オープンパス型渦相関法データの品質管理

Abstract

As the number of monitoring sites of carbon dioxide (CO2) and water vapor fluxes is increasing and long-term measurements of these fluxes are popularized, guidelines for practical procedures of quality control of measured flux data are needed to produce reliable datasets which are suitable for data analysis.Several quality control tests for eddy covariance data have been proposed. To examine whether such tests are suitable for practical use, we applied three types of quality control tests to a two-year dataset obtained by the open-path eddy covariance method at a rice field. Raw data quality control tests detected time series inappropriate for flux calculations such as runs with rainfall or instrumental failure. Steady state tests detected two kinds of non-steady state runs. Monotonic trends were detected by the instationarity test, while intermediate changes of turbulence structure were detected by the nonstationarity test. Integral turbulence characteristic tests based on flux-variance similarity were useful for confirming whether turbulence was well developed or not. Results of these quality control tests, together with additional information on the length of recording time, wind direction relative to the sonic anemometer, and precipitation records were categorized into nine classes and combined for evaluation of the final data quality of the covariances. The final data quality was then evaluated as either discard, high quality, low quality, or normal quality. By visual inspection of the daily courses of CO2 fluxes, we confirmed that anomalies and unexpected covariances had been rightly judged to be discarded in the final data quality classification. Energy imbalance ratios also supported the evaluation of the final data quality as the average of the half-hourly energy imbalance ratios decreased with the increase of final data quality. We concluded that the quality control tests examined and the final data quality evaluation proposed in the present study provide a suitable guide to data quality and can be a useful tool for flux studies.