The Arctic is experiencing unprecedented changes in its sea ice cover, characterized by thinning, reduced extent, and decreased resilience to summer melt. This phenomenon has significant implications for the global climate system. Snow plays a crucial role in this dynamic, acting as a highly reflective and insulative material. Its high albedo significantly reduces solar energy absorption by the sea ice and upper ocean, thus mitigating sea-ice melt and ocean warming. However, a critical knowledge gap exists concerning the drivers of summertime snow depth variability on Arctic sea ice. Understanding these drivers is essential for accurate predictions of sea ice conditions and their broader climatic impacts. This study aims to address this knowledge gap by investigating the relationship between summer snow depth on Arctic sea ice and the Arctic Oscillation (AO), a significant mode of natural climate variability that influences Arctic snowfall and air temperatures. The focus is on the summer melt season, when the presence of snow exerts a substantial influence on the albedo and, consequently, on surface melt. The study uses atmospheric reanalysis, snow modeling, and satellite data to analyze the role of the AO in modulating summer snow conditions on Arctic sea ice from 1980 to 2020.

Previous research has highlighted the importance of snow on sea ice for both sea-ice and climate systems. Snow's high albedo and insulation properties significantly influence energy balance and melt processes. The timing of sea-ice formation is crucial in determining snow distribution, with later formation leading to thinner snowpacks. While spatial patterns of snow distribution are relatively well-understood, significant knowledge gaps exist regarding temporal variability, particularly during the summer melt season. Limited routine sampling has hampered efforts to fully understand the drivers of summertime snow depth variability. Historical observations provide some insights but are limited in scope, primarily covering autumn and spring, and lacking sufficient data during the summer months. Therefore, this study aims to improve our understanding by investigating the role of atmospheric variability, particularly the Arctic Oscillation.

This study employed a multifaceted approach to investigate the relationship between the Arctic Oscillation (AO) and summertime snow on Arctic sea ice. The analysis focused on the June-August period, representing peak insolation and melt within the broader summer melt season (May-September). Data were averaged to monthly values, and anomalies were calculated and detrended based on the 1980-2020 mean. Spatial analyses incorporated latitude weighting (cosine of latitude) for principal component analysis, except for albedo data. Regression analyses examined the linear relationships between variables, with focus on the first principal component representing the leading pattern of variability. Composite analyses differentiated between linear and nonlinear responses to the AO, employing the extreme 5 years exceeding one standard deviation from the mean to define positive and negative AO composites. Lead-lag correlation analysis was conducted to assess preconditioning effects of sea-ice coverage, melt-freeze onset timing, and pre-existing snow conditions on summer snow depth variability. A Bayesian time series decomposition method was used to identify a breakpoint in the June-August sea ice concentration time series (1980-2020), revealing a significant shift around 2007. Multiple datasets were integrated: * **ERA5 Reanalysis:** Provided data on 1000 hPa geopotential height, sea level pressure (SLP), 2 m air temperature, and snowfall. Monthly AO indices were constructed based on established methods. * **SnowModel-LG:** A Lagrangian snow model provided snow depth estimates at 25 km resolution. The model was forced by ERA5 reanalysis. Model performance was evaluated against ice mass balance buoy data, revealing a consistent underestimation of summer snow accumulation. * **Satellite Retrievals:** Included sea ice concentration (SIC) from OSI SAF, melt-freeze onset dates from a passive microwave product, surface albedo from CLARA-A3, and melt pond fraction from MODIS-Peng. The albedo data were adjusted to account for the effects of melt ponds, highlighting the complexities involved in interpreting surface albedo changes. * **Snow Depth Observations:** Included historical snow depth observations from North Pole drifting ice stations (1954-1991) and ice mass balance buoys (1993-2017) to corroborate modeling results. * **Cyclone Tracking:** The Melbourne University cyclone tracking scheme was applied to ERA5 SLP fields to identify and analyze cyclone activity. Cyclone intensity was assessed by examining minimum SLP within cyclone areas. The analysis involved a range of statistical methods including correlation analysis, regression, principal component analysis, and composite analysis to thoroughly examine the interactions and relationships between the specified variables. The breakpoint analysis provided insight into changes in relationships pre- and post-2007, which is crucial for understanding long-term trends and the possible effects of climate change.

The study's key findings demonstrate a strong link between the Arctic Oscillation (AO) and summertime snow depth variability on Arctic sea ice over the period 1980-2020. * **AO Influence on Snow Depth:** Positive phases of the AO were associated with increased snow accumulation, up to 4.5 cm more near the North Pole compared to neutral conditions. This resulted in higher surface albedo in summer, particularly during June, due to increased snowfall. The linear response of snow depth to AO was stronger than the nonlinear response, indicating a consistent pattern of deeper snow during positive AO summers. * **Albedo and AO:** Positive AO indices in June-July correlated with higher mean July surface albedos in the continuous sea-ice zone. While the correlation was not perfect (r = 0.49, P = 0.001), correction for melt pond effects still showed a statistically significant relationship (r = 0.46, P = 0.030), suggesting other factors like variations in sea-ice coverage and open leads influenced overall albedo. The influence of melt ponds on albedo was found to be less significant than other factors. * **Melt Onset and Snow Depth:** The relationship between melt onset timing and snow depth variability was weak and not statistically significant when averaged over June-August. However, a strong correlation was observed in June for the eastern and central Arctic, with positive AO summers exhibiting later melt onset. This suggests a short-lived effect of melt onset on snow depth variability. * **Sea Ice and Snow Depth:** June sea-ice coverage significantly explained 19% of the snow depth variance, primarily in peripheral seas. This relationship weakened in July and was stronger again in August, partly attributed to the timing of sea-ice freeze-up. Previous year's freeze-up timing affected June snow depth variability in some regions, but this effect was short-lived. * **Remnant Snow:** Remnant snow from April-May significantly contributed to June snow depth variability, with this contribution decreasing in subsequent months. Sea-ice drift likely played a role in diminishing the correlation between early-summer and late-summer snow depths. * **Atmospheric Conditions:** Positive AO phases led to enhanced snowfall due to cooler temperatures aloft and a more active Atlantic storm track. Cooler 850-hPa temperatures promoted higher snowfall-to-total precipitation ratios. Increased cyclone activity, with more intense cyclones penetrating deeper into the Arctic during positive AO summers, was a key driver of increased snowfall. 2 m air temperature showed weak inverse correlations with the AO, limited to the central Arctic. * **AO Influence Weakening after 2007:** The relationship between the AO and snow depth, snowfall, air temperature, and cyclone activity weakened after 2007. Negative correlations emerged between the AO and these variables in some coastal areas of the Pacific sector, though not statistically significant. The relationship between the AO and sea-ice concentration also weakened, implying a significant shift in the AO's influence on snow-sea ice interactions. The reduced sea ice cover and increased proportion of liquid precipitation are likely contributing factors to this weakening relationship. * **Historical Corroboration:** Historical and contemporary snow observations support the findings, showing a positive correlation between snow depth and the AO. However, caution is necessary given the limitations of these observations.

The findings of this study address the research question by demonstrating the substantial influence of the Arctic Oscillation (AO) on summertime snow depth variability on Arctic sea ice. The strong positive relationship between positive AO phases and increased snow accumulation, leading to higher surface albedo, has significant implications for understanding sea-ice melt dynamics. The study highlights the importance of atmospheric variability in modulating summer snow conditions, which are typically underrepresented in current sea-ice models. The weakening of this relationship post-2007 is particularly noteworthy, underscoring the potential for climate change to reshape the interactions between the AO, snow cover, and sea ice. This suggests that future projections of sea-ice conditions need to account for these evolving relationships, especially considering projected declines in sea-ice extent and increased occurrence of rain events. The results highlight the complex interactions between atmospheric circulation patterns, snowfall, sea-ice conditions, and the surface energy budget in the Arctic. Further research should investigate the regional variations of these interactions and refine models to incorporate the AO's influence on summer snow depth more accurately. Future work should also focus on improving the accuracy of summer snow observations and investigating the potential impacts of the observed weakening relationship on future Arctic climate and sea-ice dynamics.

This study significantly advances our understanding of the interplay between the Arctic Oscillation, atmospheric conditions, and snow cover on Arctic sea ice during the critical summer melt season. The strong correlation between positive AO phases and increased summer snow accumulation, leading to higher albedo and potentially mitigating sea-ice melt, is a key finding. However, the weakening of this relationship after 2007 highlights the need to account for the impacts of climate change on these complex interactions. Future research should focus on refining models to incorporate these evolving relationships and better predict Arctic sea ice conditions under a changing climate. The study's results emphasize the dynamic nature of the Arctic climate system and the importance of considering atmospheric variability in predicting sea-ice conditions.

The study's findings are primarily based on reanalysis data and model output, with the limitations inherent in these approaches. The model used for snow depth estimation, SnowModel-LG, showed a consistent underestimation of summer snow accumulation, which might affect the precise quantification of the AO's influence. While efforts were made to account for factors such as melt ponds and sea-ice drift, their effects might not have been completely captured. The limited number of years post-2007 in the breakpoint analysis needs to be considered when interpreting the weakening of AO's influence. Also, the historical snow observations, though providing corroborating evidence, are limited in spatial and temporal coverage. These limitations warrant further research to refine our understanding of the intricate interactions between the Arctic Oscillation and the Arctic snow-sea ice system.