Sources of uncertainty in eddy covariance ozone flux measurements made by dry chemiluminescence fast response analysers
Abstract. Eddy covariance ozone flux measurements are the most direct way to estimate ozone removal near the surface. Over vegetated surfaces, high quality ozone fluxes are required to probe the underlying processes for which it is necessary to separate the flux into the components of stomatal and non-stomatal deposition. Detailed knowledge of the processes that control non-stomatal deposition is limited and more accurate ozone flux measurements are needed to quantify this component of the deposited flux. We present a systematic intercomparison study of eddy covariance ozone flux measurements made using two fast response dry chemiluminescence analysers. Ozone deposition was measured over a well characterised managed grassland near Edinburgh, Scotland, during August 2007. A data quality control procedure specific to these analysers is introduced. Absolute ozone fluxes were calculated based on the relative signals of the dry chemiluminescence analysers using three different calibration methods and the results are compared for both analysers. It is shown that the error in the fitted parameters required for the flux calculations provides a substantial source of uncertainty in the fluxes. The choice of the calculation method itself can also constitute an uncertainty in the flux as the calculated fluxes by the three methods do not agree within error at all times. This finding highlights the need for a consistent and rigorous approach for comparable data-sets, such as e.g. in flux networks. Ozone fluxes calculated by one of the methods were then used to compare the two analysers in more detail. This systematic analyser comparison reveals half-hourly flux values differing by up to a factor of two at times with the difference in mean hourly flux ranging from 0 to 23% with an error in the mean daily flux of ±12%. The comparison of analysers shows that the agreement in fluxes is excellent for some days but that there is an underlying uncertainty as a result of variable analyser performance and/or non-linear behaviour of disc sensitivity.