Uncertainties in using remote sensing for water use determination: a case study in a heterogeneous study area in South Africa

Abstract. South Africa is a water scarce country where it is important for water managers to have accurate information on water resource occurrence and use. A remote sensing project highlighted many uncertainties in using complex remote sensing models to determine water use in a heterogeneous study area. The severity of the uncertainties was confirmed as the results across the catchment showed a higher total evapotranspiration than precipitation. This paper illustrates some of the uncertainties and limitations using the evapotranspiration component of the water balance as calculated by the Surface Energy Balance System (SEBS) model, as an example. The introduction of uncertainties in the derivation of evapotranspiration were identified as: (1) sensitivity to land surface and air temperature gradient; (2) the choice of fractional vegetation cover formula; (3) height of wind speed measurement in relation to displacement height indicating a maximum canopy height at which the SEBS model should be used; and (4) study area heterogeneity. Uncertainties and errors are compounded when considering that the SEBS model is a complex model, requiring several image processing sequences that are combined to produce the final result. It was shown how the production and propagation of errors in the SEBS model can contribute to uncertainties in flux estimation and ultimately to uncertainties in the estimation of actual evapotranspiration.