AuthorsYasuda, Y., Watanabe, T., Ohtani, Y., Mizoguchi, Y., Mano, M.
TitleQuality Control of Flux Data Measured above a Deciduous Forest
JournalJournal of agricultural meteorology
Keywordsenergy imbalance, flux, quality control, stationarity
AbstractWe are able to obtain a huge amount of flux data using eddy covariance systems. Instrument troubles and unsuitable conditions degrade the quality of flux data, but such errors cannot be detected by visual inspections alone. Automated quality control methods for flux data are very useful for detecting errors or unsuitable data for flux calculations. In this study, we tested quality control methods for flux data of a deciduous forest. The methods were proposed by Foken and Wichura (Agric.For. Meteorol. (1996), 78, 83-105) and Vickers and Mahrt (J. Atmos. Oceanic. Technol. (1997), 14, 512-526), and can be classified into two tests: raw-data and steady-state tests. The raw-data tests check spikes, absolute limits, absolute variance, higher moment statistics and signal discontinuities, while the steady-state tests examine the stationarity of fluxes in each data record. Each test assigns flags against questionable data records when the results of a test exceed prescribed threshold values. The raw-data tests showed that the percentage of flagged data greatly differed among signals. During the growing season, signals from a sonic anemo-thermometer had low percentages of flagged data (2-5%), while signals from a CO2 sensor contained a higher number (17%). With a humidity sensor, the percentage of flags was 4% during the growing season and was much lower than that of the CO2 signals. The steady-state tests assigned flags to about 10-30% of the flux data. An inspection of the flagged fluxes in terms of the energy imbalance revealed that, on average, the energy imbalance decreased when the degree of the nonstationarity of latent heat flux decreased.