In this study, sea ice concentration (SIC) budgets were calculated for five ocean-sea ice reanalyses (CFSR, C-GLORSv7, GLORYS12v1, NEMO-EnKF and ORAS5), in the Southern Ocean and compared with observations. Benefiting from the assimilation of SIC, the reanalysis products display a realistic representation of sea ice extent as well as sea ice area. However, when applying the SIC budget diagnostics to decompose the changes in SIC into contributions from advection, divergence, thermodynamics, deformation and data assimilation, we find that both atmospheric and oceanic forcings and model configurations are significant contributors on the budget differences. For the CFSR, the primary source of deviation compared to other reanalyses is the stronger northward component of ice velocity, which results in stronger sea ice advection and divergence. Anomalous surface currents in the CFSR are proposed to be the main cause of the ice velocity anomaly. Furthermore, twice the mean ice thickness in the CFSR compared to other reanalyses makes it more susceptible to wind and oceanic stresses under Coriolis forces, exacerbating the northward drift of sea ice. The C-GLORSv7, GLORYS12v1 and NEMO-EnKF have some underestimation of the contribution of advection and divergence to changes in SIC in autumn, winter and spring compared to observations, but are more reasonable in summer. ORAS5, although using the same coupled model and atmospheric forcing as C-GLORSv7 and GLORYS12v1, has a more significant underestimation of advection and divergence to changes in SIC compared to these two reanalyses. The results of the SIC budgets of five ocean-sea ice reanalyses in the Southern Ocean suggest that future reanalyses should focus on improving the modelling of sea ice velocities, for example through assimilation of sea ice drift observations. |
