English
 
Privacy Policy Disclaimer
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT
Add to Basket
  Local TagsRelease HistoryDetailsSummary

Released
Conference Paper

Bayesian online algorithms for learning data-driven models of chaotic dynamics

Authors

Bocquet,  Marc
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Farchi,  Alban
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Malartic,  Quentin
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Finn,  Tobias
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Durand,  Charlotte
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Bonavita,  Massimo
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Laloyaux,  Patrick
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

Chrust,  Marcin
IUGG 2023, General Assemblies, 1 General, International Union of Geodesy and Geophysics (IUGG), External Organizations;

External Ressource
No external resources are shared
Fulltext (public)
There are no public fulltexts stored in GFZpublic
Supplementary Material (public)
There is no public supplementary material available
Citation

Bocquet, M., Farchi, A., Malartic, Q., Finn, T., Durand, C., Bonavita, M., Laloyaux, P., Chrust, M. (2023): Bayesian online algorithms for learning data-driven models of chaotic dynamics, XXVIII General Assembly of the International Union of Geodesy and Geophysics (IUGG) (Berlin 2023).
https://doi.org/10.57757/IUGG23-1847


Cite as: https://gfzpublic.gfz-potsdam.de/pubman/item/item_5017738
Abstract
The advent of machine and more specifically deep learning techniques significantly boosts the capabilities of data assimilation and inverse problem techniques used in the geosciences. It also spurs new, more ambitious goals for data assimilation in high dimensions. One of the key, currently very popular, area of research consists in learning data-driven models of dynamical systems. With the natural constraints of geoscience, i.e., sparse and noisy observations, this typically requires the joint use of data assimilation and neural networks. However, the vast majority of algorithms are offline; they rely on a set of observations from the physical system, which must be available before the start of the training. We propose new algorithms that update the knowledge of the surrogate (i.e., data-driven) model when new observations become available. We carry out this objective with both variational (weak-constraint 4D-Var like) and ensemble (EnKF and IEnKS like) techniques. We test these algorithms on low-order Lorenz models, on quasi-geostrophic models, the ERA5 dataset, and sea-ice dynamics. Remarkably, in several cases, the online algorithms significantly outperform the offline ones. This opens the way to adaptive surrogate models that progressively learn trends and conform to real-time constraints of operational weather forecasting.