Abstract

Using NCEP–NCAR reanalysis and Japanese 25-yr Reanalysis (JRA-25) data, this paper investigates the association between winter sea ice concentration (SIC) in Baffin Bay southward to the eastern coast of Newfoundland, and the ensuing summer atmospheric circulation over the mid- to high latitudes of Eurasia. It is found that winter SIC anomalies are significantly correlated with the ensuing summer 500-hPa height anomalies that dynamically correspond to the Eurasian pattern of 850-hPa wind variability and significantly influence summer rainfall variability over northern Eurasia. Spring atmospheric circulation anomalies south of Newfoundland, associated with persistent winter–spring SIC and a horseshoe-like pattern of sea surface temperature (SST) anomalies in the North Atlantic, act as a bridge linking winter SIC and the ensuing summer atmospheric circulation anomalies over northern Eurasia. Indeed, this study only reveals the association based on observations and simple simulation experiments with SIC forcing. The more precise mechanism for this linkage needs to be addressed in future work using numerical simulations with SIC and SST as the external forcings. The results herein have the following implication: Winter SIC west of Greenland is a possible precursor for summer atmospheric circulation and rainfall anomalies over northern Eurasia.

1. Introduction

Sea ice plays an important role in regulating climate variability because of its high albedo and insulating heat, moisture, and momentum exchanges between the atmosphere and the ocean (Honda et al. 1999; Wu et al. 1999; Rigor et al. 2002; Alexander et al. 2004; Deser et al. 2004; Magnusdottir et al. 2004; Wu et al. 2004; Balmaseda et al. 2010). Sea ice in the Nordic and Barents Seas influences the stratification and stability of the planetary boundary layer, and creates the potential for dynamical influences on the atmospheric boundary layer via vertical motion induced by boundary layer pressure anomalies (Wu et al. 2004). Decreased sea ice causes increased upward surface heat fluxes, near-surface warming, enhanced precipitation, and below normal sea level pressure (SLP), whereas increased sea ice causes the opposite of all these effects (Alexander et al. 2004). In reality, however, it is very difficult to directly detect the impact of sea ice on the atmosphere because of the dominance of atmospheric forcing of sea ice on the observed association. The large-scale indirect impact of sea ice on the atmosphere strongly depends on the interactions between anomalous surface fluxes associated with sea ice and the large-scale circulation and underlying sea surface temperature (SST) (Alexander et al. 2004; Deser et al. 2004; Balmaseda et al. 2010).

Over the past two decades, Arctic sea ice data have shown a coherent negative trend, especially during the late summer season. Since the late 1990s, September sea ice extent has frequently reached previous record lows, further enhancing the rate of decline. With decreased sea ice extent in the late summer, the ocean absorbs more heat, which delays the freezing process. Decreased Arctic sea ice enhances Arctic warming (Screen and Simmonds 2010; Kumar et al. 2010) and further affects climate variations remotely through its positive and negative feedbacks on the atmosphere (Deser et al. 2004; Magnusdottir et al. 2004; Honda et al. 1999, 2009; Wu et al. 1999).

Some studies have shown that autumn–winter Arctic sea ice anomalies impact the winter atmosphere over Eurasia (Wu et al. 1999; Honda et al. 2009; Petoukhov and Semenov 2010; B. Wu et al. 2011). Above average winter sea ice in the Barents–Kara Seas is associated with less intense cold air masses in China and a weakened East Asian winter monsoon, and vice versa (Wu et al. 1999). Using numerical experiments, Petoukhov and Semenov (2010) investigated winter atmospheric responses to winter sea ice concentration (SIC) in the Barents–Kara Seas and showed that decreased winter SIC helped cause extreme cold events over Eurasia. Decreased winter sea ice in the Greenland–Barents Seas, along with increased sea ice in the Labrador Sea, enhances the negative polarities of the Arctic Oscillation (AO) (Alexander et al. 2004). Honda et al. (2009) showed that significant cold anomalies over the Far East in early winter and zonal cold anomalies from Europe to the Far East in late winter were associated with a decrease in previous September Arctic sea ice. Further, B. Wu et al. (2011) demonstrated that the coherent variation in Arctic SIC from autumn to winter provides a means for possible seasonal prediction of the winter Siberian high (SH) and East Asian winter monsoon.

In the last two decades (1990–2009), an intensification of the winter SH, along with decreasing trends in surface air temperature in the mid- to high latitudes of Asia, has favored recent frequent cold winters over East Asia (B. Wu et al. 2011). This finding may be related to autumn SIC decline trends in the eastern Arctic and warmer SSTs in the North Atlantic, Arctic marginal seas, and North Pacific.

Relative to the observed link between autumn–winter Arctic sea ice and the winter atmosphere, the association of winter Arctic sea ice with the following summer's atmospheric circulation over northern Eurasia is not as well understood. Investigating this question will aid understanding of features of atmospheric variability over northern Eurasia, and allow us to identify any potential precursors of summer atmospheric variability. In this study, we focus on winter SIC west of Greenland, including the Labrador Sea, Davis Strait, Baffin Bay, and Hudson Bay. We select this particular region for study as the air–sea fluxes are large (Kvamstø et al. 2004) and also because sea ice is closely associated with SST in the North Atlantic on both interannual and decadal time scales (Deser and Blackmon 1993; Deser et al. 2002).

2. Data and methods

The following datasets are used: 1) the Arctic SIC dataset (1° × 1°) from 1979 to 2010, obtained from the British Atmospheric Data Centre (BADC; http://badc.nerc.ac.uk/data/hadisst/); 2) the monthly mean SLP, 850-hPa winds, and geopotential heights from 1979 to 2010, obtained from the National Centers for Environmental Prediction (NCEP)–National Center for Atmospheric Research (NCAR) Global Reanalysis 1 and the Japanese 25-yr Reanalysis (JRA-25) (Onogi et al. 2007); 3) the monthly mean SST (2° × 2°) from 1979 to 2010 (http://rda.ucar.edu/datasets/ds277.0/) (Smith and Reynolds 2003); 4) monthly mean global land precipitation data from 1979 to 2010 (http://ftp.cpc.ncep.noaa.gov/precip/50yr/gauge/2.5deg/format_bin/) (Chen et al. 2002); and 5) the monthly mean North Atlantic Oscillation (NAO) index from 1979 to 2010 (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/norm.nao.monthly.b5001.current.ascii).

Composite and regression analyses were applied to investigate the association between SIC and the prevailing atmospheric circulation. Additionally, the Monte Carlo method is applied to examine statistical field significance, as in Livezey and Chen (1983). For an anomalous field derived from linear regression, the percentage of grid points that are statistically significant at the 0.05 level is first identified over a domain. This process is then repeated 1000 times with different series of 31 numbers (or 31 winters from 1979 to 2010) randomly selected from a normal distribution. The anomalous field is deemed significant if the percentage of significant grid points exceeds that derived from 1000 experimental replications. Herein, the winter and summer respectively refer to December–February (DJF) and June–August (JJA).

Additionally, the ECHAM5 (Roeckner et al. 2003) model (T63 spectral resolution and 19 pressure levels) was applied to explore impacts of SIC on the model atmosphere. A 30-yr simulation with the climatological monthly SST and observed Northern Hemisphere monthly SIC from 1978 to 2007 as the external forcing (SIC experiment) was performed (climatological monthly SST and SIC observational data are obtained from http://www-pcmdi.llnl.gov/projects/amip/AMIP2EXPDSN/BCS/bcsintro.php), and this experiment was repeated with 12 different atmospheric initial conditions that derived from a 30-yr control run.

3. Observational relationships

The area-weighted, regionally averaged winter SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W) shows apparent interannual variability, but no significant trend during the study period (Fig. 1). It is evident that there was more SIC before the winter of 1994/95 relative to afterward. Using this time series, cases for composite analysis were first selected, and heavy (light) sea ice cases were defined as those with a standard deviation of >0.8 (<−0.8). The heavy sea ice cases include the winters of 1982/83, 1983/84, 1989/90, 1990/91, 1992/93, 1993/94, and 2007/08. The light sea ice cases are the winters of 1979/80, 1981/82, 1985/86, 1998/99, 2003/04, 2005/06, 2006/07, and 2009/10. Differences in seasonal mean SIC between heavy and light cases are illustrated in Fig. 2. Significant differences are seen in the Labrador Sea, Davis Strait, Baffin Bay, Hudson Bay, the Greenland Sea, and part of the Barents Sea. Sea ice variations in the Greenland Sea north to Iceland and the southern Barents Sea are out of phase with those west of Greenland (Fig. 2a). A similar spatial pattern is seen in the following spring (Fig. 2b). Relative to conditions in the winter and spring seasons, differences in the following summer mean SIC are less significant (not shown).

Fig. 1.

Normalized time series of winter area-weighted regionally averaged SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W); red lines represent its standard deviation of 0.8 (−0.8) and “1980” refers to the winter of 1979/80.

Fig. 1.

Normalized time series of winter area-weighted regionally averaged SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W); red lines represent its standard deviation of 0.8 (−0.8) and “1980” refers to the winter of 1979/80.

Fig. 2.

Differences in seasonal mean SIC between heavy and light sea ice cases for (a) winter and (b) the ensuing spring. Intervals are 15%, and yellow and green shading areas denote SIC differences at the 95% and 99% significance levels, respectively. The region marked by the red box is used to calculate the winter area-weighted regionally averaged SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W).

Fig. 2.

Differences in seasonal mean SIC between heavy and light sea ice cases for (a) winter and (b) the ensuing spring. Intervals are 15%, and yellow and green shading areas denote SIC differences at the 95% and 99% significance levels, respectively. The region marked by the red box is used to calculate the winter area-weighted regionally averaged SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W).

Figure 3 shows differences in seasonal mean 500-hPa geopotential heights between the heavy and light sea ice cases. In winter, positive height anomalies cover the midlatitude North Atlantic and negative anomalies over the northern high latitudes, and there is an anomalous negative center over southern Greenland (Fig. 3a). This type of spatial pattern corresponds to an anomalous cyclonic wind field in the lower troposphere. The northerly anomalies west of Greenland and southerly anomalies over the Greenland–Barents Seas (not shown) are dynamically consistent with the spatial distribution of winter SIC anomalies indicated in Fig. 2a. The circulation pattern over the North Atlantic resembles the NAO. The area-weighted regionally averaged winter SIC is significantly correlated with the winter mean NAO index (0.42 at the 95% significance level).

Fig. 3.

Differences in seasonal mean 500-hPa heights between heavy and light sea ice cases for (a) winter and the ensuing (b) spring and (c) summer, and (d) summer 500-hPa height anomalies, derived from a linear regression on the winter area-weighted regionally averaged SIC west of Greenland. (e)–(h) As in (a)–(d), respectively, but derived from the JRA-25 data. Magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively. Intervals are 10 gpm in (a)–(c) and (e)–(g) and 5 gpm in (d) and (h).

Fig. 3.

Differences in seasonal mean 500-hPa heights between heavy and light sea ice cases for (a) winter and the ensuing (b) spring and (c) summer, and (d) summer 500-hPa height anomalies, derived from a linear regression on the winter area-weighted regionally averaged SIC west of Greenland. (e)–(h) As in (a)–(d), respectively, but derived from the JRA-25 data. Magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively. Intervals are 10 gpm in (a)–(c) and (e)–(g) and 5 gpm in (d) and (h).

In the following spring, significant height anomalies appear over the Siberian marginal seas of the Arctic Ocean, Far East, and Pacific sector (Fig. 3b). Over the midlatitude North Atlantic, height anomalies display a dipole structure, and opposing anomalous centers are respectively located southeast of Newfoundland and over the eastern North Atlantic. In the following summer, height anomalies display a wave train structure, and there are two positively anomalous centers over western Europe and northwest of the Asian continent, with two negatively anomalous centers over eastern Europe and East Asia (Fig. 3c). A regression map of summer mean 500-hPa heights (Fig. 3d), regressed on the winter area-weighted regionally averaged SIC, closely resembles that indicated in Fig. 3c. This implies that winter SIC west of Greenland is significantly correlated with summer atmospheric circulation anomalies over northern Eurasia. Very similar results were noted in the JRA-25 data (Figs. 3e–h).

There are apparent differences in summer rainfall in the mid- to high latitudes of Eurasia (Fig. 4). Corresponding to a heavy SIC winter, increased summer rainfall is observed from north of Europe extending southeastwards to the midlatitude East Asia, particularly east of 100°E between 40° and 55°N, where differences exceed 60 mm. Additionally, summer rainfall differences also exceed 60 mm in part of eastern Europe. In contrast, decreased summer rainfall occurs in Europe west of 40°E and south of 60°N and the high latitudes of the Asian continent. Regions where summer rainfall differences are statistically significant coincide with anomalous centers of 500-hPa heights (Figs. 3c,g).

Fig. 4.

Differences in summer mean land rainfall between heavy and light sea ice cases; magenta and green shaded areas represent anomalies at the 95% and 99% significance levels, respectively, and intervals are 20 mm.

Fig. 4.

Differences in summer mean land rainfall between heavy and light sea ice cases; magenta and green shaded areas represent anomalies at the 95% and 99% significance levels, respectively, and intervals are 20 mm.

4. Dominant patterns of atmospheric circulation over northern Eurasia: A dynamical perspective

The anomalous patterns shown in Figs. 3c and 3g actually appear to be closely associated with the dominant pattern of atmospheric circulation over the mid- to high latitudes of Eurasia. To demonstrate this point, we construct a complex Hermitian matrix based on area-weighted monthly mean 850-hPa wind anomalies over the domain 40°–70°N and 0°–120°E (the spatial resolution is 2.5° × 2.5°) to extract dominant patterns of 850-hPa wind variability from 1979 to 2010, for a total of 384 months. This vector analysis method is similar to that used in other studies (e.g., Wu et al. 2012) and is known as complex vector empirical orthogonal function analysis (CVEOF). For more details of this method, see Wu et al. (2012).

The leading CVEOF accounts for 20% of the total anomalous kinetic energy. To illustrate the leading pattern's spatial evolution, composite analyses were conducted for the four typical leading phase ranges: the “0°” phase ( < 45° or ≥ 315°), “90°” phase (135° > ≥ 45°), “180°” phase (225° > ≥ 135°), and “270°” phase (315° > ≥ 225°). The composite analyses are carried out using all of the summer (JJA) data (96 months), and the four typical leading phase ranges respectively contain 22, 24, 22, and 28 cases.

For the “0°” phase of the leading wind pattern, there are two pairs of anomalous cyclones and anticyclones respectively over the area from the northeastern North Atlantic to 90°E and over East Asia, the latter obviously weaker than the former (Fig. 5a). Corresponding 500-hPa height anomalies show a wave train structure over northern Eurasia, and there are positive anomalous centers over both the Ural Mountains and Okhotsk Sea with a negative anomalous center between them (Fig. 6a). When the leading wind pattern reaches its “90°” phase, there are anomalous cyclones over both western Europe and northern Asian continent–Siberian marginal seas of the Arctic Ocean, with an anomalous anticyclone between them (Fig. 5b). Meanwhile, there is a weak anomalous anticyclone southeast of Lake Baikal. The spatial distribution of corresponding 500-hPa height anomalies is dynamically consistent with the 850-hPa wind anomalies (Fig. 6b). When the leading wind pattern is in its “180°” (“270°”) phase, the structure is nearly opposite to that of its “0°” (“90°”) phase (Figs. 5c,d and 6c,d).

Fig. 5.

(a) Composite of summer monthly mean 850-hPa wind anomalies (m s−1) for the “0°” phase ( < 45° or ≥ 315°) of the leading wind pattern. (b)–(d) As in (a), but for the “90°” phase (135° > ≥ 45°), “180°” phase (225° > ≥ 135°), and “270°” phase (315° > ≥ 225°), respectively.

Fig. 5.

(a) Composite of summer monthly mean 850-hPa wind anomalies (m s−1) for the “0°” phase ( < 45° or ≥ 315°) of the leading wind pattern. (b)–(d) As in (a), but for the “90°” phase (135° > ≥ 45°), “180°” phase (225° > ≥ 135°), and “270°” phase (315° > ≥ 225°), respectively.

Fig. 6.

As in Fig. 5, but for the composite of summer monthly mean 500-hPa anomalies for the four different phase ranges; intervals are 5 gpm.

Fig. 6.

As in Fig. 5, but for the composite of summer monthly mean 500-hPa anomalies for the four different phase ranges; intervals are 5 gpm.

Figure 5 shows that the leading wind pattern consists of two subpatterns, and that the corresponding 500-hPa height anomalies display different spatial structures (Fig. 6). Here, the two subpatterns are termed the Ural Mountain–Okhotsk Sea (UO) pattern and the Eurasian (EA) pattern, respectively. As in Wu et al. (2012), the real and imaginary parts of the leading complex principal component can be regarded as two indices of intensity that characterize the UO and EA patterns, respectively. Thus, the UO pattern incorporates the “0°” and “180°” phases of the leading wind pattern, and a positive (negative) phase of the UO pattern corresponds to the “0°” (“180°”) phase. Similarly, the EA pattern incorporates the “90°” and “270°” phases of the leading wind pattern, and a positive (negative) phase of the EA pattern corresponds to the “90°” (“270°”) phase. From the “0°” to “270°” phases, the positions of troughs and ridges in summer mean 500-hPa heights exhibit a clockwise migration process in the mid- to high latitudes (not shown). Thus, the leading wind pattern reflects dominant physical features of the evolution in the positions of troughs and ridges of heights.

Figure 7 shows the normalized intensity time series of the UO and EA patterns averaged for the summer months (JJA). The correlation between them is not significant, and the UO pattern does not show any trend. The EA pattern, however, shows a significant trend (at the marginal 99% significance level). This short-term trend is associated with interdecadal (or multidecadal) variability of the climate system, which indicates that the external forcing plays an important role. In fact, the EA pattern is significantly correlated with the previous winter's area-weighted regionally averaged SIC west of Greenland (49.5°–79.5°N, 49.5°–85.5°W) (r = −0.65; after removing their linear trends the correlation is −0.61 at the 99% significance level). A further analysis indicates that during the period from 1994 to 2010, the correlation coefficient between the EA pattern and the winter SIC index west of Greenland is also significant (r = −0.501). A very similar EA pattern is also detected in the JRA-25 data, and this wind pattern is also significantly negatively correlated with the previous winter's SIC index (−0.50, after removing their linear trends). Consequently, winter SIC west of Greenland is closely correlated with the dominant pattern of the prevailing atmospheric circulation in summer.

Fig. 7.

(a) Normalized intensity time series of the UO pattern (the real part of the leading complex principal component), averaged for three summer months (JJA), the red line is its trend. (b) As in (a), but for the EA pattern (the imaginary part of the leading complex principal component).

Fig. 7.

(a) Normalized intensity time series of the UO pattern (the real part of the leading complex principal component), averaged for three summer months (JJA), the red line is its trend. (b) As in (a), but for the EA pattern (the imaginary part of the leading complex principal component).

5. Discussion of possible mechanisms

In this section, we explore possible mechanisms for the observed association between summer atmospheric circulation anomalies over northern Eurasia and the previous winter SIC west of Greenland. Spring height differences exhibit a baroclinic structure north of 55°N, and positive height anomalies emerge in the lower troposphere below 850 hPa with negative anomalies aloft (Fig. 8a). Such a vertical distribution of spring height anomalies differs from that of the previous winter when height anomalies display a dominant quasi-barotropic structure (not shown). The corresponding wave activity flux shows a coherent upward propagation north of 60°N, and there is a branch of wave activity flux propagating southward in the troposphere above 700 hPa. Thus, via energy propagation the winter–spring persistent and strong SIC conditions contribute to positive height anomalies over 40°–50°N, 30°–60°W (Fig. 3b). Positive 500-hPa height anomalies then display an apparent eastward propagation after the following March (Fig. 8b), and when positive height anomalies reach the western coast of Europe, the EA pattern is excited over Eurasia (Fig. 3c). The 500-hPa height differences in the following June and July also support this point (Fig. 9).

Fig. 8.

(a) Differences in the ensuing spring mean heights between heavy and light sea ice cases and corresponding meridional and vertical (×0.17) wave activity flux (m2 s−2) along the 55°W vertical slice. Contour intervals are 5 gpm; magenta shading represents height differences at the 95% significance level. (b) Evolution of 500-hPa height differences between heavy and light sea ice cases at 50°N; magenta and green shading indicates differences at the 95% and 99% significance levels, respectively.

Fig. 8.

(a) Differences in the ensuing spring mean heights between heavy and light sea ice cases and corresponding meridional and vertical (×0.17) wave activity flux (m2 s−2) along the 55°W vertical slice. Contour intervals are 5 gpm; magenta shading represents height differences at the 95% significance level. (b) Evolution of 500-hPa height differences between heavy and light sea ice cases at 50°N; magenta and green shading indicates differences at the 95% and 99% significance levels, respectively.

Fig. 9.

Differences in the ensuing (a) June and (b) July 500-hPa heights between heavy and light sea ice cases; magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively, and intervals are 10 gpm.

Fig. 9.

Differences in the ensuing (a) June and (b) July 500-hPa heights between heavy and light sea ice cases; magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively, and intervals are 10 gpm.

In the mid- to high latitudes of the North Atlantic, spring positive SLP anomalies show a fan-shaped pattern (Fig. 10a), and the largest positive SLP anomaly is located just over the southern Labrador Sea, corresponding to heavy SIC conditions (Fig. 2b). We calculate the regionally averaged (47.5°–57.5°N, 25°–60°W) spring SLP (see Fig. 10a) as an index to approximately estimate the strength of the anomalous anticyclone in the lower troposphere (not shown). The SLP index is significantly correlated with the previous winter SIC index (r = 0.43, at the 95% significance level). Spring 500-hPa height anomalies, derived from a linear regression on the spring SLP index, are illustrated in Fig. 10b. Significantly positive height anomalies are seen over the northwestern North Atlantic and northern Europe, and a negative anomalous center is located over northern Asian continent. Spring SLP anomalies over the northern North Atlantic connect the ensuing summer atmospheric circulation anomalies over Eurasia (Fig. 10c), which shows a great similarity to the spatial distribution of summer 500-hPa height anomalies associated with the winter SIC index (Fig. 3d). Compared with the previous spring, the anomalous center in the following summer shifts downstream (Figs. 10b,c). Thus, spring atmospheric circulation anomalies over the northern North Atlantic serve as a bridge connecting the previous winter SIC and the ensuing summer circulation anomalies over Eurasia.

Fig. 10.

(a) Spring (March–May) mean SLP differences between heavy and light sea ice cases. (b) Spring and (c) summer 500-hPa height anomalies, derived from a linear regression on the spring regionally [47.5°–57.5°N, 25°–60°W, marked by the red box in (a)] averaged SLP. Magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively. Intervals are 0.5 hPa in (a) and 5 gpm in (b) and (c).

Fig. 10.

(a) Spring (March–May) mean SLP differences between heavy and light sea ice cases. (b) Spring and (c) summer 500-hPa height anomalies, derived from a linear regression on the spring regionally [47.5°–57.5°N, 25°–60°W, marked by the red box in (a)] averaged SLP. Magenta and green shading areas represent anomalies at the 95% and 99% significance levels, respectively. Intervals are 0.5 hPa in (a) and 5 gpm in (b) and (c).

To further verify the above diagnostic analysis results, numerical simulation experiments were conducted using the ECHAM5 model with 12 different atmospheric initial conditions that derived from a 30-yr control run with climatological monthly SST and SIC as the external forcing. The SIC experiment was designed as a 30-yr simulation with the climatological SST and observed Northern Hemisphere monthly SIC from 1978 to 2007 as the external forcing. We analyze the simulated differences between heavy and light SIC conditions, and the composite cases are same as that in the section 3 but from 1978 to 2007. Thus, the composite analyses respectively contain 72 heavy and 84 light SIC cases for each season. It should be emphasized that although simulated differences in the ensuing spring mean SLP and 500-hPa height between heavy and light sea ice cases (not shown) support the observational association over the North Atlantic, their amplitudes are generally too weak. Through inspecting differences between heavy and light sea ice cases in each 30-yr simulation, it is found that only six simulations, to a great extent, can reproduce the major features of the observed associations over the North Atlantic during the ensuing spring. Thus, we analyze the six integrations, and the composite analyses respectively contain 36 heavy and 42 light SIC cases for each season. This implies that initial atmospheric conditions influence the atmospheric response to the SIC forcing, and this issue needs to be investigated in future studies.

In winter, influenced by SIC forcing, significant negative surface air temperature (SAT) anomalies are seen from the Baffin Bay extending southward to east of Newfoundland with the concurrent positive SAT anomalies over the Nordic–Barents Seas (Fig. 11a). Such a spatial distribution of winter SAT anomalies is consistent with winter SIC anomalies shown in Fig. 2a. SAT anomalies are not strictly confined to the vicinity of SIC anomalies, and the model atmosphere can transfer the impact of SIC into the regions far away. Winter SLP and 500-hPa height anomalies exhibit a quasi-barotropic structure, and significant positive anomalies are located over the northern high latitudes (Figs. 11b,c). Thus, the simulated impact of SIC on the winter atmosphere is consistent with that in the previous studies (Alexander et al. 2004; Deser et al. 2004), similar to a negative phase of the AO. Apparently, the simulation results differ from the observational associations shown in Fig. 3a.

Fig. 11.

Differences in simulated winter mean (a) SAT, (b) SLP, and (c) 500-hPa height between heavy and light sea ice cases (SIC experiment). Magenta and green shading represents differences at the 95% and 99% significance levels, respectively; intervals are 1° in (a), 0.5 hPa in (b), and 5 gpm in (c).

Fig. 11.

Differences in simulated winter mean (a) SAT, (b) SLP, and (c) 500-hPa height between heavy and light sea ice cases (SIC experiment). Magenta and green shading represents differences at the 95% and 99% significance levels, respectively; intervals are 1° in (a), 0.5 hPa in (b), and 5 gpm in (c).

In the following spring, the spatial distribution of SAT anomalies over both sides of Greenland closely resembles that in the previous winter (Fig. 12a), and is consistent with spring SIC anomalies (Fig. 2b). Significant negative SAT anomalies are seen over the northern Asian continent far from the SIC forcing area, implying spring colder than normal northern Asian continent. Positive SLP anomalies cover most of the northern North Atlantic, and there is a significant positive anomalous center over the mid- to high latitudes of the North Atlantic (Fig. 12b). Spring positive 500-hPa height anomalies occupy the northwestern North Atlantic, and there is a significant positive anomalous center over east to Newfoundland (Fig. 12c). Thus, simulated SLP and 500-hPa height anomalies over the northern North Atlantic are in agreement with the observed associations shown in Figs. 10a and 3b. SLP and 500-hPa height anomalies display a baroclinic structure over the region west of Greenland, also consistent with that shown in Fig. 9a. Additionally, simulated spring 500-hPa height differences closely resemble the regression analysis result of reanalysis data over the North Atlantic and Eurasia shown in Fig. 10b. Thus, the experiment results support the observed spring associations over the northern North Atlantic, the Labrador Sea, and Davis Strait. In the ensuing summer, relative to the previous spring, simulated 500-hPa height differences further shift downstream over the mid- to high latitudes of Eurasia (Fig. 13), and the negative anomalous center migrates from west to Lake Baikal in spring to east of Japan in summer. Indeed, there are obvious differences between simulated spring anomalies and the observations over the Siberian marginal seas of the Arctic Ocean and northern Eurasia (Figs. 3b, 10a, and 12b,c). Those differences and the fact that only six simulations can reproduce the major features of observed atmospheric circulation anomalies over the North Atlantic and northern Eurasia imply that besides impacts of the atmospheric initial conditions and other external forcing factors such as SST and snow cover, the influence of internal variability and model uncertainties may be rather large.

Fig. 12.

As in Fig. 11, but for the ensuing spring; intervals are 1.0° in (a), 0.3 hPa in (b), and 5 gpm in (c).

Fig. 12.

As in Fig. 11, but for the ensuing spring; intervals are 1.0° in (a), 0.3 hPa in (b), and 5 gpm in (c).

Fig. 13.

As in Fig. 12c, but for the following summer; intervals are 2 gpm.

Fig. 13.

As in Fig. 12c, but for the following summer; intervals are 2 gpm.

Additionally, above average winter SIC west of Greenland corresponds to horseshoe-like SST anomalies east of 70°W and north of 20°N, with weak positive SST anomalies in 30°–45°N and 40°–63°W and negative SST anomalies outside this region (Fig. 14a). Similar SST anomalies are also observed in the following spring, with strengthened positive SST anomalies (Fig. 14b). Additionally, the spring SST anomalies associated with the ensuing EA pattern exhibit a similar spatial pattern (Fig. 14c). Consequently, besides the winter SIC conditions west of Greenland, horseshoe-like SST anomalies may contribute to spring–summer atmospheric circulation anomalies over the North Atlantic as well as northern Eurasia.

Fig. 14.

Differences in (a) winter and (b) the ensuing spring mean SST between heavy and light sea ice cases. (c) The spring SST anomalies, derived from a linear regression on the normalized time series of the EA pattern (after detrending). Magenta and green shading represents anomalies at the 95% and 99% significance levels, respectively.

Fig. 14.

Differences in (a) winter and (b) the ensuing spring mean SST between heavy and light sea ice cases. (c) The spring SST anomalies, derived from a linear regression on the normalized time series of the EA pattern (after detrending). Magenta and green shading represents anomalies at the 95% and 99% significance levels, respectively.

Many previous studies investigated impacts of SST anomalies in the North Atlantic on the atmospheric circulation over the Arctic and Eurasia (Peng and Whitaker 1999; Peng et al. 2003; Li 2004; Wu et al. 2009; R. Wu et al. 2011; and many others). Peng and Whitaker (1999) indicated that SST anomalies can induce two interacting anomalous forcing: diabatic heating and eddy forcing. Peng et al. (2003) further summarized the preliminary process of the eddy forcing mechanism (the eddy feedback mechanism): the SST tripole pattern in the North Atlantic initially induces an anomalous heating, which drives a heating-forced anomalous flow; the latter interacts with the storm tracks resulting in an anomalous eddy vorticity forcing; the eddy forcing drives an eddy-forced NAO-like anomalous flow, which in turn influence the heating. We can apply this eddy feedback mechanism to explain a role of SST anomalies though the horseshoe-like SST pattern in this study obviously differs from the SST tripole pattern of Peng et al. (2003) (see their Fig. 1).

The winter heavy SIC condition west of Greenland and concurrent SST anomalies in the mid- to high latitudes of the North Atlantic would persist into the ensuing spring (Figs. 14a,b), and persisting SST and SIC anomalies enhance the baroclinicity in the lower troposphere over the high latitudes of the North Atlantic (Figs. 8a, 3b, and 10a) and associated interactions between eddies and zonal flow at upper levels, leading to a barotropic pattern over the midlatitudes of the North Atlantic that strengthens the subtropical jet along the coast of North America (Losada et al. 2007).

On the other hand, the atmospheric response to the eddy forcing can, in turn, modify the heating, which influences SST anomalies. Thus, spring positive 500-hPa height anomalies correspond to anomalous anticyclone in the lower troposphere (not shown), which further enhances spring positive SST anomalies in midlatitudes of the North Atlantic (Fig. 14b). Additionally, a strengthened SST gradient near Newfoundland favors an enhanced baroclinic response in storm tracks (Lopez et al. 2000), which shifts downstream to feedback the mean flow and induce a nonlocal response (Kvamstø et al. 2004). Wu et al. (2009) proposed that anomalous NAO in spring can induce a SST tripole pattern in the North Atlantic that persists into the following summer and excites an atmospheric teleconnection pattern over the northern Eurasia. Certainly, the present study is to explore the relationship between winter SIC and the ensuing summer atmospheric circulation over Eurasia, and the more detailed interaction mechanism between SST anomalies and the atmosphere is beyond the scope of the present study.

It is noteworthy that in the midtroposphere over the northwestern North Atlantic, positive height anomalies are weak and there is only a very small region where spring height differences are at the 95% significance level (Fig. 3b). There are two possible reasons that may explain this phenomenon: 1) the atmospheric response to SST and SIC anomalies in the North Atlantic is nonlinear (Sutton et al. 2001), and 2) similar to the deduction by Cassou et al. (2004), tropical SST anomalies might counteract impacts of extratropical SST and SIC on the spring atmospheric circulation south of Newfoundland. The relationship between tropical SST and SIC anomalies and their relative contributions to spring atmospheric circulation anomalies are not yet clear. Moreover, although the SIC experiment, to a great extent, supports the observational association, the simulated anomalous amplitudes are weaker relative to the observations. A possible reason is that the experiment does not contain the impact of concurrent North Atlantic SST anomalies.

The observational analysis and simulations presented here can certainly not physically diagnose the impact mechanism of SIC on the atmosphere and differentiate the relative contributions of SIC and SST to atmospheric circulation anomalies, and these critical issues need to be further investigated by using numerical simulations with the SIC and SST as external forcings. Although these issues still remain unclear, we can conclude the following: Winter SIC west of Greenland is a potential precursor for summer atmospheric circulation patterns and corresponding rainfall anomalies over northern Eurasia.

6. Summary

We have examined the relationship between winter SIC west of Greenland and the ensuing summer atmospheric circulation anomalies over northern Eurasia from 1979 to 2010. The results show that above average winter SIC conditions west of Greenland coincide with horseshoe-like SST anomalies in the North Atlantic, with both anomalies persisting into the following spring season. Such anomalies feed back on the ensuing spring atmosphere over the North Atlantic, acting as a bridge linking winter–spring SIC and SST anomalies and summer atmospheric circulation anomalies over northern Eurasia. Thus, winter SIC west of Greenland is a precursor for the ensuing summer atmospheric circulation and rainfall anomalies over northern Eurasia. Indeed, this study provides only observational evidence and a discussion of possible mechanisms, and the more precise feedback of SIC and SST on the atmosphere needs to be investigated in the future using numerical simulations.

Acknowledgments

Comments of three anonymous reviewers have led to a significant improvement of this paper. This study was supported by Chinese Project (GYHY200906017), National Natural Science Foundation of China (Grants 41221064 and 40875052), State Oceanic Administration Project (201205007), and the Basic Research Foundation of the Chinese Academy of Meteorological Sciences (CAMS) (2010Z003).

REFERENCES

REFERENCES
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
.
Balmaseda
,
M. A.
,
L.
Ferranti
,
F.
Molteni
, and
T. N.
Palmer
,
2010
:
Impact of 2007 and 2008 Arctic ice anomalies on the atmospheric circulation: Implications for long-range predictions
.
Quart. J. Roy. Meteor. Soc.
,
136
,
1655
1664
.
Cassou
,
C.
,
L.
Terray
,
J. W.
Hurrell
, and
C.
Deser
,
2004
:
North Atlantic winter climate regimes: Spatial asymmetry, stationarity with time, and oceanic forcing
.
J. Climate
,
17
,
1055
1068
.
Chen
,
M.
,
P.
Xie
,
J. E.
Janowiak
, and
P. A.
Arkin
,
2002
:
Global land precipitation: A 50-yr monthly analysis based on gauge observations
.
J. Hydrometeor.
,
3
,
249
266
.
Deser
,
C.
, and
M.
Blackmon
,
1993
:
Surface climate variations over the North Atlantic Ocean during winter: 1900–1989
.
J. Climate
,
6
,
1743
1753
.
Deser
,
C.
,
M.
Holland
,
G.
Reverdin
, and
M.
Timlin
,
2002
:
Decadal variations in Labrador sea ice cover and North Atlantic sea surface temperature
.
J. Geophys. Res.
,
107
(
C5
),
doi:10.1029/2000JC000683
.
Deser
,
C.
,
G.
Magnusdottir
,
R.
Saravanan
, and
A. S.
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
.
Honda
,
M.
,
K.
Yamazaki
,
H.
Nakamura
, and
H.
Takeuchi
,
1999
:
Dynamic and thermodynamic characteristics of atmospheric response to anomalous sea-ice extent in the Sea of Okhotsk
.
J. Climate
,
12
,
3347
3358
.
Honda
,
M.
,
J.
Inous
, and
S.
Yamane
,
2009
:
Influence of low Arctic sea-ice minima on anomalously cold Eurasian winters
.
Geophys. Res. Lett.
,
36
,
L08707
,
doi:10.1029/2008GL037079
.
Kumar
,
A.
, and
Coauthors
,
2010
:
Contribution of sea ice loss to Arctic amplification
.
Geophys. Res. Lett.
,
37
,
L21701
,
doi:10.1029/2010GL045022
.
Kvamstø
,
N. G.
,
P.
Skeie
, and
D. B.
Stephenson
,
2004
:
Impact of Labrador Sea-ice extent on the North Atlantic Oscillation
.
Int. J. Climatol.
,
24
,
603
612
.
Li
,
S.
,
2004
:
Impact of northwest Atlantic SST anomalies on the circulation over the Ural Mountains during early winter
.
J. Meteor. Soc. Japan
,
82
,
971
988
.
Livezey
,
R.
, and
W. Y.
Chen
,
1983
:
Statistical field significance and its determination by Monte Carlo techniques
.
Mon. Wea. Rev.
,
111
,
46
59
.
Lopez
,
P.
,
T.
Schmith
, and
E.
Kaas
,
2000
:
Sensitivity of the Northern Hemisphere circulation to North Atlantic SSTs in the ARPÈGE climate AGCM
.
Climate Dyn.
,
16
,
535
547
.
Losada
,
T.
,
B.
Rodríguez-Fonseca
,
C. R.
Mechoso
, and
H.-Y.
Ma
,
2007
:
Impacts of SST anomalies on the North Atlantic atmospheric circulation: A case study for the northern winter 1995/1996
.
Climate Dyn.
,
29
,
807
819
,
doi:10.1007/s00382-007-0261-x
.
Magnusdottir
,
G.
,
C.
Derser
, and
R.
Saravanan
,
2004
:
The effects of North Atlantic SST and sea ice anomalies in the winter circulation in CCM3. Part I: Main features and storm track characteristics of the response
.
J. Climate
,
17
,
857
876
.
Onogi
,
K.
, and
Coauthors
,
2007
:
The JRA-25 Reanalysis
.
J. Meteor. Soc. Japan
,
85
,
369
432
.
Peng
,
S.
, and
J. S.
Whitaker
,
1999
:
Mechanisms determining the atmospheric response to midlatitude SST anomalies
.
J. Climate
,
12
,
1393
1408
.
Peng
,
S.
,
W. A.
Robinson
, and
S.
Li
,
2003
:
Mechanisms for the NAO responses to the North Atlantic SST tripole
.
J. Climate
,
16
,
1987
2004
.
Petoukhov
,
V.
, and
V. A.
Semenov
,
2010
:
A link between reduced Barents-Kara sea ice and cold winter extremes over northern continents
.
J. Geophys. Res.
,
115
,
D21111
,
doi:10.1029/2009JD013568
.
Rigor
,
I. G.
,
J. M.
Wallace
, and
R. L.
Colony
,
2002
:
Response of sea ice to the Arctic Oscillation
.
J. Climate
,
15
,
2648
2663
.
Roeckner
,
E.
, and
Coauthors
,
2003
: The atmospheric general circulation model ECHAM5. Part I: Model description. Max Planck Institute for Meteorology Rep. 349, 127 pp.
Screen
,
J. A.
, and
I.
Simmonds
,
2010
:
The central role of diminishing sea ice in recent Arctic temperature amplification
.
Nature
,
464
,
1334
1337
.
Smith
,
T.
, and
R.
Reynolds
,
2003
:
Extended reconstruction of global sea surface temperature based on COADS data (1854–1997)
.
J. Climate
,
16
,
1495
1510
.
Sutton
,
R. T.
,
W. A.
Norton
, and
S. P.
Jewson
,
2001
:
The North Atlantic oscillation—What role for the ocean?
Atmos. Sci. Lett.
,
1
,
89
100
.
Wu
,
B.
,
R.
Huang
, and
D.
Gao
,
1999
:
Effects of variation of winter sea-ice area in Kara and Barents Seas on East Asian winter monsoon
.
Acta Meteor. Sin.
,
13
,
141
153
.
Wu
,
B.
,
J.
Wang
, and
J. E.
Walsh
,
2004
:
Possible feedback of winter sea ice in the Greenland and Barents Seas on the local atmosphere
.
Mon. Wea. Rev.
,
132
,
1868
1876
.
Wu
,
B.
,
J.
Su
, and
R.
Zhang
,
2011
:
Effects of autumn–winter arctic sea ice on winter Siberian high
.
Chin. Sci. Bull.
,
56
,
3220
3228
,
doi:10.1007/s11434-011-4696-4
.
Wu
,
B.
,
J.
Overland
, and
R.
D'Arrigo
,
2012
:
Anomalous arctic surface wind patterns and their impacts on September sea ice minima and trend
.
Tellus
,
64A
,
18590
,
doi:10.3402/tellusa.v64i0.18590
.
Wu
,
R.
,
S.
Yang
,
S.
Liu
,
L.
Sun
,
Y.
Lian
, and
Z.
Gao
,
2011
:
Northeast China summer temperature and North Atlantic SST
.
J. Geophys. Res.
,
116
,
D16116
,
doi:10.1029/2011JD015779
.
Wu
,
Z.
,
B.
Wang
,
J.
Li
, and
F.
Jin
,
2009
:
An empirical seasonal prediction model of the East Asian summer monsoon using ENSO and NAO
.
J. Geophys. Res.
,
114
,
D18120
,
doi:10.1029/2009JD011733
.