This thesis describes the collection, analysis and interpretation of data on sensible heat transfer into the atmosphere from three sparsely vegetated sites: a vineyard and legume crop in Spain, and a savannah bush site in Niger. The viability of using measurements of the variance of air temperature to determine sensible heat flux is investigated. Monin-Obukhov similarity theory is reviewed, with particular emphasis on its limitations and asymptotes. The physical interpretation and relative magnitudes of zero-plane displacement and roughness length for heat and momentum transfer are examined. A heat flux measurement system based on measuring fluctuations of temperature is developed, and its results compared with eddy correlation measurements in both Spain and Niger. The surface temperatures of the various components of a sparsely vegetated surface are compared with the area-averaged radiometric temperature. It is shown that, for sparsely vegetated surfaces, this radiometric surface temperature may be much lower than the "aerodynamic" surface temperature required by similarity theory. Values of Z_0h, the roughness length for heat transfer, are calculated for all three sites. These are found to be smaller than for uniform vegetation, and to vary with wind speed by several orders of magnitude. A review of the physics of heat and momentum transfer is undertaken to explain this phenomenon. It is demonstrated that Z_0h is a function of both heat and momentum transfer, and that it can be expected to vary with wind speed. This effect is masked for uniform vegetation, but is amplified for sparse vegetation by the relatively low efficiency of heat transfer. This explains the observed behaviour of Z_Oh on the three field sites. Historical measurements of Z_0 and Z_0h over a variety of surfaces are reviewed and reinterpreted in terms of heat and momentum transfer properties.