Lingling Suo

Nansen Environmental and Remote Sensing Center and Bjerknes Center for Climate Research, Bergen, Norway

Keywords: Tropical sea surface temperature Arctic climate change Arctic tropospheric warming Teleconnections

ABSTRACT The contribution of the tropical sea surface temperature in the Arctic tropospheric warming during 1979–2013 was studied through simulations using the CAM6-Nor atmospheric general circulation model.Results showed that the tropical sea surface temperature explained about 30%–40% of the autumn warming and the January warming in the historical simulation.This implies that the tropical sea surface temperature could have been one of the main drivers of Arctic winter tropospheric warming between 1979 and 2013.The tropical sea surface temperature impacts generally came from a combination of the effects of the tropical central-eastern Pacific,the tropical Indo-western Pacific,and the tropical Atlantic,except for the January warming below 850 hPa,which was dominated by the tropical Indo-western Pacific impacts.

1.Introduction

The Arctic has been experiencing dramatic warming in recent decades (Cohen et al.,2014),and this warming is occurring from the surface up to the high troposphere throughout the whole year,with the most significant warming in the autumn–winter seasons (Screen et al.,2012;Screen and Simmonds,2010a).The surface warming has mainly been attributed to the sea-ice retreat and associated albedo feedback(Screen and Simmonds,2010b;Suo et al.,2015).The extended summer tropospheric warming after 1979 can be largely attributed to external forcing variations (Suo et al.,2022).However,an explanation for the extended winter warming throughout the troposphere in the Arctic is still lacking.Suo et al.(2022) compared the contributions of external radiative forcing,Arctic sea-ice retreat,interdecadal Pacific variability,and Atlantic multidecadal variability,and found none of them could explain the winter Arctic tropospheric warming in recent decades,which indicates that the sea surface temperature(SST)in other regions might play a role.

Tropical SST variations can drive simulated warming as observed based on the annual mean temperature in the past several decades over northeastern Canada and Greenland by altering the North Atlantic Oscillation,and the impacts are possibly from the tropical Pacific(Ding et al.,2014),which could be reinforced by the contribution from the tropical Atlantic SST during the winter season(McCrystall et al.,2020).The Indian Ocean SST has also been proposed to have the potential ability to change the Arctic winter climate,either by driving a long-term trend of Arctic winter circulation(Jeong et al.,2022)or by changing the Atlantic Meridional Overturning Circulation(Xu et al.,2022).

Thus,the tropical SST could be a factor contributing to the Arctic tropospheric warming,especially during the winter season.The present study is a continuation of the work of Suo et al.(2022) by trying to isolate the contributions of the tropical SST to Arctic warming.Section 2 introduces the model experiments and methods.The analysis of the results is presented in Section 3,followed by some further discussion and the conclusions in Section 4.

2.Methods and model experiments

Four sets of experiments were performed using CAM6-Nor,the atmospheric component of NorESM2 (Seland et al.,2020).The first one was the historical experiment (His),in which the boundary conditions were the daily sea ice and SST from HadISST2.2.The second one was a tropical SST (TroSST) sensitivity experiment,in which the daily climatological SST was prescribed in the tropics (20?S–20?N) and the other settings were the same as in His.The buffer zones of 25?–20?S and 20?–25?N were where the climatological SST at 20?S/20?N was linearly changing to the observed SST at 25?S/25?N.Comparing the results of the first and second experiment showed the implications of the prescribed TroSST.However,the 1979–2013 tropical SST trend is not zonally uniform,with warming in the tropical Atlantic and Indo-western Pacific and cooling in the central-eastern Pacific (Schneider and Deser,2018).To further locate where the TroSST impacts come from,two more experiments were designed with regional SST variations inhibited: (1) a tropical central-eastern Pacific SST (CEPSST)sensitivity experiment,in which the daily climatological SST was set in the region (20?S–20?N,160?E–70?W) but with the other settings the same as in His;and (2) a tropical Indo-western Pacific SST (I-WPSST) sensitivity experiment,in which the daily climatological SST was prescribed in the region(20?S–20?N,30?–160?E)but with the other settings the same as in His.The 5?buffer zones were also adopted in these two sensitivity experiments.All four experiments were forced by CMIP6 external forcing,starting from 1979.There were 30/20 ensembles for the His/three sensitivity experiments.

The impacts of TroSST,CEPSST,and I-WPSST were obtained as the differences between His and the rest respective experiment.Thet-test(α=0.05,two sided)was used to verify if the differences were statistically significant.Since a tropical Atlantic SST (TroAtlSST) sensitivity experiment was not conducted,it was not possible to isolate the impacts of TroAtlSST using the same method.Instead,the residual part,obtained as the TroSST effects minus those of CEPSST and I-WPSST,assuming the impacts from different regions could be linearly summed,was treated as an estimation of the TroAtlSST impacts,but this might contain some biases caused by the nonlinearity in the combined effects from different regions.The TroAtlSST impacts refer to the part not explained by CEPSST and I-WPSST and possibly driven by TroAtlSST,which are also shown for comparison.

ERA5 reanalysis data set (Hersbach et al.,2020),which has been widely used in studies about Arctic warming (Liang et al.,2021;Suo et al.,2022),was also used here for comparison with the simulation results.

3.Results

The Arctic tropospheric warming is generally located below the 300 hPa pressure level,as shown in previous studies(Screen et al.,2012;Suo et al.,2022).Thus,the 1979–2013 Arctic (north of 67?N) temperature trend from the surface to 300 hPa in ERA5 is shown here in Fig.1(a)for comparison with the model output in Fig.1(b).Similar to the multimodel simulation averages shown in Suo et al.(2022),His generally reproduced the seasonality and vertical structure of the patterns in ERA5,but also with an overestimation of the warming at about 0.3/0.1?C/10 yr higher in the mid-troposphere during the summer/-winter–spring seasons(Fig.1(b)).The difference in April warming under 850 hPa between the simulated results and ERA5 was discussed in Suo et al.(2022).

Generally,TroSST contributed to the Arctic warming from the surface to the high troposphere during summer and autumn(Fig.2(a)).The strongest warming driven by TroSST reached 0.2?C/10 yr in September,which accounted for about 30%–40% of the simulated warming in His(Fig.2(e)).Interestingly,none of the sub-regions’ SST in the tropics brought such significant warming (Fig.2(b–d)).This indicates that the TroSST effects during summer and autumn come from the overlapping or interactions of TroAtlSST,CEPSST,and I-WPSST impacts.

TroSST also caused a marked January Arctic warming throughout the troposphere,with the most significant warming of about 0.2?C/10 yr occurring at 400–500 hPa(Fig.2(a)),which explained up to 40%of the simulated warming there in His(Fig.2(e)).The January Arctic warming below 850 hPa driven by TroSST mainly came from the effects of IWPSST (Fig.2(b)),which were counteracted by the cooling effects of TroAtlSST (Fig.2(d)).However,none of the TroAtlSST,CEPSST,and IWPSST impacts dominated the January Arctic warming in the mid-tohigh troposphere driven by TroSST.This implies again the importance of the combined action of TroAtlSST,CEPSST,and I-WPSST.

Fig.3 shows the spatial distribution of the TroSST impacts on the 1979–2013 temperature and geopotential height trend at 400 hPa in September and January,where and when the most significant impacts appeared.In September,TroSST caused widespread warming north of 40?N,except over the northern Pacific regions,and the warming reached and covered the whole Arctic region (Fig.3(a)).The warming was associated with three anomalous highs over the eastern North Atlantic,eastern Eurasia,and Hudson and Baffin Bay,which brought more warm air flow from the south into the central Arctic(Fig.3(c)).In January,the temperature and geopotential height changes caused by TroSST were different from those in September: TroSST caused significant warming over the North Pacific and centered over the Norwegian Sea,extending to the north of 67?N (Fig.3(b)),which was associated with the significant anomalous high centered over the eastern North Pacific and North Atlantic (Fig.3(d)).

4.Discussion and conclusion

In this work,the contribution of TroSST to the Arctic tropospheric warming in 1979–2013 was studied through simulations using the CAM6-Nor atmospheric general circulation model.Results showed that TroSST can explain about 30%–40%of the autumn warming and 40%of the January warming simulated in His.This implies that TroSST could have been one of the main drivers of the Arctic winter tropospheric warming in 1979–2013.The impacts of TroSST generally came from a combination of,or the interactions among,the effects of I-WPSST,CEPSST,and TroAtlSST,except for the January warming below 850 hPa,which was dominated by the effects of I-WPSST.The effects of TroAtlSST were obtained as the residual trends based on the assumption that the effects of SST in the other regions could be summed linearly.Because of the importance of the interactions,which are possibly nonlinear,the effects of TroAtlSST presented in this paper may contain biases that need to be verified further using other methods.

Tropical SST anomalies bring tropical convection and associated precipitation and diabatic heating anomalies and drive the Rossby wave train propagating into the high latitudes(Chu et al.,2018).Meanwhile,I-WPSST tends to drive a North Atlantic Oscillation–like pattern,while CEPSST tends to drive a Pacific–North American pattern (Chu et al.,2018;Jia et al.,2009).TroAtlSST anomalies can also drive a wave train reaching Eurasia (Matsumura and Kosaka,2019).However,how the SSTs in all these tropical regions work together to drive the patterns shown in Fig.3 is not clear and should be explored further.

This work was based on one single model simulation,so the spread of the results among different models needs to be checked and compared with the result presented in this work.Furthermore,a large ensemble of simulations that can better eliminate the internal variability could supply a more solid verification of the results shown in this work.

Funding

This work was funded by the Research Council of Norway through the project COMBINED[grant number 328935].

Acknowledgments

The author acknowledges the contribution of Professor Yongqi Gao(1965–2021) to the design of the experiments.The CAM6-Nor simulations were performed on resources provided by UNINETT Sigma2—the National Infrastructure for High Performance Computing and Data Storage in Norway(nn2343k,NS9015K).