PhD Defense

The aim is to study the climate of Mediterranean region with a regional coupled model. Because the sea and the orography surrounding modulate the climate towards small-scales, regional coupled model are an essential tool to analize Mediterranean climate. LMDZ-Med/NEMO MED8 model results from the coupling between LMDZ-Med regional atmospheric model and the oceanic limited area model NEMO MED8. LMDZ-Med/NEMO MED8 is then guided outside the Mediterranean region with ERA40 dataset as boundary conditions in order to perform a dynamical downscaling for the 44 years period from 1958 to 2001 at daily scale.

At first we compare the model to observations, particularly to MEDATLAS, a climatic compilation based on observations. Results show an agreement of the downscaling with respect to MEDATLAS for surface variables. The model is also able to reproduce deep water formation in all the fundamental sites of Mediterranean Sea.

Deep convection in the Gulf of Lions is then examined. The steps of the process and the different causes are priorised. The annual chronology of the mixed layer maximum fits well with litterature. Especially, the model shows to be a very adequate tool to follow meteorological events, as the storm over this area the 11th february 1999 and its effect on the mixed layer state and deepening that follows.

Finally, the variability of heat fluxes is also studied. Model simulation results are also compared to a dataset. The two first EOF are in close agreement and represent almost 60% of the total variability. For the first time, heat fluxes are decomposed into one stochastic part and one part proportional to SST over the Mediterranean, following the method developed by Frankignoul [2002]. Feedback coefficient is found to be negative all over the basin with patterns explained by the dynamics of the different regions. Evaporation feedback drives at first order the heat flux feedback. Therefore, Mediterranean appears to be a region favourable to the Ocean-Atmosphere interaction.