Variational assimilation of tomography data, twin experiments
Fabienne Gaillard
IFREMER
In the previous analysis (Thetis 1), priority was given to the data. Theoretical knowledge of the dynamical behavior of the ocean can be added to the measurements in the form of a numerical predictive model, the analysis then takes the name of data assimilation. In order to evaluate the feasability of assimilating tomography data, a simulation has been set up in the configuration of Thetis 2 (western Mediterranean basin). The variational method, based on the model adjoint equations, has been applied first with a quasi-geostrophic model, and later with a primitive equation model. Tomography gives access to the large scale part of the spectrum and we observe in particular that the tomography data impose a constraint on the model which maintain the integral heat content at the correct value and prevent model drift. The performance of the estimation is analyzed by studying separately the baroclinic and barotropic components of the field, horizontal resolution is also considered. Finally complementarity with altimetry is explored in the case of the Q-G model.