Modeling urban evapotranspiration with large-eddy simulations: The spatial and temporal influence of eddy-covariance footprints

Vegetation is suggested to mitigate urban heat by shading and increasing evapotranspiration (ET). Understanding ET is crucial for effective urban water resources management and optimized cooling from vegetation. The most direct method to observe and study ET over larger areas is eddy-covariance (EC). EC systems measure the ET from a time-dependent source area called the footprint. The footprint depends on the meteorological conditions including wind conditions and atmospheric stability, and the sensor location and height. Although the time-dependent footprint can be modeled analytically, these analytical footprint descriptions neglect the heterogeneous urban flow field. The Large-Eddy Simulations (LES) technique explicitly resolves turbulent transport incorporating footprint dynamics in the modeled ET. Here we explore the temporal and spatial sensitivity of EC observations to the footprint at two sites in Berlin. For this purpose, we apply analytically derived footprints to analyze half-hourly footprints over a full year and LES to capture the spatial detail in footprints over one day. The LES results show realistic ET rates while differences with the observations seem to be driven by the mesoscale forcing. Our findings suggest that within a day the footprint's time dependency causes as much variation in observed ET as within a year, even when the day has relatively steady large-scale meteorological conditions. Virtual EC systems in the LES frequently show no correlation while being less than 280 m apart demonstrating how EC systems represent only their specific location. Our study shows the urgency of considering footprints when analyzing instantaneous ET observations from EC systems and the ineffectiveness of representing an EC system as measuring its average surroundings.