Random Sampling Errors in CO2 Fluxes Measured by the Open-path Eddy Covariance Method and Their Influence on Estimating Annual Carbon Budget

Masayoshi MANO, Akira MIYATA, Hideyuki NAGAI, Tomoyasu YAMADA, Keisuke ONO, Makoto SAITO, Yoshikazu KOBAYASHI
2007 Journal of Agricultural Meteorology  
Error assessment of measured CO 2 fluxes is essential both to evaluate the annual carbon budgets of flux study sites and to decide whether individual half-hourly fluxes may be accepted for further analysis. In this paper, we focus on random sampling errors in CO 2 fluxes measured at a rice field using the open-path eddy covariance method. We investigate: (1) the propagation of random sampling errors stemming from the density corrections required with this method, and (2) the influences of the
more » ... influences of the errors in half-hourly CO 2 fluxes on evaluations of the annual carbon budget. The CO 2 flux with density corrections is the sum of raw CO 2 flux and correction terms resulting from sensible and latent heat fluxes. Each flux has its own random sampling error. We first calculated the errors of the individual terms as the square root of the variance of the relevant covariance. Then we calculated the error in the corrected CO 2 flux based on a general error propagation rule. The influence of the random sampling error became important when the corrected CO 2 flux was close to zero and the sensible heat flux was large. The random sampling error in the corrected CO 2 flux was about 10 of its magnitude under favorable measurement conditions, but an error of around 0.5 to 1.0 µmol m -2 s -1 remained even when the corrected CO 2 flux was zero. These results may be utilized for post-processing of data, such as in quality control tests, and subsequent data analyses. The annual accumulation of CO 2 flux observed at the study site showed a CO 2 sink of about 200 to 300 g C m -2 yr -1 . The random sampling error for the annual flux, however, was 1 to 2 of the flux magnitude since the summation of large numbers of half-hourly random sampling errors followed the error propagation rule. This effect demonstrates one of the advantages of estimating the annual carbon budget using the eddy covariance method.
doi:10.2480/agrmet.63.67 fatcat:owqnwu4um5gbhebm2qsd7qtqre