Data Assimilation Seminar
Prof. Ronan Fablet (October 21, 2022, 13:00-14:30)
|Affiliation||IMT Atlantique, France|
|Title||4DVarNets to learn Variational Data Assimilation and Solvers: application to space oceanography|
Whereas model-driven approaches represent the state-of-the-art for the analysis, simulation and reconstruction of (geo)physical dynamical systems, learning-based and data-driven frameworks become relevant schemes for a large number of application domains, including for the study of phenomena governed by physical laws. They offer new means to fully benefit from available observation and/or simulation data. In this context, making the most of model-driven and data-driven paradigms naturally arises as a key challenge.
In this talk, we focus on data assimilation (DA) problems and introduce the joint learning of variational DA models and solvers for the reconstruction of space-time dynamics from irregularly-sampled and possibly multimodal observation data. Applications to the reconstruction of sea surface dynamics from satellite-derived observations support the relance of the proposed framework. We may discuss further embedding uncertainty quantification in the proposed framework.