The growing concern about environment has increased the number of land surface processes studies. Computer simulation models of land surface processes have been developed for a range of scales and with different levels of physical complexity. Because the interactions between soil, vegetation and atmosphere vary both spatially and temporally, regional evaporation in heterogeneous natural landscapes is difficult to predict by means of computer simulation models. Remote sensing measurements of land surface radiative properties offer however a means to indirectly measure land surface state conditions at a range of scales. A straightforward estimation of evaporation from radiative properties of the land surface is hampered by the fact that only a very few parameters of the classical flux-profile relationships can be estimated directly from remote sensing measurements. Moreover, the accuracy of surface temperature measurements necessary to solve flux-profile relationships is still poor. Inclusion of ground measurements is a possible solution, but the absence of such data on large scales and for heterogeneous land surfaces where these parameters are not measured, forms an immediate obstacle for the implementation of remote sensing algorithms. A Surface Energy Balance Algorithm for Land (SEBAL) has been developed in a way that the need for collateral measurements is partly eliminated, a very accurate surface temperature map is no longer required (although it should be as good as possible) and a land use classification to relate surface temperature to evaporation is not needed. Each pixel is characterized by a surface hemisherical reflectance, surface temperature and a vegetation index. The methodology composes of multiple flux-profile relationships for small sub-areas. Although the concept has a physical basis, the parameters are estimated by empirical relationships, for instance a relationship between near-surface vertical air temperature difference and surface temperature forms an essential component in the estimation of the sensible heat flux density. The absolute surface energy balance terms are estimated on an instantaneous time basis. Temporal integration of instantaneous surface flux densities is feasible using the evaporative fraction (latent heat flux density/net available energy): The evaporative fraction remains fairly constant during daytime hours for both homogeneous and heterogeneous areas. A physical explanation for this is given. A bulk surface resistance of a heterogeneous landscape has been related analytically to canopy and bare soil evaporation resistances. Measurements in central Spain have shown that the evaporative fraction and bulk surface resistance are suitable indicators to describe areal patterns of near-surface soil water content. Although the bulk surface resistance has a distinct diurnal variation, it is much less affected by changes in net available energy and therefore preferred to describe the energy partitioning for longer time series (weeks, months). SEBAL has been validated with data available from regional evaporation projects in Egypt and Spain. The error of high resolution evaporative fraction estimations decreases from 20% to 10% at a scale of 1 km to 5 km respectively, The error of low resolution evaporative fraction images at a scale of 100 km is approximately 1 to 5 %. Hence, the error averages out if a larger set of pixels is considered. It is concluded that the uncertainty of evaporation in regional water balances and model studies can be substantially reduced by estimating evaporation with remote sensing measurements using the proposed SEBAL technique.
