Images and Prediction for Geophical Fluids

F.-X. Le Dimet, A. Vidard, O. Titaud, I. Souopgui
INRIA-UJF

Predicting the evolution of the ocean requires gathering all the available information (models, data, statistics) in order to retrieve an initial condition. This task is carried out through a process of data assimilation. It is worthwhile to point out that the number of degrees of freedom of the model is much larger than the number of observations leading to an ill-posed problem. In recent years, many satellites have been launched and provide a lot of information in the form of images.

How to incorporate images into numerical models? Two basic approaches are proposed:

1) Extract some patterns (vortex, fronts, etc.) from a sequence of images. Tracking these patterns/features allows you to estimate horizontal velocity which can be assimilated into the model as regular data.

2)Project the patterns to model state variables using the appropriate basis functions. Then, create associated observation operators to assimilate into the model.

We will present some applications to experimental data in the lab (Coriolis Platform) and to images of a model of the Black Sea.