Journal of Geophysical Research: Atmospheres

Emissions of ozone-depleting substances in Russia during 2001



[1] There is a long-standing need for measurement-based estimates of the emissions of ozone-depleting substances (ODSs) in Russia. This need arises from >20 years of globally important ODS manufacture in Russia that reportedly ceased in December 2000, for which only aggregated, unaudited production figures are available. The integrity of these production figures is questionable because, for nearly the last decade, the global emissions of several important ODSs estimated from global production figures (production-based estimates) have been insufficient to account for their measured atmospheric burdens. Are these shortfalls in worldwide production-based estimates the result of Russian emissions that are inordinate relative to the reported production figures? We estimate Russian emissions of six ODSs (chlorofluorocarbon-11 (CFC-11, CCl3F), CFC-12 (CCl2F2), CFC-113 (CCl2FCClF2), carbon tetrachloride (CCl4), methyl chloroform (CH3CCl3), and halon-1211 (CBrClF2)) from thousands of measurements of their mixing ratios along 8500 km of the Russian trans-Siberian railway in June–July 2001. Our measurement-based estimates indicate that Russian emissions in 2001, even if grossly underestimated because of underreported production, were insufficient in magnitude to play a major role in recent global emission shortfalls. The results also corroborate the reported termination of CFC production in Russia at the end of 2000. The large CFC-12 emissions observed in Russia suggest that a recent estimate of the global CFC-12 reserve is too small.

1. Introduction

[2] The atmospheric loading of ozone-depleting halogens is currently in decline because of large reductions in the global production and emissions of ODSs by developed countries during the 1990s [Elkins et al., 1993; Montzka et al., 1996; Cunnold et al., 1997; Montzka et al., 1999; United Nations Environment Programme (UNEP), 2002a; Yokouchi et al., 2002; Montzka et al., 2003]. Production of CFCs, halons, carbon tetrachloride, and methyl chloroform ceased in most developed countries before the 1 January 1996 production deadline (1994 for halons) in compliance with the 1987 Montreal Protocol for Substances that Deplete the Ozone Layer and its subsequent amendments [UNEP, 1998, 2001]. Despite these reductions, considerable emissions of these and other ODSs continue to emanate from ongoing production in developing (Article 5) countries [UNEP, 2002a, 2002b], allowed under the Protocol until 2010–2015, and from previously produced, unemitted reservoirs (banks) around the globe [World Meteorological Organization (WMO), 1999, 2003].

[3] The six ODSs discussed in this work have wide ranging applications. CFC-11 is primarily a foam blowing agent, CFC-12 is a refrigerant, halon-1211 is a fire extinguishing agent, and CFC-113, CH3CCl3, and CCl4 are solvents. Carbon tetrachloride is also the precursor chemical (feedstock) for the manufacture of CFC-11 and CFC-12. On the basis of their primary applications these 6 ODSs are classified as banked or nonbanked. CFC-11 and CFC-12 are extensively banked in blown foam products, refrigerators, and air conditioners while halon-1211 is held in hermetic fire extinguishing systems. These compounds typically remain in sealed systems for several years before escaping to the atmosphere [Fraser et al., 1999; McCulloch et al., 2001, 2003]. The three solvents are generally released to the atmosphere within a year of their production and are consequently considered nonbanked ODSs [Fraser et al., 1996; Simmonds et al., 1998; McCulloch and Midgley, 2001], although industrial and household stockpiles may exist.

[4] Models that predict future stratospheric ozone abundance are strongly dependent on estimates of past, current, and future global ODS emissions as input. The reliability of current ozone projections is reduced by ongoing discrepancies between production-based estimates of global CFC-11, CFC-113, and halon-1211 emissions [e.g., McCulloch et al., 2001] and those based on measured atmospheric burdens (burden-based estimates) [e.g., Fraser et al., 1996; Cunnold et al., 1997; Fraser et al., 1999; Prinn et al., 2000]. The current discrepancies began in 1994 and have increased in magnitude at least through the year 2000, when production-based global emission estimates of each of these three important ODSs were 30–60% lower than burden-based estimates [WMO, 2003]. Shortfalls of similar magnitudes also occurred in the late 1980s for CFC-12 and CFC-113, and throughout the 1980s for CFC-11, but by 1990 production-based and burden-based emission estimates for these ODSs were in agreement.

[5] Uncertainties in global ODS production figures have widened since 1994–1995 when the bulk of worldwide manufacture shifted from companies who voluntarily report their compound-specific CFC production to the Alternative Fluorocarbons Environmental Acceptability Study [AFEAS, 2002] to nonreporting companies in the former Soviet Union (FSU), China, India, and the Republic of Korea. These countries do submit Protocol-mandated annual production figures to the UNEP, but these are unaudited, ozone depletion potential (ODP)-weighted, aggregated production figures for the broad Protocol ODS classifications (Annexes). Though the audited AFEAS CFC production data are reputed to be more accurate than UNEP figures, they cannot yield comprehensive global production inventories without the speculative division of aggregated UNEP production figures for countries with nonreporting companies into compound-specific production estimates. The concomitant onset of shortfalls in production-based emission estimates and the rapid shift in global production to nonreporting companies has prompted conjecture of underreported ODS production in the last decade by Russia, China, India, and the Republic of Korea.

[6] The former Soviet Union, an important producer of ODSs since 1980 or earlier [Gamlen et al., 1986], accounted for 10 and 14% of global CFC and halon production in 1986 [UNEP, 2002a]. In 1988 the FSU ratified the Montreal Protocol as a non-Article 5 (developed) country. After the 1991 political collapse of the FSU, Russia affirmed it would adhere to the Protocol production deadlines of the FSU [UNEP, 1998]. Russian CFC and halon production reportedly fell by 53% (84 to 39 ODP-Gg (109 g) yr−1) and 90% (11 to 1 ODP-Gg yr−1) between 1991 and 1995, but did not cease until December 2000 because of financial difficulties encountered during the transition to a market economy [UNEP, 1998, 2001, 2002b]. This breach of the Protocol has not served to quell conjecture that Russia might underreport its ODS production.

[7] The primary objectives of this work were to estimate the magnitudes of modern Russian ODS emissions from a geographically extensive set of in situ atmospheric measurements and to ascertain whether gross underestimation of recent Russian production and emissions might have caused the recent shortfalls in global production-based emission estimates.

2. Methods

2.1. Measurements

[8] Thousands of in situ measurements of CFC-11, CFC-12, CFC-113, CH3CCl3, CCl4, halon-1211, carbon dioxide (CO2), carbon monoxide (CO), radon 222 (222Rn), and several other trace gases were made along the Russian trans-Siberian railway during 27 June to 10 July 2001, as part of the seventh Trans-Siberian Observations Into the Chemistry of the Atmosphere (TROICA-7) expedition. The six ODSs were measured by the four-channel Airborne Chromatograph for Atmospheric Trace Species (ACATS-IV) [Elkins et al., 1996; Romashkin et al., 2001] for 250 hours of the 13-day, 17,098 km journey from Moscow to Khabarovsk and back (Figure 1). Independent analyzers were used to measure CO2 (Licor model 6262), CO (Thermo Environmental model 48S), and 222Rn (TracerLab low level radon daughter monitor) as described by Oberländer et al. [2002]. Instruments were housed in a specially equipped laboratory carriage coupled immediately behind the electric locomotive. Outside air was rapidly pumped to instruments through tubes mounted to the roof and side of the carriage approximately 2–4 m above the rail bed.

Figure 1.

Route of the trans-Siberian railway traveled during TROICA-7 from Moscow (55.8°N, 37.7°E) to Khabarovsk (48.5°N, 135.1°E) and back. Each city labeled on the map had a 1990 population between 300,000 (Chita) and 1.5 million (Novosibirsk) except for Moscow (9 million). The six source regions that influenced our measurements during the western transect are shown as alternating orange and yellow areas (see section 2.3). The conjoined source regions comprise a 611,100 km2 rail corridor (Table 1) with a population of nearly 10 million people.

[9] The railway between Moscow and Khabarovsk passes through several industrialized population centers (Figure 1) as well as pristine rural environments. It is fully electrified, permitting measurements of atmospheric constituents without discernable contamination from the locomotive, as demonstrated by six previous TROICA expeditions that examined air quality and trace gas emissions at local and regional scales [Crutzen et al., 1998; Bergamaschi et al., 1998; Röckmann et al., 1999; Oberländer et al., 2002]. ODS measurements had not been made from the train prior to TROICA-7, so it was imperative to examine the possibility that train-based sources could have influenced the ODS data.

[10] Train-based sources of CO2, CO, and the six measured ODSs included the small coal-fired samovars and refrigeration units in some of the trailing carriages. The fact that 24–57% of the TROICA-7 data for these gases were concordant with their “baseline” mixing ratios at remote, northern midlatitude sites during summer 2001 (G. Dutton et al., personal communication, 2003) rules out continual contamination of our measurements by train-based sources. The percentages of baseline data increased by 10–20% when we examined only the data obtained when the train was in motion (>30 km h−1), but decreased an average of 10% when only measurements made during stops were considered. This leaves open the possibility that train-based sources behind the laboratory carriage may have influenced measurements made from the stationary train. Baseline data percentages were an average of 13% lower at 231 station stops and 10% higher at 53 nonstation stops than the percentages for the entire expedition data set, implicating sources proximate to train stations, not aboard the train itself as the cause of higher mixing ratios when the train was stopped.

[11] Similar analyses of “polluted” mixing ratio data, those above the 80th percentile of the complete expedition data set, were performed for the same data subsets (train speeds, stop locations) described above. Polluted data percentages were 14–21% for the moving train, 16–46% for the stationary train, and were on average 20% higher at station stops than at nonstation stops. These observations support the idea that sources near train stations influenced our measurements from the stationary train. Though these analyses provide strong evidence that non-train-based sources had the greatest influences on the data, they cannot conclusively prove that all the measurements were unaffected by train-based sources. To avoid influences from sources at train stations, measurements made at train stations were not used in this analysis.

2.2. Determination of Emission Ratios and Average Fluxes

[12] One notable characteristic of the TROICA-7 data is the coincidence of increased trace gas mixing ratios and the presence of shallow, stable nocturnal boundary layers indicated by low-height (<100 m) temperature inversions (Figure 2). Almost every night of the expedition, emissions from surface sources were trapped in the nocturnal stable layer, where they accumulated until sunlight-driven convection mixed them into a deepening boundary layer. The resulting diurnal cycles in the mixing ratios of gases provide germane information about their relative emission strengths in the source regions that influenced the measurements.

Figure 2.

Radon 222 activity, CO2 mixing ratio, and temperature inversion intensity (T100m–T0m) data obtained during the western transect. Temperature measurements were made 100 and 0 m above the train by a vertically scanning radiometer mounted on the carriage rooftop [Oberländer et al., 2002]. Temperature inversions, observed predominantly at night, signify the presence of shallow boundary layers that accumulated gases emitted by surface sources. Data are plotted as a function of longitude-based local time to account for the 6.5-hour change in the daylight cycle (97° of longitude) between Khabarovsk and Moscow. The six 24-hour diurnal periods of the WT are depicted by alternating orange and yellow bands along the x axis that correspond to the source regions in Figure 1. Dashed vertical lines denote midnight local time.

[13] Emission ratios of ODSs and CO2 were calculated from their mixing ratio increases coincident with increases in 222Rn activities [e.g., Kuhlmann et al., 1998; Conen et al., 2002] during six different 24-hour diurnal periods of the 6.5-day western transect. Only western transect (WT) data were utilized because low-height temperature inversions were absent two nights of the eastern transect (ET), making gas accumulations negligible, and because 26% more ODS measurements were made during the WT than the ET. The 24-hour diurnal periods analyzed (15:00 to 15:00 local time) were chosen to encompass the complete lifetimes of the nocturnal stable layers and hence the diurnal cycles of the measured gas mixing ratios.

[14] Emissions of ODSs and CO2 were gauged against 222Rn emissions because the soil source of 222Rn, the radioactive decay of the radium 226 present in all soils, is fairly uniformly distributed and well quantified at regional and larger scales [Nazaroff, 1992]. ODS and CO2 data for each diurnal period, excluding data obtained at train stations, were plotted against coincident 222Rn data and fit with a line whose slope is the spatially averaged emission ratio for that period (Figure 3 and Table 1). The y-intercepts of fit lines were all <3% different from baseline ODS mixing ratios because the linear fits were well constrained by data obtained in nonpolluted regions during the daytime. Points with residuals >3 times the residual standard deviation were discarded and the data refit to minimize the influences of any isolated ODS or CO2 emissions from local sources. No attempt was made to correct for the <7% decay of 222Rn during the 10-hour mean lifetime of the nighttime temperature inversions during the WT.

Figure 3.

Emission ratio plots for CO2, CFC-12, halon-1211, and CFC-113 against 222Rn for the 6 diurnal periods of the western transect. Each datum for CO2 is an average of near-continuous measurements made over the 10-min period of an integrated 222Rn measurement. ODS data are averages of 4–9 discrete measurements made during the 222Rn measurement period. Data averages containing measurements at train stations were discarded. The data in each panel were fit using a linear orthogonal distance regression that accounts for errors in both the x and y variables [Press et al., 1992]. Data with residuals >3 times the RSD were discarded, and the retained data (solid circles) were refit. On average this process discarded 1.2 data points (1.3%) per panel and reduced emission ratio uncertainties by 18%.

Table 1. Emission Ratios in the Trans-Siberian Rail Corridor During the Westbound Transect
DiurnalDistance,aSource Area,bEmission Ratioc
  • a

    Distance traveled by the train during each 24-hour diurnal period. The total distance of 8067 km is shorter than the 8549 km WT because measurements were not made in the first hour after departure from Khabarovsk at 03:45 local time on 5 July, and 7 hours of data collected after 15:00 on 10 July were not used in this work because there was very little increase in 222Rn activities or ODS mixing ratios before the train arrived in Moscow at 22:00.

  • b

    Areas of the source regions that influenced our measurements during each diurnal period (see section 2.3).

  • c

    Regional-average emission ratios for CO2 and ODSs are given in units of ppm/(Bq m−3) and ppt/(Bq m−3). Values in boldface type are significantly different from zero at the 95% level of confidence shown in parentheses. Emission ratios insignificantly different from zero were not used in the calculation of ODS emissions in the rail corridor.

5 July692452007.3 (1.4)−0.006 (0.100)1.0 (0.5)0.001 (0.029)0.003 (0.004)−0.014 (0.032)
6 July13318790915.1 (3.0)−0.004 (0.244)5.7 (1.3)0.063 (0.054)0.067 (0.016)−0.024 (0.068)
7 July13178317511.2 (2.2)−0.005 (0.491)2.6 (3.0)0.036 (0.110)0.045 (0.022)0.005 (0.122)
8 July147216577510.8 (2.4)0.055 (0.040)1.0 (0.4)0.029 (0.018)0.007 (0.003)−0.024 (0.024)
9 July17079058012.6 (2.7)0.077 (0.371)2.9 (1.5)0.042 (0.062)0.039 (0.019)0.102 (0.085)
10 July15481384349.2 (1.3)0.080 (0.052)2.5 (1.2)0.030 (0.039)0.021 (0.016)−0.039 (0.058)

[15] Regional average fluxes were determined for each diurnal period as the products of emission ratios and the global average terrestrial flux of 222Rn (76 Bq m−2 h−1). This global average value is adopted in the absence of measured 222Rn fluxes in the rail corridor, and is known to differ from regional average 222Rn fluxes by up to a factor of 2 [Jacob et al., 1997]. This potential error introduces large uncertainties (+100%, −50%) in the average ODS flux values and ultimately in our regional-scale emission estimates. ODS and CO2 fluxes were determined only for those emission ratios significantly different from zero at the 95% (2σ) confidence level.

[16] These calculation methods were tested by computing regional average emission ratios and respiration fluxes of CO2, using the TROICA-7 data, for direct comparison with respiration flux values reported for summertime in temperate ecosystems. The determination of CO2 respiration fluxes required that the influences of local- to regional-scale combustion sources on CO2 mixing ratios be minimized. We used coincident CO measurements and a CO:CO2 product ratio of 0.02 mol mol−1 for fossil fuel combustion [Logan et al., 1981; Bakwin et al., 1994] to estimate and remove recent combustion source enhancements in CO2 mixing ratios. Biomass combustion, with typically larger CO:CO2 product ratios, was not considered a regional-scale CO2 source because smoke or haze from biomass fires was seldom encountered during the expedition. This treatment of the CO2 data reduced mixing ratios by an average 3.4 ppm (0.9%) and changed <5% of the data by >10 ppm.

[17] The CO2 emission ratio for each of the 6 diurnal periods of the WT was significantly different from zero (Table 1). The resultant CO2 respiration flux values ranged from 6 to 13 μmol CO2m−2 s−1 (mean ± 1σ = 9 ± 2), very much in accord with values of 4–10 μmol CO2 m−2 s−1 reported for North American mixed hardwood and boreal aspen forests during summertime [Goulden et al., 1996; Black et al., 1996]. This good agreement lends credence to our methods of calculating regional average emission ratios and fluxes from the TROICA-7 data, and suggests that differences between regional-scale 222Rn fluxes in the rail corridor and the global average flux value were less than a factor of two.

[18] In contrast to CO2, not all of the emission ratios calculated for ODSs were significantly different from zero (Table 1). Emission ratios insignificantly different from zero may indicate weak or nonexistent emissions of an ODS in the rail corridor, as is believed the case for CH3CCl3 for which no significant emission ratios were determined for TROICA-7 (see section 3.1). Statistically insignificant emission ratios may also be the consequence of considerable scatter in the ODS:222Rn correlations that would result from imprecision in the measurements of ODS and 222Rn increases, local ODS emissions, and disparate distributions of the ODS and 222Rn sources. Relative to the precision of measurements, the nighttime mixing ratio increases of CH3CCl3, CFC-11, CFC-113, and CCl4 (each with 0 to 2 statistically significant emission ratios) were smaller than those of CFC-12 and halon-1211 (each with 5). The data filter applied to reduce the influences of localized ODS sources on emission ratios (Figure 3) may not have been completely effective. Emissions of ODS and 222Rn from noncollocated sources in the upwind source regions may not have been well mixed in air masses before they reached the train. The influences of these potential contributors to correlation scatter are not readily assessable, but it is likely they all affected ODS emission ratio determinations for TROICA-7.

2.3. Estimates of Collective Source Region Areas and ODS Emission Rates

[19] The average emission ratios and fluxes determined here are representative of the collective upwind sources that influenced measurements during the 6 diurnal periods of the WT. Consequently, the area of the collective source region for each diurnal period is needed to spatially scale average flux values to average rates of emission. In this sense, the collective source region is the composite of source regions for all measurements made during that diurnal period. The collective source regions are inevitably large because the train traveled an average of 1475 km during each full 24-hour diurnal period (Table 1).

[20] Boundaries of the 6 collective source regions were defined by the position histories of near-surface air masses advected to the train, as portrayed by isentropic back trajectories calculated for the location of the train every 3 hours of the WT. Back trajectories were initialized at times <6 h different from the actual train presence at 49 different locations and run at 3-hour resolution using ECMWF wind fields. Given that the mean duration of low-height temperature inversions during the WT was 10.1 hours, it follows that 9- or 12-hour back trajectory positions would most accurately define source region boundaries. However, this assumption is valid only if the lengths of these back trajectories conform to actual nighttime wind speeds during the WT because TROICA-7 emission ratios were most profoundly influenced by the mixing ratio increases at night. Sixteen different 3-hour back trajectories that reached train locations between midnight and 06:00 local time depict an average nighttime wind speed of 3.3 ± 0.8 m s−1, a typical value for drainage (katabatic) winds in a stable nighttime boundary layer [Stull, 1988; Horst and Doran, 1986]. At this wind speed, air masses would be advected 121 km in 10.1 hours, a distance best represented by the mean 130 km length of the 9-hour back trajectories for all 49 train locations.

[21] The areas of collective source regions were determined graphically from the 49 train locations, their associated 9-hour back trajectory positions, and the 48 midpoints between train locations (Figure 4). The linear method of calculating areas from nonlinear back trajectories propagates only small errors because the average straight-line distance between a train location and its 9-hour back trajectory position was on average only 3% less than the summed linear distances of the three 3-hour segments of 9-hour back trajectories (0–3 h, 3–6 h, 6–9 h). When adjoined, the 6 collective source regions, with areas ranging from 45,000 to 166,000 km2 (Table 1), delineate a 8070 km-long strip that averages 70 km in width and alternates sides of the rail line every 600 to 1400 km (Figure 1). In 1990 nearly 10 million people, about 7% of the Russian population, lived within this 611,000 km2 strip which is hereinafter referred to as the rail corridor.

Figure 4.

The periphery of combined source regions for the 6.5-day WT was established by linearly connecting (thick solid lines) the 9-hour back trajectory positions for sequential train locations. Inner boundaries are shown as thin solid lines between contiguous train locations and their midpoints. Source regions were divided temporally by connecting every train location to its 9-hour back trajectory position (thick dashed lines). The result is forty-eight 5-sided sections (of which 4 are shown), each representing the source region for the 3 hours of data acquired between sequential train locations. Each section was divided into 3 triangles (thin dashed lines) to facilitate area calculations. Areas of the collective source regions (Table 1) were estimated by summing the triangle areas for each diurnal period and subtracting any redundant areas.

[22] Emission rates in the rail corridor were calculated as the sums of emission rates in each of the 6 collective source regions for which a statistically significant emission ratio was determined (Table 1). Emission rates were set to zero for source regions with emission ratios insignificantly different from zero.

[23] Given the reported closure of Russian ODS production facilities in December 2000, the strengths of ODS emissions in mid-2001 were dictated by the sizes and leakage rates of ODS banks in Russia. It is assumed that the bank sizes and leakage rates of CFC-11, CFC-12, and halon-1211 did not decline rapidly in the first post-production year (2001) because the average delay between manufacture and emission of these ODSs is >4 yr [McCulloch et al., 2001, 2003; UNEP, 1999]. Thus the emission rates measured during mid-2001 were presumed representative of emissions in the rail corridor during the entire year (Table 2). In contrast, CFC-113, CCl4, and CH3CCl3 are typically released within a year of their production, which we portray with a linear, 100% yr−1 decline in their emissions beginning January 2001. Our mid-2001 measurements of CFC-113 and CCl4 emission rates therefore depict the midpoint in emission rates for the entire year. Rail corridor emission estimates for all six ODSs were scaled to the whole of Russia (Table 2) based on the 15:1 ratio of Russian to rail corridor population.

Table 2. Rail Corridor, Russian, and Global ODS Emission Estimates for 2001
ChemicalRussian Rail CorridorRussianGlobal Emissionsd
  • a

    All estimates are given in Gg. Uncertainties of Russian rail corridor emission estimates, given in parenthesis, are based solely on the 2σ (95% confidence) uncertainties of emission ratios.

  • b

    Ranges of rail corridor and Russian emission estimates incorporate the potential factor of 2 error in using an average terrestrial 222Rn flux value in the absence of regional average 222Rn flux values.

  • c

    Russian emission estimates are based on rail corridor estimates scaled by population (see section 2.3).

  • d

    Global emissions estimates for 2001 are from a halocarbon emission scenario Ab [WMO, 2003] that incorporates current data for production trends, emissions of new production, bank sizes, and bank release rates.

  • e

    Upper limit for rail corridor and Russian emissions of CH3CCl3. See section 3.1 for an explanation of how these were determined

CFC-110.08 (0.04)0.04––2.378.2
CFC-124.4 (0.8)2.2–8.86432–129122
CFC-1130.05 (0.03)0.03––1.58.0
Halon-12110.08 (0.02)0.04––2.38.67
CCl40.04 (0.03)0.02––1.163.6
CH3CCl3<0.03e <0.4e 41.4

3. Results and Discussion

[24] The estimates of 2001 Russian ODS emissions presented here carry sizeable uncertainties, yet they serve several important purposes. Our measurements of CCl4 emissions in the rail corridor can confirm or refute the reported cessation of Russian CFC production before 2001 because the majority of global CCl4 emissions are attributable to the manufacture of CFC-11 and CFC-12 [Simmonds et al., 1998; WMO, 2003]. Our estimates of CFC-113 and CH3CCl3 emissions in 2001 provide a check of 2000 Russian production figures reported to UNEP since these chemicals are released within a year of their manufacture. The strong emissions of CFC-12 observed during TROICA-7 indicate the existence of a large bank in Russia which we compare to a recent estimate of the global CFC-12 bank. Most importantly, the strengths of Russian CFC-11, CFC-113, and halon-1211 emissions in 2001 are compared to the magnitudes of recent shortfalls in their production-based global emission estimates to ascertain if these shortfalls result from the underestimation of recent Russian emissions.

3.1. Emissions of CCl4, CFC-113, and CH3CCl3

[25] Little is known about Russian CCl4 production because UNEP figures exclude the CCl4 manufactured for use as feedstock. Global production figures are similarly incomplete because UNEP is the sole reporting mechanism for CCl4 manufacture, hence the only global emission estimates available are those modeled from atmospheric burden measurements. Our Russian emission estimate of 0.6 Gg CCl4 in 2001 (Table 2) is diminutive compared to the global emission estimate of 63.6 Gg for 2001 [WMO, 2003] and corroborates the reported termination of CFC production in Russia at the end of 2000. Similarly small is the Russian-reported manufacture of 0.35 ODP-Gg (0.32 Gg) nonfeedstock CCl4 in 1999 and none in 2000 relative to the reported 68 Gg produced globally in 2000 [UNEP, 2002a]. Almost 90% of the global CCl4 production in 2000 was reported by China and India, who together accounted for 48% of the reported worldwide manufacture of CFCs in that year [UNEP, 2002a]. However, Russia reported nonfeedstock consumption (production + imports − exports) of 2.0 and 4.1 ODP-Gg CCl4 in 1999 and 2000 that greatly exceed Russian production figures and indicate substantial importation of this ODS. The direct impacts of these imports on Russian emissions in 2001 are not known, but the CCl4 emissions estimated here for Russia in 2001 are only 15–35% of the reported consumption figures for these 2 previous years.

[26] The amount of CFC-113 produced in Russia during 2000 can be estimated only through a hypothetical partitioning of the aggregated, ODP-weighted CFC (Annex A, Group 1) manufacture of 25.5 ODP-Gg reported by Russia to UNEP. This figure is apportioned into an estimate of CFC-113 manufacture using the compound-specific, but globally incomplete production figures reported to AFEAS for 1997–2000. These describe ODP-weighted CFC-113 production as 2.7 ± 1.1% of the ODP-weighted sum of CFC manufacture [AFEAS, 2002]. Employing an ODP of 0.8 for CFC-113, Russian production in 2000 is estimated at 0.9 Gg CFC-113. This value is in good agreement with our measurement-based emission estimate of 0.8 Gg (Table 2) if we assume that all of the CFC-113 Russia produced in 2000 was released in 2001. The consensus between these two approaches adds confidence to our measurement-based emission estimates for the rail corridor and their extrapolation by population to all of Russia.

[27] No significant CH3CCl3 emission ratios were determined for the rail corridor, not surprising since Russia has reported zero manufacture since 1995 [UNEP, 2002a]. An upper limit of Russian CH3CCl3 emissions (Table 2) is estimated by assuming CH3CCl3 emission ratios would have been significantly different from zero if emissions were stronger. Four threshold values for the statistical significance of CH3CCl3 emission ratios were computed independently for each diurnal period from the uncertainties (2σ) of the emission ratios of CFC-11, CFC-12, halon-1211, and CFC-113, and the ratios of their measurement precision to that of CH3CCl3. The four emission ratio threshold values were consistent (mean ± σ = 0.041 ± 0.009 ppt Bq−1 m3) only for 8 July, the exclusive diurnal period when emission ratios for these 4 ODSs were all significantly different from zero and carried relatively low uncertainties (Table 1). The mean threshold value for 8 July describes an emission rate upper limit of 0.03 ± 0.01 Gg CH3CCl3 yr−1 for the rail corridor, which scaled to the Russian population (0.4 Gg yr−1) represents about 1% of global CH3CCl3 emissions in 2001 (Table 2).

3.2. Emissions of CFC-12 and the Global Bank

[28] Production- and burden-based estimates of global CFC-12 emissions are in reasonable agreement in recent years [WMO, 2003]. However, our estimate of Russian emissions in 2001 (64 Gg) is anomalously large, representing just over 50% of the global emissions for that year (Table 2). Russian CFC production did increase by 85% during 1999–2000, and in 2000 accounted for 20% of the global total, surpassed only by Chinese manufacture. If a global emission function for CFC-12, depicting 25% of last year's bank being released the subsequent year, is applied to our Russian emission estimate for 2001 (Table 2), the Russian bank in 2000 would have been ∼250 Gg. A recent reassessment of the global CFC-12 bank [McCulloch et al., 2003], with revised higher emission rates after 1992, estimates the 2000 global bank at only 60 Gg. Even if our Russian emission estimate is a factor of 2 high and the global emission function overestimates the bank by 100%, our recalculated Russian CFC-12 bank (∼60 Gg) would account for the entire global bank estimate. According to our measurement-based emission estimates for Russia the 60 Gg global bank estimate is questionably small. Though it is likely the global CFC-12 reservoir is diminishing with time, as indicated by 2001 global emissions of 122 Gg (Table 1) that exceed the 2000 global production of 81 Gg (calculated as for CFC-113 in section 3.1), our results contradict the idea that it is essentially exhausted.

3.3. Global Shortfalls of CFC-113, CFC-11, and Halon-1211

[29] Shortfalls in production-based global emission estimates of CFC-113 and CFC-11 appeared in 1994, the first year that CFC production in developed countries dropped by > 40% yr−1 [WMO, 2003; UNEP, 2002a]. These shortfalls have increased in recent years and were 25–60% in 2000 [WMO, 2003]. For the purpose of this paper we assume that shortfalls in 2001 were similar in magnitude to those in 2000. This assumption is reasonable because the reported worldwide production of CFCs and halons decreased substantially between 1997 and 2000 while their atmospheric burdens changed relatively little.

[30] Our 2001 Russian emission estimate of 0.8 Gg yr−1 CFC-113 (Table 2) represents about 10% of the global production-based emission estimate for 2001 (Table 2) which falls short of the burden-based estimate for 2000 by 7–15 Gg [WMO, 2003]. Hence, even if Russian CFC-113 emissions in 2001 were grossly underestimated from underreported production, their underestimation would be inadequate to explain the magnitude of a global shortfall similar to that documented for 2000. Given the <1 year delay between the manufacture and release of this ODS [Fraser et al., 1996] and the 85% increase in Russian CFC production during 1999–2000 relative to the near-constant levels reported for 1996–1998 [UNEP, 2002a, 2002b], it is tempting to assert that Russian CFC-113 emissions in each of 1997, 1998, and 1999 were lower than in either 2000 or 2001 and were therefore also too weak to cause annual global emission shortfalls of 7–15 Gg. However, because emissions of CFC-113 and other rapid release chemicals are closely linked to recent manufacture, this assertion relies on an assumed consistent relationship between reported production and actual emissions in Russia during 1996–2001. Though our estimates of Russian CFC-113 emissions in 2001 and production in 2000 are in agreement, this does not constitute evidence of a consistent relationship between reported production and actual emissions in previous years.

[31] Russian CFC-11 emissions of 1.2 Gg in 2001 (Table 2) were similarly inadequate in magnitude to account for a 38 Gg yr−1 emission shortfall such as that documented for 2000 [WMO, 2003]. Emissions of this ODS in Russia comprise only 3% of the global emission estimate for 2001 (Table 1), a surprisingly small fraction compared to Russia's production of 11% of the world's CFCs since 1986 [UNEP, 2002a]. Strong CFC-11 emissions from ongoing production in Article 5 countries could effectively diminish the importance of Russia's contemporary bank releases, but modern global emissions are believed dominated by a very large CFC-11 bank estimated at 656 Gg in 2000 [WMO, 2003]. The Russian CFC-11 bank presumably declined more rapidly than the global bank because Russian CFC production fell 61% between 1990 and 1993, nearly double the 34% drop in global production over that period. Though Russian CFC manufacture grew again in 1999–2000, the amount reported for 2000 was only 25% of the 1990 production figure. Unlike for CFC-113, the multiple-year banking time of CFC-11 precludes rapid variations in its emissions by changes in production. Hence it is likely that Russian emissions of this chemical during several years prior to 2001 were also too weak to play a major role in the global emission shortfalls depicted in WMO [2003].

[32] The occurrence and magnitude of halon-1211 shortfalls are more difficult to ascertain than for the CFCs because of greater uncertainty in its atmospheric lifetime and sizeable disparities between existing calibration scales for halon-1211 measurements, both of which introduce considerable uncertainty in burden-based emission estimates [WMO, 2003]. Production-based emission estimates for halon-1211 and other banked ODSs are generally less accurate than for nonbanked ODSs because their emission functions are more difficult to estimate with certainty. Despite these challenges, inventory and modeling work by Fraser et al. [1999] depicts a shortfall of halon-1211 emissions starting in 1994, when halon production in developed countries fell by 97% [UNEP, 2002a], and a widening of the shortfall to 4 Gg yr−1 (30% yr−1) by 1997. The model also portrays a sizeable reduction in emissions in 1994 that would markedly slow the atmospheric growth of halon-1211, a feature absent from the measurement record [WMO, 2003]. Another model [UNEP, 1999] employs the same production figures to deduce global halon-1211 burdens that are consistently less than atmospheric measurements by 10–25% since they began in 1979 [WMO, 2003], and does not implicate recent shortfalls in production-based emissions.

[33] The 1.1 Gg estimate for Russian halon-1211 emissions in 2001 (Table 2) represents 12% of global emissions in that year, consistent with 9% of global halon manufacture taking place in Russia during the previous decade. If modern shortfalls in global production-based emission estimates of 4 Gg yr−1 do exist, Russian halon-1211 emissions during 2001 were too weak to explain a similar shorfall in 2001. As is the case for CFC-11, the multiple-year banking of halon-1211 suggests that Russian emissions during several previous years were also inadequate to address global emission shortfalls of this magnitude.

4. Summary

[34] Our measurement-based estimates of Russian ODS emissions in 2001 attest to their global significance but cannot support any conjecture that underreporting of Russian production would lead to the recent shortfalls in global production-based estimates of CFC-113, CFC-11, and halon-1211 emissions. Only weak emissions of CCl4 were detected in mid-2001, supporting the report that Russian CFC production had ceased at the end of 2000. Our finding of strong CFC-12 emissions in Russia reveals a large bank of this chemical that may be greater than a recent estimate of the global bank.


[35] We thank all personnel involved with TROICA-7 and gratefully acknowledge Evgeny Kadygrov and Alexei Lykov for temperature profile data, Brad Hall for calibration and maintenance of ODS gas standards, and Joyce Harris for the back trajectory calculations used in this work. Steve Montzka and Jessica Neu provided insightful comments on this paper, and Valentin Koropalov assisted with the expedition. Funding for the expedition was provided by the International Science and Technology Center (EU), the Russian Foundation for Basic Research, the Volkswagen Foundation (Germany), Ruhrgas AG (Germany), the Atmospheric Chemistry and Modeling Analysis and Upper Atmospheric Research Programs (ACMAP and UARP) of the National Aeronautics and Space Administration (USA), the Atmospheric Composition and Climate Program, Office of Oceans and Atmospheres, and Arctic Research Program of NOAA (USA), and CIRES (USA).