Validation of Reference Evapotranspiration from Meteosat Second Generation (MSG) Observations

This article presents validation results of the grass Reference Evapotranspiration (ETo) product, which is provided operationally by the Satellite Applications Facility on Land Surface Analysis (LSA SAF).
ETo is considered to be the evapotranspiration from a hypothetical extensive well-watered field covered with green grass (12 cm high and with 0.23 albedo), under the given down-welling short-wave radiation. The LSA SAF product is estimated from daily solar radiation at the surface via a methodology designed to be applicable to that well-defined reference surface. In line with its definition, the LSA SAF ETo product is based on estimates of the radiative energy available at the surface.
Validation of ETo estimates is challenging, given the difficulties of finding reliable in situ measurements that meet the ETo definition. We show that for the site that matches closely the reference surface (Cabauw, The Netherlands), the LSA SAF product outperforms other commonly used methodologies, even when these use in situ observations as input. However, observations taken at other stations, located in areas that do not deviate greatly from the reference surface, put into evidence the high uncertainty in local measurements. It is shown that the LSA SAF ETo product follows well in situ values, with average differences of 0.3 mm/day or less for mid-latitude sites and of the order of 1 mm/day for the Spanish sites near Cordoba and Albacete. Local advection effects cannot be ignored in measurements performed in the latter stations, where summer conditions are mostly warm and dry. In situ ETo measurements performed in those cases with lysimeters within limited fields are higher than satellite estimates due to advection of warm dry air from the vicinity, which acts as an extra source of energy. It is shown that this effect can be parameterized as a function of near surface air temperature. However, it is argued that those local advection effects should not occur in the idealized surface referred above.
In contrast to other ETo estimates, the LSA SAF product is not influenced by local aridity or advection effects, and therefore it is particularly appropriate for large scale climate assessments, including drought monitoring (e.g. by considering the ratio of actual and reference evapotranspiration). Additionally, it provides suitable estimates of irrigation requirements in support of water management.