Parallel algorithms for variational data assimilation and coupling models

The purpose of this paper is to show how variational data assimilation can be used to couple some models and to perform parallel algorithms. In a first part we will recall the general principles of Variational Data Assimilation and we will give an usual algorithm to solve this problem. The following section deals with the problem of coupling models and more precisely how to perform data assimilation for coupled models. We also propose in this section a parallel algorithm to solve this problem. The third section concerns the parallel implementation of the proposed algorithms with some variants, and the numerical results obtained with the proposed algorithms.