EDDY-PERMITTING OCEAN CIRCULATION HINDCASTS OF PAST DECADES The DRAKKAR Group: Barnier B., Brodeau L., Le Sommer J., Molines J.-M, Penduff T., LEGI-CNRS, Grenoble, France. Theetten S., Treguier A.-M., LPO, Brest, France. Madec G., LOCEAN, Paris, France. Biastoch A., Böning C., Dengg J., IFM-GEOMAR, Kiel, Germany. Gulev S., SIO-RAS, Moscow, Russia. Bourdallé Badie R., Chanut J., Garric G., MERCA- TOR-Océan, Toulouse, France. Alderson S., Coward A., de Cuevas B., New A., NOC, Southampton, UK. Haines K., Smith G., ESSC, Reading, UK. Drijfhout S., Hazeleger W., Severijns C., KNMI, De Bilt, The Netherlands. Myers P., DEAS, Ed- monton, Canada. Corresponding author: Bernard.Barnier@hmg.inpg.fr 1. Introduction Research conducted by the DRAKKAR consortium is motivated by open questions related to the variability of the ocean circulation and water mass properties dur- ing past decades, and their effects on climate through the transport of heat. Of primary concern is the circula- tion and the day-to-decade variability in the North At- lantic Ocean, as driven by the atmospheric forcing, by interactions between processes of different scales, by exchanges between basins and regional circulation fea- tures of the North Atlantic (including the Nordic Seas), and by the influence of the world ocean circulation (in- cluding the Arctic). DRAKKAR carries out these inves- tigations using a hierarchy of high resolution model configurations based on the NEMO system (Madec, 2007). Simulation outputs are carefully evaluated by comparison with collocated existing observations (Pen- duff et al., this issue). The DRAKKAR consortium was created to take up the challenges of developing realistic global eddy- resolving/permitting ocean/sea-ice models, and of building an ensemble of high resolution model hind- casts representing the ocean circulation from the 1960s to present. The Consortium favours an integration of the complementary expertise from every member of the group; the coordination of a simulation program that builds a consistent ensemble of 50 year long hindcasts; and an increase of available manpower and computer resources. 2. DRAKKAR hierarchy of models A hierarchy of embedded model configurations of dif- ferent grid resolution (from coarse to eddy-resolving) has been constructed to make possible realistic, long term (several decades) simulations of the ocean/sea-ice circulation and variability at regional and global scale, and to perform sensitivity studies investigating key dynamical processes (requiring especially high resolu- tion) and their impact at larger scales. The DRAKKAR model configurations are used by the participating re- search teams to address their scientific objectives. All configurations are based on the NEMO Ocean/Sea-Ice GCM numerical code and use the quasi-isotropic global ORCA grid (Madec, 2007). 2.1. Global ORCAii configurations Global DRAKKAR configurations span resolutions of 2° (ORCA2), 1° (ORCA1), 1/2° (ORCA05) and 1/4° (ORCA025, Fig. 1). Fig. 1: ORCAii global configurations. The model domain and land-sea mask are shown for the ¼° ORCA025 configuration (axes correspond to grid points). Colours show a SSH snap- shot (in meter) on June 24, 1998 from one of the hindcast runs (G70), sea-ice cover being in white. Boxes show the do- main decomposition on a large number of processors, ocean processors (not marked by a cross, 186 of those) being the only ones retained in the calculation. On vector computers a more moderate parallelization (typically up to 32 processors) is used. The red box is the region where a 2-way grid refine- ment at 1/10-1/12°is being implemented. The targeted configuration for the ensemble of hind- casts is the eddy permitting ORCA025, extensively de- scribed in Barnier et al. (2006). Such eddy-permitting models are still worth exploring and enhancing, since they will be the target resolution of the next generation of climate models. The ORCA grid becomes finer with increasing latitudes, so the effective 1/4° resolution is 27.75 km at the equator and 13.8 km at 60°S or 60°N. It is ∼7 km in the center of the Weddell and Ross Seas and ∼10 km in the Arctic. In the vertical, there are 46 levels