In practice, the interactive feedback of the atmosphere and the ocean at that scale is often neglected. The necessary ocean surface data is taken from an external data set, for example, a global climate simulation or a sea surface data analysis. However, examining the atmosphere separately would yield an incomplete picture of the real climate system, because the links between the different climate system components selleck compound would be missing. The use of prescribed surface ocean data might lead to an inaccuracy of the model results. For instance, Kothe et al. (2011) studied the radiation budget in the COSMO-CLM

regional climate model for Europe and North Africa using ERA40 reanalysis data (Uppala et al. 2005) as the lower boundary forcing. The authors evaluated the model outputs against re-analysis and satellite-based data. The results show an underestimation of the net short wave

radiation over Europe, and more considerable errors over the ocean. Because the lower boundary condition was prescribed with ERA40, these errors in radiation over the ocean could be due to wrongly assumed albedo values over ocean and sea ice grids. In the same way, ocean models often use atmospheric forcing datasets without active feedback from the atmosphere. Griffies et al. (2009) investigated the behaviour of GKT137831 datasheet an ocean-sea-ice model with an atmospheric data set as the upper boundary condition. In that study, the difficulties in using a prescribed atmosphere to force ocean-sea-ice models are recognised. First of all, it is very often the case that atmospheric forcing datasets may not be ‘tuned’ specifically for the purpose of an ocean-sea-ice model experiment. For example, the above study used global atmospheric forcing data for the ocean and sea-ice model from Large & Yeager (2004). However, this dataset was originally evaluated over the ocean, not over sea ice and, thus, gives better results over open water. Moreover, the authors also demonstrated that the error consequent upon decoupling the ocean and sea ice from the interactive atmosphere could be large. One problem that is very likely to crop up is the error in the ocean salinity, due to the fresh water inflow,

especially precipitation. The prescribed Fenbendazole precipitation can cause a dramatic drift in ocean salinity. The second problem is the error in sea-ice area, which can lead to a wrong balance of the Earth’s radiation and an unrealistic heat transfer between atmosphere and ocean. The findings from this paper show the necessity of giving an active atmosphere feedback to the ocean instead of using a forcing dataset. The ocean-atmosphere interaction has been taken into account in many AOGCMs (Atmosphere-Ocean General Circulation Models), as shown in Giorgi (2006). However, on a global scale, the local characteristics of marginal seas cannot be resolved (Li et al. 2006) and these seas are, in fact, not well represented by AOGCMs (Somot et al. 2008).