Assessing the uncertainty of the estimation of potential evapotranspiration under climate change using a pseudo-reality approach

Evapotranspiration is an important process in the water balance. It accounts for water transported from the surface of the Earth to the atmosphere. For practical applications, the estimation of evapotranspiration is limited by the potential evapotranspiration (PET), which is the maximum evapotranspiration that occurs when the availability of water on the surface is unlimited. PET can be estimated based on different climate variables. When assessing climate change projections, different climate models often project different change directions for such variables. Furthermore, the reliability of the climate model projections varies for each variable. Therefore, the uncertainty associated with the estimation of PET can be large. This study assesses this uncertainty by employing ten methods commonly used to estimate PET for ten different locations across Europe, capturing different climate and physical conditions. An ensemble of ten Euro-CORDEX climate models is used to assess projected PET changes through the century under the RCP 8.5 scenario. Different climate model bias correction methods are employed to reduce the biases in the climate model outputs when compared to the reference climate. A pseudo-reality experiment is set where each climate model acts as reference to train the correction methods, which are applied to the climate models that remain in the ensemble. The uncertainty of the projected PET is compared to the reference-climate-PET using evaluation metrics. Results are relevant for decision and policy makers and professionals developing impact studies.