Influence of snow on surface air temperature in the Northern Hemisphere and its impact on subseasonal prediction

Snow cover (SC) is considered an important contributor to atmospheric predictability on subseasonal to seasonal time scales. However, the causal relationship in the snow–atmosphere interaction and its impact on subseasonal to seasonal (S2S) predictability remains not fully understood. This study evaluates the causal relationship between SC and surface air temperature (SAT) at a submonthly timescale in the Northern Hemisphere throughout all seasons, using a novel technique known as Liang–Kleeman information flow analysis. By applying information flow analysis, we identify 'cold spots' where SC conditions effectively exert influence on SAT. The cold spots vary by month and region, including North Eurasia in September and October, East Siberia in October and May, Canada in November, East Asia in November and March, Central Asia in October and November, and Eastern Europe in March.

Next, we focus on the causal relationship over Eurasia, specifically examining the November east–west dipolar SC pattern, a dominant large-scale SC variability. The information flow analysis indicates a one-way causality from SAT to SC around the west pole (Europe) during boreal autumn (November), suggesting that SC has minimal influence on the time evolution of SAT. In contrast, causality from SC to SAT is significant around the east pole (the Mongolian Plateau). Sensitivity experiments using an atmospheric model support the SC–SAT causal relationship obtained by the information flow analysis and suggest that the SAT response to SC can endure for up to a month due to snow-albedo feedback, although the remote response of the upper atmosphere in the model is not significant.

Furthermore, we assess the SC–SAT causal relationship in operational S2S prediction models. The results indicate that the S2S models underestimate the influence of SC on SAT to varying degrees, suggesting deficiencies in the models. This study underscores that snow conditions are a source of SAT predictability on the S2S timescale and highlights the importance of accurately representing the effect of SC on SAT in S2S models for potential enhancements in sub-seasonal prediction skill. These findings contribute to a more fundamental understanding of the snow–atmosphere interaction and sub-seasonal predictability influenced by land snow conditions.

Komatsu, K. K., Y. Takaya, T. Toyoda, and H. Hasumi (2023) A submonthly scale causal relation between snow cover and surface air temperature over the autumnal Eurasian continent. Journal of Climate, 36, 4863–4877, doi:10.1175/JCLI-D-22-0827.1.
Takaya, Y., K. K. Komatsu, N. G. Ganeshi, T. Toyoda, H. Hasumi (2024) A sub-monthly timescale causality between snow cover and surface air temperature in the Northern Hemisphere inferred by Liang–Kleeman information flow analysis. Climate Dynamics, 62, 2735–2753, doi:10.1007/s00382-024-07112-6.