Abstract

The authors explore possible temperature modifications of the Atlantic Water Layer (AWL) induced by climate change, performing simulations for 1970 to 2099 with a coupled ice–ocean Arctic model (CIOM). Surface fields to drive the CIOM were provided by the Canadian Regional Climate Model (CRCM), driven by outputs from the Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled Global Climate Model, version 3 (CGCM3) following the A1B climate change scenario. In the present climate, represented as 1990–2009, the CIOM can reliably reproduce the AWL compared to Polar Science Center Hydrographic Climatology (PHC) data. For the future climate, assuming the A1B climate change scenario, there is a significant increase in water volume transport into the central Arctic Ocean through Fram Strait due to the weakened atmospheric high pressure system over the western Arctic and an intensified atmospheric low pressure system over the Nordic seas. The AWL temperature tends to decrease from 0.36°C in the 2010s to 0.26°C in the 2060s. In the vertical, the warm Atlantic water core slightly expands before the 2030s, significantly shrinks after the 2050s, and essentially disappears by 2070–99, in the southern Beaufort Sea. The temperature decrease after 2030 is mainly due to the reduced heat fluxes in the Kara and Barents Seas. In the northeastern Barents and Kara Seas, the loss of sea ice increases the heat loss from the Atlantic water and reduces the water temperature near the bottom, contributing to decreased heat fluxes into the central Arctic Ocean, as well as decreased AWL temperature at central Arctic Ocean intermediate layers. In addition, the vertically integrated heat loss also plays an important role in the AWL cooling process.

1. Introduction

In the present climate, Atlantic water (AW) enters the Arctic Ocean through Fram Strait and the Barents Sea (Figs. 1a,b). The Fram Strait branch along West Spitsbergen region is relatively warm and continues eastward along the southern margin of the Nansen basin, with a shallow core distinguished by a temperature maximum (Karcher et al. 2012). However, the Barents Sea branch is relatively cold and less saline, and experiences cooling and freshening in the Barents and Kara Seas. While the surface water, with salinity smaller than 34.8, in the Barents and Kara Seas recirculates to Fram Strait because of its small density, the deep dense AW, with salinity about 35, enters the Nansen basin through Victoria Channel and the St. Anna Trough. Along the north slope of the Kara Sea, the water from Fram Strait and the northeastern Barents and Kara Seas undergoes limited mixing and continues to move cyclonically along the slope of the Arctic basin, accompanied by bathymetrically steered flows returning toward Fram Strait along the Lomonosov and Mendeleyev Ridges (Figs. 1a,b; Aagaard 1989; Rudels et al. 1994; Woodgate et al. 2001; Karcher and Oberhuber 2002; Karcher et al. 2012). In addition, the AW spreads over the whole Arctic Ocean, at depths that are often 200–900 m, through the cyclonic boundary current and thermohaline intrusion (McLaughlin et al. 2009), forming a warm Atlantic Water Layer (AWL) in the central Arctic Ocean. However, the AW is quite variable. The upper depth of AW can vary in space from about 150 m in the Eurasian basin to close to 250 m in the Canada basin.

Fig. 1.

(a) Model domain for CIOM. Units are in meters (shaded), and the thick black line across the Arctic Ocean shows the location of hydrographic sections referred to later in the text. The lines A, B, C, and D are used to compute the heat fluxes into the central Arctic Ocean from the Kara Sea, the Barents Sea, downstream of Fram Strait, and Fram Strait, respectively. (b) Schematic patterns for the AW flow in the central Arctic Ocean.

Fig. 1.

(a) Model domain for CIOM. Units are in meters (shaded), and the thick black line across the Arctic Ocean shows the location of hydrographic sections referred to later in the text. The lines A, B, C, and D are used to compute the heat fluxes into the central Arctic Ocean from the Kara Sea, the Barents Sea, downstream of Fram Strait, and Fram Strait, respectively. (b) Schematic patterns for the AW flow in the central Arctic Ocean.

The water temperatures associated with the AWL are dominated by low-frequency oscillations at time scales of 50–80 years, and changes in the AWL can have long-term consequences for sea ice and ocean circulation in the Arctic Ocean (Dmitrenko et al. 2010; Polyakov et al. 2010; Karcher et al. 2011). Observations show that the AW in the central Arctic Ocean was relatively warmer in the 1930s–1940s, and also in recent decades, but colder in the early twentieth century and in the 1960s–1970s (Polyakov et al. 2004). However, the temperature anomalies associated with the decadal variations have been abrupt and pulse-like (Karcher et al. 2003; Polyakov et al. 2005; Bourgain and Gascard 2012), progressing from the eastern North Atlantic into the Nordic seas and propagating farther toward the central Arctic Ocean (Carmack et al. 1995, 1997; Karcher et al. 2003; Dmitrenko et al. 2008; Holliday et al. 2008). Observations also suggest that the AW warming is associated with a substantial shoaling of the upper AW boundary in the central Arctic Ocean and weakening of the Eurasian basin upper-ocean stratification; the heat flux from the intermediate-depth Atlantic water to the upper surface layer in the central Eurasian basin is about 1 W m−2 (Polyakov et al. 2013). Therefore, changes in the Eurasian basin can increase the upward transfer of AW heat to the ocean surface layer and help to precondition the polar ice cap for the extreme ice loss events (Dmitrenko et al. 2010; Polyakov et al. 2010). In addition, the large-scale atmospheric circulation plays an important role in the AWL decadal variations. For example, the anomalous cyclonic wind stress in the 1990s not only had a strong impact on the intensity of the cyclonic boundary currents at intermediate layers, but also was associated with an increased Atlantic water inflow through Fram Strait (Karcher et al. 2003).

The changes in the Arctic atmospheric conditions under high CO2 climate change scenarios can have significant impacts on the Arctic Ocean (Deser et al. 2010; Vavrus et al. 2012; Koenigk et al. 2013; Long and Perrie 2013). Because of the positive feedbacks involving snow and sea ice, and other related processes, the largest lower tropospheric warming is expected to occur in the Arctic. As surface air temperature increases, Arctic ice cover will retreat, with the biggest decreases projected to occur in summers. In addition, model simulations show an increasing trend in the freshwater content (FWC) and sea surface height (SSH) in the Beaufort Sea due to enhanced ice melting and Ekman transport. The maximum increases in FWC and SSH are expected to occur near the center of the Beaufort Gyre, where the maximum FWC and SSH are located (Long and Perrie 2013). Moreover, recent studies suggest that the increased water temperature associated with climate change can enhance the northward heat flux in the Barents Sea (Koenigk and Brodeau 2013).

In terms of the AW long-term variability, previous studies mainly focus on the present climate (Karcher et al. 2003; Polyakov et al. 2004, 2005; Dmitrenko et al. 2008; Holliday et al. 2008; Dmitrenko et al. 2010; Polyakov et al. 2010; Karcher et al. 2011; McLaughlin et al. 2011; Bourgain and Gascard 2012), and no studies on estimated AW projections under climate change scenarios have been reported. Since the AW is an important heat and mass source for the Arctic Ocean, any changes in the AW can have significant impacts on the heat and mass balance in the Arctic. In this study, we will use a coupled ice–ocean model (CIOM) to simulate the AW in the central Arctic Ocean to understand the impacts of climate change on the AWL.

The objective of this paper is to investigate the changes in the water temperature associated with the AWL in a warmer climate. Section 2 describes the model and the experimental design. Section 3 shows the changes of water temperature at intermediate layers. Air–sea interaction effects in the northeastern Barents and Kara Seas are considered in section 4. Discussions are given in section 5, and section 6 presents the conclusions.

2. Model description and experiment design

The coupled ice–ocean model used in this study consists of two components, the Princeton Ocean Model (POM; Blumberg and Mellor 1987) and a multicategory ice model (Hibler 1979, 1980). Long et al. (2012) implemented the CIOM in the Arctic Ocean to study the interannual variations of FWC and SSH in the Beaufort Sea, based on earlier versions by Yao et al. (2000) and Wang et al. (2005). This model system has also been successfully applied in a recent climate change study (Long and Perrie 2013).

a. Ocean model

POM is a three-dimensional, primitive equation model with complete thermohaline dynamics, a sigma (σ) vertical coordinate and a free surface. A second-order turbulence closure scheme (Mellor and Yamada 1982) is used to represent mixed layer dynamics. In the experiments described here, 25 vertical sigma levels are used with higher resolution in the upper mixed layer and lower resolution in the deep ocean. In the central Beaufort Sea where the depth is about 3500 m, the vertical resolution decreases from about 6 m for the upper seven layers to about 380 m near the bottom, with a vertical resolution of about 60 m at the depths of 200–400 m; this resolution is too coarse to resolve the processes associated with double diffusion and the strong vertical temperature gradients near 200 m. In addition, to minimize pressure gradient errors, the bottom topography in the model was smoothed such that the difference in the depths of adjacent grid points divided by their means is less than 0.2 (Mellor et al. 1994). This smoothing process may cause underestimates in the cyclonic boundary circulation. The model grid is distributed on a rotated spherical surface with the North Pole at 8°N, 131.5°E (Fig. 1a). The horizontal resolution is 0.29° × 0.25°, and the time step is 20 min. The background vertical eddy viscosity varies from 1 × 10−5 to 2 × 10−5 m2 s−1, while the background vertical eddy diffusivity is set in the range from 1 × 10−6 to 1 × 10−5 m2 s−1, depending on the local ice concentration conditions. To take into account of the effects of eddies and tides, an extra 5 × 10−5 m2 s−1 is added to the background viscosity and diffusivity for the water above 70 m, based on observations (Fer 2009). Without the inclusion of the impacts of eddies and tides, the simulated AWL is too warm, compared to PHC data (not shown). In addition, the Neptune effect is implemented in the model to represent the interactions between eddies and bottom topography (Holloway and Wang 2009). However, because of the relatively coarse resolution, the currents associated with the Neptune effect are underestimated, in particular the currents along the Lomonosov and Mendeleyev Ridges.

Along open boundaries, we use radiation boundary conditions for the baroclinic current, and volume transports are specified, based on observations (Beszczynska-Möller et al. 2011; Fig. 1a). The inflow through Bering Strait is prescribed as 0.8 Sv (1 Sv ≡ 106 m3 s−1), and 2.3 Sv is prescribed as the outflow through the Canadian Archipelago. We also prescribe an inflow of 8.5 Sv of Atlantic water into the Arctic Ocean via the Norwegian Sea, and an outflow of 7.0 Sv along the east coast of Greenland. Although changes in lateral water volume transport can impact simulations of the AWL, in fact recent studies suggest that no significant changes are expected in the water volume transported into the Arctic Ocean along the lateral boundaries (Jahn and Holland 2013). Moreover, the enhanced northward heat flux into the Arctic Ocean, related to climate change, is mainly associated with increased water temperature, whereas the contributions from the changes in the water volume transport are relatively small, in particular in the Barents Sea (Koenigk and Brodeau 2013; Beszczynska-Möller et al. 2012).

Water temperature T and salinity S at open boundaries are prescribed using the Polar Science Center Hydrographic Climatology (PHC3.0, hereafter PHC) by Steele et al. (2001), plus monthly anomalies from the Canadian Centre for Climate Modelling and Analysis (CCCma) Coupled Global Climate Model, version 3 (CGCM3). The T and S anomalies are computed from CGCM3 outputs, minus their long-term means (1979–2008). The anomalies are added to the PHC climatology (Steele et al. 2001) to get final estimates for T and S (for 1970 to 2099) for the open boundaries. No restoration or “nudging” is applied for T and S during the simulation.

b. Ice model

The sea ice component of CIOM was developed at Bedford Institute of Oceanography (Yao et al. 2000) using a thermodynamic model based on a multicategory ice thickness distribution function (Thorndike et al. 1975; Hibler 1980) and a viscous-plastic sea ice dynamics model (Hibler 1979). The ice thickness has seven categories (in terms of their upper thickness limits: 0.4, 0.8, 1.2, 2, 3, 5, and 7 m). Heat and salt fluxes at the ice–ocean interface are governed by appropriate boundary processes as discussed by Mellor and Kantha (1989).The surface heat flux is given as

 
formula

where ρ0 is1026 kg m−3, Cp is the specific heat of seawater, T0 is the temperature at the bottom of ice, and T is the temperature at the uppermost model grid point. The heat transfer coefficient Cz is defined by , where u* = is surface friction velocity in terms of surface stress τ and water density , is the turbulent Prandtl number, k = 0.4 is the von Kármán constant, and z is the vertical coordinate corresponding to T. Here, z0 is a roughness parameter, specifically the sum of under-ice roughness length (z0i) and open ocean roughness length (z0o) weighted by the ice area A, given by lnz0 = A ln(z0i) + (1 − A) ln(z0o), and the factor BT parameterizes a molecular sublayer, given by , where is the molecular viscosity.

Similarly, the salt flux at the ice–water interface is given as

 
formula

where S0 is the salinity at the ice–ocean interface, S is the salinity at the uppermost grid point, and is the transfer coefficient for salinity, , where . The background details for this formulation can be found in Mellor and Kantha (1989). An empirical formulation is used to estimate clear-sky incoming shortwave radiation following Shine and Crane (1984), with the cloud correction formulation of Reed (1977). The formulation for longwave radiation is given by the Smith and Dobson (1984) parameterization, and bulk formulas are used to estimate latent and sensible heat fluxes and wind stress. Air–ice and air–water drag coefficients are set to 2.75 × 10−3 (Prinsenberg and Peterson 2002) and 1.3 × 10−3 (http://www.whoi.edu/page.do?pid=30576), respectively, while the ice–water drag coefficient is 5.5 × 10−3 (Shirasawa and Ingram 1991). A sea ice–albedo scheme is based on the formula suggested by Køltzow (2007).

c. Experiment

CGCM3 is a coupled atmosphere–ice–ocean global climate model. Its atmospheric output has about 2.5° horizontal resolution, and the CGCM3 ocean resolution is about 1.85°, both of which are too coarse to simulate the detailed features of the ice edge and Arctic Ocean circulation. The Canadian Regional Climate Model (CRCM) is a limited-area atmospheric model, which is used to downscale CGCM3 simulations for 1970–2099 to provide high-resolution (45 km) surface fields to drive CIOM, following the A1B climate change scenario (Solomon et al. 2007). The A1B scenario is one of the three A1 scenario groups, which include fossil fuel energy intensive (A1FI), non–fossil fuel energy intensive (A1T), and balanced across all energy sources (A1B) scenarios. The A1B scenario predicts carbon dioxide emissions that are increasing until around 2050 and then decreasing after that. The CGCM3 simulations provide CRCM with initial and lateral boundary conditions (6-hourly wind, geopotential height, temperature, and specific humidity). The sea ice and sea surface temperature used in CRCM simulations are specified by the CGCM3 simulations. Surface air temperature, surface specific humidity, precipitation rate, total cloud cover, sea level pressure (SLP), and 10-m winds (from CRCM simulations) are linearly interpolated to the CIOM grid to provide surface driving fields for CIOM. Long and Perrie (2013) used the same surface fields to drive CIOM, to study the impacts of climate change on the sea surface height and freshwater content in the Beaufort Sea. Comparisons between CRCM simulations and NCEP reanalyses suggest that CRCM can capture the main features of sea level pressure and surface air temperature in NCEP data. However, CRCM outputs tend to slightly overestimate the Beaufort Sea high (BSH) and to underestimate the Icelandic low pressure system over the Barents and Norwegian Seas (Long and Perrie 2013). Additional related details (e.g., concerning the surface fields from the CRCM simulations) are given by Long and Perrie (2013).

CIOM was integrated for 130 years from 1970 to 2099. The initial conditions for both the ocean and the sea ice come from the run driven with CRCM outputs. In this approach, the CRCM surface variables for 1970–99 are used to drive CIOM for 60 years to get the initial conditions, including sea ice, water temperature, and salinity. Climatological river runoff observations are prescribed as precipitation minus evaporation (PE) along the Arctic coast for 1970–99 (Prange and Lohmann 2004); moreover, an increasing trend, with an amplitude of 0.12% yr−1, is added for 2000–99 with a total increase of 4.15 km3 yr−1 (Wu et al. 2005). In this study, our analyses focus only on the CIOM output from 1979 to 2099. However, since the analyses are based on a one-member ensemble for the CIOM simulation, which is subject to uncertainties, the results from this study are more qualitative than quantitative.

3. Impacts of climate changes on the waters at intermediate layers

a. Current climate

1) Spatial patterns of water temperature

Simulated water temperature at 200 m is compared to PHC data in Fig. 2. In the PHC data, the water temperature in the West Spitsbergen region is 1°–3°C (Fig. 2f), but it gradually decreases as it moves westward into the interior Arctic Ocean, decreasing from above 0.5° to below 0°C as it moves westward along the Siberian shelf. It is in the range from 0° to −1°C in the Canada basin. In addition, the water is relatively warm near the Barents Sea Opening (3°–5°C) but rapidly decreases in the Barents Sea due to the atmospheric cooling, and reaches from −1° to 0°C near St. Anna Trough. Comparisons between the CIOM simulations in 1990–2009 (Fig. 2a) and the PHC data (Fig. 2f) show that CIOM generally simulates water temperature well (e.g., the cold temperatures in the Canada basin and warm temperatures in the Eurasian basin). However, CIOM slightly overestimates (by about 0.3°–0.9°C) the water temperature at 200 m, failing to reproduce the temperature minimum (in the range from −1° to −0.5°C) in the Canada basin (Fig. 2g). Moreover, CIOM underestimates the water temperature, by about 0.3°–0.6°C, in the eastern part of Fram Strait and by about 0.3°–1.5°C in the central Greenland Sea (Fig. 2g) and fails to capture the water temperature minimum (Fig. 2f). In addition, the area where the temperature is below 0°C in the Canada basin is smaller in the CIOM simulations than that in the PHC data.

Fig. 2.

Annual water temperature (°C) at 200 m, showing (a)–(e) decadal averages for CIOM results and (f) PHC data. Bidecadal averages are indicated for CIOM results. (g) The difference between the CIOM simulation (1990–2009) and PHC data (°C).

Fig. 2.

Annual water temperature (°C) at 200 m, showing (a)–(e) decadal averages for CIOM results and (f) PHC data. Bidecadal averages are indicated for CIOM results. (g) The difference between the CIOM simulation (1990–2009) and PHC data (°C).

The water temperature at 450 m is relatively uniform in the central Arctic Ocean (Fig. 3). For example, the water temperature is in the range from 0° to 0.5°C in the central Arctic Ocean and in the range from 1° to 1.5°C in the region near West Spitsbergen in the PHC data (Fig. 3f). By comparison, CIOM generally reproduces the PHC water temperature in the central Arctic Ocean (Fig. 3a), but slightly underestimates water temperatures in Canada basin by about 0°–0.3°C (Fig. 3g). In addition, CIOM overestimates the water temperature by about 0.3°–1.8°C in Greenland Sea and underestimates PHC water temperatures near the Norwegian coast by 0.9°–12°C, as indicated in Fig. 3g.

Fig. 3.

As in Fig. 2, but at 450 m.

Fig. 3.

As in Fig. 2, but at 450 m.

The vertical structure of water temperature in the central Arctic Ocean suggests a three-layer system in the PHC data (Fig. 4). While the AWL at 200–900 m is relatively warm (0°–1.5°C), the water temperature is near the freezing point above 120 m, and about 0°C below 900 m. In addition, the AWL is slightly shallower in the Eurasian basin than in the Canada basin, with a temperature maximum along the Barents Sea shelf at 200–450 m, and the surface water in the Barents and Beaufort Seas is relatively warm (Fig. 4f). Compared to the PHC data (Fig. 4f), the model simulation reproduces the three-layer vertical structure (Fig. 4a) but overestimates the vertical extent of the AWL by about 100 m and underestimates the water temperature near the surface in the Barents Sea by about 2.5°–3°C (Fig. 4g). In addition, CIOM underestimates the strong vertical temperature gradients near 200 m in the PHC data (Figs. 4a,f,g).

Fig. 4.

Cross sections for annual water temperature (°C) as a function of depth (m) along the line indicated in Fig. 1. The x-axis units are 103 kilometers, showing (a)–(e) bidecadal averages for CIOM results and (f) PHC data. (g) The difference between the CIOM simulation (1990–2009) and PHC data (°C).

Fig. 4.

Cross sections for annual water temperature (°C) as a function of depth (m) along the line indicated in Fig. 1. The x-axis units are 103 kilometers, showing (a)–(e) bidecadal averages for CIOM results and (f) PHC data. (g) The difference between the CIOM simulation (1990–2009) and PHC data (°C).

2) Atlantic water in the central Arctic Ocean

Warm saline Atlantic water enters the Barents Sea through the Barents Sea Opening at a rate of about 1.8 Sv (Blindheim 1989; Skagseth et al. 2011; Rudels et al. 2013). However, before flowing into the central Arctic Ocean through the Victoria Channel and the St. Anna Trough, it rapidly loses its heat as a result of atmospheric cooling, constituting more than 50% of the heat loss in the Arctic Ocean (Serreze et al. 2007). Moreover, the cold water inflow into the Barents and Kara Seas provides a significant portion of the Atlantic water occurring at intermediate layers, which plays an important role in the formation of the AWL (Rudels et al. 1994; Schauer et al. 1997). In the Barents and Kara Seas, the Atlantic water enters the central Arctic Ocean mainly through the St. Anna Trough and the Victoria Channel. The observed net volume transport through the St. Anna Trough is 1.5 Sv (Simonsen and Haugan 1996; Schauer et al. 2002), and recent model simulations also suggest a 1.4 Sv water volume transport into the central Arctic Ocean from the Kara Sea (Aksenov et al. 2010). However, CIOM slightly overestimates the water inflow, with a volume transport of 1.7 Sv in 1980–2000 (Fig. 5a). For the Victoria Channel, although there are no available observations, model simulations suggest about 0.4 Sv cold dense water inflow into Nansen basin from the Barents Sea (Maslowski et al. 2004; Aksenov et al. 2010; Årthun et al. 2011). Similarly, the CIOM simulation shows a volume transport of 0.4 Sv into Nansen basin (Fig. 5a).

Fig. 5.

(a) Average water volume transports (Sv) at 100–900-m depth into the central Arctic Ocean from the Kara Sea (green, averaged along line A in Fig. 1), the Barents Sea (red, averaged along line B in Fig. 1), downstream of Fram Strait (broken red, averaged along line C), Fram Strait inflow (black, averaged along line D), and Fram Strait outflow (broken black). The convention is that for A, B, and D, positive is into the Arctic Ocean while negative is out of the Arctic Ocean. For section C, positive indicates counterclockwise boundary flow. (b) Average water temperature (°C) at 100–900-m depth in the Kara Sea (green, averaged along line A in Fig. 1), the Barents Sea (red, averaged along line B in Fig. 1), at Fram Strait (black, averaged along line D), and downstream of Fram Strait (broken red, averaged along line C). (c) Simulated water volume inflow above 700 m (Sv, black) at Fram Strait (78°50′N, 5°E–8°40′E) are compared with observations (Sv, red; https://www.whoi.edu/page.do?pid=30914).

Fig. 5.

(a) Average water volume transports (Sv) at 100–900-m depth into the central Arctic Ocean from the Kara Sea (green, averaged along line A in Fig. 1), the Barents Sea (red, averaged along line B in Fig. 1), downstream of Fram Strait (broken red, averaged along line C), Fram Strait inflow (black, averaged along line D), and Fram Strait outflow (broken black). The convention is that for A, B, and D, positive is into the Arctic Ocean while negative is out of the Arctic Ocean. For section C, positive indicates counterclockwise boundary flow. (b) Average water temperature (°C) at 100–900-m depth in the Kara Sea (green, averaged along line A in Fig. 1), the Barents Sea (red, averaged along line B in Fig. 1), at Fram Strait (black, averaged along line D), and downstream of Fram Strait (broken red, averaged along line C). (c) Simulated water volume inflow above 700 m (Sv, black) at Fram Strait (78°50′N, 5°E–8°40′E) are compared with observations (Sv, red; https://www.whoi.edu/page.do?pid=30914).

Fram Strait is the only deep water passage for Atlantic water entering the central Arctic Ocean. The northward transport of Atlantic water in the West Spitsbergen Current has been found to range between 3 and 5 Sv, with large variations. Moreover, part of the inflow recirculates within the strait, and only 2–3 Sv is contributed to the eastward boundary current along the Eurasian continental slope (Rudels et al. 2013). In addition, previous model simulations also suggest 2–3 Sv water volume transport into the central Arctic Ocean (Karcher et al. 2003; Aksenov et al. 2010). Here, our simulated inflow of the West Spitsbergen Current by CIOM gives about 4 Sv, whereas the outflow along the east coast of Greenland is about 3 Sv (Fig. 5a). Moreover, CIOM simulations suggest about 2 Sv volume transport at 100–900 m along the Eurasian continental slope in 1980–2000 (Fig. 5a), which is weaker than what the observations and previous model simulations suggest. To further validate the CIOM results, Fig. 5c shows comparisons of simulated water volume transport above 700 m and observations. On average, observations suggest about 4 Sv Atlantic water inflow to the Arctic Ocean via the West Spitsbergen Current; however, the corresponding CIOM simulated water volume transport is about 3.5 Sv and fails to reproduce the increased water volume transport during 1998–2000. Moreover, the AW is warmer downstream of the West Spitsbergen region than in the Barents and Kara Seas. On average, the average temperature at 100–900 m is about 1.3°C across Fram Strait, about 2.2°C downstream of the West Spitsbergen Current region, 1.4°C in the Barents Sea, and 0.8°C for the Kara Sea (Fig. 5b).

After entering the central Arctic Ocean from Fram Strait, AW gradually loses its heat (Fig. 6). The maximum heat content integrated between 0°C surfaces at intermediate layers is located in the Eurasian basin, and the heat content in the Canada basin is relatively small (Fig. 6f). Compared to PHC data (Fig. 6f), CIOM reproduces the basic heat content patterns (Fig. 6a), but overestimates the heat content along the shelf of Eurasian basin and the Siberian shelf by about 1–2 × 109 J m−2 (Fig. 6g).

Fig. 6.

(a)–(e) Annual heat content between AWL 0°C surfaces (1010 J m−2), as indicated in Fig. 4, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. Straight boundaries are chosen in the southeast portion of the map to delineate the central Arctic Ocean boundaries, in estimates of the average. (g) The difference between the simulation (1990–2009) and PHC data (1010 J m−2).

Fig. 6.

(a)–(e) Annual heat content between AWL 0°C surfaces (1010 J m−2), as indicated in Fig. 4, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. Straight boundaries are chosen in the southeast portion of the map to delineate the central Arctic Ocean boundaries, in estimates of the average. (g) The difference between the simulation (1990–2009) and PHC data (1010 J m−2).

Figure 7 shows annual maximum temperature at 100–900 m. Similar to the spatial pattern in Fig. 6, the AW core temperature is warmer in eastern Arctic Ocean than in western Arctic Ocean (Fig. 7f). The core temperature is about 0.8°–2.8°C in the Eurasian basin and 0°–0.8°C in the Canada basin (Fig. 7f). While CIOM is able to reproduce the spatial patterns in PHC data (Fig. 7a), it underestimates the AW core temperature by 0.2°–0.4°C in the Canada basin and by about 0.2°–1°C in the central Eurasian basin (Fig. 7g).

Fig. 7.

(a)–(e) Annual maximum water temperature (°C) at 100–900 m, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. (g) The difference between the simulation (1990–2009) and PHC data (°C).

Fig. 7.

(a)–(e) Annual maximum water temperature (°C) at 100–900 m, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. (g) The difference between the simulation (1990–2009) and PHC data (°C).

Moreover, the AW is relatively shallower in the Eurasian basin than in the Canada basin (Fig. 8). The minimum depth is about 80–280 m in the Eurasian basin, and about 280–400 m in the western Arctic Ocean (Fig. 8f). Compared to PHC data, the CIOM simulates the spatial patterns well (Fig. 8a) but underestimates the minimum depth by about 40 m along the shelves of the Barents and Kara Seas, and also overestimates the minimum depth in the central Eurasian basin and Canada basin by about 50–150 m (Fig. 8g).

Fig. 8.

(a)–(e) Minimum depth of Atlantic water, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. (g) The difference between the simulation (1990–2009) and PHC data. (All units are in meters.)

Fig. 8.

(a)–(e) Minimum depth of Atlantic water, showing CIOM results and (f) PHC data. Bidecadal averages are indicated. (g) The difference between the simulation (1990–2009) and PHC data. (All units are in meters.)

3) Heat balance of Atlantic water in the central Arctic Ocean

The AWL heat fluxes into the central Arctic Ocean are computed as , where A is the lateral area at 100–900 m along the section lines shown in Fig. 1, υ is the current perpendicular to the lines which is positive (negative) if the water flows into (out of) the Arctic Ocean, T is water temperature, and T0 is the reference temperature which is set to 0°C. In this study, the reference temperature depends on the minimum AW temperature in the central Arctic Ocean, which is set to 0°C, following previous studies by Polyakov et al. (2004, 2010). If the inflow Atlantic water temperature is above 0°C, there would be a positive heat flux into the central Arctic Ocean. Moreover, if the outflow has temperature below 0°C, the heat fluxes are also positive, because cold water is transported out of the central Arctic Ocean.

For the present climate, the average AW temperature simulated by CIOM gradually increases from about 0.35°C in the 1990s to 0.37°C in the 2000s (Fig. 9). Meanwhile, the heat content associated with the AW also increases from about 5.4 to 5.9 × 1021 J during this period (Fig. 9). Compared to PHC data, the CIOM simulation underestimates the AW temperature by about 0.1°C, but overestimates the heat content by about 9 × 1020 J (Fig. 9). The discrepancy may be related to the overestimated vertical extent of the AWL, as shown in Fig. 4.

Fig. 9.

Simulated (solid red) and PHC (broken red) water temperature (°C) as well as simulated (solid black) and PHC (broken black) heat content (unit is 1.5 × 1022 J, relative to 0°C), averaged between 0°C surfaces in the central Arctic Ocean.

Fig. 9.

Simulated (solid red) and PHC (broken red) water temperature (°C) as well as simulated (solid black) and PHC (broken black) heat content (unit is 1.5 × 1022 J, relative to 0°C), averaged between 0°C surfaces in the central Arctic Ocean.

The heat fluxes associated with the AWL mainly come from Fram Strait and the Kara Sea (Fig. 10a). In Fram Strait, observations and previous model simulations suggest 20–30 TW heat transport into the central Arctic Ocean (Fig. 10c; Karcher et al. 2003; Aksenov et al. 2010). However, CIOM simulations suggest about 20 TW heat transport at 100–900 m via the West Spitsbergen Current in 1980–2000, which is weaker than the observations and previous model simulations suggest (Fig. 10a). In the Barents and Kara Seas, the observations suggest a wide range of variability, with heat transports ranging between 4 TW southward and 4 TW northward through the St. Anna Trough (Simonsen and Haugan 1996; Schauer et al. 2002), and recent model simulations suggest a net heat transport of 5TW into the central Arctic Ocean from the Kara Sea (Aksenov et al. 2010). By comparison, CIOM simulates a total heat transport of about 3 TW (Fig. 10a), which is within the range of previous observations and model simulations. For the Victoria Channel, previous model simulations suggest a weak heat transport into the Nansen basin from the Barents Sea (Maslowski et al. 2004; Aksenov et al. 2010; Årthun et al. 2011). Similarly, the CIOM simulation shows no significant heat flux from the Barents Sea into the central Arctic Ocean (Fig. 10a). Moreover, the simulated heat fluxes from both Fram Strait and the Kara Sea tend to increase during the decades from the 1990s to the 2000s, which is responsible for the AW warming, as seen in Fig. 9.

Fig. 10.

(a) Heat flux into the central Arctic Ocean from the Kara (green, averaged along line A in Fig. 1) and Barents Seas (red, along line B in Fig. 1) and Fram Strait inflow (black, along line D) and outflow (broken black, along line D), averaged at 100–900 m, relative to 0°C and integrated vertical heat flux averaged between depths of 0°C surface in the central Arctic Ocean (blue indicated as C). (b) Total vertically integrated heat flux rate averaged between 0°C surfaces in the central Arctic Ocean, and (c) comparisons of simulated inflow heat flux above 700 m (black) at Fram Strait (78°50′N, 5°–8°40′E) and observations (red; https://www.whoi.edu/page.do?pid=30914) in (c). (Unit is terawatt.)

Fig. 10.

(a) Heat flux into the central Arctic Ocean from the Kara (green, averaged along line A in Fig. 1) and Barents Seas (red, along line B in Fig. 1) and Fram Strait inflow (black, along line D) and outflow (broken black, along line D), averaged at 100–900 m, relative to 0°C and integrated vertical heat flux averaged between depths of 0°C surface in the central Arctic Ocean (blue indicated as C). (b) Total vertically integrated heat flux rate averaged between 0°C surfaces in the central Arctic Ocean, and (c) comparisons of simulated inflow heat flux above 700 m (black) at Fram Strait (78°50′N, 5°–8°40′E) and observations (red; https://www.whoi.edu/page.do?pid=30914) in (c). (Unit is terawatt.)

The heat influxes into the central Arctic Ocean, from Fram Strait and the Kara Sea, are approximately balanced by the heat loss due to the outflow along the east coast of Greenland and vertical heat transport (Fig. 10a). On average, there is about 20 TW of heat transport into the central Arctic Ocean from Fram Strait and about 3 TW from the Kara Sea, and a similar heat flux is transported out of the central Arctic Ocean through the outflow along the east coast of Greenland and the vertical heat transport (Fig. 10a). In addition, recent observations show that the flux from intermediate-depth AW is about 1 W m−2 in 2009/10 (Polyakov et al. 2013), but the vertically integrated heat loss in the CIOM simulation is about 2 W m−2, which suggests that CIOM overestimates the vertical heat loss of the AW in the central Arctic Ocean (Fig. 10b). Moreover, comparisons of observations with the simulated heat flux above 700 m via the West Spitsbergen Current suggest that CIOM underestimates the heat flux into the central Arctic Ocean (Fig. 10c).

b. Future climate

1) Spatial patterns of water temperature

In the future climate, the CIOM simulation suggests a significant change in the water temperature at 200 m in the central Arctic Ocean, as shown in Fig. 2. In the western Arctic Ocean, while no significant trends in temperature can be seen before 2030 and after 2069, rapid cooling occurs in 2030–69. For example, the area with temperatures below 0°C expands from the Canada basin in 1990–2009 to the Makarov basin in 2050–69. These results are consistent with the decreasing trend in the average water temperature (shown in Fig. 9). In addition, although the water temperature near the West Spitsbergen region tends to increase slightly from 1990 to 2099, it remains largely stable in the Eurasian basin, over this time period. By comparison, changes in water temperature at 450 m are relatively weak (Fig. 3). Although the temperature (0°–0.5°C) remains uniform from 1999 to 2049 in the central Arctic Ocean, it gradually decreases and falls below 0°C in the Beaufort Sea and along the northern coast of the Canadian Archipelago in 2080–99. In addition, no significant changes can be seen in the Greenland Sea.

Under the climate change scenario, model simulations show significant changes in the vertical structure of water temperature in the Arctic Ocean (Fig. 4). In the vertical direction, the warm layer at 200–900 m slightly expands from 1990 to 2029, but shrinks from 2059 to 2099. For example, the 0°C lower limit isosurface is about 900 m in 1990–2009, decreasing to about 1000 m in 2010–29, but then rising to above 750 m in 2080–99. During 2070–99, this warm AWL becomes significantly colder in the Beaufort Sea relative to conditions during 1990–2009. In addition, significant surface warming can be seen in the Barents Sea and the Beaufort Sea. Moreover, the warming along the slopes of the Barents and Beaufort Seas is probably related to the intensified ventilation, which brings more warm AW water to the upper layers. In the Barents Sea, the water temperatures in shallow coastal areas are expected to increase, from the surface to the bottom. However, as suggested by Fig. 4, temperatures in deep waters near Franz Josef Land are estimated to decrease in the future climate scenario.

2) Atlantic water in the central Arctic Ocean

Under the A1B climate change scenario, there is a significant increase in the water volume transport into the central Arctic Ocean through Fram Strait (Fig. 5a). The inflow via the West Spitsbergen Current increases from 4 Sv in 1980s to about 5 Sv in 2050s. Moreover, the simulated transport downstream of Fram Strait shows a steady increase from about 2 Sv in the 1980s to 3 Sv in the 2050s, suggesting an intensified cyclonic boundary current along the slope of the Barents Sea. However, there is no significant trend in the transports after the 2050s along the West Spitsbergen region and downstream of Fram Strait. Meanwhile, the water temperature downstream of Fram Strait increases from about 2.2°C in 1980 to about 3°C in 2090s, but there is no significant trend in the water temperature averaged across Fram Strait (Fig. 5b).

Moreover, a slight increase (about 0.5 Sv) in the water volume transport through the St. Anna Trough can be seen from 1980 to 2020, whereas there is a slight decreasing trend through the Victoria Channel (Fig. 5a). In the Victoria Channel, the water temperature decreases from about 1.4°C in 1980s to about 0.3°C in 2090. The cooling in the St. Anna Trough is relatively weaker, with the average temperature decreasing from about 0.8°C in 1980s to 0.5°C in the 2090s (Fig. 5b).

The increased transport through Fram Strait is associated with the changes in the Arctic atmospheric circulation. In the present climate, there is a low pressure system over the Barents Sea and Nordic seas, shown in Fig. 11, associated with the North Atlantic Oscillation (NAO), which is responsible for transporting warm, moist Atlantic air into the Arctic. During the high NAO events, the low pressure systems over the Barents Sea and Nordic seas are strong, and observations show that an enhanced low pressure system can lead to increased Atlantic water flow into the central Arctic Ocean through Fram Strait and the Barents Sea (Grotefendt et al. 1998; Dickson et al. 2000). In addition, the dominant feature in the western Arctic is an anticyclonic atmospheric system, whereby a surface high is embedded within a pressure ridge, extending from northeastern Eurasia to northwestern Canada (Long and Perrie 2013). Compared to NCEP data, the CRCM simulation overestimates the high pressure system in the western Arctic and underestimates the low pressure system in the Nordic seas (shown in Fig. 11). Model simulations suggest that the strong (weak) high pressure systems in the western Arctic Ocean tend to suppress (enhance) the cyclonic boundary current of the AWL (Karcher et al. 2012). Under our simulation, using the A1B climate change scenario, the low pressure system in the eastern Arctic will become stronger and expand eastward (Fig. 11) while the high pressure system in the western Arctic will become weaker (Long and Perrie 2013). Furthermore, there will be a cyclonic wind stress anomaly over the western and eastern Arctic Ocean in the future climate scenario, which will increase the Atlantic water inflow into the central Arctic Ocean through Fram Strait, as shown in Fig. 5a (see also Karcher et al. 2003, 2007).

Fig. 11.

Average (bidecadal) annual sea level pressure (hPa), comparing NCEP averaged on 1980–99 with results CRCM regional climate model.

Fig. 11.

Average (bidecadal) annual sea level pressure (hPa), comparing NCEP averaged on 1980–99 with results CRCM regional climate model.

In a warmer climate, changes occur in the horizontal patterns of the AWL. In a warmer climate, the main heat loss associated with the AWL occurs in the Eurasian basin (Fig. 6). In the present climate, whereas the heat content averaged between the AWL 0°C surfaces in the eastern central Arctic Ocean ranges from 0.2 to 0.6 × 1010 J m−2 in the Eurasian basin, in the future climate these values slowly decrease. In 2080–99, the expected heat content is relatively uniform over the Arctic at about 0.1–0.2 × 1010 J m−2, except along the shelves of the Barents and Kara Seas (Fig. 6). Similarly, significant decreases in AW core temperatures can be seen in the eastern central Arctic Ocean. In particular, in the Eurasian basin, the average core temperatures decrease by about 0.3°C (Fig. 7). Moreover, the AWL is expected to become shallower in the eastern central Arctic Ocean in the future climate (Fig. 8).

3) Heat balance of Atlantic water in the central Arctic Ocean

For the temperature averaged between the AWL 0°C surfaces, at the intermediate depths in the central Arctic Ocean, no significant trend can be seen in 2060–99; however, the average temperature tends to decrease from 0.36°C in the 2010s to 0.26°C in the 2060s. In addition, there is a significant decadal variation in the average temperature, but it is relatively weak during the period 2010–40, when the decreasing trend is dominant and probably masks the variability. Similarly, the average heat content decreases from 6 × 1021 J in the 2010s to 2.5 × 1021 J in the 2060s; there is no significant trend after 2070.

The cooling of the AWL is mainly associated with decreased heat fluxes from the Kara and Barents Seas and increased vertical heat loss of AW in the central Arctic Ocean (Fig. 10). Consistent with the increases in water volume transport and water temperature (Fig. 5a), the heat flux along the West Spitsbergen region shows an increasing trend from 20 TW in the 1980s increasing to 30 TW by about the 2090s. Meanwhile, the heat fluxes transported out of the central Arctic Ocean along the east coast of Greenland decrease from about −11 TW in the 1980s to about −5 TW in the 2090s (Fig. 10a). However, in the Kara Sea, the heat fluxes tend to decrease after 2010. The heat fluxes decrease from about 3 TW in the 2010s to about −5 TW in the 2090s. In addition, there is a decreasing trend in heat fluxes from the Barents Sea by about 3 TW from the 1980s to the 2090s (Fig. 10a). Moreover, the vertically integrated heat fluxes of AW in the central Arctic Ocean decrease from about −10 TW in the 2010s to about −15 TW in the 2090s (Fig. 10a), and the heat flux rate associated with the vertical heat loss decreases from about −2 TW m−2 in the 1990s to about −4 TW m−2 in the 2090s (Fig. 10b). Therefore, the cooling in the Kara and Barents Sea is a major factor for the AWL cooling in the central Arctic Ocean, and the vertical heat loss of AW also plays an important role in the process.

4. Air–sea interactions in the Kara and northeastern Barents Seas

Sea ice is an important component in the Arctic Ocean. When the Arctic Ocean freezes over, sea ice effectively prevents the heat exchange between the atmosphere and the ocean. In the present climate, the northeastern Barents Sea is covered with sea ice during the Arctic autumn while the southern Barents Sea remains largely open water. Under the A1B scenario, the ice cover in the Barents and Kara Seas will gradually decrease, and the northeastern Barents Sea will become largely ice free in autumns in the 2060s or earlier (Fig. 12; see also Wang and Overland 2009). In autumn, over this time period, the AW is expected to lose most of the heat in the southern Barents Sea, the surface water temperature is near the freezing point, and solar radiation is weak. Therefore, the ice loss in the Kara and northeastern Barents Seas can increase surface water density and enhance vertical mixing. As shown in Fig. 13, there is a significant increase in the surface heat fluxes in the northeastern Barents and Kara Seas. In 1980–99, the heat flux in the northeastern Barents Sea is about 40 W m−2, and it is estimated to almost double in 2080–99, suggesting an increased heat loss of AW (Fig. 13).

Fig. 12.

Ice concentration in the eastern Arctic Ocean in November.

Fig. 12.

Ice concentration in the eastern Arctic Ocean in November.

Fig. 13.

Annual surface heat flux (W m−2).

Fig. 13.

Annual surface heat flux (W m−2).

The increased surface heat fluxes result in significant cooling near the bottom of the northeastern Barents Sea. On average, the water is relatively colder near the surface than at the bottom (Fig. 4) due to the heat loss in the southern Barents Sea. Therefore, the enhanced vertical mixing increases the heat loss and reduces water temperature at the bottom (Figs. 4 and 14). In 1980–99, the water temperature is above 0°C at 150-m depth in the northeastern Barents Sea. However, it gradually decreases in the future climate scenario, falling below an estimated 0°C in 2020–39, which is consistent with the cooling trend near Franz Josef Land (Fig. 4). After the 2080s, the water temperature tends to increase (Fig. 14). It is notable that the water from the northeastern Barents Sea forms about 50% of the total flow to the Kara Sea (Schauer et al. 2002), and therefore, a decrease in water temperature near the St. Anna Trough (Fig. 14) can contribute to reduced heat flux into the Arctic Ocean via the St. Anna Trough.

Fig. 14.

Water temperature (°C) at 150 m.

Fig. 14.

Water temperature (°C) at 150 m.

5. Discussion

To understand the sensitivity of the AW cooling in the future climate to the changes in the Atlantic water inflow, a separate CIOM experiment has been conducted, whereby the Atlantic water inflow was increased from 8.5 to 9.5 Sv, and the outflow along the east Greenland shelf was set at 8 Sv. Figure 15 presents the cross section for annual water temperature (degree Celsius) as a function of depth (meter) along the trans-Arctic line indicated in Fig. 1. Compared to Fig. 4, the enhanced Atlantic water inflow results in only slight increases in the water temperature at the intermediate layers; however, the cooling trends at intermediate depths are quite similar for the two experiments. These results suggest that the overall findings are not very sensitive to slight changes in the water volume transports along the lateral boundaries. Moreover, Fram Strait, the Victoria Channel, and the St. Anna Trough are not resolved in the CGCM3 simulation because of its coarse resolution (about 1.8°). Compared to CIOM simulations and PHC data (Fig. 4), CGM3 is incapable of reproducing the warm AW at intermediate depths, and there is no significant warm water along the Beaufort Sea shelf, suggesting that the ventilation process is not represented in CGCM3 (Fig. 15).

Fig. 15.

Cross section for annual water temperature (°C) as a function of depth (m) along the trans-Arctic line indicated in Fig. 1. Units of x axis are 103 kilometers, showing CIOM and CGCM3 results where the barotropic transport into the Arctic Ocean increased by 1 Sv.

Fig. 15.

Cross section for annual water temperature (°C) as a function of depth (m) along the trans-Arctic line indicated in Fig. 1. Units of x axis are 103 kilometers, showing CIOM and CGCM3 results where the barotropic transport into the Arctic Ocean increased by 1 Sv.

In terms of the impacts of climate change in the Arctic, the main feedback between the atmosphere and the Arctic Ocean is related to the loss of sea ice. In this study, the impacts of sea ice on CRCM simulations mainly depend on the global climate model simulations as provided by CGCM3. As mentioned in section 2, CGCM3 is a coupled global climate model and has the parameterizations to represent these processes. Moreover, the CRCM simulation is driven by the CGCM3 outputs, specifically, at the lateral boundaries and the specification of the surface conditions. Therefore, the overall feedbacks are largely included in the CRCM results. On small scales, although there are still some discrepancies in the sea ice as simulated by CIOM, compared to that of CGCM3, their impacts are relatively weak.

The intermediate layers of the Arctic Ocean are mainly ventilated through boundary convection processes that transport brine-enriched water formed over the shelf to the deep central basins along the continental slope, entraining the surrounding water during these processes (Tanhua et al. 2009). In the southern Beaufort Sea, these processes bring heat from the AWL to the surface along the Beaufort Sea slope regions. Compared to the PHC data, the CIOM simulations clearly underestimate the ventilation process (Fig. 4), which can therefore be a cause for its overestimation of the AW temperature in the Canada basin (Figs. 2 and 3). However, under A1B scenario, the ventilation along the Beaufort Sea slope tends to intensify, toward the end of the simulation, by 2080–99, which helps to cool the water temperatures at the intermediate depths (Figs. 2 and 3).

In terms of salinity, under the climate change scenario, the enhanced sea ice melting and Ekman transport result in an increasing trend in freshwater accumulation in the Beaufort Sea, as shown in Fig. 16 and Long and Perrie (2013). Climatologically, Ekman convergence associated with the Beaufort Gyre transports cold, fresh surface water into the Beaufort Sea, maintaining a salinity minimum in the region. However, CIOM underestimates the salinity minimum compared to PHC data. This result suggests possible weak Ekman convergence, which might also contribute to the overestimated water temperatures in Canada basin (Figs. 2 and 3). In the future climate scenario, the Ekman convergence will intensify, which can enhance cooling in the Beaufort Sea (Long and Perrie 2013). Moreover, the salinity minimum tends to intensify from the 1990s to the 2060s but becomes weaker after the 2080s (Fig. 16).

Fig. 16.

As in Fig. 4, but for water salinity (psu).

Fig. 16.

As in Fig. 4, but for water salinity (psu).

6. Conclusions

A coupled ice–ocean model (CIOM) is implemented in the Arctic Ocean to simulate the impacts of climate change on the water temperature associated with the Atlantic water (AW). Driven by the output from the Canadian Regional Climate Model (CRCM), CIOM was integrated from 1970 to 2099, following the A1B climate change scenario (Solomon et al. 2007). Compared to PHC data, CIOM is shown able to capture the warm AWL, approximately defined at 200–900 m, showing a three-layer system in the central Arctic Ocean. In addition, we show that the observed water volume transports into the central Arctic Ocean from Fram Strait and the Kara and Barents Seas are reasonably reproduced in CIOM simulations.

Under the AIB climate change scenario, the water temperature associated with the AWL has a decreasing trend. While there is no significant trend in 1980–2020 and 2060–99, the average temperature between 0°C surfaces tends to decrease from 0.36°C in the 2010s to 0.26°C in the 2060s, and the most significant cooling occurs in the western Arctic Ocean. Although the water temperature near the West Spitsbergen region tends to increase slightly from 1990 to 2099, it remains largely stable in the Eurasian basin during this time period. Vertically, within the water column, the warm AWL appears to slightly expand during the period from 2009 to 2029 and to significantly shrink from 2059 to 2099. Moreover, there appears to be no significant warm layer in the Beaufort Sea in the period 2070–99. In addition, there is a significant increase in the water volume transport into the central Arctic Ocean through Fram Strait, from the 1980s to the 2050s, due to the weakened atmospheric high pressure system in the western Arctic and the intensified atmospheric low pressure system in the Nordic seas, in the future climate scenario.

Loss of sea ice in the northeastern Barents Sea also has impacts on the heat balance in the central Arctic Ocean. Climatologically, until the most recent decade, the northeastern Barents Sea has been covered with sea ice. However, according to the A1B climate change scenario, the ice cover will gradually decrease in the future climate, and the northeastern Barents Sea will become largely ice free during the autumn by the 2060s, if not earlier. Moreover, the loss of sea ice will increase the heat loss of Atlantic water and reduce the water temperature near the bottom of the northeastern Barents Sea in autumn seasons. For example, in the simulation described here, the water temperature at 150-m depth in the northeastern Barents Sea is above 0°C in 1980–99 but gradually decreases in future decades. By 2020–39, the estimated water temperature falls below 0°C, whereas by 2080–99 it tends to increase. Consistent with the decreased water temperature in the Kara and Barents Seas, there are significant decreases in the heat fluxes into the central Arctic Ocean, which are mainly responsible for the AWL cooling in the central Arctic Ocean. In addition, the vertically integrated heat loss of the AWL also plays an important role in the AWL cooling. The integrated vertical heat fluxes decrease from about −10 TW in the 2010s to about −15 TW in the 2090s, and the associated heat flux rate decreases from about −2 TW m−2 in 1990s to about −4 TW m−2 in the 2090s.

Acknowledgments

Support for this research comes from the DFO’s Aquatic Climate Change Adaptation Service Program (ACCASP) Initiative. The authors thank three anonymous reviewers for very helpful comments.

REFERENCES

REFERENCES
Aagaard
,
K.
,
1989
:
A synthesis of the Arctic Ocean circulation
.
Rapp. P.-V. Reun. Cons. Int. Explor. Mer
,
188
,
11
22
.
Aksenov
,
Y.
,
S.
Bacon
,
A. C.
Coward
, and
A. J. G.
Nurser
,
2010
:
The North Atlantic inflow to the Arctic Ocean: High-resolution model study
.
J. Mar. Syst.
,
79
,
1
22
, doi:.
Årthun
,
M.
,
R. B.
Ingvaldsen
,
L. H.
Smedsrud
, and
C.
Schrum
,
2011
:
Dense water formation and circulation in the Barents Sea
.
Deep-Sea Res. I
,
58
,
801
817
, doi:.
Beszczynska-Möller
,
A.
,
R. A.
Woodgate
,
C.
Lee
,
H.
Melling
, and
M.
Karcher
,
2011
:
A synthesis of exchanges through the main oceanic gateways to the Arctic Ocean
.
Oceanography
,
24
,
82
99
, doi:.
Beszczynska-Möller
,
A.
,
E.
Fahrbach
,
U.
Schauer
, and
E.
Hansen
,
2012
:
Variability in Atlantic water temperature and transport at the entrance to the Arctic Ocean, 1997–2010
.
ICES J. Mar. Sci.
,
69
,
852
863
, doi:.
Blindheim
,
J.
,
1989
:
Cascading of Barents Sea bottom water into the Norwegian Sea
.
Rapp. P.-V. Reun. Cons. Int. Explor. Mer
,
188
,
49
58
.
Blumberg
,
A. F.
, and
G. L.
Mellor
,
1987
: A description of a three-dimensional coastal ocean circulation model. Three-Dimensional Coastal Ocean Models, N. S. Heaps, Ed., Amer. Geophys. Union, 1–16.
Bourgain
,
P.
, and
J. C.
Gascard
,
2012
:
The Atlantic and summer Pacific waters variability in the Arctic Ocean from 1997 to 2008
.
Geophys. Res. Lett.
,
39
, L05603, doi:.
Carmack
,
E. C.
,
R. W.
Macdonald
,
R. G.
Perkin
,
F. A.
McLaughlin
, and
R. J.
Pearson
,
1995
:
Evidence for warming of Atlantic water in the southern Canadian basin of the Arctic Ocean: Results from the Larsen-93 expedition
.
Geophys. Res. Lett.
,
22
,
1061
1064
, doi:.
Carmack
,
E. C.
, and Coauthors
,
1997
:
Changes in temperature and tracer distributions within the Arctic Ocean: Results from the 1994 Arctic Ocean section
.
Deep-Sea Res.
,
44B
,
1487
1502
.
Deser
,
C.
,
R.
Tomas
,
M.
Alexander
, and
D.
Lawrence
,
2010
:
The seasonal atmospheric response to projected Arctic sea ice loss in the late twenty-first century
.
J. Climate
,
23
,
333
351
, doi:.
Dickson
,
R.
, and Coauthors
,
2000
:
The Arctic Ocean response to the North Atlantic Oscillation
.
J. Climate
,
13
,
2671
2696
, doi:.
Dmitrenko
,
I. A.
, and Coauthors
,
2008
:
Toward a warmer Arctic Ocean: Spreading of the early 21st century Atlantic Water warm anomaly along the Eurasian basin margins
.
J. Geophys. Res.
,
113
, C05023, doi:.
Dmitrenko
,
I. A.
, and Coauthors
,
2010
:
Impact of the Arctic Ocean Atlantic water layer on Siberian shelf hydrography
.
J. Geophys. Res.
,
115
, C08010, doi:.
Fer
,
I.
,
2009
:
Weak vertical diffusion allows maintenance of cold halocline in the central Arctic
.
Atmos. Oceanic Sci. Lett.
,
2
,
148
152
.
Grotefendt
,
K.
,
K.
Logemann
,
D.
Quadfasel
, and
S.
Ronski
,
1998
:
Is the Arctic Ocean warming?
J. Geophys. Res.
,
103
,
27 679
27 687
, doi:.
Hibler
,
W. D.
, III
,
1979
:
A dynamic thermodynamic sea ice model
.
J. Phys. Oceanogr.
,
9
,
815
846
, doi:.
Hibler
,
W. D.
, III
,
1980
:
Modeling a variable thickness sea ice cover
.
Mon. Wea. Rev.
,
108
,
1943
1973
, doi:.
Holliday
,
N. P.
, and Coauthors
,
2008
:
Reversal of the 1960s to 1990s freshening trend in the northeast North Atlantic and Nordic seas
.
Geophys. Res. Lett.
,
35
, L03614, doi:.
Holloway
,
G.
, and
Z.
Wang
,
2009
:
Representing eddy stress in an Arctic Ocean model
.
J. Geophys. Res.
,
114
, C06020, doi:.
Jahn
,
A.
, and
M. M.
Holland
,
2013
:
Implications of Arctic sea ice changes for North Atlantic deep convection and the meridional overturning circulation in CCSM4-CMIP5 simulations
.
Geophys. Res. Lett.
,
40
,
1206
1211
, doi:.
Karcher
,
M.
, and
J. M.
Oberhuber
,
2002
:
Pathways and modification of the upper and intermediate waters of the Arctic Ocean
.
J. Geophys. Res.
,
107
, 3049, doi:.
Karcher
,
M.
,
R.
Gerdes
,
F.
Kauker
, and
C.
Köberle
,
2003
:
Arctic warming: Evolution and spreading of the 1990s warm event in the Nordic seas and the Arctic Ocean
.
J. Geophys. Res.
,
108
, 3034, doi:.
Karcher
,
M.
,
F.
Kauker
,
R.
Gerdes
,
E.
Hunke
, and
J.
Zhang
,
2007
:
On the dynamics of Atlantic Water circulation in the Arctic Ocean
.
J. Geophys. Res.
,
112
, C04S02, doi:.
Karcher
,
M.
,
A.
Beszczynska-Möller
,
F.
Kauker
,
R.
Gerdes
,
S.
Heyen
,
B.
Rudels
, and
U.
Schauer
,
2011
:
Arctic Ocean warming and its consequences for the Denmark Strait overflow
.
J. Geophys. Res.
,
116
, C02037, doi:.
Karcher
,
M.
,
J. N.
Smith
,
F.
Kauker
,
R.
Gerdes
, and
W. M.
Smethie
Jr.
,
2012
:
Recent changes in Arctic Ocean circulation revealed by iodine-129 observations and modeling
.
J. Geophys. Res.
,
117
, C08007, doi:.
Koenigk
,
T.
, and
L.
Brodeau
,
2013
:
Ocean heat transport into the Arctic in the twentieth and twenty-first century in EC-Earth
.
Climate Dyn.
,
42
,
3101
3120
, doi:.
Koenigk
,
T.
,
L.
Brodeau
,
R. G.
Graversen
,
J.
Karlsson
,
G.
Svensson
,
M.
Tjernström
,
U.
Willén
, and
K.
Wyser
,
2013
:
Arctic climate change in 21st century CMIP5 simulations with EC-Earth
.
Climate Dyn.
,
40
,
2719
2743
, doi:.
Køltzow
,
M.
,
2007
:
The effect of a new snow and sea ice albedo scheme on regional climate model simulations
.
J. Geophys. Res.
,
112
, D07110, doi:.
Long
,
Z.
, and
W.
Perrie
,
2013
:
Impacts of climate change on fresh water content and sea surface height in the Beaufort Sea
.
Ocean Modell.
,
71
,
127
139
, doi:.
Long
,
Z.
,
W.
Perrie
,
C. L.
Tang
,
E.
Dunlap
, and
J.
Wang
,
2012
:
Simulated interannual variations of freshwater content and sea surface height in the Beaufort Sea
.
J. Climate
,
25
,
1079
1095
, doi:.
Maslowski
,
W.
,
D.
Marble
,
W.
Walczowski
,
U.
Schauer
,
J. L.
Clement
, and
A. J.
Semtner
,
2004
:
On climatological mass, heat, and salt transports through the Barents Sea and Fram Strait from a pan-Arctic coupled ice-ocean model simulation
.
J. Geophys. Res.
,
109
, C03032, doi:.
McLaughlin
,
F.
,
E. C.
Carmack
,
W. J.
Williams
,
S.
Zimmermann
,
K.
Shimada
, and
M.
Itoh
,
2009
:
Joint effects of boundary currents and thermohaline intrusions on the warming of Atlantic water in the Canada basin, 1993–2007
.
J. Geophys. Res.
,
114
, C00A12, doi:.
McLaughlin
,
F.
,
E.
Carmack
,
A.
Proshutinsky
,
R. A.
Krishfield
,
C.
Guay
,
M.
Yamamoto-Kawai
,
J. M.
Jackson
, and
B.
Williams
,
2011
:
The rapid response of the Canada basin to climate forcing: From bellwether to alarm bells
.
Oceanography
,
24
,
146
159
, doi:.
Mellor
,
G. L.
, and
T.
Yamada
,
1982
:
Development of a turbulence closure model for geophysical fluid problems
.
Rev. Geophys.
,
20
,
851
875
, doi:.
Mellor
,
G. L.
, and
L. H.
Kantha
,
1989
:
An ice-ocean coupled model
.
J. Geophys. Res.
,
94
,
10 937
10 954
, doi:.
Mellor
,
G. L.
,
T.
Ezer
, and
L. Y.
Oey
,
1994
:
The pressure gradient conundrum of sigma coordinate ocean models
.
J. Atmos. Oceanic Technol.
,
11
,
1126
1134
, doi:.
Polyakov
,
I. V.
, and Coauthors
,
2004
:
Variability of the intermediate Atlantic water of the Arctic Ocean over the last 100 years
.
J. Climate
,
17
,
4485
4497
, doi:.
Polyakov
,
I. V.
, and Coauthors
,
2005
:
One more step toward a warmer Arctic
.
Geophys. Res. Lett.
,
32
, L17605, doi:.
Polyakov
,
I. V.
, and Coauthors
,
2010
:
Arctic Ocean warming contributes to reduced polar ice cap
.
J. Phys. Oceanogr.
,
40
,
2743
2756
, doi:.
Polyakov
,
I. V.
,
A. V.
Pnyushkov
,
R.
Rember
,
L.
Padman
,
E. C.
Carmack
, and
J. M.
Jackson
,
2013
:
Winter convection transports Atlantic water heat to the surface layer in the eastern Arctic Ocean
.
J. Phys. Oceanogr.
,
43
,
2142
2155
, doi:.
Prange
,
M.
, and
G.
Lohmann
,
2004
: Variable freshwater input to the Arctic Ocean during the Holocene: Implications for large-scale ocean-sea ice dynamics as simulated by a circulation model. The Climate in Historical Times: Towards a Synthesis of Holocene Proxy Data and Climate Models, H. Fischer et al., Eds., Springer, 319–335.
Prinsenberg
,
S.
, and
I. K.
Peterson
,
2002
:
Variations in air-ice drag coefficient due to ice surface roughness
.
Int. J. Offshore Polar Eng.
,
12
,
121
125
.
Reed
,
R. K.
,
1977
:
On estimating insolation over the ocean
.
J. Phys. Oceanogr.
,
7
,
482
485
, doi:.
Rudels
,
B.
,
E. P.
Jones
,
L. G.
Anderson
, and
G.
Kattner
,
1994
: On the intermediate depth waters of the Arctic Ocean. The Polar Oceans and Their Role in Shaping the Global Environment, Geophys. Monogr., Vol. 85, Amer. Geophys. Union, 33–46, doi:.
Rudels
,
B.
,
U.
Schauer
,
G.
Björk
,
M.
Korhonen
,
S.
Pisarev
,
B.
Rabe
, and
A.
Wisotzki
,
2013
:
Observations of water masses and circulation with focus on the Eurasian basin of the Arctic Ocean from the 1990s to the late 2000s
.
Ocean Sci.
,
9
,
147
169
, doi:.
Schauer
,
U.
,
R. D.
Muench
,
B.
Rudels
, and
L.
Timokhov
,
1997
:
Impact of eastern Arctic shelf waters on the Nansen basin intermediate layers
.
J. Geophys. Res.
,
102
,
3371
3382
, doi:.
Schauer
,
U.
,
H.
Loeng
,
B.
Rudels
,
V. K.
Ozhigin
, and
W.
Dieck
,
2002
:
Atlantic water flow through the Barents and Kara Seas
.
Deep-Sea Res. I
,
49
,
2281
2298
, doi:.
Serreze
,
M. C.
,
A. P.
Barrett
,
A. G.
Slater
,
M.
Steele
,
J.
Zhang
, and
K. E.
Trenberth
,
2007
:
The large-scale energy budget of the Arctic
.
J. Geophys. Res.
,
112
, D11122, doi:.
Shine
,
K. P.
, and
R. G.
Crane
,
1984
:
The sensitivity of a one-dimensional thermodynamic sea ice model to changes in cloudiness
.
J. Geophys. Res.
,
89
,
10 615
10 622
, doi:.
Shirasawa
,
K.
, and
R. G.
Ingram
,
1991
:
Characteristics of the turbulent oceanic boundary layer under sea ice. Part 1: A review of the ice-ocean boundary layer
.
J. Mar. Syst.
,
2
,
153
160
, doi:.
Simonsen
,
K.
, and
P.
Haugan
,
1996
:
Heat budgets of the Arctic Mediterranean and sea surface heat flux parameterizations for the Nordic seas
.
J. Geophys. Res.
,
101
,
6553
6576
, doi:.
Skagseth
,
Ø.
,
K. F.
Drinkwater
, and
E.
Terrile
,
2011
:
Wind- and buoyancy-induced transport of the Norwegian coastal current in the Barents Sea
.
J. Geophys. Res.
,
116
, C08007, doi:.
Smith
,
S. D.
, and
F. W.
Dobson
,
1984
:
The heat budget at ocean weather station Bravo
.
Atmos.–Ocean
,
22
,
1
22
, doi:.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
Z.
Chen
,
M.
Marquis
,
K.
Averyt
,
M.
Tignor
, and
H. L.
Miller
Jr.
, Eds.,
2007
: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.
Steele
,
M.
,
R.
Morley
, and
W.
Ermold
,
2001
:
PHC: A global ocean hydrography with a high-quality Arctic Ocean
.
J. Climate
,
14
,
2079
2087
, doi:.
Tanhua
,
T.
,
E. P.
Jones
,
E.
Jeansson
,
S.
Jutterström
,
W. M.
Smethie
Jr.
,
D. W. R.
Wallace
, and
L. G.
Anderson
,
2009
:
Ventilation of the Arctic Ocean: Mean ages and inventories of anthropogenic CO2 and CFC-11
.
J. Geophys. Res.
,
114
, C01002, doi:.
Thorndike
,
A. S.
,
D. A.
Rothrock
,
G. A.
Maykut
, and
R.
Colony
,
1975
:
The thickness distribution of sea ice
.
J. Geophys. Res.
,
80
,
4501
4513
, doi:.
Vavrus
,
S. J.
,
M. M.
Holland
,
A.
Jahn
,
D. A.
Bailey
, and
B. A.
Blazey
,
2012
:
Twenty-first-century Arctic climate change in CCSM4
.
J. Climate
,
25
,
2696
2710
, doi:.
Wang
,
J.
,
Q.
Liu
,
M.
Jin
,
M.
Ikeda
, and
F. J.
Saucier
,
2005
:
A coupled ice-ocean model in the pan-Arctic and North Atlantic Ocean: Simulation of seasonal cycles
.
J. Oceanogr.
,
61
,
213
233
, doi:.
Wang
,
M.
, and
J. E.
Overland
,
2009
:
A sea ice free summer Arctic within 30 years?
Geophys. Res. Lett.
,
36
, L07502, doi:.
Woodgate
,
R. A.
,
K.
Aagaard
,
R. D.
Muench
,
J.
Gunn
,
G.
Björk
,
B.
Rudels
,
A. T.
Roach
, and
U.
Schauer
,
2001
:
The Arctic Ocean boundary current along the Eurasian slope and the adjacent Lomonosov Ridge: Water mass properties, transports and transformations from moored instruments
.
Deep-Sea Res. I
,
48
,
1757
1792
, doi:.
Wu
,
P.
,
R.
Wood
, and
P.
Stott
,
2005
:
Human influence on increasing Arctic river discharges
.
Geophys. Res. Lett.
,
32
, L02703, doi:.
Yao
,
T.
,
C. L.
Tang
, and
I. K.
Peterson
,
2000
:
Modeling the seasonal variation of sea ice in the Labrador Sea with a coupled multicategory ice model and the Princeton ocean model
.
J. Geophys. Res.
,
105
,
1153
1165
, doi:.