Evapotranspiration (ET) is dominated by transpiration (T) in the terrestrial water cycle. However, continuous measurement of transpiration is still difficult, and the effect of vegetation on ET partitioning is unclear. The concept of underlying water use efficiency (uWUE) was used to develop a new method for ET partitioning by assuming that the maximum, or the potential uWUE is related to T while the averaged or apparent uWUE is related to ET. T/ET was thus estimated as the ratio of the apparent over the potential uWUE using half‐hourly flux data from 17 AmeriFlux sites. The estimated potential uWUE was shown to be essentially constant for 14 of the 17 sites, and was broadly consistent with the uWUE evaluated at the leaf scale. The annual T/ET was the highest for croplands, i.e., 0.69 for corn and 0.62 for soybean, followed by grasslands (0.60) and evergreen needle leaf forests (0.56), and was the lowest for deciduous broadleaf forests (0.52). The enhanced vegetation index (EVI) was shown to be significantly correlated with T/ET and could explain about 75% of the variation in T/ET among the 71 site‐years. The coefficients of determination between EVI and T/ET were 0.84 and 0.82 for corn and soybean, respectively, and 0.77 for deciduous broadleaf forests and grasslands, but only 0.37 for evergreen needle leaf forests. This ET partitioning method is sound in principle and simple to apply in practice, and would enhance the value and role of global FLUXNET in estimating T/ET variations and monitoring ecosystem dynamics.

Water use efficiency is a critical index for describing carbon‐water coupling in terrestrial ecosystems. However, the nonlinear effect of vapor pressure deficit (VPD) on carbon‐water coupling has not been fully considered. To improve the relationship between gross primary production (GPP) and evapotranspiration (ET) at the subdaily time scale, we propose a new underlying water use efficiency (uWUE = GPP · VPD^0.5/ET) and a hysteresis model to minimize time lags among GPP, ET, and VPD. Half‐hourly data were used to validate uWUE for seven vegetation types from 42 AmeriFlux sites. Correlation analysis shows that the GPP · VPD^0.5 and ET relationship (r = 0.844) is better than that between GPP · VPD and ET (r = 0.802). The hysteresis model supports the GPP · VPD^0.5 and ET relationship. As uWUE is related to CO₂ concentration, its use can improve estimates of GPP and ET and help understand the effect of CO₂ fertilization on carbon storage and water loss.

The Budyko hypothesis states that the ratio of the actual evapotranspiration over precipitation (E/P) is fundamentally related to the ratio of the potential evapotranspiration over precipitation (E₀/P). A number of Budyko functions have been proposed to describe such a relationship between E₀/P and E/P. There is, however, no simple method to generate Budyko functions that meet the water and energy constraints. This study showed analytically that for any Budyko function, the sum of elasticity of evapotranspiration with respect to potential evapotranspiration and that with respect to precipitation is equal to unity. This complementary relationship for sensitivity of evapotranspiration has important implications for evaluating hydrologic impact of change in climate and/or catchment characteristics. More importantly, this study found a function that is monotonically increasing with simple limiting properties. This function can be used to generate numerous valid Budyko functions and can also be used to test the validity of the existing Budyko functions.

Water use efficiency (WUE) is a crucial parameter to describe the interrelationship between gross primary production (GPP) and evapotranspiration (ET). Incorporating the nonlinear effect of vapor pressure deficit (VPD), underlying WUE (uWUE = GPP · VPD^0.5/ET) is better than inherent WUE (IWUE = GPP · VPD/ET) at the half-hourly time scale. However, appropriateness of uWUE has not yet been evaluated at the daily time scale. To determine whether uWUE is better than IWUE, daily data for seven vegetation types from 34 AmeriFlux sites were used to validate uWUE at the daily time scale. First, daily mean VPD was shown to be a good substitute for the effective VPD that was required to preserve daily GPP totals. Second, an optimal exponent, k*, corresponding to the best linear relationship between GPP · VPD^k* and ET, was about 0.55 both at half-hourly and daily time scales. Third, correlation coefficient between GPP · VPD^k and ET showed that uWUE (k = 0.5 and r = 0.85) was a better approximation of the optimal WUE (k = k* and r = 0.86) than IWUE (k = 1 and r = 0.81) at the daily scale. Finally, when yearly uWUE was used to predict daily GPP from daily ET and mean VPD, uWUE worked considerably better than IWUE. Comparing observed and predicted daily GPP, the average correlation coefficient and Nash-Sutcliffe coefficient of efficiency were 0.81 and 0.59, respectively, using yearly uWUE, and only 0.59 and −0.83 using yearly IWUE. As a nearly optimal WUE, uWUE consistently outperformed IWUE and could be used to evaluate the effects of global warming and elevated atmosphere CO₂ on carbon assimilation and evapotranspiration.