The use of mean atmospheric parameters in the calculation of modeled mean surface heat fluxes over the world's oceans

Abstract

There are a number of atlases that display the distribution of ocean-atmosphere sensible and latent heat fluxes over various regions. Many are based on the "classical" method, in which time mean quantities are used in the bulk aerodynamic formulas, rather than the more accurate "sampling" method, which computes the mean of the instantaneous fluxes. Much of the justification for the use of this approximation comes from some studies conducted in the North Pacific and Atlantic oceans. How valid is it when applied globally? In this study we use large datasets, generated by January and July simulations of a general circulation model of the atmosphere, to examine comprehensively the differences between the two methods. We find that the ocean zonal averages of the two fluxes differ by less than 10 W m-2 at most latitudes in both months. However, at high southern latitudes in winter the sensible heat fluxes north of the Antarctic ice pack are up to 17 W m-2 (25%) greater when calculated with the sampling method. We show that the two methods differ due to four temporal covariances of various atmospheric quantities Some of the covariances contribute up to 12 W m-2 to the zonal mean fluxes. The fact that the classical and sampling methods give similar results over most of the globe is due, in large part, to the near cancellation of the covariances. However, this is not always true around the periphery of Antarctica.