Skip to Main content Skip to Navigation

Assimilation variationnelle de données dans un modèle couplé océan-biogéochimie

Abstract : This thesis is concerned with the implementation of the optimal control numerical method applied to a data assimilation problem in ocean biogeochemistry. We first present the system of non-linear partial differential equations which modelizes the evolution over space and time of physical and biological variables. A first part of the work is mathematical and consists of proving existence, uniqueness and positivity of the solution to the biological model. The second part of the work is numerical. The model is discretized using finite differences. Then, the tangent linear and adjoint codes are derived using automatic differentiation. Once these tools are developped the inverse data assimilation problem can be dealt with. The control variables are the parameters included in the non-linear biological reaction terms. We seek an optimal parameter set minimizing a cost function which measures the distance between model and observations in a least square sens. A preliminary sensitivity study using the tangent linear model as well as twin experiments using the adjoint model are conducted. We then use the method to assimilate real data from the JGOFS-Dyfamed station in the Northwestern Mediterranean Sea.
Complete list of metadatas
Contributor : Blaise Faugeras <>
Submitted on : Thursday, December 19, 2002 - 5:22:34 PM
Last modification on : Friday, November 6, 2020 - 4:04:00 AM
Long-term archiving on: : Tuesday, September 11, 2012 - 7:10:09 PM


  • HAL Id : tel-00002176, version 1



Blaise Faugeras. Assimilation variationnelle de données dans un modèle couplé océan-biogéochimie. Modélisation et simulation. Université Joseph-Fourier - Grenoble I, 2002. Français. ⟨tel-00002176⟩



Record views


Files downloads