Search

Long-term tall-tower eddy-covariance (EC) measurements have been recently established in three European pilot cities as part of the ICOS-Cities project. We conducted a comparison of EC software to ensure a reliable generation of interoperable flux estimates, which is the prerequisite for avoiding methodological biases and improving the comparability of the results. We analyzed datasets covering 5 months collected from EC tall-tower installations located in urbanized areas of Munich, Zurich, and Paris. Fluxes of sensible heat, latent heat, and CO2 were calculated using three software packages (i.e., TK3, EddyPro, and eddy4R) to assess the uncertainty of flux estimations attributed to differences in implemented postprocessing schemes. A very good agreement on the mean values and standard deviations was found across all three sites, which can probably be attributed to a uniform instrumentation, data acquisition, and preprocessing. The overall comparison of final flux time series products showed a good but not yet perfect agreement among the three software packages. TK3 and EddyPro both calculated fluxes with low-frequency spectral correction, resulting in better agreement than between TK3 and the eddy4R workflow with disabled low-frequency spectral treatment. These observed flux discrepancies indicate the crucial role of treating low-frequency spectral loss in flux estimation for tall-tower EC systems.

The eddy-covariance method provides the most direct estimates for fluxes between ecosystems and the atmosphere. However, dispersive fluxes can occur in the presence of secondary circulations, which can inherently not be captured by such single-tower measurements. In this study, we present options to correct local flux measurements for such large-scale transport based on a non-local parametric model that has been developed from a set of idealized large-eddy simulations. This method is tested for three real-world sites (DK-Sor, DE-Fen, and DE-Gwg), representing typical conditions in the mid-latitudes with different measurement heights, different terrain complexities, and different landscape-scale heterogeneities. Two ways to determine the boundary-layer height, which is a necessary input variable for modelling the dispersive fluxes, are applied, which are either based on operational radio soundings and local in situ measurements for the flat sites or from backscatter-intensity profiles obtained from co-located ceilometers for the two sites in complex terrain. The adjusted total fluxes are evaluated by assessing the improvement in energy balance closure and by comparing the resulting latent heat fluxes with evapotranspiration rates from nearby lysimeters. The results show that not only the accuracy of the flux estimates is improved but also the precision, which is indicated by RMSE values that are reduced by approximately 50 %. Nevertheless, it needs to be clear that this method is intended to correct for a bias in eddy-covariance measurements due to the presence of large-scale dispersive fluxes. Other reasons potentially causing a systematic underestimated or overestimation, such as low-pass filtering effects and missing storage terms, still need to be considered and minimized as much as possible. Moreover, additional transport induced by surface heterogeneities is not considered.

The observing system design of multidisciplinary field measurements involves a variety of considerations on logistics, safety, and science objectives. Typically, this is done based on investigator intuition and designs of prior field measurements. However, there is potential for considerable increases in efficiency, safety, and scientific success by integrating numerical simulations in the design process. Here, we present a novel numerical simulation–environmental response function (NS–ERF) approach to observing system simulation experiments that aids surface–atmosphere synthesis at the interface of mesoscale and microscale meteorology. In a case study we demonstrate application of the NS–ERF approach to optimize the Chequamegon Heterogeneous Ecosystem Energy-balance Study Enabled by a High-density Extensive Array of Detectors 2019 (CHEESEHEAD19). During CHEESEHEAD19 pre-field simulation experiments, we considered the placement of 20 eddy covariance flux towers, operations for 72 h of low-altitude flux aircraft measurements, and integration of various remote sensing data products. A 2 h high-resolution large eddy simulation created a cloud-free virtual atmosphere for surface and meteorological conditions characteristic of the field campaign domain and period. To explore two specific design hypotheses we super-sampled this virtual atmosphere as observed by 13 different yet simultaneous observing system designs consisting of virtual ground, airborne, and satellite observations. We then analyzed these virtual observations through ERFs to yield an optimal aircraft flight strategy for augmenting a stratified random flux tower network in combination with satellite retrievals. We demonstrate how the novel NS–ERF approach doubled CHEESEHEAD19's potential to explore energy balance closure and spatial patterning science objectives while substantially simplifying logistics. Owing to its modular extensibility, NS–ERF lends itself to optimizing observing system designs also for natural climate solutions, emission inventory validation, urban air quality, industry leak detection, and multi-species applications, among other use cases.

Accurate measurements of turbulence statistics in the atmosphere are important for eddy-covariance measurements, wind energy research, and the validation of atmospheric numerical models. Sonic anemometers are widely used for these applications. However, these instruments are prone to probe-induced flow distortion effects, and the magnitude of the resulting errors has been debated due to the lack of an absolute reference instrument under field conditions. Here, we present the results of an intercomparison experiment between a CSAT3B sonic anemometer and a high-resolution bistatic Doppler lidar, which is inherently free of any flow distortion. This novel remote sensing instrument has otherwise very similar spatial and temporal sampling characteristics to the sonic anemometer and hence served as a reference for this comparison. The presented measurements were carried out over flat homogeneous terrain at a measurement height of 30 m. We provide a comparative statistical analysis of the resulting mean wind velocities, the standard deviations of the vertical wind speed and the friction velocity and investigate the reasons for the observed deviations based on the turbulence spectra and co-spectra. Our results show an agreement of the mean wind velocity measurements and the standard deviations of the vertical wind speed with a comparability of 0.082 and 0.020 m s−1, respectively. Biases for these two quantities were 0.003 and 0.012 m s−1, respectively. Slightly larger differences were observed for friction velocity. Analysis of the corresponding co-spectra showed that the CSAT3B underestimates this quantity systematically by about 3 % on average as a result of co-spectral losses in the frequency range between 0.1 and 5 s−1. We also found that an angle-of-attack-dependent transducer-shadowing correction does not improve the agreement between the CSAT3B and the Physikalisch-Technische Bundesanstalt (PTB) lidar effectively.

Three-dimensional sonic anemometers are the core component of eddy covariance systems, which are widely used for micrometeorological and ecological research. In order to characterize the measurement uncertainty of these instruments we present and analyse the results from a field intercomparison experiment of six commonly used sonic anemometer models from four major manufacturers. These models include Campbell CSAT3, Gill HS-50 and R3, METEK uSonic-3 Omni, R. M. Young 81000 and 81000RE. The experiment was conducted over a meadow at the TERENO/ICOS site DE-Fen in southern Germany over a period of 16 days in June of 2016 as part of the ScaleX campaign. The measurement height was 3 m for all sensors, which were separated by 9 m from each other, each on its own tripod, in order to limit contamination of the turbulence measurements by adjacent structures as much as possible. Moreover, the high-frequency data from all instruments were treated with the same post-processing algorithm. In this study, we compare the results for various turbulence statistics, which include mean horizontal wind speed, standard deviations of vertical wind velocity and sonic temperature, friction velocity, and the buoyancy flux. Quantitative measures of uncertainty, such as bias and comparability, are derived from these results. We find that biases are generally very small for all sensors and all computed variables, except for the sonic temperature measurements of the two Gill sonic anemometers (HS and R3), confirming a known transducer-temperature dependence of the sonic temperature measurement. The best overall agreement between the different instruments was found for the mean wind speed and the buoyancy flux.

A comparison of two popular eddy-covariance software packages is presented, namely, EddyPro and TK3. Two approximately 1-month long test data sets were processed, representing typical instrumental setups (i.e., CSAT3/LI-7500 above grassland and Solent R3/LI-6262 above a forest). The resulting fluxes and quality flags were compared. Achieving a satisfying agreement and understanding residual discrepancies required several iterations and interventions of different nature, spanning from simple software reconfiguration to actual code manipulations. In this paper, we document our comparison exercise and show that the two software packages can provide utterly satisfying agreement when properly configured. Our main aim, however, is to stress the complexity of performing a rigorous comparison of eddy-covariance software. We show that discriminating actual discrepancies in the results from inconsistencies in the software configuration requires deep knowledge of both software packages and of the eddy-covariance method. In some instances, it may be even beyond the possibility of the investigator who does not have access to and full knowledge of the source code. Being the developers of EddyPro and TK3, we could discuss the comparison at all levels of details and this proved necessary to achieve a full understanding. As a result, we suggest that researchers are more likely to get comparable results when using EddyPro (v5.1.1) and TK3 (v3.11) – at least with the setting presented in this paper – than they are when using any other pair of EC software which did not undergo a similar cross-validation. As a further consequence, we also suggest that, to the aim of assuring consistency and comparability of centralized flux databases, and for a confident use of eddy fluxes in synthesis studies on the regional, continental and global scale, researchers only rely on software that have been extensively validated in documented intercomparisons.

The objective of this study is to assess the feasibility and quality of eddy-covariance flux measurements from a weight-shift microlight aircraft (WSMA). Firstly, we investigate the precision of the wind measurement (σu,v ≤ 0.09 m s−1, σw = 0.04 m s−1), the lynchpin of flux calculations from aircraft. From here, the smallest resolvable changes in friction velocity (0.02 m s−1), and sensible- (5 W m−2) and latent (3 W m−2) heat flux are estimated. Secondly, a seven-day flight campaign was performed near Lindenberg (Germany). Here we compare measurements of wind, temperature, humidity and respective fluxes between a tall tower and the WSMA. The maximum likelihood functional relationship (MLFR) between tower and WSMA measurements considers the random error in the data, and shows very good agreement of the scalar averages. The MLFRs for standard deviations (SDs, 2–34%) and fluxes (17–21%) indicate higher estimates of the airborne measurements compared to the tower. Considering the 99.5% confidence intervals, the observed differences are not significant, with exception of the temperature SD. The comparison with a large-aperture scintillometer reveals lower sensible heat flux estimates at both tower (−40 to −25%) and WSMA (−25–0%). We relate the observed differences to (i) inconsistencies in the temperature and wind measurement at the tower and (ii) the measurement platforms' differing abilities to capture contributions from non-propagating eddies. These findings encourage the use of WSMA as a low cost and highly versatile flux measurement platform.