Evapotranspiration (ET) is a critical process in the global terrestrial water cycle, which describes the actual water consumption from land surface. ET comprises the biotic process of transpiration (T) via stomata, and the abiotic processes of evaporation from wet leaves (interception) and bare soil. The Penman-Monteith (PM) equation, together with the surface energy balance theory, provides a direct way to estimate the ET. In the PM equation, the surface conductance is an important variable that cannot be detected by remote sensing. Traditional methods estimate the surface conductance by empirical or parameterization methods, which requires calibration and, therefore, produces uncertainty during model application.
At the core of our universal productivity model, the P model, is a description of optimal stomatal behaviour as a function of environment that is equally applicable to all C3 plants. We can then predict canopy conductance based on the P model and a physical model (the Fick’s Law) that describes the fusion velocity of H2O molecule. ET is finally estimated by the PM equation and an empirical function of T/ET that is based on local environment conditions. The comparison with weekly FLUXNET observation (Fig. 1) supports the robustness of our ET estimating method.
Figure 1. Scatter plots between ET observation and estimation. ET observation is from 108 global distributed FLUXNET sites. Point density (number of points within each 0.25X0.25 grid) is represented by colour. Red line represents the linear fitting function, black dashed line represents the 1:1 line. Red line is the linear fitting result with the intercept with y-axis to be 0. The symbol N in the panel represents the sample quantity.
Since our model predicts the canopy conductance and evapotranspiration from the optimized stomata behaviour, it requires no type-based parameters to be calibrated for specific biome or land cover types. The parameterization method increases the uncertainty during extrapolating to none-calibrated regions or periods. Our model bypasses this problem by employing the universal plant behaviour to replace the parameterization strategy. LPICEA people will keep going with the eco-hydrological topics.