Atmosphere and ocean both use incremental updates to incorporate data assimilated analyses, largely to avoid "shocks" that can degrade or fail simulations.
Currently, CICE does a "direct insertion" of the analysis, as it seems to work and can be done outside the model.
However, this leads to
- mismatches between ice and atm-ocn as the next forecast cycle begins with "old" background atm-ocn and "new" analysis ice
- responsibilities for DA to generate a complete, updated model restart that model integrates
Useful references:
https://doi.org/10.1002/qj.2555
https://tc.copernicus.org/articles/20/3299/2026/
https://spj.science.org/doi/10.34133/olar.0025
Atmosphere and ocean both use incremental updates to incorporate data assimilated analyses, largely to avoid "shocks" that can degrade or fail simulations.
Currently, CICE does a "direct insertion" of the analysis, as it seems to work and can be done outside the model.
However, this leads to
Useful references:
https://doi.org/10.1002/qj.2555
https://tc.copernicus.org/articles/20/3299/2026/
https://spj.science.org/doi/10.34133/olar.0025