Evaluation on the Impacts of Different Background Determination Methods on CO2 Sources and Sinks Estimation and Seasonal Variations

To accurately determine background conditions or extract sources and sinks information from the observed atmospheric carbon dioxide (CO2) concentration is crucial for quantitative estimation of regional and global carbon budget and future trends of atmospheric CO2. In this study, the synchronized observed surface winds and carbon monoxide (CO) concentration have been examined to test their effectiveness as filter factors to determine CO2 background conditions at Waliguan site. The results show that the surface winds and CO concentrations can be used as filter factor in winter, but they are not very effective in summer. Three statistical methods, robust estimation of background signal (REBS), Fourier transform algorithm (FTA) and a new developed moving average filtering (MAF), are applied to atmospheric CO2 background selection. The result suggested that our new developed MAF method, which can well estimate the elevated and sequestered CO2 concentrations due to using changing and adjusting filter criteria at every two-week fitting window, is thus better than the other two statistical methods. A good consistency is indicated by the three methods for estimating the elevated CO2 caused by local or regional emissions, but it showed large discrepancies when determining the sequestered CO2. The result suggested that the three methods can reasonable extract those anthropogenic influenced episodes, but only MAF method would well identify those episodes due to terrestrial CO2 fluxes. Mean seasonal amplitude of atmospheric CO2 at Waliguan during 1995-2008 is 10. 3 x 10(-6) estimated by MAF method, which is in good agreement with previous studies. Whereas, the seasonal amplitudes derived by REBS method are much lower, only with a value of ~9. 1 x 10(-6) during 1995-2008, which will result in an underestimation of regional or global CO2 fluxes.