1. Introduction
Over the last several decades, the Arctic climate has undergone substantial and rapid changes. The warming trend in the Arctic is more than twice that of the global mean (Serreze and Francis 2006; Serreze and Barry 2011), a phenomenon referred to as the Arctic amplification. Although ubiquitous, the greatest warming has been observed in the cold season (from October to April), especially during winter when the contribution of shortwave radiation is negligible (Graversen et al. 2008; Screen and Simmonds 2010a; Dai et al. 2019). Coincident with Arctic amplification, Arctic sea ice has experienced an extensive decline in both its extent and thickness at an accelerating rate (Stroeve et al. 2007; Kinnard et al. 2011; Lang et al. 2017). Many mechanisms have been proposed to explain the abnormal warming in the Arctic. In particular, Dai et al. (2019) showed that sea ice loss, which is necessary for the other mechanisms, is the main cause of the Arctic amplification of surface warming under increasing greenhouse gases. However, some studies have also highlighted the importance of local feedbacks, as well as heat and moisture transport for Arctic amplification (e.g., Zhang et al. 2008, 2012; Pithan and Mauritsen 2014; Screen and Francis 2016; Messori et al. 2018; Goosse et al. 2018; Mewes and Jacobi 2019; Russotto and Biasutti 2020). Since changes in climate variability are arguably more important for society and ecosystems than changes in mean climate (Katz and Brown 1992; Screen 2014), there is growing concern about how Arctic temperatures fluctuate on different time scales and the underlying mechanisms of variability.
Despite a large focus on the changing Arctic surface air temperature (SAT), previous studies tend to be concerned with decadal and multidecadal time scales (Jungclaus and Koenigk 2009; Levitus et al. 2009; Day et al. 2012; Miles et al. 2014; Van der Linden et al. 2016) and have paid less attention to interannual variability. However, to accurately estimate the long-term warming trends, assessing the interannual variability superimposed on such trends is necessary. Moreover, the interannual SAT variability is important in assessing the occurrence and magnitude of yearly extremes and developing seasonal to annual climate predictions. Especially in winter, SAT variability in the Arctic may induce anomalous climate conditions at midlatitudes (Dai and Deng 2021; Cohen et al. 2020; He et al. 2020; Li et al. 2020; Gao et al. 2014). Enhanced knowledge about the features of interannual variability in wintertime Arctic SAT and the associated mechanisms is therefore vital.
On the interannual time scale, sea ice–related surface processes may also play an important role in SAT variability. Although most of the Arctic is covered by sea ice in winter, there is still significant interannual variability at the margins of the Arctic Ocean on the Atlantic side, and the interannual variability there is greater because the ice layer becomes thinner under rapid Arctic warming (Kay et al. 2011; Holland and Stroeve 2011; Mioduszewski et al. 2019). Due to the larger ocean–atmosphere temperature gradient in winter, sea ice can impact the surface heat fluxes by acting as a lid that opens and closes, thereby modulating the SAT (Screen and Simmonds 2010a,b; Yim et al. 2016; Dai et al. 2019). In modeling studies, the most robust and intuitive response to sea ice loss is local warming (e.g., Deser et al. 2010, 2016; Smith et al. 2017; Ogawa et al. 2018; Screen et al. 2018; Sun et al. 2018; Blackport and Screen 2019; Liang et al. 2020). However, some recent studies have indicated that sea ice variations only have a limited impact on interannual SAT variations in most northern middle- and high-latitude regions (Koenigk et al. 2018; Blackport et al. 2019; He et al. 2020). A better understanding of the role of sea ice in the interannual SAT variability in the wintertime Arctic is thus needed.
In addition to sea ice forcing, atmospheric conditions associated with anomalous atmospheric moisture and downward longwave radiation (DLR) also have important influences on the interannual variations in wintertime Arctic SAT. In winter, clouds have a warming impact on the surface because they reduce surface energy losses by reemitting longwave radiation back to the surface (Shupe and Intrieri 2004). Both synoptic and interdecadal variations in the wintertime Arctic SAT are closely related to water vapor variations. On the synoptic time scale, there is a stark surface temperature difference between the warm opaque cloudy-sky state and the cold radiative clear-sky state (Stramler et al. 2011). On the interdecadal time scale, Rinke et al. (2019) confirmed a robust relative moistening trend in Arctic winter, and many studies demonstrated that the wintertime Arctic warming trend is coincident with DLR enhancement due to poleward atmospheric moisture transport (Zhang et al. 2012; Woods and Caballero 2016; Gong et al. 2017; Lee et al. 2017; Pithan et al. 2018; Mewes and Jacobi 2019). However, the relationship between water vapor and Arctic SAT on the interannual time scale has received less attention. Although Woods et al. (2013) suggested that the interannual variability in moisture intrusions is strongly correlated with variability in the winter-mean surface DLR and skin temperature averaged over the Arctic, further work is required to understand to what extent the interannual variability in wintertime Arctic SAT is driven by the variation in water vapor conditions. Furthermore, some studies show that the Arctic DLR increase is driven by horizontal atmospheric water flux and warm air advection into the Arctic rather than evaporation from the Arctic Ocean (Morrison et al. 2012; Park et al. 2015; Nygård et al. 2019). These studies suggest that atmospheric conditions, especially water vapor and DLR, may play an important role in regulating the interannual variability in the wintertime Arctic SAT.
Motived by the lack of scientific understanding of the relative contributions of atmospheric and sea ice variability to the interannual variability in wintertime Arctic SAT, this study extends previous studies in three main aspects: (i) we examine the characteristics of the interannual variability in wintertime Arctic SAT, and the related characteristics in Arctic sea ice, atmospheric water vapor, radiation, vertical temperature structures, and larger-scale atmospheric circulation; (ii) we also use physically motivated approaches, lead–lag correlations, and partial correlations to assess to what extent the interannual variability in wintertime Arctic SAT is driven by sea ice or atmospheric conditions; and (iii) on the basis of these statistical links, we examine the cause-and-effect relationships among SAT, sea ice, and atmospheric conditions by conducting atmospheric numerical experiments to further disentangle the relative contributions of sea ice and the internal atmospheric variability to the interannual variability in wintertime Arctic SAT.
The remainder of this paper is organized as follows. Details of the data, methods, model, and experimental design are introduced in section 2. The characteristics of the interannual variability in wintertime Arctic SAT and its statistical links with sea ice, water vapor, surface heat flux, radiation, and larger-scale atmospheric circulation are discussed in section 3. Section 4 investigates the physical mechanisms and relative contributions of sea ice–related surface processes and internal atmospheric variability to the interannual variability in wintertime Arctic SAT. A summary and discussion are presented in section 5.
2. Data, methods, and model experiments
a. Reanalysis data
In this study, we used atmospheric variables from 1979 to 2019 from the ERA-Interim reanalysis dataset (Dee et al. 2011). These variables include monthly mean SAT, total column water vapor (TCWV), sea level pressure (SLP), geopotential height, and wind fields at 850, 500, and 200 hPa. Following previous studies (Lee et al. 2017; Gong et al. 2017), we computed the forecasted surface and radiative fluxes using the daily accumulated ERA-Interim values at time steps 3 and 6 for both 0000 and 1200 UTC. To study the effect of heat fluxes on the SAT, the surface turbulent (sensible + latent) heat flux (THF) is defined as positive in the upward direction, whereas the DLR is defined as positive in the downward direction. The data utilized in this study have a horizontal resolution of 0.5° × 0.5°. Additionally, we analyzed temperature and specific humidity data at the pressure level to provide the characteristics of the vertical structure. The pressure-level data are provided at 25-hPa vertical intervals from 1000 to 900 hPa, at 50-hPa intervals from 900 to 700 hPa, and at 100-hPa intervals from 700 to 300 hPa.
Furthermore, the Met Office Hadley Centre sea ice concentration (SIC) and sea surface temperature (SST) dataset (HadISST) from 1979 to 2019 gridded at a 1.0° × 1.0° resolution was also employed in this study (Rayner et al. 2003). Since the main objective of this study is to investigate processes regulating wintertime SAT interannual variations, 9-yr moving averages were removed from both the ERA-Interim and HadISST data at each grid box before further analyses. The period analyzed in this study was 40 winters from 1980 to 2019. Wintertime means were constructed from the monthly means by averaging the data for December–February (DJF). Here, we used the convention that the winter of 1980 ends in February 1980.
b. Methods
The present study employs an empirical orthogonal function (EOF) analysis based on SAT data to extract the dominant patterns of interannual SAT variability. To understand the mechanisms involved in connections among the Arctic SAT, SIC, and atmospheric circulation, composite, correlation, and regression analyses are applied. Particularly, the partial correlation method, which is often used to find the correlation between two variables after removing the effects of another variable (e.g., Ashok et al. 2003; Nicholls 2009; Feng et al. 2010; Li et al. 2013), is imported to describe the relationship between winter SAT, SIC, and TCWV. Furthermore, we also apply the same two independent approaches as Blackport et al. (2019) to infer causality. One is a physically motivated approach based on the direction of surface THF over the Arctic regions, and another approach utilizes a lead–lag analysis. The methods are described in more detail in section 3. Statistical significance was calculated using two-tailed Student’s t tests, and the wind was considered significant only when one of the U and V wind components was significant.
c. Model and experimental design
We use an atmospheric general circulation model (AGCM) to uncover the relative roles of SIC forcing and atmospheric conditions in driving the interannual variability in SAT. The AGCM experiments are conducted by using the Max Planck Institute (MPI) ECHAM5.4 model (Roeckner et al. 2003) at a T63 horizontal resolution (1.875°) with 19 vertical levels extending from the surface to 10 hPa. SST and SIC are specified as surface boundary conditions based on HadISST. The control experiment (CTL) is integrated for 105 years with 1979–2010 monthly climatology for SIC and SST as the lower boundary condition. Based on the studies using the ECHAM5 model, a 3-month spinup is sufficient (Bergman et al. 2012; Nam and Quaas 2012). Here, to ensure the spinup period is enough to balance the inner feedback of the AGCM, the first 5 years are discarded as a spinup period and last 100 years are analyzed.
The impacts of the diminishing Arctic sea ice on air temperature and atmospheric circulation are assessed by comparing two perturbation experiments with different wintertime (DJF) monthly-varying Arctic sea ice and associated SST distributions, with all other external variables held fixed. Both perturbation experiments are started from initial conditions taken from the CTL corresponding to 1 January of the 6th year and are run for 100 years to 1 March of the 106th year so that it spans 100 entire winters (1 December of the 6th year to 28 February of the 106th year, where the first 11 months are removed because of the spinup interval). In the first perturbation experiment, which we call sen_warm, the wintertime SIC over the Arctic (66.5°–90°N) is specified using the average of the eight warm years discussed in section 3 while the ocean surface conditions outside the Arctic are the same as in the CTL run. The second perturbation experiment, referred to as sen_cold, is the same as sen_warm except that Arctic SIC is specified using the average of the nine cold years discussed in section 3. SSTs are set to the monthly climatology, except for regions with substantial sea ice changes from the climatology where the averaged SSTs from the same years as the SIC are used (Screen et al. 2013; Peings and Magnusdottir 2014; Osborne et al. 2017). In DJF, if the SIC in the sen_warm experiment differs by 10% or more from its value in the sen_cold experiment, the SSTs in the sen_warm and sen_cold experiments are changed to the average of warm and cold years in section 3, respectively. Otherwise, the SSTs are set to the climatological values. Thus, only changes in the Arctic SIC and the SST that are directly associated with SIC changes are accounted for. The prescribed SIC and SST differences between the Sen_warm and Sen_cold experiments for the individual winter months are shown in Fig. 1.
Prescribed (a)–(c) SIC and (d)–(f) SST differences between the Sen_warm and Sen_cold experiments (Sen_warm minus Sen_cold) for the individual winter months (from December to the following February). SST differences here are those due to local SIC differences (see section 2).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Prescribed (a)–(c) SIC and (d)–(f) SST differences between the Sen_warm and Sen_cold experiments (Sen_warm minus Sen_cold) for the individual winter months (from December to the following February). SST differences here are those due to local SIC differences (see section 2).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Prescribed (a)–(c) SIC and (d)–(f) SST differences between the Sen_warm and Sen_cold experiments (Sen_warm minus Sen_cold) for the individual winter months (from December to the following February). SST differences here are those due to local SIC differences (see section 2).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
With wintertime Arctic SIC and directly associated SST changes as the only forcing, the impact of wintertime Arctic SIC is assessed by comparing the two perturbation experiments, and the CTL can be used to analyze the impact of internal atmospheric variability on SAT without the presence of SIC and SST variations. It must be clarified that the internal atmospheric variability referred to here also includes the part due to land surface anomalies such as soil moisture. Thus, the so-called internal atmospheric variability here is not purely internal to the atmosphere but also refers to the part of variability independent of SST/sea ice forcing, which is prescribed in the AGCM simulations. While the SAT and other variations in the CTL run are purely due to atmospheric (and land) internal variability as the ocean surface conditions are fixed to monthly climatology, in contrast to those seen in ERA-Interim data (which include contributions from SST and SIC variations), SAT variations in the CTL (and the two perturbation experiments) are likely dampened significantly by the fixed ocean surface conditions. This simulation method has been widely used in previous studies to distinguish the relative roles of surface boundary condition forcing and internal atmospheric variability (e.g., Dong et al. 2018; He et al. 2017), although the use of fixed SST and SIC may dampen the variability in the simulated SAT. This dampening effect is especially important in comparison with observations and reanalysis data, which do not have this dampening effect and include the impact from varying ocean surface conditions.
3. Results from ERA-Interim
a. Interannual Arctic SAT variability and the associated SIC and TCWV anomalies
An EOF analysis of the SAT over the Arctic (65°–90°N) is performed to reveal the dominant features of the Arctic winter SAT at year-by-year time scales. The first leading pattern accounts for 41% of the total variance (Fig. 2a). Spatially, positive anomalies cover most parts of the Arctic Ocean and its marginal seas, and the greatest warming has occurred in the eastern Arctic Basin. To examine the physical mechanism driving the observed SAT interannual variability, years with absolute standard deviation (SD) of the principal component time series of the first leading EOF mode greater than 0.8 are selected (Fig. 2b). Eight winters with SDs greater than 0.8 (1984, 1985, 1990, 1995, 2000, 2005, 2006, and 2012) are selected as warm Arctic winters, and nine winters with SDs lower than −0.8 (1982, 1988, 1997, 1998, 2003, 2004, 2010, 2011, and 2019) are selected as cold Arctic winters.
(a) The spatial distribution of the first leading mode of the EOF of winter (DJF) mean SAT in the Arctic (65°–90°N), and (b) its principal component time series. The SAT data were removed the 9-yr running mean on their own grid before EOF analysis. The red and blue dots in (b) represent the selected eight warm (larger than 0.8 positive SD) and nine cold (smaller than 0.8 negative SD) years for the computation of the composites, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
(a) The spatial distribution of the first leading mode of the EOF of winter (DJF) mean SAT in the Arctic (65°–90°N), and (b) its principal component time series. The SAT data were removed the 9-yr running mean on their own grid before EOF analysis. The red and blue dots in (b) represent the selected eight warm (larger than 0.8 positive SD) and nine cold (smaller than 0.8 negative SD) years for the computation of the composites, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
(a) The spatial distribution of the first leading mode of the EOF of winter (DJF) mean SAT in the Arctic (65°–90°N), and (b) its principal component time series. The SAT data were removed the 9-yr running mean on their own grid before EOF analysis. The red and blue dots in (b) represent the selected eight warm (larger than 0.8 positive SD) and nine cold (smaller than 0.8 negative SD) years for the computation of the composites, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Next, we analyze the composites and correlations to examine the relationship between the SAT and Arctic sea ice and moisture. We first display the differences in SAT, SIC, and TCWV between the warm and cold Arctic winters. The SAT composite pattern (Fig. 3a) is very similar to the leading EOF mode (Fig. 2a). The entire Arctic area is anomalously warm, with the greatest warming occurring in the eastern Arctic Basin, especially over the northern parts of the Barents and Kara Seas extending to the North Pole, and the maximum regional warming is approximately 10°C. It is evident that there is strong spatial coherence between the SAT and TCWV composites; in all warming regions, the overlying TCWV has strongly increased (Fig. 3c). However, SIC reductions are limited in extent to a narrow region near the climatological 15% SIC line (Fig. 3b). Over the eastern Arctic Basin, where the SAT anomalies are most remarkable, with the exception of small parts of the Barents and Kara Seas, the decreases in SIC are almost negligible. These patterns suggest that moisture is likely to be the main driver of SAT interannual variation, whereas the role of sea ice may be relatively minor.
Composites of (a) SAT, (b) SIC, and (c) TCWV. The composites show the differences between warm and cold Arctic winters (warm minus cold). Stippling denotes composites that are significant at the 95% confidence level. The magenta contour in (b) indicate 15% SIC for wintertime climatology (1980–2019).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of (a) SAT, (b) SIC, and (c) TCWV. The composites show the differences between warm and cold Arctic winters (warm minus cold). Stippling denotes composites that are significant at the 95% confidence level. The magenta contour in (b) indicate 15% SIC for wintertime climatology (1980–2019).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of (a) SAT, (b) SIC, and (c) TCWV. The composites show the differences between warm and cold Arctic winters (warm minus cold). Stippling denotes composites that are significant at the 95% confidence level. The magenta contour in (b) indicate 15% SIC for wintertime climatology (1980–2019).
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
To check the impacts of the SIC and TCWV on the Arctic SAT in winter, we computed correlation and partial correlation values between the SAT field and the SIC and TCWV time series. The SIC and TCWV time series are the cosine latitude-weighted area-mean SIC and TCWV over the Eurasian sector of the Arctic (65°–90°N, 0°–140°E; Fig. 4, outlined with black boxes), respectively, and then the 9-yr running means of the SIC and TCWV time series are removed. The reason for the selection of this domain is that the SIC and TCWV, which have the most significant correlations with SAT, are mainly located in this region. It is clear from Fig. 4 that the correlations are strong between both the SIC and TCWV (Figs. 4a,b), and there are few changes in the SAT–TCWV relationship, except for a weaker positive correlation over the Barents and Kara Seas when the effect of sea ice is removed (Fig. 4b vs Fig. 4d). However, the partial correlations between the SAT and SIC after removing the TCWV effect, which is mostly confined to the eastern Arctic Basin, show significantly weaker and smaller patterns than the total correlations (Fig. 4a vs Fig. 4c). This result suggests that the strong correlation between the SAT and SIC, to a large extent, may depend on the TCWV, whereas the relationship between the SAT and TCWV is not heavily influenced by the SIC.
Correlation between the wintertime Arctic SAT and (a) the SIC time series, (b) the TCWV time series, (c) the SIC time series after the removal of the TCWV time series, and (d) the TCWV time series after the removal of the SIC time series. The SIC and TCWV time series are the cosine latitude-weighted area-mean SIC and TCWV over the Eurasian sector of the Arctic (65°–90°N, 0°–140°E, outlined in black), respectively. The 9-yr running means of all of the SAT field, SIC, and TCWV time series are removed. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Correlation between the wintertime Arctic SAT and (a) the SIC time series, (b) the TCWV time series, (c) the SIC time series after the removal of the TCWV time series, and (d) the TCWV time series after the removal of the SIC time series. The SIC and TCWV time series are the cosine latitude-weighted area-mean SIC and TCWV over the Eurasian sector of the Arctic (65°–90°N, 0°–140°E, outlined in black), respectively. The 9-yr running means of all of the SAT field, SIC, and TCWV time series are removed. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Correlation between the wintertime Arctic SAT and (a) the SIC time series, (b) the TCWV time series, (c) the SIC time series after the removal of the TCWV time series, and (d) the TCWV time series after the removal of the SIC time series. The SIC and TCWV time series are the cosine latitude-weighted area-mean SIC and TCWV over the Eurasian sector of the Arctic (65°–90°N, 0°–140°E, outlined in black), respectively. The 9-yr running means of all of the SAT field, SIC, and TCWV time series are removed. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
We also consider time lags to examine the lead–lag relationship among the Arctic SAT, SIC, and TCWV by performing lead–lag regressions. Following Blackport et al. (2019), lead–lag regressions were performed using monthly averaged sea ice indices for December, January, and February and monthly mean SAT and TCWV data from November through March. The regressions were performed after removing the 9-yr running mean for each month and combining (not averaging) the monthly averaged data. A lag of −1 (+1) month corresponds to the regressions between November (January) SAT/TCWV and December sea ice; December (February) SAT/TCWV and January sea ice; and January (March) SAT/TCWV and February sea ice. As shown in Fig. 5, 1 month ahead of the sea ice reductions, there are strong positive SAT and TCWV anomaly patterns similar to the 0-month lag regressions (Fig. 5a vs Fig. 5b, Fig. 5d vs Fig. 5e). In contrast, 1 month after the sea ice reduction, a weak warming and wet anomaly is found only in a narrow region (Figs. 5c,f). This result, combined with the above composite and partial correlation analysis, provides evidence that variations in the TCWV is intimately related to the variation in the Arctic winter SAT on an interannual time scale. In contrast, the striking coherence between sea ice and SAT is limited in extent to the ice retreat regions rather than the widespread area of interannual warming.
SAT and TCWV lead–lag regressions with Arctic (65°–90°N) ice: (a)–(c) SAT and (d)–(f) TCWV regressed on the Arctic winter SIC index at (left) −1-month, (center) 0-month, and (right) +1-month lags. The Arctic winter SIC index was created by calculating the SIC over the Arctic region and averaging over the winter season. The SIC time series were then divided by the SD, and the sign was reversed so that regression maps represent the field associated with a 1-SD reduction in the SIC. A negative lag indicates that the SAT and TCWV lead the sea ice. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
SAT and TCWV lead–lag regressions with Arctic (65°–90°N) ice: (a)–(c) SAT and (d)–(f) TCWV regressed on the Arctic winter SIC index at (left) −1-month, (center) 0-month, and (right) +1-month lags. The Arctic winter SIC index was created by calculating the SIC over the Arctic region and averaging over the winter season. The SIC time series were then divided by the SD, and the sign was reversed so that regression maps represent the field associated with a 1-SD reduction in the SIC. A negative lag indicates that the SAT and TCWV lead the sea ice. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
SAT and TCWV lead–lag regressions with Arctic (65°–90°N) ice: (a)–(c) SAT and (d)–(f) TCWV regressed on the Arctic winter SIC index at (left) −1-month, (center) 0-month, and (right) +1-month lags. The Arctic winter SIC index was created by calculating the SIC over the Arctic region and averaging over the winter season. The SIC time series were then divided by the SD, and the sign was reversed so that regression maps represent the field associated with a 1-SD reduction in the SIC. A negative lag indicates that the SAT and TCWV lead the sea ice. Stippling denotes values that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
b. Surface energy balance
Since the ice cover insulates the relatively warm near-surface ocean from the colder atmosphere above, removing the ice cover reduces the insulating effect, leading to a greater THF entering the atmosphere and thus warming the air near the surface. According to this scenario, if the THF plays a dominant role in the Arctic winter SAT, the regions experiencing the greatest increase in THFs should coincide with the regions of greatest warming. However, composite analysis (Fig. 6a) reveals that there is little spatial coherence between SAT and THF anomalies. This result indicates that the larger SAT anomalies observed over this region are not mainly caused by surface THF. In contrast, it is clear that the DLR anomaly pattern matches reasonably well with not only the SAT (Fig. 3a) but also the TCWV (Fig. 3c) anomaly pattern. Moreover, it is also noteworthy that the THF patterns do not match the SIC pattern, which suggests that the mechanisms of the link between SIC and THF are more than a weaker insulating effect, and the physical mechanisms are discussed in more detail in section 4.
As in Fig. 3, but for composites of the (a) surface turbulent (sensible + latent) heat flux (positive values for flux indicate upward) and (b) downward longwave radiation (positive values for flux indicate downward). The unit is watts per meter squared. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 3, but for composites of the (a) surface turbulent (sensible + latent) heat flux (positive values for flux indicate upward) and (b) downward longwave radiation (positive values for flux indicate downward). The unit is watts per meter squared. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 3, but for composites of the (a) surface turbulent (sensible + latent) heat flux (positive values for flux indicate upward) and (b) downward longwave radiation (positive values for flux indicate downward). The unit is watts per meter squared. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
To further investigate the physical processes by which the surface energy drives SAT variability, we show the monthly composite differences in SAT, SIC, TCWV, DLR, and THF from November to the following March in Fig. 7. Consistent with the earlier results, strong spatial pattern correlations were also found between the SAT and TCWV and DLR for each month from November to March. However, the SIC anomaly remains weak in December even with significant warming, and there is an absence of warming over the Arctic in the following March, despite the presence of SIC and THF anomalies. This result provides a confirmation of the earlier lead–lag regressions and again implies that clouds and water vapor exert significant controls on the interannual variability in the Arctic winter SAT by regulating the DLR, but the variance in the SIC does not substantially contribute to variability in wintertime Arctic SAT on a year-by-year time scale. These findings are in agreement with previous studies (e.g., Lee et al. 2017; Gong et al. 2017; Kim and Kim 2017; Mortin et al. 2016; Woods and Caballero 2016; Park et al. 2015), which found that the Arctic surface temperature trend is driven primarily by downward infrared radiation and not by surface turbulent fluxes.
Monthly composite differences in SAT, TCWV, SIC, DLR, and THF between warm and cold Arctic winters (warm minus cold) from November to the following March. For DLR, positive flux values indicate downward flux; for THF, positive flux values indicate upward flux. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Monthly composite differences in SAT, TCWV, SIC, DLR, and THF between warm and cold Arctic winters (warm minus cold) from November to the following March. For DLR, positive flux values indicate downward flux; for THF, positive flux values indicate upward flux. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Monthly composite differences in SAT, TCWV, SIC, DLR, and THF between warm and cold Arctic winters (warm minus cold) from November to the following March. For DLR, positive flux values indicate downward flux; for THF, positive flux values indicate upward flux. Stippling denotes composites that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
c. Vertical structure and large-scale atmospheric circulation
The above results imply that the question of what processes drive the interannual SAT variability, to a large extent, can be relegated to the same question about moisture. What then drives the interannual variability of moisture? Motivated by the close relationship between moisture, cloud cover, and longwave radiation with atmospheric pressure fields (Morrison et al. 2012; Nygård et al. 2019), we assess how the vertical structure of temperature, moisture, and large-scale atmospheric circulation varies between warm and cold Arctic winters.
The composite differences in the vertical structure of temperature and moisture between warm and cold Arctic winters are presented in Fig. 8. As can be seen, although the largest warming is in the lower level of the troposphere, there is still a significant warming of approximately 2°C at 400 hPa (Fig. 8a). Similarly, significant specific humidity anomalies can also be found as high as 400 hPa (Fig. 8b). In other words, the temperature and specific humidity anomalies are not confined to the near-surface atmosphere but rather extend from the surface to the mid-troposphere. Since retreating sea ice is associated with energy input in the lowermost part of the atmosphere, these vertical structures are likely driven by large-scale atmospheric circulations rather than surface feedbacks.
Composites of the winter meridional vertical section of (a) temperature (K) and (b) specific humidity (0.1 g kg−1) along 0°–140°E. The black solid and dashed contours indicate composites that are significant at the 99% or 95% confidence level, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of the winter meridional vertical section of (a) temperature (K) and (b) specific humidity (0.1 g kg−1) along 0°–140°E. The black solid and dashed contours indicate composites that are significant at the 99% or 95% confidence level, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of the winter meridional vertical section of (a) temperature (K) and (b) specific humidity (0.1 g kg−1) along 0°–140°E. The black solid and dashed contours indicate composites that are significant at the 99% or 95% confidence level, respectively.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of the large-scale atmospheric circulation patterns are shown in Fig. 9. Concomitant with Arctic warming are robust anomalies in atmospheric circulation, such as the high pressures observed over northern Eurasia and the low pressures observed over northeastern Canada and Greenland (Fig. 9a). These patterns favor winds blowing across the Arctic Basin from the North Atlantic to the Beaufort and Chukchi Seas, which enables increased meridional moisture transport from lower latitudes into the Arctic. It is worth noting that positive differences in geopotential heights are also found at 850, 500, and 200 hPa, with centers in northern Eurasia and significant anticyclonic winds from lower latitudes flowing into the Arctic (Figs. 9b–d). Again, sea ice loss does not seem to be the dominant cause of such deep atmospheric circulation anomalies.
(a) As in Fig. 3, but for the composites of SLP (shading; hPa) and winds at 10 m (vectors; m s−1). (b)–(d) As in (a), but for upper-level geopotential heights (m) and winds. The green contours indicate composites of SLP or geopotential heights that are significant at the 95% confidence level, and vectors are shown only for wind anomalies that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
(a) As in Fig. 3, but for the composites of SLP (shading; hPa) and winds at 10 m (vectors; m s−1). (b)–(d) As in (a), but for upper-level geopotential heights (m) and winds. The green contours indicate composites of SLP or geopotential heights that are significant at the 95% confidence level, and vectors are shown only for wind anomalies that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
(a) As in Fig. 3, but for the composites of SLP (shading; hPa) and winds at 10 m (vectors; m s−1). (b)–(d) As in (a), but for upper-level geopotential heights (m) and winds. The green contours indicate composites of SLP or geopotential heights that are significant at the 95% confidence level, and vectors are shown only for wind anomalies that are significant at the 95% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
The above results indicate that water vapor transport caused by large-scale circulation anomalies may be the main cause of the interannual warming of the Arctic, while sea ice loss affects only local warming through surface feedback. Next, we provide further confirmation of the robustness of our conclusions by complementary analyses using a physically motivated approach based on the direction of the THF over the Arctic regions. First, using the same method as Blackport et al. (2019), we classify the winters into two regimes: atmosphere forces sea ice (22 winters) and sea ice forces atmosphere (18 winters). Winters when the atmosphere is primarily driven by sea ice are defined as those with a lower than average SIC and a positive THF anomaly (where positive is defined as from the ocean to the atmosphere) or a higher than average SIC and a negative THF anomaly. Conversely, winters during which the atmosphere is driving the sea ice are defined as those with a lower than average SIC and a negative THF anomaly or a higher than average SIC and a positive THF anomaly. After classifying each winter, we performed a linear regression of the atmospheric fields onto the SIC separately for these two regimes to determine the large-scale circulation and temperature patterns associated with a reduction in sea ice, both when the ice is driving the atmosphere and when the atmosphere is driving the ice.
The regression map between SIC and SAT (Fig. 10a) reveals that in association with the decrease in SIC, the winter SAT is substantially higher over the Arctic Ocean, which has a similar spatial pattern to the composite anomalies (Fig. 3a). Moreover, the regression patterns of THF, TCWV, and DLR (Figs. 10b–d) closely resemble those of the composite anomalies (Figs. 3c and 6a,b). However, these results do not mean that the relationships are causal. During winters when the sea ice is driving the atmosphere, the SAT, TCWV, and DLR anomalies (Figs. 10i–l) are mostly confined to the ice-retreat area along the southern margin of the sea ice cover. In contrast, during winters when the atmosphere is driving sea ice, the SAT, THF, TCWV, and DLR patterns (Figs. 10e–h) strongly resemble those shown in Figs. 10a–d. These results again suggest that reduced sea ice has a weak and local influence on the wintertime Arctic SAT but that anomalous large-scale circulation simultaneously cause increased water vapor, enhanced DLR, and reduced sea ice and surface warming.
Winter (a) SAT, (b) THF, (c) TCWV and (d) DLR values regressed on the standardized SIC index (created by calculating the SIC over the Arctic region and averaging over the winter season and then divided by the SD). The sign is reversed so that the maps represent the field associated with a 1-SD reduction in SIC. (e)–(h) As in (a)–(d), but for winters when the atmosphere is driving the sea ice. (i)–(l) As in (e)–(h), but for winters when the sea ice is driving the atmosphere.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Winter (a) SAT, (b) THF, (c) TCWV and (d) DLR values regressed on the standardized SIC index (created by calculating the SIC over the Arctic region and averaging over the winter season and then divided by the SD). The sign is reversed so that the maps represent the field associated with a 1-SD reduction in SIC. (e)–(h) As in (a)–(d), but for winters when the atmosphere is driving the sea ice. (i)–(l) As in (e)–(h), but for winters when the sea ice is driving the atmosphere.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Winter (a) SAT, (b) THF, (c) TCWV and (d) DLR values regressed on the standardized SIC index (created by calculating the SIC over the Arctic region and averaging over the winter season and then divided by the SD). The sign is reversed so that the maps represent the field associated with a 1-SD reduction in SIC. (e)–(h) As in (a)–(d), but for winters when the atmosphere is driving the sea ice. (i)–(l) As in (e)–(h), but for winters when the sea ice is driving the atmosphere.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
4. Results from numerical experiments
a. Physical processes associated with sea ice variations
To verify the causal relationship among the Arctic SAT, SIC, and atmospheric conditions discussed in section 3, model experiments with ECHAM 5.4 were conducted (see section 2c). Furthermore, the DJF-mean atmospheric variations discussed in section 3 result not only from internal atmospheric variability, but also from underlying ocean and land surface conditions. This differs from the AGCM CTL experiment discussed here, in which the surface conditions are fixed and thus all the atmospheric variations are due to its internal variability.
The effects of sea ice anomalies on SAT are depicted in Fig. 11a. It is consistent with the expectation that the warming forced by sea ice is mostly confined to the northern Barents and Kara Seas, where the sea ice is removed and SSTs are warmer. However, this confinement of the sea ice–induced warming is likely a result of the use of the same surface conditions in the two experiments outside the regions with significant sea ice change, as the warming induced by sea ice loss is much more widespread in coupled model simulations (e.g., Deser et al. 2016; Dai and Song 2020). During warm Arctic winters, there is a substantial sea ice reduction at the sea ice margins on the Atlantic side. As the newly exposed ocean water is warmer than the cold sea ice surface that existed previously, removing the ice cover allows the ocean to release more heat to warm the lower troposphere, as shown previously (Dai et al. 2019). Such a scenario is supported by the perturbation experiments. As shown in Fig. 11b, we find that THF has increased substantially over areas with substantial sea ice loss, which indicates a warm ocean surface and extra heating of the air. However, in our experiments, the surface heat flux differences are small outside of the ice retreat region.
Composites of differences in winter (a) SAT, (b) THF, and (c) DLR values between the sen_warm and sen_cold experiments (sen_warm minus sen_cold). Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of differences in winter (a) SAT, (b) THF, and (c) DLR values between the sen_warm and sen_cold experiments (sen_warm minus sen_cold). Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composites of differences in winter (a) SAT, (b) THF, and (c) DLR values between the sen_warm and sen_cold experiments (sen_warm minus sen_cold). Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
The ERA-Interim results reveal that the SAT interannual variations are closely associated with significant DLR anomalies, which leads to the following question: to what extent are these DLR anomalies related to sea ice? As shown in Fig. 11c, the pattern of the sea ice-forced DLR increases coincides with the regions of sea ice reductions. Since evaporation increases from the sea surface in the region of sea ice reductions, water vapor in these regions increases (Fig. 12a), which in turn leads to an increase in DLR. That is, Arctic sea ice loss allows the ocean to release more heat and moisture and adds to Arctic warming, but this warming effect of sea ice reduction is confined to the ice-retreat area. The SLP responses to sea ice forcing (Fig. 12b) also indicate that although there are some anomalous anticyclones over and around the Arctic, the pattern is statistically insignificant and substantially different from that in the composite analysis of ERA-Interim (Fig. 9a). Moreover, the vertical structures of both temperature (Fig. 12c) and water vapor (Fig. 12d) exhibit a “bottom heavy” pattern, which is similar to the vertical structure in ERA-Interim (Fig. 8), but the responses are substantially weaker in the perturbation experiments than in the ERA-Interim. This deficiency is seen in other AGCM simulations with specified surface conditions (e.g., Kumar et al. 2010), and the atmospheric response to sea ice loss penetrates deeper into the troposphere in coupled model simulations (Deser et al. 2016; Dai and Song 2020). The lack of upper-level response in our experiments indicates that without coupling to the oceans, sea ice reduction cannot cause most of the interannual variations in large-scale atmospheric circulations and water vapor.
As in Fig. 11, but for (a) TCWV, (b) SLP, and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 11, but for (a) TCWV, (b) SLP, and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 11, but for (a) TCWV, (b) SLP, and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
In summary, in our AGCM experiments without coupling to the oceans, sea ice reductions only enhance the warming and moistening of the lower troposphere and allows the relatively warm ocean to interact with the colder atmosphere via surface heat flux release, but the warming caused by this interaction is limited in extent to the ice retreat regions rather than the widespread area of interannual warming. Moreover, the simulation results suggest that without coupling to the oceans, the diminishing sea ice in winter plays an unimportant role in the widespread increases in humidity and changes in the large-scale circulation.
b. Physical processes forced by internal atmospheric variability
Because of the absence of SST and SIC variations beyond the mean seasonal cycle, year-to-year atmospheric fluctuations in the CTL simulation can only be attributed to internal atmospheric and land dynamics; therefore, the CTL can be used to analyze the impact of atmospheric (and land) internal variability. We realize that in the real world or a fully coupled model, interactions with the oceans and sea ice may amplify or dampen SAT and other variations caused by atmospheric variability, while the fixed SST and SIC have a large influence on atmospheric water vapor and other conditions and thus can dampen the variations in SAT and other atmospheric fields in the CTL simulation. These are major caveats for the results discussed below.
Composite analysis is performed based on the area-mean surface air temperature over the eastern Arctic between 65° and 90°N and between 0° and 140°E. The reason for the selection of this domain is that the interannual variability signal is mainly located in this region. Based on the threshold of +1 SD of the area-mean SAT, we obtained 15 warm winter years from the 100 CTL simulation. We calculated the difference between the 15 warm Arctic winters and the mean of all the 100 winters in the CTL, and used this composite difference to quantify the impact of atmospheric (and land) internal variability on SAT.
Figure 13 shows the differences in the winter SAT, THF and DLR between the 15 warm Arctic winters and the mean of all 100 winters in the CTL. The spatial extent of the SAT anomalies in the CTL experiment is similar to that of ERA-Interim (Fig. 3a) but with a much smaller magnitude, especially over the ice retreat region, which is understandable because of the absence of SST and SIC variability beyond the mean seasonal cycle. Interestingly, despite the absence of SST and SIC forcing in the control experiment, there are still significant negative anomalous THFs found in the Barents and Kara Seas, which is opposite to the positive anomalies observed in the perturbation experiments (Fig. 13b). This can be attributed to the reduced temperature gradient between the ocean surface and lower atmosphere. Climatologically, the air temperatures are well below freezing during Arctic winter; in the ice-covered region, the ice layer insulates the relatively warm near-surface ocean from the colder atmosphere above, whereas over the open-water region, latent and sensible fluxes are released from the ocean to the atmosphere associated with the temperature and humidity gradient. As the SST and SIC do not change in the control experiment, the THF is suppressed in association with the reduced temperature gradient during warm Arctic winters. This is the main reason for the significant negative THF anomaly in Fig. 13b.
Composite differences in winter (a) SAT, (b) THF, and (c) DLR values between the warm Arctic winters and the mean of all 100 winters (warm minus mean) in the CTL. Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composite differences in winter (a) SAT, (b) THF, and (c) DLR values between the warm Arctic winters and the mean of all 100 winters (warm minus mean) in the CTL. Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
Composite differences in winter (a) SAT, (b) THF, and (c) DLR values between the warm Arctic winters and the mean of all 100 winters (warm minus mean) in the CTL. Stippling denotes composites that are significant at the 90% confidence level.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
The THF responses in the perturbation and control experiments also suggest that the associations of reductions in sea ice cover with the SAT variations in the lower atmosphere are two-way. On the one hand, removing the ice cover induces a weaker insulation effect and a greater transfer of heat to the atmosphere, and the sea ice–induced atmospheric warming will increase DLR, thereby amplifying Arctic warming and driving further melting; this is an important positive feedback within the Arctic. On the other hand, when the lower atmosphere is significantly warmer, the exchange of heat from the ocean to the atmosphere is inhibited over the open water area due to the reduced temperature gradient. The negligible THF anomaly in the ice-retreat area in Fig. 6a illustrates that the SAT responds to a combination of the two effects mentioned above, whereas the significant negative THF anomaly in the open water area in Fig. 13b is attributed to the reduced temperature gradient. This is consistent with the finding by Kim and Kim (2017), who showed that Arctic warming will slow as the open water fraction increases.
Accordingly, the changes in the THF are not the main cause of warm Arctic winters in the CTL; to a greater extent, it is similar to a surface feedback response after lower atmospheric warming. In contrast, the increase in DLR appears to be dominant in modulating the Arctic surface temperature. As shown in Fig. 13c, increasing DLR exhibits strong spatial coherence with surface warming in the control experiments (Fig. 13a), which suggests an intimate relation between SAT and DLR. To further elucidate the cause of the DLR increase, we examine the composite anomalies in the SLP and TCWV between the 15 warm winters and the mean of all 100 winters in the control run. In comparison with the results in the perturbation experiments, the composite differences in TCWV (Fig. 14a) and larger-scale atmospheric circulation (Fig. 14b) in CTL are much more widely distributed and pronounced. There is a strong spatial coherence between TCWV (Fig. 14a) and DLR (Fig. 13c), which again suggests that DLR is strongly coupled with TCWV during winter. For the large-scale circulation (Fig. 14b), although the distribution in the North Pacific is quite different from the ERA-Interim results, the low pressure anomaly over the Arctic and the high pressure anomaly over Eurasia are consistent with the ERA-Interim results (Fig. 9a). Furthermore, the vertical structures of temperature (Fig. 14c) and specific humidity (Fig. 14d) have spatial patterns similar to those of the composite ERA-Interim anomalies (Fig. 8). Given that all ensemble members of the control run have the same SIC and SST conditions, the results suggest that warming extending from the surface to the midtroposphere, which has a spatial pattern similar to that of the composite ERA-Interim anomalies, is not mainly forced by sea ice loss but is more closely related to the atmospheric internal variability.
As in Fig. 13, but for (a) total column water vapor (TCWV), (b) sea level pressure (SLP), and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 13, but for (a) total column water vapor (TCWV), (b) sea level pressure (SLP), and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
As in Fig. 13, but for (a) total column water vapor (TCWV), (b) sea level pressure (SLP), and meridional vertical sections of (c) temperature and (d) specific humidity along 0°–140°E. Stippling in (a) and (b) denotes composite differences that are significant at the 90% confidence level, and the black contours in (c) and (d) indicate composite differences that are significant at the 95% (solid lines) and 90% (dotted lines) confidence levels.
Citation: Journal of Climate 34, 17; 10.1175/JCLI-D-20-0779.1
5. Summary and discussion
In this study, we have examined the interannual variability in Arctic winter SAT by analyzing ERA-Interim data and some AGCM experiments. During warm Arctic winters, substantial surface warming is widespread across the Arctic Ocean, accompanied by a SIC decrease over the southeastern edge of the ice layer and TCWV increases over the Atlantic sector of the Arctic Basin. We have investigated to what extent the interannual variability in the Arctic winter SAT is driven by surface processes as compared with internal atmospheric variability based on reanalysis datasets and numerical simulations. In both the modeling results and reanalysis data, the interannual warming caused by sea ice variations is confined to the ice-retreat area, and widespread warming across the Arctic Ocean is associated with increased TCWV and DLR, indicating the important role of internal atmospheric variability. Moreover, in our sea ice forcing simulations, there is no pronounced warming extending high into the upper troposphere and no large-scale circulation response similar to the reanalysis data. However, in the control run in which sea ice variations are not involved, we observe a vertical warming structure and a large-scale circulation pattern similar to those in the reanalysis data. These results strongly suggest that SIC forcing is not the dominant factor for modulating the interannual variances of the Arctic winter SAT. Instead, internal atmospheric variability contributes significantly more to the variations of winter SAT by regulating TCWV and DLR.
In this study, the interannual signal is extracted from the SAT data; however, we find that surface warming is accompanied by pronounced warming extending from the surface to the midtroposphere. Since the effect of sea ice loss on the Arctic temperature is mainly through enhanced surface heat flux, which is expected to primarily affect temperatures in the lowermost part of the atmosphere, sea ice variations cannot be the main cause of warming aloft. The internal atmospheric variability, on the other hand, regulates poleward atmospheric moisture and heat transport via large-scale atmospheric circulations. This effect can affect not only the temperature in the middle and upper troposphere but also the SAT and SIC. Some synoptic-scale studies have found that moisture increases and warming aloft generally occur before surface warming (Woods and Caballero 2016; Gong et al. 2017; He et al. 2020), indicating that Arctic warming aloft is not induced by surface warming and further supporting the main driver of internal atmospheric variability in interannual variations in wintertime Arctic SAT. It is plausible that the internal atmospheric variability induces an increase in moisture and DLR, which melts sea ice, resulting in an increase in THFs and further warming of the lower troposphere.
In addition, note that the climate model used in this study is an uncoupled model in which the SIC and SST are prescribed. Some studies have highlighted the importance of ocean–atmosphere coupling for simulating the response to sea ice (Deser et al. 2016; Smith et al. 2017; Sun et al. 2018; Screen et al. 2018; Blackport and Kushner 2018). The absence of ocean–atmosphere coupling here leads to a lack of sea ice–air interactions; therefore, the responses to sea ice forcing in the uncoupled climate model used in this study may be weaker than those in the coupled model. However, even in the coupled model (e.g., Deser et al. 2016; Smith et al. 2017; Screen et al. 2018), the response of midtroposphere temperature to sea ice loss does not reach the range of composite results shown in Fig. 8, which is in qualitative agreement with our results. Nevertheless, it is difficult to separate atmospheric processes from the surface processes in studying Arctic SAT variability, as they often influence SAT in a tightly coupled way. Results from the uncoupled model in this study should be interpreted with caution, and a detailed assessment of the full impact of SIC variations in a fully coupled model is necessary to pursue in future work.
This study confirms the result from previous studies that the wintertime Arctic SAT is dependent on moisture and DLR (e.g., Lee et al. 2017; Gong et al. 2017; Kim and Kim 2017; Mortin et al. 2016; Woods and Caballero 2016; Park et al. 2015). Here, we further point to the key role of atmospheric large-scale circulation induced by internal atmospheric variability in moisture regulation. Internal atmospheric variability is likely to be a common driver of the increased water vapor and enhanced DLR, thereby driving warming and sea ice losses, or vice versa. In areas of sea ice reduction, warming will be enhanced by local feedback. However, this enhancement will be inhibited, and a balance will be reached as the air–sea temperature gradient decreases. Our results resonate with findings in many recent Arctic–midlatitude studies (e.g., He et al. 2020; Guan et al. 2020; Blackport et al. 2019; Sun et al. 2016) and help us to gain better insight into the individual contributions of internal atmospheric variability and surface processes.
Acknowledgments
We thank the three anonymous reviewers for their constructive and helpful comments, which were very helpful in improving our paper. We thank Dr. Jian Cao and Dr. Boqi Liu for discussions of the paper. This research is supported by the National Key Research and Development Program of China (2018YFC1505804), the Natural Science Foundation of China (41905074), and the Natural Science Foundation of Jiangsu Province (BK20190782).
REFERENCES
Ashok, K., Z. Guan, and T. Yamagata, 2003: Influence of the Indian Ocean Dipole on the Australian winter rainfall. Geophys. Res. Lett., 30, 1821, https://doi.org/10.1029/2003GL017926.
Bergman, T., V. M. Kerminen, H. Korhonen, and K. J. Lehtinen, 2012: Evaluation of the sectional aerosol microphysics module SALSA implementation in ECHAM5-HAM aerosol-climate model. Geosci. Model Dev., 5, 845–868, https://doi.org/10.5194/gmd-5-845-2012.
Blackport, R., and P. J. Kushner, 2018: The role of extratropical ocean warming in the coupled climate response to Arctic sea ice loss. J. Climate, 31, 9193–9206, https://doi.org/10.1175/JCLI-D-18-0192.1.
Blackport, R., and J. A. Screen, 2019: Influence of Arctic sea ice loss in autumn compared to that in winter on the atmospheric circulation. Geophys. Res. Lett., 46, 2213–2221, https://doi.org/10.1029/2018GL081469.
Blackport, R., J. A. Screen, K. van der Wiel, and R. Bintanja, 2019: Minimal influence of reduced Arctic sea ice on coincident cold winters in mid-latitudes. Nat. Climate Change, 9, 697–704, https://doi.org/10.1038/s41558-019-0551-4.
Cohen, J., and Coauthors, 2020: Divergent consensuses on Arctic amplification influence on midlatitude severe winter weather. Nat. Climate Change, 10, 20–29, https://doi.org/10.1038/s41558-019-0662-y.
Dai, A., and M. Song, 2020: Little influence of Arctic amplification on mid-latitude climate. Nat. Climate Change, 10, 231–237, https://doi.org/10.1038/s41558-020-0694-3.
Dai, A., and J. Deng, 2021: Arctic amplification weakens the variability of daily temperatures over northern middle-high latitudes. J. Climate, 34, 2591–2609, https://doi.org/10.1175/JCLI-D-20-0514.1.
Dai, A., D. Luo, M. Song, and J. Liu, 2019: Arctic amplification is caused by sea-ice loss under increasing CO2. Nat. Commun., 10, 121, https://doi.org/10.1038/s41467-018-07954-9.
Day, J. J., J. C. Hargreaves, J. D. Annan, and A. Abe-Ouchi, 2012: Sources of multi-decadal variability in Arctic sea ice extent. Environ. Res. Lett., 7, 034011, https://doi.org/10.1088/1748-9326/7/3/034011.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, https://doi.org/10.1002/qj.828.
Deser, C., R. Tomas, M. Alexander, and D. Lawrence, 2010: The seasonal atmospheric response to projected Arctic sea ice loss in the late twenty-first century. J. Climate, 23, 333–351, https://doi.org/10.1175/2009JCLI3053.1.
Deser, C., L. Sun, R. A. Tomas, and J. Screen, 2016: Does ocean coupling matter for the northern extratropical response to projected Arctic sea ice loss? Geophys. Res. Lett., 43, 2149–2157, https://doi.org/10.1002/2016GL067792.
Dong, L., L. R. Leung, F. Song, and J. Lu, 2018: Roles of SST versus internal atmospheric variability in winter extreme precipitation variability along the U.S. West Coast. J. Climate, 31, 8039–8058, https://doi.org/10.1175/JCLI-D-18-0062.1.
Feng, J., J. Li, and Y. Li, 2010: A monsoon-like southwest Australian circulation and its relation with rainfall in southwest western Australia. J. Climate, 23, 1334–1353, https://doi.org/10.1175/2009JCLI2837.1.
Gao, Y., and Coauthors, 2014: Arctic sea ice and Eurasian climate: A review. Adv. Atmos. Sci., 32, 92–114, https://doi.org/10.1007/s00376-014-0009-6.
Gong, T., S. Feldstein, and S. Lee, 2017: The role of downward infrared radiation in the recent Arctic winter warming trend. J. Climate, 30, 4937–4949, https://doi.org/10.1175/JCLI-D-16-0180.1.
Goosse, H., and Coauthors, 2018: Quantifying climate feedbacks in polar regions. Nat. Commun., 9, 1919, https://doi.org/10.1038/s41467-018-04173-0.
Graversen, R. G., T. Mauritsen, M. Tjernstrom, E. Kallen, and G. Svensson, 2008: Vertical structure of recent Arctic warming. Nature, 451, 53–56, https://doi.org/10.1038/nature06502.
Guan, W., X. Jiang, X. Ren, G. Chen, and Q. Ding, 2020: Role of atmospheric variability in driving the “warm-Arctic, cold-continent” pattern over the North America sector and sea ice variability over the Chukchi-Bering Sea. Geophys. Res. Lett., 47, e2020GL088599, https://doi.org/10.1029/2020GL088599.
He, J., C. Deser, and B. J. Soden, 2017: Atmospheric and oceanic origins of tropical precipitation variability. J. Climate, 30, 3197–3217, https://doi.org/10.1175/JCLI-D-16-0714.1.
He, S., Xu, X., Furevik, T., and Gao, Y., 2020: Eurasian cooling linked to the vertical distribution of Arctic warming. Geophys. Res. Lett., 47, e2020GL087212, https://doi.org/10.1029/2020GL087212.
Holland, M. M., and J. Stroeve, 2011: Changing seasonal sea ice predictor relationships in a changing Arctic climate. Geophys. Res. Lett., 38, L18501, https://doi.org/10.1029/2011GL049303.
Jungclaus, J. H., and T. Koenigk, 2009: Low-frequency variability of the Arctic climate: The role of oceanic and atmospheric heat transport variations. Climate Dyn., 34, 265–279, https://doi.org/10.1007/s00382-009-0569-9.
Katz, R., and B. G. Brown, 1992: Extreme events in a changing climate: Variability is more important than averages. Climatic Change, 21, 289–302, https://doi.org/10.1007/BF00139728.
Kay, J. E., M. M. Holland, and A. Jahn, 2011: Inter-annual to multi-decadal Arctic sea ice extent trends in a warming world. Geophys. Res. Lett., 38, L15708, https://doi.org/10.1029/2011GL048008.
Kim, H.-M., and B.-M. Kim, 2017: Relative contributions of atmospheric energy transport and sea ice loss to the recent warm Arctic winter. J. Climate, 30, 7441–7450, https://doi.org/10.1175/JCLI-D-17-0157.1.
Kinnard, C., C. M. Zdanowicz, D. A. Fisher, E. Isaksson, A. de Vernal, and L. G. Thompson, 2011: Reconstructed changes in Arctic sea ice over the past 1,450 years. Nature, 479, 509–512, https://doi.org/10.1038/nature10581.
Koenigk, T., and Coauthors, 2018: Impact of Arctic sea ice variations on winter temperature anomalies in northern hemispheric land areas. Climate Dyn., 52, 3111–3137, https://doi.org/10.1007/s00382-018-4305-1.
Kumar, A., and Coauthors, 2010: Contribution of sea ice loss to Arctic amplification. Geophys. Res. Lett., 37, L21701, https://doi.org/10.1029/2010GL045022.
Lang, A., S. Yang, and E. Kaas, 2017: Sea ice thickness and recent Arctic warming. Geophys. Res. Lett., 44, 409–418, https://doi.org/10.1002/2016GL071274.
Lee, S., T. Gong, S. B. Feldstein, J. A. Screen, and I. Simmonds, 2017: Revisiting the cause of the 1989–2009 Arctic surface warming using the surface energy budget: Downward infrared radiation dominates the surface fluxes. Geophys. Res. Lett., 44, 10 654–10 661, https://doi.org/10.1002/2017GL075375.
Levitus, S., G. Matishov, D. Seidov, and I. Smolyar, 2009: Barents Sea multidecadal variability. Geophys. Res. Lett., 36, L19604, https://doi.org/10.1029/2009GL039847.
Li, F., X. Wan, H. Wang, Y. J. Orsolini, Z. Cong, Y. Gao, and S. Kang, 2020: Arctic sea-ice loss intensifies aerosol transport to the Tibetan Plateau. Nat. Climate Change, 10, 1037–1044, https://doi.org/10.1038/s41558-020-0881-2.
Li, X.-F., J. Yu, and Y. Li, 2013: Recent summer rainfall increase and surface cooling over northern Australia since the late 1970s: A response to warming in the tropical western Pacific. J. Climate, 26, 7221–7239, https://doi.org/10.1175/JCLI-D-12-00786.1.
Liang, Y.-C., and Coauthors, 2020: Quantification of the Arctic sea ice–driven atmospheric circulation variability in coordinated large ensemble simulations. Geophys. Res. Lett., 47, e2019GL085397, https://doi.org/10.1029/2019GL085397.
Messori, G., C. Woods, and R. Caballero, 2018: On the drivers of wintertime temperature extremes in the high Arctic. J. Climate, 31, 1597–1618, https://doi.org/10.1175/JCLI-D-17-0386.1.
Mewes, D., and C. Jacobi, 2019: Heat transport pathways into the Arctic and their connections to surface air temperatures. Atmos. Chem. Phys., 19, 3927–3937, https://doi.org/10.5194/acp-19-3927-2019.
Miles, M. W., D. V. Divine, T. Furevik, E. Jansen, M. Moros, and A. E. J. Ogilvie, 2014: A signal of persistent Atlantic multidecadal variability in Arctic sea ice. Geophys. Res. Lett., 41, 463–469, https://doi.org/10.1002/2013GL058084.
Mioduszewski, J. R., S. Vavrus, M. Wang, M. Holland, and L. Landrum, 2019: Past and future interannual variability in Arctic sea ice in coupled climate models. Cryosphere, 13, 113–124, https://doi.org/10.5194/tc-13-113-2019.
Morrison, H., G. de Boer, G. Feingold, J. Harrington, M. D. Shupe, and K. Sulia, 2012: Resilience of persistent Arctic mixed-phase clouds. Nat. Geosci., 5, 11–17, https://doi.org/10.1038/ngeo1332.
Mortin, J., G. Svensson, R. G. Graversen, M.-L. Kapsch, J. C. Stroeve, and L. N. Boisvert, 2016: Melt onset over Arctic sea ice controlled by atmospheric moisture transport. Geophys. Res. Lett., 43, 6636–6642, http://doi.org/10.1002/2016GL069330.
Nam, C. C. W., and J. Quaas, 2012: Evaluation of clouds and precipitation in the ECHAM5 general circulation model using CALIPSO and CloudSat satellite data. J. Climate, 25, 4975–4992, https://doi.org/10.1175/JCLI-D-11-00347.1.
Nicholls, N., 2009: Local and remote causes of the southern Australian autumn–winter rainfall decline, 1958–2007. Climate Dyn., 34, 835–845, https://doi.org/10.1007/s00382-009-0527-6.
Nygård, T., R. G. Graversen, P. Uotila, T. Naakka, and T. Vihma, 2019: Strong dependence of wintertime Arctic moisture and cloud distributions on atmospheric large-scale circulation. J. Climate, 32, 8771–8790, https://doi.org/10.1175/JCLI-D-19-0242.1.
Ogawa, F., and Coauthors, 2018: Evaluating impacts of recent Arctic sea ice loss on the Northern Hemisphere winter climate change. Geophys. Res. Lett., 45, 3255–3263, https://doi.org/10.1002/2017GL076502.
Osborne, J. M., J. A. Screen, and M. Collins, 2017: Ocean–atmosphere state dependence of the atmospheric response to Arctic sea ice loss. J. Climate, 30, 1537–1552, https://doi.org/10.1175/JCLI-D-16-0531.1.
Park, D.-S. R., S. Lee, and S. B. Feldstein, 2015: Attribution of the recent winter sea ice decline over the Atlantic sector of the Arctic Ocean. J. Climate, 28, 4027–4033, https://doi.org/10.1175/JCLI-D-15-0042.1.
Peings, Y., and G. Magnusdottir, 2014: Response of the wintertime Northern Hemisphere atmospheric circulation to current and projected Arctic sea ice decline: A numerical study with CAM5. J. Climate, 27, 244–264, https://doi.org/10.1175/JCLI-D-13-00272.1.
Pithan, F., and T. Mauritsen, 2014: Arctic amplification dominated by temperature feedbacks in contemporary climate models. Nat. Geosci., 7, 181–184, https://doi.org/10.1038/ngeo2071.
Pithan, F., and Coauthors, 2018: Role of air-mass transformations in exchange between the Arctic and mid-latitudes. Nat. Geosci., 11, 805–812, https://doi.org/10.1038/s41561-018-0234-1.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407–4435, https://doi.org/10.1029/2002JD002670.
Rinke, A., and Coauthors, 2019: Trends of vertically integrated water vapor over the Arctic during 1979–2016: Consistent moistening all over? J. Climate, 32, 6097–6116, https://doi.org/10.1175/JCLI-D-19-0092.1.
Roeckner, E., and Coauthors, 2003: The atmospheric general circulation model ECHAM5. Part I: Model description. Max-Planck-Institute for Meteorology Rep. 349, 140 pp., https://pure.mpg.de/pubman/faces/ViewItemOverviewPage.jsp?itemId=item_995269.
Russotto, R. D., and M. Biasutti, 2020: Polar amplification as an inherent response of a circulating atmosphere: Results from the TRACMIP aquaplanets. Geophys. Res. Lett., 47, e2019GL086771, https://doi.org/10.1029/2019GL086771.
Screen, J. A., 2014: Arctic amplification decreases temperature variance in northern mid- to high-latitudes. Nat. Climate Change, 4, 577–582, https://doi.org/10.1038/nclimate2268.
Screen, J. A., and I. Simmonds, 2010a: The central role of diminishing sea ice in recent Arctic temperature amplification. Nature, 464, 1334–1337, https://doi.org/10.1038/nature09051.
Screen, J. A., and I. Simmonds, 2010b: Increasing fall–winter energy loss from the Arctic Ocean and its role in Arctic temperature amplification. Geophys. Res. Lett., 37, L16707, https://doi.org/10.1029/2010GL044136.
Screen, J. A., and J. A. Francis, 2016: Contribution of sea-ice loss to Arctic amplification is regulated by Pacific Ocean decadal variability. Nat. Climate Change, 6, 856–860, https://doi.org/10.1038/nclimate3011.
Screen, J. A., I. Simmonds, C. Deser, and R. Tomas, 2013: The atmospheric response to three decades of observed Arctic sea ice loss. J. Climate, 26, 1230–1248, https://doi.org/10.1175/JCLI-D-12-00063.1.
Screen, J. A., and Coauthors, 2018: Consistency and discrepancy in the atmospheric response to Arctic sea-ice loss across climate models. Nat. Geosci., 11, 155–163, https://doi.org/10.1038/s41561-018-0059-y.
Serreze, M. C., and J. A. Francis, 2006: The Arctic amplification debate. Climatic Change, 76, 241–264, https://doi.org/10.1007/s10584-005-9017-y.
Serreze, M. C., and R. G. Barry, 2011: Processes and impacts of Arctic amplification: A research synthesis. Global Planet. Change, 77, 85–96, https://doi.org/10.1016/j.gloplacha.2011.03.004.
Shupe, M. D., and J. M. Intrieri, 2004: Cloud radiative forcing of the Arctic surface: The influence of cloud properties, surface albedo, and solar zenith angle. J. Climate, 17, 616–628, https://doi.org/10.1175/1520-0442(2004)017<0616:CRFOTA>2.0.CO;2.
Smith, D. M., N. J. Dunstone, A. A. Scaife, E. K. Fiedler, D. Copsey, and S. C. Hardiman, 2017: Atmospheric response to Arctic and Antarctic sea ice: The importance of ocean–atmosphere coupling and the background state. J. Climate, 30, 4547–4565, https://doi.org/10.1175/JCLI-D-16-0564.1.
Stramler, K., A. D. Del Genio, and W. B. Rossow, 2011: Synoptically driven Arctic winter states. J. Climate, 24, 1747–1762, https://doi.org/10.1175/2010JCLI3817.1.
Stroeve, J., M. Holland, W. Meier, T. Scambos, and M. Serreze, 2007: Arctic sea ice decline: Faster than forecast. Geophys. Res. Lett., 34, L09501, https://doi.org/10.1029/2007GL029703.
Sun, L., J. Perlwitz, and M. Hoerling, 2016: What caused the recent “warm Arctic, cold continents” trend pattern in winter temperatures? Geophys. Res. Lett., 43, 5345–5352, https://doi.org/10.1002/2016GL069024.
Sun, L., M. Alexander, and C. Deser, 2018: Evolution of the global coupled climate response to Arctic sea ice loss during 1990–2090 and its contribution to climate change. J. Climate, 31, 7823–7843, https://doi.org/10.1175/JCLI-D-18-0134.1.
Van der Linden, E. C., R. Bintanja, W. Hazeleger, and R. G. Graversen, 2016: Low-frequency variability of surface air temperature over the Barents Sea: Causes and mechanisms. Climate Dyn., 47, 1247–1262, https://doi.org/10.1007/s00382-015-2899-0.
Woods, C., and R. Caballero, 2016: The role of moist intrusions in winter Arctic warming and sea ice decline. J. Climate, 29, 4473–4485, https://doi.org/10.1175/JCLI-D-15-0773.1.
Woods, C., R. Caballero, and G. Svensson, 2013: Large-scale circulation associated with moisture intrusions into the Arctic during winter. Geophys. Res. Lett., 40, 4717–4721, https://doi.org/10.1002/grl.50912.
Yim, B. Y., H. S. Min, B.-M. Kim, J.-H. Jeong, and J.-S. Kug, 2016: Sensitivity of Arctic warming to sea ice concentration. J. Geophys. Res. Atmos., 121, 6927–6942, https://doi.org/10.1002/2015JD023953.
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, https://doi.org/10.1029/2008GL035607.
Zhang, X., A. Sorteberg, J. He, J. Zhang, I. Polyakov, R. Gerdes, J. Inoue, and P. Wu, 2012: Enhanced poleward moisture transport and amplified northern high-latitude wetting trend. Nat. Climate Change, 3, 47–51, https://doi.org/10.1038/nclimate1631.