Isotopic signatures of snow, sea ice, and surface seawater in the central Arctic Ocean during the MOSAiC expedition

The Arctic is experiencing warming that is four times the global rate, with cascading consequences affecting major biogeochemical cycles (Vihma et al., 2016; Arctic Monitoring and Assessment Programme, 2017; Pörtner et al., 2022). For instance, the Arctic water cycle, encompassing precipitation, evaporation, and water transport through rivers, oceans, and sea ice, is undergoing notable changes which may, in turn, impact global climate and ocean circulation (Polyakov et al., 2018). In the past 20 years, the Arctic’s surface air temperature has risen by approximately 1.7°C compared to the long-term average (Overland et al., 2019), having distinct effects on the cryosphere, especially in major reductions in sea ice (Parkinson and DiGirolamo, 2021), and variations in precipitation patterns (Rieke et al., 2023). In particular, the loss of sea ice is critical, as sea ice regulates the energy exchange between the ocean and the atmosphere (Serreze et al., 2006; Stroeve et al., 2012; Sun et al., 2022). A reduced sea-ice cover leads to increased evaporation which, in turn, fuels moisture transport into and out of the Arctic (Klein et al., 2015; Mellat et al., 2021). These sea-ice changes also amplify atmospheric blocking patterns over Europe, resulting in extreme snowfall across Scandinavia and Western Europe (Bailey et al., 2021). The consequences of cryospheric changes in the Arctic water cycle are far-reaching and have global implications, given that the melting of sea ice has the potential to modify ocean currents and weather patterns, ultimately influencing climate conditions in middle and lower latitudes (Puntsag et al., 2016; Screen et al., 2018). Essential to improving our understanding of the causes and consequences of the Arctic changes is to collect direct year-round observations and to use these data to benchmark climate models that can then be used to forecast future Arctic scenarios, such as one that is ice-free in the summer (Lindsay and Schweiger, 2015).

The lack of observations in the central Arctic Ocean limits our understanding of the Arctic water cycle and hampers the projection of its future changes. The Multidisciplinary drifting Observatory for the Study of Arctic Climate (MOSAiC) expedition aimed to improve models of the Arctic climate system and fill knowledge gaps by gathering extensive field data (Nicolaus et al., 2022; Rabe et al., 2022; Shupe et al., 2022). This year-long expedition started in September 2019 with the German research icebreaker Polarstern (Knust, 2017) frozen into the sea ice north of the Laptev Sea (Krumpen et al., 2020). The ship and surrounding observatory then drifted with the ice within the Transpolar Drift (TPD) for the following 12 months. This expedition provided the opportunity to collect an extensive number of discrete water samples from multiple hydrological components to learn about the exchange processes between the ocean, cryosphere, ecosystems, and atmosphere around the Polarstern during an entire hydrological year, including the transition from a frozen ocean state to open seawater conditions (Nicolaus et al., 2022).

Stable water isotopologues, H₂O¹⁶, H₂O¹⁸, and HD¹⁶O, hereafter referred to as water isotopes, are excellent proxies for past climate changes and the modern water cycle due to their temperature-dependent fractionation, which occurs during water phase transitions (Dansgaard, 1964; Rozanski et al., 1992; Gat, 1996; Galewsky et al., 2016; Bowen et al., 2019). Isotopic fractionation is a term describing the preferential partitioning of isotopes during physical and chemical processes (Dansgaard, 1964). Equilibrium fractionation occurs when a system achieves a state of isotopic equilibrium, wherein isotopic exchanges occur with the surrounding environment. This equilibrium fractionation process not only is influenced by the specific conditions of the system but also exhibits a pronounced temperature dependence. Non-equilibrium fractionation, on the other hand, occurs when the isotopic composition of water molecules changes due to kinetic processes (i.e., one-way reactions), such as evaporation, condensation, or other phase transitions. These processes can cause a separation between the isotopes of hydrogen and oxygen, leading to a deviation from the expected equilibrium relationship. Essentially, this deviation provides information about the non-equilibrium conditions that exist during the evaporation process, where the slower diffusion of heavy isotopes results in the enrichment of deuterium in the evaporated water vapor. These non-equilibrium conditions are influenced by factors such as humidity gradients and wind advection, which prevent isotopic equilibrium from being reached between the liquid and gas phases (Pfahl and Sodemann, 2014).

The measurement of δ¹⁸O and δ²H involves determining the ratio of heavy to light isotopes relative from a standard, typically Vienna Standard Mean Ocean Water (V-SMOW standard). The results are expressed in delta (δ) notation, which quantifies the proportion of the isotope ratio of the sample and the standard in per mil (‰). The Global Meteoric Water Line (GMWL) is a linear relationship between the isotopic composition of precipitation and temperature on a global scale (Gat, 1996). To quantify the extent of deviation from the GMWL, a second-order parameter called deuterium excess (d-excess = δ² H – 8 * δ¹⁸O; Craig, 1961) is used.

Isotopes are highly sensitive tracers of moisture exchanges between the atmosphere, ocean, and cryosphere, with each component exhibiting a unique isotopic composition (Jouzel, 2003; Klein and Welker, 2016). The isotopic composition of water is affected by origin and transport of the moisture which results in altitudinal and latitudinal gradients (Johnsen et al., 1989; Dutton et al., 2005; Welker, 2012; Terzer-Wassmuth et al., 2021) that reflect the origin and transport of moisture from various sources, including oceanic and continental regions (Hoffmann et al., 2000; Steen-Larsen et al., 2014; Akers et al., 2020; Mellat et al., 2021). The isotopic composition of water is affected by evaporation, condensation due to rainout processes during transport from oceans to continents (Winnick et al., 2014), and by the formation and melting of ice (Eicken, 1998; Granskog et al., 2017; Tian et al., 2018). Therefore, water isotopes provide an integrated tracer of moisture exchange between phases, conditions, and across different components of the water cycle (Aron et al., 2023).

The isotopic values of winter precipitation are typically depleted in δ¹⁸O and δ²H values in contrast to summer precipitation (Welker, 2000; Bailey et al., 2019), primarily as a result of temperature-dependent isotope fractionation (Dansgaard, 1964). During winter, the air is colder, and the amount of moisture in the atmosphere is lower. Additionally, winter storms tend to originate from more northerly latitudes (Puntsag et al., 2016; Bailey et al., 2019; Bailey et al., 2021; Brunello et al., 2023), where the moisture sources have already undergone considerable fractionation due to lower temperatures. Hence, the effect of seasonality on the isotopic composition of precipitation should be most pronounced at the surface, where snow is added. Furthermore, frost flowers, icy crystals resulting from water vapor recrystallization, form on nascent ice (Style and Worster, 2009). Composed of delicate needle-like crystals, these structures extend several centimeters from the ice surface. Notably, the isotopic composition of frost flowers offers a compelling measure due to their emergence process, wherein they provide a means to track the impact of vapor isotopes. This tracking ability is because frost flowers form when warm air encounters colder ice. In Arctic regions, they emerge on newly frozen leads during warmer intervals, allowing freezing from brine channels (Barber et al., 2014). Consequently, these frost flowers possess higher sea salt concentrations than surrounding ice, snow, and air (Douglas et al., 2012).

The isotopic composition of Arctic Ocean water is influenced by sea-ice formation and sea-ice and snow melting, precipitation, and meteoric freshwater input from rivers and glaciers (Laukert et al., 2017; Paffrath et al., 2021; Solomon et al., 2021; Wefing et al., 2022). The Arctic Ocean is characterized by a more depleted water in heavy isotopes than compared to other ocean regions (Östlund and Hut, 1984) due to its high latitude and the westerly circulation system over Eurasia. Melting of sea ice and snow during summer causes the δ¹⁸O and δ²H values of the surrounding water to decrease due to the addition of freshwater depleted in δ¹⁸O (Lehmann and Siegenthaler, 1991; Csank et al., 2019). Furthermore, the Arctic Ocean receives freshwater from major rivers like the Ob, Yenisei, and Lena. These rivers have lower salinity and lower δ¹⁸O and δ²H values, signifying a substantial riverine influx and thus underscoring the significant influence of freshwater on the seawater geochemistry of the Arctic Ocean, particularly at the surface (Cooper et al., 2005; Bauch et al., 2011a; Dubinina et al., 2017).

The Arctic hydrological system experiences significant influence from sea ice, a critical and dynamic element within the water cycle, which also influences processes of ocean-atmosphere exchange (Eicken, 1998; Deser and Teng, 2008; Toyota et al., 2013; Chemke et al., 2021; Smith et al., 2022). Within this context, stable water isotopes offer valuable insights into freezing mechanisms, where the degree of kinetic isotopic fractionation is directly linked to the isotopic composition of surface water and the freezing rate, as demonstrated by Jouzel et al. (1999). To illustrate, regions in closer proximity to continental freshwater sources, exemplified by the Siberian rivers in the Kara and Laptev Sea areas, foster the formation of relatively isotopically light, ¹⁸O-depleted sea ice. Consequently, this phenomenon contributes to the formation of sea ice characterized by lower δ¹⁸O values. This specific type of sea ice has the potential to be transported across the central Arctic through mechanisms like the TPD, as evidenced by the findings of Granskog et al. (2017). In this manner, the interplay of water isotopes and freezing dynamics emerges as a crucial factor in the intricate processes of the Arctic hydrological system.

Further, water balance calculations suggest that besides sea ice, meltwater from the snow cover and direct precipitation can contribute significantly to the evolution of melt ponds and leads during the sea-ice melt season (Webster et al., 2015; Nomura et al., 2023). While the contribution of melt ponds and leads as local sources of water vapor remains a topic of debate (Tian et al., 2018), clarifying their roles is important. Leads are narrow and often linear cracks in the sea ice, generated by wind and ocean currents (Wang et al., 2016). During a lead formation, the seawater underneath the ice cover is suddenly exposed to the surface. In winter, leads typically refreeze in a few hours depending on the air temperature (Perovich and Richter-Menge, 1994). Melt ponds, on the other hand, are reservoirs of water formed from surface melt of snow and sea ice (Webster et al., 2022). These ponds form on the sea ice during the spring and summer seasons, corresponding to rising ambient air temperatures (Hanisco et al., 2007). Therefore, quantifying the isotopic composition of sea ice, its snow cover, and the processes affecting its formation and melting is crucial for assessing the complex interactions between individual components of the Arctic hydrological system.

This article aims to explore the isotopic (δ¹⁸O, δ²H, d-excess) composition of various components in the Arctic Ocean system during the MOSAiC campaign, with a focus on snow, sea ice, and seawater. To better understand water sources and mixing processes for the individual components of the Arctic water cycle, we use the overall statistics of the isotope dataset as well as the spatial-temporal changes of each hydrological component during the MOSAiC expedition. Finally, we characterize the key factors influencing the isotopic composition and the interaction between different hydrological components throughout the year. By doing so, we aim to provide a background dataset of the isotopic signatures of different water components in the central Arctic Ocean and their spatio-temporal changes for different types of sea ice, snow, and seawater. Our goal is to learn which key mechanisms control the fractionation of isotopes during sea-ice formation and melting, what are the contributions of different water sources to the Arctic Ocean’s isotopic composition, and how changes in these sources affect the Arctic water cycle over time.

The MOSAiC expedition took place from October 2019 to September 2020. During this time, the Polarstern drifted across the central Arctic Ocean, frozen in the sea ice (Nicolaus et al., 2022; Rabe et al., 2022; Shupe et al., 2022). The MOSAiC expedition was divided into 5 phases, so-called legs (Figure 1). On September 20, 2019, Polarstern left Tromsø and reached its destination on the selected ice floe, north of the Laptev Sea (85°N, 136°E), where the ship was anchored to an ice floe on October 4, and later became frozen into the sea ice. Until May 2020, the selected ice floe remained mostly intact while moving with the TPD across the central Arctic toward Fram Strait (during Legs 1–3; Figure 1). From mid-May to mid-June 2020, Polarstern temporarily left the original ice floe, transited to Svalbard and back, and reached the edge of the sea ice in Fram Strait on July 31 (78.9°N, 2°E). During Leg 3, Polarstern was carried across the Arctic faster than anticipated. Hence, during Leg 5, the ship returned north, to a position near the North Pole (87.7°N, 104°E), where additional measurements were carried out in the central Arctic during the summer (Krumpen et al., 2021). Throughout the entire campaign, observational activities were performed onboard in the proximity of Polarstern, in a Central Observatory ice camp and across a Distributed Network of autonomous observing systems. Moreover, in-situ observations were complemented by coordinated satellite and aircraft observations. The master track of Polarstern during the MOSAiC expedition was extracted from the PANGAEA repository (Haas, 2020; Kanzow, 2020; Rex, 2020, 2021a, 2021b).

Discrete samples of the different hydrological components (e.g., snow, sea ice, seawater, leads, melt ponds, Figure 2) were collected during all legs of the MOSAiC expedition. More than 2200 samples were analyzed for their isotopic (δ¹⁸O, δ²H) composition. However, certain limitations in the sampling process during this extensive campaign must be acknowledged. The collection of precipitation samples posed significant challenges due to consistently high wind speeds, which made recognizing snowfall events and collecting samples difficult. Additionally, the diverse teams and individuals involved in collecting seawater, sea ice, and snow samples worked across various locations on the ice floe, introducing complexities in comparing samples and maintaining uniformity in collection methods. Interruptions in sampling occurred periodically due to safety concerns or adverse conditions on the ice floe, which affected the regularity of data collection at certain locations.

Samples of snow deposited on the sea ice were collected by the snow pit team (Nicolaus et al., 2022) during Legs 1–5 (N = 306) from different locations and sites (i.e., Snow2-A2 and Snow2-A8) in the Central Observatory. Each sample consisted of 3–6 cm of snow collected throughout the vertical snow profile. Snow was typically sampled at 3 different depths (hereafter referred to as bottom, middle, and surface samples). The bottom layer was defined as the snow laying directly above the ice surface, while the surface layer was defined as the surface snow in contact with the air. The middle layer was collected approximately at half the height of the snow profile. The depths of the individual samples in the snow profiles were variable, as the thickness of the snow cover varied spatially and temporally. Snow samples were collected by shovel and were kept frozen in sampling bags until they arrived in the laboratory, where they were thawed at room temperature before analysis. In addition to snow, 10 frost flower samples were collected opportunistically during Legs 1–3, transferred to gas-tight Tedlar^TM bags, melted, and stored in narrow neck bottles at 4°C before analysis.

Sea-ice cores were collected from different sites in the Central Observatory. Level sea ice on the floe was categorized into three types based on age: sea ice that grew during the same winter, referred to as first-year ice (FYI); sea ice that had survived one summer melting period, referred to as second-year ice (SYI); and the new sea ice growing at the bottom of the SYI layer from the freezing of the seawater, referred to as insulated first-year ice (iFYI; Angelopoulos et al., 2022). Some samples were also collected from false bottoms and sea-ice ridges and are presented elsewhere (Smith et al., 2022; Salganik et al., 2023).

Sea-ice cores were collected using a Kovacs Mark II 9 cm diameter corer. The core was extracted and placed in a cradle equipped with a metric ruler. Using a standard Kovacs ice thickness gauge, the freeboard was taken and the length of the core was measured (Angelopoulos et al., 2022). The snow on top of the sea ice was brushed off the top of the cores to minimize snow affecting the ice surface. Due to the harsh winter conditions, cutting and sampling the cores in the field was not feasible. Instead, the cores were transported horizontally to a shipboard laboratory where the temperature was maintained at –20°C. Onboard Polarstern, the cores were cut into 10 (Legs 1–3) and 5 (Leg 4) cm sections within the next 30 minutes to 48 hours in the cold room by hand (Leg 1) or using an electric band saw (Legs 2–4). Each section was transferred into a gas-tight Tedlar^TM bag. The closed bags were degassed carefully with a vacuum pump (KNF Neuberger, type N035). Melting occurred within 12–15 hours at 4°C. After shaking the melted ice within the Tedlar^TM bags, discrete sampling started by first drawing the melt-water upward through a Tygon tube connected with the opened valves of the gas-tight bags and then delivering it into prepared brown narrow-neck 50 ml glass vials. The sea-ice dataset consists of 1043 samples, with 339 from Leg 1, 128 from Leg 2, 328 from Leg 3, 209 from Leg 4, and 39 from Leg 5.

On a daily basis, seawater samples were obtained from beneath the keel of the Polarstern via a valve system connected to a pump situated at a maximum depth of 10 m below sea level. To avoid contamination, the inlet was kept open for 2 minutes to rinse the pipe thoroughly before the water sample was filled into a 50 ml glass screw cap, sealed, and stored at 4°C before being transferred to the laboratory for isotope analysis. Salinity was not measured directly at the time of collection. Instead, salinity values were obtained at 1 h resolution from the ship-based conductivity-temperature profiling through depth (CTD) sensors (Tippenhauer et al., 2021); daily averages were calculated and attributed to the seawater isotopic measurements. In this study, salinity is reported on the practical salinity scale. The resulting “underway seawater” dataset consists of 302 continuous observations from November 2019 to October 2020 (Mellat et al., 2022b).

Additional seawater samples were obtained during Legs 1–3 of the expedition at the site called Ocean City (OC) (Rabe et al., 2022), from here on referred to as the “OC seawater” dataset. Seawater profiles were collected from the ice floe once a week at variable depths. A 12-bottle CTD rosette was used for water sampling from the sea surface to the ocean bottom, and water from the Niskin bottles at each water depth was sampled. This study presents the results from 43 surface samples (from 2–5 m depth). Storage and transport were identical to the “underway seawater” samples.

Two types of lead samples were collected and are presented in this work. During Legs 1–3, ice samples were obtained from refrozen leads using an ice saw or ice corer, while during Legs 4 and 5, the surface layer of water from open leads surrounding Polarstern was collected. In total, 137 samples were taken from leads, from which 80 were lead ice and 57 were lead water. In addition, water samples from melt ponds surrounding Polarstern were collected. Our dataset includes 109 samples collected from melt ponds, during Legs 4 and 5.

All discrete water samples presented here from the MOSAiC expedition were stored at 4°C before being analyzed for stable water isotopes at the ISOLAB facility of the AWI in Potsdam, Germany. The isotope composition (δ¹⁸O and δ²H) was determined using a Finnigan MAT Delta-S mass spectrometer fitted with two equilibrium units following a procedure described in Meyer et al. (2000). The δ¹⁸O and δ²H values are expressed as parts per mil deviation from the V-SMOW standard. For estimating the accuracy of the measurements, a quality control standard (HDW) was used, yielding a mean and standard deviation (SD) of –12.6‰ ± 0.1‰ (N = 54) and of –95.3‰ ± 0.6‰ for δ¹⁸O and δ²H, respectively.

The statistical summary of the isotopic composition of all discrete samples from different components of the Arctic water cycle is presented in Table 1 and Figure 3. The dataset exhibits a range of δ¹⁸O values from –39.6‰ for snow to 1.4‰ for iFYI. Meanwhile, the d-excess values vary between a minimum of –26.6‰ and a maximum of 26.5‰, both for snow. A more detailed analysis of the isotope measurements for snow, sea ice, seawater, and the interactions between different components is presented in the following subsections.

Snow samples were collected from different locations on the MOSAiC ice floe to analyze the variability of δ¹⁸O, δ²H, and d-excess values. In general, snow had the most δ¹⁸O-depleted isotope signatures of all hydrological components, with mean δ¹⁸O and δ²H values of –16.3‰ and –123.5‰, respectively. Moreover, snow also had the widest range of δ¹⁸O values (–0.6‰ to –39.6‰, N = 306; Table 1; Figure 3a). Frost flowers were the second-most δ¹⁸O-depleted hydrological component, with a mean δ¹⁸O of –9.55‰ and the highest SD (9.4‰), similar to the SD of snow (9.0‰). Snow and frost flower samples were characterized by high mean d-excess values of 7.3‰ and 6.6‰, respectively. However, the most negative d-excess values were also observed among snow samples (–26.6‰; Figure 3b).

On average, the δ¹⁸O values in the snow displayed an increasing trend from the top layer (–22.5‰ ± 6.7‰; N = 98) to the middle layer (–17.3‰ ± 8.2‰; N = 73) to the bottom layer (–9.7‰ ± 6.1‰; N = 92; as shown in Table 2). This trend corresponds to a 5.2‰ increase in δ¹⁸O between the surface and middle layers and another 7.6‰ offset between the middle and bottom layers. The δ¹⁸O-δ²H linear regressions exhibited distinct slopes and intercepts for the top (δ²H = 7.9 * δ¹⁸O + 11.1), middle (δ²H = 7.4 * δ¹⁸O − 1.7), and bottom layers (δ²H = 7.0 * δ¹⁸O − 11.0; Table 2). The slope of the top layer (7.89) was slightly lower than the GMWL slope (8.00), while the slopes of the middle (7.4) and bottom layers (7.0) were considerably lower than the GMWL slope. Hence, the change of the isotopic composition with depth is accompanied by a decrease in slope and intercept from top to bottom. This decrease indicates that the isotopic composition of snow in the study area deviates from the GMWL and that this relationship depends on the height of the snow profile, indicating that there are secondary processes (i.e., sublimation, vapor diffusion within the snowpack) underway in the snowpack (Ala-aho, 2021).

Opposite to the δ¹⁸O signal, the d-excess values decreased from the top to the bottom layer of snow (Figure 4). Comparing the mean values and SDs, we observed that snow from the top layer had the highest mean d-excess value (13.4‰) and the lowest SD (5.4‰). In contrast, snow at the bottom had the lowest mean d-excess value (−1.7‰) and the highest SD (8.9‰), indicating a higher variability in the values with increasing depth. Snow in the middle fell between these two, with an intermediate mean d-excess value of 8.3‰ and SD of 6.9‰.

Given the physical proximity between snow and sea ice, the results of the isotope composition of the surface sea-ice samples were analyzed together with the snow (Figure 4). The top 5–10 cm of sea ice exhibited a relatively constant mean δ¹⁸O of −4.4‰ ± 3.5‰ (N = 93) throughout the sampling period with a low SD (1.9‰), indicating relatively consistent values across this sample set. Hence, the surface layer of sea ice was, on average, 5.4‰ more enriched in ¹⁸O than the bottom layer of snow. Compared to FYI and ocean water samples, however, the surface layer of sea ice had a lower δ¹⁸O value (Table 1).

Figure 5 presents a time series spanning October 2019 to July 2020, depicting weekly averaged air temperatures and surface snow δ¹⁸O values, which suggests a relationship between these variables. The cyclic patterns in the upper and middle snow layers showcase lower δ¹⁸O values (−40‰ to −10‰) in January and February 2020 (Leg 2) and higher values (−27‰ to 5‰) from October to December 2019 (Leg 1). Interestingly, the lowest δ¹⁸O values (−25‰ to −15‰) in the bottom snow layer correspond to the June–August 2020 period (Leg 4), contrasting with the highest δ¹⁸O values aligning with Leg 2. These findings underscore the role of temperature as a primary driver of seasonal isotopic shifts, a concept initially introduced by Dansgaard (1964).

The Arctic Ocean receives most of its precipitation from the Atlantic and Pacific Oceans, as well as from ice-free areas within the Arctic Ocean itself. Throughout each season, a combination of factors including temperature, humidity, and air mass trajectories influence the isotopic characteristics of snow, supporting previous findings (Mellat et al., 2021). While the overall trajectory of mean δ¹⁸O in the upper snow surface samples mirrors air temperature fluctuations during winter, a closer examination reveals intricate dynamics. However, a direct and statistically significant correlation between the daily isotopic composition of surface snow samples and concurrent daily air temperatures does not exist (Figure S1). The correlation coefficient between these variables is notably low, indicating a complex interplay (slope = 0.04, R² = 0.02). This divergence is particularly evident in April 2020, as highlighted in Figure 5. However, the relationship between air temperature and surface snow δ¹⁸O values shows a subtle yet noticeable upward trend from September 2019 to April 2020 (slope = 0.47, R² = 0.13, p-value = 0.04).

More detailed time series of snow isotope profiles sampled at two sites (Snow2-A2 and Snow2-A8; Figure S2) are provided in Figure 6. They exhibit similar patterns in both δ¹⁸O and d-excess values across the three layers of snow (Figure 6a, b). The changes in δ¹⁸O observed in the top layer of snow at both stations could be linked to the changes in snow depth (up to 38 cm), as the snow became increasingly depleted in δ¹⁸O with each increase in depth. Notably, d-excess values lower than −15‰ were observed primarily at the snow2-A2 site (Figure 6a), where a negative trend from −9.3‰ to −26.6‰ is obvious between January 10, 2020, and April 15, 2020 (also visible in Figure 4). This continuous change in the d-excess values from Leg 1 until the end of Leg 4 was most pronounced in the bottom layer of the snow that is in contact with the sea ice, due to vapor transport during snow metamorphism (Sturm et al., 1997).

The mean δ¹⁸O values showed a gradual increase of approximately 13‰ from the surface down to the bottom of snow profiles. Correspondingly, there was a 12‰ decrease in d-excess as one moved toward the base of the snow profiles. This pattern becomes evident when contrasting the δ¹⁸O and d-excess values of the surface snow with those found in the middle and bottom layers. This consistency in findings supports the conventional composition of the surface snow samples, where fresh snowfall typically exhibits lower δ¹⁸O and higher d-excess values. This alignment makes sense considering that the introduction of new snow through precipitation significantly influences the overall isotopic composition of the snow cover (Ala-aho et al., 2021). The factors affecting isotopic compositions in snow are well-captured in the work by Stichler et al. (1981), which underscores the influence of temperature, air moisture, and air mass origin on the precipitation isotopes. These variables collectively give rise to the observed seasonal variation in heavy isotope concentrations in precipitation, with higher values during summer and lower values during winter. The high d-excess values at the snow surface support this hypothesis. The correlation of δ¹⁸O changes in top layer snow with snow depth changes (Figure 5) also suggests that these isotopic changes were caused by added snow at the surface, for example, by precipitation or wind drift, and is supported by findings of Zuhr et al. (2023). However, the weak to non-existing correlation between surface temperatures, precipitation amounts, and the isotopic values of the top snow layers (Figure 4) indicate that the contribution of meteoric waters (e.g., fresh snowfall) was either very limited or that snow undergoes metamorphism after deposition. Such isotopic changes of the top snow layer by postdepositional sublimation have recently been suggested by Wahl et al. (2021) and cannot be excluded from our samples. To quantify the impact of fresh snowfall on the isotopic composition of the snowpack and the influence or absence of post-depositional sublimation effects, a higher resolution event-based snow sampling is required. Unfortunately, such sampling was not performed during the MOSAiC campaign.

The process of vapor sublimation and vertical diffusion can take place between the relatively warmer sea-ice surface and the comparatively colder overlying snow layer (Jafari et al., 2020). As a consequence of this sublimation, water vapor originating from the sea ice tends to possess a higher δ¹⁸O in comparison to the snow cover above it (refer to Table 1 and Figure 3). This phenomenon has the potential to lead to an enrichment of δ¹⁸O within the snow through metamorphism. The subsequently enriched isotopic signature of vapor can then be transported upward within the snowpack along the thermal gradient, facilitated by additional vapor advection as explained by Sommerfeld et al. (1991) and Stichler et al. (1981) and upward movement within the snow layer as described by Friedman et al. (1991). Vertical vapor transport helps to explain the consistent isotope gradient observable in snow profiles.

The Arctic region’s snowpack undergoes substantial temperature variations with time, particularly notable during winter, due to the marked differences between air and ocean temperatures. This climatic characteristic leads to the development of recrystallization processes and the formation of columnar snow structures, known as depth hoar (Pinzer et al., 2012). These features were noted frequently during the MOSAiC expedition (Nicolaus et al., 2022). A reasonable assumption is that the vertical isotopic changes within the sea ice’s snow cover are closely interconnected with these observed recrystallization effects. During the summer months, vapor sublimation from the sea ice is suppressed due to the isothermal temperature.

Compared to the top snow layer, the middle and bottom layers of the snow profiles show higher δ¹⁸O and lower d-excess values, which may be linked with postdepositional snow metamorphic processes. These observations during MOSAiC are similar to other snowpack isotope studies in the Arctic, with variation from the surface to the bottom layers (Ala-aho, 2021). For instance, in Alaskan tundra snowpack, profiles of δ¹⁸O values for the most recent storm were recognizable; however, postdepositional processes modified the remaining isotope profiles from near the top to the bottom of the snowpack. In comparison, in taiga snowpack δ¹⁸O profiles in the eastern Arctic, where there is significantly less wind-driven redistribution than in the open Alaskan tundra, layering and preservation of divergent moisture sources and isotopic memory were much more apparent. Snow metamorphism involves both sublimation and recrystallization, which can cause changes in the isotopic composition of snow, leading to the further enrichment of heavy isotopes in the remaining snow (Friedman et al., 1991; Ebner et al., 2017; Wahl et al., 2022). Enrichment by sublimation in snow isotopes has also been reported by Steen-Larsen et al. (2014) and Casado et al. (2021), albeit in different environmental settings.

At the bottom boundary of the snow layer, the snow was more depleted in δ¹⁸O values than the underlying sea-ice surface. However, the surface layer of sea ice itself was also more δ¹⁸O-depleted than between FYI and ocean water samples. These isotopic differences together suggest that an additional isotopic exchange between the bottom layer of snow and the sea-ice surface layer occurred and that this exchange altered both the bottom snow and sea-ice surface layer samples.

Besides sublimation and diffusion processes, changes in the vertical isotopic composition of the snow layer might be caused by the partial melting of snow, especially when temperatures are close to the melting point. When the snow melts, the lighter isotopes of oxygen are more likely to diffuse into the vapor phase, leading to the remaining snow becoming slightly enriched in its δ¹⁸O values (Beria et al., 2018; Aemisegger et al., 2022). However, in the context of MOSAiC, the possibility of melting, percolation, and refreezing of snow was unlikely due to continuously negative temperatures during Legs 1–3. This interpretation was additionally supported by the absence of ice layers in the individual snow profiles, which would be indicative of melting and refreezing. However, during Leg 4, a partial melting of snow occurred, which could be responsible for the more uniform isotope composition between the different layers of snow due to percolation of meltwater through the snowpack.

In summary, the isotopic gradients observed in snow profiles from Legs 1–3 of the MOSAiC expedition result from the complex interplay of various processes. These processes differ for each profile and depend on factors like air temperature, wind, and sea-ice type beneath the snowpack. Notably, the dominant influences on the lower snowpack in the central Arctic are the mixing of material from the sea-ice boundary and moisture transfer from the ocean. These effects overprint interactions at the snow-atmosphere interface, such as precipitation and wind-driven snow accumulation. The findings of this study align with a similar dataset and analysis (AR Macfarlane, unpublished data), which delves deeper into snow and sea-ice metamorphism processes during MOSAiC.

The different types of sea ice that were measured had distinctly different δ¹⁸O values (Table 1). FYI had a mean δ¹⁸O value of −0.7‰ and was isotopically more enriched compared to SYI (δ¹⁸O = −4.6‰) which, surviving one melt season, had the lowest δ¹⁸O among all sea-ice samples. The mean δ¹⁸O value of 0.2‰ for iFYI was the highest, whereas lead ice δ¹⁸O averaged −1.18‰. Although, the mean d-excess values of FYI and iFYI were similar at 0.5‰ and 0.4‰, respectively, SYI had a slightly higher d-excess of 1.6‰.

The main coring sites (MCS) were established on first-year ice (MCS-FYI) and second-year ice (MCS-SYI), where coring activities were conducted from late October 2019 until the end of Leg 3 in May 2020 (Angelopoulos et al., 2022). To gain a better understanding of the temporal sea-ice development, Figure 7 presents the combined data of salinity and δ¹⁸O at the MCS-FYI and MCS-SYI sites. These data allow a more detailed consideration of sea-ice isotope variability. At the MCS-FYI site, the δ¹⁸O values exhibited a broad range from −6.7‰ to 1.6‰, with a mean value of −0.2‰ ± 1.4‰. At the MCS-SYI, the δ¹⁸O values fluctuated between −8.0‰ and 1.4‰, with a mean value of −2.8‰ ± 2.3‰. Salinity at MCS-FYI varied widely, ranging from 2.1 to 14.9, with a mean value of 5.2 ± 1.4, whereas MCS-SYI showed a smaller range of variation, with salinity levels between 0 and 7.5, and a mean value of 2.1 ± 2.0 (Figure 7). Generally, the MCS-FYI δ¹⁸O values were distributed mainly around 0‰. However, near the surface, that is, above 20–30 cm, distinctly lower δ¹⁸O values to about −4‰ were observed. These lower values were accompanied by a slight increase in salinity (Figure 7). MCS-SYI δ¹⁸O values were also lowest near the surface but showed an increase in δ¹⁸O toward greater depth, which was in line with a rise in salinity. Below about 70 cm depth (iFYI, Figure 8a, b), some of the MCS-SYI cores tend to show an isotope-salinity pattern similar to MCS-FYI. Generally, the MCS-SYI δ¹⁸O and salinity values were much lower to a greater depth. Sea-ice d-excess values were distributed mainly around 0.0‰ with some positive and negative deviations, the latter especially for SYI samples.

The growth of sea ice in the Arctic Ocean is a complex process (Notz and Worster, 2008) that is influenced by various environmental factors, such as temperature, salinity, wind, and ocean currents. Valuable insights into the formation and evolution of sea ice can be gained by analyzing its isotopic composition (Toyota et al., 2013; Granskog et al., 2017; Tian et al., 2018). As sea ice forms, it extrudes salt, which is then released into the seawater (so-called brine release). The process of freezing is accompanied by isotope fractionation processes also due to diffusion at the ice-water interface (Lehmann and Siegenthaler, 1991), favoring heavier isotopes to remain in the sea ice, whereas lighter isotopes are released to the seawater (Beck and Münnich, 1988). This fractionation results in the depletion of the surface seawater in δ¹⁸O (Table 1), indicative of the rate of sea-ice growth as well as of brine release into the ocean (Toyota et al., 2013).

Compared to SYI, FYI was more enriched in heavy isotopes, attributable to multiple factors. One primary distinction is that FYI originated from the freezing of ocean water with a relatively consistent isotopic composition around 0‰ during the MOSAiC expedition (Table 1). In contrast, SYI was formed in the Laptev Sea or its vicinity, particularly near the Laptev Sea shelf break (Krumpen et al., 2020). In this region, surface water salinity and δ¹⁸O values are notably lower due to freshwater inputs from surrounding snow-dominated watersheds (Bauch et al., 2009; Bauch et al., 2016; Juhls et al., 2020). Further, SYI has undergone a freeze-thaw cycle, involving the melting of surface snow and subsequent contribution of meltwater into the remaining SYI. This process leads to both lower salinity and lower δ¹⁸O.

The isotopic composition of SYI could be influenced by periods of partial melting, further altering its isotopic composition. Additionally, heavy isotopes may be enriched in the pond water due to evaporation from melt ponds. During winter, a newly formed ice layer known as iFYI develops beneath SYI. We follow the terminology presented in Angelopoulos et al. (2022), which slightly deviates from basic sea-ice definitions (World Meteorological Organization, 2014) but facilitates the explanation of our time series. iFYI generally exhibits higher δ¹⁸O values compared to the overlying SYI, influenced by similar factors that are responsible for the heavier isotope composition of FYI compared to SYI. However, iFYI has a slower growth rate than FYI, resulting in less pronounced kinetic fractionation during the freezing process. While SYI might be exposed to the atmosphere and exchange water vapor, the isotopic composition of underlying seawater primarily influences iFYI. Therefore, the isotopic composition of sea ice provides useful information about its formation and can help distinguish between ocean-sourced FYI and iFYI from SYI with a larger meteoric contribution. Our analysis of the isotopic composition of MCS-FYI and MCS-SYI in the Arctic Ocean is supported by earlier findings of studies by Granskog et al. (2017) and Lange et al. (2021). The authors observed that FYI had a higher isotopic composition compared to SYI (Figure 8a), concluding on differences in the sources of water for ice formation. Furthermore, our data reveal that the isotopic composition of both SYI and FYI had slightly lower δ¹⁸O toward the surface (Figure 8a), which suggests a snow contribution in the upper approximate 20 cm of the FYI cores. For the SYI, the top 70 cm displayed a lighter isotope composition, also reflected in low salinities at MCS-SYI throughout the expedition (Figure 8b). At the MCS-SYI site, the iFYI layer had grown at the bottom with slow brine release toward the bottom of the ice (Angelopoulos et al., 2022). The changes in our δ¹⁸O data aligned with the changes in salinity (or brine volume fraction; Angelopoulos et al., 2022) with a threshold of about −2‰ as an indicator for separating the iFYI from the upper SYI layer (Figure 8b).

OC seawater and underway seawater samples had similar mean δ¹⁸O values of −1.7‰ and −1.5‰, respectively, while lead water samples had a mean δ¹⁸O of −2.3‰. These values were associated with mean d-excess values of 2.0‰, 1.2‰, and 1.1‰, respectively (Table 1). The variability of δ¹⁸O in the underway seawater samples collected during all five legs of the expedition (−3.4‰ to 0.4‰) was slightly higher than that of the OC seawater samples (−2.7‰ to −0.4‰) collected only during Legs 1–3. This difference was mainly due to the difference in the sampling periods, as the underway seawater dataset consisted of samples from Legs 4 and 5 with generally more depleted in δ¹⁸O. The largest isotopic range was found for lead water samples (δ¹⁸O of −7.8‰ to 1.3‰; d-excess of −7.7‰ to 5.8‰).

Near-surface seawater samples collected daily from November 2019 to October 2020 for measuring the isotopic signature of the seawater along the Polarstern track were combined with continuous measurements of seawater salinity (Schmithüsen, 2021a). We found that, generally, the changes in seawater δ¹⁸O were closely mimicked by changes in seawater salinity (Figure 9b).

At the beginning of the expedition in the Amundsen Basin at about 85°N, surface seawater had a δ¹⁸O composition of around −1.5‰ and a salinity of 32.2. As the Polarstern drifted toward Fram Strait, seawater became increasingly depleted in δ¹⁸O and less saline until mid-February (at 87°N, 91°W), when the lowest δ¹⁸O value (−2.4‰) and salinity (31.7) were observed during Legs 1–3 (Figure 9b). After this minimum, δ¹⁸O and salinity values increased to a maximum of −0.4‰ and 34.3, at the end of Leg 3 (May 16, 2020). The weekly sampled OC seawater δ¹⁸O values show an almost identical pattern compared to the more highly resolved time series for underway seawater δ¹⁸O.

As Polarstern left the ice floe (Central Observatory) to Svalbard on May 17 for a 1-month break until June 17, δ¹⁸O values during transit to Svalbard increased to an absolute maximum of 0.4‰, together with a maximum surface salinity of 33.9. At the start of Leg 4, Polarstern navigated back to the original ice floe and drifted south toward Fram Strait. During this time, δ¹⁸O values became progressively lower, down to −2.2‰ at the end of Leg 4 (July 31; Figure 9b). Similarly, the seawater salinity during this period decreased to 31.3 (Figure 9a, b).

At the end of July, the ice floe broke apart in Fram Strait (79°N, 2.5°W) and Polarstern transited northward until it reached a location close to the North Pole (87°N, 104°E) in the Amundsen Basin, where the final part of the expedition, Leg 5, took place. Here, seawater δ¹⁸O decreased from −1.1‰ to −3.0‰ during the first 10 days and after that ranged between −3.0‰ and −3.3‰ until the end of the campaign. Likewise, seawater salinity decreased from 32.1 to 29.2 in the first 10 days and then varied around 29. A more detailed study, focusing on ocean column properties such as temperature, salinity, and density, was conducted during MOSAiC by Schulz et al. (2023).

The relationship between surface seawater δ¹⁸O and mean daily salinity correlated significantly (R² = 0.80, p-value < 0.0001; Figure 9) throughout the entire MOSAiC period, with low δ¹⁸O values associated with lower salinities. Except for Leg 2, the regression lines between δ¹⁸O and salinity had similar slopes (0.7 to 0.9) but variable intercepts (−33‰ to −12‰; Figure 9). Throughout MOSAiC, seawater δ¹⁸O values of −1.2‰ to −2.4‰ were observed consistently during all 5 legs, while the salinities varied between 30 and 34 (Figure 9b), contributing to different δ¹⁸O-salinity regressions. These seasonally and geographically changing relationships suggest that seawater isotopes were influenced by more than two sources in the Arctic Ocean, which had been investigated extensively (Östlund and Hut, 1984; Bauch et al., 1995; Ekwurzel et al., 2001; Yamamoto-Kawai et al., 2008; Bauch et al., 2011b; Newton et al., 2013).

The Arctic Ocean receives water from various sources, including the Pacific and Atlantic Oceans, river inflow, glaciers, sea-ice meltwater, and local precipitation. Atlantic Water flows into the Arctic through Fram Strait, as well as through the Barents and Kara Seas (Aagaard and Carmack, 1989; Rudels et al., 2005). The inflow of relatively warm saline Atlantic water is associated with a relatively higher isotopic composition (Östlund and Hut, 1984; Legrande and Schmidt, 2006). We, therefore, assume a relatively strong influence of Atlantic waters in our data at the beginning of the expedition in the Amundsen Basin (85°N, 135°E), where seawater δ¹⁸O was relatively high (about −1.5‰). The contribution of Atlantic Water to the Arctic Ocean was also noticeable during the transition period between Legs 3 and 4, when Polarstern left the ice floe and moved into the open waters to reach Svalbard (May 16 to June 17, −0.8 < δ¹⁸O < 0.1‰, 31.1 < salinity < 33.9). The other marine water source to the Arctic Ocean is Pacific inflow through the Bering Strait. However, because the MOSAiC expedition took place in the eastern sector of the Arctic Ocean, the potential effect of this water source may not be traceable in this study (Bauch et al., 2011b; Paffrath et al., 2021).

River discharge provides a freshwater source with a distinctively low isotopic composition (Yi et al., 2012). Siberian rivers (i.e., Lena, Yenisey, Ob) channel meteoric waters falling over the continent and release them into the Arctic Ocean through the Kara, Laptev, and East Siberian seas. The freshwater input from these rivers provides regionally low saline waters through the shelf seas to the Arctic Ocean. The depletion of seawater in δ¹⁸O in the Arctic Ocean is known to be associated with river discharge (Solomon et al., 2021). In their seminal study, Cooper et al. (2008) conducted extensive observations of major Arctic Ocean inflowing rivers, revealing remarkably low isotopic compositions with δ¹⁸O values as follows: −14.6‰, Ob; −17.5‰, Yenisey; −20‰, Lena; and −21.9‰, Kolyma. In October–November 2018, during the year preceding the MOSAiC expedition and, in a region, coinciding with the proximity of Polarstern, the δ¹⁸O values of Lena River water were primarily within the range of −18‰ to −20‰ (Juhls et al., 2020). Freshwater was subsequently introduced into the Arctic Ocean, thereby contributing to the overall isotopic composition of the Arctic Ocean’s water. When Polarstern drifted into the Nansen Basin, the surface seawater δ¹⁸O and salinity rose to −0.4‰ and 34.3 (April 29), respectively. We attribute the isotopic enrichment to a gradual increase in the influence of Atlantic Water (and a decreasing influence of riverine meteoric water) during Leg 3.

The largest and most abrupt changes in seawater isotopic composition and salinity during the MOSAiC campaign occurred during Leg 4, when δ¹⁸O and salinity decreased progressively (from around −0.3‰ to −2.2‰; Figure 9) as Polarstern passed through Fram Strait on the edge of the sea ice into the East Greenland Current (EGC). The simultaneous decreasing change in salinity suggests the mixing of freshwater with ocean water in this region. The EGC flows along the continental slope of the Greenland Sea Shelf and transports freshwater and sea ice with low δ¹⁸O from the Arctic Ocean toward the North Atlantic (Aagaard and Carmack, 1989; Rabe et al., 2014; Steur et al., 2018). In addition, the Greenland Ice Sheet releases meltwater and has been described as one of the freshwater sources contributing to low δ¹⁸O (Dodd et al., 2012; Laukert et al., 2017) and salinity (Rudels et al., 2005). Greenland Ice Sheet meltwater is discharged into the North Atlantic at remarkable rates through the 79°N glacier, as documented by Huhn et al. (2021). The seawater collected in the vicinity of this release of glacier meltwater exhibited consistently low δ¹⁸O values, with an average value of approximately −2.9‰ (±0.1, N = 7), and a low salinity mean of 31.3 (±0.4). Moreover, during Leg 4, the ice floe experienced an increase in melt ponds from sea ice and snow melt at the surface, accounting for up to 20% of the surface area by the end of July and accompanied by a drastic reduction in snow-covered areas (Webster et al., 2022). The continuous release of this fresh and isotopically light water throughout Leg 4 contributes to the gradual depletion of seawater in δ¹⁸O while Polarstern was moving southward with the EGC. Due to similar δ¹⁸O values of freshwater, contributions from river water, snow melt, or glacial melt cannot be distinguished (Dodd et al., 2012; Laukert et al., 2017).

The lowest salinity and δ¹⁸O values during MOSAiC were recorded during Leg 5 (August and September 2020) close to the North Pole in the Amundsen Basin (87°N to 89°N). Compared to Atlantic Waters (Legrande and Schmidt, 2006), seawater in the sea-ice-covered area around the North Pole had a low (approximately −3‰) δ¹⁸O composition. This low value points to a large contribution of freshwater to this region in the heart of the Arctic Ocean. Potential sources of light isotopic freshwater in the Arctic Ocean in Leg 5 were similar to those described before; that is, the Siberian riverine influx, local precipitation, and the snow on sea ice. However, slopes and intercepts between stable isotopes and salinity during Leg 5 are distinctly different from those of Legs 1–4 (Figure 9), suggesting at first glance the mixing of different endmembers. An endmember refers to a pure and distinct component within a mixture, representing an extreme or specific source with unique characteristics, often used in isotopic analysis to identify relative contributions from different sources or processes. But y-intercepts in the Arctic cannot be interpreted as endmembers due to the influence of sea-ice formation, which adds salt to the water column. The observed differences in correlations and slopes between datasets can be attributed to differences in geographical regions, as well as the seasonality of sea-ice melting and formation (Bauch et al., 2011a). However, the distinction between freshwater sources such as rivers and snow melt cannot be based solely on stable isotopic water signatures and cannot be further disentangled in this study.

In addition to snow, sea ice, and seawater, melt ponds, lead water, lead ice, and water vapor were sampled for isotopic measurements during the MOSAiC expedition. Each of these hydrological components provide additional and distinctive contributions to an understanding of the Arctic water cycle as a whole.

Melt ponds play a crucial role in the energy budget of Arctic sea ice, especially in spring and summer, as they absorb more solar radiation than the surrounding ice (Perovich et al., 1998). The isotopic composition of melt ponds in this study changed over time, due to the relative contributions of snow and sea ice melt, the evaporation from the pond surface, and refreezing. Generally, refrozen and open melt ponds had an intermediate isotopic signature compared to the surrounding sea ice and snow (Table 1), suggesting that they were a mixture of sea ice and snow melt. Similarities in the isotope composition of melt ponds and SYI suggest that this mixture in ponds could be involved as a first step in the formation of SYI (Figure 10). As outlined above, the isotopic composition of sea ice and snow changed over time due to the relative contributions of different water sources and due to the interaction between snow and sea ice, and these changes may be transferred into melt ponds as they became larger and deeper due to the continuous sea-ice melt (Zhang et al., 2018).

Lead water had slightly lower mean δ¹⁸O (−2.3‰; Table 1) compared to lead ice (−1.2‰), seawater (−1.7‰), and FYI (−0.7‰), which could be a result of the contribution of freshwater at the surface during summer as suggested by Nomura et al. (2023). They studied lead effects on water structure during Leg 5 in the central Arctic, analyzing lead width, refreezing, and mixing. Melt ponds had similar δ¹⁸O values to lead water samples and surrounding sea ice. The melting of sea ice resulted in the production of liquid water that displayed reduced δ¹⁸O values compared to the remaining ice. This disparity arises from the selective liberation of lighter isotopes, specifically ¹⁶O, in comparison to the heavier isotopes (¹⁸O), which are released preferentially at a higher rate. Ice formation and melting change the salinity of seawater, influencing the δ¹⁸O values of melt ponds. Melt pond formation causes dissolved oxygen concentration in seawater to decrease, changing the δ¹⁸O values due to oxygen isotope exchange with the atmosphere. In our study, we observed that the mean d-excess of melt ponds (0.9‰) fell between the values for FYI (0.5‰) and SYI (1.6‰), with the rather low melt pond δ¹⁸O values (mean = −2.1‰) suggesting the contribution of snow to melt ponds.

Brunello et al. (2023) investigated the continuous water vapor isotopic composition in the Arctic during MOSAiC, revealing seasonal variations in δ¹⁸O and d-excess. The authors found that the changes in the isotopic composition of the summer moisture correlated with temperature and humidity over open ocean sectors, and that they were controlled largely by evaporative conditions over the source regions. In contrast, winter moisture could not be explained solely by source conditions and appears to be influenced by moisture exchange processes during the transport and the long residence over the sea ice. Coupling these results with isotopic composition data from discrete samples of sea ice, seawater, snow, and melt ponds can lead to a better understanding of the regional impacts of Arctic water exchange processes.

Climate change causes profound alterations of the Arctic region. The isotopic composition of the hydrological components informs us on the dynamics and processes involved in the Arctic water cycle, contributing to an assessment of the current state and future changes in response to ongoing warming. In this study, we have presented the isotopic composition (δ¹⁸O, δ²H) of the different water components in the central Arctic Ocean, as observed during the MOSAiC drifting expedition in 2019–2020. The overall dataset includes 2200 samples and provides a statistical overview of the isotopic composition of sea ice, seawater, snow, melt ponds, and other parts of the Arctic hydrological cycle. Both spatial and temporal gradients were studied with a focus of surface seawater, sea ice, and snow on top of the sea ice.

We found that the δ¹⁸O values of snow gradually increased from the surface to the bottom of the snowpack, whereas d-excess values decreased. This gradient is visible in most snow profiles, independently of time and location of sampling. The higher δ¹⁸O and lower d-excess values in the middle and bottom layers of the snowpack indicate a smaller contribution from meteoric water with depth and a significant impact from the underlying sea ice surface. Even at the snow surface, where fresh precipitation would be added, air temperatures were not correlated with the surface snow δ¹⁸O values. Thus, the contribution of meteoric waters (e.g., fresh snowfall) was either very limited or had been modified after deposition. Additionally, the continuous depletion of d-excess from October 2019 through July 2020 (e.g., Figure 5a) was associated with the snow-sea ice interaction and the sublimation processes at this interface. These findings suggest the need for further investigation into the effects of snow metamorphic processes on the isotopic composition of snow on top of the sea ice.

The surface seawater isotopic composition and salinities across the Arctic Ocean during the MOSAiC expedition indicate different water sources that contributed with distinct isotopic signatures. Our analysis showed a strong correlation between salinity and δ¹⁸O. The geographical as well as seasonal shifts in salinity and δ¹⁸O correlation suggest that seawater isotopes can be used to infer the relative contributions of distinct water masses to the whole Arctic Ocean. Significant contributions are Atlantic Ocean waters mixing with fresh water influxes from Siberian rivers (Legs 1, 2), and the melt of snow together with freshwater input from sea-ice melt and influx from the Greenland ice sheet (Legs 4, 5). These contributions lead to regionally and seasonally different, specific Arctic surface seawater isotopic compositions and salinities.

The isotopic composition of sea ice provides useful information about different formation mechanisms, and the key role of sea ice in the Arctic Ocean’s water cycle. FYI was typically more enriched in δ¹⁸O than SYI due to various factors, including freezing and post-freezing modifications. The sea-ice isotopic composition was influenced by factors such as lower surface seawater δ¹⁸O values at the time of formation and evaporation from melt ponds, resulting in lower salinity and δ¹⁸O. In contrast, iFYI, which forms during the winter season at the bottom of SYI, generally had a higher isotopic composition than SYI due to similar factors that contribute to the heavier isotope composition of FYI compared to SYI. A threshold of approximately −2‰ separated the iFYI from the upper SYI layer. Our results show that the isotopic compositions help to distinguish between ocean-sourced FYI and iFYI from SYI with a meteoric contribution. Moreover, they offer crucial insights into the isotopic dynamics between snow and sea ice in the central Arctic.

In summary, seawater, sea ice, and snow as key components of the Arctic water cycle interact with each other as well as with more transient constituents such as melt ponds, leads, and frost flowers to provide insight into local exchange processes with the atmosphere. These findings deepen our understanding of the intricate interactions between atmosphere, sea ice, and ocean in this region and underscore the need for comprehensive and coordinated isotopic data collection across diverse Arctic locations and seasons. Such a collaborative approach is vital as we face complex challenges posed by climate change in the Arctic.

All data in this manuscript are publicly available from online repositories. The data can be found under the following references:

• Stable water isotopes of underway seawater during the MOSAiC expedition (Mellat et al., 2022b).

• Stable water isotopes of snow during MOSAiC expedition (Mellat et al., 2022a).

• Stable water isotopes and conductivities of a lead case study during Leg 5 of the MOSAiC expedition (Meyer et al., 2022).

• Stable water isotopes of seawater at site Ocean City during MOSAiC expedition (Mellat et al., 2023a).

• Stable water isotopes of sea ice at multiple coring stations (MCS) during the MOSAiC expedition (Mellat et al., 2023b).

• Overview of physical oceanography measurements during Polarstern cruise PS122—MOSAiC (Tippenhauer et al., 2021).

• Continuous meteorological surface measurement during Polarstern cruise PS122/1-2-3-4-5 (Schmithüsen, 2021a, 2021b, 2021c, 2021d, 2021e).

• Master tracks in different resolutions of Polarstern cruise PS122/1-2-3-4-5 (Haas, 2020; Kanzow, 2020; Rex, 2020, 2021a, 2021b).

The supplemental files for this article can be found as follows:

Figures S1 and S2. PDF

The authors would like to express their gratitude to the International Multidisciplinary Drifting Observatory for the Study of the Arctic Climate (MOSAiC) with the tag MOSAiC20192020 for providing the opportunity to conduct this research and for producing the data used in this report. Specifically, they would like to thank all people who were involved in this expedition of the Research Vessel Polarstern (Knust, 2017) and helped to collect the samples during MOSAiC in the 2019–2020 project (Project ID: AWI_PS122_00) (Nixdorf et al., 2021). They extend their sincere appreciation to all the persons involved in the expedition of the R/V Polarstern during MOSAiC in 2019–2020 (AWI_PS122_00). They also thank Mikaela Weiner and Andreas Marent for analyzing the isotopic compositions (δ¹⁸O, δ²H) of the samples in the ISOLAB Facility at AWI in Potsdam. They also extend their thanks to the two anonymous reviewers whose insightful and constructive feedback contributed to the improvement of the manuscript’s structure and clarity. They also acknowledge support by the Open Access Publication Funds of Alfred-Wegener-Institut Helmholtz-Zentrum für Polar- und Meeresforschung.

MM, CFB, MW, and HM are grateful for the support provided by the German Federal Ministry of Education and Research (project MOSAiC-CiASOM, grant number 03FO869A). DB received support from the German science foundation (DFG grant BA1689/4). MAG received support from The Research Council of Norway (project HAVOC, grant no. 280292). DN received support from the Japan Society for the Promotion of Science (grant numbers: JP18H03745; JP18KK0292; JP17KK0083; JP17H04715; JP20H04345) and a grant from the Joint Research Program of the Japan Arctic Research Network Center. SA was funded by the German Research Council (DFG) in the framework of the priority program “Antarctic Research with comparative investigations in Arctic Ice areas” (Grant Nos. SPP1158 and AR1236/1), the Emmy Noether Grant of the DFG (AR 1236/2-1/493362232) and by the Alfred-Wegener-Institut, Helmholtz-Zentrum fuür Polar-und Meeresforschung.

All authors declare that they have no competing interests.

Contributed to conception and design: MM, HM, CFB, MW, DB, MAG, JMW.

Contributed to acquisition of data: MM, HM, CFB, MW, DB, ED, MA, DN, JMW, MS, MAG, MH, ARM, SA.

Contributed to analysis and interpretation of data: MM, HM, CFB, MW, DB, MAG, MA, JMW.

Drafted and/or revised the article: MM, HM, MW, CFB, DB, MAG, ARM, DN, MA, JMW, MS, SA, MH, ED.

Approved the submitted version for publication: All Authors.

How to cite this article: Mellat, M, Brunello, CF, Werner, M, Bauch, D, Damm, E, Angelopoulos, M, Nomura, D, Welker, JM, Schneebeli, M, Granskog, MA, Hoerhold, M, Macfarlane, AR, Arndt, S, Meyer, H. 2024. Isotopic signatures of snow, sea ice, and surface seawater in the central Arctic Ocean during the MOSAiC expedition. Elementa: Science of the Anthropocene 12(1). DOI: https://doi.org/10.1525/elementa.2023.00078

Domain Editor-in-Chief: Jody W. Deming, University of Washington, Seattle, WA, USA

Guest Editor: Gunnar Spreen, Institute of Environmental Physics, University of Bremen, Bremen, Germany

Knowledge Domain: Ocean Science

Part of an Elementa Special Feature: The Multidisciplinary Drifting Observatory for the Study of Arctic Climate (MOSAiC)