Different responses of surface air temperature over Eurasia in early and late winter to the autumn Kara–Laptev Sea ice

The Arctic climate is changing rapidly, along with intensified melting of sea ice, which has significant impacts on surface air temperature (SAT) in Eurasia. This study reveals that the subseasonal response of SAT to the autumn Kara–Laptev Sea ice (KLSIC) differs significantly between early and late winter. The response of SAT to KLSIC forms a warm Arctic–cold Eurasia pattern in early winter. Conversely, the negative anomaly response of SAT to KLSIC in late winter is only distributed in the band range of Eurasia, without significant positive SAT anomaly over the Arctic Ocean. After further examination of the separate physical mechanisms involved in early and late winter, it is found that a decrease in KLSIC in autumn can lead to a strengthened Ural high and Siberian high in the Arctic–Eurasia region, which is conducive to cold events in the mid-latitudes of Eurasia in early winter. For late winter, a westward shift in the response of atmospheric circulation to KLSIC leads to a negative anomaly feedback of North Sea surface temperature, which triggers the propagation of Rossby waves to the Sea of Japan through the wave activity flux. Meanwhile, the deep trough of East Asia is strengthened and extends to the southeast, guiding northern cold air to the western Pacific. Our results highlight that different subseasonal effects of sea ice should be considered in Eurasian climate prediction, rather than only consider the effects in winter mean.


Introduction
In recent years, due to the Arctic amplification effect and the ongoing reduction in Arctic sea ice, there has been an increased emphasis on studying the relationship between the Arctic and mid-latitudes, along with the potential mechanisms, in climate variability research (Fan et al 2018, Cohen et al 2020).Changes in Arctic sea ice and its interplay with the atmosphere not only affect local thermal and dynamic conditions, but also have intricate impacts on the midand even low-latitude through complex interactions and feedback processes (Cohen et al 2021, Zhang et al 2022, Fan et al 2023).The Kara-Laptev Sea serves as a critical region for ice-atmosphere interactions (Chen et al 2021, Wang et al 2021).The preceding autumn reduction of sea ice in the Kara-Laptev Sea (KLSIC) is closely linked to winter Ural blocking events and strengthening of the Siberian high, subsequently leading to significant cold anomalies over a broad Eurasian region (Li and Wang 2012, Zhang et al 2019, Pang et al 2023), which receives a great deal of attention.
Several previous studies have investigated the effect of Arctic sea ice on the mid-latitude climate (Wang and Liu 2016).The loss of autumn Arctic sea ice is conducive to greater variation in the jet stream, stratosphere-troposphere coupling, and significant cooling over Eurasia during winter (Wu et al 2011, Mori et al 2014, Cohen et al 2021).For example, © 2024 The Author(s).Published by IOP Publishing Ltd Arctic warming over the Atlantic Ocean sector in winter can influence Eurasia by producing circulation anomalies similar to the Arctic Oscillation and North Atlantic Oscillation (NAO; Alexander et al 2004).An anomalous stationary wave train can be stimulated by the warming around the Barents-Kara Sea, which can be due to positive anomalies of the Siberian high, and ultimately contributes to the cold condition in Eurasia (Nakamura et al 2015).The pattern of the jet stream and related weather extremes can also be caused by Arctic warming (Francis and Vavrus 2015).The Ural blocking plays an important role in the connection between a warm Arctic and cold Eurasia (Luo et al 2016, Zhang et al 2021).There has also been some research carried out indicating that the loss of Arctic sea ice can cause the stratospheric polar vortex to weaken or move (Kim et al 2014, Hoshi et al 2017, Kim and Kim 2020) and result in downward anomalies from the stratosphere to the troposphere (Screen 2017, Zhang et al 2018a, Chripko et al 2021).In this respect, there might be two potential mechanisms involved: the horizontal or vertical propagation of quasi-stationary planetary waves (Zhang et al 2018a(Zhang et al , 2018b)), and diminished high-latitude westerly winds (Francis and Vavrus 2012) as a result of the reduced meridional temperature differential.
Previous research has tended to focus on investigating the relationship between the autumn sea ice and Eurasian surface air temperature (SAT) in winter (Zuo et al 2016, Wegmann et al 2018).However, an increasing number of studies have found that the subseasonal variations in Eurasian winter SAT exhibit more intricate changes (Dai et al 2019, Yin et al 2023), presenting significant challenges for climate forecasting (Xu et al 2023(Xu et al , 2024)).For instance, the warm Arctic-cold Eurasia (WACE) pattern in Eurasia can maintain a strong intensity on the subseasonal timescale, which leads to extreme climate events.The record-breaking cold waves in eastern China in the winter of 2020/2021 (Zhang et al 2021) and a super sandstorm in North China in the spring of 2021 (Yin et al 2022) were both heavily influenced by the subseasonal change in the WACE pattern.Additionally, in terms of the subseasonal variability situated between the scales of weather forecasting and long-range climate prediction, improving the subseasonal predictability remains a formidable challenge (Li et al 2021), particularly within the mid-to-high latitudes (Wang et al 2022).
In the above context, the present study focuses on the subseasonal connections between sea ice and Eurasian SAT and finds that the response of Eurasian SAT to KLSIC exhibits significantly different characteristics in early and late winter, which may provide a possible preceding signal for predicting Eurasian SAT at the subseasonal scale.

Data
Daily reanalysis data over the period 1979-2021 for autumn and winter (September-February), including SAT, temperature at 500 hPa (T500), sea level pressure (SLP), sea surface temperature (SST), surface sensible and latent heat fluxes, zonal and meridional winds, and geopotential height (Hgt), at various vertical levels, and only for autumn (September-November), including sea ice concentration (SIC), were provided by the fifth major global reanalysis produced by ECMWF (ERA5; Hersbach et al 2020).The spatial resolution of ERA5 is 1 • × 1 • .The monthly gridded SIC data in autumn with a resolution of 1 • × 1 • are also obtained from the Hadley Centre Sea Ice and SST data set of the UK Met Office (Rayner et al 2003).
The KLSIC index was defined as the regionally averaged SIC over the Kara-Laptev Sea (75 • -85 • N, 35 • -150 • E) in autumn, multiplied by −1, representing the reduction in SIC.The LLSAT index was defined as the regionally averaged SAT over midand low-latitude Eurasia (20 , and the MLSAT index was defined as the regionally averaged SAT over mid-latitude Eurasia (47 The NorSST index was defined as the regionally averaged SST over the North Sea (52 • -62 • N, 4 • W-10 • E) in late winter, multiplied by −1.The turbulent heat flux (THF) is defined as the sum of surface sensible and latent heat fluxes.In this paper, the early winter of a particular year refers to the period from 1 December of that year to 13 January of the following year, while late winter is the period from 14 January to 28 February of the following year (Yin et al 2023).All the data and indices in this study had their daily linear trends removed prior to use.Plumb (1985) defined the horizontal wave activity flux (WAF) to indicate the propagation of planetary wave packets in horizontal space:

Method
where p represents pressure, u and v are the zonal and meridional winds, Φ is the Hgt, Ω is the rate of Earth's rotation, α is Earth's radius, and ϕ and λ are the latitude and longitude, respectively.Also, u and v in equation ( 1) are geostrophic winds: . (2)

KLSIC has different effects on the early and late winter SAT in Eurasia
We begin by examining the response of winter Eurasian SAT to KLSIC, as depicted in figure S1.
When KLSIC decreases in autumn, SAT shows significant positive anomalies over the east Siberian Sea, Laptev Sea, Kara Sea, and the north part of the Barents Sea in winter.Meanwhile, there are significant negative SAT anomalies over most of mid-and low-latitude Eurasia, which extend east to the western Pacific.The significant positive SAT anomalies over the Arctic and the negative SAT anomalies over the Eurasia collectively constitute the WACE pattern (Kug et al 2015, Jin et al 2020).Consistent with existing research, the reduced sea ice in autumn can lead to more frequent formation of blocking high pressure in the Arctic-Eurasia region, which is conducive to extreme cold events in mid-latitude Eurasia (Mori et al 2014, Zhang et al 2022).However, on a subseasonal timescale, the response of SAT to KLSIC differs significantly between early and late winter.As shown in figure 1(a), the SAT of early winter shows significant positive anomalies over the Arctic Ocean from the Barents Sea to East Siberian Sea, and negative anomalies over mid-and low-latitude Eurasia.It exhibits a WACE pattern distribution, which is similar to the responses of winter mean SAT (figure S1).Besides, there is a positive temperature anomalies center over the Barents Sea and the north part of Ural Mountain at 500 hPa.The deep-layer warmth in the Arctic has a greater tendency to cause below-average temperatures and more extreme cold days to occur in early winter over central Eurasia (figure S2; He et al 2020).Nevertheless, compared with the responses of winter mean SAT, the positive SAT anomalies in early winter over the Arctic Ocean are stronger and wider, and there are no significant negative anomalies in the mid-latitudes of Europe.In addition, the negative anomalies over Eurasia south of 55 • N are stronger than that of the winter mean.
Interestingly, for late winter, the significant negative anomaly responses of SAT are located in the East European Plain, the north side of Lake Baikal, and the west Pacific Ocean (figure 1(b)), and the impact is stronger and more widespread over the plains of eastern Europe.Besides, the positive anomalies over the Arctic Ocean cannot extend into late winter.Therefore, the response of the seasonal WACE pattern to KLSIC in winter may only originate from the response of SAT to KLSIC in early winter, and have little connection with late winter.The time series of LLSAT, MLSAT and KLSIC show that there is a significant negative correlation between KLSIC and both LLSAT and MLSAT (figure S3).After further verification, we find that the correlation coefficient between KLSIC and LLSAT is −0.51 (exceeding the 99% confidence level) in early winter and −0.24 in late winter.On the contrary, the correlation coefficient between KLSIC and MLSAT is −0.25 in early winter and −0.46 (exceeding the 99% confidence level) in late winter (figures 1(c)).This demonstrates that, statistically, the KLSIC anomalies in autumn result in significant SAT negative anomalies over mid-and low-latitude Eurasia in early winter and over mid-latitude Eurasia in late winter, which suggests the response to KLSIC in early and late winter are significantly different.
In order to further verify the different effects of sea ice on SAT in early and late winter, the SAT was composited for the KLSIC high-value years (figure S4).Years in which the standardized KLSIC are greater than one standard deviation are defined as highvalue years.In early winter of high-value years, the SAT presents a significant dipole pattern between the Arctic (positive) and Eurasia (negative).The distribution of SAT anomalies and significant areas is the same as in figure 1

Associated physical mechanisms in early winter
Corresponding to the different responses of SAT to KLSIC in early and late winter, there are also obvious differences in the atmospheric circulation in early and late winter under the influence of KLSIC.Influenced by KLSIC, the THF of the Kara Sea, Laptev Sea, and the west part of the Barents Sea in autumn shows significant negative anomalies (figure 2(a)), indicating that KLSIC influences the atmosphere through upward THF.The effect of KLSIC on the atmosphere can extend up to 200 hPa, and the strongest anomaly center is located in the area between 75 • E and 125 As shown in figure 3, the influence of KLSIC on the atmosphere can last to the early winter.By the effect of KLSIC, the zonal wind at 200 hPa over mid-and high-latitude Eurasia (between 45 • N and 75 • N, 0 • E and 120 • E) presents significant negative anomalies.Compared with the climatological state, the negative anomalies further weaken the zonal wind in the northern regions of Caspian Sea (figure 3(a)).There are also significant positive zonal  wind anomalies over mid-and low-latitude Eurasia (between 25 • N and 55 • N, 70 • E and 120 • E), which exhibit a negative/positive pattern from north to south over Eurasia.At 500 hPa, the significant positive anomaly center moves westward over the Barents Sea and the Kara Sea and extends southward over the Ural Mountains to the north of the Caspian Sea, strengthening the Ural high (figure 3(b)).The influence of KLSIC on the upper atmosphere propagates southeastward through the WAF to mid-and lowlatitude Eurasia, where it forms a significant negative Hgt anomaly center.Two significant anomaly centers form a north-positive-south-negative pattern distribution on the Eurasian continent, corresponding well to the negative/positive pattern of zonal wind at 200 hPa.The response of the atmosphere to KLSIC also becomes stronger and propagates down to near the ground (figures 3(c) and (d)).The Siberian high is significantly strengthened by KLSIC and controls the Eurasian continent, which leads to an increase in the pressure gradient.In response to the positive anomalies of the Siberian high, there is a significant northerly wind in the west and north of the Qinghai-Tibet Plateau, which is conducive to the southbound cold air from the north.This leads to the lower SAT in the west and north of the Qinghai-Tibet Plateau (figure 1(c)).
In addition, East Asia is situated in front of the enhanced Siberian high (figure 3(d)).The structure of Hgt is beneficial to the spread of cold air southward and the outbreak of a cold wave, which leads to the significant negative SAT anomalies over East Asia (Wu et al 2011, Cheung et al 2012).Therefore, the influence of KLSIC on the Hgt strengthens in early winter, which directs cold air southward and leads to the outbreak of a cold wave.As a result, negative SAT anomalies appear over mid-and low-latitude Eurasia.

Associated physical mechanisms in late winter
Different from early winter, the influence of KLSIC on the atmosphere has a clear westward shift in late winter.Also, it mainly affects the Eastern European Plain, and there are no significant abnormalities in Eurasia (figure 4).The negative anomalies of zonal wind move west over the North Atlantic, between 50 • N and 65 • N. In comparison to the climatological state, the zonal wind, originally reaching 20 m s −1 , is significantly weakened in late winter (figure 4(a)).The positive anomalies of Hgt at 500 hPa also move west over the Greenland Sea and the Norwegian Sea (between 60 • N and 85 • N; figure 4(b)).Together with the negative Hgt anomalies at 40 • N and 20 • W, they form a north-positive-south-negative pattern.
The Hgt anomalies can also propagate downward to near the ground (figures 4(c) and (d)).In the SLP field, the positive anomaly area lies over the Greenland Sea and northwestern Europe, forming an anticyclonic circulation.Meanwhile, the negative anomaly area is located over the Iberian Peninsula and creates a cyclonic circulation (figure 4(d)).The SLP anomalies exhibit a pattern similar to the negative phase of the NAO (Trigo et al 2002), and the positive anomalies are stronger and wider than the negative anomalies.Over the Eastern European Plain, which is located in the southeastern part of the anticyclonic circulation, there are significant northerly wind anomalies extending southward to 55 • N, which is conducive to the southward movement of cold air from the north (figure 4(d)) and results in significant negative SAT anomalies over western Europe in late winter (figure 1(b)).
Since the atmospheric circulation over the north side of Lake Baikal and the western Pacific Ocean does not respond significantly to KLSIC in late winter (figure 4), we further investigate the physical mechanism by which KLSIC affects the mid-latitude SAT anomalies in Eurasia.Because KLSIC cannot directly affect the SAT through atmospheric circulation, we check for an intermediate factor affecting the SAT over mid-and low-latitude Eurasia.As can be seen in figure 4(d), the response of the late winter atmospheric circulation to KLSIC shows significant easterly wind anomalies in the SLP field between 55 • N and 65 • N.Under its influence, the North Atlantic warm current, which can reach the North Sea, is weakened, resulting in significant negative anomalies of NorSST (Oziel et al 2020).Correspondingly, KLSIC results in significant negative NorSST anomalies (figure 5(a)).The distribution of the correlation coefficients between NorSST and late-winter  SAT is very similar to that of KLSIC with late-winter SAT.The high-value correlation area spans midlatitude Eurasia and extends to the western Pacific (figures 1(d) and 5(c)).This indicates that KLSIC affects the late-winter SAT on the north side of Lake Baikal and in the western Pacific Ocean through NorSST.
Under the influence of NorSST, there are significant negative THF anomalies over the North Sea, indicating that NorSST affects the atmospheric circu- The Eastern European plain is located between them, which is conducive to the transport of cold air from north to south.This further leads to a strengthening of negative SAT anomalies over the Eastern European Plain.
The WAF at 500 hPa shows that the Hgt anomalies over the North Atlantic can be transmitted to the Sea of Japan through two great circle paths.There are negative anomalies over the western Mediterranean, the Arabian Sea, and the Sea of Japan to the western Pacific Ocean, and positive anomalies over the Persian Gulf and eastern China.However, the statistical significance of the positive anomalies over eastern China does not exceed the 95% confidence level.At 500 hPa, there are negative Hgt anomalies extending from the Lake Baikal area through the Sea of Japan to the western Pacific, strengthening the deep trough of East Asia (figure 6(b)).With decreasing height, the intensity and scope of the negative anomalies gradually decrease and move toward the western Pacific.In the SLP field, the significant negative area basically moves east over the western Pacific (figure 6(d)).At 850 hPa, there is a significant negative center over the Sea of Japan, forming a significant cyclonic circulation (figure 6(c)).The winds are distributed from the Lake Baikal region through the Bohai Sea region to the western Pacific region, causing cold air from the north to be transported from the Lake Baikal region to the western Pacific, bypassing eastern China (figure 6(d)).This leads to significant negative SAT anomalies in the Lake Baikal and western Pacific regions in late winter, without significant SAT anomalies in eastern China (figure 1(b)).Therefore, NorSST, as an intermediate factor, transmits the Hgt anomalies over the North Atlantic to East Asia through the great circle path, causing negative SAT anomalies to appear in the Lake Baikal and western Pacific regions, but no significant SAT anomalies appear in eastern China.

Conclusions and discussion
The subseasonal connections between KLSIC and Eurasian SAT were investigated in this study and it was found that the response of SAT to KLSIC differs significantly between early and late winter in Eurasia.The response of SAT to KLSIC shows significant positive anomalies over the Arctic Ocean and negative anomalies over mid-and low-latitude Eurasia in early winter, which forms the WACE pattern and is similar to the response of the winter mean SAT (Sung et al 2018).Conversely, the negative anomaly response of SAT to KLSIC in late winter is mainly distributed in the band range of Eurasia and there is no significant SAT response in the Arctic region.In addition, the WACE signal in the winter mean may only originate from the response of SAT to KLSIC in early winter, rather than that in late winter.This may be helpful for studying WACE from a subseasonal perspective.
The associated physical mechanisms were also investigated in this study, the findings of which are summarized schematically in figure 7. The decreased KLSIC results in locally significant negative THF anomalies and positive Hgt anomalies extending up to 200 hPa.The influence of KLSIC on the atmosphere can last to early winter.Accordingly, there are significant positive Hgt anomalies over the Kara-Laptev Sea, which are transmitted to low-latitude Eurasia by the WAF, indicating positive Ural high and Siberian high anomalies.The Hgt gradient is strengthened, which is conducive to the southward propagation of Arctic cold air (Cheung et al 2012).Finally, there are significant negative SAT anomalies over mid-and low-latitude Eurasia.The response of SAT in winter mean to KLSIC is consistent with that in early winter, forming the WACE pattern.The response of atmospheric circulation is also similar to that in early winter (figure S5).However, the response in late winter is significantly different from that in winter mean and early winter.
The positive Hgt anomalies over the Arctic Ocean caused by KLSIC move westward and propagate downward to near the surface, which forms an anticyclone over the Norwegian Sea.The northerly airflow on its southeast side is conducive to the movement of cold air from the Arctic to the plains of eastern Europe.Meanwhile, the strong near-surface easterly winds on its south side result in a weakening of the North Atlantic warm current and a decrease in NorSST.The anomalies of NorSST can affect the atmosphere vertically and form a northpositive-south-negative pattern at 500 hPa, which can spread to East Asia via the WAF.This causes the eastward movement and southward extension of the deep trough of East Asia, which can guide cold air from the north to spread from the east side of Lake Baikal to the western Pacific Ocean and cause negative SAT anomalies.The results obtained from using the Hadley Centre are identical (figures S6-S9).
It is important to note that this study mainly used statistical methods to research the different responses of SAT in Eurasia to the autumn Kara-Laptev Sea ice in early and late winter.Thus, further model simulation experiments are needed to verify our findings (Cheung andZhou 2015, Song andZhou 2023).At the same time, subseasonal temperature forecasts for Eurasia pose significant challenges (Lin 2018).Therefore, whether the early sea-ice factors we selected can be used to enhance the accuracy of subseasonal forecasting will be the focus of our future investigations.
(a), and the MLSAT reaches −1.56 • C. In late winter, there are the same significant areas as in figure 1(b), except for the west Siberian plain.The LLSAT can reach −1.26 • C. The correlation and composite analysis results both indicate that KLSIC influences different areas in early and late winter, the underlying physical mechanisms of which we further explore next.

Figure 1 .
Figure 1.(a), (b) The correlation coefficients between the KLSIC and SAT of (a) early winter and (b) late winter.The green boxes represent (a) the area of LLSAT and (b) the area of MLSAT.(c) The correlation coefficients between KLSIC and both LLSAT and MLSAT in early winter (blue) and late winter (orange).The white dots in (a), (b) and the grey dotted line in (c) indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 2 .
Figure 2. The correlation coefficients between KLSIC and the (a) THF and (b) Hgt from 100 hPa to 1000 hPa between 80 • N and 60 • N in autumn.The linear trend has been removed.The green boxes represent the area of the North Sea.The area between the grey imaginary lines in (b) represents the same longitudinal range of KLSIC.The white dots in (a) and (b) indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 3 .
Figure 3.The correlation coefficients between KLSIC and (a) u-wind at 200 hPa, (b) Hgt (shading) and WAF (arrows) at 500 hPa, (c) Hgt (shading) and winds (arrows) at 850 hPa, and (d) SLP (shading) and winds at 10 m (arrows) in early winter.The green contour lines in (a) represent the climatological state of the u-wind at 200 hPa in early winter from 1991 to 2020.The linear trends have been removed.The white dots indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 4 .
Figure 4.The correlation coefficients between KLSIC and (a) u-wind at 200 hPa, (b) Hgt (shading) and WAF (arrows) at 500 hPa, (c) Hgt (shading) and winds (arrows) at 850 hPa, and (d) SLP (shading) and winds at 10 m (arrows) in late winter.The green contour lines in (a) represent the climatological state of the u-wind at 200 hPa in late winter from 1991 to 2020.The linear trends have been removed.The white dots indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 5 .
Figure 5. (a) The correlation coefficients between KLSIC and late-winter SST.(b), (c) The correlation coefficients between NorSST and late-winter (b) THF and (c) SAT.The linear trends have been removed.The green boxes represent the area of the North Sea.The white dots indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 6 .
Figure 6.The correlation coefficients between NorSST and (a) u-wind at 200 hPa, (b) Hgt (shading) and WAF (arrows) 500 hPa, (c) Hgt (shading) and winds (arrows) at 850 hPa, and (d) SLP (shading) and winds at 10 m (arrows) in late winter.The green contour lines in (a) represent the climatological state of the u-wind at 200 hPa in late winter from 1991 to 2020.The linear trends have been removed.The white dots indicate that the correlation coefficients are significant above the 95% confidence level.

Figure 7 .
Figure 7. Schematic illustrating the different effects of KLSIC on the SAT of early and late winter.The correlation coefficients in early winter between KLSIC and SAT are shown by the shading, and those between KLSIC and Hgt at 500 hPa are denoted by the contours.Likewise, the correlation coefficients in late winter between KLSIC and SAT are shown by the shading, and those between NorSST and Hgt at 500 hPa are denoted by the contours.The dark-green arrow represents the direction of influence of KLSIC on NorSST.The dark-yellow arrows represent the direction of the correlation coefficients between the 500 hPa WAF in early (late) winter and KLSIC (NorSST).The green boxes represent the area of North Sea.