In-situ measurement of CO2 at the Xinglong regional background station over North China

YANG Yng, WANG Ting, WANG Pui, ZHOU Minqing n YAO Bo

aKey Laboratory of Middle Atmosphere and Global Environment Observation, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China; bCollege of Earth and Planetary Science, University of Chinese Academy of Sciences, Beijing,China; cRoyal Belgian Institute for Space Aeronomy,Brussels,Belgium; dMeteorological Observation Center of China Meteorological Administration,Beijing,China

ABSTRACT High-accuracy continuous measurements of atmospheric concentrations of CO2 were made from May 2016 to December 2017 using the Picarro G2301 analyzer at Xinglong station(40°24′N,117°30′E,940 MSL),150 km northeast of Beijing.The near-ground CO2 measurements were calibrated by standards based on WMO procedures. The regional background measurements were filtered by the robust extraction of baseline signal method to study the seasonal and diurnal cycles. The regional background CO2 concentrations were low in summer.The maximum difference between the local sources and regional background CO2 concentrations occurred in summer and autumn,indicating a strong influence from local sources. Cluster analysis and potential source contribution function analysis showed that the long-distance transport of anthropogenic emissions in the Beijing-Tianjin-Hebei metropolitan area influenced the CO2 concentrations in Xinglong, especially in summer. The diurnal variation of CO2 was mainly influenced by the various vertical transport conditions of the tropospheric atmosphere in a day.

KEYWORDS Xinglong station; CO2 concentration; variation;source analysis

1. Introduction

The continuous increases in the concentrations of greenhouse gases,particularly CO2,have attracted much attention from the international community.The global average concentration of atmospheric CO2has increased by 145%since the pre-industrial era and was up to 403.3±0.1 ppm in 2016(WMO 2017).CO2has a long lifetime in the atmosphere and contributes>60%of the total radiative forcing of greenhouse gases, although it accounts for a relatively small proportion of the total amount of gas in the atmosphere (IPCC 2013). The increase in atmospheric concentrations of CO2is mainly attributed to the burning of fossil fuels and changes in land use (Peters et al. 2011). The sources and sinks of CO2in the atmosphere are unevenly distributed and subject to a variety of transport processes.High-precision measurements of the concentrations of CO2provide a foundation for the assessment of carbon emissions and studies of the carbon cycle(Andrews et al.2014;Fang et al.2014).

Governments and research groups have put much effort into monitoring and quantifying the magnitude and distribution of greenhouse gases at both domestic and international levels.To date,there are more than 150 sites worldwide in the World Meteorology Organization/Global Atmosphere Watch (WMO/GAW) network, with the aim of detecting changes in greenhouse gas concentrations and source-sink analysis.The USA has conducted long-term observations of atmospheric CO2concentrations at the Mauna Loa site in Hawaii since 1958.The US National Oceanic and Atmospheric Administration(NOAA) Earth System Research Laboratory have been working to build a network of tall-tower CO2measurement sites since the early 1990s(Bakwin et al.1998;Zhao,Bakwin,and Tans 1997)and a reliable and precise in-situ CO2analysis system has been developed and deployed at eight sites in the NOAA Earth System Research Laboratory Global Greenhouse Gas Reference Network.The Integrated Carbon Observation System in Europe is committed to providing high-precision data to determine the spatial distributions of long-term changes in CO2concentrations(Yver Kwok et al.2015).

The concentrations of greenhouse gases in China are increasing quickly due to rapid economic growth and the increased combustion of fossil fuels. Since the 1990s, seven CO2/CH4background stations have been set up by the China Meteorological Administration(CMA), and four of them - Waliguan in Qinghai Province, Shangdianzi in Beijing, Lin’an in Zhejiang Province, and Longfeng Mountain in Heilongjiang Province (Fang et al. 2014; Liu et al. 2009) - belong to the WMO/GAW global network.

Measurements of CO2in economically developed and densely populated areas of China are important due to the impact of human activities, especially in the Beijing-Tianjin-Hebei metropolitan area. Xinglong station is located in Xinglong County, Hebei Province,and can be used as a background station for the Beijing-Tianjin-Hebei metropolitan area (Cheng et al.2003). A Picarro cavity ring-down spectroscopy (CRDS)analyzer is used for monitoring carbon emissions.

2. Data and methods

2.1. In-situ measurement methods

Xinglong station (40°24′N, 117°30′E, 940 MSL) (red spot in Figure S1) is located in Xinglong County, Hebei Province, south of the Yanshan Mountains, about 150 km northeast of Beijing. It is surrounded by mountains with shrubland and grassland. Also, there are no industrial activities within 20 km.Xinglong station is far from megacities and major transport routes and the atmospheric conditions may, therefore, represent the regional atmosphere (Wu et al. 2010).

Based on CRDS techniques, the Picarro instrument is designed as a low drift, high-precision greenhouse gas analyzer.The intensity of the laser introduced into a highprecision optical cavity decays exponentially with time.The decay time constant, τ, depends on the loss during the transport between different endoscopes in the optical cavity and the absorption and scattering of the sample being measured. The concentration of the gas sample is related to τ(Crosson 2008).CO2concentrations have been observed at Xinglong station using the Picarro G2301 high-precision CRDS analyzer since May 2016.The instrument sampling inlet is located on the roof of a three-story building and has been operated for more than 1 year.

2.2. Data processing

2.2.1. Calibration

Eight calibrations were performed from 14 November to 14 December 2017, with a duration of 15 min for each calibration,to determine the stability of the G2301 highprecision CRDS analyzer. Standard gas, with a CH4concentration of 2036.5±0.3 ppb and a CO2concentration of 399.95 ± 0.02 ppm under dry and clean atmospheric conditions was calibrated by the Meteorological Observation Center of the CMA based on the WMO X2007 scale for CO2.

The response of the analyzer is unstable after the start of calibration as a result of the time required for flushing the internal volume of the analyzer(Yan et al.2011).We set the flushing time to 5 min and therefore only data obtained 5 min after the start of the calibration were retained. The temporal resolution of the raw data was 5 s,but 1-min average values were used for the calibration analysis and 1-h average values elsewhere.

2.2.2. Data filtering

The robust extraction of baseline signal method(Thoning,Tans,and Komhyr 1989;Fang et al.2015)is always used to classify the regional background and non-background CO2measurements. The regional background measurements showed that the air was well mixed and not affected by local anthropogenic emissions and regional transport. The concentrations higher than the regional background values were governed by regional or local sources,usually reflecting significant changes in concentration over short periods of time - for example, the rapid increase in concentrations during daytime due to human activities (Zhang, Zhou, and Xu 2013).Concentrations lower than the regional background values represent local sinks.

The data were filtered with a method similar to that of Ruckstuhl et al. (2012): (1) all the outliers introduced by system failure or other factors were flagged;(2)the 30-day running average and standard deviation σ of the measurements were calculated; and (3) the absolute difference between each measurement and the running average was calculated.The measurements were flagged and considered as non-background data if this difference exceeded σ.The final extracted data were considered to represent the fluctuations in regional background CO2concentrations at Xinglong station. The remaining measurements were either sources or sinks, tagged, respectively,by values higher or lower than that of the regional background(Zhang,Zhou,and Xu 2013).

The total number of hourly mean CO2samples was 13 077 during the observation period, including 9680 regional background measurements (74.0%).

3. Results and discussion

3.1. Instrument stability

The upper panels in Figure S2 show the differences(Andrews et al. 2014) between the CO2concentrations in atmosphere and calibration tanks. The average differences for CO2are stable around±0.1 ppm during the calibration period,confirming that the accuracies of our measurements meet the WMO/GAW criteria of ±0.1 ppm for CO2(Tuomas 2007).

As the differences in the values for CO2were small relative to their concentrations, we considered the calibration coefficients during each calibration to be the ratio of the measured value and the standard value, as shown in the lower panel.The calibration coefficients for CO2were stable at 0.992-0.993. We used the average calibration coefficients for CO2to calibrate the raw data.

3.2. Seasonal and diurnal variations

Figure 1 shows the time series of CO2concentrations in Xinglong during the period June 2016 to December 2017.The black dots are the filtered regional background values based on the robust extraction of the baseline signal.Figure 2 shows the monthly cycles of local sources and regional background CO2concentrations over a 12-month period and the upper panel shows their differences.Figure 2 shows that both the local sources and regional background CO2concentrations are low in summer, the season with the strongest photosynthetic uptake.The high regional background values in winter and early spring are due to positive biological fluxes(i.e.,stronger plant respiration and weaker plant photosynthesis). The seasonal variations in Shangdianzi, which is also located in the Beijing-Tianjin-Hebei urban agglomeration on the North China Plain, are almost identical compared to that in Xinglong,with the lowest occurring in August and highest in March (Xia et al. 2016). In addition to weaker plant photosynthesis,CO2occupies 6.13%of the total emissions from straw burning in China in autumn (Cao et al. 2008),influencing the regional background signal in Xinglong in October. The increase in anthropogenic sources of CO2from domestic heating contributes to the higher CO2values during winter (Wang et al. 2010). The peak-totrough value of the monthly regional background CO2concentration is 32.43 ± 10.95 ppm, proving the obvious seasonal variation in CO2in Xinglong station.Differences in the local sources and regional background CO2concentrations (upper panel in Figure 2) are large in summer and early autumn, reaching 21.63 ± 7.40, 35.56 ± 7.40, and 26.72±7.40 ppm in July,August,and September,respectively(Fang et al.2017).This indicates the strong influence of strong local sources in these two seasons.The relatively higher discrepancies in November and February also illustrate higher signals from local sources.

Figure 2. Monthly cycles of local sources and regional background CO2 concentrations over a 12-month period. The upper panel shows the differences between the local sources and regional background CO2 concentrations over a 12-month period.The lower panel shows the monthly cycles of local sources and regional background CO2 concentrations over a 12-month period.

Figure 1. Time series of regional background (black) and non-background (gray) CO2 concentrations in Xinglong.

The least-squares method was used to fit the hourly average data of CO2to evaluate the diurnal cycles in the regional background CO2concentrations in different seasons that were not affected by linear trends.Figure 3 shows four diurnal cycles of CO2.

In general, the diurnal variations in CO2concentrations are significant in all seasons, with low values around 1000 to 1400 local time. This is because the stronger convection at noon mixes the near-ground high CO2concentrations with the low concentrations in the upper-level atmosphere. The weaker convective diffusion in the boundary layer limits the accumulation of CO2in the afternoon and high values usually occur at night.The peak-to-trough diurnal amplitudes are 2.82±0.93, 3.97 ± 1.26, 3.58 ± 1.26, and 4.10 ± 1.06 ppm for spring, summer, autumn, and winter, respectively.

3.3. Influence of long-distance transport

Figure 3.Diurnal cycles of regional background concentrations of CO2 by season.

Xinglong station is built on a relatively high hill and CO2concentrations are likely due to large-scale advection(Fang et al. 2016). We calculated the backward 72-h airflow trajectories arriving in the Xinglong area in all four seasons from May 2016 to December 2017 based on MeteoInfo software to analyze the long-distance transport contributing to CO2concentrations. Cluster analysis groups a large number of trajectories according to the horizontal movement velocity as well as the direction of the air masses and is applied to the analysis of the classification of different transportation(Shi et al.2008).Based on the potential source contribution function (PSCF) algorithm, we calculated the quantitative contributions of potential areas for CO2in all four seasons in Xinglong (Begum et al. 2005).

Figure 4 shows the patterns of cluster analysis for the major 72-h backward trajectories in each season and the potential source contribution of CO2in Xinglong from May 2016 to December 2017. The numbers indicate the trajectory clustering type.The PSCF analysis is also shown in Figure 4 as a supplement to the backward trajectory analysis.The shading represents the probability distribution of emission sources.Table 1 shows the cluster analysis of trajectories,the average CO2concentrations in each cluster,and the regions the clusters run through.

Most of the clusters in spring have pathways northwest of the station, where Mongolia, Russia and Zhangjiakou (in Hebei Province) are located. However,these regions are not expected to be a major contributor for clusters 1, 2, 3, and 6 (covering about 74.32%),which have lower mean CO2concentrations. Cluster 4 may transport anthropogenic emissions with relatively high CO2concentrations of 418.68 ± 10.95 ppm,from Tangshan and Tianjin. The slower speed of cluster 4 obstructs the dilution of strong emissions and hence CO2is accumulated before arriving in Xinglong.

The CO2concentrations in summer are more influenced by clusters 1 and 5 with higher CO2concentrations running through Beijing-Tianjin-Hebei and Shandong.Clusters 1 and 5 account for about 47.79%. Clusters 2 and 4, from Mongolia and Zhangjiakou, together with cluster 3 from the east of Inner Mongolia and Chifeng,are considered as relatively clean areas with lower CO2concentrations.

In autumn and winter, with the strong invasion of westerly wind, the west-north-west clusters through Mongolia, Urumqi, and Zhangjiakou all transport large amounts of CO2. The urban agglomerations of Beijing-Tianjin-Hebei and Shandong also make a positive contribution to the concentrations of CO2in Xinglong in autumn,with the highest values of 418.11 ± 13.20 ppm for CO2carried by cluster 1. Although there is no obvious airflow from the Beijing-Tianjin-Hebei metropolitan area, the PSCF analysis shows that the anthropogenic emissions from these areas also make a large contribution to CO2concentrations in winter.Liu et al.(2009)also studied the relationship between the winds and the non-background CO2in Shangdianzi. They concluded that transport from Beijing-Tianjin-Hebei makes apparently positive contributions to the non-background CO2in Shangdianzi station,which acts similar to Xinglong station.

Figure 4.Cluster analysis of 72-h backward trajectory and PSCF analysis of CO2 in Xinglong from May 2016 to December 2017.The numbers indicate the trajectory clustering type and the color depth represents the probability distribution of the emission sources:the deeper the color, the greater the contribution.

Table 1. Statistical analysis of cluster trajectories, including trajectory numbers, ratios, average CO2 concentrations in each cluster,and the regions each cluster runs through.

The PSCF(shaded areas in Figure 4)was also studied in addition to the cluster analysis of the 72-h backward trajectory. In spring, in addition to the regions determined from the cluster analysis, there is also a possible contribution from the cities in eastern Hebei Province.In summer, the potential sources of CO2are restricted to the south of the station, mainly in Hebei Province.The potential contributing areas in autumn for CO2are mainly located in two main directions: south to Henan Province and west-northwest to Inner Mongolia.Potential source areas for CO2in winter are concentrated in Beijing, north of Shanxi Province and westnorthwest to Inner Mongolia.

4. Conclusions

This paper reports continuous measurements of atmospheric CO2in Xinglong from May 2016 to December 2017. The seasonal and diurnal variations in CO2concentrations, together with long-distance transport, are discussed.

The accuracy of the Picarro instrument in Xinglong meets the requirements of the WMO/GAW global network. The CO2concentrations are high in winter and low in summer. The seasonal variation in the regional background concentrations of CO2is mainly due to the photosynthesis and respiration of plants. Regional transport is another important factor. The airmass from southwest, south, and southeast of the station(e.g., from Beijing, Hebei Province, and Shandong Province) contributes to the high CO2concentrations,especially in summer. Transport from Inner Mongolia is another potential source for CO2in autumn and winter.

The diurnal patterns in CO2concentrations show relatively low values around 1000-1400 local time. For CO2, the strong sunshine at noon results in stronger mixing of the near-surface atmosphere with that at higher levels, which dilutes the near-ground CO2concentrations. The more stable and lower boundary layer in the afternoon increases the concentrations of CO2.

The measurements at Xinglong are made continuously and with regular calibration. With more collected data, we will be capable of analyzing the changes/trends of CO2in the near future. Meanwhile, instead of the regional background CO2measurements, we intend to study the relatively high and low CO2measurements and try to better understand the sources and sinks of CO2around Xinglong.

Acknowledgments

We express our great thanks to the staffat Xinglong station who contributed to the instrument’s maintenance at the station.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This research was funded by the National Key Research and Development Program of China [grant numbers 2017YFB0504000 and 2017YFC1501701] and the National Natural Science Foundation of China[grant number 41575034].