Ensemble spectra for thirteen different eddy-covariance tower sites were calculated using high frequency turbulence data provided by eleven groups, all part of the FLUXNET community. The normalized spectra of vertical wind speed (w) in neutral conditions collapsed onto a single curve for all the forested sites when wavenumbers were normalized with measurement height. This initial result led to the comparison of ensemble spectra of all the measured variables to known standard spectral forms for a wide range of surface/canopy types.

Standard spectra are often used as a gauge of what eddy-covariance signals should look like. Deviations from that spectral shape are then attributed to shortcomings in instrumentation or measurement conditions. The most common set of standard spectra used were derived by Kaimal and co-workers from a set of 45 hourly runs of eddy-covariance measurements taken over an extensive stretch of flat Kansas farmland. Here, we investigate how the Kaimal spectra are realised in long-term eddy-covariance measurements that span many thousands of hours.

While the initial agreement of forest ensemble spectra of w in neutral conditions is quite good, the canopy heights of the sites used in the full analysis span two orders of magnitude and a wide variety of canopy types as well as meteorological conditions. Not surprisingly, the behavior of ensemble spectra is more complex for the full range of sites, especially when scalar tracers or conditions other than neutral are investigated. However, it is demonstrated how the careful evaluation of the statistics of the ensemble spectra can be used to extract representative shapes that can be compared to the Kaimal spectra.