Long-term measurements of the CO column content in the atmosphere in both Beijing and Moscow have been conducted using identical grating infrared spectrometers with a spectral resolution of 0.2 cm^-1. The record in Moscow is from time earlier than 1992 to 2012, while in Beijing it is from 1992 to 2012 and mainly during the late autumn and early winter seasons. The CO column content is calculated from the spectral absorption of the solar radiation received at the Earth's surface by CO in the atmosphere in a selected spectral interval. The uncertainty of the measurement and derivation of the atmospheric CO column content is about 5%-7% ( Rakitin et al., 2011). Details of the instruments' specifications and the methodology for deriving the CO content are described in Dianov-Klokov et al. (1989), Wang et al. (1998b), Fokeeva et al. (2007), and Yurganov et al. (2011).

The most significant issue is to derive the anthropogenic portion (or city emission part) of the CO content, denoted here as Δ U. It is defined as the difference between the measured CO content, U, at the site of study and that of its background site, U_B.

Δ U = U - U_B.

It can be seen that obtaining the background value of the CO content is vital for deriving the anthropogenic CO content. For Moscow, the rural Zvenigorod observation station (50 km northwest of Moscow, with few habitants and factories) is selected as its background site; while for Beijing, the minimum CO content observed on the cleanest days is used as the CO background value as there is no good background station that is not affected by human activities. Such an approach for determining the U_B in Beijing is more reasonable, because the lowest concentration of all polluted species in the atmosphere is observed during the cleanest days and, in that case, the anthropogenic part of CO, Δ U, is close to zero.

The anthropogenic CO content in the atmosphere, Δ U, in both Beijing and Moscow is derived from the dataset of 1992-2012 and the results are given in Fig. 1. It can be seen that there is an evident decreasing trend of CO city emissions for Moscow from 2003 to 2012 in spite of the continuous increase of vehicles in the city. On average, the rate is decreasing at 1.4% per year. The higher anthropogenic CO contents are observed mainly during the cold season, when a continuous anticyclone takes place, as well as in cases with a low wind speed in the boundary layer or a stagnant synoptic pattern ( Wu et al., 2005; Rakitin et al., 2011). It should be noted that a high anthropogenic part of CO content greater than 0.11 atm·cm is never observed in Moscow during 2008-2012, except for summer 2010, when a natural fire occurred.

Figure 1

	Figure Option ViewDownloadNew Window
	Figure 1 Comparison of anthropogenic parts of CO content, Δ U, in the atmosphere in Beijing and Moscow during 1992-2012. The broken lines indicate the average Δ U in Beijing for 1992-2003 and 2006-2012, respectively.

In Beijing, meanwhile, conditions influencing the CO content in the atmosphere are more complicated. During the observational period, haze pollution appears quite often on calm days and on days with a weak southern wind. On the pollution days, the CO column content is several times higher than its background level. At the same time, for half the observation days or more, the CO content is close to its background. Statistical analysis shows that the ratio of "dirty days" (Δ U ≥ 0.08 atm·cm) to "clean days" (Δ U ≥ 0.03 atm·cm) is, on average, 1.05 for Beijing and 0.18 for Moscow for the whole observation period of 1992-2012.

An obvious decrease of CO content in Beijing during 1992-2012 can also be seen in Fig. 1. There is no significant trend in anthropogenic CO. However, it can be clearly seen that the CO content decreases from the first period of 1992-2003 to the second period of 2004-2013, as marked by the broken lines in Fig. 1. This behavior corresponds to the other pollutants, such as PM₁₀, in Beijing according to the long-term measurements taken beside the meteorological tower of the Institute of Atmospheric Physics, Chinese Academy of Sciences.

To gain more understanding about the variation characteristics of the CO content in the atmosphere in different megacities, the frequency distribution of the anthropogenic part of the CO content (Δ U) in the atmosphere for both Moscow and Beijing are presented in Fig. 2. It can be seen that, for Moscow, the anthropogenic part, Δ U, is smaller than 0.05 atm·cm for more than 80% of the data points; while it is at this level for less than 50% in Beijing. In Moscow, the extreme values of CO content (Δ U ≥ 0.110 atm·cm) are usually observed in the cold season (November to April), and they appear for less than 5% of the whole observation period. For Beijing, meanwhile, Δ U ≥ 0.110 atm·cm for about 20% of the time for the same period. Moreover, another difference in the Δ U frequency distribution between Moscow and Beijing is a second peak at 0.07-0.11 atm·cm for the latter, which is strongly related to air pollution events there.

Figure 2

	Figure Option ViewDownloadNew Window
	Figure 2 Frequency distribution of anthropogenic parts of CO content, Δ U, in the atmosphere in Beijing and Moscow for the autumns of 1992-2012.

To discuss the variation properties of CO content in both Moscow and Beijing for different time periods, the frequencies of Δ U for different time periods for both megacities are compared in Fig. 3. It can be seen from the figure for Moscow, that the frequency with high Δ U during 2006-2012 decreases, when compared with that during 1993-2005. Moreover, extreme Δ U (≥ 0.110 atm·cm) do not appear in Moscow since 2008, except for July to August 2010, when natural fires occurred in Moscow's suburban region. A parallel comparison is made for Beijing. It shows that the frequency of higher Δ U ranging from 0.075 to 0.175 atm·cm during 2001-2012 decreases when compared with the frequency during 1992-2000 (Fig. 3), implying that the CO pollution in Beijing has reduced somewhat over the last 10 years.

Figure 3

	Figure Option ViewDownloadNew Window
	Figure 3 Comparison of frequency of ∆ U in the atmosphere in Moscow (top) for the periods 1993-2005 and 2006-2012; and in Beijing (bottom) for the periods 1992-2000 and 2001-2012.

The CO column contents in Beijing and Moscow during recent years are compared in Fig. 4. The prominent feature is the very high CO values up to three times as much as the normal values in the summer of 2010. This corresponds to the severe natural fires in the Moscow region at that time. If 0.2 atm·cm is maintained as a threshold value of CO column content for extreme air pollution situations in Moscow, then it can be seen that such situations are observed quite frequently in Beijing.

The annual variation of the CO column content in the atmosphere for both Beijing and Moscow during the period of 2010-2012 is presented in Fig. 4, to give a general understanding of the difference between Beijing and Moscow. Four evident features can be found: First, during the whole period, for more than 50% of days the CO column contents in Beijing have the same level as in Moscow. Second, the minimum CO column contents observed during the whole period in Beijing and Moscow is almost same, so it can be treated as their background values. This implies that Beijing and Moscow have the same background value. Third, the extremely high CO column contents imply that air pollution takes place quite frequently in Beijing during all seasons, while in Moscow such extreme CO column contents are rarely observed except for the period of natural fire events mentioned above. Fourth, the CO column content is characterized by a maximum in winter and a minimum in summer for both Beijing and Moscow.

Figure 4

	Figure Option ViewDownloadNew Window
	Figure 4 CO column content in Moscow and Beijing for 2010-2012.

As mentioned above, a high CO concentration in Beijing ( Dianov-Klokov et al., 1989; Wang et al., 1998b; Wu et al., 2005) may not be interpreted only by specific meteorological conditions and CO emission in the city region; there should be a reasonable difference in pollution sources between Moscow and Beijing. To better understand the possible anthropogenic emission sources in both Beijing and the surrounding area that influence the air pollution state in Beijing, the potential source contribution function (PSCF) method with the Hybrid Single-Particle Lagrangian Integrated Trajectory Model (www.arl. noaa.gov/ready) is used to determine the back trajectory of pollutants in Beijing. The meteorological field of Global Data Assimilation System (GDAS) with a spatial resolution of 1° × 1° and a temporal resolution of three hours is used. The results are presented in Fig. 5, where the darkness of each grid represents the probability of pollution sources. The darker the grid, the higher the probability. It can be seen that many darker grids appear to the south of Beijing within 500 km, implying frequently severe air pollution. A statistical analysis for the quantitative estimation of the pollution contribution is made for each trajectory and shown in Fig. 6. Heavy pollution cases with M_sa ≥ 300 μg m^-3 are presented in Table 1, where M_sa is the mass concentration of submicron aerosol. It can be seen that in all 413 heavy air pollution cases in Beijing, more than 80% can be attributed to trajectories 1, 2, and 7, of which trajectories 2 and 7 (mainly anthropogenic emission sources) account for about 48%. This implies that emission sources in the surrounding areas, particularly areas to the south, are responsible for heavy air pollution in Beijing, and the second peak in frequency distribution in Fig. 2 can be well explained here.

Figure 5

	Figure Option ViewDownloadNew Window
	Figure 5 Spatial distribution of emission sources and trajectories of pollutants.

Table 1

Table 1

Table 1 Statistics for heavy pollution cases ( Msa ≥ 300 μg m-3). Number Total number of trajectories Average Msa 1 136 482.04 2 120 638.56 3 41 392.67 4 5 402.20 5 0 0.00 6 32 388.22 7 78 451.03 8 1 393.85 Total 413 504.34

Table 1 Statistics for heavy pollution cases ( Msa ≥ 300 μg m-3).

For Beijing, haze events with a very high concentration of the submicron aerosol are mostly responsible for air pollution. We can see in Fig. 6 that Δ U and D (45°, 510 nm), which measures the scattering intensity at an angle of 45° and 510 nm by the ground submicron aerosol, are positive correlated, indicating that a high CO concentration appears simultaneously in haze events when it is contributed mainly by both local and regional emission sources. Moreover, an obvious difference in the slope of the correlation regression line can be found for the period before and after 2000. This indicates a change in the features of anthropogenic pollution sources in Beijing and the surrounding area for the last ten years.

Figure 6

	Figure Option '>ViewDownloadNew Window
'>	Figure 6 Correlation between ∆ U, anthropogenic part of CO content and D (45°, 510·nm), measured scattering intensity at angle of 45° and 510 nm of submicron aerosol near the Earth's surface in Beijing.