Actual evapotranspiration differences between measurements of eddy covariance and lysimeters over grasslands

Accurate measurements of actual evapotranspiration (ETa) play an important role in understanding land surface processes and agricultural management. Two of the most commonly used and established methods for quantifying ETa are eddy covariance (EC) and weighable lysimeters measurements. Previous studies on hourly or daily basis indicated sometimes large differences between the ETa of the two methods (Δ_ly-EC). It is still unclear which factors influence these differences. Here, we examine and compare half-hourly ETa measurements from EC (ET_EC) and weighable lysimeters (ET_ly) at four different sites. The four sites span a climatic gradient from humid conditions at a pre-alpine (Fendt, DE) and a mid-mountain grassland (Rollesbroich, DE) to semi-arid conditions at two sites with a natural grass and shrub (Els Plans, ES) and a tree-grass ecosystem (Majadas de Tiétar, ES). We used a boosted regression tree method to identify environmental drivers of Δ_ly-EC during day and night at the half-hourly resolution.

Our results revealed that substantial differences were found with a mean annual Δ_ly-EC of 117 mm, and Δ_ly-EC displayed obvious spatiotemporal variabilities across the sites. Energy balance non-closure of EC was found to be the most important factor contributing to the large annual Δ_ly-EC, especially at Majadas de Tiétar site. With the distinct climatic gradient, Δ_ly-EC was negatively correlated with mean annual wind speed and vapor pressure deficit when they reached a specific level. Monthly ET_EC and ET_ly agreed well with Δ_ly-EC peaking in summer at the sites in Germany, while Δ_ly-EC peaked earlier due to the different climate in Spain. Differences in grass height caused by field management and EC footprint also affected Δ_ly-EC, especially at daily timescales for the pre-alpine and the mid-mountain grassland ecosystem. The relative impacts of different environmental variables to half-hourly ET_EC and ET_ly were almost the same with soil water content (SWC) being more important for nighttime ET_ly. Meanwhile, we found that the dominant controlling factors of daytime Δ_ly-EC changed with climatic conditions, but nighttime Δ_ly-EC were mainly regulated by SWC. These findings provide a critical evaluation for the roles of climatic and land surface conditions on turbulent flux dynamics from different measurements, which has important implications for ecosystem water and energy balance.