Linking Arctic stratospheric polar vortex weakening to rising CO2-induced intensification of the Indo-Pacific warm pool during the past five decades

Accompanying the global rise in greenhouse gas emissions, a warming trend in the Indo-Pacific warm pool (IPWP) has exerted a discernible influence on tropical atmosphere–ocean interactions. However, the impact of this intensification of the IPWP on the Arctic stratospheric polar vortex (ASPV) remains unclear. In this study, we revealed a link between the changes in the IPWP and ASPV during the early winter months, with nearly half of the weakening in the ASPV attributable to the intensification of the IPWP from 1968 to 2020. Wave trains triggered by the elevated SST in the IPWP region lead to enhanced eastward-propagating flux convergence in the northern high-latitude stratosphere, ultimately resulting in a diminishing ASPV. With increasing atmospheric greenhouse gas, the ASPV is poised to further weaken in the future, particularly in the context of a more intense IPWP. Our finding has significant implications for early winter ASPV strength and location prediction and seasonal weather forecasting.


Introduction
The Arctic stratospheric polar vortex (ASPV) refers to a circumpolar westerly wind system that typically resides in the Arctic stratosphere spanning altitudes of 10-50 km during the boreal winter months (Waugh et al 2017).Existing research has demonstrated that the ASPV's strength and position influence not only the movement of the tropospheric jet and storm tracks (Kidston et al 2015), but also surface weather patterns across northern Eurasia and North America (Kolstad et al 2010, Gerber et al 2012, Huang and Tian 2019).Furthermore, the state of the ASPV serves as a pivotal factor in seasonal weather predictions (Scaife et al 2016, Jia et al 2017).Consequently, accurate characterization of the ASPV's location and strength is important for enhancing the accuracy of seasonal forecasts (Tripathi et al 2015, Scaife et al 2016).
Prior studies have indicated that the variability of the wintertime ASPV arises from a multitude of large-scale tropical climate phenomena operating across intraseasonal to interannual timescales, including the El Niño-Southern Oscillation (ENSO) (Garfinkel andHartmann 2008, Hurwitz et al 2014), the Madden-Julian Oscillation (Garfinkel et al 2014, Liu et al 2014), the Indo-Pacific warm pool (IPWP) (Zhou et al 2019), and the Quasi-Biennial Oscillation (Liu et al 2014, Watson and Gray 2014, Rao et al 2020).
However, understanding the trajectory of wintertime ASPV trends remains a subject of ongoing debate, with outcomes varying depending on the study period considered.Garfinkel et al (2017) suggested a weakening trend in the ASPV since 1980, which intensified notably after 1990, owing to rising concentrations of greenhouse gases.Some researchers also attributed the decline in ASPV intensity from 1979 to 2012 to preceding Arctic Sea ice loss, particularly in the Barents and Kara Seas (Jaiser et al 2013, Kim et al 2014, Yang et al 2016, Zhang et al 2016, Chen and Wu 2018).Sun et al (2015), through idealized numerical experiments, highlighted divergent effects of sea ice loss in the Atlantic and Pacific sectors on the ASPV and documented a shift of the vortex towards the Eurasian continent since 1979 in response to sea ice loss in the Barents and Kara Seas.The aforementioned Arctic Sea ice loss has been associated with global warming resulting from increasing greenhouse gas emissions (Screen andSimmonds 2010, Stroeve et al 2012).However, Seviour (2017) contended that the weakening and shift of the ASPV stemmed from internal forcing rather than a response to greenhouse gas increase.In contrast, Hu et al (2018) identified an increasing trend in ASPV strength from 1998 to 2016, attributable to warming in the central North Pacific.Moreover, warming in the Indian and Atlantic Oceans, coupled with cooling in the tropical Pacific Ocean, was found to contribute to colder ASPV conditions in late winter and early spring (Garfinkel et al 2015).
While prior studies have explored trends in ASPV intensity and their relationship to greenhouse gas emissions, there remains variability in their conclusions regarding the role of such emissions.In this study, we aim to dissect the influence of rising greenhouse gas emissions on changes in ASPV intensity.Specifically, we examine the indirect impact of rising greenhouse gases through the IPWP (remote effect), in contrast to the direct and localized influence of Arctic Sea ice loss as a response to increasing CO 2 emissions (Notz and Stroeve 2016).This approach has been motivated by recent research that has primarily associated the warming and expansion of the IPWP with the increase in greenhouse gases (Weller et al 2016, Bai et al 2022).While prior studies have explored the association between ASVP intensity and IPWP strength, with a primary focus on their interannual relationship (Zhou et al 2018(Zhou et al , 2019)), the present study seeks to fill a critical gap in the existing literature by quantifying the impact of IPWP intensification on the long-term weakening trend of the ASVP, thereby making a novel and significant contribution to this field of study.
Additionally, it is important to note that the ASPV exhibits distinct trends in early winter (November-December) and late winter (January-February) (Fu et al 2010, Young et al 2012).Zhao et al (2022) and Tian and Fan (2020) have also noticed the differential impact of ENSO on the East Asian winter monsoon in early and late winters.They demonstrated that SST anomalies in the tropical Indian Ocean during early winter can trigger an atmospheric wave train propagating from the northern Indian Ocean to East Asia and the North Pacific.In contrast, they found that SST anomalies in the Indian Ocean during late winter do not induce such a wave train.For the scope of this study, we focus solely on investigating the indirect effect of external forcing through IPWP on ASPV variability in early winter.

Dataset and methods
The analysis in this study covers the period from 1968 to 2020, selected to ensure the availability of long-term, reliable data for our investigation.The variables essential to our investigation were sourced from the European Centre for Medium-Range Weather Forecasts' fifth-generation reanalysis (ERA5, Hersbach et al 2020), which included monthly sea ice concentration and geophysical height data in lower stratosphere at a spatial resolution approximately 0.25 • latitude × 0.25 • longitude.For early winter sea surface temperature (SST) data, we relied on the U.S. National Oceanic and Atmospheric Administration (NOAA) Extended Reconstructed SST V5 dataset, with a spatial resolution of 2.0 • latitude × 2.0 • longitude (Huang et al 2017).The Intensity of the IPWP was quantified as the areaaveraged SST within the region spanning 60 • E to 170 • E and 15 • S to 15 • N. The IPWP index used in our analysis is available for download from the following website: https://psl.noaa.gov/gcos_wgsp/Timeseries/Data/pacwarmpool.ersst.data.
Data for monthly outgoing longwave radiation (OLR) at the top of the atmosphere, which serves as a proxy for convective activity strength, were obtained from NOAA Interpolated OLR data (Liebmann and Simth 1996).
To elucidate the generation and propagation of planetary waves, we employed the Rossby wave source (RWS) as defined by Sardeshmukh and Hoskins (1988) and the wave activity flux described by Takaya and Nakamura (2001).Additionally, to visualize the propagation of stationary waves in a latitude-altitude profile, we utilized the Eliassen-Palm (EP) flux, defined following the approach outlined by Edmon et al (1980).
Statistical analyses in this study relied on linear correlation and regression methods.Significance levels were assessed using a two-tailed Student's t-test.Prior to conducting correlation and regression analyses, we removed any underlying trends present in the variables to ensure the robustness of correlation results.
To further validate our statistical findings, we carried out numerical experiments employing version 5 of the Community Atmosphere Model (CAM5), which serves as the atmospheric component of the Community Earth System Model (CESM) and is comprehensively detailed in Neale et al (2011).Our CAM5 simulations used a horizontal resolution of 1.9 • latitude × 2.5 • longitude and 30 vertical levels.Specifically, we conducted a 50 year control simulation driven by the climatological annual cycle of SST and sea ice concentration data from the Hadley Center.Additionally, we executed a 20 year sensitivity simulation, mirroring the final two decades of the control simulation, with the sole exception of introducing a 2 • C warm SST anomaly within the IPWP region (80 • -160 • E, 0 • -20 • N).This SST anomaly was imposed annually from 1 November to 31 December.

The relation between ASPV and IPWP
Over the past five decades, a discernible weakening of the ASPV has been observed, characterized by a positive trend in 150 hPa geopotential height across the Arctic region, with exceptions in the Canadian Arctic Archipelago and areas near Greenland (figure 1(a)).To quantify the intensity of the ASPV, we employed geopotential height anomalies averaged over 70 • N-90 • N at 150 hPa, wherein a positive (negative) averaged height anomaly corresponds to a weakened (strengthened) ASPV, as per the definition by Garfinkel et al (2017).Our analysis reveals a notable increasing trend in 150 hPa height, with a rate of 1.28 gpm yr −1 (figure 1(b)), indicative of a weakening ASPV.
Concurrently, the intensity of the IPWP has exhibited a discernible increasing trend of 0.015 • C yr −1 (figure 1(d)).Notably, the intensity of the IPWP displays a significant correlation with the intensity of the ASPV, with correlation coefficients of 0.41 (p < 0.01) considering their trends and 0.33 (p < 0.05) with their trends removed.This stronger IPWP results in widespread positive height anomalies across the Arctic region, with exception of negative height anomalies over the Norwegian Sea and North Europe (figure 1(c)).While the correlation between the IPWP and ASPV intensity is not exceptionally high, the significant trend in IPWP intensity contributes to 50% of the trend in height anomalies at grid points with significant trends.This observation aligns with the findings of Zhou et al (2019), underscoring the role of the strengthened IPWP in driving a weakened ASPV on the interannual time scales.

Anomalous SST and atmospheric circulations
In the subsequent analysis, we employ linear regression to elucidate how the IPWP influences the ASPV during early winter.Regressing atmospheric and oceanic variables onto the normalized time series of IPWP intensity, we observe SST anomalies associated with a stronger IPWP that exhibit positive values across tropical oceans, flanked by negative values in subtropical and mid-latitude oceans (figure 2(a)).Specifically, SST anomalies in the tropical Pacific and Indian Oceans signify an El Niño state and a positive ocean basin mode, respectively.These SST anomalies correspond to convective precipitation anomalies with increased values over the equatorial and subtropical Pacific Ocean and decreased values to the north and south (figure 2(b)).In the tropical Indian Ocean, there is a contrasting pattern of anomalous convective rainfall.This results in stronger divergent winds and negative RWS (figure 2(c)) due to intensified convective rainfall over the northern Indian Ocean and South China Sea.Consequently, this excites a planetary wave train (figure 2(d)) that propagates northeastwards and eastwards into the North Pacific Ocean.A portion of this wave train enters the Arctic, some reflects back into the tropics, and the remainder continues eastwards into North America.A portion of the wave train also extends into the tropical Atlantic Ocean before predominantly propagating into Greenland, resulting in positive height anomalies over the Arctic.The wave train further propagates southeastwards into Europe and western Asia.
This wave train, projected in vertical-meridional profiles as the divergence (convergence) of EP flux at low (high) latitudes in the stratosphere and upperlevel troposphere (figure 3(a)), generates weaker zonal winds at 50 • N-70 • N (figure 3(b)) and warmer 150 hPa air temperatures in high latitudes (figure 3(c)).These conditions facilitate the weakening of the ASPV.Consequently, over the past five decades, the strengthened IPWP, linked to increased greenhouse gas emissions, has facilitated the weakening of the ASPV and its shift towards Greenland and North America (figure 1(d)).
It is noteworthy that this planetary wave train aligns with findings in recent studies (Tian andFan 2020, Zhao et al 2022).In particular, Zhao et al (2022) provided evidence that SST anomalies in the tropical Indian Ocean during early winter can instigate the propagation of a planetary wave train from the northern Indian Ocean to East Asia and the North Pacific.In contrast, the study also revealed that SST anomalies in the Indian Ocean during late winter do not exhibit a similar capability in inducing such a wave train.This, in conjunction with the distinct ASPV trend patterns observed between early and late winter, provided the impetus for our study to exclusively focus on early winter.

Numerical simulation results
To corroborate the statistical findings above, we conducted numerical simulations using the CAM5 atmospheric model to further demonstrate how warm SST anomalies in the IPWP region may alter atmospheric circulations and contribute to the changes in ASPV.As depicted in figure 4(a), the difference in the anomalous 200 hPa geopotential height between the numerical simulation where a 2 • C warm SST anomalies is introduced in the IPWP region (the

The role of planetary waves and Arctic Sea ice loss
Previous studies have underscored the significance of wavenumber 1 and 2 disturbances in influencing the winter stratosphere and ASPV.In figure 5, we evaluate the vertical and horizontal patterns of planetary waves related to IPWP, decomposing them by wavenumber using a Fourier transformation along latitude circles.During the positive phase of the IPWP, the wave-1 response is approximately in quadrature with the background stationary wave, with the quadrature tending to be in phase as altitude increases (figures 5(a) and (c)).Conversely, the wave-2 response is out of phase with the background wave (figures 5(b) and (d)).The combined effects of wave 1 and 2 interact constructively with the background stationary wave, contributing to the weakening of the ASPV.
Prior investigations have examined the role of Arctic Sea ice depletion in driving trends in the ASPV.To assess this, we computed the mean sea ice concentration north of 60 • N during the early winter period.Our analysis revealed a noteworthy declining trajectory in domain-averaged Arctic Sea ice concentration, characterized by a rate of −0.026 per decade (figure S1(a)).Subsequently, we quantified the influence of Arctic Sea ice loss on changes in the ASPV by multiplying the trend in domain-averaged Arctic Sea ice concentration with the regression pattern of the 150 hPa height anomalies onto the normalized time series of Arctic Sea ice concentration.The resultant 150 hPa height anomalies, linked to Arctic Sea ice loss, exhibited positive anomalies across the Arctic expanse, with the exceptions of northwestern Canada, the Norwegian Sea, and North Europe (figure S1(b)).This comprehensive analysis unveiled a correlation between the intensity of the ASPV and domain-averaged Arctic Sea ice concentration, yielding coefficients of −0.38 (p < 0.01) when considering their trends, and −0.29 (p < 0.05) when trends are removed.Remarkably, in grid points with discernible trends, Arctic Sea ice loss accounted for a substantial 60% of the trend in height anomalies.Consequently, over the past five decades, the diminishing extent of Arctic Sea ice has contributed substantially to the weakening of the ASPV, aligning with the findings of prior studies (Jaiser et al 2013, Kim et al 2014, Yang et al 2016).
However, it is important to note that the relationship between the domain-average Arctic Sea ice concentration and the intensity of the IPWP exhibited a distinctive profile.While their correlation coefficient is strong at −0.84 (p < 0.01), indicating a robust negative correlation, this relationship lost significance when trends are removed, with a coefficient of −0.12 (p > 0.05).This outcome underscores that, on interannual timescales, the intensity of the IPWP remains largely unconnected to the broader context of Arctic Sea ice cover.

Conclusion and discussion
We have conducted a comprehensive examination of the influence of the intensified IPWP on the ASPV during the early winter, utilizing global reanalysis data and idealized numerical simulations with the CAM5 model spanning the period from 1968 to 2020.
Over the past five decades, our findings reveal a compelling relationship between the escalating IPWP intensity and a weakened ASPV, explaining nearly half of the trend observed in the 150 hPa geopotential height anomalies.The mechanism underlying this relationship is a cascade of events initiated by the presence of positive SST anomalies in the northern Indian Ocean and South China Sea, which, in turn, act as catalysts for increased convective activity.This increased convection triggers a propagating wave train that travels from the northern Pacific and Atlantic Oceans into the Arctic region (Zhou et al 2019).An examination of EP flux in vertical-meridional profiles, closely linked to this wave train, reveals a convergence in the high-latitude stratosphere-a key driver behind the observed weakening of the ASPV.This phenomenon is further substantiated by the findings of This study makes pioneering contributions by being the first to systematically explore the influence of an intensified IPWP on the weakening ASPV over the past five decades, extending beyond the conventional focus on interannual timescales (Zhou et al 2018(Zhou et al , 2019)).Furthermore, our examination centers on the early winter period, offering fresh insights compared to a prior study that focused on late winter effects (Zhou et al 2018).In addition, while prior research has predominantly underscored the influence of Arctic Sea ice loss resulting from greenhouse gas increases on the ASPV, our study accentuates the remote impact of the strengthening IPWP, a consequence of increasing greenhouse gas concentration, on the changes in ASPV strengths.This nuanced distinction delineates the indirect influence on ASPV from its local and direct effects.Notably, we find that the direct impact of overall Arctic Sea ice loss on the weakened ASPV differs from but is comparable to that of the IPWP.Importantly, we acknowledge the potential influence of internal forcings within the climate system on sea ice loss and ASPV intensity, as suggested by numerous studies in the literature (Jaiser et al 2013, Kim et al 2014, Garfinkel et al 2015, Yang et al 2016, Seviour 2017, Yu et al 2017, 2019, 2022, Hu et al 2018, Yu and Zhong 2018).Furthermore, Bai et al (2022) revealed that over the period from 1953 to 2012, more than 95% of the rapid warming of the observed IPWP was detectable and attributable to human influence.Notably, the impact of the Interdecadal Pacific Oscillation (IPO) on the weakening of the ASPV has remained relatively minor over the past five decades.
Prior studies have examined the impact of Arctic Sea ice loss on the weakening of the Arctic polar vortex (Jaiser et al 2013, Kim et al 2014, Yang et al 2016, Chen and Wu 2018).However, a pivotal question arises: is there a dynamic interplay between Arctic Sea ice loss and the intensified IPWP?Our investigation into this interplay has unveiled an existing connection, albeit one of relatively small magnitude.The correlation analysis between domain-average Arctic Sea ice concentration and IPWP intensity unveiled a robust negative correlation of −0.84 (p < 0.01), signifying a substantial association.Nonetheless, it is imperative to note that this correlation loses statistical significance when underlying trends are removed, yielding a coefficient of −0.12 (p > 0.05).Moreover, the 150 hPa height anomalies linked to Arctic Sea ice loss exhibited positive anomalies throughout the Arctic region, except for select regions such as northwestern Canada, the Norwegian Sea, and North Europe (figure S1(b)).These observed height anomalies are notably distinct from those correlated with the intensified IPWP (figure 5).Our findings suggest that while there may indeed be an interplay between Arctic Sea ice loss and IPWP, the resulting impact on the ASPV may not be significantly pronounced.This nuanced relationship is a fundamental aspect of our study, as it underlines the complex dynamics that influence the behavior of the Arctic polar vortex.Our finding on the IPWP and ASPV relationship holds significant implications for the prediction of ASPV strength and location during the early winter.As atmospheric greenhouse gas levels continue to rise, our results suggest that the ASPV is poised to further weaken in the future, particularly in the context of a more robust IPWP.This weakening trend is notably concentrated in the northern regions of North America and the Arctic Ocean, increasing the likelihood of cold air masses intruding into North America.Consequently, predictions of extreme cold events in North America should take into account the anomalies in the IPWP.
However, it is important to note that our study solely examines the impact of a strengthened IPWP on the early winter ASPV, leaving the assessment of late winter effects to be explored in future research endeavors.
The data that support the findings of this study are openly available at the following URL/DOI: http:// nsidc.org/data/NSIDC-0051.

Figure 1 .
Figure 1.Trends in 150 hPa geopotential height in early winter (November-December) (gpm yr −1 ) (a), the time coefficients of geopotential height anomalies averaged over 70 • N-90 • N at 150 hPa (b), the contributions of IPWP intensity change to 150 hPa geopotential height (gpm yr -1 ) (c), The intensity of the IPWP ( o C yr -1 ) (d).The dotted regions in panel (a) indicate above 95% confidence level.The dashed line in panels (b) and (d) denotes the trend.