Changes in the Surface Salinity Gradient and Transport of the Irminger Current: The Climate Perspective

. Here we use a new analysis schema, the Freshening Length, to study the transport in the Irminger Current on the east and west sides of Greenland. The Freshening Length schema relates the transports on either side of Greenland to the corresponding surface salinity gradients by analyzing climatological data from a data assimilating global ocean model. Surprisingly, the warm and salty waters of the Current are clearly identiﬁed by a salinity maximum that varies nearly linearly with distance 5 along the Current’s axis. Our analysis of the climatological salinity data based on the Freshening Length schema shows that only about 20% of the transport east of Greenland navigates the southern tip of Greenland to enter the Labrador Sea in the west. The other 80% disperses into the ambient ocean. This independent quantitative estimate based on a 37-year long record complements seasonal to annual ﬁeld campaigns that studied the connection between the seas east and west of Greenland more synoptically. A temperature-salinity analysis shows that the Irminger Current east of Greenland is characterized by a 10 compensating isopycnal exchange of temperature and salinity, while west of Greenland the horizontal convergence of less dense surface water is accompanied by downwelling/subduction.

2 https://doi.org/10.5194/os-2021-100 Preprint. Discussion started: 9 November 2021 c Author(s) 2021. CC BY 4.0 License. charges as described by Bacon et al. (2002) and Sutherland and Pickart (2008). These southward currents all converge as they approach Cape Farewell. Also of interest are the southward currents off East Greenland, which appear to continue northward off West Greenland in Figure 1b only. Most of these waters turn cyclonically near 65 N latitude, however, a fraction extends northward into Baffin Bay as a prominent Atlantic temperature maximum that can be traced beyond 78 N latitude (Münchow, 2016;Rignot et al., 2021). The western limb of the Irminger Current contains an unknown fraction of the eastern limb (Myers et al., 2009). 30 It appears then that the transport of salty and warm Atlantic water of subtropical origin plays an important role in the AMOC. As is evident from figure 1 the Irminger Current is an important component of this climate system as it is a conduit for relatively warm and salty waters to mix with colder and fresher waters. Panel a of figure 1 suggests that this current blends into the Labrador Sea and does not flow past Cape Farewell. In contrast, panel b of this figure shows that some fraction of the Irminger Current continues past Cape Farewell as the West Greenland Current, (Pacini et al., 2020) The latter view is supported 35 by Holliday et al. (2007) who estimated that about 5.1 Sv or 80 % of the Irminger Current retroflects back into the Irminger Sea.
The large number of observational studies conducted in the Irminger and Labrador Seas off southern Greenland inform the kinematics of subpolar boundary currents, deep convection, and climate impacts, (Pickart et al., 2003;Holliday et al., 2007;Myers et al., 2007Myers et al., , 2009Le Bras et al., 2020). These studies describe in considerable detail the seasonal and interannual 40 variability of the hydrography and currents from moorings and hydrographic expeditions along cross-sections of the currents, but except for Holliday et al. (2007) generally provide limited information of synoptic transport variations of the Irminger Current east and west of Cape Farewell.
In order to provide additional insight on synoptic variations of the Irminger Current we use nearly 37 years of observations obtained from a data assimilating global model (Carton et al., 2018). Consequently our salinity data set is dynamically con-45 sistent with the velocity fields. From this we are able to quantify the fraction of the Irminger Current to the east of Greenland that successfully negotiates the sharp turn at Cape Farewell, and which contributes to the West Greenland and Baffin Island Currents (Münchow et al., 2015); and eventually the Labrador Current. As was concluded by Pickart et al. (2003) some of the deep Labrador Sea Water must originate in the Irminger Sea. The West Greenland and Irminger Currents are the main conduits for the transport of this water. While much of the Irminger Sea water detaches from the boundaries as it flows into the Labrador 50 Sea near Cape Farewell, we emphasize that an unknown fraction of the Irminger Current continues to flow northward along West Greenland (Münchow et al., 2015).
To quantify the fraction of the Irminger Current that flows into the Labrador Sea and remains at its surface, i.e. does not subduct, we appeal to the new paradigm of Evaporation Length developed by Berman et al. (2019). That approach utilizes the relative changes in the salinity, due to the net evaporation that a column of water undergoes along a trajectory, to calculate a new 55 parameter, the Evaporation Length. The method was applied to salinity and geochemical data from the Red and Mediterranean Seas, the Indian Ocean, and the Gulf Stream (Berman et al., 2019). The Evaporation Length estimates the (hypothetical) distance that a moving column of surface water has to travel in order to evaporate all its water. As was shown in (Berman et al., 2019) the Evaporation Length reliably quantifies the volume transport per unit length of the cross-stream direction of the current.
We extend this paradigm to the case where the salinity of the water in the column decreases along the trajectory, and in our application the length parameter is termed Freshening Length. In the scenario considered here the Freshening Length estimates the (hypothetical) distance that the column has to travel in order for its water to become fresh (i.e. zero salinity). Save for a trivial change of sign the subsequent quantification of the volume transport (per unit length of the cross-stream direction) of the current in the freshening scenario is identical to that in the evaporation scenario. The paradigm is particularly suitable for 65 use in connection with changes in routinely observed variables (salinity in our case) stored in all climatological data archives since it is measured in all observations and reported in all model calculations. We also use the surface temperature data of this reanalysis data archive to analyze changes in the properties of surface Temperature-Salinity (T-S) diagrams along the current to examine the exchange that takes place in the currents east and west of Greenland.
Our approach is based on long-term observations. This differs from previous studies in that the analysis follows the mean 70 flow and is not limited to brief snapshots in time and not constrained to a specific geographical region. Furthermore, the new metric is insensitive to interannual variations associated with large decadal freshening of the subpolar North Atlantic as described most recently by Holliday et al. (2020). Thus, this climatological data base is appropriate for analysis using the Freshening Length, as it encapsulates long-term trends in the along-flow characteristics of salinity and temperature in this region.

75
The next section describes the data used in the analysis. This is followed by the results of our analysis. We conclude with a synopsis of our findings and a discussion of new issues raised by this research.

Data and Methods
We use data from the "Simple Ocean Data Assimilation" or SODA project. The technical details of these data are given in Carton et al. (2018) and the data are freely accessed at https://rda.ucar.edu/datasets/ds650.0/. The spatial resolution of the 80 gridded data assimilation product is 0.5 degrees in latitude and longitude while the temporal resolution is 5 days. Time series of salinity and temperature data span the period from January 3, 1980 to December 19, 2017, nearly 37 years. The surface salinity and temperature values presented here are averages over the entire 37-year record taken at a depth of about 5 m. The geographical region used here is 45 • W and 35 • W longitude between 55 • N and 65 • N latitude. It covers the ocean adjacent to the southern part of Greenland.

85
In the first step of the analysis we determine, at each latitude, the longitudes of the maximum salinity. Then we identify the maximum salinity along the axis of the Irminger Current by calculating a 5-point zonal average (90 km  distance between two adjacent salinity maxima. Figure 2 shows the location of the zonal salinity maxima to the east (in blue symbols) and west (in red symbols) between Cape Farewell and 63.75 • N. (where S 0 is the salinity at the origin of the trajectory and ∆S ∆X is the salinity gradient along the trajectory) yields the value of the Evaporation Length denoted by L. The Evaporation Length schema focuses on a water column of depth H that moves at the ocean surface with speed U and whose salinity change is determined by q, the rate at which fresh water is removed (say, by net evaporation) while its salt content remains unchanged. The conservation of mass of salt and water in the moving column relates L and q to the column's depth, H, and speed, U , via the linear relation: where F (=HU ) is the volume transport per unit length in the cross-stream direction.
In the present study we extend the original Evaporation Length paradigm to the freshening scenario where the salinity decreases along the trajectory (i.e. ∆S < 0) since fresh water is added to the column e.g. by entrainment. In both cases the salinity change of the water in a column that extends from the surface to a depth z = −H yields a length that estimates either 120 the (hypothetical) distance that the column has to travel for all its water to evaporate (in the evaporation paradigm) or for its water to loose all its salinity and become fresh water (in the freshening paradigm).
The Freshening Length paradigm is applied here to the Irminger Current that bifurcates from the salty North Atlantic Current.  paradigm is also applied to the West Greenland Current, which is also marked by a local salinity maximum and that flows northwestward east of Greenland.

135
We find a 5-fold decrease in the value of L from the eastern to the western limbs of the Irminger Current. According to Equation (1) this change in L results from a combination of changes in the current transport, F , and the rate of fresh water entrainment, q which is discussed in the next section.

Summary and Conclusions
In view of the tremendous impact that AMOC has on Earth's climate, it important to clarify which of the flows proposed in 140 the two panels of 1 is correct. To answer this question we applied a new analysis technique, the Freshening Length paradigm (Berman et al., 2019), to 37 years of assimilated modeled hydrographic data. As noted above, this paradigm compliments traditional observational studies that utilize data from moorings and surveys of the Irminger and the West Greenland Currents.
Consider first the T-S plot in figure 3. Note that the southernmost stations on the east and west Greenland have nearly identical T-S values, even though they are over 75 km apart. This implies that the West Greenland Current is a continuation of 145 the Irminger Current. Figure 3 also indicates different freshening mechanisms of the Irminger Current east of Greenland and the West Greenland Current. The Irminger Current exhibits typical isopycnal mixing from the northern start to the southern end, a distance of about 700 kilometers. We interpret this isopycnal mixing as horizontal entrainment along a density surface combined with release of heat to the atmosphere. This conclusion is consistent with both (Pickart et al., 2003) andSmith (1937). While the former employed modern moorings and cross-sectional hydrography, early American (Smith et al., 1937) 150 and Danish (Riis-Carstensen, 1936) studies provided similar descriptions as part of the International Ice Patrol. As opposed to these quasi-synoptic survey and seasonal mooring data, we here used climatological data that average properties over 37 years.
In contrast to the Irminger Current, the Western Greenland Current is characterized by much more intense cooling and freshening. The intense cooling and freshening west of Greenland cannot result from the localized and slow release of heat to the atmosphere accompanied by slow entrainment of surrounding water as on the eastern Irminger Current. Instead, it probably 155 results from horizontal convergence of cold and fresh water accompanied by downwelling (subduction of the current). This, too, is consistent with Pickart et al. (2003) who concluded that the source of the North Atlantic Deep Water originates in the Irminger Sea rather than the Labrador Sea. A comparison between the T-S properties in the two limbs of the Irminger Current shows that the slight freshening along the eastern limb is density-compensated by cooling. In contrast, the more substantial freshening by polar waters along the western limb is not density-compensated by cooling, which result in a decrease of the water The difference between the values of L can be explained by a combination of two extreme cases.
In the first extreme case we assume that the fresh water entrainment rates, q, into the Irminger and West Greenland Currents 170 are equal. According to Eq. (1) the fact that q is constant implies that the equatorward transport, F , of the Irminger Current is 5-times larger than the poleward transport of the West Greenland Current. In this uniform q extreme case the 80% decrease in L from 1 × 10 5 km in the east to 0.2 × 10 5 km in the west implies that 80% of the transport of the Irminger Current detaches from it and does not negotiate the sharp turn at Cape Farewell. Some of the 80% may cool, sink (i.e. subduct) or contribute to the NADW as suggested by the purple cyclone labeled "LS" in Figure 1a. Alternatively, some of these 80% could retroflect 175 and form the southern branch of the Irminger Gyre as depicted in 1b and labeled "IC". Holliday et al. (2007) quantifies such a retroflection from a single snapshot of velocity observations, however, they find that 80% navigates the bend of Cape Farewell from east to west while 20% moves offshore and to the east. This suggests that a large fraction of the 80% is shed as eddies in the area (Bracco et al., 2008).
In the second extreme case, the volume transport, F , is assumed constant, which implies that the fresh water entrainment 180 rate, q, is 5 times smaller in the West Greenland Current compared to the Irminger Current. This extreme case requires that the speed of the current decreases in the West Greenland Current, to allow for the higher (temporal) rate of entrainment, while the depth of the current increases to maintain the same volume transport.
In addition to the 37-year mean that yields the Freshening Lengths discussed above, the SODA3 data contain annual and seasonal variations of surface salinity. We repeated the above calculations of L based on annual data and found that the root- Though the surface salinity data used here is dynamically consistent with SODA3's velocity data, the Freshening Length 195 estimate of the transport is much more robust and informative than direct estimates based on velocity profiles. The reason is that even though cross sections of along-stream velocity in the two Currents yield similar maximal velocities of about 0.2 m·s −1 , the vertical and horizontal extents of the currents are not accurately determined. Consequently, velocity profiles cannot yield a reliable estimate of the transports in the Currents.
Our result that only 20% of Irminger Current waters navigate around Cape Farewell could be tested by placing cross-200 sectional current mooring and hydrographic surveys east and west of Greenland. We hypothesize that the 5-fold difference in the value of L in the Irminger and West Greenland Currents is due to changes in volume transport and entrainment rate. This could also be combined with an effort to estimate changes of L directly from concurrent drifter deployments that would also reveal along-track salinity and temperature variations.