Potential impacts of reduced winter Kara Sea ice on the dipole pattern of cold surge frequency over the tropical western Pacific

The impact of Arctic Sea ice melting on weather and climate extremes in the Northern Hemisphere has garnered widespread attention. Existing research has convincingly demonstrated the importance of this impact in mid-high latitudes, while its influence in areas beyond remains controversial. This study reveals the indirect influence of Kara Sea ice reduction on cold surge (CS) over the tropical western Pacific (TWP), with the East Asian jet stream serving as the connecting link. The leading mode of CSs over the TWP exhibits a zonal dipole characteristic, which is associated with cyclonic anomaly over the Philippine Sea. The enhanced cyclonic anomaly is caused by strengthened and northward-moved subtropical East Asian jet stream and weakened polar jet stream, which can lead to more CSs over the South China Sea and fewer CSs over the Philippine Sea. Such variations in the jet stream are contributed by the facilitated atmospheric blockings west of the Ural Mountains, which suppressed the circumpolar westerly winds and increased meridional temperature gradient in Northeast Asia. The connection between atmospheric blockings and Kara Sea ice can be confirmed through local vertical energy exchange. Simulations of the atmospheric response to the forcing of decreased Kara Sea ice support the proposed connection. Although there is no statistically significant correlation between tropical CSs and Kara Sea ice, this study highlights the potential impacts of Arctic climate change signal on weather and climate extremes over tropical regions.


Introduction
Previous studies indicate that the reduction of Arctic sea ice is the reason for the Arctic amplification (Comiso et al 2008, Cavalieri and Parkinson 2012, Wu et al 2012).This has a profound impact on extreme weather and climate events in mid-to-high latitudes, garnering significant attention and sparking extensive discussions (Chen et al 2017, Sun et al 2018, England et al 2020).
Especially in cold seasons, the weakening of the polar vortex and westerly winds around the pole significantly diminish their control over polar cold air (Kim et al 2014, Wyrwoll et al 2016, Luo et al 2019).Additionally, the reduction of sea ice also has a beneficial impact on the development of atmospheric blocking in high-latitude regions (Gong andLuo 2017, Luo et al 2018).These factors in combination, have led to a notable increase in the amplitude of Rossby waves in mid-to-high latitude regions.The strengthening of high-latitude troughs and ridges guides cold air from high latitudes to move southward, intensifying cold surge (CS) in mid-latitude areas (Park et al 2014, Park et al 2015, Yang et al 2018).Consequently, despite the increase in the global average temperature reducing the number of cold days (Wang and Ding 2006, Wei and Lin 2009, Qu et al 2015), the sudden outbreak of cold air causes intense cooling (Park et al 2011), continuing to pose a significant threat.
Compared to atmospheric blocking in other regions, Ural blocking exhibits a more prominent increase and enhancement in terms of frequency, intensity, and duration after mid-1990s (Davini et al 2012).Such phenomenon is influenced by the melting of sea ice in the Barents-Kara Seas and the unique topography of the Ural Mountains (Luo et al 2016(Luo et al , 2018(Luo et al , 2019)).This makes it more challenging investigate the variations in East Asian CSs significantly influenced by the Ural blocking under global warming.For East Asian CSs, cold air generally follows three paths: source from the western North Atlantic via a western route or from the east of the Ural Mountains through a northwestern path, or from the western side of the Ural Mountains through a northern route (Li 1955, Cai et al 2019).All three paths of cold air accumulation and strengthening occur in the vicinity of Lake Baikal before eventually invading East Asia.Therefore, the Ural blocking plays a crucial role in the variation of East Asian CSs paths.In the presence of weak blocking and strong zonal westerly winds, CSs are more likely to follow the western path.In contrast, with strong blocking, CSs tend to favor the northern path (Yang et al 2020a(Yang et al , 2020b)).
Current researches have mostly focused on the variations of East Asian CSs in mid-high latitudes and their connection to Arctic Sea ice melting (Mori et al 2019, Screen and Blackport 2019, Yang et al 2020a, Cai and Zeng 2022).Due to the vast distance between the Arctic and the tropical regions, as well as the predominant activities in the atmospheric teleconnections over Eurasia during winter, there is limited research on the influence of the Arctic on tropical CSs in the East Asian region.A recent study (Lian et al 2014) has particularly emphasized the importance of investigating CSs over the tropical western Pacific (TWP).It indicates that the outbreaks of these CSs can lead to zonal wind stress anomalies, which may have implications for phase transitions of the El Niño-Southern Oscillation.This study also proposes the key influence of signals from high latitudes on CSs over TWP.Therefore, the main objective of this study is to reveal the primary spatiotemporal variations of CSs over the TWP and explore its potential causal connections with Arctic Sea ice.
This paper is structured as follows: section 2 introduces the data and methods employed; section 3 includes the results obtained and section 4 provides a comprehensive summary and discussion of this study.

Data
This study used the daily reanalysis dataset from the NCEP/NCAR Reanalysis 1 dataset for the period 1979-2021 (Kalnay et al 1996) to identify the winter (December to February of the following year) CSs over the TWP and the associated circulation anomalies, which consists of geopotential height (Z), horizontal winds (U and V), and temperature (T) in multiple levels.Monthly sea ice concentrations are from the Met Office Hadley Centre with a horizontal resolution of 1

Definition of CSs
In contrast to the method of defining CSs in midhigh latitudes through daily temperature drops and daily minimum temperatures (Park et al 2008, 2014, Yang et al 2020a), tropical CSs are typically identified by strong northerly wind anomalies.Consistent with previous research (Xavier et al 2020, Feng et al 2022, Pang et al 2022), here we mark the day with daily meridional wind at 925 hPa below −1 standard deviation as a CS day at each grid.There is also a significant negative correlation between the number of CS days obtained from this definition and surface air temperature (figures 1 and S1).

Definition of atmospheric blocking
The atmospheric blocking is identified by northward meridional gradient (GHGN) and southward meridional gradient (GHGS) of the Z at 500 hPa at each grid point following by Tibaldi and Molteni (1990): where ϕ = 35 • , 37.5 • …, 75 In equation ( 4), (φ , λ) are the latitude and longitude; p represents the pressure; p 0 = 1000 hPa; U = (U, V) is the basic flow; a denotes the radius of the earth; ψ ′ = Φ ′ f is the disturbance of geostrophic stream function relative to basic flow, in which the f means the coriolis parameter and the Φ indicates the geopotential.
The intensity of transient disturbance activity is represented by the transient kinetic energy (k e ) on synoptic scale (Murakami 1979).The formula for k e is: u and v are zonal and meridional winds respectively, and the superscript ' ′ ' represents the disturbance component, which is obtained from 2.5-6 d bandpass filtering of daily wind data (Chen et al 2012).
The overbar represents the winter averaged value.The high value of k e reflects the strong transient wave activity.
The Eady growth rate (EGR) could be used to measure the atmospheric baroclinicity (Eady 1949), which can lead to synoptic-scale transient EGR activities.The following equation calculates the EGR: where f is the Coriolis parameter, N is the Brunt-Vä ˙lsälä frequency, V is the horizontal wind velocity, and z is the vertical coordinate.Negative EGR anomalies are associated with weakening atmospheric baroclinicity, favoring the development of anticyclonic anomalies (Crawford andSerreze 2016, Wu andFrancis 2019).

Results
According to the definition of the CSs in section 2. Figure 2 depicts the spatial distribution of CS days associated with the first EOF mode and the corresponding circulation anomalies.In years with positive PC1, the South China Sea experiences an average of 14 CS days per year, while the Philippine Sea witnesses 4-8 CS days.Conversely, in years with negative PC1, the South China Sea sees around 8 CS days per year, while the Philippine Sea witnesses 10 CS days.The dipole pattern of the frequency of CS days is closely related to the low-level cyclonic anomaly over the Philippine Sea.A positive PC1 is associated with stronger north wind anomaly over the South China Sea and notably intensified southerly wind anomaly over the Philippine Sea.Here, we define the Cyclonic Index as the average Z at 850 hPa over the region (5-25 • N, 105-135 • E).The results indicate that the correlation coefficient between the Cyclonic Index and PC1 has reached a value of 0.50, which is significant at the 99% confidence level.A recent study (Zhang et al 2023) has shown that the opposite trends CSs in the South China Sea and the Philippine Sea are related to their upstream cold air activity pathways.Therefore, this study mainly focuses on their relationship with high latitude circulation.
The variations of these crucial low-level cyclonic anomalies are closely linked to the modulation of large-scale circulation patterns (figure 3(a)).The upper-level circulation pattern manifests as two wave trains propagating eastward, originating from Europe.The northern wave train propagates along the Ural Mountains-Baikal Lake, while the southern wave train follows the path from the Middle East to southern China (figure 3(b)).The polar front jet stream and the subtropical jet stream guide the propagation   1984,2000,2010,2011,2017,2020), there is a notable increase in the zonal wind component in the northern region of the climatological subtropical jet stream location, while both the high-latitudes and northern part of the tropical regions experience a decrease in the zonal wind.Negative zonal wind anomalies (easterly anomalies) over southern China and the Indochinese Peninsula are conducive to promoting cyclonic circulation anomalies over the Philippines.In years with a low Cyclonic Index values (anomalies lower than −1 standard deviation ; 1982, 1986, 1991, 1994, 1997, 2001, 2015, 2018), the opposite circulation conditions are observed (figure 3(d)).
Consistent with previous studies (Xiao et al 2016, An et al 2020, Zhang et al 2022), here the EOF analysis is employed to assess the co-variability of the East Asian jet stream's displacement and intensity (figure 4).The leading EOF mode of zonal wind from 80 • E to 150 • E contributes more than 50% of the total variance, showing a triple pattern with positive zonal wind anomalies in mid-latitudes, negative anomalies in low-latitudes and high-latitudes in high PC years.With reference to the climatological position of the vertical jet stream axis, the positive phase of this mode reflects the strengthening and northward of the subtropical jet stream and the weakening of the polar front jet stream.The identified  mode is significantly correlated with the Cyclonic Index (correlation coefficient reaches a high value of 0.71), highlighting the crucial connection between winter jet streams and the climate anomalies over the TWP region.This relationship still exists in nonstrong El Niño-Southern Oscillation years (figure S2).The partial correlation coefficients among the Nino 4 index (5 • N-5 • S, 160 • E-150 • W), Jet Stream Index, and Cyclonic Index is also calculated to eliminate the impact of El Niño-Southern Oscillation on the Cyclonic Index.The result also revealed a notable positive correlation (with the partial correlation coefficient of 0.34, surpassing the 95% confidenceimilar results can be obtained level) between the Jet Stream Index and Cyclonic Index, after excluding the influence of El Niño-Southern Oscillation on the Cyclonic Index.Similar results can be obtained using Nino 3 and Nino 3.4 index.Despite a diminished correlation, this finding reaffirms the robustness of the relationship between the Cyclonic Index and Jet Stream Index.This suggests that while tropical ocean signals play a crucial role in CSs over the South China Sea Here, we designate the PC1 of the zonal wind as the Jet Stream Index, which serves as a comprehensive measure of the displacement and intensity variations in the East Asian jet stream.Due to its strong statistical correlation with the Cyclonic Index, the circulation patterns related to the Jet Stream Index (figure 5) exhibit a spatial distribution similar to that shown in figure 3.However, it should be noted that the results in figure 5 highlight the role of positive Z anomalies over the Urals region in weakening the polar front jet and enhancing the northward shift of the subtropical jet.The subtropical jet enhances the upperlevel easterly wind anomalies at low latitudes and the lower-level northerly wind anomalies over the South China Sea and the lower-level southerly wind anomalies over the Philippine Sea.This configuration is consistent with the circulation structure on the south side of the jet stream exit area during the strengthened subtropical jet stream (Ren et al 2008, Albern et al 2021).The results of dynamical diagnostics further provide evidence supporting these conclusions (figure 6).
The strengthening and northward shift of the subtropical jet stream, along with the weakening of the polar front jet stream, are accompanied by concurrent variations in regional synoptic scale transient eddy and atmospheric baroclinicity (figures 6(a) and (b)).Positive EGR anomalies appear in most of East Asia, which enhances the activity of cold vortices, favoring the formation of East Asian CSs.The meridional temperature gradient in East Asia exhibits a dipole distribution with a higher gradient in the north and lower in the south (figure 6(c)).The pattern structured with reverse omega formed by the Ural blocking causes the high-pressure ridge to extend northward, leading to the accumulation of cold air in front of the ridge and a decrease in temperatures in northern Asia, resulting in positive temperature gradient anomalies.Furthermore, the cyclonic anomalies over Japan lead to negative temperature advection near the South China Sea and cause a decrease in temperatures over the South China Sea (figure 6(d)).Similarly, as can be observed in figure 6(d), the anticyclonic anomaly over the Ural Mountains region significantly contributes to the weakening of the polar front jet stream.Therefore, further investigation was conducted here to explore the relationship between winter atmospheric blocking and Jet Stream Index (figure 7).The results indicate that the increased frequency of winter atmospheric blocking in the east of the Ural Mountains region (50-70 • N, 55-80 • E) corresponds to the weakening of the polar front jet stream, with a correlation coefficient of 0.37 between the two.
The results of dynamical diagnostics indicate that the increased frequency of winter atmospheric blocking over the Ural Mountains region weakens the high-altitude transient eddy kinetic energy in the polar front jet region, resulting in a zonal distribution of negative anomalies in mid-latitude kinetic energy and the weakening of atmospheric baroclinicity (figures 8(a) and (b)).The increased blocking is associated with warming in high-latitude  regions, leading to a decrease in meridional temperature gradients in polar areas and an enhancement of meridional temperature gradients in the region ahead of the high-pressure ridge (near Lake Baikal).The dipole distribution of temperature gradients with lower gradients in the north and higher gradients in the south favors negative (positive) zonal wind anomalies on the polar (equatorial) side of the tropospheric polar front jet axis (figure 8(c)).This leads to an overall northward shift of both the polar front jet stream and the subtropical jet stream, accompanied by concurrent changes of a weakened polar front jet stream and a strengthened subtropical jet stream.Furthermore, the increased frequency of Ural blocking corresponds to the strengthening of winter surface continental high-pressure systems (figure 8(d)), providing favorable conditions for the accumulation of cold air and the outbreak of tropical CSs.
Atmospheric blocking is not only a significant manifestation of internal atmospheric variability but is also influenced by external forcing factors, such as the notable impact of Arctic Sea ice (Luo et al 2016(Luo et al , 2018(Luo et al , 2019)).To eliminate the linear influence of global warming on the calculation of statistical correlation coefficients, the analysis here involves calculating the correlation coefficient distributions between the Jet Stream Index, Blocking Index, and Arctic Sea ice after removing the linear trends from the data (figure 9).The results indicate that the reduced sea ice in the Kara Sea (70-78 • N, 52-78 • E) is closely associated with the increased frequency of Ural blocking as well as the strengthening and northward shift of the subtropical jet stream.The reduction in sea ice over the Kara Sea is accompanied by positive anomalies in turbulent heat flux and upward longwave radiation (figure 10), indicating an unusual process of heat transfer from the surface to the overlying atmosphere.This contributes to the formation of local atmospheric blocking patterns.Table 1 provides the correlation coefficients among all the indices used in this study.It can be observed that the main variations in CS days over the TWP and Cyclonic Index have a very weak direct correlation with high-latitude circulation systems.However, the jet stream plays a crucial role by acting as a significant link, connecting highlatitude circulation with tropical climate anomalies.These conclusions emphasize the significant potential impact of Arctic warming on tropical climate.In addition, the negative correlation between the sea ice in Kara Sea and jet stream or PC1 has becoming more significant after mid-1990s (figure S3).
We also used Community Atmosphere Model version 5.3 (select the finite-volume dynamical core configured with a horizontal resolution of 1.9 • latitude ×2.5 • longitude (f19_f19) and 30 vertical hybrid levels, Neale et al 2012) to design corresponding numerical simulations (table S1 and figure S4    decrease in CSs over the Philippine Sea, confirming observations of the potential impact of Arctic warming on tropical CSs.

Conclusion and discussion
This study elucidates the leading variation mode of winter CSs over TWP from 1979-2021 and investigates its potential connection with Arctic Sea ice through statistical analysis, dynamic diagnostics, and numerical simulations.The results of EOF analysis indicate that the dominant mode of CSs over TWP exhibits a zonal dipole structure and displays significant interannual variations.Despite the absence of a robust established direct statistical correlation to prove the association between this mode and the reduction in Arctic Sea ice, the decrease in Kara Sea ice could potentially promote atmospheric blocking west of the Ural Mountains through meridional energy exchange with the atmosphere, which could further suppress the polar front jet stream and enhance the meridional temperature gradient in Northeast Asia.Such dynamic processes result in the northward shift and strengthening of the East Asian subtropical jet stream.Consequently, the northward-moved subtropical jet stream has resulted in a strengthening of high-level easterly winds over the TWP, which favors the intensification of low-level cyclonic anomaly over the Philippine Sea.The series of processes ultimately influences and regulates the dipole mode of CSs over the TWP.Furthermore, numerical simulations reproduced the indirect causal link between the melting of Kara Sea ice and variations in CSs in the tropical Pacific.
An increasing body of research has confirmed the great impact of Arctic warming on extreme events in mid-to-high latitudes.However, exploring its influence on tropical regions still requires more attentions.Serving as a bridge between Arctic Sea ice and tropical climate anomalies, atmospheric blocking and the East Asian jet stream exhibit a dual nature: respond to external forcings and substantial internal atmospheric variability.This adds an element of instability to the potential impact of Kara Sea ice on CSs over the TWP.Recent studies also show that increasing member size helps reveal the robust response of the atmosphere to sea ice by converging internal variability, and it is recommended that future studies use super large member simulations to further explore the potential impact of Arctic sea ice on extreme weather at mid-low latitudes (Ye et al 2024).

Figure 1 .
Figure 1.(a) Standard deviation of meridional wind anomalies at 925 hPa in winter (unit: m s −1 ).(b) The annual averaged spatial distribution of winter CS frequency from 1979 to 2021.The (c) first EOF mode of winter CS frequency and its (d) principal component time series (PC1).

Figure 2 .
Figure 2. Composite of CS frequency over TWP in years with (a) positive and (b) negative PC1. Linear regression of winter (c) Z anomalies (shading, unit: gpm) and horizontal wind anomalies (vector, unit: m s −1 ), and (d) meridional wind anomalies (unit: m s −1 ) at 850 hPa on PC1.Dotted areas are significant at the 95% confidence level.(e) Time series of Cyclonic Index and PC1 from 1979 to 2021.The black box in (c) represents the key region of Cyclonic Index.

Figure 3 .
Figure 3. Linear regression of winter (a) Z anomalies (shading, unit: gpm) and horizontal wind anomalies (vector, unit: m s −1 ), and (b) WAF (vector, unit: m 2 s −2 ) and stream function (shading, unit: 10 5 kg s −1 ) at 300 hPa on Cyclonic Index.Composite of zonal wind anomalies (shading, unit: m s −1 ) at 300 hPa in years with normalized Cyclonic Index (c) higher than 1 and (d) lower than −1.The thick black solid line represents the climate state position in the 30 meter wind speed area.Dotted areas are significant at the 95% confidence level.

Figure 4 .
Figure 4.The (a) first EOF mode of winter zonal wind spanning from 80 • E to 150 • E and its (b) PC time series (gray bars).The black solid and dashed lines in (a) represent the climatic state positions of the subtropical jet axis and the polar front jet axis respectively.The black line in (b) indicates the normalized Cyclonic Index.

Figure 5 .
Figure 5. Linear regression of winter Z anomalies (shading, unit: gpm) and horizontal wind anomalies (vector, unit: m s −1 ) at (a) 300 hPa and (b) 850 hPa on Jet Stream Index.(c) Linear regression of winter WAF (vector, unit: m 2 s −2 ) and stream function (shading, unit: 10 5 kg s −1 ) at 300 hPa on Jet Stream Index.Dotted areas are significant at the 95% confidence level.

Figure 7 .
Figure 7. (a) Linear regression of winter atmospheric blocking frequency (unit: day) on Jet Stream Index.(b) Normalized time series of winter Jet Stream Index and Blocking Index from 1979 to 2021.
), and the simulation results reproduced the response of circulation (figures S5 and S6) and CSs frequency over the TWP (figureS7) to Kara Sea ice reduction.Considering the nonlinear response of circulation to sea ice(Zhang and Screen 2021), only low sea ice experiments (LSIC) and control experiments (CTRL) were included in the simulation scheme.The loss of Kara Sea ice promotes a positive anomaly in local Z, with its quasi-barotropic structure extending from lower to upper levels.The enhancement of blockings suppresses the westerly wind anomalies at high latitudes, weakening the polar front jet stream, and indirectly forming cyclonic anomalies to stimulate the subtropical westerly jet stream to strengthen and move northward.In general, the climate state distribution of CS frequency over the TWP in the simulation results is slightly lower than the observation results, while their spatial pattern is similar to the observation results.It indicated that there are still other driving factors that regulate the patterns of CS frequency over the TWP, such as El Niño-Southern Oscillation and Indian Ocean Dipole (Corporal-Lodangco et al 2016,Zhou et al 2019).In particular, the difference between LSIC and CTRL shows the characteristics of a zonal dipole structure.The reduction of sea ice leads to an increase in CSs over the South China Sea and a

Figure 9 .
Figure 9. Correlation coefficient (without trend) between winter sea ice and (a) Jet Stream Index and (b) Blocking Index.Dotted areas are significant at the 95% confidence level.(c) Normalized time series (without trend) of winter Sea Ice Index (black line), Jet Stream Index (blue line), and Blocking Index (red line) from 1979 to 2021.

Figure 10 .
Figure 10.Linear regression of winter (a) turbulent heat flux anomalies (shading, unit: W m −2 ) and (b) upward longwave radiation flux anomalies (shading, unit: W m −2 ) on Sea Ice Index.Dotted areas are significant at the 95% confidence level.
(Yin and Zhang 2021) 2003).Notably, previous research has indicated that the circulation anomalies associated with CSs in TWP and the main characteristics of the jet stream remain consistent across different reanalysis datasets(Yin and Zhang 2021).
such as the South China Sea and the Philippine Sea.The number of CS days in the South China Sea region is approximately 10-13 d per year, while in the Philippine Sea region, it is approximately 7-10 d per year.To capture the primary spa- 2, we calculated the distribution of annual averaged winter CS frequency over TWP.Figures1(a) and (b) indicate that the climatological distribution of winter CS days over the TWP region decreases from northwest to southeast, with greater variability over the maritime areas, tiotemporal variations of the CS days, the empirical orthogonal function (EOF) analysis is conducted (figures 1(c) and (d)).The first mode of EOF accounts for 31.47% of the total variance and exhibits a zonal dipole pattern, with its time series showing significant interannual variation.
a Significant at the 99% confidence level.b Significant at the 95% confidence level.