Vegetation is responsible for more than 50% of the land-atmosphere water vapour fluxes.It affects both the atmospheric moisture content and heat by its control on flux partitioning into latent and sensible heat. The convective exchange of latent and sensible heat is generally assumed to be driven by wind (forced convection). To test this assumption, we analyzed the dependence of the heat transfer coefficient (HTC) on wind speed using eddy covariance datafrom sites along the North Australian Tropical Transect. Here we deduced the HTC from observed sensible heat flux, air temperature and radiometric surface temperature. In most cases, we did not find a significant correlation between HTC and wind speed. In contrast, many cases seemed to suggest a significant dependence of HTC on the surface-air tem-perature gradient (DT) itself. This would implicate buoyancy-driven vegetation atmosphere exchange rather than wind-driven. However, we found that the correlation between HTC and DT is very sensitive to the formulation used for radiometric surface temperature estimation and the assumed surface emissivity. In fact, in many cases, surface temperatures deduced from radiometric data were unrealistic when compared with measured air temperatures and sensible heat flux, while realistic values could only be achieved with very low surface emissivity values that are well outside the commonly used range. This analysis raises questions about our fundamental understanding of radiative surface energy exchange and/or the measurements of sensible and radiative heat fluxes.