Skip to Main content Skip to Navigation
Conference papers

Completing physics-based models by learning hidden dynamics through data assimilation

Abstract : Data Assimilation remains the operational choice when it comes to forecast and estimate Earth's dynamical systems. The analogy with Machine Learning has already been shown and is still being investigated to address the problem of improving physics-based models. Even though both techniques learn from data, machine learning focuses on inferring models while data assimilation concentrates on hidden system state estimation with the help of a dynamical model. In this work, we exploit the complementarity of these methods in a twin experiment where the system is partially observed and the known dynamics are incomplete. Finally, we succeed in partially retrieving the dynamics of a fully-unobserved variable by training a hybrid model through variational data assimilation.
Complete list of metadata
Contributor : Dominique Béréziat Connect in order to contact the contributor
Submitted on : Friday, December 11, 2020 - 9:43:36 PM
Last modification on : Wednesday, November 17, 2021 - 12:33:27 PM
Long-term archiving on: : Friday, March 12, 2021 - 8:31:32 PM


Files produced by the author(s)


  • HAL Id : hal-03004938, version 1


Arthur Filoche, Julien Brajard, Anastase Alexandre Charantonis, Dominique Béréziat. Completing physics-based models by learning hidden dynamics through data assimilation. NeurIPS 2020, workshop AI4Earth, Dec 2020, Vancouver (virtual), Canada. ⟨hal-03004938⟩



Record views


Files downloads