Topography-Based Estimation of Evapotranspiration at High Altitudes in Semi-arid Regions

Assessing the surface water balance of mountains is a real challenge given notably the extreme variability of meteorological conditions and the sparsity of in-situ monitoring. While mountains are recognized as water towers feeding the surrounding plains, there is only unconsolidated knowledge about the individual water balance components especially the evapotranspiration (ET). Satellites land surface temperature (LST) along with air temperature (Ta) and incoming solar radiation (Rg) can be used to assess the energy budget and provide a reasonable estimation of instantaneous ET. Nevertheless, over mountains, the Ta and Rg, respectively, undergo strong topographical changes due to elevation and sun exposure effects. Moreover, upscaling the instantaneous ET to its daily value is expected to be uncertain in mountains as the evaporative fraction (EF, defined as the ratio of ET to available energy ratio) of a given pixel can no longer be considered constant during daytime until proven otherwise. In this context, this contribution focuses on a topography-based estimation of ET using the two-source energy balance (TSEB) model. We also examine the variability of hourly and daily EF estimates through both satellite and in-situ monitoring. An eddy covariance tower was installed at 3850 m.a.s.l over the High Atlas Mountains in central Morocco and has been operating since September 2020 to present. The 30 m resolution LST is derived from thermal data collected by Landsat-7, 8, and 9 on clear sky days. Rg is estimated at the Landsat (30 m) resolution from the SRTM’s digital elevation model (DEM) and two different topography-based approaches: a physically based model (DART) and a simplified semi-empirical model. The 9 km resolution ERA5-Land’s air temperature product is spatialized at the same (30 m) resolution by applying the environmental lapse rate (ELR) retrieved at the Landsat overpass time over a 9 km2 area including the eddy covariance tower. Satellite-derived estimates of ET and EF are compared to instantaneous station measurements for three and nine dates in 2020 and 2021, respectively. The variability during daytime of the in-situ EF is also assessed to evaluate the potential for upscaling instantaneous remotely sensed ET to a daily scale.