Examples of the effects of different averaging methods on carbon dioxide fluxes calculated using the eddy correlation method

Abstract. Three hours of high frequency vertical windspeed and carbon dioxide concentration data recorded over tropical forest in Brazil are presented and discussed in relation to various detrending techniques used in eddy correlation analysis. Running means with time constants 100, 1000 and 1875s and a 30 minute linear detrend, as commonly used to determine fluxes, have been calculated for each case study and are presented. It is shown that, for different trends in the background concentration of carbon dioxide, the different methods can lead to the calculation of radically different fluxes over an hourly period. The examples emphasise the need for caution when interpreting eddy correlation derived fluxes especially for short term process studies.

Keywords: Eddy covariance; detrending; running mean; carbon dioxide; tropical forest