How representative are measurements from a single tower? Implications from a highly resolved wind tunnel study

The use of eddy-covariance measurements to infer the surface-atmosphere exchange of trace constituents is formally expressed in terms of the time- and spatially-averaged mixing ratio, wind vector and turbulent co-variances. In contrast the measurements used to infer the exchange are generally taken from a single tower. For logistical reasons towers are rarely positioned randomly within the canopy, tending more towards the gaps between individual trees. There is then a possibility that a systematic bias arising from tower positioning can impact the accuracy and unbiased nature of eddy-covariance estimates of surface-atmosphere exchange.

Using data from a highly resolved (80 lateral positions, 22 heights between 0 and 3 canopy heights, approximately 2kHz) wind tunnel study of the flow around a model canopy we illustrate that tower positioning can lead to systematic differences in the time-mean wind vector observed at a single location compared to the spatial-average. The mean wind vector observed at a tower positioned furthest away from any canopy element is seen to overestimate the spatially-averaged, time-mean streamwise velocity from the ground up to approximately twice canopy height, as well as recording a systematic non-zero mean vertical velocity. We also show that some, but not all, of the second and higher order moments of the flow are significantly correlated (in space) with the time-mean wind vector implying that such variables measured on individual towers are also likely to be biased with respect to their spatial average. We expect that the spatial variation imposed on the flow in a wind tunnel experiment is more pronounced than that in the field, given the regularity of the model canopy. However, as this variation is imposed through proximity to the canopy elements and the unique nature of canopy turbulence (which is well reproduced by the wind tunnel experiment), we argue that these biases will, qualitatively, be reflected in real world canopies, particularly those with structure such as forest plantations or orchards. The wind tunnel experiment, therefore, provides a useful upper bound on the potential errors that could be imposed through tower positioning.

Finally we briefly discuss the associated spatial variability of scalar exchange and the implications for aerodynamic estimates of eddy-covariance estimates of surface-atmosphere exchange.