Reconstruction of Antarctic Sea Ice State since 1979 using data assimilation of sea ice concentration in NEMO-SI3

Recent decades have been marked by pronounced changes in Antarctic sea ice, including record-breaking highs and lows and significant regional contrasts. This variability is puzzling and not captured by climate models; furthermore, while long-term records of sea ice extent are available since 1979, estimates of sea ice thickness are more difficult to obtain. Such estimates are, however, essential for understanding underlying processes, regional dynamics, for evaluating model performance, and for supporting climate studies in polar regions. Data assimilation (DA) offers a robust framework to combine satellite observations with numerical models to generate estimates of sea ice evolution over multi-decadal periods.

Here, we present a reconstruction of the sea ice state over the period 1979–2025 based on the assimilation of satellite-derived sea ice concentration (SIC) into the NEMO–SI³ sea ice–ocean model using an Ensemble Kalman Filter (EnKF). The ensemble consists of 25 members generated through perturbations of ERA5 atmospheric forcing. Monthly SIC observations from the OSI SAF dataset are assimilated throughout the satellite era, yielding a dynamically consistent reconstruction of Antarctic sea ice variability. The reconstructed product is primarily analysed for the Antarctic, with a detailed regional assessment across the main sea ice sectors.

This reconstruction provides a physically consistent description of Antarctic sea ice evolution over the last four decades and offers a basis for regional process studies, climate variability analyses. In parallel, this work represents a first step toward improving polar ocean and sea ice initial conditions for coupled prediction systems, including the Earth System Model EC-Earth. Future developments will involve the assimilation of additional satellite observations, such as sea ice freeboard, with the aim of extending this reconstruction toward coupled ocean–sea ice reanalyses and prediction applications.