Investigating a potential new Antarctic sea ice state during 2021–2023 using an atmosphere-based reconstruction

The persistent below normal Antarctic sea ice extent (SIE) since 2017 after decades of increases possibly suggests a new Antarctic sea ice state, from largely wind-driven variability prior to 2015, with more oceanic influence after 2015. To investigate this further, here we make use of a new large ensemble of seasonal Antarctic SIE statistical reconstructions calibrated during 1979–2023, as well as the original reconstructions calibrated during 1979–2020, before many of the observed record lows. Since these reconstructions are primarily based on atmospheric predictor variables (and indirect oceanic influences via teleconnections), their performance in predicting recent SIE extremes (2021–2023) diagnoses whether atmospheric or oceanic processes dominate the new state, while also robustly assessing reconstruction skill. The first group of new reconstructions performs only slightly lower than the original version and generally remains stationary through time. However, the original reconstruction only reliably predicts the SIE anomalies during 2021–2023 in the Amundsen-Bellingshausen and Weddell Seas. The maintained higher predictive power and model performance near the Antarctic Peninsula supports the possibility of a continued wind-driven role across much of West Antarctica for recent extremes, and a stronger oceanic thermodynamic role in much of East Antarctica. While some ensemble members, primarily in the Ross-Amundsen Sea sector, produce an extreme low SIE in 1958 connected to a strong tropically-driven component, the reconstructions rarely produce negative Antarctic SIE anomalies as low as observed in winter 2023. The analysis suggests the reconstructions may under-represent SIE variations which have a stronger oceanic influence, unless specifically calibrated to these relationships.