Quantifying the uncertainty in estimates of surface- atmosphere fluxes through joint evaluation of the SEBS and SCOPE models

Abstract. Accurate estimation of global evapotranspiration is considered of great importance due to its key role in the terrestrial and atmospheric water budget. Global estimation of evapotranspiration on the basis of observational data can only be achieved by using remote sensing. Several algorithms have been developed that are capable of estimating the daily evapotranspiration from remote sensing data. Evaluation of remote sensing algorithms in general is problematic because of differences in spatial and temporal resolutions between remote sensing observations and field measurements. This problem can be solved by using Soil Vegetation Atmosphere Transfer (SVAT) models, because on the one hand these models provide evapotranspiration estimations also under cloudy conditions and on the other hand can scale between different spatial resolutions. In this paper, the Soil Canopy Observation, Photochemistry and Energy fluxes (SCOPE) model is used for the evaluation of the Surface Energy Balance System (SEBS) model. SCOPE was employed to simulate remote sensing observations and to act as a validation tool. The advantages of the SCOPE model in this validation are (a) the temporal continuity of the data, and (b) the possibility of comparing different components of the energy balance. The SCOPE model was run using data from a whole growth season of a maize crop. It is shown that the original SEBS algorithm produces significant uncertainties in the turbulent flux estimations due to the misparameterizations of the ground heat flux and sensible heat flux. In the original SEBS formulation the fractional vegetation cover is used to calculate the ground heat flux. As this variable saturates very fast for increasing LAI, the ground heat flux is underestimated. It is shown that a parameterization based on LAI greatly reduces the estimation error over the season from RMSE = 25 W m−2 to RMSE = 18 W m−2. The uncertainties in the sensible heat flux arise due to a misparameterization of the roughness height for heat. In the original SEBS formulation the roughness height for heat is only valid for short vegetation. An additional parameterization for tall vegetation was implemented in the SEBS algorithm to correct for this. This improved the correlation between the latent heat flux predicted by the SEBS and the SCOPE algorithm from −0.05 to 0.69, and led to a decrease in error from 123 W m−2 to 94 W m−2 for the latent heat, with SEBS latent heat being consistently lower than the SCOPE reference. In addition the stability of the evaporative fraction was investigated.