Summary
Progress in data assimilation allows:
- noisy, irregular and indirect observations to be combined with
- models that can predict observations to give a
- global, realistic and dynamically consistent state.
|
Improved data assimilation, the use of new observation types, and better models all improve forecast skill.
|
Problems and challenges:
- Initialization
- Adjoint construction (especially forecast model parametrizations)
- Kalman filter
- Allowing for model error
- New observation systems
- Representations of B ...
|
|