Evaporation of water from soil and its transpiration by vegetation together form a ux between the land and the atmosphere called evapotranspiration (ET). ET is a key factor in many natural and anthropogenic processes. It forms the basis of the hydrological cycle and has a strong inuence on local climate, weather and numerous biophysical processes, such as plant productivity. As energy is required for ET to occur, it also forms a link between the land-surface energy uxes and water uxes.
Therefore, to be able to obtain reliable estimates of ET, reliable estimates of the other land-surface energy uxes, such as sensible heat ux, ground heat ux and net radiation, are also necessary. While it is possible to measure those uxes with ground-based instruments at local scales, at region scales they usually need to be modelled or estimated with the help of satellite remote sensing data. Even though this has been a topic of research for many years, a number of issues still persist. Not least, the mismatch between the spatial scales of measured and modelled uxes and the ability to model uxes consistently with input from different satellite sensors provided at varying spatial resolutions.

The objective of this study was to look at, and improve, various approaches for modelling the land-surface energy uxes at different spatial scales. The work was done using physically-based Two-Source Energy Balance (TSEB) approach as well as semi-empirical \Triangle" approach. The TSEB-based approach was the main focus of this study and the model was applied in the Skjern river catchment, in the west of Denmark's Jutland peninsula, in area covered by the Danish Hydrological Observatory (HOBE). Modelling was performed at a range of spatial scales, from a nominal resolution of 30 m to the whole river catchment, and the resultant uxes were compared to field based measurements and to the output of a well calibrated, physically-based distributed hydrological model. The "Triangle" approach was applied in semi-arid Spanish landscape at spatial resolutions ranging from 30 m to 4 km.
The study resulted in a number of improvements to the existing modelling approaches and provided enhanced understanding of the modelled processes. The TSEB -based Dual-Temperature-Difference (DTD) model was modified to allow its use in areas at high latitudes, such as Denmark, by removing its dependence on geo-stationary satellite observations. The performance of the DTD model was improved in forested ecosystems and during senescence, by taking into account the fraction of the vegetation that is green, as well as during dry conditions and in temperate climates, by modifying certain model formulations. A disaggregation algorithm was
also developed to increase the spatial resolution of the reliable DTD-modelled fluxes from 1 km to 30 m. Furthermore, synergies between remote sensing based models and distributed hydrological models were studied with the aim of improving spatial performance of the hydrological models through incorporation of remote sensing model outputs. Knowledge was also gained on the optimal choice of conditions and model options to achieve the most accurate results at different spatial scales with the "Triangle" modelling approach.