Introduction

As the world’s largest warm pool region, the tropical Indo-Pacific anchors the ascending branch of Walker circulation and local meridional cell. The convective activity in this region releases substantial latent heat, which plays a critical role in driving atmospheric teleconnection worldwide1,2 and thus acts as the lynchpin of the tropics and the extra-tropical climates3.

The tropical Indo-Pacific convective activity influences precipitation and temperature spatial patterns over East Asia by modulating local meridional circulations4. It also generates a Gill-type response on its northwest side, affecting climate anomalies in South Asia5. It not only influences nearby climate variability but also exerts remote impacts by inducing atmospheric teleconnections. On interannual timescales, the boreal winter convection over the tropical Pacific causes Pacific-North America and Pacific-South America teleconnections, influencing climate anomalies both in North and South America6,7 and even in Antarctic2,8. Similarly, the boreal summer convection over the Maritime Continent triggers the Pacific-Japan pattern, influencing climates in East Asia and the Pacific Rim9,10. Besides, intraseasonal convective activities over the Indian Ocean and western Pacific also significantly influence global weather and climate through inducing transient Rossby waves11,12,13,14,15,16,17,18. Therefore, the tropical Indo-Pacific convection is an important predictability source of extratropical climate variations. However, whether and how tropical Indo-Pacific convection affects the Arctic climate remains elusive. Understanding the Tropical-Arctic connection is critical for reliable predictions and future projections of Arctic climate.

The Canadian Arctic Archipelago (hereafter the Arctic Archipelago) is situated at the northernmost point of the North America Continent and consists of numerous islands and waterways19,20. The Northwest Passage in the region is one of the most important international trade routes, and the summer sea ice in the region is expected to largely decrease under global warming21,22. The Arctic Archipelago also plays a crucial role in freshwater transportation, serving as a primary conduit for freshwater flow from the Arctic Ocean to the North Atlantic Ocean23. Deficient freshwater flux in this region seriously affects freshwater exchange between the two oceans24,25, leading to the occurrence of large salinity anomalies26,27 in the North Atlantic. With rising temperature and melting sea ice in the Arctic Archipelago, the increased freshwater flux to the North Atlantic impacts the Atlantic Meridional Overturning Circulation and therefore influences global climate28. Additionally, the melting of sea ice in this region also releases micronutrients (iron, manganese) and transport macronutrients (nitrogen, silica, phosphorus) to the surface ocean, alternating the local ecological environment29. Therefore, advancing the understanding and prediction of climate variability and changes in the Arctic Archipelago are of enormous socioeconomic importance.

Various atmospheric circulation factors influence the Arctic Archipelago climate. For instance, sea-level pressure anomalies in the Arctic Archipelago can significantly affect sea surface height differences and through-flow transport30, while the Arctic Oscillation and El Niño-Southern Oscillation are also significantly related to the Arctic Archipelago sea ice extent31. The tropical convection could influence climate variability in the Arctic Archipelago through inducing the Pacific-Arctic teleconnection32. It is noticed that surface air temperature (SAT) over the Arctic Archipelago is an important indicator reflecting climate variability and climate change in the region. However, the drivers of the Arctic Archipelago SAT variability have not yet been fully elucidated33,34,35,36,37. A natural question is whether tropical Indo-Pacific convection in boreal summer steadily affects the Arctic Archipelago SAT through atmospheric teleconnections or such a Tropical-Arctic connection is modulated by background mean state changes. Motivated by this science question, in this paper, we intend to examine the temporal evolution of tropical convection impact in boreal summer on the Arctic Archipelago SAT and explore the underlying physical mechanism for the non-stationary impacts. The results can provide scientific reference for seasonal predictions and future projections of Arctic Archipelago climate.

Results

Observed changing influence of IPDC

Since a zonal dipole convection pattern dominates the tropics from the eastern Indian Ocean to Pacific (Fig. 1a), we define a tropical Indo-Pacific dipole convection (IPDC) index as the negative sign of the correlation coefficient weighted OLR (Supplementary Fig. 1) averaged over (15°S–15°N, 80°E–100°W) (Fig. 1a). This west-east dipole convection pattern can also be extracted by an Empirical Orthogonal Functions analysis (Supplementary Fig. 2) or based on the GPCP precipitation field (Supplementary Fig. 3). The year-to-year IPDC index exhibits a pronounced interannual variation (Fig. 1c). The regression of surface air temperature onto the IPDC index (Fig. 1b) suggests a weak relationship between SAT in Arctic Archipelago and IPDC. While the correlation coefficient between the Arctic Archipelago SAT index (Fig. 1c, defined as the area-mean SAT within the blue box in Fig. 1b) and the IPDC index is only 0.18 (p = 0.235) for the entire analyzed period of 1979–2021, the running correlation coefficients between them vary over the past four decades (Fig. 1d). Two regime shifts around 1992 and 2004 in the running correlation coefficients are identified (Supplementary Fig. 4). Therefore, the entire analysis period is divided into three periods: the first period of 1979–1991 (P1), the second period of 1992–2004 (P2), and the third period of 2005–2021 (P3). In P2, the IPDC and Arctic Archipelago SAT are highly correlated with a correlation coefficient of 0.634 (p < 0.05), while the correlation coefficients are 0.275 (p > 0.05) and −0.275 (p > 0.05) in P1 and P3, respectively (Fig. 1d). Based on the three different periods, we next examine the changing impacts of the IPDC on the Arctic Archipelago SAT and physical mechanism responsible for the changes.

Fig. 1: The decadal changes in relationship between tropical Indo-Pacific convection (IPDC) and Arctic Archipelago surface air temperature (AA_SAT).
Fig. 1: The decadal changes in relationship between tropical Indo-Pacific convection (IPDC) and Arctic Archipelago surface air temperature (AA_SAT).
Full size image

a Correlation coefficient between IPDC index and Outgoing Longwave Radiation (OLR) field. The slash region represents the correlation coefficient significant at 95% confidence level. The blue dashed box is (15°S–15°N, 80°E–100°W) for defining IPDC index. b Regressed surface air temperature (shading, units: °C) onto IPDC index. The dotted region represents the regression coefficient significant at 90% confidence level. The blue box is (72°–80°N, 80°–130°W) for defining AA_SAT index. c The year-to-year IPDC and AA_SAT indices. d The sliding correlation coefficient between IPDC and AA_SAT indices with a sliding windows of 11-yr. The correlation coefficient significant at 90% confidence level are with blue dots. The horizontal dashed line in d represents the threshold of correlation coefficient passing the 90% confidence level considering effective degrees of freedom.

We first examine the upper- and lower-tropospheric circulation response to the IPDC forcing (Fig. 2). A clear northeastward Rossby wave train appeared, with alternated cyclonic-anticyclonic-cyclonic-anticyclonic centers. This wave train pattern is in accordance with the great circle theory proposed by Hoskins et al. 38, who suggested that tropical forcing-induced Rossby wave activity propagates northward and eastward, forming a great circle route. The strength and northward extent of this Rossby wave train depends mainly on the meridional gradient of background potential vorticity (PV) and zonal wind. As seen from Fig. 2d, a well-organized wave train appeared in P2, and it extended all the way to the Arctic Archipelago, where a strong and significant anomalous anticyclone was located. The Rossby wave train exhibits an equivalent barotropic structure (Fig. 2d, e). The occurrence of this strong anomalous anticyclone (high-pressure) was crucial for inducing pronounced positive SAT anomalies in situ (Fig. 2f). However, the Rossby wave train was weak and less organized in P1, so that the Northern Hemisphere circulation anomalies associated with the IPDC were primarily featured by anomalous anticyclones in the North Pacific and south of the Arctic Archipelago, and a cyclone from the northeast Pacific to western Canada (Fig. 2a, b). Influenced by the weak anomalous anticyclone south of the Arctic Archipelago, the SAT anomalies were positive only to the south of the Arctic Archipelago (Fig. 2c). During P3, the Rossby wave train stemming from IPDC appeared weak and exhibited an obvious southward-shift compared with P2 (Fig. 2g and h). Although similar anomalous anticyclone was also observed over the North Pacific, the anomalous cyclone to the north was confined to the southern Aleutian Islands, and the anomalous anticyclone downstream was located in the Alaskan region. The Arctic Archipelago was largely influenced by cyclone anomalies over the Arctic, leading to a weak negative SAT anomaly in the Arctic Archipelago (Fig. 2i). The Rossby wave train features above are consistent with those based on different reanalysis datasets (Supplementary Fig. 5).

Fig. 2: The atmospheric circulation anomalies related to tropical Indo-Pacific convection (IPDC) during different periods.
Fig. 2: The atmospheric circulation anomalies related to tropical Indo-Pacific convection (IPDC) during different periods.
Full size image

Regressed a 250hPa geopotential height (shading, units: gpm) and wind (vectors, units: m s-1), b 850hPa geopotential height (shading, units: gpm) and wind (vectors, units: m s-1), and c surface air temperature (SAT) (shading, units: °C) onto IPDC index during P1. The dotted region represents the regression coefficient significant at 90% confidence level. The blue box in c indicates the area defining Arctic Archipelago SAT index. The letters A/C in blue/red represent the center of anomalous barotropic anticyclone/cyclone, while letters A/C in magenta denotes the center of anomalous anticyclone/cyclone as the Gill-type response to the tropical convection. d, e, f and g, h, i are same as in a, b, c, but for P2 and P3, respectively.

Comparing with the circulation characteristics in other two periods, it is evident that IPDC had a closer linkage with the Arctic Archipelago SAT in P2 (Fig. 1d). Although the IPDC could induce upper-level anomalous cyclone over tropical North Pacific, serving as the Rossby wave sources during all the three periods (Supplementary Figs. 6 and 7), only in P2, the IPDC-triggered Rossby wave train propagated to the northernmost latitude (around 80°N). Conversely, in P1 and P3, the Rossby wave train stemming from IPDC was confined to south of 70°N, leading to the southward shift of the anomalous anticyclone, and consequently, a diminished impact on the Arctic Archipelago SAT. A further question is raised as to why the IPDC could induce a teleconnection signal into the Arctic Archipelago and led to the significant positive Arctic Archipelago SAT anomalies during P2? In the following section we intend to reveal the underlying physical mechanism.

Possible cause of the changing teleconnection

Since the upper-level westerly jet and associated meridional PV gradient play the vital role in shaping the Rossby wave train39,40,41, the mean state differences of upper-level zonal wind in the three periods are investigated. Figure 3a and b illustrate a significant weakening of the westerly jet over East Asia during P2 in comparison to P1 and P3. Along with the weakened westerly is the enhanced meridional PV gradient in the North Pacific (purple box in Fig. 3c). While the climatological Rossby waveguides in boreal summer are well separated along the subtropical jet stream and the polar front jet (Fig. 3d), the enhanced meridional PV gradient over the North Pacific fills the gap between the subtropical jet and polar front jet (Fig. 3c and d) and facilitates a more poleward-propagating Rossby wave and thus the Tropical-Arctic teleconnection in P2 (Fig. 2d).

Fig. 3: The composite differences of thermal and thermal-dynamical fields affecting the teleconnection toward Arctic Archipelago.
Fig. 3: The composite differences of thermal and thermal-dynamical fields affecting the teleconnection toward Arctic Archipelago.
Full size image

The composite difference of 250hPa zonal wind (shading, units: m s-1) between a P2 and P1 (P2 minus P1), and between b P2 and P3 (P2 minus P3) and the climatological 250hPa wind speed (contours) of 20 m s-1 and 25 m s-1. c Regressed 250hPa meridional potential vorticity gradient (shading, units: 10-12 K m2 kg-1 s-1) onto East Asian westerly jet strength index (with reversed sign), and d The climatological 250hPa meridional potential vorticity gradient (shading, units: 10-12 K m2 kg-1 s-1). e, f are same as a, b but for the composite difference of sea surface temperature. The green box in a, b indicates the area defining East Asian westerly jet strength index (EAJ; 35°N–42°N, 80°E–170°E), the purple box (45°N–65°N, 100°E–140°W) in c shows the key region of meridional potential vorticity gradient increased by weaker East Asia westerly jet, while the red box in e, f represents the region defining Russian Arctic sea surface temperature index to the north of Eurasian Continent (70°N–83°N, 30°E–170°E). The regression/composition coefficient significant at 90% confidence level are dotted.

To test the hypothesis, we compared the atmospheric teleconnections associated with the IPDC between the years with strong and weak East Asian westerly jet. Under a weak East Asian jet condition, the teleconnection induced by IPDC extends further north (Supplementary Fig. 8a, b). A barotropic high-pressure (anticyclone) anomaly appears over the Arctic Archipelago, which leads to local positive SAT anomalies (Supplementary Fig. 8c). In contrast, under a stronger East Asian jet condition, the pathway of the IPDC-induced teleconnection shifts southward, and a high-pressure (anticyclone) anomaly is confined to the south of the Arctic Archipelago, leading to insignificant impact on SAT in the Arctic Archipelago (Supplementary Fig. 8d, f).

What causes the changes in the strength of the East Asian westerly jet? We hypothesize that the negative Russian Arctic SST anomalies could induce local low-pressure (cyclonic) anomalies in northern Eurasia, further triggering a southeastward-propagating Rossby wave train42,43,44 (Supplementary Fig. 9), resulting in two high-pressure (anticyclonic) anomalies to the north of 40°N over East Asia and the North Pacific. Indeed, Russian Arctic SST shows significant negative anomalies in P2 compared with the other two periods (Fig. 3e, f), corresponding to the weakened westerly jet over East Asia (Fig. 3a, b). The southern flank of the anticyclonic anomalies ultimately weakened the East Asian westerly jet (Fig. 3a, b and Supplementary Fig. 9). In view of the close link between the change of the westerly jet intensity and the IPDC-induced teleconnection, here we propose that the changes in the East Asian westerly jet intensity driven by the Russian Arctic SST anomalies is the primary contributor to the alternate tropical teleconnections toward the Arctic Archipelago.

Numerical evidences for the impact of Russian Arctic SST on teleconnections

Consistent with the observed relationship (Fig. 4a), the difference between the sensitivity experiment with additional 2 K negative Russian Arctic SST anomalies (Exp_Arctic-2K) and the control experiment (Exp_Ctrl) reveals that negative Russian Arctic SST anomalies lead to a weakening of the East Asian westerly jet (Fig. 4b). In Exp_Ctrl, when the Russian Arctic SST is relatively warm, the IPDC-induced wave train propagates more zonal and less northward, leading to an anomalous anticyclone south of Arctic Archipelago (Fig. 4c). In Exp_Arctic-2K, when Russian Arctic SST is much colder and the East Asian westerly jet becomes weaker, the IPDC-triggered Rossby wave train propagates more northward and leads to a circulation anomaly right on the Arctic Archipelago (Fig. 4d). The simulated wave train patterns and anomalous circulation centers resemble well those observed during P2 (Fig. 2d). Therefore, comparing the two experiments, when the Russian Arctic SST is colder (warmer) and the East Asian westerly jet is weaker (stronger), the center of the barotropic anomalous anticyclone dominates (locates to the south of) the Arctic Archipelago, resulting in positive (insignificant) Arctic Archipelago SAT anomalies (Fig. 4e-f). This conclusion is further supported by additional model experiments with a realistic Russian Arctic SSTA pattern but different amplitudes (see Supplementary Figs. 10 and 11). Furthermore, the impact of the colder Russian Arctic SSTA on the Tropical-Arctic teleconnection can be further demonstrated in a linear baroclinic model (see Supplementary Figs. 9b, 12 and 13).

Fig. 4: The numerical simulations demonstrating the underlying mechanism of the alternate teleconnection toward Arctic Archipelago.
Fig. 4: The numerical simulations demonstrating the underlying mechanism of the alternate teleconnection toward Arctic Archipelago.
Full size image

a Regressed 250hPa zonal wind (shading, units: m s-1) onto Russian Arctic sea surface temperature (SST) index (to show the influence of the negative Russian Arctic SST anomalies, the regression field is multiplied by -1). b The differences of 250hPa zonal wind (shading, units: m s-1) using outputs of Exp_Arctic-2K minus Exp_Ctrl. c Regressed 250hPa geopotential height (shading, units: gpm) and wind (vectors, units: m s-1) onto the tropical Indo-Pacific dipole convection (IPDC) index in Exp_Ctrl, d same as c but for that in Exp_Arctic-2K. e, f same as c, d but for the regressed surface air temperature (shading, units: °C). The regression/composition significant at 90% confidence level are dotted. The black contours in a and b denotes the wind speed of 20 m s-1, 25 m s-1 in observation and Exp_Ctrl, respectively. The letter A/C represents the center of the anomalous anticyclone/cyclone.

In summary, a number of numerical simulations confirm the hypothesis that Russian Arctic SSTA are the moderator of the IPDC-induced teleconnections toward the Arctic Archipelago. Colder Russian Arctic SST leads to a weakened East Asian westerly jet, which acts as a switch allowing less zonal but more northward-propagating Rossby wave train to affect SAT in Arctic Archipelago.

Discussion

The present study unveils the notable changing impact of the tropical Indo-Pacific dipole convection on SAT in the Canadian Arctic Archipelago through distinct poleward teleconnections in boreal summers during 1979–2021. Based on observational diagnoses and numerical simulations, it is evident that the Tropical-Arctic connection is controlled by the westerly jet intensity over East Asia. When the East Asian westerly jet is relatively weak, the tropical convection-induced Rossby wave train propagates more poleward, leading to a significant influence on SAT in the Arctic Archipelago. Conversely, when the westerly jet is relatively strong, the Rossby wave train propagates more zonal, so that the tropical influence on the Arctic SAT diminishes.

It is interesting to notice that the major cause of the change in the westerly jet strength is attributable to the Russian Arctic SST anomalies (Fig. 5). A negative (positive) Russian Arctic SST anomaly triggers a southeastward-propagating Rossby wave train and thus an anomalous anticyclone (cyclone) north of the westerly jet, leading to a weakening (strengthening) of the jet. The weakened (strengthened) westerly jet enhances (reduces) the meridional PV gradient over the North Pacific, thereby favoring (impeding) the northward propagation of tropical convection-induced Rossby wave train (Fig. 5). Consequently, an anticyclonic circulation anomaly dominates (is away from) the Arctic Archipelago, leading to a significant (negligible) impact on the Arctic Archipelago SAT. Therefore, the Russian Arctic SST acts as a moderator controlling the tropical impact on the SAT over the Arctic Archipelago.

Fig. 5: Diagram of Tropic-Arctic connection switched by the Russian Arctic Sea Surface Temperature (SST) in P2.
Fig. 5: Diagram of Tropic-Arctic connection switched by the Russian Arctic Sea Surface Temperature (SST) in P2.
Full size image

The shadings in tropic indicates the Indo-Pacific dipole convection. The red shading represents the positive surface air temperature anomalies over the Canadian Arctic Archipelago. The blue shading indicates the negative Russian Arctic SST anomalies. The blue slashed shading over North Pacific denotes the positive anomaly of meridional potential vorticity gradient. The letter A/C represents the center of the anomalous anticyclone/cyclone. Dots in the tropical North Pacific denote the Rossby wave source induced by Indo-Pacific dipole convection.

One may suspect that the change of the tropical mean state may also play a role. To test this hypothesis, we examined the background state difference of the tropical heating and SST between P2 and other two periods (Supplementary Fig. 14). It turns out that there are no significant SST and convection changes between these periods over most of tropics, except the enhanced convection over tropical Indian Ocean and negative SST anomalies over the southern Indian Ocean where has little role in affecting the westerly jet and the poleward Rossby wave pattern comparing with that of the Russian Arctic SSTA.

Unlike most of the previous studies, which underpinned a one-way impact either from Tropics to Arctic45,46,47 or from Arctic to Tropics48,49,50,51, the present study proposes a two-way interaction between Tropics and Arctic. Specifically, the Arctic thermal condition changes the background mean states, thus influencing the atmospheric teleconnections linking the Tropics and the Arctic Archipelago. Although the non-stationarity of the Tropical-Arctic teleconnection poses a challenging for seasonal predictions of the Arctic Archipelago SAT using tropical forcing as a predictor, this study sheds light on the causes of the changing Tropical-Arctic connection, thereby providing clues for improving the seasonal predictions of Arctic climate.

Since our conclusion is derived from the datasets in a relatively short period of the satellite era, the proposed mechanism for the changing connection between the tropics and the Arctic needs more upcoming observational evidences. Furthermore, in the context of global warming, how will the Arctic Archipelago SAT change and whether the proposed mechanism will continue to be at play remain open questions.

Methods

Observation and reanalysis datasets

The dataset used in this study includes: 1) Monthly interpolated outgoing longwave radiation (OLR) provided by the National Oceanic and Atmospheric Administration (NOAA)52, with a horizontal resolution of 2.5°×2.5°; 2) Monthly precipitation data from the Global Precipitation Climatology Project version 2.353, with a horizontal resolution of 2.5°×2.5°; 3) Monthly atmospheric reanalysis data from the fifth generation European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis54, including horizontal wind, geopotential height, potential vorticity, etc., with a horizontal resolution of 1°×1°; 4) Monthly atmospheric reanalysis data from the Japanese Meteorological Agency (JMA) JRA-55 dataset55, with a horizontal resolution of 1.25°×1.25°; 5) Monthly sea surface temperature and sea ice data from the Met Office Hadley Centre56, with a horizontal resolution of 1°×1°; 6) Monthly sea surface temperature data from NOAA ERSST version 557, with a horizontal resolution of 2°×2°. The present study covers the period from 1979 to 2021, and boreal summer denotes the average mean of June, July, August and September (JJAS).

Diagnostics

The methods employed in this study include correlation and regression analyses. The significance is determined using a standard two-tailed Student’s t-test. The effective degrees of freedom (Nedof) are considered when calculate running correlation. The Nedof are calculated as follows58:

$${{N}}_{{edof}}={N}(1-{{r}}_{1}{{r}}_{2})/(1+{{r}}_{1}{{r}}_{2})$$
(1)

where N indicates the sample number, r1 and r2 represent the lag -1 autocorrelations of two series.

To analyse the propagation of Rossby wave, the wave activity flux is calculated based on the following formula59:

$${{\bf{W}}}=\frac{1}{2\left|\bar{{{\boldsymbol{U}}}}\right|}\left[\begin{array}{c}\bar{u}\left({{\psi }^{{\prime} }}_{x}^{2}-{{\psi }^{{\prime} }{\psi }^{{\prime} }}_{{xx}}\right)+\bar{v}\left({{\psi }^{{\prime} }}_{x}{{\psi }^{{\prime} }}_{y}-{{\psi }^{{\prime} }{\psi }^{{\prime} }}_{{xy}}\right)\\ \bar{u}({\psi ^{\prime} }_{x}{\psi ^{\prime} }_{y}-{\psi ^{\prime} \psi ^{\prime} }_{{xy}})+\bar{v}({{\psi }^{{\prime} }}_{y}^{2}-{\psi ^{\prime} \psi ^{\prime} }_{{yy}})\end{array}\right]$$
(2)

where an overbar and a prime represent the climatological mean and anomaly, respectively; ψ and \(\bar{{{\boldsymbol{U}}}}\) = (\(\bar{u}\), \(\bar{v}\)) represent the stream function and the horizontal wind, respectively; and W denotes the two-dimensional Rossby WAF.

Model experiments

To explore the impact of Arctic SST on the changing Rossby wave train, the atmospheric general circulation model ECHAM version 5.3 developed by the Max Planck Institute for Meteorology60 is utilized. ECHAM v5.3 was run in the Atmospheric Model Intercomparison Project (AMIP) configuration, which includes interactive atmosphere and land-surface components with prescribed sea surface temperature (SST) and sea-ice. The horizontal resolution of T42 (approximately 2.88°×2.88°) and 31 vertical layers was employed in this study. We conducted two sets of experiments using ECHAM v5.3 to examine the climate impact of Russian Arctic negative SST anomalies on modulation of the poleward propagation of tropical convection-induced Rossby wave trains.

In the first set, the control experiment (Exp_Ctrl) is run by monthly SST and sea ice data derived from HadISST dataset54 from 1981 to 2010 with different SST and initial conditions. In the sensitivity experiment (Exp_Arctic-2K), an idealized negative 2 K SST anomaly is added over Russian Arctic Ocean (70°–83°N, 30°–170°E) during boreal summer (JJAS), while SST elsewhere is same as that in Exp_Ctrl.

In the second set, we conducted experiments with a realistic Russian Arctic SST anomaly distribution and with hierarchical forcing (i.e., different SSTA amplitudes). A realistic SST anomaly pattern over Russian Arctic Ocean (70°–83°N, 30°–170°E), obtained by averaging the observed composite SST differences between P2 and P1 (P2 minus P1) and between P2 and P3 (P2 minus P3) (Supplementary Fig. 10), is prescribed in the model during boreal summer (JJAS). The SST anomalies pattern is prescribed with different multiples from -3.0 to +3.0 with a 0.4 interval (i.e., -3.0, -2.6, -2.2, -1.8, -1.4, -1.0, -0.6, -0.2, 0.2, 0.6, 1.0, 1.4, 1.8, 2.2, 2.6, 3.0 times of the observed Russian Arctic SSTA), while SST elsewhere is same as that in Exp_Ctrl. Therefore, there are 16 sensitivity experiments in total in this set of experiments.

Additionally, a linear baroclinic model (LBM) is employed to simulate the associated physical processes. The LBM is constructed based on a dynamical core of AGCM cooperatively developed at the Center for Climate System Research (CCSR), University of Tokyo, and National Institute for Environmental Studies (NIES), namely the CCSR/NIES AGCM version 5.4g61. The LBM removed the nonlinearity in the dynamical atmosphere to understand the complex feedback in the numerical model better. The dry version model with a T42 horizontal resolution and 20 vertical sigma levels. The damping coefficient is set at 0.5 per day for the lowest three sigma level (σ = 1.00, 0.98 and 0.95), at 1 per day for the upper two sigma level (σ = 0.02 and 0.01), and at 20 per day for other levels. The e-folding decay time of 0.2 day for the largest wave number. The observed climatological summer mean (JJAS) is prescribed as the model basic state, which was taken from the long-term mean of 1981–2010.