Lunch seminar: How well can we quantify evapotranspiration?

Evapotranspiration represents the link between the terrestrial water cycle and the energy balance. Accurate knowledge of this key variable is therefore essential to predict the evolution of weather, climate, and ecosystems.

Two of the most direct methods for measuring evapotranspiration in the field are the eddy-covariance method and lysimeters. However, due their different underlying assumptions and their resulting limitations it remains a challenge to match the results from these two methods for a given site. What can we do if they do not agree?

One way to examine the validity of turbulence measurements is the closure of the energy balance at. Measurements from eddy-covariance sites all over the world show that the sum of turbulent energy fluxes (e.g. convective heat) between the biosphere and the atmosphere generally underestimate the non-convective terms by 10% to 30%. Hence, considering the law of energy conservation, it follows that even state-of-the-art measurements are generally fraught with a substantial closure problem.

Several studies on this long-standing energy balance closure problem indicate that atmospheric motion in the meso-γ range inherently cannot be captured by single point-measurement systems. Hence, several approaches have been proposed in order to adjust the measured fluxes for this apparent systematic error by distributing the residual of the energy balance, assuming all other relevant term are measured accurately. However, there are uncertainties how exactly this partitioning should be done and whether such a correction should be applied on 30-minute data or longer time scales.

The data for this study originate from two grassland sites in southern Germany, where lysimeter measurements are available as reference. We found that a Bowen-ratio conserving adjustment method using a daily correction factor results in a better comparability. However, a second method that attributes the majority of the residual to the sensible heat flux leads to a smaller overall bias. These results are similar between both sites despite considerable differences in terrain complexity and grassland management. Additional idealized large-eddy simulations show that this partitioning depends on the measurement height, which may help to explain contradicting findings about this issue in the literature.