The evaluation of the suitability of potential evapotranspiration models for drought monitoring based on observed pan evaporation and potential evapotranspiration from eddy covariance

Abstract

Potential evapotranspiration (PET), representing atmospheric evaporative demand, is a critical variable in drought monitoring and prediction. Currently, more than 100 PET models are available, and the choice of models can lead to large uncertainty in drought monitoring. Here, we are aiming to evaluate the suitability of widely used 33 PET models for drought monitoring against the pan evaporation measurements (Epan) and water-free stress evapotranspiration from eddy covariance observations (EC ET_unstr), for different climate zones of China. We found that the optimal PET models are different for different climate zones as well as drought types. The temperature-based models such as Romanenko and Schendel, along with the mass-transfer-based Brockamp-Wenner model, were most effective for meteorological drought monitoring (monthly Taylor Skill Score, TSS > 0.72) in arid and semi-arid zones, while the radiation-based Irmak model demonstrated high accuracy for agricultural drought monitoring (TSS > 0.67). The radiation-based Jensen-Haise model (monthly TSS > 0.83) and the McGuinness-Bordne model (TSS > 0.47) were suitable for meteorological and agricultural drought monitoring in the humid and semi-humid zones, respectively. The combination and radiation-based models proved more effective for agricultural drought monitoring than the temperature- and mass-transfer-based models because they consider more vegetation effects on PET. The validation through historical drought events further confirmed that the optimal PET models can capture drought events and dynamics. We also found that weaker meteorological droughts in China may also lead to higher agricultural drought risk. This study offers critical guidance on selecting PET models for drought monitoring, and emphasizes the need for further optimization of PET model.