Abstract：As to the atmospheric CO2 mole fractions measured by the in-situ system from 10m above the ground at Longfengshan station in Heilongjiang province from January 2009 to December 2011, applicability of SWDV (Surface Winds and Diurnal Variations) method and REBS (Robust Extraction of Baseline Signal) technique in data flagging was studied. Except summer, both flagging methods could be applied to the observed data at Longfengshan regional station. The background data flagged by both methods could reflect the CO2variation trend and the non-background data flagged by both methods could reflect the influence of the local sources/sinks. But REBS was more disadvantageous to be used in summer with heavy vegetation coverage at Longfengshan regional station. About 30.7% and 58.9% valid hourly data were flagged as regional representative by SWDV method and REBS technique, respectively. 21.5% and 32.0% of the observed data were commonly classified as background and non-background events, respectively. The difference of background seasonal averaged CO2 mole fractions between the two methods was relatively small in spring with a value of (0.1±0.3)×10-6, larger in winter and autumn, and the largest difference in summer with a value of (4.2±1.0)×10-6. Case studies proved that the data during daytime when the airflow was from the southwest direction by SWDV method and the data at calm conditions by REBS method were occasionally mis-flagged as background events. In summer, it was difficult to distinguish whether the CO2 mole fractions were affected by anthropogenic pollution in the daytime by REBS technique or not, because the photosynthesis was strengthened in that period, which leaded to the large difference of background values between two methods.
栾天, 方双喜, 周凌晞, 姚波, 刘钊. 龙凤山站大气CO2浓度2种筛分方法对比研究[J]. 中国环境科学, 2015, 35(2): 321-328.
LUAN Tian, FANG Shuang-Xi, ZHOU Ling-Xi, YAO Bo, LIU Zhao. Comparison of two flagging approaches to the observed CO2 mole fractions at the Longfengshan WMO/GAW Regional Station in China. CHINA ENVIRONMENTAL SCIENCECE, 2015, 35(2): 321-328.
