ECMWF | Reading | 10-13 September 2018
Earth System Assimilation
In 2018 it is 20 years since the original introduction of 4D-Var at ECMWF. To commemorate this achievement there was a one-day symposium on 4D-Var in January and then in September the Annual Seminar will be on 'Earth System Assimilation'.
4D-Var has brought many very significant benefits over the past 20 years. In recent years the importance of having an ensemble component to the data assimilation has also become apparent and effort has focused on optimising the ensemble configuration, including hybrid techniques. The 2018 Annual Seminar will present the state of the art in data assimilation methods as applied in different Earth system components, as well as progress and strategies for coupled data assimilation. In particular the seminar will give an opportunity to report and identify common themes in assimilation for different Earth system components (atmosphere, including composition; land; ocean; and cryosphere) with a particular emphasis on coupling. The Seminar will also present the status and evolution of the global observing system for Earth system assimilation.
The seminar began with an overview of motivation and methods in Earth system assimilation, followed by a review of the evolving global observing system for the Earth system.
It reported on ideas aiming at coupled assimilation in operational systems. This was followed by more detailed talks on aspects of data assimilation methodology relevant to multiple Earth system components, such as the handling of non-linearity and non-Gaussianity in variational systems, ensemble and particle filter data assimilation, modelling of background errors, model error and parameter estimation, observation error modelling and issues in data sparse systems (such as pre-satellite era reanalysis).
The seminar briefly presented forecast model developments that may drive requirements for new data assimilation capability as well as issues with the linear model required by the 4D-Var algorithm. The seminar moved on to a more detailed review of coupling strategies, including outer-loop coupling, full covariance coupling and coupling through the observation operator.
The seminar also considered actual progress in complex systems, including in the context of 20th century reanalysis, atmospheric composition, ocean and land.
Presentations and recordings
Monday, 10 September
|
Introduction |
 |
 |
|
Motivation and Methods in Earth System Data Assimilation |
|
|
|
Evolution of global observing systems |
|
|
|
Towards coupled assimilation in operational systems |
|
|
Tuesday, 11 September
|
Non-linearity and non-Gaussianity in 4D-Var (and beyond) |
|
|
|
Ensemble Data Assimilation and Particle Filters for NWP |
|
|
|
Characterising and modelling background error |
|
not available |
|
Model error and parameter estimation |
|
|
|
Issues specific to data sparse systems |
|
|
|
Characterising and modelling observation errors |
|
|
Wednesday, 12 September
|
Land, ocean, sea ice, wave coupled model developments |
|
|
|
Atmospheric composition coupled model developments and surface flux estimation |
|
|
|
Tangent-linear and adjoint models in data assimilation |
|
|
|
Comparison of data assimilation coupling strategies for Earth system models |
|
|
|
Coupling through the observation operator |
|
|
|
Coupled re-analysis |
|
|
Thursday, 13 September
|
Assimilation of atmospheric composition |
|
|
|
Ocean data assimilation for numerical weather prediction |
|
|
|
Land data assimilation in atmospheric systems |
|
|
|
Towards operational earth system assimilation : challenges and priorities |
|
|
Organising committee
Massimo Bonavita, Niels Bormann, Rossana Dragani, Stephen English, Johannes Flemming, Hans Hersbach, Elias Holm, Lars Isaksen, Sebastien Massart, Patricia Rosnay, Hao Zuo