Davis Strait, situated between western Greenland and Baffin Island (Fig. A), is a key gateway for the exchange of water masses between the Arctic Ocean and the North Atlantic . The sill of Davis Strait, with a depth of approximately 1060 m, defines the southern boundary of Baffin Bay (maximum depth 2300 m) and the northern boundary of the Labrador Sea (maximum depth 3500 m). Through this passage, cold and fresh southward-flowing waters from Baffin Bay meet the northward-flowing warm Atlantic waters entering from the Labrador Sea. Both the fluxes and the water mass properties in Davis Strait exhibit pronounced seasonal variability . The largest changes in temperature and salinity originate from modifications within Baffin Bay, which receives Arctic-derived waters via the main channels of the Canadian Arctic Archipelago (CAA). These include Nares Strait, connecting to the Eurasian Basin, as well as Lancaster Sound and Jones Sound, which link to the Canada Basin . At the same time, Baffin Bay receives an Atlantic inflow from the Labrador Sea, linking Baffin Bay to the western subpolar gyre and key regions of deep convection, which contribute to the Atlantic Meridional Overturning Circulation (AMOC), the ventilation of deep waters, and the oceanic uptake of anthropogenic CO₂ .

The exchanges of water masses within Baffin Bay, and their associated variability and impacts, underscore the importance of understanding the complex circulation system of the region. The general circulation in Baffin Bay (Fig. A) is largely influenced by boundary currents. From eastern Davis Strait, the northward-flowing West Greenland Current system (WGC, dark red arrows, Fig. A) transports two components along the Greenland shelf. The fresh West Greenland Shelf Water (WGSW, dark red), transported by the West Greenland Coastal Current, and the warm and saline West Greenland Irminger Water (WGIW, light green arrow Fig. A), which is confined within a shelf-break jet . The WGSW originates from the East Greenland Current (EGC, dark red arrow east of Greenland, Fig. A), which outflows Fram Strait and flows south along the Greenland shelfbreak, carrying fresh and relatively warm Polar Surface Water (PSW) from the Arctic Ocean . After rounding the southern tip of Greenland at Cape Farewell, the EGC is joined by WGIW at depth. The warm and saline WGIW originates from the North Atlantic Current (NAC, black arrows Fig. A), carrying water from the subtropics . Towards the northern Labrador Sea, the current system of west Greenland becomes baroclinically and barotropically unstable due to the steep continental slope, leading to large anticyclonic eddies. The eddies are known as Irminger Rings and carry both the WGSW and WGIW offshore into the Labrador Sea . In the northern Labrador Sea, the WGC bifurcates into two branches, with one continuing north into Baffin Bay, and a larger branch following the bathymetry of the Labrador Sea, turning westward towards the Labrador Shelf .

Western Davis Strait is dominated by the southward-flowing surface Baffin Island Current (BIC, orange arrow Fig. A), which transports fresh water of Arctic origin, such as Arctic Water (AW) and cold Arctic Water (cold AW), along the Baffin Island shelf and slope to the Labrador Sea . While Arctic Water is strongly influenced by glacial run-off and sea ice melt , cold Arctic Water (similar to the hydrographic properties of cold Polar Water; ) represents a subset of cold and more saline water within Arctic Water, which experienced winter convection by stronger cooling and salinification . Waters exiting the Arctic Ocean enter Baffin Bay through the shallow sills of Nares Strait 220 m,, Lancaster Sound 125 m, and Jones Sound 125 m,. These water masses with low salinities (<33.5) and low temperatures (<1 °C) are present primarily at the surface of Baffin Bay and along the BIC.

In central Davis Strait, at intermediate depths between 300 and 600 m, relatively warm Transition Water (dark blue Fig. A) flows throughout the year and at low velocities out of Baffin Bay . described Transition Water as an intermediate layer formed by mixing Atlantic water from the south and colder water from the north. More recently, identified two different types of Transition Water, each resulting from the mixing of several water masses. Further, identified a temperature maximum in western Baffin Bay aligning with the warmest and deepest fraction of Transition Water. Below Transition Water, down to 1600 m depth, the water column is dominated by Baffin Bay Mode Water (BBMW), also known as Baffin Bay Deep Water, whose formation processes remain unclear . In central Baffin Bay, depths greater than 1600 m are filled by old Baffin Bay Bottom Water (BBBW) , which is not observed in the Davis Strait outflow .

South of Davis Strait, the cold, fresh surface water from the BIC joins the Labrador Current, where they mix with cool, fresh outflow from the Hudson Strait and water from the WGC that turned westward, bathymetrically steared at the Labrador Sea .

In the Labrador Sea, the Irminger Rings detach from the boundary current and drift into the central basin. This water contributes to the formation of Labrador Sea Water (LSW, light blue arrows in Fig. A), together with fresh water from the Labrador Current, which accounts for approximately 6 %–8 % of LSW . The annual deep winter convection that forms the LSW represents a key component of the AMOC . Other water masses in the Labrador Sea include the North East Atlantic Deep Water (NEADW, yellow arrow in Fig. A), which follows the cyclonic Deep Western Boundary Current and is found below LSW at approximately 2000 m depth. The NEADW forms by mixing of multiple source water inflows from east of Greenland, including LSW, Denmark Strait Overflow Water (DSOW, dark green in Fig. A), and Iceland Scotland Overflow Water (ISOW, light green in Fig. A) . The bottom depths are occupied by DSOW and carried primarily by the Deep Western Boundary Current from its formation region in the Nordic Seas towards the Grand Banks .

Decades of oceanographic studies have significantly improved our understanding of water mass composition and volumetric transport in Davis Strait, Baffin Bay, and the Labrador Sea . However, climate-driven changes are altering the circulation and freshwater dynamics of the region . Under climate change, the Arctic is warming at nearly four times the global average, yet current ocean models appear to under-represent this rapid evolution . Observational evidence shows that weakened stratification in the Eurasian Basin of the Arctic Ocean is driving structural changes in the water column – a process known as “Atlantification”. This shift enables warm Atlantic water to penetrate further north, accelerating sea ice melt and reducing winter sea ice formation . Historically, approximately half of the exported Arctic freshwater has been exported through the CAA and into Baffin Bay . In recent years, increased glacial melt from Greenland and the CAA has led to more frequent ice-free channels during summer (Canadian Ice Service; http://ice-glaces.ec.gc.ca/, last access: 26 May 2026), likely modifying freshwater exchange between the Arctic and Baffin Bay . Furthermore, Baffin Bay now receives increasing volumes of glacier meltwater from Greenland, a trend accelerated by the warming influence of Atlantic water .

The established pathways of water mass exchange between high latitudes and the subpolar North Atlantic may also be changing . Freshwater discharge (water with practical salinity below 34.6, ) from Davis Strait into the subpolar North Atlantic is projected to increase, potentially affecting deep convection in the Labrador Sea . Despite Davis Strait's role as a freshwater source to the subpolar North Atlantic , the freshwater dynamics of both Davis Strait and Baffin Bay remain poorly constrained. This is largely due to strong seasonal variability in freshwater release and the limited access during winter imposed by sea ice cover .

Additionally, the contribution of water outflowing Davis Strait on the water mass formation in the Labrador Sea remains uncertain. This is likely due to intense winter convection in the Labrador Sea, which strongly modifies temperature and salinity, complicating efforts to trace the origins of water using only these properties .

Furthermore, the role of Baffin Bay intermediate and deep water (e.g. Transition Water) in the formation of LSW remains poorly understood compared to the better-known origin and pathways of surface water . Also, recent studies have emphasised the importance of Arctic freshwater export to the subpolar North Atlantic and the role of cross-density lateral mixing in the Labrador Sea, contributing approximately 60 % to the annual formation of LSW . However, most observational studies continue to focus on boundary current systems, leaving off-boundary water exchanges understudied . Therefore, a better understanding of the off-shelf and intermediate circulation between Baffin Bay and the Labrador Sea is still needed , particularly regarding the origin and formation of Transition Water in Baffin Bay and the potential contributions of Arctic origin freshwater to the formation of LSW.

To advance on these complex water-mass interactions, we propose a novel approach in the region that combines the two long-lived artificial radionuclides ¹²⁹I (t1/2=15.7 Ma) and ²³⁶U (t1/2=23.4 Ma), with hydrographic properties such as salinity and temperature. We expect ¹²⁹I and ²³⁶U to suitably trace circulation patterns and mixing in the Davis Strait region because inflowing water presents contrasting tracer concentrations measured for the Pacific, Atlantic and Arctic oceans .

¹²⁹I and ²³⁶U originate primarily from nuclear sources and are assumed to be conservative in seawater . Their input into the ocean was dominated by a peak in ²³⁶U from the nuclear weapons test global fallout in the 1960s, and liquid releases from nuclear fuel reprocessing plants in Sellafield (UK) and La Hague (France) (Fig. A, black factory symbols), peaking in the 1980s for ²³⁶U and ramping up after 1990s for ¹²⁹I. Figure  in Appendix B represents their input function defined at 70° N (blue star in Fig. A), a combination of global fallout and liquid releases from the nuclear reprocessing plants . In both cases, the released radionuclides flow north from the North Sea and join the Norwegian Coastal Current (NCC), entering the Arctic Ocean. From the entrance of the Arctic Ocean, the radionuclides can recirculate within the Eurasian Basin in a short loop or a longer path (Canada Basin) before exiting the Arctic Ocean via Fram Strait or the CAA . From Fram Strait, the radionuclides are transported south and ultimately reach the Labrador Sea and Baffin Bay as part of the WGC . Up to date, there are no measurements of these two tracers in either Nares Strait or the CAA, so their transport through these routes is still unknown.

In the Arctic Ocean, the transport times and mixing of Atlantic Water can be calculated using different models that are all based on tracer input functions (Fig. ) . However, in sub-Arctic regions, the simultaneous mixing of multiple water masses with distinct input functions complicates the accurate estimation of water age and mixing when relying solely on these tracers. Nevertheless, the combination of ¹²⁹I and ²³⁶U has already proven to be a suitable tool to study the formation and origin of water masses in the subpolar North Atlantic, where it can be assumed that both tracers are in a steady state for each water mass .

In the study area, we expected to find a wide representation of the historical temporal variation of the tracers, allowing us to disentangle the origin of Atlantic water. While the “old” Arctic-Atlantic Water (AAW) from the Canada Basin would be characterised by high ²³⁶U and low ¹²⁹I , the “young” Polar Surface Water (PSW) recirculated in the Eurasian Basin would carry a higher signal of ¹²⁹I . Low tracer concentrations should be characteristic of the WGIW, which enters the study region from the subtropics without direct contact with the radionuclides . Also, Pacific Water (salinity of 32.5) entering Baffin Bay through the CAA would carry low concentrations of both radionuclides, as their only source was the atmospheric weapon tests . Finally, it is important to note that the freshest water (i.e. glacier melt, sea ice melt, and river run-off: ) should contain almost no tracer, hereinafter denoted as tracer-free water .

In this study, we address critical knowledge gaps in understanding water mass transformations and exchanges between Baffin Bay, Davis Strait and the Labrador Sea by applying a novel approach based on the two long-lived artificial radionuclides, ¹²⁹I and ²³⁶U. The tracers, which exhibit distinct distributions across Arctic and Atlantic origin water, offer a complementary tool to traditional hydrographic parameters such as temperature and salinity, enabling the identification of water mass sources in regions characterised by complex mixing and strong seasonal variability. Using observations collected between 2022 and 2024 across key Arctic gateways – including Davis Strait, Nares Strait, Lancaster Sound – and the Labrador Sea, we apply a binary mixing model to characterise tracer signatures of inflowing water masses and quantify their contributions to key intermediate and deep water masses. This approach allows us to explore the origin of the Transition Water, BBMW, BBBW and (cold) Arctic Water and to follow the evolution of WGSW and WGIW along their cyclonic journey through the Baffin Bay. It also enables us to assess the potential influence of Transition Water on water mass formation in the Labrador Sea.

Seawater samples were collected from multiple sites (Fig. B). In May 2022, along the AR7W Line across the Labrador Sea, seawater was collected from the surface and six depth profiles (depth range: surface–3560 m), including four surface samples from southwest of the AR7W Line, aboard R/V Atlantis as a part of the Atlantic Zone Off-Shelf Monitoring Program (AZOMP) by Bedford Institute of Oceanography (red triangles). In October 2022, additional sampling was conducted in the Davis Strait region aboard R/V Neil Armstrong as part of Davis Strait Observing System Program. This survey included a depth profile in central Baffin Bay (Fig. B, red square, depth range: surface–2377 m), and full transects along the Northern Line (9 stations, Fig. B red circles, depth range: surface–1665 m), Davis Strait (10 stations, red diamonds Fig. B, depth range: surface–1013 m) and Northern Labrador Sea Line (9 stations, Fig. B red stars, depth range: surfac–2614 m). Further sampling was conducted in September–October 2024 during the Refuge Arctic Cruise Legs 4 and Transforming Climate Action Leg 5, both onboard CCGS Amundsen, targeting Nares Strait and Lancaster Sound. In northern Nares Strait, two stations were sampled: one depth profile (50–345 m, RA28) and one surface station (RA34). Both are represented in Fig. B as one unfilled orange square. In the southern Nares Strait, three stations were sampled: RA44, RA48, RA50, of which only RA48 is a depth profile (70–600 m). These stations are represented in Fig. B as unfilled orange circles. In Lancaster Sound, two sites were sampled in the western part of the Archipelago (Keb/TCA-S3, depth range: 2–140 m, unfilled red triangles in Fig. B), and the eastern part close to Baffin Bay (TCA S3, depth range: 2–890 m, unfilled red diamond in Fig. B). All cruises used a CTD-Rosette equipped with 24–12 L Niskin bottles to collect samples for the purpose of this study.

Seawater samples for ¹²⁹I analysis were collected in 250 mL brown plastic bottles, and pre-rinsed three times with water from the Niskin bottle, before being filled. For ²³⁶U, 2–3 L of seawater were collected in Nalgene cubitainers, also pre-rinsed three times before filling them. All samples were stored onboard without any further processing, sent to ETH Zurich for the analysis of iodine and uranium isotopes.

The radiochemistry of ¹²⁹I and ²³⁶U was performed following and for a total of 307 samples. The samples were analysed using the TANDY AMS system at the Laboratory of Ion Beam Physics (LIP), ETH Zurich . The ¹²⁹I concentrations were calculated using the measured ¹²⁷I/¹²⁹I atom ratio and the known amount of spiked ¹²⁷I. The reproducibility of ¹²⁹I data was estimated based on repeated measurements of an internal seawater standard (n=14, average ¹²⁹I = 10.4 ± 0.5 × 10⁷ at kg⁻¹). Blanks for ¹²⁹I (n=16, 7.6 ± 1.5 × 10⁵ at) were obtained using deionised water treated following the same procedure as for seawater samples.

The samples for ²³⁶U were spiked with 1 pg of ²³³U following the radiochemistry described by and . The MILEA AMS system at LIP facility was used to measure ²³³U/²³⁸U and ²³⁶U/²³⁸U in the samples and the in-house standard “ZUTRI”. Correction procedures were applied as described by . The concentrations of ²³⁶U and ²³⁸U were calculated based on the known amount of ²³³U that was spiked in the sample. Each batch included one blank (n=11, ²³⁶U: 0.01 ± 0.01 × 106 at), consisting of deionised water that was treated like the samples.

The results of ¹²⁹I and ²³⁶U are reported and plotted in atoms per kg of seawater (at kg⁻¹). The combined uncertainty of chemical processing and measurements for both ¹²⁹I and ²³⁶U was below 6 %. All results and detailed analytical uncertainties are reported in the Zenodo database (10.5281/zenodo.16914587, ).

The water masses are summarised in Table , and were assigned according to previous water mass classifications in the literature and identified using conservative temperature (CT) and absolute salinity (SA, see T–S diagram in Fig.  in Appendix B). The conversion from practical to absolute salinity and from potential to conservative temperature was performed according to TEOS-10 . Temperature and salinity are also represented along sections in Fig.  to facilitate the comprehensive distribution of water masses. West Greenland Irminger Water (WGIW; CT > 3.5 °C, SA > 34.7) is the warmest and most saline water mass, confined to intermediate depths along the Greenland shelf. West Greenland Shelf Water (WGSW; CT < 5 °C, SA < 34.2) is fresher than WGIW and located at the surface along the Greenland shelf. Arctic Water (AW; CT: -0.8 to 1.1 °C, SA > 32.9) is the freshest water mass, confined to the Baffin Island and Canadian shelf. The CT and SA ranges used here differ significantly from those in ; we applied a narrower range for Arctic Water and distinguished it from cold Arctic Water (cold AW; CT < -0.8 °C, 32.5 < SA < 33.8), which is colder and more saline. Below the cold Arctic Water, Transition Water (CT < 1.8, SA: 34–34.6) is characteristic of mid-depth in Baffin Bay and Davis Strait. Within Transition Water, temperature increases with depth to a temperature maximum (CT: 1.4–1.8 °C, SA = 34.64 ± 0.03, ). In Baffin Bay, at depths of 600–1000 m, Baffin Bay Mode Water (BBMW; CT: 0.7–1.2 °C, SA = 34.6; ) was observed overlying Baffin Bay Bottom Water (BBBW; CT < 0.4 °C, SA = 34.6; , Fig. ). In the Labrador Sea, Labrador Sea Water (LSW; CT: 3.1–3.8 °C, SA > 35.0; ) resides at intermediate depth. Beneath LSW, North East Atlantic Deep Water (NEADW; CT: 2.0–3.3 °C, SA = 35.07 ± 0.2; ) overlies Denmark Strait Overflow Water (DSOW; CT < 1.3 °C, SA = 35.0 ± 0.1; ). In Nares Strait and Lancaster Sound, water masses were classified as Arctic Water CAA (AW_CAA; CT < 0 °C, SA < 33) and Arctic Atlantic Water CAA (AAW_CAA; CT > -1 °C, SA > 33), entering through the Canadian Arctic Archipelago. Samples that do not fall within the defined water mass categories are considered mixtures of the above water types. While all samples are presented in the results section, the discussion focuses on those assigned to specific water masses.

In the study area, several potential water masses are adjacent to those observed in Baffin Bay. These include tracer-free water, Pacific Water , WGIW , WGSW , Polar Surface Water from the Eurasian Basin , and Arctic-Atlantic Water originating from the Canada Basin . Further assessment of their formation processes will be based on the conservative tracer signatures of ¹²⁹I and ²³⁶U. However, the number of water masses considered exceeds the number of available conservative tracers. Therefore, the complexity of the model is reduced by modelling the mixing fractions of samples between the two most dominant water mass sources (hereinafter referred as endmembers), instead of employing an Optimum Multiparameter Analysis. Before quantifying the origin and estimating the mixing fractions, the water masses are classified according to their temperature and salinity. Then, a binary mixing model combining both ²³⁶U and ¹²⁹I is used, as previously described by and . The water masses in the study area are considered as conservative mixtures of different endmembers. The endmembers are source components of a mixture and set the boundaries of the mixing model. In this study, each endmember has a distinct and characteristic value for temperature, salinity, and ¹²⁹I and ²³⁶U concentrations (see details on each endmember in Sect. ).

In most cases, the ¹²⁹I and ²³⁶U concentrations of the endmember consist of an average concentration of several samples collected in this and previous studies, with an associated uncertainty that corresponds to the standard deviation of the average values (see Table ). Considering the two endmembers as the primary water-mass sources, the resulting water mass is treated as a mixture of the two. The mixing fraction f is then estimated as: 1f=‖a‖‖b‖ where b is the vector connecting the two endmembers (mixing line), and a is the vector connecting a given sample to one of the endmembers. The resulting value of f represents the percentage contribution of each endmember to a given watermass. This study focuses on different contributions of Atlantic-derived water, and not on the differentiation between different freshwater sources, which are more precisely addressed using nutrients and δ18O . A key limitation of this model is the assumption that mixing occurs exclusively between two endmembers. If a sample falls outside the defined mixing line, its composition is estimated by projecting orthogonally onto the mixing line and neglecting potential contributions from a third endmember. The model also assumes a steady state for transient tracers, which was previously discussed and justified by and . In the study region, changes in water mass formation and mixing ratios between endmembers are believed to have a greater influence on tracer concentrations than recent changes in the tracer input function (Fig. ), supporting the steady-state assumption. The resulting mixing fractions will only be discussed in Sect. (not in results) and summarised in Table .

Depth profiles collected in Nares Strait are represented in Fig. A and B as orange symbols. Both tracers show low concentrations at the surface Arctic Water (AW, ¹²⁹I: 60–85 × 107 at kg⁻¹, ²³⁶U: 11–14 × 106 at kg⁻¹) and a local maximum in the Arctic Atlantic Water at about 250 m (AAW, ¹²⁹I: 180–210 × 107 at kg⁻¹, ²³⁶U: 23 × 106 at kg⁻¹). Below that depth, ¹²⁹I seems to slightly decrease towards the bottom, while ²³⁶U remains more constant. Furthermore, the concentrations of both isotopes appear to be higher in the northern part of Nares Strait (squares) compared to the south (circles).

The two profiles in Lancaster Sound are represented in Fig. A and B as red symbols. Similar to Nares Strait, the highest concentrations of ¹²⁹I were associated with AAW_CB (Canada Basin) located at 200 m in eastern Lancaster Sound (¹²⁹I: 130 × 107 at kg⁻¹, SA: 33.85, CT: -0.10 °C) and at depths below 140 m in the western station (¹²⁹I: 120 × 107 at kg⁻¹, SA: 33.6, CT: -0.44 °C). In western Lancaster Sound, the highest concentration of ²³⁶U was observed at 10 m depth in Arctic Water (²³⁶U: 19×106 at kg⁻¹, SA: 30.58, CT: -0.04 °C), and the lowest at the surface (²³⁶U: 15 × 106 at kg⁻¹). A similar distribution was observed in the eastern Lancaster Sound, with the lowest concentration at the surface (²³⁶U: 13 × 106 at kg⁻¹), and a maximum at 200 m (²³⁶U: 22 × 106 at kg⁻¹), which then decreased slightly with increasing depth to 15 × 106 at kg⁻¹ at 886 m.

Along all transects (Fig. ), the highest ¹²⁹I concentrations were observed in West Greenland Shelf Water (WGSW), at the surface of the Greenland shelf (Fig. A–D). The concentrations of ¹²⁹I decreased along the pathway of the West Greenland Current (WGC), with WGSW dropping from 320 × 107 at kg⁻¹ at AR7W in the Labrador Sea (Fig. D) to 170 × 107 at kg⁻¹ at the Northern Line in Baffin Bay (Fig. A). In contrast, the warmest and most saline West Greenland Irminger Water (WGIW) generally showed increasing ¹²⁹I concentrations from 50 × 107 at kg⁻¹ at AR7W to a maximum of 90 × 107 at kg⁻¹ at the Northern Line. In Arctic Water, the freshest water along the Baffin Island Current (BIC) and Labrador Current (LC), ¹²⁹I concentrations increased slightly from central Baffin Bay (140 × 107 at kg⁻¹, blue squares Fig. C) to the Northern Labrador Sea Line (150 × 107 at kg⁻¹, Fig. C). But, Arctic Water showed lower concentrations (100 × 107 at kg⁻¹) at the AR7W Line in the southern Labrador Sea. Cold Arctic Water exhibited elevated ¹²⁹I concentrations in the 100–170 × 107 at kg⁻¹ range and showed no clear north-south trend. While Arctic Water, WGIW and WGSW were present in all sections, Transition Water was only present along the Northern Line, Davis Strait (Fig. A, B) and central Baffin Bay (Fig. C, blue squares), with concentrations in the range of 50–120×107 at kg⁻¹, respectively. BBMW was observed along the Northern Line, Davis Strait and central Baffin Bay only, with ¹²⁹I concentrations of 29–37 × 107 at kg⁻¹. ¹²⁹I concentrations at the levels of the blanks (<1.2±1.5×107 at kg⁻¹) were measured in BBBW present below 1600 m in central Baffin Bay and the Northern Line. Labrador Sea Water (LSW) and North East Atlantic Deep Water (NEADW) (only sampled at AR7W, Fig. D) covered an ¹²⁹I range of 30–70 × 107 at kg⁻¹ at intermediate depth in the Labrador Sea, while the near-bottom Denmark Strait Overflow Water (DSOW) presented concentrations up to 120 × 107 at kg⁻¹.

The distribution of ²³⁶U concentrations (Fig. E–H) differs from that of ¹²⁹I. The highest ²³⁶U was observed in the Transition Water (Fig. E, F), with up to 19 × 106 at kg⁻¹ in central Baffin Bay (Fig. D, blue squares), followed by cold Arctic Water, with ²³⁶U in the 14–17 × 106 at kg⁻¹ range between Davis Strait (Fig. F) and central Baffin Bay. While the ²³⁶U concentration in Arctic Water was generally in the range of 13–15 × 106 at kg⁻¹, the AR7W Line recorded the maximum value (16 × 106 at kg⁻¹, Fig. H). In water inflowing Davis Strait, such as WGSW, ²³⁶U concentrations decreased slightly towards the north, from 15 × 106 at kg⁻¹ at the AR7W Line to 13 × 106 at kg⁻¹ at the Northern Line (Fig. E). Similarly to ¹²⁹I, low ²³⁶U concentrations were measured in WGIW, with an increasing trend from 9 × 106 at kg⁻¹ at AR7W to 11 × 106 at kg⁻¹ on the Northern Line.

In the Labrador Sea, LSW presented ²³⁶U concentrations on the 9–11 × 106 at kg⁻¹ range on the AR7W and Northern Labrador Sea lines (Fig. G, H). NEADW, only sampled along AR7W, showed ²³⁶U concentrations slightly above LSW (10–12 × 106 at kg⁻¹). The sampling density of DSOW was smaller in the Northern Labrador Sea Line than for AR7W and confined at slightly lower concentrations (AR7W: 12–14 × 106 at kg⁻¹, Northern Labrador Sea Line: 12 × 106 at kg⁻¹). Moving towards the Baffin Bay, the BBMW had similar ²³⁶U concentrations of 10–12 × 106 at kg⁻¹. Finally, in BBBW occupying depths below 1600 m in central Baffin Bay (blue squares Fig. D) and the Northern Line, ²³⁶U concentrations reached the analytical limit of detection (1.2±1.5×106 at kg⁻¹).

Both tracers have concentrations well above global fallout, which would be about 2.2 × 107 at kg⁻¹ for ¹²⁹I and 5.4 × 106 at kg⁻¹ for ²³⁶U, according to recent estimates at the Bering Strait . Only the bottom water at Baffin Bay (BBBW) presents tracer concentrations lower than expected from the global fallout.

Figure provides an overview of the water masses observed in Baffin Bay, Davis Strait and the Labrador Sea and their distribution in T–S diagram (Fig. A) and ¹²⁹I–²³⁶U tracer space (Fig. B) while the sampling location is indicated by the symbols and can be referred to in Fig. B.

WGSW (dark-red symbols) stands out in both panels due to its relatively high temperature and low salinity (Fig. A), as well as its elevated ¹²⁹I concentrations (from 180 to >300 × 107 at kg⁻¹) (Fig. B). Its seasonal variability in temperature (Fig. A) is consistent with observations by and and clearly expressed in the broad temperature range with CT < 0 °C along the AR7W Line (dark red triangles) sampled in spring and CT > 2 °C at the other sections sampled in autumn. In the tracer space, WGSW forms a well-defined linear relationship, despite the seasonality. This indicates that the tracers are largely insensitive to seasonal changes. The saline and warm WGIW (Fig. A, light green symbols) is located at the lower left in tracer space (Fig. B, light green symbols) due to its low ¹²⁹I and ²³⁶U concentrations. Radionuclide tracers are especially low for samples collected in the Labrador Sea, then they increase slightly together with a freshening and cooling experienced on the northward flow of WGIW.

Arctic Water (Fig. , light brown symbols) is fresh and cold and shows tracer concentration slightly below cold Arctic Water (Fig. , teal symbols), which is characterised by its distinct temperature minimum and confined to the upper 100 m. Both water masses are positioned in the tracer space at the mixing interface between the relatively warm Transition Water and WGSW.

Within the Transition Water (light blue symbols), the samples located at the temperature maximum (TrW_Tmax, dark-blue symbols) exhibit the highest ²³⁶U concentrations. This TrW_Tmax lies between cold Arctic Water (teal symbols) and WGIW (light green symbols) in T–S space, consistent with its intermediate hydrographic character (Fig. A). In tracer space, however, the TrW_Tmax shows the highest ²³⁶U levels within the region (Fig. B), second only to the concentrations observed in Nares Strait (orange symbols) and Lancaster Sound (red symbols).

At depth below Transition Water and low temperatures is the BBMW (light purple squares and circles) confined at low ¹²⁹I and intermediate ²³⁶U concentrations, leading to the even deeper and colder BBBW (purple squares and circles). The BBBW in the Northern Line and central Baffin Bay is found in the bottom left corner of the tracer space and its concentrations of ¹²⁹I and ²³⁶U approach the analytical limit of detection.

In the Labrador Sea, surface water (light red symbols) fills the gap tracer between WGSW and WGIW and bridges WGSW and LSW (blue trangles) in the T–S diagram. LSW is colder than WGIW but warmer than the deeper NEADW (yellow triangles). In the tracer space, both LSW and NEADW are located close to WGIW but shifted towards slightly higher ²³⁶U concentrations.

Finally, DSOW (Fig. , green symbols), is characterised by low temperatures, high salinity, and intermediate tracer concentrations.

Figure A shows the geographical locations of the defined endmembers, each represented with distinct symbols and colours. Their corresponding ¹²⁹I and ²³⁶U tracer signatures are displayed in Fig. B. The endmembers are defined using a combination of published data and new tracer measurements. Water mass assignments are based on temperature–salinity characteristics, and for endmembers taken from the literature, the original classifications are retained. Table  summarises all endmembers used in this study, including their geographic locations, depth ranges, tracer concentrations, temperature and salinity values, the number of samples included, and the relevant references.

Based on previous studies, the initial mixing scenarios (Fig. B) are constructed using the following endmember pairs: AAW_CB–Pacific Water , Nares Strait outflow–Pacific Water and WGSW–NAC North Atlantic Current;.

West Greenland Shelf Water (WGSW) and West Greenland Irminger Water (WGIW) undergo strong seasonal hydrographic variability. They cool and become more saline in winter and freshen in summer, which complicates quantifying their role in water mass formation in Baffin Bay . In contrast, radionuclide tracers remain unaffected by these processes, making them a powerful tool for tracking WGSW and WGIW in their journey to northern latitudes. In our dataset, the largest hydrographic variations are observed in WGSW: samples from the AR7W Line show low temperatures in spring (overlapping dark red triangles in Fig. A), while autumn samples adjacent to Davis Strait display higher temperatures and a broad salinity range (all dark red symbols in Fig. A, with CT above 2 °C). Despite these seasonal changes, tracer concentrations (Fig. B) remain remarkably stable, enabling a robust quantification of the contribution of WGSW to water mass formation. In the model, mixing between PSW_EGC–NAC is considered in Fig. A for the formation and evolution of WGSW (dark red symbols) and WGIW (light green symbols).

Along the AR7W Line (dark red triangles in Fig. A) the WGSW retained up to 70 % of the PSW_EGC. As the WGSW flows northward along the Greenland shelf, it becomes progressively entrained by tracer-free water and WGIW (light green symbols in Fig. A). By the time WGSW reaches Davis Strait and the Northern Line (diamonds and circles in Fig. A), the PSW_EGC fraction has declined to approximately 40 %. This dilution could be compared to the decrease in the thickness of warm Polar Water (equivalent to WGSW) as they flow from eastern Davis Strait to eastern Northern Line, observed by .

At central Baffin Bay, water classified as Arctic Water (light brown square in Fig. A), preserved a significant fraction (about 30 %) of PSW_EGC compared to the endmember at Denmark Strait. This finding contrasts with and , who did not consider WGSW as a major contributor to the formation of water masses in the region. The substantial presence of WGSW may underscore the role of eddies, the off-branching character of the WGC, and the significant transformation of surface water in Baffin Bay.

On the contrary, WGIW is positioned closer to the NAC endmember than WGSW, reflecting a stronger influence from southern-sourced water and thus exhibits lower tracer concentrations. An increase in its tracer content from the Labrador Sea to Baffin Bay confirms its mixing with WGSW, preserving approximately 15 % PSW_EGC, consistent with the decrease in the thickness observed by , while contrasting , who did not observe WGIW on the Greenland shelf in Baffin Bay. Furthermore, observed WGIW in central Baffin Bay, which could not be identified here, probably due to the limitation of the mixing of two endmembers in the tracer analysis.

The formation and origin of Transition Water (TrW) remain subjects of ongoing debate. characterised Transition Water (their TrW1) as a mixture of several water masses, including WGIW, cold Arctic Water and WGSW. In contrast, described it as a blend of “Atlantic” water from the south and denser, colder northern water. Complementing these perspectives, nutrient analyses by and identified WGIW, Pacific Water, and Arctic-Atlantic Waters as the primary contributors of Transition Water formation.

The broad range of ¹²⁹I and ²³⁶U concentrations within Transition Water are consistent with the contribution from multiple endmembers . The temperature maximum within Transition Water (TrW_Tmax, dark blue symbol in Figs. and  A) clusters near the Lancaster Sound endmember (red cross in Fig. A), whereas the remaining Transition Water (light blue symbols in Figs. and A) exhibits a spread towards lower temperatures and higher ¹²⁹I, indicative of additional mixing processes. However, as mentioned above, the Lancaster Sound outflow is known to be highly variable and might shift along the Pacific–AAW_CB mixing line. Therefore, the contribution of Lancaster Sound outflow to waters in Baffin Bay (TrW_Tmax and BBMW) is not calculated using this geographical endmember but from the mixing line between Pacific–AAW_CB endmembers. Although the Lancaster Sound outflow might not be considered as a fixed endmember, its influence remains apparent in the composition of Transition Water and TrW_Tmax. TrW_Tmax reflects contributions of cold, ²³⁶U-rich AAW_CB mixed with Pacific Water from Lancaster Sound, while its elevated temperatures are likely caused by mixing with warm, ¹²⁹I- and ²³⁶U-poor WGIW within Baffin Bay. This results in a contribution of the AAW_CB of 40 %–60 % if calculated using Pacific water as the lower tracer endmember (i.e. this is a lower bound contribution, as TrW_Tmax also mix with WGIW). This finding aligns with observations by , but emphasises the outflow of Lancaster Sound as more important than contributions by the Nares Strait endmember. This finding differs from , who attributed the AAW component to Nares Strait outflow only. The Lancaster Sound outflow might be transported from shallow to intermediate depths by deep convection and cascading driven by intense air–sea fluxes and polynya activity (; ), being subsequently redistributed through lateral exchange between slope and basin waters, as modelled by . At intermediate depths, particularly along the 27.5 kg m⁻³ isopycnal , this dense water may encounter warm WGIW, where diffusive instability and cabbeling promote further mixing and downward propagation , forming TrW_Tmax.

Ultimately, Transition Water forms through the evolution of TrW_Tmax toward colder, fresher waters, which results in less ²³⁶U and higher ¹²⁹I due to mixing with WGSW (Figure 6A, dashed line connecting TrW_Tmax with WGSW). WGSW may entrain Transition Water through cascading along the Greenland Shelf . Additionally, Transition Water might also have contributions from the Nares Strait outflow , as observed by the increased concentrations ²³⁶U that bring some of the Transition Water to the Nares Strait endmember in Fig. A.

Despite the limitations of a two-endmember model to constrain the origin and mixing of water masses, the radionuclide tracers strongly suggest the presence of AAW_CB outflowing Lancaster Sound and subsequently entraining Transition Water, a process that could not be quantified when using nutrient-based tracers alone . Furthermore, radionuclide tracers show negligible contributions of WGSW to TrW_Tmax.

Moving to greater depths, the origin and formation of BBMW still remain unclear . BBMW (purple symbols in Fig. A) has been suggested to be older than other water masses in the region, with tracer ages estimated to be up to 90 years , which aligns well with a tracer signal of “old” water according to Fig. . The relatively high ²³⁶U and very low ¹²⁹I displayed by BBMW indicate the presence of AAW_CB and align with observations by and . However, based on the available tracer data, it is not possible to differentiate if the AAW_CB originated from Nares Strait, as indicated by and or from Lancaster Sound.

Finally, the deepest waters in Baffin Bay, also known as Baffin Bay Bottom Water (BBBW), remain tracer-free (Fig.B, dark purple squares). Low tracer concentrations are consistent with observations by , which show that convection or cascading do not contribute to BBBW formation, and is further consistent with its long ventilation time of up to 455 years based on tritium and ³He measurements . Other studies have posited that Atlantic Water from the Canada Basin may contribute to BBBW formation , but this interpretation is incompatible with high tracer concentrations in the Canada Basin , which would manifest as elevated ¹²⁹I and ²³⁶U concentrations in BBBW.

Arctic Water and cold Arctic Water are significant sources of freshwater to the subpolar North Atlantic, and they are strongly influenced by sea ice, glacial meltwater and Pacific water . However, the composition of these freshwaters as they mix with other water masses in the region remains uncertain. Here, ²³⁶U and ¹²⁹I demonstrate that cold Arctic Water is a mixture of Arctic-Atlantic Water outflowing mostly through Nares Strait, Pacific Water, and WGSW, with minor contributions from Lancaster Sound. Arctic Water has a similar origin, but with a stronger influence of tracer-free freshwater. Both Arctic Water and cold Arctic Water are presented in Fig. B as light brown and teal symbols, respectively. The new endmember for WGSW is calculated as the mean of the values shown in Fig. A, with the error bars representing the corresponding standard deviation. This endmember includes all samples with concentrations of ¹²⁹I below 240 × 107 at kg⁻¹, selected to best characterise the core of WGSW advecting north beyond Davis Strait. We consider mixing between Lancaster Sound (to best represent the surface and subsurface inflow) and WGSW ; Nares Strait outflow (NS_South) and Pacific Waters . Additionally, potential mixing between Lancaster Sound and NAC and WGSW with NAC is indicated as grey solid lines.

Cold Arctic Water generally has higher tracer concentrations than Arctic Water and is confined to a cold (CT ∼ -1.5 °C) and rather narrow temperature range, while it covers a salinity range between 32.5 and 33.6 (Figs.  and ). Although observed mixing between cold Arctic Water (their cold Polar Water), meteoric water, and WGIW, our mixing model (Fig. B) points to additional water masses involved in its formation. In the following, we examine each contributor in detail, starting with WGSW.

In the binary mixing model, cold Arctic Water stretches from the mixing line of Nares Strait South (NS_South) and tracer-free waters towards the WGSW endmember, suggesting the presence of WGSW in cold Arctic Water as well. The largest contribution of WGSW, approximately 80 %, is observed in a sample located at 54 m depth in the middle of the Northern Line (circle in Fig. B). WGSW generally follows the main cyclonic circulation , which may cool through air-sea heat exchange in winter (especially in the North Water Polynya) and cascade along the Greenland shelf , thus reaching the temperature of the cold Arctic Water . In central Baffin Bay (teal square in Fig. B), the Nares Strait outflow contributes up to 70 % to cold Arctic Water, in agreement with previous observations by and .

Finally, Pacific Water and NAC (Fig. B) with low ¹²⁹I and ²³⁶U, may contribute to cold Arctic Water as well. While focused primarily on contributions of WGIW, the radionuclide tracer data is less conclusive, as mixing with either source produces a similar low-tracer signal. At this stage, the binary mixing model reaches its limit, unable to distinguish between these two low-tracer endmembers. Finally, mixing with Arctic Water, as previously noted by , is likely but cannot be precisely quantified here due to similar ¹²⁹I and ²³⁶U signatures of cold Arctic Water and Arctic Water. Overall, the cold Arctic Water cluster reflects a varying blend of water masses described above, showing no such clear trend when looking at hydrographic properties only.

Following the assessment of the cold Arctic Water cluster, we turn to Arctic Water to explore its distinct properties and its close association with cold Arctic Water. Although Arctic Water is fresher than cold Arctic Water, both water masses plot closely together in tracer space (light brown and teal symbols in Fig. B). Arctic Water appears to be the result of similar sources as cold Arctic Water but with higher contributions of low-tracer water, probably of tracer-free and Pacific origin . In Arctic Water, contributions of WGSW are observed in central Baffin Bay (light brown square in Fig. B), highlighting the strong stratification in this region. On the western flank of the Northern Line, there is an influx of low tracer water (light brown circles in Fig. B), probably due to mixing with 50 % of Pacific Water (if mixing between Pacific and NS_South is considered). These mixed waters remain confined within the BIC and are transported southward. Stronger mixing with cold Arctic Water likely occurs near Davis Strait (diamond symbols), illustrating the dynamic conditions in this area . Further south in the Northern Labrador Sea Line (star symbols), Arctic Water was likely sampled within the bifurcation of the WGC, showing a WGSW contribution of up to 70 % (if mixing with Lancaster Sound outflow is considered). attributed a southward transport of water confined to the Greenland Coast to an instability of the WGC, which leads to shedding eddies in the northern Labrador Sea. In comparison, at the AR7W Line (triangles), Arctic Water carried by the Labrador Current transports a lower ¹²⁹I signal, probably originating from the Nares Strait outflow (up to 70 %, again if mixing NS_South with Pacific Water is considered). As the Labrador Current continues southward along the Canada Shelf, two surface samples (red triangles Fig. B) originating south west of the AR7W Line suggest further mixing between WGSW, Nares Strait and/or Lancaster Sound outflow and a low tracer component, probably of North Atlantic origin (i.e. NAC). The two surface samples are warmer (CT: 1.5 °C) and more saline (SA: 33.8) than Arctic Water, but colder and fresher than the remaining samples classified as Labrador Sea surface (Fig.). Possible influences on the main water masses observed in the Labrador Current might include: seasonal variability in southward velocities at Davis Strait, which are generally highest in summer and lowest in winter ; variability in the outflow of AAW_CB via Hudson Strait south of the Northern Labrador Sea Line ; the presence of eddies and Irminger Rings .

The formation of Labrador Sea Water (LSW) is driven by deep winter convection and associated physical processes , and is modulated by multiple freshwater sources (salinity < 34.6; ) that alter its properties . The relative contributions of these sources remain uncertain, with most studies focusing on freshwater inputs from boundary currents . In contrast, artificial radionuclides highlight the potential importance of off-boundary current sources in the formation of both LSW and North East Atlantic Deep Water (NEADW). Insights from ¹²⁹I and ²³⁶U tracers suggest that the properties of LSW and NEADW may be shaped by mixing between BBMW, Transition Water (TrW), and the North Atlantic Current (NAC).

In the previous study by using ¹²⁹I and ²³⁶U in the Labrador Sea region, the mixing model reached its limitations when trying to understand the origin of LSW, due to low tracer concentrations and the absence of critical endmembers. Nonetheless, that study showed that ¹²⁹I-rich WGSW eddies contribute to LSW as they branch off the WGC and move into the interior of the Labrador Sea . Labrador Sea Water (blue symbols in Fig. C) appears to have a significant contribution of WGSW, estimated to be up to 30 % (if mixing with the NAC endmember is considered). The influence of WGSW is even stronger, up to 70 %, on surface water of the eastern Labrador Sea (light red triangles in Fig. C).

Another source of ¹²⁹I and ²³⁶U in the LSW could be the Labrador Current, carrying Arctic Water and cold Arctic Water, but contributions to the LSW formation might be relatively small (< 10 %), as previously discussed in , , and . However, the main LSW cluster observed in Fig. C (blue triangles) is located at comparable low ¹²⁹I and elevated ²³⁶U concentrations. This suggests that another water mass, richer in ²³⁶U relative to ¹²⁹I, is required to account for the observed composition of LSW. This feature, first noted by and later confirmed by , can now be explained by the influence of the Transition Water outflowing Davis Strait and entraining into LSW.

Among the southward-flowing water entering the Labrador Sea, the Transition Water is the only water mass with sufficiently high ²³⁶U concentrations to explain the tracer signature of LSW. To our knowledge, this entrainment of the Transition Water outflowing Baffin Bay into LSW has not been previously considered in the literature. To estimate the contributions of Transition Water a new endmember is derived from the maximum temperature observed in Transition Water (TrW_Tmax, dark blue cross in Fig. C) with error bars reflecting the variability between samples. As shown by the blue triangles in Fig. C, the contribution of TrW_Tmax could be as high as 20 %. In Davis Strait, long-term moored measurements of temperature and salinity (Fig. A, B), and across-strait velocity (2004–2024) at 60° W and 500 m depth support this view (Fig. C). They show a persistent southward flow of Transition Water from Davis Strait into the Labrador Sea from 2004 to 2024, complementing gridded transports for the 2004–2010 period . These measurements coincide with the ²³⁶U maximum in Davis Strait (²³⁶U ∼ 17.5 × 106 at kg⁻¹, Fig. ), demonstrating that the transport towards the Labrador Sea of water fresher than LSW (SA < 34.9 g kg⁻¹) provides both a ²³⁶U and freshwater source for LSW. Once the Transition Water enters the Labrador Sea, it may be entrained with LSW through mixing along-density surfaces . Furthermore, similar concentrations of ¹²⁹I and ²³⁶U between BBMW and LSW (Fig. ) suggest that substantial mixing also occurs between these water masses. As a result, with an average SA < 34.49 g kg⁻¹, BBMW can also act as a further freshwater source for LSW. These previously unreported freshwater contributions from Transition Water and BBMW to LSW can help shed light on potentially unresolved physical processes in ocean models that contribute to deep convection variability in the Labrador Sea.

Finally, NEADW is a water mass that originates from multiple sources, including ISOW and LSW . In the mixing model (yellow triangles in Fig. C), NEADW occupies a low-tracer area influenced by several of these water sources, and known to be strongly entrained by ISOW (yellow-green diamond Fig. C, from ) while the contribution of DSOW (dark green triangles) is minor . However, the elevated ²³⁶U relative to ¹²⁹I is not explained by the endmembers established so far. Our results suggest that Transition Water and/or BBMW contribute up to 25 % to NEADW when mixing with NAC is considered. During autumn and winter, BBMW can enter the deep Labrador Sea as a weak overflow through Davis Strait . Once there, it joins the deeper layers and mixes with NEADW. Although the simple two-endmember model used here cannot fully resolve every contribution, particularly under low tracer concentrations and multiple sources, it still provides a valuable first-order approximation. Future work could build on this foundation with a more comprehensive multiparameter analysis.