Concurrence of blowing snow and polynya enhances arctic surface–atmosphere interaction: a modeling study with an extreme wind event in 2018

Snow, a critical element influencing surface energy/mass balance of the Arctic, can also drift in the air to complicate the surface–atmosphere interaction. This complexity can be further enhanced when the surface includes polynya. These processes, however, have not been well studied and are often unrepresented in climate and weather models. We address this by applying a snow/ice-enhanced version of the Weather Research and Forecasting model to examine the impacts of blowing snow and polynya on the surface–atmosphere interaction during an extreme Arctic wind event in February 2018, when an unprecedented polynya occurred off the north coast of Greenland. The results indicate that blowing snow and the polynya contribute opposite signs to the changes of surface sensible/latent heat fluxes, but both cause enhanced downwelling longwave radiation. Process analysis shows that the thermodynamic moistening/cooling effects due to the blowing snow sublimation are amplified by increased surface winds, reduced temperature inversion, and upward wind anomaly associated with the polynya. Enhanced surface–atmosphere interaction over a polynya due to blowing snow sublimation can potentially sustain the continuing development of the polynya.


Introduction
The Arctic climate system has experienced substantial changes in recent decades, including a faster warming than anywhere else of surface/lower tropospheric air and ocean water temperatures (Serreze and Barry 2014, Walsh 2014, Huang et al 2017, Cohen et al 2020, Druckenmiller et al 2021, Rantanen et al 2022 and a drastic retreat of sea ice coverage along with dramatic thinning of sea ice thickness (Comiso and Hall 2014, Meier et al 2014, Overland et al 2014, Mallett et al 2021. In conjunction with these changes, multiscale atmospheric circulation systems have also exhibited various changes, such as a large temporal fluctuation and a radical spatial shift of the surface pressure pattern (Thompson and Wallace 1998, Zhang et al 2008, Overland et al 2012 as well as poleward-shifted extratropical cyclone tracks and intensified Arctic cyclone activity (Zhang et al 2004, Simmonds et al 2008, Day et al 2018, Vessey et al 2020, Valkonen et al 2021.
Changes in the atmospheric circulation systems can cause increases in air temperature, wind, and precipitation, which may enhance the feedback processes and, in turn, amplify Arctic changes. For example, intense cyclones have more frequently occurred during recent decades, including two remarkable Arctic cyclones in August 2012 and 2016, respectively, drastically decreasing sea ice cover and contributing to the minimum summer sea ice extent on record (Simmonds and Rudeva 2012, Zhang et al 2013, Peng et al 2021. During late February and early March 2018, the concurrence of an intense cyclone centered in the north of the Canadian Arctic Archipelago with a strong anticyclone over the northeast/east of Iceland resulted in strong southerlies greater than 25 m s −1 at the surface across the north coast of Greenland, causing an unprecedented polynya event (Moore et al 2018, Ludwig et al 2019. In addition to the impacts of atmospheric circulation on extreme sea ice and temperature events, strong winds can also induce and magnify another important physical process-blowing snow, an uplift and horizontal transport of surface snow by winds. On the one hand blowing snow affects surface mass balance and hydrological processes via the redistribution of surface snow accumulation (Gallée et al 2001, Das et al 2013, Grazioli et al 2017; on the other hand blowing snow also impacts atmosphere thermal structure and stability, as well as cloud properties and the associated shortwave and longwave radiation fluxes (Déry and Yau 2001, Lenaerts and van den Broeke 2012, Barral et al 2014, Yang et al 2014, Hofer et al 2021, Luo et al 2021, Luo and Zhang 2022. For instance, blowing snow sublimation can increase the specific humidity of lower atmosphere, weaken the temperature inversion, and therefore affect the vertical transport of heat and moisture (Luo et al 2021, Luo andZhang 2022). However, these processes have not been well investigated in the Arctic sea ice environment. Notably, most reanalyses and global climate models do not have the blowing snow included as a process. Divergent ice motion driven by strong winds can lead to the development of polynya even during deep winter over climatologically thick ice area. An example is the striking Greenland polynya event in February 2018 (Moore et al 2018), and many coastal polynyas in Antarctic (Massom et al 1998, Tamura et al 2008. A polynya allows the air and sea interact directly without the presence of insulating sea ice. This interaction is of great importance physically, biogeochemically and biologically (Stirling 1997, Arrigo et al 1999, Arrigo 2003, Maqueda et al 2004. Moreover, clouds, fog, and convective plumes frequently form above and downwind of the polynya, resulting in strong radiative effects (Andreas and Cash 1999, Dare and Atkinson 1999, Moore et al 2002, Weijer et al 2017. In addition, large heat loss to the air can also cause deep ocean convection beneath the polynya (Cheon et al 2014). The presence of a polynya complicates the interaction of air-ice-sea processes. A new research question, therefore, naturally arises: how does the concurrence of blowing snow and a polynya impact the Arctic surface-air interaction that may contribute to Arctic climate change? To address this problem, we conducted a modeling study on an Arctic blowing-snow/polynya event to improve process-level understanding with a snow-ice enhanced Weather Research and Forecasting (WRF) modeling system (WRF-ice; Zhang et al 2013, Luo et al 2021, Luo and Zhang 2022).

WRF-ice model and modeling experiment design
The WRF model (Skamarock et al 2008) is the backbone of WRF-ice, in which improved model physics for sea ice, snow, and ice sheets have been implemented. Specifically WRF-ice couples with the thermodynamic sea ice module of Zhang and Zhang (2001), the two-moment blowing snow formulation of Déry and Yau (2001), and the snow-firn physics for an ice sheet/shelf (Yao et al 2016, Luo andZhang 2022). These extra modules, particularly the blowing snow physics within WRF-ice, enable modeling studies of how blowing snow and the polynya impact the surface-air interaction during the extreme Arctic wind event in February 2018.
A total of four simulation experiments have been set up as shown in table 1. Among these experiments, experiment PB serves to collectively combine the impacts of polynya and blowing snow; experiment IB models the blowing snow impact individually over a fully ice-covered Arctic; experiment P serves to examine the individual impact from the polynya; and experiment I eliminates altogether the polynya and blowing snow impacts.
The model configuration for the simulation experiment PB has been summarized in table 2. The three-hourly fifth ECMWF (European Centre for Medium-range Weather Forecasts) reanalysis (ERA5, Hersbach et al 2020) is used to provide the initial, lateral, and lower boundary conditions for the simulations. Throughout the WRF simulations, sea ice concentration and sea surface temperature are updated daily from the ERA5 reanalysis. Two one-way nested domains at 20 and 5 km horizontal resolution are configured. While a nesting ratio of 4 is used here, very similar results were obtained with the ratio of 5. Vertically there are 49 levels with the top at 10 hPa and 14 levels located below 1 km. The snow/ice modules mentioned above are implemented within the NOAH land surface scheme (Chen and Dudhia 2001) for capturing the thermodynamic processes of blowing snow, sea ice, and an ice sheet. The cooling and moistening effects of blowing snow sublimation are included in the 2.5-order Mellor-Yamada-Nakanishi-Niino (MYNN-2.5) boundary layer parameterization scheme (Nakanishi and Niino 2006). Twenty-four layers from the surface to 1 km above ground are set up within the blowing snow module. Radiation processes are parameterized with the Rapid longwave and shortwave Radiative Transfer Model (RRTMG) (Iacono et al 2008). The Morrison two-moment scheme (Morrison and Gettelman 2008) is used for the microphysics processes. The cumulus parameterization used in the 20 km resolution domain is the Grell-3D ensemble scheme (Grell and Devenyi   Grell-3D 2002) and no cumulus scheme is used in the 5 km domain. Note that no nudging is used in the simulation experiments and the surface roughness is defined from the land-use table of WRF with 1.0 mm over sea ice and 0.1 mm over ocean but modified by the ocean surface wind. The nested modeling domain with the initial sea ice coverage and wind field is shown in figure S1 of the supplementary figures. The outer domain (Domain-1) with a resolution of 20 km covers the central and Atlantic side of the Arctic Ocean, and the inner domain (Domain-2) with 5 km resolution covers the northeastern coast of Greenland, in which a significant wind-driven coastal polynya occurred. Modeling results for the entire simulation period from Domain-2 are analyzed to explore how blowing snow and the polynya interact in the following section.

Evaluation of WRF-ice simulation with station observations
Modeled surface wind speeds from Domain-2 of the PB experiment are first evaluated with the collected station observations as shown in figure S2 of the supplementary figures. There are two stations within Domain-2 but one of them (Nord, GL) does not have hourly data. The temporal evolution of observed wind speed is well captured by the PB experiment, with a correlation of 0.79 with the station observations. Magnitudes of extreme wind speed also match reasonably well, except that the relatively calm winds around 11-18 UTC February 24 are not captured by the model. During the final 12 h, modeled wind speeds are too weak. The overall bias and root-mean-squared error are 0.28 m s −1 and 3.71 m s −1 , respectively. The wind field plays a critical role in the onset of blowing snow, so the agreement between the modeled and observed wind speeds of both magnitude and timing gives us confidence for further analysis of how blowing snow and the polynya interact during this extreme wind event.
In addition, considering that WRF tends to have relatively larger bias in the longwave radiation (LW) over the polar region (Hines et al 2019(Hines et al , 2021, we also compared the PB modeled downwelling and upwelling LW with station observation from the World Radiation Monitoring Center-Baseline Surface Radiation Network (BSRN). For the study period, the BSRN network has only one station Ny-Ålesund (Maturilli 2018) within the Domain-1. The comparison of modeled and observed surface downwelling and upwelling LW, as shown in figure S3 of the supplementary figures, indicates that the model basically captures the observed LW although the timing of a drop in the downwelling LW around the end of the third day simulation is earlier than the observed one.

Individual and collective impacts of blowing snow and polynya on the surface-air interaction
The impacts of blowing snow and the polynya on the surface-air interactions and associated energy fluxes are examined by comparing the Domain-2 simulation results among three paired sensitivity experiments: P and I, IB and I, and PB and IB (figure 1). Figures 1(a)-(d) show the differences of sea ice concentration, surface sensible (SH) and latent (LH) heat fluxes (both SH and LH are positive upward), and downwelling longwave radiation (LW) at the surface between the experiments P and I, in which sea ice concentration is defined using the ERA5 reanalysis and as a uniform value of 100%, respectively. The differences are averaged over the entire simulation period, allowing us to explore the overall impacts of the polynya on the surface-air interaction. Sea ice concentration shows a reduction up to 60% off the northeastern coast of Greenland during the polynya event from February 22 to 26. Over the study period, sensible heat flux is downward in general due to the temperature inversion. The competing effects between increased wind and weakened temperature inversion associated with the occurrence of the polynya cause spatially heterogenous changes in surface sensible heat flux. Over areas with significant sea ice reduction, such as the polynya center, the effect of reduced temperature inversion offsets or outweighs the effect of increased wind, resulting in little changed or reduced sensible heat flux results. Over areas with less sea ice reduction, the effect of increased wind outweighs the effect of reduced inversion, resulting in enhanced downward sensible heat flux. Therefore the magnitude of downward sensible heat flux decreases (increases) over the polynya area with relatively larger (smaller) sea ice reduction. Open water exposure during the polynya always helps increase upward latent heat flux. A slight increase in downwelling LW is also observed over the downstream area of the polynya, which could be a result of an increase in atmospheric water vapor due to the enhanced latent heat flux. Overall, the domain-averaged changes in sensible and latent heat fluxes and downwelling LW caused by the polynya are −1.88 w m −2 , 0.65 w m −2 , and 0.33 w m −2 , respectively.
Comparison between the experiments IB and I provides insight into how blowing snow affects the surface-air interaction in a full sea ice covered condition. Averaged blowing snow sublimation from IB and the averaged differences of surface sensible and latent heat fluxes and downwelling LW between IB and I are examined with figures 1(e)-(h). Note that sensible heat flux is basically downward (negative) and latent heat flux is upward (positive) in the study area. Therefore we find that blowing snow reduces the magnitudes of both downward sensible and upward latent heat fluxes, which is basically opposite to the effects of the polynya. However, consistent with the polynya's impact, blowing snow also enhances downwelling LW at the surface. The reduction in downward sensible heat flux is attributed to the cooling effect of blowing snow sublimation, which cools the air above surface and, therefore, weakens the temperature inversion. The change in latent heat flux exactly reflects the moistening effect of blowing snow sublimation. Note that, although the maximum blowing snow sublimation mainly occurs over the Greenland ice sheet, particularly the northern part of Greenland plateau, significant changes in surface heat fluxes appear over the sea ice off the northern coast of Greenland. This difference can be ascribed to the significantly different surface winds. The surface wind is stronger over the sea ice than that over the icesheet during the study period. When the cooling and moistening effects of blowing snow sublimation work together, relative humidity increases efficiently. This increase can enhance cloud formation particularly over the relatively warm and humid area, such as the lower right corner of the study domain. As a consequence, downwelling LW increases. The domain-averaged sensible and latent heat fluxes and downwelling LW change by about 4.20 w m −2 , −3.85 w m −2 , and 1.00 w m −2 , respectively, due to blowing snow. These changes are greater than those produced by the polynya.
The analysis above comparing the experiments P and I, IB and I reveals the roles that blowing snow and the polynya play individually. Next we address how the concurrence of blowing snow and polynya together influences the surface-air interactions. We first examine the blowing snow impacts when the polynya presents by comparing the experiments PB, IB, and I. The averaged difference throughout the study period between PB and IB shows that blowing snow sublimation is stronger when the polynya is present ( figure 1(i)). This enhanced sublimation can be attributed to stronger surface winds and warmer temperature over the polynya, which will be analyzed and discussed in section 3.3 below. Differences between PB and I exhibit an overall decrease in the magnitude of downward sensible and upward latent heat fluxes and an increase in downwelling LW (figures 1(j)-(l)). Differences between PB and I (PB-I) basically reflect the sums of the differences between P and I (P-I) and between IB and I (IB-I). However, the magnitudes of the PB-I differences are greater than the sums. For instance, the domain-averaged difference of sensible heat flux in PB-I is 2.52 w m −2 , which is greater than the sum (2.32 w m −2 ) of P-I (−1.88 w m −2 ) and IB-I (4.20 w m −2 ).
To gain a picture of how blowing snow impacts the polynya development, we compare the surface energy budget averaged over the polynya area between the experiments PB and P (figure S4 of the supplementary figures). Again, we have the same results as in the comparison between IB and I that blowing snow reduces the magnitudes of downward sensible and upward latent heat fluxes, but enhances downwelling LW. As a result, the net downward energy flux, i.e. the sum of sensible and latent heat fluxes and downwelling LW in the PB experiment shows a positive anomaly, which would melt more ice, suggesting that blowing snow favors further development of the polynya. Taken together, the concurrence of blowing snow and polynya amplifies the surface-air interaction, evidenced by the fact that the collective impact of the polynya and blowing snow is greater than the linear combination of individual impacts induced by polynya and blowing snow, respectively. This suggests that blowing snow can potentially help maintain the polynya for a longer period.

Processes associated with air-blowing snow-polynya interaction
To fully understand the surface-air interactions associated with blowing snow and the polynya, we further explore the three-dimensional dynamic and thermodynamic processes in the designed simulation experiments. As described in the conceptual model of blowing snow effects (Luo et al 2021), strong blowing snow sublimation only occurs when the air is relatively dry. During the study period here, there are three dips of the domain-averaged air humidity (figure S5(a) of the supplementary figures). The last one is very close to the end of the simulation period, so we only examine blowing snow sublimation and associated impacts for the first two dry periods. The surface wind speed is relatively strong during the first dry period, but weak during the second dry period ( figure S5(b)). This allows us to compare blowing snow's impacts and interactions with the polynya under different wind conditions.
We first look at the horizontal spatial characteristics of blowing snow and associated atmospheric properties influenced by the polynya using the experiments PB and IB. During the first dry period, strong blowing snow sublimation occurs right over the polynya with the vertically integrated sublimation rate close to 8 × 10 −6 mm s −1 in PB ( figure 2(a)). This is about 2 × 10 −6 mm s −1 greater than the sublimation rate in IB ( figure 2(b)). Southerlies over the polynya are stronger. The area with wind enhanced by the polynya (the gray stippling in figure 2(b)) matches the area with increased blowing snow sublimation. This suggests a strengthening role of the polynya in the wind field and, in turn, blowing snow sublimation, which can be further confirmed in the vertical profile analysis below.
Based on the horizontal spatial characteristics, we define a vertical cross-section (the green dashed line AB in figure 2(b)) to examine changes in the vertical structure of the lower atmosphere induced by blowing snow and the polynya. The vertical extent is from the surface to around 1 km above ground. As mentioned in section 3.2, a temperature inversion is present in the study area as depicted by the temperature profile in PB ( figure 3(a)). The near surface-temperature is about 5 • C colder than the upper air and the depth of the inversion becomes greater along the offshore direction. At the same time, a relatively large vertical gradient of the water vapor mixing ratio also occurs from the surface up to about 250 m, which is about the top of planetary boundary layer.   To further explore the impacts of the polynya on blowing snow sublimation as revealed by figure 2(b), we examine the vertical structure of the differences of air temperature, water vapor and wind vector profiles between PB and IB (figures 3(c) and (d)). It is found that the polynya reduces the intensity of the temperature inversion. The near-surface air temperature increases by up to 2 • C and the upper air temperature decreases by around 0.5 • C. The temperature decrease in the upper air is likely caused by the weakened temperature inversion due to the polynya-induced warming effect. The weakened inversion allows enhanced upward mixing of cold surface air. Horizontally the air temperature is colder along the onshore direction, i.e. an onshore temperature gradient exists. This gradient is stronger near surface as well as above the boundary layer, and becomes stronger in the near-surface layer when the polynya is present. This polynya-enhanced temperature gradient can act like the coastal front generating an upward wind anomaly nearshore. This anomalous upward wind causes a larger vertical extent of upper air cooling and can also bring more blowing snow particles upward to a drier environment, favoring stronger blowing snow sublimation. Water vapor increases in the lower atmosphere (mainly confined within the layer from the surface to about 300 m) due to enhanced latent heat flux from the polynya and because warmer air temperature is able to hold more moisture. The increase of air temperature and moisture becomes greater in the offshore direction, resulting from larger surface heat and moisture fluxes due to stronger surface winds.
Next we examine the cooling and moistening effects of blowing snow sublimation over the underlying surface conditions of the polynya and fully covered sea ice. Comparisons between two paired experiments PB and P (figures 3(e) and (f)) and IB and I (figures 3(g) and (h)) indicate that the intensity and vertical extent of the cooling and moistening effects due to blowing snow sublimation are larger when the polynya is present, particularly over the colder nearshore zone. In contrast to the upward wind anomaly caused by the polynya (figure 3(c)), the wind vector differences in PB-P and IB-I show downward anomalies. This suggests that the effect of blowing snow on its own is to generate a downward wind anomaly due to the cooling effect of sublimation. In addition, as the blowing snow sublimation is stronger nearshore (figure 2(a)), the cooling and moistening effects are greater along the onshore direction, which is opposite to the offshore maxima induced by the polynya (figures 3(c) and (d)). The further cooling of the cooler onshore near-surface air acts to enhance downward wind anomalies. Moreover, due to stronger sublimational cooling, the downward anomaly is stronger when the underlying surface is the polynya (figure 3(e) vs. 3(g)).
In contrast to the first dry period analyzed above, the second dry period is characterized by relatively weak wind. As indicated by the experiment PB, significant blowing snow sublimation mainly occurs over the Greenland ice sheet (figure 2(c)), which is probably caused by the dry downslope winds over the Greenland plateau. Off the north coast of Greenland where the most significant polynya is present, there is a pattern of high-low sublimation paired regions, as can be seen in cloud streets. These sublimation streets are likely formed due to gravity waves. While cloud streets form from rising air over the crests of gravity waves, sublimation streets detected here form over the troughs of gravity waves, which are characterized by stronger wind speed and warmer temperature (figure 4). Similar to those revealed in the first dry period, stronger wind always favors stronger blowing snow sublimation. On the other hand, when the wind speeds are similar, such as around distances of 100-110 km and 140-150 km along the magenta profile CD of figure 2(d), relatively large sublimation occurs with warmer air temperature ( figure 4(d)). The reason for this is that saturated water vapor mixing ratio at the surface of blowing snow particles is greater when the air is warmer, which helps blowing snow sublimate more efficiently. Therefore, stronger wind and warmer near-surface air temperature brought by polynya synergistically result in stronger sublimation streets in the experiment PB compared to IB ( figure 2(d)).

Summary
The extreme wind event of February 2018 in the Arctic resulted in an extraordinary polynya event north of Greenland (Moore et al 2018, Ludwig et al 2019. A strong wind event like this also magnifies another important process-blowing snow. In this study, we applied a snow-ice enhanced WRF-ice model (Zhang et al 2013, Luo et al 2021, Luo and Zhang 2022 to explore the impacts of blowing snow and the polynya on the surface-air interaction during the 2018 extreme wind event. A total of four simulation experiments with and without the polynya and blowing snow process were set up to conduct the study. The impacts of blowing snow and the polynya on the surface-air interaction are first examined individually. It is found overall that the polynya results in increases of the downward sensible heat flux and the upward latent heat flux, as well as downwelling longwave radiation (LW). The cooling and moistening effects of blowing snow sublimation lead to a reduction in the downward sensible heat flux and the upward latent heat flux, but increased downwelling LW. The signs of changes in the sensible and latent heat fluxes caused by the polynya and blowing snow are opposite, though both the polynya and blowing snow help increase downwelling LW at surface. The magnitudes of changes in surface energy fluxes caused by blowing snow are larger than those caused by the polynya.
Blowing snow sublimation is stronger in the presence of the polynya, compared to that when the surface is fully ice-covered. As a result, the collective impact of blowing snow and the polynya on the surface-air interaction is greater than the sum of impacts induced by the polynya and blowing snow, separately. A surface energy budget analysis over the polynya area demonstrates that blowing snow helps increase the net downward energy flux, which is the sum of sensible and latent heat fluxes, and downwelling LW. This net energy increase is mainly attributable to increased downwelling LW and reduced upward latent heat flux, suggesting that blowing snow has the potential to enhance the polynya development.
Because blowing snow sublimation occurs only when the air humidity is relatively dry, interactive processes among blowing snow sublimation, polynya, and atmospheric dynamics and thermodynamics are examined during two dry periods with different wind conditions. Under dry and windy condition, significant blowing snow sublimation occurs. At the same time, the presence of the polynya increases surface wind, weakens the near surface temperature inversion, and generates an upward wind anomaly, which prompts further enhanced blowing snow sublimation. The combined effects of polynya-induced stronger wind and warmer temperature by enhancing blowing snow sublimation can also be found under dry and relatively calm conditions. In the presence of gravity waves, sublimation streets occur over the troughs of gravity waves, where there are stronger winds and warmer temperatures. As a summary, a schematic diagram exemplifying the relationship between the polynya and blowing snow is shown in figure 5.
This study represents the first effort in modeling the air-blowing snow-polynya interaction and associated impacts. As already pointed out, blowing snow not only affects surface snow redistribution (Gallée et al 2001, Das et al 2013, Grazioli et al 2017, but it also impacts atmosphere thermal structure, cloud and associated radiation fluxes (Déry and Yau 2001, Lenaerts and van den Broeke 2012, Barral et al 2014, Yang et al 2014, Luo et al 2021, Hofer et al 2021, Luo and Zhang 2022. The processes related to surface snow redistribution and the thermodynamic effects of blowing snow sublimation have been carefully taken into consideration within WRF-ice. Using this model, we have revealed interactive processes between air, blowing snow, and polynya and their impacts on the surface energy budgets. However, there are also limitations of the study. A direct interaction between blowing snow particles and cloud and the associated radiation process has not been fully developed. Further model development will help to enable more comprehensive study of blowing snow induced feedback processes in the climate system. Nevertheless the absence of these blowing snow induced processses are a notable deficiency in most state-of-art weather and climate models. In order to better capture the20162021 surface-atmosphere interactions over regions such as the Arctic, Antarctic, Figure 5. Schematic diagram of blowing snow and polynya relationship. Sign of ± represents the positive/negative change, respectively. The polynya induced stronger wind, warmer temperature, and upward anomaly help enhance the blowing snow sublimation. Increased downwelling longwave radiation (LW) and reduced upward latent heat flux (LH) caused by the enhanced sublimation moistening effect favor further development of the polynya. SH is the sensible heat flux. and high mountain area, where there is extensive loose snow and strong winds, a full consideration of various processes and impacts induced by blowing snow is imperative.

Data availability statement
The data cannot be made publicly available upon publication because they are not available in a format that is sufficiently accessible or reusable by other researchers. The data that support the findings of this study are available upon reasonable request from the authors.