Remote Sensing (SR) is the most efficient way to monitor the Earth's surface and obtain regional evapotranspiration estimates (ET). ET has been obtained by combining satellite observations in the various electromagnetic spectrum bands with values of climate variables measured at the surface level. In large-scale applications, physically-based methods are preferred, but they often depend on excessive simplifications of aerodynamic resistance or on locally tuned parameters. These solutions are efficient, but they limit the performance of SR models spatially. We developed and evaluated two RS models for estimating evapotranspiration in non-rainy days, and rainfall interception loss in rainy days. We modelled the aerodynamic resistance for heat transfer (rah) and identified the differences between using physical or empirical simplifications. The impact on the evapotranspiration component of the intercepted loss was obtained by analysing the vegetation storage capacity (Sv) and canopy cover fraction (c). Our analyses exhibited contrasting performances. For the non-rainy days the parameters of momentum roughness length (z0m), the excess resistance between heat and momentum transfer (kB-1), and the representation of plant structure by the Plant Area Index improved ET estimates. However, the parametrization of the rainfall interception model using the equations identified in the literature for determining Sv and c using RS data exhibited poor performance. Our results support that using physically meaningful parameters in RS modelling can yield better results rather than empirical parameters. Although our interception model had poor performance, its development is critical to achieve ET models that provide accurate ET partitioning estimates.
The 28th IUGG General Assembly (IUGG2023) (Berlin 2023)