Atmospheric water vapor fluxes were derived from a 1-year data set of horizontal wind speed and specific humidity assimilated from meteorological observations by the European Center for Medium-Range Weather Forecast (ECMWF). Vertically integrated horizontal freshwater fluxes were compared to those of two data sets based on a climatology [Oort, 1983] and on simulations with an atmospheric general circulation model (AGCM). Zonal transports agree fairly well at all latitudes outside the tropics, where fluxes are about double for the AGCM data set. Meridional fluxes of the AGCM and ECMWF data sets show close agreement, while the cliniatological fluxes are generally smaller with a considerable northward shift in the southern hemisphere. Atmosphere-to-ocean freshwater fluxes were derived from the three data sets. Not only is there substantial disagreement between the data sets, but their zonal averages over the Atlantic, Pacific, and Indian Ocean basins show Little resemblance to the respective restoring freshwater fluxes from a 2-dimensional ocean model. If the ocean model is forced with the observed and modeled atmospheric fluxes, we find that the mode of ocean circulation is determined mostly by the net flux to the high-latitude oceans and the amount of freshwater exported from the Atlantic basin. The latitudinal structure of the freshwater fluxes in low-latitudes and midlatitudes has little influence on the modeled thermohaline circulation. The fluxes derived from the climatology and ECMWF permit North Atlantic Deep Water (NADW) formation, but a strong freshwater input to the Southern Ocean inhibits Antarctic Bottom Water formation. The AGCM transports so much moisture to the Arctic Ocean that NADW formation is shut down, resulting in a ocean circulation mode of southern sinking in all three ocean basins. If NADW is formed in the model, the strength of the Atlantic meridional overturning is determined by the net freshwater export from the Atlantic basin. When this export is artificially increased in the model over a range from 0.2 to 1 Sv, the ratio of overturning to freshwater forcing decreases almost linearly.
Nt504Times Cited:28Cited References Count:37