Two-source energy balance modelling of evapotranspiration over complex terrain

In recent years the Adige basin, in northeastern Italy, has experienced extreme drought events which resulted in agricultural and water management issues. Evapotranspiration (ET), one major indicator of water use, plays an important role under drought conditions. On the other side, reliable estimates of ET are currently missing over mountainous and heterogeneous areas such as the Adige basin. This study aims to provide an estimate of ET which will be of benefit both to research and agricultural services. In the framework of the project RETURN (Multi-risk science for resilient communities under a changing climate) and of the Italian National Drought Hydrological Monitoring System, we estimate daily ET at high spatial resolution (below 100 m) over the Adige catchment for the period 2017-2022. Remote sensing has been widely used to compute spatially distributed ET maps from thermal infrared datasets. In particular, the two-source energy balance (TSEB) model has proven to perform well over different land types and climates. An implementation of TSEB was already developed to estimate high-resolution ET from Copernicus globally available products (Sen-ET), adopting the Sentinel-3 and Sentinel-2 constellations for estimating fine-scale land surface temperature. Moreover, the ERA5 reanalysis data and the Climate Change Initiative land cover map have been used to retrieve solar radiation and vegetation structural parameters. However, the use of these datasets presents some shortcomings over such a complex area as the Adige basin, mainly due to their coarse spatial resolution. In this study, Sen-ET is adapted for complex terrains by replacing the ERA5 solar radiation data, available at 30 km, with the Meteosat Second Generation (MSG) radiation dataset (3.5 km) and, additionally, substituting the current land cover map (300 m) with the 100 m grid size Corine Land Cover product. Furthermore, a correction factor is applied to the radiation dataset to consider topographic shading, slope, and aspect. A comparison with daily aggregated global solar radiation from 79 weather stations in the Alpine region, covering a wide range of elevations, resulted in an R2 of 0.95 and 0.75 for MSG and ERA5 respectively, showing that this approach could greatly improve the reliability of ET estimation. Future steps will focus on the impact of changing radiation and land cover input data on the accuracy of modelled ET. The results will be validated against observations at Eddy-covariance sites.