Why climate models are unable to reproduce the observed Antarctic sea-ice expansion

Antarctic sea-ice has expanded over the period of continuous satellite monitoring, which seemingly contradicts ongoing global warming resulting from increasing concentrations of greenhouse gases. A variety of hypotheses have been proposed to explain the observed Antarctic sea-ice expansion and corresponding model–observation discrepancy, but the issue remains unresolved. In a new study published in Nature Climate Change, an international team of scientists led by Eui-Seok Chung at the Korea Polar Research Institute in South Korea shows that Antarctic sea ice has expanded since the late 1970s due to ocean surface cooling associated with multidecadal variability in the Southern Ocean that temporarily outweighs the opposing forced response. In both observations and model simulations, Southern Ocean multidecadal variability is closely linked to internal variability in the tropics. The linkages are, however, distinctly weaker in simulations than in observations resulting from potential model bias in the forced response and observed tropical variability. Thus, the forced response dominates in simulations, resulting in an apparent model–observation discrepancy.

Figure 1 from Chung et al. (2022). Observed and model-simulated changes in annual-mean SIE and SST over the SO (south of 50° S): a, Box plots of model-simulated SIE trends over 29-year (yellow green) and 36-year (dark blue) periods for three cases: Hist, trends over 1950−1978 and 1979−2014 under historical forcing; PI, trends for all possible overlapping 29-year and 36-year segments of pre-industrial control runs; and PI + forced, PI trends with the corresponding ensemble-mean values for 1950−1978 and 1979−2014 added. The box covers the inter-quartile range with the line inside the box representing the median value across multi-ensemble models and whiskers denoting the maximum and minimum values. The red solid line denotes the satellite-observed 1979−2014 SIE trend with the accompanying dashed lines representing the standard error of the trend. b, Same as in a, but for SST trends. The orange solid line denotes the observed 1950−1978 SST trend averaged over four SST datasets: Extended Reconstructed Sea Surface Temperature (ERSST), Hadley Centre Sea Ice and Sea Surface Temperature (HadISST), Centennial in situ Observation-Based Estimates (COBE) and European Centre for Medium-Range Weather Forecasts Reanalysis v.5 (ERA5). The accompanying dashed lines represent minimum and maximum trends. The solid and dashed lines in red denote the corresponding observed SST trends over 1979−2014. c, Time series of SIE anomaly relative to the 1979−2020 means. The red dot denotes the SIE anomaly for September 1964 from the Nimbus-1 satellite. For model simulations, lines denote the ensemble-mean anomaly for individual models. The shading indicates inter-ensemble variability for the Community Earth System Model version 2 (CESM2) Large Ensemble with one and two standard deviations represented, respectively, by dark and light grey.

Chung, ES., Kim, SJ., Timmermann, A., Ha, KJ., Lee, SK., Stuecker, MF., Rodgers, KB., Lee, SS. & Huang, L. (2022). Antarctic sea-ice expansion and Southern Ocean cooling linked to tropical variability. Nature Climate Change, https://doi.org/10.1038/s41558-022-01339-z