Eurasian snow cover variability and links to winter climate in the CMIP5 models

Abstract

Observational studies and modeling experiments illustrate that variability in October Eurasian snow cover extent impacts boreal wintertime conditions over the Northern Hemisphere (NH) through a dynamical pathway involving the stratosphere and changes in the surface-based Arctic Oscillation (AO). In this paper, we conduct a comprehensive study of the Eurasian snow–AO relationship in twenty coupled climate models run under pre-industrial conditions from the Coupled Model Intercomparison Project Phase 5 (CMIP5). Our analyses indicate that the coupled climate models, individually and collectively, do not capture well the observed snow–AO relationship. The models lack a robust lagged response between October Eurasian snow cover and several NH wintertime variables (e.g., vertically propagating waves and geopotential heights). Additionally, the CMIP5 models do not simulate the observed spatial distribution and statistics of boreal fall snow cover across the NH including Eurasia. However, when analyzing individual 40-year time slices of the models, there are periods of time in select models when the observed snow–AO relationship emerges. This finding suggests that internal variability may play a significant role in the observed relationship. Further analysis demonstrates that the models poorly capture the downward propagation of stratospheric anomalies into the troposphere, a key facet of NH wintertime climate variability irrespective of the influence of Eurasian snow cover. A weak downward propagation signal may be related to several factors including too few stratospheric vortex disruptions and weaker-than-observed tropospheric wave driving. The analyses presented can be used as a roadmap for model evaluations in future studies involving NH wintertime climate variability, including those considering future climate change.