Mercury in the atmosphere around Japan, Korea, and China as observed during the 2001 ACE-Asia field campaign: Measurements, distributions, sources, and implications



[1] We report on mercury in the atmosphere of east Asia (Japan, Korea, and China) as measured from sea level to ∼7000 m during 16 research flights in the National Science Foundation (NSF)/National Center for Atmospheric Research (NCAR) C-130 aircraft during the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) campaign ( The air at all altitudes contained concentrations of atmospheric mercury above the global background. The atmosphere was highly stratified with plumes originating from massive dust storms carried out of China, from local industrial pollution, from volcanoes, and, less well defined, from biomass burning. Most often the air masses were mixtures, e.g., dust layers contained anthropogenic emissions or volcanic plumes were embedded in anthropogenic pollution; thus the total data set showed no significant correlations of gaseous mercury with the most common anthropogenic pollutants in the area (CO and SO2), but good correlations were observed for identifiable plumes. Highest mixing ratios for gaseous elemental mercury (GEM) were found in industrial plumes exiting China (∼6.3 ng/m3), Korea (∼3 ng/m3), and Japan (∼3 ng/m3). The core of the plume from Miyake Jima volcano contained ∼3.7 ng/m3 of GEM. Crustal mercury was also present, emitted and subsequently deposited during the outbreak of the spring dust storms. Some of the nondust aerosols contained soluble mercury in highly variable ratios with the gas-phase mercury contained in the same plume. Preliminary estimates for the export from China are 5–15 t of crustal mercury during the dust storms, ∼150 t/yr of gas-phase mercury from biomass/biofuel combustion, and ∼600 t/yr from industrial sources, mostly from coal combustion. Gaseous mercury is a useful tracer for industrial, volcanic, and biomass-burning sources, but in most cases, robust plume identification required one or more cotracers. Because of the inertness of GEM and the ease of its measurement it is well suited for the tracking of long-range transport.

1. Introduction

[2] The mercury concentration in surface air is generally higher in east Asia than in Europe and in the United States [Lin et al., 2002; Nakagawa and Hiromoto, 1997]. This is the result of high primary anthropogenic emissions and evasion from soil, which is enriched by wet and dry deposited mercury from local and global sources [Kim et al., 2002; Kim and Kim, 1999]. China, Korea, and India are thought to release an estimated 1900 t/yr of mercury, which is equal to about half of that globally emitted from anthropogenic sources [Pacyna and Pacyna, 2002]. In 1995, coal-burning power plants in China were reported as emitting 214 and 303 t by Zhang et al. [2002] and Wang et al. [2000], respectively. Mercury emissions for this region have increased over that past several decades as a result of rapid industrialization. However, more recently, this anticipated growth is being moderated in part by improved coal preparation and combustion technologies [Carmichael et al., 2002]. Municipal waste incineration extensively practiced in Japan [Takaoka et al., 2002] is an additional source of anthropogenic emissions. Also contributing to atmospheric mercury at these regional scales are geothermal sources (volcanoes in Japan; Nakagawa [1999]) and advected wildfire plumes known to contain mercury species [Brunke et al., 2001; Friedli et al., 2001a, 2003a, 2003b].

[3] Atmospheric mercury is of significant interest because it is a hazardous pollutant and its potential value as a conserved tracer for the study of long-range transport. Elemental mercury at global background levels (1.2–1.4 ng/m3) has insignificant acute toxicity since the lowest adverse effect observed was at 15–30 μg/m3 [Kazantzis, 2002]. However, after biological conversion in watersheds into methyl mercury [Downs et al., 1998] and bioaccumulation [Rolfhus et al., 1995; Gnamus et al., 2000] into top predators (piscivorous fish or mammals) methyl mercury becomes a hazard to populations consuming large quantities of such species because of its neurotoxicity, particularly in utero toxicity to fetuses [Mahaffey, 1999].

[4] Gaseous elemental mercury is by far the dominant atmospheric mercury species [Lindberg and Stratton, 1998] and can serve as a conserved tracer because of its lifetime estimated at about one year [Schroeder and Munthe, 1998]. The long lifetime is mainly the result of low dry deposition rate, low water solubility, and slow conversion with the major oxidants HO and O3 in the free troposphere and in clouds [Lin and Pehkonen, 1999; Hall, 1995; Sommar et al., 2001]. The mercury cycle and transport is described in regional and global models [Seigneur et al., 2001; Petersen et al., 2001; Bergan et al., 1999; Bergan and Rohde, 2001].

[5] The objective of the Asian Pacific Regional Aerosol Characterization Experiment (ACE-Asia) field experiment ( was the integration of satellite, aircraft, ship, and ground station measurements of chemical, physical, and radiative properties of particulates emanating from the Asian continent. Flight planning for the National Science Foundation (NSF)/National Center for Atmospheric Research (NCAR) C-130 aircraft flights was in support of the project objectives and gas-phase mercury was measured as a piggyback add-on. Throughout the paper, flights are labeled as research flight (RF) with digits 01–16 designating individual flights. The objective of this paper is to describe the measurement of gaseous mercury and its distribution between sea level and about 6.6 km in the atmosphere around Japan, Korea, and China, and to assign sources and their contribution to mercury budgets.

2. Methods

2.1. Measurement of Total Gaseous Mercury (TGM)

2.1.1. Tekran Operation

[6] Total gaseous mercury (TGM) was measured using a Tekran 2537A Mercury Vapor Analyzer (Tekran Inc., Toronto, Ontario). The operation of the Tekran instrument is based on collecting mercury from air by amalgamation on one of two gold cartridges while thermally desorbing elemental mercury from the second and assaying the elemental mercury by cold vapor atomic fluorescence spectroscopy (CVAFS). Minimum warm-up time for the instrument before the flights was 15 min, but generally >1 hour was allowed. The output of this instrument is commonly designated as “total gaseous mercury or TGM,” which includes gaseous elemental mercury (GEM) and the fraction of reactive gaseous mercury (RGM), if present and not scavenged on inlet and filter surfaces. RGM (Hg + 2) is small for most ambient backgrounds [Lindberg and Stratton, 1998] but can be large close to point sources or during Arctic or Antarctic depletion events [Lindberg et al., 2002]. Since we did not sample near known RGM emission sources and filters were in place to remove particulate mercury (pHg), the measured TGM in our case is essentially all GEM and is considered as such.

2.1.2. Sample Collection

[7] Air was aspirated continuously by the Tekran analyzer via a backward facing inlet and discharged through a dedicated backward facing outlet by the pump, which is integral to the analyzer. Ebinghaus and Slemr [2000] have demonstrated that at sample pressures above 500 mbar the Tekran instrument is collecting TGM quantitatively but that the response of the CVAFS detector must be corrected for the pressure, in our case, cabin pressure, to which the detector cell effluent is discharged.

[8] The inlets were located in the free airstream 25 cm outside the skin at the bottom center of the C-130 aircraft aft of the wheel wells. Particulates in the sample air were removed by two filters: a 50 mm diameter 5 μm Teflon filter in a Teflon housing located about 60 cm from the inlet tip and a 47 mm diameter 0.2 μm Teflon filter installed at the Tekran analyzer. The distance between inlet and Tekran was 2.6 m, and all tubing was made of 1/4 inch OD Teflon. The Tekran was operated with ultrahigh-purity Argon (Scott Specialty Gases) and zero air (Scott Specialty Gases) from gas cylinders. Zero air, which is required for instrument start-up and protection of the cartridges during take-off and landings, was purified by a Resisorb canister (Tekran Inc.) containing activated carbon. The Tekran instrument was operated on 300 s adsorption/desorption cycles at 1.5 L/min (STP) or 7.5 L (STP) total sample up to about 4.5 km MSL. At higher altitude the Tekran pump could not deliver the full 7.5 L and the flow dropped to about 3.9 L (STP) at the highest altitude during transit flights. The integrity of the sampling system was tested by manual injections of GEM at the inlet and at the Tekran injection port: measurements ranged from 95 to 105% of the values injected at the Tekran indicating that mercury was not lost in the tubing or on the two filters. Since the sample pressure was >500 mbar for all research flights (but could drop to ∼400 mbar during transit flights) the assumption was made that all GEM was captured by the cartridges. The 300 s time resolution combined with the limited time spent in a distinct air mass, particularly during rapid profile flights, limited the spatial resolution of the mercury measurements.

2.1.3. Calibration

[9] Mercury calibrations were performed using an internal permeation cell at the beginning and end of the field deployment when at least 24 hours of conditioning time for the permeation cell were available. Comparisons between the permeation cell and manual injections using a 2505 Mercury Vapor Calibration Unit (Tekran Inc., Toronto, Ontario) were conducted before and after deployment and agreed within 8%. In the field, where permeation cell calibrations were not possible, all calibration was performed by manual injections.

2.1.4. Poisoning and Regeneration

[10] We experienced unexpected severe loss of detector sensitivity due to poisoning of the gold cartridges during this field experiment. Poisons block mercury adsorption sites such that mercury is only partially collected. Suspected poisons are contaminants associated with the airbase operation (Iwakuni Marine Corps Air Station) or, more likely, sulfur compounds from a close-by wood-pulping plant. After deactivation was first noted, exposure of the cartridges to ambient air near the airbase was avoided by operating the Tekran on zero air during instrument warm-up and take-off and landing. Poisoning occurred early in the experiment, but continued once it had started. Sensitivity was partially maintained by high-temperature treatment of the cartridges with zero air according to the provisions of the Tekran manual. The instrument was manually calibrated between some of the flights and always before and after regenerations. It is noteworthy that the original sensitivity was fully restored after several days of operation on zero air as the sample after the campaign; that is, the poisoning was fully reversible. Figure 1 documents the massive change of sensitivity and the effect of regeneration as observed during the course of the field experiment. The decay rate of B cartridge remained higher during the experiment. The poisoning was ongoing once the cartridges had been damaged. Regeneration effectiveness was about the same for the A and B cartridges. At the end of the campaign upon return to Boulder the response factors had declined by 36.0 and 32.8% for the A and B cartridges, but during RF 15 (the worst case) the response factors had dropped ∼50 and ∼80% for the A and B cartridges.

Figure 1.

Tekran operation and calibration: the effect of poisoning and regeneration of the cartridges experienced during the field campaign and associated in-field calibrations.

2.1.5. Quantitation

[11] The area integral (counts) for each cycle was adjusted for the instrument background counts and corrected for the pressure dependence of the detector based on the recorded cabin pressure at each data point. The pressure dependence was calculated from calibrations at the home base at 823 mbar before the ferry flights and at the deployment location at Iwakuni, Japan, at 1020 mbar. The pressure correction for both cartridges was 5.162 cts/pg/mbar or 0.106%/mbar. This compares with the values of Ebinghaus and Slemr [2000] of 0.122 and 0.120%/mbar for their A and B cartridges and the same model Tekran instrument. The pressure-corrected area values were then converted to concentrations by applying the calibrations in counts/pg and the corresponding sample volumes. Response factors for individual flights are linear interpolations between the calibration responses of the six on-site calibrations (see Figure 1). When the individual response factors were applied to A and B, adjacent measurements agreed within about 10% for all flights except for flights RF 8–11 where there was an unexplained larger discrepancy. We elected to arbitrarily adjust the response factors to bring the A and B relationship into the same range as all other measurements, assuming A was correct. We estimate the accuracy of the measurements to be ±10%.

2.2. Measurement of Particulate Mercury (pHg)

[12] Particles were sampled isokinetically and collected in the total aerosol sampler (TAS) on a Teflon cone followed by a Teflon filter. The collected material was washed from the cone with 30 mL 10−5 m triflouroacetic acid (TFA), and the filter was washed with 10 mL of a mixture of a 9 mL 10−5 m TFA and 1 mL ethanol. The combined extracts (less the small amount used for liquid chromatography analysis) and the corresponding blanks were analyzed for total mercury by Frontier Geoscience (E. Prestbo) using clean techniques in a class 100 clean bench. The samples and blanks were transferred into trace-clean vials and added up to 40.0 mL with 0.06 N solution of BrCl in double deionized water. The samples were heated for two hours with lid tightly closed at 70oC and analyzed using SnCl2 reduction, dual gold amalgamation and cold vapor atomic fluorescence spectroscopy (CVAFS) detection. The detection limit (eMDL) was 0.013 ng/digestion.

3. Results and Discussion

3.1. Distribution of TGM

[13] The ACE-Asia flights began 31 March 2001 during a synoptically active period, followed by massive dust storms, then a calm, synoptically inactive period and finally a flight through a weak trough with precipitation (30 April 2001). Note that the mercury sampling was restricted to the flight tracks selected to satisfy the primary mission objectives and may be biased because of limited spatial coverage. Figure 2 illustrates the TGM distribution along 16 flight tracks. The highest concentrations of TGM were observed in the East China Sea (industrial, ∼6.3 ng m−3), the Sea of Japan (industrial, ∼3 ng m−3) and in the North Pacific Ocean (Miyake Jima volcano plume ∼3.7 ng m−3, and industrial, ∼3 ng m−3).

Figure 2.

TGM distribution along the flight tracks for flights RF 01–16. The highest concentrations are in the South China Sea (industrial China), the Sea of Japan (industrial Korea, China), and the North Pacific Ocean (industrial Japan and Miyake Jima volcano). Flight segments discussed in detail are identified.

[14] The geographic distribution of TGM by 2o latitude/longitude and 500 m altitude bins as illustrated in Figure 3 reveals the complex nature of the mercury distribution in this region and consistently enhanced concentrations above the well-mixed background of 1.2–1.4 ng m−3 [Temme et al., 2003] observed for the Northern Hemisphere. Unique situations with especially high TGM levels can be identified; e.g., the volcanic emissions from Miyake Jima (see section 3.2.2) at (140°–142°E, 34°–38°N), highly processed air (see section 3.2.6) at high altitude (129°E, >6 km) and low altitude (25°–26°N, <1 km), and the low-altitude outflow from eastern China (see section 3.2.5) at 124°–130°E and 24°–36°N.

Figure 3.

Geographic distribution of TGM by longitude (between latitudes 24°N and 44°N) and latitude (between 124°E and 146°E) in 500 m altitude bins for all flights.

[15] The mean vertical distribution of TGM obtained by averaging over latitude and longitude is depicted in Figure 4. High TGM is indicated at <1 km and between 6 and 7 km. Back trajectories trace the low-altitude contribution to outflow from regional emissions, and the high-altitude component is associated with lofted pollution with back trajectories extending beyond Mongolia from unknown sources or, we speculate, they may be returning Asian plumes after having circled the globe. Banic et al. [2003] observed no mercury gradients with altitude over continental North America. Their observation is not replicated here, presumably because of strong point sources carried aloft and remaining relatively intact. Considering the full profile, however, there appears to be a decrease in mean mercury with altitude, an observation we have also observed over the eastern North Pacific (L. F. Radke et al., Atmospheric mercury over the northeast Pacific during ITCT 2K2: Gradients, residence time, stratosphere-troposphere exchange, and long-range transport, submitted to Journal of Geophysical Research, 2004).

Figure 4.

Mean TGM concentrations and their standard deviations for all flights binned for eight vertical segments between sea level and 8000 m.

[16] The age of the observed air masses, i.e., the time since emission, was estimated using their SO4/(SO4 + SO2) ratio, where values close to 0 represent air masses in source regions, and values close to 1 indicate well-aged air masses. Ages associated with these ratios can be assigned using a chemical conversion rate, but this is only approximate because the conversion rate depends on the oxidant levels and atmospheric conditions. Assuming a constant 24 hour average conversion rate of SO2 to sulfur of 1%/h, the observed air masses were grouped into fresh, aged and old: fresh <1 days, SO4/total sulfur <0.2, aged 1–4 days, SO4/total sulfate 0.2–0.6, and old air masses >4 days, SO4/total sulfur >0.6. The frequency of occurrence of these three age groupings in four different altitude bins is given in Table 1. The age distribution data indicate that fresh air was only present at low altitudes and that air lofted above 3000 m was generally old.

Table 1. Age Distribution of Air Sampled From Sea Level to >5000 m
Altitude, mFraction of Age Group, %
<2 Days2–4 Days>4 Days

3.2. Sources of Mercury

3.2.1. Overview

[17] We anticipated substantial emissions for TGM in this geographic area from at least four sources: industrial/anthropogenic (mostly coal-burning power plants), volcanic, biomass and biofuel burning, and dust. The existence of multiple sources of TGM is substantiated by the fact that the mercury (TGM) versus CO scatterplot for the ACE-Asia data set as given in Figure 5 shows no clear correlation between the two variables. This is explained by the fact that CO is the tracer for various types of combustion (fossil, biofuel, wildfires, etc.), all of which were active in the project area and have different emission ratios (e.g., power plants versus biomass burning). Amazingly the data cover the whole range from very polluted high-CO/background TGM to very clean low-CO/high-TGM air masses. The presence of sources with different emission ratios is illustrated by coplotting biomass-burning data, where TGM is well correlated with CO [Friedli et al., 2003b] and identifiable anthropogenic plume data from the ACE-Asia experiment (“Shanghai plume” in RF 16). The data indicate that the atmosphere in this geographic area at this time of the year is defined by poorly mixed emissions with different emission ratios. Frequently flown gradient profiles substantiated the presence of multiple layers representing the different sources for mercury.

Figure 5.

Scatterplot for TGM/CO for RF 02 to RF 16 sampled during the ACE-Asia flights. Data points below 3 km and above 3 km are identified. The lines represent known wildfire [Friedli et al., 2003b] and RF 16 “Shanghai plume” emission. Flight segments for RF 8 and RF 11 with biomass/biofuel-burning signals are also highlighted.

[18] To more clearly identify sources of mercury in east Asia, we used additional tracers from the ACE-Asia data sets ( measured during the C-130 flights. Expected cotracers with TGM for emissions from power plants burning coal are SO2 and SO4 representing fresh and aged emissions. A consistent tracer in the proximity of volcanoes is SO2. For biomass-burning plumes, Ma et al. [2003] proposed K+ and CO as useful cotracers. For dust plumes, soluble Ca is an appropriate tracer. Figure 6 is an overview of anthropogenic, volcanic, biomass-burning, and dust tracers for 16 flights. This graph clearly shows the major dust storms (Ca in RF 05, 06, 10, and 13), many interceptions of distinct anthropogenic or volcanic plumes (SO2 and SO4) and isolated biomass signals (K+). The graph illustrates the variability and extraordinary complexity encountered during the ACE-Asia experiment. While the total data set showed no significant correlations of gaseous mercury with the most common anthropogenic pollutants in the area, good correlations were observed for identifiable plumes. The analysis of the flight segments with identifiable correlations is discussed below.

Figure 6.

Overview of all flights: concentration of TGM and tracers associated with dust storms (Ca), anthropogenic and volcanic emissions (SO2 and SO4), and biomass burning (K). The top two panels are for flights RF 01–08. The bottom two panels are for RF 09–16.

3.2.2. Mercury Emissions From Volcanoes

[19] The Miyake Jima volcano was in a posteruptive phase during the ACE-Asia field experiment and air masses influenced by its emissions were observed on several occasions. One example is RF 11 as shown in Figure 2. The passage across the plume of Miyake Jima volcano (139.5°E, 34°N) lasted 90 s as indicated by 1 Hz SO2 measurements. The 40-min-old plume was intercepted over the open sea at 450 m and consisted of a mixture of polluted air (∼290 ppbv CO) with a back trajectory indicating that it had crossed the Sea of Japan and mainland Japan, and the volcano effluent. The relationship between SO2 and TGM is given in Figure 7. The 300 s TGM instrument cycle overlapped the plume symmetrically; that is, 105 s each were sampled at the background before and after the peak. The SO2 plume was very heterogeneous with an average over the 90 s passage of 222 ppbv. The TGM concentration in the plume was calculated from the TGM concentrations measured in the cycles before, during and after passage of the plume and was 3.7 ng m−3. The pHg collected during an overlapping aerosol collection cycle was below detection limit, but since the plume passage was short (6% of the 25 min aerosol collection cycle), pHg emitted from the volcano, unless very large, would remain undetected. RGM, not reported in the literature to be present in volcanic plumes, was not measured. The emission factor measured for the Miyake Jima plume, ΔTGM/ΔSO2 (weight/weight (w/w)), is 2.4 × 10−6 as calculated from the average SO2 concentration in the plume and the difference between the TGM concentrations in the plume (the calculated 90 s value) and the average of the TGM concentrations in the cycles before and after plume passage.

Figure 7.

TGM and SO2 measured during the 90 s passage through the Miyake Jima plume at 450 m during RF 11.

[20] Identification of air masses containing volcanic emissions is difficult because of the fact that there are other large anthropogenic emissions of SO2 in east Asia. We employed a modeling approach to help identify mercury from volcanic sources by a model simulation of SO2 using only emissions from volcanoes, i.e., the difference between measured total sulfate and sulfate associated with known anthropogenic sources. The model used is described by Uno et al. [2003] and Carmichael et al. [2003]. The model results identified individual flight segments impacted by volcanic emissions. Figure 8 shows the sections of the flight tracks with TGM attributed by the model to volcanic emissions, most clearly seen for Miyake Jima during RF 11 and RF 12. The observed values of TGM and SO2 for these segments are plotted in Figure 9. This approach yields an emission factor of 4.2 × 10−6 (w/w), a value about 40% larger than the emission factor (2.4 × 10−6) derived from the direct measurement of the Miyake Jima plume.

Figure 8.

Flight tracts for all flights with locations of segments impacted by volcanic emissions as identified by using the three-dimensional model with volcano emission tracer. Most pronounced is the emission from Miyake Jima during RF 11 and RF 12. The ratio of TGM/SO2 (w/w) is 4.2 × 10−6.

Figure 9.

Emission factor for TGM relative to SO2 for the Miyake Jima using the flight segments identified in Figure 8. Points associated with segments of RF 11 and 12 are highlighted.

[21] The mercury emission from Miyake Jima was estimated from the experimental TGM/SO2 ratio of 2.4 × 10−6 (w/w) and the published average SO2 emission flux (http://www:// For 2000, 2001, and 2002 the fluxes were 42,000, 21,000, and 10,000 t/d, respectively, yielding estimated annual mercury fluxes of 36.3, 18.1, and 8.7 t/yr for 2000, 2001, and 2002, respectively. The flux reported for geothermal activity in Japan was previously reported to be 1.4 t/yr [Nakagawa, 1999], less than 20% of our estimate for Miyake Jima alone, but note that the Nakagawa estimate did not include Miyake Jima. Varekamp and Buseck [1986] concluded from their investigations that young and active volcanoes have higher emission ratios than inter eruptive volcanoes, which might be one of the reasons why the current Miyake Jima mercury flux estimate is so high. The Miyake Jima mercury emissions are expected to decline as the rapid decline of SO2 flux continues. It is not known, however, if the Hg/SO2 ratio is also a function of SO2 flux, which could further modify the mercury flux. Reported volcanic emission ratios relative to SO2 (Hg/SO2, w/w) vary by 5 orders of magnitude depending on geographic location, observational techniques, time since last eruption and degree of activity of the volcano: North American volcanoes: 3.9 × 10−5 to 3.0 × 10−3, European volcanoes: 6.0 × 10−8 to 2.5 × 10−5 [Richardson, 2001]. Miyake Jima falls within the range for European volcanoes. Nriagu and Becker [2003] have recently summarized regional and global volcanoic emissions and reported smaller values than we measured in these experiments. There is a need for additional measurements.

3.2.3. Mercury in Dust Plumes (Gaseous and Crustal)

[22] Several large dust storms occurred during the ACE-Asia experiment. The dust plumes encountered in RF 05, 06, 10, and 13 (see Ca in Figure 6) contained TGM above background. With the exception for RF 10 there is no significant correlation (R2 > 0.5) between TGM and Ca ++ and we attributed the TGM present to anthropogenic emissions mixed in with the dust plume along the path of the storm. For example, on the basis of back trajectories, the dust plume observed in RF 06 also contained emissions from the Quingdao industrial area. Dust from Chinese deserts contains crustal mercury of 89 ng/g (50 ng/g from Nriagu [1989]) as measured in a standard sample (Reference CJ-1/6, M. Uematsu, Ocean Research Institute, University of Tokyo). If all of the crustal mercury in dust were converted to TGM by some photochemical or chemical processes and using an average value of ambient dust concentration during these events of 260 μg/m3, this still could not account for the observed TGM levels; that is, there is more gaseous mercury in the plume than all mercury contained in the dust mass.

[23] Crustal mercury in dust may, however, be important for regional and global deposition and transport. Using measurements at Zhenbeitai, located in the northern China desert, Zhang et al. [2003] estimated a dust deposition of 189 g/m2 for the 92 days of the spring 2001 dust storm season. Using the 89 ng/g mercury content as representative for Chinese dust, 16.4 μg/m2 (in 92 days) mercury would have been deposited during the dust events at this site. This compares to an annual wet and dry deposition rate of 2.0–4.7 μg/m2/yr modeled by Seigneur et al. [2001], implying that deposition of crustal mercury in this region is much more important than the atmospheric deposition from global sources.

[24] Model estimates summarized by Huebert et al. [2003] range from 250 to 650 t (±500% uncertainty) for the dust mass mobilized during the ACE-Asia storms. Using the 89 ng/g crustal mercury content yields a crustal mercury mobilization estimate of 21–60 t. Satake et al. [2004] estimated that 25% of the total dust emissions were exported out of China. This yields an estimate for the crustal mercury flux out of China of about 5–15 t during ACE-Asia. This indicates that the crustal mercury flux out of China during ACE-Asia of about 5–15 t is relatively minor when compared to published Chinese power plant emissions of 200 to 300 t/yr [Zhang et al., 2002; Wang et al., 2000].

3.2.4. TGM From Biomass Burning

[25] CO is a tracer for mercury for the combustion of fossil, biofuel and biomass fuels, which are the three most important combustion sources in east Asia. CO contribution from biomass burning in Asia accounts for ∼25% of combustion CO in Asia on an annual basis, and ∼40% in the spring [Kasibhatla et al., 2002; Streets et al., 2003b; J.-H. Woo et al., Asian mercury emissions from man-made and natural sources, manuscript in preparation, 2004]. The emission ratio of TGM/CO in biomass-burning plumes measured for fires in temperate forests [Friedli et al., 2003b] is included in Figure 5. The emission ratio is much smaller than the corresponding emission ratio for the industrial “Shanghai” plume. This suggests that the TGM/CO ratio can be used to distinguish biomass (or biofuel) from fossil fuel combustion, provided that there is not much variability among biomass/biofuel combustion. However, in the period during which ACE-Asia took place, the biomass plumes were usually very dilute by the time they passed over the study region, leading to low concentration of TGM that made it difficult to capture biomass-burning signals. On the basis of these facts it is necessary to include additional “bioemitted” tracers other than TGM and CO to identify biomass-burning plumes. The data set was searched for simultaneously high correlation of TGM with a second tracer, K+. Flight segments were identified which likely contained biomass-burning signatures, i.e., RF 8 (TGM/CO, R2 = 0.77−0.81; TGM/K+, R2 = 0.59−0.66) and RF 11 (TGM/CO, R2 = 0.65; TGM/K+, R2 = 0.49). Observations for these flight segments are identified in Figure 5. The TGM/CO ratios associated with these points fall between the slope for biocombustion and the industrial “Shanghai” plume values, reflecting the mixing of air from different sources as pollution is transported off of east Asia. Additional tracers for wildfire emissions, such as methyl halides [Friedli et al., 2001b; Ma et al., 2003] or acetonitrile [Holzinger et al., 1999] would be helpful for more definite assignments of biomass-burning plumes.

[26] Using the observed TGM/CO ratio shown in Figure 5, and estimates of the total CO from biofuel burning and biomass burning (assumed to have the same TGM/CO ratio), yields estimates of mercury emissions from these two sources of ∼50–100 t/yr and 35–100 t/yr, respectively (based on CO emissions estimates from biofuel combustion of 95.7 × 103 t/yr from Streets et al. [2003b] and 196 × 103 t/yr from Ludwig et al. [2003], and from biomass burning of 67.1 × 103 t/yr from Streets et al. [2003b] and 211.4 × 103 t/yr from Ludwig et al. [2003]). These estimates are consistent with values derived from “bottom-up” mercury emissions inventory for Asia; 60.8 t/yr for biofuel and 71.t/yr for biomass burning, respectively (J.-H. Woo et al., Asian mercury emissions from man-made and natural sources, manuscript in preparation, 2004).

[27] Collectively biomass and biofuel emissions (80–200 t/yr) represent a significant source of TGM from Asia. Further studies focused on mercury emissions from biofuel and biomasses burning are needed to better characterize emission factors under Asian combustion conditions.

3.2.5. Mercury in Industrial Plumes

[28] Industrial plumes originating in China, mostly from the industrial areas around Shanghai, Qingdao, and Beijing, were sampled in several flights. Qingdao emissions were contained in the dust storms described in section 3.2.3. On several occasions distinct plumes exiting the east China coast were sampled (flights RF 01, RF 02, RF 08, and RF 16). These measurements provide the best documented emissions for plumes exiting the coastal industrial areas (“Shanghai plume”). They are multisource emitters for mercury as indicated by the good correlations with typical industrial sources: CO (R2 = 0.66–0.81), SO2 (R2 = 0.50–0.72), and C2H6 (R2 = 0.76). A distinctive encounter with the “Shanghai plume” occurred during RF 16 (see Figure 2), when it was sampled in the marine boundary layer (MBL) and contained the highest mercury concentrations measured during ACE-Asia: 6.3 ng m−3 for TGM and 0.92 ng m−3 for pHg. The measured TGM/CO ratio from RF 16 using the points attributed to Shanghai on the basis of trajectory analysis is shown in Figure 5 and has a TGM/CO ratio of 6.4 × 10−6 w/w. This ratio represents a strong fossil fuel emissions signal. For example, the overall fossil fuel combustion TGM/CO ratio for Asia is 7.8 × 10−6 (calculated as the amount of TGM emitted from industrial and power sectors divided by the amount of CO emitted from these sectors) [Streets et al., 2003a; J.-H. Woo et al., Asian mercury emissions from man-made and natural sources, manuscript in preparation, 2004].

3.2.6. Mercury in Highly Processed Plumes

[29] Two air masses high in TGM but very low in CO, SO2, and particulates were encountered under two totally different regimes. During RF 16 one air mass sampled in the MBL was essentially sulfur-free, with CO ∼75 ppbv, low particulates, but high in TGM. It represents very clean, exhaustively processed air that had, according to its back trajectory, resided for at least nine days in the western Pacific MBL. It likely is the remnant of polluted Chinese air with most pollutants removed, except for TGM, which is more resistant to oxidation and removal by wet precipitation. Figure 10 depicts RF 16 during which both very clean, highly processed air and a highly polluted “Shanghai plume” were encountered in the same flight. The flight path and sample location is shown in Figure 2. The clean aged air was associated with flows from the southeast, while the “Shanghai” plume was associated with continental outflow. This flight illustrates that in the same region on the same day, one can find air masses with dramatically different mercury scenarios.

Figure 10.

Examples of clean processed and polluted air masses observed in the South China Sea during RF 16: One represents MBL pollution emanating from continental outflow from China’s industrial coastal areas, and the other represents a processed air mass, possibly from the same origin, but having been processed in the MBL for many days.

[30] A second very clean air mass (sulfur and particulates removed, low CO) was intercepted twice at 6500 m during the RF 07 transit flights in and out of Iwakuni. The location of these data points is shown in Figure 2. Back trajectories indicated that this fast moving air mass had passed over Mongolia at 8000 m about 3 days earlier. It may be a synoptically lofted plume of mercury that circled the globe episodically at concentrations well above background or be the result of warm conveyor belt processing as described by Cooper at al. [2004]. These observations suggest that TGM is a useful conserved tracer for highly processed plumes and for long-range transport.

3.2.7. An Asian Flux of Mercury

[31] Prior to participating in this experiment we estimated, on the basis of published mercury emission data [Zhang et al., 2002; Wang et al., 2000], that our Tekran instrument would barely be able to distinguish an expected Asian signal from local background. In contrast, we found that the mean concentrations downwind of Asia generally exceeded the expected global background of 1.2–1.4 ng m−3 by at least 0.5 ng m−3 or >30% This prompted a simple calculation of the net eastward flux between 24° and 44°N along 124°E, using the mean mercury distribution shown in Figure 2 and the mean observed westerly component of the horizontal winds. The mercury budget estimates are shown in Table 2. The total net eastward outflow estimated in this way between 0 and 7 km is 410 t/month; the result of both China local emissions and global mercury transfer. The net flux below 3 km is 50 t/month. This outflow is due predominantly to Chinese emissions and amounts to an extrapolated 600 t/yr on an annual basis. These estimates represent an upper limit, as the sampling during ACE-Asia was biased toward strong outflow conditions.

Table 2. Estimated TGM Enhancement Flux Out of Asia Along Longitude 124°E and Between 24°N and 44°N for April 2001
Altitude, mAverage TGM,a ng/m3Average East Wind Speed,b m/sOutflow Flux,c Gg/month
  • a

    The TGM values in this column were calculated as the mean of the observed values along this longitude/latitude slice (shown in Figure 4).

  • b

    The mean easterly wind speed was calculated from the observed wind speeds along this longitude/latitude slice.

  • c

    The enhancement flux is defined as the flux of TGM above background, calculated using the concentrations above background obtained by subtracting a background of 1.2 ng/std m3 from the mean observed values shown above. These values were multiplied by the mean easterly observed wind component, the cross-sectional area of each layer.

Sum (0–7 km)  0.41
Sum (0–3 km)  0.05

3.3. Mercury Speciation

[32] Mercury in the troposphere is mostly GEM, but point and processed sources can contain high fractions of RGM or pHg. In their model calculations, Seigneur et al. [2001] used 50, 30, and 20% for GEM, RGM, and pHg, respectively, as representative emission speciation for Asian sources. Pacyna and Pacyna [2002] estimated the Asia emissions distribution to be 54, 36, and 10% GEM, RGM, and pHg, respectively. Fang et al. [2001a, 2001b] measured up to 1.984 ng m−3 pHg for different locations in the city of Changchun, China, compared to only 1–86 pg m−3 background values. For this city the high pHg concentration was attributed to coal combustion and soil dust. From a Japanese power plant burning precleaned coal, only 0.5% of the mercury contained in the coal was released as pHg [Yokoyama et al., 2000], and Sakata and Marumoto [2002] calculated that from a Tokyo municipal solid waste incinerator, 2.9% of the mercury contained in the waste material was released in the form of particulates. They concluded that the mercury in the particulates was present in the form of elemental mercury, presumably GEM adsorbed onto particulate matter upon cooling of the effluent containing ∼9 ng/m3 GEM.

[33] Few data have been published about the emission of RGM in this geographic area. In a case where high-mercury-content coal (0.53 ng Hg g−1) was burned, Feng et al. [2002] measured 450 pg m−3 of RGM, compared a background of <20 pg m−3 in pristine areas. Our pHg measurements are limited to samples from flights RF 02 (Sea of Japan), RF 10 (Gosan, Yellow Sea), RF 11 (circuit around Japan), and RF 16 (off South China coast). They were selected because of their proximity to known anthropogenic or volcanic sources. The samples were collected on the total aerosol sampler (TAS) cone/filter combination assembly operated by the B. J. Huebert group on the C-130 to measure the soluble anionic and cationic composition of particulates. The dissolved particulates and the sample air volumes and sampling location (25–45 min sampling time) were made available to us for the determination of total mercury concentrations. The results are presented in Table 3. The data attest to the great variability pHg concentration and the fraction of pHg of the total mercury during this experiment. The pHg was below detection limit (eMDL = 0.013 ng/digest) in 6 out of 15 samples but reached 0.917 ng m−3 in the anthropogenic plume out of south China (“Shanghai plume”), the plume that also contained the highest TGM values (∼6.3 ng m−3) measured during the campaign. The fraction of pHg of the total mercury ranged from 0.4 to 31.6%. We propose that the variability reflects different sources and different degrees of processing. The particulate data reported here are preliminary and incomplete and may be subject to pitfalls associated with sample preparation, sample storage and limited spatial coverage. However, J. Schauer (personal communication, 2002) also made speciated mercury measurements at the Gosan ground site on Jeju Island, Korea, during the ACE-Asia experiment with results equally surprising and supporting our airborne observations of variability of pHg concentrations and of pHg fraction of total gaseous mercury in the gas phase. A summary of their results is included in Table 3. In view of the occasionally large fractions of pHg observed during ACE-Asia and in the absence of RGM measurements, it is evident that robust budgets must include GEM, RGM, and pHg, for instance measured by the techniques recently described by Landis et al. [2002]. Clearly, an understanding of speciation at the source and its change during plume aging will be required to better define the mercury cycle.

Table 3. TGM and pHg Measurements for Flights RF 02, 10, 11, and 16 and for Surface Measurements at Gosan, Korea
FlightpHg SampleTGM,a ng/m3pHg, ng/m3Percent pHg
  • a

    N.A., not applicable.

RF 02 (Sea of Japan)allN.A.below detection limit
RF 10 (Sea of Japan)1N.A.below detection limit
RF 10 (Sea of Japan)21.550.71531.6
RF 10 (Sea of Japan)31.920.31714.2
RF 11 (around Japan)1, 2, 3N.A.below detection limit
RF 11 (around Japan)41.60.1287.41
RF 16 (Shanghai plume)1, 2N.A.below detection limit
RF 16 (Shanghai plume)33.50.0140.4
RF 16 (Shanghai plume)46.40.91712.53
RF 16 (Shanghai plume)52.50.2118.4
Gosan (surface) 2.15–6.340.036–1.5580.52–26.55

4. Conclusions and Summary

[34] East Asian air at all altitudes contains mercury concentrations above the global well-mixed background of 1.2–1.4 ng m−3 expected for the Northern Hemisphere. Mercury is emitted from large local sources and from soil evasion resulting from mercury previously deposited from local sources and from the global pool. We measured TGM (essentially gaseous elemental mercury, GEM) during 16 flights and particulate mercury (pHg) for a few flight segments close to known sources; no measurements of gaseous ionic mercury (RGM) were made. From the limited data for particulate mercury (pHg) we found that its concentration and the ratio of pHg to TGM varied over a broad range, reflecting speciated sources and/or differences in processing. Most air masses contained emissions from different sources, often found in distinct altitude bands. In two rare cases at ∼6500 m and in the MBL, very clean air with most pollutants reacted or scrubbed out was found to still contain high concentrations of TGM. The main source of TGM is industrial, mostly coal burning. The age of (most) air masses sampled was >1 day as estimated from their SO4/(SO2 + SO4) ratios, and well supported by back trajectories. Volcanoes were identified as mercury sources, measured by sampling a concentrated plume directly (Miyake Jima) and by modeling total sulfate along the flight track. Our estimate for the 2001 TGM flux from Miyake Jima is 8.6 t/yr. Biomass-burning signals identified by TGM and K+ measurements were less definitive, understandably so because the fires burnt far away and because K+ is not a unique tracer for wildfires. On the basis of published CO values for biomass and biofuel burning in Asia and the emission factors estimates derived from our measurements, as much as 150 t/yr of mercury can be attributed to this source. Episodic dust storms transported and deposited more crustal mercury than the modeled dry and wet deposition expected for the same area. The outflow of crustal mercury during the 92 days of the ACE-Asia campaign was 5–15 t while the annual outflow of TGM from China is in the range of 600 t/yr. Uncertainties remain very large, but east Asia currently is the major contributor to global mercury pollution, i.e., ∼50% of the global anthropogenic emissions [Pacyna and Pacyna, 2002].

[35] TGM was found to be a useful tracer for anthropogenic, biomass-burning, and geothermal sources when considered in combination with appropriate cotracers. The 1-year estimated lifetime of TGM and the ease and sensitivity of its measurement make TGM a useful conserved tracer for long-range transport. Because mercury is transported globally, is a health concern in the form of methyl mercury, and east Asia is the major source of gaseous as well as crustal mercury, a quantification of these sources and mercury cycling is essential.

[36] This investigation must be considered as an initial look at the very complex sources, distribution, and transport of mercury in east Asia. Targeted additional research should address the following areas: speciated mercury measurements replacing TGM-only measurements; Lagrangian experiments to investigate the interconversion and deposition of mercury species; and long-range transport studies of mercury in gaseous and crustal form. Finally, modeling for east Asia mercury emissions based on robust source inventories and emission factors should be done.


[37] This research is a contribution to the International Global Atmospheric Chemistry (IGAC) Core Project of the International Geosphere Biosphere Program (IGBP) and is part of the IGAC Aerosol Characterization Experiments (ACE). Our appreciation goes to Barry Huebert for accepting mercury measurements as a piggyback experiment and for providing particulate samples. Many thanks go to Teresa Campos, Stan Hall, and Ian Faloona, who operated the Tekran instrument when no seat was available for the PI. The use of the measurements by Rodney Weber (SO4, K, and Ca), Byron Blomquist (SO2), and Teresa Campos (CO) is gratefully acknowledged. Eric Prestbo (Frontier Geoscience, Inc.) provided valuable analytical guidance and a gratis mercury analysis of the Chinese dust standard. Jamie Schauer provided unpublished speciated mercury measurements taken at Gosan, Korea, during the ACE-Asia mission. Mike Coffey and Andrew Weinheimer were most helpful in reviewing the manuscript. The C-130 flight crew, technicians, and project and data managers provided much appreciated professional services. The National Center for Atmospheric Research, sponsored by the National Science Foundation, and Electric Power Research Institute (EPRI) under contract P 2044, funded the mercury measurements and data analysis. The work at the University of Iowa was supported in part by grants from the NSF Atmospheric Chemistry Program, NSF grant Atm-0002698, NASA GTE, and ACMAP programs.