Projections of Global MercuryEmissions in 2050D A V I D G . S T R E E T S , * , † Q I A N G Z H A N G , †

A N D Y E W U ‡

Decision and Information Sciences Division, ArgonneNational Laboratory, Argonne, Illinois 60439, and Departmentof Environmental Science and Engineering, TsinghuaUniversity, Beijing 100084, China

Received September 2, 2008. Revised manuscript receivedDecember 11, 2008. Accepted January 5, 2009.

Global Hg emissions are presented for the year 2050 under avariety of assumptions about socioeconomic and technologydevelopment. We find it likely that Hg emissions will increasein the future. The range of 2050 global Hg emissions isprojected to be 2390-4860 Mg, compared to 2006 levels of2480 Mg, reflecting a change of -4% to +96%. The main drivingforce for increased emissions is the expansion of coal-firedelectricity generation in the developing world, particularly Asia.Our ability to arrest the growth in Hg emissions is limited bythe relatively low Hg removal efficiency of the current generationofemissioncontrol technologies forcoal-firedpowerplants (flue-gas desulfurization). Large-scale deployment of advancedHg sorbent technologies, such as Activated Carbon Injection,offers the promise of lowering the 2050 emissions range to1670-3480 Mg, but these technologies are not yet in commercialuse. The share of elemental Hg in total emissions willdecline from today’s levels of ∼65% to ∼50-55% by 2050,while the share of divalent Hg will increase. This signals a shiftfrom long-range transport of elemental Hg to local depositionof Hg compoundssthough emissions of both species couldincrease under the worst case.

1. Introduction

The dangerous effects of mercury (Hg) in the environmentare well established, and efforts to curtail atmosphericreleases of Hg have begun in many countries of the developedworld. Knowing how much to reduce emissions to achievea desired level of Hg in the environment is complicated,however, by the potential of elemental mercury (Hg0) toundergo long-range transport at hemispheric scale. Therelative contributions of local sources and distant sources toHg concentrations and deposition at particular locations are,therefore, not well-known. Further complexity is added bydoubt about the relative contributions of man-made andnatural sources of Hg, as well as the magnitudes of the varioussinks of Hg. Indeed, the whole global Hg cycle is still underinvestigation.

Present-day anthropogenic sources of global Hg havebeen studied for more than a decade. The first global emissioninventories were compiled by Pacyna and Pacyna and co-workers, yielding estimates of 2140 Mg for 1990 (1) and 1910Mg for 1995 (2, 3). Subsequent updates (4, 5) generated an

estimate of 2190 Mg for the year 2000. According to the mostrecent global inventory (4), about 65% of emissions camefrom stationary fuel combustion in 2000, 11% from goldproduction, 7% from nonferrous metals production, 6% fromcement production, and the rest from a variety of smallercontributing source types. Geographically, about 54% of theemissions came from Asia, 18% from Africa, 8% from Europe,and 7% from North America.

There have also been a number of studies of Hg emissionsfrom particular regions and countries, e.g., Australia (6), China(7, 8), Europe (9), the Mediterranean (10), North America(11, 12), and South Africa (13), as well as from particularsource types such as artisanal gold mining (14), petroleumproducts (15), and biomass burning (16-19). In addition,inferences about the magnitude of global (20, 21) and regional(22, 23) Hg emissions have been made from chemicaltransport modeling studies constrained by field measure-ments. These special studies have sometimes confirmed andsometimes disputed the values in the global inventories, butin all cases they have expanded our knowledge base aboutthe many sources of Hg. Unification and reconciliation ofcurrent thinking about anthropogenic Hg emissions hasrecently been sponsored by the United Nations EnvironmentProgramme (24).

There have been no published estimates to date of futureglobal Hg emissions. We do not know whether to expectglobal Hg emissions to increase or decrease, nor do we knowhow the spatial patterns of releases might change over time.The main reason for this is the difficulty in matching upcurrently available projections of the future use of energy,fuels, and materials with currently available Hg emissioncalculation schemes. This paper remedies the situation bydeveloping global anthropogenic Hg emission estimates forthe year 2050 under four alternative IPCC scenarios of futuredevelopment.

2. MethodologyIn previous work we studied future patterns of energy andfuel use under alternative IPCC scenarios and used them toestimate present and future emissions of the carbonaceousaerosols, black carbon (BC) and organic carbon (OC) (25).We also reported estimates of Hg emissions in the presentand the future for one of the most difficult of regions tocharacterize, China (7, 8). In this work we marry these twoschemes to project global Hg emissions in 2050 underalternative IPCC scenarios of development. The first step inthe process is to develop a global Hg inventory for 1996, thebase year of our BC/OC inventory (25). The BC/OC inventoryconsists of a detailed accounting of combustion-relatedenergy and fuel use, disaggregated among many differentsector/fuel/technology options for 17 world regions. Theoriginal list of 112 combustion options has been presentedpreviously (25); at the time of the commencement of this Hgwork, the model consisted of 144 distinct combustion options,the previous list having been augmented by the addition ofsome new categories like municipal waste combustion, openwaste burning, liquefied petroleum gas, etc. For the purposeof calculating Hg emissions, we have added to this combus-tion set the eight industrial process categories that weconsider the most significant for Hg release: pig ironmanufacture, cement manufacture, copper smelting, leadsmelting, zinc smelting, artisanal gold extraction, Hg mining,and caustic soda production. Other industrial process sour-ces are not considered at this time.

Four types of open biomass burning are included: tropicalforests, extra-tropical forests, savanna/grassland, and crop

* Corresponding author tel: 630-252-3448; fax: 630-252-6500;e-mail: dstreets@anl.gov.

† Argonne National Laboratory.‡ Tsinghua University.

Environ. Sci. Technol. 2009, 43, 2983–2988

10.1021/es802474j CCC: $40.75 2009 American Chemical Society VOL. 43, NO. 8, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2983

Published on Web 03/04/2009

residue burning in fields. Following precisely the samemethod as we used for BC and OC emissions (25), we takeour forecasts of the amounts of open biomass burned (asdistinct from biofuels) directly from IPCC forecasts (26-28);therefore, they are scenario-specific and consistent with theamounts of fossil fuels and biofuels burned. For biomassburning, the IPCC considers only those emissions that aredirectly associated with human activities. For example, forestburning is specifically associated with the need to expandagricultural land at the expense of forested land to provideadditional food supplies (managed forest in IPCC terminol-ogy). The IPCC forecasts do not include the contributionfrom natural wildfires. We use the IMAGE managed forestprojections directly and supplement them with estimates ofwildfire emissions using the mature forest projections of theIPCC. Changes in mature forest area over time are assumedto be proportional to changes in wildfire emissions. The IPCCprojections of grassland and crop residue burning are useddirectly. The total amounts of biomass burned globallydecline from 6.02 Gt in 1996 to 5.71 Gt in 2050 A1B, 4.30 Gtin 2050 A2, 4.47 Gt in 2050 B1, and 4.18 Gt in 2050 B2 (seebelow for scenario descriptions).

Using the 1996 energy and fuels data set, we thenreparameterize the model to yield Hg emissions, rather thanBC/OC emissions. This requires only minor modificationsto the model structure, e.g., to calculate the Hg content offuels and to add Hg removal by flue-gas desulfurization (FGD)units. Mostly it requires new data on Hg-relevant parameterssuch as initial release rate, ash retention efficiency, captureefficiency, etc., for each element of the [(144+8) × 17] sourcematrix. In doing this we make use of data from our Chinainventory (7, 8), the global compilations (2-5), and U.S. EPApublications (29, 30). Table 1 summarizes our assumptionsabout present-day Hg emission factors. This proceduregenerates detailed regional Hg emission estimates in theformat of our forecasting model (25). As described in thenext section, our 1996 results compare well with otheremission estimates, especially when recent new estimatesfor Australia (6) and South Africa (13) are incorporated (24).The second step of the process is to update the 1996 Hginventory to 2006 to give a more current representation ofemissions and to make the data congruent with inventoriesbeing used in the latest Hg modeling studies. We thereforereplace the 1996 energy and fuels data with data for 2006 andincorporate developments in Hg capture, e.g., the penetrationof FGD units between 1996 and 2006.

The final step is to calculate future Hg emissions. Weproject the 2006 inventory forward to 2050 using the sameprocedure reported earlier for BC/OC emissions (25). Weuse the IPCC forecasts developed by the IMAGE group atRIVM, the National Institute for Public Health and theEnvironment in The Netherlands (26). We examine four ofthe scenarios developed by the IPCC for its Third AssessmentReport: A1B, A2, B1, and B2 (27, 28). These scenarios specifythe regional energy use and fuel mix associated withalternative future pathways of human development, whichcan be directly used in the calculation of future emissions.The scenarios also embody different rates of technologydevelopment and technology transfer among countries.

The A1B scenario is characterized by rapid energy andeconomic growth, low population growth, continued glo-balization in its western form, increasing privatization, andthe rapid introduction of new and more efficient technologies.The designation “B” in this scenario connotes a balance ofenergy technologies across all source types. The A2 scenarioreflects a heterogeneous world of self-reliance and preserva-tion of local identities. Population growth is high. Techno-logical change is slow and fragmented. The B1 scenario ischaracterized by high economic growth, low populationgrowth, and continuing globalizationssimilar to A1B. How-ever, two major differences from A1B are a change ineconomic structure toward a service and informationeconomy and an emphasis on global solutions to socialand environmental problems with the introduction of cleanand resource-efficient technologies. Finally, the B2 scenariodescribes a future world in which the emphasis is on localsolutions to economic, social, and environmental problems.Population growth and economic growth are moderate. Thereis less rapid and more diverse technological change than inA1B and B1.

The rates of adoption of FGD technology by 2050 on coal-fired power plants follow IMAGE assumptions by region andscenario, as shown in Table 2. The difficult parameterizationto make is the rate of Hg removal that will be achieved bysuch systems by 2050. For our base-case assumptions, wehave followed the review by Brown et al. (31). We assumethat in the A1B and A2 scenarios, no significant advance ismade over reported levels in the U.S. of 40% (30). For the B1and B2 scenarios, we assume that Hg reduction efficiencyimproves to 70% in the developed world (North America,OECD Europe, and Japan) and to 51.6%sthe more optimisticview of current reduction measurements at a number of U.S.power plants (31)sin the rest of the world. Because of theuncertainty in this parameter, we also examine a range ofalternative values of Hg removal, including the introductionof advanced sorbent technologies. For the eight industrialprocess categories, we use the IMAGE industrial process SO2

forecasts as a guide to project Hg emission growth rates.

TABLE 1. Emission Factors Used for Mercury Emissions in1996 and 2006

emitting source type unitemission

factora source

coal combustionpower plants g/t fuel 0.04-0.3 (2, 4)b

industrial, residentialand commercial boilers g/t fuel 0.1-0.5 (2, 4)b

oil combustion g/t fuel 0.058 (29)biofuel combustion g/t fuel 0.02 (17)biomass burningcrop residue g/t biomass 0.037 (18)extra-tropical forest g/t biomass 0.113 (18)savanna g/t biomass 0.080 (18)tropical forest g/t biomass 0.113 (18)pig iron manufacture g/t product 0.04 (2, 4)cement manufacture g/t product 0.065 (30)copper smelting g/t product 5-6 (2, 4)lead smelting g/t product 3 (2, 4)zinc smelting g/t product 7-80 (2, 4, 7)

a Ranges indicate values for different countries, basedon level of development. b Calculated from mercury con-tent of coals.

TABLE 2. Degree of FGD Penetration in Coal-Fired PowerPlants in 2050 by Scenarioa

A1B A2 B1 B2

Region Ab 0.95 0.95 0.95 0.95Region Bc 0.6-0.85 0.4-0.8 0.95 0.5-0.95Region Cd 0.1-0.2 0-0.05 0.6 0-0.05

a Ranges indicate values for different countries withinthese regions, based on level of development. Source: ref(26). b Region A: Canada, U.S., OECD Europe, Japan, andAustralia. c Region B: Central America, South America,Eastern Europe, Former USSR, and Asia. d Region C:Africa.

2984 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 8, 2009

3. Results and DiscussionThe results of the 1996 and 2006 model calculations are shownin Table 3 and Figure 1. We estimate that global mercuryemissions in 1996 were 2128 Mg, coming mainly frombiomass burning (614 Mg, 28% of total), power-plant fuel

combustion (321 Mg, 15%), artisanal gold extraction (300Mg, 14%), industrial fuel combustion (266 Mg, 13%), zincsmelting (191 Mg, 9%), and residential fuel combustion (106Mg, 5%). By 2006, global emissions had risen to 2480 Mg, anincrease of 17%. The shares of contributing sources had alsochanged, reflecting a large increase in power sector andindustrial emissions. Biomass burning had fallen to 586 Mg(25%), power-plant fuel combustion had risen sharply to 459Mg (18%), and zinc smelting emissions had risen dramaticallyto 384 Mg (15%). Without good information on the timevariation of emissions associated with artisanal gold extrac-tion, we have assumed they remained constant at 300 Mg(12%). Emissions from industrial fuel combustion rose to278 Mg (11%). Emissions from residential fuel combustiondeclined, because there was less use of coal for domesticheating and cooking in China and other developing countries.These results reveal the growing roles of coal combustion for

TABLE 3. Mercury Emissions by Sector and World Region in 1996 and 2006 (Mg/yr)

emitting sector North America Central and South America Africa Europe, Russia, Middle East Asia and Oceania world

1996residential 3.8 3.0 9.7 21.5 68.3 106.3industry 15.9 10.9 5.8 64.8 169.1 266.5power 55.9 4.8 15.8 106.8 138.2 321.4transportation 20.7 4.8 1.9 18.5 13.3 59.2biomass burning 28.6 156.5 252.7 49.1 127.2 614.1industrial processes 66.4 105.6 43.0 98.2 459.1 760.5total 179.4 285.6 328.9 358.9 975.2 2128.0

2006residential 3.3 3.5 11.6 16.7 37.0 72.1industry 15.1 14.8 7.8 61.8 179.0 278.5power 63.1 8.2 22.6 123.5 241.1 458.6transportation 24.5 7.3 2.9 23.7 17.8 76.1biomass burning 28.7 146.9 229.0 49.2 132.2 585.9industrial processes 43.6 110.1 46.4 97.7 710.6 1008.4total 178.3 290.8 320.3 372.6 1317.7 2479.6

other estimates1995 (2)a 213.5 59.2 210.6 249.7 1179.8 1912.82000 (4)a 145.8 92.1 398.4 247.8 1305.9 2189.9“most recent” (24)a 144.7 159.7 21.8b 277.8c 1041.6d 1645.6e

a Estimates do not include biomass burning. UNEP best estimate for Hg emissions from global biomass burning is 675Mg/yr (24). b South Africa only. c Europe and Russia only. d China, India, and Australia only. e Includes only major worldregions and countries; therefore, not globally complete.

FIGURE 1. Distributions of global Hg emissions among major contributing source types in 1996 and 2006.

TABLE 4. Mercury Emissions in 2050 by Scenario and WorldRegion (Mg/yr)

scenarioNorth

America

Centraland SouthAmerica Africa

Europe,Russia,

Middle EastAsia andOceania world

2050 A1B 225.9 473.6 509.6 676.5 2970.0 4855.6

2050 A2 239.1 415.6 375.5 667.3 2208.5 3905.9

2050 B1 121.9 340.4 357.0 358.1 1208.9 2386.2

2050 B2 131.3 331.2 308.1 398.0 1461.4 2629.9

VOL. 43, NO. 8, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2985

electricity generation and of nonferrous metals productionin East Asia (China). In other world regions growth was moremoderate in this period (e.g., South Asia), or environmentalregulations and technology renewal held Hg emissions incheck (e.g., OECD Europe, U.S., and Japan).

By 2050 global energy use will increase dramatically overpresent-day levels. Primary energy use increases by a factorof 3.6 under the A1B scenario, 2.6 under A2, 2.3 under B2,and 2.1 under B1. This does not mean that Hg emissions willincrease by these same amounts, however; the penetrationof improved technologies and fuels, sectoral transformations,and socioeconomic development all combine to moderatethe growth of Hg emissions. In addition, the adoption of newemission control technologies for other speciessparticularlyFGD for SO2 controlshas a mitigating effect on Hg emissionsas well. Table 4 summarizes 2050 Hg emissions under eachscenario.

We estimate that global Hg emissions will grow from 2480Mg in 2006 to 4856 Mg in 2050 under the A1B scenario, to3906 Mg under A2, and to 2630 Mg under B2. Under the B1scenario, Hg emissions decline slightly to 2386 Mg. Thechange relative to 2006 levels ranges, therefore, from +96%to -4%. Developments in Asia largely dictate the variationacross scenarios. Rapid expansion of coal-fired electricity

generation in Asia is the major determinant of the growth,as shown in Figure 2, where the share of coal-fired powerplants in global Hg emissions could rise to almost 50%.

The increase in Hg emissions from coal combustionis difficult to mitigate. Though rapid deployment of FGD ispartially successful in the environmentally strong scenarioslike B1 and B2, the limited ability of FGD to remove Hg inhibitsthe effectiveness of emission control. Figure 3 shows the fullrange of emissions, present and future, by major emittingcategory. It also shows the sensitivity of total Hg emissionsto the efficiency of Hg removal by FGD in the range of30-100% removal, for the highest (A1B) and lowest (B1)scenarios. The range of total emissions covered by thissensitivity analysis is 1580-5110 Mg, or a range from areduction in total Hg emissions of 36% to an increase of106%. This sensitivity analysis shows how important it willbe to develop ways to improve the Hg removal efficiency ofcoal-fired power plants in future developing-country ap-plications. And though the effect on absolute emission levelsis greatest in the developing world, Figure 4 shows that,relative to total emission levels in each region, the greatestbenefits to improving Hg removal efficiency in coal-firedpower plants are obtained in the U.S. and OECD Europewhere electricity generation comprises the largest share of

FIGURE 2. Distributions of global Hg emissions among major contributing source types in the 2050 A1B and 2050 B1 scenarios.

FIGURE 3. Comparison of global Hg emissions for each of the six cases, by major contributing sector, showing sensitivity of the 2050A1B and 2050 B1 cases to alternative values of the Hg removal efficiency of FGD in the ranges of 30-100% and 50-100%,respectively.

2986 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 8, 2009

total emissions. In these two regions, improving Hg removalefficiency to the extreme value of 100% reduces total Hgemission levels in 2050 by almost 70%.

At present, a number of add-on sorbent injection tech-nologies are under development with the specific objectiveof capturing Hg released during coal combustion. Onepromising technique is Activated Carbon Injection (ACI), inwhich powdered activated carbon is injected into the flue-gas stream before the particulate matter control device inorder to bind the Hg and allow it to be captured (32).Demonstrations of ACI suggest that a performance of 95%Hg removal may be achievable, though actual removalefficiencies in some of the tests were significantly less than95% (33). If we assume that all coal-fired power plants employACI at 95% removal in 2050, then the range of Hg emissionsover the four scenarios would be 1670-3480 Mg, or a changerelative to 2006 levels of -33% to +40%. However, becauseACI has not yet reached the stage of commercial deployment,it would be inappropriate to include it in the main forecasts.

Finally, we examine the effect of these future scenarioson relative emissions of the different species of Hg, as thiswill have an important bearing on the trends for long-rangetransport vs local Hg deposition. Applying the source-specificHg speciation profiles listed in Table 6 of Streets et al. (7) toeach of the 152 source categories, we obtain speciatedemissions of Hg0, divalent Hg (Hg2+), and particulate Hg(Hgp) for each emissions case. The results are illustrated inFigure 5. For 1996 the shares of primary emitted species atglobal scale were: 67% Hg0, 25% Hg2+, and 7% Hgp. We seethat by 2006 the share of elemental Hg has already begun todecline slightly. By 2050 we see a large decline in the shareof elemental Hg and an increase in the share of divalent Hg,though there is some variation across scenarios. Shares ofHgp remain relatively constant at 5-7%. Under 2050 A1B wefind the following shares: 47% Hg0 and 49% Hg2+. Under2050 B1, the shares are 56% Hg0 and 40% Hg2+. This shiftfrom Hg0 to Hg2+ is largely driven by the adoption ofpollution control technology on coal-fired installations. Theimplications are clear: there will be a relative trend towardreduced long-range transport and enhanced local depositionof Hg. In absolute terms, the change in emissions from 2006to 2050 for Hg0 is from -19% to +38% (also shown in Figure5), whereas for Hg2+ the range is from +38% to +243%.

The forecasting of future emissions is a challenging task.There are significant uncertainties in each of the stepsinvolved and some of them are hard to quantify. In this workwe have utilized the IPCC socioeconomic scenarios and their

associated energy and fuels requirements, imposed on themthe technology characteristics that affect emissions of Hg,and used our own parametrizations of Hg emission rates.The uncertainty in estimates of present-day global Hgemissions is believed to be on the order of ( 25-30% (4, 5).Consistent with this, we have estimated the uncertainty ofestimates of Hg emissions in China at ( 44% (7) withuncertainties as high as a factor of 3-4 for some industrialprocesses. The uncertainty in estimates of Hg emissions isperhaps less than might be expected. This is because thequantity of Hg (just like sulfur) in the original fuels and oresconstrains the final emission estimate. Uncertainties arehigher for species that are mostly formed in the combustionzone and therefore sensitive to combustion conditions (e.g.,CO, NOx, BC). Future developments in complex humansystems are inherently unpredictable (27). Two significantsources of uncertainty in this regard are the evolution of theunderlying driving forces and the nature of mitigation policiesthat governments might adopt in response to environmentaldegradation. Thus, the scenario approach is the commonlyused way to bound the range of possible futures. However,we can be sure that the uncertainty in the estimates of futureemissions along any of the trajectories is greater than the (25-30% estimate for current emissions.

In this work we have chosen four IPCC scenarios thateffectively bound future conditions. The A1B scenario,reflecting as it does unfettered economic growth with nospecial attention to environmental protection, represents anupper bound on emissions, while the B1 scenario, with itsemphasis on global solutions to environmental problems,represents a lower bound. Using these brackets, we can saythat Hg emissions will likely be higher in 2050 than in 2006.The worst case would be roughly a doubling in emissionsfrom 2480 Mg to 4860 Mg. The best case would be a reductionof just 4% under base-case conditions. This could beimproved to 33% by full application of ACI or to 36% bycomplete elimination of Hg emissions from coal-fired powerplants. Coal-fired power plants, we feel, are the key target forHg emission control. With a great expansion of electricitydemand throughout the developing world, much of whichis likely to be fueled by coal (according to the IPCC scenarios),this source category dominates Hg emissions in 2050. Thoughflue-gas desulfurization for SO2 control will achieve somereduction in Hg emissions as a cobenefit, it is insufficient.We recommend the urgent development of specific tech-niques for Hg control, such as ACI, before these new powerplants are built.

AcknowledgmentsThis work was funded by the U.S. Environmental ProtectionAgency’s STAR Program on the Consequences of GlobalChange for Air Quality, as part of collaboration with Harvard

FIGURE 4. Sensitivity of total Hg emissions under the 2050 A1Bscenario in eight regions and the world to Hg removal efficiencyof FGD in the range 30-100%, indexed to the base 2050 A1Bassumption of 40% removal. Note that the lines for the U.S. andOECD Europe are essentially superimposed.

FIGURE 5. Variation of species fractions in each of the sixcases (left axis, lines) and absolute values of Hg0 emissions(right axis, bars).

VOL. 43, NO. 8, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 2987

University. We are grateful for the support and collegial adviceof Darrell Winner (U.S. EPA) and Daniel Jacob (HarvardUniversity). The submitted manuscript has been created byUChicago Argonne, LLC, Operator of Argonne NationalLaboratory (“Argonne”). Argonne, a U.S. Department ofEnergy Office of Science laboratory, is operated underContract DE-AC02-06CH11357.

Literature Cited(1) Pacyna, J. M.; Pacyna, E. G. Global Emissions of Mercury to the

Atmosphere. Emissions from Anthropogenic Sources; ArcticMonitoring and Assessment Programme (AMAP): Oslo, Norway,1996.

(2) Pacyna, E. G.; Pacyna, J. M. Global Emissions of Mercury fromAnthropogenic Sources in 1995. Water Air Soil Pollut. 2002,137, 149–165.

(3) Pacyna, J. M.; Pacyna, E. G.; Steenhuisen, F.; Wilson, S. Mapping1995 Global Anthropogenic Emissions of Mercury. Atmos.Environ. 2003, 37 (S1), 109–117.

(4) Pacyna, E. G.; Pacyna, J. M.; Steenhuisen, F.; Wilson, S. GlobalAnthropogenic Mercury Emission Inventory for 2000. Atmos.Environ. 2006, 40, 4048–4063.

(5) Wilson, S. J.; Steenhuisen, F.; Pacyna, J. M.; Pacyna, E. G. Mappingthe Spatial Distribution of Global Anthropogenic MercuryAtmospheric Emission Inventories. Atmos. Environ. 2006, 40,4621–4632.

(6) Nelson, P. F. Atmospheric Emissions of Mercury from AustralianPoint Sources. Atmos. Environ. 2007, 41, 1717–1724.

(7) Streets, D. G.; Hao, J.; Wu, Y.; Jiang, J.; Chan, M.; Tian, H.; Feng,X. Anthropogenic Mercury Emissions in China. Atmos. Environ.2005, 39, 7789–7806.

(8) Wu, Y.; Wang, S.; Streets, D. G.; Hao, J.; Chan, M.; Jiang, J. Trendsin Anthropogenic Mercury Emissions in China from 1995 to2003. Environ. Sci. Technol. 2006, 40, 5312–5318.

(9) Pacyna, E. G.; Pacyna, J. M.; Pirrone, N. European Emissions ofAtmospheric Mercury from Anthropogenic Sources in 1995.Atmos. Environ. 2001, 35, 2987–2996.

(10) Pirrone, N.; Costa, P.; Pacyna, J. M.; Ferrara, R. MercuryEmissions to the Atmosphere from Natural and AnthropogenicSources in the Mediterranean Region. Atmos. Environ. 2001,35, 2997–3006.

(11) Pai, P.; Heisler, S.; Joshi, A. An Emissions Inventory for RegionalAtmospheric Modeling of Mercury. Water Air Soil Pollut. 1998,101, 289–308.

(12) Gbor, P. K.; Wen, D.; Meng, F.; Yang, F.; Sloan, J. J. Modelingof Mercury Emission, Transport and Deposition in NorthAmerica. Atmos. Environ. 2007, 41, 1135–1149.

(13) Dabrowski, J. M.; Ashton, P. J.; Murray, K.; Leaner, J. J.; Mason,R. P. Anthropogenic Mercury Emissions in South Africa: CoalCombustion in Power Plants. Atmos. Environ. 2008, 42, 6620–6626.

(14) Telmer, K.; Veiga, M. World Emissions of Mercury from SmallScale Artisanal Gold Mining and the Knowledge Gaps aboutThem. In Mercury Fate and Transport in the Global Atmosphere:Measurements, Models and Policy Implications; Pirrone, N.,Mason, R., Eds.; United Nations Environment Programme:Geneva, Switzerland, 2008.

(15) Wilhelm, S. M. Estimate of Mercury Emissions to the Atmospherefrom Petroleum. Environ. Sci. Technol. 2001, 35, 4704–4710.

(16) Friedli, H. R.; Radke, L. F.; Lu, J. Y. Mercury in Smoke fromBiomass Fires. Geophys. Res. Lett. 2001, 28, 3223–3226.

(17) Friedli, H. R.; Radke, L. F.; Lu, J. Y.; Banic, C. M.; Leaitch, W. R.;MacPherson, J. I. Mercury Emissions from Burning of Biomassfrom Temperate North American Forests: Laboratory and

Airborne Measurements. Atmos. Environ. 2003, 37, 253-267.

(18) Friedli, H. R.; Radke, L. F.; Prescott, R.; Hobbs, P. V.; Sinha, P.Mercury Emissions from the August 2001 Wildfires in Wash-ington State and an Agricultural Waste Fire in Oregon andAtmospheric Mercury Budget Estimates. Global Biogeochem.Cycles 2003, 17, 1039. doi:10.1029/2002GB001972.

(19) Cinnirella, S.; Pirrone, N. Spatial and Temporal Distribution ofMercury Emissions from Forest Fires in Mediterranean Regionand Russian Federation. Atmos. Environ. 2006, 40, 7346–7361.

(20) Selin, N. E.; Jacob, D. J.; Park, R. J.; Yantosca, R. M.; Strode, S.;Jaegle, L.; Jaffe, D. Chemical Cycling and Deposition ofAtmospheric Mercury: Global Constraints from Observations.J. Geophys. Res. 2007, 112, D02308, doi:10.1029/2006JD007450.

(21) Selin, N. E.; Jacob, D. J.; Yantosca, R. M.; Strode, S.; Jaegle, L.;Sutherland, E. M. Global 3-D Land-Ocean-Atmosphere Modelfor Mercury: Present-Day Versus Preindustrial Cycles andAnthropogenic Enrichment Factors for Deposition. GlobalBiogeochem. Cycles 2008, 22, GB2011, doi10.1029/2007GB003040.

(22) Jaffe, D.; Prestbo, E.; Swartzendruber, P.; Weiss-Penzias, P.; Kato,S.; Takami, A.; Hatakeyama, S.; Kajii, Y. Export of AtmosphericMercury from Asia. Atmos. Environ. 2005, 39, 3029–3038.

(23) Weiss-Penzias, P.; Jaffe, D.; Swartzendruber, P.; Hafner, W.;Chand, D.; Prestbo, E. Quantifying Asian and Biomass BurningSources of Mercury Using the Hg/CO Ratio in Pollution PlumesObserved at the Mount Bachelor Observatory. Atmos. Environ.2007, 41, 4366–4379.

(24) United Nations Environment Programme. Mercury Fate andTransport in the Global Atmosphere: Measurements, Models andPolicy Implications; Pirrone, N., Mason, R., Eds.; UNEP: Geneva,Switzerland, 2008.

(25) Streets, D. G.; Bond, T. C.; Lee, T.; Jang, C. On the Future ofCarbonaceous Aerosols. J. Geophys. Res. 2004, 109, D24212,doi:10.1029/2004JD004902.

(26) RIVM. The IMAGE 2.2 Implementation of the SRES Scenarios: AComprehensive Analysis of Emissions, Climate Change andImpacts in the 21st Century; RIVM CD-ROM publication481508018; National Institute for Public Health and the Envi-ronment (RIVM): Bilthoven, The Netherlands, 2001.

(27) Nakicenovic, N., et al. Emissions Scenarios: A Special Report ofWorking Group III of the Intergovernmental Panel on ClimateChange; Cambridge University Press: Cambridge, UK, 2000.

(28) Intergovernmental Panel on Climate Change (IPCC). ClimateChange 2001: The Scientific Basis. Contribution of Working GroupI to the Third Assessment Report of the Intergovernmental Panelon Climate Change; Houghton, J. T., et al., Eds.; CambridgeUniversity Press: UK and New York, 2001.

(29) U.S. Environmental Protection Agency. Compilation of AirPollutant Emission Factors, AP-42, Fifth Edition, Vol. I: StationaryPoint and Area Sources; EPA: Washington, DC, 1995.

(30) U.S. Environmental Protection Agency. Mercury Study Reportto Congress, Vol. II: An Inventory of Anthropogenic Emissions inthe United States; EPA-452/R-97-004; EPA: Washington, DC,1997.

(31) Brown, T. D.; Smith, D. N.; Hargis, R. A.; O’Dowd, W. J. MercuryMeasurement and Its Control: What We Know, Have Learned,and Need to Further Investigate. J. Air Waste Manage. Assoc.1999, 1–97; 29th Annual Critical Review.

(32) U.S. Environmental Protection Agency. Controlling Power PlantEmissions: Mercury-Specific Activated Carbon Injection (ACI);available at http://www.epa.gov/hg/control_emissions/tech_merc_specific.htm.

(33) U.S. Environmental Protection Agency. Control of MercuryEmissions from Coal-Fired Electric Utility Boilers; available athttp://www.epa.gov/ttn/atw/utility/hgwhitepaperfinal.pdf.

ES802474J

2988 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 8, 2009