Multiscale ocean data assimilation

Titre : Multiscale ocean data assimilation

Laboratoire de rattachement : LEGI

Encadrant : Emmanuel Cosme emmanuel.cosme hmg.inpg.fr

Téléphone : 04.76.82.50.51

Mots clés : ocean dynamics, data assimilation

Contexte et objectifs de la mission de stage :

Ocean data assimilation (DA) consists in merging observations of the ocean state with a numerical model of the ocean circulation, with the purpose of constraining the model and make its solution more realistic. In DA, dealing with the multiple spatial scales involved in the ocean dynamics is particularly tricky. Models perform their computation on discretized grids unable to represent the dynamics at scales smaller than their grid meshes. Observations often reflects dynamical scales much smaller than models grid meshes. This characteristic is referred to as the representativity of observations.
In DA, observations representativity must be accounted for. But in practice, this is generally done in a rather crude way. The work proposed here consists in implementing a theoretically rigorous approach recently developed at MEOM to take representativity into account properly. A simplified model mimicking the Gulf Stream dynamics will be run at 2 different resolutions (1/4° and 1/12°). Synthetic observations will be extracted from the 1/12° simulation, and assimilated into the 1/4° model. The new, "optimal" multiscale method will be compared to the standard one.

—