An air relative‐humidity‐based evapotranspiration model from eddy covariance data

Abstract

We present an air relative‐humidity‐based model of evapotranspiration (ARM‐ET) using relative humidity (RH) as a water stress factor to modify the Priestley and Taylor (1972) model potential evapotranspiration (PET) to actual evapotranspiration (ET). This model estimates ET using RH, air temperature (Ta), surface net radiation (Rn), and ground heat flux (G). The ARM‐ET model was calibrated with 8 day averaged ET observations at a flux site and validated with independent ET observations at 14 flux sites from a wide range of climates and vegetation types. Estimates of ET(Rn−G) with available energy defined as (Rn − G) had an R2 = 0.71 with a root mean square error = 23.62 W m−2 and a bias = 8.02 W m−2. As the eddy covariance method for ET measurements can suffer from an energy imbalance problem, we quantitatively analyzed the impact of such energy imbalance by replacing the available energy of Rn‐G with the sum of latent heat flux (λE) and sensible heat flux (H) in the modified ARM‐ET model, denoted ARM‐ET(λE+H). Further validation at 16 flux sites indicates that ARM‐ET(λE+H) had a more accurate estimate (R2 = 0.83, RMSE = 16.98 W m−2, and bias = 5.42 W m−2) compared with ARM‐ET(Rn−G), which shows that the energy imbalance problem did affect model accuracy. This evaluation also showed RH was more closely related to the evaporation fraction than to the vapor pressure deficit (VPD), soil water content, and RHVPD indicating that RH was superior to other water stress metrics in ET estimation. As a pure meteorological model with a simple parameterization, the ARM‐ET model is applicable over large spatial scales under a variety of climates and land cover types.