We use the Community Earth System Model (CAM5, version 1.1.2) simulations from PDRMIP (Myhre et al., 2017) (http://www.cicero.uio.no/en/PDRMIP/PDRMIP-data-access). The model is the same as Kay et al.(2015) but with a coarser resolution (Myhre et al., 2017); 2.5 × 1.9° in latitude and longitude with 30 vertical levels. CESM1 (CAM5) consists of coupled atmosphere, ocean, land, and sea ice component models. The model uses an emission database (Lamarque et al., 2010) for aerosols and calculates the life-cycle of aerosols along with their optical and physical properties. The past studies show that the model has shown good ability in simulating the mean climate state and variabilities (Myhre et al., 2018; Smith et al., 2018). The CAM5 simulations are chosen due its fair agreement to the magnitude and sign of the multi-model mean (MMM) for Arctic amplification calculation (Stjern et al., 2019).

We analyzed three sensitivity experiments for increased anthropogenic sulfate concentration 1) globally, and over 2) Asia, 3) Europe, and a control experiment (Myhre et al., 2017; Stjern et al., 2019). In the global experiment, anthropogenic sulfate concentration has been increased five times all over the globe (SUFL5). On the other hand, in another two regional sensitivity experiments, anthropogenic sulfate concentration (sulfate aerosols hereafter) has been increased by ten-times over Asia (10°–50°N, 60°–140°E) (SUFLASIA10) and Europe (35°–70°N, 10°W–40°E) (SULFEUROPE10). All the enhancements are instantaneous and a setup with fixed sea surface temperature (fixed-SST hereafter). Fixed-SST simulations were widely used to study the fast response (also known as a rapid adjustment) (on a time scale of a month to a season) of the climate system to the initial enhancement/perturbations (Hansen 2005; Bala et al., 2010; Ganguly et al., 2012; Myhre et al., 2017). All the experiments were carried out for 25 years with aerosol emissions of the year 2005. Out of 25 years of simulations, we use the only last 15 years of simulation output to investigate the rapid adjustment/fast response of Arctic temperatures to sulfate enhancement. A brief description of each of the experiments used is provided in Table 1. The spatial extent and global sulfate burden for control as well as emission enhancements are depicted in Supplementary Figure S1.

To study the fast response of the Arctic temperature to regional and global sulfate emission enhancements, we estimated surface temperature changes from EXPT-CTRL (surface temperature anomalies hereafter). Figure 1 shows surface temperature anomalies averaged over 60–80°N. A striking feature seen in Figure 1 is that sulfate aerosol has induced cold spring and summer (March-August). There is a slow transition in Arctic surface temperature; colder summer is followed by warmer autumn-winter (October-February) that passes through a cold phase at the end of summer month (September). The warmer winters are then followed by colder autumn-summers, passing through a warm phase at the end of the winter month (February). Asian emissions play a role in the cooling of the spring (March-April), while the dominant cooling effect due to European and global emissions is evident in summer. The warming effect is most pronounced for Asian and global emissions in winters, except in January where the global emissions produce higher warming than European emissions. A deep in cooling ∼ −0.8 K was seen in summer (July) due to global emissions, while a peak in warming (∼0.6 K) is seen in winter (January) due to Asian emissions. The European emissions also produce significant cooling (−0.6 K) in summer (July).

Figure 1 shows that the seasonal response varies in magnitude for the different emission scenarios, but the tendency of cooling (spring-summer) and warming (autumn-winter) is the same for global and regional emission enhancements. The reason for the different seasonal responses for sulfate emission enhancement; cooling in spring-summer and warming in autumn-winter seasons are investigated further in the following sections. The mechanism is the same for 1) cooling in spring and summer and 2) warming in autumn and winter. Hence, we show in the analysis below for the two seasons; summer (June, July, and August months) and winter (November, December, and January), where the emission enhancement response on surface temperature is strongest.

Figure 2 shows the spatial distribution of surface temperature anomalies during the summer and winter seasons for different emission enhancement experiments. During the summer season, the surface temperature anomalies are negative over a large part of the polar region irrespective of emission sources (Figures 2A–C). The magnitude of the anomalies varies spatially for each of the emission sources. It may be noted that large spatial cooling is observed over Greenland, part of the Russian Arctic and Siberian region for a change in global and European sulfate emission. However, global emission enhancement impacts are more pronounced over the Canadian Arctic, whereas it is well marked over part of Northern Europe and the Russian Arctic for European emissions changes. Interestingly, the surface temperature anomalies during the winter season highlight the dominant effect of Asian emissions over global and European emissions (Figure 2D–F). Unlike the summer season, the winter season shows a positive anomaly (warming) over the polar and adjacent regions.

The magnitude and spatial extent of warming in winter (over 60–80°N) are larger for regional sulfate emission enhancements than for the global emission enhancement. The spatial map of anthropogenic emissions enhancements for all the scenarios (global, Asia and Europe) (Supplementary Figure S2) shows that the anthropogenic sulfate emissions are confined in the northern hemisphere. Thus, their impact on Arctic regions is largely modulated by seasonal transport processes, forcing, and through feedback and response mechanisms.

Figure 3 depicts the latitudinal distribution of zonal mean surface temperature anomalies for the three sulfate enhancement experiments considered in this study. It shows gradual diversification of surface temperature anomalies, over the subtropics (∼20–40°N) that amplify in the polar region. In the subtropics, the surface temperature anomalies are negative irrespective of season and forcing. The seasonal effects are opposite north of 50°N, the magnitude of summer cooling is highest due to global and European emission (∼67°N), and the winter warming is dominant due to Asian and global emissions (∼65°N). Warming during the winter season shows a maximum of ∼78°N due to European emissions. The relative strength of cooling is comparatively less than that of warming for Asian emission enhancement. It is interesting to note that the surface temperature response to anthropogenic sulfate emission is clearly visible over sub-tropical to polar latitudes. However, the changes are maximum for the polar latitudes (60–80°N) irrespective of seasons. For this reason, we have chosen the above latitude range as a representation of the Arctic region and used it for calculating spatial averaging in this study.

To examine the vertical extent of seasonal warming and cooling, we have analysed the latitude height cross-sections of temperature anomalies (Figure 4).

The temperature changes are prominent beyond the north of 40°N (Figure 3); hence, we show vertical temperature anomalies between 40–90°N. Interestingly, the temperature response to emission enhancement extends from the surface to the lower stratosphere (150 hPa) at high latitudes. It may be noted that there exists a widespread anomalous cooling pattern during the summer season in all three experiments. During the summer, cooling is strongest in the lower troposphere (1,000–800 hPa) for the global and European emission experiments. Maximum cooling (>1 K) is observed north of 66° for the global and European experiment (Latitudinal Temperature Changes Due to Sulfate Forcing). The vertical extent of the magnitude of cooling is lowest for Asian emission enhancements compared to the global and European emissions.

The warming in winter, however, mostly observed over the high latitudes, are quite distinct in each experiment. The results for the different emission experiments differ in the maximum temperature anomaly and in the vertical extent of the signal. Unlike the other two experiments, the Asian emission experiment shows a high magnitude of warming extending to Upper Troposphere—Lower Stratosphere (UTLS ∼100 hPa) in the region 75–90°N. The strength of winter warming over high latitudes is moderate for global and European sulfate perturbations. The warming in the UTLS in the SULFASIA10 experiment is due to the transport of Asian sulfate aerosols to the Arctic through the UTLS (Shindell et al., 2008; Fadnavis et al., 2019). The Community Atmosphere Model, version 5 equipped with Explicit Aerosol Source Tagging (CAM5-EAST) simulations also show that increases in emissions from South and East Asia led to positive trends (1980–2018) in Arctic sulfate in the upper troposphere (Ren et al., 2020). On the other hand, the warming is mostly observed close to the Arctic surface (60–80ºN) for European emissions since a higher amount of sulfate aerosols are transported from Europe than for global/Asia to the Arctic through the lower troposphere in winter (Shindell et al., 2008). A source attribution study (Ren et al., 2020) also shows that surface concentrations of sulfate in the Arctic are mainly dominated due to emissions from Europe, Russia, and the local Arctic. On the other hand, South and East Asia sulfate emissions enhancement mainly causes an increase in the upper tropospheric sulfate concentration in the Arctic.

The magnitude and spread of emissions perturb energy balance by altering radiative forcing. As a response, the transport of moisture and energy from the tropics to the polar regions and clouds also change. The clouds are one of the important forcing agents regulating the Arctic climate (Bennartz et al., 2013; Willis et al., 2018). Cloud radiative forcing may be playing a role during summer and winter that has not been reported hitherto. In the next section, we deliberate the impact of sulfate emissions enhancement on Arctic clouds.

Clouds are critical to Arctic climate feedbacks. However, Arctic clouds and their radiative impacts are difficult to observe and evaluate in climate models (Kay and Gettelman 2009). The community atmospheric model (CAM) version 5 has a more realistic representation of cloud properties than the previous version. This version exhibits substantial improvement in three long-standing climate model cloud biases like the underestimation of total cloud, the overestimation of optically thick cloud, and the underestimation of midlevel cloud (Kay et al., 2012). The Arctic cloud fraction seasonal cycle and annual mean cloud fraction are found to be comparable with the CALIPSO observation (Kay et al., 2012). Hence, the selection of CAM5 to evaluate the response of Arctic clouds to aerosol changes may add more confidence to the findings presented in the subsequent section. However, there are other biases in the CESM-I, e.g., biases in the relative distributions of cloud cover, microphysics including phase, and the characteristics of the underlying surface. These biases are the same in the CTRL and sensitivity simulations that lenient the biases in the estimated change in cloud parameters.

The vertical distribution of anomalies in the cloud fraction/cover (Figure 5) shows a pronounced difference between the summer and winter seasons. The clouds considered are low clouds when formed between 1,000–600 hPa, mid clouds: 600–300 hPa, and high clouds: 300–50 hPa. The column average of cloud fraction/cover is depicted in Supplementary Figure S3. During summer, cloud fraction anomalies for all three experiments exhibit higher values of positive anomalies in the lower-mid troposphere (800–500 hPa) indicating higher amounts of low-mid clouds than high clouds over most of the Arctic region. The global and European emissions contribute more strongly than the Asian emissions towards the low cloud enhancement in the Arctic (also see Supplementary Figure S3). In all three experiments, during winter, cloud fraction anomalies are positive in the lower stratosphere region in the Arctic, indicating high clouds. The cloud anomalies are mostly negative in the lower and mid-troposphere (except near 80–85°N for Asian emissions), indicating low-mid cloud’s absence. Thus, during winter, European, Asian, and global emissions result in the formation of high clouds in the Arctic. The high cloud amounts are found to be higher in magnitude for Asian and global emissions than for European emissions. Vertically integrated cloud fractions show a maximum over parts of Greenland, Canadian Arctic, and the Russian Arctic regions (Supplementary Figure S3).

The enhancement on low/high clouds in summer/winter seen in Figure 5 is linked with the layer of the atmosphere through which transport of sulfate aerosols to the Arctic occurs. Higher amounts of low clouds in summer for European and global emission enhancements are associated with the transport of sulfate aerosols through the lower-mid troposphere (Shindell et al., 2008). The prevailing meteorological conditions in summer, i.e. increase in solar radiation, melting of ice, warm-air advection from adjacent open water or land over the melting ice, increase in cloud liquid water path and surface evaporation rate, upward heat fluxes, together further enhance the low-mid cloud formation (Sotiropoulou et al., 2016; Yu et al., 2019). High clouds in winters for Asian and global emission enhancement are associated with the transport of sulfate aerosols through the UTLS (Shindell et al., 2008; Ren et al., 2020). Sulfate aerosols are known to play an important role in the formation of high clouds (Aloyan et al., 2016). Previous studies have also reported high cloud formation in the lower stratosphere (between 50 and 200 hPa) in the Arctic both during the winter and summer seasons (Fadnavis et al., 2019; Cairo and Colavitto 2020). Arctic clouds in the stratosphere in winter (polar stratospheric clouds) occur because of low stratospheric temperatures (e.g., Grooß and Müller, 2021; Tritscher et al., 2021). During summer, the negative anomalies in the high cloud in the upper troposphere (300–150 hPa) may be associated with the transport of sulfate aerosols mostly occurring above the tropopause (Fadnavis et al., 2019). However, detailed analysis in this regard is warranted.

Clouds interact with radiation depending on their height and optical properties. In general, low clouds have a net cooling effect on the Earth’s surface and atmosphere due to their large reflection in the shortwave (overcompensating their absorption in the long-wave). On the other hand, high clouds which are much colder than the low clouds, warm the Earth’s surface and atmosphere by trapping and re-emitting the outgoing infrared radiation (Shupe and Intrieri, 2004). Hence, clouds and the associated radiative effects induced by sulfate emission enhancements will affect the Arctic temperature. Figures 4, 5 together show a strong association between seasonal change in temperature and low/high clouds. To quantify the effects of clouds on radiation and energy balance, we compute cloud radiative forcing, which largely depends on cloudiness. We have quantified the change in cloud radiative forcing for summer and winter corresponding to each forcing experiment in the subsequent sections.

Cloud radiative forcing (CRF hereafter) is defined as the difference between the net radiation flux with and without all/cloud and clear sky (Bennartz et al., 2013; Cox et al., 2016). CRF LW  = ( LW ↓ − LW ↑ ) all   sky − ( LW ↓ − LW ↑ ) clear   sky (1) CRF SW  = ( SW ↓ − SW ↑ ) all   sky − ( SW ↓ − SW ↑ ) clear   sky (2)

The LW and SW represent the shortwave and long-wave radiation flux, respectively. The up/down arrows correspond to the direction (outgoing/incoming) of radiation. The total CRF can hence be written as CRF = CRF SW + CRF LW (3)

The CRF can further be decomposed into Top of the atmosphere (TOACRF) and Surface/Bottom of the atmosphere (BOACRF). We have demonstrated the change at CRF for the top/bottom of the atmosphere for the summer and winter seasons to evaluate the response of emission enhancements to radiation through clouds. During the summer season, top of the atmosphere cloud radiative forcing anomalies (Figures 6A–C, top panel) show a widespread cooling for global and European emission enhancements. Most of the polar region exhibits a negative CRF (<−5 Wm−²) for these emission scenarios. On the other hand, for the Asian emission experiments, there is a mixed response observed for TOACRF. In this case, there is both cooling and warming, with the warming effect (positive CRF) mostly at higher latitudes. The ∼10–15% reduction in TOACRF (when averaged over the polar region) is seen in the summer season for European emission and global emission enhancements (also see Figure 7). The overall effect of Asian emissions to TOACRF is negative when averaged over the area 60–80°N (see Figure 7). Additionally, the CRF anomalies in BOACRF are negative irrespective of emission sources (Figure 6G–I). Though there is varying spatial distribution, the area-averaged values show a maximum reduction of radiative forcing for European emissions. During summer, low cloud enhancement (Impacts of Sulfate Forcing on Clouds) could be the reason behind these anomalous negative TOACRF and BOACRE values caused by the cloud induced higher reflection of shortwave insolation to space compared to the reflection in CTRL. During winter, TOACRF and BOACRF anomalies show positive radiative forcing values (i.e., long-wave absorption). Interestingly, the magnitude of the TOACRF is more distinct than that of BOACRF, which could be due to the enhancement of high clouds (Figure 5D–F). Please note that negative cloud radiative forcing during summer and positive during winter at Arctic sites are also reported (Dong et al., 2010; Ebell et al., 2020) in the past which is in agreement with our study. Long-term satellite observation revealed that in general, Arctic ocean clouds warm the surface and cool the TOA with a tendency of shortwave TOA cooling and longwave TOA warming (Kay and L’Ecuyer 2013). Recently, the negative cloud radiative effect during northern summer and warming during winter was also reported over the Arctic from the Clouds and the Earth’s Radiant Energy System (CERES) Energy Balanced and Filled (EBAF) observations by L’Ecuyer et al. (2019) in line with our findings.

It is important to note that there is no shortwave insolation in the Arctic region during winter, and thus the terrestrial radiation is the dominant component of the energy budget (Ebell et al., 2020). The positive CRF warming anomalies at the top and bottom are due to the LW radiation trap in the atmospheric column. Further, during winter, there is a significant increase in the high cloud fractions over the polar latitudes, which is seen in all emission experiments (Figure 5). The high clouds absorb the terrestrial radiation that would otherwise escape to space and re-radiate back, while the absorption and the re-radiation are not independent. This further enhances the warming in the atmosphere (Figure 4).

Here, we show linkages of the fast response of clouds and Arctic surface temperature due to regional and global enhancement of sulfate emissions. Figures 7A,B depicts the seasonal mean area-averaged values (60–80°N, zonal mean) for surface temperature anomalies and low cloud fractions. It shows that cooling in summer is associated with an enhancement in low cloud fractions and a reduction in high cloud fractions for all emission scenarios. In the summer season, the maximum enhancement in low cloud fraction (∼7%) has been found for global sulfate emission enhancements followed by enhancements of European emissions (>5%). The winter warming shows association with a reduction in low could fraction and enhancement in high cloud fraction for all emission scenarios (Figures 7B–C). During winter, the impact of Asian and European emission enhancements is higher than the global emissions in enhancing the high clouds.

During the summer season, an increase (decrease) in low (high) clouds is favourable to the solar dimming effect that reduces the downwelling radiation to the surface. Further, these conditions help to allow more terrestrial long-wave radiation to escape to space, which accelerates the surface and atmospheric cooling. On the other hand, during the winter season, the Arctic energy balance is dominated by long-wave radiation (due to the absence of sunlight). The presence of anomalous high clouds cover for all emission enhancements will result in absorbing and reradiating the outgoing long-wave radiation back to the surface.

To find the association of seasonal mean cloud change with radiation, we calculated the area-averaged top of the atmosphere cloud radiative forcing anomalies for summer and winter (Figure 7D). There is a negative cloud radiative forcing at the top of the atmosphere for the summer season. It is 10 times higher than clear sky (aerosols) radiative forcing at the top of the atmosphere for the summer season (Supplementary Figure S4). The negative radiative forcing is an indication of net cooling. Though our findings are in phase with the change in high and low clouds and associated changes in surface temperature, this generalization may not always be true in the Polar Regions. The warming or cooling caused by the low clouds depends on the cloud phase, water path, and the albedo of the underlying surface (Kravitz et al., 2014). A few studies reported that low clouds over bright surfaces induce warming even in the summer months (Bennartz et al., 2013; Van Tricht et al., 2016). However, a detailed analysis of the cloud phase, water path, and surface albedo is beyond the scope of the present study.

Additionally, there exists an anomalous positive CRF during winter; however, the net percentage change in TOACRF exceeds BOACRF in both seasons (Figure 6). This cloud-radiation change is consistent with the simulated surface temperature changes as a response to sulfate emission enhancements. Figure 7 also shows that the seasonal mean cloud fraction and CRF are more sensitive to European emissions and global emissions than Asian emissions in summer. In winter, Asian and global emissions show higher impacts on CRF than European emissions. This is consistent with summer (winter) surface temperature cooling (warming) showing a dominant impact by global and European (Asian and global) emission changes (also see Figure 1). Note that there is a slight increase in summertime high clouds for the Asian emission enhancement scenario (Figure 7C, blue). Additionally, the low cloud enhancement for the Asian emission scenario is comparatively smaller than the rest of the cases (Figure 7B, blue). This might be one of the reasons for the comparative smaller Arctic summer temperature response to the Asian emissions experiment. Cloud phase and, specifically, the presence of super-cooled water play a significant role in the net radiative effects of clouds at the surface. It is to be noted that there are substantial biases in liquid holding clouds over the Arctic in the CESM1 (Cesana et al., 2012) and the associated cloud microphysical processes (Tan and Storelvmo 2016) that might affect the downwelling longwave radiation and, in turn to the surface temperature (McIlhattan et al., 2017). In addition, it is worth mentioning that Arctic cloud characteristics and representation have substantially improved from CESM1 to CESM2 that has impacted their effects on the surface radiation budget (McIlhattan et al., 2020).