Comparison of Mediterranean sea level variability as given by three baroclinic models

We compare the results of three baroclinic models with the aim of evaluating their skills in reproducing Mediterranean long-term sea level variability. The models are an ocean-ice coupled forced global model (ORCA), a regional forced ocean model (OM8) and a regional coupled atmosphere-ocean model (MITgcm). Model results are compared for the period 1961-2000 against hydrographic observations for water mass properties and steric sea level, and against satellite altimetry data and a reconstruction for sea level. All models represent the temperature variability of the upper layers reasonably well, but exhibit a considerable positive drift in the temperature of the deep layers due to an imbalance between the surface heat flux and the heat flux through Gibraltar. OM8 and MITgcm simulate the process of dense water formation better than ORCA thanks to their higher resolution in the model grid and in the atmospheric forcings. Concerning sea level variability, MITgcm is the only model that simulates well the inter-annual sea level variability associated with the Eastern Mediterranean Transient. However, none of the models is able to reproduce other features that have clear signatures on sea level. The inter-annual variability of Mediterranean mean sea level is better reproduced by the ORCA model because it is the only one considering the mass contribution from the Atlantic. The lack of that component in the regional models is a major shortcoming to reproduce Mediterranean sea level variability. Finally, mean sea level trends are overestimated by all models due to the spurious warming drift in the deep layers.