Climate change in the Arctic, manifested in the increasing freshwater content over the past several decades (Prowse et al., 2015), affects the Atlantic overturning circulation with possible implications for global climate (Rahmstorf et al., 2015). Recent increases in Arctic freshwater flux (de Steur et al., 2013; Karpouzoglou et al., 2022; Karpouzoglou et al., 2023) are primarily attributed to precipitation, river runoff, and sea-ice melt (Rabe et al., 2011; Rabe et al., 2014; Haine et al., 2015), but also to Pacific water (PW; Alkire et al., 2017). Low-salinity PW entering the Arctic Ocean through Bering Strait is the second main source of fresh water to the Arctic Ocean (Carmack et al., 2016) and serves to increase primary production by supplying nutrients that are otherwise limited (Mills et al., 2018). After entering the Arctic Ocean through the Bering Strait, PW is spread along two major pathways: a Transpolar branch crossing the Arctic Ocean to Fram Strait, and an Alaskan branch that flows along the Beaufort Sea continental slope through the Canadian Arctic Archipelago (CAA) and eventually into Baffin Bay – Figure 1A (Jones, 2001; Rudels, 2012; Hu and Myers, 2013; Woodgate, 2013; Aksenov et al., 2016; Hu et al., 2019). As a result, PW is a primary source of fresh water for Baffin Bay (Münchow et al., 2007), and comprises up to 20% of the freshwater inventory in the upper 300 m in western Fram Strait (Dodd et al., 2012). Note that the orientation of the transpolar branch varies with atmospheric forcing associated with the Arctic Oscillation and Beaufort High variability (Wang and Danilov, 2022) and may be deflected to the southwest, advecting more PW towards the CAA and subsequently into Baffin Bay during some years ( Figure 1 ; Hu et al., 2019). Regardless of the path, the outflow of PW from the Arctic Ocean through the CAA and into Baffin Bay is poorly resolved due to insufficient data coverage, specifically over the heavily ice covered CAA continental slope and the northern Canadian Arctic.

Over the coastal domains of northern Greenland, Baffin Bay, and the CAA, there are additional contributions to the freshwater budget from glacial melt (Gardner et al., 2011; Castro de la Guardia et al., 2015; Bendtsen et al., 2017; Bamber et al., 2018) superimposed on the Arctic freshwater flux (Dukhovskoy et al., 2019). Meltwater discharge from the Greenland ice sheet has accelerated since the 1990s (Rignot et al., 2011; Hugonnet et al., 2021) and is predicted to cause fundamental changes to the hydrography, biogeochemical cycles, and marine productivity of the downstream regions (Hendry et al., 2019; Seifert et al., 2019; Kanna et al., 2022). However, PW complicates direct estimates of the glacier meltwater fraction and therefore the freshwater balance. The general contribution of PW to the freshwater inventory remains poorly understood, partly due to seasonal and interannual variability (Falck et al., 2005; Melling et al., 2008; Dodd et al., 2012; Woodgate et al., 2012; de Steur et al., 2013; Münchow, 2015; Shroyer and Pickart, 2019; Lin et al., 2021), but also because of the different pathways of PW in the Arctic Ocean.

It seems that PW outflow from the Arctic Ocean is impacted by intense interaction between the Canada Basin and shelf water of the Beaufort Sea (Dmitrenko et al., 2018) and the CAA (Melling et al., 1984). This interaction is largely controlled by wind-forced water dynamics over the continental slope. Key processes affecting the shelf-basin interaction in the western Arctic include wind-driven up- and down-welling (Carmack and Chapman, 2003; Pickart et al., 2009; Dmitrenko et al., 2016; Dmitrenko et al., 2018), enhanced vertical mixing over the sloping topography (Guthrie et al., 2013), and processes associated with instability of the shelf break current transporting PW eastward along the Beaufort Sea continental slope (Lin et al., 2016; Dmitrenko et al., 2018). However, the volume transport by this shelfbreak current has decreased by approximately 80% between 2002-2004 and 2008-2011 due to the weakening of summer westerly winds (Brugler et al., 2014), demonstrating the importance of wind forcing.

Several studies have focused on modification of PW through the interaction between the shelf break current and surrounding waters along the Alaskan continental slope (Spall et al., 2008; Schulze and Pickart, 2012; von Appen and Pickart, 2012; Lin et al., 2016; Foukal et al., 2019). In contrast to the Alaskan Beaufort Sea, the PW modification through the shelf-basin interaction in the Canadian Beaufort Sea has received less attention, and the shelf-basin interaction over the CAA continental slope is completely unknown. Ocean-glacier interactions also have the potential to modify PW through cooling and freshening (Münchow et al., 2016; Dmitrenko et al., 2017; Kirillov et al., 2017; Dmitrenko et al., 2019), though this process has not previously been considered for modifying PW.

Overall, studies suggest that climate change accelerates the Arctic and subarctic freshwater cycle (Carmack et al., 2016; Rodell et al., 2018), which highlights the need for a clear understanding of the variability of PW outflow from the Arctic Ocean. In particular, this knowledge is required to address one of the central questions posed by Carmack et al. (2016): “Will the melting of glacial ice (including Greenland) substantially alter the freshwater budget and flow regimes of the Arctic Ocean and surrounding subarctic seas?” This is of special importance for Baffin Bay where the PW outflow through the CAA and Nares Strait dominates the freshwater inventory over the surface water layer (Alkire et al., 2010). The increase in freshwater in Baffin Bay elevates the sea surface height due to steric effects, which impacts the Baffin Bay circulation (Grivault et al., 2017).

This paper is focused on the analysis of a unique set of conductivity–temperature–depth (CTD) and nitrate profiles collected in the northern CAA, specifically Expedition Fjord and Iceberg Bay around Axel Heiberg Island, from the landfast ice in April–May 2022 ( Figure 2 ). The field campaign was based out of the McGill Arctic Research Station (MARS) located near the Thompson Glacier outlet ( Figure 2 ). Prior to the field program in April-May 2022, there were no oceanographic data collected in Iceberg Bay. CTD observations in Expedition Fjord were limited to several profiles taken in August 1988 in the inner part of the fjord near the mouth of Expedition River (Gilbert, 1990). Here we build on results of Melling et al. (1984), who showed that the Atlantic-derived water layer in the western CAA is modified due to enhanced vertical mixing over the rough bottom topography of the CAA margins. Our main goal is to infer modification of water masses as they enter the CAA. To assess this goal, our observations are set within the context of upstream, downstream and local conditions, with a focus on modifications of Pacific- and Atlantic-derived waters of the Arctic Ocean imposed by vertical mixing and ocean-glacier interactions during their advection through the CAA. Note that within the context of upstream observation, we assess the modification of PW only relative to an Alaskan branch transporting PW along the Beaufort Sea continental slope ( Figure 1A ). The other transport route via the central Arctic Ocean is not included in our analysis due to deficiency of the long-term observations.

Expedition Fjord and Iceberg Bay represent a glacial inlet of Axel Heiberg Island, which opens to the Sverdrup Channel connecting the continental slope of the Arctic Ocean with the CAA interior ( Figure 1B ). Expedition Fjord and Iceberg Bay are covered by landfast sea ice from October to June with significant interannual variability in the coverage and duration of open water ( Figure 3 ). The landfast ice edge in Iceberg Bay roughly delineates the extension of pack ice inflowing to the CAA from the Arctic Ocean through Sverdrup and Peary Channels (Melling, 2022). The bottom topography of Expedition Fjord and Iceberg Bay is poorly known. The most comprehensive bathymetric data set in Expedition Fjord, reported by Gilbert (1990) and Aitken and Gilbert (1996), revealed a gradual sloping of the seafloor from the mouth of Expedition River to ~110 m depth at ~91.8°W longitude with a sill of ~70 m depth located at ~91.6°W. Further west of the entrance to Expedition Fjord, the seafloor slopes down to >300 m at ~92.3°W (Aitken and Gilbert, 1996). The seafloor of Iceberg Bay is largely unknown, though it seems to be impacted by Iceberg Glacier, which is a tidewater outlet glacier branching off from the Müller Ice Cap ( Figure 2 ) that is over 500 m thick in the interior (Priergaard Zinck and Grinsted, 2022). Calving of the Iceberg Glacier terminus generates numerous icebergs that are commonly observed in the interior of Iceberg Bay and in the outer part of Expedition Fjord to the 70-m depth sill at ~91.6°W ( Figure 2 ). Iceberg Bay opens to Sverdrup Channel, which is connected to the Arctic Ocean and generally covered by landfast ice through much of the year (Melling, 2002). There is a sill at ~450 m depth across the entrance from the Arctic Ocean to Sverdrup Channel (Melling et al., 1984), though this is deep enough to allow both Pacific-derived and Atlantic‐derived Arctic water with temperatures >0°C to flow into the CAA. In what follows, we examine the properties of the water column in Iceberg Bay and Expedition Fjord to trace the origins of the water and identify any local modifications.

Between 24 April and 1 May 2022, 47 CTD profiles were collected from the landfast sea ice (1.0 to 1.6 m thick) in Expedition Fjord and Iceberg Bay in the CAA ( Figure 2 ). The CTD observations were carried out with a Sea-Bird Scientific SBE-19plus CTD that was calibrated prior to the expedition and was accurate to ±0.005°C and ±0.0005 S m^–1. Throughout the manuscript, we used practical salinity calculated directly from the conductivity and temperature of seawater as defined by Practical Salinity Scale 1978 (PSS 78; Lewis, 1980). All CTD casts were taken down to the seafloor. The CTD was outfitted with a Sea-Bird Scientific SUNA V2 Optical Nitrate Sensor that measured nitrate based on the absorption characteristics of nitrate in the UV light spectrum. The nitrate profiling was conducted at 12 stations (#12, 14, 17, 22, 24, 27, 29, 31, 36, 39, 42, and 46; Figure 2 ), and was accurate to ± 2 mmol m^-3 (mmol m^-3 = millimoles of nitrate per cubic meter of water) or ± 10% of reading according to the manufacturer’s specifications. However, the SUNA nitrate data were not calibrated. In what follows, we limited our analysis based on nitrate data only to PW qualitative tracing.

Three sets of complementary CTD profiles are used to provide context for the water masses we observed in Expedition Fjord and Iceberg Bay. (i) CTD profiles collected in the eastern Beaufort Sea between 134°W and 135°W from June to October from 2002 to 2011 by ArcticNet ( Figure 1A ) were averaged and reveal the mean summer water profile in the eastern Beaufort Sea (Dmitrenko et al., 2017), which we consider the “upstream” area. (ii) CTD profiles collected by Ice Tethered Profilers (Toole et al., 2011; Ice Tethered Profilers, 2022) off the CAA continental slope and in the Canada Basin are used to represent the ambient water masses. ITPs are autonomous profilers that are deployed on perennial sea ice in the polar oceans to measure changes in upper ocean temperature and salinity (Krishfield et al., 2006). In the eastern Canada Basin, we use CTD profiles from ITP #104 (profile #2926) and #105 (profile #2467) taken on 8 March and 18 February 2019, respectively ( Figure 1A ; Toole and Krishfield, 2016). Over the lower continental slope of the CAA (depth >2000 m), ITPs #98 and #103 sampled the water column on 14 October 2016 (profile #29) and 2 April 2019 (profile #3366), respectively ( Figure 1A ). These are the only ITPs providing CTD data along the lower continental slope of the CAA in proximity to Peary and Sverdrup Channels ( Figure 1A ). To our knowledge, there are no oceanographic observations conducted over the CAA continental slope of the Sverdrup Islands. All ITP-derived CTDs were measured from 12 m to 760 m depth. We used level 2 real time ITP data with salinity accurate to 0.04. (iii) Finally, we adopted 3 CTD profiles (#CAA6, CAA9, and WC01) taken during the CCGS Amundsen GEOTRACES expedition in September-October 2015 (Hughes et al., 2018; Colombo et al., 2021) north of Parry Channel around Cornwallis Island in the central CAA, which we consider the “downstream” area ( Figure 1A ).

Ice conditions over Iceberg Bay and Expedition Fjord during summer were characterized by optical imagery from Sentinel-2 ( Figures 2 , 3 ), which has 10-meter spatial resolution and was acquired through Sentinel Hub (https://apps.sentinel-hub.com; Gascon et al., 2017). In general, open water, sea-ice, large icebergs, and glacier termini are easily distinguishable in cloud-free Sentinel-2 imagery. Sea-ice thickness was measured manually at each CTD station with an ice thickness tape with accuracy ±1 cm. The seasonal mean (July to October; open water season) fields of sea level atmospheric pressure (SLP) were retrieved from the ERA5 atmospheric reanalysis (Copernicus Climate Change Service, 2017; Hersbach et al., 2020) for 2020, 2021, and from 2001 to 2021 using Web-Based Reanalysis Intercomparison Tools (WRIT; Smith et al., 2014; NOAA Physical Sciences Laboratory, 2022). The horizontal resolution of ERA5 is 31 km.

Finally, ice-penetrating radar data acquired over Iceberg Glacier in 2014 and 2017 by the Multichannel Coherent Depth Sounder as part of NASA’s Operation Ice Bridge (Paden et al., 2010; IceBridge MCoRDS L2 Ice Thickness, 2023) were used to characterize the thickness of Iceberg Glacier ( Figure 4 ). Data were elevation-corrected so that the surface matched simultaneous measurements from the Airborne Topographic Mapper (Studinger, 2013). Plotting and tracing in Figure 4B were done using ImpDAR (Lilien et al., 2020). Glacier height above flotation ( Figure 4C ) was estimated using the bed reflector. For this calculation, we assumed zero firn air, similar to other glaciers in the area (Müller, 1962), and a constant radar wave speed in ice of 1.69 10⁸ m s^–1. The bed reflector was converted to a height above flotation assuming that the density of ice is 917 kg m^–3 and the density of ocean water is 1028 kg m^–3. Where the height above flotation is at or below 0, the glacier is floating or only lightly grounded.

The water column at Iceberg Bay and Expedition Fjord shows four distinct layers, which are consistent across the entire study area ( Figure 5 ). The four layers, from top to bottom are described below:

(i) 0–27 m: The layer immediately below the sea-ice down to ~27 m depth comprises the subsurface halocline conditioned by snow, sea-ice and glacier meltwater inflow during summer ( Figure 5B ). The subsurface halocline is characterized by a relatively strong vertical salinity gradient that on average increases by 0.22 m^–1, but peaks at 0.8 m^–1 immediately below the ice. This layer shows two intermediate temperature maxima at ~3 m and 15–20 m depth with temperature increasing from –1.45°С to –0.41°С ( Figures 5A , 6 ). The intermediate temperature maximum at ~3 m depth is recorded mainly in the inner part of Expedition Fjord with the magnitude of the temperature maximum gradually decreasing towards the mouth of Expedition Fjord ( Figure 6 ). This maximum is not traced in Iceberg Bay ( Figure 6 ). The deeper temperature maximum at ~15–20 m depth is present over the entire study area; however, the highest temperature of the intermediate maximum (–0.41°С) was recorded at station #2 in the shallowest (~40 m depth) inner part of Expedition Fjord ( Figures 2 , 6 ). The deeper temperature maximum gradually diminishes towards the outer part of Iceberg Bay and the Iceberg Glacier terminus ( Figure 6 ).

(ii) 27–180 m: A layer with a weaker vertical salinity gradient (salinity 0.016 m^–1) and temperatures increasing from –1.65°С to –1.25°С underlays the subsurface halocline and extends to a depth of ~180 m ( Figures 5A, B ). We refer to this layer as the Pacific-modified polar water following its similarity to the Pacific-derived water layer over the adjoining Canada Basin (e.g., Steele et al., 2004; Shimada et al., 2005). This layer is further split into two sub-layers around approximately 110 m depth by an increase in the intermediate maximum of nitrates (~18-19 mmol m^-3; light blue vs. dark blue shading in Figure 5C ), which indicates the presence of Pacific winter water (Jones and Anderson, 1986; Alkire et al., 2019). However, the temperature of the Pacific-modified polar water in Iceberg Bay and Expedition Fjord does not show the patterns typical for PW in the Arctic Ocean. The Arctic Ocean Pacific summer water with temperatures above −1.2°C for salinities between 31 and 32 overlays a deeper layer of Pacific winter water with salinities between 32 and 33 with temperature as cold as −1.45°C (Steele et al., 2004; Shimada et al., 2005; Timmermans et al., 2017). In contrast to the western Arctic Ocean, in Expedition Fjord and Iceberg Bay the shallower portion of the Pacific-modified polar water (associated with Pacific-derived summer water) has a mean temperature of ~ −1.4°C, while the deeper portion of this layer (related to the Pacific-modified winter water) exhibits an increase in temperature with depth from ~ −1.2°C to −0.5°C ( Figures 5A and 6 ). Away from the Iceberg Glacier terminus and the inner part of Iceberg Bay, which was packed with icebergs, water temperature shows a gradual increase with depth from –1.5°С at 27 m depth to –0.5°С at 180 m depth indicating the influence of warm Atlantic water (AW). In the inner part of Iceberg Bay, and close to the Iceberg Glacier terminus, the temperature of Pacific-modified water shows intrusions of cooler water with negative temperature anomalies (relative to the ambient CTD profiles) from 0.05°C to ~0.3°C ( Figures 5A and 6 ). In the σ₀ plane, these intrusions are isopycnical meaning they are entirely driven by temperature anomalies with no deviations in salinity ( Figure 7 ). The majority of the intrusions were recorded within the salinity range from 31.7 to 33.5 ( Figure 7 ) from ~30 m down to ~130–140 m depth.

(iii) 180–240 m: Below the Pacific-modified polar water (>180 m depth), temperatures increased steadily through the Atlantic-modified polar water with salinity from 34 to 34.5 ( Figures 5A, B ). The temperatures eventually exceeded 0°C (the commonly accepted upper boundary of the Atlantic layer) at ~240 m depth, indicating the presence of AW ( Figure 5A ).

(iv) 240 m to bottom: AW occupies depths exceeding 240 m. The thermohaline characteristics of the AW layer are spatially uniform over the entire study area ( Figures 5A, B ). The temperature increases with depth from 0°C at ~240 m to 0.4°C at ~495 m and 478 m depth (stations #43 and #45, respectively; Figures 2 , 5A ). The salinity of the AW layer ranges from 34.4 to 34.78 ( Figures 5B , 7 ). The thermohaline stratification of a deeper portion of the AW layer from 400 m to 495 m depth is weak, with vertical temperature and salinity gradients of 0.0002°C m^–1 and 0.0002 m^–1, respectively ( Figures 5A, B ).

Finally, stations #1, 4, and 11 show an anomalous increase in temperature and salinity by 0.02°C to 0.05°C and 0.2 to 0.4, respectively, through the layer 5-20 m above the seafloor ( Figures 5A, B , and 6 ). These stations are located in the inner part of Expedition Fjord (#1 and #4) and at the Iceberg Glacier terminus (station #11, Figure 2 ).

Below we discuss the potential origin and modifications of the water masses identified in Iceberg Bay and Expedition Fjord. We also use our findings to trace PW flow through the northwestern CAA to Parry Channel, putting our results into the context of upstream observations in the Canada Basin and the eastern Beaufort Sea and downstream observations in the CAA.

A portion of the freshwater transport through the CAA consists of low‐salinity Pacific‐derived Arctic water flowing southeastward to Baffin Bay and onwards to the Labrador Sea. However, the processes that influence and modify this transport are not well understood. Here, a unique set of conductivity–temperature–depth (CTD) and nitrate profiles collected in April–May 2022 in Expedition Fjord and Iceberg Bay at Axel Heiberg Island, are used to identify the properties and spreading pathways of the Pacific-derived Arctic water in the northwestern CAA. The CTD profiles are examined within the context of upstream observations in the Arctic Ocean and downstream observations in the central CAA and reveal the origin of water masses and their interactions with ambient water from the continental slope and the nearby tidewater glacier outlet. CTD casts collected in Iceberg Bay and Expedition Fjord show that the subsurface water (25–180 m depth) is associated with the PW outflow from the Arctic Ocean. The underlying halocline separates the PW from a deeper layer of polar water that has interacted with warm (>0°C) AW that is detected below 240 m depth.

Notably, the water masses in Expedition Fjord and Iceberg Bay are significantly modified compared to the adjoining Arctic Ocean, as suggested from:

(i) Subsurface temperature maxima at ~3 m depth (salinity ~29.7) and ~17 m depth (salinity ~31) are attributed to solar radiative heating during summer 2021 and 2020, respectively. The preservation of these temperature features to spring 2022 clearly indicates weak water circulation and a lack of vertical mixing in Iceberg Bay and Expedition Fjord assuming that the revealed modification of PW and AW occurs upstream to this area. These findings are critically important for interpreting modifications of Pacific-derived and AW while they are “en route” from the Arctic Ocean to Baffin Bay;

(ii) The upper boundary of PW is traced at ~25 m depth in the study area compared to ~70 m depth in the adjoining Arctic Ocean. This indicates less freshwater content of the surface layer in the fjord system where the salinity of the 5-m thick surface layer is <30. We speculate that the anticyclonic atmospheric forcing over the Canada Basin favors surface water outflow to the Arctic Ocean, and Pacific and Atlantic water inflow through the CAA;

(iii) There is no signature of Pacific summer water; from ~30 m to 90 m depth, the mean water temperature is –1.4°C compared to –0.85°С over the lower continental slope of the CAA. At the front of the tidewater glacier outlet and near the surrounding icebergs, cold water intrusions of ~0.25°C are present through the Pacific water layer down to a depth of ~140 m, which corresponds to the estimated draft of the glaciers terminus;

(iv) The temperature of the deeper PW layer with salinity between 33 and 34 exceeds the temperature of Pacific winter water in the adjoining Arctic Ocean by ~0.5°C. The salinity of the PW layer exceeds that in the adjoining Arctic Ocean by 0.3–0.5. In contrast, AW is ~0.4°C cooler compared to the adjoining Arctic Ocean;

(v) CTD data suggests there is a geothermal heat flux at several spots in Expedition Fjord and near the terminus of Iceberg Glacier in Iceberg Bay. We speculate that the hypothetical extension of the geothermal springs to the Iceberg Glacier subglacial valley can generate sub-glacial discharge impacting sea-ice breakup along the glacier terminus.

Overall, our results indicate that Pacific and Atlantic waters in the CAA are modified due to enhanced vertical diffusivity over a rough bottom topography of the upper continental slope, which is consistent with upstream СTD observations in the Beaufort Sea continental slope, eastern Canada Basin and over the lower continental slope of the CAA, and preceding results by Melling et al. (1984). Taking into account numerous sills and narrow straits through the CAA, this leads us to conclude that tracing the initial thermohaline signature of the Pacific and Atlantic water flow through the CAA seems to be hardly possible. Both water masses are strongly modified due to disruptions imposed by interactions with a rough bottom topography of the CAA. This is consistent with results of downstream CTD observations and numerical simulations by Hughes et al. (2017).

Dedicated to the memory of our colleague Prof. David G. Barber, who inspired this research but passed away suddenly on 15 April 2022.

The original contributions presented in the study are included in the article/ Supplementary material . Further inquiries can be directed to the corresponding author.

ID, SK, and BR guided the overall research problem and developed methodology. ID conceptualized this research. ID, SK, NG, and DL conducted formal analysis and data curation. ID, SK, and NG performed the investigation. DD-J allocated resources. ID, BR, and DL wrote the original draft. ID, BR, NG, DL, JE, and DB revised and edited the original draft. ID, SK, and DL generated figures. ID and DD-J supervised and administrated this project. DD-J and DB accomplished the funding acquisition. All authors contributed to the article and approved the submitted version.