Skip to Main content Skip to Navigation
Theses

Assimilation de données et identification de paramètres : une application en hydrologie

Abstract : Some hydrodynamic parameters for unsaturated flow models cannot be measured, they have to be tuned with respect to the observations and the ouput of the model. The objective of this work is to propose a variational data assimilation method allowing to retrieve these parameters by combining the model and the observation. The method is based on optimal control theory. The work deals with the problems of parameters identification with monodimensional Richard's model. As a preliminary, an investigation of the physics related to the problem is explored. The data to be assimilated are measurements of cumulative infiltration and the control vector selected is the initial condition, the boundary conditions and the hydrodynamic parameters. Thus according to certain distributions of the observations (observed cumulative infiltration), the control parameter is retrieved. This phase is followed by an a posteriori study based on the use of second order adjoint which allows to estimate the discrepancy, the influence of the temporal configuration of the observations on the quality of the identification as well as a good understanding of the minimization process. Automatic derivation through the differentiator ODYSSEE is used to compute first and the second order information. Lastly, by using previous second order studies realized, Newton algorithm is applies to the optimality system.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00004695
Contributor : Thèses Imag <>
Submitted on : Monday, February 16, 2004 - 5:40:43 PM
Last modification on : Friday, November 6, 2020 - 4:08:49 AM
Long-term archiving on: : Friday, April 2, 2010 - 7:23:30 PM

Identifiers

  • HAL Id : tel-00004695, version 1

Collections

Citation

Pierre Désiré Ngnepieba. Assimilation de données et identification de paramètres : une application en hydrologie. Modélisation et simulation. Université Joseph-Fourier - Grenoble I, 2001. Français. ⟨tel-00004695⟩

Share

Metrics

Record views

492

Files downloads

479