Causality of the Link between Autumn Arctic Sea Ice and the Winter Extratropical Atmosphere

Abstract

Changes in Arctic sea ice have been proposed to affect the mid-latitude winter atmospheric circulation, often based on observed variability. However, causality of this relationship remains unclear. This thesis investigates the link between autumn sea ice and the extratropical winter atmosphere, clarifying the role of internal variability and demonstrating a role for tropical variability. Single model experiments can simulate an apparent link between autumn sea ice and the winter North Atlantic Oscillation (NAO) through unforced internal variability, but the ensemble average relationship is weak, suggesting a large role for internal variability. Additionally, longer, free-running simulations also indicate this link is highly non-stationary in time. These results question the robustness of proposed sea ice-NAO links based solely on short observational reanalysis. Multiple linear regression and causal effect network analysis indicate that the tropical west Pacific plays a role in the link between sea ice and the NAO. This is supported by multi-model simulations containing large ensembles, which demonstrate minimal causal influence of sea ice variability on the winter NAO, and that winter extratropical patterns connected with sea ice variability partly originate in the tropical Pacific. Tropical nudging experiments in autumn reveal that while tropical information is not sufficient to directly recreate interannual BK ice variability, tropical information can reproduce the autumn stratospheric polar vortex and NAO variability, which are strongly linked to the winter NAO and autumn Barents-Kara sea ice respectively. The signal-to-noise issue present throughout dynamic model simulations may lead to overly weak extratropical teleconnections, which may inhibit the detection any direct tropical to Arctic link. These results provide new evidence of a non-causal link between sea ice and the extratropical circulation, stemming from tropical sources, and further clarify the role of internal variability.