1. Introduction
In the present climate regime, the warm and salty Atlantic water that enters the Nordic (Greenland–Iceland–Norwegian and Barents) seas across the Greenland–Scotland Ridge in the upper branch of the Atlantic meridional overturning circulation maintains the winter Arctic sea ice edge exceptionally far north relative to other sub-Arctic seas. It streams toward the Arctic Ocean through the shallow Barents Sea and deep Fram Strait, and partly recirculates southward in the Greenland Sea (Fig. 1a). Strong air–sea interactions that occur in the Nordic seas area (e.g., Simonsen and Haugan 1996) lead to high ocean buoyancy losses and convert the Atlantic water into a dense water that feeds the sinking branch of the Atlantic meridional overturning circulation through overflows across the Greenland–Scotland Ridge. These processes are regulated by the heat and salt fluxes from the south and by the freshwater and sea ice export from the Arctic Ocean, and are further modified by local sea ice formation in winter (e.g., Aagaard and Carmack 1989).

(a) Bathymetry (102 m) in the Nordic seas area and (b) the summer (June–September) index of AWT variability in BSO area (box in Fig. 1a) in the period 1982–2005 (circles) and the following winter (December–March) index of Nordic seas SIA variability based on SIC data integrated over the full domain of Fig. 1a (squares). In Fig. 1a, shading denotes the area with the long-term mean of winter SIC above 15%, the arrows depict major pathways of Atlantic water through the Nordic seas, and HTR and WNZC stand for the Hopen Trench Recirculation and the West Novaya Zemlya Current, respectively.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

(a) Bathymetry (102 m) in the Nordic seas area and (b) the summer (June–September) index of AWT variability in BSO area (box in Fig. 1a) in the period 1982–2005 (circles) and the following winter (December–March) index of Nordic seas SIA variability based on SIC data integrated over the full domain of Fig. 1a (squares). In Fig. 1a, shading denotes the area with the long-term mean of winter SIC above 15%, the arrows depict major pathways of Atlantic water through the Nordic seas, and HTR and WNZC stand for the Hopen Trench Recirculation and the West Novaya Zemlya Current, respectively.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
(a) Bathymetry (102 m) in the Nordic seas area and (b) the summer (June–September) index of AWT variability in BSO area (box in Fig. 1a) in the period 1982–2005 (circles) and the following winter (December–March) index of Nordic seas SIA variability based on SIC data integrated over the full domain of Fig. 1a (squares). In Fig. 1a, shading denotes the area with the long-term mean of winter SIC above 15%, the arrows depict major pathways of Atlantic water through the Nordic seas, and HTR and WNZC stand for the Hopen Trench Recirculation and the West Novaya Zemlya Current, respectively.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
The inflow of Atlantic water to the Nordic seas varies in time (e.g., Zhang et al. 2004), as does the seasonal sea ice extent in the Nordic seas. Observational records of the sea ice concentration (SIC) from recent decades exhibit strong interannual-to-decadal variability superimposed on a long-term decreasing trend of total sea ice area (SIA) in the Nordic seas in all seasons (Kvingedal 2005). While the summertime SIC anomalies strongly influence the transfer of solar energy to the ocean, the wintertime SIC anomalies strongly affect the turbulent surface heat flux (SHF). Consequently, the wintertime sea ice variations in the Nordic seas are associated with large surface air temperature (SAT) anomalies, and presumably induce a dynamic atmospheric perturbation in the marginal ice zone (MIZ) and adjacent areas (e.g., Wu et al. 2004). Interpretation of these links is difficult, however, as sea ice responds to changing conditions in both the atmosphere and ocean.
Forcing of sea ice variability in the Nordic seas by large-scale atmospheric teleconnections is suggested by simultaneous SIC variations in the entire Arctic region. In particular, a dipole of one polarity anomalies in the Nordic seas and opposite polarity anomalies in the Labrador Sea is often attributed to the North Atlantic Oscillation (NAO), the most prominent mode of wintertime extratropical atmospheric variability in the Northern Hemisphere (e.g., Deser et al. 2000; Ukita et al. 2007). However, the NAO’s centers of action are nonstationary (e.g., Zhang et al. 2008) and, consequently, there is a strong time dependence of the NAO influence on the sea ice extent in the Nordic seas (e.g., Vinje 2001; Strong and Magnusdottir 2010). Of course, this dependence does not invalidate the paradigm of atmospheric forcing of the Arctic climate variability, which may be shaped by fortuitous phasing of the NAO and other intrinsic patterns of atmospheric variability (e.g., Overland et al. 2008). In particular, the interannual variability of the wintertime sea ice extent in the Barents Sea over recent decades was not strongly linked to the NAO but was strongly affected by local wind anomalies (Sorteberg and Kvingedal 2006). Oceanic heat anomalies coming from the south could also contribute to this variability (e.g., Francis and Hunter 2007). Moreover, a positive feedback mechanism may exist in the Arctic climate system in which a key role is played by anomalous wind-driven inflow of Atlantic water to the Barents Sea. This anomalous inflow should affect the sea ice cover and consequently generate SHF anomalies in the Barents Sea MIZ, which then drive atmospheric temperature anomalies and sustain the anomalous winds (e.g., Ikeda 1990; Mysak and Venegas 1998; Bengtsson et al. 2004).
Feedbacks in the Arctic climate system may provide some predictability of atmospheric conditions in the Arctic region and beyond. In particular, recent studies indicate that autumn-to-winter patterns of large-scale atmospheric variability can “remember” summer Arctic sea ice extent anomalies, which precondition the sea ice formation and SHF anomalies at the onset of the cooling season (e.g., Francis et al. 2009; Overland and Wang 2010). The autumn-to-winter atmospheric conditions should also remember earlier oceanic heat anomalies, at least in the Nordic seas area. This is suggested by a strong connection of recent variability in the wintertime Nordic seas SIA to the previous summer’s anomalies of Atlantic water temperature (AWT) in the Barents Sea Opening (BSO) area (Fig. 1a, box) reported by Schlichtholz (2011). The summer’s AWT anomalies in the BSO area are a good predictor of the coherent wintertime sea ice variability in the Barents and Greenland seas as they do not only incorporate signals coming from the south, but also contain information on air–sea interactions occurring on the eastern and western sides of the Nordic seas at the end of the previous winter (Schlichtholz and Houssais 2011). These interactions should generate persistent oceanic heat anomalies that influence the sea ice extent in the following winter (Schlichtholz 2011).
Namias and Born (1970, 1974) analyzed the temporal coherence of sea surface temperature (SST) anomalies in the North Pacific and came up with the idea that autumn-to-winter surface reemergence of thermal anomalies stored in the previous winter-to-spring mixed layer and persisting at depth through summer could be useful in seasonal climate prediction. Alexander and Deser (1995) were the first to show that the process of oceanic reemergence exists in observations of vertical temperature profiles from the North Pacific and North Atlantic. The reemergence of SST anomalies in northern midlatitudes was then investigated in more detail (e.g., Deser et al. 2003) and shown to contribute significantly, albeit modestly, to the winter-to-winter persistence of the NAO (e.g., Kushnir et al. 2006; Cassou et al. 2007). Here, based on lead–lag relationships between observed oceanic variables and atmospheric reanalysis data, we provide evidence for a strong impact of oceanic heat anomalies, mainly reemerging SST anomalies, on the wintertime surface atmospheric variability in the Nordic seas area.
The study is organized as follows. Data and methods are described in section 2. Then, in section 3, general features of the recent wintertime climate variability in the Nordic seas area are presented. The thermodynamic and dynamic atmospheric response to oceanic heat anomalies in the Nordic seas is investigated in section 4 based on links of wintertime SAT and wind anomalies to the previous summer AWT anomalies in the BSO area. Next, in section 5, mechanisms that make the wintertime atmospheric response persistent are elucidated based on the heat budget of the ocean surface mixed layer. Then, in section 6, the role of reemerging SST anomalies in the climate variability in the Nordic seas area is investigated. Concluding remarks are given in section 7.
2. Data and methods


Links between anomalies of the seasonal-mean variables are investigated using the linear regression analysis. The calculations are made for linearly detrended time series in the period 1982–2006 unless stated differently. Fields of anomalies are regressed onto normalized indices of climate variability defined as the anomalies of selected variables divided by their standard deviations. Some indices are constructed for preselected seasons while some other indices used in the time-lagged correlation analysis are constructed for overlapping 4-month-long seasons with an interval of 1 month. Basic seasons are referred to as summer (June–September), autumn (September–December), winter (December–March), and spring (March–June). The seasons starting 1 month earlier and later than the basic seasons are referred to as “early” and “late,” respectively (e.g., early winter for the November–February season and late winter for the January–April season, etc.). All indices are introduced below and summarized in Table 1.
Seasonal indices of climate variability in the Nordic seas area used in the present study. The indices are defined as normalized anomalies of key variables constructed either for preselected 4-month-long seasons or for overlapping 4-month-long seasons with an interval of 1 month. The sign convention adopted for the OIW, ZW, open water SHF, and MIZ Ekman suction indices is that their positive values correspond to geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack, westerly geostrophic wind anomalies in the BSO area, anomalous heat gain by the atmosphere in the Hopen Trench area, and anomalous divergence of the average ocean surface Ekman transport in the MIZ area north of 75°N, respectively. See the text for more details.


Subsurface ocean heat anomalies are characterized by the summer AWT index constructed by Schlichtholz and Houssais (2011) for the period 1982–2005 (Fig. 1b, circles) using temperature data from the International Council for the Exploration of the Sea Oceanographic Database (http://ocean.ices.dk/) and from NOAA’s National Oceanographic Data Center (http://www.nodc.noaa.gov/). The temperature data were averaged over the Atlantic water core (100–300 m) in the BSO area (70°–76°N, 13°–17°E; box in Fig. 1a). Ocean surface temperature anomalies are described by the “open water SST” index based on the SST data averaged over a broader area in the zone of Atlantic water flow toward the Barents Sea and Fram Strait (70°–74°N, 10°–25°E; box in Fig. 9b, figure is discussed in greater detail below). Air–sea interactions on the open water side of the Barents Sea ice edge are characterized by the “open water SHF” index constructed from the SHF anomalies in the Hopen Trench area at about 75°N, 32°E (cross in Fig. 8d, figure is discussed in greater detail below). This is an area with the largest long-term mean winter SHF over the Barents Sea, reaching ~350 W m−2 in the NCEP–NCAR reanalysis (Schlichtholz and Houssais 2011). Therefore, negative (downward) wintertime values of the open water SHF index correspond to a local reduction of the ocean surface heat loss to the atmosphere in the Hopen Trench area.
Regional variability of SIC and SAT is described by the “Nordic seas SIA” and “Nordic seas SAT” indices obtained from the integrated SIC and averaged SAT data over the full domain of Fig. 1a (65°–82°N, 25°W–60°E), respectively. The indices constructed from the SAT data averaged separately over the eastern and western parts of the Nordic seas, east and west of 15°E, are referred to as the “Barents Sea SAT” and the “Greenland Sea SAT” indices, respectively. In addition, the SIC data are integrated over the entire Barents Sea and its eastern and western parts (east and west of 45°E) to obtain the “Barents Sea SIA,” “eastern Barents Sea SIA,” and “western Barents Sea SIA” indices, respectively.
Regional wind variability is investigated using the “on ice wind” (OIW) index calculated following Schlichtholz and Houssais (2011) from the SLP difference between the southern tip of Novaya Zemlya (70°N, 57.5°E) and the southern tip of Spitsbergen (77.5°N, 17.5°E) (crosses in Fig. 3a, figure is discussed in greater detail below). Positive values of the OIW index correspond to geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack. An index characterizing the strength of geostrophic westerlies into the Barents Sea, referred to as the zonal wind (ZW) index, is obtained from the SLP difference between the points at 70° and 75°N on the 17.5°E meridian (crosses in Fig. 4a, figure is discussed in greater detail below). This index corresponds approximately to the index introduced by Bengtsson et al. (2004) as the SLP difference between northernmost Norwegian coast and Spitsbergen. An index of wind variability in the MIZ area, referred to as the “MIZ vorticity” index, is obtained from ζa averaged over an area north of 75°N (75°–83°N, 25°W–60°E; box in Fig. 7a, figure is discussed in greater detail below). The index based on wE averaged over the same area is referred to as the “MIZ Ekman suction” index. The large-scale atmospheric variability is characterized by the winter NAO index constructed by Hurrell (1995) as the SLP difference between Lisbon (Portugal) and Stykkisholmur (Iceland). The index was updated (from http://www.cgd.ucar.edu/cas/jhurrell/), and then detrended and renormalized for the period under study.
The statistical significance of all correlations r is assessed using a two-tailed t test carried out with an effective number of degrees of freedom based on the same formula as used by Schlichtholz and Houssais (2011). Statistical field significance is determined following the procedure introduced by Livezey and Chen (1983). First, local correlations of a given scalar field F with the given index of climate variability I are obtained. Then, the total area AFI where these correlations are statistically significant at the 95% confidence level is calculated for a given mapping domain. Next, the corresponding area AFN is obtained for 500 Monte Carlo trials in which the index I is replaced with the Gaussian noise N. Finally, the percent s of the Monte Carlo trials with AFN < AFI is found. The larger s is, the lower the probability is that the observed pattern of significant correlations between F and I occurred by chance. The field significance parameter (i.e., s) is also calculated for the zonal sλ and meridional sφ components of the vector quantities.
3. Wintertime surface climate variability in the Nordic seas area
In the period under study, the winter SAT anomalies regressed onto the concurrent (nonlagged) NAO index are statistically significant (at the 95% confidence level) over broad areas in the extratropical Northern Hemisphere (Fig. 2a). However, they are not significant in the Nordic seas region (Fig. 2a, box) in spite of a strong NAO-associated anomalous surface wind circulation in this region (Fig. 2b, arrows). Over the Barents Sea, the NAO-associated surface wind anomalies generally follow the isotherms of the mean SAT front along the ice edge (Schlichtholz and Houssais 2011), so that the corresponding anomalous heat advection is small and cannot maintain significant SAT anomalies. On the western side of the Nordic seas, the NAO-associated meridional wind anomalies should result in significant anomalous atmospheric heat advection as well as onshore ice drift, leading to significant SHF anomalies in the Greenland Sea MIZ (Schlichtholz and Houssais 2011). However, none of these or other NAO-associated effects is strong enough to drive significant concurrent SAT anomalies in this area.

Winter SAT anomalies from 1982/83 to 2005/06 in (a) the extratropical Northern Hemisphere regressed onto the concurrent NAO index and (b) the Nordic seas area [box in (a)] regressed onto the concurrent Nordic seas SIA index (Fig. 1b, squares). The contour interval (CI) is 0.5°C per the corresponding unit index. Red and blue contours represent positive and negative anomalies, respectively. The zero contour is omitted, and pink and aquamarine shading denote positive and negative anomalies statistically significant at the 95% confidence level, respectively. The thick black line shows the climatological mean position of the ice edge (15% SIC contour) in the season of the regressed scalar field while the value of s on the rhs is the field significance parameter for the regressed scalar. In (b), the arrows show the winter anomalies of surface wind regressed onto the concurrent NAO index (m s−1 per unit NAO index; scaled as in the bottom right corner and masked if both components are nonsignificant at the 95% confidence level) while the values of s = (sλ, sφ) on the rhs of the subplot are the field significance parameters for the components of the regressed vector field.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Winter SAT anomalies from 1982/83 to 2005/06 in (a) the extratropical Northern Hemisphere regressed onto the concurrent NAO index and (b) the Nordic seas area [box in (a)] regressed onto the concurrent Nordic seas SIA index (Fig. 1b, squares). The contour interval (CI) is 0.5°C per the corresponding unit index. Red and blue contours represent positive and negative anomalies, respectively. The zero contour is omitted, and pink and aquamarine shading denote positive and negative anomalies statistically significant at the 95% confidence level, respectively. The thick black line shows the climatological mean position of the ice edge (15% SIC contour) in the season of the regressed scalar field while the value of s on the rhs is the field significance parameter for the regressed scalar. In (b), the arrows show the winter anomalies of surface wind regressed onto the concurrent NAO index (m s−1 per unit NAO index; scaled as in the bottom right corner and masked if both components are nonsignificant at the 95% confidence level) while the values of s = (sλ, sφ) on the rhs of the subplot are the field significance parameters for the components of the regressed vector field.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Winter SAT anomalies from 1982/83 to 2005/06 in (a) the extratropical Northern Hemisphere regressed onto the concurrent NAO index and (b) the Nordic seas area [box in (a)] regressed onto the concurrent Nordic seas SIA index (Fig. 1b, squares). The contour interval (CI) is 0.5°C per the corresponding unit index. Red and blue contours represent positive and negative anomalies, respectively. The zero contour is omitted, and pink and aquamarine shading denote positive and negative anomalies statistically significant at the 95% confidence level, respectively. The thick black line shows the climatological mean position of the ice edge (15% SIC contour) in the season of the regressed scalar field while the value of s on the rhs is the field significance parameter for the regressed scalar. In (b), the arrows show the winter anomalies of surface wind regressed onto the concurrent NAO index (m s−1 per unit NAO index; scaled as in the bottom right corner and masked if both components are nonsignificant at the 95% confidence level) while the values of s = (sλ, sφ) on the rhs of the subplot are the field significance parameters for the components of the regressed vector field.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
The wintertime SAT anomalies over the Nordic seas are strongly coupled to sea ice variability, as shown by a very high correlation between the winter Nordic seas SAT and Nordic seas SIA indices (r = −0.93; see Table 2). Local SAT anomalies associated with the winter Nordic seas SIA index reach a slightly larger magnitude over the Barents Sea MIZ (~4°C) than over the Greenland Sea MIZ (~3°C) but are significant across the entire Nordic seas area (Fig. 2b, contours). As previously noted (e.g., Deser et al. 2000), the magnitude of SAT anomalies in the MIZ obtained from the NCEP–NCAR reanalysis might be somewhat exaggerated because of a binary representation of the MIZ (0% or 100% SIC) in the reanalysis. However, a significant association of wintertime SAT anomalies over the Nordic seas with the concurrent sea ice variability in which below normal atmospheric temperature is related to anomalous sea ice advance while above normal atmospheric temperature corresponds to anomalous sea ice retreat appears in the patterns of SAT anomalies obtained from the NCEP–NCAR reanalysis in different periods and regressed onto different indices of sea ice variability (e.g., Deser et al. 2000; Wu et al. 2004; Sorteberg and Kvingedal 2006). This association also appears in the patterns of SAT anomalies obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analyses (Germe et al. 2011) and is corroborated by climate model simulations (e.g., Bengtsson et al. 2004; Koenigk et al. 2009).
Lagged correlations of the winter Nordic seas SAT, Barents Sea SAT, Greenland Sea SAT, MIZ vorticity, and MIZ Ekman suction, indices [columns rNS, rBS, rGS, r(ζa), and r(wE), respectively] with some others defined in Table 1. Correlations are for detrended data in the 1982–2006 period except for row 8 (nondetrended data in the 1982–2006 period), and rows 9 and 10 (detrended data in the 1982–2011 period). Correlations significant at the 95% (99%) confidence level are in boldface (boldface and italic). Lags (months) are negative for the SAT, MIZ vorticity, and MIZ Ekman suction indices lagging the other indices. Positive values of r for the OIW index indicate that warm anomalies of SAT, cyclonic anomalies of surface wind vorticity


In the Barents Sea area, the coupled winter SAT and sea ice variability is to a large extent driven by anomalous local meridional winds that reach ~1 m s−1 per unit OIW index (Fig. 3a, arrows). In this area, SAT should respond directly to anomalous atmospheric heat advection across the ice edge (e.g., Schlichtholz and Houssais 2011) and also respond indirectly to this advection through feedbacks from the ocean. Indeed, in the positive phase of the OIW index, a lobe of relatively weak upward SHF anomalies in the Barents Sea MIZ coexists with a lobe of relatively strong downward SHF anomalies in the nearby open water (Fig. 3b). The downward lobe of SHF anomalies in the open water, with a maximum magnitude of ~80 W m−2 per unit OIW index (r = −0.87) in the Hopen Trench area, appears in response to advection of warm air by the southerly wind anomalies and maintains warm SST anomalies of ~0.6°C per unit OIW index at the ice edge (Fig. 3d). The lobe of upward SHF anomalies in the MIZ is maintained by negative SIC anomalies (Fig. 3c, contours) and contributes to large, warm SAT anomalies over the central Barents Sea (Fig. 3a, contours). The negative SIC anomalies should subsequently result from sea ice response to warm oceanic heat anomalies generated by the downward SHF anomalies at the ice edge. This scenario is supported by a higher correlation of the winter western Barents Sea SIA index with the OIW index (r = −0.87) 1 month earlier than with the concurrent OIW index (r = −0.74), and its very high correlation with the SHF anomalies at the ice edge (r = 0.92 at the cross in Fig. 3b) 1 month earlier. It is also consistent with a very strong link of the winter SST anomalies at the Barents Sea ice edge to the OIW index (r = 0.86 at the cross in Fig. 3d) 1 month earlier. The maximum correlation between the Nordic seas SAT and OIW indices occurs at lag 0 months. This correlation is quite high in winter (r = 0.81; see Table 2) as well as in late winter and early spring (r = 0.84), suggesting that a strong local atmospheric forcing of oceanic heat anomalies in the Nordic seas occurs throughout the winter-to-spring season.

Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, and (d) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent OIW index (positive for geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack) based on the SLP difference between the crosses in (a). In (c), the arrows show early-winter (November–February) anomalies of surface wind regressed onto the eastern Barents Sea SIA index 1 month later [i.e., in winter (December–March)]. In (a)–(d) the CI are 0.5°C, 15 W m−2, 5%, and 0.1°C per unit OIW index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, and (d) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent OIW index (positive for geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack) based on the SLP difference between the crosses in (a). In (c), the arrows show early-winter (November–February) anomalies of surface wind regressed onto the eastern Barents Sea SIA index 1 month later [i.e., in winter (December–March)]. In (a)–(d) the CI are 0.5°C, 15 W m−2, 5%, and 0.1°C per unit OIW index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, and (d) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent OIW index (positive for geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack) based on the SLP difference between the crosses in (a). In (c), the arrows show early-winter (November–February) anomalies of surface wind regressed onto the eastern Barents Sea SIA index 1 month later [i.e., in winter (December–March)]. In (a)–(d) the CI are 0.5°C, 15 W m−2, 5%, and 0.1°C per unit OIW index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
The wintertime climate variability in the Nordic seas area depends not only on the meridional wind anomalies over the Barents Sea, but also on zonal wind anomalies in the Iceland–Barents Sea corridor, as shown by the pattern of early-winter surface wind anomalies associated with the eastern Barents Sea SIA index 1 month later (Fig. 3c, arrows). In the next section, it will be shown that these zonal wind anomalies should be mainly forced by the ocean.
4. Atmospheric response to oceanic heat anomalies
The wintertime sea ice variability in the Nordic seas is strongly related not only to the concurrent SAT anomalies (see the previous section), but also to earlier oceanic heat variations (Schlichtholz 2011). In particular, the winter maximum of the postsummer correlation of the Nordic seas SIA index with the previous summer AWT index (Fig. 6a, triangles, lag 6 months, figure is described in greater detail below) is very high (r = −0.85; see Fig. 1b for comparison of the time series). This indicates that a thermodynamic atmospheric response to oceanic heat anomalies should occur in the Nordic seas MIZ. Indeed, significant winter SAT anomalies associated with the previous summer AWT index appear in the MIZ on both sides of the Nordic seas (Fig. 4a, contours). The largest AWT-associated SAT anomalies (~3.5°C) are found in the eastern Barents Sea MIZ near Novaya Zemlya, in the area of large concurrent SIC anomalies (Fig. 4c) and significant SHF anomalies of ~60 W m−2 per unit AWT index (Fig. 4b). Two-thirds of the latter are contributed by the sensible SHF anomalies, which should appear in response to the SIC anomalies as the ice surface temperature in winter is several degrees lower than the freezing point (e.g., Overland and Guest 1991). Indeed, the postsummer evolution of the SAT anomalies in the MIZ associated with the summer AWT index closely follows the corresponding evolution of the surface temperature anomalies, as shown in Fig. 5a for the location where the AWT-associated SAT anomalies are extreme. At this location, the AWT-associated SAT anomalies increase from weak summertime values to a winter maximum. More generally, summertime atmospheric variability over the Nordic seas is weakly related to concurrent subsurface ocean heat anomalies, as shown by nonsignificant correlations of the Nordic seas SAT index with the summer AWT index at lags from −1 to 3 months in Fig. 6a (squares), in which the sign of r for the Nordic seas SAT index is reversed for the sake of clarity.

Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, (d) SST, (e) ocean surface Ekman transport (arrows), and (f) mixed-layer heating from temperature advection by the Ekman flow in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a)–(d) and (f), the CI are 0.5°C, 15 W m−2, 5%, 0.1°C, and 5 W m−2 per unit AWT index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, (d) SST, (e) ocean surface Ekman transport (arrows), and (f) mixed-layer heating from temperature advection by the Ekman flow in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a)–(d) and (f), the CI are 0.5°C, 15 W m−2, 5%, 0.1°C, and 5 W m−2 per unit AWT index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Anomalies of (a) SAT (contours) and surface wind (arrows), (b) SHF, (c) SIC, (d) SST, (e) ocean surface Ekman transport (arrows), and (f) mixed-layer heating from temperature advection by the Ekman flow in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a)–(d) and (f), the CI are 0.5°C, 15 W m−2, 5%, 0.1°C, and 5 W m−2 per unit AWT index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Time-lagged regression coefficients of the anomalies of the seasonal-mean (a) SAT (circles) and surface (ice or ocean) temperature (squares) in the eastern Barents Sea MIZ at the circle in Fig. 4a, and (b) SST in the eastern Barents Sea at the circle in Fig. 4d (circles) regressed onto the summer AWT index in the 1982–2005 period (Fig. 1b, circles). The anomalies are in degrees Celsius per unit AWT index while filled symbols denote anomalies statistically significant at the 95% confidence level. Positive lags correspond to the AWT index leading the surface variables calculated as 4-month averages with the interval of 1 month. In (b), the squares represent the annual cycle of the long-term mean of the seasonal-mean SST (°C) divided by 10 at the circle in Fig. 4d, with the summer (June–September) value at lag 0 months, the late-summer (July–October) value at lag 1 month, and so on.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Time-lagged regression coefficients of the anomalies of the seasonal-mean (a) SAT (circles) and surface (ice or ocean) temperature (squares) in the eastern Barents Sea MIZ at the circle in Fig. 4a, and (b) SST in the eastern Barents Sea at the circle in Fig. 4d (circles) regressed onto the summer AWT index in the 1982–2005 period (Fig. 1b, circles). The anomalies are in degrees Celsius per unit AWT index while filled symbols denote anomalies statistically significant at the 95% confidence level. Positive lags correspond to the AWT index leading the surface variables calculated as 4-month averages with the interval of 1 month. In (b), the squares represent the annual cycle of the long-term mean of the seasonal-mean SST (°C) divided by 10 at the circle in Fig. 4d, with the summer (June–September) value at lag 0 months, the late-summer (July–October) value at lag 1 month, and so on.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Time-lagged regression coefficients of the anomalies of the seasonal-mean (a) SAT (circles) and surface (ice or ocean) temperature (squares) in the eastern Barents Sea MIZ at the circle in Fig. 4a, and (b) SST in the eastern Barents Sea at the circle in Fig. 4d (circles) regressed onto the summer AWT index in the 1982–2005 period (Fig. 1b, circles). The anomalies are in degrees Celsius per unit AWT index while filled symbols denote anomalies statistically significant at the 95% confidence level. Positive lags correspond to the AWT index leading the surface variables calculated as 4-month averages with the interval of 1 month. In (b), the squares represent the annual cycle of the long-term mean of the seasonal-mean SST (°C) divided by 10 at the circle in Fig. 4d, with the summer (June–September) value at lag 0 months, the late-summer (July–October) value at lag 1 month, and so on.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Time-lagged correlation of (a) the summer AWT index with the ZW (circles), MIZ vorticity (diamonds), OIW (reversed triangles), Nordic seas SAT (squares), and Nordic seas SIA (triangles) indices; (b) the winter Nordic seas SAT index with the Nordic seas SIA (circles) and open water SST (squares) indices; (c) the winter eastern Barents Sea SIA (squares) and MIZ vorticity (circles) indices with the open water SHF index; and (d) the winter eastern Barents Sea SIA index with the western Barents Sea SIA index (circles) and the winter Barents Sea SIA index with this index in other seasons (squares) in the 1982–2006 period. The filled symbols denote correlations statistically significant at the 95% confidence level and positive lags correspond to the specified summer or winter indices leading the other indices, which are all based on 4-month mean variables and calculated with the interval of 1 month. All indices are based on detrended data and defined in Table 1. In (a), the sign of r for the SAT anomalies is reversed. In (c), positive r for the eastern Barents Sea SIA index and negative r for the MIZ vorticity index indicate that positive sea ice area anomalies in the eastern Barents Sea and anticyclonic surface wind vorticity anomalies in the marginal ice zone area correspond to upward surface heat flux anomalies in the Hopen Trench area.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Time-lagged correlation of (a) the summer AWT index with the ZW (circles), MIZ vorticity (diamonds), OIW (reversed triangles), Nordic seas SAT (squares), and Nordic seas SIA (triangles) indices; (b) the winter Nordic seas SAT index with the Nordic seas SIA (circles) and open water SST (squares) indices; (c) the winter eastern Barents Sea SIA (squares) and MIZ vorticity (circles) indices with the open water SHF index; and (d) the winter eastern Barents Sea SIA index with the western Barents Sea SIA index (circles) and the winter Barents Sea SIA index with this index in other seasons (squares) in the 1982–2006 period. The filled symbols denote correlations statistically significant at the 95% confidence level and positive lags correspond to the specified summer or winter indices leading the other indices, which are all based on 4-month mean variables and calculated with the interval of 1 month. All indices are based on detrended data and defined in Table 1. In (a), the sign of r for the SAT anomalies is reversed. In (c), positive r for the eastern Barents Sea SIA index and negative r for the MIZ vorticity index indicate that positive sea ice area anomalies in the eastern Barents Sea and anticyclonic surface wind vorticity anomalies in the marginal ice zone area correspond to upward surface heat flux anomalies in the Hopen Trench area.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Time-lagged correlation of (a) the summer AWT index with the ZW (circles), MIZ vorticity (diamonds), OIW (reversed triangles), Nordic seas SAT (squares), and Nordic seas SIA (triangles) indices; (b) the winter Nordic seas SAT index with the Nordic seas SIA (circles) and open water SST (squares) indices; (c) the winter eastern Barents Sea SIA (squares) and MIZ vorticity (circles) indices with the open water SHF index; and (d) the winter eastern Barents Sea SIA index with the western Barents Sea SIA index (circles) and the winter Barents Sea SIA index with this index in other seasons (squares) in the 1982–2006 period. The filled symbols denote correlations statistically significant at the 95% confidence level and positive lags correspond to the specified summer or winter indices leading the other indices, which are all based on 4-month mean variables and calculated with the interval of 1 month. All indices are based on detrended data and defined in Table 1. In (a), the sign of r for the SAT anomalies is reversed. In (c), positive r for the eastern Barents Sea SIA index and negative r for the MIZ vorticity index indicate that positive sea ice area anomalies in the eastern Barents Sea and anticyclonic surface wind vorticity anomalies in the marginal ice zone area correspond to upward surface heat flux anomalies in the Hopen Trench area.
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
The atmosphere responds to oceanic heat anomalies in the Nordic seas not only thermodynamically but also dynamically. Indeed, strong (~1 m s−1) significant winter surface wind anomalies over the open water in the Iceland–Barents Sea corridor are associated with the previous summer AWT index (Fig. 4a, arrows). When the summer AWT anomalies in the BSO area are warm, the anomalous winds in the following winter blow into the Barents Sea along the common rim of a cyclonic vortex with positive anomalies of the surface wind vorticity

Anomalies of (a) surface wind vorticity and (b) atmospheric Ekman pumping velocity (positive upward) in the extratropical Northern Hemisphere in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a) and (b), the CIs are 0.5 × 10−6 s−1 and 1 × 10−4 m s−1 per unit AWT index, respectively, while the box delineates the area for calculation of the MIZ vorticity and Ekman suction indices defined in Table 1. The contour and shading colors and s value are explained in the caption to Fig. 2. In (b),
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Anomalies of (a) surface wind vorticity and (b) atmospheric Ekman pumping velocity (positive upward) in the extratropical Northern Hemisphere in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a) and (b), the CIs are 0.5 × 10−6 s−1 and 1 × 10−4 m s−1 per unit AWT index, respectively, while the box delineates the area for calculation of the MIZ vorticity and Ekman suction indices defined in Table 1. The contour and shading colors and s value are explained in the caption to Fig. 2. In (b),
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Anomalies of (a) surface wind vorticity and (b) atmospheric Ekman pumping velocity (positive upward) in the extratropical Northern Hemisphere in winters from 1982/83 to 2005/06 regressed onto the previous summer AWT index (Fig. 1b, circles). In (a) and (b), the CIs are 0.5 × 10−6 s−1 and 1 × 10−4 m s−1 per unit AWT index, respectively, while the box delineates the area for calculation of the MIZ vorticity and Ekman suction indices defined in Table 1. The contour and shading colors and s value are explained in the caption to Fig. 2. In (b),
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
The relevant wintertime indices of dynamic atmospheric variability in the Nordic seas area are related to earlier oceanic heat anomalies as strongly as the corresponding indices of thermodynamic atmospheric variability in this area. In particular, the maximum correlation of the postsummer MIZ vorticity index based on ζa averaged over the box in Fig. 7a with the previous summer AWT index is the same (r = 0.77) as the corresponding maximum correlation for the Nordic seas SAT index. Both occur in winter at lag 6 months (Fig. 6a, diamonds and squares, respectively). In the case of the ZW index describing the variability of geostrophic westerlies in the BSO area (between the crosses in Fig. 4a), the corresponding maximum correlation (r = 0.80) occurs 1 month earlier (i.e., in early winter at lag 5 months) (Fig. 6a, circles). However, both dynamic (ZW and MIZ vorticity) indices remain significantly correlated with the previous summer AWT index to early spring (lag 8 months)—that is, as long as the postsummer Nordic seas SIA index remains significantly linked to the summer AWT index (Fig. 6a, triangles). These relations and the fact that the postsummer surface wind vorticity anomalies south of the Iceland–Barents Sea corridor are less significantly linked to the summer AWT index than the corresponding anomalies north of this corridor suggest that the dynamic atmospheric response to oceanic heat anomalies in the Nordic seas is mainly driven through SHF anomalies in the MIZ. This response should mainly be sustained in the eastern Barents Sea MIZ, as indicated not only by the significant link between the winter eastern Barents Sea SIA index and the quasi-concurrent surface wind anomalies in the Iceland–Barents Sea corridor (Fig. 3c), but also the patterns of winter SIC and SST anomalies associated with the concurrent MIZ vorticity index. In these patterns, significant SIC anomalies appear only in the eastern Barents Sea (Fig. 8a, contours) and the largest SST anomalies align with the ice edge in this area (Fig. 8b), where also significant AWT-associated SST anomalies are found (Fig. 4d).

Anomalies of (a) SIC and (b) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent MIZ vorticity index based on ζa averaged over the box in Fig. 7a; and anomalies of (c) early-spring (February–May) SST and (d) late-winter (January–April) SHF (contours) and surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto the following-winter MIZ vorticity index. In (a), the arrows show the winter anomalies of surface wind regressed onto the previous late-winter open water SHF index based on the surface heat flux at the cross in (d). In (a)–(d), the CI are 5%, 0.1°C, 0.1°C, and 15 W m−2 per unit MIZ vorticity index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Anomalies of (a) SIC and (b) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent MIZ vorticity index based on ζa averaged over the box in Fig. 7a; and anomalies of (c) early-spring (February–May) SST and (d) late-winter (January–April) SHF (contours) and surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto the following-winter MIZ vorticity index. In (a), the arrows show the winter anomalies of surface wind regressed onto the previous late-winter open water SHF index based on the surface heat flux at the cross in (d). In (a)–(d), the CI are 5%, 0.1°C, 0.1°C, and 15 W m−2 per unit MIZ vorticity index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Anomalies of (a) SIC and (b) SST in the Nordic seas area in winters from 1982/83 to 2005/06 regressed onto the concurrent MIZ vorticity index based on ζa averaged over the box in Fig. 7a; and anomalies of (c) early-spring (February–May) SST and (d) late-winter (January–April) SHF (contours) and surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto the following-winter MIZ vorticity index. In (a), the arrows show the winter anomalies of surface wind regressed onto the previous late-winter open water SHF index based on the surface heat flux at the cross in (d). In (a)–(d), the CI are 5%, 0.1°C, 0.1°C, and 15 W m−2 per unit MIZ vorticity index, respectively, while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1





The air pumped out of the atmospheric boundary layer in the Nordic seas MIZ area in response to positive oceanic heat anomalies may, at least partly, return to this layer in the area south of the Iceland–Barents Sea corridor. In this area, the significant anticyclonic surface wind vorticity anomalies in winters following the summers with warm AWT anomalies in the BSO area (Fig. 7a) indeed coexist with significant anomalous suction of air from the free atmosphere (Fig. 7b). However, the full atmospheric response should be more complex, as suggested by hemispheric teleconnections in the patterns of the AWT-associated winter anomalies of ζa and wEa (Fig. 7). These teleconnections indicate that eddy–mean flow interactions discussed in some of the abovementioned studies (Alexander et al. 2004; Magnusdottir et al. 2004; Deser et al. 2004) may play a role in the atmospheric response to oceanic heat anomalies in the Nordic seas. However, a detailed analysis of this effect is beyond the scope of the present study.
5. Mechanisms of oceanic forcing
a. General remarks
The postsummer SAT anomalies in the eastern Barents Sea MIZ, and generally over the Nordic seas, become significantly linked to the summer AWT index in late autumn at lag 4 months [Fig. 5a (circles) and Fig. 6a (squares)]. At that time, the AWT-associated surface wind anomalies are not significant yet (Fig. 6a, circles and diamonds). Therefore, in this early stage of the atmospheric response to oceanic forcing, the SHF anomalies in the MIZ should be sustained by entrainment of subsurface heat anomalies into the deepening ocean surface mixed layer, and possibly by some nondynamic effects related to the entrainment process. However, the entrainment should not be important in winter, when the AWT-associated postsummer SST anomalies stop growing at the eastern Barents Sea ice edge (Fig. 5b, circles), and generally in the Nordic seas (Schlichtholz 2011). In this stage, the SHF anomalies in the MIZ and the corresponding anomalous atmospheric circulation triggered by oceanic heat anomalies can be sustained by feedbacks from this anomalous circulation onto ocean currents. This scenario will be elaborated based on estimates of the order of magnitude of relevant terms in the postsummer anomalous heat balance of the ocean mixed layer associated with the summer AWT index.





In the MIZ, the anomalous heat loss from the ocean
b. Early stage
The mixed-layer heating from entrainment of subsurface temperature anomalies is described by the term
The horizontal redistribution of heat anomalies by the mean currents, expressed by the contribution
c. Developed stage
The mixed-layer heating from temperature advection by the anomalous Ekman flow [i.e., the term
The anomalous Ekman flow may contribute to oceanic forcing of the SHF anomalies in the Nordic seas MIZ indirectly through inducing anomalous geostrophic currents. Such currents should indeed appear in response to the significant AWT-associated basinwide anomalies of the suction velocity at the bottom of the ocean surface Ekman layer,
6. Late-winter-to-next-winter climate feedback
The Nordic seas SAT index is significantly linked to the summer AWT index not only in the postsummer season at lags from 4 to 7 months, but also in the presummer season at lags from −6 to −2 months (Fig. 6a, squares). A simple interpretation of this relation could be that the wintertime sea ice and atmospheric variability in the Nordic seas area is strongly influenced by persistent oceanic temperature anomalies advected from the south in subsurface layers and coming up to the surface as the fall mixed layer deepens after the summer. Forcing by a broad-scale, low-frequency oceanic oscillation would be consistent with a sequence of subperiods with same sign anomalies observed in the time series of the summer AWT index (Fig. 1b, circles), but other mechanisms should also play a role. Indeed, the maximum presummer correlation with the summer AWT index is higher, and occurs earlier, for the Nordic seas SAT index (r = 0.85 in late winter at lag −5 months; see Fig. 6a, squares) than for the Nordic seas SIA index (r = −0.79 at lags −4 and −3 months; see Fig. 6a, triangles). Moreover, as already reported by Schlichtholz and Houssais (2011), the summer AWT index correlates highly with the previous late-winter OIW index (r = 0.84; see Fig. 6a, reversed triangles). Furthermore, the significant links of the summer AWT index to the previous late-winter OIW and Nordic seas SAT indices, and also to the following early-winter ZW (Fig. 6a, circles) and winter Nordic seas SAT indices, are largely accounted for by the strictly interannual components of the time series obtained by subtracting the curves smoothed with a five-point binomial filter from the original time series (Table 3). These facts indicate that a local climate feedback should exist in the Nordic seas area in which a key role is played by oceanic heat anomalies that are generated by presummer atmospheric forcing, survive below the shallow summer mixed layer and, after reemerging on the surface, strongly influence the postsummer sea ice and atmospheric variability. This feedback will be discussed below based on relations of the summer AWT anomalies in the BSO area and of the following-winter atmospheric variability over the Nordic seas to the previous winter-to-spring air–sea interactions.
Lagged correlations of the summer (June–September) AWT index in the 1982–2005 period with the previous late-winter (January–April) OIW and Nordic seas SAT indices (columns OIW−5 and SAT−5, respectively), and with the following early-winter (November–February) ZW and winter (December–March) Nordic seas SAT indices (columns ZW5 and SAT6, respectively) defined in Table 1. Correlations denoted as r (first row), rs (second row), and rr (third row) are for the original (detrended) time series, the corresponding time series obtained by smoothing with a 5-point binomial filter, and the residuals obtained by subtracting the smoothed time series from the original time series, respectively. The significance (p value) of each correlation is given in parentheses. Positive correlations for the OIW and ZW indices indicate that warm Atlantic water temperature anomalies in the BSO area correspond to geostrophic wind anomalies across the Barents Sea MIZ directed toward the ice pack and westerly geostrophic wind anomalies in the BSO area, respectively.


The summer AWT index not only correlates highly with the previous late-winter OIW index, but also correlates moderately with the previous winter NAO index (Schlichtholz and Houssais 2011). Consistent with these relations, warm summer AWT anomalies in the BSO area are preceded by significant late-winter southerly wind anomalies over the Barents Sea and northerly wind anomalies along the Greenland coast (Fig. 9a, arrows), and consequently by warm early-spring SST anomalies that are significant in the open water and MIZ on both sides of the Nordic seas but are largest at the Barents Sea ice edge (Fig. 9a, contours). The strong link of the summer AWT anomalies in the BSO area to the presummer air–sea interactions over the Barents Sea was attributed by Schlichtholz and Houssais (2011) to formation of ocean temperature anomalies in a deep mixed layer by SHF anomalies resulting from anomalous atmospheric heat advection, and to their subsequent westward export by a recirculating branch of Atlantic water (HTR arrow in Fig. 1a). Consistent with this attribution, a tongue of warm SST anomalies spreading westward from the Hopen Trench along the slope of the Spitsbergen Bank appears in the patterns of winter-to-spring SST anomalies associated with the concurrent OIW index (Fig. 3d) and the following-summer AWT index (Fig. 9a).

Anomalies of early-spring SST (contours) and late-winter surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto (a) the following-summer AWT index (Fig. 1b, circles) and (b) the following-winter Nordic seas SAT index. The CI is 0.1°C per the corresponding unit index while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Anomalies of early-spring SST (contours) and late-winter surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto (a) the following-summer AWT index (Fig. 1b, circles) and (b) the following-winter Nordic seas SAT index. The CI is 0.1°C per the corresponding unit index while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Anomalies of early-spring SST (contours) and late-winter surface wind (arrows) in the Nordic seas area in the period 1982–2005 regressed onto (a) the following-summer AWT index (Fig. 1b, circles) and (b) the following-winter Nordic seas SAT index. The CI is 0.1°C per the corresponding unit index while
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
While some oceanic heat anomalies generated at the western Barents Sea ice edge in the winter-to-spring season influence the SIC and SAT variability already in that season (see section 3), some oceanic heat anomalies generated in the winter-to-spring season at (or advected to) more southern locations in the western Barents Sea should be transported eastward below the summer mixed layer, and drive the sea ice and atmospheric variability on the eastern side of the Barents Sea in the following winter. Supporting this scenario, significant winter SST anomalies associated with the previous summer AWT index appear in the eastern Barents Sea, but not in the western Barents Sea (Fig. 4d), where the SST anomalies depend significantly on the concurrent meridional wind anomalies (Fig. 3d). This scenario is also consistent with the fact that the winter eastern Barents Sea SIA index correlates significantly with the previous late-winter-to-early-summer western Barents Sea SIA index (Fig. 6d, circles, lags from −11 to −7 months) while the Barents Sea SIA index in winter is not significantly autocorrelated with its previous winter-to-spring values (Fig. 6d, squares).
The scenario of a strong postsummer atmospheric response to oceanic heat anomalies driven by the presummer air–sea interactions in the western Barents Sea area is corroborated by significant links of the winter MIZ vorticity index to anomalies observed in this area during the previous winter-to-spring season. Indeed, the cyclonic wintertime anomalies of the surface wind vorticity north of the Iceland–Barents Sea corridor that correspond to positive values of the MIZ vorticity index are preceded by warm early-spring SST anomalies in the western Barents Sea (Fig. 8c). These warm SST anomalies are generated by southerly wind anomalies (Fig. 8d, arrows) through downward SHF anomalies over the open water (Fig. 8d, contours). In late winter, the latter exhibit a maximum magnitude of ~75 W m−2 per unit of the following-winter MIZ vorticity index in the Hopen Trench area, at the cross in Fig. 8d. The late-winter open water SHF index based on the SHF data at this location correlates highly with the following-winter eastern Barents Sea SIA index and the corresponding MIZ vorticity index (r = 0.86 and −0.86, respectively; see Fig. 6c, lag −11 months). It also correlates highly with the following-winter MIZ Ekman suction index based on wE averaged over the box in Fig. 7b (r = −0.87; see Table 2). The pattern of winter surface wind anomalies associated with the previous late-winter open water SHF index (Fig. 8a, arrows) is strikingly similar to the corresponding pattern associated with the previous summer AWT index (Fig. 4a). This similarity is consistent with the strong link of the summer AWT anomalies in the BSO area to the previous late-winter SHF anomalies in the Hopen Trench area reported by Schlichtholz and Houssais (2011). All these relations show that the wintertime feedback between anomalous winds and currents discussed in the previous section is an integral part of the late-winter-to-next-winter climate feedback caused by reemerging SST anomalies.
The wintertime atmospheric variability over the Nordic seas may be influenced by locally forced oceanic heat anomalies which reemerge on the surface not only on the eastern side, but also on the western side of these seas. This is indicated by, for instance, a significant, albeit moderate, correlation of the winter Greenland Sea SAT index with the previous winter NAO index (r = 0.46; see Table 2) and supported by a significant imprint left by the wintertime NAO on springtime SST anomalies in the Greenland Sea (not shown) and the following-winter SIC anomalies in this area (Schlichtholz 2011). Consistent with the scenario of a postsummer atmospheric response to oceanic heat anomalies driven by the presummer air–sea interactions on both sides of the Nordic seas area, the winter Nordic seas SAT index correlates significantly with the previous late-winter surface wind anomalies over the Barents, Norwegian, and Greenland seas (Fig. 9b, arrows), and with the previous early-spring SST anomalies in the western Barents Sea open water and all around the Greenland Sea (Fig. 9b, contours), but not with the previous winter-to-spring Nordic seas SIA index (Fig. 6b, circles). Its maximum correlation with the open water SST index based on the SST data averaged over an area extending from the eastern Greenland Sea to the western Barents Sea (box in Fig. 9b) is very high (r = 0.86) and occurs just for the previous early-spring SST anomalies (Fig. 6b, squares, lag −10 months).
7. Conclusions
a. General remarks
Previous studies indicate that the extratropical climate feedback through which the atmosphere responds to oceanic heat anomalies that are driven by atmospheric forcing in the deep mixed layer at the end of winter and then reentrained into the deepening mixed layer at the beginning of the following winter is rather weak. However, the investigations were mainly focused on the effects of large-scale patterns of oceanic heat anomalies, such as the sea surface temperature tripole in the North Atlantic driven by the North Atlantic Oscillation (e.g., Cassou et al. 2007), and so do not preclude a strong impact of other reemerging SST anomalies on the local atmosphere in some regions. Such an impact may occur in some sub-Arctic seas, where the reemerging SST anomalies may induce large surface heat flux anomalies by affecting the sea ice advance in the autumn-to-winter season. Motivated by this hypothesis and by the recent finding that the wintertime sea ice concentration anomalies in the Nordic (Greenland–Iceland–Norwegian and Barents) seas in the period 1982–2006 were strongly linked to the previous spring’s locally forced SST anomalies, we have carried out a study of atmosphere–ocean feedbacks in this region. The analysis is based on time-lagged relationships between the anomalies of seasonal-mean fields of observed SST and SIC from Reynolds et al. (2007), the corresponding anomalies of surface atmospheric variables from the NCEP–NCAR reanalysis, and the summer (June–September) Atlantic water temperature index constructed by Schlichtholz and Houssais (2011) from observed temperature anomalies in the western Barents Sea Opening area over the 100–300-m-depth layer.
We have shown that, in the 1982–2006 period, 86% (r = −0.93) of the variance of the average surface air temperature anomalies over the Nordic seas (Nordic seas SAT index) in winter (December–March) is associated with the concurrent variations in the total SIA in this region. This coupled atmosphere–ice variability is completely disconnected from the concurrent NAO characterized by the station-based index of Hurrell (1995). Our primary finding is that 1) this variability is to a large extent driven by the locally forced, reemerged SST anomalies, 2) the atmosphere responds to the reemerged SST anomalies not only thermodynamically but also dynamically, and 3) the dynamic atmospheric response is maintained mainly by oceanic forcing in the Barents Sea MIZ. The dynamic response includes significant surface wind anomalies appearing all along the Iceland–Barents Sea corridor, on the common rim of two anomalous vortices that resemble the atmospheric response to heavy wintertime sea ice conditions in the Barents Sea reported by Koenigk et al. (2009) from sensitivity experiments with a climate model.
The above conclusions are supported by high correlations between key indices of the regional climate variability (Fig. 6) and high confidence levels for field significance (see the numbers on the right-hand side of the regression maps in Figs. 4, 8, and 9). They are corroborated by nondetrended as well as detrended data in the 1982–2006 period, and also by the SST and atmospheric time series extended to 2011 (Table 2). In the 1982–2006 period, as much as 74% of the variance of the winter Nordic seas SAT index is explained by the previous early-spring (February–May) SST anomalies averaged over an open water area extending from the eastern Greenland Sea to the western Barents Sea (70°–74°N, 10°–25°E). The same fraction (74%) of the variance of the winter average surface wind vorticity in the MIZ area north of 75°N is explained by the previous late-winter (January–April) SHF anomalies on the open water side of the Barents Sea ice edge in the Hopen Trench area. In this period, the regional climate feedback from reemerging SST anomalies is clearly reflected by strong links of the summer subsurface ocean temperature anomalies in the BSO area to the presummer and postsummer air–sea interactions. In particular, a large fraction of the variance of the winter Nordic seas SAT index (59%) and early-winter (November–February) anomalies of zonal geostrophic winds over the BSO area (64%) is explained by the previous summer AWT index, which in turn has a large fraction of its variance explained by the previous late-winter Nordic seas SAT index (72%) and the corresponding “on ice wind” index (71%) introduced by Schlichtholz and Houssais (2011) to characterize the surface winds across the Barents Sea MIZ. All these relationships show that the climate variability in the Nordic seas area is to some extent predictable. It should be noted that the summer oceanic heat anomalies in the Nordic seas (AWT index) are also significantly correlated with the following-winter atmospheric variability in remote regions (Fig. 7), but the causality of this link remains to be demonstrated.
Oceanic heat anomalies that influence the sea ice and atmospheric variability in the Barents Sea area are traditionally viewed as being advected from the south (e.g., Francis and Hunter 2007) or generated by wind-driven anomalies of the Atlantic water inflow through the BSO (e.g., Bengtsson et al. 2004). Our results show that there are at least three climate feedback mechanisms related to oceanic heat anomalies formed locally in the Barents Sea by air–sea interactions in the winter-to-spring season: 1) a thermodynamic atmospheric response to these anomalies during the season of their formation, 2) the thermodynamic and dynamic atmospheric response when these anomalies are reentrained into the mixed layer in the following autumn-to-winter season, and 3) a dynamic atmosphere–ocean feedback that makes the latter response persistent. These feedbacks are depicted schematically in Fig. 10 and summarized below.

Schematic of climate feedbacks in the Nordic seas area. (left) Generation of temperature anomalies in the ocean surface mixed layer (i.e.,
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1

Schematic of climate feedbacks in the Nordic seas area. (left) Generation of temperature anomalies in the ocean surface mixed layer (i.e.,
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
Schematic of climate feedbacks in the Nordic seas area. (left) Generation of temperature anomalies in the ocean surface mixed layer (i.e.,
Citation: Journal of Climate 26, 9; 10.1175/JCLI-D-11-00594.1
b. Summary of atmosphere–ocean feedback mechanisms
As depicted in the left panel of Fig. 10, heating from temperature advection by anomalous winter-to-spring winds
Some of the oceanic heat anomalies generated in the western Barents Sea at the end of winter are transported eastward below the summer mixed layer. They influence the atmospheric variability over the Nordic seas in the following winter by driving SHF anomalies in the MIZ on the eastern side of the Barents Sea. As depicted in the right panel of Fig. 10, upward SHF anomalies
The wintertime atmospheric response to oceanic heat anomalies in the Nordic seas occurs in two stages. In the early, short, purely thermodynamic stage, when significant oceanically driven SAT anomalies begin to appear while the corresponding surface wind anomalies are not significant yet, the SHF anomalies in the MIZ are triggered by anomalous heating from entrainment of subsurface ocean temperature anomalies (
Acknowledgments
The Physical Sciences Division (PSD) of the Earth System Research Laboratory (ESRL) of NOAA, Boulder, Colorado, is acknowledged for providing the Optimum Interpolation (OI.v2) SST and SIC fields and the NCEP Reanalysis data derived from their website (http://www.esrl.noaa.gov/psd/). Ocean temperature data were provided by the Oceanographic Database of the International Council for the Exploration of the Sea (http://ocean.ices.dk/) and the World Ocean Database (WOD05) of the National Oceanographic Data Center of NOAA (http://www.nodc.noaa.gov/). The author thanks three anonymous reviewers for their helpful comments.
REFERENCES
Aagaard, K., and E. C. Carmack, 1989: The role of sea ice and other fresh water in the Arctic circulation. J. Geophys. Res., 94, 14 485–14 498.
Alekseev, V. A., I. Esau, I. V. Polyakov, S. J. Byam, and S. Sorokina, 2012: Vertical structure of recent arctic warming from observed data and reanalysis products. Climate Change,111, 215–239, doi:10.1007/s10584-011-0192-8.
Alexander, M. A., and C. Deser, 1995: A mechanism for the recurrence of wintertime midlatitude SST anomalies. J. Phys. Oceanogr., 25, 122–137.
Alexander, M. A., U. S. Bhatt, J. E. Walsh, M. S. Timlin, J. S. Miller, and J. D. Scott, 2004: The atmospheric response to realistic Arctic sea ice anomalies in an AGCM during winter. J. Climate, 17, 890–905.
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.
Bengtsson, L., V. A. Semenov, and O. M. Johannessen, 2004: The early twentieth-century warming in the Arctic—A possible mechanism. J. Climate, 17, 4045–4057.
Cassou, C., C. Deser, and M. A. Alexander, 2007: Investigating the impact of reemerging sea surface temperature anomalies on the winter atmospheric circulation over the North Atlantic. J. Climate, 20, 3510–3526.
Deser, C., J. E. Walsh, and M. S. Timlin, 2000: Arctic sea ice variability in the context of recent atmospheric circulation trends. J. Climate, 13, 617–633.
Deser, C., M. A. Alexander, and M. S. Timlin, 2003: Understanding the persistence of sea surface temperature anomalies in midlatitudes. J. Climate, 16, 57–72.
Deser, C., G. Magnusdottir, R. Saravanan, and A. Phillips, 2004: The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part II: Direct and indirect components of the response. J. Climate, 17, 877–889.
Francis, J. A., and E. Hunter, 2007: Drivers of declining sea ice in the Arctic winter: A tale of two seas. Geophys. Res. Lett., 34, L17503, doi:10.1029/2007GL030995.
Francis, J. A., W. Chan, D. J. Leathers, J. R. Miller, and D. E. Veron, 2009: Winter Northern Hemisphere weather patterns remember summer Arctic sea-ice extent. Geophys. Res. Lett., 36, L07503, doi:10.1029/2009GL037274.
Frankignoul, C., 1985: Sea surface temperature anomalies, planetary waves, and air–sea feedback in the middle latitudes. Rev. Geophys., 23, 357–390.
Gammelsrød, T., Ø. Leikvina, V. Lienand, W. P. Budgell, H. Loeng, and W. Maslowski, 2009: Mass and heat transports in the NE Barents Sea: Observations and models. J. Mar. Syst., 75, 56–69.
Germe, A., M.-N. Houssais, C. Herbaut, and C. Cassou, 2011: Greenland Sea sea ice variability over 1979–2007 and its link to the surface atmosphere. J. Geophys. Res.,116, C10034, doi:10.1029/2011JC006960.
Hurrell, J. W., 1995: Decadal trends in the North Atlantic Oscillation: Regional temperature and precipitation. Science, 269, 676–679.
Ikeda, M., 1990: Decadal oscillations of the air–ice–ocean system in the Northern Hemisphere. Atmos.–Ocean, 28, 106–139.
Ingvaldsen, R., L. Asplin, and H. Loeng, 2004: The seasonal cycle in the Atlantic transport to the Barents Sea during the years 1997–2001. Cont. Shelf Res., 24, 1015–1032.
Ivanov, V. V., and G. I. Shapiro, 2005: Formation of a dense water cascade in the marginal ice zone in the Barents Sea. Deep-Sea Res. I, 52, 1699–1717.
Johannessen, O. M., K. Lygre, and T. Eldevik, 2005: Convective chimneys and plumes in the northern Greenland Sea. The Nordic Seas: An Integrated Perspective, Geophys. Monogr., Vol. 158, Amer. Geophys. Union, 251–272.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471.
Koenigk, T., U. Mikolajewicz, J. H. Jungclaus, and A. Kroll, 2009: Sea ice in the Barents Sea: Seasonal to interannual variability and climate feedbacks in a global coupled model. Climate Dyn., 32, 1119–1138.
Kolstad, E. W., 2008: A QuikSCAT climatology of ocean surface winds in the Nordic seas: Identification of features and comparison with the NCEP/NCAR reanalysis. J. Geophys. Res., 113, D11106, doi:10.1029/2007JD008918.
Kushnir, Y., W. A. Robinson, P. Chang, and A. W. Robertson, 2006: The physical basis for predicting Atlantic sector seasonal-to-interannual climate variability. J. Climate, 19, 5949–5970.
Kvingedal, B., 2005: Sea-ice extent and variability in the Nordic Seas, 1967–2002. The Nordic Seas: An Integrated Perspective, Geophys. Monogr., Vol. 158, Amer. Geophys. Union, 39–49.
Livezey, R. E., and W. Y. Chen, 1983: Statistical field significance and its determination by Monte Carlo techniques. Mon. Wea. Rev., 111, 46–59.
Magnusdottir, G., C. Deser, and R. Saravanan, 2004: The effects of North Atlantic SST and sea ice anomalies on the winter circulation in CCM3. Part I: Main features and storm track characteristics of the response. J. Climate, 17, 857–876.
Midttun, L., 1985: Formation of dense bottom water in the Barents Sea. Deep-Sea Res., 32, 1233–1241.
Mysak, L. A., and S. A. Venegas, 1998: Decadal climate oscillations in the Arctic: A new feedback loop for atmosphere–ice–ocean interactions. Geophys. Res. Lett., 25, 3607–3610.
Namias, J., and B. M. Born, 1970: Temporal coherence in North Pacific sea-surface temperature patterns. J. Geophys. Res., 75, 5952–5955.
Namias, J., and B. M. Born, 1974: Further studies of temporal coherence in North Pacific sea surface temperatures. J. Geophys. Res., 79, 797–798.
Overland, J. E., and P. S. Guest, 1991: The Arctic snow and air temperature budget over sea ice during winter. J. Geophys. Res., 96 (C3), 4651–4662.
Overland, J. E., and M. Wang, 2010: Large-scale atmospheric circulation changes are associated with the recent loss of Arctic sea ice. Tellus, 62A, 1–9.
Overland, J. E., M. Wang, and S. Salo, 2008: The recent Arctic warm period. Tellus, 60A, 589–597.
Reynolds, R. W., T. M. Smith, C. Liu, D. B. Chelton, K. S. Casey, and M. G. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 5473–5496.
Schlichtholz, P., 2011: Influence of oceanic heat variability on sea ice anomalies in the Nordic Seas. Geophys. Res. Lett., 38, L05705, doi:10.1029/2010GL045894.
Schlichtholz, P., and M.-N. Houssais, 2011: Forcing of oceanic heat anomalies by air–sea interactions in the Nordic Seas area. J. Geophys. Res., 116, C01006, doi:10.1029/2009JC005944.
Simonsen, K., and P. M. Haugan, 1996: Heat budgets of the Arctic Mediterranean and sea surface heat flux parameterizations for the Nordic Seas. J. Geophys. Res., 101, 6553–6576.
Sorteberg, A., and B. Kvingedal, 2006: Atmospheric forcing on the Barents Sea winter ice extent. J. Climate, 19, 4772–4784.
Strong, C., and G. Magnusdottir, 2010: Modeled winter sea ice variability and the North Atlantic Oscillation: A multi-century perspective. Climate Dyn., 34, 515–525.
Ukita, J., M. Honda, H. Nakamura, Y. Tachibana, D. J. Cavlieri, C. L. Parkinson, H. Koide, and K. Yamamoto, 2007: Northern Hemisphere sea ice variability: Lag structure and its implications. Tellus, 59A, 261–272.
Vinje, T., 2001: Anomalies and trends of sea-ice extent and atmospheric circulation in the Nordic Seas during the period 1864–1998. J. Climate, 14, 255–267.
Wu, B., J. Wang, and J. Walsh, 2004: Possible feedback of winter sea ice in the Greenland and Barents Seas on the local atmosphere. Mon. Wea. Rev., 132, 1868–1876.
Zhang, J., M. Steele, D. A. Rothrock, and R. W. Lindsay, 2004: Increasing exchanges at Greenland–Scotland Ridge and their links with the North Atlantic Oscillation and Arctic Sea Ice. Geophys. Res. Lett., 31, L09307, doi:10.1029/2003GL019304.
Zhang, X., A. Sorteberg, J. Zhang, R. Gerdes, and J. C. Comiso, 2008: Recent radical shifts of atmospheric circulations and rapid changes in Arctic climate system. Geophys. Res. Lett., 35, L22701, doi:10.1029/2008GL035607.