Can internal variability explain observed Antarctic sea ice expansion?

Abstract. Expansion of Antarctic sea ice was observed over the pre-2014 satellite record, and most climate models are not able to reliably simulate this expansion. Here, we explore the hypothesis that internal variability is enough to explain the expansion by analyzing two ensembles of the Community Earth System Model version 1.3, which have ocean/ice model resolutions of ∼10 km (high-resolution; HR) and ∼100 km (low-resolution; LR). The HR configuration often simulates Antarctic sea ice area (SIA) expansion in September of similar magnitude to observations, while LR rarely does. However, the zonal structure of the SIA expansion in HR has important discrepancies with observations, overemphasizing sea ice increases within the Weddell Sea ice pack. In the rare cases when LR has multidecadal SIA expansion similar to observations, it appears to get the zonal structure correct. Recovery from open-ocean deep convection explains the relative frequency of SIA expansion in HR, while the SIA expansion in LR is associated with variability in the Amundsen Sea low and the Pacific–South American pattern. We show that high-resolution modeling has the potential to help reconcile simulated Antarctic sea ice trends with observations, but the spatial patterns of the trends must be taken into consideration when assessing model performance.