IPSL-CM1

The first version of the model (Braconnot et al. 1997) was build with the LMD5.3 version of the LMD atmospheric model, the OPA 7 version of the ocean model developed at LODYC, and a simple sea ice component. The first coupled (Braconnot et al. 1997) simulations exhibited a large drift in surface air temperature, which has been attributed to an energetic imbalance of the atmospheric model. Several adjustments were performed. A first set of modifications has improved the atmosphere boundary layer over mixed sea ice and ocean grid cells. A second set was designed to equilibrate the atmospheric model, by adjusting the threshold for vertical diffusion in stable cases, which allowed for better simulation of temperature inversions in high latitude (Krinner et al. 1997). Other adjustments concern the drag coefficient in stable cases and the balance between long wave and short wave radiation at the top of the atmosphere (reduction of the water droplet size from 15 µm to 8 µm . Although these adjustments allow stable coupled simulations, the energy absorbed in the tropical region was underestimated, which explains the cold tropical bias in the tropics (Braconnot et al. 2000). Mid-latitudes experienced a reversed bias. These characteristics improved with a new version of the Morcrette et al. (1986) radiative scheme. The hydrological cycle was also closed in this revised version, thanks to a simple routing scheme that considered the 46 major rivers (Le Clainche 1996). Following the work of Guilyardi et al. (2001), isopycnal diffusion was implemented in the ocean model. A revised version of the interpolation scheme (Marti et al. 2005) also contributed to the conservation of energy at the air-sea interface.