Continuous measurements of O3 and CO were made during 1997–1999 at Mondy, a remote mountain site in East Siberia, in order to quantify their mixing ratios and their climatology in the “background” troposphere of continental Eurasia. The seasonal cycles of O3 and CO show the spring maximum-summer minimum similar to that previously reported in the remote Northern Hemisphere. The influences of Siberian forest fires on the variations of CO mixing ratios at Mondy were observed both on a local and a regional scale during spring 1997 and fall 1998, respectively. We further evaluate the possible impact of European pollution export to the remote atmosphere of Siberia using trajectory analysis. It was found that the O3 and CO mixing ratios in the air masses transported from Europe are higher than those from Siberia and high-latitude regions for most of the year. The medians of O3 and CO mixing ratios associated with the European air masses are 44.2 and 134 ppb, respectively, in comparison with 42.7 and 128 ppb in the Siberian air masses, and 41.0 and 110 ppb in the high-latitude air masses. The residence time analysis of air masses transported from the European continent indicates that CO mixing ratios significantly decrease with longer transport time of air masses from Europe, while rapid air motion retains higher CO mixing ratios in every season due to the admixture of polluted European air into the continental background air during air mass transport over Eurasia and photochemical loss by OH. Because of a shorter lifetime in summer, CO mixing ratios decrease at a rate of 6–7 ppb per day, while they decrease at a rate of 2–4 ppb per day in winter and spring. The similar trend is found for O3 but only in summer, at a rate of 2–3 ppb per day. From this analysis, we are able to identify that European pollution exerts an influence, though not very strong, on the background O3 and CO at Mondy in remote Siberia/East Asia.
 In a global and regional perspective, the background region for atmospheric pollutants would be referred to as the area where the atmosphere is not directly perturbed by local and regional anthropogenic sources. Atmospheric observations in such a background region are very important as a reference for evaluating the anthropogenic impact of other regions. In the Northern Hemisphere, however, such regions are very limited due to the extensive human activities in most of the continental area. The possible background region thus defined may include the central Pacific, middle Atlantic, and central Eurasia. Among these three regions, atmospheric chemistry measurements in the continental eastern Eurasia are very scarce in contrast to the central Pacific and middle Atlantic, where substantial amounts of data have been obtained by both ground-based and aircraft measurements [see e.g., Fehsenfeld et al., 1996; Hoell et al., 1996, 1997, 1999; Penkett et al., 1998, and references therein]. It is known that even measurements of trace species, especially ozone and carbon monoxide, in the Pacific and the Atlantic are sometimes unavoidably influenced by the long-range transport of air pollutants on the hemispherical scale [Parrish et al., 1993, 1998; Jaffe et al., 1997]. For O3 and CO measurements in remote Siberia/East Asia, only a ground observation at a midlatitude Chinese WMO station at Waliguan (J. Tang, personal communication) and train measurement by trans-Siberian railway [Crutzen et al., 1998] have been carried out. No other continuous measurement at a remote site in East Siberia has been reported so far to our knowledge.
 In this work, we report for the first time the 3-year continuous data set of O3 and CO at Mondy, a remote mountain station in East Siberia. In addition, enhanced levels of O3 and CO due to the local/regional-scale biomass burning in Siberia are also reported.
2. Observations and Method
2.1. Monitoring Site
 The O3 and CO data used in this work were obtained from multiyear continuous monitoring at Mondy observatory site (51° 39N, 100° 55E, 2006 m above sea level) in central Eastern Siberia. The site is solar-terrestrial observatory station in the remote mountain ranges near a borderline of Russia-Mongolia southwest of Lake Baikal as shown in Figure 1. The nearest major city, Irkutsk, is about 250 km to the east with a population of 600,000. There are no other major cities within a 500-km radius from Mondy except small villages which spread lightly all over the region. The nearest residence to the observatory site is Mondy village which is located downhill at a distance of 15 km at 800 m asl elevation. A population of Mondy village is about 2200. This area of Siberia is known to be one of the very lightly populated areas of the world. No major anthropogenic pollutant, neither from point sources nor mobile sources, has been found to strongly affect this area. Crutzen et al.  defined Siberia as a pristine region where surface ozone levels should be close to their natural level. Mondy, thus, ideally represents one of the first true representative continental remote stations for monitoring background atmospheric trace components in the unpolluted region of Siberia/East Asia.
2.2. O3 and CO Measurements
 Measurement of O3 at Mondy has been conducted continuously since October 1996 and CO measurement has been carried out from March 1997. Mixing ratios of O3 and CO were measured using an UV absorption analyzer (Dylec Model 1007-AHJ) and a nondispersive infrared spectrometer (Kimoto Model 541), respectively. The similar criteria for instruments calibration and data handling were applied as for our other monitoring sites in East Asia. Details can be found elsewhere [Pochanart et al., 1999; Tanimoto et al., 2000]. In brief, O3 analyzer was calibrated by a primary O3 standard analyzer (TEII Model 49 Ps). During O3 measurement, a change in zero level of O3 analyzer was constantly checked using a zero air generator (TEII Model 111). As a calibration at site could not be carried out on a frequent basis, the analyzer was regularly cross-checked against three more of the same model O3 analyzers, one in the laboratory and two at other sites in East Asia, which were initially set to the same conditions as the Mondy instrument, and constantly calibrated. The CO instrument was set to automatically sample ambient air for 40 min following by 20 min of zero air in one hour cycle. The analyzer was calibrated using certified CO standard gas (Nippon Sanso, 1.2 ppm and/or 1.8 ppm), manually every day on a presence of staff and automatically once every 10–15 days when the site is left unattended. The detection limit of the O3 measurements is about the same as at the other sites, 1 ppb with the overall uncertainty of 5% at a mixing ratio of 20 ppb [Tanimoto et al., 2000]. The detection limit for CO, defined as three times the standard deviation of baseline fluctuation over 40 min period, can reach 2–3 ppb under the best circumstances but normally it is up to 10 ppb, which gives 10% overall uncertainties of hourly averaged CO data at 100 ppb mixing ratio. The raw data used in this work are 1-hour averaged mixing ratios. Examples of time series plots of the 1-hr averaged mixing ratios of O3 and CO during 1-year period from July 1997 to June 1998 are shown in Figure 2. For most of the year, the raw data at Mondy do not reveal any strong influence from local/regional pollution (Figures 2a–2l). Only for a few cases when the site experienced a rare pollution episode do the Mondy data show strong O3 and CO fluctuations, e.g., the plumes from local forest fires which were observed in the area during March–April 1997 (Figure 2m, see details later in text). We discard data from such events and do not further use them in routine data processing (e.g., calculation of daily average and monthly average). All times reported in this work are local winter times (+8 hours ahead of UT) except for trajectory calculation, which is at 0000 and 1200 UT.
2.3. Trajectory Analysis
 In order to identify the origins of the air masses arriving at Mondy and investigate the influence of the long-range transport on O3 and CO characteristics in Siberia, the 10 days backward isentropic air mass trajectory analysis has been conducted for two years periods during 1997–1998. The trajectory model, developed at National Institute for Environmental Studies (NIES), is the same as that used in our previous works [Pochanart et al., 1999, 2001]. The basic meteorological data set is the ECMWF/TOGA Basic Level III-A Upper Air provided by European Center for Medium-Range Weather Forecasts with the spatial resolution of 2.5 degrees in longitude and latitude and 15 levels in vertical resolution from 1000 to 10 hPa. We used clusters of five trajectories in one run, one started at the exact location and the other four being displaced by ±0.5 degree of latitude and longitude, to reduce the uncertainty in trajectory calculation [Pochanart et al., 1999]. Trajectories were calculated twice a day with the initial level at 2000 m elevation and the time step of 1 hour.
3. Results and Discussion
3.1. The Typical Air Mass Transport Patterns Arriving at Mondy
 From the trajectory analysis, we see that most of the air masses (>80%) at Mondy arrived from western section as a result of rapid air circulation in the midlatitudes of the Northern Hemisphere. The trajectory studies at 100°E longitude line by Newell and Evans  showed similar results (Mondy is coincidentally located at the exact longitude of 100° 55E). Trajectories are known to be good indicators of large-scale flow and can be useful for studying the potential regional sources of O3 and CO but a caution must be made here that uncertainties exist in trajectory model and any trajectory should not be viewed as the exact pathway of an air mass [Harris and Oltmans, 1997; Liu et al., 1999; Harris et al., 2000]. We further classified the typical types of air mass reaching Mondy into four groups according to their origins and transport pathways, namely Europe (EU), Siberia (SI), High-latitude (HL), and Southwest (SW) air masses. Figure 3 shows examples of the typical EU, SI, and HL trajectory categories during early spring and summer. The EU category is for the air masses originating in or passing through Europe (defined here as the west of Ural mountain at 60°E longitude) before arriving at Mondy (Figures 3a and 3b). The SI category is for long-range transport of air mass within Siberia (east of Ural Mountain) for the whole period of 10 days (Figures 3c and 3d). The HL air masses originated in the high-latitude and Arctic polar region (higher than 75°N) and were transported southward to Mondy (Figures 3e and 3f). The EU, SI, and HL categories, respectively, account for 27, 19, and 15% of all air masses arriving at Mondy for two years periods in 1997 and 1998 with the total numbers of 1460 runs (7300 trajectories). The altitudes of transport courses, as can be seen in the figure, normally varied from 2–6 km. The median value of the trajectory maximum altitude is about 680 hPa (3.2 km). We found only a few cases that air masses are directly transported from the higher level of troposphere, as the example is shown in Figure 3f. In total, only 4.4% of all air masses descend from a level above 450 hPa (around 6.3 km). The SW type (not shown) is the air mass transported with a much shorter range, usually less than several hundred kilometers from southwest direction. It significantly counts for 18% of the total frequency but most of the air masses in this type dropped to the trajectory based level (955 hPa) in the early days of backward calculation and were not further used in interpretation. The rest, 21%, is of unclassified type and other minor types. Most of the unclassified type comprises the mix pattern of the typical types showing dispersion into different directions of the five trajectories. The minor types are the transports with distinct pattern but infrequently found, such as air masses transport from the coastal regions of East Asia, or across the pole from Alaska. The percentages of minor transport types are very low and not considered significant for statistical analysis.
3.2. General Behaviors of O3 and CO at Mondy
3.2.1. O3 Seasonal Cycle
 The O3 seasonal variation observed at Mondy during late 1996–1999 is shown in Figure 4a. The O3 mixing ratios increase gradually from fall and winter to reach a maximum in April–May, 52.9 ppb on average, and then decrease to a minimum, 38.1 ppb on average, during the summer–fall. In winter, its mixing ratios are almost constant and the lowest variability is observed (also see, Figures 2f–2h). In summer, O3 exhibits the highest variability despite the low averaged mixing ratios. This variability is due to the diurnal and temporal change of summertime O3 (Figures 2a, 2b, and 2l). The 3-year annual averaged mixing ratio is 43.5 ± 5.2 ppb. The averaged mixing ratios of O3 (and CO) for each season are shown in Table 1. The interannual variations of O3 at Mondy, excluding O3 data influenced by local/subregional fires in spring 1997, are not found very large for each season. The averaged O3 mixing ratios during winter, summer, and fall in each year are almost constant. A slightly higher interannual variation is found in spring.
Table 1. Seasonal Averaged O3 and CO Mixing Ratios at Mondy During 1997–1999a
Mixing Ratio, ppb
Indicated values are average ±1 standard deviation. “Hours” indicates total hours of valid data from each year. Only winter data are not continuous, taken from January, February, and December of the same year.
 During summer and early fall, the averaged mixing ratios of O3 at Mondy are lowest. It was proposed that the photochemical loss process in the very clean atmosphere can be responsible for the summer minimum of O3 observed at some high-latitude sites in Europe [Solberg et al., 1997]. At Mondy, we suggest that the low summer O3 mixing ratios are due to the O3 sink by surface deposition and the more active vegetation in summer. It has been pointed out that the natural environment such as the huge forest can act as a net sink for O3 [Kirchhoff, 1988] although most of the observable cases take place in the tropics where suitable conditions for photochemical activities (high temperature, high humidity, and high solar radiation) exist. From a 3D-Global CTM study, a short photochemical lifetime of O3 and a lower transport flux of O3 during summer than spring and fall are found in the remote Eurasia [Wild and Akimoto, 2001]. Their results agree with Mondy O3 behavior, implying that the lower background troposphere of the continental Eurasia can be a sink of O3 in summer.
 It is noteworthy that some recent results from O3 measurements in the upper troposphere by the commercial airliners show the summer O3 maximum over Siberia, which is the opposite pattern to summer O3 minimum at Mondy [Brunner et al., 2001; Stohl et al., 2001]. Stohl et al.  analyzed O3 data from MOZAIC program using trajectory statistics and suggested that summer O3 maximum in the upper troposphere can be connected to a possible large O3 source, most likely the biomass burning, in the Central Asian/Eurasian continental boundary layer in summer. Brunner et al.  explain that a broad maximum of O3 in the upper troposphere observed during NOXAR program is associated with the seasonal cycle in tropospheric photochemical activity. A discrepancy between these results and Mondy data would be due to the tropospheric levels of O3 measurements. Mondy is a surface site at 2 km asl and can be strongly affected by surface O3 deposition. Large variations of O3 mixing ratios at Mondy in summer (see Figures 2a, 2b,2l, and 4a) imply that both O3 sources and sinks are strong but its averaged levels indicate that O3 sinks dominate over the sources compared to other seasons. The upper troposphere of Central Asia/Eurasia, however, is a different situation. Once O3 and its precursors, presumably emitted by forest fires, are convectively uplifted to the higher level of troposphere, their photochemical lifetimes extend as there is no strong O3 deposition in such level and the photochemical activities could become very effective. Thus, a summer O3 maximum in the upper troposphere over Siberia can be possibly observed in contrast with the summer minimum at surface levels.
3.2.2. CO Seasonal Cycle
 As shown in Figure 4b, CO mixing ratios decrease from a maximum in spring to a minimum in summer–fall and increase successively during winter similarly to O3 mixing ratios. The 3-year monthly averaged CO mixing ratios range between 90 and 174 ppb with an annual average of 127 ± 20 ppb. While both O3 and CO show the spring maximum and summer minimum, phase shift of the maximum period is observed. Normally, the maximum of CO seasonal cycle appears around March, one month earlier than for O3. The spring maximum of CO shows about a factor of two higher CO mixing ratios than in summer–fall minimum as can be seen in Figure 4b and Table 1. The difference of CO mixing ratios between spring and summer at Mondy can be explained partially by a destruction process with OH radicals that favors summer [Novelli et al., 1992, 1998a; Holloway et al., 2000]. The patterns of seasonal cycle of background CO mixing ratios at remote sites in the middle latitudes of the Northern Hemisphere do not show strong differences [Novelli et al, 1992, 1998a, 1998b].
 However, it should be noted here that the CO mixing ratios during summer–fall at Mondy are significantly lower than those previously reported in the middle latitudes of the Northern Hemisphere. In direct comparison with other of our monitoring sites (using the same instrumental system) in East Asia, apparently higher CO mixing ratios are observed in Japan [Kajii et al., 1998; Narita et al., 1999; Pochanart et al., 1999; Tanimoto et al., 2000] during fall, winter, and spring which can be interpreted as a direct effect of large-scale emissions in East Asia. The nearest NOAA/CMDL CO monitoring site to Mondy at Ulaan Uul, Mongolia (UUM, 44°N, 111°E, 914 m asl) [Novelli et al, 1998a; Holloway et al., 2000] shows slightly higher CO mixing ratios, by about 10–20 ppb. From trace gas measurements using the trans-Siberian railroad during Trans-Siberian Observations on Chemistry of the Atmosphere 2 (TROICA 2) campaign in Siberia/Russia in summer 1996, Crutzen et al.  found the lowest surface CO level of near 110 ppb over the sparsely populated area in west Siberia [see also Bergamaschi et al., 1998]. This value is about 15 ppb higher than summer–fall averaged CO mixing ratios at Mondy. These low CO mixing ratios would probably reflect the location of the site where the elevation represents closely the free troposphere, is less affected by the CO emission within the boundary layer, and is very far from regional anthropogenic sources of CO.
3.3. Evidence of Siberian Forest Fires at Mondy
 Recently, the biomass burning and forest fires in the midlatitude of the Northern Hemisphere have gained more attention in the field of global biomass burning. The large areas of burning in middle and high latitudes of the Northern Hemisphere (including Siberia, China, Canada and Alaska) were verified [Carhoon et al., 1994; Levine et al., 1995; Tanimoto et al., 2000; Wotawa and Trainer, 2000; Wotawa et al., 2001]. It was suggested that these boreal forest fires could significantly enhance the mixing ratios of O3 and CO in the region downwind from the fire [Phadnis and Carmichael, 2000]. In East Asia, the Siberian forest fires were suspected to cause a shift of CO spring maximum to the later period based on isotopic CO observations and model calculation at Happo, Japan [Kato et al., 2000]. In addition, trajectory analysis implied that the Siberian forest fires may have caused higher CO mixing ratios during spring and summer 1998 at Happo [Kato et al., 2002]. The direct capture of an elevated level of CO mixing ratios resulting from Far Eastern Siberian Forest fires was observed at Rishiri Island in northern Japan during summer–fall 1998 [Tanimoto et al., 2000].
 Mondy observatory site experienced the interference from forest fires in Siberia to the O3 and CO monitoring both on a local/subregional scale and a regional scale. In spring 1997, forest fires occurred in the area around Mondy and sometimes they could be detected just within a visible distance from the site. These fires started from late March and lasted for several weeks. Abrupt changes of CO and O3 mixing ratios at Mondy were usually observed when the fire plumes reached the site. The time series plots of 1-hour averaged O3 and CO mixing ratios in Figure 2m show such an example in April 1997. The arrows indicate the periods when forest fire plumes were transported downwind to the site and abruptly elevated CO mixing ratios were noticed. O3 mixing ratios also appeared to be influenced by these forest fires though not as distinct as CO mixing ratios. At the beginning of April when the fires were located far from the site, encounters of the plumes show O3 increase simultaneously with CO as a result of photochemical production by fire emissions (see examples during 4–5, 9–10 April in Figure 2m) but in late April when the fires moved closer to the site, O3 decrease was often seen with the abrupt increase of CO as a result of titration by NO emitted from fires (22–23, 25–26 April, in Figure 2m). For comparison, Figure 2j shows the O3 and CO mixing ratios in April 1998. In spring 1998, the local forest fires were not directly observed in the area and CO mixing ratios at Mondy remained constant for most of time.
 In contrast, although the satellite advanced very high resolution radiometer (AVHRR) revealed very active fire spots over large areas of Siberia from spring to fall of 1998 [Kato et al., 2002], we do not find a direct influence of those fires on CO mixing ratios at Mondy during spring and early summer. Enhancement of CO mixing ratios at Mondy was not observed in spring and early summer of 1998 (unfortunately, we lost CO data from the middle of July and the whole month of August 1998). However, during fall 1998 CO at Mondy shows about 40 ppb higher mixing ratios than its baseline levels in 1997 and 1999 (see Table 1 and Figure 4). This CO enhancement is likely associated with the extensive Siberian forest fires in 1998. The intense Siberian forest fires in 1998 were suggested to burn up larger areas in the Far East Siberia in summer–fall than areas in central and east Siberia [Kato et al., 2002]. Unlike the local/subregional fires of spring 1997 which were noticed together with the abrupt changes of CO signal in a short timescale, the impact of large-scale Siberian forest fires in 1998 would rather be seen as the increase of CO baseline levels as in case of CO data during fall 1998 due to the long range transport and atmospheric mixing in a regional scale. The CO anomaly at Mondy in 1998 agrees very well with the recent reports of CO enhancement in the Extratropical Northern Hemisphere [Wotawa et al., 2001] and in the high-latitude sites [Röckmann et al., 2002] by the large-scale forest fires in the Northern Hemisphere during summer and fall of 1998. Although the forest fires observation in the area around Mondy and the AVHRR data do not seem to be entirely compatible in 1998, the detection of forest fires in Siberia in this work supports the contention that the fires, burning areas, their emissions, and their impact on a background troposphere of the remote Siberia on both local and regional scale could be substantial and varied among different regions of Siberia and times of the year.
3.4. Influence From Transport of European Anthropogenic Emissions Toward Siberia/East Asia
 Based on trajectory analysis, the frequency of air masses transported from Europe to Mondy is about 27% annually. This value is comparable with the trajectory studies by Newell and Evans . Considering that the contribution of EU air masses is relatively large, the ability of regional scale anthropogenic emissions from Europe to pollute the atmosphere of Siberia/East Asia is of interest. Europe is one of the three largest source regions in the world and the influence of this large-scale anthropogenic emission within its own territory is well-recognized [see, e.g., Derwent et al., 1998; Ryall et al., 2001; Pochanart et al., 2001, and references therein]. However, the study of the impact of European emission transport toward other regions based on observational data received little attention, probably due to a deficiency of data in the region downwind. In this section, the following three analysis techniques were employed in order to identify this potential influence from European pollution export. First, we applied a clustering analysis of O3 and CO data in different air mass regions based on trajectories. Second, a probability distribution frequency of these trajectory categorized O3 and CO data was studied. The third technique concerns a residence time analysis of air masses leaving Europe. Details and results from the three approaches are presented below.
3.4.1. Cluster Analysis
 Hourly averaged O3 and CO data were paired with the corresponding trajectory categorized EU, SI, and HL air masses for each month during 1997–1998. These were done regardless of the air mass vertical distribution in each group. As more than 95% of air masses are transported below 450 hPa, a stratospheric influence is not thought to be a major concern for CO and O3 data analysis at Mondy. Seasonal variations of O3 and CO associated with different air mass regions are displayed in Table 2. The O3 and CO are shown as monthly averaged mixing ratios. Seasonal variations of O3 and CO in all three categories show the spring maximum-summer minimum, an analogous pattern to the overall seasonal cycle. This implies that seasonal cycles of O3 and CO in Siberia do not significantly depend on the different air masses regions, which are in contrast to the East Asian sites [Pochanart et al., 1999, 2001; Lam et al., 2001] and the European sites [Derwent et al., 1994, 1998; Simmonds et al., 1997]. The annual averaged O3 mixing ratios in EU, SI, and HL air masses, derived from monthly averages, are 44.3 ± 4.1, 42.4 ± 4.4, and 42.2 ± 3.8, ppb, respectively. Slightly higher O3 mixing ratios are observed in EU air masses than in SI and HL air masses, by about 2 ppb annually and 3 ppb during spring–summer. For a given error range (±1 standard deviation) one might argue that the differences of O3 mixing ratio in the three air mass regions are not significantly large. However, for the intercontinental scale transport, these values are comparable with the impact of trans-Pacific transports of Asian O3 to the US from a model study [Jacob et al., 1999]. For CO mixing ratios, the results also do not reveal large difference among the three categories. The annual averages in EU, SI, and HL air masses, derived from monthly values, are 127 ± 18, 128 ± 17, and 122 ± 13, ppb for 1998, respectively.
Table 2. Monthly Averaged O3 and CO Mixing Ratios Classified by Trajectoriesa
European Air Masses
Siberian Air Masses
High-Latitude Air Masses
Indicated values are average ±1 standard deviation in a unit of parts per billion mixing ratios. O3 and CO mixing ratios were calculated from January 1997 to December 1998, and July 1997 to December 1998, respectively.
40.8 ± 1.4
145 ± 14
42.3 ± 1.9
139 ± 15
42.0 ± 2.3
133 ± 13
46.0 ± 2.2
145 ± 13
42.3 ± 2.3
148 ± 15
45.1 ± 1.2
156 ± 11
47.0 ± 2.2
151 ± 18
47.9 ± 3.0
161 ± 18
48.3 ± 5.6
173 ± 10
51.8 ± 3.6
154 ± 16
51.2 ± 3.6
167 ± 19
49.1 ± 3.2
172 ± 11
53.4 ± 5.6
156 ± 23
50.1 ± 4.9
151 ± 19
46.5 ± 3.9
134 ± 9
46.9 ± 6.8
120 ± 21
38.4 ± 7.6
111 ± 20
39.8 ± 3.8
90 ± 10
43.3 ± 4.3
83 ± 13
36.7 ± 7.0
87 ± 13
40.6 ± 6.7
77 ± 15
36.7 ± 6.4
91 ± 12
37.7 ± 8.1
86 ± 14
38.1 ± 6.7
83 ± 12
41.8 ± 5.9
109 ± 21
40.7 ± 5.7
114 ± 19
34.8 ± 4.1
100 ± 10
40.5 ± 4.4
111 ± 18
37.1 ± 3.6
129 ± 19
40.1 ± 2.8
122 ± 17
41.4 ± 3.5
126 ± 25
41.8 ± 2.7
127 ± 21
40.5 ± 4.1
109 ± 16
41.8 ± 2.5
131 ± 18
43.3 ± 2.6
127 ± 23
42.1 ± 1.3
121 ± 16
 From the seasonal variations of O3 and CO in different air mass regions, the impact of long-range transport of polluted air masses from Europe may not be clearly seen. However, it is generally agreed that this cluster analysis will be useful only if there is a distinct difference in magnitude of the contribution from the different source regions (or trajectories). When the differences in magnitude among categories of trajectory are small, showing monthly averaged values might not prove to be useful. In case of Mondy, a very long distance between the possible sources (Europe) and the site could also prevent any clear evidence of pollution transport episode. The next approach was presented to confirm this possibility of European pollution affecting Siberia/East Asia.
3.4.2. Probability Distribution Study
 While the monthly averaged mixing ratios of O3 and CO do not reveal a large difference for EU, SI, and HL air masses, investigation of a distribution of 1-hour averaged data in each air mass region could provide some useful information about long range transport of the European pollution to Siberia/East Asia. Figure 5 shows the frequency distribution of O3 and CO mixing ratios at Mondy. As CO monitoring started in March 1997 and was influenced by the local forest fires in spring 1997, the data used were from July 1997 to December 1998. The histograms show more numbers of O3 and CO data with higher mixing ratios in EU air than in SI and HL air. O3 mixing ratios in the three air mass regions appear to be a normal distribution with the highest numbers of data in EU air. Below 35 ppb of O3, the amounts of data for each air mass region are approximately the same but the numbers of data in EU air increase notably at a higher than 35 ppb of O3. CO data show broader distribution but their probability distribution functions still appear to be a normal distribution. From CO frequency distribution plots, the numbers of CO data below 120 ppb are about the same for each air mass region but significantly increase in EU air for CO higher than 120 ppb. As shown in Figure 5b, at 50th percentile (median) of the entire data in EU, SI, and HL group, the O3 mixing ratios are 44.2, 42.7, and 41.0 ppb, respectively. The corresponding CO mixing ratios in Figure 5d are 134, 128, and 110 ppb.
 Cumulative probability distributions for each group are further plotted as shown in Figure 6 for the entire data and seasonally separated data. For the entire data set, the higher O3 and CO mixing ratios in EU air at most cumulative percentages are noticed. The same feature is also found with the seasonal data. Discarding the upper and lower ends of the data (<5th percentile and >95th percentile), O3 and CO at any percentile in EU air show significantly higher mixing ratios than those in SI and HL air masses in most seasons. The orders from the highest mixing ratios appear to be EU, SI, and HL air, respectively. The largest difference of O3 mixing ratios among air regions is observed in summer, suggesting that the EU air mass is more photochemically active than the SI and HL air in this season of the year. For CO, the largest difference is found during fall but summer and winter also show the significant difference, especially between EU and HL air. The only exception is seen in spring when CO in HL and SI air show higher mixing ratios than EU air at the same percentiles. At the moment, we still have no evident explanation for the higher springtime CO transported by HL air mass but it is generally known that the high CO mixing ratios are often found in the Arctic region during late-winter and early spring as a result of transport of anthropogenic pollutants from northern Europe into the polar vortex [Novelli et al., 1998a]. For SI air mass, the forest fires during spring 1998 in western and central Siberia might induce the higher CO baseline. These two events, in combination with an uncertainty due to fewer hours of valid data (spring data were from 1998 only), may bring about higher CO mixing ratios in HL and SI air masses during spring, respectively.
 Based on this frequency distribution, it can be concluded that for most seasons of the year the EU air masses reveal higher O3 and CO mixing ratios than the SI and HL air masses, likely due to the long-range transport of pollution from Europe although the magnitude of mixing ratios differences is not very large. We have provided here the evidence that Europe pollution does have influence on the background O3 and CO at Mondy. In the following part, we further investigate this possible influence of European pollution using trajectory with residence time analysis.
3.4.3. Residence Time Analysis of Air Mass From Europe
 As mentioned, EU air masses are those originating in or passing through Europe before arriving at Mondy. The relationship between transport times of air masses after leaving Europe until arriving at Mondy and CO and O3 mixing ratios in the EU air region is investigated. We applied a technique similar to that described by Pochanart et al.  to CO and O3 mixing ratios at Mondy. This simple technique assumes that for the same group of EU air mass, one would expect to see a difference or gradient of trace gases mixing ratios between those by a rapid air mass transport from Europe and a slowly transported one if an export of European pollution and a long range transport are the only contributing factors for those trace gases variations.
 Europe in this investigation is defined as the areas between west of Ural mountain (60°E longitude) and the west coast of the European continent. These large areas cover the western and eastern European countries, and parts of Russia and FSU countries. The averaged transport times of the clusters of 5 trajectories after leaving this “European wall” until arriving at Mondy (or to say so, the residence times over the remote Eurasian continent) were then calculated. Data from July 1997 to December 1998 are applied. This one and a half year period gives total numbers of 1098 trajectory clusters in which 293 clusters were from Europe. With a removal of about 10% possible outliers, 268 clusters are used in calculation. The trajectories are then matched with 12 hours of CO and O3 mixing ratios at the times air masses arrived at Mondy on a seasonal basis. It must be noted that because of the rather limited amounts of trajectories (about 60–70 per season), we do not further stratify whether the air masses come from the boundary layer or free troposphere of Europe. A general impression is that the free tropospheric motion shares a larger fraction in total transport frequencies (see also, Newell and Evans ) but the altitudes of trajectory courses are not found to be very high. It is found that 53% of the EU air masses descend from levels between 400 and 700 hPa, and 44% are from the levels below 700 hPa. On average, the maximum elevation of EU air masses transport routes during 10-day period is about 3.9 ± 1.4 km. Examples of the relationship between CO mixing ratios and averaged transport times of the air mass clusters after leaving Europe in summer and winter are shown in Figure 7.
 It can be seen that CO mixing ratios gradually decrease when it takes longer time for air masses to reach Mondy. The rapid transport of air masses from Europe results in higher mixing ratios of CO. This appears to be a common feature for CO in every season. We also see the same trend for O3 in some seasons. To further simplify the relationship between trace gases and residence times, we grouped the transport times of EU air masses after leaving Europe into unit of days, namely, 1 day, 2 days, …, up to 8 days. These transport times were then plotted against the averaged mixing ratios of the corresponding CO and O3 data in each group. The results are shown in Figure 8. The CO mixing ratios exhibit a significant correlation with the transport time of air masses from Europe in every season. For O3 mixing ratios, the correlation is seen only in summer and spring. CO is known to be good anthropogenic tracer [Parrish et al., 1998]. Decrease of CO mixing ratios over longer transport times of air masses from Europe primarily reflects a dilution of European polluted air, high in CO, into the well-mixed background troposphere of lower CO over Eurasia, and the photochemical loss of CO assuming that no additional CO from other emission sources is advected into air during transport of the polluted EU air to Mondy. This admixture of CO in European air into continental background air would occur on the timescale of days.
 Linear regressions have been plotted to estimate CO decreasing rates. In Figure 8, we compared the results from a simple linear regression presented as dotted lines with those of Reduced Major Axis (RMA) regression presented as solid lines. It was suggested that the evaluation of atmospheric traces data would be better achieved using the RMA analysis [Ayers, 2001]. In our case, the results between two methods do not show a strong difference. The slope represents the CO decreasing rate and the Y axis intercept gives the CO mixing ratio when the air masses leave Europe, and presumably represents the CO mixing ratio under “European regionally polluted” condition.
 For the estimated CO in European regionally polluted condition (Y axis intercept), its mixing ratios are lowest in summer, about 136 ppb, and increase to highest in spring, 168 ppb. As no ground-based CO monitoring around this “European wall” was reported, we cannot verify whether these estimated values properly represent the “European regionally polluted” CO mixing ratios. In comparison with spectroscopic column CO at Zvenigorod (37°E), Russia [Yurganov et al., 1997] and CMDL/NOAA surface European sites [Novelli et al., 1998a; Holloway et al., 2000], our estimated values agree quite well with data from Zvenigorod, Iceland, and Mace Head.
 The slopes of CO-residence time plots are steeper in summer and fall, about 6–7 ppb per day. Less steep slopes are observed in spring and winter, 2–4 ppb per day. The faster CO decreasing rate in summer would be explained by the shorter CO lifetime due to OH loss, and would reflect the inhomogeneous CO distribution among different regions in Eurasia and the effective mixing of European regionally polluted air into Eurasian continental background air. With the loss rate expressed as k[CO][OH]; with k = 2.5 × 10−13 cm3 molecule−1s−1 [DeMore et al., 1994]; [CO] = 136 ppb (estimated European regionally polluted CO mixing ratio in summer, Y-intercept); [OH] = 6.8 × 105 molecule.cm−3 (case 1: low CO loss scenario, July averaged OH concentration at 52°N, 1000 hPa) or 12.4 × 105 molecule.cm−3 (case 2: high CO loss scenario, July averaged OH concentration at 52–68°N, 1000–700 hPa) [Spivakovsky et al., 2000], the CO loss by OH in summer could be roughly estimated as 1.8 ppb/day (case 1) −3.6 ppb/day (case 2). It appears that the reaction with OH accounts for about 27–54% of total CO loss during transit. In winter, the lifetimes of CO are much longer. Our estimation indicates that reaction with OH accounts for only 4–6% of CO loss during air transport in winter. Therefore, it appears that the atmospheric diffusion is a main explanation for this feature. A more homogeneous CO distribution in the Northern Hemispheric middle latitudes gives a less steep slope of CO-residence time plots in the winter.
 O3 mixing ratios show negative correlation with transport times of air masses from Europe during summer and spring but no correlation during fall and winter. The higher slope, 2–3 ppb per day, is observed during summer. The springtime slope is about 1 ppb per day but with much lower correlation coefficient (R = 0.79) suggesting that this O3 decrease rate may not be very significant. A decrease of O3 in summer with longer air mass residence time would be explained by the O3 deposition owing to shorter photochemical lifetime during transport over Eurasia. A noncorrelation during winter is not unexpected as the photochemistry is suppressed in this period. It is noteworthy that while Mondy shows O3 seasonal cycle with a spring maximum and summer minimum, the extrapolation of O3 to day 0 (Y axis intercept) which could be regarded as “European regionally polluted” condition gives the O3 seasonal cycle with a spring–summer broad maximum, identical to the O3 behaviors normally observed in polluted regions of Europe [see e.g., Low et al., 1990; Beekmann et al., 1994; Kley et al., 1994; Marenco et al., 1994; Staehelin et al., 1994; Laurila, 1999; Monks, 2000, and references therein].
 In addition to the above explanation by air mass dilution and photochemical loss, another alternative possibility concerning the uncertainties of trajectories is also suggested. Trajectories are known to increase an inaccuracy over times. Therefore, for the EU air masses with longer residence time over Eurasia there is higher probability that their actual transport courses might be deviated and the air masses might be transported from other regions where the emissions are less intense than Europe, and thus, the lower CO mixing ratios would be observed. However, though this possibility exists, it still demonstrates the influence of European pollution on the background troposphere of Siberia since the EU air masses that actually come from Europe (those with shorter residence times over Eurasia) clearly show higher CO mixing ratios than those that may not pass over Europe (those with longer residence times).
 From the above analysis, we conclude that the export of European pollution does have impact on the background CO and O3 mixing ratios in Siberia/East Asia. Continental background O3 and CO levels in Siberia are at least partly influenced by the admixture of long-range transport of European emissions. Validation of these results is confirmed by reexamination of the data set with a stringent criterion. We confine the boundary of “European continent (60°E)” to only “Western Europe (region from the Atlantic Ocean coast to 30°E, 35°N to 60°N)” which is known to be the most intense anthropogenic region in Europe based on emission inventory report [EMEP, 1996]. The negative correlation between the transport times of air masses after leaving Western Europe and CO mixing ratios at Mondy is confirmed although there is a decrease of correlation coefficients, from about 0.90–0.98 to 0.62–0.78, in every season.
 This evidence supports the contention that the remote troposphere of Siberia/East Asia is partly perturbed by European anthropogenic sources. Nonetheless, the impact of European pollution toward Siberia may not be strong as what was observed from the long range transport of air mass from North America to the Atlantic, or East Asia to the Pacific rim region [see, e.g., Parrish et al., 1998; Nagao et al., 1999; Pochanart et al., 1999; Lam et al., 2001]. A lack of evidence that European anthropogenic emissions exert a strong influence on the atmosphere of Siberia/East Asia from this work may have two reasons. One is the long residence times of air masses transported over Eurasia and the well-mixed condition of air masses in Siberia which could reduce any impact of pollution transport. The other is that unlike the boundary layer pollutions of East Asia or North America which are known to potentially affect the downwind regions [Jaffe et al., 1997; Parrish et al., 1998], European boundary layer pollution may contribute much less in term of long-range transport. Newell and Evans  pointed out that a contribution from European boundary layer to East Asian atmosphere is less than the European free tropospheric transport. They further noted that European air masses influence on Asia is maximum in winter and minimum in summer. The polluted boundary layer in winter, especially in Europe, is suggested to be an O3 net sink [Derwent et al., 1998; Solberg et al., 1997]. Meanwhile, the CO lifetime in winter is in order of several months. Thus, the ability of pollution from Europe to influence Siberia/East Asia by the long-range transport may be not dominant. The preliminary FRSGC/UCI three dimensional global CTM analysis suggests that European sources may contribute from 2 ppb in winter to 5 ppb of O3 at Mondy in summer, and about 10 ppb to 25 ppb of CO from summer to spring, respectively. (see model details given by Wild and Akimoto ; O. Wild, personal communication). In order to be able to capture the pollutant episode from Europe, we suggest that more continuous data would be essential and the collaborate O3 and CO monitoring in the far western section of Siberia (closer to Europe) is needed.
 From the analysis of O3 and CO data obtained at Mondy, a remote mountain site in Siberia, the following conclusions are found.
The O3 and CO mixing ratios at Mondy show a seasonal variation with a spring maximum and summer minimum, similar to those previously observed in the remote Northern Hemisphere. The observed O3 and CO mixing ratios are thought to be the representative continental background O3 and CO in remote Eurasia. Monthly averaged continental background O3 mixing ratios range between 38.1 and 52.9 ppb during 1997–1999. The 3-year monthly averaged continental background CO mixing ratios range between 90 and 174 ppb.
The evidences of forest fires, both on a local/subregional scale and regional scale, were observed at Mondy. The local forest fires during spring 1997 were found to have an abrupt disturbance on the variations of O3 and CO mixing ratios at Mondy on a short timescale. The extensive regional-scale Siberian forest fires in 1998 were found to enhance CO mixing ratios at Mondy during fall 1998, which increased by 40 ppb from the 1997 and 1999 levels. Our finding supports the significance of the impact of boreal forest fires in Siberia and their emissions to a continental background atmosphere. However, AVHRR fire spots over Siberia and the ground-based CO observation at Mondy during spring and summer 1998 did not show a good agreement.
Trajectory analysis confirms that O3 and CO in the air masses transported from Europe show higher mixing ratios than those transported from the high latitudes and within Siberia based on 10-day backward air mass transport. It was further found that using the residence time analysis, CO mixing ratios in the European air masses decrease significantly when the transport time to the site is longer. The decrease rate is found highest in summer, 6–7 ppb per day, and lowest in winter, 2–4 ppb per day. The similar trend is observed for O3 but only in summer, at a rate of 2–3 ppb per day. Decreases of CO and O3 over the longer transport times are explained by the admixture and dilution of European pollution into the continental background troposphere, and a seasonal change of the photochemical lifetimes of O3 and CO. Our results are among the first observation-based evidences proving that the export of large-scale European anthropogenic pollutions does have impact on the remote atmosphere of Siberia. Continental background O3 and CO levels in Siberia are thought to be partly influenced by an admixture of these European emissions during the long-range transport of air masses over Eurasia.
 The authors would like to thank J. Hirokawa of the University of Tokyo and V. A. Obolkin of the Limnological Institute for their help in setting up the monitoring site at Mondy. We acknowledge the local staff at Mondy station for their support. S. Maksyutov of Frontier Research System for Global Change (FRSGC) is acknowledged for providing trajectory analysis program. We thank O. Wild of FRSGC for his helpful discussion. We are grateful to two anonymous reviewers for their comments and corrections. The observational part of this research has been funded by Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Corporation (JST).