Antarctic Sea Ice Trends: Insights from a Suite of Climate Models

Abstract

Background: Antarctic sea ice concentration has been observed to increase from 1978 to 2015, in contrast with the decrease that most climate models show. Here, we aim to examine the respective roles of natural variability and anthropogenic forcing in shaping Antarctic sea ice trend. Method: To do so, we use the GFDL-CM2 coupled climate model with varying horizontal resolutions in the ocean (1°, 0.25° and 0.10°) that displays a range of behaviours in natural variability with the representation of Weddell Sea polynyas, and different intensities in the decrease of sea ice under climate change. Results: In the 0.10° model, a sea ice trend of similar sign and magnitude to that observed over the satellite record is found between two occurrences of the Weddell Sea polynya. In the 1° and 0.25° models, which do not simulate any polynya, no equivalent trend of what the satellite record shows is found. Under increasing CO2 forcing, all models show a surface cooling on a short time scale (years) south of 50°S, followed by a warming on a longer time scale (decades), consistent with the delayed warming mechanism of Ferreira et al. (2015). Of all models, the higher resolution model shows the strongest surface warming and decrease in sea ice, suggesting an important role for mesoscale eddies in the response of Antarctic sea ice to climate change. Conclusion: We conclude that the Weddell Sea polynya is key to the representation of the sea ice trend and that the disagreement between models and observations might partly arise from a desynchronization of the polynya cycles or a too weak natural variability of sea ice in models compared to observations.