Partitioning of Evapotranspiration at hourly resolution in a temperate grassland using a mobile water isotope laboratory

Quantifying and partitioning the evapotranspiration (ET) of agricultural ecosystems in various environmental settings allows the study of the determinants of field site-specific plant water use and water stress conditions. ET flux, as determined from eddy covariance measurements, was partitioned into its component fluxes, soil evaporation (E) and plant transpiration (T), with a range of independent methods, i.e. with water stable isotope analysis (δ²H and δ¹⁸O), using lysimeter data, by applying the water use efficiency concept, and from process-based numerical modelling with the Community Land Model (CLM) 5.0. Data was collected with a mobile water stable isotope laboratory (IsoMobile) in the vicinity of the ICOS DE-RuR climate station (Rollesbroich, Germany) in an intensively managed temperate grassland ecosystem between May 5 and September 24, 2025. Isotopic partitioning was calculated at sub-daily resolution from mass balance on basis of ET, E, and T isotopic compositions (δ_ET, δ_E, and δ_T, respectively). δ_ET was determined statistically with the Keeling-plot approach and non-destructive measurements of atmospheric water vapor inside and above the plant canopy. δ_E was calculated from the isotopic composition of the atmospheric water vapor and that of soil water, which was either determined destructively and a posteriori in the laboratory or non-destructively and in situ using gas-permeable tubing placed in the soil. Finally, δ_T was estimated destructively from stem water extracted from composite grass samples (Alopecurus pratensis, Lolium perenne, Poa trivialis, Rumex acetosa) and under the assumption of isotopic steady state transpiration. The collected standardized ICOS data was used additionally to set up both the water use efficiency partitioning approach and the CLM. All partitioning results were confronted with time series of environmental variables measured by the local weather station. Sub-daily T/ET responded to daily and seasonal changes of environmental conditions, as well as farming practices applied to the grassland. T/ET decreased significantly after the plants were cut, followed by an increase during the subsequent period of plant regrowth. T/ET estimates range between 21 to 98 % for δ¹⁸O over the course of the seasons, δ²H-based partitioning shows similar temporal developments as δ¹⁸O, while overestimating T/ET by ~13 %. Water stress was not detected during the campaign period, as ET did not decrease while T/ET remained high, even during the dryest and hottest period in summer. From a technical view, non-destructive soil water vapor sampling was found to be a good alternative to destructive soil water sampling for the purpose of ET partitioning. It provides similar δ_E estimations while reducing the need for fieldwork, laboratory time and resources. In conclusion the high-resolution partitioning results presented in this study provide an opportunity to investigate field scale water fluxes in a variety of environments and can aid in improving water flux estimations embedded in large-scale environmental models.