龙凤山站大气CO_2浓度2种筛分方法对比研究

Comparison of two flagging approaches to the observed CO_2 mole fractions at the Longfengshan WMO/GAW Regional Station in China

针对黑龙江龙凤山区域本底站2009年1月~2011年12月大气CO2在线观测数据,研究基于地面风、日变化等大气本底/非本底数据筛分方法（SWDV）和稳健局部近似回归大气本底/非本底数据筛分方法（REBS）在龙凤山区域本底站的适用性.研究表明：2种筛分方法在春、秋和冬季都能很好反映龙凤山大气CO2浓度的趋势变化及局地源汇对观测CO2浓度的影响,对于高浓度的非本底数据都能够较好的识别,但在夏季使用REBS方法会影响筛分的准确性,不建议在龙凤山区域本底站使用REBS筛分方法.SWDV和REBS法筛分出的本底数据分别占总数据量的30.7%和58.9%.2种方法均筛分为本底浓度和非本底浓度的数据分别占总数据量的21.5%和32.0%.二者筛分的本底季平均浓度在春季相差最小为（0.1±0.3）×10^-6（摩尔比,下同）,冬季和秋季次之,在夏季相差最大为（4.2±1.0）×10^-6.典型个例分析表明,SWDV法会将白天一些受西南污染气流影响的CO2浓度误筛分为本底浓度,REBS法会将个别在静稳天气条件下受局地影响大的CO2观测值误筛分为本底浓度.夏季局地污染状况可能被植被强烈的光合作用抵消,CO2浓度变化不大,使得REBS误筛分为本底浓度,以及对于一些较低的CO2浓度值,REBS误筛分为非本底浓度,这些因素导致2种筛分方法在夏季本底浓度差别较大.

As to the atmospheric CO2 mole fractions measured by the in-situ system from 10 m 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 CO2 variation 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^-6larger 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.