Greenland Submarine Melt Water Observed in the Labrador and Irminger Sea
Abstract
Helium and Neon distributions from 1994 to 2015 are used to identify and quantify submarine melt water (SMW) from the Greenland Ice Sheet in the Labrador and Irminger Sea. SMW fractions >0.2% (maximum 0.62 ± 0.075%) are confined in the upper 400 m of the Greenland shelf and slope and account for 12 ± 6% of the total east Greenland freshwater flux. SMW is embedded in water of different mixtures of cold and fresh Polar Water and warm and saline Atlantic Water. The resulting salinity range from that mixture exceeds the effect of SMW addition by one order of magnitude. Up to now, SMW is not detectable in the formation region of the Labrador Sea Water. Episodically, He and Ne excess anomalies are found in the deep ocean, probably introduced by the East Greenland Current spill jets.
Key Points
- The noble gases He and Ne elucidate the distribution of submarine melt water (SMW) off Greenland
- SMW fractions up to 0.6% are confined in the upper 400 m of the Greenland boundary current
- SMW is not detectable in the formation region of Labrador Sea Water
Plain Language Summary
In a warming world, ice sheets from Greenland do not only melt at the surface, but at the part swimming in seawater also from below. This fraction of the melt is tracked in the ocean by measuring the concentrations of the noble gases Helium and Neon. They have been trapped in the ice and are released into the seawater during melting. We found out that most of the submarine melt water remained close to Greenland and has yet not intruded into the interior of the ocean.
1 Introduction
Mass loss from the Greenland Ice Sheet (GrIS) increased significantly over the past two decades. Besides the growing influence on sea level rise (Dieng et al., 2017), the pathways and the amount of GrIS melt water imported into the ocean's interior could have crucial consequences for the ventilation and formation of deep water and hence the strength of the climate-relevant Atlantic meridional overturning circulation (AMOC). In most climate and coupled ocean-sea ice models (Danabasoglu et al., 2016; IPCC, 2013), weak convection leads to a weaker AMOC and vice versa. It is thought that the arrival of additional melt water in the deep-water formation region could strengthen the density stratification and thus weaken or even suppress deep convection. Rahmstorf et al. (2015) and Yang et al. (2016) concluded that the GrIS melt has already slowed down the AMOC in the 20th century. In contrast, the analysis of a high-resolution ocean-sea ice model showed that additional freshwater from GrIS has not exerted any influence on deep water formation yet but might do so in the next decades (Böning et al., 2016).
None of the usually applied methods to estimate ocean forced mass loss of GrIS are able to directly quantify the amount of GrIS melt water and to trace its spreading in the ocean. Although salinity is measured with very high precision, many processes change salinity (i.e., Rhein et al., 2013). Salinity anomalies such as the Great Salinity Anomaly in the 1970s were described as advective features originating north of Iceland (Dickson et al., 1988), while anomalies in the 1990s seem to reflect an increased Arctic freshwater flow through the Canadian Archipelago into the Labrador Sea (Belkin, 2004).
In short, it is difficult to attribute a local salinity decrease uniquely to the presence of GrIS melt water, especially when more than one water mass is present. It is up to now not clear, how much of the melt water remains on the Greenland shelf and in the boundary current and whether and where melt water is released into the Labrador Sea and especially into the key region for deep convection.
Based on studies carried out around Antarctica, stable noble gases Helium and Neon distributions in the ocean provide a useful tool to identify and to quantify the components of glacially modified water (e.g., Hellmer et al., 2016; Hohmann et al., 2002; Huhn et al., 2018, 2008; Kim et al., 2016; Loose et al., 2016; Schlosser, 1986; Schlosser et al., 1990). Atmospheric air is trapped in the ice matrix during formation of the meteoric ice. When the glacial ice on the ocean melts from below, these gases are completely dissolved in the water due to the enhanced hydrostatic pressure at the base of the ice shelf or glacier. Hence, submarine melt water (SMW) is highly enriched in He and Ne.
Owing to the year-long below freezing-point surface temperatures, surface melt does not play a role at Antarctica, and only SMW needs to be identified. This is different around Greenland, where summer temperatures are higher. Here surface melting linked to atmospheric warming, and submarine melting due to warm ocean water reaching the marine terminating glaciers are equally important (e.g., Straneo & Cenedese, 2015). For the ice-tongue glaciers, most of the SMW is released via melting beneath the ice tongue (Wilson et al., 2017). For tidewater glaciers, a minor part is discharged by melting of the vertical tidewater face, but the major part is discharged through melting of icebergs, so that here the SMW injection points are mostly inside the fjords along the trajectories of the icebergs (Enderlin et al., 2016; Moon et al., 2018).
For the three remaining largest ice tongues all located in northern Greenland, the contribution of submarine melting to the glacial runoff is about 80% (Wilson et al., 2017). In 1960–1990, that is, a time period with a stable GrIS, the ice discharge (iceberg calving and submarine melting) from both glaciers with ice tongue and tidewater glaciers covered about 54% of the total freshwater flux (Bamber et al., 2012).
In summer 2015, we took 178 He and Ne samples at two hydrographic sections, reaching from the southern Greenland outer shelf into the convection region of the Labrador Sea (Figure 1a). The more northwestern section follows the WOCE AR7W line, and the southern section runs parallel to 44°W. The sections are downstream of the main drainage systems of eastern Greenland (Bamber et al., 2012). Additionally, we discuss noble gas measurements taken in November 1994 at AR7W and WOCE line A1E from the Irminger Sea toward Ireland (Figure 1a). Owing to bad weather, the 1994 AR7W section has a gap in the interior Labrador Sea.
In a fjord (Beaird et al., 2015), salinities lower than the inflowing water at the entrance can uniquely be attributed to be GrIS melt water inserted into the fjord, and the authors applied an Optimum Multiparameter Analysis (OMP) to quantify the GrIS melt in the fjord. However, SMW is rapidly diluted and far off the source region not anymore associated with a fresh anomaly (e.g., Beaird et al., 2018). The He and Ne signals of surface runoff from GrIS are not significantly different from freshwater of Arctic origin. Consequently, it is not possible in the open ocean to unambiguously distinguish Greenland runoff from for instance advection of freshwater from the Arctic Ocean (e.g., Belkin, 2004; Dickson et al., 1988). The fraction of freshwater attributed to surface runoff would mainly depend on the salinities chosen for the source water masses in the OMP. Therefore, we focus here on the distribution of SMW. We identify SMW by the measured excess of He and Ne relative to the noble gas supersaturations observed in the mixed layer (ML). This has the advantage that the calculated SMW fractions are independent from the choice of the oceanic water masses and from the relative weights attributed to individual parameters of the OMP.
2 Data
The presented He and Ne samples were taken from 10-L Niskin bottles attached to a conductivity-temperature-depth rosette during the cruise MSM43 with RV MERIAN (25 May to 27 June 2015). The uncertainties for the salinity (S) and temperature (T) measurements are ±0.002 for salinity and ±0.002 °C (Mertens et al., 2017). Velocities were measured with a vessel mounted 75- and 38-kHz Acoustic Doppler Current Profiler (Mertens et al., 2017). The isotopes 3He, 4He, 20Ne, and 22Ne are measured. 3He is not a tracer for submarine melt and henceforth not discussed. For this study we omit the suffix and use He instead of 4He and Ne instead of 20Ne. The He and Ne samples were collected in copper tubes sealed with steel clamps and analyzed in the University of Bremen mass spectrometer Helis (Sültenfuß et al., 2009). Based on replicates, the reproducibility of the 2015 data is He ± 0.19% and Ne ± 0.26%. One hundred ninety-four samples have been extracted, and 178 could be analyzed in the mass spectrometer. From these, 11 had been identified as obvious outliers and removed. From the 1994 data, seven samples were already flagged “questionable,” and additional 25 obvious outliers (although flagged “good”) have been removed, most of them due to a clear lack of helium. On most noble gas data sets, the percentage of outliers is typically between 5% and 20% (Well & Roether, 2003), and the two data sets used here are on the lower side of this range.
3 Methods
3.1 Calculation of He and Ne excess anomaly attributed to SMW
Away from sources and sinks, He and Ne concentrations are found to be in excess to the atmospheric equilibrium (e.g., Hamme & Emerson, 2002; Loose & Jenkins, 2014; Loose et al., 2016; Well & Roether, 2003). The main process for He and Ne supersaturation is air injection through air bubbles being introduced by wave breaking. Small air bubbles dissolve completely. Gas transfer or exchange with large air bubbles occurs by gas pressure that is enhanced hydrostatically or by surface tension (Fuchs et al., 1987; Well & Roether, 2003). Wave intensity and wave breaking are dependent on wind speed, so that air injection could vary regionally and seasonally. Regional Ne and He differences in saturation of up to few percent can be induced where atmospheric pressure persistently differs from its standard value. Other processes are temperature (and salinity) changes as well as ML deepening.
For the 2015 data, the samples from the upper 20 m are used. For November 1994, we used data from the upper 25 m (owing to the typically bad weather in fall, the uppermost samples at that cruise were taken between 20 and 25 m). We assume that these samples represent the noble gas concentrations in the ML and calculate for each cruise the mean supersaturation in the ML, HesupersatmeanML, from equation 1.
In June 2015, the mean supersaturation in the ML was 5.0 ± 0.52% for He and 4.1 ± 0.65% for Ne. He was supersaturated by 2.2 ± 0.5% in November 1994. The uncertainties reported here are always the standard deviations of the data sets and reflect the variability of the ML supersaturations. No dependence of the ML saturations with temperature (Figure 1b) or salinity or location was found. ML supersaturations near the glacier tongue could be higher than outside the fjords, mainly caused by upwelling of SMW containing water. This part of the SMW cannot be observed remote from the fjords, since air-sea gas exchange removes the signal quickly from the surface layer and only subsurface signals remain.
The differences of the mean ML supersaturations between November 1994 and June 2015 are of the same order of magnitude as the modeled seasonal signal (Hamme & Severinghaus, 2007). In November, their modeled ML supersaturation was lower by about 4% compared to December/January and lower by about 2% compared to April. Our results show, however, that Labrador Sea Water (LSW)—ventilated in spring—has supersaturations similar to the ones found in the ML in fall, pointing toward none or small differences in ML supersaturations between the two seasons. We cannot exclude that a bias in the absolute calibration between the data sets exists, but this would not affect the results presented here. The ML supersaturations are compatible to other regions, for instance the South Atlantic (2–3%; Well & Roether, 2003) and at Bermuda station Bermuda Atlantic Time-series Study in the suptropical Atlantic (about 2%; Hamme & Emerson, 2002).
The water in the ocean interior could be mixed with several water masses that had different T and S at the surface. However, for both gases, the nonlinearity of the temperature-solubility function is negligible (e.g., Roether et al., 1998), so that this is of no concern here.
The largest uncertainty in calculating the excess anomalies is the variability of the mean ML supersaturations. We choose this uncertainty to be about twice as large as the observed standard deviation of the data set, that is, ±1.0%. Together with the measurement errors, this results in an overall uncertainty of the calculated SMW fraction of 0.075%. Only SMW fractions higher than this threshold are considered as significant.
The excess anomalies (and thus the calculated SMW fractions) in 2015 from He and Ne measurements are similar within the uncertainties (Figure 1c), meaning that addition of crustal He from the bedrock (e.g., Beaird et al., 2015) that only affect He but not Ne does not play a role here. We adopt this finding from 2015 to the data from 1994 and interpret the observed He excess to be caused either by surface processes or by the presence of SMW. If both He and Ne are available, the calculated SMW fractions are a mean of both.
4 Results
To identify the water masses in the study region, a salinity section from June 2015 is presented (Figure 1d). The fresh water at and near the surface reaches from the Greenland slope/shelf into the interior of the Labrador Sea. This is Polar Water (PW), originating in the Arctic Ocean, with contributions from Greenland melt. The saltier (and warm) Atlantic Water (AW), a remnant of the subtropical water imported by the North Atlantic Current, spreads in the East Greenland Current (EGC) mainly along Greenland. AW delivers heat into the fjords and in contact with the outlet glaciers initiates submarine melting (e.g., Schaffer et al., 2017; Straneo & Cenedese, 2015). The definition of the deep water masses upper and deep LSW (dLSW), Iceland-Scotland-Overflow-Water (ISOW), and Denmark-Strait-Overflow-Water (DSOW) follow, for example, Rhein et al. (2017). In the eastern part of the A1E section (Figure 2e), water in the density range of DSOW is modified Antarctic Bottom Water (AABW; McCartney, 1992).
In general, SMW fractions are below the threshold of 0.075%. SMW signals are mainly confined in the upper ocean close to the Greenland and Canada slope/shelf. Occasionally, significant SMW fractions are also found in the deep ocean, that is, at both sections in 2015 in dLSW/ISOW around σθ = 27.80 (Figures 2c and 2d) and at AR7W in 1994 at the northern part at σθ = 27.78 (Figure 2b). SMW signals seem also present in DSOW close to the bottom and in the bottom water in the Northeast Atlantic at densities σθ > 27.88.
In the upper ocean (Figures 2f–2i), SMW is found below the ML and mostly clustered above 400-m depth. The largest horizontal and vertical extents of SMW were found at both sides of the AR7W section in 1994 (Figures 2f and 2g). The water-containing SMW is a mixture of AW (warm and salty) and PW (cold and fresh) with varying contributions of both. This is to be expected since the warm and salty AW is needed to provide the energy for submarine melting, but in the fjord and in the boundary current, it is subject to mixing with the colder and fresher ambient water there. Surprisingly, the maximum SMW signal at the Greenland slope (0.62 ± 0.075% in 2015) is comparable to the maximum SMW fractions observed in a fjord at Disco Bay (0.66 ± 0.09%; Beaird et al., 2015), although the SMW in the boundary current had more opportunities to dilute. In 1994, the A1E section in the Irminger Sea did not reach far onto the Greenland Shelf, so SMW was present only at the station closest to Greenland (Figure 2e).
5 Discussion
5.1 Origin of the SMW in the Upper Ocean
Upstream of the sections discussed here, several fjords with outlet glaciers (Figure 1a) are able to produce SMW (Straneo & Cenedese, 2015), that subsequently could be exported into the southward flowing EGC. Recent studies at the Nioghalvfjerdsfjorden Glacier at 79°N in Northeast Greenland showed that warm water enters the cavity below the ice tongue (Schaffer et al., 2017). A SMW source closer to the study region is the outflow of the Kangerdlugssuaq Fjord north of Denmark Strait. Heat delivered to the calving front of this glacier is equivalent to a melting of 10 m/day (Inall et al., 2014). This also occurs further downstream at the Helheim Glacier (Beaird et al., 2018; Straneo et al., 2011). The authors found that glacial melt is exported out of the fjord at the surface and at the interface between the cold, fresh PW and the warm, saline AW. The exported water (at about 150–250-m depth) has densities between σθ = 27.0 and 27.5. The highest SMW portions are found mostly at densities between σθ = 27.0 and 27.6 (Figure 3), that is, close to the densities of the SMW signal of the Helheim glacier. SMW could also enter the Greenland boundary current downstream of the AR7W section (e.g., Beaird et al., 2015, 2017; Straneo & Cenedese, 2015), so that it is not clear which source fed the SMW signal found in the boundary current at the Canadian side of the Labrador Sea.
In the last 20 years, mass loss of GrIS has increased (Bamber et al., 2012) and most likely increased the amount of SMW in the ocean. However, the noble gas data coverage is not sufficient to address temporal changes, although fractions >0.45% were only observed in 2015. All data miss the main part of the shelf, where SMW might be expected.
A rough estimate of the SMW transport in the Greenland boundary current is calculated for the onshore part of the Labrador Sea sections shown in Figures 2f–2h. The SMW fractions of the single data points are area-weighted and multiplied with the mean shipboard Acoustic Doppler Current Profiler velocity of the two sections (0.2 and 0.23 m/s, respectively; C. Mertens et al., personal communication, August 2017). The SMW transport is then about 0.001 Sv at AR7W and 0.003 Sv at 44°W in 2015 and 0.002 Sv at AR7W in 1994 (assuming the same velocity at AR7W-1994 as in 2015). The mean SMW transport is equivalent to 12 ± 6% of the total freshwater fluxes for the Greenland region that drain upstream of the A1E section into the Nordic Seas and the Irminger Sea (time period 1992–2010; Bamber et al., 2012). By assuming that half of the total freshwater flow from Greenland into the Nordic Seas and the Irminger Sea consists of SMW (Van den Broeke et al., 2009, 2016), the observed SMW transport in the Greenland boundary current covers about 24% of that flow.
5.2 Influence of SMW on LSW Formation
Although only confirmed on the four sites of the noble gas measurements analyzed here, SMW seems to be present in the upper 400 m in the Greenland and Canadian boundary current. One important implication of this finding is that SMW does not spread into the basin's interior at the surface—as most models analyzing a connection between GrIS melt and AMOC assume. Instead, SMW is distributed over several hundred meters below the ML and arrives in the LSW formation region as an already strongly diluted water mass. Even if SMW would be included in the LSW formation, the signal would be further diluted after being vertically mixed down to 1,000–1,500-m depth and could then no longer be detected with our method. One also has to bear in mind that the significant SMW portions are related to salinities between S = 33.2 and S = 34.9 (Figure 3b), and the highest SMW fractions are mostly found at the higher salinities. The admixture of 0.1% SMW would decrease the salinity by about 0.035, and this effect on the salinity is an order of magnitude smaller than the effect of variable percentages of PW and AW. Figure 3c shows, that, especially in 2015, fresh LSW is not necessarily related with a high SMW fraction.
The result of Yang et al. (2016) that GrIS melt has affected LSW formation relied on the fact that the LSW layer thickness (between 27.74 and 27.80) decreased till 2000, while the GrIS freshwater flux increased and overwhelmed the salinity flux of the AW. The correlation between GrIS melt and LSW layer thickness however breaks down in 2013/2014 and following years, when ventilation of these densities resumed (e.g., Rhein et al., 2017) with the layer thickness increasing since then by about 600 m.
5.3 Import of SMW Into dLSW and ISOW
In the Irminger Sea, the so-called EGC spill jets introduce water from the Greenland shelf into the deeper water masses (e.g., Jochumsen et al., 2015; Pickart et al., 2005). For instance, 180-km downstream of Denmark Strait, the EGC contribution to DSOW was between 20% and 40% (Jochumsen et al., 2015). It is however unknown if and how frequently and how much of SMW is part of these spill jets. It is plausible that spill jets not always reach DSOW but occasionally contain water with densities closer to LSW (e.g., Pickart et al., 2005) or ISOW and so leading to the observed signals in 2015 and 1994. As a clear evidence for sporadic SMW injection in both years, the significant SMW fractions in dLSW and ISOW are linearly correlated with salinity (Figure 3c). A similar linear relationship is found in water shallower than upper LSW (σθ < 27.68) and denser than σθ = 27.5 (Figure 3b). For this density range, the SMW fractions are located along the mixing line between AW (S = 35, SMW = 0) and pure SMW (S = 0). For less dense water, PW (S = 32.5, SMW = 0) is also part of the mixture. If SMW-containing water with densities > σθ = 27.5 is part of the spill jet, SMW signals with a salinity relation as observed in dLSW and ISOW in 1994 and 2015 could be created (Figures 3b and 3c).
5.4 SMW in DSOW
Due to the sparseness of SMW in DSOW and the higher salinities in 2015 compared to 1994, the correlation between SMW and S is not significant (Figure 3d). The SMW-bearing samples could have been caused by SMW in the spill jet, but we cannot exclude that some samples in this water mass have been contaminated by air.
5.5 SMW in the Deep Northeast Atlantic
Of all deep and bottom water, the AABW exhibits globally the highest He and Ne excess anomalies (Loose et al., 2016; Schlosser, 1986) due to entrainment of SMW from the Antarctic ice shelves. Noble gas excess anomalies in AABW are also found away from the Southern Ocean (e.g., Well & Roether, 2003). The coherent SMW signal in the deep eastern Atlantic below the isopycnal σθ = 27.88 (Figure 2d) has been imported most likely from the Southern Ocean with the AABW (e.g., McCartney, 1992).
6 Conclusions
The analysis of He and Ne distributions in the ocean turned out to be a powerful tool to quantify and trace SMW. The Greenland boundary current transports about 24% of the total Greenland ice discharge into the Nordic and Irminger Seas. However, the sections most likely missed some SMW onshore of our measurements, and some SMW could have left the EGC further upstream. SMW fractions are the highest in the upper 400 m below the ML and are mostly confined to the Greenland and Canadian slope and shelf. The strong dilution of SMW in the upper 400 m and the relatively small SMW percentages compared to the abundance of PW indicate that SMW has not played an active role in the LSW formation yet. The mechanism leading to the episodic occurrence of He excess anomalies in the deep water masses is uncertain. Based on the salinity—SMW relation, it could herald episodic EGC spill jet events that contain SMW-bearing water with densities > σθ = 27.5. The coherent SMW signal found in the deep northeastern Atlantic does not originate from Greenland but is owed to the presence of AABW, containing SMW from the Antarctic ice sheets.
Acknowledgments
We thank Christian Mertens (IUP Bremen) for providing the velocities of the Greenland boundary current. M. Rhein received funding from the DFG in the framework of the Priority program SPP1889 (Rh25/43) and from the BMBF in the RACE program (03F0729A). The data from November 1994 have been taken from the CCHDO repository (https://cchdo.ucsd.edu). The 2015 noble gas data are archived in PANGAEA.