Towards a universal evapotranspiration model based on optimality principles

Abstract

Natural resource management requires knowledge of terrestrial evapotranspiration (ET). Most existing numeric models for ET include multiple plant- or ecosystem-type specific parameters that require calibration. This is a significant source of uncertainty under changing environmental conditions. A novel ET model with no type−specific parameters was developed recently. Based on the coupling the diffusion (via stomata) of water and carbon dioxide (CO2), this model predicts canopy conductance based on environmental conditions using eco-evolutionary optimality principles that apply to all plant types. Transpiration (T) and ET are calculated from canopy conductance using the Penman-Monteith equation for T and a universal empirical function for the T:ET ratio. Here, the model is systematically evaluated at globally distributed eddy-covariance sites and river basins. Site-scale modelled ET agrees well with flux data (r = 0.81, root mean square error = 0.73 mm day–1 in 23,623 records) and modelled ET in 39 river basins agrees well with the ET estimated by monthly water budget using two runoff datasets (r = 0.62 and 0.66, respectively). Modelled global patterns of ET are consistent with existing global ET products. The model's universality, parsimony and accuracy combine to indicate a broad potential field of application in resource management and global change science.