Confidential manuscript submitted to JGR-Oceans
1 Transformation of the North Iceland Irminger Current waters
2 in the Nordic Seas and its link to the Denmark Strait Overflow
3 Water
1 1 1,2 1,3 4,5 4 Yarisbel Garcia Quintana , Colin More , Pia Wiesner , Xianmin Hu , Dagmar Kieke ,
1 1 5 Bruce R. Sutherland and Paul G. Myers
1 6 Department of Earth and Atmospheric Sciences, University of Alberta
2 7 Department of Geosciences, Kiel University
3 8 Bedford Institute of Oceanography, Dartmouth, Nova Scotia, Canada
4 9 Institute of Environmental Physics, University of Bremen
5 10 MARUM - Center of Marine Environmental Sciences, University of Bremen
11 Key Points:
• 12 Strong light-to-dense Atlantic Water transformation driven by heat loss occurs in
13 the boundary currents of the Nordic Seas, with densities reaching that of Denmark
14 Strait Overflow Water
• 15 Water entering the Nordic Seas within the North Iceland Irminger Current (NIIC)
16 is transformed into overflow waters in the shelf system of the Nordic Seas within 6
17 years
• 18 Along the north Icelandic shelfbreak the transformation is faster with export of the
19 transformed waters occurring in the North Icelandic Jet within 1 year
Corresponding author: Yarisbel Garcia Quintana, [email protected]
–1– Confidential manuscript submitted to JGR-Oceans
20 Abstract
21 Classically, the Nordic Seas are often considered the headwaters for the Meridional Over-
22 turning Circulation (MOC), for it is where the densest component of the Deep Western
23 Boundary Current is formed. In spite of at least two decades of observations measuring
24 the transport of the overflows across the Greenland-Scotland Ridge into the North Atlantic
25 Ocean, questions exist about the reservoir that drives them. We address this subject us-
26 ing two eddy permitting configurations of an ocean general circulation model and the La-
27 grangian tracking tool Ariane to explore the Atlantic Water transformation in the Nordic
28 Seas and its influence on driving the North Icelandic Jet. Transformation to greater den-
29 sities is found to occur in the boundary currents of the Nordic Seas. These waters leaving
30 at depth through Denmark Strait are found to do so within six years. A faster transfor-
31 mation occurs in a loop along the north Iceland shelfbreak with export occurring in the
32 North Iceland Jet within one year. Despite the transformation to denser water occurring in
33 the boundary currents, the maximum densities reached by the particles are consistent with
34 the maximum densities observed in the Denmark Strait Overflow Water. Thus it is possi-
35 ble that even the most dense parts of the Denmark Strait Overflow Water could come from
36 boundary current transformations, rather than deep convection in the interior of the Nordic
37 Seas.
38 1 Introduction
39 The Nordic Seas are a grouping of three small basins: the Iceland, Norwegian and
40 Greenland Seas. These three seas are separated from the Atlantic Ocean by the Denmark
41 Strait, Iceland-Faroe Ridge, and the Faroe Faroe-Shetland Channel, and from the Arctic
42 Ocean by the Fram Strait and Barents Sea Opening (Figure 1). As well as having im-
43 portant local climatic and ecological consequences [Drinkwater et al., 2013], the region
44 plays an important role in the global oceanographic system (e.g, Yashayaev and Seidov
45 [2015]). The Nordic Seas allow for the exchange of warm salty waters from the Atlantic
46 to the Arctic Ocean. They receive cold low-salinity waters and sea ice from the Arctic
47 [Eldevik et al., 2009], which are later exported to the southern oceans through the different
48 straits located between the Greenland-Scotland Ridge (Figure 1).
49 Additionally, the Nordic Seas have historically been known as a deep water forma-
50 tion site [Aagaard et al., 1985]. This production forms a dense water reservoir within
51 the basin, which has been argued to overflow the sills at Denmark Strait and the Faroe-
–2– Confidential manuscript submitted to JGR-Oceans
52 Shetland Channel. Subsequent sinking and entrainment of surrounding waters leads to the
53 production of North Atlantic Deep Water, which flows south in the Deep Western Bound-
54 ary Current (DWBC) to become a key component in the lower limb of the Meridional
55 Overturning Circulation [Mauritzen, 1996]. The Denmark Strait Overflow Water (DSOW)
56 is the largest and densest of these contributions [Dickson and Brown, 1994; Send et al.,
57 2011; Mertens et al., 2014; Fischer et al., 2015]. Measurements of DSOW transport have
58 been carried out continuously since 1996 [Macrander et al., 2005; Jochumsen et al., 2012,
59 2017], but a detailed description of where and how the DSOW is formed, is still missing.
60 Many numerical modelling studies have linked changes in freshwater, such as from
61 the Greenland Ice Sheet, to changes in deep water formation and the MOC (e.g Gerdes
62 et al. [2006]; Stouffer et al. [2006]; Swingedouw et al. [2007]; Hu et al. [2011]; Weijer et al.
63 [2012]). These studies can be considered in the context of a reduction of deep convection
64 in the Nordic Seas since 1982 [Rhein, 1991; Schlosser et al., 1991; Somavilla et al., 2013],
65 leading to an increase of salinity and temperature in the deep Greenland Sea due to the
66 advection of deep Arctic Ocean waters [Somavilla et al., 2013]. Given that the deep over-
67 flows are driven by pressure gradients set up by transformation of Atlantic Water (AW)
68 to increased densities in the Nordic Seas [Hansen and Østerhus, 2000], these changes led
69 Hansen et al. [2001] to suggest that, based on an upward-looking Acoustic Doppler Cur-
70 rent Profiler (ADCP) moored at the Faroe Bank Channel sill (from November 1995 to
71 June 2000), the overflow through the Faroe Bank Channel was decreasing, with potential
72 impacts on the MOC. However, with a longer dataset (1995 to 2005) from three ADCPs
73 moored also at the Faroe Bank Channel sill, Hansen and Østerhus [2007] suggested that
74 the overflow transport was steady. Furthermore Jochumsen et al. [2012] suggested there
75 was no significant trend in the overflow transport at Denmark Strait, although the tempera-
76 ture did decrease. A number of studies [e.g. Mauritzen [1996]; Eldevik et al. [2009]; Våge
77 et al. [2011]] have suggested that the transformation of inflowing AW in the Nordic Seas
78 is a gradual process that occurs within the boundary currents. In fact, Eldevik et al. [2009]
79 clearly state that the overflows are not linked to convective mixing.
80 Based on data from 11 oceanographic cruises covering a period of 10 years, Mau-
81 ritzen [1996] argued against this widely-held paradigm of deep convection feeding the
82 dense overflows from the Nordic Seas. Instead, she proposed that inflowing AW is grad-
83 ually densified via heat loss as it circulates in the Norwegian Atlantic Current. The dense
84 water is then transported to the outflows by the boundary currents surrounding the Ice-
–3– Confidential manuscript submitted to JGR-Oceans
85 land and Greenland Seas at both shallow and intermediate depths. Eldevik et al. [2009]
86 supported this idea, arguing that the AW circulation and transformation within the bound-
87 ary currents of the Nordic Seas was the main source for the overflows. Their circulation
88 scheme suggested that anomalies passing through Denmark Strait had travelled along the
89 rim of the Nordic Seas, while the Faroe-Shetland Channel overflows reflected variability
90 within an overturning loop within the Norwegian Sea as well as a shorter pathway through
91 the Jan Mayen Channel. Using a theoretical model, Yang and Pratt [2013] showed that the
92 dense water reservoir in the Nordic Seas available to drive the overflows is limited to the
93 boundary currents, with the dense water in the interior basins kept isolated within closed
94 geostrophic contours.
–4– Confidential manuscript submitted to JGR-Oceans
FS Barents Sea
K. RidgeWSC BSO
Greenland GS
JMFZ M. Ridge EGC LB IS DS NIJ NS Iceland
NIIC IFR NwAC DWBC
FSC Norway
NAC
Figure 1: Schematic of the Nordic Seas, showing the main basins, inflows, outflows and geograph- ical features, following Orvik and Niiler [2002] and Våge et al. [2013]. For the currents, red colors represents light surface waters which are transformed into cold, dense overflow waters represented in blue colors. Main current acronyms: NAC - North Atlantic Current; NIIC - North Iceland Irminger Current; NwAC - Norwegian Atlantic Current; WSC - West Spitsbergen Current; EGC - East Greenland Current; DWBC - Deep Western Boundary Current; NIJ - North Icelandic Jet.
Black dashed lines represent the main straits that separate the Nordic Seas form the Atlantic and Arctic Oceans: Denmark Strait (DS), Iceland Faroe Ridge (IFR), Faroe Shetland Channel (FSC), Barents Sea Opening (BSO) and Fram Strait (FS). Main basins: Norwegian Sea (NS), Iceland Sea (IS), Lofoten Basin (LB) and the Greenland Sea (GS). For each of the basins its circulation is illus- trated by black dashed circles with the arrow heads pointing the direction of the circulation. In red appear the Knipovich Ridge (K. Ridge), Mohns Ridge and the Jan Mayen Fracture Zone (JMFZ).
–5– Confidential manuscript submitted to JGR-Oceans
95 Våge et al. [2011] pointed out that although traditionally the East Greenland Current
96 (EGC) was considered the primary pathway for supplying the Denmark Strait overflow,
97 there is now evidence for a significant role of the North Icelandic Jet (NIJ), which sup-
98 plies around a third of the overflow water transport at the sill [Våge et al., 2011; Harden
99 et al., 2016]. Våge et al. [2011] used a simplified numerical model to suggest a dynami-
100 cal link between the inflowing North Iceland Irminger Current (NIIC) and the outflowing
101 NIJ, with the majority of transformation and sinking of the inflowing AW occurring in the
102 boundary current north of Iceland. According to their model, the NIIC sheds eddies into
103 the Iceland Sea interior where, due to heat loss during winter, the water carried within the
104 eddies is made dense, returning later to the boundary, where it sinks and forms the NIJ.
105 Våge et al. [2013] extended the understanding of the region, showing that the cir-
106 culation upstream of Denmark Strait is complex, with the source waters for the overflow
107 mainly approaching the strait on the Icelandic side. They also confirmed that the NIJ is
108 a distinct current from the East Greenland Current. The former one bifurcates 450 m up-
109 stream of Denmark Strait at the northern end of the Blosseville Basin, and the diverted
110 branch is known as the separated EGC which flows south parallel to the NIJ [Våge et al.,
111 2013].
112 Recently Behrens et al. [2017] supported the existence of the hypothesized over-
113 turning loop [Våge et al., 2011] along the shelfbreak north of Iceland. By using the La-
114 grangian tracking tool Ariane in a high-resolution model hindcast, they investigated the
115 upstream sources of the DSOW and argued that the water carried by the NIIC transforms
116 to feed the NIJ. What is more, while Behrens et al. [2017] claim the overturning cell to
117 be located on the shelfbreak north of Iceland, Våge et al. [2011] and Våge et al. [2013]
118 hypothesized the cell to occur in the Iceland Sea interior.
119 Our study aims to shed further light towards the understanding of the NIIC trans-
120 formation and its role on driving the NIJ. For that purpose we use the fields from two
121 different eddy permitting ocean general circulation model experiments to drive the integra-
122 tion of Lagrangian virtual floats. The transformation pathways for the AW as it enters the
123 Nordic Seas via Denmark Strait within the NIIC and its potential contribution to DSOW
124 are examined. Emphasis is made on the transformation occurring north of Iceland as a po-
125 tential driver of the NIJ. The manuscript is structured as follows. The model description,
126 experiment setup and details of the Lagrangian tool Ariane are topics in Section 2. Then
–6– Confidential manuscript submitted to JGR-Oceans
127 the possible pathways of the NIIC as it enters the Nordic Seas and its transformation into
128 overflow-like waters is examined in Section 3. Section 4 focuses on the transformation
129 that occurs north of Iceland and its contribution to the NIJ. Discussion and conclusions
130 are presented in Section 5.
131 2 Analysis Methods
132 The model used is the Nucleus for European Modelling of the Ocean (NEMO) nu-
133 merical framework [Madec, G. and the NEMO team, 2008]. NEMO is composed of five
134 major components: ocean dynamics, sea ice, biogeochemistry, adaptative mesh refinement
135 software, and an assimilation component. We use two configurations in order to analyse
136 the transformation pathways of the AW once it enters the Nordic Seas via Denmark Strait:
137 1. ORCA025: The global hindcast experiment (ORCA025-KAB001, hereafter referred
138 to as KAB001, carried out in Kiel) ran from 1958 to 2004 with a temporal resolu-
139 tion of 5 days, and is based on the eddy permitting ORCA025, a global ocean/sea-
140 ice configuration of NEMO 3.1 [Bernard et al., 2006] implemented by the Euro-
141 pean DRAKKAR collaboration [DRAKKAR Group, 2007]. The model is coupled
142 with the Louvain-la-Neuve (LIM2) sea ice thermodynamic and dynamic numerical
143 model [Fichefet and Maqueda, 1997]. The ORCA grid becomes finer with increas-
144 ing latitudes, so the effective 1/4° resolution is 27.75 km at the equator and 13.8
145 km at 60°S or 60°N. There are 46 unevenly spaced vertical levels, increasing in
146 thickness from 6 m near the surface to 250 m at depth, with partial steps in the
147 lowest level.
148 Atmospheric forcing was set up in the Co-ordinated Ocean-ice Reference Experi-
149 ments (CORE) framework [Griffies et al., 2009], using the forcing fields developed
150 by Large and Yeager [2009]. Surface forcing for KAB001 uses the standard CORE
151 forcing data. Depending on the field, the resolution is 6-hourly, daily or monthly.
152 Surface damping of sea surface salinity is weak (300 days for 10 m depth). In ad-
153 dition, a full three-dimensional restoring is performed for both T and S in the polar
154 regions with a timescale of 181 days [Biastoch et al., 2008]. This configuration has
155 been used in numerous studies, including one by Behrens et al. [2013] that exam-
156 ined the role of subarctic freshwater forcing on trends in the meridional overturning
157 circulation, linking it to the density of the Denmark Strait overflow.
–7– Confidential manuscript submitted to JGR-Oceans
158 2. ANHA4: The regional configuration Arctic and Northern Hemisphere Atlantic
159 (ANHA) with a 1/4° resolution, hereafter referred to as ANHA4, is also used.
160 NEMO 3.4 is used as a framework in ANHA4. The model is coupled with the
161 Louvain-la-Neuve (LIM2) sea ice thermodynamic and dynamic numerical model
162 [Fichefet and Maqueda, 1997]. ANHA4 covers the whole Arctic Ocean, North
163 Atlantic and a part of the South Atlantic with two open boundaries, one close to
164 Bering Strait and the other one at 20°S. The mesh grid is extracted from the 1/4°
165 global tripolar grid, ORCA025. The highest horizontal resolution (∼ 6km) is in
166 Dease Strait, an east-west waterway between the mainland of Kent Peninsula and
167 Victoria Island in Nunavut, Canada. The lowest resolution (∼ 28km) is at the Equa-
168 tor. It has 50 vertical levels, with the layer thickness smoothly increasing from 1.05
169 m at the surface, to 453.13 m in the last level.
170 Initial and monthly open boundary conditions are provided by Global Ocean Re-
171 analyses and Simulations 2 version 3 (GLORYS2v3) [Ferry et al., 2016]. The at-
172 mospheric forcing data, provided by Canadian Meteorological Centre’s global de-
173 terministic prediction system reforecasts (CGRF) data set [Smith et al., 2014], has
174 hourly 33 km resolution for: 10 m surface wind, 2 m air temperature and humid-
175 ity, downward shortwave and longwave radiation, and total precipitation. Interan-
176 nual monthly river runoff is from Dai et al. [2009], while melt-water discharge from
177 Greenland Ice Sheet is provided by Bamber et al. [2012]. The simulation ran with
178 ANHA4 covers the period from the beginning of 2002 to the end of 2016 and has
179 a temporal resolution of 5 days. Not temperature or salinity restoring is apply leav-
180 ing the model to evolve freely.
181 The regional configuration ANHA4 has been used in the past by Holdsworth and
182 Myers [2015] to explore the influence of high frequency atmospheric forcing on
183 the circulation and deep convection in the Labrador Sea. Dukhovskoy et al. [2016]
184 used the same configuration to look at the spreading of Greenland freshwater in
185 the sub-Arctic Seas. Gillard et al. [2016] explored the pathways of the melt-water
186 from marine terminating glaciers of the Greenland ice sheet, also by making use
187 ANHA4. Müller et al. [2017] made use of the ANHA4 configuration to explore the
188 temperature flux carried by eddies across 47°N in the Atlantic Ocean.
189 To analyse the pathways of inflowing AW and its export, we used a well-tested [Lique
190 et al., 2010; Hu and Myers, 2013; Gillard et al., 2016; Behrens et al., 2017] off-line La-
–8– Confidential manuscript submitted to JGR-Oceans
191 grangian tool, Ariane [Blanke and Raynaud, 1997; Blanke et al., 1999]. To compute three-
192 dimensional (3-D) trajectories, Ariane is provided with ocean velocity fields. For the study
193 carried with KAB001, approximately 2500 virtual particles were initialized at the entrance
194 of Denmark Strait each season (January, April, July, and October) every four years from
195 1960 to 2000 inclusive. In the case of the study carried with ANHA4, approximately
196 10000 virtual particles were initialized at the entrance of Denmark Strait each season (Jan-
197 uary, April, July, and October) every year from 2005 to 2014 inclusive. The section where
198 the virtual particles were released is represented in Figure 1 as a dashed black line.
199 The actual number of particles inflowing within the AW is determined by:
200 1. Particles are inserted based on:
Vi vi n = N( )( ), i ∈ 1..k, (1) i Ík v j=1 Vj
201 where ni is the number of particles to initialize around the i-th grid cell, k is the
202 number of grid points which meet the criteria, vi is the velocity of each identified
203 (as AW) grid cell, v¯ is the mean velocity of each identified grid cell, Vi is the vol-
204 ume of each identified grid cell, Vj is the total volume of each identified grid cell, a
205 given grid cell and N is the total number of particles to be initialized.
206 2. We define AW as: a water mass with a temperature higher than 3°C, salinity higher
3 3 207 than 34.9 and within the density range from 26.8 kg/m to 27.5 kg/m . However
3 208 there are very few months where densities as high as 27.5 kg/m are seen. If no
209 grid points meet this criteria at a given initialization time, no particles are seeded
210 that month.
211 We use two different models, covering two time periods and with different air sea
212 forcing to show that our results are robust and independent of the setup of a given model
213 configuration.
214 3 NIIC waters pathways into the Nordic Seas
215 In this section we describe the pathways of the virtual floats entering the Nordic
216 Seas within the NIIC and also leaving the Nordic Seas through Denmark Strait. First, we
217 will describe the results obtained by using KAB001, followed by a description of the re-
218 sults obtained by using ANHA4.
–9– Confidential manuscript submitted to JGR-Oceans
219 3.1 KAB001 experiment
220 Overall, 72% of the particles inflowing with the AW are estimated to leave the Nordic
221 Seas within 5 years of their entry. After 10 years, this increases to 92%. There is no de-
222 pendence on the month of particle release. Most of the exported particles (61%) exit to
223 the north, either into the Arctic Ocean via Fram Strait (45% of the northern export), or
224 into the Barents Sea between Svalbard and Norway (55% of the northern export). The two
225 pathways are correlated with r=0.57, but significant only at the 90 % level. This weak cor-
226 relation is stronger earlier in the analysis and breaks down in the 1990s. The mean tran-
227 sit time from Denmark Strait to the Barents Sea Opening is 2.3 years, and 3.4 years for
228 Fram Strait. The majority of the particles, 78%, remain within the top 300 m and thus in
229 the warm Atlantic layer. Significantly fewer particles exit the Nordic Seas by the southern
230 straits. About 10% of the total inflow leaves through Denmark Strait (17% after 10 years).
231 Although there is some inter-annual variability in these percentages, there are no
232 significant correlations between the number of particles exported and the Arctic or North
233 Atlantic Oscillations. Thus, most of the inflowing AW transits the Nordic Seas to the Bar-
234 ents Sea and Arctic before being transformed and returning to the Atlantic. Some wa-
235 ter could conceivably enter the Barents Sea or Fram Strait and then recirculate into the
236 Nordic Seas before being transformed.
237 Water that is transformed in the Nordic Seas tends to be saltier than water trans-
238 formed within the Arctic Ocean, but also warmer. For example, Hansen and Østerhus
239 [2000] found that the main water mass (which they call Modified East Iceland Water,
240 or MEIW) formed over winters in the Iceland Sea from the North Iceland Irminger Cur-
3 241 rent has a density range of 27.65 ≤ σθ ≤ 27.97 kg/m . Therefore, to track how the water
242 masses in this experiment were transformed in the Nordic Seas, we search for particles
243 with the density of MEIW and which are exported to the south below 200m. According
244 to our criteria, approximately 82% of the AW water imported through Denmark Strait is
245 transformed into MEIW within the Nordic Seas. An additional 10 % is transformed into
246 even denser water. Overall, water transformed within the Nordic Seas before being ex-
3 247 ported through the Denmark Strait experiences a mean increase in density of 0.4 kg/m ,
3 3 248 from 27.39 ± 0.03 kg/m to 27.82 ± 0.05 kg/m . The associated increase in depth (mea-
249 sured at the sill) is 293m, from 109 ± 10 m to 402 ± 26 m. Figures 2.a and 2.b shows the
250 relevant distributions for these properties, illustrating that outflowing particles are clearly
–10– Confidential manuscript submitted to JGR-Oceans
a) b)
c) d)
Figure 2: For the experiment KAB001: a) Depth distributions, in terms of relative percentages, for floats that enter through Denmark Strait (red) and then leave through Denmark Strait (blue) at depths greater than 200m. Purple shading is used to indicate the percentage for whichever of the
inflow/outflow has a smaller percentage in a given bin. b) As for a), but for density. c) Particle tracks for particles entering through Denmark Strait and then also leaving through Denmark Strait at depths greater than 200 m. d) As for c) but for particles leaving at all depths. Track colour (in c) and d)) indicates the decade in which the particle was released (Dark Blue - 1960s; Light Blue -
1970s; Green - 1980s; Orange - 1990s and Red - 2000s). The green boxes show the actual regions that were used to define the strait exits when a particle was determined to have entered the given box. For these two panels only, 1 in every 50 trajectories is plotted to reduce clutter.
251 part of a different water mass than the one they entered with and dense enough to feed the
252 overflow waters. The particles fill all the deeper levels of the overflow down to near the
253 sill.
254 To examine the pathways of the transformed particles and see where the transfor-
255 mation of inflowing AW to denser water is occurring, we plot a subset of particles, by
–11– Confidential manuscript submitted to JGR-Oceans
256 decade, that exit at depths greater than 200 m (Figure 2.c). Although some particles do
257 circulate around the Nordic Seas, the majority are transformed in the north of Iceland. We
258 note many particles take a very short route, recirculating north of Iceland and entering
259 the North Icelandic Jet to flow back into Denmark Strait (as first suggested by Våge et al.
260 [2011]). Other particles follow Jan Mayen Ridge to circulate around the Iceland Sea. Of
261 those particles not transformed which remain at shallow depths, significantly more circu-
262 late the Nordic Seas before leaving in the East Greenland Current (Figure 2.d). Very few
263 particles enter the interiors of the gyres in the Iceland, Norwegian and Greenland Seas,
264 but that might be a consequence of the model resolution and underestimation of eddy ex-
265 change.
266 3.2 ANHA4 experiment
267 Figure 3 shows the number of particles that after entering the Nordic Seas through
268 Denmark Strait in a given month and year, also leaves the Nordic Seas through Denmark
st 269 Strait at some point between their time of release and December 31 , 2014. On average,
270 out of the 10000 floats released every January (Figure 3) 5223 (52%), entered and left
271 the Nordic Seas through Denmark Strait. In the case of the particles released every April,
272 5133 (51%) particles leave later through Denmark Strait (Figure 3). For those released
273 in July, 5184 (52%), while 5677 (57%) of those released in October leave through Den-
274 mark Strait (Figure 3). The average number of particles seeded in October is larger com-
275 pared to the releases in other months. The reason behind this behaviour will be addressed
276 shortly. Given that the details of different release times are qualitatively similar, we shall
st 277 use the particles released on January 1 2005 for a case study. We consider this case as it
278 is the one that offers the longest time series in order to capture properly the floats evolu-
279 tion from their release date up to December 2014.
st 280 Figure 4.a shows the trajectories of all the floats released on January 1 , 2005, that
281 enter the Nordic Seas through Denmark Strait, and leave also through Denmark Strait. Out
282 of the 10000 floats released, 4722 floats enter and leave the Nordic Seas through Denmark
283 Strait. In this study we will focus on these floats to study the AW that gets transformed
284 in the Nordic Seas and then leaves through Denmark Strait, potentially contributing to
285 the DSOW. Different possible pathways (20 in total) were identified by using the seven
286 boxes displayed in Figure 4.a. The boxes and their positions were chosen to capture all
287 possible different routes through and around the Nordic Seas. The color scale shows how
–12– Confidential manuscript submitted to JGR-Oceans
7500 January April July October 7000
6500
6000
5500
5000
4500 Number of particles
4000
3500 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Years
Figure 3: Total number of particles entering the Nordic Seas through Denmark Strait every Jan-
uary (solid blue line), April (solid orange line), July (solid yellow line) and October (solid purple line), from 2005 to 2014 and which leave the Nordic Seas through Denmark Strait (not necessarily within the DSOW). Their respective averages are represented with a dashed line of the same color.
288 many particles would take a certain path by considering the boxes they pass through after
289 entering at Denmark Strait.
290 The majority of the particles follow four major routes (Figure 4.b). Figures S1-S5
291 (Supplementary Information) from show the trajectories of the floats released in other
292 years and months. Although there is some variability the overwhelming feature is that the
293 same four major routes are dominant.
294 We now focus on those four routes which are separately represented in Figures 4.c,
295 4.d, 4.e and 4.f. Few particles enter the interior of the Greenland, Iceland and Norwegian
296 basins. Most particles remain within the boundary currents. From the 4722 particles en-
297 tering and leaving the Nordic Seas through Denmark Strait, 3874 (82%) pass only through
298 box 4 before they leave (Figure 4.c). Thus these floats only recirculate north of Iceland
299 before leaving the Nordic Seas. There is little exchange into the interior of the Iceland
300 Sea. There is a higher incidence of particles that follow this route if they are released in
301 Denmark Strait in October (Figure S.5 - Supplementary Information). This is a function of
302 the predominant wind pattern over the region at this time of the year (not shown) which
303 is favourable for a strong recirculation of the AW as it enters through Denmark Strait, en-
304 hancing exchange out of the Nordic Seas.
–13– Confidential manuscript submitted to JGR-Oceans
a) b)
c) d)
e) f)
Figure 4: For the experiment ANHA4: a) Trajectories of all the particles released on January 1st , 2005 and integrated up to December 31st 2014, entering and leaving the Nordic Seas through Den- mark Strait. b) shows the main four routes found for the particles that enter and later leave through
Denmark Strait (not necessarily within the DSOW): c) Route 1, d) Route 2, e) Route 3 and f) Route 4. The numbers such as 4 or 1-5-6, represent the number of the boxes the float travels through while the number within brackets tells us how many particles follow the route.
305 A second major route (678 particles, 14.4% of total), 1-5-6-4 (Figures 4.b and 4.d),
306 represents floats that travel around the entire Nordic Seas within the rim current system
307 circulating around one or more of the Norwegian, Iceland or Greenland gyres. There is
308 still only limited exchange into the interior of the gyres. Two minor routes are shown in
–14– Confidential manuscript submitted to JGR-Oceans
309 Figures 4.e and 4.f. 65 floats (1.4% of total) enter through Denmark Strait before circulat-
310 ing around the southern Nordic Seas, feeding into the Icelandic Sea gyre and returning via
311 the Jan Mayen Current. 33 floats (0.7% of total) follow the rim current without entering
312 any of the three gyres within the Nordic Seas.
a) b)
Figure 5: a) Depth distributions, in terms of relative percentages, for floats that enter through Den- mark Strait (red) and then leave through Denmark Strait (blue) at depths greater than 200 m (as for Figure 2.a but for the ANHA4 experiment). Purple shading is used to indicate the percentage
for whichever of the inflow/outflow has a smaller percentage in a given bin. b) As for a, but for density.
313 Figures 5.a and 5.b (just like in Figure 2.a and 2.b) illustrate the transformation the
314 AW undergoes. Figure 5 (a and b) clearly shows again how the AW that leaves the Nordic
315 Seas below 200 m is transformed to densities that are high enough to contribute to the
316 DSOW, and the depth at where they are exported is above the depth of the Denmark Strait
317 sill. Following the MEIW criteria (like in the experiment KAB001), approximately 70%
318 of the water imported by the NIIC gets transformed in the Nordic Seas into MEIW and
319 exported through the Denmark Strait within 10 years. The mean density increase expe-
3 3 320 rienced is 0.3 kg/m , from 27.64 ± 0.14 to 27.93 ± 0.04 kg/m . The associated depth
321 increase (measured at the sill) is approximately 314 m, from 86 ± 1.3 m to 400 ± 25 m.
322 Where within the basin is this transformation occurring? We separate from all the
323 particles following the main four routes those that get transformed to a density high enough
324 to contribute eventually to the DSOW once they leave through Denmark Strait. In order to
325 do so, we first identify the characteristics of the DSOW in the ANHA4 experiment. A TS
326 diagram (Figure S.6 - Supplementary Information) of the water masses in Denmark Strait
–15– Confidential manuscript submitted to JGR-Oceans
327 showed the modelled DSOW as a water mass with density equal to or higher than 27.93
3 328 kg/m , colder than 1.5°C and a salinity between 34.9 and 34.95.
329 Figure 6.a shows the density of the selected floats. The floats are found to follow the
330 rim of the deepest basins within the Nordic Seas. There are two regions where the den-
3 331 sities reach values between 28.02 - 28.08 kg/m : north of Iceland and west and south of
332 Svalbard. The particles transformed north of Iceland appear to be densified as they travel
333 around the rim of the Iceland Sea and while travelling southward within the East Green-
334 land Current. The second region shows floats that are transformed as they travel within
335 the NwAC and the West Spitsbergen Current (WSC).
336 The depth of the selected floats (Figure 6.b) confirms that the transformation is oc-
337 curring on the continental slope. The deepest floats are found above 900 m, mainly those
338 travelling within the WSC. A few floats following the EGC along the rim of the Iceland
339 Sea and over the Jan Mayen Fracture Zone reach ∼ 800 m depth. There are two regions
340 where the floats are found at depths shallower than 100 m: north-west of Vestfirðir (north-
341 west of Iceland), mostly in the interior of the Ísafjarðjúp Fjord, Iceland, and around Spits-
342 bergen, south and north of Prins Karls Forland, and north of Widgefjorden Fjord, in north-
343 ern Spitsbergen. The transformation occurring north-west of Vestfirðir is most likely cap-
344 tured in Figure 5.b which shows that there is a small percent of AW that gets transformed
3 345 to densities higher than 28 kg/m before entering the Nordic Seas. The majority of the
346 floats travelling within the rim current system reach between 300 - 700 m depth.
347 The floats flowing along the Icelandic shelfbreak show a salinity decrease from
348 34.95 and 34.92 (Figure 6.c). This slight freshening is most likely due to some entrain-
349 ment from the EGC. The floats that separate from the Icelandic shelfbreak (from north-
350 east of Iceland to the Faroe Islands) show a salinity increase from 34.92 to 34.95. This
351 could occur due to entrainment from the branch of the North Atlantic Currents (NAC)
352 which enters the Nordic Seas through the Iceland-Scotland Ridge. The salinity of most
353 of the floats flowing within the NwAC and the WSC is between 34.945 to 34.95. As we
3 354 are only showing those floats whose density is equal or denser than 27.93 kg/m , the gap
355 without floats between the Faroe Islands and the NwAC suggests that the already trans-
356 formed AW is transformed back to lighter densities driven by an increase in the salinity
357 and temperature (Figure 6.c) due to entrainment.
–16– Confidential manuscript submitted to JGR-Oceans
358 As the floats recirculate south in Fram Strait and enter the EGC, their salinity de-
359 creases and is found to be between 34.9 and 34.915 (Figure 6.c). Nevertheless there are a
360 few floats within the EGC whose salinity reaches 34.95. Along the EGC, from Scoresby
361 Sund, south-east Greenland, to Denmark Strait, the salinity of the floats decreases down
362 to 34.9. For all the floats flowing within the rim current, it is in this region of the Iceland
363 Plateau where the floats show the lowest salinity. The evolution of the salinity fields point
364 out that oceanographic features like the NwAC and the EGC play a role in the transforma-
365 tion of the AW entering through Denmark Strait.
a) b) ] 3 Depth [m] Density [kg/m
c) d) Salinity Temperature [°C] Temperature
e) [Months] Transformation time Transformation
Figure 6: Properties of the floats that enter the Nordic Seas within the NIIC and are transformed to DSOW with densities higher than 27.93 kg/m3 (ANHA4): (a) density, (b) depth, (c) salinity, (d) temperature and (e) transformation time.
–17– Confidential manuscript submitted to JGR-Oceans
366 The shallower floats found north-west of Vestfirðir and south and north of Prins
367 Karls Forland have salinities of 34.95, while those located north of the Widgefjorden Fjord
368 are a bit fresher with a salinity of 34.9. Floats located over the Jan Mayen Fracture Zone
369 have salinities mostly between 34.9 and 34.915. Nevertheless, some of the floats show a
370 salinity of 34.95.
371 The temperature of the selected floats as they transit around the Nordic Seas is shown
372 in Figure 6.d. Over the Iceland Plateau the temperature of the floats goes from 0 to 1.5°C.
373 The floats closer to the Icelandic shelfbreak have a temperature between 1.5 - 2 °C. The
374 floats located along the eastern part of the Icelandic shelfbreak to the Faroe Islands have a
375 temperature between 1 - 2 °C, getting warmer (2.5°C) the closer they get to the Faroe Is-
376 lands. In general, floats travelling within the NwAC and the WSC have a temperature be-
377 tween 2 - 2.5°C. Nevertheless a few particles along the WSC have a lower temperature be-
378 tween 0.25 - 1.25°C. The floats flowing within the EGC have a temperature between 0.25
379 - 2°C. The floats found north-west of Vestfirðir have a temperature between 2.25 - 2.5°C.
380 The floats north of Prins Karls Forland, as well as those located north of the Widgefjorden
381 Fjord have a temperature between 0.25 - 0.75°C.
382 It takes up to 110 months for the selected floats to enter the Nordic Seas, get trans-
383 formed to higher densities and leave later through Denmark Strait, potentially as DSOW
384 (Figure 6.e). Within 20 months the floats flowing along the Icelandic shelfbreak travel
385 from Denmark Strait all the way to the Faroe Islands. Those floats following the NwAC
386 and the WSC take between 60 to 70 months to get transformed to overflow-like water. A
387 few floats, mostly those travelling closer to Norway and Spitsbergen shelfbreak, get trans-
388 formed within 20 to 40 months. That is also the case for the floats found north of Prins
389 Karls Forland, as well as those located north of the Widgefjorden Fjord in Iceland.
390 Along the EGC the floats show a transformation time around 60 to 80 months. A
391 small number of floats show a travel time up to 100 months, mostly those located over
392 and around Jan Mayen Fracture Zone. The older floats are found around the Iceland Sea,
393 taking up to 110 months.
394 4 NIIC waters transformation north of Iceland
395 In this section we consider only those floats transformed to densities equal or higher
3 396 than 27.93 kg/m along the Iceland shelfbreak within 20 months. The main purpose here
–18– Confidential manuscript submitted to JGR-Oceans
397 is to link the transformation of the AW north of Iceland with the NIJ which feeds the
398 DSOW.
399 We noticed that, in spite of their set-up differences, the analysis from KAB001 and
400 ANHA4 have similar outcomes with the majority of the particles travelling around the
401 shelf system of the Nordic Seas rather than going into the deeper part of the basin (Fig-
402 ures 2.c and 2.d for KAB001 and Figure 4 for ANHA4). It was also shown that both ex-
403 periments have a similar percentage of particles leaving through Denmark Strait at depth
404 greater than 200 m with overflow-like densities (Figures 2.a and 2.b for KAB001 and Fig-
405 ures 5.a and 5.b for ANHA4). At the same time both experiments show that part of the
406 particles are getting transformed in a loop north of Iceland where they reach densities
3 407 higher than 27.9 kg/m . Thus, we will focus only on the results from the ANHA4 ex-
408 periment in order to explore the transformation occurring north of Iceland of the waters
409 carried by the NIIC. Even when the experiment KAB001 provides a longer time series,
410 however considering that the travel time of the particles taking this route is 1 year from
411 the moment they enter to their export (already transformed) through Denmark Strait, the
412 integration time on the ANHA4 experiment is enough to explore the process.
413 Figure 7 shows the property evolution of the selected floats along the Icelandic
414 shelfbreak, which are found to follow the 1000-1500 m isobaths. Figure 7.a shows the
3 415 density of those floats which can be as high as 28.04 kg/m . North and north-east of Ice-
416 land the floats are found to be not deeper than 600 m, while north-west of Iceland and
417 closer to Denmark Strait the floats are found between 200 - 700 m deep (Figure 7.b). The
418 floats located north-west of Vestfirðir (mostly in the interior of the Ísafjarðjúp Fjord, Ice-
3 419 land) are as dense as 27.98 kg/m and not deeper than 100 m (Figure 7.b).
420 As the floats enter the Nordic Seas their salinity decreases during their travel along
421 the Icelandic shelfbreak (Figure 7.c). North-west of Iceland and closer to Denmark Strait,
422 the floats located closest to the Icelandic shelfbreak have a salinity around 34.95, while
423 those farther from the shelfbreak have a salinity that goes from 34.9 to 34.92. That dif-
424 ference represents the in-flowing (saltier) and the out-flowing (fresher and transformed)
425 floats. North of Iceland and along the shelfbreak the salinity decreases gradually to 34.91.
426 As the floats continue their travel along the north-eastern Icelandic shelfbreak their salinity
427 increases again up to 34.94.
–19– Confidential manuscript submitted to JGR-Oceans
a) b) ] 3 Depth [m] Density [kg/m
c) d) Salinity Temperature [°C] Temperature
e) [Months] Transformation time Transformation
Figure 7: Properties of the floats that enter the Nordic Seas within the NIIC and are transformed to densities higher than 27.93 kg/m3 in less than 1 year (ANHA4): (a) density, (b) depth, (c) salinity,
(d) temperature and (e) transformation time.
428 The along-shelfbreak transformation can also be observed in the evolution of the
429 temperature of the floats (Figure 7.d). As the floats enter their temperature decreases from
430 2.5°C to 0.5°C north of Iceland, where the coldest temperatures are found. North-east of
431 Iceland and along the shelfbreak the temperature increases again up to 2 °C. All the trans-
–20– Confidential manuscript submitted to JGR-Oceans
432 formation mentioned above occurs within 12 months from the moment the floats enter
433 through Denmark Strait (Figure 7.e).
434 Our results point out that, as the water gets transformed to higher densities along
435 the shelfbreak, it eventually cascades into deeper layers of the Iceland Sea driven by the
436 density increase. The transformed waters that contribute to the DSOW find their way to-
437 wards the strait thanks to the predominant flow pattern along the north Iceland shelfbreak
438 (Figure 8). The seasonal climatology of the current velocity (Figures 8.a to 8.d) shows
439 that, 100 to 200 km from the Iceland coast the predominant current direction is towards
440 the west. This will imply that, once that the transformed waters reach a depth greater
441 than 500 m they will be trapped within the westward flow and hence exported out of the
442 Nordic Seas as part of the DSOW.
443 Figure 9 shows the monthly climatology of total heat loss for the months October
444 through March, when the heat loss is the highest in the ANHA4 experiment. For each
445 month, the largest heat loss occurs on the shelf surrounding the Nordic Seas (except the
446 Greenland side most likely due to ice cover conditions), while less heat loss occurs in
447 the interior of the Iceland, Norwegian and Greenland gyres. Also interesting is the strong
448 heat loss that takes place in Denmark Strait. This heat loss is largest near the coast of Ice-
449 land, which coincides with the pathway of the NIIC as it enters the Nordic Seas. Along
450 the west coast of Svalbard and coinciding with the pathway of the WSC, the heat loss is
451 also strong mostly during January to March.
452 Either in Denmark Strait or along the west coast of Svalbard, the heat loss can be
2 453 as high as 400 W/m in March. This would induce a decrease in the temperature of the
454 AW as it transits, for example, along the Icelandic Shelfbreak. This water mass with an
455 already high salt content, would then gain in density as its temperature decreases induced
456 by the strong heat loss.
–21– Confidential manuscript submitted to JGR-Oceans
a) b)
Iceland Iceland Depth [m]
Velocity [m/s] Velocity
300 200 100 0 300 200 100 0
c) d)
Iceland Iceland Depth [m]
Velocity [m/s] Velocity
300 200 100 0 300 200 100 0 Distance [km] Distance [km]
Figure 8: Seasonal flow velocity normal to a cross section of the north Iceland shelfbreak: a) Winter (January to March), b) Spring (April to June), c) Summer (July to September) and d) Fall
(October to December). Location of the section is indicated by the map on the lower right corner of each panel. The length of the section is indicated in km.
–22– Confidential manuscript submitted to JGR-Oceans
a) b)
October November
c) d)
December January
e) f)
February March
Figure 9: Monthly climatology (2005-2014) of total heat loss diagnosed from the experiment ANHA4, for the months: a) October, b) November, c) December, d) January, e) February, and f) March
–23– Confidential manuscript submitted to JGR-Oceans
457 5 Discussion and Conclusions
458 We have explored the transformation of the NIIC as it enters the Nordic Seas through
459 Denmark Strait as well as the role played by this transformation in driving the NIJ. We
460 consider this question using two different configurations of an eddy-permitting ocean gen-
461 eral circulation model and the off-line Lagrangian tracking tool, Ariane. This tool has
462 been used in other studies examining both the upper and lower limbs of the MOC [e.g.
463 Desbruyères et al. [2013]; Lique et al. [2010]; Gary et al. [2011, 2012]; Lozier et al. [2013]],
464 and is ideal for examining the pathways and properties transformation of a given water
465 mass [e.g. Gillard et al. [2016]]. We thus insert a large numbers of particles at Denmark
466 Strait to track the transformation pathways of the NIIC into the Nordic Seas.
467 The two configurations used have different setups whose main features appear in
468 Table 1. This was done in order to show that our results are robust and independent of
469 the setup of a given model configuration. Both experiments, KAB001 and ANHA4, show
470 that the majority of the particles travel around the shelf system of the Nordic Seas rather
471 than going into the deeper part of the basin (Figures 2.c and 2.d for KAB001 and Fig-
472 ure 4 for ANHA4). Both experiments also have a similar percentage of particles leaving
473 through Denmark Strait at depth greater than 200 m with overflow-like densities (Figures
474 2.a and 2.b for KAB001 and Figures 5.a and 5.b for ANHA4). For the case of the experi-
475 ment KAB001 we do not explore in details the transformation of the NIIC waters found
476 to occur north of Iceland. We focus on ANHA4 here because of the higher resolution
477 and more recent atmospheric forcing. However as in the case of ANHA4, the experiment
478 KAB001 shows that part of the particles are getting transformed north of Iceland to densi-
3 479 ties higher than 27.9 kg/m .
480 The experiment KAB001 shows that, based on the time-line for the particles to leave
481 the Nordic Seas, the residence time for AW carried into the basin within the NIIC is 5
482 (72 % particles departed) to 10 years (92 % particles departed). However, since the ma-
483 jority of the particles that leave do so untransformed through the Barents Sea Opening
484 and Fram Strait, the residence time for AW in the combined Arctic Mediterranean may be
485 much larger [Rudels et al., 2015]. However, we did not consider the role of those particles
486 that entered the Arctic Ocean.
487 In the case of the ANHA4 experiment, 47% of the total particles (around 10000
488 each time) released in January 2005, were found to leave the Nordic Seas through Den-
–24– Confidential manuscript submitted to JGR-Oceans
Table 1: Table highlighting the main differences between the KAB001 and the ANHA4 experi- ments.
KAB001 ANHA4
Domain Global 20° S to Bering Strait Integration period 1958-2004 2002-2016
Radiation fluxes CORE CGRF Precipitation CORE CGRF 3-D restoring in polar regions 181 days None
# of particles released 2500 10000
489 mark Strait within 10 years. 34% of these particles transformed to densities higher than
3 490 27.93 kg/m and fed the DSOW within 6 years. This transformation was found to oc-
491 cur in a rim current overturning loop around the Nordic Seas. A faster transformation oc-
492 curred along the north Icelandic shelfbreak. We found that 30% of the particles inflowing
3 493 within the NIIC transformed to densities higher than 27.93 kg/m north of Iceland in less
494 than a year. This transformation leads to a contribution to the NIJ which is consistent with
495 the new circulation scheme proposed by Våge et al. [2011].
496 Despite the statement by Våge et al. [2013] that the transformation of the NIIC oc-
497 curs inside the Iceland Sea gyre, we find in our analisys that much of the transformation
498 is happening along the north Icelandic shelfbreak. As a matter of fact, both of the experi-
499 ments used in this study showed that most of the particles inflowing within the NIIC stay
500 on the shelf system rather than entering the Iceland, Norwegian and Greenland gyres. We
501 found this outcome to be in agreement with Jeansson et al. [2008] who, by using water
502 mass analysis from tracer observations, found that the contribution from the Greenland
503 Sea to the DSOW is only 10%, while the remaining 90% was found to have its source
504 in the eastern Nordic Seas and the Arctic Ocean. Our results also agree with Behrens
505 et al. [2017]. By using Eulerian and Lagrangian diagnostics they analyse the upstream
506 sources of the DSOW in a global simulation at 1/20° resolution, from 1948 to 2009. They
507 found that the sources of the NIJ are mainly originated from the transformation of the
508 NIIC north of Iceland. This transformation, they argue, occurs mainly along the north-
–25– Confidential manuscript submitted to JGR-Oceans
509 ern Icelandic shelfbreak, contrary to recent findings from Våge et al. [2013] and Våge et al.
510 [2015].
3 511 None of the particles transformed to densities higher than 27.93kg/m within one
512 year were found to pass through the interior of the Iceland Sea gyre (ANHA4 experiment;
513 Figure 7). Similarly the particles transformed to the same densities within 6 years along
514 the shelf system of the Nordic Sea, did not enter the Greenland, Norwegian or the Iceland
515 gyres (Figure 6). The transformation in less than one year that was found to occur on the
516 Icelandic shelfbreak suggests that the in-phase interannual variability in salinity of the in-
517 flowing NIIC with that of the outflowing NIJ [Pickart et al., 2017] might actually be given
518 by the NIIC salinity signal dictating that of the NIJ.
519 The transformation of the AW circulating around the Nordic Seas was found to be
520 driven by heat loss. Strong heat loss occurs at Denmark Strait, along the Icelandic shelf-
521 break, which coincides with the pathway followed by the NIIC. During this transit the
3 522 density of the Atlantic inflow increases on an average of 0.3 kg/m , from 27.64 ± 0.14
3 523 to 27.93 ± 0.04 kg/m , leading to an increase in depth by around 314 m, from 86 ± 1.3
524 m to 400 ± 25 m. A strong heat loss also occurs for those particles transiting within the
525 NwAC and the WSC, leading to the further transformation of the AW.
526 This result confirms the hypothesis by Mauritzen [1996] and agrees with the results
527 of Isachsen et al. [2007]. By using air-sea fluxes from reanalysis products and climatolog-
528 ical hydrographic data, Isachsen et al. [2007] found that a strong light-to-dense transfor-
529 mation takes place east of the Mohn-Knipovich Ridge system, which coincides with the
530 region where we found (Figures 9.a - 9.f) a strong heat loss happening during the win-
531 ter months together with light-to-dense water transformation (Figures 6.a - 6.e). By using
532 air-sea fluxes from reanalysis products and hydrographic data from Argo floats Latarius
533 and Quadfasel [2016] investigated the water mass transformation exclusively in the deepest
534 basins of the Nordic Seas. They found that, in fact, the Greenland Sea does not impact the
535 density of the overflow water, given that much of the density transformation was found to
536 occur, agreeing with Isachsen et al. [2007], in the eastern basins of the Nordic Seas.
537 We do not expand here on to what causes the heat loss pattern in the Nordic Seas
538 as it is not under the scope of our study. However, given its role on driving the AW trans-
539 formation it is a topic that needs further study. An increase in the air temperature over the
540 Nordic Seas will imply, eventually, a reduction in the oceanic heat loss and hence a de-
–26– Confidential manuscript submitted to JGR-Oceans
541 creased in the light-to-dense transformation of the AW occurring north of Iceland and in
542 the eastern shelf system of the Nordic Seas. This will be reflected in the overflow strength
543 and the Atlantic MOC.
544 One limitation of the simulations used in this study is that the underlying model
545 used is only eddy-permitting, with resolution insufficient to resolve all the eddies [Hall-
546 berg, 2013]. One of the main mechanisms by which AW and therefore particles, is ex-
547 changed out of the boundary currents is thus missing. However similar analysis carried
548 out using 1/12° resolution (not shown) confirmed that the transformation does occur along
549 the shelf system in the Nordic Seas, and not within the Iceland, Norwegian and Green-
550 land gyres. Given that the particles leaving the Nordic Seas through Denmark Strait reach
3 551 densities of ≥ 27.92 kg/m confirms that even the most dense parts of the overflows can
552 come from boundary current transformations, rather than deep convection in the interior
553 of the Nordic Seas. That said, we can not rule out other sources, including interior deep
554 waters, contributing to the overflow.
555 In a model inter-comparison study using passive tracers to track pathways of melt
556 waters from the Greenland Ice Sheet, Dukhovskoy et al. [2016] showed that the take-up
557 of the tracer in the interior of the Nordic Seas was about double, over 10 years, using a
558 1/12° degree model as compared to a 1/4° simulation. However, the same study showed
559 that some of the Greenland melt passive tracer reached the interior of the Greenland Sea
560 in the 1/4° model, mainly via the East Greenland Current. Therefore, although it would
561 be useful to perform a study similar to ours at higher resolution, we believe that our nar-
562 rative of the majority of inflow transformation occurring in the Nordic Seas boundary
563 currents (as opposed to gyres interior) is not unrealistic. Thus, provided that AW is still
564 transformed by winter cooling in the boundary currents, even a significant influx of fresh
565 water to the interior of the Nordic Seas (leading to a prolonged cessation of convection)
566 may not have a significant impact on the local overflows strength and global MOC.
567 Acknowledgments
568 We gratefully acknowledge the financial and logistic support of grants from the Natu-
569 ral Sciences and Engineering Research Council (NSERC) of Canada. These includes a
570 Discovery Grant (rgpin227438-09) awarded to P.G. Myers, Climate Change and Atmo-
571 spheric Research Grant (VITALS - RGPCC 433898), and an International Create (Arc-
572 Train - 432295). We are grateful to the NEMO development team and the Drakkar project
–27– Confidential manuscript submitted to JGR-Oceans
573 for providing the model and continuous guidance, and to Westgrid and Compute Canada
574 for computational resources. We would also like to thank G. Smith for the CGRF forc-
575 ing fields, made available by Environment and Climate Change Canada. Output from the
576 KAB001 run of ORCA025 was provided by C. Boening’s group at GEOMAR, Kiel, Ger-
577 many. To gain access to the dataset used in this study please contact Paul G. Myers.
578 References
579 Aagaard, K., J. H. Swift, and E. C. Carmack (1985), Thermohaline circulation in the Arc-
580 tic Mediterranean Seas, Journal of Geophysical Research: Oceans, 90(C3), 4833–4846,
581 doi:10.1029/JC090iC03p04833.
582 Bamber, J., M. den Broeke, J. Ettema, J. Lenaerts, and E. Rignot (2012), Recent large
583 increases in freshwater fluxes from Greenland into the North Atlantic, Geophysical Re-
584 search Letters, 39(19), doi:10.1029/2012GL052552.
585 Behrens, E., A. Biastoch, and C. W. Böning (2013), Spurious AMOC trends in global
586 ocean sea-ice models related to subarctic freshwater forcing, Ocean Modelling, 69, 39
587 – 49, doi:https://doi.org/10.1016/j.ocemod.2013.05.004.
588 Behrens, E., K. Våge, B. Harden, A. Biastoch, and C. W. Böning (2017), Composition and
589 variability of the Denmark Strait Overflow Water in a high-resolution numerical model
590 hindcast simulation, Journal of Geophysical Research: Oceans, 122(4), doi:10.1002/
591 2016JC012158.
592 Bernard, B., G. Madec, T. Penduff, J.-M. Molines, A.-M. Treguier, J. L. Sommer,
593 A. Beckmann, A. Biastoch, C. Böning, J. Dengg, C. Derval, E. Durand, S. Gulev,
594 E. Remy, C. Talandier, S. Theetten, M. Maltrud, J. McClean, and B. D. Cuevas (2006),
595 Impact of partial steps and momentum advection schemes in a global ocean circula-
596 tion model at eddy-permitting resolution, Ocean Dynamics, 56(5-6), 543–567, doi:
597 10.1007/s10236-006-0082-1.
598 Biastoch, A., C. W. Böning, J. Getzlaff, J.-M. Molines, and G. Madec (2008), Causes
599 of Interannual–Decadal Variability in the Meridional Overturning Circulation of
600 the Midlatitude North Atlantic Ocean, Journal of Climate, 21(24), 6599–6615, doi:
601 10.1175/2008JCLI2404.1.
602 Blanke, B., and S. Raynaud (1997), Kinematics of the Pacific Equatorial Undercurrent: An
603 Eulerian and Lagrangian approach from GCM results, Journal of Physical Oceanogra-
604 phy, 27(6), 1038–1053, doi:10.1175/1520-0485(1997)027<1038:KOTPEU>2.0.CO;2.
–28– Confidential manuscript submitted to JGR-Oceans
605 Blanke, B., M. Arhan, G. Madec, and S. Roche (1999), Warm water paths in the Equa-
606 torial Atlantic as diagnosed with a General Circulation Model, Journal of Physical
607 Oceanography, 29(11), 2753–2768, doi:10.1175/1520-0485(1999)029<2753:WWPITE>
608 2.0.CO;2.
609 Dai, A., T. Qian, K. E. Trenberth, and J. D. Milliman (2009), Changes in continental
610 freshwater discharge from 1948 to 2004, Journal of Climate, 22(10), 2773–2792, doi:
611 10.1175/2008JCLI2592.1.
612 Desbruyères, D., V. Thierry, and H. Mercier (2013), Simulated decadal variability of the
613 meridional overturning circulation across the A25-Ovide section, Journal of Geophysical
614 Research: Oceans, 118(1), 462–475, doi:10.1029/2012JC008342.
615 Dickson, R. R., and J. Brown (1994), The production of north atlantic deep water:
616 Sources, rates, and pathways, Journal of Geophysical Research: Oceans, 99(C6),
617 12,319–12,341, doi:10.1029/94JC00530.
618 DRAKKAR Group (2007), Eddy-permitting ocean circulation hindcasts of past decades,
619 CLIVAR Exchanges, 12(3), 8 – 10.
620 Drinkwater, K., E. Colbourne, H. Loeng, S. Sundby, and T. Kristiansen (2013), Compar-
621 ison of the atmospheric forcing and oceanographic responses between the Labrador
622 Sea and the Norwegian and Barents seas, Progress in Oceanography, 114, 11 – 25, doi:
623 https://doi.org/10.1016/j.pocean.2013.03.007.
624 Dukhovskoy, D. S., P. G. Myers, G. Platov, M.-L. Timmermans, B. Curry, A. Proshutin-
625 sky, J. L. Bamber, E. Chassignet, X. Hu, C. M. Lee, and R. Somavilla (2016), Green-
626 land freshwater pathways in the sub-Arctic Seas from model experiments with pas-
627 sive tracers, Journal of Geophysical Research: Oceans, 121(1), 877–907, doi:10.1002/
628 2015JC011290.
629 Eldevik, T., J. E. Ø. Nilsen, D. Iovino, K. A. Olsson, A. B. Sandø, and H. Drange (2009),
630 Observed sources and variability of Nordic seas overflow, Nature Geoscience, 2(6),
631 406–410, doi:10.1038/ngeo518.
632 Ferry, N., L. Parent, S. Masina, A. Storto, H. Zuo, and M. Balmaseda (2016), Product
633 user manual for Global Ocean Reanalysis Products GLOBAL-REANALYSIS-PHYS-001-
634 009, GLOBAL-REANALYSIS-PHYS-001-011 and GLOBAL-REANALYSIS-PHYS-001-017,
635 EU Copernicus Marine Service.
636 Fichefet, T., and M. Maqueda (1997), Sensitivity of a global sea ice model to the treat-
637 ment of ice thermodynamics and dynamics, Journal of Geophysical Research: Oceans
–29– Confidential manuscript submitted to JGR-Oceans
638 (1978–2012), 102(C6), 12,609–12,646, doi:10.1029/97JC00480.
639 Fischer, J., J. Karstensen, R. Zantopp, M. Visbeck, A. Biastoch, E. Behrens, C. Böning,
640 D. Quadfasel, K. Jochumsen, H. Valdimarsson, S. Jónsson, S. Bacon, N. Holliday,
641 S. Dye, M. Rhein, and C. Mertens (2015), Intra-seasonal variability of the DWBC in
642 the western subpolar North Atlantic, Progress in Oceanography, 132, 233 – 249, doi:
643 10.1016/j.pocean.2014.04.002.
644 Gary, S. F., M. S. Lozier, C. W. Böning, and A. Biastoch (2011), Deciphering the path-
645 ways for the deep limb of the meridional overturning circulation, Deep Sea Research
646 Part II: Topical Studies in Oceanography, 58(17), 1781 – 1797, doi:10.1016/j.dsr2.2010.
647 10.059.
648 Gary, S. F., M. S. Lozier, A. Biastoch, and C. W. Böning (2012), Reconciling tracer and
649 float observations of the export pathways of Labrador Sea Water, Geophysical Research
650 Letters, 39(24), doi:10.1029/2012GL053978.
651 Gerdes, R., W. Hurlin, and S. M. Griffies (2006), Sensitivity of a global ocean model to
652 increased run-off from Greenland, Ocean Modelling, 12(3), 416 – 435, doi:10.1016/j.
653 ocemod.2005.08.003.
654 Gillard, L. C., X. Hu, P. G. Myers, and J. L. Bamber (2016), Meltwater pathways from
655 marine terminating glaciers of the Greenland ice sheet, Geophysical Research Letters,
656 43(20), 10,873–10,882, doi:10.1002/2016GL070969, 2016GL070969.
657 Griffies, S. M., A. Biastoch, C. Böning, F. Bryan, G. Danabasoglu, E. P. Chassignet,
658 M. H. England, R. Gerdes, H. Haak, R. W. Hallberg, et al. (2009), Coordinated ocean-
659 ice reference experiments (COREs), Ocean Modelling, 26(1), 1–46, doi:10.1016/j.
660 ocemod.2008.08.007.
661 Hallberg, R. (2013), Using a resolution function to regulate parameterizations of oceanic
662 mesoscale eddy effects, Ocean Modelling, 72, 92 – 103, doi:10.1016/j.ocemod.2013.08.
663 007.
664 Hansen, B., and S. Østerhus (2000), North Atlantic–Nordic Seas exchanges, Progress in
665 Oceanography, 45(2), 109–208, doi:10.1016/S0079-6611(99)00052-X.
666 Hansen, B., and S. Østerhus (2007), Faroe Bank Channel Overflow 1995–2005, Progress
667 in Oceanography, 75(4), 817–856, doi:10.1016/j.pocean.2007.09.004.
668 Hansen, B., W. R. Turrell, and S. Østerhus (2001), Decreasing overflow from the Nordic
669 seas into the Atlantic Ocean through the Faroe Bank channel since 1950, Nature,
670 411(6840), 927–930, doi:10.1038/35082034.
–30– Confidential manuscript submitted to JGR-Oceans
671 Harden, B., R. Pickart, H. Valdimarsson, K. Våge, L. de Steur, C. Richards, F. Bahr,
672 D. Torres, E. Børve, S. Jónsson, A. Macrander, S. Østerhus, L. Håvik, and T. Hatter-
673 mann (2016), Upstream sources of the Denmark Strait Overflow: Observations from a
674 high-resolution mooring array, Deep Sea Research Part I: Oceanographic Research Pa-
675 pers, 112, 94 – 112, doi:10.1016/j.dsr.2016.02.007.
676 Holdsworth, A. M., and P. G. Myers (2015), The influence of high-frequency atmospheric
677 forcing on the circulation and deep convection of the Labrador Sea, Journal of Climate,
678 28(12), 4980–4996, doi:10.1175/JCLI-D-14-00564.1.
679 Hu, A., G. A. Meehl, W. Han, and J. Yin (2011), Effect of the potential melting of the
680 Greenland Ice Sheet on the Meridional Overturning Circulation and global climate in
681 the future, Deep Sea Research Part II: Topical Studies in Oceanography, 58(17), 1914 –
682 1926, doi:10.1016/j.dsr2.2010.10.069.
683 Hu, X., and P. G. Myers (2013), A Lagrangian view of Pacific water inflow pathways in
684 the Arctic Ocean during model spin-up, Ocean Modelling, 71, 66–80, doi:10.1016/j.
685 ocemod.2013.06.0077.
686 Isachsen, P. E., C. Mauritzen, and H. Svendsen (2007), Dense water formation in the
687 Nordic Seas diagnosed from sea surface buoyancy fluxes, Deep Sea Research Part I:
688 Oceanographic Research Papers, 54(1), 22–41, doi:10.1016/j.dsr.2006.09.008.
689 Jeansson, E., S. JutterstrÃűm, B. Rudels, L. G. Anderson, K. A. Olsson, E. P. Jones,
690 W. M. Smethie, and J. H. Swift (2008), Sources to the east greenland current and its
691 contribution to the denmark strait overflow, Progress in Oceanography, 78(1), 12 – 28,
692 doi:10.1016/j.pocean.2007.08.031.
693 Jochumsen, K., D. Quadfasel, H. Valdimarsson, and S. Jónsson (2012), Variability of the
694 Denmark Strait overflow: Moored time series from 1996–2011, Journal of Geophysical
695 Research: Oceans, 117(C12), doi:10.1029/2012JC008244.
696 Jochumsen, K., M. Moritz, N. Nunes, D. Quadfasel, K. M. H. Larsen, B. Hansen,
697 H. Valdimarsson, and S. Jonsson (2017), Revised transport estimates of the Denmark
698 Strait overflow, Journal of Geophysical Research: Oceans, 122(4), 3434–3450, doi:
699 10.1002/2017JC012803.
700 Large, W. G., and S. G. Yeager (2009), The global climatology of an interannually
701 varying air–sea flux data set, Climate Dynamics, 33(2), 341–364, doi:10.1007/
702 s00382-008-0441-3.
–31– Confidential manuscript submitted to JGR-Oceans
703 Latarius, K., and D. Quadfasel (2016), Water mass transformation in the deep basins of
704 the nordic seas: Analyses of heat and freshwater budgets, Deep Sea Research Part I:
705 Oceanographic Research Papers, 114, 23 – 42, doi:10.1016/j.dsr.2016.04.012.
706 Lique, C., A. M. Treguier, B. Blanke, and N. Grima (2010), On the origins of water
707 masses exported along both sides of Greenland: A Lagrangian model analysis, Journal
708 of Geophysical Research: Oceans, 115(C5), doi:10.1029/2009JC005316.
709 Lozier, M. S., S. F. Gary, and A. S. Bower (2013), Simulated pathways of the overflow
710 waters in the North Atlantic: Subpolar to subtropical export, Deep Sea Research Part
711 II: Topical Studies in Oceanography, 85, 147 – 153, doi:http://dx.doi.org/10.1016/j.dsr2.
712 2012.07.037.
713 Macrander, A., U. Send, H. Valdimarsson, S. Jónsson, and K. R. H. (2005), Interannual
714 changes in the overflow from the Nordic Seas into the Atlantic Ocean through Denmark
715 Strait, Geophysical Research Letters, 32(6), doi:10.1029/2004GL021463.
716 Madec, G. and the NEMO team (2008), NEMO ocean engine, Note du Pôle de modélisa-
717 tion, Institut Pierre-Simon Laplace (IPSL), France, No 27, ISSN No 1288-1619.
718 Mauritzen, C. (1996), Production of dense overflow waters feeding the North Atlantic
719 across the Greenland-Scotland Ridge. Part 1: Evidence for a revised circulation scheme,
720 Deep Sea Research Part I: Oceanographic Research Papers, 43(6), 769 – 806, doi:
721 10.1016/0967-0637(96)00037-4.
722 Mertens, C., M. Rhein, M. Walter, C. W. Böning, E. Behrens, D. Kieke, R. Steinfeldt,
723 and U. Stöber (2014), Circulation and transports in the newfoundland basin, western
724 subpolar north atlantic, Journal of Geophysical Research: Oceans, 119(11), 7772–7793,
725 doi:10.1002/2014JC010019.
726 More, C. (2010), Circulation patterns and outflow characteristics across the Greenland-
727 Scotland Ridge of North Atlantic Current water in the Nordic Sea, Master’s thesis, Uni-
728 versity of Alberta, Edmonton, Alberta, Canada.
729 Müller, V., D. Kieke, P. G. Myers, C. Pennelly, and C. Mertens (2017), Temperature flux
730 carried by individual eddies across 47°N in the Atlantic Ocean, Journal of Geophysical
731 Research: Oceans, 122(3), 2441–2464, doi:10.1002/2016JC012175.
732 Orvik, K. A., and P. Niiler (2002), Major pathways of atlantic water in the northern north
733 atlantic and nordic seas toward arctic, Geophysical Research Letters, 29(19), 2–1–2–4,
734 doi:10.1029/2002GL015002.
–32– Confidential manuscript submitted to JGR-Oceans
735 Pickart, R. S., M. A. Spall, D. J. Torres, K. Våge, H. Valdimarsson, C. Nobre, G. W. K.
736 Moore, S. Jonsson, and D. Mastropole (2017), The North Icelandic Jet and its relation-
737 ship to the North Icelandic Irminger Current, Journal of Marine Research, 75(5), 605–
738 639, doi:10.1357/002224017822109505.
739 Rhein, M. (1991), Ventilation rates of the Greenland and Norwegian Seas derived from
740 distributions of the chlorofluoromethanes F11 and F12, Deep Sea Research Part A.
741 Oceanographic Research Papers, 38(4), 485 – 503, doi:10.1016/0198-0149(91)90048-K.
742 Rudels, B., M. Korhonen, U. Schauer, S. Pisarev, B. Rabe, and A. Wisotzki (2015), Circu-
743 lation and transformation of Atlantic water in the Eurasian Basin and the contribution of
744 the Fram Strait inflow branch to the Arctic Ocean heat budget, Progress in Oceanogra-
745 phy, 132, 128 – 152, doi:10.1016/j.pocean.2014.04.003.
746 Schlosser, P., G. Bönisch, M. Rhein, and R. Bayer (1991), Reduction of deepwater for-
747 mation in the greenland sea during the 1980s: Evidence from tracer data, Science,
748 251(4997), 1054–1056, doi:10.1126/science.251.4997.1054.
749 Send, U., M. Lankhorst, and T. Kanzow (2011), Observation of decadal change in the At-
750 lantic meridional overturning circulation using 10 years of continuous transport data,
751 Geophysical Research Letters, 38(24), doi:10.1029/2011GL049801.
752 Smith, G. C., F. Roy, P. Mann, F. Dupont, B. Brasnett, J.-F. Lemieux, S. Laroche, and
753 S. Bélair (2014), A new atmospheric dataset for forcing ice–ocean models: Evaluation
754 of reforecasts using the Canadian global deterministic prediction system, Quarterly Jour-
755 nal of the Royal Meteorological Society, 140(680), 881–894, doi:10.1002/qj.2194.
756 Somavilla, R., U. Schauer, and G. Budéus (2013), Increasing amount of Arctic Ocean
757 deep waters in the Greenland Sea, Geophysical Research Letters, 40(16), 4361–4366,
758 doi:10.1002/grl.50775.
759 Stouffer, R. J., J. Yin, J. M. Gregory, K. W. Dixon, M. J. Spelman, W. Hurlin, A. J.
760 Weaver, M. Eby, G. M. Flato, H. Hasumi, A. Hu, J. H. Jungclaus, I. V. Kamenkovich,
761 A. Levermann, M. Montoya, S. Murakami, S. Nawrath, A. Oka, W. R. Peltier, D. Y. Ro-
762 bitaille, A. Sokolov, G. Vettoretti, and S. L. Weber (2006), Investigating the causes of
763 the response of the thermohaline circulation to past and future climate changes, Journal
764 of Climate, 19(8), 1365–1387, doi:10.1175/JCLI3689.1.
765 Swingedouw, D., P. Braconnot, P. Delecluse, E. Guilyardi, and O. Marti (2007), The
766 impact of global freshwater forcing on the thermohaline circulation: adjustment of
767 North Atlantic convection sites in a CGCM, Climate Dynamics, 28(2-3), 291–305, doi:
–33– Confidential manuscript submitted to JGR-Oceans
768 10.1007/s00382-006-0171-3.
769 Våge, K., R. S. Pickart, M. A. Spall, H. Valdimarsson, S. Jonsson, D. J. Torres, S. Øster-
770 hus, and T. Eldevik (2011), Significant role of the North Icelandic Jet in the forma-
771 tion of Denmark Strait overflow water, Nature Geoscience, 4, 723 – 727, doi:10.1038/
772 ngeo1234.
773 Våge, K., R. S. Pickart, M. A. Spall, G. Moore, H. Valdimarsson, D. J. Torres, S. Y. Ero-
774 feeva, and J. E. Ø. Nilsen (2013), Revised circulation scheme north of the Denmark
775 Strait, Deep Sea Research Part I: Oceanographic Research Papers, 79, 20 – 39, doi:
776 10.1016/j.dsr.2013.05.007.
777 Våge, K., G. Moore, S. Jónsson, and H. Valdimarsson (2015), Water mass transformation
778 in the Iceland Sea, Deep Sea Research Part I: Oceanographic Research Papers, 101, 98
779 – 109, doi:10.1016/j.dsr.2015.04.001.
780 Weijer, W., M. E. Maltrud, M. W. Hecht, H. A. Dijkstra, and M. A. Kliphuis (2012), Re-
781 sponse of the Atlantic Ocean circulation to Greenland Ice Sheet melting in a strongly-
782 eddying ocean model, Geophysical Research Letters, 39(9), doi:10.1029/2012GL051611,
783 l09606.
784 Yang, J., and L. J. Pratt (2013), On the Effective Capacity of the Dense-Water Reservoir
785 for the Nordic Seas Overflow: Some Effects of Topography and Wind Stress, Journal of
786 Physical Oceanography, 43(2), 418–431, doi:10.1175/JPO-D-12-087.1.
787 Yashayaev, I., and D. Seidov (2015), The role of the Atlantic Water in multidecadal ocean
788 variability in the Nordic and Barents Seas, Progress in Oceanography, 132, 68–127, doi:
789 10.1016/j.pocean.2014.11.009.
–34–