Deriving potential evapotranspiration from satellite-based reference evapotranspiration, Upper Tekeze Basin, Northern Ethiopia

Study region:
Zamra catchment, Upper Tekeze Basin, Northern Ethiopia.

Study focus:
Direct field estimates of potential evapotranspiration (PET) remain a challenge, because they are expensive and have limited spatial representativeness. An alternative is to derive PET by remote sensing as a product of reference evapotranspiration (ETo) and land use-land cover (LULC) factor (Kc). The aims of this study were to: i) validate satellite-derived Daily Reference Evapotranspiration (DMETREF-ETos); ii) correct the advection-bias of DMETREF-ETos as compared to in-situ FAO Penman-Monteith ETo (FAO-ETog); and iii) convert the bias-corrected DMETREF-ETos into PET. Four ATMOS 41 weather stations were installed and 1-year daily data was used to validate the DMETREF-ETos. The wet season actual evapotranspiration from FAO Water Productivity Open-access portal (WaPOR) was used to validate PET.

New hydrological insights for the region:
Evaluation of DMETREF-ETos showed large underestimation biases when compared to FAO-ETog derived from in-situ weather stations, particularly in warm conditions, due to unaccounted advection-bias. The bias correction was performed as a function of near-surface air temperature obtained from ERA5-Land, merged with in-situ air temperature; the results showed substantial improvement of ETo. Finally, the daily PET was estimated by multiplying the corrected DMETREF-ETos by a spatio-temporally variable Kc, defined as linearly dependent on the NDVI obtained from Sentinel 2. This study emphasized substantial difference between ETo and PET, implying that the ETo cannot substitute PET, which in many hydrological studies is still a common practice.