A global heat and freshwater forcing dataset for ocean models

A global dataset based on the ECMWF Re-Analyses (ERA) is presented that can be used as surface boundary conditions for ocean models with sea-ice components. The definition of these conditions is based on bulk formulae. To study the mean ocean circulation, a mean annual cycle on a daily basis was constructed from ERA for all relevant parameters including wind stress. Continental runoff is considered by using information about the catchment areas of the rivers and about the main drainage basins. The bulk formulae were extended by using sea ice concentration.To estimate meridional heat transports (MHT) and to avoid any drift in ocean model simulations, the heat and fresh water budgets have been closed by applying an inverse procedure to fine-tune the fluxes towards observed transports. To improve the MHTs on the Southern Hemisphere the winds and the short wave radiation at southern higher latitudes should be corrected. Furthermore, tests were performed concerning short wave radiation which was increased in the tropics and decreased in the subsidence zones.The heat and fresh water fluxes are assessed by using a scheme of Macdonald and Wunsch based on hydrographic sections. The net heat fluxes of ERA and of the forcing dataset are consistent with the heat flux divergences and convergences estimated by this scheme except for parts of the South Atlantic and the Indian Ocean sector of the Southern Ocean where none of these datasets is consistent with these estimates. In the subtropical South Indian Ocean the forcing dataset is consistent with these estimates while ERA are not. The flux components of ERA and the forcing dataset were compared to several observational datasets (SRB, SOC, HOAPS, GPCP, and CMAP). For each component, at least one of these datasets (especially HOAPS) supports the effects of the inverse procedure and the bulk formulae almost globally with some regional exceptions: short wave radiation in the tropical oceans and the subtropical North Atlantic, latent heat flux at higher latitudes, and precipitation in the northern North Atlantic.Comparisons to the NCAR/NCEP Re-Analyses (NRA) (versions 1 and 2) and the ECHAM model in place of ERA lead to similar results. In the North Atlantic the net heat fluxes of the model based datasets approach the hydrographic estimate with increasing resolution. Applied to any ocean/sea-ice model and compared to ERA, the forcing dataset would induce only a relative small net sea-surface buoyancy loss.A comparison of the forcing dataset to measurements made using one buoy deployed in the western Pacific warm pool and five buoys deployed in the subduction region of the Northeast Atlantic shows that at the site of the first buoy the net heat fluxes of the forcing dataset are in poorer agreement than those of ERA. At the sites of two subduction buoys both datasets show the same level of agreement within the error bars specified. At the sites of the three remaining subduction buoys the forcing dataset shows a marginal improvement on ERA.