information and map extract with location
of the profiles and ice edge for the time span of the section (see Sect. 2.4
for details); numbers of original dives on the section, numbers of profiles
from the final data set on the section, time span, and direction of the
section:
Sections: 2014, glider 558, sections 1–7. Shown is the upper 55m
of the water column. Left column: information and map extract with location
of the profiles and ice edge for the time span of the section (see Sect. 2.4
for details); number of original dives on the section, number of profiles
from the final data set on the section, time span, and direction of the
section:
Sections: 2015, glider 127, sections 1–9. Shown is the upper 55 m
of the water column. Left column: information and map extract with location
of the profiles (no ice was observed during the whole mission time in the map
extract); number of original dives in the section, number of profiles from
the final data set in the section, time span, and direction of the section:
The adjustment of absolute values to ship-based CTD data was at least an order of magnitude larger than the accuracy of an SBE 37 and thus demonstrates the urgent need of high-quality ship-based CTD data in the vicinity of a glider mission.
Values for the different inaccuracies are summarized in Table 4.
Possibilities and constraints for using the glider data presented here are briefly discussed at the end of Sect. 5.
The glider data are available from the PANGAEA data
archive. The digital open-access data library PANGAEA is a publisher for
Earth system science and hosted by the Alfred Wegener Institute, Helmholtz Centre for Polar and Marine Research (AWI) and the Center of Marine Environmental Sciences, University of Bremen (MARUM). The glider
data are stored in two ASCII
The main link is
Sub-links are as follows:
This section provides a brief description of the temperature and salinity distributions measured during the two summer glider missions.
Typical hydrographic conditions in the GS reflect the major circulation features in the Nordic Seas. The deep Greenland Basin is bounded by the EGC on the western side. Cold and fresh Polar Surface Waters are transported with the current from the Arctic Ocean into the subpolar North Atlantic. On the eastern side, the Greenland Basin is bounded by the West Spitsbergen Current. Waters of Atlantic characteristic – warm and salty – flow along the Norwegian coast and shelf break to the north and undergo cooling on the way. In the Fram Strait, part of the Atlantic Water continues into the Arctic and part recirculates (Rudels et al., 2005; Hattermann et al., 2016). The recirculated Atlantic Water joins the EGC and partly subducts below the Polar Surface Water (see Fig. 1) (Rudels et al., 2002). The central GS is dominated by Arctic Intermediate Waters, which are relatively cold and salty (Blindheim and Rey, 2004; Rudels et al., 2005, 2012). The deep basin of the GS is filled with this water mass, formed by local convection during winter. A very weak stratification is characteristic for the inner GS, reflecting the winter convection. Strong seasonal variations are only observed on top of the Arctic Intermediate Water. The near-surface layer is dominated by summer heating and winter cooling. Additionally during late spring and summer occasional freshening is observed (de Steur et al., 2015; Latarius and Quadfasel, 2016).
Our observations with the gliders captured the Arctic Intermediate Water in
the central GS and the Polar Surface Water as well as the Recirculating
Atlantic Water near the ice edge in the west, thus confirming the typical
hydrographic conditions. However, only in 2015 was the core of the EGC with cold
(below
The glider observations give insight into the interannual variability close
to the surface. The warm near-surface layer was around 20 m thick in 2014 but
up to twice that thick in 2015 (see Fig. 7). We expected that due to a much
more extended ice coverage the summer heating started later in 2014 (see
Sect. 2.4). However, the most obvious difference between the summers 2014 and
2015 was the near-surface salinity distribution. In summer 2014 waters with
very low salinities (31–33) reached up to 3
Also in 2015 the lowest salinities in the near-surface layer were observed at the northwestern end of the section. However, the signature was different. The water was restricted to the western end of the section and was not as fresh as in 2014 and the freshwater was not that concentrated near the surface. As described above it reflects the Polar Surface Water flowing with the EGC from the Arctic Ocean through the Fram Strait to the south and continuing its way along the shelf break to the subpolar North Atlantic (Rudels et al., 2005). Also the development of the ice coverage during summer 2015 (Fig. 3) suggests that this water mass was not a signature of recent ice melt. Since the beginning of August, the shelf break was not covered with ice. The ice retreated to the shallow shelf close to the coast until mid-September.
The presented distributions of temperature and salinity, measured along sections from the inner GS to the EGC during summer 2014 and summer 2015, show signs of freshwater intrusions close to the surface. The development within a single summer as well as the interannual differences are demonstrated. The freshwater intrusions are not masked by the inaccuracies of the measurements, as we described in detail in Sect. 3, as the absolute difference between the Polar Surface Water and the Arctic Intermediate Waters is of the order of 4–6 K for temperature and 2–4 for salinity. For further analyses, one has to take into account that opposite to ship-based CTD sections, glider sections are never “quasi-synoptic”. Thus, the combination of low time resolution and high spatial resolution provided by glider measurements must be considered, when deriving quantitative conclusions from the observed distributions.
For Figs. 7 to 10 all profile data of the final data set along a specific
section were gridded in the horizontal at 0.05
For details see Sect. 3.3.3.
Glider 127 2014: Dive nos.: 10–13, 17, 11, 24, 76, 82, 206–208, 212–214, 220–227, 229–231, 233–234;
Glider 558 2014: Dive nos.: 1, 3–13, 15–25, 85–86, 91–93, 101–103, 110–112, 116–121, 125–127, 390;
Glider 127 2015: Dive nos.: 2–7, 9–17, 19–32, 34–67, 75–77, 106–107, 109–115, 117–124, 167–226, 230, 233, 329–420.
The dive no. is named
The authors declare that they have no conflict of interest.
We thank Bastian Queste from the University of East Anglia
for developing and making the UEA toolbox publicly available and
supporting us in using it.
During the mission piloting we got prompt, detailed and friendly
support from Harald Rohr, Optimare, Kongsberg Germany and
Kongsberg USA.
During the missions, Thomas Krumpen and his colleagues from
Drift+Noise provided high-resolution ice information.
We would like to thank the captain and crew of RV
This research has been supported by the German Research Association (DFG) within the framework of the Research Group FOR1740: “A new approach toward improved estimates of Atlantic Ocean freshwater budgets and transports as part of the global hydrological cycle” (grant no. SCHA 376/3-2).
This paper was edited by Robert Key and reviewed by two anonymous referees.