The Impact of Projected Arctic Sea Ice Loss on Large-Scale Patterns of Atmospheric Variability in the Polar Amplification Model Intercomparison Project

Arctic Amplification, the phenomena in which the high latitudes of the Northern Hemisphere

experience higher rates of warming than the global average, is a robust and well-known impact

of Arctic sea ice loss. In addition to this mean impact, studies have shown that future Arctic sea

ice loss leads to a decrease in temperature variance. A recent study by Gervais et al. (2024)

examined this in the Community Earth System Model from the perspective of large-scale 500hPa

height patterns over North America and found a general de-amplification of these patterns and

their associated lower tropospheric sensible weather with sea ice loss. Most notably, they found

that a large-scale pattern associated with deep cold anomalies with current sea ice conditions was

associated with warm anomalies in simulations with future Arctic sea ice loss. The mechanisms

responsible included thermodynamic, dynamic and diabatic processes. Furthermore, this single

pattern played an outsized role in the mean warming response to future Arctic sea ice loss. The

analysis in this study was conducted using a single model which leaves open the question of

model dependence of the findings.

This current work utilizes the Polar Amplification Model Intercomparison Project (PAMIP) to

determine whether the results in Gervais et al. (2024) are consistent between models.

Atmosphere-only simulations across multiple models with present-day and projected future

Arctic sea ice loss conditions are used as inputs to a machine learning algorithm, self-organizing

maps, to categorize large-scale patterns of winter daily 500hPa geopotential height anomalies.

Composite analysis is conducted to investigate the sensible weather associated with these large-

scale patterns. Results from this work show how patterns of atmospheric variability and

associated sensible weather compare across models. Furthermore, they assess the robustness of

the de-amplification of large-scale patterns owing to Arctic sea ice loss, including that with deep

cold anomalies, across models. Finally, we explore the mechanisms responsible for these model

differences.