New perspectives on the skill of modelled sea ice trends in light of recent Antarctic sea ice loss

Most climate models do not reproduce the 1979–2014 increase in Antarctic sea ice area (SIA). This was a contributing factor in successive Intergovernmental Panel on Climate Change reports allocating low confidence to model projections of sea ice over the 21st century. However, due to the rapid declines in Antarctic sea ice since 2016, the linear trend in annual mean Antarctic SIA is no longer positive. We therefore investigate what impact this has on the evaluation of trends from the CMIP6 multi-model ensemble and show that the recent rapid declines bring observed SIA trends back into line with the models. More generally, the level of agreement between observed and modelled linear trends depends both on the length of the time series examined ('timescale') and the exact years ('time period').  Our novel result that trends over the full satellite era 1979–2023 do not disagree between observations and models could imply that models are better able to represent changes over longer timescales than previously thought. However, this is not the only interpretation. One confounding aspect is the abrupt nature of recent change, as a result of which the full time series does not appear particularly linear. This presentation will discuss these aspects and the implications for future research priorities.