1. The role of Mediterranean mesoscale eddies on the climate of the Euro-Mediterranean region by A. Bellucci 1, S. Gualdi 1,2, E. Scoccimarro 2, A. Sanna 1, P. Oddo 2, and A. Navarra 1,2 contact: alessio.bellucci@cmcc.it 1. CMCC – Centro Euro-Mediterraneo per i Cambiamenti Climatici (Euro-Mediterranean Centre for Climate Change), Bologna, Italy 2. INGV – Istituto Nazionale di Geofisica e Vulcanologia (National Institute of Geophysics and Volcanology), Bologna, Italy Istituto Nazionale di Geofisica e Vulcanologia Introduction and Motivations Within the CIRCE (Climate Change and Impact Research: The Mediterranean Environment) EU Project, substantial efforts were devoted to enhance the representation of the oceanic system in the Mediterranean region. This was achieved by developing coupled general circulation models with ocean components which either explicitly resolve, or simply permit, mesoscale circulation features. The inclusion of the eddy variability tail in the spectrum of the processes resolved by the modelled system represents a particularly relevant step forward with respect to the previous CMIP3 generation of climate models, as these were systematically based on coarse resolution ocean components, leading in turn to an extremely rough representation of the Mediterranean Sea sub-system. In this study the role of mesoscale oceanic features on the air-sea interactions over the Mediterranean region was analysed, in the context of one of the CIRCE ensemble of climate models. To this aim, two different simulations of the 20th Century climate, performed with two distinct configurations of the CMCC coupled general circulation model featuring radically different horizontal resolutions in the Mediterranean Sea domain, were compared. This comparison highlights the implications deriving from the inclusion of energetic ocean mesoscale structures in the variability spectrum of the coupled ocean-atmosphere system and points to the need for high- resolution ocean components in the development of next generation climate model. Fig.1 : Daily SST (1 st Jan 1960) snapshots over the Med Sea from experiments (top) H, and (bottom) L. OGCM: OPA 8.2 (GLOBAL) NEMO (Med Sea) AGCM:ECHAM5 Global SST Models and experimental setup Two 20C3M simulations forced with historical timeseries of GHG, aerosol & volcanoes are performed, using two global CGCMs, only differing by the ocean space resolution over the Med Sea region (see Fig.1): Exp. H : ECHAM5 T159L31 + OPA8.2 (global ocean) + LIM2 + NEMO 1/16 o (Med Sea) Exp. L : ECHAM5 T159L31 + OPA8.2 (global ocean) + LIM2 Exp. H (turbulent) Exp. L (laminar) OGCM: OPA 8.2 (GLOBAL) Med Sea SST AGCM:ECHAM5 Global SST Exp. H (turbulent) H L In this study, two numerical simulations of the 20 th Century climate performed with two global GCMs are analysed. In the first experiment (L), a T159 atmosphere (equivalent to ∼ 80 Km horizontal resolution) is coupled to a 2x2 o global ocean model, with a locally enhanced 1 o resolution over the Mediterranean Sea region. In the second experiment (H), the same T159 atmosphere is coupled to a global ocean model, except over the Mediterranean Sea where a regional high-resolution 1/16 o ( ∼ 7 Km) ocean model is used, which is in turn coupled to the low-resolution global OGCM at Gibraltar Strait (CMCC-Med; Gualdi et al. 2011). Thus, in H, as far as the Mediterranean area is concerned, the atmosphere is locally coupled to an ocean model which resolves mesoscale features (turbulent ocean), whereas in L the atmosphere interacts with a laminar oceanic system. Since these two experiments are identical except for the resolution of the ocean model over the Mediterranean Sea, the systematic comparison of H and L allows the assessment of the net effects on the climate of the Euro-Mediterranean region from explicitly resolving mesoscale oceanic features in the coupled model. Spectral analysis in the wavenumber domain (Fig.2). Power spectra of surface temperature in the wavenumber domain were computed for both H and L experiments using daily zonal transects in the Eastern Mediterranean basin, over a 4 years long period (Fig.2). The spectra were diagnosed from both the ocean model (sea-surface temperature; SST) and the atmospheric model (surface air temperature over ocean grid-points; SAT) counterpart. The simple comparison between SST and SAT power spectra for experiment H highlights the existence of an upper cut-off wavenumber set by the atmospheric resolution, which inhibits the direct transfer of spatial SST variance from the ocean to the atmosphere for wavelenghts shorter than the smallest spatial scale resolved by the atmosphere (80 Km). Surface temperature fields display a typical k -m power-law shape, i.e. with energy decaying for larger wavenumbers. In H, m approximately fits the theoretical -5/3 law of two-dimensional turbulence within the 500-100 Km range, while a steeper slope is revealed for the smaller-scale dissipative range (SST power spectrum). On the other hand, in L the ocean and the atmosphere share a much similar horizontal resolution (80 and 111 Km, for the atmosphere and the ocean GCM, respectively). Interestingly, SST (not shown) and SAT power spectra display a steeper slope and consistently lower energy in the 100-500 Km range with respect to H. Thus, the absence of a developed ocean eddy field in L seems to affect the long- wave part of the common ocean-atmosphere variability spectrum. The inclusion of a vigorous oceanic eddy field in the coupled system appears to indirectly affect the large scale part of the variability spectrum. This may possibly occur through the non-linear eddy-large scale interactions taking place in the high-resolution ocean component. In particular, the upscale energy transfer, which typically takes place in two-dimensional turbulent fluids (such as the ocean) may play a role in this process. Fig.2 : Wavenumber spectra for SST and atmospheric surface temperature from experiments L and H from daily zonal transects in the Eastern Mediterranean basin, from both the ocean and atmospheric model. A constant slope theoretical k -5/3 spectrum is also shown. Patterns of H-L mean state differences (Fig.3 Left) reveal an overall 1 o K warming impact of the enhanced ocean horizontal resolution over the Med Sea,. Consistent H-L patterns of enhanced evaporation (not shown) and latent heat losses also emerge. The comparison of model SST climatology with HadISST over the Mediterranean region reveals a substantial SST bias reduction in the high-resolution H experiment, with respect to experiment L. Fig.3 : Left H-L difference between long term climatologies for (top) surface air temperature (colour; o C) (bottom) latent heat flux (W/m 2 ). Right SST climatology Model-OBS difference for (top) H and (bottom) L. HadISST data were used as SST OBS. Impact of Med Sea horizontal resolution on surface european climate:bias reduction. Acknowledgements This work was funded by the EU FP7 CIRCE (Climate Change and Impact Research: the Mediterranean region and the global climate system ) Integrated Project. Med Sea interannual variability Fig.4 Power spectra of Mediterranean basin-averaged SSTs reveal enhanced variability around interannual time-scales in the eddy resolving H experiment (black), while the control L experiment (green) shows a red-noise shaped structure, with enhanced power at lower frequencies. AR1 95% confidence levels are also shown. Concluding remarks The inclusion of a vigorous eddy field in the ocanic component of a coupled climate model substantially alters both the mean state of the system and its space and time variability. This comparison points to the need for high-resolution ocean components in the development of next generation climate models. References Gualdi and Coauthors, 2011: The CIRCE simulations: a new set of regional climate change projections performed with a realistic representation of the Mediterranean Sea, to be submitted to BAMS.